Conservation Genetics of the Tiger Rattlesnake (Crotalus Tigris) in the Context of Long-Term Ecological Data

Conservation Genetics of the Tiger Rattlesnake (Crotalus tigris) in the Context of Long-term Ecological Data

Item Type text; Electronic Dissertation

Authors Goode, Matt

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 11/10/2021 08:11:22

Link to Item http://hdl.handle.net/10150/578413

CONSERVATION GENETICS OF THE TIGER RATTLESNAKE (CROTALUS TIGRIS) IN THE CONTEXT OF LONG-TERM ECOLOGICAL DATA

Matt Goode

______

A Dissertation Submitted to the Faculty of the

SCHOOL OF NATURAL RESOURCES AND THE ENVIRONMENT

In Partial Fulfillment of the Requirements for the Degree of

DOCTOR OF PHILOSOPHY

With a Major in Wildlife and Fisheries Science

In the Graduate College of

THE UNIVERSITY OF ARIZONA

2015

THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE

As members of the Dissertation Committee, we certify that we have read the dissertation prepared by Matt Goode, titled Conservation Genetics of the Tiger Rattlesnake (Crotalus tigris) in the Context of Long-term Ecological Data, and recommend that it be accepted as fulfilling the dissertation requirement for the Degree of Doctor of Philosophy.

______Date: May 15, 2015 Melanie Culver

______Date: May 15, 2015 John Koprowski

______Date: May 15, 2015 Mike Rosenzweig

______Date: May 15, 2015 Cecil Schwalbe

Final approval and acceptance of this dissertation is contingent upon the candidate’s submission of the final copies of the dissertation to the Graduate College.

I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.

______Date: May 15, 2015 Dissertation Director: Melanie Culver

STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of the requirements for an advanced degree at the University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that an accurate acknowledgement of the source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author.

SIGNED: Matt Goode

ACKNOWLEDGEMENTS

I have stood on the shoulders of many people throughout the years it took to complete my PhD. Literally dozens of people have helped me with virtually every aspect of this work, and I would be remiss if I failed to acknowledge their important contributions. I thank Dave Duvall for inspiring me to become a scientist, which required a rearrangement of my thought process, having been an English major who was more turned on by literature than natural history and ecology. I am indebted to Cecil Schwalbe and his graduate students, Don Swann, Roy Averill-Murray, and Dale Turner for helping me understand the Sonoran Desert in ways that I may have never come to understand on my own. I am especially thankful for the help of many dedicated field assistants who worked long hours under often difficult conditions. Those deserving special mention are Kristin Albert, Melissa Amarello, Chip Cochran, Tony Dee, Ed Kuklinski, Jeff Smith and Mike Wall. Many other people captured rattlesnakes for genetic analyses, including Danny Brower†, Phil Healey, and Larry Jones. A few of the people who worked with me contributed intellectually to the research, including Melissa Amarello, and Jeff Smith. And without the help of one person in particular, Mickey Parker, I’m pretty sure my research program would have completely come off the rails. I could always count on Mickey’s hard work and dedication matter what else might go wrong. I greatly appreciate the help of Anna Carlson, Rulon Clark, Taylor Edwards, Bob Fitak, Caren Goldberg, Matt Kaplan, Adrian Munguia-Vega, Ashwin Naidu, and Karla Pelz-Serrano who helped with lab work and interpretation of genetic data. Above all, Alex Ochoa took the time to make sure that what I was doing made sense and was in line with current approaches in the rapidly changing field of conservation genetics. I had the great fortune to have several people who supported my research program by providing me with the infrastructure needed to make it work. Larry Norris, National Park Service, was very generous in his support, providing funding that allowed me to establish a sustainable operation. Brooks Jeffery, College of Architecture, Planning and Landscape Architecture, provided me with office and lab space even though I had no official affiliation with his academic unit. Mike Rosenzweig, Department of Ecology and Evolutionary Biology, provided me with lab space on Tumamoc Hill. John Koprowski welcomed me into his lab even though he and his students work primarily on mammals, and solidified my belief that behavioral ecology was critical to conservation. Cecil Schwalbe took me in as a Master’s student, and his belief in me provided the encouragement I needed to pursue my passion for herpetology. And finally, I extend my sincerest thanks to Melanie Culver, my PhD advisor, who supported me no matter what. Melanie’s unquestioned dedication to her graduate students is something I will never forget. I thank the staff at Saguaro National Park, Dick Maes of Vistoso Partners, the Stone Canyon Golf Club, and the residents of Stone Canyon for allowing us to conduct research on their land. Financial support came from Arizona Game and Fish Department, National Park Service, Western National Parks Association, United States Golf Association, National Fish and Wildlife Foundation, Reid Park Zoological Society, and Denver Zoological Foundation.

DEDICATION

I dedicate this dissertation, and all that went into completing it, to one person above all others, my life partner, Andrea Tuijl. I know, beyond a shadow of a doubt, that I would never have succeeded, in academia or in life, without her unfailing support.

I would like to dedicate this work to my daughter, Remi Tuijl-Goode, whose very existence has taught me more about life than anything else ever could. Remi, you are an amazing person, destined to live a life that will shine a bright light on all who know you.

In addition, I dedicate this document to my parents, Harvey and Mimi Goode, together for 65 years as of this writing. These humble people taught me, purely by example, that honesty, integrity, hard work, and above all, generosity, are the hallmarks of a fulfilled life.

And finally, I dedicate this work to Charles W. Painter (1949-2015) who passed away a few days before my defense after a long and courageous battle with cancer. Charlie had a huge impact on my work and on the way I view life, as he did on so many others. No finer man will ever walk this earth.

TABLE OF CONTENTS

ABSTRACT…………………………………………………………………………………….....7

SUMMARY……………………………………………………………………………………….8

LITERATURE CITED…………………………………………………………………………..13

APPENDIX A Population genetic structure in a habitat specialist rattlesnake (Crotalus tigris); manuscript prepared for publication – Conservation Genetics…………………………………..15

APPENDIX B Sex-biased dispersal, seasonal habitat use and mating system dynamics contribute to fine-scale genetic structure in the Tiger Rattlesnake (Crotalus tigris); manuscript prepared for publication – Molecular Ecology...………………………………………………………………38

APPENDIX C Movement patterns and reproductive behavior drive sex-based differences in fine- scale genetic structure and parentage in a polygynandrous rattlesnake (Crotalus tigris); manuscript prepared for publication – PLoS ONE………………………………………………72

ABSTRACT

I combined long-term ecological data and population genetic data using microsatellite DNA markers to examine among- and within-population genetic structure and parentage in Tiger Rattlesnake (Crotalus tigris) populations from the Tucson Basin of southern Arizona located in the northern Sonoran Desert. Based on long-term data from radio telemetry, I determined that C. tigris show strong fidelity to both their home range and winter shelter sites, remaining in close proximity to rocky habitats within mountain ranges, which leads to apparent natural isolation of populations. Therefore, I predicted that C. tigris populations would show substantial genetic differentiation among mountain ranges. However, Bayesian clustering analyses revealed a surprising pattern of extensive admixture among mountain ranges, indicating the presence of gene flow among populations. This pattern of genetic admixture can likely be explained by historical changes in climate and physiognomy in the Sonoran Desert. Analyses of pack rat midden remains clearly show that mountain ranges were previously connected by mesic woodland habitats that may have led to panmixia in C. tigris populations as recently as 5,000- 8,000 years ago. At present, C. tigris show a strong preference for xeroriparian washes, which allows individuals to occasionally move relatively long distances, likely resulting in contemporary gene flow. To maintain connections among mountain ranges, I recommend effective management, protection, and restoration (if needed) of wash habitats, which also act as corridors for a suite of other species. At the within population scale, genetic clustering analyses revealed the existence of fine-scale genetic structure in C. tigris subpopulations located in the Rincon Mountains. Further analyses based on location data of individuals indicated the existence of a potential barrier to gene flow, which corresponded to a watershed divide. Although the watershed divide would appear not to present a physical barrier to gene flow, it likely acts to segregate populations based on habitat and movement preferences associated with wash habitats. Data on spatial ecology and reproductive behavior, indicate that C. tigris distribute gametes across the landscape in the absence of actual displacement of individuals due to fidelity to home ranges and winter shelter sites. Analyses of parentage were constrained by the difficulty in obtaining offspring from gravid female C. tigris that give birth deep in rock outcrops. However, I did conduct analyses on over 30 offspring from known mothers and nearly 60 free-ranging offspring found while conducting ecological research. Surprisingly, not a single male C. tigris found courting or copulating with a female was identified as the father, indicating that reproductive behavior is a poor predictor of parentage, and therefore, fitness. Interestingly, males identified as fathers were found up to 2 km distance from their offspring, demonstrating that males from surrounding areas may move relatively long distances to mate. The mating system of C. tigris, which is characterized by promiscuity in both sexes, appears to drive dramatic differences in spatial ecology between males and females, and may lead to fine-scale genetic structure among females and not males who spend a great deal of time searching for receptive females.

SUMMARY

The field of conservation genetics is relatively new, beginning with the recognition of the diverse role that genetics plays in understanding the biology of species that are threatened with extinction (Schonewald-Cox et al. 1983). The rise of conservation genetics roughly coincided with an increasing awareness that human activities were leading to a biodiversity crises (Soulè 1985). Importantly, the potential role of genetics in dealing with this crisis was explicit in early treatments (Frankel & Soulè 1981). However, scientists argued about the role of genetics in conservation, contending that ecology, and demography in particular, were more important in endangered species management, perhaps resulting from the failure to realize that genetics had much more to offer than simply describing the genetic diversity of a population (Lande 1988, Caro & Laurenson 1994). Ultimately, the advent of molecular techniques, largely in the last decade of the 20th century, led to the rise of conservation genetics as a mainstream discipline, marked by the publication of several influential books (Avise 1994, Avise & Hamrick 1996) and review chapters (Hedrick & Miller 1992, Frankham 1995), and the appearance of influential journals such as, Molecular Ecology in 1992 and Conservation Genetics, in 2000. However, one can argue that the application of genetics to the management of threatened species in the wild is still in its infancy (Frankham 2010). As we move into an era of genomics, the role of genetics in conservation will only grow in importance (Ouborg et al. 2010)

In the past two decades, the use of increasingly powerful genetic tools to inform endangered species management has revolutionized the study of organisms in nature (DeSalle & Amato 2004). We now know that molecular approaches enable biologists to ask fundamental questions about important aspects of a species’ biology that would be difficult, if not impossible, to answer using traditional demographic approaches alone. In fact, attempting to answer questions in the absence of genetic information can lead to erroneous conclusions regarding prioritization of species needing attention (Allendorf & Luikart 2009). Today, many species of conservation concern have been the focus of some genetic analyses, ranging from effects of inbreeding on small populations, management of captive populations and assessing the suitability of species for reintroduction, phylogenetic analyses directed at elucidating taxa in need of conservation attention, and determining population genetic structure at varying geographic scales (Frankham et al. 2009). The role of genetics in conservation is now considered to be of critical importance, because it is entirely feasible to “recover” a species demographically without any understanding of whether or not the species possesses the genetic variability needed to respond to changes in the environment over evolutionary time (Mills 2013).

Clearly, genetic approaches to conservation are needed, but just as demographic approaches are likely to be less effective in the absence of genetic information, the same can be said for genetic approaches in the absence of ecological data. For this reason, and in response to the obvious increase in conservation genetics research, I decided to obtain tissue samples (for genetics study) from individuals as I embarked on an ecological study of Tiger Rattlesnake (Crotalus tigris) populations in the Sonoran Desert of Arizona in 1997. At that time, very little was know about the species, so my goal was to learn as much as possible about the natural history of C. tigris

through intensive fieldwork. My experience working with other rattlesnake species provided me with the tools I needed to conduct the research, including radiotelemetry, mark-recapture, habitat analyses, reproductive behavior, and feeding ecology. However, I knew from the beginning that some of the questions that had eluded snake ecologists could only be addressed through genetic analyses. These questions included fundamental aspects of any species’ biology, such as within and among population relatedness, dispersal, reproductive success, mating system dynamics, and parentage. Only a few studies had even attempted to answer these kinds of questions in snakes (cf., Prior et al. 1997), but the advent of more rapidly evolving, neutral, microsatellite DNA markers made it clear that we had entered a new era of genetic-based inquiry.

In my early efforts to conduct long-term research, the fact that C. tigris was not considered a species of conservation concern was both a detractor and an advantage. The common status of C. tigris made it more difficult to obtain funding, but the advantage was studying a species that was still common and apparently doing well in the wild. Indeed, many endangered species that have been studied have already been subjected to the potentially negative genetic consequences of small population size due to anthropogenic change (Spielman et al. 2004). Furthermore, studying a species without special status would allow me to obtain baseline data that could be used to compare to subsequent data if the species did become threatened. And finally, results from my research could be applied to at-risk species with similar habitat requirements and life histories.

By the time I completed this dissertation, I had amassed a large data set on C. tigris ecology based on long-term (19 years) research involving multiple populations. Although I learned a great deal about the species during this time, many questions remained. For this reason, I turned to genetic analyses to address these questions, with the critical advantage of being able to interpret genetic results within the context of long-term ecological data. The questions I attempted to address in my dissertation were at both among- and within-population scales. I was fortunate enough to work with colleagues who had expertise in genetic analyses that I lacked, and it was their involvement that allowed me to more effectively address these important questions.

In the first paper (Appendix A), we examined among-population genetic structure in C. tigris populations from mountain ranges surrounding the Tucson Basin. These mountain ranges are relatively isolated from each other by intervening lower elevation desert environments with which C. tigris is not typically associated. Furthermore, C. tigris is a habitat specialist, rarely found far from rocky foothill environments that are isolated below by hotter and drier desert scrub and above by colder and wetter semi-desert grassland and oak-dominated interior chaparral. Given the natural isolation of C. tigris populations, we predicted that genetic structuring should occur among mountain ranges in the absence of substantial gene flow among populations. Although C. tigris is associated with rocky habitats, data from radio telemetry revealed extensive use of relatively mesic xeroriparian wash habitats during part of the active season. These critical habitats appear to provide not only a more mesic climate, which likely minimizes water loss, they are also richer in prey resources and vegetative cover. Importantly, washes are also critical for reproduction, because both males and females use them during the mating season. The primary question stemming from detailed information on spatial ecology and

habitat use was the degree to which use of wash environments might facilitate gene flow among populations. Although movement data indicated that individuals, especially males, occasionally move long distances, these distances were less than the distances among populations.

Using the Bayesian clustering software packages, we determined that C. tigris populations showed considerable genetic admixture among mountain ranges. This result was somewhat surprising given the apparent isolation of populations within mountain ranges, and the overall low vagility of the species. One explanation for observed genetic admixture among populations is related to changes in climate and physiognomy of the Sonoran Desert region during the mid- Holocene. Based on evidence from woodrat midden deposits, we know that the present day environment was very different as recent as 5,000-8,000 years ago. The valleys separating mountain ranges consisted of pine-oak woodland environments, which likely facilitated movement among mountain ranges. As the climate warmed and became drier, these woodland elements receded to higher elevation, and desert scrub communities took their place. It is likely that C. tigris populations were interconnected during this time, and that the genetic signature we see today is, at least in part, the result of greater connectivity resulting in regional populations. However, at present, it seems likely that extensive use of washes, especially during the mating season, apparently results in gene flow among mountain ranges. The fact that we occasionally observed individuals equipped with radio transmitters moving relatively long distances (up to 2.7 km), followed by establishment of a new winter shelter site, indicates that contemporary gene flow may take place in a stepping-stone fashion over the course of multiple active seasons. To tease apart the relative influence of contemporary and historical gene flow on standing genetic structure, future work should focus on the use of mitochondrial DNA markers, which evolve over longer time scales, providing a signal of among-population relatedness at deeper temporal scales. Clearly, in the absence of genetic data, we would have been unable to adequately address these questions.

The implications of this study are apparent. From a biogeographic perspective, our results add to an understanding of the effects of historic climates on contemporary patterns of the distribution of species, especially less vagile groups, such as snakes. From a conservation standpoint, our results emphasize the importance of maintaining contemporary connectivity among populations to minimize the potential for decreased gene flow due to anthropogenic alterations to wash environments. This is especially critical given the recent explosive development of the Tucson Basin. A clear management recommendation coming out of this research is effective management, protection, and restoration (if needed) of higher elevation xeroriparian washes, and lower elevation and valley bottom riparian areas, both of which likely provide population connectivity for a suite of species.

In the second paper (Appendix B), we turn our focus to within-population genetic structure and the role played by spatial ecology, habitat use, and mating system dynamics on C. tigris fine- scale genetic structure. The impetus for this paper came out of results from Bayesian clustering analyses performed on the entire data set, which revealed three distinct subpopulations, separated by relatively short distances (4-8 km), within the Rincon Mountains (i.e., Loop Road, Tanque Verde Ridge, Rocking K). In addition, GENELAND identified a potential barrier to gene flow

between C. tigris subpopulations corresponding to a watershed divide separating the Loop Road subpopulation from both the Tanque Verde Ridge and Rocking K subpopulations. Although theTanque Verde Ridge and Rocking K subpopulations showed genetic subdivision, an obvious landscape feature that could be expected to act as a barrier to gene flow was not evident.

To interpret these results, we turned to our intensive radio telemetry data set, which indicated that C. tigris maintain relatively small home ranges, to which they show strong fidelity. Furthermore, individuals use the same winter shelter sites year after year, often returning to the exact same locations during the active season, which is presumably mediated through chemoreception. We also determined that movement patterns and space use are decidedly sex- biased, characterized by males that maintain home ranges 3-4 times larger than females, moving far greater distances than females during the mating season.

Our data on C. tigris mating systems and reproduction also proved to be informative in explaining sex differences in spatial ecology and ultimately, in interpreting the observed pattern of fine-scale genetic subdivision. First, it is important to recognize that female C. tigris typically do not reproduce annually, presumably due to an inability to obtain sufficient prey resources to develop a body condition that is conducive to giving birth. Although the resulting mating system may best be described as polygynandrous, because both males and females mate with multiple partners, the spatial distribution of females, and the fact that they typically do not reproduce annually, indicates that the mating system also contains a component of scramble competition polygyny, where selection favors males that can effectively locate females. In addition, the fact that males are known to engage in combat (ritualized fighting) over access to females demonstrates that the mating system also has characteristics of female defense polygyny.

By combining information on spatial and reproductive ecology, we can arrive at a plausible explanation for the existence of fine-scale genetic structure in a landscape that might be expected to support a continuous population. In the case of the Loop Road and Rocking K subpopulations, the physical presence of a steep and massive ridge would be expected to limit gene flow between these two genetic clusters. The maximum elevation at which C. tigris has been found is 1,555 m (Goode et al. in press), and the maximum elevation at which the ridge separates the Loop Road and Rocking K subpopulations is 1,600 m. Our data on winter shelter sites indicates that individuals typically overwinter in large outcrops situated on mid- to lower-elevation slopes. When temperatures become sufficiently warm, C. tigris move downslope into washes (as revealed by habitat analyses), as opposed to upslope, which only serves to increase the effectiveness of the ridge as a barrier to gene flow.

The genetic subdivision between the Loop Road and Tanque Verde Ridge subpopulations is more surprising, because the watershed divide created by the ridge at lower elevations does not present a physical barrier in the way that it does at higher elevations. Therefore, genetic clustering is likely related to the manner in which C. tigris are distributed across the landscape, both spatially and temporally. Given the species’ preference for washes, especially during the mating season, combined with strong home range and winter shelter site fidelity, we hypothesized that proximity of individuals based on wash distance may be a better predictor of

genetic structure than Euclidian distance. To test this hypothesis, we compared genetic distance between with both wash and Euclidian distance using a Mantel test of matrix correspondence. Interestingly, our results failed to support the prediction that wash distance was more highly correlated with genetic distance, but they also failed to support isolation-by-distance, indicating that something other than geographic distance alone was needed to explain genetic clustering among subpopulations. It is important to note that we measured wash distances that included lower elevation segments running through valley bottom habitats; however, these lower elevation habitats are less likely to be used by C. tigris due to their distance from favored rocky foothill habitats. Therefore, it seems more likely that the complex network of washes associated with higher elevation slopes where C. tigris are more likely to occur may provide stepping-stone connections among subpopulations, with individuals moving short distances over upland habitat to reach adjacent washes. Future analyses should include a more sophisticated approach that incorporates movement resistance values of various landscape features, which should lead to better discrimination of genetic structure due to habitat use.

