ABSTRACT

Biological invasions are a global threat to aquatic biodiversity. Of particular concern are invasive freshwater fishes because they have high establishment rates, and introductions can result in the displacement and extirpation of native species through a range of processes including competition and hybridization. Though it is well known that invasive freshwater fishes commonly spread following introduction events, little is known about how fast and far they may move. Additionally, observations of hybridization involving invasive stream fishes have been linked to elevated turbidity; however, the extent to which impaired water clarity influences reproductive isolation among invasive and native species remains poorly understood. To better understand how invasive freshwater fishes disperse, and how turbidity affects reproductive isolation between native and non-native species, I carried out a series of three related studies. First,

I evaluated genetic variation across the native and invasive ranges of red shiner

(Cyprinella lutrensis), throughout the United States. Second, I characterized genetic variation and clinal stability across a hybrid swarm involving native blacktail shiner

(Cyprinella venusta stigmatura) and invasive red shiner in the Upper Coosa River Basin

(UCRB), USA. Third, I examined whether turbidity influences pre-mating social interactions between invasive red shiner and native blacktail shiner. MtDNA haplotypes from native range populations of red shiner form four divergent lineages and suggest that introduced populations in the western and eastern US originate from dissimilar genetic

lineages. I also recovered a previously undescribed lineage of Cyprinella that has been cryptically introduced into the western US. Examination of the hybrid swarm in the

UCRB revealed that the proportion of hybrids increased between 2005 and 2011, and that the hybrid swarm is continuing to expand both upstream and downstream. Under turbid conditions, I found that pre-mating social interactions increased, and that native blacktail shiner females are especially likely to interact with invasive red shiner males. Localized control or removal may be effective in managing non-native red shiner; further monitoring, however, is needed to help identify additional factors contributing to hybrid swarm movement. Furthermore, integrating knowledge of species behavior into management planning could help deter the further establishment and spread of invasive red shiner.

ACKNOWLEDGEMENTS

Many people contributed to the success of this dissertation. First, I wish to thank my advisor, Michael J. Blum for all of the opportunities provided, assistance given, and for his unwavering support. I wish to also thank my other dissertation committee members - Jordan Karubian, Cori Richards-Zawacki, and David Walters, for their comments and suggestions that improved this work. During my tenure at Tulane I was very fortunate to have worked with many great people. However, I owe a very special thanks to Fernando Alda and Jessica Ward; who both mentored me throughout my dissertation, taught me many skills, and provided countless hours of assistance. I would like to thank Erick Gagne and Travis Haas for their friendship, encouragement, and support. I would also like to thank Sabrina Hunter for assisting with molecular data collection, as well as the following people for all of their assistance in the lab and in the field: Fernanda Sa, Brandon Policky, Lee Attaway, Brittany Bernick, and Graham

Derryberry. Finally, I would like to thank my loving wife, Ashley, for all of her support and encouragement. Ashley and I met towards the end of this long journey, and she deserves extra thanks for supporting me through some of my toughest days. I would also like to thank my loving family, Ron, Cheryl, and Paul Glotzbecker, who believed in me every step of the way. And to my brother, Paul, thanks for frequently visiting me in New

Orleans.

ii

TABLE OF CONTENTS

ACKNOWLEDGEMENTS ...... ii

LIST OF TABLES ...... iv

LIST OF FIGURES ...... vi

Chapter

1. GEOGRAPHIC INDEPENDENCE AND PHYLOGENETIC DIVERSITY OF

RED SHINER INTRODUCTIONS ...... 1

2. RAPID MOVEMENT AND INSTABILITY OF AN INVASIVE HYBRID

SWARM ...... 39

3. TURBIDITY ALTERS PRE-MATING SOCIAL INTERACTIONS BETWEEN

NATIVE AND INVASIVE STREAM FISHES ...... 70

LIST OF REFERENCES ...... 94

iii

LIST OF TABLES

Table 1-1. Collection records of first occurrences of non-native red shiner ...... 28

Table 1-2. Locations and genetic diversity estimates for 46 populations of Cyprinella

lutrensis across native and invasive ranges ...... 29

Table 1-3. Additional Cyprinella sequence data used in genetic analysis ...... 31

Table 2-1. Summary data for 47 locations where Cyprinella were sampled in the Upper

Coosa River Basin...... 61

Table 2-2. Comparison of genetic and phenotypic clines models across the C. lutrensis x

C. venusta hybrid swarm from 2005 to 2011 ...... 62

Table 3-1. GEE models tests examining the effects of species and treatment on the

number of approaches made by male and female subjects towards opposite-sex

subjects in mixed-species aggregations ...... 89

iv

Table 3-2. GEE model tests examining the effects of species and treatment on the total

time (s) spent by male and female subjects associating with opposite-sex subjects

in mixed-species aggregations ...... 90

v

LIST OF FIGURES

Figure 1-1. ABC model scenarios of colonization for non-native red shiner populations

across the US...... 32

Figure 1-2. Phylogeny reconstruction of cytochrome b haplotypes using maximum

likelihood analysis ...... 33

Figure 1-3. Statistical-parsimony haplotype frequency network of cytochrome b for

Cyprinella lutrensis samples across US ...... 35

Figure 1-4. ABC analysis invasion scenarios for western and eastern US ...... 37

Figure 1-5. ABC analysis for within invasive western and eastern ranges ...... 38

Figure 2-1. 47 locations sampled between 2005 and 2011, along a 477-km transect of the

Coosa River, Oostanaula River, and Conasauga River (Alabama, Georgia,

Tennessee; USA) ...... 63

vi

Figure 2-2. Proportion of parental Cyprinella venusta (BT), Cyprinella lutrensis (RS),

and hybrid (F1, F2, BCbt, and BCr) individuals at 47 sites across sampling

transect in: A: 2005, B: 2008, and C: 2011...... 64

Figure 2-3. Clinal changes in frequencies of Cyprinella venusta phenotype, mtDNA

haplotypes, and microsatellite multilocus genotype between 2005 and 2011. A:

phenotype cline models, B: mtDNA haplotype cline models, C: multilocus

genotype cline models...... 66

Figure 2-4. Clinal changes in frequencies of Cyprinella venusta phenotype, mtDNA

haplotypes, and microsatellite multilocus genotype. A: 2005 combined cline

models, B: 2008 combined cline models, C: 2011 combined cline models ...... 68

Figure 3-1. A diagram of the three-chambered experimental tank ...... 91

Figure 3-2. Mean (± sem) behavioral responses of male blacktail shiners (A, C) and male

red shiners (B, D) towards conspecific and heterospecific females under clear and

turbid visual conditions ...... 92

Figure 3-3. Mean (± sem) behavioral responses of female blacktail shiners (A, C) and

female red shiners (B, D) towards conspecific and heterospecific males under

clear and turbid visual conditions ...... 93

vii

1

CHAPTER 1

GEOGRAPHIC INDEPENDENCE AND PHYLOGENETIC DIVERSITY

OF RED SHINER

ABSTRACT

The spread of non-native species can displace native biota and irreversibly alter community composition. Identifying areas at risk of invasion can be difficult, however, when the distribution of a non-native species encompasses geographically disjunct regions. Understanding genealogical relationships among native and non-native populations can clarify the origins and nature of fragmented distributions, which in turn can clarify how fast and far a non-native species may spread. We evaluated genetic variation across the native and invasive ranges of red shiner (Cyprinella lutrensis), a minnow known to displace and hybridize with native species, to reconstruct invasion pathways and better understand its spread across the United States (US). Examination of mitochondrial cytochrome-b variation found that haplotypes from native range populations of red shiner fall into four highly divergent lineages that likely warrant species recognition. The introduced red shiner populations in the eastern and western US are derived from only two of these lineages. Introduced populations in the western US originate from the mid-western and western genetic lineages, whereas eastern introductions derive only from the mid-western lineage. Introduced populations in the western US exhibited fewer, but more diverse haplotypes compared to introduced

2

populations in the eastern US. We also recovered a previously undescribed, divergent lineage of Cyprinella that has been cryptically introduced into the western US. This raises the possibility that hybridization has proceeded following secondary contact between previously allopatric lineages of red shiner. Approximate Bayesian Computation modeling suggests that the disjunct distribution of red shiner across North America is an agglomeration of independent regional invasions with distinct origins, rather than a stepwise advance of a unified invasion front or secondary introductions. Thus localized control or removal may be effective in managing non-native red shiner, including further spread to areas of conservation concern.

INTRODUCTION

Biological invasions are a global threat to aquatic biodiversity (Wilson 1992; Sala et al. 2000). Human transport and introduction of non-native species has contributed to widespread restructuring and biotic homogenization of aquatic communities (Moyle and

Leidy 1992; Allan and Flecker 1993; Olden et al. 2004). Of particular concern are non- native freshwater fishes, which have higher establishment rates than other taxa (Jeschke and Strayer, 2006). Non-native freshwater fishes also are of concern because introductions can result in the displacement and extirpation of native biota through a range of processes including predation (Witte et al. 1992), competition (Callaway and

Aschehoug 2000) and hybridization (Olden et al. 2004). Non-native fish introductions can devastate populations of at-risk native species and irrevocably restructure entire assemblages. For example, the introduction of non-native trout to historically fish-less, high elevation lakes has contributed to the decline of threatened amphibians in the Sierra

Nevada mountains in the western United States (US) and the introduction of predatory

3

Nile perch (Lates nilotica) precipitated the loss of nearly 200 endemic cichlids in Lake

Victoria (Kaufman 1992; Knapp and Matthews 2000).

Freshwater aquatic invasions, including the advance of predatory and highly aggressive fish species, are often a result of intentional or accidental introductions related to aquaculture and angling (Lever 1996; Moyle and Light 1996). Aquaculture and angling have been identified as primary sources for fish introductions worldwide, with

50% of cultivated freshwater fish species having become established in the wild (Casal

2006). For example, as the most commonly used bait fish in many regions, the golden shiner (Notemigonus crysoleucas) is now widespread across western North America (Dill and Cordone 1997; Pflieger 1997). The spread of three species of Asian carp

(Hypophthalmichthys nobilis, H. molitrix, Mylopharyngodon piceus) across the

Mississippi River Basin serves to illustrate how releases from aquaculture facilities can reconfigure the trophic structure of aquatic ecosystems (Chick and Pegg 2001). Though invasions can result in a cohesive distribution that reflects expansive spread of a unified invasive front, like the progression of Asian carp in the Mississippi River drainage, many non-native freshwater fishes exhibit patchy distributions (Townsend and Crowl 1991;

Leprieur et al. 2006). Despite being so prevalent, the origins and interactions that may occur across disjunct populations are not well understood, which in turn can complicate efforts to determine how fast and far non-native freshwater fish may spread.

Several processes may give rise to fragmented distributions of invasive species.

Multiple independent introductions from distinct sources within the native range can result in disjunct static populations with little to no interaction, though post-introduction interactions with native species can trigger unification through expansive spread. For

4

instance, the sheepshead minnow (Cyprinodon variegatus) was introduced to the Pecos

River drainage in western Texas and New Mexico via bait bucket releases between 1968 and 2000. Sheepshead minnow eventually hybridized with the imperiled Pecos pupfish

(Cyprinodon pecosensis), resulting in the spread of hybrids across more than half of the range of the Pecos pupfish in less than five years (Echelle and Connor 1989; Childs et al.

1996; Rosenfield et al. 2004). Disjunct populations can also arise from stochastic secondary spread following a single introduction event. For example, the spread of invasive European green crabs (Carcinus maenas) across the northeastern Pacific coast resulted from stochastic transport by climate-mediated currents following a single introduction event from a small founding population derived from the Atlantic coast of

North America (Darling et al. 2008, Tepolt et al. 2009). The northeastern Pacific coast range of invasive C. maenas ballooned in less than ten years with the establishment of disjunct populations isolated across distances as large as 1000 km (Tepolt et al. 2009).

Similarly, the progression of Asian topmouth gudgeon (Pseudorasbora parva)- which predates on eggs and transmits the parasite Sphaerothecum destruens to native fishes- across 32 European countries is likely the result of a single introduction into Romania or

Hungary followed by stochastic secondary spread (Gozlan et al. 2010, Simon et al. 2011).

Reconstructing genealogical relationships among non-native populations can clarify the origins and nature of fragmented distributions (Estoup and Guillemaud 2010).

For example, Blum et al. (2007) were able to validate historical evidence that smooth cordgrass (Spartina alterniflora) was introduced into Willapa Bay (WA) from multiple source populations, and found evidence suggesting S. alterniflora secondarily spread to a nearby embayment on the Pacific coast. Similarly, studies of the invasive European green

5

crab in the Canadian Maritimes revealed that at least five new lineages had been cryptically introduced to the area within a few decades, and differences in genetic diversity between Canadian and southern populations were best explained by multiple introduction events spanning a period of several decades (Roman 2006). Subsequent work (Blakeslee et al. 2010) determined that introduced populations in Newfoundland originated from Nova Scotia via counter-current spread, rather than directly from Europe.

Finally, studies of the western mosquito fish (Gambusia affinis) across the North Island of New Zealand demonstrated that the establishment and spread of G. affinis reflected a combination of localized dispersal and human-assisted colonization (Purcell et al. 2012;

Purcell and Stockwell 2015).

Understanding the relative contributions of different processes that give rise to disjunct distributions can help identify areas at risk of invasion as well as improve the selection and impact of management approaches. In this study, we evaluated genetic variation across the native and invasive ranges of red shiner (Cyprinella lutrensis) to infer relationships among invasive populations and pathways of spread across the United

States (US). Non-native red shiner populations are representative of the legacy of intentional and accidental introductions from aquaculture and angling in North America.

Historical collection data suggest that red shiner were introduced to the western US at least 20 years before being introduced to the eastern US (Table 1-1). However, it is not clear whether the emergence of non-native red shiner populations across the US is (1) an outcome of multiple, independent introductions from the native range; or (2) stepwise advance through successive secondary introductions from one part of the invasive range to another. We evaluated both possibilities by first examining mtDNA haplotype

6

variation across the native range to characterize sources that could have contributed to introduced populations. We then examined mtDNA haplotype variability among representative samples from the invasive range to infer the source and number of introductions across the eastern and western US. Finally, comparisons of native and invasive regions enabled us to infer the sequence and pathways of invasion, which in turn offered novel perspectives on risk and prevention of further spread to areas of conservation concern.

MATERIALS AND METHODS

Study species

Red shiner now occur in 11 states beyond native regions in North America (Fuller at al. 1999). With a native range encompassing Great Plains and Central Lowland tributaries of the Mississippi River and western Coastal Plain drainages of the Rio

Grande River, red shiner occur in the US from California to North Carolina. The species thrives under harsh conditions (e.g., low flow, high turbidity, poor water quality) and aggressively colonizes severely degraded habitats (Cross and Cavin1971; Matthews

1985; Matthews and Hill 1977, 1979). For example, introduced red shiner have become the most abundant species in degraded, urban streams in metropolitan Atlanta, Georgia

(Devivo and Freeman 1995). Due to high fecundity and aggressive behavior, introduced populations can spread rapidly, often displacing native species and hybridizing with native congeners (Hubbs and Strawn 1956; Minckley and Deacon 1968; Page and Smith

1970; Greger and Deacon 1988; Larimore and Bayley1996; Moyle 2002; Walters et al.

2008). Dill and Cordone (1997) referred to the red shiner as the second greatest threat to

7

the welfare of indigenous south-western fishes (after the mosquito fish, Gambusia affinis). Red shiner populations in the Moapa River and Virgin River (Nevada) have been implicated in the regional decline of native fish including spikedace (Meda fulgida), woundfin (Plagopterus argentissimus), and Virgin River chub (Gila seminuda; Moyle

1976; Deacon 1988; U.S. Fish and Wildlife Service 1990a, 1995). The decline of redside shiner (Richardsonius balteatus) also coincided with the rise of red shiner in the Green

River outside of Dinosaur National Monument (Utah; Holden and Stalnaker 1975).

Additionally, over 40% of congeners in the southeastern US, widely considered to be a global hotspot of aquatic biodiversity, are known to hybridize with introduced red shiner in the wild (C. analostana, C. camura, C. callitaenia, C. spiloptera, C. venusta cercostigma, C. v. stigmatura, C. v. venusta, and C. whipplei; Hubbs and Strawn 1956;

Page and Smith 1970; Wallace and Ramsey 1982; Johnson 1999; W.C. Starnes personal communication) or under laboratory conditions (C. caerulea, Burkhead unpublished data).

Time series collection records indicate that red shiner were first introduced to the western US in the Lower Colorado River in the 1950’s and then later introduced to the eastern US in the Lower Yadkin and Upper Coosa Rivers in the early 1970’s (Fuller at al.

1999; Walters et al. 2008; Table 1-1). Given the distance between these geographic regions, it is likely that non-native populations in the western and eastern US are a result of multiple independent introductions from the native range. Some survey records also suggest that the progression of red shiner invasions within drainage basins may be the result of multiple independent introduction events (Walters et al. 2008). However, it is possible that non-native populations originated from secondary introductions (e.g.,

8

Darling et al. 2008) as a consequence of human mediated bait bucket or aquarium releases deriving from the invasive range (Jennings and Saiki 1990; Walters et al. 2008).

Predictive models and inventory records suggest that red shiner could spread throughout the conterminous US (Poulos et al. 2012; Poulos and Chernoff 2014) as a consequence of rapid population growth, dispersal, and aggressive colonization (Hubbs and Lagler 1958; Minckley and Deacon 1968; Minckley 1973). Based on the premise that red shiner have broad environmental tolerances, forecasting models indicate that red shiner can potentially invade habitats with mean minimum water temperatures above freezing and mean maximum summer water temperatures up to 35ºC (Matthews 1977;

Poulos et al. 2012). According to some climate change projections, the range of native and non-native red shiner populations could increase by 42%. Expansion into the western

US, southeastern US, and much of Canada (Poulos and Chernoff 2014) could precipitate widespread declines in the abundance and distribution of other stream fishes (Gido et al.

1999; Marsh-Matthews and Matthews 2000; Douglas et al. 1994).

Native and invasive range collections

Between 2004 and 2010, specimens were obtained from 43 rivers (34 in the native range, 9 in the invasive range) across the US. Specimens of native red shiner were collected from the Mississippi River between Louisiana and Iowa. Specimens also were sampled from basins in Texas, New Mexico, Oklahoma, Kansas, Nebraska, Illinois, and

Arkansas (Table 1-2). Specimens of non-native red shiner were sampled from western

US populations in California, Nevada, and Arizona (Table 1-2). Non-native red shiner from the eastern US were collected in Alabama, Georgia, and North Carolina (Table 1-2).

Though we obtained samples representing all known clades and lineages of red shiner

9

within the native range, we were unable to obtain samples from non-native populations in the northern reaches of the Colorado River and Green River, as well as small populations recorded in Virginia and Massachusetts. On average, 10 individuals (SD=8.87) were sampled at each site via seine netting (Table 1-2). Once collected, whole fish were immediately preserved in 95% ethanol for genetic analysis.

DNA extraction and mtDNA sequencing

Genomic DNA was extracted from approximately 0.05 g of preserved fin tissue from each specimen using DNeasy kits (Qiagen, Inc., Valencia, CA). Approximately 10-

50 ηg of DNA was used as template for 15 µl polymerase chain reaction (PCR) mixtures that also included 2.5 mM MgCl2, 2.5 mM each dNTP, 0.5 units Taq DNA polymerase

(Invitrogen, Carlsbad, CA), 0.5 µM each of a pair of oligonucleotide primers and 1X

PCR buffer (Invitrogen, Carlsbad, CA). The complete cytochrome b (cyt b) gene

(1140bp) was amplified with primers GLU and THR as described in Schmidt et al.

(1998) under a thermal regime of 35 cycles of 94°C for 30 seconds, 49°C for 30 seconds, and 72°C for 90 seconds, followed by a final extension stage at 72ºC for 5 minutes with a

MJ Dyad thermocycler (MJ Research, Inc., Waltham, MA). PCR amplicons were purified with ExoSAP-IT (Affymetrix, Santa Clara, CA) following manufacturer protocols and served as templates in sequencing reactions utilizing ABI BigDye Terminator v.3.1 sequencing kits (Applied Biosystems, Inc., Foster City, CA). The sequencing reactions were run on an ABI 3730xl automated sequencer (Applied Biosystems, Inc., Foster City,

CA). Raw sequence files were edited, assembled and aligned with Sequencher 5.0 (Gene

Codes Corp., Ann Arbor, MI) in preparation for analysis. All sequences were translated into amino acids to check for stop codons, sequencing errors and misalignments. No

10

indels were detected in any of the sequences. All newly obtained sequences were deposited in GenBank under accession numbers KR004184-KR004610. Separate accession numbers for previously acquired sequence data (Broughton et al. 2011; Diver

2013) are provided in Table 1-3.

