Genetic, physiological, and ecological consequences of sexual and kleptogenetic reproduction in salamanders
DISSERTATION
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University
By
Robert Daniel Denton
Graduate Program in Evolution, Ecology and Organismal Biology
The Ohio State University
2017
Dissertation Committee:
H. Lisle Gibbs, Advisor
Bryan Carstens
William Peterman
Copyrighted by
Robert Daniel Denton
2017
Abstract
Every year, there is at least one widespread news story documenting a “virgin birth” in a variety of animals as diverse as snakes and sharks. These events capture our attention because they represent departures from an assumed necessity of vertebrate life: having sex. Yet, vertebrates do not always reproduce via sex, and biologists have long studied the evolutionary costs and benefits of this type of reproduction. One of the main costs of sex are males, who cannot directly generate offspring and use up resources from reproductive females that cannot be put towards additional offspring. Eastern North
America is home to one of the strangest vertebrates that lack males and appear to be sexual and asexual at the same time: an all-female group of salamanders that appear to
“steal” sperm from males of other species. These all-female salamanders can potentially retain the advantage of gaining new genetic diversity from other species without males of their own. However, the extent and flexibility of this mating systems is still not understood, and the factors that promote the coexistence of all-female salamander lineages and the sexual species from which they use reproductive material are mysterious. I have investigated three primary questions concerning these unusual animals. First, how do we identify an all-female salamander in Ohio? Because of their cryptic morphology compared to similar sexual species, all-female Ambystoma salamanders are only reliably identified by sequencing their mitochondrial DNA, which is independently transferred maternally. However, mitochondrial sequences that closely
ii resembled those of all-female salamanders were previously found in salamander individuals across Ohio that were identified morphologically as either the smallmouthed or streamside salamander. I gathered microsatellite data from these potentially misidentified animals and evaluated three hypotheses for why the mitochondrial data does not match the nuclear DNA or morphological species identity. The best supported hypothesis was one of mitochondrial introgression, where the mitochondrial haplotypes of one species (streamside salamanders) were introgressed into populations of another species (smallmouth salamanders). This chapter describes the evolutionary basis for the discordance between mitochondrial and nuclear DNA markers and provides necessary diagnostic information to correctly identify all-female salamander lineages in Ohio.
Second, can differences in dispersal between breeding environments dictate the coexistence between all-female salamanders and other salamander species, preventing mutual extinction? I inferred salamander movements across a fragmented agricultural landscape in Ohio with genetic data and treadmill endurance trials, and I found that sexual species traveled significantly greater distances between breeding sites and fatigue much more slowly than unisexuals, contrary to a hypothesis that would explain the ecological coexistence between these groups. Third, what environmental factors determine the coexistence of all-female salamanders and their sexual relatives at broad scales? I gathered rangewide occurrence data for blue-spotted, Jefferson’s salamanders, and all-female salamanders that vary by the number of nuclear genomes possessed from either sexual species. By comparing the niche overlap from ecological niche models representing all-female salamander groups and each sexual species, I found that neither the total number of genomes or the composition of all-female salamander genomes iii explained the amount of niche overlap. Instead, niche overlap was significantly greater than expected for all comparisons between sexual species and the all-female lineage. I additionally used joint species modeling techniques to suggest that species interactions, in this case the all-female salamanders’ reproductive requirement for sperm, are the most powerful predictor of ecological niche in all-female salamanders and limit their ability to differentiate from their sexual relatives. Taken together, these chapters have expanded our understanding of how all-female salamanders have evolved to navigate the tradeoffs of intermediacy between sexual and asexual reproduction while providing new avenues of research for the future, including physiological limitations in members of the all- female lineage and molecular discordance between their separate mitochondrial and nuclear genomes.
iv
Dedication
For my Grandfather and Mother, my most relentless supporters. You told me to chase a
dream while generously shielding me from the consequences of doing so.
For Ashley and Wallace, who both believe I can do anything. But without them, anything
feels like nothing.
v
Acknowledgments
Scientific work cannot happen in isolation, and I have been fortunate to benefit from a truly outstanding collection of colleagues and friends who have devoted their time, perspective, and effort to helping me succeed.
Lisle Gibbs brought me to his lab and gave me the chance to become the scientist
I wanted to be. His carefully crafted advice and commitment to professionalism has influenced me greatly, and I will likely repeat his words to myself and others many times during the course of my career.
I thank the members of my dissertation and defense committees, Bryan Carstens,
Steve Matthews, Bill Peterman, and Joe Williams. Whether when initially forming my thoughts for PhD research or looking back on these projects after completion, my committee members have struck a perfect balance of providing both challenge and support.
I am grateful for my cohort of colleagues in the Department of EEOB. Some of these individuals have provided relief with empathetic ears, some have given me perspective across discipline and culture, and some have inspired me with their own incredible work. Some of these individuals do all three on a daily basis. In alphabetical order, I thank Mike Broe, Paul Blischak, Tony Fries, Jenn Hellman, Laura Kenyon, Isaac vi Ligocki, Erin Lindstedt, Jason Macrander, Eric McCluskey, Tara Pelletier, David Salazar,
Jordan Satler, Sarah Smiley-Walters, Mike Sovic, Ben Titus, and Jamin Wieringa. I especially thank Matt Holding, my closest collaborator and steadfast companion, who has responded to years of sarcasm with only laughter and unrelenting assistance. Finally, I am grateful for all of the undergraduates who have worked with me in the lab and field.
They have likely never realized the extent that their enthusiasm, dedication, and spirit has affirmed my desire to help others do exciting science. I especially thank Collin Ries,
Meghan Parsley, Paul Hudson, Mónica Saccucci, Abby Pomento, Katherine Costello, and
Blair Perry. I also acknowledge Bobby Arnold for providing quiet support for many years.
Other individuals have played crucial roles in generating the following thesis chapters and more: Jose Diaz and Josh Dyer. Jose’s assistance in the lab has provided me additional freedom that is rare for a graduate student, and his personality is the core of what makes working in our lab special. Josh’s dedication to amphibian conservation has made my fieldwork in Crawford County possible, and his talent as an educator has been a yearly source of inspiration. Finally, I truly appreciate the work by Chanelle Kinney,
Corey Ross, and Sue Meier for making the most offending scheduling, administrative, or financial issue disappear without a single complaint.
I am especially grateful towards Katy Greenwald, who told me about weird salamanders over a beer in Providence, Rhode Island. Without having that conversation, the last seven years would have looked much different and likely not been as fun.
My work has been generously supported by the many landowners of Crawford
County and their desire to learn more about the animals with which they share their vii fertile land. I have received financial support from the Ohio Biodiversity Conservation
Partnership between Ohio State University and the Ohio Division of Wildlife, the
National Science Foundation, the Ohio State University Graduate School, the American
Society of Ichthyologists and Herpetologists, those who participated in the SciFund
Challenge, the Society for the Study of Amphibians and Reptiles, and the Ohio State
Council of Graduate Students.
viii
Vita
2005...... Eastbrook High School, Marion IN
2005-2009 ...... B.S. Biology, Ball State University
2009-2011 ...... M.S. Biological Sciences
...... Eastern Kentucky University
2011-2012 ...... University Fellow, The Ohio State University
2012-2015 ...... Graduate Teaching Assistant, Department of
Evolution, Ecology, and Organismal Biology,
The Ohio State University
2012-2016 ...... Graduate Research Assistant, Department of
Evolution, Ecology, and Organismal Biology,
The Ohio State University
2016-2017 ...... Presidential Fellow, Department of Evolution,
Ecology, and Organismal Biology, The Ohio
State University
ix
Publications
McElroy K, RD Denton, J Sharbrough, L Bankers, M Neiman, and HL Gibbs. 2017
Genome balance in a triploid trihybrid vertebrate. Genome Biology and Evolution.
9:968-980
Denton RD, KR Greenwald, and HL Gibbs. 2017. Locomotor endurance predicts
differences in realized dispersal between sympatric sexual and unisexual salamanders.
Functional Ecology. 31:915-926.
Hudson P, RD Denton, ML Holding, and HL Gibbs. 2016. Repeatability of Locomotor
Endurance in the Small-mouthed Salamander, Ambystoma texanum. Herpetological
Review. 47(4):583-586.
Greenwald KR, RD Denton, and HL Gibbs. 2016. Niche partitioning among sexual and
unisexual Ambystoma salamanders. Ecosphere. 7:1-17.
Pomento AM, BW Perry, RD Denton, HL Gibbs, and ML Holding. 2016. No safety in
the trees: Evidence of local and species-level adaptation of the Eastern gray squirrel
(Sciurus carolinensis) to resisting the venom of sympatric rattlesnakes. Toxicon.
118:149-155.
x Saccucci MJ, RD Denton, ML Holding, and HL Gibbs. 2016. Polyploid unisexual
salamanders have higher tissue regeneration rates than diploid sexual relatives.
Journal of Zoology. 300:77-81.
Denton, RD. 2016. Response to the comment on ‘polyploid unisexual salamanders have
higher tissue regeneration rates than diploid sexual relatives’. Journal of Zoology.
300:239-240.
Gibbs HL and RD Denton. 2016. Cryptic Sex? Estimates of genome replacement in
asexual mole salamanders (Ambystoma sp.). Molecular Ecology. 30(12): 2805-2815.
Holding ML, EH Kern, RD Denton, and HL Gibbs. 2016. Fixed prey cue preference
among Dusky Pigmy Rattlesnakes (Sistrurus miliarius barbouri) raised on different
long-term diets. Evolutionary Ecology. 30: 1-7.
Denton RD, HL Gibbs, and TC Glenn. 2015. Development of 31 new microsatellite loci
for two mole salamanders (Ambystoma laterale and A. jeffersonianum). Conservation
Genetics Resources. 7(1):167-170.
Holdingǂ ML, RD Dentonǂ, AE Kulesza, and JS Ridgway. 2014. Confronting scientific
misconceptions by fostering a classroom of scientists in the introductory biology lab.
American Biology Teacher. 76:518-523.
Denton RD, LJ Kenyon, KR Greenwald, and HL Gibbs. 2014. Evolutionary basis of
mitonuclear discordance between sister species of mole salamanders (Ambystoma
sp.). Molecular Ecology. 23:2811-2824.
Denton RD and SC Richter. 2013. Relationship between amphibian community and
habitat similarity in natural and constructed ridge-top wetlands with implications for
wetland construction and management. Journal of Wildlife Management. 77:886-896. xi
Nunziata SO, SC Richter, RD Denton, JM Yeiser, DE Wells, KL Jones, C Hagen, and SL
Lance. 2012. Fourteen novel microsatellite markers for the gopher frog, Lithobates
capito (Amphibia: Ranidae). Conservation Genetics Resources. 4:201-203.
Denton RD and SC Richter. 2012 A quantitative comparison of two common larval
amphibian sampling techniques for wetlands. Herpetological Review. 43:44-47.
Denton RD and MJ Bernot. 2011. Examining the effects of multiple agricultural
chemicals on amphibian development. Proceedings of the Indiana Academy of
Science. 120:39-44.
Fields of Study
Major Field: Evolution, Ecology and Organismal Biology
Minor Field: College and University Teaching
xii
Table of Contents
Abstract ...... ii
Dedication ...... v
Acknowledgments...... vi
Vita ...... ix
Publications ...... x
Fields of Study ...... xii
Table of Contents ...... xiii
List of Tables ...... xvii
List of Figures ...... xix
Chapter 1: General Introduction ...... 1
Chapter 2: Evolutionary basis of mitonuclear discordance between sister species of mole salamanders (Ambystoma sp.) ...... 9
Abstract ...... 9
Introduction ...... 10
Methods ...... 16
Samples and Genotyping ...... 16
xiii Population and Phylogenetic Analyses...... 17
Morphological and Environmental Analyses ...... 23
Results ...... 24
mtDNA and nuDNA Genotyping ...... 24
Bayesian Clustering ...... 25
Estimates of Population Size and Gene Flow ...... 26
Phylogenetic Analyses ...... 28
Morphological and Environmental Analyses ...... 29
Discussion ...... 30
Value of multiple methods of analysis ...... 31
Mitonuclear discordance in Ambystoma ...... 33
Mechanisms of mitochondrial introgression ...... 35
Conclusions ...... 38
Chapter 3: Locomotor endurance predicts differences in realized dispersal between sympatric sexual and unisexual salamanders ...... 40
Abstract ...... 40
Introduction ...... 41
Materials and Methods ...... 46
Locomotor Endurance ...... 46
xiv Collection of Genetic Data ...... 47
Genetic Analyses ...... 48
Landscape Analyses ...... 50
Results ...... 51
Locomotor Endurance ...... 51
Genetic Analyses of Dispersal ...... 54
Landscape Analyses ...... 59
Discussion ...... 60
Causes of dispersal differences...... 61
Dispersal in Ambystoma salamanders ...... 64
Coexistence of sexuals and asexuals ...... 66
Chapter 4: Sperm dependence limits niche divergence in unisexual salamanders of varying ploidy and genome composition ...... 68
Abstract ...... 68
Introduction ...... 69
Methods ...... 75
Occurrence points ...... 75
Climate data ...... 77
Ecological Niche Modelling ...... 78
xv Quantifying niche overlap ...... 80
Biotic Interactions between unisexual and sexual salamanders ...... 81
Results ...... 83
Niche models ...... 83
Niche overlap ...... 86
Inferring biotic interactions ...... 94
Discussion ...... 95
Ploidy as a driver of niche divergence ...... 96
Genome composition as a driver of ecological niche ...... 97
Biological interactions as a driver of ecological niche ...... 100
Conclusion ...... 104
Chapter 5: General Conclusions ...... 105
Appendix A: Supplemental Tables ...... 111
Appendix B: Supplemental Figures ...... 113
References ...... 115
xvi
List of Tables
Table 1. Proposed hypotheses to explain the presence of salamander individuals with
“mismatched” mtDNA haplotypes in central Ohio...... 14
Table 2. Distances traveled during locomotor endurance trials by unisexual and sexual salamanders collected from sites within Crawford County Ohio with associated morphological measurements...... 53
Table 3. Number of individuals sampled from each site and the proportion of individuals assigned to the site at which they were sampled...... 55
Table 4. Individuals identified as dispersers (mismatch between sampled site and genetically-assigned site)...... 57
Table 5. Summary of rangewide occurrence data used for analyses of two sexual
Ambystoma salamanders and five unisexual Ambystoma biotypes...... 77
Table 6. Parameter summary of the highest ranked ecological niche models for each species or unisexual biotype ...... 84
Table 7. Summary of ENMtools background tests. Each niche comparison between groups was made twice: one involving a comparison of the occurrence points for group 1
xvii versus the background area of group 2 and vice versa (Mean background 1 vs 2, Mean background 2 vs 1)...... 90
xviii
List of Figures
Figure 1. A unisexual Ambystoma salamander...... 2
Figure 2. Sampling locations and partial range maps for Ambystoma barbouri, A. texanum, and unknown individuals identified with mismatched mtDNA haplotypes. .... 17
Figure 3. Phylogenetic tree with posterior probabilities based on a 346 bp section of control region mtDNA...... 19
Figure 4. STRUCTURE bar plot for K = 2 with q values (SE) that represent proportion of ancestry to each group...... 26
Figure 5. Probabilities of three primary hypotheses regarding the identification of mismatched mtDNA haplotypes in central Ohio Ambystoma...... 27
Figure 6. Result of maximum clade credibility species tree analysis with *BEAST including 10 nuclear loci from all unknown, Ambystoma texanum, and A. barbouri individuals...... 29
Figure 7. Unisexual Ambystoma salamander (top) and Small-mouthed Salamander (A. texanum, bottom)...... 45
Figure 8. Univariate plots (Weissgerber et al. 2015) for the distance travelled by dispersed sexual salamanders (circles) and unisexual salamanders (triangles) as measured
xix by Euclidean distance (A) and the log Distance travelled during locomotor endurance trials by unisexuals and three sexual species (B)...... 53
Figure 9. Predicted least-cost paths traveled by genetically-identified dispersers (top: A. texanum, bottom: unisexual Ambystoma)...... 60
Figure 10. Geographic distribution of occurrence data for two sexual Ambystoma species and five unisexual Ambystoma biotypes...... 76
Figure 11. The subset of occurrence points from Ontario only for A. laterale (“LL”, N =
246, bottom) and A. jeffersonianum (“JJ”, N = 123, top) and the unisexual biotypes with mainly A. jeffersonianum genomes (320 points for LJJ, 54 points for LJJJ; top) or A. laterale genomes (406 for LLJ, 30 for LLLJ; bottom)...... 79
Figure 12. Ecological Niche Models plotted in geographic space for two sexual
Ambystoma species (A. laterale and A. jeffersonianum) and five unisexual Ambystoma biotypes...... 86
Figure 13. Summary of Principal Component (PC) 1 and 2 scores for each sexual species shown as squares (A. laterale “LL”; A. jeffersonianum “JJ”) and associated unisexuals of varying ploidy and genome composition shown as colored circles...... 88
Figure 14. A summary of niche overlap values based on ENMtools background tests for niche overlap between each sexual salamander species (A. laterale: LL, A. jeffersonianum: JJ) and unisexual salamanders grouped by ploidy (A, B) or genome composition (C, D)...... 91
Figure 15. A summary of the magnitude of niche similarity/differentiation between sexual salamanders species (A. laterale: LL, A. jeffersonianum: JJ) and unisexual salamanders grouped by ploidy (middle) or genome composition (bottom)...... 93 xx Figure 16. Summary of residual correlations from joint species distribution models using the Ontario only data for occurrences of Ambystoma laterale (left column), A. jeffersonianum (right column) and five unisexual Ambystoma biotypes...... 95
xxi
Chapter 1: General Introduction
Possessing two sexes that both pass their genes to the next generation is one of the fundamental assumptions that we make about animal life, and modern evolutionary biology has revealed advantages of sexual reproduction that Charles Darwin could have only imagined when he was writing The Origin of Species. For example, the mixing of genetic material can foil co-evolving parasites and provide a diversity of evolutionary building blocks to adapt to one’s environment (Lively & Dybdahl 2000; Lively 2010).
Yet sexual reproduction also comes with a significant cost. The production of males, who themselves do not directly make offspring, requires resources that could be dedicated to females (Maynard Smith 1978). From an economic prospective, males are preventing the maximal production of a female’s offspring. There are a small minority of plants and animals that reproduce using a single sex and exist without the frequent shuffling of genes. However, most of these lineages are relatively short-lived in evolutionary time and often referred to as “evolutionary dead ends” (Neiman et al. 2009). Thus organisms which are asexual but are also evolutionarily long-lived offer a special opportunity to understand the evolutionary advantages and disadvantages of sex.
Ohio is home to such an unusual lineage in vertebrates: unisexual Ambystoma
1 salamanders (“unisexuals”, Figure 1). These all-female salamanders are the oldest known unisexual lineage (~5 million years) and reproduce in a way that is unique among vertebrates, called kleptogenesis1 (Bi & Bogart 2010a). When a female arrives at a pond to breed in the spring, she attempts to mate with a male of another salamander species.
The unisexual female uses the male’s sperm to initiate egg development, but her offspring are generally identical clones of the mother. However, the genetic material from the male is occasionally incorporated into the unisexual’s eggs, resulting in the addition or substitution of an entire genome into the resulting offspring. This process has produced more than twenty genome combinations among unisexuals, which can have three to five whole genomes derived from up to five other species (Bogart et al. 2009). Given the strange nature of their reproductive mode, unisexual Ambystoma salamanders may be expected to be rare or geographically isolated, but this group is widespread across North
America and often more abundant than sympatric Ambystoma (Bogart & Klemens 1997).
Figure 1. A unisexual Ambystoma salamander. This individual is a tetraploid with genomes representing four different sexual Ambystoma species: A. laterale, A. texanum, A. jeffersonianum, and A. tigrinum.
1Othewise referred to as “sala-madness” 2
Unisexual salamanders play a unique role in the study of sex. These animals could potentially be described as intermediates between sexual and asexual reproduction, and they provide a new system to test the costs and benefits of sexual reproduction that have been established in other species that are completely sexual or asexual. However, since unisexuals’ reproductive method is recently proposed, there are still many unanswered question concerning the mechanism and maintenance of this unusual mode of reproduction. I have approached three of these primary questions using an integrative set of techniques, ranging from genetics to functional physiology: 1) What is the cause of the mitochondrial ambiguity in Ohio Ambystoma salamanders?, 2) Can differential dispersal explain how sexual and unisexual salamanders coexist?, and 3) Does genomic composition or the total number of genomes predict the range-wide ecological niche relationships of unisexual Ambystoma and their sexual hosts?
What explains the discordance between mitochondrial haplotypes in
Ohio Ambystoma? - Differentiating unisexual Ambystoma from their sexual sperm donor species cannot be completed using morphological differences, necessitating the use of
DNA identification. Identifying unisexual individuals via mitochondrial markers has been previously successful, as the mitochondrial haplotypes of unisexual Ambystoma have sufficiently diverged from their closest relative and the species that was likely the female member of the original hybridization that initiated the all-female lineage, Ambystoma barbouri (Bogart et al. 2009; Greenwald & Gibbs 2012). However, previous surveys of unisexual individuals in Ohio had provided many examples of animals displaying
3 mitochondrial haplotypes from A. barbouri in unexpected areas of the state and with mismatched morphology. Because of this discordance between mitochondrial haplotype and expected species distribution, understanding how A. barbouri-like mitochondrial were found far outside of the established range of A. barbouri was a necessary step to identifying the distribution of unisexual Ambystoma and the sexual species that they interact with (Greenwald & Gibbs 2012).
I tested three hypotheses for this discordance (undetected range expansion, mitochondrial DNA introgression, and hybridization) using data from both nuclear DNA and mitochondrial DNA analyzed with methods that varied in the parameters estimated and the timescales measured. The best supported hypothesis was one of mitochondrial introgression, where the mitochondrial haplotypes of one species (A. barbouri) were introgressed into populations of another sexual species (A. texanum). I strengthened this result by confirming that individuals with mismatched mitochondrial haplotypes were A. texanum through examining species-specific tooth morphology. Further, environmental data from all collection sites showed a pattern of environmental differentiation between sites with introgressed mitochondria and those without, providing a potential role of selection for certain environmental conditions in this process.
Specifically, this chapter provides the necessary distributional information to correctly identify sexual and unisexual salamanders in Ohio using a single mitochondrial locus and describes the evolutionary basis for the observed mitonuclear discordance between A. barbouri and A. texanum. Generally, this chapter provides a template example of the value of using complimentary analyses to make inferences of the directionality, time scale, and source of mtDNA introgression in animals, which is increasingly 4 recognized as a long term consequence of species hybridization (Linnen & Farrell 2007;
Bolnick et al. 2008) .
Can dispersal explain coexistence? - If unisexual salamanders are able to gather new genetic material without making males, how can sexual and unisexual organisms coexist? Unisexuals are essentially “sexual parasites” of sexual species because sexual males are wasting sperm that could be used to mate with females of their own species. In contrast, even though unisexuals do not make males, they still require the input of sperm to make their eggs. This tension between reliance and independence is difficult to explain, but other researchers have provided a mathematical framework that may predict which life history characteristics promote the coexistence between unisexual and sexual groups (Hellriegel & Reyer 2000). Of these characteristics, dispersal, the ability to move across the landscape, is potentially instrumental in explaining coexistence. If unisexuals excel at dispersing, they can exploit temporary habitats and take advantage of sexual males that are naïve to the presence of unisexuals.
For my second thesis chapter, I compared unisexual and sexual salamander dispersal in a fragmented agricultural landscape in Crawford County Ohio using genetic data. I used genetic markers to perform a novel assignment test method that identifies individual animals that were not born in the same wetlands in which they were sampled.
The proportion of these dispersers was similar between unisexuals and the local sexual species (A. texanum), but the unisexuals traveled a significantly smaller distance from the wetlands in which they were born. This result runs opposite to the prediction of greater dispersal ability in unisexuals, raising even more questions about what factors determine coexistence. I expanded this analysis by measuring the actual functional physiological 5 differences between unisexual and sexual salamanders that may be reflected in the genetic patterns by measuring their walking endurance on customized treadmills. These results were consistent with the genetic data: unisexual salamanders become fatigued earlier and walk less than half the distance travelled by sexual salamanders.
What drives niche relationships between unisexual and sexual Ambystoma salamanders? - Unisexual taxa that require reproductive output from other species are not exempt from the competition that these species provide, and characterizing the presence of ecological diversification between competing sexual and asexual taxa has been a foundational debate in understanding ongoing coexistence (Kirkendall 1990;
Kokko et al. 2008). Studies that investigate how reproductive mode, whether completely asexual or gynogenetic, relates to ecological niche overlap have described a wide variety of patterns, from niche intermediacy to niche transgression (Glennon et al. 2014; Ficetola
& Stöck 2016). However, the plant and animal systems used in these studies are often unable to make conclusions on whether the observed patterns are due to differences in ploidy between unisexual and sexual taxa (more genomes promote neofunctionalization;
Renny-Byfield & Wendel 2014) or imbalance between the numbers of genomes from parental species (more net genomes from a given species promote similarity to that species). Few wild lineages provide sufficient variation in these traits that are necessary to make appropriate comparisons. For my third chapter, I took advantage of a range-wide dataset of occurrence points for two sexual Ambystoma species (A. laterale and A. jeffersonianum) and unisexual salamanders of varying ploidy and genome composition in order to investigate if ploidy or genome similarity best predicts differences in niche overlap. 6 Using rigorous niche model construction, I found that the realized niches of both sexual species were overly similar to all unisexuals, regardless of ploidy or genome composition. This results suggests a third hypothesis: the reproductive requirement of sexual males limits the niche divergence of unisexual salamanders. Additional results from joint species distribution modelling on a regional subset of the data support this third hypothesis by suggesting that biological interactions between sexual and unisexual salamanders play a significant role in their co-occurrence. This chapter is the first study to investigate the role of both ploidy and genome composition in the realized niche relationships of sexual and unisexual taxa. However, the data support reproductive behavior as a more powerful force for shaping the niche of unisexuals that still require sperm from congeneric males, limiting the role niche divergence can play in promoting the coexistence of closely-related sexual and kleptogenetic taxa.