The role of mating systems dynamics and its effects on movement patterns appears to reinforce the existence of fine-scale genetic structure among subpopulations of C. tigris in the Rincon Mountains. Clearly, females show a greater degree of genetic differentiation among subpopulations than males based on results from spatial autocorrelation analyses, and the fact that pairwise FST values were greater for males than females. This leads us suggest that male- biased dispersal in C. tigris is likely more a function of gamete dispersal than dispersal of actual individuals. Supporting this idea are data on home range size during the mating season, when males dramatically increase their home range size as a function of mate searching. However, because males typically return to the same winter shelter sites year after year, there is very little actual displacement of individuals among subpopulations.

In the third paper (Appendix C), we examined parentage in a single population of C. tigris in the Tortolita Mountains, where we conducted intensive ecological research from 2002-2008, providing us the opportunity to interpret genetic findings in the context of a rich ecological data set. Although we radio tracked a large number of gravid females over the course of the study, we were unable to conduct an extensive analysis of parentage, because females gave birth deep inside large rock outcrops, making it difficult to obtain offspring. In spite of this difficulty, we were able to determine paternity in a subset of offspring for which the mother was known, and we were able to assign parentage of both mothers and fathers to a subset of free-ranging neonates that were found after dispersing from their natal sites.

The most striking result of this study was the fact that not a single male equipped with a radio transmitter was identified as a father, in spite of the fact that we observed many of these males engaged in reproductive behavior with females whose offspring were included in our analyses. This is an important finding, because it clearly shows that mating behavior is a poor predictor of parentage in snakes. Because our marker set had an exceedingly high probability of paternal exclusion, we speculated that the true fathers were simply not included in our sample of genotyped individuals. Supporting this conjecture was the finding that assigned males were found relatively long distances (>2 km) the mothers of their offspring, or the offspring

themselves. Indeed, our study area was characterized by high densities of C. tigris, and was surrounded by suitable habitat for the species. Furthermore, we continued to capture unmarked males throughout the study.

In conclusion, the research that comprises my dissertation has made several significant contributions to our understanding of population genetics in a critically understudied group of organisms. Our findings related to both among- and within-population genetic structure across geographic scales add to a body of literature that is still in its infancy compared to other vertebrates, such as mammals and birds. The discovery of fine-scale genetic structure and male- biased dispersal in C. tigris is consistent with a handful of similar studies, but like those studies, indicates that these may be common attributes of snake populations worldwide. Our research stands out in that results from genetic analyses were interpreted within a rich context of long- term ecological data, providing more robust conclusions about the interaction of genetics and demography in snake populations. And finally, the importance of these findings for management, and ultimately, conservation of this poorly understood group make it clear that combining genetic and demographic research can lead to more effective delineation of threats and their consequences on not only snakes, but on wildlife in general. In the process of completing my dissertation I generated results that are novel and applicable to the conservation and ecology of snakes today.

LITERATURE CITED

Allendorf FW, Luikart G (2009) Conservation and the Genetics of Populations. John Wiley & Sons, Hoboken, NJ.

Avise JC (1994) Molecular Markers, Natural History, and Evolution. Chapman & Hall, New York, NY.

Avise JC, Hamrick JL (1996) Conservation Genetics. Springer Science & Business Media.

Caro TM, Laurenson MK (1994) Ecological and genetic factors in conservation: a cautionary tale. Science-AAAS-Weekly Paper Edition-including Guide to Scientific Information, 263, 485- 486.

DeSalle R., Amato G (2004) The expansion of conservation genetics. Nature Reviews Genetics, 5.9, 702-712.

Frankel OH, Soulè ME (1981) Conservation and Evolution. Cambridge University Press, Cambridge, UK.

Frankham R (1995) Conservation genetics. Annual Review of Genetics, 29, 305-327.

Frankham R, Ballou JD, Briscoe DA (2009) Introduction to Conservation Genetics, 2nd Edition. Cambridge University Press, Cambridge, UK.

Frankham R (2010) Where are we in conservation genetics and where do we need to go? Conservation Genetics, 11, 661-663.

Hedrick PW, Miller PS (1992) Conservation genetics: techniques and fundamentals. Ecological Applications, 1992, 30-46.

Lande R (1988) Genetics and demography in biological conservation. Science, 241, 1455-1460.

Mills LS (2013) Conservation of Wildlife Populations: Demography, Genetics, and Management, 2nd Edition. Wiley-Blackwell, Oxford, UK.

Ouborg NJ, Pertoldi C, Loeschcke V, Bijlsma RK, Hedrick PW (2010) Conservation genetics in transition to conservation genomics. Trends in Genetics, 26, 177-187.

Prior KA, Gibbs HL, Weatherhead PJ (1997) Population genetic structure in the black rat snake: implications for management. Conservation Biology, 11, 1147-1158.

Schonewald-Cox CM, Chambers SM, MacBryde B, Thomas L (1983) Genetics and conservation. Benjamin Cummings, Menlo Park, CA.

Soulé ME (1985) What is conservation biology? A new synthetic discipline addresses the dynamics and problems of perturbed species, communities, and ecosystems. BioScience, 35, 727- 734.

Spielman D, Brook BW, Frankham R (2004) Most species are not driven to extinction before genetic factors impact them. Proceedings of the National Academy of Sciences, 101, 15261- 15264.

APPENDIX A

Manuscript prepared for publication – Conservation Genetics

Population genetic structure in a habitat specialist rattlesnake (Crotalus tigris)

Matt Goode1, Mickey Ray Parker1, Alex Ochoa1 and Melanie Culver1,2

1Wildlife Conservation and Management Program, School of Natural Resources and the Environment, University of Arizona, 325 Biological Sciences East, Tucson, AZ 85721, USA 2US Geological Survey, Arizona Cooperative Fish and Wildlife Research Unit, School of Natural Resources & Environment, University of Arizona, 213 Biological Sciences East, Tucson, AZ 85721, USA

Corresponding author: Matt Goode 325 Biological Sciences East School of Natural Resources and the Environment University of Arizona Tucson, AZ 85721 [email protected]

ABSTRACT

Studies dealing with population genetic structure are decidedly biased towards species of conservation concern that are threatened by anthropogenic habitat alterations (e.g., habitat fragmentation). However, many common species may be affected by human-caused disturbance, especially species occurring in naturally isolated populations that must move through disturbed environments to maintain gene flow among populations. The Tiger Rattlesnake (Crotalus tigris) is a habitat specialist associated with rocky foothills of isolated desert mountain ranges in the Sonoran Desert. These cryptic snakes make use of relatively mesic washes during the active season to avoid temperature and humidity extremes, and to forage and mate. Although C. tigris exhibits strong home range fidelity, typically returning to the exact same winter shelter year after year, males will occasionally move relatively long distances during the summer mating season in search of receptive females, a scarce resource due to sometimes extreme reproductive periodicity. Although genetic analyses using Bayesian clustering techniques support population differentiation, the pattern is not consistent with underlying landscape structure of isolated mountain ranges. Furthermore, considerable genetic admixture is not fully consistent with predictions based on potential habitat connectivity within and among mountain ranges. We hypothesize that C. tigris populations were previously connected when more mesic environments supported woodland vegetation associations at lower elevations (ca. 8,000 years ago). But today, populations must rely on suitable wash habitat to maintain gene flow among mountain ranges. Therefore, we stress the importance of enhancing, or at least maintaining, population connectivity through proactive management of riparian and xeroriparian movement corridors. Importantly, these specialized habitats are critical to a suite of other species.

INTRODUCTION

Microevolutionary processes combine to produce patterns of population genetic structure on both spatial and temporal scales (Templeton 2006). Simply put, the random process of genetic drift and the deterministic process of selection both lead to population differentiation. These processes are countered by gene flow, which is a combination of dispersal and reproduction (Slatkin 1987). Gene flow can be mediated through actual movement of individuals among populations, or dispersal of gametes of species that show strong fidelity to a particular habitat feature (e.g., roosting or denning sites). If dispersal capabilities of the species are limited, inbreeding may occur, especially in isolated populations. Indeed, many vulnerable species show a pattern of isolation and inbreeding that can result in reduced genetic variation that may ultimately lead to a classic extinction vortex (Gilpin & Soulé 1986).

The overwhelming majority of conservation genetic studies across taxa have understandably focused on vulnerable species, many of which previously occurred in continuous populations, or in connected habitat patches that have become increasingly isolated due to recent anthropogenic disturbance (Storfer et al. 2010). The usual approach involves correlating standing genetic structure with changes in landscape structure; however, contemporary patterns of genetic divergence among populations are likely the result of both recent and historic changes in landscape structure (Manel et al. 2003).

Detailed data on the ecology and life history of a species allow for predictions to be made about both mechanisms involved in producing genetic differentiation among populations, and potential response to changes in landscape structure (Holdregger & Wagner 2008). For example, dispersal capabilities of a species are critical for maintaining gene flow among populations. For species that exhibit long-distance dispersal relative to population density, the effects of barriers to gene flow (or removal of barriers in a conservation context) on genetic structure can be detected relatively rapidly (as little as 10-15 generations in annually reproducing species). However, genetic differentiation among populations of species with limited dispersal and greater-than- annual reproductive cycles, can take hundreds, if not thousands of generations to detect (Landguth et al. 2010). Because the primary focus of landscape genetic studies has been on species with relatively large dispersal capabilities, the genetic consequences of anthropogenic habitat alteration have already become manifest.

But what about populations of animals with low dispersal capabilities and slow life histories that live in naturally isolated populations, with seemingly little or no migration among populations? For these species, population differentiation may be more dependent on dispersal ability than on anthropogenic barriers to gene flow. Ideally, examination of genetic structure among populations that are naturally isolated will increase our understanding of anthropogenic influences on genetic structure of populations that have been separated by human activities, while simultaneously providing a baseline for the analyses of future changes in genetic structure due to anthropogenic change, especially for non-threatened taxa that have not already experienced negative genetic consequences due to habitat fragmentation and a subsequent reduction in genetic diversity.

Therefore, we examined genetic structure of Tiger Rattlesnakes (Crotalus tigris), a cryptic, habitat specialist found in close association with rocky habitats within isolated mountain ranges of the Sonoran Desert in southern Arizona (Brennan & Holycross 2009). Locally abundant in rocky foothills, C. tigris occur in a band of elevation separated by hotter and drier desertscrub below and colder and wetter semidesert grassland and oak-dominated interior chaparral above (Goode et al. in press). Although C. tigris are rarely found far from rocks, they do make extensive use of xeroriparian wash habitats during the summer monsoon period, which coincides with the mating season. During this time, individuals may travel relatively long distances along washes, which likely facilitates gene flow among populations (Goode et al. 2008).

Given the narrow habitat requirements of C. tigris, we predicted that populations naturally separated by relatively large tracts of unsuitable habitat would exhibit genetic differentiation. This led to the related prediction that isolated populations would exhibit a decrease in within- population genetic variability. To test these predictions, we examined population genetic data from analyses of polymorphic microsatellite loci (N = 10) within the rich context of hard-won ecological data obtained from radiotelemetry and mark-recapture analyses at two long-term study sites (Goode et al. 2008, Goode et al. 2011).

METHODS

Study System

We studied C. tigris from four mountain ranges surrounding the Tucson Basin in the Sonoran Desert of southern Arizona (Fig. 1) over a 12-year (1997-2008). We conducted long-term, intensive, population-level research in the Rincon Mountains (Rincons) and Tortolita Mountains (Tortolitas), and we collected tissue samples from the Tucson Mountains (Tucsons) and Santa Catalina Mountains (Catalinas). In general, the study sites were characterized by steep upper slopes with extensive rock outcroppings, gently sloping lower alluvial fans containing a network of small, poorly developed washes, and relatively flat lower elevation desert scrub containing larger washes with relatively well developed soil terraces and xeroriparian vegetation. Vegetation was typical Sonoran Desertscrub, Arizona Upland Subdivision (Turner & Brown 1994). Common plants included saguaro (Carnegia gigantea), foothill paloverde (Cercidium microphyllum), brittlebush (Encelia farinosa), prickly pear (Opuntia spp.), cholla (Cylindropuntia spp.), and velvet mesquite (Prosopis velutina).

A relatively small, isolated mountain range, the Tortolitas reach a maximum elevation of 1431 m. Although C. tigris likely occurs throughout the entire range, our long-term research was confined to lower elevation (900-1000 m), rocky foothills, within and adjacent to a high-end, low-density residential development (Stone Canyon) dominated by a world-class golf course.

Separated from the Tortolitas by Cañada del Oro a major watershed that runs through Oro Valley (~5-7 km wide, elevation ~750-850 m), the Catalinas are a large range, rising to an elevation of 2791 m. We obtained samples from C. tigris throughout the moderately urbanized foothills (850-

1000 m), most of which came from local fire departments, which provide a snake removal service to residents.

Another large range, the Rincons reach an elevation of 2641 m. Although isolated from the Catalinas at higher elevations, and divided by Tanque Verde Creek and Reddington Pass, the Rincons and Catalinas are connected at lower elevations (below ~1250 m), providing continuous suitable habitat for C. tigris. We conducted long-term research on C. tigris at two sites within, and adjacent to, the Rincon Mountain District of Saguaro National Park in the foothills of the Rincons, and we obtained tissue samples from individuals found while conducting road cruising surveys on the Loop Road, much of which runs through the lower foothills. We found our highest elevation C. tigris in the Rincons at 1147 m.

The Tucsons consist of a series of peaks, with a maximum elevation of 1429 m. Some peaks are connected at middle to higher elevations, and others are separated by intervening, lower elevation desert scrub. We collected tissue samples primarily from the Tucson Mountain District of Saguaro National Park, which is being increasingly surrounded by mostly low- to medium- density urban development. We found our lowest elevation C. tigris in the Tucsons at 696 m.

Field Data

We obtained C. tigris during foot searches, road cruising surveys, while radio tracking individuals implanted with transmitters, from colleagues conducting research in overlapping study areas, from golf course maintenance personnel, from residents who found rattlesnakes on their property, and from local fire departments that provide rattlesnake removal services to homeowners.

We obtained tissues samples by extracting ~50 µl of blood from the caudal vein. On occasion, especially for smaller individuals, we were unable to obtain blood, so we collected skin tissue from ventral scale clips. We also obtained DNA from shed skins of neonates born in captivity to females collected at our study sites.

To examine spatial ecology of C. tigris, we surgically implanted radio transmitters into a subset of adult males and females at our study sites in the Rincons (Goode et al. 2008) and Tortolitas (Goode & Parker 2011). We used the Animal Movement Analysis extension (USGS Alaska Biological Science Center) in ArcView 3.2 (ESRI, Inc.) and the Geospatial Modeling Environment (Beyer 2012) in ArcMap 10.2 (ESRI, Inc.) to calculate parameters pertaining to rattlesnake seasonal movements, including minimum convex polygon (MCP) home range size, length of major MCP axis, total distance moved, average distance moved/day, and maximum distance moved from winter shelter sites over the course of the active season.

We used mark-recapture data to determine the maximum distance moved by adult male and female rattlesnakes that were not implanted with radio transmitters.

Genetic Analyses

We developed 18 microsatellite loci for C. tigris (Goldberg et al. 2003, Munguia-Vega et al. 2009). We initially confined our analyses to the 12 most informative loci based on their combined ability (PID) to discriminate among individual rattlesnakes. We further reduced the number of loci to 10, because <80% of individuals scored at two loci.

We used GENALEX (Peakall & Smouse 2012) to obtain standard indices of genetic diversity. We used GENODIVE (Miermans & Van Tienderen 2004) to screen loci for deviations from Hardy-Weinberg equilibrium using GENODIVE, which implements exact probability tests (10,000 permutations). We controlled for increased Type I error rates due to multiple testing with sequential Bonferroni correction. We used the program FSTAT (Goudet 1995) to test for evidence of linkage disequilibrium between all pairs of loci in each population, and we assessed significance after adjusting for multiple tests. We examined loci for the presence of null alleles in GENEPOP (Rousset 2008). We used FSTAT to estimate inbreeding coefficients (FIS) and allelic richness (AR). We evaluated the power of the entire microsatellite panel to distinguish among individuals by estimating probability of identity (PID) in GENALEX.

We used STRUCTURE (Pritchard et al. 2000) and GENELAND (Guillot et al. 2005), both of which use Bayesian assignment methods, to infer genetic clusters (K). We ran STRUCTURE without location priors and assumed admixture and correlated allele frequencies among putative genetic clusters. Markov Chain Monte Carlo (MCMC) parameters consisted of 10 independent runs per K with 500,000 generations each. We eliminated the first 100,000 generations of each run as burn-in. We used STRUCTURE HARVESTER (Dent & vonHoldt 2012), which applies the ΔK method (Evanno et al. 2005), to estimate the most likely number of genetic clusters. We used CLUMPP to summarize individual assignment probabilities to genetic clusters across all runs for the most likely K value, and we used DISTRUCT to plot the results (Rosenberg 2004). To assess genetic divergence among populations, we computed pairwise population estimates of FST (Wright 1969) and G’ST (Hedrick 2005) in FSTAT. We used a Mantel test, as implemented in GENALEX to test for the presence of genetic isolation by distance among populations (9,999 permutations), which determines the correlation between pairwise genetic and Euclidean geographic distance. We used geographic coordinates obtained from initial capture locations for individuals with multiple captures.

Unlike STRUCTURE, GENELAND incorporates the geographic locations of samples, but has no information pertaining to underlying landscape structure. GENELAND attempts to detect barriers to gene flow based on genetic discontinuities associated with spatial structure. We performed three independent runs of GENELAND by varying the number of putative populations from 1-9. We ran 1,000,000 MCMC iterations, with thinning set at 1,000 and burn- in set at 500. We used ArcMap 10.2 to overlay the map of population membership produced by GENELAND onto a map of the study system, which allowed for examination of underlying landscape features associated with potential barriers. We also used high-resolution aerial imagery

(©Pima County Association of Governments, available at http://www.pagnet.org) to further examine landscape features that may act as barriers to gene flow.

RESULTS

Field Data

We captured 869 C. tigris at all sites combined (Table 1). Overall, we captured adult males (68%) more than twice as often as adult females (32%), and we captured adults (88%) more than 7 times as often as subadults (12%). We classified individuals as subadults if they were smaller than the minimum size (snout-vent length) for reproduction (i.e., males ≤512 mm, females ≤509 mm), which we based on observations of reproductive behavior (Goode et al. in press), combined with histological findings from preserved specimens (Goldberg 1999). The overwhelming majority of subadults were young-of-the-year (“neonates”); we rarely found second-year juveniles.

We implanted transmitters into 106 adult C. tigris (Tortolitas = 71, Rincons = 35), which we radio tracked 9,151 times over the 12-year period of the study. We excluded 12 individuals from analyses, because they did not meet our minimum sample size requirements of ≥20 fixes during the pre- and post-mating seasons, and ≥30 fixes during the mating season.

Published results from the Rincons (Goode et al. 2008) revealed that male active season home ranges (µ = 13.1 ± 2.0 ha) were more than four times larger than female home ranges (µ = 3.9 ± 1.2 ha). In addition, males (µ = 27.2 m) moved significantly greater distances per day than females (µ = 11.3 m) during the mating season. The longest distance moved by a male equipped with a radiotransmitter from its previous winter shelter site was 2,709 m, which took place over a six-week period during the post-mating season. This particular individual overwintered in a new location and, to our knowledge, did not return to its previous winter shelter site. After removing this outlier male from subsequent analyses, the maximum distance (µ = 490 m) that the remaining males moved from their winter shelter sites was similar to females (µ = 447 m).

