Social Behavior and Kinship in the Four-Toed Salamander, Hemidactylium Scutatum

Home , Salamandrina, Streamside salamander

SOCIAL BEHAVIOR AND KINSHIP IN THE FOUR-TOED SALAMANDER, HEMIDACTYLIUM SCUTATUM

ABIGAIL JOY MALEY BERKEY

DISSERTATION

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Ecology, Evolution, and Conservation Biology in the Graduate College of the University of Illinois at Urbana-Champaign, 2015

Urbana, Illinois

Doctoral Committee:

Professor Andrew V. Suarez, Chair Associate Professor, Marlis R. Douglas, Co-Director of Research, University of Arkansas Associate Professor Christopher A. Phillips, Co-Director of Research Professor Jeffrey D. Brawn Associate Professor Robert Lee Schooley ABSTRACT

Amphibians are among the taxa experiencing the largest global decline in biodiversity.

Many salamanders, especially North American members of the family Plethodontidae, are at risk of local extirpation because they persist in small, isolated populations due to specialized habitat requirements and limited dispersal ability. To effectively conserve and manage such species, factors influencing population connectivity and dynamics, such as dispersal and behavior, must be understood across various spatial scales. The Four-Toed Salamander, Hemidactylium scutatum, a species of concern in eastern North America, is a small-bodied plethodontid with a biphasic life cycle. Despite its broad geographic range, its occurrence is patchy due to specific breeding habitat requirements. Little is known about how distance between these habitat patches affects gene flow and which physical and biotic landscape features may act as barriers to dispersal. The population structure, i.e. gene flow and genetic diversity, of other plethodontid species is characterized by high levels of genetic divergence over relatively short distances and often reflects an isolation by distance pattern. In this study, I examined dispersal and population structure in H. scutatum from a local to a regional scale, focusing on the interaction between relatedness and dispersal (kin discrimination) and the potential for phenotypic traits to be used as a mechanism for kin recognition. I combined empirical observations with lab experiments to (1) examine population structure in H. scutatum (Chapter 1), (2) determine if phenotypic similarity between individuals is associated with genotypic similarity (Chapter 2), and (3) examine if kin discrimination is exhibited and potentially triggered by scent (Chapter 3).

In Chapter 1, I used microsatellites to estimate genetic variation and gene flow within and among populations, calculate effective population size, and evaluate the possibility of population bottlenecks on both local and regional scales. The genetic data recorded a pattern of isolation-by-

ii distance, characteristic of plethodontid salamanders. However, H. scutatum showed low genetic divergence among neighboring sites (1,000-2,000m). Several explanation exist for the low genetic divergence among populations including greater gene flow in H. scutatum compared to other plethodontids, relatively recent range expansion following the last Pleistocene glaciation, and a reduction in allele loss as a result of reduced whole clutch mortality in an indirect developing species.

In Chapter 2 I examined the geographic and genotypic variation in the ventral spot pattern of H. scutatum. Specifically, if variation in color pattern is genetically controlled, color pattern may be used as a mechanism for kin recognition between conspecifics or be utilized in conservation decisions as an indicator of genetic diversity within a population. Additionally, I investigated the potential use of phenotypic traits as an indicator of stress or genetic diversity within populations. Fluctuating asymmetry, differences in the bilateral symmetry of a character due to disturbances in internal or external environments, influences color pattern in many in amphibian taxa and has the potential to be used as an indicator of at-risk populations under stress. Hemidactylium scutatum is characterized by a white ventral surface patterned with distinctive black spots. I used quantitative image analysis and microsatellite markers to investigate the potential influence of geographic variation on spot pattern and the potential relationship of phenotypic and genetic similarity. I also examined the possible influence of body condition, via fluctuating asymmetry, on spot pattern. While spot pattern exhibited significant variation on the regional scale, no relationship was found between spot pattern similarity and degree of kinship between individuals, suggesting that spot pattern is not used as a mechanism for kin recognition in H. scutatum. I also found no relationship between spot pattern symmetry and body condition. Combined, these findings suggest that, while spot pattern may not be useful

iii for assessing genetic variation or population stress, it may be useful for taxonomists and allow recent immigrants to populations to be identified.

In Chapter 3, I examined if isolated populations have increased degrees of kinship among individuals, potentially resulting in selection for kin recognition and kin discriminatory behaviors. Understanding the role of kinship in social behavior of at risk species such as H. scutatum may help in conservation and management efforts, especially those focusing on reintroduction and captive propagation programs. I investigated if there was an association between relatedness and aggressive behavior in H. scutatum by conducting behavioral trials in the lab and examined if related individuals may be spatially aggregated within a wild population.

No relationship was found between the straight line distance between individuals and their relatedness. I also did not find a relationship between relatedness and aggressive behavior in the lab. It is possible that kin discriminatory behaviors differ between demographic groups or kin discrimination does not occur under the conditions observed in this study. Finally, the indirect development of H. scutatum may impact juvenile dispersal, resulting in a decreased rate of encounters between kin and decreased selection for kin recognition compared to other, direct developing plethodontid species.

ACKNOWLEDGEMENTS

First and foremost, I would like to acknowledge my advisers, Drs. Christopher A. Phillips and Marlis R. Douglas. Their guidance and wisdom have been invaluable to my research. I would also like to my thank my other committee members, Drs. Andrew V. Suarez, Jeffrey D.

Brawn, and Robert Lee Schooley who contributed greatly to the development and refining of this project. Dr. Mark Davis, Dr. Paul Tinerella, Dr. Whitney Banning Anthonysamy, Dr. Carla

Cáceres, Sarah Baker-Wylie, Dan Wylie, Dr. Michael Douglas, Dr. Michael Dreslik, Ellen

Lawrence, John MacGregor, Susan King, Dr. Stephen Richter, Dr. Reid Harris, and Tom

Biebighauser have all lent their expertise and offered constructive feedback and ideas. Tim

Herman provided me with H. scutatum tissues invaluable information on sampling locations for this species in Kentucky. Various agencies have been involved with this research including the

Program in Ecology, Evolution, and Conservation Biology and the Department of Integrative

Biology at the University of Illinois in Champaign-Urbana, the Museum of Vertebrate Zoology at the University of California, the Illinois Natural History Survey, the Illinois Department of

Natural Resources, the USDA National Forest Service, and the Kentucky Fish and Wildlife

Service. Specimens were collected and handled under the Illinois Department of Natural

Resources Endangered and Threatened Species Permits #09-11S and #11-10S, the Kentucky Fish and Wildlife Service Scientific Collecting Permit #SC1211096, and the Institutional Animal

Care and Use Committee Protocol #12016. This project greatly benefited from the help of many field and lab assistants including Joe Frumkin, Ryan Jorgensen, Brittany Perrotta, Kevin Murray,

Alex Remigio, Courtney Romolt, David Bozman, Noah Horsley, Austin Meyers, Christopher

Maley, and Gracie Berkey. This project was generously funded by numerous sources, including the Illinois Wildlife Preservation Fund, the Endangered Species Protection Board, the Chicago

Herpetological Society, the University of Illinois, and Tom Beauvais. I would also like to acknowledge Dr. Carol A. Augspurger for her kind words and guidance through the stormy waters of graduate school. Finally I would like to thank my parents, Charles and Joyce Maley, my aunt Gail Steck, and my husband Daniel Berkey for their support and never failing confidence in me.

TABLE OF CONTENTS

CHAPTER 1. POPULATION GENETICS OF THE FOUR-TOED SALAMANDER, HEMIDACTYLIUM SCUTATUM, AT LOCAL AND REGIONAL SCALES …………..1

CHAPTER 2. SOURCES OF VARIATION IN THE VENTRAL SPOT PATTERN OF THE FOUR-TOED SALAMANDER ………………………………………………………...36

CHAPTER 3. ROLE OF KINSHIP AND INTRASPECIFIC AGGRESSION IN THE FOUR- TOED SALAMANDER………………………………………………………………....69

APPENDIX A.………………………………………………………………………………….103

APPENDIX B.………………………………………………………………………………….104

APPENDIX C.………………………………………………………………………………….105

APPENDIX D.………………………………………………………………………………….106

APPENDIX E.………………………………………………………………………………….107

APPENDIX F.………………………………………………………………………………….108

vii

CHAPTER 1. POPULATION GENETICS OF THE FOUR-TOED SALAMANDER, HEMIDACTYLIUM SCUTATUM, AT LOCAL AND REGIONAL SCALES

Introduction

Global declines in biodiversity place us amid a major extinction event in the geological record. Causes implicated in this decline include climate change (Bellard et al. 2012, Hannah et al. 2002, Thomas et al. 2004), habitat fragmentation and destruction (Fischer and Lindenmayer

2007, Krauss et al. 2010, Tilman et al. 2001), introduction of exotic species (Vellend et al. 2007,

Vilà et al. 2011), and environmental pollution (Hsu et al. 2006, Johnston and Roberts 2009,

Zvereva and Kozlov 2010). Amphibians are particularly vulnerable to these agents, as well as additional factors associated with global change, including increased UV radiation, over exploitation, and emerging diseases (Beebee and Griffiths 2005, Collins and Storfer 2003). As a result, amphibians are one of the vertebrate taxa most affected by this global loss in biodiversity

(Stuart et al. 2004, Wake and Vredenburg 2008).

Persistence of a species is influenced by short- and long-term processes and depends on its ability to cope with environmental change. Long-term, directional shifts in natural phenomena have historically occurred slowly, allowing mobile species to disperse to more suitable regions, such as refugia during periods of glaciation. Alternatively, a species may adapt in response to long-term pressures, but its adaptive capacity depends largely on inherent genetic diversity

(Barrett and Schluter 2007, Lande and Shannon 1996). Genetic diversity can be eroded through drift as populations become increasingly isolated and/or inbreeding as they decrease in size, potentially resulting in an intensified genetic load (Reed and Frankham 2003, Rowe and Beebee

2003). This process, which ultimately leads to reduced fitness and further decreased population size, is known as the small population paradigm (see Caughley 1994). Thus for declining species understanding the magnitude of population isolation is important for effective conservation and

1 management and this is best accomplished by assessing genetic structure within and among habitat patches.

Due to specialized habitat requirements and/or limited dispersal ability (Welsh and

Droege 2001) many salamander species, especially the small-bodied members of the family

Plethodontidae, persist in small, isolated populations characterized by high levels of genetic divergence. A review by Larson et al. (1984) across 21 species found a mean FST value of 0.53 among populations, suggesting that plethodontids typically form units not connected by gene flow. While studies on population structure generally have focused on large-scale differentiation

(>100km distance; Niemiller et al. 2008, Tilley and Mahoney 1996), genetic structuring has also been observed on local scales (<10km; Kozak et al. 2006) and across distances as small as 200m

(Cabe et al. 2007). Moreover, individual movements in plethodontids appear to be limited, even among nearby populations (Gillette 2003, Mathis 1991), resulting in isolation-by-distance patterns (Crespi et al. 2003, Jackman and Wake 1994, Martínez-Solano et al. 2007). Gene flow can be further reduced in these groups by both natural barriers, such as streams (Marsh et al.

2007), or man-made ones, such as roads (deMaynadier and Hunter 2000, Marsh et al. 2005,

Marsh et al. 2008). Additionally, genetic structure of many salamander species reflects post-

Pleistocene colonization, and is characterized by low diversity and consequently weak population divergence (Hewitt 2004). This pattern is consistent across many salamander families, including Plethodontidae (Highton and Webster 1976), Salamandridae (Kutcha and Tan

2005), and Ambystomatidae (Demastes et al. 2007, Phillips et al. 2000, Steele and Storfer 2006).

Life history characteristics may also modulate genetic structure in plethodontids. Most species exhibit direct development with few retaining the ancestral mode of an aquatic larval stage.

Direct development may increase the potential for whole clutch mortality that in turn could

2 reduce genetic variation within, but increase distinctness among populations through genetic drift

(Dubois 2004). In addition, low rates of gene flow in the direct-developing Plethodon cinereus

(Cabe et al. 2007) indicateindicate low dispersal rates. Combined effects of life history characteristics and scant dispersal could be a mechanism that led to the relatively high number of species in the Plethodontidae.

The Four-toed Salamander, Hemidactylium scutatum, is a small-bodied plethodontid with a biphasic life cycle and specific habitat requirements. The species has a broad geographic distribution that extends north to Nova Scotia, south to Florida, and as far west as Oklahoma and

Missouri (Petranka 1998), but with patchy occurrences in the southern and western areas.

Populations tend to centralize around patches of forested areas surrounding spring or seep-fed pools. Eggs are laid at the edge of ponds and larvae hatch at a late stage and wriggle into the pond where they metamorphose within 3-6 weeks. Among 36 states, districts, and territories H. scutatum is designated as either “imperiled” or “critically imperiled” in 9 of these regions and

“vulnerable” within an additional 13 (NatureServe 2011). Persistence of isolated populations makes H. scutatum an intriguing candidate for investigating how distance affects gene flow and what physical and biotic landscape features act as barriers to dispersal. Additionally, H. scutatum is an upland, terrestrial species with indirect development, a life history unusual among plethodontids and may exhibit unique patterns of population structure that require specific conservation strategies.

Due to the small-body size of plethodontid salamanders, genetic studies are best accomplished using non-lethal methods for tissue sampling. Microsatellite analysis is especially useful to assess gene flow because it requires only small amounts of tissue (i.e. tail clips), is relatively inexpensive, and yields generally highly polymorphic loci facilitating fine-scale

3 examination of genetic structure (Selkoe and Toonen 2006). Gene flow has been evaluated in a variety of salamander species, but few studies used microsatellite and most focused on the relatively large bodied ambystomatid salamanders (Giordano et al. 2007, Spear et al. 2005,

Zamudio and Wieczorek 2007), save a couple studies on the Eastern Red-backed Salamander, P. cinereus (Cabe et al. 2007, Noël et al. 2007), a smaller plethodontid species.

This study investigated the population structure of H. scutatum on both local and regional scales by evaluating several questions. (1) Does H. scutatum have patterns of population structure similar to those found in other plethodontid species? If H. scutatum’s dispersal capabilities are similar to those recorded for other plethodontid species, then comparable patterns of population structure, including high levels of population differentiation over relatively short distances, limited gene flow between populations, and an isolation-by-distance pattern indicated by a positive correlation between inter-population distance and genetic divergence (Wright 1943) should be observed. (2) Does H. scutatum have patterns of population structure that reflect recolonization from refugia following glaciation events? If the present patterns of genetic diversity in H. scutatum reflect the influence of glaciation events, then genetic differences should be low between sites located within previously glaciated regions as a result of re-colonization events. (3)

Has the unusual life history of H. scutatum resulted in patterns of population structure that differ from those observed in other plethodontid species? If indirect development in H. scutatum results in decreased whole clutch mortality and subsequent allele loss, then less genetic divergence should observed between sites than recorded for other plethodontid species. I used microsatellite markers to estimate genetic variation, gene flow, and genetic differences between populations, and to evaluate the possibility of previous population bottlenecks on both local and regional

4 scales. I further compared these results to those previously recorded in the literature for other plethodontid species.

Materials and Methods

Samples and Collection Sites

I selected study sites at two spatial scales: (1) the local scale consisted of two sites (I-AN and I-AS) in the Middle Fork State Fish and Wildlife Area in Vermilion County, IL; (2) the regional scale was represented by 11 sites in central Kentucky (K-A, K-B, K-C, K-D, K-E, K-F,

K-H, K-J, K-L, K-M, K-N) and one site in Jo Daviess County in northwestern Illinois (I-B)

(Figure 1.1). Sites I-AN and I-AS, separated by ~1,244 m and a stream, were analzyed separately in this analysis. Previous studies on H. scutatum documented relatively small maximum straight- line dispersal distances of only 201 m from a wetland source (Windmiller 2000) and streams appear to be a barrier to dispersal in a similar sized plethodontid, P. cinereus (Marsh et al. 2007).

I conducted visual encounter surveys for H. scutatum at local sites from March -

November for four years (2009-2012) and at regional sites in KY from 18-23 March 2011. For each capture, I recorded the GPS location in UTM coordinates using a Garmin GPS 12, body size (mm) as Snout-Vent-Length (SVL), Total-Length (TL) and mass (g), and photographed the underside of the individual to identify potential recaptures. For genetic analyses, I collected small tail clips (1-5 mm) from individuals weighing more than 0.18 g using sterilized clippers and tissues were stored in ethanol at -80 °C. Samples from site I-B, were collected by Tim

Herman in May 2004 using similar procedures. For the regional analyses, additional tissues from the local site I-AN collected by Tim Herman in April-May 2005 (I-AN05) were included.

Molecular Methods and Genetic Analyses

I isolated whole genomic DNA from tail clips using the Qiagen DNEasy Extraction kit following the manufacture’s protocol. For genetic analysis I screened 28 microsatellite loci developed for four salamander species, three of which were plethodontid and one an ambystomatid, for their potential usefulness in H. scutatum. Seven loci (HS3a, HS3b, HS5, HS7,

HS8, HS14, and HS15) were developed for H. scutatum by the Reid Harris lab at James Madison

University. The remaining loci were initially designed for other salamander species, including 11 for P. elongatus [PE0, PE1, PE3, PE4, PE5, PE7, PE8, PE9, PE10, PE11, and PE12 (Degross et al. 2004)], seven for P. cinereus [PC1, PC2, PC3, PC4, PC5, PC6, and PC7 (Connors & Cabe

2003)], and three for Dicamptodon tenebrosus [DT4, DT5, and DT8 (Curtis & Taylor 2001)]. I set up the final optimized PCR reactions in 10 µl volumes, each with 2 µl of the extracted DNA solution, 10 µM of each primer, 1.25 mM of each dNTP, 25 mM MgCl2, 2 µl 1X reaction buffer, and 1.5-2.0 units Taq and conducted amplifications on either an Applied Biosystems 2720

Thermocycler or an Applied Biosystems Veriti Thermocycler. With the exception of loci DT4 and HS7, amplifications consisted of a 3 minute denaturation step at 95 °C, 15 cycles of 45 s at

95 °C, 60 s at the annealing temperature, 45 s at 72 °C, followed by 25 cycles of 30 s at 95 °C,

45 s at the annealing temperature, 30 s at 72 °C. The amplification for DT4 consisted of a 3 minute denaturation step at 95 °C, 15 cycles of 45 s at 95 °C, 60 s at the annealing temperature,

60 s at 72 °C, followed by 25 cycles of 30 s at 95 °C, 45s at the annealing temperature, 45 s at

72 °C and the amplification for HS7 consisted of a 3 minute denaturation step at 95 °C, 15 cycles of 45 s at 95 °C, 60 s at the annealing temperature, 90 s at 72 °C, followed by 25 cycles of 30 s at

95 °C, 45 s at the annealing temperature, 30 s at 72 °C.