Genetic data analysis

Phylogenetic relationships among haplotypes recovered in the native and invasive range were inferred using Maximum Likelihood (ML) analysis with the package phangorn (Schliep 2011) in R v.2.15.1 (R Development Core Team 2009). ML analyses were implemented using the TN93+I+Γ substitution model, which was selected according to log-likelihood values and Akaike information criterion (AIC) values in model tests run with PhyML v. 3.0 (Guindon et al. 2010) and ape (Paradis et al. 2004) in

R. Node support was assessed from 500 bootstrap replicates. ML trees were rooted using cyt b sequences from Nocomis raneyi, N. micropogon, N. biguttatis and N. asper

(Accession Nos. GQ275147, GQ275148, GQ275149 and GQ275150; Schönhuth and

Mayden 2010). In addition to including sequence data from other native range populations examined in Broughton et al. (2011) and Diver (2013), data from 33 recognized and proposed species of Cyprinella as well as representative species from 9 genera of North American cyprinids were included in the analyses (Table 1-3). Haplotype networks also were constructed for each of the major lineages containing both native and non-native populations using the median-joining algorithm (Bandelt et al. 1999) in

Network v.4.612 (Fluxus Technology Ltd., http://www.fluxus-engineering.com) to further explore genealogical relationships as well as haplotype distributions and frequencies in the native and invasive ranges of red shiner.

11

Estimates of genetic diversity were calculated for every sampled site as well as for the major lineages recovered in phylogenetic analyses. DnaSP v.5.10.1 (Librado and

Rozas 2009) was used to calculate the number of haplotypes and to estimate mitochondrial haplotype diversity (Hd), nucleotide diversity (π), the number of polymorphic sites (S) and the average number of nucleotide differences between sequences (k) for each river and major lineage. We conducted AMOVAs in Arlequin v.3.5.1.2 (Excoffier et al. 2010) to characterize patterns of hierarchical (collection site, river, and drainage) genetic variation within and between the native and invasive ranges.

Genetic variation in the native range was tested independently for each of the major lineages to account for the unresolved phylogeny of Cyprinella, as some lineages are thought to include multiple species (Schönhuth and Mayden 2010). Comparisons between the non-native ranges included all samples. Additionally, pairwise estimates of genetic differentiation (ΦST) were calculated between rivers using Arlequin v.3.5.1.2.

Testing models of invasion history

Competing scenarios of introductions were evaluated using Approximate

Bayesian Computation (ABC) with the program DIYABC v2.0 (Cornuet et al. 2014).

Initially, we tested hypotheses describing the general pattern of invasion from the native into the disjunct invasive distributions in the western and eastern US. We first examined five competing scenarios of colonization following collection records and records of introductions and the observed phylogeographic pattern. Therefore only the major lineages occurring in the native and non-native ranges were included in the analyses (i.e., mid-western and western lineages; Figure 1-1). The first two scenarios assumed a single introduction of C. lutrensis from the native range into the western invasive range, and

12

subsequent secondary introduction from the western range to the eastern US. Scenario 1 considers all non-native populations as derived from populations of the mid-western lineage, while Scenario 2 considers an ancestral population of mixed mid-western and western lineage individuals as the source of non-native populations (i.e., an admixed aquarium trade or bait fish stock). Two scenarios were proposed to assess the hypothesis of an independent origin of the western and eastern invasive ranges. Scenario 3 assumes that the eastern invasive populations are derived from a single introduction of the native mid-western lineage, whereas the western invasive populations are derived from an introduction of the mid-western lineage and a second introduction of individuals from the native western lineage. Scenario 4, on the other hand, proposes that western non-native populations were initially founded by individuals from the western lineage and from a subsequent invasion of the mid-western lineage. As in the previous scenario, non-native eastern US populations would have been founded exclusively from a mid-western lineage. Finally, Scenario 5 proposed the existence of an ancestral admixed population

(i.e., an aquaculture fish stock) constituted by individuals from the mid-western and western lineages, which was the source for a single invasion of the western non-native range. In this scenario, the invasive range in the eastern US would have been founded exclusively by individuals from the mid-western lineage (Figure 1-1).

Based on the results from testing competing scenarios of invasion history across the US, we tested hypotheses of the invasion pathway within the eastern and western invasive ranges, respectively. We assessed whether the presence of red shiner within each region is a consequence of population expansion derived from one introduction or whether it reflects independent introductions at different times corresponding to historical

13

records (Figure 1-1). For the western region, Scenario W1 assumes an initial introduction into the San Joaquin River, California in 1950 (i.e., our sampling site with the earliest record) and subsequent expansion in chronological order from a single introduction to other basins in the region: San Joaquin River, California in 1950; Virgin River, Nevada in

1967; Virgin River, Arizona in 1967; Gila River, Arizona in 1991 (Nico et al. 2015).

Scenario W2, on the other hand, assumes multiple independent introductions from the native range into each sampled area within the western region following the same chronological order. Similarly, for the eastern region, Scenario E1 proposes that the initial colonization of Terrapin Creek, Alabama in 1992 resulted in subsequent expansion invasions according to chronological records (Terrapin Creek, Alabama in 1992; Dead

River, Alabama in 1992; Buffalo Creek, North Carolina in 1994; Peach Tree Creek,

Georgia in 1995; Conasauga River, Georgia in 2000; Rocky River, North Carolina in

2004; Nico et al. 2015), whereas Scenario E2 assumes that eastern populations are derived from multiple independent introductions from the native range following the same chronological order (Figure 1-1).

According to the historical records available, the prior distribution of demographic parameters considered all introduction events to have occurred over the last

100 years. Also, we assumed that the effective population size was the same for all non- native populations and used a uniform distribution between 10-10,000 individuals, except for the native and ancestral populations which were bounded between 100-1,000,000 individuals. A reduction in effective population size was modeled in all introduction events from the native to the non-native populations. Genetic variation within and between populations was characterized using a set of summary statistics. For each

14

population and each population pair, we used the number of haplotypes, the number of segregating sites and the mean number of pairwise differences. We also used pairwise

ΦST values as well as Tajima’s D for the one sample summary statistics. For the comprehensive and each regional analysis, we used 2 x 106 simulated datasets per scenario to build reference tables. Posterior probabilities of the competing invasion scenarios were calculated using 1% of the simulated datasets to estimate the relative posterior probability of each scenario with a logistic regression over a linear discriminant analysis of the summary statistics (Estoup et al. 2012). Confidence in the model of choice was assessed by calculating Type I (the probability of rejecting a scenario when it is true) and Type II errors (the probability of choosing a scenario when it is false) for each scenario.

RESULTS

Haplotype diversity and distributions

We examined mitochondrial cyt b sequence data from 480 individuals from 46 locations, with data derived from 352 specimens sampled in the native range and 128 specimens collected in the non-native range. Though the majority of these data encompassed the full cyt b region, we constrained comparisons to a subsection of the gene in order to incorporate existing mtDNA sequence data from geographically important regions. Thus, we reduced our sequence data set to an alignment targeting a

574bp fragment for subsequent analysis. The resulting cyt b dataset yielded 132 haplotypes, with 103 haplotypes found in the native range and 29 recovered in the invasive range.

15

The phylogenetic hypothesis recovered from ML analysis of mtDNA sequence variation indicated that our samples constitute four highly supported clades (i.e., ≥98% bootstrap support; Figure 1-2). Although the relationships among the lineages are not well resolved, it is clear the four lineages do not form a monophyletic group and that each lineage corresponds to a distinct geographic region: the mid-west, which includes rivers in the Mississippi River drainage; the south-west, which includes rivers in the Rio Grande

River drainage; the west, which includes rivers from the Gulf Coast, Rio Grande and

Mississippi River drainage; and the Brazos, which includes rivers draining to the Gulf of

Mexico largely within Texas. The range size of the four lineages is variable, with the mid-western lineage being the largest and most widespread, followed by the western, south-western, and Brazos. Genetic divergence among lineages ranges between 8.23% ±

1.05% (uncorrected p-distance between mid-west and Brazos lineages) and 12.36% ±

1.27% (between the west and south-west lineages). In addition, genetic diversity varies across the lineages, with the mid-western lineage encompassing 78 haplotypes, the south- western clade encompassing 23 haplotypes, the western clade encompassing 16 haplotypes, and the Brazos clade encompassing 13 haplotypes (Table 1-2). The most widespread and abundant haplotypes were members of the mid-western lineage.

Examination of introduced populations yielded two distinct clades in the western

US, and one clade in the eastern US. Only the mid-western and western lineages included haplotypes recovered in invasive regions (Figures 1-2 and 1-3). The mid-western lineage included haplotypes recovered in invasive populations in the eastern and western US, including: Dead River (AL), Terrapin Creek (AL), Virgin River (AZ), Conasauga River

(GA), Peachtree Creek (GA), Buffalo Creek (NC), and Rocky River (NC; Figures 1-2

16

and 1-3). The western clade included haplotypes only recovered in invasive populations sampled in California and haplotypes shared between California and native range populations (Figures 1-2 and 1-3).

The haplotypes recovered in invasive populations were not prevalent within the mid-western lineage, whereas the western clade haplotypes recovered in invasive populations dominated the clade (Figures 1-2 and 1-3). Of these, six haplotypes were shared between native and invasive ranges in the mid-western clade, and only two haplotypes were shared between native and invasive ranges in the western clade. A single haplotype (haplotype #12) was recovered in approximately one quarter of all samples (n

= 109), and was found in eastern US and western US invasive ranges, as well as the native range. This haplotype was by far the most abundant and widespread of all the haplotypes recovered; individuals exhibiting it were collected in 21 rivers, across 13 states, and in both the US and Mexico. The second most common haplotype, haplotype

#45, was found in 4.8% of individuals (n = 21) from 10 different rivers across 7 states.

Across the native range, the western lineage (Figure 1-3) exhibited the highest haplotype diversity (Hd = 0.981, SD = 0.031), followed by the lineage encompassing the mid-western US (Hd = 0.976, SD = 0.004), and the lineage with a south-western geographical footprint (Hd = 0.954, SD = 0.017). Despite encompassing two lineages, the invasive western US populations exhibited an average haplotype diversity comparatively lower than invasive eastern US populations (Hd = 0.913, SD = 0.014; Hd = 0.951, SD =

0.014). Maximum haplotype diversity scores of 1 were recovered in seven different populations across the native range (Table 1-2), while maximum haplotype diversity scores were only recovered in two invasive populations: Terrapin Creek (AL), and Dead

17

River (AL; Table 1-2). The native population with the lowest haplotype diversity was in

Boyer River (IA), while the invasive population with the lowest haplotype diversity was in Virgin River (NV). Overall, haplotype diversity was not significantly different between native range populations (Hd = 0.963, SD = 0.019) compared to non-native range populations (Hd = 0.901, SD = 0.020; p = 0.137).

Other measures of genetic diversity suggest that invasive populations harbor greater levels of genetic diversity than do native range populations. Notably, the average number of nucleotide differences in invasive populations (k = 11.9 bp SD = 5.445) was over a third larger than that of native range populations (k = 7.8 bp SD = 3.233; Table 1-

2). Higher nucleotide diversity was also found in introduced populations (π = 0.080, SD =

0.044) compared to native range populations (π = 0.014, SD = 0.007; Table 1-2).

However, some of these trends vary when comparisons were drawn between individual rivers and sample sites. For example, invasive range samples from the Virgin River (NV) exhibited relatively low measures of genetic diversity (k = 0.142 bp, π = 0.00025), while native range samples from the Canadian River (NM) exhibited comparably high levels of genetic diversity (k = 7.667 bp, π = 0.010).

Genetic structure

AMOVA comparisons revealed that most of the observed genetic variation across our sampled populations reflects a combination of local and regional heterogeneity. Most of the genetic variation among the native range populations from the mid-western lineage was attributable to differences among drainages (64.29%, Fct =0.642, p = 0.000), although genetic differentiation among populations within drainages and within populations were also highly significant (21.99%, Fsc=0.616, p = 0.000 and 13.72%,

18

Fst=0.862, p = 0.000, respectively). Native populations from the south-western lineage did not show significant differentiation among drainages, but harbored a higher percentage of genetic variation among populations within drainages (95.91%, Fsc=0.963, p = 0.000) and a lower percentage within populations (3.69%, Fst=0.963, p = 0.000) than in the mid-west. Neither the western lineage nor the Brazos lineage showed significant genetic structure at any level. In the invasive ranges, the extent of variation attributable to differences among collection sites differed between the eastern and western US. A large portion of genetic variation in invasive western US drainages reflects heterogeneity among collection sites (Fst =0.409, P < 0.001), whereas variation in invasive eastern US drainages does not reflect significant differences among collection sites (6.11%, Fst

=0.061, P = 0.099). Comparing invasive eastern and western US ranges explained the largest amount of variation across invasive populations (75.54%, Fct =0.755, P < 0.001).

Overall, genetic differentiation between native and invasive populations, either considered altogether, or separately by western and eastern US drainages, was non- significant (-6.35%, Fct =-0.063, p = 0.672 and -14.48%, Fct =-0.145, p=0.897, respectively). However, considering drainages separately for each lineage showed significant genetic differentiation between native and invasive populations (36.04%, Fct

=0.360, p = 0.001 for the mid-western lineage and 24.64%, Fct =0.246, p = 0.030 for the west lineage).

Invasion scenarios

Approximate Bayesian Computation (ABC) analyses support an invasion scenario with geographically and temporally independent introductions into the eastern and western US. The best supported model was Scenario 5 (PP= 0.6901, CI=0.6355-0.7446),

19

which proposes the existence of an ancestral admixed population with co-occurring haplotypes from the mid-western and western lineages from which the non-native range in the western US was founded, and an independent invasion exclusively from mid- western lineage populations into the non-native range in the eastern US. The second and third best supported scenarios were Scenario 4 and 3, which showed much lower relative posterior probabilities and no overlapping confidence intervals with Scenario 5 (PP=

0.2032, CI=0.1544-0.2521 and PP= 0.1065, CI=0.0798-0.1332) (Figure 1-4). The probability of rejecting Scenario 5 when true (i.e., Type I error) was 0.564, and the probabilities of choosing it when Scenarios 3 and 4 were true (i.e., Type II errors) were

0.274 and 0.258. In the non-native ranges, the scenarios proposing one introduction event followed by within-range expansions (Scenarios W1 and E1) showed the highest support

(W1: PP= 0.9309, CI= 0.9236-0.9382, and E1: PP= 0.6639, CI=0.6553-0.6725; Figure 1-

5). In all cases the confidence intervals for the relative posterior probabilities of the best- fit models did not overlap with the other models tested. Type I errors for Scenarios W1 and E1were 0.276 and 0.308, and Type II errors were 0.176 and 0.208.

DISCUSSION

Biological invasions generally follow a pattern of introduction, establishment, and range expansion (Kolar and Lodge 2001). Yet introduced species often exhibit fragmented distributions within invasive ranges, possibly as the result of multiple independent introductions from the native range or as the result of stochastic secondary spread. Prior studies documenting red shiner introductions have provided observational records of first accounts and localized dispersal within individual river systems (Page and

20

Smith 1970, Moyle 2002, Walters et al. 2008) that offered limited information on the origin and nature of its disjunct distribution. Through the analysis of mtDNA sequence variation, we reconstructed genealogical relationships among native and non-native populations to understand the history of red shiner invasions and to clarify the potential for future spread across the US. ABC analyses suggest that introduced populations of red shiner across the eastern and western US resulted from multiple independent introductions from the native range, which is consistent with prior suggestions that introductions have largely been the result of unrelated bait bucket releases that have given rise to regionalized spread (Fuller et al. 1999, Walters et al. 2008). However, we also found that a highly divergent (western clade) but phenotypically cryptic evolutionary lineage (which likely warrants species recognition) has been introduced to the western

US and that multiple evolutionary lineages co-occur within the Virgin River (AZ). Thus current estimates of potential spread under current and future climate conditions (Poulos et al. 2012, Poulos and Chernoff 2014) should be revisited to account for phylogenetic diversity within and among invasive populations. Management and monitoring of invasive populations also should account for the possibility of post-introduction hybridization among distinct evolutionary lineages in addition to hybridization with native congeners.

Phylogenetic diversity and genetic variation of native range red shiner

The taxonomy and phylogeny of Cyprinella lutrensis are not well resolved; since the species was originally described, different forms of “lutrensis” have been distinguished and more than a dozen names have been given to local variants across the native range (Contreras-Balderas 1975; Matthews 1987; Mayden 1989; Broughton and

21

Gold 2000; Schönhuth and Mayden 2010). Presently, the widely distributed C. lutrensis is considered a heterogeneous species complex that consists of at least four monophyletic groups occurring in four distinct geographic regions (Richardson and Gold 1995;

Schönhuth and Mayden 2010). Two of the four monophyletic groups that occur in the southern US and Mexico have been considered to be different species (C. forlonensis, in central-east Mexico; C. sp. 1 in northern Mexico-southern US). We also discovered a previously undescribed and well supported western lineage from samples collected in north-central Texas, south-west Arkansas, and north-central Mexico, that is strongly divergent compared to other native range red shiner lineages. For example, approximately 11% sequence divergence differentiates the newly detected lineage from the mid-western lineage. Assuming a mtDNA sequence divergence rate of approximately

1.4-1.9% per million years (Brown and Chapman 1991; Bentzen et al. 1993; Martin and

Palumbi 1993), these two lineages diverged ~5.8-7.9 million years ago.

Like in many other North American fishes, the phylogeographic structure of red shiner lineages appears to reflect Pleistocene glaciation (Bermingham and Avise 1986,

Richardson and Gold 1995, Soltis et al. 2006). Prior studies suggest that the evolutionary lineages ascribed to C. lutrensis arose well before the onset of Pleistocene glaciation, and support the hypothesis that one or more lineages entered the region directly affected by glaciation during or after the Pleistocene (Richardson and Gold 1995). Similar to prior findings (Richardson and Gold 1995), our phylogenetic analysis of C. lutrensis within the native range revealed that the mid-western lineage exhibits the largest number of haplotypes and geographical extension within the species complex. However, our findings indicate that genetic variation within the mid-western and south-western lineages

22

largely reflects local scale heterogeneity, and that significant genetic structure also occurs among drainages within the mid-western clade. This pattern of genetic structure could be the result of range expansion by means of stream capture and interconnecting drainages, following isolation in southern refugia during the last glacial maximum (Conner and

Suttkus 1986; Richardson and Gold 1995; Schönhuth et al. 2008). The southwestern lineage, on the other hand, appears to exhibit a more restricted distribution despite encompassing six closely related clades with at least five named species that are likely of

Pleistocene origin (Schönhuth and Mayden 2010). The high degree of heterogeneity and species co-occurrence in these drainages suggests that limited range expansion and secondary contact occurred following divergence in allopatry due to geographical isolation and divergence during the last glaciation. Conversely, we did not recover any significant genetic structure within the western or the Brazos lineages. In both cases, a pattern of geographically restricted lineages with low genetic diversity was observed, which could be attributable to low dispersal and long term isolation (Conner and Suttkus

1986; Richardson and Gold 1995).

Phylogenetic diversity and invasion history of non-native red shiner

Genetic analysis of invasive species can provide valuable information on invasive history including demographic and evolutionary processes contributing to invasion success as well as colonization pathways and source populations (Muirhead et al. 2008,

Estoup and Guillemaud 2010, Guillemaud et al. 2010). Despite relatively large probabilities of both Type I and Type II errors, our ABC analyses provide an additional line of supporting evidence, suggesting that the invasion of red shiner across the eastern

US is different and distinct from the invasion of the western US. Our findings indicate

23

that the western US introduction originated from an admixed population including individuals form the mid-west and western lineages in the native range, while eastern US introductions derive exclusively from the mid-west lineage. Only haplotype #12, the most common haplotype (n=109) was found in both the western and eastern invasive ranges.

Haplotypes contributing to the western invasions mostly belonged to rivers harboring the western lineage, and matched native range haplotypes found in the Canadian and Pecos

River (NM), Wichita River (TX), and San Pedro River (Mexico), and one mid-western haplotype found in the Kaskaskia River, (IL). The eastern introductions, however, represent a wider geographical sample from the native range with shared haplotypes from five states and Mexico. The repeated patterns of shared haplotypes provide some indication of potential native sources of introduction including the Arkansas River (AR) and the Salt River (MO), each of which shared the same two haplotypes with five non- native populations in North Carolina, Alabama and Georgia. Similarly, two haplotypes from the Verdigris River (KS) were also found in four populations in Alabama and

Georgia. ABC models suggest, however, that secondary spread occurred within both the eastern and western invasive ranges following initial introduction events. This indicates either that secondary introductions from sources within the invasive ranges have occurred or that there is regional connectivity among drainages, possibly reflecting long-distance dispersal or local transplantation. This finding contrasts with some prior studies (Walters et al. 2008) suggesting that regional spread, even within a specific basin (e.g., the upper

Mobile basin), is possibly an outcome of multiple independent introductions from the native range to geographically proximate basins.