In the Tortolitas, C. tigris home ranges were smaller than in the Rincons, but the pattern of differences in home range size between males and females was similar. Male (µ = 11.6 ha) annual home range size was more than three times greater than females (µ = 3.3 ha). In addition, annual MCP major axes for males (µ = 1,492 m) approached twice that for females (µ = 860 m). Over the course of the active season, males (µ = 1,186 m) moved more than twice the total distance moved by females (µ = 564 m), and more than twice the distance per day (µmales = 19.2 m/day, µfemales = 9.1 m/day). Discrepancy in home range size and movements between males and females was even greater during the mating season. Male (µ = 4.39 ha) home ranges were more than seven times greater than females (µ = 0.6 ha), and MCP major axes were nearly three times longer (µmales = 456 m, µfemales = 173 m). In addition, males (µ = 34.6 m/day) moved 2.5 times farther per day than females (µ = 13.7 m/day) during the mating season. Unlike the Rincons, where males and females moved similar maximum distances from winter shelter sites,

the maximum distance moved from winter shelter sites for males (µ = 713 m) approached twice that females (µ = 380 m) in the Tortolitas.

We obtained 177 recaptures of the 571 individuals from the Rincons and Tortolitas that were not included in the radiotelemetry data set. The large majority of recaptures were from the Stone Canyon study site in the Tortolitas, where we obtained 154 recaptures of 389 individuals. We obtained 23 recaptures of 182 individuals at the Rincon study site.

Recapture data revealed that the maximum distance moved by a recaptured males was 1,881 m. Furthermore, 13 males moved > 1,000 m maximum distance between recaptures. In contrast, the maximum distance moved by a female was 1,032 m, but the second farthest distance was 436 m. Combined results from radiotelemetry and mark-recapture clearly show that males are by far the more vagile sex, moving significantly farther than females regardless of site, season, or year.

Genetic Analyses

Of the 869 rattlesnakes captured, we included 586 in our genetic analyses. The majority of individuals omitted were captured before we began collecting tissue samples. We removed additional individuals captured by the fire department, for which we were unable to obtain reliable capture locations. And finally, we removed additional individuals from analyses, because we either failed to obtain sufficient quantities of DNA or they failed to score at ≥80% of loci.

Genetic diversity of C. tigris was similar among mountain ranges (Table 2), with among- population heterozygosity varying by only 2% (µ = 0.578). After corrections for multiple tests (Bonferroni adjusted significance criterion = 0.00125), microsatellite screening revealed deviations from Hardy-Weinberg equilibrium at one locus in the Catalinas (CR47), three loci in the Rincons (CR06, CR10, CR47), four loci in the Tortolitas (CR06, CR09, CR47, CR95), and three loci in the Tucsons (CR06, CR09, CR47). Total number of migrants among all three subpopulations was estimated at 9.3 over the period of time encompassing genetic sampling.

We found evidence of linkage disequilibrium (LD) at four of a possible 210 loci combinations, but there was no consistent pattern among alleles or populations. We found no evidence of null alleles at any locus from any population.

Combined results from STRUCTURE (Fig. 2) and STRUCTURE HARVESTER (Fig. 3) revealed the strongest support for K = 2 subpopulations (Rincons and Tortolitas; ΔK = 65.5), followed closely by K = 3 subpopulations (addition of Catalinas, ΔK = 60.6). At K = 4, there is no discernable pattern of population differentiation (ΔK = 0.2).

Pairwise population relatedness based on both FST and G’ST support STRUCTURE results in that the Tortolitas and Rincons are the most divergent; however, all pairwise combinations show significant differences in relatedness as expressed by both estimators (Table 3). Analyses revealed a positive correlation between genetic and geographic distance for all four populations combined (Fig. 4). A Mantel matrix correspondence test revealed a significant

[P(rxy-random ≥ rxy-observed) = 0.01] correlation between genetic and geographic distance, suggesting that genetic differentiation among populations follows a pattern of isolation-by- distance (IBD).

In contrast to STRUCTURE, results from GENELAND (Fig. 5) provided support for seven genetic demes. However, closer examination of the population membership map reveals the presence of three so-called “ghost populations,” to which no individuals were assigned. The presence of “ghost populations” is due to idiosyncrasies of specific Bayesian clustering algorithms (for technical discussion, see Guillot 2008). Therefore, the most likely number of genetic clusters based on GENELAND results is K = 5. These clusters do not necessarily correspond to the geographic locations of sampled subpopulations, indicating that genetic substructure within mountain ranges is likely, which is the subject of current research in the Rincon Mountains (Goode et al., in progress).

DISCUSSION

Long-term field data clearly show that C. tigris are habitat specialists, strongly associated with rocky foothills of mountain ranges that are isolated by inhospitable lowland desert scrub. Furthermore, mark-recapture and radiotelemetry data demonstrate that C. tigris maintain relatively small home ranges, and show strong fidelity to winter shelter sites, to which they return year after year. The combination of these ecological attributes leads to the two related predictions that regional populations will exhibit genetic divergence, and local populations will show reduced genetic variability. Although genetic analyses support population differentiation, the pattern is not consistent with underlying landscape structure of isolated mountain ranges. Furthermore, considerable genetic admixture is not fully consistent with predictions based on potential habitat connectivity within and among mountain ranges.

Bayesian clustering analysis in STRUCTURE, with no prior information on the geographic locations of individuals, supported two distinct populations in the Tortolitas and Rincons. This result was not surprising, because these populations are farthest apart (ca. 36 km). Both populations show genetic admixture from the Catalinas, which makes sense from a geographic perspective, because the Catalinas lie between the Tortolitas and Rincons. Although the Tucsons are the most geographically isolated mountain range, results indicated that C. tigris found there are the least genetically differentiated, with extensive admixture among the other three populations. Clearly, the pattern of genetic divergence among populations does not correspond well to the degree of isolation among mountain ranges, but does show evidence of isolation by distance (see below).

Further examination of among-population relatedness, based on pairwise estimates of population differentiation (FST and G’ST), indicated that C. tigris from the Catalinas appear to be less divergent from conspecifics from the Tortolitas than those from the Rincons. However, this pattern does not make ecological sense, because the Catalinas and Rincons provide essentially continuous suitable habitat for the species, as opposed to the Catalinas and Tortolitas, which are separated by low-elevation desertscrub with which the species is not normally associated. With

regard to the Tucsons, both FST and G’ST values indicate a greater degree of allelic sharing with the Rincons than either the Catalinas or Tortolitas, in spite of the greater geographic distance separating the two least genetically divergent populations.

Based on the absence of clear and consistent patterns of population genetic differentiation related to landscape structure between population pairs, we tested for the presence of isolation by distance (IBD) using a Mantel test, which compares matrices of genetic and geographic distance. Although the test was statistically significant, the strength of the relationship between genetic and geographic distance was weak, suggesting that a combination of interacting factors (e.g., distance, degree of isolation, sampling design) are likely responsible for the lack of a consistent pattern of genetic divergence among populations.

Results from GENELAND provide additional information regarding genetic differentiation among populations. Presumably, the use of location priors should result in more power to assign individuals to genetic clusters. If this is the case, then genetic differentiation among C. tigris populations is more subdivided than results from STRUCTURE would indicate. Consistent with results from STRUCTURE, GENELAND also indicated that the geographically isolated population in the Tortolitas is part of a genetic cluster that contains individuals from the Catalinas. In addition, the program identified another genetic cluster that includes individuals from the Catalinas and Rincons. In sharp contrast to STRUCTURE, GENELAND results provided support for a distinct genetic cluster in the Tucsons.

Although analyses of genetic differentiation among populations do not provide a clear, easily explainable pattern, examination of landscape structure and apparent genetic discontinuities, may provide some possible explanations. Results from GENELAND indicate the presence of a putative barrier between the Tortolitas and Tucsons that corresponds reasonably well with the location of the Santa Cruz River and Interstate Highway 10. While it is certainly possible that Interstate Highway 10 could present a formidable barrier to gene flow, it seems more likely that the physical expanse of unsuitable habitat between the two mountain ranges would maintain separation of individuals long before encountering the highway. This is especially true between the Tortolitas and Tucsons, where alluvial debris from erosion of the two mountain ranges has accumulated more than anywhere else in the Tucson Basin (McAuliffe 1994). Drainage of this large alluvial fan is primarily contained on the surface of the alluvium, which would further preclude movements by C. tigris using wash corridors.

The putative barrier separating genetic clusters in the Catalinas is more difficult to explain. We speculate that C. tigris from the south side of the Catalinas are separated from their counterparts on the north side of the range simply be elevation. The fact that C. tigris are rarely found above 1,500 m (Goode et al. in press) prevents population connectivity at higher elevations. It seems plausible that the putative barrier in the Catalinas reflects habitat specialization by C. tigris, which are largely confined to a band of suitable habitat that eliminates gene flow at higher elevations and limits gene flow at lower elevations.

The most surprising result from GENELAND was the identification of three subpopulations, separated by distances of only 2-4 km, in the Rincons. This unexpected fine-scale genetic differentiation is the topic of current research (Goode et al. in progress), but preliminary results indicate that these populations may be separated by local drainage patterns, with washes acting as dispersal corridors within, but not among watersheds. By extension, it is possible that connections among populations at the scale of the entire Tucson Basin are also based on watersheds, especially given the preference of C. tigris for riparian habitats during the summer mating period.

Data from radiotelemetry and mark-recapture indicate that individuals do not move great enough distances to travel between mountain ranges. However, males will occasionally move 2-3 km, and have been known to establish new winter shelter sites relatively far removed from previous overwintering sites. Therefore, it is conceivable that an individual could continue to move increasingly greater distances from its natal population over the course of a few to several years, eventually ending up in a neighboring mountain range. Even if a small number of individuals were able to disperse long distances, and breed successfully, it would likely be enough to leave a genetic signature (Wright 1969, Mills & Allendorf 1996). Alternatively, an individual could come into contact with an opposite-sex conspecific that also moved an unusually long distance from its natal range; however, this seems less likely given the fact that females move considerably less than males.

In any case, documented maximum movement distances are still less than the distances between mountain ranges, which leads us to speculate that observed genetic differentiation among mountain ranges may be partly explained by historic gene flow. Extensive research, based on examination of remains found in woodrat (Neotoma spp.) middens, clearly demonstrates that Sonoran Desert environments were characterized by more mesic, woodland vegetation associations (as low as 600 m) as recently as ca. 8,000 years ago (Van Devender 1977). Furthermore, the herpetofauna in this region has been stable for at least the last 15,000 years (Van Devender 1978). Therefore, it seems likely that environmental conditions in the recent past provided connections among mountain ranges that no longer exist today. Indeed, it is entirely possible that C. tigris were panmictic in the not-so-distant past and that contemporary genetic structure can be partially explained by these previous conditions.

Based on extensive field data, we estimate that C. tigris reach sexual maturity at three years, leading to an estimated 2,666 generations since mountain ranges were likely connected in the mid-Holocene. However, given the expected mutation rate of microsatellite DNA (10-4 per locus per generation; Chakraborty et al. 1997), contemporary exchange of genetic material among mountain ranges must have occurred, and is likely still occurring, along xeroriparian washes in a stepping-stone fashion. This is supported by within-population patterns of genetic diversity, which indicate that isolated populations are apparently exhibiting migration-drift equilibrium, with no evidence of inbreeding.

Because the foothills and valleys separating mountain ranges have been developed at an alarming rate over the past several decades (Huckleberry 2002, US Census Bureau 2013),

movement among mountain ranges has likely been impeded to some extent. Therefore, to prevent future isolation of C. tigris populations, emphasis should be placed on maintaining wash corridors between mountain ranges. This includes small xeroriparian washes at higher elevations where C. tigris are abundant, and not just larger, lower elevation washes that typically receive the most management attention. Maintenance of xeroriparian and riparian washes, will not only facilitate gene flow among otherwise isolated populations, but will also benefit a plethora of species (some of which are threatened and endangered) that depend on these habitats for various aspects of their life history (Naiman et al. 1993, Webb et al. 2007)

ACKNOWLEDGEMENTS

This study would not have been possible without the hard work and dedication of numerous field assistants, lab technicians, and volunteers. In particular, Scott Breeden, Ed Kuklinski, Ryan Gann†, James Borgmeyer, Chris Davis, Jeff Smith, Melissa Amarello, Josh Capps, Danny Brower†, Larry Jones, and Phil Healey went above and beyond the call of duty. We thank Taylor Edwards, Caren Goldberg, Matt Kaplan, Adrian Munguia-Vega, and Karla Pelz-Serrano for advice on genetic techniques and analyses. We are grateful to the staff at Saguaro National Park, especially Pam Swantek, Kristin Beaupre, Natasha Kline, Don Swann, Rich Hayes and Meg Weesner, for permitting us to conduct research in the park, and for providing equipment and technical support. We also thank Dick Maes, Vistoso Partners LLC, Stone Canyon Golf Club staff, Stone Canyon residents for allowing us to conduct fieldwork, literally, in their backyards. The senior author extends his gratitude to Larry Norris, National Park Service (retired) for his generous support of his research program. We greatly appreciate the generosity of local residents, especially Joel Peterson and Shawnee Riplog-Peterson, who provided access to their private land. We thank the Rocking K Ranch for allowing us to work on their land. Mike Rosenzweig, John Koprowski and Cecil Schwalbe provided advice, support, and critical reviews of the manuscript. Our research was supported by grants from Arizona Game and Fish Department Heritage Fund, National Park Service, Western National Parks Association, Reid Park Zoological Society, and Denver Zoological Society.

LITERATURE CITED

Beyer, HL (2012) Geospatial Modeling Environment (Version 0.6.0.0). http://www.spatialecology.com/gme.

Brennan TC, Holycross AT (2009) Amphibians and Reptiles in Arizona. Arizona Game and Fish Department, Phoenix, AZ.

Chakraborty R, Kimmel M, Stivers DN, Davison LJ, Deka R (1997) Relative mutation rates at di-, tri- and tetranucleotide microsatellite loci. Proceedings of the National Academy of Sciences, USA, 94, 1041-1046.

Dent EA, vonHoldt, BM (2012) STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genetics Resources, 4, 359-361.

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

Gilpin ME, Soulé ME (1986) Minimum viable populations: processes of species extinction. In: M. E.. Conservation Biology: The Science of Scarcity and Diversity (ed. Soulé ME), pp. 19-34. Sinauer, Sunderland, MA.

Goldberg CS, Edwards T, Kaplan ME, Goode M (2003) PCR primers for microsatellite loci in the tiger rattlesnake (Crotalus tigris, Viperidae). Molecular Ecology Notes, 3, 539-541.

Goldberg SR (1999) Reproduction in the tiger rattlesnake, Crotalus tigris (Serpentes:Viperidae). The Texas Journal of Science, 51, 31-36.

Goode M, Smith JJ, Amarello, M (2008) Seasonal and annual variation in home range and movements of Tiger Rattlesnakes (Crotalus tigris) in the Sonoran Desert of Arizona. In: The Biology of Rattlesnakes (eds. Hayes WM, Beaman KR, Cardwell MD, Bush SP), pp. 327-334. Loma Linda University Press, Loma Linda, CA.

Goode M, Parker MR (2011) Microgeographic variation in Tiger Rattlesnake ecology and life history: the importance of long-term, natural history-based, multiple-population research. Reptile & Amphibian Conservation and Natural History, 18, 84-91.

Goode M., Parker MR, Feldner MJ., Repp RA. (In Press) Tiger Rattlesnake (Crotalus tigris) species account. In: The Rattlesnakes of Arizona (eds. Schuett GW, Reiserer RS, Smith CF, Feldner MJ). Eco Press, Rodeo, NM.

Goudet J. (1995) FSTAT : a computer program to calculate F-statistics. Journal of Heredity, 86, 485-486

Guillot G, Mortier F, Estoup A (2005) GENELAND: a computer package for landscape genetics. Molecular Ecology Resources, 5, 712-715.

Guillot G, Santos F, Estoup A (2008). Analysing georeferenced population genetics data with GENELAND: a new algorithm to deal with null alleles and a friendly graphical user interface. Bioinformatics, 24, 1406–1407.

Hedrick PW (2005) A standardized genetic differentiation measure. Evolution, 59, 1633–1638.

Holderegger R, Wagner HH (2008) Landscape genetics. BioScience, 58, 199–207.

Huckleberry C (2002) The Sonoran Desert Conservation Plan. Endangered Species Bulletin, 2, 12-15.

Jakobsson M, Rosenberg NA (2007) CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics, 23, 1801-1806.

Landguth EL, Cushman SA, Schwartz MK, McKelvey KS, Murphy M, Luikart G (2010) Quantifying the lag time to detect barriers in landscape genetics. Molecular Ecology, 19, 4179- 4191.

Manel S, Schwartz MK, Luikart G, Taberlet P (2003) Landscape genetics: combining landscape ecology and population genetics. Trends in Ecology and Evolution, 18, 189–197.

McAuliffe JR (1994) Landscape evolution, soil formation, and ecological patterns and processes in Sonoran Desert bajadas. Ecological Monographs, 1994, 112-148.

Mills LS, Allendorf FW (1996) The one-migrant-per-generation rule in conservation and management. Conservation Biology, 10, 1509-1518.

MunguiaVega A, PelzSerrano K, Goode M, Culver M (2009). Eleven new microsatellite loci for the tiger rattlesnake (Crotalus tigris). Molecular Ecology Resources, 9, 1267-1270.

Meirmans PG, Van Tienderen PH (2004) GENOTYPE and GENODIVE: two programs for the analysis of genetic diversity of asexual organisms, Molecular Ecology Notes, 4, 792-794.

Naiman RJ, Decamps H, Pollock M (1993) The role of riparian corridors in maintaining regional biodiversity. Ecological Applications, 3, 209–212.

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.

Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics, 155, 945–959.

Rosenberg NA (2004) DISTRUCT: a program for the graphical display of population structure. Molecular Ecology Notes, 4, 137-138.

Rousset F (2008) GENEPOP 2007: a complete reimplementation of the GENEPOP software for Windows and Linux. Molecular Ecology Resources, 8, 103-106.

Slatkin M (1987) Gene flow and the geographic structure of natural populations. Science, 236, 787-792.

Storfer A, Murphy MA, Spear SF, Holdregger R, Waits LP (2010) Landscape genetics: where are we now? Molecular Ecology, 19, 3496-3514.

Templeton AR (2006) Population Genetics and Microevolutionary Theory. John Wiley & Sons, Hoboken, NJ.

Turner RM, Brown DE (1994) Tropical-subtropical desertlands. In: Biotic Communities of the American Southwest – United States and Mexico. University of Utah Press, Salt Lake City, UT.

U.S. Census Bureau (2013) Quick facts. Retrieved from http://quickfacts.census.gov/qfd/states/04000.html.

Van Devender TR (1977) Holocene woodlands in the southwestern deserts. Science, 198, 189- 192.

Webb RH, Leake SA, Turner RM (2007) The Ribbon of Green: Change in Riparian Vegetation in the Southwestern United States. University of Arizona Press, Tucson, AZ.

Wilson GA, Rannala B (2003) Bayesian inference of recent migration rates using multilocus genotypes. Genetics, 163, 1177-1191.

Wright S (1969) Evolution and the Genetics of Populations, Vol. 2. University of Chicago Press, Chicago.

FIGURES

Tortolitas

Santa Catalinas

Tucsons

Rincons

Fig. 1. Map showing study system and capture locations of 586 C. tigris from four mountain ranges surrounding the Tucson Basin in southern Arizona, USA from 1997-2008.

A C RI TO TU K=2

K=3

K=4

Figure 2. Results from STRUCTURE showing barplots for K=2, K=3, and K=4 C. tigris “populations” from four mountain ranges surrounding the Tucson Basin.

Figure 3. Results from STRUCTURE HARVESTER, indicating strongest support for two populations, followed closely by three populations. K = population number, ΔK = mean(|L’’(K)|)/SD(L(K))

Fig 4. Correlation of genetic and geographic distance for C. tigris among all four mountain ranges. A Mantel test of matrix correspondence was significant [P(rxy-random ≥ rxy-observed) = 0.01]

Figure 5. Output from GENELAND showing population membership map for C. tigris from four mountain ranges surrounding the Tucson Basin (upper right) superimposed on a map of the study system to examine putative barriers associated with genetic discontinuities found by the program. GENELAND was provided with the geographic coordinates of individuals, but has no information pertaining to underlying landscape structure.