I evaluated the amplification success of loci on agarose gels stained with GelGreen

(Biotium, Inc., Hayward CA) and visualized fragments using a DarkReader Transluminator DR-

88M (Clare Chemical Research, Inc., Dolores CO). For efficient screening of genetic diversity, I grouped selected loci into pool-plex panels where the amplification products of multiple loci for the same individual are combined for data generation. I labeled forward primers with a fluorescent dye and further evaluated each locus in terms of polymorphism, amplification specificity, and signal strength. Genotyping was conducted on an Applied Biosystems 3730xl

Analyzer at the W. M. Keck Center for Comparative and Functional Genomics, University of

Illinois. An internal size standard (LIZ 500) was included with each sample to determine fragment length. I scored alleles using GENEMAPPER v5.0 and assessed genotype scoring errors, the presence of null alleles, and the occurrence of large allele dropout via MICROCHECKER v2.2.3

(Oosterhout et al. 2004). I also employed POWSIM v4.1 to determine the statistical power of using this set of microsatellites for analyses (Ryman and Palm 2006).

Local Analysis

For evaluation of genetic structure at the local scale, I analyzed 157 samples collected between 2009-2012 from Vermilion County, Illinois (sites I-AN and I-AS). I used GENEPOP v3.4

(Raymond and Rousset 1995) to test each locus for Hardy-Weinberg equilibrium and linkage disequilibrium and to assess the overall allelic richness and heterozygosity (genetic diversity) using Markov-chain parameters with 10000 dememorization steps, 999 batches, and 9999 iterations per batch. GENEPOP was also used to assess the number of migrants (Nm) per generation between populations following the private allele method outlined by Barton and

Slatkin (1986) and correcting for sample size. Population structure was evaluated by determining

7 numbers of distinct gene pools employing a Bayesian assignment test as implemented in

STRUCTURE v2.3.4 (Hubisz et al. 2009) using uncorrelated allele frequencies, the LOCPRIOR model, a 10000 burn in period with 10000 reps, and testing K values from 1 to 6 with 10 runs at each K. Preliminary runs with STRUCTURE suggested that λ= 0.5 would be appropriate for this system and STRUCTURE HARVESTER v0.6.94 (Earl and vonHoldt 2012) was used toassess the appropriate number of clusters following the Evanno method (Evanno et al. 2005). Under K=2,

STRUCTURE identified most individuals as belonging solely to one main gene pool across the I-

AN and I-AS sites with a second group of individuals with whole or partial membership in a second distinct gene pool. To determine whether the individuals with membership in the second gene pool represented recent migrants, I conducted a two-tailed t-test with unequal variances comparing the mean relatedness value of individuals, calculated by following the estimate of relatedness (r) defined by Queller and Goodnight (1989), in the main gene pool, to the mean relatedness value of these potential immigrants. I used LDNE v1.31 (Waples and Do 2008) to estimate the effective population size (Ne) for gene pools identified by these analyses using a minimum allele frequency 0.01 and to construct confidence intervals for each Ne estimate using parametric methods. I performed an Analysis of Molecular Variance (AMOVA) using ARLEQUIN v3.5.1.2 (Excoffier and Lischer 2010) to examine population structure and to investigate the proportion of genetic variation contributed by variation between the two sampling sites (among populations), within each sampling site (within populations), and among individuals and to calculate F-statistics.

Regional Analysis

For evaluation of genetic structure at the regional scale, I analyzed the entire set of samples, including all individuals from the local analysis, as well as tissues collected from

Vermillion County in 2005, from Jo Daviess County in 2004, and Kentucky in 2011. I used

GENEPOP v3.4 to check for deviations from Hardy-Weinberg expectations and linkage disequilibrium using Markov-chain parameters with 10,000 dememorization steps, 999 batches, and 9,999 iterations per batch. I used STRUCTURE v2.3.4 to examine overall population structure using uncorrelated allele frequencies, the LOCPRIOR model, a 10,000 burn in period with 10,000 reps, testing K values from 1 to 13 with 10 runs at each K, and using λ= 0.5. STRUCTURE

HARVESTER v0.6.94 was again used to determine the appropriate number of population clusters.

In order to evaluate population designations, I used RXC (www.marksgeneticsoftware.net) to conduct multilocus contingency tests of allele frequency heterogeneity following Waples and

Gaggioti (2006) with 99 batches, 9,999 demorization runs, and 9,999 replicates for each randomization test. I used ARLEQUIN v3.5.1.2 to perform an AMOVA examining relative genetic variation regionally (among populations), within populations, and within individuals and to calculate F-statistics. To test for an isolation-by-distance pattern of genetic differences among sampling sites, I conducted a Mantel test examining association between linear distance and FST value for each pairwise comparison. I used BOTTLENECK v1.2.02 (Cornuet and Luikart 1997) to test for recent population bottlenecks at sites with more than 10 individuals (K-A, K-D, K-L, and

I-A) using 10,000 repetitions to conduct Wilcoxon rank sign tests under the infinite alleles model.

Results

Samples and Collection Sites

To assess genetic structure at the local level, 157 salamanders were genotyped, with 111 from site I-AN and 46 from I-AS (Table 1.1). Most sampled individuals were adults; only 13 were juveniles. Sample size for each site is listed by year in Appendix A. For the regional analysis 279 unique samples were used, including the 157 individuals from the local analysis, seven from I-B, eight from I-A collected in 2005, and 107 samples collected from central

Kentucky.

Microsatellite Loci

Of the 28 microsatellite loci screened (Table 1.2), 20 initially amplified fragments in H. scutatum and were tested in pool-plex panels. When screened using dye-labeled primers, nine loci (Table 1.2), showed sufficient variability to assess genetic diversity in H. scutatum. The remaining 13 loci were excluded from analyses due to scoring problems, inconsistent or weak amplification or artifacts. The size range, final pool-plex panels, dye assignments, and number of alleles for each of the nine loci are listed in Table 1.2. Additional optimization could potentially render the excluded loci useful in future studies of H. scutatum. Linkage disequilibrium was found for the locus pair HS7 and HS8 in both I-AN and I-AS populations. While no large allele dropout was detected for any of the nine loci, homozygote excess, often suggestive of null alleles, was detected for DT8, HS3a, HS3b, HS7, HS8, and PC1. This homozygote excess is likely due to the Wahlund effect, wherein the admixture of two or more populations with differing allele frequencies results in heterozygote deficiency if analyzed as a combined

10 population (Hartl 2000). Because such a high proportion of the loci were out of Hardy-Weinberg

Equilibrium, it is likely that sampling issues, rather than scoring errors and null alleles, are responsible. Because many plethodontid salamanders utilize underground retreats [i.e. P. cinereus (Jaeger 1980)], it is likely that the individual H. scutatum sampled at a given time represent only a subset of the proportion of the population active and above ground during the sampling period. In addition, it is possible that a large proportion of samples represent closely related individuals, also affecting Hardy-Weinberg expectations and linkage disequilibrium.

However, locus HS7 was excluded from further analyses because it was found to be in linkage disequilibrium with HS8. Tests for the statistical power of these eight loci using POWSIM gave an expected Type I error rate of α=0.048.

Local Analysis

Hardy-Weinberg expectations for allele frequencies were violated, with a Bonferroni corrected p-value needed for significance of p<0.003, for all loci but HS3b in I-AN, and for four

(out of eight) loci in I-AS (i.e., DT4, HS3b, HS8, and HS5). Substantial amounts of immigration and emigration could explain these violations of Hardy-Weinberg equilibrium. It is more likely, however, that these deviations from HWE are a result of the Wahlund effect, wherein the admixture of two or more populations with differing allele frequencies results in heterozygote deficiency if analyzed as a combined population (Hartl 2000). Indeed, the subsequent Bayesian assignment analysis suggests that some population admixture has occurred from recent immigrants into the population.

Genetic Diversity. —Expected heterozygosity (HE) was high for both I-AN and I-AS, with 0.64 ± 0.18 and 0.65 ± 0.17, respectively. Loci originally developed for H. scutatum

11 showed the greatest genetic diversity with per loci heterozygosity ranging from 0.82 to 0.87 compared to a maximum heterozygosity among the cross-amplified loci of 0.54 (Table 1.2,

Appendix B). The overall FIS value was 0.137 ± 0.047, suggesting that inbreeding occurs infrequently. Expected heterozygosity for I-AN and I-AS varied very little between years

(Appendix C), with a range of 0.61 to 0.66 and 0.61 to 0.71 respectively.

Genetic Structure. —The results of the Bayesian Assignment test conducted in

STRUCTURE indicated that the best overall value for number of gene pools was K=1, with a Ln likelihood = -3571.1. This suggests that individuals from the two sites, I-AN and I-AS, are actually part of one larger population without subdivision. STRUCTURE runs conducted with K=2 indicated that most individuals belong solely to one large gene pool (cluster 1) with a handful of samples either partially or fully being assigned to another gene pool (cluster 2) (Figure 1.2). The t-test comparison of the mean relatedness (Queller and Goodnight 1989) of individuals belonging to these two gene pools indicated a significantly greater relatedness of individuals assigning to cluster 1 than those assigning to cluster 2 with a p-value of 0.027, indicating individuals in cluster 2 represent recent immigrants (or their offspring) from various source populations. The results of the AMOVA indicate that most of the observed variance (86%) is the result of within individual variation with the remaining variance (14%) attributed to among individual variation

(Table 1.3). The calculated pairwise FST value between the I-AN and I-AS sites was very low and indicated no divergence (FST=0.004, p-value=0.11). The mean number of migrants per generation (Nm) between these two localities was 7.78. The effective population size, NE, for the combined I-AN and I-AS sites (i.e. site I-A) was 593 with a 95% confidence interval of 263 to

5151.

Regional Analysis

Genetic Diversity.—The I-AN and I-AS sites exhibit low genetic divergence, therefore I lumped them together (I-A) for the regional analyses focusing on within-population genetic diversity. However, I continued to analyze them separately for analyses focusing on population differences and genetic structure. This approach allowed the local I-AN and I-AS data to be included in analyses focusing on the role of distance in population structure in H. scutatum.

Expectations for Hardy-Weinberg genotype frequencies were violated in all populations except

K-F and K-H (Bonferroni corrected p-value <0.003). Expected heterozygosity (HE) was high for populations in the regional analysis and ranged from 0.52 ± 0.04 to 0.74 ± 0.08.

Genetic Structure.—The STRUCTURE assignment test indicated that four gene pools (K=4) best reflect the genetic structure in the regional analysis (ln likelihood = -7333.7, Figure 1.3).

Under this scenario, I-AN and I-AS are grouped together, with K-D, K-F, and K-N forming a second, and K-H, K-L, and K-M comprising a third group. The remainder of the Kentucky sites

(K-A, K-B, K-C, K-E, K-J) and I-B showed a high proportion of admixed gene pools. Unlike

AMOVA results for the local analysis, the regional AMOVA analysis indicated that 9% of the overall genetic diversity is due to among population variation, 18% due to among individual variation, and 73% due to within individual variation (Table 1.3). The overall FIS value for the regional analysis was 0.202 ± 0.044. Significant genetic differences were detected between all three regions (I-A, I-B, and KY) (Table 1.4). Within the KY sites, the only significant differences were found between site K-D and sites K-J, K-L, and K-M (Appendix D). The results of the RxC analysis found four major gene pools: cluster 1 contains individuals from I-AN and I-

AS, cluster 2 from I-B, cluster 3 from K-D, and cluster 4 from the remaining Kentucky sampling sites (K-A, K-B, K-C, K-E, K-F, K-H, K-J, K-L, K-M, and K-N) (Figure 1.4 and Appendix E).

The Mantel test found a significant positive relationship between geographic distance and FST (p- value = 0.01, Figure 1.5). The results of the Wilcoxon tests for population bottlenecks suggest a recent population bottleneck for population K-L (p-value = 0.004), but not for populations K-A,

K-D, or I-A (Table 1.5).

Discussion

The global decline in amphibian biodiversity has created a need for a better understanding of how short- and long-term processes impact the population genetics of amphibian species at different spatial scales. I examined the genetic structure of 13 populations of H. scutatum to test alternative hypotheses for local and regional scale patterns. Specifically, I investigated whether H. scutatum exhibits limited gene flow and isolation-by-distance patterns that characterize most other plethodontid species. I also examined the potential impact of post- glacial range expansion on genetic structure in H. sctuatum. Finally, I evaluated the hypothesis that life history characteristics in H. scutatum, especially developmental mode, may be influencing the genetic structure of this species. I found evidence for high levels of gene flow between neighboring populations (1000-2000 m), but an isolation-by-distance pattern on a

larger, regional scale. Genetic differences between populations, i.e. FST values, were low relative to those observed in other plethodontid species.

Population structure and dispersal at the local scale

On a local scale, dispersal appears to be much greater in H. scutatum than in other plethodontids. Neither FST values or the RxC analysis indicated significant genetic differences between I-AN and I-AS. The mean number of migrants between these two areas is relatively

14 high suggesting extensive gene flow - a surprising result considering the relatively large distance

(1,224 m) that separates these two sites. Interestingly, a similar pattern was observed among northern populations in Kentucky; while some of these populations are separated by short distances (e.g., K-A and K-H), and therefore would be expected to show little divergence, most are distant from the others by several kilometers (Appendix F). Small sample sizes could influence these results, but stochastic sampling issues would be expected to result in exaggerated, rather than lower genetic divergence. In addition, this pattern is repeated among sites with large sample sizes (i.e. I-AN and I-AS). Phylogeographic work on H. scutatum has suggested that the patterns of genetic structure observed in this species reflect a post-glacial expansion from refugia in the Appalachian Mountains (Herman 2009). A post-glacial recolonization would account for the lack of genetic differences observed between I-AN and I-

AS, but does not explain the similar pattern observed among the northern populations in

Kentucky, an area that remained unglaciated during the Pleistocene.

Potential for high dispersal is further supported by other signals in my data. Bayesian

Assignment tests conducted in STRUCTURE (Figure 1.2) identified two distinct gene pools.

However, they were not associated with the two sampling sites. Instead, the population appears to be quite homogenous across the entire area, with the exception of some individuals that have very distinct genotypes. These individuals are likely recent immigrants or the offspring of recent immigrants and thus provide evidence that migration to these sites from an outside source is occurring. Hence, while I find support for the post-glacial hypothesis, I cannot exclude the possibility that H. scutatum possesses greater dispersal abilities than expected.

Population structure and dispersal at the regional scale

Results of my regional analyses are more congruent with observations in other plethodontid species and best fit an isolation-by-distance model, a common pattern among members of Plethodontidae [Batrachoseps attenuatus (Martínez-Solano et al. 2007),

Desmognathus ochrophaeus (Tilley and Mahoney 1996), D. wrighti (Crespi et al. 2003),

Ensantina sp. (Jackmann and Wake 1994), and P. cinereus (Cabe et al. 2007)]. Thus it is not surprising that such a pattern was observed in this study and this supports the hypothesis that H. scutatcum follows a pattern of genetic structure similar to that of other plethodontids on a regional scale (Figure 1.5). In general, salamanders from sites in Kentucky were not genetically divergent from each other, with the exception of site K-D that was genetically distinct from the remaining populations in Kentucky. Divergence of the K-D population likely represents a deep biogeographic signal and can be explained by vicariance due to topography and underlying geology of Kentucky; the site is located on the Mississippian Plateau, whereas the remainder of the sampling sites is located on the Cumberland Plateau. These two plateaus are separated by the

Pottsville escarpment, a series of sandstone cliffs and steep-sided valleys (Newell 1986).

While the Bayesian assignment and RxC analyses are largely in agreement, some differences do exist in the population clusters identified by each method. Specifically, the RxC analysis indicates K-D as distinct from all other populations, whereas the Bayesian assignment analysis clusters K-D with K-C, K-F, and K-N. These seemingly contradictory results may be due to the isolation-by-distance pattern. Bayesian assignment analysis is known to overestimate population structure when there is an isolation-by-distance signal in the data (Frantz et al. 2009) and is sensitive to small sample sizes (Waples and Gaggioti 2006), such as those in populations

K-C, K-F, and K-N (Table 1.1).

Population Size and Recent Bottlenecks

The estimate of effective population size of 593, for site I-A is low relative to that recorded for other plethodontid species, but the upper limit of the 95% confidence interval for this estimate (5151) places Ne estimates for H. scutatum within range of other plethodontids (i.e.

Ne ranging from 4000 to 64000, see review by Larson et al. 1984). Among the four sites with sufficient sample size for testing (K-A, K-D, K-L, and I-A), only site K-L was identified as having recently undergone a population bottleneck. It is unlikely that this is due to population size reduction during the last Pleistocene glaciation that ended about 10000 years ago (see Black

1974) as site only I-A was located within the limits of the last glacial maximum (Figure 1.1,

Inset B). It is possible that some site specific, local variable is responsible for this pattern, as any factor at a regional scale, such as post-Pleistocene climatic fluctuations, would have also impacted the nearby (~9 km) site K-A.

Statistical validity of data

Shallow divergence could potentially be an artifact due to insufficient power of the markers used to detect genetic structure. However, results based on the eight loci should be relatively robust as suggested by the POWSIM expected Type I error rate of α=0.048 and should be sufficient to assess population structure. While additional loci would likely be informative, microsatellite cross-amplification is notoriously difficult in amphibians as a result of their large genome size (Garner 2002). It is not surprising that so many of the loci from species other than

H. scutatum failed to amplify or could not be reliably scored. Additionally, a number of other studies have successfully used similar numbers of microsatellites to assess salamander population structure. Cabe et al. (2007) and Noël et al. (2007) investigated genetic diversity in P.

17 cinereus populations and found observed levels of heterozygosity ranging from 0.32 to 0.85

(based on 6 loci) and from 0.38 to 0.69 (based on 7 loci), respectively. Among members of the genus Ambystoma, Giordano et al. (2007) and Spear et al. (2005) found a negative relationship between elevation and genetic differences with observed heterozygosity ranging from 0.03 to

0.81 for A. tigrinum (derived from 7 loci) and of 0.32 for A. macrodactylum (from 8 loci), respectively. In contrast, Purrenhage et al. (2009) found no isolation-by-distance patterns and strong overall connectivity among populations of A. maculatum using 8 loci with observed heterzygosity ranging from 0.49 to 0.70.

Conclusions

Small isolated populations may be a common occurrence in plethodontid species and may have contributed to the long-term evolution of this group. My study found that H. scutatum follows a similar pattern of genetic structuring as other plethodontid species on a regional scale.

On a local scale, however, very little genetic divergence was observed between sites. This may be a result of high levels of gene flow and dispersal, reduced genetic variation between populations due to post-glacial recolonization from a common source, and/or a reduction in allele loss as a result of reduced whole clutch mortality in an indirect developing species. While genetic patterns of post-glacial recolonization have been found in other studies on H. sctutatum

(Herman 2009), increased dispersal cannot be ruled out because reduced genetic divergence was similarly found in populations from unglaciated areas and Bayesian assignment analyses suggested the presence of recent immigrants in sites I-AN and I-AS. Moreover, it is possible that indirect development, the ancestral condition in plethodontids, persists in H. scutatum because it prevents the loss of genetic variation and counters any loss of gene flow this species experiences

18 due to its need for specialized breeding habitat. Whether as a result of increased dispersal or developmental mode, reduced loss in genetic variation could translate into a reduced susceptibility in H. scutatum to the small population paradigm. Future conservation and management decisions for this species should consider the implications of life history traits such as developmental mode on the short-term survival of populations and the long-term survival of species.