The unisexual salamander system is a promising avenue of research for discovering why sexual reproduction is so widespread among animal life. The unisexuals’ reproductive mode, potentially possessing the benefits of both sexual reproduction (new genetic variation) and asexual reproduction (all individuals directly reproduce), allows for the examination of why this “perfect” way of reproducing has not resulted in the extinction of related species. I have integrated genetic, ecological, and physiological analyses to form a better understanding for how unisexual salamanders maintain co-existence with the sexual species that they take advantage of. While the advantages of kleptogenesis are clear on paper, the unisexual lineage is at a disadvantage in regards to mate choice, fecundity, and dispersal. Together, the costs of maintaining an independent mitochondrial lineage while acquiring nuclear genomes from other species 7 may result in a downstream effects on metabolism and reproductive efficiency that limit the competitive ability of unisexual salamanders. The extent and timing of these consequences provides an exciting research avenue to further advance research in the evolution and maintenance of sex.
8
Chapter 2: Evolutionary basis of mitonuclear discordance between sister species of mole salamanders (Ambystoma sp.)
Note: This chapter has been published as below, and benefitted from contributions of the co-authors:
Denton RD, LJ Kenyon, KR Greenwald, and HL Gibbs. 2014. Evolutionary basis of
mitonuclear discordance between sister species of mole salamanders (Ambystoma
sp.). Molecular Ecology. 23:2811-2824.
Abstract
Distinct genetic markers should show similar patterns of differentiation between species reflecting their common evolutionary histories yet there are increasing examples of differences in the biogeographic distribution of species-specific nuclear (nuDNA) and mitochondrial DNA (mtDNA) variants within and between species. Identifying the evolutionary processes that underlie these anomalous patterns of genetic differentiation is an important goal. Here we analyze the putative mitonuclear discordance observed between sister species of mole salamanders (Ambystoma barbouri and A. texanum) in which A. barbouri-specific mtDNA is found in animals located in the range of A. texanum. We test three hypotheses for this discordance (undetected range expansion, 9 mtDNA introgression and hybridization) using nuDNA and mtDNA data analyzed with methods that varied in the parameters estimated and the timescales measured. Results from a Bayesian clustering technique (STRUCTURE), bi-directional estimates of gene flow
(MIGRATE-N and IMa2), and phylogeny-based methods (*BEAST, BUCKy) all support the conclusion that the discordance is due to geographically restricted mtDNA introgression from A. barbouri into A. texanum. Limited data on species-specific tooth morphology match this conclusion. Significant differences in environmental conditions exist between sites where A. texanum with and without A. barbouri-like mtDNA occur, suggesting a possible role for selection in the process of introgression. Overall, our study provides a general example of the value of using complimentary analyses to make inferences of the directionality, time scale, and source of mtDNA introgression in animals.
Introduction
Conceptually, different classes of genetic markers should show similar patterns of differentiation both within and between species as a result of their shared evolutionary history. This assumption allows for the inference of phylogenetic relationships among taxa. In practice, various markers often show different patterns of differentiation due to a variety of evolutionary processes (Avise 1994). For instance, genetic comparisons of closely related animal taxa show discordance between nuclear genes (nuDNA) and mitochondrial DNA (mtDNA). This lack of congruence (termed mitonuclear discordance) results from the introgression of mitochondrial genes from one population or species to another combined with low levels of nuclear introgression (Avise 1994).
Studies that report mitonuclear discordance have become more common as researchers 10 increasingly use mtDNA and nuDNA loci concertedly for phylogeographic and phylogenetic analyses in a range of taxa (Funk & Omland 2003; Chan & Levin 2005;
Gompert et al. 2008; Parham et al. 2013; Zieliński et al. 2013).
Identifying cases of mitonuclear discordance and understanding the underlying mechanisms is an important step in understanding evolutionary and ecological relationships between species or populations (Toews & Brelsford 2012). For example, different types of genetic introgression can alter ecological relationships among organisms (Ryan et al. 2009), create independent lineages (Robertson et al. 2006), or result in taxonomic misidentifications (reviewed in Funk & Omland 2003). Mitonuclear discordance can also result in a loss of genetic distinctiveness between species that results in uncertainty in specifying species’ ranges, leading to the potential misidentification of cryptic species (Rohwer et al. 2001; Zieliński et al. 2013). Mitonuclear discordance has been explained by a variety of evolutionary mechanisms, including adaptive sweeps of mtDNA haplotypes, sex-biased hybridization, or demographic influences such as genetic drift (reviewed in Toews & Brelsford 2012). Although it is a well-recognized phenomenon, the methods used to detect and explain mitonuclear discordance vary in their approaches and assumptions. Thus, there is an active focus on evaluating the current methods used to detect cases of mitonuclear discordance and to identify the evolutionary mechanisms responsible (Funk & Omland 2003; Toews et al. 2013).
Here, we investigate a putative case of mitonuclear discordance within sister species of mole salamanders in Ohio, Ambystoma barbouri (Streamside Salamander) and
A. texanum (Smallmouth Salamander). Amphibians display diverse patterns of mitonuclear discordance that result from multiple processes, including asymmetrical 11 mtDNA introgression (reviewed in Toews & Brelsford 2012), asymmetrical nuDNA introgression (Di Candia & Routman 2007; Johanet et al. 2011), and the introgression of both mtDNA and nuDNA (Chatfield et al. 2010; Veith et al. 2012). Ambystoma barbouri and A. texanum are two morphologically similar salamander species but can be identified using species-specific patterns of tooth morphology and inhabit different breeding environments (Kraus & Petranka 1989). Despite this distinctiveness, these animals are known to interbreed in several sympatric areas of their ranges, likely leading to locations where mitonuclear discordance exists (Niedzwiecki 2005). For example, based on three mtDNA loci and morphological characters, Niedzwiecki (2005) identified a single population of A. texanum in southwestern Ohio (Greene County) with a mtDNA haplotype most similar to that of A. barbouri. Eastman et al. (2009) identified two individuals that were potential hybrids based on mismatched mtDNA haplotypes.
Greenwald and Gibbs (2012) subsequently discovered several individuals in central Ohio with A. barbouri mtDNA haplotypes that are >100 km from the nearest sample within the established range of A. barbouri (“Unknown” individuals; Figures 1 and 2, Supporting information Table S1). However, all examples involved a small number of samples, lack of information on nuDNA or morphological variation, and a limited set of methods to analyze the data. Additionally, A. barbouri displays multiple satellite populations (USGS
2012), leaving the possibility that mtDNA haplotypes have identified previously unknown populations of A. barbouri. Thus, understanding the extent of mitonuclear discordance and its possible causes requires detailed sampling and more comprehensive analyses of both mtDNA and nuDNA markers.
12 For this purpose, we collected samples from a broad range of sites (Figure 2) and analyzed them using a diagnostic mtDNA marker and 10 nuDNA sequence-based markers (c.f. Greenwald and Gibbs 2012). Initially, we sought to confirm that mitonuclear discordance was present in these salamanders and identify its geographic extent. Next, we tested three hypotheses that provide demographic or potentially adaptive explanations for the observed genetic pattern (Table 1). First, the presence of
“mismatched” mtDNA haplotypes could be explained as a consequence of A. barbouri having a larger range than previously recognized and it has gone undocumented due to the difficulty in identifying the two species (“misidentification” hypothesis). Second, the
“unknown” individuals may be A. texanum with introgressed A. barbouri mtDNA from a historical hybridization event (“introgression” hypothesis; Toews & Brelsford 2012).
Finally, the mitonuclear discordance could be due to ongoing but geographically restricted hybridization between A. barbouri and A. texanum, with the result that the
“unknown” individuals in central Ohio with mismatched genomes are hybrids
(“hybridization” hypothesis).
13 Table 1. Proposed hypotheses to explain the presence of salamander individuals with “mismatched” mtDNA haplotypes in central Ohio. Each hypothesis is presented along with supporting predictions for each of four analysis types.
Analysis Type and Example Program/Procedure Bayesian Gene flow Phylogeny Clustering estimation estimation Morphology Hypothesis (STRUCTURE) (MIGRATE-N, IMa2) (*BEAST, BUCKy) (Maxillary teeth) Unknown individuals Unknown Symmetrical gene Unknown display rounded cusps A. barbouri individuals flow between A. individuals group (A. barbouri misidentification group with A. barbouri and with A. barbouri phenotype) on barbouri unknown group maxillary teeth
14 Unknown individuals Unknown Symmetrical gene mtDNA Unknown display pointed cusps individuals flow between A. introgression individuals group (A. texanum group with A. texanum and into A. texanum with A. texanum phenotype) on texanum unknown group maxillary teeth Unknown individuals Admixture Gene flow between Admixture within potentially display Hybridization within unknown unknown group and unknown group intermediate group both parental groups phenotype
Each of these hypotheses can be tested by comparing patterns of variation in nuDNA markers in the two parental species and in mismatched individuals (Table 1). We used three types of methods for such tests: Bayesian clustering methods (Beaumont et al.
2001; Susnik et al. 2004; Grant et al. 2007; Pastorini et al. 2009; Bohling et al. 2012),
Isolation-Migration (IM) methods (Barrowclough et al. 2005; Ackermann & Bishop
2010; Nevado et al. 2011; Austin et al. 2011) and species tree-based phylogenetic techniques (Sequeira et al. 2011; Melo-Ferreira et al. 2012; Parham et al. 2013). Each method varies in terms of assumptions made, the parameters estimated, and the time- scale over which the estimation occurs. For example, genetic clustering techniques identify sets of genetically similar samples under minimal assumptions but provide no estimates of gene flow or effective population sizes. In contrast, many IM methods provide estimates of the direction and magnitude of gene flow but assume that populations are at genetic equilibrium and that retained ancestral polymorphism does not impact estimates of variation shared between populations. Finally, species tree-based phylogenetic techniques can account for retained ancestral polymorphism in estimates of polymorphism between species but assume that gene flow between species is limited.
Utilizing a range of techniques allows for a more comprehensive assessment of the pattern of genetic discordance and its possible causes in these salamanders than is usually completed as past studies generally use a limited set of methods (Barrowclough et al.
2005; Parham et al. 2013). Overall, this study provides a general example of the value of using complimentary analyses to make inferences of the directionality, time scale, and source of mtDNA introgression in animals.
15 Methods
Samples and Genotyping
We collected DNA samples from the tail tips of 39 individual salamanders from across
Ohio that spanned the previously-described ranges of A. barbouri and A. texanum (Figure
1, Supporting information Table S1). Within the putative ranges of each species, A. barbouri samples were collected from streams (N = 11) whereas A. texanum samples were collected in ponds (N = 26). DNA was extracted from tail tips using Qiagen DNeasy tissue kits (Qiagen, Valencia, CA). Each sample was then analyzed in two ways. First, to identify species, each sample was sequenced for a 346 bp section of control region mtDNA from an amplicon generated using primers F-THR and R-651 (Shaffer &
McKnight 1996; Bogart et al. 2007) that contains species-specific polymorphisms
(Bogart et al. 2007; Greenwald & Gibbs 2012). Individuals were classified as having either A. barbouri or A. texanum mtDNA based on the presence/absence of specific mtDNA polymorphisms (Supporting information Table S2) and whether they clustered with reference A. texanum or A. barbouri samples (Figure 2). Samples collected in the putative range of A. texanum that were identified as having A. barbouri mtDNA were designated as “Unknown”. We also sequenced all samples at 10 nuDNA loci (see
Supporting information Table S3 for primer sequences) consisting of seven anonymous
DNA loci (Smith et al. 2005; Greenwald & Gibbs 2012) and partial sequences from three protein-coding loci (Vieites et al. 2007). We followed PCR protocols as described in
Vieites et al. (2007) and Greenwald & Gibbs (2012). All sequences were aligned with
MUSCLE v.3.8.31 (Edgar 2004) and nuDNA sequences were phased using Phase
(Stephens et al. 2001; Stephens & Donnelly 2003) as implemented in DNAsp v5.10.1 16 (Librado & Rozas 2009). Tests for Hardy-Weinberg equilibrium (HWE) across loci and groups were conducted with GenAlEx 6.501 (Peakall & Smouse 2006, 2012), and linkage disequilibrium between pairs of loci was evaluated using Genepop 4.2 (Raymond
& Rousset 1995; Rousset 2008).
Figure 2. Sampling locations and partial range maps for Ambystoma barbouri, A. texanum, and unknown individuals identified with mismatched mtDNA haplotypes. Range data is taken from the USGS National Amphibian Atlas (USGS 2012).
Population and Phylogenetic Analyses
First, we used the Bayesian clustering program STRUCTURE (version 2.3.3;
Pritchard et al. 2000) to evaluate the assignment of the “Unknown” group of samples
(putative A. texanum individuals with A. barbouri mtDNA) to either potential parental species. We followed the method of Gibbs et al. (2010) by first confirming that the A. barbouri and A. texanum groups are clearly detected as distinct clusters and then 17 including the unknown group while specifying K = 2 genetic clusters in the analysis. All
STRUCTURE analyses were run using all 39 individuals in an admixture model with allele frequencies correlated and without sample location priors (Hubisz et al. 2009). Each run included a burn-in period of 5 x 105 repetitions followed by 7.5 x 105 MCMC repetitions.
Convergence was determined based on the examination of ln likelihood graphs and the consistency of results across three separate runs. Lastly, we calculated q-values and associated 95% confidence intervals to interpret the proportion of unknown individuals’ genomic assignment to the two parental species.
18
Figure 3. Phylogenetic tree with posterior probabilities based on a 346 bp section of control region mtDNA (Primers F-THR, R-651; Shaffer & McKnight 1996; Bogart et al. 2007). The 39 individuals and reference samples are identified by letter (U = unknown, T = Ambystoma texanum, B = A. barbouri, J = A. jeffersonianum) and reference samples are included that have had species identity confirmed by morphology. See supporting information table S1 for more information about samples.
Second, we used the program MIGRATE-N (version 3.3.1; Beerli 2006) to estimate directional migration rates and effective population sizes using all 39 individuals and the
10 nuclear loci described above. An advantage of MIGRATE-N is that it provides a framework to evaluate the fit of different multiple migration models to the data by comparing Bayes factors. To test hypotheses about the evolutionary origin of the
“Unknown” samples we compared seven a priori migration matrix models (Beerli &
Palczewski 2010, Supporting information Figure S1). Three of the proposed models
19 reflect gene flow scenarios that would support the three main explanations for mitochondrial-range mismatch: misidentification, mtDNA introgression, and hybridization. Two additional models were included that added a one-way migration rate between the Unknowns and A. barbouri. These models reflect the potential directionality of a historical hybridization event that would result in mtDNA introgression into A. texanum. Finally, two global models were included that reflect either symmetric gene flow or unidirectional gene flow between the three groups. For each MIGRATE-N analysis, initial theta and migration values were generated using the default FST calculation and the initial genealogies were sampled starting from a random tree. Since no previous information concerning migration rates or theta values for these species was available, we used uniform priors and slice sampling for parameter distributions. Static heating was used with temperatures of 1, 1.5, 3, and 1 x 106. Three long chains with 1 x 104 burn-in repetitions followed by 1 x 104 recorded steps for every 100 steps, resulting in a total of 1 x 106 sampled genealogies. Convergence was determined by investigating the smoothness of parameter histograms and the consistency of results across three separate runs. The probability of different models in fitting the data was then compared using
Bayes factors as described in Beerli & Palczewski (2010).
Third, we used the program IMa2 (Hey & Nielsen 2004) to estimate parameters
(e. g. divergence times) not estimated with MIGRATE-N. Unlike MIGRATE-N, IMa2 explicitly accounts for shared retained ancestral polymorphism between taxa and does not assume that the taxa under study are at genetic equilibrium (Hey & Nielsen 2004). Based on the results from STRUCTURE (see below), we assumed that the unknown samples contained true A. texanum nuclear genomes. Therefore, we pooled these samples with the 20 designated A. texanum samples and compared this sample (n= 39 individuals) with the A. barbouri group (n=11). We ran all sequences through the Perl script IMGC (Woerner et al. 2007) to confirm that the data represented single non-recombining blocks of sequence which conformed to an infinite allele mutation model. We found deviations from this model for seven loci. Based on program recommendations, we deleted between 1–7 individual sequences (mean = 3.8) to satisfy the mutation and recombination assumptions and used this modified data set in subsequent IMA2 analyses.
We analyzed these data under a two population isolation-migration model in which we estimate effective population size of the ancestor to both salamander species, time at which this ancestor diverged into the two present-day taxa, the effective population sizes of these contemporary taxa, and levels of ongoing gene flow between A. barbouri and A. texanum since their origin. We set up MCMC runs of IMA2 using a variety of starting random seeds, heating schemes, and run lengths until we obtained repeatable estimates of all parameters. Our final run consisted of an MCMC run of 3 x
107 generations with a moderate level of chains (30) run with an aggressive heating scheme. We report parameter estimates based on the high point for the parameter distribution, along with error estimates based on the high and low values that define 95% of the distribution for each parameter estimated. To translate model estimates into demographic estimates we followed the IMA2 manual using an estimated mutation rate of 8.5 x 10-9 per base per year derived from available estimates of mutation rates in single copy nuclear DNA in vertebrates (see Kubatko et al. 2011) and a generation time of 2.5 years based on observed age of first reproduction in Ambystoma (Petranka 1998).
21 Finally, to incorporate a long-term evolutionary perspective with phylogenetic analyses, we used two multispecies coalescent species tree estimators, *BEAST (version
1.7; Heled & Drummond 2010; Drummond et al. 2012) and BUCKy (Larget et al. 2010) to estimate relationships among “taxa” defined as the two groups of samples designated a priori as Ambystoma sp. and the unknown samples. The *BEAST analysis uses information from multiple gene trees to estimate a species tree. The BUCKy analysis similarly combines information from multiple gene trees, but within the framework of
Bayesian concordance analysis (Larget et al. 2010). As outlined in Table 1, each hypothesis predicts different phylogenetic relationships among samples. Under the misidentification hypothesis, unknowns should group with A. barbouri samples, while the introgression hypothesis predicts the opposite pattern (unknowns with A. texanum samples). Finally, if ongoing hybridization is present the unknowns should show no clear grouping of unknowns with either A. barbouri or A. texanum samples.
For both phylogenetic analyses, we generated gene trees for each of the 10 loci with the MRBAYES plugin (version 3.2.1; Huelsenbeck & Ronquist 2001) implemented within the program GENEIOUS (version 5.6, Drummond, Ashton, et al. 2012) using a generalized time-reversible model, unconstrained branch lengths, 1 x 106 chain length, and a 1 x 105 burn-in period (Supporting information Figure S2). Locus-specific nucleotide substitution models were chosen using JMODELTEST (version 2.1.2; Guindon &
Gascuel 2003; Darriba et al. 2012) and implemented in *BEAST using BEAUTi version
1.6.1. Molecular clocks were imposed on all loci, the species tree prior was set to the
Yule process, and the population size model was set to piecewise constant. We ran three independent runs with a chain length of 5 x 108 and parameters were logged every 5 x 104 22 generations. The log files from these three runs were combined using LOGCOMBINER
(version 1.6.1). We used TRACER (version 1.5) to examine effective sample size values and posterior estimate graphs. The resulting species tree was visualized using FIGTREE
(version 1.4.0; http://tree.bio.ed.ac.uk/software/figtree/). For BUCKy analyses, gene trees were summarized with MBSUM by randomly selecting 1,001 trees from each locus and the summary trees were implemented using default parameters. We completed three independent BUCKy analyses at alpha values of 0, 1, and infinity. The alpha value indicates the level of discordance among gene tree topologies, with 0 and infinity representing no discordance and complete independence, respectively.
Morphological and Environmental Analyses
To supplement the genetic analyses, we also conducted a limited comparison of morphology, mtDNA assignment, and geographic location. The purpose of examining morphology was to potentially confirm the presence of individuals in central Ohio with a mismatch between mtDNA haplotype and species-specific morphology as suggested by the genetic analyses (see below). For this, we collected five individual salamanders from the same Crawford County (Ohio) sites as above during May 2012. Animals were sacrificed using an overdose of Tricaine (MS-222, 5 grams/liter) according to
Institutional Animal Care and Use Committee (Protocol #2012A00000039) protocols.
The maxillary tooth morphology was then examined and classified as either A. barbouri- or A. texanum -specific based on criteria described in Kraus and Petranka (1989).
Second, we were interested in whether there were differences in environmental conditions between sites occupied by unknowns and pure A. texanum. To limit the 23 effects of spatial autocorrelation, we only included the samples from the A. texanum and unknowns that were sympatric, resulting in the inclusion of 22 individuals across 11 contiguous counties in central Ohio. For each of these sites, we extracted site-specific
Worldclim variables (Hijmans et al. 2005) with DIVA-GIS v. 4.0 (Hijmans et al. 2001) from locations containing each type of sample and used a principal component analyses
(PCA) with SPSS v. 17 (IBM Corporation, Somers, NY, USA) to summarize the variation for each site. These factor scores for PCA 1-3 were then used as dependent variables for a MANOVA comparing the environmental characteristics of between sites with the A. texanum and unknown groups.
Results mtDNA and nuDNA Genotyping
Mitochondrial genotyping identified 15 unique haplotypes across the 39 samples from 31 sites across Ohio and Indiana. The phylogenetic tree produced by MRBAYES with all of the samples, including six reference A. barbouri individuals and 2 reference A. texanum individuals, clearly separated into two well-supported clades (Figure 3).
Fourteen individuals grouped with the A. texanum reference samples, whereas all other samples clustered with the reference A. barbouri individuals. Twelve of the samples with
A. barbouri-like mtDNA haplotypes were well outside of the known range of A. barbouri, similar to the “unknown” samples identified by Greenwald and Gibbs (2012).
Following the mtDNA haplotype classification of the samples into A. barbouri, A. texanum, and Unknowns, all 39 individuals were successfully sequenced and aligned at the ten nuclear loci (Supporting Information Table S1, S2, S3) and these data were used 24 for subsequent analyses. Six of the ten loci significantly differed from HWE after using a
Bonferroni-corrected significance level (p < 0.002). Departures from HWE were group- specific, as only a single locus (E13E02) was in violation within all three groups
(Supporting Information Table S4). No loci were found to be in linkage disequilibrium.
Bayesian Clustering
Samples classified a priori as Ambystoma barbouri and A. texanum were perfectly segregated under a K = 2 model in STRUCTURE with all individual assignment probabilities > 0.99 (results not shown). When the analyses were repeated under a K = 2 model with the Unknowns included, all Unknown individuals were assigned to the A. texanum cluster (Figure 4). The mean assignment coefficient to cluster one (q1 ± 95% confidence interval) for A. barbouri was 0.994 ± 0.002. The mean assignment coefficients to cluster two (q2 ± 95% confidence interval) for A. texanum and the
Unknowns were 0.969 ± 0.023 and 0.940 ± 0.057, respectively. The log likelihood plots showed convergence in all runs and three repeated runs generated the same results. To confirm that our findings were robust to the assumption made by STRUCTURE of HWE within populations we reanalyzed our data using an alternative method which does not assume HWE (k-sample clustering implemented in the R package adegenet Jombart
2008; R Development Core Team 2011) and obtained the same result (results not shown).
25 Figure 4. STRUCTURE bar plot for K = 2 with q values (SE) that represent proportion of ancestry to each group. Plot was created with distruct (version 1.1; Rosenberg 2003)
Estimates of Population Size and Gene Flow
The “from A. barbouri” MIGRATE-N model had the highest probability among the models considered (Bezier lmL = -6077.44, model probability = 1.0, Figure 4). This model includes symmetric gene flow between the A. texanum and unknown groups and one-way gene flow from A. barbouri to the unknown group. Theta values for all three groups were similar in magnitude, with A. barbouri having a slightly lower value (Θ =
0.00139) compared to either A. texanum (Θ = 0.00168) or the unknown group (Θ =
0.00162). The migration statistic M (immigration rate/mutation rate), which measures the relative importance of immigration over mutation as a source of novel variation in a population, was more than an order of magnitude higher for the symmetric gene flow parameter between the unknowns and A. texanum (M = 9,534.1) than the one-way gene flow parameter from the A. barbouri to the unknown group (M = 347.0). These results indicate that A. texanum and unknown samples form a single panmictic unit with the addition of a relatively small amount of gene flow from A. barbouri to the unknown group. This supports the hypothesis that the mitonuclear divergence found in the
26 unknowns is due to mtDNA introgression from A. barbouri into central Ohio populations of A. texanum.
Figure 5. Probabilities of three primary hypotheses regarding the identification of mismatched mtDNA haplotypes in central Ohio Ambystoma. Migration rates ( M) and Bezier log likelihood values were produced with migrate-n and probabilities were calculated as in Beerli & Palczewski (2010).
IMA2 analyses identify a small ancestral population of ~4,000 individuals which split into existing populations of A. texanum and A. barbouri roughly 400,000 years before present (ybp), although error estimates of this value are large (95% of estimated
27 values: 261,538 – 3,750,000 ybp). Contemporary populations of both species have effective population sizes that are 3 – 8 times larger than the ancestral population [point estimate of Ne for A. texanum: 12,462 individuals (95% value range: 6,808 – 21,432); Ne for A. barbouri: 32,885 (95% range: 20,500 – 50,846)]. In contrast with population size estimates produced by MIGRATE-N, the more northerly species (A. texanum) has a smaller effective population size. Additionally, there are low levels of ongoing gene flow between these species after considering shared similarity due to retained ancestral polymorphism. Coalescent-based estimates of the number of effective migrants moving from A. texanum to A. barbouri is 0.164 (95% range: 0.026 – 0.673) while the same value for gene flow in the opposite direction is slightly higher (0.285 [95% range: 0.018 – 1.26) although the 95% ranges of values for each point estimate substantially overlap. These results show that although A. barbouri and A. texanum are a recently-evolved pair of sister species, they have been isolated for a substantial period of time (> 100,000 generations) but also continue to experience limited (but nonzero) amounts of gene flow.
Phylogenetic Analyses
The 10 gene trees produces by MRBAYES provided varying levels of resolution for the three groups (Supporting information Figure S2). The *BEAST analysis including all
10 nuclear loci produced a species tree topology with a highly supported clade (P = 1.0) consisting of A. texanum and the unknowns with A. barbouri as a sister group (Figure 5).
This result supports the conclusion of mtDNA introgression from A. barbouri into A. texanum (Table 1). In contrast, the BUCKy analysis produced a single poorly resolved tree
28 with low concordance factors (Supporting information Figure S3) and hence is uninformative in discriminating among the hypotheses in Table 1.
Figure 6. Result of maximum clade credibility species tree analysis with *BEAST including 10 nuclear loci from all unknown, Ambystoma texanum, and A. barbouri individuals. Numbers on branches indicate posterior probabilities.
Morphological and Environmental Analyses
The mtDNA analysis of five adult salamanders collected for morphological analyses showed they contained a mix of A. barbouri haplotypes (N=3) and A. texanum haplotypes (N=2). However, all five individuals displayed maxillary teeth with pointed cusps that are diagnostic of A. texanum (see Figure 2 in Kraus & Petranka 1989; data not shown).