TABLES

Table 1. Total number of C. tigris captured by site, sex and age class. Parenthetical values refer to the total number of individuals captured from each mountain range.

Category Rincons (217) Catalinas (129) Tortolitas (460) Tucsons (63) Total (869) Sex Males 161 101 291 40 593 Females 56 28 169 23 276 Age Class Adults 206 117 372 51 746 Subadults 11 12 88 12 123

Table 2. Measures of genetic diversity by population. Total mean and SE are calculated across all four populations. N = sample size, Na = number of alleles, Ne = number of effective alleles, AR = allelic richness, Ho = observed heterozygosity, He = expected heterozygosity, FIS = inbreeding coefficient. Mountain Range N Na Ne AR Ho He F IS Catalinas Mean 37.2 12.2 7.2 9.5 0.57 0.64 0.12 SE 0.9 3.4 2.4 2.4 0.11 0.10 0.08

Rincons Mean 143.8 17.1 7.9 9.3 0.57 0.64 0.16 SE 2.4 5.5 3.2 2.4 0.11 0.09 0.09

Tortolitas Mean 336.1 20.8 8.0 9.5 0.58 0.66 0.17 SE 8.2 6.5 2.8 2.4 0.12 0.10 0.08

Tucsons Mean 31.2 12.5 7.3 10.2 0.59 0.62 0.09 SE 1.6 3.6 2.4 2.6 0.11 0.11 0.09

Table 3. Population differentiation among C. tigris sampled from four mountain ranges surrounding the Tucson Basin based on FST and G’ST (bold values). Mountain Range Catalinas Rincons Tortolitas Tucsons Catalinas 0.001/0.002 0.001/0.002 0.001/0.002

Rincons 0.026/0.061 0.001/0.002 0.001/0.002

Tortolitas 0.017/0.041 0.040/0.096 0.001/0.002

Tucsons 0.039/0.094 0.030/0.069 0.035/0.085

APPENDIX B

Manuscript prepared for publication – Molecular Ecology

Sex-biased dispersal, seasonal habitat use and mating system dynamics contribute to fine- scale genetic structure in the Tiger Rattlesnake (Crotalus tigris)

Matt Goode1, Ashwin Naidu1, Alex Ochoa1, Mickey Ray Parker1 and Melanie Culver1,2

Corresponding author: Matt Goode 325 Biological Sciences East School of Natural Resources and the Environment University of Arizona Tucson, AZ 85721 [email protected]

ABSTRACT

We examined population genetic differentiation in Tiger Rattlesnakes (Crotalus tigris) from the Rincon Mountains in the Sonoran Desert of Arizona. Results from traditional F-statistics and Bayesian clustering analyses supported fine-scale population substructure, presumably caused by an apparent barrier to gene flow that corresponds to a watershed divide. This finding, combined with a documented preference for drainage systems (washes) during the mating season, emphasizes the importance of xeroriparian habitat to C. tigris population ecology. Further analyses, using spatial autocorrelation, indicated that females exhibited positive spatial structure up to ~200 m, but males did not. This pattern is consistent with long-term data from radiotelemetry, which clearly demonstrated that males moved significantly farther during the mating season, presumably in search of more philopatric females. The combination of low vagility, sex-biased dispersal, and seasonal habitat preference appeared to be responsible for observed fine-scale genetic structure, even among subpopulations separated by relatively short straight-line distances (i.e., 1-4 km). Although C. tigris showed strong home range and winter shelter site fidelity, males occasionally moved long distances (max. = 2.7 km), which likely facilitates gene flow among subpopulations, as revealed by relatively high estimates of migration. Interestingly, male C. tigris exhibited significant negative spatial genetic structure at distances up to ~ 100 m. We discuss this finding in the context of kin selection and its potential role in C. tigris mating system dynamics.

INTRODUCTION

Most studies that examine population genetic structure have focused on species with wide- ranging dispersal capabilities, such as birds and mammals (Dobson 1994, Driscoll et al. 2014). Although more powerful techniques to evaluate fine-scale genetic structure (Smouse & Peakall 1999) have led to an increase in studies that focus on less vagile species (e.g., plants, insects, amphibians, and reptiles), one group that remains understudied is snakes (Parker & Plummer 1987, Dorcas & Willson 2009). Perhaps owing to cryptic lifestyles that make them difficult to study (Gibbs & Weatherhead 2001), or simply due to taxonomic bias (Bonnet et al. 2002), snakes are a good choice for studies of fine-scale genetic structure, because many species exhibit limited dispersal capabilities (Gibbs & Weatherhead 2001), complex mating systems (Duvall et al. 1993, Shine 2003), seasonal habitat use (Reinert 1983), and even apparent behavioral adaptations that function to minimize inbreeding (Clark 2004, Clark et al. 2012).

Dispersal is a fundamental life history trait that has profound effects on population dynamics and spatial genetic structure (Clobert 2001). Furthermore, many critical questions in conservation biology depend on knowledge of dispersal patterns across taxa (Driscoll et al. 2014). Regarding snakes, a strong tendency towards philopatry and resulting limited dispersal capabilities are associated with fine-scale genetic structure (Gibbs & Weatherhead 2001); however studies are limited to a small number of taxa.

Many animals show sex-biased dispersal. For example, birds (Greenwood 1980) tend to exhibit female-biased dispersal, while in mammals (Dobson 1982), males are typically the more vagile sex (but see Mabry et al. 2013). Sex-biased dispersal in reptiles in general, and snakes in particular, has not been well studied, but recent research (Rivera et al. 2006, Keogh et al. 2007, Dubey 2008, Pernetta et al. 2011, Hofman et al. 2012) indicates that male-biased dispersal may be more common (but see Lukoschek et al. 2008, Lane & Shine 2011).

Mating system dynamics clearly play an important role in the genetic structure of populations, as evidenced by a plethora of molecular genetic studies focused primarily on fish, birds and mammals (Storz 1999, Avise et al. 2002, Akçay & Roughgarden 2007). Mating systems in snakes have not been well studied empirically, but theoretical treatments have gone so far as to consider snakes as strictly polygynous (Duvall et al. 1993). However, both males and females of most species studied mate with multiple partners, which is more properly termed polygynandry (Rivas & Burghardt 2005). Regardless of terminology, studies have shown that males tend to move more than females, especially during the mating season.

Habitat use is expected to have strong influences on within-population level processes, including reproduction and associated genetic effects (Cushman et al. 2012). This is especially true for organisms that exhibit preferential use of specific landscape features for aspects of their life history. In this case, the relationship between genetic distance and Euclidian distance may be low compared to the relationship between genetic distance and landscape distance (e.g., dispersal associated with approximately linear riparian habitats; Banks & Peakall 2012). Habitat selection in snakes is decidedly multivariate (Reinert 1993), with many species relying on

geomorphological substrata and structural complexity of vegetation to optimize ecological (e.g., predator avoidance) and physiological processes (e.g., thermoregulation). In temperate zone desert environments, where extreme seasonal variation presents a challenge for ectotherms in particular, snakes must find suitable habitats to avoid high temperatures and aridity during the active season, and cold temperatures during the winter months (Mayhew 1968).

Here, we examine the interacting effects of limited dispersal, mating system dynamics, and seasonal habitat use on fine-scale genetic structure of Tiger Rattlesnakes (Crotalus tigris) in the Sonoran Desert of southern Arizona. Previous analyses of genetic differentiation among C. tigris populations from mountain ranges surrounding the Tucson Basin, using Bayesian clustering analyses, provided surprisingly strong support for fine-scale population substructure at one of our long-term study sites in the Rincon Mountains. The discovery of this pattern led us to further examine the potential causes and consequences of fine-scale genetic structure in the context of published data on C. tigris spatial ecology (Goode et al. 2008).

METHODS

Study Sites

We studied C. tigris at three sites in and adjacent to Saguaro National Park (SNP), near Tucson, Arizona, USA (Fig. 1); all three sites were situated in the foothills of the Rincon Mountains. In general, sites were characterized by steep rocky slopes with large outcrops (upper bajada), gentle alluvial slopes (lower bajada) containing a network of small, poorly developed washes, and relatively flat terrain consisting of larger washes with relatively well developed soil terraces and xeroriparian vegetation. Vegetation is typical of Sonoran Desertscrub, Arizona Upland Subdivision (Turner & Brown 1982). Common plants include saguaro (Carnegia gigantea), foothill paloverde (Cercidium microphyllum), brittlebush (Encelia farinosa), prickly pear (Opuntia spp.), cholla (Cylindropuntia spp.) and velvet mesquite (Prosopis velutina), with valley bottom associated species (e.g., creosote bush, Larrea tridentata) at lower elevations.

The Tanque Verde Ridge (TVR) site is located in the extreme southwest corner of the park. This site is bordered by a busy commuter road that runs along the west boundary of the park, and a low-density housing area immediately adjacent to the south boundary. The TVR site is situated at the “toe” of Tanque Verde Ridge where a network of smaller washes drain into a major wash that eventually feeds into Pantano Wash, the main drainage for the Rincon Basin. The elevation (ca., 880-960 m) where we studied C. tigris corresponds to the lower limit of the species’ distribution in the Rincon Mountains. Therefore, C. tigris at this site are confined to the lower rocky slopes and wash network.

The Rocking K (RK) site is situated along the park’s south boundary approximately 4 km east of TVR. This area is slated to become a large-scale development that will include up to 5,000 dwelling units, a resort, golf course, and small commercial zone. Compared to TVR, C. tigris at this site were found further upslope and surrounded by suitable habitat on all sides. This upper bajada location is higher in elevation (ca., 940-1160 m), with numerous deeply cut arroyos that

drain the steep slopes of Tanque Verde ridge, running roughly parallel to each other as they descend into Rincon Creek, which eventually flows into Pantano Wash.

The Loop Road (LR) begins at the SNP Visitor Center and traverses mostly rocky terrain as it winds through the lower bajada of the northwest-facing slope of the Tanque Verde Ridge (elevation range ca., 880-990 m). This area is drained by several larger washes that eventually lead into Tanque Verde Creek, which is a separate watershed from the one that encompasses the TVR and RK C. tigris populations. Capture-Recapture

We searched for C. tigris on foot at TVR and RK from 1997-2001. We also obtained rattlesnakes from colleagues who were conducting tortoise research in the same area. We acquired additional rattlesnakes while radio tracking individuals implanted with transmitters, including those found in association with our study animals, which occurred on several occasions during the summer mating season. And finally, we obtained rattlesnakes while conducting road cruising surveys on LR, which is nine miles in length, and winds through relatively pristine upland desert. We found the large majority of rattlesnakes along an ~6 km stretch of the road that winds through rocky foothill areas within SNP.

For details on handling and processing of snakes see Goode et al. (2008). For individuals not included in the radiotelemetry study, we calculated the maximum distance moved for all recaptured individuals.

Radiotelemetry

We used data obtained from radiotelemetry to examine C. tigris spatial ecology at RK and TVR (cf., Goode et al. 2008 for detailed methods and results). For the present study, we used a variety of parameters to characterize movement patterns by site, sex, and season, including minimum convex polygon (MCP) home range size, length of major MCP axis, total distance moved, average distance moved/day, and maximum distance moved from winter shelter sites over the course of the active season. We defined the pre-mating season as the period of time (ca., April- June) during which individuals moved away from winter shelter sites in route to their summer activity areas. The mating season (ca., July-September) is encompassed by the summer rainy season, which is the period of peak reproductive activity of the species. The post-mating season (ca., October-November) corresponds with the period during which individuals return to their winter shelter sites. These climatic seasons sync well with observations of mating behavior, as indicated by the earliest (July 15) and latest (September 22) dates that we have observed copulation. During the winter months (December-March), C. tigris retreat into rock outcrops and cease above-ground activity. Analyses only included rattlesnakes that were radio tracked ≥30 times.

Habitat Analysis

To examine the potential role of habitat on spatial ecology, we compared habitat use versus availability. To conduct the analysis, we defined the study areas as polygons containing all known radiotelemetry locations at RK and TVR. Because habitat is a species-specific concept, we classified habitat types based on functional aspects of C. tigris ecology and life history. This led to the identification of four habitat types, which included: 1) exposed rock outcrops on upper slopes and ridges, primarily used as winter shelter sites; 2) steep upper slopes connecting winter shelters with lower slopes and washes; 3) gradual alluvial slopes (bajada) containing a network of small, rocky washes; 4) larger washes consisting of a relatively well developed soil terrace, sandy active channel, and dense xeroriparian vegetation. To create habitat maps, we walked the entire perimeter of each habitat type using a Trimble GPS unit. We generated line data, which were differentially corrected using carrier phase post-processing to obtain ≤2 m accuracy. We analyzed habitat data by overlaying the habitat map with radiotelemetry locations and comparing the proportion of habitat types used versus habitat types available using χ2 analysis. Genetic Analyses

Initially, we constructed an enriched genomic library to isolate six microsatellite markers (Goldberg et al. 2003). To increase the number of informative loci, we developed an additional 11 microsatellites (Munguia-Vega et al. 2009). Of the 18 loci available, we confined our analyses to the 12 most informative loci, because their combined probability of identity (PID) was sufficient to discriminate among individual rattlesnakes.

We computed values for standard indices of genetic diversity in GENALEX (Peakall & Smouse 2012). We obtained estimates of inbreeding coefficients using FSTAT (Goudet 1995). We examined loci for deviations from Hardy-Weinberg equilibrium, and evidence of linkage disequilibrium in FSTAT. We tested for the presence of null alleles in GENEPOP (Rousset 2008).

We used three methods to examine genetic differentiation among subpopulations. We computed pairwise population estimates of FST (Wright, 1943) and G’ST (Hedrick 2005) in FSTAT. We then used STRUCTURE (Pritchard et al. 2000), which implements a Bayesian assignment method (Markov Chain Monte Carlo, MCMC) to infer genetic clusters (K). We ran STRUCTURE without location priors and assumed admixture and correlated allele frequencies among putative genetic clusters. Markov Chain Monte Carlo (MCMC) parameters consisted of 10 independent runs per K with 500,000 generations each. We eliminated the first 100,000 generations of each run as burn-in. We used STRUCTURE HARVESTER (Earl & vonHoldt 2012), which applies the ΔK method (Evanno et al. 2005), to estimate the most likely number of genetic clusters. We used CLUMPP to summarize individual assignment probabilities to genetic clusters across all runs for the most likely K value. We used DISTRUCT to plot the results (Rosenberg 2004). And finally, we used GENELAND (Guillot et al. 2008), which also uses Bayesian algorithms to define genetic clusters. However, the program incorporates geographic locations of samples to aid in cluster assignment. GENELAND attempts to detect barriers to gene flow based on genetic discontinuities associated with spatial structure, but has no

information pertaining to underlying landscape structure. We ran 106 MCMC iterations, and 103 thinning, with burn-in set at 500 steps. We used ArcMap 10.2 to overlay the map of population membership produced by GENELAND onto a high-resolution map of the study system, which allowed for examination of underlying landscape features associated with potential barriers.

To test the hypothesis that C. tigris from TVR and RK show less genetic differentiation, compared to LR due to movement along washes that act as dispersal habitat for C. tigris, we compared a pairwise genetic distance matrix to matrices of both Euclidian and wash distance. To calculate wash distances, we first determined the centroid of all samples from both subpopulations in ArcMap 10.2. We then used the path tool and high resolution aerial imagery (©Pima County Association of Governments) to measure distances along main washes originating in the sample space for each subpopulation.

We conducted among- and within-population spatial autocorrelation analyses in GENALEX. We first produced a global spatial autocorrelation plot (correlogram) for the entire dataset, followed by separate analyses of males and females to examine potential differences in spatial genetic structuring due to observed sex-based differences in movement patterns. To test the null hypothesis of no spatial genetic structure, we determined significance of correlation coefficients, r, at each distance interval based on the bootstrapped 95% confidence intervals generated in GENALEX. The x-intercept value is thought to represent the true extent of spatial genetic structure.

We calculated estimates of Nm among subpopulations based on Wright’s formula (FST ≈ 1/(1 + 4Nm), and we obtained estimates of Nm using the private allele method (Barton & Slatkin 1986), as implemented in GENEPOP. We further examined gene flow in BAYESASS (Wilson & Rannala 2003), which employs a MCMC Bayesian approach to estimate recent migration rates, m, among subpopulations.

RESULTS

Capture-Recapture

We captured 217 C. tigris at all sites combined (Table 1, Fig. 2). We captured nearly three times as many males (n = 161; 74.2%) as females (n = 56; 25.8%). The vast majority of rattlesnakes captured were adults (n = 206; 95.0%), but we did find a small fraction of juveniles (n = 6; 2.7%) and neonates (n = 5; 2.3%).

We recaptured 8 individuals 11 times on LR, 5 individuals 5 times at RK and 6 individuals 7 times at TVR. The longest distance moved by a recaptured rattlesnake was by a male that was originally captured on the LR on 08/02/98 and was recaptured one month later (09/03/08) 2.28 km away.

Radiotelemetry

We radio tracked 35 adult C. tigris, four of which we excluded from our analyses, because they did not meet our minimum sample size requirement of ≥30 location fixes. Of the 31 remaining rattlesnakes, 15 were females and 16 were males. We located these 31 snakes a total of 2,453 times over the five-year period of the study.

Published results from the Rincons (Goode et al. 2008) revealed that male active season home ranges (µ = 13.1 ± 2.0 ha) were more than four times larger than female home ranges (µ = 3.9 ± 1.2 ha). Mating season home ranges in males (µ = 10.7 ± 2.5 ha) accounted for 82% of annual home ranges, while female mating season home ranges (µ = 1.84 ± 0.72 ha) only accounted for 47% of annual home ranges. In addition, males (µ = 27.2 m) moved significantly greater distances per day than females (µ = 11.3 m) during the mating season. The longest distance moved from its previous winter shelter site by a male equipped with a radio transmitter was 2,709 m, which took place over a four-week period during the post-mating season in 1998, a year characterized by increased precipitation and reproductive activity. This particular individual overwintered in a new location and, to our knowledge, did not return to its previous winter shelter site. After removing this outlier male from subsequent analyses, the maximum distance (µ = 490 m) that the remaining males moved from their winter shelter sites was similar to females (µ = 447 m).

Home range configuration appeared to be determined by differences in relative location of riparian wash habitat to winter shelter sites. Specifically, C. tigris studied at the RK site tended to use relatively long, linear washes emanating from the steep slopes where they spent the winter. This pattern of wash use resulted in narrow, elongated home ranges. In contrast, C. tigris at the TVR site tended to use a network of washes that drained both sides of Tanque Verde Ridge at the point where it descends into the relatively flat valley bottom. This pattern of wash use resulted in more blocky home ranges. As expected, examination of the primary axes (longest distance between radio tracked locations) of MCP polygons, revealed that males from RK (µ = 1215.7 ± 299.6 m) had narrower and more elongated home ranges than males from TVR (µ = 777.0 ± 112.5 m), as revealed by a significant t-test (t = 3.97, df = 6, p = 0.004).

Habitat Analysis

Because it is not practical to display entire habitat maps for both sites, we provide a portion of the TVR habitat map for purposes of illustration (Fig. 3). Analyses revealed pronounced seasonal differences in habitat use by C. tigris (Fig. 4). The typical pattern consisted of low activity levels in spring and dry summer (pre-mating season), as snakes remained mostly on rocky slopes with some individuals moving onto the bajada and into upper portions of rocky washes. We observed a dramatic increase in movement during the wet summer (mating season) as snakes primarily used main washes and lower portions of rocky washes on the bajada to forage and mate. In fall (post-mating season), rattlesnakes continued to move, although less so than during the mating season, eventually making more directed movements back to their winter shelters located in outcrops along ridges and upper slopes. Chi-squared analyses revealed significant differences in

habitat use compared to habitat availability across all seasons (Spring: χ2 = 147.3, df =3, p < 0.0001; Dry Summer: χ2 = 148.0, df =3, p < 0.0001; Wet Summer: χ2 = 74.7, df =3, p < 0.0001; Fall: χ2 = 157.6, df =3, p < 0.0001; Winter: χ2 = 303.4, df =3, p < 0.0001).