24

Evidence of phylogenetic diversity clarifies long-standing questions about the identity and origins of red shiner introductions to the western US (Moyle 1976, Jennings and Saiki 1990). We recovered two evolutionary lineages of Cyprinella within the Virgin

River (AZ), including one that has not been previously described. The presence of two separate evolutionary lineages (one of which likely warrants species recognition) of

Cyprinella in the Virgin River highlights the possibility that putative red shiner introductions (i.e., introductions of fish that exhibit comparable phenotypes) across other regions of the US may encompass cryptic invasions of multiple Cyprinella lineages or species. Given the distribution of the two lineages within the native range, it is possible that co-occurrence in the Virgin River is an outcome of two or more transplant events.

Yet, ABC modeling suggests that a multi-species “pool” of Cyprinella were introduced to this area in a single introduction event. This scenario is consistent with the history of aquaculture and use of red shiner as a bait fish as well as commercial sale of red shiner in the aquarium trade. The presence of the undescribed western lineage of Cyprinella in the

San Joaquin River (CA) further illustrates that red shiner invasions are a mosaic of phylogenetic diversity. Our findings affirm that a genetically divergent lineage of

Cyprinella was introduced to the San Joaquin River, which is consistent with evidence of phenotypic differences noted by Jennings and Saiki (1990), who questioned whether fishes introduced to California were red shiner or a closely related congener like C. suavis (Schönhuth and Mayden 2010).

The success of aquatic invasive species has often occurred despite what some have labeled a ‘genetic paradox’ (Sakai et al. 2002; Allendorf and Lundquist 2003;

Roman and Darling 2007). Small founding populations of introduced species are

25

expected to have genetic diversity that is lower than that of native populations as a result of demographic bottlenecks (Allendorf and Lundquist 2003). Small population size increases the risk of extinction and reduced genetic variation due to founder effects is expected to limit the ability of a newly established population to respond to novel conditions (Frankham 2005). Yet many introduced species either overcome or simply do not experience the challenge of low genetic diversity (Roman and Darling 2007). Some invasive populations, for example, do not exhibit comparably low genetic diversity as a consequence of successive introductions from a single or multiple native population (e.g.,

Blum et al. 2007). Admixture and interspecific hybridization can also promote invasion success by elevating genetic diversity, enabling introgression of locally adapted functional traits from native congeners, or by stimulating post-invasion evolution of novel genotypes (Ellstrand and Schierenbeck 2000, Lee 2002). Our findings further illustrate that invasive species are not necessarily constrained by low genetic diversity.

Though local populations of non-native red shiner exhibited comparably lower levels of haplotype diversity, overall levels of diversity were higher in the invasive rather than native range. This is likely an outcome of multiple independent introductions as well as transplanted populations originating from mixed source pools. The occurrence of multiple evolutionary lineages in the Virgin River also raises the possibility of post- introduction hybridization between naturally allopatric evolutionary lineages. Though hybridization between red shiner and native congeners is a well-recognized threat to native biodiversity (Walters et al. 2008; Blum et al. 2010; Ward et al. 2012), it is possible that cryptic hybridization could be an additional threat. Invasions could be driven by hybrids of non-native evolutionary lineages (e.g., Lack et al. 2012), or novel outcomes of

26

reticulate hybridization between native congeners and multiple non-native evolutionary lineages or hybrids thereof (McDonald et al. 2008).

Disjunct distributions and forecasting spread

Understanding the origins and potential interactions between disjunct populations of invasive species can enable more accurate forecasting of future spread, particularly when models are intended to reflect physiological responses to changing climate conditions (Buckley et al. 2011, Araujo and Peterson 2012, Poulos et al. 2012). Forecasts of red shiner distributions under contemporary and future climate conditions (Poulos et al. 2012, Poulos and Chernoff 2014) are based on the dual premise that red shiner are a single evolutionary lineage that exhibits broad environmental tolerances and that red shiner are well adapted to succeed in novel, harsh or demanding conditions (Matthews and Hill 1977, Matthews 1985, Yu and Peters 2002, Walters et al. 2008). Our findings suggest that models forecasting spread of red shiner could be improved by accounting for the possibility of physiological or niche variation across the breadth of phylogenetic diversity that occurs across native and non-native populations (e.g., Compton et al. 2010,

Kelley et al. 2011). This could afford more accurate perspectives on the likelihood of range contraction, expansion, or shifts across regions occupied by different evolutionary lineages. Additionally, predictive models likely could be improved if based on the understanding that some introduced populations encompass multiple evolutionary lineages and that regional invasions have proceeded via secondary spread.

The introduction of invasive red shiner across the US serves as an example of how disjunct distributions can originate from multiple, independent introductions rather than the result of secondary spread (e.g., Blum et al. 2007, Darling et al. 2008). Our

27

findings suggest that the disjunct distribution of red shiner across North America is a collection of regional invasions with distinct origins rather than the stepwise advance of one or more unified invasion front. Thus localized control or removal may be effective in managing non-native red shiner, including further spread to areas of conservation concern. The risk of continued spread could be better managed, and perhaps reduced, if forecasting models account for the distribution of evolutionary lineages that may occupy distinct fundamental niches. Further monitoring and genetic assessments of non-native populations could also help define invasion risk, particularly in the western US where co- occurring evolutionary lineages could hybridize.

28

Table 1-1. Collection records of first occurrences of non-native red shiner.

State Year Drainage or River California 1950 Lower Colorado & Whitewater River Arizona 1953 Lower Colorado Illinois 1958 Lake Michigan Utah 1962 Upper Colorado Nevada 1967 Lower Colorado & Lower Virgin River Colorado 1969 Lower Yampa Alabama 1970 Chattahoochee River North Carolina 1974 Lower Yadkin New Mexico 1980 Gila River Wyoming 1982 Great Divide-Upper Green River South Carolina 1986 Lower Pee Dee Virginia 1986 Roanoke Georgia 1992 South Atlantic Gulf

29

Table 1-2. Locations and genetic diversity estimates for 46 populations of Cyprinella lutrensis across native and invasive ranges. Asterisks indicate non-computable values due to the presence of only one haplotype.

30

π k River Drainage State Lat. (N) Long. (W) Coll. Year n h Hd Hd SD π S k SD SD Native MW Lineage 271 78 0.976 0.004 0.010 0.005 106.000 5.457 2.635 Brazos Gulf Coast Texas 29.19 -95.58 2004 3 1 * * * * * * * Guadalupe Gulf Coast Texas 29.87 -98.49 2004 1 1 * * * * * * * Little Brazos Gulf Coast Texas 29.60 -95.76 2004 3 1 * * * * * * * Rio Grande Gulf Coast Texas 29.51 -103.40 2004 1 1 * * * * * * * Rio Grande Gulf Coast Texas 29.04 -103.33 2004 2 2 1.000 0.500 0.005 0.006 3 3.000 2.449 Arkansas River Mississippi Arkansas 35.30 -93.14 2012 23 12 0.901 0.041 0.009 0.005 17 4.905 2.480 Red River Mississippi Arkansas 33.54 -93.80 2012 5 4 0.900 0.161 0.012 0.008 15 6.800 3.859 Boyer River Mississippi Iowa 42.04 -95.35 2012 17 3 0.228 0.130 0.000 0.001 2 0.235 0.286 Skunk River Mississippi Iowa 42.03 -93.59 2012 19 5 0.737 0.069 0.003 0.002 9 1.579 0.982 Kaskaskia River Mississippi Illinois 39.57 -88.54 2012 23 6 0.625 0.109 0.002 0.002 8 1.249 0.819 Sangamon River Mississippi Illinois 39.86 -89.67 2012 16 3 0.575 0.112 0.001 0.001 2 0.650 0.531 Fox Creek Mississippi Kansas 39.83 -95.35 2005 11 4 0.800 0.075 0.003 0.002 5 1.964 1.201 Kansas River Mississippi Kansas 38.98 -95.06 2005 8 2 0.536 0.123 0.002 0.002 2 1.071 0.786 Prong Creek Mississippi Kansas 37.72 -94.99 2005 5 4 0.900 0.161 0.002 0.002 3 1.200 0.908 Verdigris River Mississippi Kansas 37.06 -95.60 2005 4 2 0.500 0.265 0.003 0.003 4 2.000 1.405 Brush Creek Mississippi Kentucky 36.73 -89.00 2009 17 3 0.588 0.094 0.001 0.001 2 0.662 0.535 Auxvasse Creek Mississippi Missouri 38.95 -91.84 2002 5 4 0.900 0.161 0.002 0.002 3 1.200 0.908 Crows Fork Creek Mississippi Missouri 38.87 -91.88 2005 1 1 * * * * * * * Cuivre River Mississippi Missouri 38.95 -90.91 2006 3 3 1.000 0.272 0.010 0.009 9 6.000 3.928 Salt River Mississippi Missouri 39.63 -91.24 2003 8 7 0.964 0.077 0.010 0.006 16 5.786 3.097 Platte River Mississippi Nebraska 41.20 -96.34 2012 37 6 0.459 0.096 0.001 0.001 3 0.508 0.440 Canadian River Mississippi New Mexico 36.32 -104.50 2012 1 1 * * * * * * * Bird Creek Mississippi Oklahoma 36.21 -95.91 2006 5 2 0.400 0.237 0.001 0.001 2 0.800 0.681 Canadian River Mississippi Oklahoma 35.24 -97.57 2006 5 3 0.800 0.164 0.004 0.003 4 2.200 1.450 Deep Fort River Mississippi Oklahoma 35.67 -97.18 2012 3 1 * * * * * * * Indian Creek Mississippi Tennessee 35.57 -89.86 2012 13 4 0.744 0.091 0.002 0.002 4 1.231 0.834 Bear Creek Mississippi Tennessee 35.45 -89.96 2009 19 3 0.503 0.113 0.001 0.001 2 0.550 0.472 Arroyo Correras Rio Grande Mexico 24.45 -98.33 2006 1 1 * * * * * * * San Juan Rio Grande Mexico 31.81 -106.56 2006 2 2 1.000 0.500 0.002 0.002 1 1.000 1.000 San Pedro Rio Grande Mexico 28.26 -105.48 2006 1 1 * * * * * * * Pecos River Rio Grande New Mexico 33.61 -104.37 2012 3 3 1.000 0.272 0.010 0.009 9 6.000 3.928 Rio Grande Rio Grande New Mexico 31.79 -106.53 2006 6 3 0.733 0.155 0.009 0.006 9 4.867 2.761 Native SW Lineage 40 23 0.954 0.017 0.030 0.015 59 16.673 7.580 Guadalupe Gulf Coast Texas 29.87 -98.49 2004 1 1 * * * * * * * Rio Grande Gulf Coast Texas 29.04 -103.33 2004 2 2 1.000 0.500 0.005 0.006 3 3.000 2.449 Rio Grande Gulf Coast Texas 29.51 -103.40 2004 2 1 * * * * * * * Concho River Gulf Coast Texas 31.90 -101.12 2010 3 2 0.667 0.314 0.003 0.003 3 2.000 1.512 Rio Grande Gulf Coast Texas 29.04 -103.33 2004 4 3 0.833 0.222 0.003 0.002 3 1.500 1.121 Rio Grande(suavis) Gulf Coast Texas 29.04 -103.33 2004 2 2 1.000 0.500 0.003 0.004 2 2.000 1.732 Canadian River Mississippi Oklahoma 35.24 -97.57 2006 2 2 1.000 0.500 0.005 0.006 3 3.000 2.449 Morris Creek Mississippi Oklahoma 35.11 -94.61 2010 1 1 * * * * * * * Morris Creek(lepida) Mississippi Oklahoma 35.11 -94.61 2010 1 1 * * * * * * * Cottonwood River Mississippi Kansas 38.26 -96.54 2006 1 1 * * * * * * * Arroyo Correras Rio Grande Mexico 24.45 -98.33 2006 2 1 * * * * * * * Rio Grande Rio Grande New Mexico 31.79 -106.53 2006 5 1 * * * * * * * Pecos River Rio Grande New Mexico 33.61 -104.37 2012 3 1 * * * * * * * San Pedro Rio Grande Mexico 28.26 -105.48 2006 2 1 * * * * * * * Pecos River Rio Grande New Mexico 29.68 -101.37 2006 6 1 * * * * * * * Pecos River(suavis) Rio Grande New Mexico 29.68 -101.37 2006 3 1 * * * * * * * Native West sp.2 Lineage 15 9 0.981 0.031 0.009 0.005 21 5.314 2.718 Wichita River Gulf Coast Texas 33.77 -99.09 2004 2 2 1.000 0.500 0.012 0.013 7 7.000 5.292 Red River Mississippi Arkansas 33.54 -93.80 2012 2 2 1.000 0.500 0.005 0.006 3 3.000 2.449 Canadian River Mississippi New Mexico 36.32 -104.50 2012 4 4 1.000 0.177 0.014 0.010 15 7.667 4.533 San Pedro Rio Grande Mexico 28.26 -105.48 2006 1 1 * * * * * * * Pecos River Rio Grande New Mexico 33.61 -104.37 2012 6 4 0.867 0.129 0.006 0.004 9 3.600 2.121 Native Brazos Lineage 24 13 0.942 0.025 0.007 0.004 15 3.848 2.003 Brazos Gulf Coast Texas 29.19 -95.58 2004 22 11 0.931 0.028 0.007 0.004 13 3.831 2.003 San Pedro Rio Grande Mexico 28.26 -105.48 2006 1 1 * * * * * * * Pecos River Rio Grande New Mexico 29.68 -101.37 2006 1 1 * * * * * * * Invasive West 81 16 0.851 0.025 0.035 0.017 71 20.059 8.959 San Joaquin California California 38.06 -121.68 2012 18 7 0.843 0.056 0.009 0.001 14 5.111 2.599 Gila River Colorado Arizona 32.93 -113.40 2005 7 3 0.524 0.209 0.003 0.001 5 1.810 1.180 Virgin River Colorado Arizona -113.80 36.95 2005 29 4 0.638 0.050 0.058 0.003 67 33.296 14.960 Virgin River Colorado Nevada 36.80 -114.06 2005 27 2 0.143 0.086 0.000 0.000 1 0.142 0.213 Invasive East 46 31 0.951 0.014 0.125 0.071 30 3.760 1.931 Terrapin Creek South Atlantic Alabama 34.03 -85.61 2008 5 5 1.000 0.127 0.006 0.004 7 3.400 2.080 Dead River South Atlantic Alabama 34.13 -85.79 2008 3 3 1.000 0.272 0.006 0.005 5 3.333 2.323 Peach Tree Creek South Atlantic Georgia 33.82 -84.40 2008 19 11 0.906 0.047 0.123 0.073 14 3.696 1.954 Conasauga South Atlantic Georgia 34.79 -84.86 2008 8 7 0.964 0.077 0.144 0.091 12 4.321 2.390 Buffalo Creek South Atlantic North Carolina 36.10 -79.77 2004 8 4 0.786 0.113 0.007 0.004 12 3.857 2.165 Rocky River South Atlantic North Carolina 35.17 -80.47 2004 3 1 * * * * * * *

31

Table 1-3. Additional Cyprinella sequence data used in genetic analysis.

Species GenBank Accession numbers C. whipplei GQ275230-231 C. suavis DQ324095, GQ275175-176, GQ275189-190 C. garmani EU082522-23, DQ324102 C. sp 1 GQ275192-193 C. venusta GQ275201-217, GQ275242 C. spiloptera GQ275218-223, GQ275232-233 C. panarcys GQ275195-196 C. lutrensis KR061540 - KR061569 C. lutrensis (Awaiting numbers from Tom Turner)

32

Figure 1-1. ABC model scenarios of colonization for non-native red shiner populations across the US. MW = mid-west lineage, W = western lineage. S1-S5 = general models of invasion into western and eastern US. W1-W2 = models of invasion within the western invasive range. E1-E2 = models of invasion within the eastern invasive range.

33

Figure 1-2. Phylogeny reconstruction of cytochrome b haplotypes using maximum likelihood analysis. Blue = western US invasive population; White = native range populations; Pink = eastern US invasive population; Green = found in both western and eastern invasive ranges; Black outline = shared between native and invasive ranges.

34

35

Figure 1-3. Statistical-parsimony haplotype frequency network of cytochrome b for

Cyprinella lutrensis samples across US. (A) the mid-western clade occurs in native populations and non-native populations in the western and eastern US; (B) the western clade occurs in native and non-native populations in the western US. White / blue outline

= western US invasive population; White / black outline = native range populations;

White / pink outline = eastern US invasive population. The most common haplotypes are labeled by number.

36

37

Figure 1-4. ABC analysis invasion scenarios for western and eastern US. Changes in line color represent assumed changes in population size post colonization.

38

Figure 1-5. ABC analysis for within invasive western and eastern ranges. Changes in line color represent assumed changes in population size post colonization. GR = Gila River,

VR = Virgin River, SJ = San Joaquin River, RR = Rocky River, CR = Coosa River, PT =

Peachtree Creek, BC = Buffalo Creek, DR = Dead River, TC = Terrapin Creek.

39

CHAPTER 2

RAPID MOVEMENT AND INSTABILITY OF AN INVASIVE HYBRID SWARM

ABSTRACT

Expansion and movement of hybrid swarms following the introduction of non- native species can potentially overwhelm native congeners, particularly narrowly distributed endemic or threatened species. Though it is well known that invasive hybrid swarms can be highly dynamic, little is known about rates of movement or expansion over time. Here we examine an invasive hybrid swarm to understand temporal dynamics and to clarify how fast and far an invasive species may spread through hybridization with a native congener. We characterized genetic variation and clinal stability across the recently formed hybrid swarm involving native blacktail shiner (Cyprinella venusta stigmatura) and introduced non-native red shiner (Cyprinella lutrensis) in the Upper

Coosa River basin (Alabama, Georgia, Tennessee, USA). Examination of phenotypic meristic traits, multilocus microsatellite genotypic, and mitochondrial haplotype variability between 2005 and 2011 revealed that the proportion of hybrids has increased over time, with nearly one quarter of all sampled individuals exhibiting an intermediate genotype in our final year of sampling. Models of mitochondrial haplotype, microsatellite genotype, and phenotypic clines provided evidence that the hybrid swarm has been progressing upstream, but at a declining and slower pace than rates estimated from collection records. Additionally, it appears that microsatellite alleles and mtDNA

40

haplotypes are introgressing farther than phenotypic traits, and cline models indicate that the hybrid swarm is expanding and contracting over time. We also documented the presence of red shiner and hybrids farther downstream than prior studies have detected, which suggests that congeners in the Coosa River basin, including all remaining populations of the Federally Threatened blue shiner, are at greater risk than previously thought. Further monitoring, particularly of southerly reaches, would help identify factors contributing to hybridization and perhaps mitigate continued spread of invasive red shiner and hybrids in areas of conservation concern.

INTRODUCTION

Hybrid zones are geographic regions of phenotypic or genotypic change that separate genetically distinct populations. Hybrid zones are classified according to the distribution of parental and hybrid genotypes (Harrison and Bogdanowicz 1997). Hybrid swarms, for example, are characterized by a unimodal distribution reflecting the dominant presence of intermediate hybrid genotypes (Endler 1977; Harrison 1993;

Jiggins and Mallet 2000). Examples of hybrid swarms have been documented in an array of taxa, ranging from butterflies (e.g., Heliconius erato; Blum 2002, 2008) to freshwater minnows (e.g., Cyprinella lutrensis x Cyprinella venusta; Walters et al. 2008; Ward et al.

2012). Taxa that form hybrid swarms often exhibit weak assortative mating and incomplete prezygotic isolation (Jiggins and Mallet 2000). Selection against hybrids also is often weak or absent, increasing the potential for introgressive hybridization to erode species boundaries, which can lead to rapid loss of genetic and species diversity

(Seehausen et al 1997; Wolf et al. 2001; Taylor et al. 2006). Additionally, hybrid swarms

41

can overwhelm parental species through genetic homogenization, precipitating species collapse (Perry et al. 2001; Behm at al. 2010). Yet the temporal dynamics of hybrid swarms remain poorly documented, including the tempo of movement and expansion

(Taylor et al. 2006; Seehausen et al 2008; Gilman and Behm 2011).

The formation and evolution of a hybrid swarm can be a dynamic and rapid process involving the progression of initially discrete parental trait distributions to a single distribution of intermediate, hybrid forms over space and time (Jiggins and Mallet

2000; Taylor et al. 2006). The propensity of discrete species distribution (i.e., boundaries) to erode can vary according to the nature and extent of interactions between parental species and hybrids following secondary contact (Harrison 1993; Arnold 1997). At the onset of secondary contact, trait distributions typically resemble steep coincident spatial transitions from one parental species to another (Barton and Hewitt 1985; 1989; Arnold

1997). If reproductive isolation is incomplete and barriers to gene flow are weak, the frequency of later-generation hybrids and backcrossed individuals increases, leading to the loss of distinct phenotypic and genetic entities (Forbes and Allendorf 1991; Taylor et al 2006; Gilman and Behm 2011). Over time, the initially steep clines established by secondary contact erode (Endler 1977; Barton and Gale 1993), flattening individual genotype and trait distributions and expanding the spatial extent of introgression. The potential for elevated fitness in hybrid genotypes and the rise of advantageous traits in admixed populations can subsequently hasten the erosion of species boundaries or genetic assimilation of parental species (Arnold 1997; Barton 2001; Coyne and Orr 2004;

Hall et al. 2006).