The PCA procedure generated three components that together explained 94.47% of the total variation. Principal Component 1 (43.28% of total variation explained) had
29 high loadings (> 0.093) from five Bioclim variables related to rainfall (Annual
Precipitation, Precipitation of Wettest Quarter, Precipitation of Driest Quarter,
Precipitation of Warmest Quarter, and Precipitation of Coldest Quarter). Therefore, PC1 was interpreted as capturing the variation in precipitation. The second component
(33.75% of total variation explained) captured variation in temperature (Bioclim variables with loadings > 0.92: Annual Mean Temperature, Maximum Temperature of
Warmest Month, Mean Temperature of Wettest Quarter, Mean Temperature of Warmest
Quarter, and Mean Temperature of Coldest Quarter). Principal Component 3 (17.44% of total variation explained) reflects annual variation in temperature and precipitation
(variable with high (> 0.80) loadings: Temperature Seasonality and Precipitation
Seasonality and Annual Temperature Range).
Overall, PCA components were significantly different between the sympatric A. texanum and unknowns (MANOVA, F = 9.507, hypothesis df = 3, p < 0.001). However, a component-by-component analysis shows that only PC 1 scores were significantly different between the two groups (t = -3.457, 95% confidence interval = - (1.924 –
0.476), p = 0.002). Therefore, unknown individuals that were sympatric with A. texanum individuals were associated with sites that had higher amounts of annual precipitation.
However, the differences in mean annual precipitation between the A. texanum (942.5 mm ± 4.24 SE) and unknown samples (973.5 mm ± 8.27 SE) are relatively small.
Discussion
Our genetic analyses show that the unknown salamanders in central Ohio are A. texanum individuals with introgressed A. barbouri mtDNA. Our work has both 30 methodological and evolutionary implications for studying mitonuclear discordance in nature. Below we discuss issues to do with using multiple analyses of genetic data to investigate mitonuclear discordance, their relationship to past observations of introgression in Ambystoma, and what mechanisms may have led to the observed patterns in these salamanders and the general implications of our work for understanding the causes of mitonuclear discord in vertebrates.
Value of multiple methods of analysis
Our study illustrates the power of using multiple analyses to investigate presumed cases of mitonuclear discordance and offers more comprehensive approach than many recent studies based on mtDNA and nuDNA variation (Sequeira et al. 2011; Nevado et al. 2011; Melo-Ferreira et al. 2012). The methods we used represent novel approaches to investigating discordance and were chosen to complement each another in light of the strengths and limitations of each technique. For example, our use of model selection within MIGRATE-N was novel in that it allowed us to statistically compare the likelihood of different models accounting for patterns of discordance as described in Table 1. These results also provide information concerning the directionality of the mtDNA introgression, as there was well-supported unidirectional gene flow from A. barbouri to the unknown group. This suggests the mtDNA introgression resulted from A. barbouri individuals invading populations of A. texanum. While model selection has not been commonly used in investigations of mitonuclear discordance, it is increasingly becoming an important tool for evaluating demographic hypotheses about phylogeography
(Carstens et al. 2013), hybridization (Kubatko 2009), and species delimitation (Csilléry et 31 al. 2010; Camargo et al. 2012) and we see an important role for it in future investigations of genetic discordance in natural populations. A weakness of MIGRATE is that it assumes a migration-drift equilibrium in each population and does not account for the potential impact of incomplete lineage sorting (ILS) on levels of genetic similarity between populations. We feel this is not a significant issue for two reasons: first, the estimated average time for lineage sorting to occur after divergence between populations is 2–3 Ne generations (Neigel & Avise 1986). Based on our estimates of Ne (< 40, 000) and generation time (2.5 years – see above) there has been sufficient time for lineage sorting to have taken place in these species since they diverged. Second, the results of the
IMa2 program, which takes into account ILS, match the best supported model from
MIGRATE-N in terms of directions and magnitude of migration. When investigations of mitonuclear discordance have relied solely on observing conflicting patterns of mtDNA- and nuDNA-based phylogenetic trees (Di Candia & Routman 2007; Bossu & Near 2009;
Spinks & Shaffer 2009; Chen et al. 2009), the effects of ILS have been difficult to address. The same can be said for analyses that make interpretations based on a clustering method such as STRUCTURE (Gompert et al. 2008; Veith et al. 2012). Methods such as
STRUCTURE are valuable in that they have few assumptions, but are limited in their ability to produce specific parameter estimates. However, recent work has shown that both
IMA2 (Strasburg & Rieseberg 2010) and species tree analyses (Knowles & Carstens
2007) are relatively robust to violation of assumptions and do account for ILS. Here we show that leveraging analyses that do and do not account for ILS can provide a thorough and complete evaluation of the timing and direction of introgression.
32 Mitonuclear discordance in Ambystoma
Amphibians are one of the most common groups in which mitonuclear discordance has been identified, although there are examples from other animals
(reviewed in Toews & Brelsford 2012). However, the majority of identified cases of mitonuclear discordance have been recognized in frogs, with many fewer cases in salamanders (Chan & Levin 2005). Ambystoma salamanders provide many examples of genetic introgression, including adaptive introgression from invasive into native species
(Ryan et al. 2009) and extensive introgression between multiple species within the unisexual Ambystoma complex (Bi & Bogart 2006; Bogart et al. 2007; Bi et al. 2009).
Specifically, mitochondrial haplotypes originally derived from A. barbouri, similar to those described here in A. texanum, are found within the entire unisexual Ambystoma complex (Robertson et al. 2006; Bogart et al. 2007). Unisexuals are hypothesized to be the result of an ancient hybridization involving a common ancestor most similar to A. barbouri. The persistence of this independent mitochondrial lineage, given the cytonuclear interactions of up to five genomes from other Ambystoma species, suggests that there is some property of A. barbouri-like haplotypes or Ambystoma nuclear genomes that allow for reduced cytonuclear conflict after introgression (Bogart et al.
2007, 2009).
This study is not the first to discover A. barbouri-like mtDNA haplotypes within
A. texanum populations, yet it is the first to characterize discordance and evaluate putative causes. A range-wide genetic survey by Niedzwiecki (2005) revealed a single A. texanum individual from Greene County Ohio that contained an A. barbouri mtDNA haplotype and a single individual from southern Indiana was identified by Eastman et al. 33 (2009). Greene County is one county north of Warren County, where we identified all specimens sampled as pure A. barbouri. The individual identified by Niedzwiecki suggests that the sampling gap between central and southwestern Ohio in our study likely contains populations of A. texanum with A. barbouri-like mtDNA. Because the range of
A. texanum is many times larger than that of A. barbouri, it is surprising that no other mtDNA mismatches have been found, especially near the other recognized zone of introgression between the two species in western Kentucky. This geographic pattern suggests that A. texanum may have carried mtDNA northward from previous introgression events during glacial maxima. This scenario is supported from other studies which show that species with expanding ranges are more likely to carry introgressed mitochondria from other sympatric species with more stable distributions (Petit &
Excoffier 2009; Keck & Near 2010). Finally, our results from MIGRATE-N and IMA2 do support low levels of historical gene flow from A. barbouri into A. texanum which is counter to reports of these species being strongly isolated from each other even when in close proximity (Kraus & Petranka 1989). While having separate breeding habitats could limit the chances of hybridization between these species, anti-predator adaptations against fish predation on A. barbouri may reinforce the reproductive barrier between A. texanum and A. barbouri (Storfer & Sih 1998).
A unidirectional pattern of mtDNA introgression from A. barbouri into A. texanum is biologically likely for two reasons. First, even though these species breed in different habitats, there are more observations of A. barbouri using A. texanum habitat than the opposite. While A. barbouri primarily breed in headwater streams, there are multiple accounts of A. barbouri breeding in ponds (Kraus & Petranka 1989; Venesky & 34 Parris 2009). In contrast, there are fewer reports of A. texanum breeding in streams
(Petranka 1984), supporting a higher likelihood that the original source of genetic introgression was from A. barbouri individuals invading A. texanum populations.
Secondly, the dispersal of A. barbouri into the range of A. texanum could be explained by a combination of habitat connectivity and reduced landscape resistance. The northeastern extent of the A. barbouri range lies within the same major river drainage (The Scioto
River) of the “Unknown” samples which may have provided a likely corridor for the movement of A. barbouri individuals.
Mechanisms of mitochondrial introgression
Multiple processes have been hypothesized to be responsible for cases where mtDNA has introgressed from one species into another and this work takes a novel approach to testing these hypotheses. Most mechanisms fall in to the categories of adaptive introgression, demographic differences, and sex-biased asymmetries (Toews &
Brelsford 2012). While patterns of mitonuclear discordance have been identified across many taxa, few studies have explicitly linked a pattern of discordance to a particular process. Instead, many authors have proposed mechanisms as determined by the geographic patterns of discordance (extent of mtDNA introgression, frequency of introgressed haplotype) and characteristics of the focal species (sex determination, mating strategies, relative abundances). In this light the mitochondrial introgression in A. barbouri and A. texanum is unusual in the distance within the range of A. texanum that the A. barbouri-like haplotypes have spread. When foreign mtDNA haplotypes appear at a distance >50% of the total range, these foreign haplotypes tend to be at fixation, 35 suggesting an adaptive introgression of mtDNA (Quesada et al. 1999; McGuire et al.
2007; Melo-Ferreira et al. 2009; Brelsford et al. 2011). In the case of these two
Ambystoma species, A. barbouri-like mtDNA haplotypes have been detected in far less than 50% of the range of A. texanum, but the distance from the nearest area of sympatry that these introgressed haplotypes are found is relatively large (~150 km). While the geographical extent of discordance suggests that the mtDNA haplotypes may provide an adaptive advantage, the frequency of introgressed haplotypes is not near fixation as one would predict. Recent preliminary sampling within the transitional gradient of mtDNA haplotypes (Crawford County) identifies wetlands with ~50-75% introgressed haplotypes less than 2 km from wetlands with 100% A. texanum haplotypes (Denton, unpublished data).
Although adaptive introgression has been demonstrated in other amphibians
(Pfennig 2007; Fitzpatrick et al. 2010), determining the adaptive value of introgressed mtDNA is difficult (Toews & Brelsford 2012; Toews et al. 2013). Ambystoma texanum individuals with A. barbouri-like mtDNA were significantly more likely to be present at localities with higher levels of precipitation. Even though the average difference in annual precipitation of sites with mitonuclear discordance was small (~3% of an average year’s total), the statistical significance of this pattern within such a small geographic area lends support to this being a real biological phenomenon. Higher levels of precipitation at the sites were mitonuclear mismatch is present suggesting a role for differences in moisture in the environment as a driver of selection for the A. barbouri mtDNA haplotype populations of A. texanum. One potential explanation could involve the temporal components of each species’ breeding strategies. Ambystoma texanum are 36 explosive breeders that rely on the sudden filling of temporary wetlands in the spring, while A. barbouri breed during an overlapping period of 4-5 months from December-
April (Petranka 1984). Because of the semi-permanence of A. barbouri breeding streams, they may be more adapted to the wetter environment of stream sides and A. texanum with
A. barbouri-like mtDNA are limited to wetland environments with higher precipitation.
While this is not an adaptive advantage of having A. barbouri-like mtDNA, precipitation variables may predict the extent of introgression. This association with wetter environments does not exclude the potential of some other beneficial property of the introgressed haplotypes for which the limitation of wetter environments is a trade-off.
While more investigation is needed to determine the process behind the mtDNA introgression, recent studies have successfully uncovered the adaptive significance of introgressed mtDNA haplotypes, especially with more recent techniques to assay mitochondrial metabolism and efficiency (Ruiz-Pesini et al. 2004; Moyer et al. 2005;
Grant et al. 2006; Toews et al. 2013).
Demographic differences and sex-biased asymmetries between A. texanum and A. barbouri provide less convincing explanations for the mitonuclear discordance between these species that is observed in central Ohio. A large shift in the range of A. barbouri that would leave behind a wake of mtDNA (Rohwer et al. 2001) is unlikely due to the environmental specificity of A. barbouri. Large discrepancies in relative abundance between species that have influenced introgression in other systems (Chan & Levin 2005;
Linnen & Farrell 2007) are also unlikely due to the narrow range of sympatry and the separation of breeding habitats for each species. Another potential mechanism responsible for mitochondrial introgression would be an extension of Haldane’s rule 37 (Haldane 1922), which predicts that during hybridization, the heterogametic sex is most likely to suffer a fitness loss. In a XY sex determination system, this would predict a higher fitness for females. However, Ambystoma display a ZW sex determination system
(reviewed in Hillis & Green 1990). This contradicts the observed pattern of introgressed mitochondria in A. texanum, but conclusions are difficult to make due to a lack of clarity concerning the sex determination system in Ambystoma and amphibians as a whole (see
Robertson et al. 2006). Finally, female-biased dispersal could potentially initiate mitochondrial introgression, but there is no support for sex-biased dispersal in
Ambystoma (Trenham et al. 2001). If any demographic differences have influenced the mitochondrial introgression between A. texanum and A. barbouri, it may be the asymmetrical behavioral reproductive isolation described above. Because breeding densities of A. barbouri would be predicted to be lower than A. texanum due to a longer breeding season, A. barbouri females may be less likely to discriminate against a male A. texanum.
Conclusions
Mitonuclear discordance is a widespread phenomenon that is likely an important force in the shaping of genetic diversity between species. Our work makes three general contributions to the study of this process in natural populations. First, it provides an example of a comprehensive methodological framework for investing this phenomenon that that is based on a diverse set of approaches. In particular, Table 1 provides a model testing framework in which specific results from different analyses can be used to infer the processes underlying mitonuclear discord in any animal. Second, our results provide 38 an example of the extent to which species boundaries are genetically permeable and a possible example how selection acting through environmental variation may constrain mitochondrial introgression between species (Ballard & Melvin 2010). Finally, our results provide yet another caution of the sole use of mtDNA for species identification
(e.g. DNA barcoding) in taxa with poorly known geographic distributions (Rubinoff
2006).
39 Chapter 3: Locomotor endurance predicts differences in realized dispersal between sympatric sexual and unisexual salamanders
Note: This chapter has been published as below, and benefitted from contributions of the co-authors:
Denton RD, KR Greenwald, and HL Gibbs. 2017. Locomotor endurance predicts
differences in realized dispersal between sympatric sexual and unisexual salamanders.
Functional Ecology. 31:915-926.
Abstract
Dispersal is the central mechanism that determines connectivity between populations yet few studies connect the mechanisms of movement with realized dispersal in natural populations. To make such a link, we assessed how physiological variation among individuals predicted dispersal in natural populations of unisexual (all-female) and sexual Ambystoma salamanders on the same fragmented landscape in Ohio. Specifically, we assessed variation in a trait that influences long-distance animal movement
(locomotor endurance) and determined whether variation in endurance matched patterns of realized dispersal assessed using genetic assignment tests. A possible mechanism for
40 why unisexuals would have lower locomotor endurance than a sympatric sexual species
(A. texanum) is the potential energetic cost of evolutionarily-mismatched mitochondrial and nuclear genomes within polyploid unisexuals. We found that sexuals walked four times farther than unisexuals during treadmill endurance trials that mimic the locomotor endurance required for dispersal. We then applied landscape genetic methods to identify dispersed adults and quantify realized dispersal. We show that the differences in locomotor endurance between unisexual and sexual salamanders scale to realized dispersal: dispersing sexual individuals traveled approximately twice the distance between presumed natal wetlands and the site of capture compared to dispersing unisexuals. This study links variation in individual performance in terms of endurance with realized dispersal in the field and suggests a potential mechanism (physiological limitation due to mitonuclear mismatch) for the reduced endurance of unisexual individuals relative to sexual individuals although we discuss other possible explanations.
The differences in dispersal between these two types of salamanders also informs our understanding of sexual/unisexual coexistence by suggesting that unisexuals are at a competitive disadvantage in terms of colonization ability under a extinction-colonization model of coexistence.
Introduction
Dispersal differences within and between species shape patterns of diversity from local to range-wide scales (Berdahl et al. 2015). Integrating dispersal rates or distances with the phenotypes that drive these patterns is an important step toward understanding how species influence each other’s dispersal behavior (Fronhofer et al. 2015) and how 41 these differences produce patterns of biodiversity across landscapes (Lowe & McPeek
2014). Intraspecific differences in dispersal behavior can scale up to shifts in species distributions (Bestion et al. 2015), and dispersal asymmetries between species can have strong effects on the outcomes of competition (Amarasekare 2003). Despite the impact that dispersal has on evolutionary and ecological processes, there is little data that connects variation in traits that facilitate dispersal of individuals with actual movement of individuals among populations. However, there are a multitude of behavioral, morphological, physiological, and genetic traits that can influence such movements
(Bowler & Benton 2005; Nathan et al. 2008). Understanding how specific traits contribute to successful dispersal events is critical to predicting how changes in environment, phenotypic diversity, and species interactions may alter species persistence through time (Salomon et al. 2010).
Studies that link the physiology of dispersal with observed dispersal patterns in the wild can provide important ecological insights. For example, differences in metabolic rate and locomotion propensity can account for differences in dispersal among demes in the Glanville fritillary butterfly (Melitaea cinxia; Hanski 2012). Studies of the spread of invasive Cane Toads (Rhinella marina) across Australia have also shown how greater dispersal ability is correlated with specific phenotypes (Shine et al. 2011). For instance, rapidly-dispersing range-edge toads have longer limbs (Phillips et al. 2006b), greater locomotor endurance (Llewelyn et al. 2010), and upregulation of genes involved in metabolism and cell repair (Rollins et al. 2015). These toads have also revealed patterns that are contrary to predictions of expected dispersal for specific phenotypes. For example, there is a lack of cellular metabolic differences between toads with differing 42 dispersal ability (Tracy et al. 2012) and physiological performance may not be related to the magnitude of in situ dispersal (Olson & McPherson 1987). Physiology is clearly important for the determination of dispersal ability, but there are few studies that link differences in physiological phenotypes between competing species with their realized capacity for dispersal (Salomon et al. 2010)
Here, we provide an integrated assessment of both a mechanism of dispersal and realized natural dispersal within sympatric populations of sexual and unisexual salamanders (genus Ambystoma). Unisexual Ambystoma salamanders are the oldest lineage of unisexual vertebrate and reproduce through kleptogenesis, in which polyploid unisexual female salamanders produce clonal offspring after stimulation from a sexual male’s sperm but can occasionally “steal” sperm from the males of congeneric sexual salamanders (Bogart et al. 2007; Gibbs & Denton 2016). The result of this mating strategy is a single, distinct mitochondrial lineage combined with 2-5 haploid nuclear genomes from other sexual Ambystoma species (A. laterale, A. jeffersonianum, A. texanum, A. tigrinum, A. barbouri; Robertson et al. 2006; Bi & Bogart 2010a). The majority of unisexuals contain nuclear genomes from A. laterale and A. jeffersonianum, whereas the mitochondrial genome of unisexuals is most closely related to A. barbouri, an extremely rare sperm donor to the lineage (Robertson et al. 2006; Bogart & Klemens
2008). Unisexuals are currently widespread across northeastern North America, and their range corresponds to large areas of agricultural land (Bogart et al. 2007; Bogart &
Klemens 2008), which are some of the most challenging environments for salamander movement and persistence (Compton et al. 2007; Greenwald et al. 2009a).
43 A compelling reason why there may be differences in dispersal ability in unisexual salamanders compared to related sexual species is that unisexuals are a lineage that shows cyto-nuclear discordance similar to that observed in experimentally-generated nucleocytoplasmic “cybrids”, which combine the mitochondrion of one species with the nuclear genome of another (Narbonne et al. 2011). Specifically, the nuclear genomes within unisexual individuals are more evolutionarily distinct from their mitochondrial genomes than those found in sexual species because the unisexual mitochondrial genome is most closely related to a species (A. barbouri) whose nuclear genomes are extremely rare in unisexuals (see above). Genome exchange between sexual males and unisexual females is relatively common (Gibbs and Denton 2016), and therefore presents more opportunity for coevolution between heterospecific mitochondrial and nuclear genomes compared to strictly asexual taxa. But even small levels of cyto-nuclear mismatch may have a significant impact on physiological processes involved in energy production at the cellular level. A higher likelihood of molecular mismatch within protein complexes that require coding information from both mitochondrial and nuclear genomes in unisexuals and a resultant reduction in the efficiency of ATP production (Harrison & Burton 2006).
Although there is evidence that mitochondrial introgression can also be associated with greater mitochondrial respiration (Toews et al. 2013), mitonuclear mismatch often causes oxidative stress (Monaghan et al. 2009) and can lead to generalized physiological limitations (Wolff et al. 2014)
We assessed potential and realized differences in dispersal capacities between unisexual Ambystoma and a closely related, sympatric sexual species (Small-mouthed salamander, A. texanum, Figure 6) using a two-step procedure. First, we measured 44 walking endurance using treadmill trials on wild-caught individuals from a fragmented landscape in central Ohio. Second, we confirmed differences in endurance by using genetic assignment tests to identify dispersed animals at the same field sites (reviewed in
Broquet & Petit 2009). Together, these methods can connect the capacity to disperse with the distribution of dispersed animals on a landscape in relation to their natal population, linking a potential mechanism for differential dispersal to the patterns in realized dispersal inferred from genetic data. These results also have important implications for understanding mechanisms of coexistence between sexual and unisexual forms
(Hellriegel & Reyer 2000).
Figure 7. Unisexual Ambystoma salamander (top) and Small-mouthed Salamander (A. texanum, bottom).
45
Materials and Methods
Locomotor Endurance
We collected 38 individuals (17 A. texanum, 21 unisexuals) from five sites within a largely agricultural landscape in Crawford County, Ohio (~200 km2) during spring
2015. The mean number of individuals per site was 4.25 (range: 3-5; sites C10, C13, C22,
C29) for A. texanum and 7 for unisexuals (range: 6-9; sites C13, C29, C1303). All individuals were acclimated individually in a cold room kept at 13° C and fed three adult crickets weekly for a month. In addition to sympatric unisexual and A. texanum individuals, we included two additional sympatric, sexual species, A. jeffersonianum (N =
5) and A. laterale (N = 2), that had been held captive in the same conditions as the wild- caught animals since 2010. Because of the potential confounding factors associated with the length of captivity (diet quality, lack of seasonality, acclimation to human interaction), we do not include these two sexual species in any statistical analysis.
However, the data from A. jeffersonianum and A. laterale individuals are useful for providing a qualitative comparison in levels of endurance between the unisexuals and their parental, sexual species that constitute their nuclear genomes.
We conducted treadmill endurance trials at the same acclimation temperature following the protocol of Johnson et al. (2010). Briefly, each randomly-selected animal was fed three days before their trial, then weight, snout to posterior vent (SVL) length, and femur length was recorded. The treadmill used for trials was the same as used by
Johnson et al. (2010), and rotation speed was maintained at a near constant speed that matched the walking speed of each individual salamander. A metal spatula was used to 46 gently prod animals in order to maintain initial walking speed. Every three minutes, the animal was removed from the treadmill and tested for fatigue by a righting response test
(Johnson et al. 2010). If an animal could not right itself after three seconds, the trial was terminated. If an animal refused to continue after 10 minutes of gently tapping or pinching the tail, the trial was concluded. After a successful righting response, the animal was rehydrated with a spray bottle and returned to the treadmill. Distance traveled was calculated using the speed maintained by each individual multiplied by the duration walked and an Analysis of Covariance (ANCOVA) was conducted using weight and femur length as covariates and distance traveled as the response variable.
Collection of Genetic Data
We collected salamander tissue samples from the same sites described above and all other breeding wetlands in the ~200 km2 section of southwestern Crawford County,
Ohio to infer dispersal distance in wild individuals using genetic assignment tests (Berry et al. 2004). To improve the confidence associated with identifying dispersed animals using genetic data (Cornuet et al. 1999), we chose well-documented sites that have been previously mapped (Weyrauch and Grubb 2004) such that all known breeding sites within the study area were sampled. All sites were characterized as vernal wetlands embedded in small woodlots (< 1 km2). We surveyed the 28 sites, including those above, identified as amphibian breeding habitat by Weyrauch & Grubb Jr. (2004) over a four year period (2012-2015). In 2012, we visited all sites to confirm the presence of wetland habitat and conduct preliminary salamander surveys. From 2013-2015, we sampled each site in at least two consecutive years using aquatic minnow traps and constrained 47 searches during the early spring. We sampled each wetland over multiple years to avoid effects from potential relatedness within breeding cohorts (Semlitsch et al. 1996). For analysis, we chose only sites in which salamanders were detected over multiple seasons and where at least two consecutive nights of trapping yielded >20 individuals of either group.
Genetic Analyses
We extracted genomic DNA from the tail tips of 294 A. texanum and 151 unisexual salamanders using Qiagen DNeasy kits (Qiagen, Velencia, CA). We identified unisexual individuals based on longer snouts, slimmer bodies, and longer limbs in comparison to the primary sperm-donor species of the area, A. texanum (R. Denton, unpublished data). We confirmed the field identity of the first 200 individuals using a 346 bp section of control region mtDNA (THR, McKnight & Shaffer 1997). We correctly confirmed the identity of all 200 individuals, and so further samples were classified based only on field identification. We amplified 10 species-specific microsatellite loci for A. texanum using the PCR conditions recommended by Williams & DeWoody (2003)
(Supplemental Table 8) and determined the ploidy and genome composition of unisexual samples using a single nucleotide polymorphism (SNP) assay (Greenwald & Gibbs
2012). After we identified the genomes present in the unisexual individuals, we amplified a combination of species-specific microsatellite loci designed for A. laterale and A. jeffersonianum (Julian, King & Savage 2003; Denton, Gibbs & Glenn 2015,
Supplemental Table 8). A total of 19 loci were amplified: six specific to A. laterale, nine specific to A. jeffersonianum, and two that amplify in both species at different size 48 ranges. The genotypes for unisexuals were determined using the ploidy expectation from the SNP assay. If an animal had more than two haploid genomes from either A. laterale or A. jeffersonianum, alleles were scored based on comparative peak height with any ambiguity coded as missing alleles. The final genotype for a unisexual individual consisted of the combined alleles from all haploid genomes (A. jeffersonianum and A. laterale). All loci were scored using Geneious (v 7.1.8; Drummond et al. 2012).