Genetic Analyses

After genotyping all individuals, we determined that ≥80% failed to score at two of the 12 loci; therefore, we dropped these two loci from further analyses, resulting in a final marker set of 10 loci. Of the 217 rattlesnakes captured, we included 151 in genetic analyses. We captured the majority of omitted individuals before we began collecting tissue samples. We removed additional individuals, because we either failed to obtain sufficient quantities of DNA, or they scored at <80% of loci. The probability of identity of the final marker set, based on the chance that the genotypes of two individuals drawn at random would be equivalent to the genotype of two full-sibs (PIDsibs) was LR = 0.00049, RK = 0.00041, and TVR = 0.00033. This is a conservative approach to determining PID of a microsatellite marker panel (Waits et al., 2001).

Standard measures of genetic diversity (Table 2), revealed a similar number of effective alleles (NE), levels of observed and expected heterozygosity (HO and HE, respectively) and inbreeding (FIS) among subpopulations. We found 2-3 loci per subpopulation that did not conform to Hardy- Weinberg equilibrium proportions based on a Bonferroni-adjusted p-value (0.0017) to minimize α inflation due to multiple tests.

We found no evidence of linkage disequilibrium (LD) for any combination of loci across all three subpopulations. We found evidence of one null allele at one locus (CR47) for one subpopulation (LR).

Previous analyses in STRUCTURE, based on C. tigris populations from four mountain ranges surrounding the Tucson Basin, indicated that the Rincon Mountains represented a distinct genetic cluster (Goode et al. in progress). Subsequent analyses in STRUCTURE, confined to the Rincon Mountains, identified two subpopulations with some admixture among sites (Fig. 5). When we performed a separate analysis of the Rincon Mountains in GENELAND, results revealed further subdivision into three distinct genetic clusters, with LR separated from RK and TVR by an apparent barrier, while RK and TVR were distinct, but more closely related to each other than to LR (Fig. 6a). Further analyses in GENELAND, based on posterior probabilities of population membership, confirmed the presence of three distinct subpopulation segments (Fig. 6b-d).

Although overall pairwise FST and G’ST values among subpopulations were low, they supported the existence of genetic discontinuity associated with the putative barrier identified in GENELAND (Fig. 7). In addition, genetic differentiation between RK-TVR was not significant, in contrast to significant differences found between LR-RK and LR-TVR. There was less genetic divergence between LR-RK than LR-TVR, despite the fact that LR-RK are separated by longer distances and a steep, massive ridge.

Based on the private allele method, total number of migrants among all three subpopulations was estimated at 9.1 per generation. Calculated values for Bayesian estimates of migration rates among subpopulations were highly similar, varying from 9.7-10.6% (Table 4); hence, estimates of non-migrants were also highly similar. Based on Wright’s statistic, the number of female migrants among subpopulations was 4.2 per generation, and the number of male migrants was 15.4 per generation.

Combined spatial autocorrelation analyses for all individuals across the three subpopulations revealed significant negative relatedness among individuals at the 100 m distance class, and significant positive relatedness among individuals at the 200 m distance class (Fig. 8). We found no spatial genetic structure at all other distance classes (400-6,400 m). The x-intercept (first point at which positive spatial genetic structure became negative) was 315 m. Spatial autocorrelation analysis performed on adult male and female C. tigris from all subpopulations combined (Fig. 9) revealed significant negative genetic relatedness among males up to 100 m, and significant positive genetic relatedness among females up to 200 m. The x-intercept was 536 m for males, and 268 m for females.

DISCUSSION

Initially, we were interested in potential genetic differentiation among C. tigris populations from mountain ranges surrounding the Tucson Basin. Species accounts (Brennan & Holycross 2009, Goode et al. in press) and our own field data from long-term mark-recapture and radiotelemetry, clearly demonstrate that C. tigris is strongly associated with isolated rocky foothill habitats. Furthermore, data on movement patterns demonstrated strong home range and winter shelter site fidelity (Goode et al. 2008, Goode & Parker 2011). These findings led us to predict that isolated populations of C. tigris may show genetic differentiation among mountain ranges, which could lead to potentially negative effects of genetic drift and inbreeding depression in the absence of gene flow.

To examine these predictions, we used Bayesian clustering techniques, which showed that some populations were comprised of distinct genetic clusters, while other populations showed within- population substructure. In particular, C. tigris in the Rincon Mountains, where we previously conducted long-term field research, showed fine-scale genetic structure among subpopulations associated with an apparent barrier to gene flow that corresponded to a watershed divide. This surprising pattern prompted us to further investigate potential factors that might lead to microgeographic genetic substructure, leading to additional genetic analyses and interpretation in the context of existing ecological data.

We first performed separate analyses, confined to C. tigris from the Rincon Mountains, in STRUCTURE and GENELAND, both of which confirmed subpopulation genetic structure between LR and TVR-RK. Based on examination of high-resolution aerial imagery, the apparent barrier identified in GENELAND appeared to correspond well with a watershed divide separating the Rincon Creek/Pantano and Tanque Verde Creek watersheds. The watershed divide separating these two sub-basins is obvious where it is formed by Tanque Verde Ridge, which is

both massive and steep. However, the divide separating C. tigris subpopulations at lower elevations is subtle, making it difficult to distinguish on imagery. Moreover, the “top” of the watershed along much of Tanque Verde Ridge, and especially where it reaches its lowest elevation, is situated in suitable habitat for C. tigris, which likely facilitates gene flow among subpopulations. Indeed, results from genetic clustering analyses revealed limited admixture between LR and RK, which are more completely separated by Tanque Verde Ridge. In contrast, LR and TVR are separated at lower elevations, where the watershed divide would not appear to present enough of a physical barrier to prevent movement between subpopulations. Results from both estimates of genetic differentiation (FST and G’ST) and migration estimates support our findings from Bayesian clustering analyses.

So what aspects of C. tigris ecology and life history contribute to fine-scale genetic structure in a landscape that might be expected to support a continuous population? In the case of LR and RK, the physical presence of Tanque Verde Ridge could be expected to limit gene flow between these two subpopulations. The maximum elevation at which C. tigris has been found is 1,555 m (Goode et al. in press). The maximum elevation at which Tanque Verde Ridge separates LR and RK ranges from ~1,200-1,600 m. Therefore, we would expect C. tigris to occur, albeit in lower numbers, along most of the watershed divide. However, our data on winter shelter sites, indicates that individuals typically overwinter in large outcrops situated on mid- to lower-elevation slopes. When temperatures become sufficiently warm, C. tigris move downslope into washes, as revealed by habitat analyses.

The genetic subdivision between LR and TVR is more surprising, because the watershed divide does not present a physical barrier in the way that Tanque Verde Ridge does. Therefore, genetic clustering is likely related to the manner in which C. tigris are distributed across the landscape, both spatially and temporally. Given the preference for washes, especially during the mating season, combined with strong home range and winter shelter site fidelity, we hypothesized that proximity of individuals based on wash distance may be a better predictor of genetic structure than Euclidian distance. Although researchers have used numerous methods to identify and quantify the influence of landscape variables on genetic structure and connectivity (Manel et al. 2003), including natural (e.g., topography) and anthropogenic features (e.g., roads), only a few studies have examined the influence of riparian habitat on non-aquatic terrestrial vertebrates (Peakall et al. 2003, Vignieri 2005).

Non-significant Mantel matrix correspondence tests, comparing genetic distance with both wash and Euclidian distance, failed to support the prediction that wash distance was correlated with genetic distance, but also failed to support isolation-by-distance (results not shown). We measured wash distances that included lower elevation segments running through valley bottom habitats. These lower elevation habitats are less likely to be used by C. tigris due to their distance from favored rocky foothill habitats. It seems more likely that the complex network of washes associated with higher elevation bajada habitat may provide stepping-stone connections among subpopulations, with individuals moving short distances over upland habitat to reach adjacent washes. Therefore, we feel that a more sophisticated approach, that incorporates movement resistance values of various landscape features (Segelbacher et al. 2010, Zeller et al. 2012,

Manel & Holderegger 2013), may lead to better discrimination of genetic structure due to habitat use.

The preference for washes reaches its peak during the mating season, which also corresponds to the time period during which males move the most. Although females are more philopatric, they also prefer washes, probably due to stark differences in resource availability (e.g., prey, cover, increased humidity) compared to upland habitats. Furthermore, receptive females are in short supply, because they must secure resources over multiple years to develop adequate body condition for reproduction (i.e., capital breeding; Bonnet et al. 1998), the costs of which can be dramatic in snakes (Shine 2003). The resulting effect on the operational sex ratio (Clutton-Brock & Vincent 1991) likely leads to increased selection favoring males that move more, which is reminiscent of scramble competition polygyny (Alcock 1980). We caution that labeling C. tigris as possessing a scramble competition mating system would seem reductionistic in light of observations of male-male combat in the species (Goode et al. unpublished data), which is consistent with female-defense polygyny (Duvall et al. 1993). Furthermore, both males and females mate multiply, leading us to more accurately term the mating system as polygynandry. Regardless, seasonal use of washes clearly plays a critical role in reproduction, and by extension, within- and among-subpopulation genetic structuring.

Because C. tigris show strong home range fidelity, and typically return to the exact same overwintering sites year after year, dispersal of individuals appears to be rare. This is supported by our field data, which revealed occasional long-distance movements of males. In contrast, females maintain much smaller home ranges than males, and previous research shows that the location of their home ranges remains constant from year to year, as opposed to males who exhibit annual variation in home range placement (Goode et al. 2008). The dramatic difference in movements between males and females is common in snakes (Gregory et al. 1987). From a demographic perspective, we would conclude that male-biased dispersal is the dominant pattern. This conclusion is supported by recent studies that incorporate genetic methods to determine sex- bias in dispersal patterns (Rivera et al. 2006, Keogh et al. 2007, Dubey 2008, Pernetta et al. 2011, Hofman et al. 2012). Our data on C. tigris are suggestive of male-biased dispersal based on results from spatial autocorrelation analyses, as evidenced by significant positive spatial genetic structure up to ~ 200 m in females, but not in males. In addition, pairwise FST values were greater for males than females, indicating less among-population genetic differentiation in the more mobile sex.

In contrast to other studies of snakes, we suggest that male-biased dispersal in C. tigris is likely more a function of gamete dispersal than dispersal of actual individuals. Supporting this idea are data on home range size during the mating season, which comprises 82% of annual home range size in males, whose movement increases as a function of mate searching. However, because males typically return to the same winter shelter sites year after year, there is very little actual displacement of individuals among subpopulations. We did observe one male that established a new winter shelter site after moving ~ 2.7 km over a four-week period during the post-mating season. Although purely speculative, perhaps this male represents a case of outbreeding, brought on by the recognition that previously encountered females were too closely related to result in

optimal reproduction. Although usually thought of as being asocial, increasing evidence indicates that kin recognition may be more prevalent than previously expected in snakes (Shine 2005, Pernetta et al. 2009). This is particularly true in rattlesnakes, a highly derived group that appears to possess surprising (to some) cryptic sociality as mediated through kin selection and chemoreception (Graves & Duvall 1988, Reinert & Zappalorti 1988, Greene et al. 2002, Clark 2004, Lutterschmidt & Roth 2011, Amarello 2012, Clark et al. 2012).

The possibility that males avoid other males seems reasonable if avoiding competitive interactions increases reproductive fitness. Time spent in mate “handling,” which includes physical contests over access to females, would be expected to reduce time spent searching for receptive females (Duvall et al. 1993). However, this does not account for the observed significant negative spatial genetic structure among more closely related males at short distance lags (i.e., ~ 100 m). One plausible explanation is the possibility that natal dispersal in males represents an inbreeding avoidance mechanism (Gandon 1999, Perrin & Mazalov 1999). Male- biased natal dispersal in snakes has been implicated in the lack of spatial genetic structure among males in an Australian colubrid (Dubey et al. 2007). Another, more speculative, explanation for the apparent repulsion of more closely related males may be choosiness on the part of females that discriminate among males based on kin selection (Lehmen & Perrin 2003). As the number of studies combining behavioral ecology and population genetics continues to grow, our understanding of chemical communication and kin recognition is expected to fundamentally change our interpretation of snake ecology (Hatchwell 2010, Scott et al. 2013).

A clear pattern found in snake population genetics studies is an emphasis on temperate species that aggregate in communal hibernacula to survive cold winters associated with higher latitudes. Some dens contain hundreds, and even thousands of individuals. Many well studied snake species, from a genetic perspective, make use of communal dens, which are known to have an effect of population genetic structure (Lougheed et al. 1997, Bushar et al. 1998, Clark et al. 2008, Clark et al. 2010, King 2011). Presumably, snakes from northern climates overwinter in large aggregations, because suitable sites to escape freezing temperatures are a limiting feature of the environment. Sociality, as alluded to above, may also play a role, but it seems more likely to be secondarily adapted.

In contrast, C. tigris overwinter singly, typically in large rock outcrops that apparently provide enough thermal inertia to protect against occasional freezing temperatures. Winters in the northern Sonoran Desert, where we studied C. tigris, are still sufficiently cold to prevent above ground activity from ca. December-April. Furthermore, some individuals, especially males, will remain in their winter shelter sites until May, probably because the species does not exhibit a spring mating period. In contrast, gravid females will emerge as early as late February to bask, thereby maximizing the development of ovarian follicles. Therefore, it seems reasonable to expect that the absence of communal denning may lead to differences in ecology that can have an effect on population genetic structure. Because most long-term studies of snakes have been confined to a small number of species from more northern climates, where extreme seasonality shapes patterns of reproduction, it seems clear that studies of more southerly distributed species,

such as C. tigris, will add greatly to our understanding of the effects of dispersal, habitat use and mating system dynamics on population genetic structure.

Our results have potentially important conservation implications. Populations of C. tigris occur in isolated mountain ranges, and some of these populations appear to form fairly distinct genetic clusters (Goode et al. in progress). Given overall low vagility for the species, combined with preferred use of riparian habitats, it seems likely that gene flow among populations have been, and may continue to be, impeded by anthropogenic barriers (e.g., roads, urban development). This places importance on effective management of higher elevation xeroriparian habitats, and lower elevation riparian corridors to maintain among-population connectivity, which likely occurs in the form of gene flow due to occasional long-distance movements by males over multiple years. Maintaining riparian habitats is especially critical if expected effects of climate change result in increased temperature and aridity (Stromberg et al., Martin et al. 2012). However, in the absence of among-population connectivity, patterns of within-population genetic structure may facilitate “genetic rescue” (Ingvarsson 2001) in subpopulations confined to isolated mountain ranges, as is the case with C. tigris in the Rincon Mountains.

ACKNOWLEDGEMENTS

This study would not have been possible without the hard work and dedication of numerous field assistants, lab technicians, and volunteers. In particular, Scott Breeden, Ed Kuklinski, Ryan Gann†, James Borgmeyer, and Chris Davis went above and beyond the call of duty. We thank Melissa Amarello and Jeff Smith for their many contributions, both in the field and lab, and to previous work cited herein. Taylor Edwards, Caren Goldberg, Matt Kaplan, Adrian Munguia-Vega, Karla Pelz-Serrano, and Anna Carlson provided advice on genetic techniques and contributed to analyses, including genetic processing in the lab. We are grateful to the staff at Saguaro National Park, especially Pam Swantek, Kristin Beaupre, Natasha Kline, Rich Hayes and Meg Weesner, for permitting us to conduct research in the park, and for providing equipment and technical support. The senior author extends his gratitude to Larry Norris, National Park Service (retired) for his generous support of his research program. We greatly appreciate the generosity of local residents, especially Joel Peterson, Shawnee Riplog-Peterson and the Eisentraut family, who provided access to their private land. We thank the Rocking K Ranch for allowing us to work on their land. Mike Rosenzweig, John Koprowski and Cecil Schwalbe provided advice, support, and critical reviews of the manuscript. Our research was supported by grants from Arizona Game and Fish Department Heritage Fund, National Park Service, Western National Parks Association, Reid Park Zoological Society, and Denver Zoological Society.

LITERATURE CITED

Akçay E, Roughgarden J (2007) Extra-pair paternity in birds: review of the genetic benefits. Evolutionary Ecology Research, 9, 855.

Alcock J (1980) Natural selection and the mating systems of solitary bees. American Scientist, 1980, 146-153.

Amarello, M (2012) Social Snakes? Non-random association patterns detected in a population of Arizona black rattlesnakes (Crotalus cerberus). Master’s Thesis, Arizona State University.

Avise JC, Jones, AG, Walker D, DeWoody JA (2002) Genetic mating systems and reproductive natural histories of fishes: lessons for ecology and evolution. Annual Review of Genetics, 36, 19-45.

Balkenhol N, Gugerli F, Cushman SA, Waits LP, Coulon A, Arntzen JW, Holderegger R, Wagner, H. H. (2009) Identifying future research needs in landscape genetics: where to from here? Landscape Ecology, 24, 455-463.

Banks SC, Peakall ROD (2012) Genetic spatial autocorrelation can readily detect sex-biased dispersal. Molecular Ecology, 21, 2092-2105.

Barton NH, Slatkin M (1986) A quasi-equilibrium theory of the distribution of rare alleles in a subdivided population. Heredity, 56, 409-415.

Beyer, HL (2012) Geospatial Modeling Environment (Version 0.6.0.0). http://www.spatialecology.com/gme.

Bonnet X, Bradshaw D, Shine R (1998) Capital versus income breeding: an ectothermic perspective. Oikos, 1998, 333-342.

Bonnet X, Shine R, Lourdais O (2002) Taxonomic chauvinism. Trends in Ecology & Evolution, 17, 1–3.

Brennan TC, Holycross AT (2009) Amphibians and Reptiles in Arizona. Arizona Game and Fish Department, Phoenix, AZ.

Bushar LM, Reinert HK, Gelbert, L (1998) Genetic variation and gene flow within and between local populations of the timber rattlesnake, Crotalus horridus. Copeia, 1998, 411–422.

Clark RW (2004) Kin recognition in rattlesnakes. Proceedings of the Royal Society of London Series B , 271, S243–S245.

Clark RW, Brown WS, Stechert R, Zamudio KR (2008) Integrating individual behaviour and landscape genetics: the population structure of timber rattlesnake hibernacula. Molecular Ecology, 17, 719–730.

Clark R W, Brown W S, Stechert R, Zamudio K R (2010) Roads, interrupted dispersal, and genetic diversity in timber rattlesnakes. Conservation Biology, 24, 1059-1069.

Clark RW, Brown WS, Stechert R, Greene, HW (2012) Cryptic sociality in rattlesnakes (Crotalus horridus) detected by kinship analysis. Biology Letters, 8, 523-525.

Clobert J, Nichols JD, Danchin E, Dhondt A (2001) Dispersal. Oxford University Press, Oxford, UK.

Clutton-Brock TH, Vincent AC (1991) Sexual selection and the potential reproductive rates of males and females. Nature, 351, 58-60.

Cushman SA, Shirk A, Landguth EL (2012) Separating the effects of habitat area, fragmentation and matrix resistance on genetic differentiation in complex landscapes. Landscape Ecology, 27, 369-380.

Dobson FS (1982) Competition for mates and predominant juvenile male dispersal in mammals. Animal Behaviour, 30, 1183-1192.

Dorcas ME, Willson JD (2011) Innovative methods for studies of snake ecology and conservation. In: Snakes: Ecology and Conservation, (eds. Mullin SJ, Seigel RA), Cornell University Press, Ithaca, NY.

Driscoll DA, Banks SC, Barton PS, Ikin K, Lentini P, Lindenmayer DB, Smith AL, Berry LE, Burns EL, Edworthy A, Evans MJ, Gibson R, Heinsohn R, Howland B, Kay G, Munro N, Scheele BC, Stirnemann I, Stojanovic D, Sweaney N, Villasen͂or N, Westgate MJ (2014) The trajectory of dispersal research in conservation biology: a systematic review. PloS ONE, 9, e95053.

Dubey S, Brown GP, Madsen T, Shine R (2008) Male-biased dispersal in a tropical Australian snake (Stegonotus cucullatus, Colubridae). Molecular Ecology, 17, 3506-3514.