42

The formation and evolution of hybrid swarms can result in loss of biodiversity through the collapse of species boundaries (Seehausen 2006; Taylor et al. 2006;

Seehausen et al. 2008; Behm et al 2010). For example, in less than thirty years, a sympatric species pair of threespine stickleback (Gasterosteus aculeatus) in Enos Lake collapsed into a hybrid swarm possibly due to the breakdown of premating isolation following the introduction of a non-native crayfish capable of disrupting habitat preferences for nesting (Taylor et al. 2006). Morphological analysis showed a clear breakdown between benthic and limnetic forms, resulting in a single phenotypic cluster.

Genetic analysis confirmed this finding, illustrating that two previously detectable genetic clusters had collapsed into a single cluster. Similarly, hybridization within thirteen sympatric populations of coastal cutthroat trout and rainbow trout has resulted in high hybrid frequencies and a diverse array of backcrossed and higher-order hybrid genotypes, indicating that the breakdown of reproductive isolation resulted in repeated collapse of distinct lineages into hybrid swarms (Bettles at al. 2005). Post-introduction interactions between non-native and native species can also result in the formation and expansive spread of hybrid swarms. For instance, the sheepshead minnow (Cyprinodon variegatus), which was introduced to the Pecos River drainage in western Texas and New

Mexico between 1968 and 2000, eventually hybridized with the imperiled Pecos pupfish

(Cyprinodon pecosensis), resulting in the spread of hybrids across more than half of the range of the Pecos pupfish in less than five years (Echelle and Connor 1989; Childs et al.

1996; Rosenfield et al. 2004). However, because the erosion of species boundaries can be rapid, many aspects of the process remain poorly understood, including the nature of hybrid swarm movement and expansion.

43

Biological invasions involving hybridization present exceptional opportunities to examine hybrid swarms because timelines and outcomes of interactions between native and invasive congeners are often well known (e.g., Ward et al. 2012). Hybridization between non-native and native congeners is a well-recognized invasion pathway

(Ellstrand and Schierenbeck 2000; Sakai et al. 2001), particularly in aquatic and estuarine ecosystems (e.g., Echelle and Connor 1989; Childs et al. 1996; Rosenfield et al. 2004;

Ayres et al. 2008; Walters et al. 2008, Ward et al. 2012; Strong and Ayres 2013). In this study, we examined the clinal movement of a recently formed and rapidly expanding hybrid swarm (Walters et al. 2008) involving native blacktail shiner (Cyprinella venusta stigmatura) and introduced non-native red shiner (Cyprinella lutrensis) in the Upper

Coosa river basin (Alabama, Georgia, Tennessee, USA). Though prior studies have shown that the introduction of C. lutrensis into the Coosa River basin resulted in a dynamic swarm dominated by later-generation and backcrossed hybrids, descriptions of shifts over time have largely been inferred from historical collection records (Walters et al. 2008; Blum et al. 2010; Ward el al. 2012). Clinal discordance in the distribution of multilocus microsatellite genotypes, mtDNA haplotypes, and phenotypic traits further suggests that historical records do not fully capture the extent or rate of spread. Clinal discordance suggests that the spread of hybrids within the mainstem Coosa River is more geographically expansive than documented from observable patterns of phenotypic variation (Ward et al. 2012). For example, most hybrid individuals collected near the point of initial introduction exhibit a C. lutrensis phenotype (Ward et al. 2012), which indicates that retention of parental species’ phenotypes has likely masked the full extent of hybridization in the system (Ward et al. 2012). Here we examine temporal dynamics of

44

clinal variation in nuclear and mitochondrial genetic markers, as well as phenotypic traits, to better understand the progression of the C. lutrensis x C. venusta hybrid swarm in the

Coosa River. Doing so involved comparing clinal variation at and among three points of time spanning six years. This enabled us to infer the tempo and pace of change according to observable variation and cryptic introgression, which in turn enabled us to better assess the risk the invasion poses to remnant populations of vulnerable congeners.

MATERIALS AND METHODS

Study system and collections

Collection records suggest that C. lutrensis were introduced to the upper Coosa

River basin in Lake Weiss (Alabama) in 1974. Annual surveys first documented a large

C. lutrensis x C. venusta hybrid swarm in the mainstem Coosa River in 1998 (Burkhead and Huge 2002; Walters et al. 2008), with records suggesting that the leading edge has since progressed upstream at rates of up to 31km/year (Walters et al. 2008). Following collection records and prior genetic studies (Walters et al. 2008; Ward et al. 2012), we sampled 2,384 Cyprinella in the summer months of 2005, 2008, and 2011 to infer the tempo and pace of change across 47 sites spanning a 477 km transect of the mainstem

Upper Coosa River system (Table 2-1, Figure 2-1). Sampling locations included sites south of Lake Logan Martin in northern Alabama to sites on the Conasauga River beyond the Georgia-Tennessee state line (Table 2-1, Figure 2-1). The southerly extent of the transect increased over time, however, to acquire information on the potential downstream expansion of the hybrid swarm in the system. For example, in 2008 the transect began 100 km south of Lake Weiss at H. Neely Dam, whereas in 2011 we

45

sampled 100 km south at Logan Martin Lake. River distances between collection sites were measured from satellite imagery using Google Earth software. At each site, fish were collected by seine netting and then immediately placed in 95% ethanol for morphological and genetic analysis.

Phenotypic trait measurements

For all specimens, we measured four phenotypic traits that identify and differentiate C. lutrensis and C. venusta (Boschung and Mayden 2004). These traits include: standard length (SL), maximum body depth (BD), lateral line scale count, and caudal spot intensity. All measurements were taken on the left side of each specimen, and only individuals larger than 30mm SL were measured due to the difficulty of obtaining accurate lateral line scale counts from smaller individuals. For the 2005 collections, site

37 was excluded from phenotypic analysis because all specimens were <30mm in length.

However, these fish were included in genetic analysis. Individuals were measured for SL from the tip of the snout to the end of the caudal peduncle, using digital calipers calibrated to 0.01 mm accuracy. Maximum BD was measured from the anterior junction of the dorsal fin at the dorsal midline to the anterior junction of the anal fin at the ventral midline. We used the ratio of SL to BD as an index of body shape in subsequent analyses, obtained as residuals for each specimen from the linear regression of BD on SL. The intensity of the melanic caudal spot was scored on an increasing linear scale of 0 to 3, with 0 representing the complete absence of the caudal spot and 3 representing maximum intensity. Because color intensity varies among hybrids (Walters et al. 2008), scores of 1 and 2 were included to characterize weak and intermediate spot expression, respectively.

It has been previously documented (Ward et al. 2012) that the three phenotypic traits

46

(lateral line scales, size, and caudal spot intensity) are highly correlated for both species of Cyprinella; therefore we also implemented principal component analysis (PCA) to derive an overall phenotypic score for each individual. Trait decomposition yielded a single principal component that explained a high percentage of phenotypic variation

(81.6%).

Microsatellite genotyping and mtDNA-RFLP assays

We extracted and amplified DNA following Walters et al. (2008). For each individual, genomic DNA was extracted from approximately 0.05 g of preserved fin tissue using DNeasy kits (Qiagen, Valencia, CA, USA). With a few exceptions as noted below, polymerase chain reaction (PCR) mixtures for the amplification of both the complete cytochrome b gene (cyt b) and seven microsatellite loci included 2.5 mm

MgCl2, 2.5 mm of each dNTP, 0.5 units Taq DNA polymerase (Invitrogen, Carlsbad, CA,

USA), 0.5 µm of the oligonucleotide primers HA and LA (Schmidt et al. 1998), and 0.5

µm PCR buffer (Invitrogen). For some individuals, PCR mixtures for the amplification of cytochrome b (cyt b) were prepared using 0.1 units Paq (Stratagene, Santa Clara, CA,

USA) instead of Taq, with no addition MgCl2.

Restriction of cyt b with HinfI (New England Biolabs, Ipswich, MA, USA) generates species-specific fragment size profiles in C. venusta and C. lutrensis that can be reliably scored from fluoresced agarose electrophoresis gels. HinfI restriction of cyt b amplified from C. venusta generates 130, 480, and 530-bp fragments versus 95, 130, 350, and 570-bp fragments from C. lutrensis. Following Walters et al. (2008), all individuals were assigned species-level mtDNA ancestry from restriction of cyt b amplicons.

Individuals were also genotyped at seven polymorphic microsatellite markers developed

47

for other species that have been shown to distinguish parental C. lutrensis and C. venusta

(Walters et al. 2008). Four of the loci (Nme 25C8.208, Nme 18C2.178, Nme 24B6.191, and Nme 24B6.211) were developed for Notropis mekistocholas (Burridge and Gold

2003). Two loci (Rhca20 and Rhca24) were developed for Rhinichthys cataractae

(Girard and Angers 2006). The remaining locus (Can6EPA) was developed for

Campostoma anomalum (Dimsoski et al. 2000). PCR annealing temperatures were modified according to Walters et al. (2008) with reactions run using fluorescently labeled forward primers. Microsatellite PCR products were subsequently characterized on an

ABI 3730xl (Applied Biosystems Inc., Foster City, CA) and GeneMarker v9.0 software

(Softgenetics, State College, PA) against a LIZ 600 size standard (Applied Biosystems®,

Waltham, MA).

Analysis of admixture and genetic differentiation from microsatellite variation

Multilocus admixture profiles for all individuals were generated using the program Structure v2.3.1 (Pritchard et al. 2000). As in Ward et al. (2012), we undertook preliminary analyses to evaluate the relative contribution of individual loci to admixture profiles by comparing the results of runs generated using all seven loci with results of runs involving sequential removal of individual loci in the analysis. No loci were found to bias the results, and all loci were found to be informative. Five independent runs with

K= 2 (representing the parental species) were subsequently executed to characterize admixture profiles. For all runs, data were collected over 100,000 iterations, following a

50,000 iteration burn-in, under an admixture model of co-ancestry and correlated allele frequencies (Falush et al. 2003). Individuals were considered to be pure when the posterior probability of assignment to a parental class averaged over all runs was >90%.

48

For all hybrid individuals, admixture was assessed based on the average assignment values to the first parental cluster following the criteria outlined in Walters et al. (2008).

Clinal analysis of phenotype and molecular data

We used the R package HZAR (Derryberry et al. 2014) to fit clines to admixture profiles based on multilocus microsatellite genotypes, the relative frequencies of mtDNA haplotype assignments, and phenotype according to PCA scores. HZAR fits clines using the Metropolis–Hastings Markov chain Monte Carlo (MCMC) algorithm. Autofit functions allowed for automated model selection using Akaike’s Information Criterion

(AIC; Akaike 1973), from a set of nested cline models. We constructed a maximum likelihood profile for cline width (i.e., to assess concordance) and cline center (i.e., to assess coincidence) for multilocus genotype, mtDNA, and phenotype cline models. From the generated ML profiles, estimates of cline center and width corresponding to the largest logLik values were selected and used to calculate AIC scores. Next, for each sample year, AIC scores for cline center and width were calculated for each of the three cline models using the equation: AIC = -2(logLik)+2K. Concordance and coincidence for individual clines for each sample year (e.g., the 2005 genotype, mtDNA, and phenotype clines) and across sample years (e.g., the 2005, 2008, and 2011 mtDNA clines) were then assessed by comparing differences in AIC scores (ΔAIC). For example, if the AIC score of one cline center differed by at least two points, compared to another cline center, then the clines were considered non-coincident (Burnham and Anderson 2002; Anderson

2008). For all collection years we excluded samples obtained from sites south of Hokes

Bluff Ferry to avoid complications that can arise from cline fitting across upstream and downstream transitions. Thus we only modeled upstream clinal variation across the

49

hybrid swarm, where Hokes Bluff Ferry represents the southern terminus of a 352 km transect for estimates of cline center and clinal transitions.

RESULTS

Southern extent of red shiner and hybrids in the Upper Coosa River basin

In 2008 and 2011 we expanded our sampling to areas south of Hokes Bluff Ferry

(Site #9), to document the downstream distributional extent of red shiner and hybrids in the Upper Coosa River. In 2008, we collected red shiner and C. lutrensis x C. venusta hybrids at distances up to 54 km south of Hokes Bluff Ferry. The prevalence of red shiner and hybrids varied from site to site downstream of Hokes Bluff Ferry in 2008. For example, 22 km to the south at H. Neely Dam (Site #6), we collected fewer red shiner

(24%) than at a distance of 32 km south (Rainbow Landing, site #7), where 50% of individuals collected were red shiner. A greater proportion of admixed individuals were also collected from site #7 (Figure 2-2b). In an attempt to find the downstream distributional limit of red shiner and hybrids, in 2011 we extended our sampling transect another 56 km further downstream to include five sites around Lake Logan Martin (AL).

We found that the downstream distributional limit of red shiner occurs within the reservoir; red shiner were only found at the most northern site in the reservoir (Stemley

Bridge, Site #5). Sites further downstream did, however, harbor low frequencies of hybrids (Figure 2-2c). For example, we recovered one putative F1 hybrid and one backcross at our collection site furthest to the south (Glover’s Ferry, Site #1) just past

Lower Logan Martin Dam.

50

Phenotype variation across the Upper Coosa River basin

Across all sample years, phenotypic traits exhibited among populations at the southern (i.e., Hokes Bluff Ferry) and northern terminus sites were significantly different

(P = 0.000). The model based on 2005 collections estimated that the center of the phenotypic cline in 2005 was located at 117km from the southern terminus of the transect, and that the cline exhibited a width of 221 km (Table 2-2, Figure 2-3a). The

2008 cline model estimated that the center shifted northwards to 214 km from the southern terminus, and that the cline width narrowed to 121 km (Table 2-2, Figure 2-3a).

The 2011 cline model estimated that the center of the phenotypic cline had shifted even farther northward to 250 km from the southern terminus, and that it exhibited a width of

168 km (Table 2-2, Figure 2-3a). Cross-year comparisons found that the 2005 phenotypic cline models were significantly different than the 2008 and 2011 cline models in both center and width (Table 2-2).

Mitochondrial haplotype and genotypic variation across the Upper Coosa River basin

In all years, the frequency of the C. venusta mtDNA haplotype increased with northward distance from the southern terminus of the transect. At the southern terminus, approximately 20% of specimens exhibited a C. venusta haplotype in 2005, whereas only

4% of individuals exhibited a C. venusta haplotype in 2008. The frequency of individuals exhibiting a C. venusta haplotype rebounded, however, to 66% in 2011. Across all years, all individuals exhibited a C. venusta haplotype at distances greater than 300 km to the north of the southern terminus (Figure 2-3b). Models of haplotype variation estimated the cline center to be 156 km from the southern terminus in 2005, 202 km in 2008, and 197

51

km 2011 (Figure 2-3b). Models estimated the width of the mtDNA cline to be 264 km in

2005, 175 km in 2008 and 203 km in 2011 (Figure 2-3b).

The seven microsatellite loci exhibited a mean of 21 alleles among the fish genotyped from 47 locations along the sampled transect (Table 2-1, Figure 2-1). For all three collection years, we detected spatial structure in the relative frequencies of pure parental and hybrid multilocus admixture profiles based on assignment values (Figure 2-

2). Individuals exhibiting parental C. lutrensis genotypes were numerically dominant at the southern end of the transect and decreased in frequency towards the northern end of the transect. Cline centers from the southern terminus shifted northward across the sampling period: 136 km in 2005, 197 km in 2008, and 208 km in 2011 (Table 2-2;

Figure 2-3c). No pure parental C. lutrensis genotypes were detected at distances upstream of 252 km from the southern terminus in 2005, 315 km in 2008, and 340 km in 2011

(Figure 2-2). However, hybrid genotypes were recovered at sites reaching more than 50 km beyond the northernmost extent of parental C. lutrensis phenotypes (Figure 2-2). The proportion of admixed individuals averaged 15% in 2005, 17% in 2008, and 22% in

2011. Estimates of admixture at individual collection sites ranged as high as 36% in

2005, 38% in 2008 and 39% in 2011. The majority of hybrids were later-generation and backcrossed individuals constituting 84% of hybrids in 2005, 79% in 2008, and 80% in

2011. Backcrosses were more frequently observed in the direction of C. venusta for all three collection years.

Across sample years, mtDNA and multilocus genotype cline models did not significantly differ in either centers or widths (Table 2-2). However, model estimates indicate that the centers of both clines shifted northward over time (Table 2-2). The

52

mtDNA cline center estimate for 2005 was 156.2 km from the southern terminus, while center estimates for 2008 and 2011 were approximately 50 km further upstream (202 km and 197.3 km respectively; Table 2-2). Similarly, the multilocus genotype cline center estimate for 2005 was 136.2 km from the southern terminus, and center estimates for

2008 and 2011 were about 70 km further upstream (197.9 km and 207.8 km respectively;

Table 2-2). Model estimates indicate that the width of the mtDNA cline decreased from

2005 to 2008 (264.3 km vs. 175 km), whereas the estimated cline width increased from

2008 to 2011 (175 km vs. 202.6 km; Table 2-2). This same trend was also observed in the estimates of multilocus genotype cline widths; width estimate for 2005 was 283.3 km, and 187.5 km in 2008. In 2011, the estimated cline width increased to 192.1 km (Table 2-

2).

Comparison of phenotypic and genetic clines across years

The cline models describing genetic and phenotypic distributions along the Coosa mainstem transect exhibited both concordance and discordance across sample years

(Figures 2-3, 2-4). In each sample year, mtDNA and multilocus genotype cline models did not statistically differ in either center or width (Table 2-2) whereas both consistently differed from the estimated phenotypic cline centers and widths (Table 2-2). In 2005, for example, the center of the phenotype cline was more southerly and the cline was narrower than either the mtDNA or multilocus genotype cline (Table 2-2; Figure 2-4a).

Comparisons of clines over time indicate that the centers of all three clines shifted northward over time. However, the phenotypic clines exhibited a higher rate of northward advancement; in 2008, the center of the phenotypic cline advanced to the north of the mtDNA and multilocus cline centers (Table 2-2; Figure 2-4b). A similar pattern was

53

detected in 2011, when the center of the phenotypic was estimated to be even farther upstream from the mtDNA and multilocus genotype cline centers (Table 2-2; Figure 2-

4c). However, the width of the phenotypic cline remained narrower than the other clines in 2008 and 2011 (Table 2-2; Figure 2-4).

DISCUSSION

Hybrid swarms formed following the introduction of a non-native species have the potential to decrease species diversity through genetic extirpation of native taxa

(Huxel 1999; Epifanio and Philipp 2000; Hall et al. 2006; Ward et al. 2012). The potential for loss of native biota through biological invasions involving hybridization continues to rise as human-mediated introduction and invasions of non-native species have exponentially increased over time (Allendorf et al. 2001; Epifanio and Philipp 2000;

Crispo et al. 2011). Though prior studies examining the C. lutrensis x C. venusta hybrid swarm in the Upper Coosa River basin (Walters et al. 2008; Blum et al. 2010; Ward et al.

2012) have characterized the formation and spatial extent of the swarm, as well as factors contributing to hybridization, little information has been available on the stability and evolution of the swarm over time. Empirical analysis of phenotypic and genetic clines across hybrid zones and hybrid swarms over time is arguably the most reliable method for detecting movement and expansion (Buggs 2007). Here, we examined the spatio- temporal dynamics of the C. lutrensis x C. venusta hybrid swarm to assess prior inferences of movement and expansion based on historical collection records. According to the findings of previous studies (Walters et al. 2008; Ward et al. 2012) we expected to see progressive northward movement and expansion. Across a six year period, we found

54

evidence of significant northward shifts in phenotypic and genetic cline centers, and detected a modest increase in the percent of genetic admixture over time. Additionally, individuals with hybrid genotypes were detected further north as years progressed, and similarly, we detected the presence of hybrids farther downstream from the historical introduction site than found by prior surveys. Notably, we did not find progressive expansion of the swarm according to both phenotypic and genetic attributes. Rather, we detected a signature of contraction and expansion, suggesting that the stability and size of the swarm fluctuates, likely as a consequence of temporal shifts in extrinsic and intrinsic drivers of hybridization.

Genetic admixture over time

Elevated fitness of hybrid genotypes and the rise of advantageous traits in admixed populations can hasten the erosion of species boundaries or genetic assimilation of one parental species (Arnold 1997; Barton 2001; Coyne and Orr 2004; Hall et al.