We tested for null alleles among the A. texanum loci using MicroChecker (v 2.2.3;
Van Oosterhout et al. 2004). We performed tests for linkage and Hardy Weinberg equilibrium (HWE) on the A. texanum loci using GenePop (v 4.2, Raymond & Rousset
1995; Rousset 2008). Because unisexuals are polyploid, population-level genetic data do not meet the assumptions of most analytical methods that are based on population genetics theory. As such we used assumption-free, multivariate genetic analyses within the statistical package adegenet (v 1.4-1; Jombart 2008) to genetically cluster putative populations and identify individuals dispersed from their natal populations. Populations were described using the find.clusters function and evaluated using BIC scores and by evaluating plots of population assignment for all individuals.
We conducted a discriminant analysis of principle components (DAPC; Jombart et al. 2010) to produce posterior probabilities of individual assignment to the sampled sites (Kraus et al. 2013; Gotzek et al. 2015). The DAPC procedure consists of a principle components analysis (PCA) as a prior step for a discriminant analysis (DA). This allows for the discrimination of individuals to pre-defined groups using the simplified and uncorrelated variables produced by a PCA. Most importantly, this procedure remains assumption-free (HWE or LD) and can be used to infer the group assignment of mixed 49 ploidy (James et al. 2013). To determine an optimal number of principle components
(PCs) for the DAPC, we used both the optim.a.score and the xval.dapc procedures. The retained PCs were then used for a final DAPC run. To choose a posterior probability threshold for genetic assignment of a dispersed individual, we followed the guidelines developed for assigning birds to natal populations using stable isotope data (Rocque et al.
2006; Wunder 2012). Specifically, posterior probability thresholds for assignment to the groups identified in the find.clusters analysis ranged from 0.55 to 0.99, which translate to odds ratios from 1.7:1 to 198:1. Jonker et al. (2013) refer to a posterior probability of
>0.3 as evidence of a recent instance of gene flow in migratory waterfowl using discriminant analysis on genetic data. Using the sample sizes in our study for A. texanum
(N = 294) and unisexuals (N = 151), we choose a posterior probability threshold of 0.70.
This translates to odds ratios of 28.8:1 and 16.6:1 for A. texanum and unisexuals, respectively; meaning that an individual that is assigned to a population other than the site where it was sampled at a posterior probability of 0.7 is 28.8 or 16.6 more likely than at random. To validate the use of DAPC for detecting migrant individuals, we also conducted a detection of first generation migrants test in Geneclass2 (Piry et al. 2004) using A. texanum individuals and compared these results with those obtained using
DAPC.
Landscape Analyses
To identify biologically-relevant dispersal distances for each individual identified as a disperser , we constructed least-cost paths (LCPs) among assigned and sampled sites using the Landscape Genetics Toolbox developed for ArcGIS 9.3 (Etherington 2011). 50 This requires a base layer indicating the relative cost of each landcover type. To generate such a layer, we used LANDSAT imagery based on 30m-resolution landcover data from the National Land Cover Database (NLCD 2011; Homer et al. 2015), with encoded costs based on empirical data (Compton et al. 2007). The toolbox then calculates every pairwise route among sites in such a way as to minimize the accumulated cost of the journey, and then provides the length of these cost-minimizing paths. Least-cost paths constructed using landscape resistance values derived from natural history knowledge have outperformed other geographic distances in their ability to fit patterns of genetic structure (Michels et al. 2001; Coulon et al. 2004). However, this has not been true in all cases (e.g., Charney 2012a), and there is ongoing discussion about the optimal way to parameterize such models (Zeller et al. 2012). Therefore, we also calculated straight-line
(Euclidean distance) as an assumption-free means to assess pairwise distances among sites.
Results
Locomotor Endurance
Salamanders showed pronounced differences in performance during treadmill trials. Unisexual individuals had significantly longer femurs and body length than A. texanum individuals (Femur length: t = -2.49, p = 0.018; SVL: t = -4.00, p < 0.001), whereas the groups were not statistically different in weight (t = -1.74, p = 0.09, Table 2).
We converted the duration of each trial into a distance based on the speed of the treadmill, which stayed relatively consistent for both groups at a rate of 19.4 seconds/rotation. We log-transformed the distance values to correct for non-normality. 51 All covariates measured (weight, SVL, and femur length) were significantly, positively correlated with one another (Pearson correlation range = 0.447 – 0.673, all p < 0.006,
Supplemental Figure 17). Therefore, we only retained the covariate with the largest correlation with distance (SVL) in further analyses. After controlling for SVL, A. texanum had significantly greater walking endurance compared to unisexuals (F = 29.0, p
< 0.001, Figure 7). Unisexuals traveled approximately 25% of the average distance travelled by A. texanum individuals (159.25 ± 86.4 m for A. texanum, 34.47 ± 28.2 m for unisexuals). Unisexual individual performance did not vary by genome composition, as the single individual composed of one genome from A. laterale, A. texanum, and A. jeffersonianum (abbreviated at LTJ) was near the mean for the group (28.9 m traveled).
Individuals of two additional species that represent the composite genomes of the unisexuals (A. jeffersonianum and A. laterale) showed similar treadmill endurance to the
A. texanum. Finally, all but two unisexual trials (90%) were concluded following a lack of righting reflex. In contrast, only three A. texanum individuals (18%) lacked righting response at the end of their trials and were instead concluded due to a >10 min refusal period.
52 Figure 8. Univariate plots (Weissgerber et al. 2015) for the distance travelled by dispersed sexual salamanders (circles) and unisexual salamanders (triangles) as measured by Euclidean distance (A) and the log Distance travelled during locomotor endurance trials by unisexuals and three sexual species (B). Each point represents one individual and black horizontal bars represent mean values. Standard errors for (A) are 22.25 for A. texanum and 6.31 for Unisexuals. Standard errors for (B) are 0.07, 0.07, 0.18, 0.03 for A. texanum, Unisexuals, A. jeffersonianum, and A. laterale.
Table 2. Distances traveled during locomotor endurance trials by unisexual and sexual salamanders collected from sites within Crawford County Ohio with associated morphological measurements. Group Means ± one standard deviation
Weight Snout-Vent Length Femur length Distance Walked
Group N (g ± SD) (mm ± SD) (mm ± SD) (m ± SD)
A. texanum 14 10.05 ± 1.97 63.6 ± 5.0 5.73 ± 0.79 159.25 ± 86.4
Unisexual Ambystoma 19 11.42 ± 0.60 70.7 ± 6.0 6.36 ± 0.73 34.47 ± 28.2
A. laterale 2 6.18 ± 0.88 63.0 ± 5.0 4.46 ± 0.23 161.20 ± 16.09
A. jeffersonianum 4 8.47 ± 0.70 61.0 ± 4.2 5.54 ± 0.87 172.21 ± 133.49
53 Genetic Analyses of Dispersal
Salamanders were collected from 16 and 9 sites for A. texanum and unisexuals, respectively. Samples sizes per site ranged from 10-30 individuals for A. texanum and from 8-24 individuals for unisexuals (Table 3). Unisexuals and A. texanum were the two most commonly-captured salamander species at the majority of sites, but both A. maculatum (Spotted Salamander) and A. tigrinum (Tiger Salamander) were detected in a minority of sites. While A. maculatum were abundant when present, this species does not contribute to the unisexual complex. In contrast, A. tigrinum does occasionally contribute genomes to unisexuals, but this species was rare in surveys (less than twenty individuals captured over four years).
For A. texanum, two loci showed evidence for null alleles in more than half of the populations (t133 and t87). However, we retained these loci in further analyses based on the robustness of assignment testing (Carlsson 2008) and the lack of assumptions made by DAPC assignment methods. No loci showed evidence for linkage disequilibrium based on a Bonferroni-adjusted critical P value (0.005). For A. texanum, the mean number of alleles (± SE) was 8.74 ± 0.32 (Ho = 0.55, He = 0.75). For unisexuals, the mean number of alleles was lower for A. laterale loci (1.59 ± 0.114) compared to A. jeffersonianum loci (3.58 ± 0.15). Mean pairwise Fst for A. texanum among all populations was 0.08.
54 Table 3. Number of individuals sampled from each site and the proportion of individuals assigned to the site at which they were sampled. Proportion assigned Site Group N to sampled site C03_C10 A. texanum 46 0.87 C13 A. texanum 20 0.85 unisexual 19 1.00 C22 A. texanum 20 0.60 unisexual 8 0.88 C29 A. texanum 17 1.00 unisexual 15 0.53 C57_C58 A. texanum 33 0.85 C60 A. texanum 18 0.56 unisexual 20 0.95 C61 A. texanum 17 0.59 C63 A. texanum 20 0.40 C63_C64 unisexual 36 0.94 C64 A. texanum 10 0.20 C77 A. texanum 19 0.37 unisexual 9 0.89 C84 A. texanum 19 0.16 C86 A. texanum 16 1.00 C96A A. texanum 19 0.58 unisexual 20 0.85 C99 A. texanum 20 0.35 unisexual 24 1.00
The BIC scores produced by the find.clusters procedure in adegenet suggested fewer genetic groups than the number of sampled sites (6 groups from 16 populations for
A. texanum; 5 groups from 9 populations for unisexuals). We evaluated each sampled population’s assignment probabilities and chose to combine populations that were closest in geographic distance, had extensive overlap in the assignment of individuals, and had pairwise Fst < 0.04 (suggested Fst for number of loci from Paetkau et al. [2004]). This resulted in four of the total populations being collapsed into two new sites for A. texanum
(“C03_C10” and “C57_C58”) and one newly-combined site for unisexuals (“C63_C64”).
55 Both sexual and unisexual salamanders showed similar percentages of individuals that were identified as dispersers. For A. texanum, the DAPC procedure accounted for 60% of the variance in the dataset and assigned 64% (188/294) of the total individuals to a population (28 PCs and 6 DFs retained). The percentage of individuals from a given population that were assigned to their sampled site (non-dispersers) ranged from 16-
100%. Thirteen individuals (4% of total; Table 4) were assigned to a population other than where they were sampled (disperser) with a posterior probability of ≥ 0.70 (4% of total, posterior probabilities = 0.70-0.94). Seven of these thirteen were also identified as migrants by Geneclass2 (p < 0.01) and 50% of individuals were confidently assigned to a population. For unisexuals, the DAPC procedure accounted for 79% of the variance in the dataset and assigned 90% of the total individuals (136/151) to a population (14 PCs and 6 DFs retained). The percentage of individuals assigned to their sampled site (non- dispersers) ranged from 50-100%. Eleven individuals were identified as dispersers (7% of total, posterior probabilities = 0.83-1.00, Table 4). Nine of the eleven unisexual dispersers were successfully genotyped, and the majority of all sampled individuals were tetraploids consisting of a single A. laterale genome and three A. jeffersonianum genomes
(LJJ = 32, LJJJ = 54, LTJ = 16, LTJJ = 11). Three of the nine dispersers also included a single genome derived from A. texanum, but the proportion of A. texanum genomes within the identified dispersers was similar to that of the total population (0.33 vs 0.24, respectively).
56 Table 4. Individuals identified as dispersers (mismatch between sampled site and genetically-assigned site). Discriminant Analysis of Principal Components (DAPC) was used to identify animals with a ≥ 0.70 posterior probability of assignment and GeneClass2 was used to identify first generation migrants. 57
Population Distance between sampled and assigned population (m) Assigned Assigned Least-cost path Least-cost path Individual Group Biotype Sampled (DAPC) (Geneclass2) Euclidean (canopy cover) (NLCD) A1977 A. texanum C03_C10(C10) C57_C58 C57_C58 8,013 9,520 10,111 A1980 A. texanum C03_C10(C10) C57_C58 8,013 9,520 10,111 A1549 A. texanum C22 C61 7,743 9,097 14,914 A1964 A. texanum C57_C58(C58) C03_C10 6,407 7,061 7,504 A2324 A. texanum C60 C96A C96A 2,474 2,830 2,989 A2365 A. texanum C61 C03_C10 C03_C10 9,275 11,511 11,686 A2366 A. texanum C61 C57_C58 C57_C58 5,108 5,276 5,850 A2367 A. texanum C61 C03_C10 C03_C10 9,275 11,511 11,686 A2516 A. texanum C61 C86 3,578 4,080 3,997 A2255 A. texanum C84 C57_C58 8,299 14,058 14,358 A2259 A. texanum C84 C57_C58 C57_C58 8,299 14,058 14,358 A2475 A. texanum C84 C86 C99 2,020 2,529 2,473
58 A2424 A. texanum C96A C29 10,228 12,010 18,998 A1510 Unisexual LJJJ C29 C22 2,390 2,651 2,693 A1511 Unisexual LTJJ C29 C22 2,390 2,651 2,693 A2013 Unisexual LJJJ C29 C13 2,718 6,383 5,865 A2015 Unisexual LJJJ C29 C13 2,718 6,383 5,865 A2016 Unisexual LJJJ C29 C13 2,718 6,383 5,865 A2017 Unisexual LJJJ C29 C13 2,718 6,383 5,865 A2539 Unisexual C63_C64(C63) C99 4,149 5,944 5,740 A2418 Unisexual LJJ C63_C64(C64) C22 9,429 15,165 15,399 A2511 Unisexual C77 C63_C64 2,118 2,286 2,260 A2445 Unisexual LTJJ C96A C60 2,474 2,830 2,979 A2446 Unisexual LTJ C96A C60 2,474 2,830 2,979
Landscape Analyses
Sexual salamanders travelled farther between their sampled site and their presumed natal pond when compared to unisexual salamanders. The average Euclidean distance between sites was significantly farther among A. texanum dispersers (N = 13, mean = 6826 m) compared to unisexual dispersers (N = 11, mean = 3300 m; t = 3.5, p <
0.01, Table 4). The least-cost paths generated based on landscape resistance values for amphibians (Compton et al. 2007; Greenwald et al. 2009b) displayed a similar difference between groups (A. texanum mean = 9926 m, unisexual mean = 5291 m, t = 2.5, p = 0.02,
Figures 7, 8). However, due to the homogeneous landscape of the study area, many inferred individual routes favored the use of roads as corridors in order to avoid travelling across row crop agriculture. To address this biologically-unrealistic result, we built additional LCPs using a resistance layer composed of 30-meter resolution percent tree canopy cover (TCC, Homer et al. 2015). This resistance layer produced LCPs that were more likely to reflect travel among drainage ditches and tree lines but still maintained the same relationship between groups (A. texanum mean = 8697 m, unisexual mean = 5444 m, t = 2.0, p = 0.05). However, this change in resistance layer still produced LCPs that followed roads to avoid agricultural land cover, even suggesting the unlikely scenario of unisexual animals traversing urban areas.
59 Figure 9. Predicted least-cost paths traveled by genetically-identified dispersers (top: A. texanum, bottom: unisexual Ambystoma). Paths were constructed by using resistance surfaces derived from canopy cover (orange lines) or using national land cover database values from other Ambystoma species (yellow lines). All lines are scaled by the number of individuals that moved between the same wetland pairs.
Discussion
Unisexual Ambystoma salamanders are inferior dispersers compared to a sympatric sexual species, A. texanum. This conclusion was supported by both the 60 superior endurance by sexual salamanders in treadmill trials and the greater distance from home populations displayed by sexual dispersers. If only the locomotor trials were considered, it would be unclear if the differences between the groups translated to differential dispersal and subsequent breeding in natural populations, but here we demonstrate such a link. The addition of genetic data to the locomotor endurance measurements provides strong support for a physiological basis to dispersal asymmetry between these taxa. The unisexuals’ limitation documented here is consistent with negative impacts on cellular metabolism due to mismatches between mitochondrial and nuclear-encoded proteins in unisexuals compared to sexuals, although other mechanisms are possible (see below). This study adds to the growing literature (see Bowler & Benton
2005; Lowe & McPeek 2014) that emphasizes the complexity of animal dispersal and the requirement for integrative approaches in understanding the mechanisms and ecological and evolutionary consequences of animal dispersal.
Causes of dispersal differences
Despite morphological features that favor locomotor ability (larger body size, longer limbs; Bennett et al. 1989), unisexuals performed poorly in treadmill endurance trials compared to A. texanum. While this difference in endurance could be due to the influence of either A. jeffersonianum or A. laterale genomes within unisexuals, trials with captive adults of each parental species closely resemble those of the A. texanum individuals. Additionally, A. texanum individuals travelled similar distances as another sexual species, A. tigrinum, using the same treadmill methodology (Johnson et al. 2010).
Further, these trials likely underrepresent the true endurance capacity of A. texanum due 61 to the large proportion (82%) of individuals that maintained a righting response but refused to continue a trial, suggesting that the majority of A. texanum individuals had the capacity to walk for even greater distances.
Given the similar performance by the three sexual species in this study and one
Ambystoma species from other work, the reduction of locomotor endurance that is specific to the unisexuals could potentially be related to limitations of cellular metabolism imposed by their unique mode of reproduction that reduce the efficiency of the oxidative phosphorylation (OXYPHOS) system due to a mismatch in proteins separately encoded by mitochondrial and nuclear genes (Lane 2011). Mitochondrial- and nuclear-encoded subunits that produce basic energetic functions such as the OXYPHOS pathway show evidence for positive selection that indicates coevolution (Gershoni et al.
2010; Zhang & Broughton 2013). Mitochondrial replacement experiments in both rodents
(McKenzie et al. 2003) and primates (Kenyon & Moraes 1997) show a negative correlation between cellular respiration and the phylogenetic distance between donor mitochondria and recipient cells. The evolutionary mismatch between mitochondrial and nuclear genomes of unisexual salamanders is expected to be high: all unisexual individuals harbor a phylogenetically-distinct mitochondrial haplotype (Robertson et al.
2006; Bogart et al. 2007) compared to the species from which they derive their nuclear genomes (Bi & Bogart 2010b; Gibbs & Denton 2016). This mismatch may produce a physiological disadvantage that manifests here through limited locomotor endurance compared to sexual species. The quantification of the mitonuclear mismatch at a genotype level in unisexual salamanders and the relationship between mitonuclear
62
mismatch in unisexuals and impacts on their physiology are important goal for future research.
While mitonuclear mismatch is a plausible explanation for dispersal differences, other mechanisms not explored in this study could also account for this difference. Two possibilities are 1) differences in behavioral motivation to disperse or 2) differences in other life history characteristics that promote dispersal. In terms of motivation, the environmental cues or genetic basis for dispersing is not well understood in salamanders
(Semlitsch 2008), and we have no evidence for differences in salamander density between our study sites that could lead to greater density-dependent dispersal effects on sexuals (Bitume & Bonte 2014). However, it is possible that there are unmeasured behavioral variables that may help explain the discrepancy in realized dispersal between
A. texanum and the unisexuals. Differences in realized dispersal could also be affected by other life history differences between unisexuals and sexuals. For example, unisexuals could have a greater site fidelity compared to A. texanum. However, this is unlikely because most Ambystoma species show similar levels of site fidelity (Petranka 1998;
Gamble et al. 2007). Alternatively, this result could be an artifact of sex biased dispersal.
All but one dispersed A. texanum were male, and unisexuals may share the dispersal behavior of female A. texanum. However, both unisexuals and A. texanum showed a similar proportion of dispersed individuals (7% and 4% for unisexuals and A. texanum, respectively). This indicates that, if present, the effect of sex is more likely associated with dispersal distance and not frequency. However, differences in dispersal ability between sexes seems to be driven by morphological dimorphism that affects locomotion or the additional weight of reproductive females that are carrying eggs (Bowler & Benton 63 2005). In the case of our study animals, male and female Ambystoma are not sexually dimorphic beyond a slightly swollen cloacal area in males during the breeding season and both types of females used in the locomotor trials were collected after depositing eggs when such differences are minimal. Finally, the connection between locomotor endurance and realized dispersal distance may be obscured by potential performance differences between unisexuals and A. texanum on realistic landscapes that may not be represented by the wet treadmill (e.g. grass, forest, soil). The direction of this bias is not understood because there are currently no comparative studies between species (Stevens et al. 2004;
Lee-Yaw et al. 2015). Because the landscape of Crawford Country is largely homogeneous (field, road, small forest patches), we expect that there is little opportunity for differential dispersal among substrate types that could account for differences in movement between sexuals and unisexuals.
Dispersal in Ambystoma salamanders
The magnitude of dispersal identified by our genetic methods is similar to those described for other species of Ambystoma salamanders. At spatial scales similar to those in this study, Ambystoma salamanders display high levels of gene flow between populations (Newman & Squire 2001; Zamudio & Wieczorek 2007; Purrenhage et al.
2009; Coster et al. 2015). However, this connectivity is spatially-dependent, with greater population differentiation at larger scales, as would be predicted by an isolation-by- distance model. Large Ambystoma, such as A. tigrinum, can regularly move 1-3km from breeding habitat (Searcy & Shaffer 2008), and other Ambystoma have been documented dispersing > 1 km (Smith & Green 2005; Gamble et al. 2007). Local colonization and 64 extinction of Ambystoma populations can be affected by landscape resistance (Cosentino et al. 2011a), and even different crop species affect movement decisions by salamanders
(Cosentino et al. 2011b). However, A. texanum is mainly associated with open habitats
(fields, bottomland forests, farmland) and displays some of the lowest dispersal distances among Ambystoma (Parmelee 1993; Petranka 1998; Smith & Green 2005). Because small differences in estimated dispersal distance (~350 m; Peterman et al. 2014) can produce large differences in genetic structure of sympatric salamanders, both the magnitude
(>3,000 m for both species) and differences (at least 3,000 m between species) of dispersal distance are significant in the context of previous research. Since the spatial genetic structure of Ambystoma varies both between and within species and is related to various landscape factors or life history characters, this makes comparisons of dispersal ability difficult across studies.
Importantly, no landscape genetics studies of Ambystoma salamanders include the combination of 1) species that are sexually parasitized by unisexuals and 2) geographic areas where unisexuals are present. Further analyses of A. laterale, A. jeffersonianum, and A. texanum in areas of allopatry/sympatry would better reveal if differences in the sexual species dispersal patterns are a consequence of unisexual presence. Finally, it is important to caution that all of the locomotor endurance trials were conducted on adult animals, whereas the majority of dispersal takes place in the juvenile stage (Gamble et al.
2007). Because long-term repeatability in locomotor endurance is low for salamanders
(Austin & Shaffer 1992), our locomotor endurance trials might not represent the life history stage most relevant to dispersal. In contrast, the dispersal differences calculated from genetic data represent realized dispersal unbiased by life stage effects. 65 Coexistence of sexuals and asexuals
If unisexual Ambystoma can obtain the benefits of both asexual and sexual reproduction through kleptogenesis, they would be predicted to quickly outcompete other
Ambystoma species (Lampert & Schartl 2010). However, unisexuals are rarely identified as the only Ambystoma species in amphibian communities across their range (Noël et al.
2011). Primary life history traits that are predicted to dictate the coexistence between a sexual parasite and their host favor sexual Ambystoma, including mate choice (Dawley &
Dawley 1986), fecundity (Uzzell 1969), competitive ability of larvae (Brodman &
Krouse 2007), and now dispersal ability. Using these four traits within the model of coexistence developed in a similar frog system (Hellreigel and Reyer 2000) predicts a local extinction scenario for unisexual salamanders. Not only is the inferior dispersal ability of unisexuals surprising within the context of these coexistence models, two other lines of evidence would also predict greater dispersal ability in the unisexuals. First, vertebrates that lack sexual reproduction are often all-female, and differential dispersal in relation to competing sexual species can be a stabilizing force for unisexual population persistence (Kokko et al. 2008). In cases where the unisexual lineage is a sexual parasite, colonization ability can drive persistence because the unisexual lineage is likely to outcompete local hosts and rely on locating new sexual populations (Kerr et al. 2006).
Second, unisexual salamanders are polyploids, a commonly-hypothesized driver of increased dispersal ability (Linder & Barker 2014) due to an association with marginal habitats (Greenwald et al. 2016). However, in our study, ploidy did not appear to be a driver of dispersal since the ratio of triploids to tetraploids among unisexuals was similar 66
between dispersing and non-dispersing individuals (57% tetraploids in total sampled individuals versus 64% tetraploids in dispersers).
In combination with previous comparisons of life history traits, inferior dispersal ability suggests that unisexuals should not be able to coexist with sexuals based on current models of coexistence. Yet, unisexuals populations are abundant in this and many locations across a large area of North America (Bogart & Klemens 2008), suggesting coexistence may be better explained by other models or factors. For example, A. texanum possesses both higher fecundity and longer dispersal range compared with unisexuals, two life history traits associated with an “inferior competitor” that is traditionally predicted to exploit resources in novel habitats before superior competitors arrive (the fugitive strategy; Bolker & Pacala 1999; Amarasekare 2003). Alternatively, if unisexual
Ambystoma are considered sexual parasites, these results support a model of coexistence similar to that of host-parasite systems, in which over-exploitation of the host (A. texanum) can be prevented by mating selection against the parasite (Dawley & Dawley
1986) in combination with low levels of dispersal by the parasite (shown in this study, discussed by Kokko et al. 2008, and demonstrated by Kerr et al. 2006). Unisexual
Ambystoma may be intermediate between these different models of coexistence due to the frequency of genome exchange between sexual males and female unisexuals (Ramsden
2008; Charney 2012b; Gibbs & Denton 2016). The predictions for a taxon that reproduces clonally while occasionally incorporating novel genetic material from another species, such as in unisexual Ambystoma, may differ than other unisexual taxa with limited or no introgression.
67
Chapter 4: Sperm dependence limits niche divergence in unisexual salamanders of varying ploidy and genome composition
Abstract
Unisexual lineages that are dependent on closely-related sexual species for reproduction face the conflicting challenges of promoting coexistence through niche divergence while remaining in sympatry for breeding. Niche relationships between unisexual lineages and their sexual relatives show a wide range of patterns, from niche intermediacy to niche transgression, yet the causes of this variation are unclear. How fundamental characteristics of these unisexual taxa, such as ploidy level and genome composition, influence the resolution of these conflicting demands is an important step in understanding the causes of this variation. Deciphering the influence of ploidy level versus genome composition is especially challenging because few wild lineages provide sufficient variation in these traits that are necessary to make appropriate comparisons.