Duvall D, Schuett GW, Arnold SJ (1993) Ecology and evolution of snake mating systems. In: Snakes: ecology and behavior, (eds. Seigel RA, Collins JT), pp. 165-200. McGraw-Hill, Inc., New York, NY.

Earl DA, vonHoldt BM (2012) STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genetics Resources, 4, 359-361.

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

Gandon S (1999) Kin competition, the cost of inbreeding and the evolution of dispersal. Journal of Theoretical Biology, 200, 345–364.

Gibbs HL, Weatherhead PJ (2001) Insights into population ecology and sexual selection in snakes through the application of DNA-based genetic markers. Journal of Heredity, 92, 173- 179.

Goode M, Smith JJ, Amarello M (2008) Seasonal and annual variation in home range and movements of Tiger Rattlesnakes (Crotalus tigris) in the Sonoran Desert of Arizona. In: The Biology of Rattlesnakes (eds. Hayes WM, Beaman KR, Cardwell MD, Bush SP), Loma Linda University Press, Loma Linda, CA.

Goode, M, Parker MR, Feldner MJ, Repp R. (in press) Crotalus tigris species account. In Rattlesnakes of Arizona (eds. Schuett GW, Reiserer RS, Smith CF, Feldner MJ), Eco Press: Rodeo, NM.

Goudet J (1995) FSTAT (version 1.2): a computer program to calculate F-statistics. Journal of Heredity, 86, 485-486.

Graves BM, Duvall D (1988) Evidence of an alarm pheromone from the cloacal sacs of prairie rattlesnakes. The Southwestern Naturalist, 1988, 339-345.

Greene HW, May PG, Hardy DL, Sciturro JM, Farrell TM (2002) Parental behavior by vipers. In: Biology of the Vipers, (eds. Schuett GW, Hoggren M, Douglas ME, Greene HW), Eagle Mountain Publishing, Utah.

Greenwood PJ (1980) Mating systems, philopatry and dispersal in birds and mammals. Animal Behaviour, 28, 1140-1162.

Gregory PT, Macartney JM, Larsen KW (1987) Spatial patterns and movements. In: Snakes: Ecology and Evolution, (eds. Seigel RA, Collins JT, Novak SS), Macmillan, New York, NY.

Guillot G, Santos F, Estoup A (2008) Analysing georeferenced population genetics data with Geneland: a new algorithm to deal with null alleles and a friendly graphical user interface. Bioinformatics, 24, 1406-1407.

Hatchwell BJ (2010) Cryptic kin selection: kin structure in vertebrate populations and opportunities for kindirected cooperation. Ethology, 116, 203-216.

Hedrick PW (2005) A standardized genetic differentiation measure. Evolution, 59, 1633– 1638.

Hofmann S, Fritzsche P, Solhøy T, Dorge T, Miehe G (2012) Evidence of sex-biased dispersal in Thermophis baileyi inferred from microsatellite markers. Herpetologica, 68, 514-522.

Holderegger R, Wagner HH (2006) A brief guide to landscape genetics. Landscape Ecology, 21, 793-796.

Ingvarsson PK (2001) Restoration of genetic variation lost – the genetic rescue hypothesis. Trends in Ecology and Evolution, 16, 62–63

Keogh JS, Webb JK, Shine R (2007) Spatial genetic analysis and long-term mark–recapture data demonstrate male-biased dispersal in a snake. Biology Letters, 3, 33-35.

King RB (2009) Population and conservation genetics. In: Snakes: Ecology and Conservation (eds. Mullin SJ, Seigel RA), Cornell University Press, Ithaca, NY.

Lane A, Shine R (2011) Intraspecific variation in the direction and degree of sex-biased dispersal among sea-snake populations. Molecular Ecology, 20, 1870-1876.

Lehmann L, Perrin N (2003) Inbreeding avoidance through kin recognition: choosy females boost male dispersal. The American Naturalist, 162, 638-652.

Lougheed SC, Gibbs HL, Prior KA, Weatherhead PJ (1999) Hierarchical patterns of genetic population structure in black rat snakes (Elaphe obsoleta obsoleta) as revealed by microsatellite DNA analysis. Evolution, 53, 1995-2001.

Lukoschek V, Waycott M, Keogh JS (2008) Relative information content of polymorphic microsatellites and mitochondrial DNA for inferring dispersal and population genetic structure in the olive sea snake, Aipysurus laevis. Molecular Ecology, 17, 3062-3077.

Lutterschmidt WI, Roth ED (2011) Experimental validation of sex differences in spatial behavior patterns of freeranging snakes: implications for social interactions. Ethology, 117, 852-858.

Mabry KE, Shelley EL, Davis KE, Blumstein DT, Van Vuren DH (2013) Social mating system and sex-biased dispersal in mammals and birds: a phylogenetic analysis. PloS ONE, 8, e57980.

Manel S, Schwartz MK, Luikart G, Taberlet P (2003) Landscape genetics: combining landscape ecology and population genetics. Trends in Ecology and Evolution, 18, 189–197.

Manel S, Holderegger R (2013) Ten years of landscape genetics. Trends in Ecology & Evolution, 28, 614-621.

Martin J, Kurc SA., Zaimes G, Crimmins M, Hutmacher A, Green D (2012) Elevated air temperatures in riparian ecosystems along ephemeral streams: The role of housing density. Journal of Arid Environments, 84, 9-18.

Mayhew WW (1968) Biology of desert amphibians and reptiles. Desert Biology. Special Topics on the Physical and Biological Aspects of Arid Regions, 1, 226-229.

Munguia-Vega A, Pelz-Serrano K, Goode M, Culver M (2009). Eleven new microsatellite loci for the tiger rattlesnake (Crotalus tigris). Molecular Ecology Resources, 9, 1267-1270.

Nomura T (2008) Estimation of effective number of breeders from molecular coancestry of a single cohort sample. Evolutionary Applications, 1, 462-474.

Parker WS, Plummer MV (1987) Population Ecology. In Snakes: ecology and evolutionary biology (eds. Seigel RA, Collins JT, Novak SS), pp. 253-301, McGraw-Hill, Inc., New York, NY.

Peakall R, Ruibal M, Lindenmayer DB (2003) Spatial autocorrelation analysis offers new insights into gene flow in the Australian bush rat, Rattus fuscipes. Evolution, 57, 1182-1195.

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.

Pernetta AP, Allen JA, Beebee TJC, Reading CJ (2011) Fine-scale population genetic structure and sex-biased dispersal in the smooth snake (Coronella austriaca) in southern England. Heredity, 107, 231-238.

Perrin N, Mazalov V (1999) Dispersal and inbreeding avoidance. American Naturalist, 154, 282–292.

Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics, 155, 945–959.

Reinert HK (1993) Habitat selection in snakes. In: Snakes: Ecology and Behavior, (eds. Seigel RA, Collins JT), McGraw-Hill, New York, NY.

Reinert HK, Zappalorti RT (1988) Field observation of the association of adult and neonatal timber rattlesnakes, Crotalus horridus, with possible evidence for conspecific trailing. Copeia, 1988, 1057-1059.

Rivas JA, Burghardt GM (2005) Snake mating systems, behavior, and evolution: the revisionary implications of recent findings. Journal of Comparative Psychology, 119, 447.

Rivera PC, Gardenal CN, Chiaraviglio M (2006) Sex-biased dispersal and high levels of gene flow among local populations in the argentine boa constrictor, Boa constrictor occidentalis. Austral Ecology, 31, 948-955.

Rosenberg NA (2004) DISTRUCT: a program for the graphical display of population structure. Molecular Ecology Notes, 4, 137-138.

Rousset F (2008) Genepop 2007: a complete reimplementation of the Genepop software for Windows and Linux. Molecular Ecology Resources, 8, 103-106.

Scott ML, Whiting MJ, Webb JK, Shine R (2013) Chemosensory discrimination of social cues mediates space use in snakes, Cryptophis nigrescens (Elapidae). Animal Behaviour, 85, 1493- 1500.

Segelbacher G, Cushman SA, Epperson BK, Fortin MJ, Francois O, Hardy OJ, Holderegger R, Taberlet P, Waits LP, Manel, S. (2010). Applications of landscape genetics in conservation biology: concepts and challenges. Conservation Genetics, 11, 375-385.

Shine R (2003) Reproductive strategies in snakes. Proceedings of the Royal Society of London. Series B: Biological Sciences, 270, 995-1004.

Shine R (2008) Tracking elusive timber rattlers with molecular genetics. Molecular Ecology, 17, 715-716.

Smouse PE, Peakall R (1999) Spatial autocorrelation analysis of individual multiallele and multilocus genetic structure. Heredity, 82, 561-573.

Storfer A, Murphy MA, Evans JS, Goldberg CS, Robinson S, Spear SF, Dezzani R, Delmelle E, Vierling L, Waits, LP (2007) Putting the ‘landscape’ in landscape genetics. Heredity, 98, 128- 142.

Storfer A, Murphy MA, Spear SF, Holderegger R, Waits, LP (2010) Landscape genetics: where are we now? Molecular Ecology, 19, 3496-3514.

Storz JF (1999) Genetic consequences of mammalian social structure. Journal of Mammalogy, 80, 553-569.

Stromberg JC, Lite SJ, Dixon MD (2010) Effects of stream flow patterns on riparian vegetation of a semiarid river: implications for a changing climate. River Research and Applications, 26, 712-729.

Turner RM, Brown DE (1994) Tropical-subtropical desertlands. In: Biotic Communities of the American Southwest – United States and Mexico. University of Utah Press, Salt Lake City, UT

Vignieri SN (2005) Streams over mountains: influence of riparian connectivity on gene flow in the Pacific jumping mouse (Zapus trinotatus). Molecular Ecology, 14, 1925-1937.

Waits LP, Luikart G, Taberlet P (2001) Estimating the probability of identity among genotypes in natural populations: cautions and guidelines. Molecular ecology, 10, 249-256.

Wilson GA, Rannala B (2003) Bayesian inference of recent migration rates using multilocus genotypes. Genetics, 163, 1177-1191.

Wright S (1943). Isolation by distance. Genetics, 28, 114-138.

Wright S (1969) Evolution and the Genetics of Populations, Vol. 2. University of Chicago Press, Chicago, IL.

Zeller KA, McGarigal K, Whiteley AR (2012) Estimating landscape resistance to movement: a review. Landscape Ecology, 27, 777-797.

FIGURES

TVR RK

Figure 1. Map showing our study sites located within and adjacent to Saguaro National Park (green shading) in the foothills of the Rincon Mountains near Tucson, Arizona, USA (1997-2001). LR=Loop Road, TVR=Tanque Verde Ridge, and RK=Rocking K.

Figure 2. Locations of all C. tigris captured (N = 217) from three sites within the foothills of the Rincon Mountains near Tucson, Arizona from 1997-2001. LR = Loop Road, TVR = Tanque Verde Ridge, RK = Rocking K. The black line represents the location of the watershed divide that separates LR from TVR and RK.

250 m Wash

Ridge Outcrops Slope Outcrops Slope

Bajada

Figure 3. Habitat map generated in GIS by walking the perimeters of habitat types with a GPS receiver. Habitat types are based on functional aspects of C. tigris ecology and life history. The view depicted was clipped from the complete habitat map to provide greater detail for purposes of illustration.

Habitat Use vs. Availability (C. tigris) Fall Wet Summer Dry Summer Spring Available 100% 90% 80% 70% 60% % 50% 40% 30% 20% 10% 0% Wash Bajada Slope Outcrops

Figure 4. Histogram depicting the proportion of C. tigris locations (N = 2,453) relative to availability of each habitat type by season. Spring and Dry Summer correspond to the pre-mating season, Wet Summer corresponds to the mating season, and Fall corresponds to the post-mating season. The use of washes far out of proportion to their availability during the mating season is of primary interest for the present study.

TVR RK LR

Figure 5. Fine-scale genetic structure among three C. tigris subpopulations from the Rincon Mountains as revealed by STRUCTURE. TVR = Tanque Verde Ridge, RK = Rincon Mountains, LR = Loop Road.

Map of posterior probability to belong to cluster 3 Map of posterior probability to belong to cluster 2 Map of posterior probability to belong to cluster 1

0.4 0.2

0.35 0.2

0.2

0.3 0.45 0.2 (a) (b) (c) (d) 0.25 0.15 0.2

0.45

0.15 0.2 0.4 0.2 0.25

0.4 0.2

0.2

0.35 0.25 0.25

0.25

0.45 0.35 0.35 0.2 0.25 0.4 0.25 0.3 0.45 0.25

0.15

y coordinates 0.25 0.3 0.4 0.15 0.3 0.25

0.3 0.3 y coordinates y coordinates

0.3 0.25 0.25 0.2 0.2 0.3 0.2 0.3 0.25 Figure 6. Output from Geneland showing population membership (a) map of three C. tigris subpopulations 0.2 0.2 from 0.2 0.25 0.25 the Rincon Mountains. Adjacent panels depict “heat” maps of subpopulation membership based on posterior 0.2

probability “topography” for LR (b), TVR (c) and RK (d). 0.3 0.25

3550000 3555000 3560000 0.2 3565000 0.2 520000 0.2 525000 530000 535000 540000 0.2 0.25 x coordinates 0.2 0.2 0.2 3550000 3555000 3560000 3565000 3550000 3555000 3560000 3565000 520000 525000 530000 535000 540000 520000 525000 530000 535000 540000 x coordinates x coordinates

FST = 0.009* G’ST = 0.013*

FST = 0.014* G’ST = 0.031*

TVR RK FST = 0.007 G’ST = 0.005

Figure 7. Diagram depicting genetic differentiation, as expressed by FST, and G’yST, among three C. tigris subpopulations (LR = Loop Road, TVR = Tanque Verde Ridge, RK = Rocking K) from the Rincon Mountains. Dashed line depicts the putative barrier identified by GENELAND.

0.060

0.040

0.020

r 0.000

-0.020

-0.040

-0.060 100 200 400 800 1600 3200 6400 Distance (m)

Figure 8. Correlogram showing spatial genetic structure of all three subpopulations combined. Results indicate significant negative genetic structure (filled squares) at 100 m and significant positive genetic structure at 200 m. Intercept = 315 m.

0.250 0.200 0.150 0.100 0.050

r 0.000 -0.050 -0.100 -0.150 -0.200 -0.250 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 Distance (m)

Figure 10. Spatial autocorrelation correlogram showing spatial genetic structure for males (squares) and females (circles). Results indicate significant negative autocorrelation at the 100 m distance interval for males (filled square), and significant positive autocorrelation for females (filled circles) at 100 m and 200 m distance intervals. Intercept for females = 268 m, intercept for males = 536 m.

TABLES

Table 1. Total number of C. tigris captured by site, sex and age class in three subpopulations from the Rincon Mountains.

Category Tanque Verde Ridge Rocking K Loop Road Sex Males 26 46 89 Females 15 20 21 Total 41 66 110 Age Class Adults 41 59 106 Juveniles 0 5 1 Neonates 0 2 3 Total 41 66 110

Table 2. Genetic diversity of three C. tigris subpopulations from the Rincon Mountains: Tanque Verde Ridge (TVR); Rocking K Ranch (RK); Loop Road (LR). Listed are number of individuals genotyped (N), number of alleles (NA), number of effective alleles (NE), observed heterozygosity (HO), expected heterozygosity (HE), and loci with significant deviation from Hardy-Weinberg equilibrium after Bonferroni corrections for multiple tests (HWE-dev).

Subpop N NA NE FIS HO HE HWE-dev TVR 24 9.90 6.23 0.32 0.60 0.66 CR47, CR95 RK 60 11.70 6.34 0.33 0.58 0.64 CR06, CR10, CR47 LR 67 13.20 6.99 0.33 0.56 0.61 CR06, CR19, CR47

Table 3. Effective population size (Ne) estimates for C. tigris subpopulations from the Rincon Mountains: Tanque Verde Ridge (TVR); Rocking K Ranch (RK); Loop Road (LR).. Subpopulation N Ne UCI LCI # AllelesIND TVR 24 23 15 43 965 RK 60 116 66 319 979 LR 67 355 96 Infinite 547

Table 4. Migration rate estimates among C. tigris subpopulations from the Rincon Mountains – Tanque Verde Ridge (TVR); Rocking K Ranch (RK); Loop Road (LR). Values in parentheses are the permuted confidence intervals (±) bounding the estimate. Diagonal values correspond to non-migrant individuals. Subpopulation TVR RK LR TVR 0.798 (0.078) 0.097 (0.078) 0.106 (0.083) RK 0.103 (0.083) 0.794 (0.079) 0.103 (0.083) LR 0.106 (0.083) 0.094 (0.079) 0.800 (0.080)

APPENDIX C

Manuscript prepared for submission – PloS ONE

Movement patterns and reproductive behavior drive sex-based differences in fine-scale genetic structure and parentage in a polygynandrous rattlesnake

Matt Goode1, Mickey Ray Parker1, Alex Ochoa1 and Melanie Culver1,2

1Wildlife Conservation and Management Program, School of Natural Resources and the Environment, University of Arizona, 325 Biological Sciences East, Tucson, AZ 85721, USA, 1,2US Geological Survey, Arizona Cooperative Fish and Wildlife Research Unit, School of Natural Resources and the Environment, University of Arizona, 213 Biological Sciences East, Tucson, AZ 85721, USA

Corresponding author: Matt Goode 325 Biological Sciences East School of Natural Resources and the Environment University of Arizona Tucson, AZ 85721 [email protected]

ABSTRACT

Species with polygynandrous mating systems, characterized by sexual promiscuity in both sexes, should exhibit non-random mating if selection favors males that are better at finding and successfully copulating with females. Even in species that exhibit male-male competition (i.e., combat), simply locating receptive females may be more important than the outcome of physical contests, especially if the frequency of such contests is relatively low. Male-biased dispersal may also be expected in a polygynandrous mating system, where females are the more philopatric sex. However, if multiple mate finding strategies are employed, and pre- and post-copulatory mechanisms (e.g., kin recognition, sperm competition) are operating, then differences in gender- based spatial ecology and reproductive behavior may not be good predictors of parentage. We combined data from a long-term field study with genetic analysis using microsatellite DNA markers to examine potential effects of movement patterns and reproductive behavior on spatial genetic structure and parentage in Tiger Rattlesnakes (Crotalus tigris).

INTRODUCTION

Reproductive strategies in snakes are both complex and highly variable. Costs and benefits associated with reproduction can be dramatically different for males than for females, presumably leading to intense sexual selection on behavior, morphology, physiology and endocrine function (Shine 2003, Smith et al. 2015). Theoretical attempts to characterize snake mating systems have led to divergent conclusions, ranging from strict polygyny (Duvall et al. 1993) to polyandry (Rivas & Burghardt 2006). Recent studies, based on genetic analyses of parentage, indicate that multiple paternity is common in snakes (Uller & Olsson 2008), and in many species, both males and females mate with multiple partners, both of which support polygynandry (sensu lato promiscuity; Arnold & Duvall 1994) as the dominant mating system in snakes (Wusterbarth et al. 2010). Adding to the complexity is the fact that many snake species are known to mate guard (Clark et al. 2014), male-male combat is common across taxa (Gillingham 1987), and long-term sperm storage and sperm competition have been well documented, implying the importance of both pre-and post-copulatory mechanisms in mating success (Duvall et al. 1993, Smith et al. 2015).

Research on snakes has clearly lagged behind other vertebrate groups, largely due to their cryptic lifestyles (Parker & Plummer 1987, Bonnet et al. 2002). However, radiotelemetry has revolutionized the study of snakes, making it possible to continuously observe individuals in the field, which leads to precise quantification of movement patterns and behavioral observations that are critical to our understanding of snake ecology (Reinert 1992, Dorcas & Willson 2009). A clear pattern that has emerged from intensive studies of snakes is the discrepancy in both frequency and magnitude of movements between males and females, particularly during the mating season when males spend a great deal of time in search of receptive females. This pattern has been commonly observed across snake taxa, especially in North American pitvipers (Duvall & King 1985, Brown 1993, Jellen et al. 2007, Goode et al. 2008, Glaudus & Rodriguez 2011, Putnam et al. 2013, Clark et al. 2014, Smith et al. 2015).