2006). Blum et al. (2010) recently showed that C. lutrensis x C. venusta hybrids exhibit comparable to higher measures of post-zygotic fitness than offspring of parental species under laboratory conditions, suggesting that the proportion of hybrids may rise over time in the Upper Coosa River basin. We found that the majority of hybrids in the system were either later-generation or backcrossed individuals (Figure 2-2) confirming prior inferences that, once formed, hybrids persist in the system (Walters et al. 2008; Blum et al. 2010; Ward et al. 2012). We also observed a 7% increase in the proportion of genetic admixture over 6 years, confirming prior predictions that hybrids would become more dominant over time (Walters et al. 2008; Blum et al. 2010). Additionally, the proportion of red shiner and hybrids at northward sites along the transect has increased across the

55

sampling period (Figure 2-2). For example, in 2005 no evidence of hybridization was found at a distance of 320 km from the southern terminus of the transect whereas F2 and backcrossed genotypes were found in 2011. This suggests that the observed increase in hybridization is not limited to the rise of hybrids within a localized nucleus of sites (i.e., the center of the hybrid swarm). In agreement with theoretical predictions (Endler 1977;

Huxel 1999; Hall et al. 2006), this finding suggests that increases in the frequency of successfully reproducing later-generation hybrids is influencing the boundaries and fueling the movement of the C. lutrensis x C. venusta hybrid swarm.

The southern extent of the swarm

Prior studies of red shiner and C. lutrensis x C. venusta hybrids in the Upper

Coosa River basin have largely disregarded the potential for downstream spread from the point of introduction (Walters et al. 2008). Prior work has shown that red shiners were first collected in Weiss Lake in 1974 and subsequently collected downstream in nearby

Terrapin Creek, a tributary of the “Dead River” arm of the Coosa River in 1982 (Walters et al. 2008). Collection records indicate that red shiner or hybrids progressed upstream of

Weiss Lake as early as 1992, when they were collected in Coahulla Creek (Walters et al.

2008). Collections in 1998 revealed an extensive hybrid swarm extending upstream from

Lake Weiss to the confluence with the Conasauga River (Walters et al. 2008).

Subsequent annual surveys (Walters et al. 2008; Ward et al. 2012) have documented progressive spread of red shiner and C. lutrensis x C. venusta hybrids to areas of the upper Conasauga River that harbor the largest remaining population of Federally

Threatened blue shiner (Cyprinella caerulea), which has been shown to hybridize with red shiner under laboratory conditions (Burkhead, unpublished data). Our recovery of red

56

shiner and hybrid genotypes at distances of >200 km downstream of Lake Weiss indicates that red shiner and C. lutrensis x C. venusta hybrids pose a comparable threat to congeners elsewhere in the system. Downstream spread in the Coosa River is thus a greater conservation concern than previously thought, particularly to remnant populations of blue shiner in tributaries that feed in to Weiss Lake or downstream in to the mainstem

Coosa River.

Clinal discordance and hybrid swarm expansion

Traits under different selection regimes are expected to introgress across species boundaries at different rates (Harrison 1990; Mallet 2005; Yuri et al. 2009), with attributes under divergent selection introgressing to a lesser extent than those under neutral or positive selection (Gay et al. 2009; Maroja et al. 2009; Ward et al. 2012).

Consistent with this expectation, we found wider clines in multilocus genotype admixture profiles and mtDNA haplotypes relative to the clines describing variation at phenotypic traits that reflect functional differences among hybridizing taxa (Ward et al. 2012).

Similar to prior findings (Ward et al. 2012) our results indicate that microsatellite alleles and mtDNA haplotypes are introgressing more extensively than phenotypic traits.

Nonetheless, comparison to theoretical values of neutral traits also suggests that diffusion of alleles, haplotypes, and morphological attributes are constrained (Endler 1977). As C. lutrensis can produce two generations per year, and given a maximum estimated dispersal rate of 31 km/year (Walters et al. 2008), the approximate width of neutral clines could be as wide as 700 km (Farringer et al. 1979; Ward et al. 2012). All of the clines estimated in this study were narrower than expected under neutral diffusion, suggesting that there are factors structuring introgression across the hybrid swarm (Endler 1977; Gay et al. 2007;

57

Ward et al. 2012). It is possible that some indirect selection is acting on the molecular markers and phenotypic traits examined here, therefore constraining genetic and phenotypic introgression. This would not be unexpected given that the hybrid swarm is moving, and there is indication that selection is favoring C. lutrensis and C. lutrensis-like hybrids in the southern region of the transect. Also, in contrast to expectations, we found that the hybrid swarm did not progressively expand over time. Rather, we detected a signature of contraction and expansion at both phenotypic and genetic clines, suggesting that the stability and size of the swarm fluctuates in response to temporal shifts in extrinsic and intrinsic drivers of hybridization.

Tempo and rate of upstream movement

Hybrid zone movement (Barton and Hewitt 1985, 1989; Blum 2002; Buggs 2007) can be attributable to a range of factors including a selective advantage of one phenotype in a given environment (Goodman et al. 1999), dominance drive (Blum 2002), asymmetric hybridization (Bronson et al. 2003; Buggs and Pannell 2006), hybrid fitness

(Klingenberg et al. 2000), anthropogenic environmental disturbance (Blum 2002), and climate change (Britch et al. 2001). Prior work (Blum et al. 2010; Ward et al. 2012) indicates that the dynamics of the C. lutrensis x C. venusta hybrid swarm nominally reflect hybrid fitness as well as selective advantages conferred by the red shiner phenotype. Parental C. lutrensis can become the dominant fish in native range and non- native range assemblages because of an aggressive disposition, short generation times, and high fecundity (DeVivo 1995; Fuller et al. 1999; Burkhead and Huge 2002).

Evidence of increasing dominance of invasive C. lutrensis-like hybrids in the southern region of the transect is consistent with the inference that C. lutrensis traits are selectively

58

favored and retained in admixed populations (Bensch et al. 1999; Balloux et al. 2000;

Rees et al. 2003; Vallender et al. 2007; Brelsford and Irwin 2009; Ward et al. 2012).

Additionally, upstream movement of the phenotypic cline, at rates faster than the movement of genetic clines, provides further support for the inference that phenotypic advantages are contributing to movement of the hybrid swarm (Ward et al. 2012).

Annual surveys of the Conasauga River completed from 2000 to 2003 indicate that red shiner and C. lutrensis x C. venusta hybrids can disperse at rates up to 31 km/year (Walters et al. 2008). Cline models based on collections completed from 2005 to

2011, however, indicate that upstream movement of the C. lutrensis x C. venusta hybrid swarm has proceeded at a slower pace than suggested by collection records. Between

2005 and 2011, the center of the phenotypic cline advanced approximately 133 km upstream at an average rate of 22 km/year. The centers of the mtDNA and multilocus genotypic clines advanced at a slower pace of approximately 8-12 km/year. Notably, comparison of cline centers over time suggests that the rate of upstream movement has been declining. The center of the phenotypic cline proceeded upstream at a rate of 32 km/year between 2005 and 2008, compared to a rate of 12 km / year between 2008 and

2011. The centers of the mtDNA and genotypic clines moved at a rate of 15-21 km/year between 2005 and 2008, and effectively remained stable between 2008 and 2011. It is possible that upstream movement is slowing because red shiner and hybrids are encountering unfavorable ecological conditions. Red shiner tend to prefer warmer, low- elevation habitats with sand or silt substrate (Matthews and Hill 1979). At a distance of approximately 350km from the southern terminus, the Conasauga River begins to transition into lower order, cooler and higher elevation reaches. Past the Georgia-

59

Tennessee state line, the substrate of the Conasauga River is also largely composed of large boulders and sedimentary limestone. Though further monitoring and experimental tests are warranted to test this inference, it is likely that habitat transitions could eventually constrain further upstream movement of the C. lutrensis x C. venusta swarm

(Buggs 2007; Blum 2008), and possibly prevent further overlap with threatened blue shiner in the Conasauga River.

Conclusions

The invasive C. lutrensis x C. venusta hybrid swarm in the Upper Coosa River basin affords exceptional opportunities to understand the formation and evolution of hybrid swarms over time. Our findings suggest that introgressive hybridization is continuing to drive the advancement of the hybrid swarm (Walters et al. 2008; Blum et al. 2010; Ward et al. 2012), and that the movement of phenotypic, mtDNA and genotypic clines can progress at different rates over time. We also found evidence that, despite directional movement, the extent of the hybrid swarm has fluctuated over time, and that ecological constraints may impede further progress. For the first time, we also documented the downstream extent of red shiner and hybrids, some 200 km south of the original introduction site at Lake Weiss. These findings suggest that, should further upstream progress be constrained, invasive red shiner and hybrids are likely to advance further downstream in the Coosa River, which could threaten additional populations of vulnerable native Cyprinella species. Further, it is curious to note that of the four native species of Cyprinella found in the Upper Coosa River system, to date there is no documentation of invasive red shiner hybridizing with two of these species, the tricolor shiner (C. trichroistia), and the Alabama shiner (C. callistia). Future investigation of

60

hybridization with these two species is warranted, and further monitoring at both the upstream and downstream leading edges of the present swarm could offer further information to better protect native Cyprinella, and perhaps mitigate future spread to areas of conservation concern in the basin.

61

Table 2-1. Summary data for 47 locations where Cyprinella were sampled in the Upper

Coosa River Basin. Population numbers correspond with the numbers on Figure 2-1.

2005 2008 2011 Latitude Longitude Phenotype mtDNA Msat Phenotype mtDNA Msat Phenotype mtDNA Msat Population (N) (W) n (n) (n) (n) n (n) (n) (n) n (n) (n) (n) 1 33.390 -86.378 23 23 22 23 2 33.427 -86.325 11 11 11 11 3 33.447 -86.290 14 14 8 14 4 33.457 -86.296 18 18 14 16 5 33.522 -86.229 9 9 9 9 6 33.786 -86.065 30 30 29 25 7 33.943 -86.028 42 42 38 40 7 7 5 7 8 34.002 -86.002 1 1 0 1 9 33.997 -85.881 35 15 35 35 3 3 3 3 10 34.113 -85.853 179 36 176 177 10 2 9 10 11 34.091 -85.744 26 17 26 26 12 34.288 -85.669 12 12 10 10 35 8 34 35 13 34.239 -85.601 29 29 23 29 14 34.162 -85.472 33 33 28 33 15 34.165 -85.396 15 10 11 11 1 1 1 1 2 1 2 2 16 34.251 -85.381 5 5 5 5 17 34.200 -85.256 20 19 20 20 16 16 15 16 46 44 37 37 18 34.255 -85.178 110 7 106 106 13 13 13 12 19 34.315 -85.118 129 123 129 128 20 34.371 -85.125 26 23 26 26 29 29 21 28 21 34.380 -85.124 19 19 16 19 108 63 94 87 22 34.411 -85.107 68 51 64 64 23 34.450 -85.027 19 12 19 19 24 34.468 -85.033 26 22 26 26 25 34.476 -85.030 44 41 43 43 26 34.494 -85.011 26 26 26 26 31 31 27 28 27 34.510 -84.958 39 34 39 39 32 32 29 31 26 16 24 26 28 34.529 -84.966 29 25 28 28 29 34.573 -84.945 45 36 45 44 7 7 7 6 30 34.577 -84.942 58 58 47 58 122 108 112 115 31 34.541 -84.901 14 14 14 14 14 14 12 14 11 11 11 11 32 34.595 -84.928 24 21 24 24 33 34.667 -84.931 43 37 43 43 27 27 27 26 37 33 31 36 34 34.667 -84.933 12 8 12 12 35 34.709 -84.868 22 0 22 22 28 28 28 28 34 34 30 33 36 34.672 -84.825 19 19 16 19 37 34.736 -84.857 26 25 25 25 54 54 49 44 29 26 22 29 38 34.783 -84.872 39 23 39 39 25 25 25 25 36 36 33 36 39 34.811 -84.861 26 26 26 26 40 34.817 -84.857 17 6 17 17 41 34.828 -84.851 16 16 14 16 29 28 29 29 42 34.853 -84.838 22 18 22 22 25 25 25 25 32 31 30 32 43 34.895 -84.829 11 8 11 11 44 34.904 -84.828 14 14 14 14 45 34.920 -84.842 11 9 10 10 1 1 1 1 22 22 14 22 46 34.992 -84.778 24 24 24 22 23 23 23 23 47 35.010 -84.734 7 7 7 7 Total 1078 670 1060 1059 542 542 485 517 782 663 697 736

62

Table 2-2. Comparison of genetic and phenotypic clines models across the C. lutrensis x

C. venusta hybrid swarm from 2005 to 2011. Number of model parameters (npar),

Akaike’s Information Criterion (AIC), center AIC score (AICc), width AIC score

(AICw). All distance measures are expressed in kilometers. Cline centers are expressed as the fluvial distance from the southern terminus at Hokes Bluff Ferry (site #9).

Model Center Width npar LogLikeC LogLikeW AICc AICw 2005 Phenotype 116.667 220.833 11 -221.396 -224.634 464.792 471.269 mtDNA 156.250 264.375 2 -2.737 -2.737 9.475 9.475 Msat 136.250 283.333 2 -3.810 -3.810 11.620 11.620 2008 Phenotype 213.542 121.250 3 -63.212 -63.194 132.424 132.388 mtDNA 202.083 175.000 2 -2.825 -2.825 9.649 9.649 Msat 197.917 187.500 2 -3.832 -3.831 11.663 11.662 2011 Phenotype 250.000 167.500 3 -60.526 -60.522 127.052 127.044 mtDNA 197.368 202.632 2 -2.390 -2.388 8.779 8.775 Msat 207.895 192.105 2 -3.367 -3.367 10.733 10.733

63

Figure 2-1: 47 locations sampled between 2005 and 2011, along a 477-km transect of the

Coosa River, Oostanaula River, and Conasauga River (Alabama, Georgia, Tennessee;

USA). Red circles = sites unique to 2005, black circles = sites unique to 2008, blue circles = sites unique to 2011, and green circles = sites sampled in two or more years.

64

Figures 2-2a-c: Proportion of parental Cyprinella venusta (BT), Cyprinella lutrensis

(RS), and hybrid (F1, F2, BCbt, and BCr) individuals at 47 sites (as listed in Table 2-1) across sampling transect in: A: 2005, B: 2008, and C: 2011. Collection site #9 (Hokes

Bluff Ferry) represents the southern terminus of the cline models, and collection site # 13

(Weiss Lake) is the original introduction site into the Upper Coosa River Basin.

A BT RS F1 F2 BCbt BCr

100%

80%

60%

40%

20% Cumulative Proportion (%) Proportion Cumulative 0% 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 Mainstem Site (South - North)

65

B BT RS F1 F2 BCbt BCr

100%

(%) 80%

60%

40%

20% Cumulaative Proportion Cumulaative 0% 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 Mainstem Site (South - North)

C BT RS F1 F2 BCbt BCr

100%

80%

60%

40%

20% Cumulative Proportion (%) Proportion Cumulative 0% 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 Mainstem Site (South - North)

66

Figures 2-3a-c. Clinal changes in frequencies of Cyprinella venusta phenotype, mtDNA haplotypes, and microsatellite multilocus genotype between 2005 and 2011. A: phenotype cline models, B: mtDNA haplotype cline models, C: multilocus genotype cline models. Hokes Bluff Ferry (site #9) is serving as the southern terminus.

67

68

Figure 2-4a-c. Clinal changes in frequencies of Cyprinella venusta phenotype, mtDNA haplotypes, and microsatellite multilocus genotype. A: 2005 combined cline models, B:

2008 combined cline models, C: 2011 combined cline models. Hokes Bluff Ferry (site

#9) is serving as the southern terminus.

69

70

CHAPTER 3

TURBIDITY ALTERS PRE-MATING SOCIAL INTERACTIONS BETWEEN

NATIVE AND INVASIVE STREAM FISHES

ABSTRACT

Environmental degradation can result in the loss of aquatic biodiversity if impairment promotes hybridization between non-native and native species. Though aquatic biological invasions involving hybridization have been attributed to elevated water turbidity, the extent to which impaired clarity influences reproductive isolation among non-native and native species is poorly understood. We examined whether turbidity influences intra- and interspecific pre-mating social interactions between invasive red shiner (Cyprinella lutrensis) and native blacktail shiner (Cyprinella venusta) from the Upper Coosa River Basin (USA). We found that the number or duration of conspecific and heterospecific interactions increased under turbid conditions.

Additionally, we found evidence indicating that native blacktail shiner females are especially likely to interact with invasive red shiner males due to species and sex specific responses to turbid conditions. These findings suggest that elevated turbidity can increase pre-mating social interactions between native and invasive species, which could result in greater hybridization and promote the genetic assimilation of native species following species introductions. Therefore integrating knowledge of species behavior into

71

conservation and management planning can help deter the establishment and spread of invasive species.

INTRODUCTION

Environmental degradation can result in the loss of biological diversity in aquatic ecosystems, particularly in areas that harbor non-native species. Disturbance can precipitate the displacement of native fishes by non-native fishes, which are often more tolerant of degraded conditions (Moyle and Light 1996; Gido and Brown 1999; Marchetti et al. 2004; Walters et al. 2008). Degradation may also result in the loss of aquatic biological diversity if impairment results in hybridization between non-native and native species (Rhymer and Simberloff 1996). However, the extent to which environmental impairment promotes hybridization following species introductions is not well understood (Walters et al. 2008; Ward and Blum 2012; Coleman et al. 2014). Elevated turbidity has been linked to observations of hybridization involving invasive stream fishes (e.g., Page and Smith 1970; Larimore and Bayley 1996; Walters et al. 2008), but no tests have been carried out to determine whether impaired water clarity influences reproductive isolation between invasive and native species.

Anthropogenic habitat modification that reduces water clarity can erode reproductive isolation between closely related species by disrupting the transmission and reception of signals used for species recognition and mate identification (Seehausen

1997; Järvenpää and Lindström 2004; Fisher et al. 2006; Van der Sluijs et al. 2011).

Turbid conditions inhibit light transmission, which can dampen visual signals such as body coloration or courtship behavior involving movement (Seehausen 1997; Järvenpää

72

and Lindström 2004). Alteration or interruption of communication during reproduction can weaken sexual selection and reproductive isolation, enough to result in species loss

(Seehausen 1997; Järvenpää and Lindström 2004; Taylor et al. 2006; Candolin et al.

2007; Heuschele et. al. 2009). For example, by interfering with mate choice, algal turbidity relaxes sexual selection and reduces reproductive isolation among African rift lake cichlids (Seehausen et al. 1997). Work on threespine stickleback (Gasterosteus aculeatus) in estuarine environments further suggests that algal turbidity reduces the evolutionary potential of sexual selection by diminishing the efficacy of visual displays and by weakening socially enforced signals of male quality (Wong et al. 2007).

The introduction of non-native red shiner (Cyprinella lutrensis) into the Upper

Coosa River Basin (Alabama, Georgia, Tennessee; USA) provides an opportunity to determine whether impaired water clarity can lead to the loss of biological diversity by promoting hybridization between non-native and native species. Repeated surveys indicate that a hybrid swarm involving introduced red shiner and native blacktail shiner

(C. venusta) formed around 1998, following at least 6 years of intermittent hybridization

(Walters et al. 2008). Since 2001, the hybrid swarm has been rapidly expanding upstream into a region that is widely considered to be a global hotspot of aquatic biodiversity

(Warren et al. 2000; Abell et al. 2008; Walters et al. 2008; Ward et al. 2012). Though colonization remains largely limited to mainstem reaches (Walters et al. 2008; Ward et al.

2012), the presence of hybrids in tributaries is associated with elevated turbidity from agricultural land use (Walters et al. 2008). This is consistent with observed episodes of hybridization between naturally sympatric red shiner and blacktail shiner in the

Guadalupe River (Texas) and San Marcos River (Texas), which coincided with increased

73

turbidity resulting from excessive sedimentation (Jurgens 1951, Hubbs et al. 1953).

Hybridization in the San Marcos River declined as water quality improved, leading

Hubbs and Strawn (1956) to hypothesize that turbidity disrupts species recognition during spawning, and by extension, that hybridization between C. lutrensis and C. venusta abates because prezygotic isolation is reinstated as turbidity declines.

While the inference that turbidity disrupts species recognition during courtship in

Cyprinella (Hubbs and Strawn 1956) is consistent with field surveys and studies that show increased turbidity can weaken or alter sexual selection in other fishes (Järvenpää and Lindström 2004; Candolin et al. 2007; Engström-Öst and Candolin 2007), no tests have been carried out to explicitly assess the influence of visual impairment on social interactions between invasive red shiner and native blacktail shiner (Walters et al. 2008).