Literature Cited

Barrett, R., and D. Schluter. 2007. Adaptation from standing genetic variation. Trends in Ecology & Evolution 23:38-44.

Barton, N. H. and M. Slatkin. 1986. A quasi-equilibrium theory of the distribution of rare alleles in a subdivided population. Journal of Heredity 56:409-415.

Beebee, T. J. and R. A. Griffiths. 2005. The amphibian decline crisis: a watershed for conservation biology? Biological Conservation 125:271-285.

Bellard, C., C. Bertelsmeier, P. Leadley, W. Thuiller, and F. Courchamp. 2012. Impacts of climate change on the future of biodiversity: Biodiversity and climate change. Ecology Letters 15:365-377.

Black, R. F. 1974. Geology of the Ice Age National Scientific Reserve of Wisconsin. National Park Service Monograph Series 2:1-234.

Cabe, P. R., R. B. Page, T. J. Hanlon, M. E. Aldrich, L. Connors, and D. M. Marsh. 2007. Fine- scale population differentiation and gene flow in a terrestrial salamander (Plethodon cinereus) living in continuous habitat. Heredity 98:53-60.

Caughley, G. 1994. Directions in conservation biology. Journal of Animal Ecology 63:215-244.

Collins, J. P. and A. Storfer. 2003. Global amphibian declines: sorting the hypotheses. Diversity and Distributions 9:89-98.

Connors, L. M. and P. R. Cabe. 2003. Isolation of dinucleotide microsatellite loci from red- backed salamander (Plethodon cinereus). Molecular Ecology Notes 3:131-133.

Cornuet, J. M. and G. Luikart. 1997. Description and power analysis of two tests for detecting recent population bottlenecks from allele frequency data. Genetics 144:2001-2014.

Crespi, E. J., L. J. Rissler, and R. A. Browne. 2003. Testing Pleistocene refugia theory: phylogeographical analysis of Desmognathus wrighti, a high-elevation salamander in the southern Appalachians. Molecular Ecology 12:969–984.

Curtis, J. M. R. and E. B. Taylor. 2001. Isolation and characterization of microsatellite loci in the Pacific giant salamander Dicamptodon tenebrosus. Molecular Ecology Notes 9:116-118.

Degross, D. J., L. S. Mead, and S. J. Arnold. 2004. Novel tetranucleotide microsatellite markers from the Del Norte salamander (Plethodon elongates) with application to its sister species the Siskiyou Mountain salamander (P. stormi). Molecular Ecology Notes 4:352-354.

Demastes, J. W., J. M. Eastman, J. S. East, and C. Spolsky. 2007. Phylogeography of the blue- spotted salamander, Ambystoma laterale (Caudata: Ambystomatidae). The American Midland Naturalist 157:149–161. deMaynadier, P. G. and M. L. Hunter, Jr. 2000. Road effects on amphibian movements in a forested landscape. Natural Areas Journal 20:56-65.

Dubois, A. 2004. Developmental pathway, speciation, and supraspecific taxonomy in amphibians. 1. Why are there so many frog species in Sri Lanka? Alytes 22: 19-37

Earl, D. A. and B. M. vonHoldt. 2012. STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genetics Resources 4:359-361.

Ehlers, J., P. L. Gibbard, and P. D. Hughes. (eds.) 2011. Quaternary glaciations - extent and chronology: a closer look. Developments in Quaternary Science, Vol. 15. Elsevier, Amsterdam.

Evanno, G., S. Regnaut, and J. Goudet. 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology 14:2611-2620.

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.

Fischer, J. and D. B. Lindenmayer. 2007. Landscape modification and habitat fragmentation: a synthesis. Global Ecology and Biogeography 16:265-280.

Frantz, A. C., S. Cellina, A. Krier, L. Schley, and T. Burke. 2009. Using spatial Bayesian methods to determine the genetic structure of a continuously distributed population: clusters or isolation by distance? Journal of Applied Ecology 46:493-505.

Garner, T. W. J. 2002. Genome size and microsatellites: the effect of nuclear size on amplification potential. Genome 45:212-215.

Gillette, J. R. 2003. Population ecology, social behavior, and intersexual differences in a natural population of red-backed salamanders: a long-term field study. PhD Dissertation: University of Louisiana.

Giordano, A. R., B. J. Ridenhour, and A. Storfer. 2007. The influence of altitude and topography on genetic structure in the long-toed salamander (Ambystoma macrodactylum). Molecular Ecology 16:1625-1637.

Hannah, L., G. F. Midgley, T. Lovejoy, W. J. Bond, M. Bush, J. C. Lovett, D. Scott, and F. I. Woodward. 2002. Conservation of biodiversity in a changing climate. Conservation Biology 16:264-268.

Hartl, D. L. 2000. A Primer of Population Genetics. 3rd Edition. Sinauer Associates, Inc., Sunderland, M.A., USA.

Herman, T. A. 2009. Range-wide phylogeography of the four-toed salamander (Hemidactylium scutatum): out of Appalachia and into the glacial aftermath. Master’s Dissertation: Bowling Green State University.

Hewitt, G. M. 2004. Genetic consequences of climatic oscillations in the Quaternary. Philosophical Transactions of the Royal Society B: Biological Sciences, 359:183–195.

Highton, R. and T. P. Webster, T. P. 1976. Geographic protein variation and divergence in populations of the salamander Plethodon cinereus. Evolution 30:33.

Hubisz, M. J., D. Falush, M. Stephens, and J. K. Pritchard. 2009. Inferring weak population structure with the assistance of sample group information. Molecular Ecology Resources 9:1322-1332.

Hsu, M. J., K. Selvaraj, and G. Agoramoorthy. 2006. Taiwan’s industrial heavy metal pollution threatens terrestrial biota. Environmental Pollution. 143:327-334.

Jackman, T. R. and D. B. Wake. 1994. Evolutionary and historical analysis of protein variation in the blotched forms of salamanders of the Ensatina complex (Amphibia: Plethodontidae). Evolution 48:876.

Jaeger, R. G. 1980. Microhabitat of a terrestrial forest salamander. Copeia 1980:265-268.

Johnston, E. L. and D. A. Roberts. 2009. Contaminants reduce the richness and evenness of marine communities: A review and meta-analysis. Environmental Pollution 157:1745- 1752.

Kozak, K. H., R. A. Blaine, and A. Larson. 2005. Gene lineages and eastern North American palaeodrainage basins: phylogeography and speciation in salamanders of the Eurycea bislineata species complex: drainage history and salamander phylogeography. Molecular Ecology 15:191–207.

Krauss, J., R. Bommarco, M. Guardiola, R. K. Heikkinen, A. Helm, M. Kuussaari, R. Lindborg, E. Öckinger, M. Pärtel, J. Pino, J. Pöyry, K. M. Raatikainen, A. Sang, C. Stefanescu, T. Teder, M. Zobel, and I. Steffan-Dewenter. 2010. Habitat fragmentation causes immediate and time-delayed biodiversity loss at different trophic levels: Immediate and time- delayed biodiversity loss. Ecology Letters 13:597-605.

Kuchta, S. R. and A. M. Tan. 2004. Isolation by distance and post-glacial range expansion in the rough-skinned newt, Taricha granulosa: phylogeography of the rough-skinned newt. Molecular Ecology 14:225–244.

Lande, R. and S. Shannon. 1996. The role of genetic variation in adaptation and population persistence in a changing environment. Evolution 50:434.

Larson, A., D. B. Wake, and K. P. Yanev. 1984. Measuring gene flow among populations having high levels of genetic fragmentation. Genetics 106:293-308.

Marsh, D. M., G. S. Milam, N. P. Gorham, and N. G. Beckman. 2005. Forest roads as partial barriers to terrestrial salamander movement. Conservation Biology 19:2004-2008.

Marsh, D. M., R. B. Page, T. J. Hanlon, H. Bareke, R. Corritone, N. Jetter, N. G. Beckman, K. Gardner, D. E. Seifert, and P. R. Cabe. 2007. Ecological and genetic evidence that low- order streams inhibit dispersal by Red-backed Salamanders (Plethodon cinereus). Canadian Journal of Zoology 85:319-327.

Marsh, D. M., R. B. Page, T. J. Hanlon, R. Corritone, E. C. Little, D. E. Seifret, and P. R. Cabe. 2008. Effects of roads on patterns of genetic differentiation in red-backed salamanders, Plethodon cinereus. Conservation Genetics 9:603-613.

Martínez-Solano, I., E. L. Jockusch, and D. B. Wake. 2007. Extreme population subdivision throughout a continuous range: phylogeography of Batrachoseps attenuatus (Caudata: Plethodontidae) in western North America. Molecular Ecology 16:4335-4355.

Mathis, A. 1991. Large male advantage for access to females: evidence of male-male competition and female discrimination in a territorial salamander. Behavioral Ecology and Sociobiology 29:133–138.

McGrath, K. M. 1996. Analysis of trinucleotide microsatellite DNA in the genome of the four- toed salamander, Hemidactylium scutatum. Undergraduate Thesis: James Madison University.

NatureServe. 2011. NatureServe Explorer: An online encyclopedia of life [web application]. Version 7.1. NatureServe, Arlington, Virginia. Available http://www.natureserve.org/explorer. (Accesssed: January 2, 2012).

Newell, W. L. 1986. Physiography. In R. C. McDowell (Ed.), The geology of Kentucky. United States Government Printing Office, Washington, D.C. USA.

Niemiller, M. L., B. M. Fitzpatrick, and B.T. Miller. 2008. Recent divergence with gene flow in Tennessee cave salamander (Plethodontidae: Gyrinophilus) inferred from gene genealogies. Molecular Ecology 17:2258-2275.

Noël, S. M. Ouellet, P. Galois, and F. J. Lapointe. 2007. Impact of urban fragmentation on the genetic structure of the eastern red-backed salamander. Conservation Genetics 8:599-606.

Oosterhout, C. V., W. F. Hutchinson, D. P. M. Wills, and P. Shipley. 2004. Micro-checker: software for identifying and correcting genotyping errors in microsatellite data. Molecular Ecology Notes 4:535-538.

Petranka, J. W. 1998. Salamanders of the United States and Canada. Smithsonian Institution Press, Washington, D.C., USA

Phillips, C. A., G. Suau, and A. R. Templeton. 2000. Effects of Holocene climate fluctuation on mitochondrial DNA variation in the ringed salamander, Ambystoma annulatum. Copeia, 2000:542-545.

Purrenhage, J. L., P. H. Niewiarowski, and F. B. G. Moore. 2009. Population structure of spotted salamanders (Ambystoma maculatum) in a fragmented landscape. Molecular Ecology 18: 235-247.

Queller, D. C. and K. F. Goodnight. 1989. Estimating relatedness using genetic markers. Evolution 43:258-275.

Raymond, M. and F. Rousset. 1995. GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. Journal of Heredity 86:248-249.

Reed, D. H. and R. Frankham. 2003. Correlation between fitness and genetic diversity. Conservation Biology 17:230-237.

Reid, L. A. 1994. Characterization of a microsatellite DNA locus in the four-toed salamander, Hemidactylium scutatum. Undergraduate Thesis: James Madison University.

Rowe, G. and T. J. Beebee. 2003. Population on the verge of a mutational meltdown? Fitness costs of genetic load for an amphibian in the wild. Evolution 57:177-181.

Ryman, N. and S. Palm. 2006. POWSIM: a computer program for assessing statistical power when testing for genetic differentiation. Molecular Ecology Notes. 6:600-602.

Schrecengost, A. C. 1998. The isolation and characterization of genetic markers for use in a kinship study of the four-toed salamander. Undergraduate Thesis: James Madison University.

Selkoe, K. A. and R. J. Toonen. 2006. Microsatellites for ecologists: a practical guide to using and evaluating microsatellite markers. Ecology Letters 9:615-629.

Spear, S. F., C. R. Peterson, M. D. Matocq, and A. Storfer. 2005. Landscape genetics of the blotched tiger salamander (Ambysoma tigrinum melanostictum). Molecular Ecology 14:2553-2564.

Steele, C. A. and A. Storfer. 2006. Coalescent-based hypothesis testing supports multiple Pleistocene refugia in the Pacific Northwest for the Pacific giant salamander (Dicamptodon tenebrosus): testing Pleistocene refugia hypotheses. Molecular Ecology 15:2477–2487.

Stuart, S. N., J. S. Chanson, N. A. Cox, B. E. Young, A. S. L. Rodrigues, D. L. Fischman, and R. W. Waller. 2004. Status and trends of amphibian declines and extinctions worldwide. Science 306:1783-1786.

Thomas, C. D., A. Cameron, R. E. Green, M. Bakkenes, L. J. Beaumont, Y. C. Collingham, B. F. Erasmus, M. F. De Siqueira, A. Grainger, L. Hannah, L. Hughes, B. Huntley, A. S. van Jaarsveld, G. F. Midgley, L. Miles, M. A. Ortega-Huerta, A. Townsend Peterson, O. L. Phillips, and S. E. Williams. 2004. Extinction risk from climate change. Nature 427:145- 148.

Tilley, S. G. and M. J. Mahoney. 1996. Patterns of genetic differentiation in salamanders of the Desmognathus ochrophaeus complex (Amphibia: Plethodontidae). Herpetological Monographs 10:1.

Tilman, D., J. Fargione, B. Wolff, C. D’Antonio, A. Dobson, R. Howarth, D. Schindler, W. H. Schlesinger, D. Simberloff, and D. Swackhammer. 2001. Forecasting agriculturally driven global environmental change. Science 292:281-284.

Vilà, M., J. L. Espinar, M. Hejda, P. E. Hulme, V. Jarošík, J. L. Maron, J. Pergl, U. Schaffner, Y. Sun, P. Pyšek. 2011. Ecological impacts of invasive alien plants: a meta-analysis of their effects on species, communities and ecosystems: Ecological impacts of invasive alien plants. Ecology Letters 14:702-708.

Vellend, M., L. Harmon, J. Lockwood, M. Mayfield, A. Hughes, J. Wares, and D. Sax. 2007. Effects of exotic species on evolutionary diversification. Trends in Ecology & Evolution 22:481-488.

Wake, D. B. and V. T. Vredenburg. 2008. Are we in the midst of the sixth mass extinction? A view from the world of amphibians. Proceedings of the National Academy of Sciences 105:11466-11473.

Waples, R. S. and C. Do. 2008. LDNE: a program for estimating effective population size from data on linkage disequilibrium. Molecular Ecology Resources 8:753-756.

Waples, R. S. and O. Gaggiotti. 2006. What is a population? An empirical evaluation of some genetic methods for identifying the number of gene pools and their degree of connectivity. Molecular Ecology 15:1419-1439.

Welsh, H. H., Jr. and S. Droege. 2001. A case for using plethodontid salamanders for monitoring biodiversity and ecosystem integrity of North American forests. Conservation Biology15:558-569.

Windmiller, B. 2000. Study of Ecology and Population Response to Construction in Upland Habitat by Vernal Pool Amphibians. Report for the Massachusetts Natural Heritage Program. Sudbury, MA.

Wright, S. 1943. Isolation by distance. Genetics 28:114.

Zamudio, K. R. and A. M. Wieczorek. 2007. Fine-scale spatial genetic structure and dispersal among spotted salamander (Ambystoma maculatum) breeding populations. Molecular Ecology 16:257-274.

Zvereva, E. L. and M. V. Kozlov M.V. 2010. Responses of terrestrial arthropods to air pollution: a meta-analysis. Environmental Science and Pollution Research 17:297-311.

Figures

Figure 1.1. Encountered individual and site locations for the Four-toed Salamander, Hemidactylium scutatum, plotted in ArcGIS 10.2.1 using GPS coordinates. Inset A shows locations of encountered individuals at sites I-AN and I-AS at the Middle Fork State Fish and Wildlife Area, Vermilion County, IL from 2009 to 2012. Inset B depicts 14 sampling locations in North America with respect to the last glacial maximum (Ehlers et al. 2011). Inset C shows a magnified view of seven sampling sites in northern. Sample size, GPS coordinates, and acronym for each site are provided in Table 1.1.

Figure 1.2. Probabilistic allocation of 157 H. sctuatum to two gene pools as determined by STRUCTURE (K=2). Colors represent gene pools. Data are based on eight microsatellite loci. Samples were collected from two sites in Illinois I-AN and I-AS (see Figure 1.1). Sample size and acronym for each site is provided in Table 1.1.

Figure 1.3. Probabilistic allocation of genotypes of 279 H. scutatum to four gene pools as determined by STRUCTURE (K=4). Colors represent gene pools. Data are based on eight microsatellite loci. Samples were collected from 14 sites across Illinois and Kentucky (see Figure 1.1 and 1.2). Sample size and acronyms for each site is provided in Table 1.1.

Figure 1.4. Results from a RxC analysis of 279 H. scutatum genotypes. Data are based on eight microsatellite loci. Samples were collected from 14 sites across Illinois and Kentucky (see Figure 1.1). Samples size and acronyms for each site are provided in Table 1.1. Each circle represents site or cluster of sites that the RxC analysis indicated as belonging to the same population. Lines between circles indicate nonsignificant results for a multilocus contingency test of heterogeneity of allele frequencies among pairs of samples. Sites that can be linked via a series of multiple nonsignificant tests are considered to be part of the same population.

Figure 1.5. Relationship of pairwise geographic distance (km) and genetic divergence (FST) between 14 sample sites of H. scutatum located across Illinois and Kentucky (see Figure 1.1). Sample size for each site is provided in Table 1.1.

Tables Table 1.1. Summary of 279 individual H. scutatum tissue samples collected during visual encounter surveys across 14 sites in central Illinois and Kentucky. Samples from I-AN and I-AS were examined for patterns at the local scale, whereas all samples were included in the regional analysis. Listed are: Code = site acronym, N = number of samples, State = site state, KY Cluster = membership of the site in the north or south cluster in Kentucky, County = site county, UTM Coordinates = coordinates of mean centroid calculated from all sampled individual locations. Geographic location of sites is depicted in Figure 1.1.