Although the use of hypervariable microsatellite DNA loci has lead to a revolution in our understanding of behavioral ecology, population biology and kinship, the promise of this approach (Gibbs & Weatherhead 2001) has not been realized in snakes compared to other taxa (Clark et al. 2014). However, a growing body of literature on genetic analyses of parentage in snakes indicates that combining intensive field research and molecular genetic techniques can lead to insights that would not have been possible based on field research alone (Prosser et al. 2002, Weatherhead et al. 2002, Blouin-Demers et al. 2005, Dubey et al. 2009, Ursenbacher et al. 2009, Clark et al. 2012, Meister et al. 2013, Clark et al. 2014, Levine et al. 2015).

Patterns of dispersal are fundamental to an understanding of population and mating system dynamics, and the role they play in spatial genetic structure (Clobert 2001). Many critical questions in ecology and conservation depend on knowledge of dispersal patterns across taxa (Driscoll et al. 2014). Regarding snakes, a strong tendency towards philopatry and resulting limited dispersal capabilities are associated with fine-scale genetic structure (Gibbs & Weatherhead 2001); however, studies are limited to a small number of taxa. Many animals show

sex-biased dispersal. For example, birds (Greenwood 1980) tend to exhibit female-biased dispersal, while in mammals (Dobson 1982), males are typically the more vagile sex (but see Mabry et al. 2013 for a phylogenetic perspective). Recent research indicates that male-biased dispersal occurs in some snakes (Rivera et al. 2006, Keogh et al. 2007, Dubey 2008, Pernetta et al. 2011, Hofmann et al. 2012), but not in all species (Lukoschek et al. 2008, Lane & Shine 2011).

Previous studies of the Tiger Rattlesnake, Crotalus tigris, (Goode et al. 2008, Goode & Parker 2011) clearly demonstrates that males move more often and farther than females, especially during the mating season, when they appear to spend a great deal of time in search of receptive females, which are in short supply due to greater-than-annual female reproductive cycles. In fact, males are so intent on searching for females, they may forego feeding, as evidenced by the relative absence of fecal material found in the hindgut of males captured during the mating season (Goode et. al. in press). We have also observed both males and females mating with multiple partners; however, reproductive behaviors observed in the field may not correlate well with actual parentage (Prosser et al. 2002, Clark et al. 2014). In addition, we have observed male-male combat on multiple occasions (this study, Goode et al. in press,).

We combined long-term (2002-2008) radiotelemetry and mark-recapture data with molecular genetic analyses to examine potential effects of movement patterns and reproductive behavior on spatial genetic structure, and parentage in a population of C. tigris from the Tortolita Mountains near Tucson, Arizona.

METHODS

Study Site

We studied C. tigris at Stone Canyon (SC) in the foothills of the Tortolita Mountains near Tucson, Arizona, USA (Fig. 1). The study site is situated at the urban fringe and consists of a low-density residential development and golf course interspersed with intact Sonoran Desert upland desert. The landscape is characterized by steep, rocky slopes and scrub- dominated flats interspersed with isolated rock outcrops. The golf course, roads and homes are landscaped with native vegetation, which has been augmented by run off and drip irrigation, resulting in an obviously more productive and humid environment that supports relatively dense populations of C. tigris (among other snake species). The landscape surrounding the human-occupied area has not been developed, and we found a significant proportion of the rattlesnakes in these immediately adjacent areas. Vegetation is typical of Sonoran Desertscrub, Arizona Upland Subdivision (Turner & Brown 1982). Common plants include saguaro (Carnegia gigantea), foothill paloverde (Cercidium microphyllum), brittlebush (Encelia farinosa), prickly pear (Opuntia spp.), cholla (Cylindropuntia spp.), and velvet mesquite (Prosopis velutina). Elevation ranges from approximately 900-1,000 m.

Rattlesnake Capture-Recapture

We searched for rattlesnakes on foot, during road and golf cart path surveys, and while radio tracking individual rattlesnakes implanted with transmitters, including those found in association with radio-tracked animals. We obtained a small number of rattlesnakes from golf course maintenance personnel, from residents who called us when they encountered rattlesnakes on their property, and from the fire department that provides a snake-removal service.

We anesthetized all rattlesnakes captured to obtain accurate measurements of mass (to the nearest 0.1 g), snout-vent length (to the nearest 1 mm), and a suite of other variables (e.g., head size, tail length, presence of feces and/or food bolus) not included in the present analysis. We sexed individuals with a clean sexing probe inserted into the cloaca to determine the presence or absence of hemipenes. We obtained blood (~ 50 µl) from the caudal vein using a blood collection set with a 0.5 ml sterile needle. Blood samples were stored in sterile blood collection tubes containing EDTA (Na2) buffer at -20° C. We released rattlesnakes at their exact capture locations 6-36 h after capture.

For individuals not included in radiotelemetry, we used an unpaired t-test to compare mean distances moved between successive capture locations, and maximum distance moved from initial capture locations of males and females, many of which we recaptured several times over the course of the study.

Radiotelemetry

We surgically implanted transmitters (Holohil Ltd., Model SI-2T) into a subset of adult male and female C. tigris (see Goode et al. 2008 for details). We obtained geographic coordinates of rattlesnake locations (accurate to <5 m), and imported them into ArcMap v. 10.2.2 (Esri, Inc.), and Geospatial Modeling Environment (Beyer 2012) for display and spatial analyses.

We compared total distance moved, maximum distance moved from winter shelter sites, home range area, and length of home range major axis between male and female rattlesnakes before, during, and after the mating season using two-way ANOVAs for each parameter (©JMP v. 11). We defined the mating season as the period of time between the first and last dates we observed rattlesnakes engaged in either courtship or copulation (July 15 – September 22). The mating season corresponds well with the summer monsoon season (Adams & Comrie 1997), which is a period of increased humidity and precipitation, leading to peak activity in C. tigris, especially for males (Goode et al. 2008). We used the minimum convex polygon (MCP) method to calculate home ranges, because it is thought to be more appropriate for snakes and other herpetofauna (Row et al. 2006).

Reproductive Behavior

Each time we captured or radio tracked a rattlesnake, we recorded whether or not it was with a conspecific of the opposite sex. We classified each conspecific pairing into one of three

categories (i.e., accompaniment, courtship, copulation). We defined accompaniment as a male and female located within 2 m of each other. Courtship consisted of obvious behaviors, such as chin pressing and tail searching, as described by Klauber (1972). Copulation, which can be difficult to discern from courtship (Schuett & Gillingham 1988), was only scored if we were certain that intromission had occurred, as evidenced by a distended cloaca of the female, or the female dragging the male around while in copulo.

Genetic Analyses

We genotyped individuals at 10 loci developed specifically for C. tigris (Goldberg et al. 2003, Munguia-Vega et al. 2009). We used GENALEX (Peakall & Smouse 2012) to obtain standard indices of genetic diversity. We screened loci for deviations from Hardy-Weinberg equilibrium using GENODIVE (Miermans & Van Tienderen 2004), which uses exact probability tests (10,000 permutations). We controlled for increased Type I error rates due to multiple testing using a sequential Bonferroni correction. We tested for evidence of linkage disequilibrium between all pairs of loci in each population using FSTAT (Goudet 1995), and we assessed significance after adjusting for multiple tests. We examined loci for the presence of null alleles in GENEPOP (Rousset 2008). We used FSTAT to estimate inbreeding coefficients (FIS) and allelic richness (AR). We obtained estimates of effective population size (Ne) in NeESTIMATOR v. 2.1 (Do et al. 2014) using the linkage disequilibrium method. We evaluated the power of the entire microsatellite panel to distinguish among individuals by estimating probability of identity (PID) in GENALEX.

We conducted parentage analyses using COLONY 2 (Jones & Wang 2009). We chose COLONY 2 for several reasons. First, it can deal with departures from Hardy-Weinberg equilibrium. Second, it uses a full-pedigree maximum-likelihood approach, which is appropriate for our data set, because we genotyped offspring from known mothers to inform downstream analyses of parentage. Third, COLONY 2 can account for genotyping errors due to allelic drop out, null alleles and false alleles (Wang 2004). And fourth, COLONY 2 outperformed other software programs that use different assignment methods (Harrison et al. 2012).

To optimize parentage assignment in COLONY 2, we determined locus-specific error rates for all 10 loci using MICROERRORANALYZER (Wang 2010). Based on estimated rates of null alleles, allelic dropout and false alleles, we removed two loci from subsequent parentage analyses, because their combined error rates exceeded 20%. We then used estimated locus- specific error rates from the remaining eight loci used in the final analysis. To obtain confidence in the ability of COLONY 2 to successfully assign parentage, we initially ran the program without providing information on known mother-offspring pairs. Because the program identified all known mother-offspring pairs correctly, we conducted subsequent analyses using the data on known pairs to maximize parentage assignment. To further ensure that our results were valid, we re-ran the program, which resulted in identical sibship assignments across runs. We retained all parentage assignments with maximum likelihood probabilities of >85%. We performed a two- sample, unpaired t-test to examine differences between offspring to mothers and offspring to fathers.

We performed spatial autocorrelation analyses in GENALEX v. 6.5 (Peakall & Smouse 2012), which implements a method that computes an autocorrelation coefficient (r) obtained from multiallelic, codominant loci, thereby strengthening the spatial relatedness signal by reducing noise associated with stochastic processes that arise from classic allele-by-allele and locus-by- locus analyses (Smouse & Peakall 1999). We first ran “global” spatial autocorrelation analyses for the spatial extent of the entire study area (ca., 3200 m) at distance classes ranging from 25- 400 m. We then conducted analysis of spatial autocorrelation at multiple distance classes to further examine the geographic extent of spatial autocorrelation for the entire population. Because movements and home range size vary among male and female C. tigris, we used the “Pops as DClass” function in GENALEX to examine potential differences in spatial autocorrelation between the sexes, with the prediction that females, the more philopatric sex, would exhibit a greater degree of genetic relatedness at shorter distance lags. We performed the same analysis comparing adults and offspring to examine the potential for age-biased dispersal. We assessed significance of computed r values using 95% confidence intervals generated from both permuted and bootstrap resampling of r-values. An r-value that exceeds the 95% permuted CI, combined with non-overlap of the x-axis (r = 0) is considered significant at the α = 0.05 level, thereby failing to support the null hypothesis of no spatial structure at that particular distance interval.

RESULTS

Rattlesnake Capture-Recapture

We captured 460 (290 males, 170 females) C. tigris at Stone Canyon, of which 357 were used in genetic analyses (304 adults, 22 juveniles, and 31 neonates [young-of-the-year]). We included an additional 31 neonates born to known mothers in parentage analyses.

Not including rattlesnakes implanted with transmitters (n = 71), which we recaptured in the spring and fall of each year to obtain growth data and assess reproductive condition, we recaptured the remaining 389 rattlesnakes 154 times. The mean distance moved between successive captures was 424 m for males, and 181 m for females, but this difference was not significant (df = 177, t = 1.2, p = 0.2491). The mean number of months between recaptures was 18.9 ± 1.2 (n = 155 individuals, range = 1.0 – 61.5 months) for males, and 19.3 ± 3.4 (n = 24 individuals, range = 1.1 – 58.7 months) for females. The maximum distance moved from the original capture location to the most distant recapture point was 1,880 m for males and 496 m for females. In total, 11 males moved over 1,000 m (range = 1,020-1,881 m) from their original capture locations. We recaptured one male, originally captured as a neonate, 1,799 m from its original capture location nearly two years later. We also captured three juveniles that we later captured as adults; one male moved 1,124 m over an 11-month period, one female moved 10 m over a 22-month period, and the other female moved 178 m over an 11-month period.

Radiotelemetry

We implanted transmitters into 71 (i.e., 34 males, 37 females) adult C. tigris,, which we located 6,704 times over the seven-year period of the study. Movements and home range size varied by sex and season (Table 1). Analyses revealed a significant effect of season and sex for total distance moved, and maximum distance moved from winter shelter sites. For MCP size, as well as MCP major axis length, there was a significant effect of sex, but not season (Table 2).

Reproductive Behavior Over the course of the study, we observed copulation on 12 occasions, courtship on four occasions, and accompaniment (male-female pair separated by <2 m) on 48 occasions. We observed male-male combat on four occasions. Of the pairings we observed, seven of them lasted from 3-19 days, although we could not be sure if the two snakes remained together the entire time, because we did not radio track the individuals every day. Furthermore, when we did radio track them, the individual without a transmitter could have been present but undetected. We observed copulation in four of these seven extended pairings., including one pair that copulated four times over a nine-day period. We observed six females that paired with different males in a single year, although we did not observe copulation or courtship in any of these pairings. We only observed one female that was revisited by the same male in the same year (unless males left and returned during protracted pairings, as alluded to above); these observations were separated by ~2 months. We observed two females that were accompanied by the same male in successive years, but we did not observe courtship or copulation in any of these pairings. And finally, one female was accompanied by four different males in four successive breeding seasons, but we did not observe courtship or copulation at any time. We observed all reproductive behaviors, most of which involved radio-tracked individuals, between July 15- September 22.

Genetic Analyses

We removed 103 individuals from the 460 total snakes captured, because we either failed to obtain high-quality DNA (usually from scale clips and shed skins from neonates), or they scored at less than 80% of loci.

Number of alleles, number of effective alleles, and allelic richness were 20.8, 8.0 and 9.5, respectively. Observed and expected heterozygosity were 0.58 and 0.66, respectively. The inbreeding coefficient (FIS) was 017. After corrections for multiple tests (Bonferroni-adjusted significance criterion = 0.00125), microsatellite screening revealed deviations from Hardy- Weinberg equilibrium (HWE) at five of 10 loci. We found no evidence of linkage disequilibrium or null alleles, indicating that deviations from HWE were likely due to biological reasons, and not technical issues.

The probability of paternal exclusion for each locus, when the mother was known, ranged from 0.05 to 0.92, and the combined probability of paternal exclusion for all 10 loci was 1.00. The cumulative probability that a genotype drawn at random would be the same as the genotypes of

full-sibs (Waits et al. 2001) was 0.00025, indicating that the marker set was well suited to robust parentage analysis.

We genotyped 84 offspring, 31 of which were born to 10 known mothers (mean litter size = 3.10 ± 0.35, range = 2-5, M:F = 1:1.07), and identified as such in COLONY 2, and 53 of which were wild-caught (31 neonates and 22 juveniles). In total, 27 offspring were assigned to a putative parent (Table 3). Only 4 of the 31 neonates born to a known mother were assigned to a putative father, 2 of which were littermates assigned to the same male. The other two neonates were each assigned to different fathers. Of the remaining 27 offspring born to known mothers, none were assigned to a putative father.

Of the 53 wild-caught offspring, 10 were assigned to a putative mother, 9 were unassigned to a putative father, and 4 were assigned to a putative parent pair. The remaining 30 offspring were assigned to unknown parents, presumably because the real parents were not included in the sample of candidate parents. In total, only four putative fathers were assigned to 17 offspring, and only 7 putative mothers were assigned to 14 offspring. Only one of these 7 putative mothers was not the known mother of a litter. In no case was a putative father assigned to an entire litter, nor were neonates from the same litter assigned to more than one putative father, providing indirect evidence of multiple paternity. Distances between the initial locations of offspring and mothers, and offspring and fathers, varied greatly (range = 17-2,589 m), but did not differ significantly (df = 25, t = 0.32, p = 0.75).

Spatial autocorrelation analyses revealed significant spatial genetic structure for females up to the 300 m distance class; no distances classes were significant for males (Fig. 2). Analyses comparing adults and offspring (subadult neonates and juveniles) revealed significant spatial genetic structure for adults for the first two distance classes (100 m, 200 m); no distance classes were significant for offspring (Fig. 3).

DISCUSSION

Movement patterns of C. tigris from the Tortolita Mountains are similar to those found during a previous study in the Rincon Mountains (Goode et al. 2008). Males moved farther and maintained larger home ranges than females, especially during the mating season. However, the increase in distance moved did not result in significantly larger mating season home ranges for either sex. Although both sexes moved farther from their winter shelters sites during the mating season, males moved significantly farther than females. These data suggest that males search for mates by moving more, not by expanding their home ranges. This leads us to speculate that if density of receptive females is high, and locations of females are predictable, males have no need to expand their home ranges to find females. An alternative strategy may be to visit as many (relatively nearby) females as possible, perhaps facilitated by prior knowledge of the locations of potential mates. Because females maintain exceedingly small home ranges during the mating season, with little variation in interannual home range configuration (Goode et al. 2008), it seems reasonable to suggest that stationarity (and therefore predictability) in the location of females likely selects for male movement patterns that optimize encounter rates with potential

mates. A related prediction would be an increase in male-male encounters that lead to physical contests over access to females; however, we only observed male-male combat four times in 6,704 tracking occasions (0.0006%). Selection may also favor experienced males that become familiar with the locations of females (Coupe 2002). Alternatively, male snakes may employ systematic search patterns that optimize their ability to locate resources (e.g., random walk, straight-line searching; Duvall & Schuett 1997). And finally, some snakes apparently possess spatial memory of their surroundings (Pittman et al. 2014), which seems reasonable in C. tigris given strong fidelity to not only their home range and winter shelter sites, but to exact locations within their home range, to which they return year after year (Goode et al. 2008, Goode et al. in press). Whatever the strategy, research indicates that this kind of cryptic sociality is likely mediated through chemoreception (e.g., Schwenk 1995, Scott et al. 2013).

Clearly, any number of scenarios can be invoked to explain differences in movement patterns and mate location strategies, but the paucity of empirical data on snakes necessitates some degree of speculation. Furthermore, disentangling cause and effect will likely be challenging. However, data on reproductive interactions, although limited, may shed some light on the feasibility of competing hypotheses. The fact that we observed several females that remained in association with males for extended time periods, suggests that males may obtain successful matings simply by being persistent. If this is the case, a strategy that involves mating with as many females as possible to maximize reproduction may become less important. However, our observations of reproductive interactions also indicate that at least some females are accompanied by different males both within and across mating seasons. We also observed the same male revisiting the same female within and among mating seasons. Because our sample size of reproductive behavior is both small and variable, we are not inclined to make broad conclusions about C. tigris mating tactics, other than to say that multiple strategies may be operating.

Ideally, parentage analysis would provide the kind of information needed to test a variety of predictions related to form and function of the mating system in C. tigris. Although our parentage results provide useful information that could not be obtained from a traditional demographic approach, we are limited by sample size. It is clear from the increasing number of parentage analyses in snakes that the most rigorous studies have involved species from which litters are relatively easy to obtain in the field (e.g., Clark et al. 2014, Levine et al. 2015), or egg clutches that can be brought into the lab and incubated (e.g., Prosser et al. 2002, Blouin-Demers et al. 2005). When a relatively large number of offspring are assigned to parents, it is possible to examine factors leading to mating success (e.g., body size, spatial proximity of parent pairs, and fitness-related traits of offspring) with increased statistical power.

In the case of C. tigris, gravid females retreat into rock outcrops up to several weeks before giving birth, making it all but impossible to obtain neonates from wild-born litters without destroying critical habitats that may result in injury or death to post-parturient females and their young. In addition, it is uncommon to find free-ranging neonate and juvenile C. tigris (5.2% of total captures), and for those that are found, uncertainty in paternity assignment is increased, because the mother is unknown. Indeed, we only found one litter (four neonates) in the field, in

spite of intensive efforts to repeatedly radio track a large number of gravid females during the gestation period. Unfortunately, wild-caught, captive-maintained females either resorbed follicles, produced unfertilized ovules, or gave birth to stillborn or underdeveloped neonates, further hampering our efforts to increase our sample size of offspring. Apparently, gravid C. tigris are difficult to maintain in captivity (pers. comm., J. Jarchow, DVM, Arizona-Sonora Desert Museum).

So what can we learn from paternity analysis? Although we monitored a large number of individuals using radiotelemetry, we only obtained a small number of parentage assignments among radio-tracked animals, and no males found copulating in the field were identified as fathers. This finding is important, because it clearly indicates that mating behavior is a poor predictor of parentage in C. tigris, as has been observed in other snake species (Gibbs & Weatherhead 2001, Prosser et al. 2002, Clark et al. 2014).