Cyprinella venusta and C. lutrensis are crevice-spawning species that aggregate in mixed-species groups during the breeding season, which increases the likelihood that anthropogenic shifts in social interactions can alter rates of hybridization. Prior experiments demonstrated that prezygotic isolation between C. lutrensis and C. venusta is stronger than postzygotic isolation (Blum et al. 2010) and that chemical impairment of water quality (i.e., exposure to environmental estrogens) weakens prezygotic isolation between the two species (Ward and Blum, 2012). Elevated turbidity could similarly weaken reproductive isolation between the two species if visual impairment increases interactions between heterospecific members of the opposite sex.

In this study, we conducted social behavior assays with reproductively active

(tuberculated males, gravid females), wild-caught invasive red shiner and native blacktail shiner to test the hypothesis that elevated turbidity alters pre-mating social behavior of

74

individuals spawning in mixed-species aggregations. We examined whether turbid conditions increase the number and the duration of interactions between male and female shiners in reproductive condition. Based on prior field surveys and laboratory experiments (Walters et al. 2008; Blum et al. 2010; Ward and Blum 2012; Ward et al.

2012), we predicted that blacktail shiner females would exhibit comparably weak conspecific preference and that heterospecific interactions would increase under turbid conditions, particularly between blacktail shiner females and red shiner males. Assessing these predictions enabled us to evaluate the extent to which visual impairment of freshwater environments due to elevated turbidity can promote pre-mating social interactions between heterospecifics, and provide potential opportunity for increased hybridization following species introductions.

METHODS

Specimen collection, housing and maintenance

We collected live fish via seine netting in May 2010. Blacktail shiners were collected from Sugar Creek, Crandall, Georgia (34.942, -84.835), and red shiners from

Proctor Creek, Atlanta, Georgia (33.794, -84.474). Blacktail shiners and red shiners do not co-occur at either of these collection sites, and there is no evidence of hybridization at either location. Sugar Creek is located approximately 175 km north of the Coosa River confluence, and Proctor creek is located in the Chattahoochee-Flint River Basin, 83 km southeast of the Coosa River confluence. Each species was transported to the lab in a separate 378 L polyethylene tank filled with aged, chemically conditioned water (Ammo-

Lock® and Stress Coat®; API, Hackettstown, NJ, USA). In the lab, fish were segregated

75

into single-species, mixed-sex groups in 378 L opaque polyethylene holding tanks with a continuous flow-through of filtered and UV-sterilized water. Acrylic spawning towers (as described in Blum et al. 2010) were placed in the holding tanks to provide artificial spawning habitat, and to maintain a supply of reproductively active individuals. We maintained all fish in 24-26 °C water and under a 16 h:8 h light/dark regime to mimic the ambient environment during the peak spawning months of May and June. All fish were fed three times daily with flake food (Brine Shrimp Direct, Ogden, UT, USA).

Cyprinella spawning behavior and behavioral assays

During the spawning season, C. venusta and C. lutrensis males aggregate over spawning substrate and engage in aggressive male-male interactions. Females tend to remain separate until they are inclined to spawn, when they will approach males to initiate breeding (Minckley 1972). A male will court a female by circling around her, eventually leading her to a spawning site (e.g., rocks, twigs, etc.) where the female will deposit her eggs. The male quickly fertilizes the eggs after which the spawning partners separate (Minckley 1972; Gale 1986).

To test the hypothesis that turbidity alters pre-mating social interactions between the species, we observed the behavior of individual fish in four different social and environmental scenarios (n = 20 trials for each combination; 80 total trials). We constructed two different mixed-species social groups, each consisting of three subjects.

Both groups had one male red shiner and one male blacktail shiner. The third subject in the group was either a female red shiner or a female blacktail shiner. We then observed the behavior of all three subjects in clear water and turbid environments. Forty female red shiner groupings and forty female blacktail shiner groupings were subjected to trials

76

under clear (control) and turbid water conditions (n = 40 trials per species; 20 trials per treatment). Turbid trials were carried out at a turbidity level of 100 mg/L (30 NTU), which falls within the observed range of conditions (10-1,500 mg/L, 3-450 NTU) for the

Coosa River Basin (W.O. McLarney, unpublished data) and approximates the upper range of baseflow NTUs (25.7) measured in the basin (Walters et al. 2008). We used 20 different pairs of red shiner and blacktail shiner males for tests conducted under turbid conditions, and another 20 pairs for tests conducted under clear conditions (i.e., 40 pairs total). To control for heterogeneity across male pairs, we tested one female of each species (n = 20 for each species) with each set of males (Kozak et al. 2008; Ward and

Blum 2012). This experimental design permitted us to record behavioral interactions between focal subjects across all possible combinations of sex and species under conditions of mixed-species aggregation.

Clear water control trials were conducted in fresh, triple-carbon-filtered water that was free from visual impediment. For turbid trials, we altered the visual environment in the test tank by adding a 100 mg/L suspension of fine river clay particles (collected near

Gainesville, FL). Higher turbidity levels could not be tested because fish were not visible at levels above 100 mg/L. Nonetheless, this level is substantially higher than levels of turbidity examined in prior studies of estuarine and lake fish responses (Heuschele et al.

2009) and it is at the lower end of turbidity measured in the Coosa River Basin. We drained and wiped down the test tank between each group of subjects to eliminate residual chemical cues.

77

Test Procedure

We conducted tests in a three-chambered, 208 L experimental tank lit by two

15W full-spectrum bulbs suspended 10 cm above the water (Figure 3-1). We sectioned the central chamber of the experimental tank into three 18.5 cm zones demarcated by lines drawn on the external glass face (Figure 3-1). The two zones closest to the flanking compartments were designated as ‘association zones’ and the central zone a ‘neutral zone’ (Figure 3-1). Details on the testing procedure are provided in Ward and Blum

(2012). Briefly, one male red shiner and one male blacktail shiner were randomly selected from the group holding tanks and introduced to the distal compartments of the test tank. Males were permitted a one-hour acclimation period. We placed a spawning tower against the back wall of each distal chamber to encourage intersexual social behavior. The distal compartments were separated from the central compartment by clear

Plexiglas barriers that prevented the exchange of chemical cues. Black acrylic dividers, fitted over the clear Plexiglas panels, prevented visual interactions between the males during acclimation. At the end of the male acclimation period, we introduced either a female red shiner or a female blacktail shiner to the central compartment. The female was permitted a 10 min acclimation period, after which the black dividers were removed. We recorded the interactions between the female subject and both male subjects for 10 min from behind a blind using a Canon model GZ-MG555U digital video camera. Males were permitted a one-hour rest period between trials. We balanced trials according to the order of female species presentation (red shiner or blacktail shiner), as well as the relative flanking positions of the male subjects (left or right of the central compartment).

78

Social behavior

We used two well-established measures of social preference to assess intra- and interspecific interactions between subjects (Bischoff et al. 1985; Engström-Öst and

Candolin 2007; Ward and McLennan 2009). We quantified the number of times each subject approached conspecific and heterospecific members of the opposite sex. We also recorded the total amount of time (in s) that each subject spent associating with conspecific and heterospecific members of the opposite sex. Continuous focal observations were conducted for each individual based on video recordings. For female subjects, we recorded an approach as having occurred when the subject entered the association zone adjacent to a male’s compartment. We also recorded the duration of time

(s) that the female subject spent in the association zone of each male subject. For male subjects, we recorded an approach as having occurred when a male moved toward the female subject and made contact with the divider with his snout. We used the duration of time that the male’s snout was in contact with the divider as our measure of association time (s).

Data analysis

We used generalized estimating equations (GEE; Hardin and Hilbe 2012) to fit separate models for each response variable and sex: the number of approaches, social association time, and male and female subjects were analyzed separately. Each of the four models examined the main effects of species (blacktail vs red shiner) and treatment

(clear vs turbid) on subject behavior. We determined whether male or female subjects showed conspecific social preferences by including the male species x female species interaction term. We also assessed differences between the species in social behavior

79

across visual environments by including the species x treatment interaction term. Finally, we examined whether male or female subjects from either species differed in their behavior towards conspecific and heterospecific members of the opposite sex across treatments by including the three-way male species x female species x treatment interaction term. Models based on social association time or on the number of approaches were specified using identity link functions or poisson link functions, respectively. We used an unstructured correlation structure for each model, which assumes a generalized correlation matrix among repeated or non-independent responses. We validated the appropriateness of the correlation structure in preliminary models using the Quasi

Likelihood Under Independence Model Criterion (QIC; Pan 2001; Hardin and Hilbe

2012). We removed non-significant interaction terms prior to final analyses, and compared models containing significant interaction terms to those containing only main effects. We selected the best sets of model terms using the Corrected Quasi Likelihood

Under Independence Model Criterion (QICu) (Hardin and Hilbe, 2012; Pan, 2001).

Where appropriate, we compared the responses of significant terms using pairwise Least

Significant Difference (LSD) posthoc tests based on marginal means.

RESULTS

Species differences in social behavior in mixed-species aggregations

Generalized estimating equations (GEE) revealed a significant main effect of species on social behavior (species terms; Tables 3-1 and 3-2). In three out of four models, red shiners demonstrated an overall higher level of sociality than blacktail shiners. Across both treatments, male red shiners approached female subjects more frequently than did male blacktail shiners (Figures 3-2a, 3-2b). Red shiner males also

80

spent more time associating with female subjects across both treatments (Figures 3-2c, 3-

2d). Female red shiners spent more time associating with male subjects than did female blacktail shiners across both treatments (species terms; Tables 3-1 and 3-2), although female blacktail shiners approached male subjects more frequently than female red shiners (Figures 3-3a, 3-3b).

Examination of the male species x female species interaction term in each model confirmed that male and female subjects from both species demonstrated social preferences for conspecifics (Tables 3-1 and 3-2). When both treatments were considered, male and female subjects spent more time associating with conspecific subjects than with heterospecific subjects (all, posthoc LSD test, P < 0.001; Figures 3-2c,

3-2d, 3-3c, 3-3d). Similarly, female red shiners and males from both species approached conspecific subjects more frequently than heterospecific subjects (all, P < 0.007; Figures

3-2a, 3-2b, 3-3b). However, female blacktail shiners approached conspecific and heterospecific subjects with equal frequency (P = 0.403; Figure 3-3a).

Effects of turbidity on social behavior in mixed species aggregations

We observed differences in the social behavior of females and males towards members of the opposite sex across treatments. However, preliminary analyses indicated that the three-way male species x female species x treatment interaction term was not significant for either model of female behavior (association time: χ2 = 3.13, df = 2, P =

0.21; approaches: χ2 = 3.89, df = 2, P = 0.14) or for the number of male approaches towards females (χ2 = 1.47, df = 2, P = 0.48). These results indicate that observed changes in these behaviors across treatments did not depend on the species identity of the opposite sex (i.e., conspecific or heterospecific). Thus we removed these terms in further

81

analyses. Final models identifying a significant main effect of treatment on subject responses found that social behavior among subjects in mixed-species aggregations was elevated in turbid conditions compared to behavior exhibited in clear water (Tables 3-1 and 3-2).

Pooling the data for male subjects of both species indicated that male subjects spent more time associating with females in turbid conditions (treatment term; Table 3-2;

Figures 3-2c, 3-2d). Although the male species x treatment interaction term for this model was not significant (Table 3-2), the three-way male species x female species x treatment interaction term was significant (Table 3-2). Subsequent examination of the data indicated that the responses of male red and blacktail shiners towards heterospecific females differed according to treatment. In clear water, males of both species spent similar amounts of time associating with heterospecific females (P = 0.721; Figures 3-2c,

3-2d). However, under turbid conditions red shiner males spent more time associating with heterospecific females (i.e., with blacktail shiner females) than did blacktail males

(P = 0.032). The total association time (female plus male heterospecific association time) between red shiner females and blacktail shiner males declined by about half under turbid conditions compared with clear water conditions (from 188 to 110 s), but doubled (from

130 to 261 s) for red shiner males and blacktail shiner females. The number of times that male subjects approached female subjects did not differ between treatments (treatment term; Table 3-1; Figures 3-2a, 3-2b); the male species x treatment interaction term for this model was not significant, indicating that red shiner and blacktail shiner males exhibited similar changes in this pre-mating social behavior under turbid conditions.

82

Pooling the data for females of both species indicated that female subjects were more likely to approach male subjects in turbid water than in clear water (treatment term;

Table 3-1; Figures 3-3a, 3-3b). Examination of the significant female species x treatment interaction term for this model revealed that female red shiner responses to turbidity were different from responses of female blacktail shiners (Table 3-1). Female red shiners approached male subjects significantly more often in turbid than in clear water conditions

(P = 0.007; Figure 3-3b), whereas there was no difference between treatments in the number of times that female blacktail shiners approached males (P = 0.844; Figure 3-3a).

The amount of time that female subjects spent associating with male subjects did not differ between treatments (treatment term; Table 3-2; Figures 3-3c, 3-3d).

DISCUSSION

Our results support the hypothesis that elevated turbidity alters pre-mating social interactions among native and non-native fishes that spawn in mixed-species aggregations. We found that invasive red shiner demonstrated higher levels of sociality than native blacktail shiner, and as predicted, female blacktail shiner discriminated less among conspecific and heterospecific mates when compared to female red shiners. Our trials similarly showed that behavioral responses to changes in visual environment varied according to species and sex. Contrary to our other prediction, we did not observe consistently greater heterospecific interactions in turbid conditions. Rather, we observed increases in the number or duration of pre-mating social interactions with members of the opposite sex in turbid water. We did find, however, that red shiner males spent more time associating with blacktail shiner females in turbid conditions. These findings indicate that elevated turbidity can increase pre-mating social interactions between native and invasive

83

species, which could result in greater hybridization and promote the genetic assimilation of native species following species introductions.

Effects of turbidity on the behavior of Cyprinella in mixed-species aggregations

By disrupting the transmission and reception of signals, elevated turbidity impairs species recognition, alters sexual selection, and possibly breaks down prezygotic isolation among congeners. As in prior studies of fish responses to reduced water clarity

(Seehausen 1997; Järvenpää and Lindström 2004; Fisher et al. 2006; Candolin et al.

2007), we observed that female behavior expressed in turbid conditions differed from behavior expressed in clear water. Maan et al. (2010), for example, found that in Lake

Victoria cichlids of the genus Pundamilia, female mate choice preferences differed as a result of elevated turbidity, where females from areas of high turbidity failed to discriminate among conspecific and heterospecific males. Likewise, mate choice trials involving two closely related species of haplochromine chichlids (Haplochromis nyererei, Haplochrimis zebra nyererei) demonstrated that females of both species exhibited species-assortative mate choice under white light conditions, but chose non- assortatively when differences in male color were masked by monochromatic light conditions (Seehausen and Van Alphen 1998). In contrast to these studies, we found that female responses to elevated turbidity were asymmetrical. Relative to blacktail shiner females, red shiner females exhibited a stronger preference for conspecific mates and approached male subjects significantly more often in turbid conditions.

The observed asymmetry in female social behavior could be attributable to variation in male interactivity. As in many other fish (Ryan and Keddy-Hector 1992), females of both species prefer males that expend greater courtship effort (Ward and

84

Blum, unpubl. data). Thus the weaker conspecific preference exhibited by blacktail shiner females could be due to male red shiner exhibiting higher levels of sociality than blacktail shiner males and red shiner males engaging in higher levels of pre-mating social behavior with heterospecific females (i.e., with blacktail shiner females) in turbid conditions. Red shiner females approaching male subjects more often in turbid conditions could be attributable to males of both species interacting more with females under turbid conditions. The observed increase in male interactivity is comparable to findings indicating that male stickleback spend more time courting conspecific females and court more intensely in water with greater filamentous algal growth (Candolin et al. 2007).

Under turbid conditions, greater male courtship effort is required to sustain female interest (Engstrom-Ost and Candolin 2007). Similarly, elevated interactivity in red shiner and blacktail shiner males could be a compensatory response to visual cues being muted under turbid conditions, where elevated turbidity requires greater effort by males to elicit responses from females. If so, males would be expected to exhibit more active engagement to attract female attention. This may reflect a general female preference either for traits or behaviors that elicit greater sensory stimulation (Ryan and Keddy-

Hector 1992) or for a trait that indicates aspects of male quality or condition (Milinski and Bakker 1990, Bakker and Mundwiler 1994). However, changes in female C. venusta and C. lutrensis pre-mating social behavior due to anthropogenic modification of the environment are not necessarily linked to variation in male behavior or attributes (Ward and Blum 2012). Further experiments will be required to clearly determine the extent to which female responses are driven by male interactivity.

85

Conservation and management implications

Prior studies of elevated turbidity and reproductive isolation have largely focused on responses of naturally syntopic fishes to eutrophication of estuarine and lake ecosystems (Seehausen 1997; Järvenpää and Lindström 2004; Candolin et al. 2007;

Heuschele et. al. 2009; Van der Sluijs et al. 2011). Work on African rift lake cichlids and threespine sticklebacks (Seehausen et al. 1997; Taylor et al. 2006) has shown that hybridization resulting from environmental degradation can lead to the loss of native biodiversity through species collapse (Seehausen et al. 1997; Taylor et al. 2006;

Seehausen 2006; Seehausen et al. 2008). The influence of elevated turbidity on the sensory environment in riverine ecosystems has not been as well studied, though prior research suggests that impairment can influence the transmission and reception of signals, including those that help maintain species boundaries. For example, low levels of turbidity reduce responses to visual stimuli by endangered fountain darter (Etheostoma fonticola) (Swanbrow Becker and Gabor 2012), and chemically mediated species recognition in sheepshead swordtails is hindered by anthropogenic increases in humic acid content (Fisher et al. 2006). Episodes of hybridization between naturally sympatric red shiner and blacktail shiner in the Guadalupe River (Texas) and San Marcos River

(Texas) also have been observed during periods of elevated turbidity (Jurgens 1951;

Hubbs et al. 1953).

By extension, it is logical to expect that changes in water quality could impair visual signaling and disrupt social interactions among native and invasive fishes, including riverine species that are often subject to naturally turbid conditions, like red shiner and blacktail shiner. Walters et al. (2008) showed that the spread of invasive C.

86

lutrensis x C. venusta hybrids in the Coosa River basin corresponds to turbid conditions associated with agricultural land use, which parallels inferences that hybridization among naturally sympatric red shiner and blacktail shiner is linked to elevated turbidity from excessive sedimentation (Jurgens 1951; Hubbs et al. 1953; Hubbs and Strawn 1956). We found that conditions of elevated turbidity modify pre-mating social interactions between invasive red shiner and native blacktail shiner. Our results suggest that elevated turbidity escalates the intensity of pre-mating social behavior, which could increase the number of interspecific interactions resulting in hybridization, particularly under more prolonged or severe conditions of environmental degradation (McGree et al.2010; Ward and Blum

2012). Additionally, we found that red shiner males spent more time associating with heterospecific blacktail shiner females in turbid conditions. This finding, along with the finding that female blacktail shiner do not strongly discriminate between conspecific and heterospecific mates, is consistent with a prior study demonstrating that female blacktail shiner are more likely to spawn with male red shiner (Blum et al. 2010) as well as evidence of proportionally greater maternal genetic contributions from blacktail shiner in post-F1 C. lutrensis x C. venusta hybrids (Walters et al. 2008; Ward et al. 2012). Thus, the reproductive fitness of native blacktail shiner females is likely more at risk when water clarity is impaired due to natural or anthropogenic factors (e.g., sediment runoff from agricultural and urban land use), as has been inferred for species that occupy naturally oligotrophic estuarine and lake ecosystems (Seehausen 1997; Järvenpää and Lindström

2004; Candolin et al. 2007).

Understanding how environmental degradation impairs species recognition and social behavior can aide in the development of strategies to manage the spread of aquatic

87

invasive species. Environmental impairment can promote the loss of native aquatic biodiversity by weakening prezygotic reproductive isolation between native and invasive species (Rhymer and Simberloff 1996; Ward and Blum 2012). Our findings indicate that elevated turbidity gives rise to greater interspecific interactions that could increase the risk of hybridization between red shiner and blacktail shiner, which in turn suggests that elevated turbidity puts other congeners at greater risk of hybridization. Cyprinella is the second most speciose genus of freshwater fish in North America (Warren et al. 2000).

Not only has hybridization been observed between red shiner and several naturally co- occurring congeners (Jurgens 1951, Hubbs et al. 1953, Hubbs and Strawn 1956, Page and

Smith 1970, Larimore and Bayley 1996), invasive red shiner are known to hybridize with at least nine native congeners in the southeastern United States (summarized in Walters et al. 2008). Adopting new approaches that account for ecological and evolutionary processes may be most effective in reducing risks posed by red shiner– particularly in the southeastern United States where fish exhibit high rates of endemism and imperilment

(e.g., nine species of Cyprinella are endemic to a single Southeastern river drainage;

Warren et al. 2000; Ashley et al. 2003; Walters et al. 2008). Complete eradication is the desired management endpoint with nuisance species like red shiner, but it is often not feasible because of cost and logistical constraints. Coupling well-conceived control and mitigation efforts could, however, reduce the presence of red shiner while also

(re)establishing ecological functions that promote the success of native Cyprinella congeners. For example, remediating impaired water quality conditions to reduce turbidity could suppress hybridization by strengthening prezygotic isolation (Hubbs and

Strawn 1956). Though remediation may not elicit shifts in reproductive behavior

88

sufficient to prevent genetic dilution and assimilation, improving water quality could slow the pace of hybridization, which would afford more time to implement complementary control strategies to prevent further displacement of native Cyprinella by invasive red shiner or hybrid swarms (Walters et al. 2008; Ward et al. 2012).