Code N State KY Cluster County UTM Coordinates Local I-AN 111 IL N/A Vermilion 433114 4454580 I-AS 46 IL N/A Vermilion 433033 4453804 Regional I-AN05 8 IL N/A Vermilion 433114 4454580 I-B 7 IL N/A Jo Daviess 718487 4687128 K-A 12 KY North Powell 264610 4187946 K-B 4 KY North Menifee 273080 4198054 K-C 2 KY South Laurel 745448 4101073 K-D 29 KY South Adair 668775 4119508 K-E 7 KY North Powell 265866 4188298 K-F 3 KY North Powell 264949 4187610 K-H 5 KY North Powell 264672 4187874 K-J 6 KY North Powell 263948 4189143 K-L 23 KY North Wolfe 272711 4183552 K-M 7 KY North Wolfe 272752 4187911 K-N 9 KY North Menifee 264754 4191931

Table 1.2. Amplification success of microsatellite loci for application in H. scutatum. Species = taxon for which the locus was developed. GenBank # = accession number for locus sequence in GenBank. Locus = locus name. Citation = publication originally describing the locus. Amplification = success (+) or failure (-) of loci to amplify in initial tests evaluated on agarose gels. Genotyping = success (+) or failure (-) of genotyping loci resulting from variation in primer specificity, signal strength, and occurrence of locus polymorphism. Dye =fluorescent dye used to label the primer for sequencing. The remaining parameters are derived from 279 samples analyzed in this study, including: Allele Size Range (bp) = size range of alleles in base pairs, N = observed number of alleles, Na = expected number of alleles, HO = observed heterozygosity, HE = expected heterozygosity.

Dye

nge (bp)

Panel

Locus

Species

Citation

Allele Size Size Allele

GenBank # GenBank

Genotyping

Amplification

N/A HS3a + + A PET 199-268 22 7.4 0.59 0.85

(1996)

N/A HS3b McGrath + + A PET 149-197 16 7.2 0.66 0.86

N/A HS5 + + B VIC 224-256 26 8.5 0.48 0.87

N/A HS7 + + B 6-FAM 93-178 35 8.2 0.42 0.86

N/A HS8 + + B 6-FAM 109-188 28 8.0 0.54 0.84

N/A HS14 + + C NED 158-205 17 6.7 0.71 0.82

Schrecengost (1998) Schrecengost

Hemidactylium scutatum Hemidactylium

N/A HS15 + ?

Reid

(1994)

AY532595 PE0 + ?

AY532596 PE1 - -

AY532597 PE3 + ?

AY532598 PE4 - -

AY532599 PE5 - -

Plethodon elongatus Plethodon

Degross et al. (2004) al. et Degross AY532600 PE7 + ?

AY532601 PE8 + ?

Table 1.2 (continued)

AY532602 PE9 + -

AY532603 PE10 + -

AY532604 PE11 + -

AY532605 PE12 - -

AY151377 PC1 + + C VIC 427-455 10 3.1 0.45 0.54

AY151374 PC2 + ?

AY151380 PC3 - -

AY151379 PC4 - -

P. cinereus AY151372 PC5 + -

AY151376 PC6 (2003) & Cabe Connors - -

AY151373 PC7 - -

AF149305 DT4 + + A NED 265-291 8 2.5 0.24 0.32

AF150725 DT5 + ?

Curtis & Curtis

tenebrosus

Dicamptodon Dicamptodon AF150728 DT8 (2001) Taylor + + A 6-FAM 101-135 11 3.5 0.38 0.52

Table 1.3. Distribution of genetic variance in H. scutatum at both local and regional scales as determined by Hierarchical Analysis of Molecular Variance (AMOVA). Results are based on genotypes across eight microsatellite loci of (A=Local Analysis) 157 samples collected from two sites (I-AN and I-AS) in Illinois, and (B=Regional Analysis) 279 samples collected from 13 sites in Illinois and Kentucky. Variance was evaluated at three hierarchical levels (Source of Variation). Listed are: d.f. = degrees of freedom, S.S. = sums of squares, Variance Components = amount of variance due to each source of variation, % = percent of total variance. Sample size and acronyms for each site is provided in Table 1.1. Geographic locations of all 13 sites are shown in Figure 1.1.

(A) Local Analysis Source of Variation d.f. S.S. Variance Components % Among populations 1 4.01 0.01 Va 0.32 Among individuals 155 454.10 0.35 Vb 13.67 Within individuals 157 349.00 2.22 Vc 86.01 Total 313 807.11 2.58 100.00

(B)

Regional Analysis Source of Variation d.f. S.S. Variance Components % Among populations 12 131.76 0.27 Va 9.04 Among individuals 266 856.64 0.54 Vb 18.37 Within individuals 279 596.50 2.14 Vc 72.59 Total 557 1584.90 2.95 100.00

Table 1.4. Pairwise population FST values of H. scutatum between three Illinois sampling sites (I- AN, I-AS, and I-B) and 11 combined Kentucky sampling sites (KY). Data are based on eight microsatellite loci. Geographic locations of these sites are provided in Figure 1.1. Acronyms and samples size for each site are provided in Table 1.1. FST values are below diagonal. P-values are above diagonal. To reduce the chance of Type I error, the Bonferroni corrected p-value needed for significance is p<0.008. Significant p-values indicated in bold.

I-AN I-AS I-B KY I-AN 0.106 0.000 0.000 I-AS 0.004 0.000 0.000 I-B 0.135 0.103 0.000 KY 0.076 0.074 0.100

Table 1.5. Results of Wilcoxon tests (p-values) for recent bottlenecks in H. scutatum calculated under the Infinite Alleles Model (IAM) for four sampling sites of (K-A, K-D, K-L, and I-A). Data are based on eight microsatellite loci. Geographic locations of these sites are provided in Figure 1.1. Acronyms and sample size for each site are provided in Table 1.1. To reduce the chance of Type I error, for each model, the Bonferroni corrected p-value needed for significance is p<0.0125.

Wilcoxon Test K-A 0.195 K-D 0.250 K-L 0.004 I-A 0.055

CHAPTER 2. SOURCES OF VARIATION IN THE VENTRAL SPOT PATTERN OF THE FOUR-TOED SALAMANDER

Introduction

Variation in color pattern, including the hue, brightness, number, size, shape, and location of stripes, patches, or spots, is a common phenomenon in many taxa. The functions of this variation are diverse and include predator avoidance through cryptic, disruptive, or aposematic coloration, mimicry, and startle or distracting coloration (eye spots), as well as ornamental coloration for attraction of mates, photoprotection, structural support, microbial resistance, thermoregulation, and communication (see reviews in Hubbard et al. 2010, Protas and Patel

2008, Roulin 2004). In addition, color pattern may be a mechanism for the recognition of conspecifics (i.e. Secondi et al. 2010) and possibly kinship as intraspecific variation in color pattern has been detected on a fine enough spatial scale to allow populations (Costa et al.2009), genders (i.e. Davis and Grayson 2008, Pokhrel 2009, Todd and Davis 2007), and even individuals to be distinguished (Breitenbach 1982, Carafa and Biondi 2004, Eitam and Blaustein

2002). Geographic variation in color pattern, moreover, is widespread among amphibians and has been attributed to the interaction between environment and predation (Storfer et al. 1999), the interaction between predation pressures and morphotype frequencies (Hegna et al. 2013), environmental characteristics (Fernandez and Collins 1988), resource partitioning (Denoël et al.

2001) or random events (Gray 1983).

In intraspecific interactions, color pattern may be used as an indicator of kinship. Many studies have found color pattern in salamanders to be heritable (Highton 1975, Lipsett and Piatt

1936, Twitty 1961) and visual displays and cues are used in many interactions (i.e. Kohn and

Jaeger 2009, Secondi et al. 2010, Thaker et al. 2006). These characteristics have the potential to create selection pressure for the use of color pattern in kinship mediated social interactions in

36 salamanders. Kinship mediated behaviors have been implicated in a number of salamander species, including A. opacum (Hokit et al. 1996, Walls and Blaustein 1995, Walls and

Roudebush 1991), A. tigrinum (Pfennig and Collins 1993, Pfennig et al. 1999), Hemidactylium scutatum (Harris et al. 2003), Hynobius retardatus (Wakahara 1997), Notophthalmus viridescens

(Gabor 1996), and Plethodon cinereus (Gibbons et al. 2003, Wareing 1997). While some evidence exists that chemical cues play an important role in kin recognition among larvae

(Chivers et al. 1997, Pfennig et al. 1994, Wakahara 1997), no work has attempted to untangle the role of color pattern in kin recognition among terrestrial adults.

Intraspecific variation in color pattern might also be an indicator of variation in health among individuals, especially if the color pattern is costly to produce, normally symmetrical, and sensitive to conditional or fluctuating asymmetry. Fluctuating asymmetry (Ludwig 1932), differences in the right and left measurements of a normally symmetrical bilateral character, is generally considered to be the result of disturbances in intrinsic (e.g., cellular development) or extrinsic (e.g. habitat fragmentation, pollution, parasite load) environments (Møller and Swaddle

1997). Increasing asymmetry in dorsal spot pattern has been correlated with decreasing body condition and increasing habitat disturbance in A. maculatum (Davis and Maerz 2007, Wright and Zamudio 2002) and asymmetry in dorsal pigmentation in A. opacum has been interpreted as an indicator of differences in stress response between the sexes (Pokhrel 2009, Pokhrel et al.

2013).

The Four-toed Salamander, H. scutatum, is a relatively small plethodontid salamander

(~5-10cm Total Length; Petranka 1998), exhibiting a white ventral surface patterned with distinctive black spots (Figure 2.1). This ventral skin pigmentation pattern has been well documented as a variable character and has been used reliably to identify individuals in field

37 studies (Breitenbach 1982, Carreno and Harris 1998, Harris et al. 1995, Harris and Gill 1980,

Harris and Ludwig 2004). While it has been considered aposematic coloration in this species

Brodie et al. (1974), spot pattern has not been quantified in H. scutatum. Because aposematic color patterns experience selection for symmetry (Forsman and Herrström 2004, Forsman and

Merilaita 1999), it would be expected that the ventral spot pattern in H. scutatum is, under ideal conditions, symmetrical. Additionally, H. scutatum has a broad geographic distribution that extending north to Nova Scotia, south to Florida, and as far west as Oklahoma and Missouri

(Petranka 1998), but with patchy occurrences in the southern and western areas. Because populations tend to be disjunct and centralize around patches of forested areas surrounding spring or seep-fed pools, population structure in this species should lead to increased geographic variation in phenotypic traits. These specific habitat requirements, coupled with the small body size and susceptibility to desiccation of H. scutatum, are predictors of limited lifetime dispersal.

This should increase the rate of encounters between relatives within a population and create conditions where selection for mechanisms of kin recognition, such as the use of the ventral spot pattern, would arise. Finally, as H. scutatum is designated as either ‘imperiled’ or ‘critically imperiled’ throughout much of its range (NatureServe 2011), the ability to identify at risk populations (i.e. those experiencing heightened levels of stress) by a simple, phenotypic characteristic such as increased asymmetry of spot pattern, would be a useful conservation and management tool

This study examined ventral spot pattern in H. scutatum across different spatial scales to ask the following questions: (1) Does spot pattern vary geographically? If spot pattern varies geographically or among populations, then symmetry and/or overall abundance of spotting should vary between regions across the range of H. scutatum. Geographic variation in this

38 feature would make it a useful tool for taxonomists. (2) Is spot pattern similarity within populations correlated with genetic relatedness among individuals? If ventral spot pattern of H. scutatum is genetically determined, spot pattern similarity should increase with increased relatedness (i.e., genetic similarity). (3) Does spot pattern vary with body condition? If spot pattern is an indicator of body condition, bilateral asymmetry in spot pattern should increase as body condition decreases. I used Quantitative Image Analysis (QIA) to quantify spot pattern in terms of proportion of spots and symmetry and examined how these variables vary regionally. I used microsatellite markers to estimate relatedness between individuals and examined the potential relationship between genotypic similarity (relatedness) and phenotypic similarity

(similarity in spot pattern) between individuals. Finally, I estimated body condition using a residual index and investigated the potential relationship between spot pattern symmetry and body condition.

Methods

Samples and Collection Sites

I sampled from March to November over four consecutive years (2009-2013) while H. scutatum was active at two sampling site (I-AN and I-AS) in the Middle Fork State Fish and

Wildlife Area in Vermilion County, Illinois and from 18-23 March 2011 at eight sites (K-A, K-

D, K-E, K-H, K-J, K-L, K-M, K-N) in central Kentucky (Figure 2.2). At each site, I searched for individuals along haphazardly placed transects by looking under potential shelter objects. For each encountered individual, I recorded the GPS location in UTM coordinates using a Garmin

GPS 12, and measured the body size (mm) as Snout-Vent-Length (SVL), Total-Length (TL) and

39 mass (g). For genetic analyses, I collected small tail clips (1-5 mm) from individuals weighing more than 0.18 g using sterilized clippers and tissues were stored in ethanol at -80 °C.

I photographed the ventral spot pattern of each salamander for use in individual identification and to avoid repeated tissue samples from the same individual (Breitenbach 1982).

To immobilize individuals for photographs with a digital camera, I placed each salamander within a re-sealable plastic bag and gently flattened the bag so that the animal was held between the layers of plastic. Photographic sampling has many benefits as outlined by Todd et al. (2001) including speed, creation of an instant, permanent record, its non-invasive or destructive quality, and the opportunity for repeat measurements in the future. Digital photography allowed me to document the spot pattern of each individual for later scoring in the lab and to create a record of sampled individuals for the identification of recaptures.

Molecular Methods and Genetic Analyses

To infer relatedness, I used microsatellite analysis to determine genetic similarity of individuals using eight microsatellite markers as described in Chapter 1. I calculated relatedness r as proposed by Queller and Goodnight (1989) using the methodology described in Chapter 3.

Quantifying Body Condition

To quantify body condition for sampled individuals, I used the residuals from a regression of log-transformed mass on log-transformed total length as an index (Harris 2007,

Harris and Ludwig 2004, Schulte-Hostedde et al. 2005). This approach to measuring body condition, originally proposed by Gould (1975), controls for variation across body sizes (Jakob

40 et al. 1996) and satisfies more assumptions than indices based on adjusting body mass by length

(Schulte-Hostedde et al. 2005).

Photographic Analyses

I used the plugin STRAIGTEN (Kocsis et al. 1991) for the IMAGEJ photo analysis software

(Abramoff et al. 2004) to straighten the image for each individual along an approximated medial line. STRAIGHTEN uses an algorithm to fit a non-uniform cubic spline along user defined nodes along the approximated medial line. I rotated the images so that all were oriented vertically with the head of the salamander at the top of the image. Because missing or regrown tails would alter the ventral skin pigmentation pattern recorded for an individual, I chose a subset of the ventral area for my region of interest (ROI). This included the ventral area below the points where the gular fold reaches the lateral sides of the salamander and above the most anterior portion of the vent. The ROI width matched the maximum width of the salamander’s ventral surface as bordered by the edges of the lateral coloration. The ventral spot pattern was then quantified following Todd et al. (2005) using quantitative image analysis (QIA). This approach, QIA or the assessment of image data, is useful for examining the size, shape, number, and placement of patches or spots in a color pattern and has successfully been used in amphibian species [i.e. spectacled salamanders, Salamandrina sp. (Angelini et al. 2010) and fire-bellied toads, Bombina sp. (Biancardi and Di Cerbo 2010)]. To remove the factor of body size from analysis I shrunk/fit the ROI to fit a standard lattice of 10 x 65 cells. I scored each cell of the grid visually and subjectively as a 1 or 0 representing the presence (>50% covered) or absence (<50% covered) of a ventral spot respectively. The completed lattice for a given individual was copied to a spreadsheet and I rearranged the original grid into a single spreadsheet column so that each

41 succeeding column of the grid was stacked below the one that preceded it. The total of the values in this column gives the phenotype score for an individual and represents the amount of spotting present in the ROI and reflects the lightness (low phenotype score) or darkness (high phenotype score) of an individual.

To assess the phenotypic similarity of two individuals, I summed the absolute values of the differences of corresponding cells representing the same grid cell/region for each salamander.

I divided this sum by the total number of cells to find the proportion of cells with identical scores, i.e. the phenotypic similarity score, for each pair of individuals. This score ranges from 0 to 1 with lower values indicating greater similarity in terms of presence or absence of spotting in a given grid cell.

I evaluated the symmetry of an individual by finding the sum of the absolute values of the difference between each cell on the left side of the scoring grid with the corresponding cell in the mirrored location on the right side of the grid. I divided this value by the total number of cells to provide a proportion of cells, i.e. the symmetry score, for which an individual was bilaterally symmetrical. This score ranged from 0 (symmetrical) to 1 (asymmetrical).

Statistical Analysis

To validate the efficacy of this method of phenotypic scoring, a subset of individuals was photographed multiple times and these photos scored independently. I then calculated an F- statistic to compare the variance with individual phenotypic similarity scores with the variance between individual phenotypic similarity scores. To assess whether spot pattern varies geographically, I used two-tailed t-tests with unequal variances to compare the mean phenotype and symmetry scores between the combined populations from Illinois and combined populations

42 from Kentucky sites. I also used a two-tailed t-test with unequal variances to compare the mean body condition for individuals with complete tails in Illinois vs. individuals with complete tails in Kentucky. For the two sampling sites within Illinois (I-AN and I-AS) and using only individuals with complete tails, I modeled the impacts of SVL, mass, and sampling site as a function of phenotype scores using the LME4: LINEAR MIXED-EFFECTS MODELS USING EIGEN AND

S4 PACKAGE V. 1.1-7 (Bates et al. 2014) in R V3.1.1 (R Core Team 2014). Because I was interested in the effect of body size on phenotypes, I modeled SVL and mass as fixed effects and sampling site as a random effect. The fit of this full model was compared to a null model which contained only the site variable by calculating AIC values corrected for small sample size (AICc,

Burnham and Anderson 2002) using AICCMODAVG: MODEL SELECTION AND MULTIMODEL

INFERENCE BASED ON (Q)AIC(C) PACKAGE V. 2.0-3 (Mazerolle 2015) in R. I also calculated the likelihood of each model (-2 log likelihood), differences in AICc values (Δi) and the Akaike weights (Δwi) for these models. Models with Δi<2 should be considered competing models

(Burnham and Anderson 2002). An Aikaike weight is interpreted as the variability accounted for by the model, i.e. the best model among those evaluated will have the highest Δwi. The impact of

SVL, mass, and sampling site were similarly modeled as a function of phenotype symmetry score. To investigate the potential for allometric changes in the phenotype score, i.e. spot density, and symmetry score, I assessed potential correlations between individual size and spot pattern, using both salamander SVL and mass measurements. I conducted a Mantel test using the

VEGAN: COMMUNITY ECOLOGY PACKAGE V2.0-10 (Oksanen et al. 2013) in R V3.1.1 (R Core

Team 2014) to examine the potential relationship between phenotypic (phenotypic similarity score) and genotypic similarity (relatedness, r). I investigated the potential for fluctuating

43 asymmetry in H. scutatum by assessing potential correlations between symmetry score and body condition and between phenotype score and body condition.

Results

I collected 180 tissue samples with their corresponding photos (Figure 2.2). Of these, 86

(85 with complete tails) came from site I-AN and 48 (47 with complete tails) came from site I-

AS in Illinois and 46 (with complete tails) from 8 sites in Kentucky. An additional 44 tissue samples of individuals with complete tails but no photo data were collected from Kentucky for use in the comparison of body condition indices. Because of small sample sizes representing sites in Kentucky (range = 1–18) samples were pooled for analysis.