Here, we take advantage of a single lineage of all-female (unisexual) salamanders that vary in their number (2N-5N) and origin of nuclear genomes that are derived from sympatric sexual species to determine if niche relationships between unisexuals and sexuals change across ploidy and genome composition. Using range-wide occurrence data for unisexual salamanders with different biotypes and the two primary sexual species 68 with which they interact (Blue-spotted Salamanders – Ambystoma laterale and Jefferson
Salamanders – A. jeffersonianum), we measured realized niche overlap while correcting for the amount of shared environmental variation produced by a gradient of range overlap. We found significant niche similarity, not divergence or intermediacy, to be the dominant pattern between both sexual species and unisexual salamanders across variation in both ploidy and genome composition. Results from joint species distribution modelling on a regional subset of our data suggest that biological interactions between species may be an important cause of niche similarity at smaller scales. These findings suggest that the unisexual salamanders’ dependence on sperm from sexual males of other species may limit their niche divergence from those sexual species at a range-wide scale. Taken together, these analyses provide evidence that reproductive behavior overrides ploidy and genome diversity as drivers of realized niche in unisexual lineages that must stay reproductively tethered to sexual species, limiting the role niche divergence can play in promoting the coexistence of closely-related sexual and kleptogenetic taxa.
Introduction
The coexistence of competing sexual and unisexual taxa is a long-standing problem in evolutionary biology (Jokela et al. 2009). This coexistence is especially difficult to explain between unisexual animal taxa that compete with sexual taxa while still depending on them for reproduction. These taxa are mostly clonal, all-female lineages that require sperm from males of closely-related sexual species to generate offspring (termed "gynogenesis"; reviewed in Schlupp 2005). Gynogenetic unisexuals are considered “sexual parasites” because they use congeneric sperm to alleviate the costs of 69 producing their own males and instead impose this cost on sympatric sexual congeners
(Maynard Smith 1978). At the same time, sperm dependence means gynogens are reproductively tethered to their sexual hosts, potentially limiting their ability to avoid the competition for resources that results from using the same niche (Vrijenhoek & Pfeiler
1997; Beukeboom & Vrijenhoek 1998). To coexist with the sexual species that gynogens exploit, these lineages are expected to strike a paradoxical balance between remaining in sympatry for reproduction while showing sufficient ecological divergence from competing sexual species to minimize competition.
Direct dependence on sexual species to reproduce predicts that gynogens should display strongly overlapping distributions with potential sperm-donors but, in fact, patterns of co-distribution vary widely, suggesting a possible influence of niche divergence. Some hybridogens, such as edible frogs (Pelophylax kl. esculentus; Kuzmin
1999), show tightly-linked distributions between hybridogens and the sexual species that they utilize, while others display distributions on the periphery (Poeciliopsis fish, Moore et al. 1970; Wetherington et al. 1989) or within a subset (Ambystoma salamanders and
Amazon mollies, Bogart et al. 2007; Costa & Schlupp 2010) of the associated host species’ distribution. Gynogen distributions that are peripheral to or nested within the distribution of a progenitor species may indicate an intermediate realized niche that has evolved to reduce competition (Vrijenhoek & Parker Jr 2009), while niche differentiation by gynogens is predicted to result in both less competition and lower extinction rates
(Kirkendall 1990; Schley et al. 2004; Kokko et al. 2008). Multiple sperm-dependent unisexual groups have shown some level of ecological niche differentiation from their sexual hosts, including flatworms (Weinzierl et al. 1999), salamanders (Greenwald et al. 70 2016), and multiple fishes (Vrijenhoek 1978; Vrijenhoek & Pfeiler 1997; Schlosser et al.
1998; Gray & Weeks 2001; Mee et al. 2013). Although we lack an understanding of the causes of variation in niche relationships, much of this variation is hypothesized to be a result of species-specific interactions, variation in the level of sperm dependence, and the length of time that the sexual and unisexual groups have been in sympatry. However, no studies have comprehensively evaluated the impact of these factors in a single gynogenetic lineage.
Most gynogens are also polyploid, and so the well-known impacts of polyploidy on niche variation may also apply to gynogenetic polyploids. For example, the long- standing hypothesis that polyploid plants must diverge from the ecological niches of their sympatric diploid progenitors has been described for both polyploid animals ("the frozen niche" hypothesis; Vrijenhoek 1984; Vrijenhoek & Parker Jr 2009) and plants (the "niche shift" hypothesis; Levin 1975, 2003). However, empirical studies have shown diverse patterns in the relationship between parental and polyploid lineages, including niche divergence from parental species, conservation of the parental niche, and the occupation of intermediate space between parental niches (McIntyre 2012; Soltis et al. 2014; Mau et al. 2015; Ficetola & Stöck 2016; Marchant et al. 2016). As the majority of polyploids have equal numbers of genomes from each parental species (e.g. Tetraploid [4N] =
Parental species 1 [2N] + Parental species 2 [2N]), their phenotype could be predicted to equally represent each parent (Madlung 2012). Accordingly, intermediate polyploid niches between parental species are the most commonly observed pattern in both plants
(Glennon et al. 2014; Marchant et al. 2016) and animals (Costa & Schlupp 2010; Mee et al. 2013). However, it is unclear if niche intermediacy or other niche relationship patterns 71 are primarily an effect of an increased number of genomes or the individual contributions of genomes from different parental species (Parisod & Broennimann 2016; Soltis et al.
2016).
Disentangling the effects of genome number and genome composition is a difficult task for plants and animals for a variety of reasons. Plants that vary widely in ploidy are either formed via a doubling of the same genomes (autopolyploids) or by hybridization of closely related parental species (allopolyploids; Renny-Byfield &
Wendel 2014). Therefore, comparisons between polyploids and their parental species in plants are not limited by variation in ploidy, but by subgenome diversity since autopolyploid subgenomes are typically duplicates whereas allopolyploid subgenomes typically exhibit balanced numbers of chromosomes. As a result, studies that investigate niche relationships across ploidy levels often fail to evaluate effects of differences in subgenome composition (e.g. Theodoridis et al. 2013; Glennon et al. 2014). The opposite challenge is present in animal polyploids, in which a majority of polyploids are triploids consisting of one copy of a first parent’s genome and two copies of a second parent’s genome (Otto & Whitton 2000). In contrast to plant polyploids, animal polyploids provide opportunities to study the impact of genome diversity because of their often unbalanced genome composition, but most lineages show limited variation in ploidy level
(e.g. Costa & Schlupp 2010). Here, we study a unique system that allows for the assessment of effects from both ploidy and genome composition on niche relationships: a polyploid, sperm-dependent group of salamanders which consistently vary in ploidy and genome composition that represents introgression from two co-occurring sexual species.
72
Polyploid unisexual salamanders (Genus Ambystoma) are an all-female kleptogenetic lineage that require the sperm of congeneric males to initiate egg production (Bogart et al. 2007). However, genomic introgression from the sexual species into the unisexuals occurs at a relatively high rate in unisexual Ambystoma, often causing the addition or replacement of an entire haploid genome (Bogart et al. 2007; Gibbs &
Denton 2016). This process results in unisexual individuals with a consistent maternally- inherited mitochondrial genome and variable numbers of nuclear genomes (2N-5N) composed of haploid genomes from up to five different sexual species (Bogart et al.
2009), most commonly comprised of genomes from the Blue-spotted Salamander (A. laterale, genomes abbreviated as “L”) and the Jefferson Salamander (A. jeffersonianum, genomes abbreviated as “J”; Bogart et al. 2007). There are more than twenty distinct combinations of ploidy and genome composition recognized (defined as “biotypes”, e.g.
LJJ, LLJ, LJJJ).
If the ecological niche of unisexual Ambystoma is determined by ploidy level, we predict that unisexuals with a greater number of genomes may display a more strongly differentiated niche from both sperm-donating species regardless of genome composition
(niche novelty or niche transgression; Glennon et al. 2012; Theodoridis et al. 2013). This would provide evidence for a scenario in which additional genomes may provide more opportunity for neofunctionalization of duplicated genes that in turn can lead to increased genetic diversity and divergent ecological phenotypes (Shimizu-Inatsugi et al. 2016). In contrast, if the niche use by unisexual Ambystoma is influenced by the composition of their nuclear genomes, we predict that unisexuals with biotypes more representative of one sexual species would be more similar to that sexual species in terms of realized 73 niche. In other words, a triploid LJJ individual would have a more similar niche to A. jeffersonianum than an LLJ individual. This result would support genome composition as the primary driver for niche use in a sperm-dependent taxon and provide means by which gynogenetic polyploids could potentially escape competition from sympatric sexual species. Niche differentiation among sexual-unisexual groups has been demonstrated between two sexual species (A. texanum and A. jeffersonianum) and unisexuals with 1-2 genomes from each sexual species (Greenwald et al. 2016). However, this result is puzzling because niche differentiation such as that described by Greenwald et al. (2016) conflicts with gynogens’ reproductive dependence on the sperm of sexual species (Costa
& Schlupp 2010) and the limited dispersal ability of unisexual salamanders (Denton et al.
2017). Finally, if the requirement for sperm were to prevent significant niche differentiation between unisexuals and their sexual hosts, we would expect to observe high niche overlap between each sexual species and unisexuals, regardless of ploidy or genome composition.
Previous work on the niche relationships of these animals (Greenwald et al.
2016), was limited because of a regional focus, a lack of niche characterization of the most widespread species that donates sperm to unisexuals (A. laterale), and the need for genetically identifying phenotypically similar sexual and unisexual individuals. Here we use a larger data set over a broader geographic range combined with new methods of analyses to generate a broader understanding of the factors that underlie niche relationships in this sexual-unisexual complex of salamanders. To differentiate between the hypotheses outlined above, we first collected range-wide, genetically-confirmed occurrence points for the two most prominent species that interact with unisexuals (A. 74
laterale and A. jeffersonianum) and five unisexual biotypes (LJ, LJJ, LLJ, LJJJ, and
LLLJ). We then constructed ecological niche models (ENMs) that correct for the amount of environmental variation shared between groups to estimate their niche overlap.
Second, we constructed joint-species models to infer correlations between species occurrence beyond those that can be explained by measured environmental variation alone to examine the potential effect of biological interactions (specifically sperm dependence) on the co-occurrence between unisexual and sexual Ambystoma.
Methods
Occurrence points
We collected range-wide presence localities for A. laterale, A. jeffersonianum, and unisexual Ambystoma biotypes (LJ, LLJ, LLLJ, LJJ, LJJJ) using both published
(Bogart & Klemens 1997, 2008; Bi et al. 2008a) and unpublished data (J Bogart, K
Greenwald). Because unisexual individuals are phenotypically similar to sexual species, we only used occurrence data from individuals that had been genetically characterized to species, ploidy, and genomic composition using either allozymes and erythrocyte area
(Bogart & Klemens 1997), microsatellite DNA markers (Ramsden et al. 2006), or single nucleotide polymorphisms (Greenwald & Gibbs 2012). This initial set of occurrence points included 1,826 individual records (Table 4). To prepare these data for analysis, we first spatially rarefied all points in order to account for model overfitting due to bias in environmental similarity (Boria et al. 2014). We conducted a spatial autocorrelation analysis in the ecospat R package (Cola et al. 2017) that indicated non-significant spatial autocorrelation at distances greater than 400m. Based on this distance, we used the 75 spatially rarefy occurrence data tool in SDMtoolbox (Brown 2014) to rarefy points for each group that were less than 1 km apart and adjust which points were removed based on maximizing sampling from across a gradient of spatial heterogeneity (Table 4, Figure
9). This process was completed twice to construct two separate sets of occurrence points, one that separated each unisexual biotype (LJ, LLJ, LLLJ, LJJ, LJJJ) and one that combined biotypes into ploidy levels (2N [LJ], 3N [LJJ + LLJ], 4N [LJJJ + LLLJ]).
Figure 10. Geographic distribution of occurrence data for two sexual Ambystoma species and five unisexual Ambystoma biotypes. Each dot represents a single site and dots vary in size in order to represent the diversity present at each site. The density of sampling in Southern Ontario is presented in Figure 10.
76 Table 5. Summary of rangewide occurrence data used for analyses of two sexual Ambystoma salamanders and five unisexual Ambystoma biotypes. Occurrence points are taken from previous surveys (Bogart & Klemens 1997, 2008; Greenwald et al. 2016) and unpublished records (JP Bogart, KR Greenwald). Spatially rarefied points were generated using the spatially rarefy occurrence data tool in SDMtoolbox (Brown 2014) with a buffer distance of 1km across a gradient of spatial heterogeneity. Number of Occurrence Points Species/Unisexual All Data Spatially Rarefied Biotype A. laterale (LL) 462 336 A. jeffersonianum (JJ) 210 141 2N unisexuals 82 72 3N unisexuals 947 487 4N unisexuals 125 96 LJ 82 72 LLJ 590 365 LJJ 357 162 LJJJ 66 47 LLLJ 59 49 Total localities 1270 1172
Climate data
We gathered climate data from the Bioclim database (Hijmans et al. 2005) and trimmed the spatial extent of all 19 bioclimatic layers to fit the rectangular area around the occurrence points with a ~100km buffer to limit the selection of less informative background points (Barbet-Massin et al. 2012). Climate variables that had a correlation of r ≥ 0.80 were removed from the environmental data, leaving Bioclim variables 1, 2, 7,
8, 9, 12, 13, 15, and 18 for further analysis. To decrease model over-fit caused by the selection of unrepresentative background points (Anderson & Raza 2010; Barbet-Massin et al. 2012; Merow et al. 2013), we constructed a bias file with a local adaptive convex hull buffer (Alpha = 4, 150km buffer distance). Because of a high density of samples
77
from southern Ontario (Figure 2), we combined this bias file with a Gaussian Kernel
Density surface that increases the rate at which background points are drawn from geographic areas farther from areas of high occurrence density. This single composite bias file encompassed the occurrence points for all groups. We made this choice deliberately to standardize the correction for sampling effort across all sites, not just a single focal species.
Ecological Niche Modelling
We constructed ENMs for each unisexual Ambystoma biotype (LJ, LLJ, LLLJ,
LJJ, LJJJ), each ploidy level (2N, 3N, 4N), and both sexual species (A. laterale and A. jeffersonianum) using Maxent (Phillips et al. 2006a; Elith et al. 2011) with current best practices (Morales et al. 2017). For the unisexuals, our goal was to summarize niche variation across separate (but not independent) axes of genome and ploidy variation.
First, we used SDMtoolbox to prepare Maxent runs for spatial jackknifing, where localities are segregated into spatial groups and cross-validated to evaluate model performance. The spatially rarefied occurrence points, the composite bias file, and the nine environmental variables were used as primary input for the Maxent analysis, while regularization multiplier (0.5, 1, 2, 3, 4, 5) and feature class (linear, linear/quadratic, hinge, linear/quadratic/hinge, linear/quadratic/hinge/product/threshold) were permutated across the runs. All models were replicated five times and the best model as chosen by omission rate and area under the curve (AUC) metric was run for a final 20 replications.
This process created 4,500 niche models (5 replications x 6 multipliers x 5 feature classes x 3 spatial groups x 10 different groups/species). 78 Figure 11. The subset of occurrence points from Ontario only for A. laterale (“LL”, N = 246, bottom) and A. jeffersonianum (“JJ”, N = 123, top) and the unisexual biotypes with mainly A. jeffersonianum genomes (320 points for LJJ, 54 points for LJJJ; top) or A. laterale genomes (406 for LLJ, 30 for LLLJ; bottom). Diploid unisexual Ambystoma with a single genome from each species (LJ, N = 45) are also shown on both maps.
79 Quantifying niche overlap
To estimate overlap, we first determined which environmental variables were most influential in differentiating between the realized niches of each sexual species and associated unisexual biotypes with principal components analysis (PCA) in R (R
Development Core Team 2011) using the same nine Bioclim variables from above extracted at the spatially-rarified set of occurrence points. Second, we compared the extent of overlap between each sexual species (A. laterale and A. jeffersonianum) and corresponding unisexuals categorized by either ploidy or genome composition to the amount of overlap that would be expected given shared environmental variation using the background tests implemented in the program ENMtools (version 1.4.4; Warren et al.
2010), as the identity test approach is inappropriate given the broad and partially allopatric distribution of species in this study (Warren et al. 2010). We created background masks for each biotype and species from which to drawn comparison points and each comparison was done in both directions (e.g. A. laterale occurrence points vs.
LLJ background points and LLJ occurrence points vs. A. laterale background points) across 100 replicates. We made all statistical comparisons using Schoener’s D as a measure of niche overlap (Schoener 1968). Additionally, we calculated niche breadth using Levin’s inverse concentration measure that averages the habitat suitability score per cell of the constructed ENM (Levins 1968).
For each comparison between a sexual species and a unisexual group (biotype or ploidy level), we took the average of the two background distributions (expected niche overlap based on shared environmental variation) and subtracted this number from the actual amount of niche overlap between the two groups. This metric is defined here as the 80
magnitude of niche similarity, where positive values indicate a greater niche overlap than what would be expected based on shared environmental variation and negative values indicate niche differentiation greater than what would be expected based on shared environmental variation. The predicted relationship between the magnitude of niche similarity and the genomic similarity between a given sexual species and related unisexuals could be either positive or negative. If the addition of a sexual species’ genome in unisexual animals results in a more similar niche to that particular sexual species, we would predict a positive association between the magnitude of niche similarity and the number of genomes present from a particular sexual species (Gibbs &
Denton 2016). If the total number of genomes is a more important factor for niche differentiation between sexuals and unisexuals, we would expect a relationship between magnitudes of niche similarity and increasing ploidy regardless of genome composition.
Alternatively, if gynogenetic dependence on sperm prevents broad realized niche differentiation, we would predict no association between the magnitude of niche similarity and genome composition (Te Beest et al. 2012).
Biotic Interactions between unisexual and sexual salamanders
Niche overlap comparisons based on niche modeling cannot be used to assess direct biological interactions between groups because they only describe the relationship between the occurrence of a species and environmental variation, which can be a product of both abiotic and biotic factors. In our case, interactions between unisexual salamanders and sexual species are expected as they are required for unisexual reproduction (Bogart &
Licht 1987). To quantify biotic relationships from the co-occurrence data, we used a 81 joint-species distribution modelling method (JSDM; Pollock et al. 2014; Harris 2015).
These approaches model the distributions of multiple groups simultaneously and produces the shared environmental correlations (ℙjj’) and residual correlations (Pjj’) between species (Clark et al. 2014; Pollock et al. 2014). The shared environmental correlations between species are similar to those of other ENM approaches (Elith &
Leathwick 2009), but other ENM methodologies assume that species occurrence probabilities are uncorrelated (Calabrese et al. 2014). Because unisexual and sexual
Ambystoma have direct reproductive interactions, this assumption is likely violated and accounting for correlation in occurrence is necessary. The estimation of residual correlation between species is the other strength of JSDM methods. These values can be interpreted as either unmeasured environmental variation or interactions between species that can result in positive correlations in presence (e.g. commensalism) or negative correlations in presence (e.g. competition, Pollock et al. 2014). Because climate variables have previously been shown to successfully predict the habitat suitability of closely- related species of Ambystoma (Micheletti & Storfer 2015; Searcy & Shaffer 2016), we assume that the residual correlations produced by JSDMs are reasonably interpreted as reflecting biotic interactions between groups (Clark et al. 2014).
We conducted this analysis at a local scale (see Fig. 2) because we predicted that if any biotic interactions between sexual species and unisexuals were detected, they would be most apparent in a smaller geographic area for which environmental variation between sites is more constrained and all groups are present. We took a subset of these sites from southern Ontario (Ontario subset data, Figure 2) to run this analysis in a geographic area where all groups co-occur. To provide a fine-scale view of co-occurrence 82 and include sites where individuals of each species could disperse between breeding wetlands, the Ontario subset data was not spatially rarefied. The JSDM analysis was conducted following the procedures described in Pollock et al. (2014) using 1.0 x 106 iterations across three chains with a burnin of 1.5 x 104. Convergence was assessed by calculating the mean value for the Gelman-Rubin statistic.
Results
Niche models
Metrics for the best-ranked models generated using the SDMtoolbox workflow are shown in Table 6. The AUC values for all models were > 0.75 except for the A. laterale model (0.68). However, this species represents a problematic application of AUC as a model fit criterion because of its large geographic range and associated greater range of climatic variation compared to the other groups. Because we stringently filtered both occurrence and climate data, the best-ranked ENM for A. laterale is likely biologically realistic despite a relatively low AUC value (Lobo et al. 2008). Most important Bioclim variables were related to temperature (15 of 20 total variables; Table 6). Broadly speaking these models are biologically realistic given known distributions and habitat preferences of salamanders in the genus Ambystoma (Demastes et al. 2007; Searcy &
Shaffer 2016).
83 Table 6. Parameter summary of the highest ranked ecological niche models for each species or unisexual biotype, including the Area Under the Curve (AUC) metric, the final choices for the Maxent regularization multiplier (RM; 0.5, 1, 2, 3, 4, or 5), the Maxent feature class (FC; 1: linear, 2: linear/quadratic, 3: hinge, 4: linear/quadratic/hinge, 5: linear/quadratic/hinge/product/threshold), and measured niche breadth (Inverse Concentration). For each final model, the first and second-ranked predictor variables are listed according to their importance based on permutation analysis. Niche First-ranked predictor variable Second-ranked predictor variable Group AUC RM FC Breadth (permutational importance) (permutational importance) A. laterale (LL) 0.68 2 2 0.822 Mean Diurnal Range (28.2) Annual Mean Temp. (26.9) A. jeffersonianum (JJ) 0.86 0.5 4 0.200 Precip. Seasonality (28.5) Temp. Annual Range (21.8) 2N 0.87 0.5 4 0.374 Mean Temp. Wettest Quarter (21) Annual Mean Temp. (20.9) 3N 0.75 0.5 2 0.607 Annual Mean Temp. (29.3) Mean Temp. Driest Quarter (19.6) 4N 0.89 0.5 4 0.299 Annual Mean Temp. (33.3) Precip. Seasonality (18.1) LJ 0.87 0.5 4 0.385 Mean Temp. Wettest Quarter (21) Annual Mean Temp. (20.9) LLJ 0.80 3 4 0.501 Annual Mean Temp. (28.2) Precip. Seasonality (16.7) LJJ 0.88 0.5 4 0.257 Temp. Annual Range (49.7) Precip. of Wettest Month (17) 84 LJJJ 0.90 5 5 0.512 Temp. Annual Range (58.1) Mean Temp. Driest Quarter (29.9) LLLJ 0.90 0.5 3 0.270 Annual Mean Temp. (34.6) Precip. of Warmest Quarter (28.7)
When the ENMs for each group are viewed in geographic space (Figure 11), A. laterale and A. jeffersonianum show limited overlap in their distributions of habitat suitability. There appears to be an interaction between genome type and number in terms of niche breadth. For unisexual biotypes with greater numbers of A. laterale genomes
(LLJ and LLLJ), niche breadth decreases with increasing ploidy (0.82 for A. laterale,
0.50 for LLJ, and 0.27 for LLLJ). This pattern is reversed for A. jeffersonianum and its most similar unisexual biotypes (LJJ and LJJJ), where niche breadth increases with increasing ploidy (0.20 for A. jeffersonianum, 0.25 for LJJ, and 0.51 for LJJJ).
85
Figure 12. Ecological Niche Models plotted in geographic space for two sexual Ambystoma species (A. laterale and A. jeffersonianum) and five unisexual Ambystoma biotypes. Colors represent the logistic habitat suitability as produced by Maxent and the scale is standardized across the maps with a maximum of 1 and minimum of 0.1. All areas that have <0.1 habitat suitability are not shown.
Niche overlap
We first assessed niche relationships among these salamanders through descriptive comparisons based on PCA. The PCA identified two primary principal 86 components that explained 35% (PC1) and 31% (PC2) of the total variation among the nine Bioclim variables across the seven groups (all unisexual biotypes and both sexual species, Figure 12). The loadings for PC1 were highest for precipitation variables
(Precipitation of Wettest Month = 0.57, Annual Precipitation = 0.50, Mean Diurnal
Range = 0.46) whereas loadings for PC2 were highest for temperature variables
(Temperature Annual Range = 0.57, Temperature of Driest Quarter = 0.43, Annual
Temperature = 0.45, Precipitation Seasonality = 0.44). The separation between A. jeffersonianum and A. laterale among these two axes reflect the general environmental characteristics of each of their distributions: A. laterale is associated with higher latitudes which have relatively lower temperatures and less precipitation whereas A. jeffersonianum is associated with lower latitude, unglaciated environments with higher temperatures and more precipitation. While the unisexual biotypes show some degree of separation in a biplot of PC1 and PC2 scores, all unisexual biotypes are contained within the 95% ellipse of A. laterale (Supplemental Figure 18), demonstrating strong niche similarity to A. laterale regardless of ploidy or genomic composition.
87
Figure 13. Summary of Principal Component (PC) 1 and 2 scores for each sexual species shown as squares (A. laterale “LL”; A. jeffersonianum “JJ”) and associated unisexuals of varying ploidy and genome composition shown as colored circles. Each point represents the mean PC score and associated standard error. The loadings for PC1 were highest for precipitation variables (Precipitation of Wettest Month, BIO13 = 0.57, Annual Precipitation BIO12 = 0.50, Mean Diurnal Range, BIO2 = 0.46) whereas loadings for PC2 were highest for temperature variables (Temperature Annual Range, BIO7 = 0.57, Temperature of Driest Quarter, BIO9 = 0.43, Annual Temperature, BIO1 = 0.45, Seasonality of Precipitation, BIO15 = 0.44).
Next, we compared niche overlap using a statistical approach that involved building ENMs and then assessing patterns of overlap that takes into account spatial autocorrelation between sample locations and environmental variation in the form of background tests. For comparisons based on differences in ploidy alone between sexuals and unisexuals, niche overlap was significantly different from the mean background values for all comparisons (Table 7), but the direction of this difference was inconsistent
88 among comparisons (Figure 13A, 5B). For example, when the combined mean background values are compared, A. laterale show more distinct niches relative to background from 2N and 4N unisexuals but more similar niches to 3N unisexuals.
Ambystoma jeffersonianum also show increased niche differentiation compared to 2N unisexuals but niche similarity to 4N unisexuals. Overall, ploidy does not explain niche overlap in a consistent way between sexual and unisexual salamanders, a result which provides no support for the prediction of the niche novelty or transgression hypothesis that there should be a positive relationship between degree of overlap and ploidy level.
More broadly, there was relatively little variation in niche overlap across ploidy level when compared to either sexual species (niche overlap range = 0.11 for A. jeffersonianum comparisons and 0.29 for A. laterale comparisons). However, the average niche overlap value between A. laterale and all unisexuals (0.61 ± 0.07 SE) was substantially higher compared to A. jeffersonianum-unisexual comparisons (0.39 ± 0.03 SE). Regardless of ploidy, unisexuals generally show more similar niches to A. laterale compared to A. jeffersonianum, potentially reflecting the ubiquity of A. laterale genomes in all unisexuals regardless of their distance from the nearest A. laterale population.