89

Table 3-1. GEE models tests examining the effects of species (blacktail shiner, red shiner) and treatment (turbid, clear) on the number of approaches made by male and female subjects towards opposite-sex subjects in mixed-species aggregations. Significant effects are given in bold (α < 0.050).

Number of approaches χ2 df Sig. Male Male Species 7.618 1 0.006 Treatment 2.514 1 0.113 Male Species*Treatment 0.586 1 0.444 Male Species*Female Species 45.699 2 0.000

Female Female Species 9.594 1 0.002 Treatment 7.746 1 0.005 Female Species * Treatment 6.515 1 0.011 Female Species * Male Species 6.118 2 0.047

90

Table 3-2. GEE model tests examining the effects of species (blacktail shiner, red shiner) and treatment (turbid, clear) on the total time (s) spent by male and female subjects associating with opposite-sex subjects in mixed-species aggregations. Significant effects are given in bold (α < 0.050).

Association time χ2 df Sig. Male Male Species 9.561 1 0.002 Treatment 6.494 1 0.011 Male Species*Treatment 1.768 1 0.184 Male Species*Female Species 64.877 2 0.000 Male Species*Female Species*Treatment 6.177 2 0.046 Female Female Species 7.054 1 0.008 Treatment 1.763 1 0.184 Female Species * Treatment 1.132 1 0.287 Female Species * Male Species 65.97 2 0.000

91

Figure 3-1. A diagram of the three-chambered experimental tank. L = length, W = width,

H = height. Female association (with male A or B) was measured only in the association zones as time and approaches.

92

Figure 3-2. Mean (± sem) behavioral responses of male blacktail shiners (A, C) and male red shiners (B, D) towards conspecific and heterospecific females under clear and turbid visual conditions.

93

Figure 3-3. Mean (± sem) behavioral responses of female blacktail shiners (A, C) and female red shiners (B, D) towards conspecific and heterospecific males under clear and turbid visual conditions.

94

LIST OF REFERENCES

Abell, R., M. L. Thieme, C. Revenga, M. Bryer, M. Kottelat, N. Bogutskaya, P. Petry, et al. 2008. Freshwater Ecoregions of the World: A New Map of Biogeographic Units for Freshwater Biodiversity Conservation. BioScience 58:403–414.

Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle. In B. N. Petrov, and F. Csaki, eds. Second International Symposium on Information Theory, pp. 267–281. Akademiai Kiado, Budapest.

Allan, J.D., and A. S. Flecker 1993. Biodiversity conservation in running waters. BioScience 43:32–43.

Allendorf, F. W., R. F. Leary, P. Spruell, and J. K. Wenburg 2001. The problems with hybrids: setting conservation guidelines. Trends in ecology and evolutionary biology 16:613–622.

Allendorf, F.W., and L. L. Lundquist 2003. Introduction: population biology, evolution, and control of invasive species. Conservation Biology 17:24–30.

Araújo, M. B., and A. T. Peterson 2012. Uses and Misuses of Bioclimatic Envelope Modeling. Ecology 93:1527-1539.

Arnold, M. L. 1997. Natural Hybridization and Evolution: Oxford Series in Ecology and Evolution. Oxford University Press, Oxford.

Ashley, M. V., M. F. Willson, O. R.W. Pergams, D. J. O’Dowd, S. M. Gende, and J. S. Brown 2003. Evolutionarily enlightened management. Biological Conservation 111:115–123.

Avise, J. C., J. Arnold, R. M. Ball, E. Bermingham, T. Lamb, J. E. Neigel, et al. 1987. Intraspecific phylogeography: the mitochondrial DNA bridge between population genetics and systematics. Annual review of ecology and systematics 489-522.

Ayres, D. R., E. Grotkopp, K. Zaremba, C. M. Sloop, M. J. Blum, J. P. Bailey, C. K. Anttila, and D. R. Strong 2008. Hybridization between invasive Spartina densiflora (Poaceae) and native S. foliosa in San Francisco Bay, California, USA. American Journal of Botany 95:713-719.

95

Bakker, T. C. M., and B. Mundwiler 1994. Female mate choice and male red coloration in a natural three-spined stickleback (Gasterosteus aculeatus) population. Behavioral Ecology 5:74–80.

Balloux, F., H. Brünner, N. Lugon-Moulin, J. Hausser, and J. Goudet 2000. Microsatellites can be misleading: an empirical and simulation study. Evolution 54:1414–1422.

Bandelt, H. J., P. Forster, and A. Röhl 1999. Median-joining networks for inferring intraspecific phylogenies. Molecular Biology and Evolution 16:37.

Barton, N. H. 2001. The role of hybridization in evolution. Molecular Ecology 10:551- 568.

Barton, N. H., and G. M. Hewitt 1985. Analysis of hybrid zones. Annual review of Ecology and Systematics 113-148.

Barton, N. H., and G. M. Hewitt 1989. Adaptation, speciation and hybrid zones. Nature 341:497-503.

Barton, N. H., and K. S. Gale 1993. Genetic analysis of hybrid zones. In: Harrison, R. G. (ed.) Hybrid Zones and the Evolutionary Process pp. 13–45. Oxford University Press, Oxford.

Behm, J. E., A. R. Ives, and J. W. Boughman 2010. Breakdown in postmating isolation and the collapse of a species pair through hybridization. The American Naturalist 175:11-26.

Bensch, S., T. Andersson, and S. Akesson 1999. Morphological and molecular variation across a migratory divide in willow warblers. Phylloscopus trochilus. Evolution 53:1925–1935.

Bentzen, P., W. C. Leggett, and G. G. Brown 1993. Genetic relationships among the shads (Aloso) revealed by mitochondria1 DNA analysis. Journal of Fish Biology 43:909-917.

Bermingham, E., and J. C. Avise 1986. Molecular zoogeography of freshwater fishes in the southeastern United States. Genetics 113:939-965.

Bettles, C. M., M. F. Docker, B. Dufour, and D. D. Heath 2005. Hybridization dynamics between sympatric species of trout: loss of reproductive isolation. Journal of Evolutionary Biology 18:1220-1233.

Bischoff, R. J., J. L. Gould, and D. I. Rubenstein 1985. Tail size and female choice in the guppy (Poecilia reticulata). Behavioral Ecology and Sociobiology 17:253–255.

96

Blakeslee, A. M. H., C. H. McKenzie, J. A. Darling, J. E. Byers, J. M. Pringle, and J. Roman 2010. A hitchhiker’s guide to the Maritimes: anthropogenic transport facilitates long-distance dispersal of an invasive marine crab to Newfoundland. Diversity and Distributions 16:879–891.

Blum, M. J. 2002. Rapid movement of a Heliconius hybrid zone: evidence for phase III of Wright's shifting balance theory? Evolution 56:1992-1998.

Blum, M. J. 2008. Ecological and genetic associations across a Heliconius hybrid zone. Journal of Evolutionary Biology 21:330–341.

Blum, M. J., K. Jun Bando, M. Katz, and D. R. Strong 2007. Geographic structure, genetic diversity and source tracking of Spartina alterniflora. Journal of Biogeography 34:2055-2069.

Blum, M. J., D. M. Walters, N. M. Burkhead, B. J. Freeman, and B. A. Porter 2010. Reproductive isolation and the expansion of an invasive hybrid swarm. Biological Invasions 12:2825-2836.

Boschung, H. R., and R. L. Mayden. 2004. Fishes of Alabama. Smithsonian Books, Washington, DC.

Bradley, B. A. 2009. Regional Analysis of the Impacts of Climate Change on Cheatgrass Invasion Shows Potential Risk and Opportunity. Global Change Biology 15:196- 208.

Brelsford, A., and D. E. Irwin. 2009. Incipient speciation despite little assortative mating: the yellow-rumped warbler hybrid zone. Evolution 63:3050–3060.

Britch, S. C., M. L. Cain, and D. J. Howard 2001. Spatio-temporal dynamics of the Allonemobius fasciatus-A. socius mosaic hybrid zone: a 14-year perspective. Molecular Ecology 10:627–638.

Bronson, C. L., T. C. Grubb, G. D. Sattler, and M. J. Braun 2003. Mate preference: a possible causal mechanism for a moving hybrid zone. Animal Behavior 65:489– 500.

Broughton, R. E., and J. R. Gold 2000. Phylogenetic Relationships in the North American Cyprinid Genus Cyprinella (Actinopterygii: Cyprinidae) Based on Sequences of the Mitochondrial ND2 and ND4L Genes. Copeia 2000:1-10.

Broughton, R. E., K. C. Vedala, T. M. Crowl, and L. L. Ritterhouse 2011. Current and historical hybridization with differential introgression among three species of cyprinid fishes (Genus Cyprinella). Genetica 139:699-707.

97

Brown, B. L., and R. W. Chapman 1991. Gene flow and mitochondrial DNA variation in the killifish, Fundulus heteroclitus. Evolution 45:1147-1161.

Buckley, L. B., S. A. Waaser, H. J. MacLean, and R. Fox 2011. Does including physiology improve species distribution model predictions of responses to recent climate change? Ecology 92:2214-2221.

Buggs, R. J. A. 2007. Empirical study of hybrid zone movement. Heredity 99:301-312.

Buggs, R. J. A., and J. R. Pannell 2006. Rapid displacement of a monoecious plant lineage is due to pollen swamping by a dioecious relative. Current Biology 16:996–1000.

Burkhead, N. M., and D. H. Huge 2002. The case of the Red Shiner: What happens when a fish goes bad? United States Geological Survey. http://cars.er.usgs.gov/Southeastern_Aquatic_Fauna/Freshwater_Fishes/Shiner_R esearch/shiner_research.html (accessed on 23 march 2015).

Burridge, C. P., and J. R. Gold 2003. Conservation genetic studies of the endangered Cape Fear shiner, Notropis mekistocholas (Teleostei: Cyprinidae). Conservation Genetics 4:219–225.

Callaway, R. M., and E. T. Aschehoug 2000. Invasive plants versus their new and old neighbors: a mechanism for exotic invasion. Science 290:521–523.

Candolin, U., T. Salesto, and M. Evers 2007. Changed environmental conditions weaken sexual selection in sticklebacks. Journal of Evolutionary Biology 20:233–239.

Casal, C. M. V. 2006. Global documentation of fish introductions: the growing crisis and recommendations for action. Biological Invasions 8:3–11.

Chick, J. H., and M. A. Pegg 2001. Invasive Carp in the Mississippi River Basin. Science 292:2250-2251.

Childs, M.R., A. A. Echelle, and T. E. Dowling 1996. Development of the hybrid swarm between Pecos pupfish (Cyprinodontidae: Cyprinodon pecosensis) and sheepshead minnow (Cyprinodon variegatus): a perspective from allozymes and mtDNA. Evolution 50:2014–2022.

Coleman, R. R., M. R. Gaither, B. Kimokeo, F. G. Stanton, B. W. Bowen, and R. J. Toonen 2014. Large-scale introduction of the Indo-Pacific damselfish Abudefduf vaigiensis into Hawai'i promotes genetic swamping of the endemic congener A. abdominalis. Molecular Ecology 23:5552–5565.

98

Compton, T. J., J. R. Leathwick, and G. J. Inglis 2010. Thermogeography predicts the potential global range of the invasive European green crab (Carcinus maenas). Diversity and Distributions 16:243-255.

Cornuet, J. M., P. Pudlo, J. Veyssier, A. Dehne-Garcia, M. Gautier, R. Leblois et al. 2014. DIYABC v2. 0: A software to make approximate Bayesian computation inferences about population history using single nucleotide polymorphism, DNA sequence and microsatellite data. Bioinformatics 30:1187-1189.

Coyne, J., and H. A. Orr. 2004. Speciation. Sinaeur, Sunderland.

Crispo, E., J. S. Moore, J. A. Lee‐Yaw, S. M. Gray, and B. C. Haller 2011. Broken barriers: Human‐induced changes to gene flow and introgression in animals. BioEssays 33:508-518.

Cross, F. B., and L. M. Cavin 1971. Effects of pollution, especially from feedlots, on fishes in the upper Neosho River basin. Contribution 79. Kansas Water Resources Institute, Manhattan.

Darling, J. A., M. J. Bagley, J. Roman, C. K. Tepolt, and J. B. Geller 2008. Genetic patterns across multiple introductions of the globally invasive crab genus Carcinus. Molecular Ecology 17:4992-5007.

Deacon, J. E. 1988. The endangered woundfin and water management in the Virgin River, Utah, Arizona, Nevada. Fisheries 13:18-29.

Derryberry, E. P., G. E. Derryberry, J. M. Maley, and R. T. Brumfield 2014. HZAR: hybrid zone analysis using an R software package. Molecular ecology resources 14:652-663.

DeVivo, J. C. 1995. Impact of introduced red shiner, Cyprinella lutrensis, on stream fishes near Atlanta, Georgia. In Proceedings of the 1995 Georgia Water Resources Conference, pp. 95–98.

Dill, W. A., and A. J. Cordone 1997. History and status of introduced fishes in California, 1871-1996. California Department of Fish and Game Fish Bulletin, volume 178.

Dimsoski, P., G. P. Toth, and M. J. Bagley. 2000. Microsatellite characterization in central stoneroller Campostoma anomalum (Pisces: Cyprinidae). Molecular Ecology 9:2187–2189.

Diver, T. 2013. Disentangling Historical Biogeography and Anthropogenic Introductions: A Case Study of Red Shiner (Cyprinella lutrensis). M. S. Thesis. University of New Mexico, Albuquerque, NM, USA.

99

Douglas, M. E., P. C. Marsh, and W. L Minckley 1994. Indigenous fishes of western North America and the hypothesis of competitive displacement: Meda fulgida (Cyprinidae) as a case study. Copeia 9-19.

Dupanloup, I., S. Schneider, and L. Excoffier 2002. A simulated annealing approach to define the genetic structure of populations. Molecular Ecology 11:2571-81.

Echelle, A. A., and P. J. Connor 1989. Rapid geographically extensive genetic introgression after secondary contact between two pupfish species (Cyprinodon, Cyprinodontidae). Evolution 43:717–727.

Ellstrand, N. C., and K. A. Schierenbeck 2000. Hybridization as a stimulus for the evolution of invasiveness in plants? Proceedings of the National Academy of Sciences 97:7043–7705.

Endler, J. A. 1977. Geographic Variation, Speciation, and Clines. Princeton University Press, Princeton, New Jersey.

Engström-Öst, J., and U. Candolin 2007. Human-induced water turbidity alters selection on sexual displays in sticklebacks. Behavioral Ecology 18:393–398.

Epifanio, J., and D. Philipp 2000. Simulating the extinction of parental lineages from introgressive hybridization: the effects of fitness, initial proportions of parental taxa, and mate choice. Reviews in Fish Biology and Fisheries 10:339-354.

Estoup, A., and T. Guillemaud 2010. Reconstructing routes of invasion using genetic data: why, how and so what? Molecular Ecology 19:4113-4130.

Estoup, A., E. Lombaert, J. M. Marin, T. Guillemaud, P. Pudlo, C. P. Robert, and J. Cornuet 2012. Estimation of demo‐genetic model probabilities with Approximate Bayesian Computation using linear discriminant analysis on summary statistics. Molecular ecology resources 12:846-855.

Excoffier, L., and H. E. L. Lischer 2010. Arlequin suite ver 3.5: A new series of programs to perform population genetics analyses under Linux and Windows. Molecular Ecology Resources 10:564-567.

Falush, D., M. Stephens, and J. K. Pritchard 2003. Inference of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics 164:1567–1587.

Farringer, R. T., A. A. Echelle, and S. F. Lehtinen 1979. Reproductive cycle of the red shiner, Notropis lutrensis, in central Texas and south central Oklahoma. Transactions of the American Fisheries Society 108:271–276.

100

Fisher, H. S., B. B. M. Wong, and G. G. Rosenthal 2006. Alteration of the chemical environment disrupts communication in a freshwater fish. Proceedings of the Royal Society B: Biological Sciences 273:1187–1193.

Forbes, S. H., and F. W. Allendorf 1991. Associations between Mitochondrial and Nuclear Genotypes in Cutthroate Trout Hybrid Swarms. Evolution 1332-1349.

Frankham, R. 2005. Invasion biology – resolving the genetic paradox in invasive species. Heredity 94:385.

Frankham, R., J. D. Ballou, and D. A. Briscoe 2002. Introduction to Conservation Genetics. Cambridge University Press, Cambridge.

Fuller, P. L., L. G. Nico, and J. D. Williams 1999. Nonindigenous Fishes Introduced into Inland Waters of the United States. Special Publication 27. American Fisheries Society, Bethesda, Maryland.

Gale, W. F. 1986. Indeterminant fecundity and spawning behavior of captive red shiners—fractional, crevice spawners. Transactions of the American Fisheries Society 115:429–437.

Gay, L., G. Neubauer, M. Zagalska-Neubauer, C. Debain, J. M. Pons, P. David, and P. A. Crochet 2007. Molecular and morphological patterns of introgression between two large white-headed gull species in a zone of recent secondary contact. Molecular Ecology 16:3215–3227.

Gay, L., G. Neubauer, M. Zagalska-Neubauer, J. M. Pons, D. A. Bell, and P. A. Crochet 2009. Speciation with gene flow in the large white-headed gulls: does selection counterbalance introgression? Heredity102:133–146.

Gido, K. B., and J. H. Brown 1999. Invasion of North American drainages by alien fish species. Freshwater Biology 42:387–399.

Gido, K. B., J. F. Schaefer, K. Work, P. W. Lienesch, E. Marsh-Matthews, and W. J. Matthews 1999. Effects of red shiner (Cyprinella lutrensisis) on red river pupfish (Cyprinodon rubrofluviatilis). Southwestern Naturalist 44: 287-295.

Gilman, R. T., and J. E. Behm 2011. Hybridization, species collapse, and species reemergence after disturbance to premating mechanisms of reproductive isolation. Evolution 65:2592–2605.

Girard, P., and B. Angers 2006. Characterization of microsatellite loci in longnose dace (Rhinichthys cataractae) and interspecific amplification in five other Leusciscinae species. Molecular Ecology Notes 6:69–71.

101

Goodman, S. J., N. H. Barton, G. Swanson, K. Abernethy, and J. M. Pemberton 1999. Introgression through rare hybridization: a genetic study of a hybrid zone between red and sika deer (genus Cervus) in Argyll, Scotland. Genetics 152:355–371.

Gozlan, R. E., J. R. Britton, I. Cowx, and G. H. Copp 2010. Current knowledge on non- native freshwater fish introductions. Journal of Fish Biology 76:751–786.

Greger, P. D., and J. E. Deacon 1988. Food partitioning among fishes of the Virgin River. Copeia 314-323.

Guillemaud, T., M. A. Beaumont, M. Ciosi, J. M. Cornuet, and A. Estoup 2010. Inferring introduction routes of invasive species using approximate Bayesian computation on microsatellite data. Heredity 104:88–99.

Guindon, S., J. F. Dufayard, V. Lefort, M. Anisimova, W. Hordijk, and O. Gascuel 2010. New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0. Systematic biology 59:307-321.

Hall, R. J., A. Hastings, and D. R. Ayres 2006. Explaining the explosion: modelling a hybrid invasion. Proceedings of the Royal Society of London. Series B: Biological Sciences 273:1385–1389.

Hardin, J. W., and J. M. Hilbe 2012. Generalized Estimating Equations, 2nd ed, Chapman & Hall/CRC, Boca Raton, Florida.

Harrison, R. G. 1990. Hybrid zones: windows on the evolutionary process. Oxford Surveys in Evolutionary Biology 7:69–128.

Harrison, R. G., and S. M. Bogdanowicz 1997. Patterns of variation and linkage disequilibrium in a field cricket hybrid zone. Evolution 493-505.

Hazelton, P. D., and G. D. Grossman 2009. The effects of turbidity and an invasive species on foraging success of rosyside dace (Clinostomus funduloides). Freshwater Biology 54:1977–1989.

Heuschele, J., M. Mannerla, P. Gienapp, and U. Candolin 2009. Environment-dependent use of mate choice cues in sticklebacks. Behavioral Ecology 20:1223–1227.

Holden, P. B., and C. B. Stalnaker 1975. Distribution and abundance of mainstream fishes of the middle and upper Colorado River basins, 1967-1973. Transactions of the American Fisheries Society 104:217-231.