Variation in phenotypic similarity scores between redundant photos of the same individual (range = 2-4 photos for each of 11 individuals) was significantly lower than scores of photos taken of different individuals (F=233.24, p-value=0.00). This indicates that this method of QIA should allow reliable quantification of phenotype.

Mean phenotype scores (p-value = 0.00, Figure 2.3) and mean individual symmetry scores (p-value = 0.00, Figure 2.4) were significantly lower for individuals from Kentucky compared to individuals from Illinois. This geographic variation was characterized by lower levels of spotting and higher pattern symmetry in samples from Kentucky. This variation does not appear to be due to differences in body condition between these two areas as individuals from Kentucky did not differ significantly in body condition from individuals from Illinois (p- value=1.00, Figure 2.5). Size of individuals, measured by SVL or mass does not appear to play a significant role in their phenotype (Figures 2.6 and 2.7) or symmetry (Figures 2.8 and 2.9) scores. The AIC analysis supports this, suggesting that the null models, i.e. the models

44 containing on the site variable, as the most parsimonious model for both phenotype and symmetry score model sets (Tables 2.1 and 2.2). Combined results suggest that individuals do not experience significant ontogenetic changes in spot pattern over time.

I found no relationship between phenotypic similarity scores and inter-individual relatedness (p-value=0.71, Figure 2.10). Finally, I found no significant relationship between phenotype score (Figure 2.11) or symmetry score (Figure 2.12) and body condition, indicating that spot pattern, in terms of overall amount of spots or lateral symmetry, is not influenced by body condition.

Discussion

This study found significant geographic variation in the ventral spot pattern of H. scutatum, both in terms of overall amount of spotting and degree of bilateral symmetry. No evidence was found, however, to suggest that spot pattern is influenced by relatedness between individuals, or by body condition. This lack of association between spot pattern and these variables could be due to (a) a lack of heritability in spot pattern, (b) the use of alternative signals to assess relatedness between individuals, or (c) violations in the assumptions made by the fluctuating asymmetry analysis. Each of these scenarios is discussed in detail below.

Geographic Variation

The ventral spot pattern of H. scutatum could be a potential tool for identifying populations of taxonomic or conservation importance as it shows significant variation across the portion of its range sampled in this study. This may be due to differences in the underlying genetic variation. While low genetic diversity, i.e. stresses resulting from inbreeding, could

45 account for the increased asymmetry in spot scores observed at sites in Illinois, recent population genetics work suggest high levels of heterozygosity at these sites (Chapter 1). Alternatively, because no relationship between symmetry and body condition was found and because the mean body condition, as measured by the residual index, did not differ between Illinois and Kentucky, this might simply reflect genetic variation between the two regions (Chapter 1). The differences in phenotype and symmetry scores between Illinois and Kentucky, therefore, may have been induced by genetic drift influencing alleles responsible for spot pattern (Costa et al. 2009). Rudh et al. (2007) suggested that variation in natural or sexual selection could be responsible for the geographic variation in dorsal spot pattern they observed in Dendrobates pumilio, but it seems unlikely that these explanations apply to H. scutatum whose ventral spot pattern is less likely to be seen by predators and conspecifics.

Influence of Size and Age

No relationship was found between individual size, either measured by SVL or mass, and spot pattern, either measured by phenotype or symmetry score, suggesting that spot pattern in this species does not change with age. This was expected and reinforces the utility of these spot patterns for individual identification over long time periods (Harris and Ludwig 2004). Work on other plethodontid species has also found relatively high consistency in spot pattern over time

(Bendik et al. 2013).

Relatedness and Phenotypic Similarity

In my analyses, phenotypic similarity scores and relatedness between pairs of individual

H. scutatum were not correlated. A primary explanation could be that ventral spot pattern is

46 influenced by the environment. If spot pattern has low heritability, and assuming kin recognition occurs among adult H. scutatum, individuals must assess the identity of relatives using alternative characteristics. In H. scutatum, larvae are capable of kin recognition (Harris et al.

2003) and it is probable that, like other salamander larvae (Chivers et al. 1997, Pfennig et al.

1994, Wakahara 1997), they use chemical signals to do so. Among terrestrial plethodontids, chemosensory cues are used in different contexts to communicate a variety of information (i.e.

Jaeger and Forester 1993) and have been implicated in kin recognition (Wareing 1997). Thus, it is likely that the ability to assess kinship via chemical cues is retained in adult H. scutatum.

Additionally, the ventral location of the spot pattern may make this trait less useful in intraspecific communication. The spot pattern in H. scutatum may serve as aposematic coloration as this species produces noxious skin secretion (Brodie 1977) and, when encountered by a predator, often remains immobile and allows itself to be turned over with the ventral spot pattern exposed (Brodie et al. 1974).

Alternatively, a relationship between phenotypic and genotypic similarity may exist, but was not detected with the QIA techniques used in this study. While variation in photos taken of the same individual was significantly lower than variation in photos taken of different individuals, the degree of resolution obtained through these methodologies may only be capable of detecting broad scale patterns such as the geographic variation observed between the Illinois and Kentucky sampling sites.

Fluctuating Asymmetry

The usefulness of fluctuating asymmetry as a measure of stress or fitness in an organism varies with the trait examined (e.g. Møller and Swaddle 1997). In this study, no association was

47 observed between ventral spot pattern symmetry and body condition. There are a number of explanations for this observation. First, fluctuating asymmetry may not be associated with color pattern in H. scutatum. While fluctuating asymmetry has been observed in the dorsal spot and pigmentation patterns in some amphibian species, including A. maculatum (Davis and Maerz

2007, Pokhrel 2009, Wright and Zamudio 2002) and A. opacum (Pokhrel 2009, Pokhrel et al.

2013), a lack of fluctuating asymmetry with respect to dorsal spot pattern has been observed in

Notophthalmus viridescens (Sherman et al. 2009). Secondly, the assumption that ventral spot pattern in H. scutatum is symmetrical may have been violated. Third, spot pattern in H. sctutatum forms very early on, i.e. shortly after metamorphoses (pers. obs.). This methodology assumes that the measured body condition of an individual at the time of the study is indicative of the body condition of an individual at the time of spot formation. If an individual’s body condition has changed from the time of spot formation to the time when measured, this assumption would be violated. Finally, the relationship between fluctuating asymmetry and fitness is stronger when measured on traits undergoing growth (Aparicio 2001) or on traits that are costly to grow (Aparicio and Bonal 2002). Most of the individuals in this study were well past the point of metamorphosis and spot formation and this may have obscured any existing relationship between symmetry score and body condition.

Conclusion

In conclusion, I found that spot pattern varies geographically in H. scutatum. The significant variation in spot pattern between regions and the lack of relationship between individual spot pattern and size reinforces the practicality of using this trait to identify individuals. If variation in spot pattern exists on a finer scale (i.e. between populations), this trait

48 could also assist in conservation and management decisions by allowing recent immigrants in a population to be identified and the degree of gene flow between populations to be estimated. I did not, however, find support for the hypothesis that phenotypic similarity between individuals is an indicator of their genotypic similarity, i.e. relatedness. Spot pattern does not appear to be a viable substitute for molecular methods in evaluating relatedness among individuals. I also found no support for the hypothesis that the ventral spot pattern in H. scutatum varies with body condition (i.e. stress or fitness) of individuals or populations via the phenomenon of fluctuating asymmetry. This suggests that spot pattern is also not a useful tool for assessing relative stress among populations.

Literature Cited

Abramoff, M. D., P. J. Magalhaes, and S. J. Ram. 2004. Image Processing with ImageJ. Biophotonics International 11:36-42.

Angelini, C., C. Costa, S. Raimondi, P. Menesatti, and C. Utzeri. 2010. Image analysis of the ventral colour pattern discriminates between Spectacled salamanders, Salamandrina perspicillata and S. terdigitata (Amphibia, Salamandridae). Amphibia-Reptilia 31:273– 282.

Aparicio, J. M. 2001. Patterns of growth and fluctuating asymmetry: the effects of asymmetrical investment in traits with determinate growth. Behavioral Ecology and Sociobiology 49:273-282.

Aparicio, J. M. and R. Bonal 2002. Why do some traits show higher fluctuating asymmetry than others? A test of hypotheses with tail feathers of birds. Heredity 99:139-144.

Bates, D., M. Maechler, B. Bolker, and S. Walker. 2014. lme4: Linear mixed-effect models using Eigen and S4. R package version 1.1-7. http://CRAN.R-project.org/package=lmer4.

Bendik, N. F., T. A. Morrison, A. G. Gluesenkamp, M. S. Sanders, and L. J. O’Donnell. 2013. Computer-Assisted Photo Identification Outperforms Visible Implant Elastomers in an Endangered Salamander, Eurycea tonkawae. PloS One 8:e59424.

Biancardi, C. M. and A. R. Di Cerbo. 2010. Quantitative pattern analysis methodology in amphibians. Atti VIII Congresso Nazionale SHI (Chieti, 22-26 September 2010):383- 390.

Brede, E. 2000. A morphometric study of a hybrid newt population (Triturus cristatus/T. carnifex): Beam Brook Nurseries, Surrey, U.K. Biological Journal of the Linnean Society 70:685–695.

Breitenbach, G. L. 1982. The frequency of communal nesting and solitary brooding in the salamander, Hemidactylium scutatum. Journal of Herpetology 16:341-346.

Brodie, E. D., Jr. 1977. Salamander antipredator postures. Copeia 1977:523-535.

Brodie, E. D., Jr., J. A. Johnson, and C. K. Dodd, Jr. 1974. Immobility as a defensive behavior in salamanders. Herpetological 30:79-85.

Burnham, K. P. and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd ed. Springer-Verlag, New York.

Carafa, M. and M. Biondi. 2004. Application of a method for individual photographic identification during a study on Salamandra salamandra gigliolii in central Italy. Italian Journal of Zoology 71:181–184.

Carreno, C. A. and R. N. Harris. 1998. Lack of nest defense behavior and attendance patterns in a joint nesting salamander Hemidactylium scutatum (Caudata: Plethodontidae). Copeia 1998:183-189.

Chivers, D. P., E. L. Wildy, and A. R. Blaustein. 1997. Eastern long-toed salamander (Ambystoma macrodactylum columbianum) larvae recognize cannibalistic conspecifics. Ethology 103:187–197.

Costa, C., C. Angelini, M. Scardi, P. Menesatti, and C. Utzeri. 2009. Using image analysis on the ventral colour pattern in Salamandrina perspicillata (Amphibia: Salamandridae) to discriminate among populations. Biological Journal of the Linnean Society 96:35–43.

Davis, A. K. and K. L. Grayson. 2008. Spots of adult male red-spotted newts are redder and brighter than in females: evidence for a role in mate selection? The Herpetological Journal. 18:83–89.

Davis, A. K. and J. C. Maerz. 2007. Spot symmetry predicts body condition in spotted salamanders, Ambystoma maculatum. Applied. Herpetology 4:195-205.

Denoël, M., P. Poncin, and J. C. Ruwet. 2001. Sexual compatibility between two heterochronic morphs in the alpine newt, Triturus alpestris. Animal Behaviour 62:559–566.

Eitam, A. and L. Blaustein. 2002. Noninvasive individual identification of larval Salamandra using tailfin spot patterns. Amphibia-Reptilia 23:215-219.

Fernandez, P. J., Jr., and J. P. Collins. 1988. Effect of environment and ontogeny on color pattern variation in arizona tiger salamanders (Ambystoma tigrinum nebulosum Hallowell). Copeia, 1988:928–938.

Forsman, A. and S. Merilaita. 1999. Fearful symmetry: pattern size and asymmetry affects aposematic signal efficacy. Evolutionary Ecology 13:131-140.

Forsman, A., and J. Herrström. 2004. Asymmetry in size, shape, and color impairs the protective value of conspicuous color patterns. Behavioral Ecology 15:141-147.

Gabor, C. R. 1996. Differential kin discrimination by red-spotted newts (Notophthalmus viridescens) and smooth newts (Triturus vulgaris). Ethology 102:649-659.

Gibbons, M. E., A. M. Ferguson, D. R. Lee, and R. G. Jaeger. 2003. Mother-offspring discrimination in the red-backed salamander may be context dependent. Herpetologica 59:322-333.

Gould S. 1975. Allometry in primates, with emphasis on scaling and evolution of the brain. Pp. 244–292. In F.S. Szalay (Ed.), Approaches to Primate Paleobiology. Basel: Karger.

Gray, R. H. 1983. Seasonal, annual, and geographic variation in color morph frequencies of the cricket frog, Acris crepitans, in Illinois. Copeia 1983:300.

Harris, R. N. 2007. Body condition and order of arrival affect cooperative nesting behaviour in four-toed salamanders Hemidactylium scutatum. Animal Behaviour 75:229-233.

Harris, R. N. and D. E. Gill. 1980. Communal nesting, brooding behavior, and embryonic survival of the four-toed salamander Hemidactylium scutatum. Herpetologica 36: 141- 144.

Harris, R. N., W. W. Hames, I. T. Knight, C. A. Carreno, and T. J. Vess. 1995. An experimental analysis of joint nesting in the salamander Hemidactylium scutatum (Caudata: Plethodontidae): the effects of population density. Animal Behaviour 50:1309-1316.

Harris, R. N., T. J. Vess, J. I. Hammond, and C. J. Lindermuth. 2003. Context-dependent kin discrimination in larval four-toed salamanders Hemidactylium scutatum (Caudata: Plethodontidae). Herpetologica 59:164-177.

Harris, R. N. and P. M. Ludwig. 2004. Resource level and reproductive frequency in female four-toed salamander, Hemidactylium scutatum. Ecology 85:1585-1590.

Hegna, R. H., R. A. Saporito, and M. A. Donnelly. 2013. Not all colors are equal: predation and color polytypism in the aposematic poison frog Oophaga pumilio. Evolutionary Ecology 27:831–845.

Highton R. 1975. Geographic variation in genetic dominance of the color morphs of the red- backed salamander, Plethodon cinereus. Genetics 80:363–374.

Hokit, D. G., S. C. Walls, and A. R. Blaustein. 1996. Context-dependent kin discrimination in larvae of the marbled salamander, Ambystoma opacum. Animal Behaviour 52:17-31.

Hubbard, J. K., J. A. C. Uy, M. E. Hauber, H. E. Hoekstra, and R. J. Safran. 2010. Vertebrate pigmentation: from underyling genes to adaptive function. Trends in Genetics 26:231- 239.

Lipsett J. C. and J. Piatt. 1936. Experimental Hybridization of Triturus v. viridescens and Triturus v. symmetrica. Copeia. 1936:1.

Ludwig, W. 1932. Das Rechts-Links Problem im Tierreich und beim Menschen. Springer, Berlin.

Jaeger, R. G. and D. C. Forester. 1993. Social behavior of plethodontid salamanders. Herpetologica 49:163-175.

Jakob, E. M., S. D. Marshall, and G. W. Uetz. 1996. Estimating fitness: a comparison of body condition indices. Oikos 77:61.

Kocsis, E., B. L. Trus, C. J. Steer, M. E. Bisher, and A. C. Steven. 1991. Journal of Structural Biology 107:6-14.

Kohn N. R., and R. G. Jaeger. 2009. Male salamanders remember individuals based on chemical or visual cues. Behaviour 146:1485–1498.

Mazerolle, M. J. 2015. AICcmodavg: Model selection and multimodel inference based on (q)AIC(c). R package version 2.0-3. http://CRAN.R-project.org/package=AICcmodavg.

Møller, A. P. and J. P. Swaddle. 1997. Asymmetry, Developmental Stability, and Evolution. Oxford University Press, Inc., New York, USA.

Oksanen, J., F. G. Blanchet, R. Kindt, P. Legendre, P. R. Minchin, R. B. O’Hara, G. L. Simpson, P. Solymos, M. H. H. Stevens, and H. Wagner. 2013. vegan: community ecology package. R Package version 2.0-10. http://CRAN.R-project.org/package=vegan.

Petranka, J. W. 1998. Salamanders of the United States and Canada. Smithsonian Institution Press, Washington, D.C., USA.

Pfennig, D. W. and J. P. Collins. 1993. Kinship affects morphogenesis in cannibalistic salamanders. Nature 362: 836-838.

Pfennig, D. W., J. P. Collins, and R. E. Ziemba. 1999. A test of alternative hypotheses for kin recognition in cannibalistic tiger salamanders. Behavioral Ecology 10:436-443.

Pfennig, D. W., P. W. Sherman, and J. P. Collins. 1994. Kin recognition and cannibalism in polyphenic salamanders. Behavioral Ecology 5:225–232.

Pokhrel, L.R. 2009. Mapping the Dorsal Skin Pigmentation Patterns of Two Sympatric Populations of Ambystomatid Salamanders, Ambystoma opacum and A. maculatum from Northeast Tennessee. Thesis. Eastern Tennessee State University, Johnson City, Tennessee.

Pokhrel, L. R., I. Karsai, M. K. Hamed, and T. F. Laughlin. 2013. Dorsal pigmentation and sexual dimorphism in the marbled salamander (Ambystoma opacum). Ethology, Ecology, and Evolution 25:214-226.

Protas, M. E. and N. H. Patel. 2008. Evolution of coloration patterns. Annual Review of Cell and Developmental Biology 24:425-446.

Queller, D. C. and K. F. Goodnight. 1989. Estimating relatedness using genetic markers. Evolution 43:258-275.

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

Roulin, A. 2004. The evolution, maintenance and adaptive function of genetic colour polymorphism in birds. Biological Reviews 79:815-48.

Romano A., M. Mattoccia, S. Marta, S. Bogaerts, F. Pasmans, and V. Sbordoni. 2009. Distribution and morphological characterization of the endemic Italian salamanders Salamandrina perspicillata (Savi, 1821) and S. terdigitata (Bonnaterre, 1789) (Caudata: Salamandridae). Italian Journal of Zoology 76:422–432.

Rudh, A., B. Rogell, and J. Höglund. 2007. Non-gradual variation in colour morphs of the strawberry poison frog Dendrobates pumilio: genetic and geographical isolation suggest a role for selection in maintaining polymorphism. Molecular Ecology 16:4284-4294.

Secondi J., A. Johanet, O. Pays, F. Cazimajou, Z. Djalout, and C. Lemaire. 2010. Olfactory and visual species recognition in newts and their role in hybridization. Behaviour. 147:13–14.

Schulte-Hostedde, A. I., B. Zinner, J. S. Millar, and G. J. Hickling. 2005. Restitution of mass– size residuals: validating body condition indices. Ecology 86:155–163.

Sherman, E., K. Tock, and C. Clarke. 2009. Fluctuating asymmetry in Ichthyophonus sp. infected newts, Notophthalmus viridescens, from Vermont. Applied Herpetology 6:369-378.

Storfer, A., J. Cross, V. Rush, and J. Caruso. 1999. Adaptive coloration and gene flow as a constraint to local adaptation in streamside salamander, Ambystoma barbouri. Evolution 53:889.