89
Table 7. Summary of ENMtools background tests. Each niche comparison between groups was made twice: one involving a comparison of the occurrence points for group 1 versus the background area of group 2 and vice versa (Mean background 1 vs 2, Mean background 2 vs 1). This expected niche overlap given environmental differences between the groups was then compared to the calculated niche overlap from the ecological niche models constructed with Maxent using a one-sample t-test. Niche Mean Mean Groups Overlap background background (1 & 2) (D) (1 vs 2) t p (2 vs 1) t p JJ & LL 0.298 0.132 -114.94 <0.001 0.248 -15.74 <0.001 JJ & 2N 0.333 0.235 -34.45 <0.001 0.290 -11.65 <0.001 JJ & 3N 0.390 0.295 -94.85 <0.001 0.509 46.36 <0.001 JJ & 4N 0.452 0.292 -65.97 <0.001 0.410 -20.15 <0.001 JJ & LJJJ 0.463 0.494 7.84 <0.001 0.441 -3.24 0.002 JJ & LJJ 0.688 0.500 -54.79 <0.001 0.421 -45.79 <0.001 JJ & LJ 0.333 0.235 -34.45 <0.001 0.290 -11.65 <0.001 JJ & LLJ 0.281 0.213 -50.38 <0.001 0.232 -13.48 <0.001 JJ & LLLJ 0.317 0.283 -10.71 <0.001 0.274 -11.25 <0.001 LL & 2N 0.533 0.553 6.46 <0.001 0.387 -44.98 <0.001 LL & 3N 0.793 0.741 -44.58 <0.001 0.513 -192.19 <0.001 LL & 4N 0.499 0.723 114.09 <0.001 0.412 -65.25 <0.001 LL & LLLJ 0.503 0.600 28.42 <0.001 0.268 -66.41 <0.001 LL & LLJ 0.682 0.624 -34.57 <0.001 0.366 -113.23 <0.001 LL & LJ 0.533 0.553 6.46 <0.001 0.387 -44.98 <0.001 LL & LJJ 0.381 0.313 -36.76 <0.001 0.172 -99.77 <0.001 LL & LJJJ 0.632 0.362 -92.19 <0.001 0.228 -179.20 <0.001
For comparisons between sexuals and unisexuals of varying genome composition, all but one of the niche overlap estimates were significantly greater than the mean background distribution, demonstrating that most groups had niches that were more similar than expected based on the environments they share (Table 6). The only exception was the comparison between A. jeffersonianum and LJJJ unisexuals, where the niche overlap measurement was not significantly different from the mean background value. As expected, the background distributions for each comparison became lower as groups were spatially distinct from one another in that the expected amount of niche similarity decreases as groups become geographically separated from one another (Figure 13C and
5D). 90 Figure 14. A summary of niche overlap values based on ENMtools background tests for niche overlap between each sexual salamander species (A. laterale: LL, A. jeffersonianum: JJ) and unisexual salamanders grouped by ploidy (A, B) or genome composition (C, D). Gray bars cover the range of the 95% confidence intervals for the background similarity for each comparison, where occurrence points from one group are compared to the background area of the other group and vice versa. Colored circles represent the empirical value of niche overlap calculated by ENMtools using the final Maxent models.
Even if all niche comparisons between each sexual species and unisexual biotypes are significantly more similar than expected, the magnitude of similarity could vary according to the number of genomes shared between the unisexual biotype and the sexual species. To address this, we calculated the magnitude of niche similarity for each sexual/unisexual comparison by subtracting the mean value for both background
91
distributions (e.g. the mean of both the LL vs. LLJ and the LLJ vs. LL background tests distributions) from the measured niche overlap of the final Maxent models. The average magnitude of niche similarity was close to zero for comparisons between each sexual species and unisexuals of various ploidy (Figure 14; 0.01 ± 0.06 for A. laterale comparisons, 0.01 ± 0.04 for A. jeffersonianum comparisons). All sexual-unisexual comparisons were more positive when unisexuals are grouped by their genomic composition (mean magnitude of niche similarity ± SE = 0.16 ± 0.04 for A. laterale comparisons, 0.08 ± 0.04 for A. jeffersonianum comparisons), although this difference was non-significant (paired sample t-test, t = -1.27, p = 0.12). Broadly, the magnitude of niche similarity appears to decline as unisexual biotypes become less similar to A. laterale. With the exception of the LJJJ unisexual biotype, the magnitude of niche similarity seems to increase as unisexual biotypes become more similar to A. jeffersonianum. However, inferences based on either of these patterns are limited by sample size and high variation of niche similarity among biotypes comparisons. The overly similar ENMs between each sexual species and all unisexuals regardless of biotype indicate that variation in genome composition cannot alone explain niche similarity.
92 Figure 15. A summary of the magnitude of niche similarity/differentiation between sexual salamanders species (A. laterale: LL, A. jeffersonianum: JJ) and unisexual salamanders grouped by ploidy (middle) or genome composition (bottom). The top panel describes two predictions for the relationship between unisexual and sexual niche similarity: if sexuals are more dissimilar to polyploids as the number of net genomes from that species decreases (circles) and if sexuals and unisexuals are more similar in niche to one another regardless of ploidy (triangles). Empirical values in the middle and lower panels that are greater than the background amount of overlap represent niches that are more similar than expected based on background similarity, while empirical vales less than the background values represent niche differentiation between the groups. 93 Inferring biotic interactions
The JSDM procedure for the Ontario subset data successfully converged (Mean
Gelman-Rubin statistic ranged between 1.000795-1.005315). Generally, each sexual species had positive environmental correlations of co-occurrence with unisexuals that had similar genomes (e.g. A. laterale positively correlated with LLLJ and LLJ unisexuals) and negative environmental correlations with the other sexual species and unisexual biotypes that had dissimilar genomes (e.g. A. laterale negatively correlated with LJJ,
LJJJ, and A. jeffersonianum). However, 5 of the 12 comparisons include credible intervals that overlap zero. The residual correlation values, which represent associations beyond those explained by environmental variation, show a similar pattern (Figure 15). If these residuals are interpreted as biotic interactions between groups, this indicates that each sexual species co-occurs with unisexuals that have a greater number of similar genomes beyond what would be expected based on environmental variation alone. This result is expected if reproductive dependence outweighs ecological niche divergences as the driving mechanism responsible for niche use in sexual and unisexual salamander. For the Ontario-only data, this is significant because many sites are within typical movement distances of these salamanders (Smith & Green 2005; Zamudio & Wieczorek 2007;
Peterman et al. 2015), which strengthens the potential for the residuals to represent the result of biotic interactions in this specific case.
94 Figure 16. Summary of residual correlations from joint species distribution models using the Ontario only data for occurrences of Ambystoma laterale (left column), A. jeffersonianum (right column) and five unisexual Ambystoma biotypes. Each dot represents the mean value for residual correlations between two groups along with 95% credible intervals. These residual correlations represent the correlations in co-occurrence between groups after accounting for differences in measured environmental variation.
Discussion
We found no strong evidence that two key biological characteristics hypothesized to drive niche divergence and facilitate coexistence in other polyploids – ploidy level or genome composition – consistently account for patterns of niche divergence in sexual and polyploid unisexual salamanders. Instead, the ecological niches of unisexuals and both sexual species are significantly more similar than expected. This rejects the hypothesis that niche divergence promotes coexistence in these salamanders (the frozen niche hypothesis or the niche shift hypothesis) and supports a third novel hypothesis: unisexuals’ requirement for sperm from sexual males to initiate egg production prevents significant niche divergence from sexual hosts. Even in cases where a unisexual biotype is nearly allopatric with one of the two sexual species, such as the JJ vs. LLJ comparison,
95
the measured niche overlap is significantly higher than expected. These results provide evidence that sperm-dependent unisexual lineages are likely limited in their ability to diverge from the ecological niche of their sperm-donor species.
Ploidy as a driver of niche divergence
Our results showed no evidence for increased niche differentiation between unisexuals and sexuals with increasing ploidy. This negative result has also been found across multiple comparisons between polyploid plant species and closely related sexuals
(Glennon et al. 2012), but is at odds with the small number of polyploid plant systems in which chromosome number is associated with niche divergence of polyploid forms from parental species (Mau et al. 2015; Marchant et al. 2016). Our result suggest that additional genomes do not lead to ecologically distinct phenotypes through evolutionary mechanisms like neofunctionalization (Shimizu-Inatsugi et al. 2016) or that any influences of such mechanisms are outweighed by other factors that have a greater impact on niche use (see below).
When compared to plants that show niche divergence between diploid and polyploid species, polyploid unisexual salamanders lack some of the characteristics that may be strongly related to ecological niche differentiation from diploid relatives. First, allopolyploidy in plants generally results in immediate reproductive isolation (Mallet
2007), after which subsequent selection can drive rapid neofunctionalization and potentially lead to niche divergence. While unisexual salamanders are a relatively old lineage as measured via mitochondrial dating (~6 mya; Bi & Bogart 2010), estimating the length of time nuclear subgenomes have been within the unisexual lineage is difficult 96 because gene exchange between unisexuals and sympatric sexual species can be high
(Gibbs & Denton 2016), and introgressed genomes show little modification in patterns of gene expression (McElroy et al. 2017). Therefore, the identity of these introgressed genomes may be more influential on phenotype than the number of total genomes.
Second, plant polyploids can display distinct phenotypic traits associated with polyploidy that may influence niche relationships. For example, polyploid associated traits in plants, such as flower color (McCarthy et al. 2015) and flower scent (Gross & Schiestl 2015), could potentially promote ecological divergence without a geographic shift from parental species through pollinator shifts. In contrast to these clearly differentiated traits between polyploids and sexual relatives, the unisexual salamanders show limited morphological divergence from the sexual salamanders from which they are derived, even when unisexual individuals possess genomes from the most morphologically different sperm donors (Zeyl & Lowcock 1989). Therefore, to the extent that morphological variation mirrors ecological differences in salamanders, unisexuals are not distinct from their sexual relatives.
Genome composition as a driver of ecological niche
The ecological consequence of subgenome composition remains unclear in polyploid plants and animals. While polyploid plants commonly display patterns consistent with an intermediate niche between their diploid parental species (Glennon et al. 2014; Marchant et al. 2016), disentangling the influence of a general increase in ploidy or the individual contributions from each genome is difficult. Even in the few animal systems with unbalanced genome composition (i.e. 3N individual with two 97 haploid nuclei from one parental species and 1 haploid nucleus from a different parental species) that have been examined, niches can be intermediate between both parental species (Amazon Molly; Costa & Schlupp 2010) or more similar to the diploid species with whom they are more genetically similar (Batura toads, Bufo baturae; Ficetola &
Stöck 2016). We observed neither pattern. When we classified unisexual salamanders according to their genome composition, most sexual-unisexual comparisons showed significantly higher than expected niche overlap (Figure 13C, D). This result does not completely rule out an association between the number of genomes from a particular sexual species and the niche overlap with that species, as these comparisons could vary in their magnitude of similarity (Glennon et al. 2014). However, there was also no association between increasing genome similarity and the magnitude of positive niche overlap (Figure 14).
Despite a uniformly positive niche overlap between each sexual species and the unisexual salamander biotypes, our results do suggest potentially asymmetrical niche relationships between the unisexuals and either sexual species. The mean magnitude of niche similarity was higher for comparisons of unisexual biotypes with A. laterale compared to those with A. jeffersonianum. While not statistically significant, a more influential role for A. laterale genomes in the determination of unisexual niches is consistent with the observation that all unisexual individuals contain at least one A. laterale genome but do not necessarily contain an A. jeffersonianum genome (Bogart et al. 2007; Bogart & Klemens 2008). The ubiquity of the A. laterale genome across unisexuals is still unexplained, and attempts to attribute differences in morphology to genome composition have only been successful when biotypes consisting of the most 98
morphologically divergent sperm donor species (e.g. A. tigrinum) are included (Zeyl &
Lowcock 1989). However, these conclusions are confounded with the broad niche displayed by A. laterale when compared to A. jeffersonianum (best visualized by the
PCA, Figure 12), indicating that the magnitude of niche similarity between unisexuals and A. laterale might be a product of the niche breadth of A. laterale and not necessarily the true similarity between A. laterale and the unisexuals.
Despite the matching patterns in niche overlap between the unisexual-sexual pairs in this study, unisexuals are not limited to only A. laterale and A. jeffersonianum as sexual hosts, and their interactions with other sexual species offer a contrasting view to that presented here. For example, another species, A. texanum, is the primary sperm donor for unisexuals in Illinois, central Indiana, and northwestern Ohio (Kraus 1985).
Niche comparisons between A. texanum, A. jeffersonianum, and various unisexual biotypes (LTT, LTJ, LJJ) across Ohio show divergent niches both between unisexual- sexual comparisons and between unisexual biotypes (Greenwald et al. 2016). This result is the opposite of that described here, and may be due to a combination of the scale of investigation (Kirchheimer et al. 2016) or species-specific interactions between the unisexuals and their sexual hosts. First, Greenwald et al. (2016) focused their analysis across an ecologically diverse region that separates two sexual species (A. texanum and
A. jeffersonianum; Omernik 1987), which coincides with nearly bimodal climate variation. In our range-wide analysis, the two sexual species are less divergent from one another, likely due to the broad niche of A. laterale. Second, Greenwald et al. (2016) showed different relationships between the unisexuals and each sexual species.
Unisexuals were more associated with marginal habitat when compared to A. 99 jeffersonianum but not as associated with marginal habitat compared to A. texanum (LJJ vs A. jeffersonianum, LTT vs. A. texanum), and this type of fine-scale ecological differentiation is also supported in this study (JSDM analysis, see below). Taking these results together, we conclude that the complexity of unisexual-sexual interactions is likely to vary widely across ecotones and depend on the composition of local sexual hosts. However, when considered across their overall distributions, a pattern of overly similar niches between unisexuals and sexual hosts dominates.
Biological interactions as a driver of ecological niche
To identify potential biological interactions that may influence regional distribution patterns, we used joint species distribution modelling to quantify residual correlations within southern Ontario, where both primary sexual species and unisexual biotypes are present on the same landscape. Caution must be used when interpreting the residual correlations of JSDM analyses as unambiguous indicators of biological interactions (Harris 2015), but the small scale of the Ontario sites was intentionally chosen to limit the potential impact of climatic variation and instead emphasizes the effect of biotic interactions on distributions. In contrast to the niche overlap analyses, the environmental correlations between species show negative values between each sexual species and unisexuals with fewer similar genomes (e.g. negative environmental correlations between A. laterale and LJJ). However, the credible intervals for these estimates are broad, and even those comparing the two sexual species overlap zero. For this reason, we focused on using niche overlap analyses as an established methodology
100
(Warren 2012; Searcy & Shaffer 2016) to make inferences on niche similarity or dissimilarity between sexual and unisexuals.
The primary goal of using the JSDM analysis was to investigate the residual correlations in co-occurrence between sexual and unisexual salamanders on a regional scale. The residual correlations in co-occurrence between groups beyond those driven by environmental variation demonstrate that unisexuals with more genomes from either sexual species are more likely to co-occur with that sexual species, even on a landscape where the majority of sites are within commonly-reported dispersal distances of these salamanders (Smith & Green 2005; Peterman et al. 2015; Denton et al. 2017). We interpret this pattern as the result of a preference of unisexual individuals to take spermatophores from males that are most represented in their own genomic composition or reduced success for unisexuals that disperse to other wetlands where they must switch sperm donor species. Mate choice between unisexuals and their sexual hosts is a powerful force for coexisting with one another (Mee & Otto 2010), and male A. jeffersonianum have the ability to distinguish LJJ unisexuals compared to females of their own species during staged decision trials (Dawley & Dawley 1986). Unisexual salamanders require sperm from a male in order to initiate egg production, which requires that unisexuals remain sympatric with at least one other acceptable species of Ambystoma that can be parasitized for this function. While there are suggestions that unisexuals may not always require the sperm of sexual males to initiate embryogenesis (Charney et al. 2014), multiple studies have shown that unisexual females do not produce eggs without acquiring a spermataphore from a suitable species (Bogart & Licht 1987; Elinson et al.
1992). However, our results suggest associations in co-occurrence between either sexual 101 species and the unisexual biotypes more similar to that sexual species (Figure 13C, D). If a unisexual individual was not limited by the sperm donor species present, then dispersal to areas with a different sexual species present may afford the opportunity for less niche competition (e.g. a LLJ unisexual moving farther into the range of A. jeffersonianum).
We emphasize that it is difficult to completely rule out the effect of unmeasured environmental variation on patterns of co-occurrence, as preferences for wetland-level habitat characteristics could be misinterpreted as positive biotic interactions using the
JSDM analyses. In contrast to a laboratory choice experiment, breeding bouts in vernal wetlands are difficult to observe and document. Therefore, future work should focus on evaluating our proposed mechanisms for niche similarity by assessing if unisexuals show any degree of mate choice among sexual males and if an overabundance of spermatophores during explosive breeding allows unisexuals to avoid direct mating interactions with sexual males. Such studies could be complemented with analyses that attempt to partition the influence of species interactions using broad scale ecological data which represent an emerging subfield in ecology (Warton et al. 2015; Godsoe et al.
2017).
Two additional features of the unisexual-sexual system support the idea that reproductive interactions between unisexuals and sexuals are the driving factor leading to the homogenization of realized niche between sexuals and unisexuals. First, the male spermatophores that unisexual females acquire are not excluded from entering the unisexual lineage. The introgression of sexual genomes into the unisexual lineage is well established (Bogart et al. 2007; Bi et al. 2008b), and the rate of introgression can be as high as 0.2% genomes per generation (Gibbs & Denton 2016). This “theft” of genomes 102
by unisexuals is potentially a genetically homogenizing force that increases the genomic and hence ecological similarity between unisexuals and their local sexual host. Second, gene expression of the subgenomes with a unisexual salamander is largely balanced, indicating a lack of genome silencing and suggesting that a majority of introgressed genomes could potentially contribute to ecologically-relevant phenotypes (McElroy et al.,
2017). Therefore, the combination of sperm dependence, genomic homogenization, and limited dispersal (Denton et al. 2017) could produce unisexuals that share a similar phenotype and perhaps climatic niche to their local host, increasing the chances that unisexuals with greater numbers of genomes from a particular sexual species are found co-occurring with that species.
The introgression of parental genomes into the unisexuals is also likely constrained to some degree. Introgressed genomes appear to rarely, if ever, create diploid or polyploid unisexuals with a single parental genome (e.g. a “JJ” individual with the mitochondrial haplotype of the unisexual lineage or an “LLL” unisexual; Bogart et al.
2007; Bogart & Klemens 2008; Bi et al. 2008b). Sexual genomes derived from A. laterale and A. jeffersonianum are found in unisexual individuals that are more than 250 km from the nearest matching sperm donor, suggesting that the process of genome exchange is contextual and potentially adaptive (Charney 2012b). The balance between genome introgression as a homogenizing force and the retention of certain genomes despite this force would be a fruitful area of future research to describe the mechanisms driving the ecological coexistence between unisexual and sexual salamanders (Bi &
Bogart 2010a).
103 Conclusion
Large scale ecological data have helped facilitate a goal of understanding patterns in realized niche across the genomic composition of polyploids in terms of ploidy number or genome composition (Parisod & Broennimann 2016), but opportunities to study polyploids across a gradient of genome composition are few, as most possess equal genomic contributions from parental species. Unisexual Ambystoma salamanders provide a system to test for the asymmetrical effects of ploidy number or genome composition.
However, neither characteristic provides a satisfactory explanation for patterns of niche overlap between unisexuals and sexual Ambystoma salamanders. Instead these animals support a third, rarely considered hypothesis for the mechanism driving niche overlap between polyploids derived from related sexual forms, namely that the proximate dependence of unisexuals on sexual species for reproduction (gynogenesis) can minimize or prevent niche differentiation between asexual/polyploids and their parental species.
For animals, we suggest that unisexual salamanders are on one end of a spectrum of expected ecological similarity to their sexual relatives. Conducting similar ecological analyses across other polyploid animal lineages, from those that require sperm but more rarely display genetic introgression (e.g. Amazon Molly; Costa & Schlupp 2010) to those that do not require sexual species but still occasionally display genetic introgression
(Corbicula clams; Hedtke et al. 2011) to those that reproduce strictly asexually (New
Zealand mudsnails; Jokela et al. 2009). This reproductive diversity among unisexual animals could provide necessary perspective into the niche relationships between polyploids and their sexual relatives and play a part in explaining the longevity or ephemerality of non-sexual lineages (Neiman et al. 2009). 104
Chapter 5: General Conclusions
Unisexual Ambystoma provide an enigmatic and unique system to study the evolution of sex, but these salamanders remain understudied due to their cryptic life history and complex genomic diversity. I have expanded our knowledge of this system by solving the mitochondrial misidentification of Ambystoma in Ohio, showing that dispersal cannot explain coexistence between sexual and unisexual Ambystoma, and providing a new hypothesis for why the realized niches of unisexuals have not differentiated from their sexual competitors. These new studies provide answers for lingering questions regarding the unisexual salamander system, but also raise new possibilities in the realms of mitonuclear interactions, metabolism, and genome dosage compensation.
In addition to providing another example in a long line of literature cautioning the use of mitochondrial DNA along for species identification (Rubinoff 2006), Chapter 2 illustrates a comprehensive framework for addressing the potential causes for mitonuclear mismatch that extends the approaches of previous studies (Sequeira et al. 2011; Nevado et al. 2011; Melo-Ferreira et al. 2012). By leveraging approaches that take into account incomplete lineage sorting and those that do not, researchers can be more confident in conclusions from widely available software. Beyond the methodological approaches used 105
to describe patterns of mitochondrial discordance, a broad effort to understand the mechanisms that drive these patterns, such as adaptive introgression, demographic variation, and sex-biased asymmetries, is underway in molecular ecology (Toews &
Brelsford 2012). While mitonuclear discordance is now recognized across many taxa, only recently have efforts been made to explicitly link a pattern of discordance to a particular process. I found that A. texanum individuals with A. barbouri-like mtDNA were significantly more likely to be present at sites with greater precipitation. The statistical significance of this difference despite the relative small amount of variation in average rainfall (~3% difference between sites in average annual rainfall) suggests that there may be an adaptive basis for mtDNA introgression between these species.
Moreover, the mitochondrial introgression from A. barbouri into A. texanum is currently the only recognized example of mtDNA introgression in the Ambystoma genus
(Eastman et al. 2009). The unisexual lineage is hypothesized to be the result of ancient hybridization between a common ancestor similar to A. barbouri (most closely related mtDNA to unisexuals) and A. laterale (most representative nuclear genome in unisexuals;
Bogart et al. 2009). The persistence of this mitochondrial heritage given the variety of nuclear genomes present in any given unisexual individual suggests that A. barbouri-like haplotypes may confer reduced consequences of cytonuclear conflict following introgression.
My third chapter applied an integrative approach to quantifying the dispersal of unisexual and sexual salamanders. Without the use of endurance trials, it would be unclear if the differences in genetically-inferred dispersal were do factors other than physiological dispersal ability, such as a greater number of small movements over the 106 lifetime of an individual that might give the appearance of long distance movement. The combination of genetic and physiological data provided a link between the movement capability of animals and the mating success of dispersers, and integrative measures such as these are recommended for studying complex dispersal behaviors (see Bowler &
Benton 2005; Lowe & McPeek 2014).
Locomotor endurance by unisexual Ambystoma was significantly lower compared to all three sexual species, including those that represent all of the nuclear genomes within the unisexual individuals (A. jeffersonianum, A. laterale, and A. texanum).
Because the nuclear genomes present in unisexual are expected to show similar levels of expression compared to sexual species (McElroy et al. 2017), the difference in performance between unisexuals and sexuals is potentially due to the combination of unisexual mitochondria with nuclear genomes introgressed form sexual species. A realistic limitation may be reduced efficiency of the oxidative phosphorylation system as a result of structurally mismatch proteins produced by non-coevolved nuclear and mitochondrial genes (Lane 2011). The quantification of the mitonuclear mismatch at a genotype level in unisexual salamanders and the relationship between mitonuclear mismatch in unisexuals and the impacts on their physiology are important goals for future research.
How unisexuals salamanders maintain coexistence with sympatric sexual species remains unexplained. In addition to reduced dispersal capability, unisexuals are avoided by sexual males in mate choice trials (Dawley & Dawley 1986), have lower fecundity
(Uzzell 1969), and have no competitive advantage as larvae (Brodman & Krouse 2007).
These four traits are highlighted by Hellreigel and Reyer (2000) as the primary drivers of 107
local coexistence between sexual species and unisexual sperm parasites, and our current knowledge of the unisexual Ambystoma salamanders would lead to extinction of unisexuals according to these models. Despite these disadvantages, unisexual salamanders are abundant across many areas of northeastern North America (Bogart &
Klemens 2008), suggesting coexistence may be better explained by other models or factors. The unique reproductive mode of unisexual salamanders may not conform to the assumptions of models designed for asexual groups, and considering models designed for host-parasite systems may be more accurate. For example, over-exploitation of the local host may be prevented by mating selection against the parasite (Dawley & Dawley 1986) in combination with low levels of dispersal by the parasite (shown in this study, discussed by Kokko et al. 2008, and demonstrated by Kerr et al. 2006).
Chapter four approached the coexistence paradox of unisexual and sexual salamanders from a broader ecological perspective by testing two hypotheses that explain the ecological niche divergence between other pairs of asexual and sexual species. I found that neither of the characteristics potentially related to niche divergence, ploidy level or genome composition, could predict the niche relationships between unisexual and sexual salamanders. Instead, all niche overlap metrics were overly similar between sexuals and unisexuals, despite any differences in genome number or genome composition. This unexpected result supports a third, novel hypothesis: the requirement for sexual males limits the niche expansion of local unisexuals beyond the niche of their sperm donor species.
Unisexual salamanders lack two of the key characteristics that are related to niche divergence between diploid-polyploid lineages of plants (Mau et al. 2015; Marchant et al. 108 2016). First, allopolyploidy in these systems often results in immediate reproductive isolation (Mallet 2007), whereas the unisexual salamanders have frequent nuclear introgression from sexual species (Gibbs & Denton 2016). Because unisexual subgenomes reflect the same patterns of expression found in their sexual species of origin
(McElroy et al. 2017), frequent nuclear introgression may prevent any neofunctionalization that could lead to niche divergence (Shimizu-Inatsugi et al. 2016).
Second, and perhaps relatedly, unisexual salamanders lack the morphological diversity that can be associated with reproductive isolation, such as flower color (McCarthy et al.