Hubbs, C. L., and C. Hubbs 1953. An Improved Graphical Analysis and Comparison of Series of Samples. Systematic Zoology 2:49.

102

Hubbs, C., and K. Strawn 1956. Interfertility Between Two Sympatric Fishes, Notropis lutrensis and Notropis venustus. Evolution 10:341.

Hubbs, C. L., and K. F. Lagler 1964. Fishes of the Great Lakes region. University of Michigan Press, Ann Arbor, Michigan, pp 213.

Huxel, G. R. 1999. Rapid displacement of native species by invasive species: effects of hybridization. Biological Conservation 89:143–152.

Järvenpää, M., and K. Lindström 2004. Water turbidity by algal blooms causes mating system breakdown in a shallow-water fish, the sand goby Pomatoschistus minutus. Proceedings of the Royal Society of London. Series B: Biological Sciences 271:2361–2365.

Jennings, M. R., and M. Saiki 1990. Establishment of Red Shiner, Notropis lutrensis, in the San Joaquin Valley, California. California Fish and Game 76:46-57.

Jeschke, J. M., and D. L. Strayer 2006. Determinants of vertebrate invasion success in Europe and North America. Global Change Biology 12:1608–1619.

Jiggins, C. D., and J. Mallet 2000. Bimodal hybrid zones and speciation. Trends in Ecology & Evolution 15:250-255.

Johnson, C. E. 1999. The relationship of spawning mode to conservation of North American minnows (Cyprinidae). Environmental Biology of Fishes 55:21–30.

Jurgens, K. C. 1951. The distribution and ecology of the fishes of the San Marcos River. M.S. Thesis, The University of Texas, Austin, TX, USA.

Kaufman, L. 1992. Catastrophic change in species-rich freshwater ecosystems. BioScience 846-858.

Kelley, A. L., C. E. de Rivera, and B. A. Buckley 2011. Intraspecific variation in thermotolerance and morphology of the invasive European green crab, Carcinus maenas, on the west coast of North America. Journal of Experimental Marine Biology and Ecology 409:70-78.

Klingenberg, C. P., J. R. Spence, and C. K. Mirth 2000. Introgressive hybridization between two species of waterstriders (Hemiptera: Gerridae: Limnoporus): geographical structure and temporal change of a hybrid zone. Journal of Evolutionary Biology 13:756–765.

Knapp, F. T. 1953. Fishes Found in the Fresh Waters of Texas. Ragland Studio and Litho Printing Co., Brunswick, Georgia.

103

Knapp, R. A., and K. R. Matthews 2000. Non-Native Fish Introductions and the Decline of the Mountain Yellow-Legged Frog from within Protected Areas. Conservation Biology 14:428-438.

Kolar, C. S., and D. M. Lodge 2001. Progress in invasion biology: predicting invaders. Trends in Ecology & Evolution 16:199–204.

Kozak, H. L., L. A. Cirino, and M. B. Ptacek 2008. Female mating preferences for male morphological traits used in species and mate recognition in the Mexican sailfin mollies, Poecilia velifera and Poecilia petenensis. Behavioral Ecology 19:463– 474.

Kozak, G. M., M. L. Head, and J. W. Boughman 2011. Sexual imprinting on ecologically divergent traits leads to sexual isolation in sticklebacks. Proceedings of the Royal Society B: Biological Sciences 278:2604–2610.

Lack, J. B., D. U. Greene, C. J. Conroy, M. J. Hamilton, J. K. Braun, M. A. Mares, and R. A. Van Den Bussche 2012. Invasion facilitates hybridization with introgression in the Rattus rattus species complex. Molecular Ecology 21:3545–3561.

Larimore, R.W., and P. B. Bayley 1996. The fishes of Champaign County, Illinois, during a century of alterations of a prairie ecosystem. Illinois Natural History Survey Bulletin 35:53- 183.

Le Roux, J., and A. M. Wieczorek 2009. Molecular systematics and population genetics of biological invasions: towards a better understanding of invasive species management. Annals of Applied Biology 154:1–17.

Lee, C. E. 2002. Evolutionary genetics of invasive species. Trends in Ecology & Evolution 17:386-391.

Leprieur, F., M. A. Hickey, C. J. Arbuckle, G. P. Closs, S. Brosse, and C. R. Townsend 2006. Hydrological disturbance benefits a native fish at the expense of an exotic fish. Journal of Applied Ecology 43:930-939.

Lever, C. 1996. Naturalized Fish of the World. Academic Press, San Diego, California.

Librado, P., and J. Rozas 2009. DnaSP v5: A software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25:1451-1452.

Lodge, D. M. 1993. Biological invasions: lessons for ecology. Trends in Ecology and Evolution 8:133–137.

Maan, M. E., O. Seehausen, and J. J. M. Van Alphen 2010. Female mating preferences and male coloration covary with water transparency in a Lake Victoria cichlid fish. Biological Journal of the Linnean Society 99:398–406.

104

Mallet, J. 2005. Hybridization as an invasion of the genome. Trends in Ecology & Evolution 20: 229–237.

Marchetti, M. P., T. Light, P. B. Moyle, and J. H. Viers 2004. Fish invasions in California watersheds: Testing hypotheses using landscape patterns. Ecological Applications 14:1507–1525.

Maroja, L. S., J. A. Andre´s, and R. G. Harrison 2009. Genealogical discordance and patterns of introgression and selection across a cricket hybrid zone. Evolution 63:2999–3015.

Marsh-Matthews, E., and W. J. Matthews 2000. Spatial variation in relative abundance of a widespread, numerically dominant fish species and its effect on fish assemblage structure. Oecologia 125:283-292.

Martin, A. P., and S. R. Palumbi 1993. Body size, metabolic rate, generation time, and the molecular clock. Proceedings of the National Academy of Sciences 90:4087- 4091.

Matthews, W. J. 1987. Geographic variation in Cyprinella lutrensis (Pisces: Cyprinidae) in the United States, with notes on Cyprinella lepida. Copeia 616-637.

Matthews, W. J., and L. G. Hill 1977. Tolerance of the red shiner, Notropis lutrensis (Cyprinidae) to environmental parameters. Southwestern Naturalist 22:89-98.

Matthews, W. J., and L. G. Hill 1979. Influence of physico-chemical factors on habitat selection by red shiners, Notropis lutrensis (Pisces: Cyprinidae). Copeia 70-81.

Matthews, W. J. 1985. Distribution of Midwestern Fishes on Multivariate Environmental Gradients, with Emphasis on Notropis lutrensis. American Midland Naturalist 113:225-237.

Mayden, R. L. 1989. Phylogenetic studies of North American minnows, with emphasis on the genus Cyprinella (Teleostei: Cypriniformes). University of Kansas, Museum of Natural History Miscellaneous Publication No. 80.

McDonald, D. B., T. L. Parchman, M. R. Bower, W. A. Hubert, and F. J. Rahel 2008. An introduced and a native vertebrate hybridize to form a genetic bridge to a second native species. Proceedings of the National Academy of Sciences 105:10837- 10842.

McGree, M. M., D. L. Winkelman, N. K. M. Vieira, and A. M. Vajda 2010. Reproductive failure of the red shiner (Cyprinella lutrensis) after exposure to an exogenous estrogen. Canadian Journal of Fisheries and Aquatic Sciences 67:1730–1743.

105

Milinski, M., and T. C. M. Bakker 1990. Female sticklebacks use male coloration in mate choice and hence avoid parasitized males. Nature 344:330–333.

Minckley, W. L. 1972. Notes on the spawning behavior of the red shiner, introduced into Burro Creek, Arizona. Southwestern Naturalist 17:101-103.

Minckley, W. L. 1973. Fishes of Arizona. Arizona Game and Fish Department, Tempe, Arizona.

Minckley, W. L., and J. E. Deacon 1968. Southwestern fishes and the enigma of "endangered species". Science 159:1424-1432.

Moyle, P. B. 1976. Fish introductions in California: history and impact on native fishes. Biological Conservation 9:101-118.

Moyle, P. B. 2002. Inland fishes of California. University of California Press, Berkeley, California.

Moyle, P. B., and R. A. Leidy 1992. Loss of biodiversity in aquatic ecosystems: evidence from fish faunas. Conservation biology 127-169.

Moyle, P. B., and T. Light 1996. Biological invasions of fresh water: Empirical rules and assembly theory. Biological Conservation 78:149–161.

Moyle, P. B., and T. Light 1996. Fish invasions in California: do abiotic factors determine success? Ecology 1666-1670.

Muirhead, J. R., D. K. Gray, D. W. Kelly, S. M. Ellis, D. Heath et al. 2008. Identifying the source of species invasions: sampling intensity vs. genetic diversity. Molecular Ecology 17: 1020–1035.

Neigel, J. E., and J. C. Avise 1986. Phylogenetic relationships of mitochondrial DNA under various demographic models of speciation. In: Evolutionary Processes and Theory. Academic Press, New York, pp. 515-534.

Nico, L., P. Fuller, and M. Neilson 2015. Cyprinella lutrensis. USGS Nonindigenous Aquatic Species Database, Gainesville, Florida. http://nas.er.usgs.gov/queries/factsheet.aspx?SpeciesID=518 Revision Date: 1/27/2012.

Olden, J. D., N. L. Poff, M. R. Douglas, M. E. Douglas, and K. D. Fausch 2004. Ecological and evolutionary consequences of biotic homogenization. Trends in Ecology and Evolution 19:18-24.

106

Page, L. M., and R. L. Smith 1970. Recent range adjustments and hybridization of Notropis lutrensis and Notropis spilopterus in Illinois. Transactions of the Illinois State Academy of Science 63:264-272.

Pan, W. 2001. Akaike's information criterion in generalized estimating equations. Biometrics 57:120–125.

Paradis, E., J. Claude, and K. Strimmer 2004. APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20:289-290.

Pflieger, W. L. 1997. The fishes of Missouri. Missouri Department of Conservation, Jefferson City, Missouri. pp 372.

Poulos, H. M., B. Chernoff, P. L. Fuller, and D. Butman 2012. Mapping the potential distribution of the invasive red shiner, Cyprinella lutrensis (Teleostei: Cyprinidae) across waterways of the conterminous United States. Aquatic Invasions 7:377- 385.

Poulos, H. M., and B. Chernoff 2014. Potential range expansion of the invasive Red Shiner, Cyprinella lutrensis (Teleostei: Cyprinidae), under future climatic change. Open Journal of Ecology 4:554-564.

Pritchard, J. K., M. Stephens, and P. Donnelly 2000. Inference of population structure using multilocus genotype data. Genetics155:945–959.

Purcell, K. M., N. Ling, and C. A. Stockwell 2012. Evaluation of the introduction history and genetic diversity of a serially introduced fish population in New Zealand. Biological Invasions 14:2057-2065.

Purcell, K. M., and C. A. Stockwell 2015. An evaluation of the genetic structure and post-introduction dispersal of a non-native invasive fish to the North Island of New Zealand. Biological Invasions 17:625-636.

R Core Team 2014. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R- project.org/.

Rees, D. J., M. Dioli, and L. R. Kirkendall 2003. Molecules and morphology: evidence for cryptic hybridization in African Hyalomma (Acari : Ixodidae). Molecular Phylogenetics and Evolution 27:131–142.

Rhymer, J. M., and D. Simberloff 1996. Extinction by hybridization and introgression. Annual Review of Ecology, Evolution, and Systematics 27:83–109.

107

Richardson, L. R., and J. R. Gold 1995. Evolution of the Cyprinella lutrensis species- complex. III. Geographic variation in mitochondrial DNA of the red shiner (Cyprinella lutrensis) – influence of Pleistocene glaciation on population dispersal and divergence. Molecular Ecology 4:163-171.

Riggs, S. R. 1984. Paleoceanographic model of Neogene phosphorite deposition, US Atlantic continental margin. Science 223:123-131.

Roman, J. 2006. Diluting the founder effect: cryptic invasions expand a marine invader's range. Proceedings of the Royal Society B: Biological Sciences 273:2453-2459.

Roman, J., and J. A. Darling 2007. Paradox lost: genetic diversity and the success of aquatic invasions. Trends in Ecology & Evolution 22:454-464.

Rosenfield, J. A., S. Nolasco, S. Lindauer, C. Sandoval, and A. S. T. R. I. D. Kodric‐ Brown 2004. The role of hybrid vigor in the replacement of Pecos pupfish by its hybrids with sheepshead minnow. Conservation Biology 18:1589-1598.

Ryan, M. J., and A. Keddy-Hector 1992. Directional Patterns of Female Mate Choice and the Role of Sensory Biases. The American Naturalist 139:S4–S35.

Sakai, A. K., F. W. Allendorf, J. S. Holt, D. M. Lodge, J. Molofsky, K.A. With, S. Baughman, et al. 2001. The population biology of invasive species. Annual Review of Ecology and Systematics 32:305–332.

Sala, O. E., F. S. Chapin, J. J. Armesto, E. Berlow, J. Bloomfield, et al. 2000. Global biodiversity scenarios for the year 2100. Science 287:1770-1774.

Schliep, K. P. 2011. phangorn: phylogenetic analysis in R. Bioinformatics 27:592-593.

Schmidt, T. R., J. P. Bielawski, and J. R. Gold 1998. Morphological phylogenetics and evolution of the cytochrome b gene in the cyprinid genus Lythrurus (Actinopterygii: Cypriniformes). Copeia 14–22.

Schönhuth, S., and R. L. Mayden 2010. Phylogenetic relationships in the genus Cyprinella (Actinopterygii: Cyprinidae) based on mitochondrial and nuclear sequences. Molecular Phylogenetics and Evolution 55:77-98.

Seehausen, O. 2004. Hybridization and adaptive radiation. Trends in Ecology & Evolution 19:198-207.

Seehausen, O. 2006. Conservation: losing biodiversity by reverse speciation. Current Biology 16:R334-R337.

Seehausen, O., J. J. M. Van Alphen, and F. Witte 1997. Cichlid fish diversity threatened by eutrophication that curbs sexual selection. Science 277:1808–1811.

108

Seehausen, O., and J. J. M. Van Alphen 1998. The effect of male coloration on female mate choice in closely related Lake Victoria cichlids (Haplochromis nyererei complex). Behavioral Ecology and Sociobiology 42:1–8.

Seehausen, O., G. Takimoto, D. Roy, and J. Jokela 2008. Speciation reversal and biodiversity dynamics with hybridization in changing environments. Molecular Ecology17:30–44.

Simon, A., R. Britton, R. Gozlan, C. Van Oosterhout, F. A. Volckaert, and B. Hänfling 2011. Invasive cyprinid fish in Europe originate from the single introduction of an admixed source population followed by a complex pattern of spread. PloS one 6:e18560.

Soltis, D. E., A. B. Morris, J. S. McLachlan, P. S. Manos, and P. S. Soltis 2006. Comparative phylogeography of unglaciated eastern North America. Molecular ecology 15:4261-4293.

Strong, D. R., and D. R. Ayres 2013. Ecological and evolutionary misadventures of Spartina. Annual Review of Ecology, Evolution, and Systematics 44:389-410.

Swanbrow Becker, L. J., and C. R. Gabor 2012. Effects of turbidity and visual vs. chemical cues on anti-predator response in the endangered Fountain Darter (Etheostoma fonticola). Ethology 118:994–1000.

Tamura, K., G. Stecher, D. Peterson, A. Filipski, and S. Kumar 2013. MEGA6: Molecular Evolutionary Genetics Analysis Version 6.0. Molecular Biology and Evolution 30:2725-2729.

Taylor, E. B., J. W. Boughman, M. Groenenboom, M. Sniatynksi, D. Schluter, and J. Gow 2006. Speciation in reverse: morphological and genetic evidence of a collapse of a stickleback species pair (Gasterosteus). Molecular Ecology15:343– 355.

Tepolt, C. K., J. A. Darling, M. J. Bagley, J. B. Geller, M. J. Blum, and E. D. Grosholz 2009. European green crabs (Carcinus maenas) in the northeastern Pacific: genetic evidence for high population connectivity and current‐mediated expansion from a single introduced source population. Diversity and Distributions 15:997- 1009.

Townsend, C. R., and T. A. Crowl 1991. Fragmented population structure in a native New Zealand fish: an effect of introduced brown trout? Oikos 347-354.

U.S. Fish and Wildlife Service 1990. Spikedace recovery plan. U.S. Fish and Wildlife Service, Albuquerque, New Mexico.

109

U.S. Fish and Wildlife Service 1995. Greenback cutthroat trout recovery plan. U.S. Fish and Wildlife Service, Denver, Colorado.

Vallender, R., R. J. Robertson, V. L. Freisen, and I. J. Lovette 2007. Complex hybridization dynamics between golden-winged and bluewinged warblers (Vermivora chrysoptera and Vermivora pinus) revealed by AFLP, microsatellite, intron and mtDNA markers. Molecular Ecology16:2017–2029.

Vander Sluijs, I., S. M. Gray, M. C. P. Amorim, I. Barber, U. Candolin, A. P. Hendry, and B. B. M. Wong 2011. Communication in troubled waters: responses of fish communication systems to changing environments. Evolutionary Ecology 25:623–640.

Wallace, R. K., and J. S. Ramsey 1982. A new cyprinid hybrid, Notropis lutrensis and N. callitaenia, from the Apalachicola draingage in Alabama. Copeia 214-217.

Walters, D. M., M. J. Blum, B. Rashleigh, B. J. Freeman, B. A. Porter, and N. M. Burkhead 2008. Red shiner invasion and hybridization with blacktail shiner in the upper Coosa River, USA. Biological Invasions 10:1229-1242.

Ward, J. L., and D. A. McLennan 2009. Mate choice based on complex visual signals in the brook stickleback, Culaea inconstans. Behavioral Ecology 20:1323–1333.

Ward, J. L., and M. J. Blum 2012. Exposure to an environmental estrogen breaks down sexual isolation between native and invasive species. Evolutionary Applications 5:901–912.

Ward, J. L., M. J. Blum, D. M. Walters, B. A. Porter, N. M. Burkhead, and B. J. Freeman 2012. Discordant introgression in a rapidly expanding hybrid swarm. Evolutionary applications 5:380-392.

Warren, M. L., B. M. Burr, S. J. Walsh, H. L. Bart, R. C. Cashner, D. A. Etnier, W. C. Starnes, et al. 2000. Diversity, Distribution, and Conservation Status of the Native Freshwater Fishes of the Southern United States. Fisheries 25:7–31.

Wilson, E. O. 1992. The Diversity of Life. Belknap Press, Cambridge, Massachusetts.

Witte, F., T. Goldschmidt, J. Wanink, M. Van Oijen, K. Goudswaard, E. Witte-Maas, and N. Bouton 1992. The destruction of an endemic species flock: quantitative data on the decline of the haplochromine cichlids of Lake Victoria. Environmental biology of fishes 34:1-28.

Wolf, D. E., N. Takebayashi, and L. H. Rieseberg 2001. Predicting the risk of extinction through hybridization. Conservation Biology 15:1039–1053.

110

Wong, B. B. M., U. Candolin, and K. Lindström 2007. Environmental Deterioration Compromises Socially Enforced Signals of Male Quality in Three‐Spined Sticklebacks. The American Naturalist 170:184–189.

Yu, S. L., and E. J. Peters 2002. Diel and Seasonal Habitat Use by Red Shiner (Cyprinella lutrensis). Zoological Studies-Taipei 41:229-235.

Yuri, T., R. W. Jernigan, R. T. Brumfield, N. K. Bhagabati, and M. J. Braun 2009. The effect of marker choice on estimated levels of introgression across an avian (Pipridae: Manacus) hybrid zone. Molecular Ecology18:4888–4903.

Zink, R. M., and D. L. Dittmann 1993. Gene flow, refugia, and evolution of geographic variation in the song sparrow (Melospiza melodia). Evolution 47:717-729.

111

BIOGRAPHY

Gregory J. Glotzbecker was born in 1983 in Cincinnati, Ohio. His interest in the natural world began at a very young age when he told his mother that he wanted to be a zoologist at the age of six. His parents, Ron and Cheryl Glotzbecker both earned degrees in microbiology from Miami University, Ohio, and supported Greg in his early interests in science. Greg’s interest in fish began when his father setup a small freshwater aquarium in their home. From that point forward, he spent much of his free time raising and breeding tropical fish in his parents’ basement. Over the years his in fish continued to grow, and he majored in biological sciences at the University of Cincinnati. During his senior capstone project on invasive red shiner, Greg was introduced to Dr. Michael Blum, who was a post-doctoral fellow at the Cincinnati EPA. Greg graduated from the

University of Cincinnati in 2006 with a Bachelor of Science, and started working at the

EPA in Cincinnati, in the molecular ecology research branch. In the fall of 2007, Greg moved to New Orleans to continue his study of invasive red shiner, and began his doctoral studies at Tulane University under the direction of Dr. Michael Blum. In

December 2010, he was awarded a M.S. in ecology and evolutionary biology from

Tulane University.