Seemingly paradoxical, another important finding was the overall lack of parentage assignments. Because our marker set had high power to accurately assign parentage, based on both the ability of COLONY 2 to identify 100% of known mothers without prior information, and an essentially 100% probability of paternity exclusion, we conclude that our sample of candidate parents did not include the real parents. One interpretation of this finding is that C. tigris occur in high densities at the study site, which decreases the chances of sampling the true parent. In previous studies, we rigorously quantified effort spent searching for C. tigris in both the Tortolita and Rincon Mountains (Goode et al. 2002, Goode et al. 2003, Goode et al. 2015). Relative abundance estimates, based on encounter rates of C. tigris while conducting systematic road cruising surveys, revealed a sharp difference between the two sites. In the Rincon Mountains, it required an average of 39.6 miles to encounter a single C. tigris, compared to 20.1 miles (2002- 2003) and 27.2 miles (2009-2013) in the Tortolita Mountains. The higher index of relative abundance in the Tortolita Mountains is likely due to large amounts of water provided by residents of a low-density development and golf course, which leads to a dramatic increase in productivity (Goode & Parker 2011).

Another potential factor likely leading to low levels of parentage assignment is apparent high rates of immigration from surrounding areas. The study area in the Tortolita Mountains is nearly encircled by large rock outcrops interspersed with relatively flat desertscrub. Because C. tigris is a habitat specialist, primarily confined to rocky foothills of isolated desert mountain ranges, these surrounding areas provide not only suitable, but preferred habitat for C. tigris (Goode et al. 2002). This likely also explains why we continue to capture a large number of unmarked animals 12 years after we started studying C. tigris at this site (Goode et al. 2015).

Examination of distances between initial capture locations of successful parent pairs indicates that some individuals move relatively long distances (>2 km) to mate. Distances of this magnitude indicate that immigrant males from the periphery could easily encounter conspecifics included in the area that includes the sample of genotyped individuals. Furthermore, individuals within the study area obviously move distances that exceed the maximum home range length and maximum distances moved by recaptured snakes. This provides further support for the prediction

that C. tigris likely use multiple mate finding strategies, and it also helps to explain observed high rates of genetic diversity in the population.

We also determined that offspring were commonly found relatively long distances from assigned parents (µ = 892.4 m, S.E. = 120.2 m, range = 17-2,589 m). Given the fidelity of adult females to small home ranges, and relatively short distances between capture-recapture locations, it seems more likely that offspring are exhibiting natal dispersal. This argument cannot be made for distances between offspring and putative fathers. However, results from spatial autocorrelation, and (limited) empirical findings, do not support the prediction that C. tigris will exhibit age- biased dispersal. In contrast, significant positive spatial genetic structure among adult females, at distances up to ~300 m, and the absence of significant spatial genetic structure in males, at all distance lags, supports the presence of adult male-biased dispersal in C. tigris. Given the dramatic differences in movement patterns between males and females, we were not surprised by the existence of fine-scale genetic structure in females. However, results from observations of reproductive behavior and parentage analysis are more ambiguous with regard to predictions about mating system dynamics and reproductive strategies, pointing to the need for further research.

ACKNOWLEDGEMENTS

This study would not have been possible without the hard work and dedication of numerous field assistants, lab technicians, and volunteers. In particular, Kristin Albert, Melissa Amarello, Chip Cochran, Jeff Smith, Mike Wall, and Josh Capps were critical to our success. We thank Rulon Clark, Taylor Edwards, Caren Goldberg, Matt Kaplan, Adrian Munguia-Vega, Karla Pelz- Serrano, and Anna Carlson for advice on genetic techniques and analyses. We are grateful to the staff at the Stone Canyon Golf Club and Rancho Vistoso Partners, especially Dick Maes, for permitting us to conduct research at Stone Canyon, and for providing access to equipment and a trailer to house field personnel. We greatly appreciate the generosity of local residents who provided access to their private land, and showed a keen interest in our research. Mike Rosenzweig, John Koprowski and Cecil Schwalbe provided advice, support, and critical reviews of the manuscript. Our research was supported by grants from Arizona Game and Fish Department Heritage Fund, National Park Service Desert Southwest Cooperative Ecosystems Study Unit, United States Golf Association and National Fish and Wildlife Foundation Wildlife Links Program.

LITERATURE CITED

Adams DK, Comrie AC (1997) The North American monsoon. Bulletin of the American Meteorological Society, 78, 2197-2213.

Arnold SJ, Duvall D. 1994. Animal mating systems: a synthesis based on selection theory. American Naturalist, 143, 317–348.

Beyer, HL (2012) Geospatial Modeling Environment (Version 0.6.0.0). http://www.spatialecology.com/gme.

Blouin-Demers G, Gibbs HL, Weatherhead PJ (2005) Genetic evidence for sexual selection in black ratsnakes, Elaphe obsoleta. Animal Behaviour, 69, 225–234.

Bonnet X, Shine R, Lourdais O (2002) Taxonomic chauvinism. Trends in Ecology & Evolution, 17, 1–3.

Brown WS (1993) Biology, status, and management of the timber rattlesnake (Crotalus horridus): a guide for conservation. Herpetological Circular, 22, 1–78.

Clark RW, Brown WS, Stechert R, Greene, HW (2012) Cryptic sociality in rattlesnakes (Crotalus horridus) detected by kinship analysis. Biology Letters, 8, 523-525.

Clark RW, Schuett GW, Repp RA, Amarello M, Smith CF (2014) Mating systems, reproductive success, and sexual selection in secretive species: a case study of the western diamond-backed rattlesnake, Crotalus atrox. PLoS ONE , 9, e90616.

Clobert J, Nichols JD, Danchin E, Dhondt A (2001) Dispersal. Oxford University Press, Oxford, UK.

Coupe B (2002) Pheromones, search patterns, and old haunts: how do male Timber Rattlesnakes (Crotalus horridus) locate mates. In: Biology of the Vipers, (eds. Schuett GW. Höggren M, Douglas ME, Greene HW). Eagle Mountain Publishing, Eagle Mountain, UT.

Dobson FS (1982) Competition for mates and predominant juvenile male dispersal in mammals. Animal Behaviour, 30, 1183-1192.

Dorcas ME, Willson JD (2011) Innovative methods for studies of snake ecology and conservation. In: Snakes: Ecology and Conservation, (eds. Mullin SJ, Seigel RA), Cornell University Press, Ithaca, NY.

Dubey S, Brown GP, Madsen T, Shine R (2008) Male-biased dispersal in a tropical Australian snake (Stegonotus cucullatus, Colubridae). Molecular Ecology, 17, 3506-3514.

Dubey S, Brown GP, Madsen T, Shine R (2009) Sexual selection favours large body size in males of a tropical snake (Stegonotus cucullatus, Colubridae). Animal Behaviour, 77, 177–182.

Duvall D, King MB, Gutzwiler KJ (1985) Behavioral ecology and ethology of the prairie rattlesnake. National Geographic Research, 1, 80-111.

Duvall D, Schuett GW, Arnold SJ (1993) Ecology and evolution of snake mating systems. In: Snakes: Ecology and Behavior, (eds. Seigel RA, Collins JT), McGraw-Hill, New York, NY.

Duvall D, Schuett GW (1997) Straight-line movement and competitive mate searching in prairie rattlesnakes, Crotalus viridis viridis. Animal Behaviour, 54, 329-334.

Gibbs HL, Weatherhead PJ (2001) Insights into population ecology and sexual selection in snakes through the application of DNA-based genetic markers. Journal of Heredity, 92, 173- 179.

Gillingham JC (1987) Social behavior. In: Snakes: Ecology and Evolution, (eds. Seigel RA, Collins JT, Novak SS) Macmillan, New York, NY.

Glaudas X, Rodríguez-Robles JA (2011) Vagabond males and sedentary females: spatial ecology and mating system of the speckled rattlesnake (Crotalus mitchellii). Biological Journal of the Linnean Society, 103, 681–695.

Goldberg CS, Edwards T, Kaplan ME, Goode M (2003) PCR primers for microsatellite loci in the tiger rattlesnake (Crotalus tigris, Viperidae). Molecular Ecology Notes, 3, 539-541.

Goode M, Wall MD (2002) Tiger rattlesnake ecology and management. Final Report, Nongame and Endangered Wildlife Program, Arizona Game and Fish Department Heritage Fund, Phoenix, AZ, 62 pp.

Goode M, Smith J, Amarello M, Setser K, Favour N (2003) Effects of urban development on herpetofauna. Final Report, Nongame and Endangered Wildlife Program, Arizona Game and Fish Department Heritage Fund, Phoenix, AZ, 73 pp.

Goode M, Smith JJ, Amarello M (2008) Seasonal and annual variation in home range and movements of Tiger Rattlesnakes (Crotalus tigris) in the Sonoran Desert of Arizona. In: The Biology of Rattlesnakes (eds. Hayes, WM, Beaman, KR, Cardwell, MD, and Bush, SP), pp. 327- 334. Loma Linda University Press, Loma Linda, CA.

Goode M, Parker MR, Hauptman J (2015) Land use and wildlife at the urban fringe: integrating ecological research, public perceptions, and conservation planning. Final Report, Nongame and Endangered Wildlife Program, Arizona Game and Fish Department Heritage Fund, Phoenix, AZ, 74 pp.

Goode, M, Parker MR, Feldner MJ, Repp R. (in press) Crotalus tigris species account. In Rattlesnakes of Arizona (eds. Schuett GW, Reiserer RS, Smith CF, Feldner MJ), Eco Press: Rodeo, NM.

Goudet J. (1995) FSTAT : a computer program to calculate F-statistics. Journal of Heredity, 86, 485-486

Greenwood PJ (1980) Mating systems, philopatry and dispersal in birds and mammals. Animal Behaviour, 28, 1140-1162.

Harrison HB, Saenz-Agudelo P, Planes S, Jones GP, Berumen ML (2012) Relative accuracy of three common methods of parentage analysis in natural populations. Molecular Ecology, 22, 1158–1170.

Hofmann S, Fritzsche P, Solhøy T, Dorge T, Miehe G (2012) Evidence of sex-biased dispersal in Thermophis baileyi inferred from microsatellite markers. Herpetologica, 68, 514-522.

Jellen BC, Shepard DB, Dreslik MJ, Phillips CA (2007) Male movement and body size affect mate acquisition in the Eastern Massasauga (Sistrurus catenatus). Journal of Herpetology, 41, 451-457.

Jones OR, Wang J (2010) COLONY: a program for parentage and sibship inference from multilocus genotype data. Molecular Ecology Resources, 10, 551–555.

Keogh JS, Webb JK, Shine R (2007) Spatial genetic analysis and long-term mark–recapture data demonstrate male-biased dispersal in a snake. Biology Letters, 3, 33-35.

Klauber LM (1972) Rattlesnakes: Their Habits, Life Histories, and Influence on Mankind. University of California Press, Berkeley, CA.

Lane A, Shine R (2011) Intraspecific variation in the direction and degree of sex-biased dispersal among sea-snake populations. Molecular Ecology, 20, 1870-1876.

Levine BA, Smith CF, Schuett GW, Douglas MR, Davis MA, Douglas ME (2014) Bateman- Trivers in the 21st century: sexual selection in a North American pitviper. Biological Journal of the Linnean Society, 114, 436-445.

Mabry KE, Shelley EL, Davis KE, Blumstein DT, Van Vuren DH (2013) Social mating system and sex-biased dispersal in mammals and birds: a phylogenetic analysis. PloS ONE, 8, e57980.

Meirmans PG, Van Tienderen PH (2004) GENOTYPE and GENODIVE: two programs for the analysis of genetic diversity of asexual organisms, Molecular Ecology Notes, 4, 792-794.

Meister B, Ursenbacher S, Baur B (2012) Frequency of multiple paternity in the grass snake (Natrix natrix). Amphibia-Reptilia, 33, 308-312.

Munguia-Vega A, Pelz-Serrano K, Goode M, Culver M (2009). Eleven new microsatellite loci for the tiger rattlesnake (Crotalus tigris). Molecular Ecology Resources, 9, 1267-1270.

Parker WS, Plummer MV (1987) Population Ecology. In Snakes: ecology and evolutionary biology (eds. Seigel RA, Collins JT, Novak SS), McGraw-Hill, New York, NY.

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.

Pittman SE, Hart KM, Cherkiss MS, Snow RW, Fujisaki I, Smith BJ, Dorcas ME (2014) Homing of invasive Burmese pythons in South Florida: evidence for map and compass senses in snakes. Biology Letters, 10, 20140040.

Prosser MR, Weatherhead PJ, Gibbs LH, Brown GP (2002) Genetic analysis of the mating system and opportunity for sexual selection in northern water snakes (Nerodia sipedon). Behavioral Ecology, 43, 800–807.

Putnam BJ, Lind C, Taylor EN (2013) Does size matter? Factors influencing the spatial ecology of Northern Pacific rattlesnakes (Crotalus oreganus oreganus) in Central California. Copeia, 2013, 485-492.

Reinert H, Cundall D (1982) An improved surgical implantation method for radio-tracking snakes. Copeia, 1982, 702–705.

Reinert HK (1992) Radiotelemetric studies of pitvipers: data acquisition and analysis. In: Biology of the Pitvipers, (eds. Campbell JA, Brodie ED Jr.), Selva, Tyler, TX.

Rivas JA, Burghardt GM (2005) Snake mating systems, behavior, and evolution: the revisionary implications of recent findings. Journal of Comparative Psychology, 119, 447.

Rousset F (2008) GENEPOP 2007: a complete reimplementation of the GENEPOP software for Windows and Linux. Molecular Ecology Resources, 8, 103-106.

Row JR, Blouin-Demers G, Fox, SF (2006) Kernels are not accurate estimators of home-range size for herpetofauna. Copeia, 2006, 797-802.

Schuett GW, Gillingham, JC (1988) Courtship and mating in the copperhead, Agkistrodon contortrix. Copeia, 1988, 374-381.

Schwenk K (1995) Of tongues and noses: chemoreception in lizards and snakes. Trends in Ecology & Evolution, 10, 7-12.

Scott ML, Whiting MJ, Webb JK, Shine R (2013) Chemosensory discrimination of social cues mediates space use in snakes, Cryptophis nigrescens (Elapidae). Animal Behaviour, 85, 1493- 1500.

Shine R (2003) Reproductive strategies in snakes. Proceedings of the Royal Society of London. Series B: Biological Sciences, 270, 995-1004.

Smith CF, Schuett GW, Amarello M (2015) Male mating success in a North American pitviper: influence of body size, testosterone, and spatial metrics. Biological Journal of the Linnean Society, 115, 185–194.

Smouse, PE, Peakall, R (1999) Spatial autocorrelation analysis of individual multiallele and multilocus genetic structure. Heredity, 82, 561-573.

Turner RM, Brown DE (1994) Tropical-subtropical desertlands. In: Biotic Communities of the American Southwest – United States and Mexico. University of Utah Press, Salt Lake City, UT

Uller T, Olsson M (2008) Multiple paternity in reptiles: patterns and processes. Molecular Ecology, 17, 2566-2580.

Ursenbacher S, Erny C, Fumagalli L (2009) Male reproductive success and multiple paternity in wild, low-density populations of the adder (Vipera berus). Journal of Heredity, 100, 365-370.

Waits LP, Luikart G, Taberlet P (2001) Estimating the probability of identity among genotypes in natural populations: cautions and guidelines. Molecular Ecology, 10, 249-256.

Wang J (2004) Sibship reconstruction from genetic data with typing errors. Genetics, 166, 1963– 1979.

Wang J (2010) Effects of genotyping errors on parentage exclusion analysis. Molecular Ecology, 19, 5061–5078.

Weatherhead PJ, Prosser MR, Gibbs HL, Brown GP (2002) Male reproductive success and sexual selection in northern water snakes determined by microsatellite DNA analysis. Behavioral Ecology, 13, 808–815.

Wusterbarth TL, King RB, Duvall MR, Grayburn WS, Burghardt GM (2010) Phylogenetically widespread multiple paternity in New World natricine snakes. Herpetological Conservation and Biology, 5, 86-93.

FIGURES

Figure 1. Map showing capture locations (N = 460) of C. tigris (2002-2008) from the Stone Canyon study site situated in the foothills of the Tortolita Mountains near Tucson, Arizona, USA.

0.150 r U 0.100 L

0.050

r 0.000

-0.050

-0.100 F M F M F M F M 100 200 300 400 Distance (m) Figure 2. Spatial genetic correlations between adult male and female C. tigris. Significant values, as determined by r-values that lie above the 95% permuted CI (red dashes), and whose 95% bootstrapped CIs (gray bars) do not overlap the x-axis (r = 0), are indicated by filled squares.

0.150 r U 0.100 L

0.050

0.000

-0.050

-0.100 A O A O A O A O 100 200 400 800 Distance (m) Figure 3. Spatial genetic correlations between adult and subadult (O = offspring) C. tigris. Significant values, as determined by r-values that lie above the 95% permuted CI (red dashes), and whose 95% bootstrapped CIs (gray bars) do not overlap the x-axis (r = 0), are indicated by filled squares.

TABLES

Table 1. Summary of total distance moved (TMD), distance moved per day (DIST/DAY), maximum distance moved from winter shelter (DIST WS), home range size (MCP), and length of the major home range axis (MCP AXIS) for 71 radio tracked C. tigris. Pre-mating season corresponds to the time period encompassing egress from their winter shelter sites until the earliest observation of courtship or copulation (July 15). Mating season encompasses the earliest and latest dates we observed courtship or copulation (July 15-September 22). Post- mating season encompasses the latest date that we observed copulation (September 22) until ingress into winter shelter sites. Season Parameter Males Females Pre-Mating TDM 700 ± 97 297 ± 32 MAX DIST WS 324 ± 116 167 ± 40 MCP 3.3 ± 1.1 2.0 ± 0.6 MCP AXIS 338 ± 46 311 ± 76 Mating TDM 2119 ± 169 885 ± 64 MAX DIST WS 713 ± 171 380 ± 48 MCP 4.4 ± 1.3 0.5 ± 0.2 MCP AXIS 456 ± 78 173 ± 30 Post-Mating TDM 739 ± 63 509 ± 162 MAX DIST WS 281 ± 75 260 ± 36 MCP 4.1 ± 0.8 0.6 ± 0.1 MCP AXIS 448 ± 58 196 ± 26

Table 2. Two-way ANOVA results comparing movement parameters between male and female C. tigris listed in Table 1. Parameter Source df Sum of Squares F Ratio Prob > F TDM Season 2 23843551.0 72.1 <0.0001* Sex 1 10844281.0 65.6 <0.0001* DIST/DAY Season 2 7469.9 88.7 <0.0001* Sex 1 2714.2 64.4 <0.0001* MAX DIST WS Season 2 607437.5 6.9 0.0019* Sex 1 216365.8 4.9 0.0299* MCP Season 2 7.8 0.3 0.7722 Sex 1 299.2 19.8 <0.0001* MCP AXIS Season 2 25572.0 0.1 0.8833 Sex 1 1202339.0 11.7 0.0008*

Table 3. Identities of parents for all assigned offspring, and distances between free-ranging offspring and their assigned parents, and putative parent pairs. Offspring(Sex) Mother Father O-M (m) O-F (m) M-F (m) 297F - 857 - 835 - 298F *287 - 17 - - 304F *411 352 1512 1331 234 306M - 393 - 212 - 307M *287 566 140 466 363 313F *792 857 503 1059 1561 347M - 352 - 1329 - 432F **380 393 - - 1664 433F **380 393 - - 1664 449M *792 - 1323 - - 464M - 566 - 1274 - 467M - 566 - 1048 - 542M *792 - 724 - - 547M *287 - 928 - - 548M - 352 1063 - 557F *344 1707 - -

579F - 566 - 806 - 640M **411 352 - - 234 659M **273 566 - - 363 664F *615 566 2451 471 2053 678F *205 - 768 - - 703M - 393 - 1449 - 723F *411 - 2589 - - 737M *792 - 516 - - 894F *261 - 374 - - 896F *287 - 100 - - 914F - 566 - 270 - **denotes known mother, *denotes putative mother