2015) and flower scent (Gross & Schiestl 2015) in polyploid plants. In comparison, unisexual salamanders display limited morphological differences compared to the sympatric sexual species (Zeyl & Lowcock 1989).
Unisexual salamanders offer a unique system with the variation in ploidy and genome composition necessary to understand which characteristic is most related to differences in realized niche when compared to related diploid species. While the subgenomes of unisexual salamander may confer some ability to diverge ecologically from sympatric sexual species, this chapter supports a hypothesis that kleptogenesis tethers unisexuals to their sexual hosts. However, the generality of this scenario is not addressed due to a lack of similar analyses in other animal polyploids. Comparing these results to those from systems that require sperm but show less frequent introgression (e.g.
Amazon Molly; Costa & Schlupp 2010), those that do not require sexual species but have signatures of introgression (Corbicula clams; Hedtke et al. 2011), and those that reproduce completely asexually (New Zealand mudsnails; Jokela et al. 2009) would provide a comprehensive view of niche relationships in animal polyploids. While the 109
diversity of reproductive modes in animals has likely resulted in less research effort compared to plants, this diversity is strength for potentially explaining the longevity or ephemerality of non-sexual lineages (Neiman et al. 2009).
Overall, unisexual Ambystoma offer an exciting system to study the evolution of sexual reproduction. Although it remains difficult to explain how coexistence between unisexual and sexual salamanders is maintained, my work has provided exciting new avenues of research for the future. The metabolic costs associated with mitonuclear discordance may be limiting the physiological capability of unisexuals compared to sexual species (Havird et al. 2015), and quantifying the amount of mitonuclear discordance present in unisexuals with more and less frequent nuclear introgression could provide crucial information for designing studies to understand how a lack of mitonuclear co-evolution might influence phenotypes. More generally, a theme emerging from this research is that the genomes “captured” by unisexuals may be limited in their ability to diverge from their state in the sexual lineages, either due to frequent introgression (Gibbs
& Denton 2016) or some prevention of mutation accumulation and subsequent differentiation. Avoiding the negative effects of mutation accumulation is a major challenge for asexual lineages, and kleptogenesis might provide frequent enough gene flow to remedy the buildup of deleterious mutations at the cost of greater mitonuclear discordance. Future research that investigates the balance between these genomic limitations will provide the next step forward in understanding the evolution of this unique lineage and its role in the evolutionary persistence of sexual reproduction.
110 Appendix A: Supplemental Tables
111 Table 8. Loci used for Small-mouthed Salamanders (A. texanum) and unisexual Ambystoma individuals. Ranges of allele sizes amplified (in base pairs) is given for each locus-species combination. Primer sequences for specific loci are given in Julian et al. (2003), Denton et al. (2015), and Williams and DeWoody (2003). As discussed in Denton et al. (2015), where the same primer set is used to assay two different genomes there is a clear non-overlapping size distribution of allele sizes.
Genome amplified and range of allele sizes (in bp) Locus Ambystoma laterale A. jeffersonianum A. texanum 1003-626a (313-377) 1400-1610a (279-355) 1433-688a (224-256) D03b (220-264) D13b (216-236) D283b (144-216) D294b (244-280) D346b (298-302) (180-208) D378b (256-344) D46b (324-332) D94b (164-172) (228-264) Jef21a (366-382) Jef22a (364-376) Jef42a (356-364) Lat13a (200-212) Lat29a (132-140) Lat38a (257) Atex 102c (144-236) Atex 49c (112-196) Atex 65c (271-409) Atex 89c (204-341) Atex 1003c (428-488) Atex 133c (150-289) Atex 143c (148-202) Atex 74c (188-312) Atex 87c (107-254) Atex 141c (238-325) aDenton et al 2015, bJulian et al. 2003, cWilliams and DeWoody 2003
112
Appendix B: Supplemental Figures
90 Weight (g) Femur Length (mm) 80 Snout-Vent Length (mm) 70 60 50 40 30 20 Physical Measurement Physical 10 0 0 50 100 150 200 250 300 350 log Distance Figure 17. Correlations between log distance traveled during locomotor endurance trials and three potential morphological covariates.
113
Figure 18. Visualization of Principal Components Analyses using environmental data from nine Bioclim variables extracted at 1,172 sites. The loadings for PC1 were most representative of precipitation variables (Precipitation of Wettest Month, BIO13 = 0.57, Annual Precipitation, BIO12 = 0.50, Mean Diurnal Range, BIO2 = 0.46). The loadings for PC2 were most representative of temperature values (Temperature Annual Range, BIO7 = 0.57, Temperature of Driest Quarter, BIO9 = 0.43, Annual Temperature BIO1 = 0.45, Precipitation Seasonality, BIO15 = 0.44). A colored, normal probability ellipse for each group is drawn to match point color. 114
References
Ackermann RR, Bishop JM (2010) Morphological and molecular evidence reveals recent hybridization between gorilla taxa. Evolution, 64, 271–290.
Amarasekare P (2003) Competitive coexistence in spatially structured environments: a synthesis. Ecology Letters, 6, 1109–1122.
Anderson RP, Raza A (2010) The effect of the extent of the study region on GIS models of species geographic distributions and estimates of niche evolution: Preliminary tests with montane rodents (genus Nephelomys) in Venezuela. Journal of Biogeography, 37, 1378–1393.
Austin JD, Gorman T a., Bishop D, Moler P (2011) Genetic evidence of contemporary hybridization in one of North America’s rarest anurans, the Florida bog frog. Animal Conservation, 14, 553–561.
Austin C, Shaffer H. (1992) Short-, medium-, and long-term repeatability of locomotor performance in the tiger salamander Ambystoma californiense. Functional Ecology, 6, 145–153.
Avise JC (1994) Molecular Markers, Natural History, and Evolution. Chapman and Hall, New York.
Ballard J, Melvin R (2010) Linking the mitochondrial genotype to the organismal phenotype. Molecular Ecology, 19, 1523–39.
Barbet-Massin M, Jiguet F, Albert CH, Thuiller W (2012) Selecting pseudo-absences for species distribution models: how, where and how many? Methods in Ecology and Evolution, 3, 327–338.
Barrowclough GF, Groth JG, Mertz LA, Gutiérrez RJ (2005) Genetic structure, introgression, and a narrow hybrid zone between northern and California spotted owls (Strix occidentalis). Molecular Ecology, 14, 1109–1120.
115
Beaumont M, Barratt EM, Gottelli D et al. (2001) Genetic diversity and introgression in the Scottish wildcat. Molecular Ecology, 10, 319–336.
Beerli P (2006) Comparison of Bayesian and maximum-likelihood inference of population genetic parameters. Bioinformatics, 22, 341–345.
Beerli P, Palczewski M (2010) Unified framework to evaluate panmixia and migration direction among multiple sampling locations. Genetics, 185, 313–326.
Te Beest M, Le Roux JJ, Richardson DM et al. (2012) The more the better? The role of polyploidy in facilitating plant invasions. Annals of Botany, 109, 19–45.
Bennett A, Garland T, Else P (1989) Individual correlation of morphology, muscle mechanics, and locomotion in a salamander. The American Journal of Physiology, 256, R1200–R1208.
Berdahl A, Torney CJ, Schertzer E, Levin SA (2015) On the evolutionary interplay between dispersal and local adaptation in heterogeneous environments. Evolution, 69, 1390–1405.
Berry O, Tocher M, Sarre S (2004) Can assignment tests measure dispersal? Molecular Ecology, 13, 551–561.
Bestion E, Clobert J, Cote J (2015) Dispersal response to climate change: scaling down to intraspecific variation. Ecology Letters, 18, 1226–1233.
Beukeboom LW, Vrijenhoek R (1998) Evolutionary genetics and ecology of sperm- dependent parthenogenesis. Journal of Evolutionary Biology, 11, 755–782.
Bi K, Bogart JP (2006) Identification of intergenomic recombinations in unisexual salamanders of the genus Ambystoma by genomic in situ hybridization (GISH). Cytogenetic and Genome Research, 112, 307–312.
Bi K, Bogart JP (2010a) Time and time again: unisexual salamanders (genus Ambystoma) are the oldest unisexual vertebrates. BMC Evolutionary Biology, 10, 1–14.
Bi K, Bogart JP (2010b) Probing the meiotic mechanism of intergenomic exchanges by genomic in situ hybridization on lampbrush chromosomes of unisexual Ambystoma (Amphibia: Caudata). Chromosome Research, 18, 371–82.
Bi K, Bogart JP, Fu J (2008a) Genealogical relationships of southern Ontario polyploid unisexual salamanders (genus Ambystoma) inferred from intergenomic exchanges and major rDNA cytotypes. Chromosome Research, 16, 275–289.
Bi K, Bogart JP, Fu J (2008b) The prevalence of genome replacement in unisexual 116
salamanders of the genus Ambystoma (Amphibia, Caudata) revealed by nuclear gene genealogy. BMC Evolutionary Biology, 8, 1–9.
Bi K, Bogart JP, Fu J (2009) An examination of intergenomic exchanges in A. laterale- dependent unisexual salamanders in the genus Ambystoma. Cytogenetic and Genome Research, 124, 44–50.
Bitume E V., Bonte D, Ronce O, Olivieri I, Nieberding CM (2014) Dispersal distance is influenced by parental and grand-parental density. Proceedings of the Royal Society B: Biological Sciences, 281, 20141061.
Bogart JP, Bartoszek J, Noble DWA, Bi K (2009) Sex in unisexual salamanders: discovery of a new sperm donor with ancient affinities. Heredity, 103, 483–493.
Bogart JP, Bi K, Fu J, Noble DWA, Niedzwiecki JH (2007) Unisexual salamanders (genus Ambystoma) present a new reproductive mode for eukaryotes. Genome, 50, 119–136.
Bogart JP, Klemens MW (1997) Hybrids and genetic interactions of mole salamanders (Ambystoma jeffersonianum and A. laterale)(Amphibia, Caudata) in New York and New England. American Museum Novitates, 1–78.
Bogart JP, Klemens MW (2008) Additional distributional records of Ambystoma laterale, A. jeffersonianum (Amphibia: Caudata) and their unisexual kleptogens in northeastern North America. American Museum Novitates, 1–58.
Bogart JP, Licht LE (1987) Evidence for the requirement of sperm in unisexual salamander hybrids (genus Ambystoma). Canadian Field Naturalist, 101, 434–436.
Bohling JH, Adams JR, Waits LP (2012) Evaluating the ability of Bayesian clustering methods to detect hybridization and introgression using an empirical red wolf data set. Molecular Ecology, 74–86.
Bolker BM, Pacala SW (1999) Spatial Moment Equations for Plant Competition: Understanding Spatial Strategies and the Advantages of Short Dispersal. The American Naturalist, 153, 575–602.
Bolnick DI, Turelli M, López-Fernández H, Wainwright PC, Near TJ (2008) Accelerated mitochondrial evolution and “Darwin’s corollary”: asymmetric viability of reciprocal F1 hybrids in Centrarchid fishes. Genetics, 178, 1037–48.
Boria RA, Olson LE, Goodman SM, Anderson RP (2014) Spatial filtering to reduce sampling bias can improve the performance of ecological niche models. Ecological Modelling, 275, 73–77.
117
Bossu CM, Near TJ (2009) Gene trees reveal repeated instances of mitochondrial DNA introgression in orangethroat darters (percidae: etheostoma). Systematic biology, 58, 114–129.
Bowler DE, Benton TG (2005) Causes and consequences of animal dispersal strategies: relating individual behaviour to spatial dynamics. Biological Reviews, 80, 205–225.
Brelsford A, Milá B, Irwin DE (2011) Hybrid origin of Audubon’s warbler. Molecular Ecology, 20, 2380–9.
Brodman R, Krouse HD (2007) How blue-spotted and small-mouthed salamander larvae coexist with their unisexual counterparts. Herpetologica, 63, 135–143.
Broquet T, Petit EJ (2009) Molecular Estimation of Dispersal for Ecology and Population Genetics. Annual Review of Ecology, Evolution, and Systematics, 40, 193–216.
Brown JL (2014) SDMtoolbox: a python-based GIS toolkit for landscape genetic, biogeographic and species distribution model analyses (B Anderson, Ed,). Methods in Ecology and Evolution, 5, 694–700.
Calabrese JM, Certain G, Kraan C, Dormann CF (2014) Stacking species distribution models and adjusting bias by linking them to macroecological models. Global Ecology and Biogeography, 23, 99–112.
Camargo A, Morando M, Avila L, Sites J (2012) Species Delimitation with ABC and Other Coalescent-Based Methods: A Test of Accuracy with Simulations and an Empirical Example with Lizards of the Liolaemus Darwinii Complex (Squamata: Liolaemidae). Evolution, 66, 2834–2849.
Di Candia M, Routman E (2007) Cytonuclear discordance across a leopard frog contact zone. Molecular Phylogenetics and Evolution, 45, 564–575.
Carlsson J (2008) Effects of microsatellite null alleles on assignment testing. Journal of Heredity, 99, 616–623.
Carstens BC, Brennan RS, Chua V et al. (2013) Model selection as a tool for phylogeographic inference: an example from the willow Salix melanopsis. Molecular Ecology, 22, 4014–4028.
Chan KMA, Levin S (2005) Leaky prezygotic isolation and porous genomes: rapid introgression of maternally inherited DNA. Evolution, 59, 720–729.
Charney ND (2012a) Evaluating expert opinion and spatial scale in an amphibian model. Ecological Modelling, 242, 37–45.
118
Charney ND (2012b) Relating hybrid advantage and genome replacement in unisexual salamanders. Evolution, 66, 1387–1397.
Charney ND, Castorino JJ, Dobro MJ, Steely SL (2014) Embryo Development inside Female Salamander (Ambystoma jeffersonianum-laterale) Prior to Egg Laying. PLOS One, 9, e91919.
Chatfield M, Kozak K, Fitzpatrick B, Tucker P (2010) Patterns of differential introgression in a salamander hybrid zone: Inferences from genetic data and ecological niche modelling. Molecular Ecology, 19, 4265–4282.
Chen W, Bi K, Fu J (2009) Frequent mitochondrial gene introgression among high elevation Tibetan megophryid frogs revealed by conflicting gene genealogies. Molecular Ecology, 18, 2856–2876.
Clark JS, Gelfand AE, Woodall CW, Zhu K (2014) More than the sum of the parts: Forest Climate response from join specise distributions. Ecological Applications, 24, 990–999.
Cola V Di, Broennimann O, Petitpierre B et al. (2017) ecospat: an R package to support spatial analyses and modeling of species niches and distributions. Ecography, 40, 1– 14.
Compton BW, McGarigal K, Cushman SA, Gamble LR (2007) A resistant-kernel model of connectivity for amphibians that breed in vernal pools. Conservation Biology, 21, 788–799.
Cornuet JM, Piry S, Luikart G, Estoup A, Solignac M (1999) New methods employing multilocus genotypes to select or exclude populations as origins of individuals. Genetics, 153, 1989–2000.
Cosentino BJ, Schooley RL, Phillips CA (2011a) Spatial connectivity moderates the effect of predatory fish on salamander metapopulation dynamics. Ecosphere, 2, 1– 14.
Cosentino BJ, Schooley RL, Phillips CA (2011b) Connectivity of agroecosystems: Dispersal costs can vary among crops. Landscape Ecology, 26, 371–379.
Costa GC, Schlupp I (2010) Biogeography of the Amazon molly: ecological niche and range limits of an asexual hybrid species. Global Ecology and Biogeography, 2010, 442–451.
Coster SS, Babbitt KJ, Cooper A, Kovach AI (2015) Limited influence of local and landscape factors on fine-scale gene flow in two pond-breeding amphibians. Molecular Ecology, 24, 742–758. 119
Coulon A, Cosson JF, Angibault JM et al. (2004) Landscape connectivity influences gene flow in a roe deer population inhabiting a fragmented landscape: An individual- based approach. Molecular Ecology, 13, 2841–2850.
Csilléry K, Blum MGB, Gaggiotti OE, François O (2010) Approximate Bayesian Computation (ABC) in practice. Trends in Ecology and Evolution, 25, 410–418.
Darriba D, Taboada GL, Doallo R, Posada D (2012) jModelTest 2: more models, new heuristics and parallel computing. Nature methods, 9, 772.
Dawley E, Dawley R (1986) Species discrimination by chemical cues in a unisexual- bisexual complex of salamanders. Journal of Herpetology, 20, 114–116.
Demastes JW, Eastman J, East JS, Spolsky C (2007) Phylogeography of the blue-spotted salamander, Ambystoma laterale (Caudata: Ambystomatidae). The American Midland Naturalist, 157, 149–161.
Denton RD, Gibbs HL, Glenn TC (2014) Development of 31 new microsatellite loci for two mole salamanders (Ambystoma laterale and A. jeffersonianum). Conservation Genetics Resources, 7, 167–170.
Denton RD, Greenwald KR, Gibbs HL (2017) Locomotor endurance predicts differences in realized dispersal between sympatric sexual and unisexual salamanders. Functional Ecology, 31, 915–926.
Drummond A, Ashton B, Buxton S et al. (2012a) Geneious Pro 5.6.2.
Drummond AJ, Suchard M a, Xie D, Rambaut A (2012b) Bayesian phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and Evolution, 29, 1969–1973.
Eastman JM, Niedzwiecki JH, Nadler BP, Storfer A (2009) Duration and consistency of historical selection are correlated with adaptive trait evolution in the streamside salamander, Ambystoma barbouri. Evolution, 63, 2636–2647.
Edgar RC (2004) MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic acids research, 32, 1792–1797.
Elinson RP, Bogart JP, Licht LE, Lowcock LA (1992) Gynogenetic mechanisms in polyploid hybrid salamanders. Journal of Experimental Zoology, 264, 93–99.
Elith J, Leathwick JR (2009) Species Distribution Models: Ecological Explanation and Prediction Across Space and Time. Annual Review of Ecology, Evolution, and Systematics, 40, 677–697.
120
Elith J, Phillips SJ, Hastie T et al. (2011) A statistical explanation of MaxEnt for ecologists. Diversity and Distributions, 17, 43–57.
Etherington TR (2011) Python based GIS tools for landscape genetics: Visualising genetic relatedness and measuring landscape connectivity. Methods in Ecology and Evolution, 2, 52–55.
Ficetola GF, Stöck M (2016) Do hybrid-origin polyploid amphibians occupy transgressive or intermediate ecological niches compared to their diploid ancestors? Journal of Biogeography, 43, 703–715.
Fitzpatrick BM, Johnson JR, Kump DK et al. (2010) Rapid spread of invasive genes into a threatened native species. Proceedings of the National Academy of Sciences of the United States of America, 107, 3606–3610.
Fronhofer EA, Klecka J, Melián CJ, Altermatt F (2015) Condition-dependent movement and dispersal in experimental metacommunities. Ecology Letters, 18, 954–963.
Funk DJ, Omland KE (2003) Species-Level Paraphyly and Polyphyly: Frequency, Causes, and Consequences, with Insights from Animal Mitochondrial DNA. Annual Review of Ecology, Evolution, and Systematics, 34, 397–423.
Gamble LR, McGarigal K, Compton BW (2007) Fidelity and dispersal in the pond- breeding amphibian, Ambystoma opacum: implications for spatio-temporal population dynamics and conservation. Biological Conservation, 139, 247–257.
Gershoni M, Fuchs A, Shani N et al. (2010) Coevolution predicts direct interactions between mtDNA-encoded and nDNA-encoded subunits of oxidative phosphorylation complex i. Journal of Molecular Biology, 404, 158–71.
Gibbs HL, Denton RD (2016) Cryptic Sex? Estimates of genome replacement in asexual mole salamanders (Ambystoma sp.). Molecular Ecology, 25, 2805–2815.
Gibbs HL, Murphy M, Chiucchi JE (2010) Genetic identity of endangered massasauga rattlesnakes (Sistrurus sp.) in Missouri. Conservation Genetics, 12, 433–439.
Glennon KL, Rissler LJ, Church SA (2012) Ecogeographic isolation: A reproductive barrier between species and between cytotypes in Houstonia (Rubiaceae). Evolutionary Ecology, 26, 909–926.
Glennon KL, Ritchie ME, Segraves KA (2014) Evidence for shared broad-scale climatic niches of diploid and polyploid plants. Ecology Letters, 17, 574–582.
Godsoe W, Franklin J, Blanchet FG (2017) Effects of biotic interactions on modeled species’ distribution can be masked by environmental gradients. Ecology and 121
Evolution, 7, 654–664.
Gompert Z, Forister ML, Fordyce J a, Nice CC (2008) Widespread mito-nuclear discordance with evidence for introgressive hybridization and selective sweeps in Lycaeides. Molecular Ecology, 17, 5231–5244.
Gotzek D, Axen HJ, Suarez A V., Helms Cahan S, Shoemaker D (2015) Global invasion history of the tropical fire ant: A stowaway on the first global trade routes. Molecular Ecology, 24, 374–388.
Grant P, Grant B, Markert J, Keller L, Petren K (2007) Convergent evolution of Darwin’s finches caused by introgressive hybridization and selection. Evolution, 58, 1588– 1599.
Grant WS, Spies IB, Canino MF (2006) Biogeographic evidence for selection on mitochondrial DNA in North Pacific walleye pollock Theragra chalcogramma. The Journal of Heredity, 97, 571–580.
Gray MM, Weeks SC (2001) Niche breadth in clonal and sexual fish (Poeciliopsis): a test of the frozen niche variation model. Canadian Journal of Fisheries and Aquatic Sciences, 58, 1313–1318.
Greenwald KR, Denton RD, Gibbs HL (2016) Niche partitioning among sexual and unisexual Ambystoma salamanders. Ecosphere, 7, 1–17.
Greenwald KR, Gibbs HL (2012) A single nucleotide polymorphism assay for the identification of unisexual Ambystoma salamanders. Molecular Ecology, 12, 354– 362.
Greenwald KR, Gibbs HL, Waite TA (2009a) Efficacy of land-cover models in predicting isolation of marbled salamander populations in a fragmented landscape. Conservation Biology, 23, 1232–1241.
Greenwald KR, Purrenhage J, Savage W (2009b) Landcover predicts isolation in Ambystoma salamanders across region and species. Biological Conservation, 142, 2493–2500.
Gross K, Schiestl FP (2015) Are tetraploids more successful? Floral signals, reproductive success and floral isolation in mixed-ploidy populations of a terrestrial orchid. Annals of Botany, 115, 263–273.
Guindon S, Gascuel O (2003) A Simple, Fast, and Accurate Algorithm to Estimate Large Phylogenies by Maximum Likelihood. Systematic Biology, 52, 696–704.
Hanski I (2012) Dispersal and eco-evolutionary dynamics in the Glanville fritillary 122
butterfly. In: Dispersal Ecology and Evolution (eds Clobert J, Baguette M, Benton TG, Bullock JM), pp. 290–303. Oxford University Press, Oxford, England.
Harris DJ (2015) Generating realistic assemblages with a Joint Species Distribution Model. Methods in Ecology and Evolution, 6, 465–473.
Harrison JS, Burton RS (2006) Tracing hybrid incompatibilities to single amino acid substitutions. Molecular Biology and Evolution, 23, 559–564.
Havird JC, Whitehill NS, Snow CD, Sloan DB (2015) Conservative and compensatory evolution in oxidative phosphorylation complexes of angiosperms with highly divergent rates of mitochondrial genome evolution. Evolution, 3069–3081.
Hedtke SM, Glaubrecht M, Hillis DM (2011) Rare gene capture in predominantly androgenetic species. Proceedings of the National Academy of Sciences of the United States of America, 108, 9520–9524.
Heled J, Drummond AJ (2010) Bayesian inference of species trees from multilocus data. Molecular Biology and Evolution, 27, 570–580.
Hellriegel B, Reyer H (2000) Factors influencing the composition of mixed populations of a hemiclonal hybrid and its sexual host. Journal of Evolutionary Biology, 13, 906–918.
Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005) Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology, 25, 1965–1978.
Hijmans R, Guarino L, Cruz M, Rojas E (2001) Computer tools for spatial analysis of plant genetic resources data: 1. DIVA-GIS. Plant Genetic Resources Newsletter, 2001, 15–19.
Hillis D, Green D (1990) Evolutionary changes of heterogametic sex in the phylogenetic history of amphibians. Journal of evolutionary Biology, 64, 49–64.
Homer CG, Dewitz JA, Yang L et al. (2015) Completion of the 2011 National Land Cover Database for the conterminous United States - Representing a decade of land cover change information. Photorammetric Engineering and Remote Sensing, 81, 345–354.
Hubisz MJ, Falush D, Stephens M, Pritchard JK (2009) Inferring weak population structure with the assistance of sample group information. Molecular Ecology, 9, 1322–1332.
Huelsenbeck JP, Ronquist F (2001) MRBAYES: Bayesian inference of phylogenetic 123
trees. Bioinformatics, 17, 754–755.
James EA, Jordan R, Griffin PC (2013) Spatial genetic analysis of two polyploid macrophytes reveals high connectivity in a modified wetland. Freshwater Biology, 58, 2102–2113.
Johanet A, Secondi J, Lemaire C (2011) Widespread introgression does not leak into allotopy in a broad sympatric zone. Heredity, 106, 962–972.
Johnson JR, Johnson BB, Bradley Shaffer H (2010) Genotype and temperature affect locomotor performance in a tiger salamander hybrid swarm. Functional Ecology, 24, 1073–1080.
Jokela J, Dybdahl MF, Lively CM (2009) The maintenance of sex, clonal dynamics, and host-parasite coevolution in a mixed population of sexual and asexual snails. The American Naturalist, 174, S43–S53.
Jombart T (2008) adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics, 24, 1403–1405.
Jombart T, Devillard S, Balloux F (2010) Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. BMC Genetics, 11, 94.
Jonker RM, Kraus RHS, Zhang Q et al. (2013) Genetic consequences of breaking migratory traditions in barnacle geese Branta leucopsis. Molecular Ecology, 22, 5835–5847.
Julian SE, King TL, Savage WK (2003) Novel Jefferson salamander, Ambystoma jeffersonianum, microsatellite DNA markers detect population structure and hybrid complexes. Molecular Ecology, 3, 95–97.
Keck BP, Near TJ (2010) Geographic and temporal aspects of mitochondrial replacement in Nothonotus darters (Teleostei: Percidae: Etheostomatinae). Evolution, 64, 1410– 1428.
Kenyon L, Moraes CT (1997) Expanding the functional human mitochondrial DNA database by the establishment of primate xenomitochondrial cybrids. Proceedings of the National Academy of Sciences of the United States of America, 94, 9131–9135.