Thaker M., C. R. Gabor, and J. N. Fries. 2006. Sensory cues for conspecific associations in aquatic San Marcos salamanders. Herpetologica 62:151–155.

Todd P. A., P. G. Sanderson, and L. M. Chou. 2001. Photographic technique for the study of coral morphometrics. Raffles Bulletin of Zoology. 49:191–196.

Todd, P. A., R. J. Ladle, R. A. Briers, and A. Brunton. 2005. Quantifying two-dimensional dichromatic patterns using a photographic technique: case study on the shore crab (Carcinus maenas L.). Ecological Research 20:497–501.

Todd B. D., and A. K. Davis. 2007. Sexual dichromatism in the marbled salamander, Ambystoma opacum. Canadian Journal of Zoology 85:1008–1013.

Twitty, V. C. 1961. Second-generation hybrids of the species of Taricha. Proceedings of the National Academy of Sciences 47:1461-1486.

Walls, S. C. and A. R. Blaustein. 1995. Larval marbled salamanders, Ambystoma opacum, eat their kin. Animal Behaviour 50:537-545.

Walls, S. C. and R. E. Roudebush. 1991. Reduced aggression toward siblings as evidence of kin recognition in cannibalistic salamanders. American Naturalist 138: 1027-1038.

Wakahara, M. 1997. Kin recognition among intact and blinded, mixed-sibling larvae of a cannibalistic salamander Hynobius retardatus. Zoological Science 14:893-899.

Wareing, K. 1997. Aspects of the mother-offspring bond in the red-backed salamander (Plethodon cinereus): maternal care and recognition. Thesis. Binghamton University, Vestal, New York.

Wright, A. N., and K. R. Zamudio. 2002. Color pattern asymmetry as a correlate of habitat disturbance in spotted salamanders (Ambystoma maculatum). Journal of Herpetology 36:129–133.

Figures

Stage 1 Stage 2 Stage 3

Figure 2.1. Example of a Four-toed Salamander, Hemidactylium scutatum image being prepared for scoring. Stage 1 shows the raw image. Stage 2 shows the image manipulated to remove the background and converted to grayscale. Stage 3 shows the image after straightening with a 10x65 grid for spot pattern scoring superimposed.

Figure 2.2. Encountered individual and site locations for the H. scutatum plotted in ArcGIS 10.2.1 using GPS coordinates. Inset A shows locations of encountered individuals at sites I-AN and I-AS at the Middle Fork State Fish and Wildlife Area, Vermilion County, IL from 2009 to 2012. Inset B depicts fourteen sampling locations in North America. Inset C shows a magnified view of seven sampling sites in northern Kentucky.

Figure 2.3. Comparison of mean phenotype scores for individuals from sites in central Kentucky (KY) vs. those collected from sites I-AN and I-AS in Illinois. Scores range from 0 to 1 with patternless individuals possessing a value of 0 and completely patterned (dark) individuals possessing a value of 1. Whiskers represent the lowest and maximum values of observations (excluding outliers), edges of the boxes indicate the 25% and 75% quartiles. T-test p- value<0.000. Sampling locations shown in Figure 2.2.

Figure 2.4. Comparison of mean symmetry scores for individuals from sites in central Kentucky (KY) vs. those collected from sites I-AN and I-AS in Illinois. Scores range from 0 to 1 with laterally symmetrical individuals possessing a score of 0 and laterally asymmetrical individuals possessing a score of 1. Whiskers represent the lowest and maximum values of observations (excluding outliers), edges of the boxes indicate the 25% and 75% quartiles. T-test p- value=0.000. Sampling locations shown in Figure 2.2.

Figure 2.5. Comparison of mean body condition scores for individuals from sites in central Kentucky (KY) vs. those collected from sites I-AN and I-AS in Illinois. Body condition was calculated using a residual index based on the log transformed variables of mass on total length (TL). Whiskers represent the lowest and maximum values of observations (excluding outliers), edges of the boxes indicate the 25% and 75% quartiles. T-test p-value=1.0. Sampling locations shown in Figure 2.2.

Figure 2.6. Relationship of phenotype scores and snout-vent length or SVL (mm) for H. scutatum collected from sites I-AN and I-AS in central Illinois. Scores range from 0 to 1 with patternless individuals possessing a value of 0 and completely patterned (dark) individuals possessing a value of 1. Sampling locations shown in Figure 2.2.

Figure 2.7. Relationship of phenotype scores and mass (mg) for H. scutatum collected from sites I-AN and I-AS in central Illinois. Scores range from 0 to 1 with patternless individuals possessing a value of 0 and completely patterned (dark) individuals possessing a value of 1. Sampling locations shown in Figure 2.2.

Figure 2.8. Relationship of phenotype symmetry scores and snout-vent length or SVL (mm) for H. scutatum collected from sites I-AN and I-AS in central Illinois. Scores range from 0 to 1 with laterally symmetrical individuals possessing a score of 0 and laterally asymmetrical individuals possessing a score of 1. Sampling locations shown in Figure 2.2.

Figure 2.9. Relationship of phenotype symmetry scores and mass (mg) for H. scutatum collected from sites I-AN and I-AS in central Illinois. Scores range from 0 to 1 with laterally symmetrical individuals possessing a score of 0 and laterally asymmetrical individuals possessing a score of 1. Sampling locations shown in Figure 2.2.

Figure 2.10. Relationship of phenotypic similarity scores and relatedness between pairs of 134 individual H. scutatum collected from sites I-AN and I-AS in central Illinois. Scores range from 0 to 1 with identically patterned pairs of individuals possessing a value of 0 and entirely dissimilarly patterned pairs of individuals possessing a value of 1. Relatedness calculated following Queller and Goodnight (1989) using eight microsatellite loci (see Chapter 1). Sampling locations shown in Figure 2.2. Using a subset of 93 individuals with complete data, the Mantel test p-value=0.85.

Figure 2.11. Relationship of phenotype scores and body condition for individual H. scutatum collected from sites I-AN and I-AS in central Illinois. Scores range from 0 to 1 with patternless individuals possessing a value of 0 and completely patterned (dark) individuals possessing a value of 1. Body condition was calculated using a residual index based on the log transformed variables of mass on total length (TL). Sampling locations shown in Figure 2.2.

Figure 2.12. Relationship of symmetry scores and body condition for individual H. scutatum collected from sites I-AN and I-AS in central Illinois. Scores range from 0 to 1 with patternless individuals possessing a value of 0 and completely patterned (dark) individuals possessing a value of 1. Body condition was calculated using a residual index based on the log transformed variables of mass on total length (TL). Sampling locations shown in Figure 2.2.

Tables

Table 2.1. Relationship of phenotype score and individual characteristics for 132 individual H. scutatum from the Middle Fork State Fish and Wildlife Area in Vermilion Co., Illinois. Samples sizes by year and population cluster are provided in Table 2.1. Site refers to the effect of population cluster an individual was sampled from, SVL indicates the impact of an individual’s snout-vent length, and mass is the effect of total body mass. Models were ranked using Akaike’s information criterion (AIC) adjusted for small sample sizes (AICc), number of parameters, the -2 Log Likelihood for a given model, the difference between the AICc value for a given model and the lowest AICc value overall (Δi), and Akaike weights (wi) calculated using the AICc values.

Model # of Parameters AICc -2 Log Likelihood Δi Δwi Site 1 -383.12 194.69 0 0.385919 SVL + Site 2 -381.83 195.14 1.2889 0.202589 Mass + Site 2 -382.64 195.54 0.4824 0.303211 SVL + Mass + Site 3 -380.58 195.62 2.5418 0.108281

Table 2.2. Relationship of symmetry score and individual characteristics for 132 individual H. scutatum from the Middle Fork State Fish and Wildlife Area in Vermilion Co., Illinois. Distribution of samples sizes by year and population cluster are provided in Table 2.1. Site refers to the effect of population cluster an individual was sampled from, SVL indicates the impact of an individual’s snout-vent length, and mass is the effect of total body mass. Models were ranked using Akaike’s information criterion (AIC) adjusted for small sample sizes (AICc), number of parameters, the -2 Log Likelihood for a given model, the difference between the AICc value for a given model and the lowest AICc value overall (Δi), and Akaike weights (wi) calculated using the AICc values.

Model # of AICc -2 Log Δi Δwi Parameters Likelihood Site 1 -419.6706 212.97 0 0.319418 SVL + Site 2 -418.6602 213.55 1.0104 0.192732 Mass + Site 2 -419.8996 214.17 -0.229 0.358167 SVL + Mass + Site 3 -417.8678 214.27 1.8028 0.129684

CHAPTER 3. ROLE OF KINSHIP AND INTRASPECIFIC AGGRESSION IN THE FOUR-TOED SALAMANDER Introduction

Kin recognition, the ability to distinguish between relatives and non-relatives, plays a role in the ecology and evolution of many species. The extent and significance of its influence

(through kin selection) is subject to intense debate (e.g. Abbot et al. 2011, Breed 2014, Nowak et al. 2010) and the occurrence and importance of this process varies among species. Two forces are thought to drive the evolution of kin recognition: kin selection and mate choice. Hamilton

(1964) proposed that kin selection, the indirect fitness gained by assisting close relatives, explains the evolution of kin recognition. In social taxa, or organisms with limited dispersal, neighboring individuals are expected to be more closely related than in solitary or contrast,

Bateson (1983) suggested that mate choice might drive the evolution of kin recognition if individuals choose mates based on relatedness in order to avoid inbreeding depression.

Kin recognition is inferred when actions of an individual towards conspecifics are modulated based on relatedness (i.e., kin discrimination; Barnard 1990, Barnard and Aldhous

1991, Byers and Bekoff 1986, and Waldman et al. 1988). A failure to observe kin discrimination may either indicate that no mechanism for kin recognition exists, or kin recognition is occurring but discrimination is not (Beecher 1991). It is important to realize that the occurrence of kin recognition may vary with developmental stages during an individual’s lifetime (Waldman 1991) and the context of intraspecific encounters may influence the occurrence of kin discrimination.

Thus, the lack of kin discrimination at certain life stages may indicate a lack of selection for kin recognition at that life stage (Armitage 1989).

Salamanders of the family Plethodontidae are particularly suited for studies of kin recognition. Many plethodontid species have low vagility and high desiccation risk compared to

69 taxa with larger overall body size, and are therefore characterized by restricted gene flow among populations (see review by Larson et al. 1984). Adults rely on cover objects and, in indirect developing species, populations are centered around breeding ponds. Together, these aspects of their biology increase the likelihood of encounters between related individuals and create conditions conducive to evolution of kin discrimination in intraspecific interactions.

Among plethodontids, territoriality and courtship/mate choice are two common post- metamorphic interactions observed between conspecifics. If salamanders in high-quality territories share resources with close relatives, they increase the fitness of kin and thereby their own inclusive fitness (Waldman 1991). While no studies to date have examined the role of kin recognition in intraspecific interactions between adult plethodontids, research in other taxa suggests that the influence of kin recognition may be complex. In Rainbow Trout, Oncorhynchus mykiss, kin remained “better neighbors” in terms of aggressive territorial behavior than non-kin even when territory quality decreased so long as the territories of neighboring individuals were of similar quality (Brown and Brown 1993a, 1993b). Evidence for kin structuring within populations has been found for a diverse array of taxa (i.e. Coltman et al. 2003, MacColl et al.

2000, Støen et al. 2005) and appears to be an important factor in fitness and reproductive success

(i.e. MacColl et al. 2000).

The role of demographic class in direct aggressive encounters between conspecifics in other plethodontid species is complex and it is difficult to predict what role it may play in H. scutatum. While differences in the level of aggression between the sexes, generally with males as the more aggressive sex, has been documented for many species, (i.e. Staub 1993, Wiltenmuth

1996, Wrobell et al. 1980), a number of studies have also found no difference in aggression

70 between the sexes for some plethodontid species (i.e. Jaeger et al. 1982, Ovaska 1993, Staub

1993).

The life history traits and conservation/management needs of the Four-toed Salamander,

Hemidactylium scutatum, make it an excellent candidate for this research. The Four-toed

Salamander is a widespread, indirect-developing species that occurs in 36 states, districts, and territories in the eastern United States and Canada (Petranka 1998). It is designated as either

‘imperiled’ or ‘critically imperiled’ in nine of these regions with status of ‘vulnerable’ in another

13 areas (NatureServe 2011). Despite being widespread, H. scutatum exhibits a patchy distribution due, in part, to its requirement for a breeding habitat featuring fishless seeps, springs, or marshes (Petranka 1998). Because of its small body size, susceptibility to desiccation, and specific habitat requirements, lifetime dispersal in H. scutatum should be limited. Consequently, the probability of encounters between highly related individuals within a population is increased, creating conditions conducive to the evolution of kin recognition and kin discrimination.

The indirect development of this species may also be causing a lack of viscosity within populations, i.e. inertia to disperse from natal locations. In direct developing species, sibling groups disperse from a single, central nest location and should, all else being equal, result in clusters of related individuals, i.e. “family groups”, across a landscape. In contrast, for indirect developing species, many females in a population use the same breeding pool. Consequently, all larvae metamorphose and disperse from more or less the same location causing all individuals to be distributed at random within the population. Therefore, the assumed clustering patterns within direct developing plethodontid species, such as P. cinereus, could cause increased selection pressure for kin recognition and discrimination (Gibbons et al. 2003, Liebgold and Cabe 2008,

Wareing 1997).

This study evaluated the potential role of kinship in intraspecific encounters between post-metamorphic H. scutatum by asking several questions. (1) Does kinship, especially via the mechanism of kin selection, influence how individuals distribute themselves within a population? If closely related H. scutatum are more tolerated than non-relatives, kin should be spatially aggregated within a population. Alternatively, if H. scutatum utilize spatial distribution as a mechanism for inbreeding avoidance, then the distances between kin in a population should be greater than expected by chance. (2) Does kinship mediate antagonistic encounters between conspecifics? If relatedness mediates aggression among females, then an individual should display fewer aggressive behaviors for shorter amounts of time toward the scent of a related female compared to the scent of an unrelated female. Alternatively, if differences exist in social behavior between demographic groups, i.e. if female salamanders are nonaggressive, then there should be no difference in the number and duration of aggressive behaviors displayed by a female individual toward the scent of a female conspecific compared to an unscented control. (3)

Does indirect development, as a result of reduced within population viscosity, lead to reduced selection for kinship mediated social behaviors than that found in direct developing species? If indirect development results in reduced selection for kin recognition and kin discrimination, then no role of kinship should be observed in the spatial distribution of individuals across a landscape or in aggressive behaviors displayed by an individual toward the scent of conspecifics.

To explore these questions, I evaluated genetic relatedness (as determined from microsatellite

DNA analysis) in the context of spatial distribution among individuals encountered in the wild. I also recorded the behavior of females in the laboratory when exposed to scent/chemosensory signals of other females and tested for an association between level of relatedness and extent of aggressive behavior. Chemosensory signals, such as pheromones, are an important mode of

72 communication and may be the primary mechanism of kin recognition in plethodontid salamanders and transmit a great deal of information (i.e. Gautier et al. 2006, Jaeger and Gergits

1979). The potential for chemical signals to indicate genetic similarity, i.e. kinship, has been suggested in several studies (Madison 1975, Wakahara 1997)

Methods

Spatial Distribution of Individuals within a Population

Samples and Collection Sites.—Sampling was conducted from March to November over four consecutive years (2009-2013) while H. scutatum was active in the Middle Fork State Fish and Wildlife Area in Vermilion County, Illinois (Figure 3.1). I conducted visual surveys for H. scutatum along haphazardly placed transects by looking under potential shelter objects. For each encountered individual, I recorded the GPS location in UTM coordinates using a Garmin GPS

12, and measured the body size (mm) as Snout-Vent-Length (SVL) and Total-Length (TL), and mass (g). For genetic analyses, I collected small tail clips (1-5 mm) from individuals weighing more than 0.18 g using sterilized clippers and tissues were stored in ethanol at -80 °C. The ventral spot pattern of each salamander was photographed for use in individual identification and to avoid repeated tissue samples from the same individual (Breitenbach 1982).

Molecular Methods and Genetic Analyses.—To infer relatedness, I determined genetic similarity of individuals using eight microsatellite markers (Chapter 1). Because pedigree data was unavailable, relatedness between each pair of individuals was determined by calculating relatedness (r) (Queller and Goodnight 1989). Relatedness is defined as the proportion of alleles shared between focal individuals relative to the proportion of alleles shared between these individuals and the entire population. While this measure of relatedness has the advantage of

73 reducing bias for small sample sizes, it also is influenced by the genetic variation within the reference population. Tsutsui and Suarez (2003) noted that the group used as the reference population can influence the estimate of r if the reference population is not representative of the genetic composition across the entire range. In cases where genetic variation in the reference population is decreased compared to the variation present across the entire range, especially when most individuals are related as a result of inbreeding, estimates of r may be reduced and suggestive of a low degree of kinship even between close relatives. To avoid this potential bias, I calculated relatedness using additional samples from the extended range of this species collected from other sites in Illinois and Kentucky (Chapter 1).

Statistical Analyses.—Samples consisted of two geographically distinct clusters distributed around two breeding pools separated by a distance of 1,224 m (Figure 3.1). While these clusters are not genetically distinct (Chapter 1), it is unlikely that inter-site interactions between individuals occur frequently. Previous work by Windmiller (2000) observed a maximum straight-line distance of only 201 m from a wetland source and suggested that H. scutatum dispersal is limited. Additionally, the presence of a stream separating the two geographic clusters (Figure 3.1) is expected to further restrict movement as low order streams act as a barrier to dispersal in the similarly sized plethodontid P. cinereus (Marsh et al. 2007). Thus I analyzed the data for the north population cluster, I-AN, and the south population cluster, I-AS, separately by year. I modeled the impact of relatedness, population cluster, and year as random effects, as a function of straight-line inter-individual distance using the LME4: LINEAR MIXED-

EFFECTS MODELS USING EIGEN AND S4 PACKAGE V. 1.1-7 (Bates et al. 2014) in R V3.1.1 (R Core

Team 2014). Because I was interested in the effect of relatedness on within population distribution of individuals, I modeled relatedness as a fixed effect and population cluster and

74 year as random effects. The fit of this full model was compared to a null model which omitted the relatedness variable by calculating AIC values corrected for small sample size (AICc,

Burnham and Anderson 2002) using AICCMODAVG: MODEL SELECTION AND MULTIMODEL

(Burnham and Anderson 2002). An Akaike weight is interpreted as the variability accounted for by the model, i.e. the best model among those evaluated will have the highest Δwi. To also investigate the potential association between relatedness and spatial distribution, I conducted

Mantel tests using the VEGAN: COMMUNITY ECOLOGY PACKAGE V2.0-10. (Oksanen et al. 2013) in R V3.1.1 and compared relatedness versus geographic distance for each possible dyad of individuals for each population cluster by year. Because of repeated comparisons created by the population cluster/year combinations, the Bonferroni corrected level of significance for these tests was α=0.00625.

Kin Discrimination in Antagonistic Encounters

Sampling and Maintenance of Salamanders.—Since H. scutatum is state-threatened in

Illinois, but relatively common in Kentucky, I collected individuals for experimental trials from

18-23 March 2011 from ten sites (K-A, K-B, K-C, K-E, K-F, K-H, K-J, K-L, K-M, K-N) in

Kentucky (Figure 3.1). The target sample size from a given site was two adult females and I processed these individuals as described above. Each individual was housed separately in a clear plastic container kept in a Percival environmental chamber model I-35 LL, maintained at 20-

24 °C on a 12/12 LD cycle. Dampened, unbleached paper toweling was used as a substrate and

75 crumpled unbleached paper toweling provided as a shelter. Provosoli media (Wyngaard and

Chinnappa 1982), a synthetic pond water substitute, was used to mist and dampen these individuals as needed. I fed individuals fruit flies and pinhead crickets ad lib such that all individuals were offered the same diet each time.

Experimental Setup.—Each experimental trial utilized an acting salamander whose behavior and movements I observed in response to the chemosensory cues provided by a stimulus salamander. Bedding was scented by the stimulus salamander for a full six days before being used in an experiment (Jaeger and Gergits 1979, Mathis 1990). The experimental chamber used for each trial consisted of a clear plastic dome, 8 cm x 8 cm, placed over a square plastic base. I set this arrangement inside a four-sided, white cardboard blind in order to eliminate external visual stimuli and to maximize the visibility of the salamander on the video footage. I conducted trials at 18-25 °C under a red light to minimize stress to the salamander (Dubois 1890,

Graber 1883, Pearse 1910). Each acting salamander was only used once per day and each bedding sample used only once before being discarded. For each trial, I covered the square plastic base with scented bedding, placed the acting salamander on it, covered this individual with an opaque shelter, and contained the whole setup within the transparent plastic dome. I allowed the salamander to acclimate for a period of 5 minutes in order to minimize the impact of handling stress. After the acclimation period, I removed the shelter and allowed the individual to move freely within the plastic dome for 30 minutes. I recorded the movements and behavior of the individual using a high-resolution digital video camera (WiLife Indoor Surveillance Camera,

Logitech, Fremont, CA, USA) that remotely recorded the video to a laptop computer (WiLife

Indoor Master System, Logitech). I repeated these trials so that each salamander was used as an actor against the scent of all other individuals (“conspecific-scented”), allowing every possible

76 dyad combination of actor-scent to be tested. For control conditions, each individual was used as an acting salamander in two additional trials using bedding scented only with Provosoli media

(“Provosoli-scented”) and its own bedding (“self-scented”) respectively.

Video Analysis. —I reviewed the video footage for each trial using SITANDWAIT Event

Recorder software (Szabo and Kis 2012) to score the behaviors and measure the activity of each individual blindly with respect to the relatedness between individuals. The following behaviors following Jaeger (1984) were recorded: (1) rate of nose taps (NT: pressing the tip of the nose to the substrate) over entire trial period, (2) duration of time spent posturing in all trunk raised

(ATR) with the trunk lifted entirely off of the substrate, (3) duration of time spent posturing in flattened position (FLAT) with the entire trunk in contact with the substrate, (4) duration of time spent posturing in front trunk raised (FTR) where the anterior portion of the trunk was raised off of the substrate while the posterior remained in contact with it, and (5) the occurrence of defecating (DEF: individuals defecated during the trial). The activity of the salamander was scored as follows: (1) active (ACTD: duration of walking across the substrate), (2) inactive

(INACTD: duration of not moving at all, starting after an individual had not changed position for

>=10 seconds), and (3) climbing (CLIMB: duration of time where individuals attempted to climb the side of the experimental chamber, beginning whenever a salamander had placed both front feet on the side and ending when at least one front foot had returned to the substrate.

Statistical Analysis.—I modeled the impact of relatedness and the straight-line distance between population clusters, first as a function the proportion of time spent in ATR, secondly as a function of the proportion of time spent walking, and finally as a function of the rate of nose taps displayed by the acting individual as before, modeling relatedness, the variable of interest, as a fixed effect and straight-line distance as a random effect.

As a secondary test of the role of kinship in mediating aggression, I conducted Mantel tests to examine the influence of relatedness on each of the three main response variables: 1) percent of the trial period spent in an aggressive posture (ATR), 2) percent of the trial period spent walking, and 3) on the amount of chemosensory behavior displayed (rate of nose taps). I used Mann-Whitney U tests in R V3.1.1 (R Core Team 2014) to compare the reactions of individuals toward conspecifics to their reactions toward the control conditions, Provosoli- or self-scented bedding, and because these response variables exhibited non-normal distributions.

Results

Spatial Analysis

I first examined if there was an association between spatial distances among the capture location of individuals and their relatedness. For this spatial analysis, 128 tissue samples were collected from the Middle Fork State Wildlife Area over four years, with each individual represented by a unique GPS location (Figure 3.1). Samples were compiled separately for I-AN and I-AS (Table 3.1). Mantel tests indicated no significant relationship in the relatedness between individuals and their spatial proximity (Figure 3.2). The AIC analysis supported these results, suggesting the null model, i.e. the model containing only the population cluster and year variables, as the most parsimonious model (Tables 3.2). These combined results indicated that fine scale distribution of individuals within a population is not influenced by kinship.

Behavioral Analysis

I investigated the relationship between the intraspecific behavior and relatedness within pairs of individuals. Seventeen female adult H. scutatum were collected from the 10 sampling sites in Kentucky (Table 3.3). The relatedness between these individuals is included in Figure

3.11. Behavioral responses to the scent of conspecifics varied greatly within and between acting salamanders. In some trials the actor spent the entire period walking while in an ATR posture but when paired with a different scent, spent the entire period still and in a FLAT posture (Figure 3.6 and Figure 3.7). Mantel tests assessing the influence of kinship on activity (Figure 3.3), posture

(Figure 3.4), and rate of nose taps (Figure 3.5) showed no significant relationship with the responses of all individuals pooled. This conclusion is supported by the AIC analysis, as the most parsimonious models for all response variables were the null models (Table 3.4). Within the Mantel tests for each response variable conducted on the individual basis, only one individual showed a significant relationship for the activity (Figure 3.6) and posture variables (Figure 3.7), but not the rate of nose tapping (Figure 3.8). It is likely that this is an artifact of the data and does not represent a true impact of kinship on intraspecific interactions. The results of the Mann-

Whitney U tests (Table 3.5) indicated that the salamanders also did not exhibit significantly different behaviors when their reactions to conspecific- vs. Provosoli-scented toweling or conspecific- vs. self-scented toweling were compared for either activity and posture (Figure 3.9) or nose tap rate (Figure 3.10).

Discussion

This study found no strong evidence of association between kinship and social interactions of post-metamorphic H. scutatum, either from the indirect measurement of fine-scale spatial distribution or direct observations of the behavior exhibited by adult females in response to conspecific cues. Adult female H. scutatum, furthermore, showed no difference in response during conspecific- vs. self-scented trials or conspecific- vs. Provosoli-scented toweling. The lack of association between kinship and evaluated characteristics in this study can be interpreted

79 with several potential scenarios: (a) differences in the social behavior among demographic groups, (b) influence of developmental mode and sex on juvenile dispersal, and (c) possibility that the study did not target conditions under which kin discrimination occurs. Each of these scenarios is discussed in more detail below.

Demographic Differences in Social Behavior

It is possible that the failure to observe kin discrimination in this study is not due to lack of such behavior (i.e., kin discrimination is occurring), but instead the result of our inability to differentiate among behaviors of different demographic groups, such as males versus females.

For one, it is difficult to sex H. scutatum in the field and adults were examined as a single group comprising of both males and females. In addition, tissue samples were only obtained from individuals of sufficient size. Consequently, the evaluation of spatial distribution and genetic relatedness was limited to a general analysis with all demographic classes pooled. Male-biased dispersal, inferred from a sex-based difference in genetic and spatial association, has been found in P. cinereus (Liebgold et al. 2011), but the potential for sex-biased dispersal has not yet been investigated in H. scutatum. It is thus possible that spatial aggregation of kin does occur in this species, but might be only observable in non-dispersing sexes and/or ontogenetic stages.

The results from the experimental trials also support this hypothesis that intraspecific behavior may vary among demographic groups in H. scutatum. Individuals in these trials exhibited no differences in the frequency or duration of aggressive behaviors displayed toward the scent of a conspecific vs. an unscented controls. Because female salamanders are frequently less aggressive than their male counterparts (i.e. Staub 1993, Wiltenmuth 1996, Wrobell et al.

1980), the restriction of these trials to the use of female individuals may have led to the lack of

80 increased aggression toward the scent of conspecifics. It is possible that, if females are the least or non-aggressive sex in H. scutatum, that kinship may mediate interactions involving male individuals.

Influence of Developmental Mode and Sex on Juvenile Dispersal

I failed to reject the hypothesis that developmental mode in H. scutatum has resulted in reduced selection for kin recognition and kin discrimination. Because of their inherent need for dispersal form and migration to a breeding pool, indirect developing species like H. scutatum may have greater dispersal abilities than direct developing species. Genetic analyses on this species have revealed higher degrees of gene flow between populations of H. scutatum than would be expected for similarly sized direct developing plethodontid species (Chapter 1). This increased vagility could result in a reduced rate of encounters between close relatives, reduced selection for kin discriminatory behaviors, and ultimately be responsible for the lack of influence of kinship on the within population distribution and antagonistic interactions of individuals observed in this study.

Lack of Kin Discrimination

The lack of kin discrimination in this study may not necessarily be an indicator of a lack of kin recognition. Kin recognition has been observed in H. scutatum, occurring between mothers and their nests (Carreno and Harris 1998) as well as at larval stages, although kin discrimination in the latter instance appears to be context dependent and only occurs in environments with a high predator density (Carreno et al. 1996, Harris et al. 2003). It is possible that kin recognition does play an important role in the interactions between post-metamorphic

81 individuals, but only under the conditions outside the scope of this experiment. While kin discrimination between post metamorphic individuals was not observed in this study, H. scutatum has been shown to exhibit some degree of territoriality (Grant 1955), but individuals are frequently found sharing cover objects (pers. obs.). This suggests that under certain conditions, possibly when kin are involved, H. scutatum may allow other individuals into their territories. The design of this study also precludes the possibility that kin recognition may be a facilitator of mate choice and the avoidance of inbreeding/outbreeding depression in this species.

Madison (1975) determined that female P. jordani preferentially associate with unfamiliar males during the mating season and that familiarity might act as a surrogate for kinship.

While Grant (1955) suggested the existence of territoriality in H. scutatum, his research did not note the sex of individuals used and so no information is available on the role of sex in intraspecific aggression for this species. The lack of detecting kin discriminatory behaviors during the experimental portion of this study may also reflect that females are non- or less aggressive than males, or that individuals in general are non- or less aggressive toward females.

Finally, a lack of observed kin discrimination can occur when levels of relatedness between individuals are too low to illicit kin-related responses. Hamilton (1964) argued that kin selection should be observed when the benefit from interaction multiplied by the relatedness between the individuals is greater than the cost of interaction. This inequality, Hamilton’s Rule, depends on the degree of relatedness between the individuals involved. This estimate of relatedness ranges from 0, where individuals share no alleles and are therefore “unrelated,” to 1, where individuals share all alleles and are “closely related.” While the mean relatedness for individuals in the behavioral experiments was close to 0 (Figure 3.11), there was much greater variation in relatedness for the spatial analysis and some individuals had relatively high estimates

82 of relatedness. Therefore, although the behavioral experiments may have been hampered by very low relatedness, results for the spatial analysis that included some relatively closely related individuals, still failed to indicate that kinship does influence intraspecific encounters between post-metamorphic individuals in H. scutatum.

Conclusion

In conclusion, this study neither detected a correlation between relatedness and spatial distribution of individuals within a wild population of H. scutatum, nor between relatedness and behavior towards conspecifics in adult females. However, I did not assess if intraspecific behavior differs between demographic groups or that developmental mode may influence selection for kin recognition and kin discrimination. Genetic analyses of H. scutatum populations revealed that this species might have some mechanism for maintaining within population variation (Chapter 1). While increased genetic variation should result in a decreased rate of encounters between close relatives and decreased selection for kinship mediated behaviors, demographic differences in behavior remains a possibility and may explain the patterns observed in this study. Further investigation into the role of developmental mode in the life history and evolution of this species may assist in understanding the patterns of behavior observed across the

Plethodontidae.

Literature Cited

Abbot, P., J. Abe, J. Alcock, S. Alizon, J. A. Alpedrinha, M. Andersson, J. B. Andre, M. van Baalen, F. Balloux, S. Balshine, N. Barton, L. W. Beukeboom, J. M. Biernaskie, T. Bilde, G. Borgia, M. Breed, S. Brown, R. Bshary, A. Buckling, N. T. Burley, M. N. Burton- Chellew, M. A. Cant, M. Chapuisat, E. L. Charnov, T. Clutton-Brock, A. Cockburn, B. J. Cole, N. Colegrave, L. Cosmides, I. D. Couzin, J. A. Coyne, S. Creel, B. Crespi, R. L. Curry, S. R. X. Dall, T. Day, J. L. Dickinson, L. A. Dugatkin, C. El Mouden, S. T. Emlen, J. Evans, R. Ferriere, J. Field, S. Foitzik, K. Foster, W. A. Foster, C. W. Fox, J. Gadau, S. Gandon, A. Gardner, M. G. Gardner, T. Getty, M. A. D. Goodisman, A. Grafen, R. Grosberg, C. M. Grozinger, P. H. Gouyon, D. Gwynne, P. H. Harvey, B. J. Hatchwell, J. Heinze, H. Helantera, K. R. Helms, K. Hill, N. Jiricny, R. A. Johnstone, A. Kacelnik, E. T. Kiers, H. Kokko, J. Komdeur, J. Korb, D. Kronauer, R. Kümmerli, L. Lehmann, T. A. Linksvayer, S. Lion, B. Lyon, J. A. R. Marshall, R. McElreath, Y. Michalakis, R. E. Michod, D. Mock, T. Monnin, R. Montgomerie, A. J. Moore, U. G. Mueller, R. Noë, S. Okasha, P. Pamilo, G. A. Parker, J. S. Pedersen, I. Pen, D. Pfennig, D. C. Queller, D. J. Rankin, S. E. Reece, H. K. Reeve, M. Reuter, G. Roberts, S. K. A. Robson, D. Roze, F. Rousset, O. Rueppell, J. L. Sachs, L. Santorelli, P. Schmid-Hempel, M. P. Schwarz, T. Scott-Phillips, J. Shellmann-Sherman, P. W. Sherman, D. M. Shuker, J. Smith, J. C. Spagna, B. Strassmann, A. V. Suarez, L. Sundström, M. Taborsky, P. Taylor, G. Thompson, J. Tooby, N. D. Tsutsui, K. Tsuji, S. Turillazzi, F. Úbeda, E. L. Vargo, B. Voelkl, T. Wenseleers, S. A. West, M. J. West-Eberhard, D. F. Westneat, D. C. Wiernasz, G. Wild, R. Wrangham, A. J. Young, D. W. Zeh, J. A. Zeh, and A. Zink. 2011. Inclusive fitness theory and eusociality. Nature 471:E1-E4.

Armitage, K. B. 1989. The function of kin discrimination. Ethology Ecology & Evolution 1:111- 121.

Barnard, C. J. 1990. Kin recognition: problems, prospects and the evolution of discrimination systems. Advances in the Study of Behavior 19:29-81.

Barnard, C. J. and P. Aldhous. 1991. Kinship, kin discrimination, and mate choice. Pp. 125-147. In P. G. Hepper (Ed.), Kin Recognition. Cambridge: Cambridge University Press.

Bates, D., M. Maechler, B. Bolker, and S. Walker. 2014. lme4: Linear mixed-effect models using Eigen and S4. R package version 1.1-7. http://CRAN.R-project.org/package=lmer4.

Bateson, P. G. 1983. Optimal outbreeding. Pp. 257-309. In P. G. Bateson (Ed.), Mate Choice. Cambridge: Cambridge University Press.

Beecher, M. D. 1991. Successes and failures of parent-offspring recognition in animals. Pp. 94- 124. In P. G. Hepper (Ed.), Kin Recognition. Cambridge: Cambridge University Press.

Breed, M. D. 2014. Kin and nestmate recognition: the influence of W. D. Hamilton on 50 years of research. Animal Behavior 92:271-279.

Breitenbach, G. L. 1982. The frequency of communal nesting and solitary brooding in the salamander, Hemidactylium scutatum. Journal of Herpetology 16:341-346.

Brown, G. E. and J. A. Brown. 1993a. Social dynamics in salmonid fishes: do kin make better neighbors? Animal Behaviour 45:863-871.

Brown, G. E. and J. A. Brown. 1993b. Do kin always make better neighbors? The effects of territory quality. Behavioral Ecology and Sociobiology 33:225-231.

Burnham, K. P. and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach. 2nd ed. Springer-Verlag, New York.

Byers, J. A. and M. Bekoff. 1986. What does “kin recognition” mean? Ethology 72:342-345.

Carreno, C. A. and R. N. Harris. 1998. Lack of nest defense behavior and attendance patterns in a joint nesting salamander Hemidactylium scutatum (Caudata: Plethodontidae). Copeia 1998:183-189.

Carreno, C. A., T. J. Vess, and R. N. Harris. 1996. An investigation of kin recognition abilities in larval four-toed salamanders, Hemidactylium scutatum (Caudata: Plethodontidae). Herpetologica 52:293-300.

Coltman, D. W., J. G. Pilkington, and J. M. Pemberton. 2003. Fine-scale genetic structure in a free-living ungulate population. Molecular Ecology 12:733-742.

Dubois, R. 1890. Sur la perception des radiations lumineuses par la peau, chez les Protées aveugles des grottes de la Carniole. Comptes Rendus de l’Académie des Sciences 110:358-361.

Gautier, P., K. Olgun, N. Uzum, and C. Miaud. 2006. Gregarious behaviour in a salamander: attraction to conspecific chemical cues in burrow choice. Behavioral Ecology and Sociobiology 59:836-841.

Graber, V. 1883. Fundamentalversuche über die Helligkeits- und Farbenempfindlichkeit augenloser und geblendeter Thiere. Sitzungsberichte der Akademie der Wissenschaften in Wien, Mathematisch-Naturwissenschafliche Klasse 87:201-236.

Grant, W. C. 1955. Territorialism in two species of salamanders. Science 121:137-138.

Gibbons, M. E., A. M. Ferguson, D. R. Lee, and R. G. Jaeger. 2003. Mother-offspring discrimination in the red-backed salamander may be context dependent. Herpetologica 59:322-333.

Hamilton, W. D. 1964. The genetical evolution of social behaviour. International Journal of Theoretical Biology 7:1-16.

Jaeger, R. G. 1984. Agonistic behavior of the red-backed salamander. Copeia 2:309-314.