Kerr B, Neuhauser C, Bohannan BJM, Dean AM (2006) Local migration promotes competitive restraint in a host-pathogen “tragedy of the commons”. Nature, 442, 75– 78.
Kirchheimer B, Schinkel CCF, Dellinger AS et al. (2016) A matter of scale: Apparent 124
niche differentiation of diploid and tetraploid plants may depend on extent and grain of analysis. Journal of Biogeography, 43, 716–726.
Kirkendall LR (1990) Ecological and evolutionary stability of sperm-dependent parthenogenesis: effects of partial niche overlap between sexual and asexual females. Evolution, 44, 698–714.
Knowles LL, Carstens BC (2007) Delimiting species without monophyletic gene trees. Systematic biology, 56, 887–95.
Kokko H, Heubel KU, Rankin DJ (2008) How populations persist when asexuality requires sex: the spatial dynamics of coping with sperm parasites. Proceedings of the Royal Society B: Biological Sciences, 275, 817–825.
Kraus F (1985) Unisexual salamander lineages in northwestern Ohio and southeastern Michigan: A study of the consequences of hybridization. Copeia, 1985, 309–324.
Kraus RHS, Van Hooft P, Megens HJ et al. (2013) Global lack of flyway structure in a cosmopolitan bird revealed by a genome wide survey of single nucleotide polymorphisms. Molecular Ecology, 22, 41–55.
Kraus F, Petranka JW (1989) A new sibling species of Ambystoma from the Ohio River drainage. Copeia, 1989, 94–110.
Kubatko LS (2009) Identifying hybridization events in the presence of coalescence via model selection. Systematic Biology, 58, 478–488.
Kubatko LS, Gibbs HL, Bloomquist EW (2011) Inferring species-level phylogenies and taxonomic distinctiveness using multilocus data in sistrurus rattlesnakes. Systematic Biology, 60, 393–409.
Kuzmin SL (1999) The Amphibians of the Former Soviet Union. Pensoft, Sofia-Moscow.
Lampert KP, Schartl M (2010) A little bit is better than nothing: the incomplete parthenogenesis of salamanders, frogs and fish. BMC Biology, 8, 78.
Lane N (2011) Mitonuclear match: Optimizing fitness and fertility over generations drives ageing within generations. BioEssays, 33, 860–869.
Larget BR, Kotha SK, Dewey CN, Ané C (2010) BUCKy: gene tree/species tree reconciliation with Bayesian concordance analysis. Bioinformatics, 26, 2910–2911.
Lee-Yaw JA, Sechley TH, Irwin DE (2015) Conflicting effects of microhabitats on Long- toed Salamander (Ambystoma macrodactylum) movement: implications for landscape connectivity. Canadian Journal of Zoology, 93, 1–7. 125
Levin DA (1975) Minority Cytotype Exclusion in Local Plant Populations. Taxon, 24, 35–43.
Levin DA (2003) The ecological transition in speciation. New Phytologist, 161, 91–96. Levins R (1968) Evolution in Changing Environments. Princeton University Press, Princeton, New Jersey.
Librado P, Rozas J (2009) DnaSP v5: a software for comprehensive analysis of DNA polymorphism data. Bioinformatics, 25, 1451–1452.
Linder HP, Barker NP (2014) Does polyploidy facilitate long-distance dispersal? Annals of Botany, 113, 1175–1183.
Linnen CR, Farrell BD (2007) Mitonuclear discordance is caused by rampant mitochondrial introgression in Neodiprion (Hymenoptera: Diprionidae) sawflies. Evolution, 61, 1417–1438.
Lively CM (2010) A review of Red Queen models for the persistence of obligate sexual reproduction. The Journal of Heredity, 101, S13-20.
Lively CM, Dybdahl MF (2000) Parasite adaptation to locally common host genotypes. Nature, 405, 679–681.
Llewelyn J, Phillips BL, Alford RA, Schwarzkopf L, Shine R (2010) Locomotor performance in an invasive species: cane toads from the invasion front have greater endurance, but not speed, compared to conspecifics from a long-colonised area. Oecologia, 162, 343–348.
Lobo JM, Jiménez-valverde A, Real R (2008) AUC: A misleading measure of the performance of predictive distribution models. Global Ecology and Biogeography, 17, 145–151.
Lowe WH, McPeek MA (2014) Is dispersal neutral? Trends in Ecology and Evolution, 29, 444–450.
Madlung A (2012) Polyploidy and its effect on evolutionary success: old questions revisited with new tools. Heredity, 110, 99–104.
Mallet J (2007) Hybrid speciation. Nature, 446, 279–83.
Marchant DB, Soltis DE, Soltis PS (2016) Patterns of abiotic niche shifts in allopolyploids relative to their progenitors. New Phytologist, 212, 708–718.
Mau M, Lovell JT, Corral JM et al. (2015) Hybrid apomicts trapped in the ecological 126
niches of their sexual ancestors. Proceedings of the National Academy of Sciences of the United States of America, 112, E2357–E2365.
Maynard Smith J (1978) The Evolution of Sex. Cambridge University Press, Cambridge. McCarthy EW, Arnold SEJ, Chittka L et al. (2015) The effect of polyploidy and hybridization on the evolution of floral colour in Nicotiana (Solanaceae). Annals of Botany, 115, 1117–1131.
McElroy KE, Denton RD, Sharbrough J et al. (2017) Genome Expression Balance in a Triploid Trihybrid Vertebrate. Genome Biology and Evolution, 9, 968–980.
McGuire JA, Linkem CW, Koo MS et al. (2007) Mitochondrial introgression and incomplete lineage sorting through space and time: phylogenetics of crotaphytid lizards. Evolution, 61, 2879–97.
McIntyre PJ (2012) Polyploidy associated with altered and broader ecological niches in the claytonia perfoliata (portulacaceae) species complex. American Journal of Botany, 99, 655–662.
McKenzie M, Chiotis M, Pinkert CA, Trounce IA (2003) Functional respiratory chain analyses in murid xenomitochondrial cybrids expose coevolutionary constraints of cytochrome b and nuclear subunits of complex III. Molecular Biology and Evolution, 20, 1117–1124.
McKnight ML, Shaffer HB (1997) Large, rapidly evolving intergenic spacers in the mitochondrial DNA of the salamander family Ambystomatidae (Amphibia: Caudata). Molecular Biology and Evolution, 14, 1167–1176.
Mee JA, Noddin F, Hanisch JR, Tonn WM, Paszkowski CA (2013) Diets of sexual and sperm-dependent asexual dace (Chrosomus spp.): Relevance to niche differentiation and mate choice hypotheses for coexistence. Oikos, 122, 998–1008.
Mee JA, Otto SP (2010) Variation in the strength of male mate choice allows long-term coexistence of sperm-dependent asexuals and their sexual hosts. Evolution, 64, 2808–2819.
Melo-Ferreira J, Alves PC, Freitas H, Ferrand N, Boursot P (2009) The genomic legacy from the extinct Lepus timidus to the three hare species of Iberia: contrast between mtDNA, sex chromosomes and autosomes. Molecular Ecology, 18, 2643–58.
Melo-Ferreira J, Boursot P, Carneiro M et al. (2012) Recurrent introgression of mitochondrial DNA among hares (Lepus spp.) revealed by species-tree inference and coalescent simulations. Systematic biology, 61, 367–381.
Merow C, Smith MJ, Silander JA (2013) A practical guide to MaxEnt for modeling 127
species’ distributions: what it does, and why inputs and settings matter. Ecography, 36, 1058–1069.
Micheletti SJ, Storfer A (2015) A test of the central-marginal hypothesis using population genetics and ecological niche modelling in an endemic salamander (Ambystoma barbouri). Molecular Ecology, 24, 967–979.
Michels E, Cottenie K, Neys L et al. (2001) Geographical and genetic distances among zooplankton populations in a set of interconnected ponds: A plea for using GIS modelling of the effective geographical distance. Molecular Ecology, 10, 1929– 1938.
Monaghan P, Metcalfe NB, Torres R (2009) Oxidative stress as a mediator of life history trade-offs: mechanisms, measurements and interpretation. Ecology letters, 12, 75– 92.
Moore WS, Miller RT, Schultz RJ (1970) Distribution, Adaptation and Probable Origin of an All-Female Form of Poeciliopsis (Pisces: Poecilidae) in Northwestern Mexico. Evolution, 24, 789–795.
Morales NS, Fernandez IC, Baca-Gonzalez V (2017) MaxEnt′s parameter configuration and small samples: Are we paying attention to recommendations? PeerJ, 5, e3093.
Moyer G, Winemeller K, McPhee M, Turner T (2005) Historical demography, selection, and coalescence of mitochondrial and nuclear genes in Prochilodus species of northern South America. Evolution, 59, 599–610.
Narbonne P, Simpson DE, Gurdon JB (2011) Deficient induction response in a Xenopus nucleocytoplasmic hybrid. PLoS Biology, 9, e1001197.
Nathan R, Getz WM, Revilla E et al. (2008) A movement ecology paradigm for unifying organismal movement research. Proceedings of the National Academy of Sciences of the United States of America, 105, 19052–19059.
Neigel J, Avise J (1986) Phylogenetic relationships of mitochondrial DNA under various demographic models of speciation. In: Evolutionary Processes and Theory (eds Nevo E, Karlin S), pp. 515–534. Academic Press, New York.
Neiman M, Meirmans S, Meirmans PG (2009) What can asexual lineage age tell us about the maintenance of sex? Annals of the New York Academy of Sciences, 1168, 185– 200.
Nevado B, Fazalova V, Backeljau T, Hanssens M, Verheyen E (2011) Repeated unidirectional introgression of nuclear and mitochondrial DNA between four congeneric Tanganyikan cichlids. Molecular Biology and Evolution, 28, 2253–2267. 128
Newman RA, Squire T (2001) Microsatellite variation and fine-scale population structure in the wood frog (Rana sylvatica). Molecular Ecology, 10, 1087–1100.
Niedzwiecki JH (2005) Evolutionary history and hybridization of two mole salamander sister species from different habitats. University of Kentucky.
Noël S, Labonté P, Lapointe F (2011) Genomotype frequencies and genetic diversity in urban and protected populations of blue-spotted salamanders (Ambystoma laterale) and related unisexuals. Journal of Herpetology, 45, 294–299.
Olson RR, McPherson R (1987) Potential vs. realized larval dispersal: fish predation on larvae of the ascidian Lissoclinum patella (Gottschaldt). Journal of Experimental Marine Biology and Ecology, 110, 245–256.
Omernik JM (1987) Map Supplement: Ecoregions of the Conterminous United States. Annals of the Association of American Geographers, 77, 118–125.
Van Oosterhout C, Hutchinson WF, Wills DPM, Shipley P (2004) Micro-Checker: Software for Identifying and Correcting Genotyping Errors in Microsatellite Data. Molecular Ecology, 4, 535–538.
Otto SP, Whitton J (2000) Polyploid incidence and evolution. Annual Review of Genetics, 34, 401–437.
Paetkau D, Slade R, Burden M, Estoup A (2004) Genetic assignment methods for the direct, real-time estimation of migration rate: a simulation-based exploration of accuracy and power. Molecular Ecology, 13, 55–65.
Parham JF, Papenfuss TJ, Dijk PP Van et al. (2013) Genetic introgression and hybridization in Antillean freshwater turtles (Trachemys) revealed by coalescent analyses of mitochondrial and cloned nuclear markers. Molecular Phylogenetics and Evolution, 67, 176–187.
Parisod C, Broennimann O (2016) Towards unified hypotheses of the impact of polyploidy on ecological niches. New Phytologist, 212, 540–542.
Parmelee JR (1993) Microhabitat segregation and spatial relationships among four species of mole salamander (genus Ambystoma). Occasional Papers of the Museum of Natural History, University of Kansas, 160, 1–33.
Pastorini J, Zaramody A, Curtis DJ, Nievergelt CM, Mundy NI (2009) Genetic analysis of hybridization and introgression between wild mongoose and brown lemurs. BMC Evolutionary Biology, 9, 32.
129
Peakall R, Smouse PE (2006) GENALEX 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology, 6, 288–295.
Peakall R, Smouse PE (2012) GenAlEx 6.5: genetic analysis in Excel. Population genetic software for teaching and research-an update. Bioinformatics, 28, 2537–2539.
Peterman WE, Anderson TL, Ousterhout BH et al. (2015) Differential dispersal shapes population structure and patterns of genetic differentiation in two sympatric pond breeding salamanders. Conservation Genetics, 16, 59–69.
Petit RJ, Excoffier L (2009) Gene flow and species delimitation. Trends in Ecology and Evolution, 24, 386–393.
Petranka J (1984) Breeding migrations, breeding season, clutch size, and oviposition of stream-breeding Ambystoma texanum. Journal of herpetology, 18, 106–112.
Petranka JW (1998) Salamanders of the United States and Canada. Smithsonian Institution Press, Washington, D.C.
Pfennig KS (2007) Facultative mate choice drives adaptive hybridization. Science, 318, 965–967.
Phillips SJ, Anderson RP, Schapire RE (2006a) Maximum entropy modeling of species geographic distributions. Ecological Modelling, 190, 231–259.
Phillips BL, Brown GP, Webb JK, Shine R (2006b) Invasion and the evolution of speed in toads. Nature, 439, 803–803.
Piry S, Alapetite A, Cornuet JM et al. (2004) GENECLASS2: a software for genetic assignment and first-generation migrant detection. The Journal of Heredity, 95, 536–539.
Pollock LJ, Tingley R, Morris WK et al. (2014) Understanding co-occurrence by modelling species simultaneously with a Joint Species Distribution Model (JSDM). Methods in Ecology and Evolution, 5, 397–406.
Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics, 155, 945–959.
Purrenhage JL, Niewiarowski PH, Moore FB-G (2009) Population structure of spotted salamanders (Ambystoma maculatum) in a fragmented landscape. Molecular Ecology, 18, 235–247.
Quesada H, Wenne R, Skibinski DO (1999) Interspecies transfer of female mitochondrial DNA is coupled with role-reversals and departure from neutrality in the mussel 130
Mytilus trossulus. Molecular Biology and Evolution, 16, 655–65.
R Development Core Team (2011) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.
Ramsden C (2008) Population genetics of Ambystoma jeffersonianum and sympatric unisexuals reveal signatures of both gynogenetic and sexual reproduction. Copeia, 2008, 586–594.
Ramsden C, Beriault K, Bogart JP (2006) A nonlethal method of identification of Ambystoma laterale, A. jeffersonianum and sympatric unisexuals. Molecular Ecology, 6, 261–264.
Raymond M, Rousset F (1995) GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. Journal of Heredity, 86, 248–249.
Renny-Byfield S, Wendel JF (2014) Doubling down on genomes: Polyploidy and crop plants. American Journal of Botany, 101, 1711–1725.
Robertson AV, Ramsden C, Niedzwiecki JH, Fu J, Bogart JP (2006) An unexpected recent ancestor of unisexual Ambystoma. Molecular Ecology, 15, 3339–3351.
Rocque DA, Ben-David M, Barry RP, Winker K (2006) Assigning birds to wintering and breeding grounds using stable isotopes: Lessons from two feather generations among three intercontinental migrants. Journal of Ornithology, 147, 395–404.
Rohwer S, Bermingham E, Wood C (2001) Plumage and mitochondrial DNA haplotype variation across a moving hybrid zone. Evolution, 55, 405–422.
Rollins LA, Richardson MF, Shine R (2015) A genetic perspective on rapid evolution in cane toads (Rhinella marina). Molecular Ecology, 24, 2264–2276.
Rosenberg N a. (2003) Distruct: a Program for the Graphical Display of Population Structure. Molecular Ecology, 4, 137–138.
Rousset F (2008) GENEPOP’007: a complete re-implementation of the genepop software for Windows and Linux. Molecular Ecology, 8, 103–106.
Rubinoff D (2006) Utility of Mitochondrial DNA Barcodes in Species Conservation. Conservation Biology, 20, 1026–1033.
Ruiz-Pesini E, Mishmar D, Brandon M, Procaccio V, Wallace DC (2004) Effects of purifying and adaptive selection on regional variation in human mtDNA. Science, 303, 223–226.
131
Ryan ME, Johnson JR, Fitzpatrick BM (2009) Invasive hybrid tiger salamander genotypes impact native amphibians. Proceedings of the National Academy of Sciences of the United States of America, 106, 11166–11171.
Salomon Y, Connolly SR, Bode L (2010) Effects of asymmetric dispersal on the coexistence of competing species. Ecology Letters, 13, 432–441.
Schley D, Doncaster CP, Sluckin T (2004) Population models of sperm-dependent parthenogenesis. Journal of Theoretical Biology, 229, 559–572.
Schlosser I, Doeringsfeld M, Elder J, Arzayus L (1998) Niche relationships of clonal and sexual fish in a heterogeneous landscape. Ecology, 79, 953–968.
Schlupp I (2005) The Evolutionary Ecology of Gynogenesis. Annual Review of Ecology, Evolution, and Systematics, 36, 399–417.
Schoener TW (1968) The Anolis Lizards of Bimini: Resource Partitioning in a Complex Fauna. Ecology, 49, 704–726.
Searcy CA, Shaffer H. (2008) Calculating biologically accurate mitigation credits: insights from the California tiger salamander. Conservation Biology, 22, 997–1005.
Searcy CA, Shaffer HB (2016) Do Ecological Niche Models Accurately Identify Climatic Determinants of Species Ranges? The American Naturalist, 187, 423–435.
Semlitsch RD (2008) Differentiating Migration and Dispersal Processes for Pond- Breeding Amphibians. Journal of Wildlife Management, 72, 260–267.
Semlitsch RD, Scott DE, Pechmann JHK, Gibbons JW (1996) Structure and dynamics of an amphibian community: evidence from a 16-year study of a natural pond. In: Long-term studies of vertebrate communities (eds Cody ML, Smallwood JA), pp. 217–247. Academic Press, Sand Diego, California.
Sequeira F, Sodre D, Ferrand N et al. (2011) Hybridization and massive mtDNA unidirectional introgression between the closely related Neotropical toads Rhinella marina and R. schneideri inferred from mtDNA and nuclear markers. BMC Evolutionary Biology, 11, 264.
Shaffer H., McKnight M (1996) The polytypic species revisited: genetic differentiation and molecular phylogenetics of the tiger salamander Ambystoma tigrinum (Amphibia: Caudata) complex. Evolution, 50, 417–433.
Shimizu-Inatsugi R, Terada A, Hirose K et al. (2016) Plant adaptive radiation mediated by polyploid plasticity in transcriptomes. Molecular Ecology, 26, 193–207.
132
Shine R, Brown GP, Phillips BL (2011) An evolutionary process that assembles phenotypes through space rather than through time. Proceedings of the National Academy of Sciences of the United States of America, 108, 5708–5711.
Smith MA, Green DM (2005) Dispersal and the metapopulation paradigm in amphibian ecology and conservation: are all amphibian populations metapopulations? Ecography, 28, 110–128.
Smith JJ, Putta S, Walker J a et al. (2005) Sal-Site: integrating new and existing ambystomatid salamander research and informational resources. BMC Genomics, 6, 181.
Soltis PS, Liu X, Marchant DB, Visger CJ, Soltis DE (2014) Polyploidy and novelty: Gottlieb’s legacy. Philosophical Transactions of the Royal Society B: Biological Sciences, 369, 20130351.
Soltis DE, Visger CJ, Marchant DB, Soltis PS (2016) Polyploidy: Pitfalls and paths to a paradigm. American Journal of Botany, 103, 1–21.
Spinks PQ, Shaffer H. (2009) Conflicting mitochondrial and nuclear phylogenies for the widely disjunct Emys (Testudines: Emydidae) species complex, and what they tell us about biogeography and hybridization. Systematic biology, 58, 1–20.
Stephens M, Donnelly P (2003) A comparison of bayesian methods for haplotype reconstruction from population genotype data. American journal of human genetics, 73, 1162–1169.
Stephens M, Smith NJ, Donnelly P (2001) A new statistical method for haplotype reconstruction from population data. American journal of human genetics, 68, 978– 989.
Stevens VM, Polus E, Wesselingh R a., Schtickzelle N, Baguette M (2004) Quantifying functional connectivity: Experimental evidence for patch-specific resistance in the Natterjack toad (Bufo calamita). Landscape Ecology, 19, 829–842.
Storfer A, Sih A (1998) Gene flow and ineffective antipredator behavior in a stream- breeding salamander. Evolution, 52, 558–565.
Strasburg JL, Rieseberg LH (2010) How robust are “isolation with migration” analyses to violations of the im model? A simulation study. Molecular Biology and Evolution, 27, 297–310.
Susnik S, Berrebi P, Dovc P, Hansen MM, Snoj A (2004) Genetic introgression between wild and stocked salmonids and the prospects for using molecular markers in population rehabilitation: the case of the Adriatic grayling (Thymallus thymallus L. 133
1785). Heredity, 93, 273–282.
Theodoridis S, Randin C, Broennimann O, Patsiou T, Conti E (2013) Divergent and narrower climatic niches characterize polyploid species of european primroses in primula sect. aleuritia. Journal of Biogeography, 40, 1278–1289.
Toews D, Brelsford A (2012) The biogeography of mitochondrial and nuclear discordance in animals. Molecular Ecology, 21, 3907–3930.
Toews DPL, Mandic M, Richards JG, Irwin DE (2013) Migration, Mitochondria, and the Yellow-Rumped Warbler. Evolution, 68, 241–255.
Tracy CR, Christian KA, Baldwin J, Phillips BL (2012) Cane toads lack physiological enhancements for dispersal at the invasive front in Northern Australia. Biology Open, 1, 37–42.
Trenham P, Koenig W, Shaffer H. (2001) Spatially autocorrelated demography and interpond dispersal in the salamander Ambystoma californiense. Ecology, 82, 3519– 3530.
USGS (2012) Ambystoma texanum, Ambystoma barbouri. National Amphibian Atlas, www.pwrc.usgs.gov/naa.
Uzzell T (1969) Notes on spermatophore production by salamanders of the Ambystoma jeffersonianum complex. Copeia, 1969, 602–612.
Veith M, Baumgart A, Dubois A et al. (2012) Discordant patterns of nuclear and mitochondrial introgression in Iberian populations of the European common frog (Rana temporaria). The Journal of Heredity, 103, 240–249.
Venesky M, Parris M (2009) Intraspecific variation in life history traits among two forms of Ambystoma barbouri larvae. The American Midland Naturalist, 162, 195–199.
Vieites DR, Min M-S, Wake DB (2007) Rapid diversification and dispersal during periods of global warming by plethodontid salamanders. Proceedings of the National Academy of Sciences of the United States of America, 104, 19903–19907.
Vrijenhoek RC (1978) Coexistence of clones in a heterogeneous environment. Science, 199, 549–552.
Vrijenhoek RC (1984) Ecological differntiation among clones: the frozen niche variation model. In: Population biology and evolution (eds Wohrmann K, Loeschke V), pp. 217–231. Springer-Verlag, Berlin.
Vrijenhoek RC, Parker Jr ED (2009) Geographical Parthenogenesis: General Purpose 134
Genotypes and Frozen Niche Variation. In: Lost Sex (eds Schön I, Martens K, Dijk PP Van), pp. 99–131. Springer Publications, Berlin.
Vrijenhoek R, Pfeiler E (1997) Differential survival of sexual and asexual Poeciliopsis during environmental stress. Evolution, 51, 1593–1600.
Warren DL (2012) In defense of “niche modeling”. Trends in Ecology and Evolution, 27, 497–500.
Warren DL, Glor RE, Turelli M (2010) ENMTools: A toolbox for comparative studies of environmental niche models. Ecography, 33, 607–611.
Warton DI, Blanchet FG, O’Hara RB et al. (2015) So Many Variables: Joint Modeling in Community Ecology. Trends in Ecology and Evolution, 30, 766–779.
Weinzierl RP, Beukeboom LW, Gerace L, Michiels NK (1999) Spatial and ecological overlap between coexisting sexual and parthenogenetic Schmidtea polychroa (Tricladida; Platyhelminthes). Hydrobiologia, 392, 179–185.
Weissgerber TL, Milic NM, Winham SJ, Garovic VD (2015) Beyond Bar and Line Graphs: Time for a New Data Presentation Paradigm. PLOS Biology, 13, e1002128.
Wetherington JD, Weeks SC, Kotora KE, Vrijenhoek RC (1989) Genotypic and environment components of variation in growth and reproduction of fish hemiclones (Poeciliopsis: Poeciliidae). Evolution, 43, 635–645.
Weyrauch SL, Grubb Jr. TC (2004) Patch and landscape characteristics associated with the distribution of woodland amphibians in an agricultural fragmented landscape: an information-theoretic approach. Biological Conservation, 115, 443–450.
Williams RN, DeWoody JA (2003) Fluorescent dUTP helps characterize 10 novel tetranucleotide microsatellites from an enriched salamander (Ambystoma texanum) genomic library. Molecular Ecology, 4, 17–19.
Woerner AE, Cox MP, Hammer MF (2007) Recombination-filtered genomic datasets by information maximization. Bioinformatics, 23, 1851–3.
Wolff JN, Ladoukakis ED, Enriquez JA, Dowling DK (2014) Mitonuclear interactions: evolutionary consequences over multiple biological scales. Philosophical Transactions of the Royal Society B: Biological Sciences, 369, 20130443.
Wunder MB (2012) Determining geographic patterns of migration and dispersal using stable isotopes in keratins. Journal of Mammalogy, 93, 360–367.
Zamudio KR, Wieczorek AM (2007) Fine-scale spatial genetic structure and dispersal 135
among spotted salamander (Ambystoma maculatum) breeding populations. Molecular Ecology, 16, 257–274.
Zeller KA, McGarigal K, Whiteley AR (2012) Estimating landscape resistance to movement: A review. Landscape Ecology, 27, 777–797.
Zeyl C, Lowcock L (1989) A morphometric analysis of hybrid salamanders (genus Ambystoma) and their progenitors on Kelleys Island in Lake Erie. Canadian Journal of Zoology, 67, 2376–2383.
Zhang F, Broughton RE (2013) Mitochondrial-Nuclear Interactions: Compensatory Evolution or Variable Functional Constraint among Vertebrate Oxidative Phosphorylation Genes? Genome Biology and Evolution, 5, 1781–1791.
Zieliński P, Nadachowska-Brzyska K, Wielstra B et al. (2013) No evidence for nuclear introgression despite complete mtDNA replacement in the Carpathian newt (Lissotriton montandoni). Molecular Ecology, 22, 1884–1903.
136