Jaeger, R. G. and W. F. Gergits 1979. Intra- and interspecific communication in salamanders through chemical signals on the substrate. Animal Behaviour 27:150–156.

Jaeger, R. G., D. Kalvarsky, and N. Shimizu. 1982. Territorial behaviour of the red-backed salamander: expulsion of intruders. Animal Behaviour 30:490-496.

Larson, A., D. B. Wake, and K. P. Yanev. 1984. Measuring gene flow among populations having high levels of genetic fragmentation. Genetics 106:293-308.

Liebgold, E. B. and P. R. Cabe. 2008. Familiarity with adults, but not relatedness, affects the growth of juvenile red-backed salamanders (Plethodon cinereus). Behavioral Ecology and Sociobiology 63:277-284.

Liebgold, E. B., E. D. Brodie, and P. R. Cabe. 2011. Female philopatry and male-biased dispersal in a direct-developing salamander, Plethodon cinereus. Molecular Ecology 20:249-257.

MacColl, A. D. C., S. B. Piertney, R. Moss, and X. Lambin. 2000. Spatial arrangement of kin affects recruitment success in young male red grouse. Oikos 90:261-270.

Madison, D. M. 1975. Intraspecific odor preferences between salamanders of the same sex: dependence on season and proximity of residence. Canadian Journal of Zoology 53:1356- 1361.

Mathis, A. 1990. Territorial salamanders assess sexual and competitive information using chemical signals. Animal Behaviour 40:953–962.

Marsh, D. M., R. B. Page, T. J. Hanlon, H. Bareke, R. Corritone, N. Jetter, N. G. Beckman, K. Gardner, D. E. Seifert, and P. R. Cabe. 2007. Ecological and genetic evidence that low- order streams inhibit dispersal by red-backed salamanders (Plethodon cinereus). Canadian Journal of Zoology 85:319-327.

Mazerolle, M. J. 2015. AICcmodavg: Model selection and multimodel inference based on (q)AIC(c). R package version 2.0-3. http://CRAN.R-project.org/package=AICcmodavg.

Nowak, M. A., C. E. Tarnita, and E. O. Wilson. 2010. The evolution of eusociality. Nature 466:1057-1062.

Oksanen, J., F. G. Blanchet, R. Kindt, P. Legendre, P. R. Minchin, R. B. O’Hara, G. L. Simpson, P. Solymos, M. H. H. Stevens, and H. Wagner. 2013. Vegan: community ecology package. R Package version 2.0-10. http://CRAN.R-project.org/package=vegan.

Ovaska, K. 1993. Aggression and territoriality among sympatric western plethodontid salamanders. Canadian Journal of Zoolology 71:901-907.

Pearse, A. S. 1910. The reactions of amphibians to light. Proceedings of the American Academy of Arts and Sciences 45:161.

Petranka, J. W. 1998. Salamanders of the United States and Canada. Smithsonian Institution Press, Washington, D.C., USA.

Queller, D. C. and K. F. Goodnight. 1989. Estimating relatedness using genetic markers. Evolution 43:258-275.

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

Staub, N. L. 1993. Intraspecific agonistic behavior of the salamander Aneides flavipunctatus (Amphibia: Plethodontidae) with comparisons to other plethodontid species. Herpetologica 1993:271-281.

Støen, O., E. Bellemain, S. Sæbø, and J. E. Swenson. 2005. Kin-related spatial structure in brown bears Ursus arctos. Behavioral Ecology and Sociobiology 59:191-197.

Szabo, P. and J. Kis. 2012. Sit and Wait. Available http://sitwait.sourceforge.net/. Accessed: June 23, 2014. (Archived by WebCite® at http://www.webcitation.org/6QYI5nKaA).

Tsutsui, N. D. and A. V. Suarez. 2003. The colony structure and population biology of invasive ants. Conservation Biology 17:48-58.

Waldman, B. 1991. Kin recognition in amphibians. Pp. 162-219. In P. G. Hepper (Ed.), Kin Recognition. Cambridge: Cambridge University Press.

Waldman, B., P. C. Frumhoff, and P. W. Sherman. 1988. Problems of kin recognition. Trends in Ecology and Evolution 3:8-13.

Wakahara, M. 1997. Kin recognition among intact and blinded, mixed-sibling larvae of a cannibalistic salamander Hynobius retardatus. Zoological Science 14:893-899.

Wareing, K. 1997. Aspects of the mother-offspring bond in the red-backed salamander (Plethodon cinereus): maternal care and recognition. Unpublished Thesis. Binghamton University, Vestal, New York.

Wiltenmuth, E. B. 1996. Agonistic and sensory behaviour of the salamander Ensantina eschscholtzii during assymetrical contests. Animal Behaviour 52:841-850.

Windmiller, B. 2000. Study of ecology and population response to construction in upland habitat by vernal pool amphibians. Report for the Massachusetts Natural Heritage Program, Sudbury, MA.

Wrobell, D. J., W. F. Gergits, and R. G. Jaeger. 1980. An experimental study of interference competition among terrestrial salamanders. Ecology 61:1034.

Wyngaard, G. A. and C. C. Chinnappa. 1982. General biology and cytology of cyclopoids. Pp 495-533. In R. W. Harrison and R. C. Cowden (Eds.), Developmental Biology of Freshwater Invertebrates. A. R. Liss, New York, New York.

Figures

Figure 3.1. Sampling locations representing 128 individual H. scutatum tissue samples collected from the Middle Fork State Fish and Wildlife Area (MFSFWA), Vermilion County, IL from 2009 through 2012. GPS coordinates of each individual collected were plotted in ArcGIS 10.2.1. Distribution of samples sizes by year and population cluster are provided in Table 3.1. Upper left inset shows location of sampling sites in Illinois and Kentucky in relation to North America. Upper right inset shows the location of 9 locations from which individuals were sampled in northern Kentucky. Sample sizes for sampling sites in Kentucky are provided in Table 3.3.

Figure 3.2. Relationship of relatedness and straight-line geographic distance (m) between pairs of 128 individual H. scutatum. Relatedness calculated following Queller and Goodnight (1989) using eight microsatellite loci (see Chapter 1). Data is plotted by year and divided between the I- AN and I-AS population clusters (see Figure 3.1). Results of Mantel tests comparing relatedness and geographic distance are included as p-values in the corresponding panels. Bonferroni corrected p-value required for significance is p<0.00625. Distribution of samples sizes by year and population cluster are provided in Table 3.1.

Figure 3.3. Relationship of relatedness and activity (i.e. proportion of time spent walking) during experimental trials between pairs of 17 H. scutatum. Relatedness calculated following Queller and Goodnight (1989) using eight microsatellite loci (see Chapter 1). Mantel test p-value =0.66.

Figure 3.4. Relationship of relatedness and posture [i.e. proportion of time spent in the all trunk raised (ATR) posture (Jaeger 1984)] during experimental trials between pairs of 17 H. scutatum. Relatedness calculated following Queller and Goodnight (1989) using eight microsatellite loci (see Chapter 1). Mantel test p-value=0.606.

Figure 3.5. Relationship of relatedness and rate of nose tapping [i.e. mean nose taps/second (Jaeger 1984)] during experimental trials between pairs of 17 H. scutatum. Relatedness calculated following Queller and Goodnight (1989) using eight microsatellite loci (see Chapter 1). Mantel test p-value =0.606.

Figure 3.6. Individual responses of 17 individual H. scutatum in terms of proportion of time spent walking, and their relatedness to conspecifics during experimental trials. Relatedness calculated following Queller and Goodnight (1989) using eight microsatellite loci (see Chapter 1). Codes for individuals are listed at the top of each panel. P-values for Mantel tests of individual responses and individual regression equations and R2 values are included in each panel. Bonferroni corrected p-value required for significance is p<0.00294.

Figure 3.7. Individual responses of 17 individual H. scutatum in terms of proportion of time spent in the all trunk raised (ATR) posture (Jaeger 1984), and their relatedness to conspecifics during experimental trials. Relatedness calculated following Queller and Goodnight (1989) using eight microsatellite loci (see Chapter 1). Codes for individuals are listed at the top of each panel. P-values for Mantel tests of individual responses and individual regression equations and R2 values are included in each panel. Bonferroni corrected p-value required for significance is p<0.00294.

Figure 3.8. Individual responses of 17 individual H. scutatum in terms of mean rate of nose taps (Jaeger 1984), measured in nose taps/second, and their relatedness to conspecifics during experimental trials. Relatedness calculated following Queller and Goodnight (1989) using eight microsatellite loci (see Chapter 1). Codes for individuals are listed at the top of each panel. P- values for Mantel tests of individual responses and individual regression equations and R2 values are included in each panel. Bonferroni corrected p-value required for significance is p<0.00294.

Figure 3.9. Impact of stimulus type on posture [i.e. proportion of time (mean ± SE) spent in the all trunk raised (ATR) posture (Jaeger 1984)] and activity [i.e. proportion of time (mean ± SE) spent walking] during experimental trials using 17 H. scutatum.

Figure 3.10. Impact of stimulus type on rate of nose tapping (mean ± SE) [i.e. nose taps/second (Jaeger 1984)] during experimental trials using 17 H. scutatum.

Figure 3.11. Distribution of relatedness values for 17 individual H. scutatum used in the experimental analysis and the 128 individual H. scutatum used in the spatial analysis. Relatedness was calculated following Queller and Goodnight 1989 and using eight microsatellite loci (see Chapter 1).

Tables

Table 3.1: List of Hemidactylium scutatum collected from population clusters I-AN and I-AS within the Middle Fork State Fish and Wildlife Area by year. *Drought conditions prevented the collection of samples at cluster I-AS during 2012.

Year I-AN I-AS Total

2009 29 15 44

2010 18 12 30

2011 35 10 45

2012* 9 0 9

Total 91 37 128

Table 3.2. Relationship of relatedness and straight-line geographic distance (m) between pairs of 128 individual H. scutatum from the Middle Fork State Fish and Wildlife Area in Vermilion Co., Illinois. Relatedness calculated following Queller and Goodnight (1989) using eight microsatellite loci (see Chapter 1). Distribution of samples sizes by year and population cluster are provided in Table 3.1. Null model includes sampling year and population cluster as random effects. Relatedness model includes these and the fixed effect of relatedness. Models were ranked using Akaike’s information criterion (AIC) adjusted for small sample sizes (AICc), number of parameters, the -2 Log Likelihood for a given model, the difference between the AICc value for a given model and the lowest AICc value overall (Δi), and Akaike weights (wi) calculated using the AICc values.

Model # of Parameters AICc -2 Log Likelihood Δi Δwi Year + Site 2 17295.03 -8643.50 0.00 0.73 Relatedness + Year + Site 3 17297.03 -8643.49 2.00 0.27

100

Table 3.3. Summary of 17 individual H. scutatum collected during visual encounter surveys across 10 sites in Kentucky. Listed are: Code = site acronym, N = number of samples, County = site county, UTM Coordinates = coordinates of mean centroid calculated from all sampled individual locations. Geographic location of sites is depicted in Figures 3.1.

Code N County UTM Coordinates K-A 1 Powell 264610 4187946 K-B 2 Menifee 273080 4198054 K-C 2 Laurel 745448 4101073 K-E 2 Powell 265866 4188298 K-F 2 Powell 264949 4187610 K-H 2 Powell 264672 4187874 K-J 2 Powell 263948 4189143 K-L 1 Wolfe 272711 4183552 K-M 1 Wolfe 272752 4187911 K-N 2 Menifee 264754 4191931

101

Table 3.4. Relationship of the responses of 17 adult H. scutatum to chemosensory cues provided by conspecifics during experimental trials and the relatedness between individuals. The response variables investigated include the proportion of time in an aggressive, “all trunk raised” posture (% time ATR), proportion of time an individual was active (% time walking), and the rate of nose tapping (nose taps/second). Relatedness calculated following Queller and Goodnight (1989) using 8 microsatellite loci (see Chapter 1). Null models include straight-line distance between the individual sampling sites as a random effect. Relatedness models include this and the fixed effect of relatedness. Models were ranked using Akaike’s information criterion (AIC) adjusted for small sample sizes (AICc), number of parameters, the -2 Log Likelihood for a given model, the difference between the AICc value for a given model and the lowest AICc value overall (Δi), and Akaike weights (wi) calculated using the AICc values.

# of -2 Log Response Model Parameters AICc Likelihood Δi Δwi % Time ATR Distance 1 128.08 -61.00 0.00 0.49 Relatedness + 127.99 -59.93 Distance 2 -0.09 0.51 % Time 89.93 -41.92 Walking Distance 1 0.53 0.43 Relatedness + 89.40 -40.63 Distance 2 0.00 0.57 Nose -2283.96 1145.02 Taps/Second Distance 1 0.00 0.59 Relatedness + -2283.20 1145.67 Distance 2 0.75 0.41

Table 3.5. Results of Mann-Whitney U Tests comparing the behavior of 17 individual H. scutatum in response to bedding scented with conspecific chemosensory cues vs. self-scented bedding or bedding dampened only with Provosoli media respectively. Behaviors measured included the proportion of time an individual spent in an aggressive, “all trunk raised” posture (% time ATR), proportion of time an individual was active (% time walking), and the rate of nose tapping (nose taps/second). Bonferroni corrected p-value required for significance is p<0.008.

Comparison % Time ATR % Time Walking Nose Taps/Second Conspecific vs. Self 0.585 0.185 0.773 Conspecific vs. Provosoli 0.593 0.059 0.305

102

APPENDICES

Appendix A. Results of visual encounter surveys for H. scutatum for central Illinois sites I-AN and I-AS by year. Locations of individual encounters are shown in Figure 1.1. Total = total number of individuals sampled in both sites.

Year I-AN I-AS Total 2009 33 14 47 2010 27 17 44 2011 34 14 48 2012 17 1 18 Total 111 46 157

103

Appendix B. Overall allelic diversity in terms of frequency of each allele at each locus for eight microsatellite loci in H. scutatum based on 279 sampled individuals from 14 sites in Illinois and Kentucky (see Figure 1.1). Sample sizes for each site are included in Table 1.1. Acronyms and size ranges for each locus are provided in Table 1.1.

104

Appendix C. Expected heterozygosity of H. scutatum at the central sampling Illinois sites (I-AN and I-AS) listed by year. Data are based on eight microsatellite loci. Sample sizes for each site by year are provided in Table 1.1.

Year I-AN I-AS 2009 0.65 0.71 2010 0.61 0.61 2011 0.66 0.62 2012 0.62 N/A

105

Appendix D. Pairwise population FST values for eleven sampling sites for H. scutatum in Kentucky. Data are based on eight microsatellite loci. Acronyms and sample sizes for each site are provided in Table 1.1. Geographic locations of sites are shown in Figure 1.1. FST values are below diagonal. P-values are above diagonal. To reduce the chance of Type I error, the Bonferroni corrected p-value needed for significance is p<0.0009. Significant p-values indicated in bold. K-A K-B K-C K-D K-E K-F K-H K-J K-L K-M K-N K-A 0.5398 0.3503 0.0063 0.8047 0.2361 0.7520 0.1368 0.0011 0.0056 0.6228 K-B 0.0277 0.8007 0.0288 0.9806 0.9162 0.6838 0.8708 0.1666 0.1561 0.4076 K-C 0.0489 0.1179 0.0487 0.8595 0.6998 0.0954 0.8164 0.0759 0.2564 0.3916 K-D 0.0366 0.0764 0.1060 0.0033 0.0097 0.0010 0.0004 0.0000 0.0000 0.0101 K-E 0.0088 0.0118 0.0597 0.0644 0.6500 0.6500 0.5452 0.0036 0.0493 0.6873 K-F 0.0399 0.0265 0.0248 0.1042 0.0483 0.1966 0.9312 0.1756 0.1870 0.5301 K-H 0.0028 0.0314 0.0942 0.0899 0.0229 0.0407 0.1184 0.5958 0.2214 0.2848 K-J 0.0328 0.0190 0.0231 0.0935 0.0335 -0.0086 0.0400 0.1112 0.1563 0.3480 K-L 0.0416 0.0413 0.0982 0.1050 0.0558 0.0405 0.0058 0.0291 0.0852 0.0009 K-M 0.0565 0.0773 0.0943 0.1352 0.0680 0.0540 0.0302 0.0426 0.0287 0.0170 K-N 0.0079 0.0403 0.0448 0.0432 0.0168 0.0219 0.0223 0.0250 0.0491 0.0553

106

Appendix E. Pairwise population p-values for an RxC analysis (multilocus contingency tests for heterogeneity of allele frequencies among pairs of samples) above diagonal and standard error values below diagonal for 279 individual H. scutatum from 14 sampling sites (see Figure 1.1). Data are based on eight microsatellite loci. Acronyms and sample sizes for each site are provided in Table 1.1. Significant values indicate a pair of populations should be considered distinct from one another. To reduce the chance of Type I error, the Bonferroni corrected p-value needed for significant is p<0.0003.

I-AN I-AS I-B K-A K-B K-C K-D K-E K-F K-H K-J K-L K-M K-N I-AN 0.11 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 I-AS 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 I-B 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 K-A 0.00 0.00 0.00 0.40 0.20 0.00 0.20 0.43 0.98 0.07 0.00 0.03 0.27 K-B 0.00 0.00 0.00 0.01 0.02 0.00 0.50 0.41 0.17 0.24 0.00 0.01 0.05 K-C 0.00 0.00 0.00 0.01 0.00 0.00 0.17 0.53 0.03 0.60 0.00 0.02 0.10 K-D 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 K-E 0.00 0.00 0.00 0.01 0.01 0.01 0.00 0.18 0.29 0.04 0.00 0.00 0.03 K-F 0.00 0.00 0.00 0.02 0.02 0.02 0.00 0.01 0.20 0.84 0.00 0.10 0.34 K-H 0.00 0.00 0.00 0.00 0.01 0.00 0.00 0.01 0.01 0.03 0.50 0.25 0.22 K-J 0.00 0.00 0.00 0.01 0.01 0.02 0.00 0.00 0.01 0.00 0.00 0.05 0.08 K-L 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.02 0.00 0.01 0.00 K-M 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.01 0.01 0.00 0.00 K-N 0.00 0.00 0.00 0.01 0.01 0.01 0.00 0.00 0.01 0.01 0.01 0.00 0.00

107

Appendix F. Straight-line distances (km) between H. scutatum sampling sites in Illinois and Kentucky (see Figure 1.1).

I-AN I-AS I-B K-A K-B K-C K-D K-E K-F K-H K-J K-L K-M I-AS 1 I-B 316 316 K-A 447 447 759 K-B 448 447 758 13 K-C 473 472 789 101 114 K-D 411 410 726 143 155 79 K-E 448 448 760 1 12 102 144 K-F 448 448 760 0 13 100 143 1 K-H 448 447 759 0 13 101 143 1 0 K-J 446 446 758 1 13 101 143 2 2 1 K-L 457 456 768 9 15 102 148 8 9 9 10 K-M 454 453 765 8 10 105 150 7 8 8 9 4 K-N 445 445 757 4 10 104 145 4 4 4 3 12 9

108