Nabalus Spp.) Facing Global Environmental Change

SUNY College of Environmental Science and Forestry Digital Commons @ ESF

Dissertations and Theses

12-9-2019

Conservation Insights for Endemic Alpine Plants (Nabalus spp.) Facing Global Environmental Change

Follow this and additional works at: https://digitalcommons.esf.edu/etds

Part of the Forest Biology Commons, Genetics Commons, and the Genomics Commons

Recommended Citation Haynes, Kristen, "Conservation Insights for Endemic Alpine Plants (Nabalus spp.) Facing Global Environmental Change" (2019). Dissertations and Theses. 118. https://digitalcommons.esf.edu/etds/118

This Open Access Dissertation is brought to you for free and open access by Digital Commons @ ESF. It has been accepted for inclusion in Dissertations and Theses by an authorized administrator of Digital Commons @ ESF. For more information, please contact [email protected], [email protected]. CONSERVATION INSIGHTS FOR ENDEMIC ALPINE PLANTS (NABALUS SPP.)

FACING GLOBAL ENVIRONMENTAL CHANGE

Kristen R. Haynes

A dissertation submitted in partial fulfillment of the requirements for the Doctor of Philosophy Degree State University of New York College of Environmental Science and Forestry Syracuse, New York December 2019

Department of Environmental and Forest Biology

Approved by: Donald J. Leopold, Major Professor Douglas M. Johnston, Chair, Examining Committee Melissa K. Fierke, Department Chair S. Scott Shannon, Graduate School Dean

ii ACKNOWLEDGEMENTS

It takes a village to raise a child…and support a dissertation to completion. This work would not have been possible without the intellectual contributions, financial assistance, logistical help, and emotional support of so many people. First, I would like to thank my advisor, Donald Leopold, who took me on as a student and gave me the freedom to develop my own research project. Don always supported and encouraged my professional endeavors as well as my academic ones, and always found time to meet, review documents, provide advice, and write letters of reference in the midst of a busy schedule as department chair. Additionally, I would like to thank the members of my committee for their support throughout this project. Sean Robinson, Danilo Fernando, and Jannice Friedman gave critical guidance in molecular methods and also generously provided laboratory space, training, and other resources. Mark Lomolino helped expand my biogeographic thinking and knowledge, and initially encouraged me to enter the PhD program. John Stella provided exceedingly helpful statistical advice and fresh ideas on these chapters. Many individuals provided technical and/or logistical support for this project, including: Abrar (Abbi) Aljiboury, Sarge Boss, Sara Cairns, Paul Casson, Heather Coleman, Terry Ettinger, Julia Goren, Arthur Haines, Jean Hoekwater, Mike Jones, Mike LeBlanc, Paul Logue, Bob Popp, Bill Powell, Matthew Rubin, Kari Seagraves, Dan Sperduto, Lisa St. Hilaire, Stephen Stehman, Bill White, and Steve Young. In particular, I would like to thank Randall Grimshaw and the ESF Computing and Network Services staff who quickly arranged my access to a supercomputer for several months, which enabled me to complete critical analyses for this dissertation. My thanks also to EFB office staff members Sandra Polimino, Faith Ashmore, Joanne Rappleyea, and AnnMarie Clarke for their assistance throughout my degree process. I am deeply indebted to my research and field assistants, whose hard work was so critical to this project’s completion. They include Jared Carpentier, Hannah Kowalsky, Jani Liu, Austin Miller, Siobhan Rubsam, Elysa Smigielski, and Kyle Turchick. My thanks also go to the Adirondack High Peaks Summit Stewards, as well as ESF student volunteers Charlotte Bernhard, Amanda Christiano, Sean Cromwell, Aaron Goodell, and Alexandra Grove, who all provided short-term assistance with data collection and processing tasks. For site access, I am grateful to the Adirondack Mountain Club, Baxter State Park, Beyond Ktaadn, Jarvis Forest Management, the Mount Washington Auto Road (and Howie Wemyss), and the Whiteface Mountain Atmospheric Sciences Research Center, Veterans’ Memorial Highway and Ski Resort. This project certainly could not have been completed without financial support. My sincere thanks to the Edna Bailey Sussman Foundation, the ADKHighpeaks Foundation, the ESF College Foundation/Department of Environmental and Forest Biology (Dr. Samuel Grober ’38 Graduate Fellowship, Edwin H. Ketchledge Scholarship Fund), and the Office of Instruction and Graduate Studies (travel grant) for their funding. I extend additional thanks to Beyond Ktaadn for sponsoring my ADKHighpeaks Foundation application and managing the resulting funds. Additionally, I would like to thank the Department of Environmental and Forest Biology for providing me with funding via a teaching assistantship for four years and Stacy McNulty/the Adirondack Ecological Center for providing a graduate assistantship for one semester. Thank

iii you to Justin Fiene and Melissa Fierke for providing both funding and an incredible opportunity to work in the capacity of Business Manager at Cranberry Lake Biological Station. Finally, I am grateful to Tom Horton for offering me a paid position after I had exhausted my assistantship allotment. Most recently, I would like to thank Kamal Mohamed for supporting my degree completion as I began a new position at SUNY Oswego. Many current and former members of the Leopold Lab and other plant ecology labs at ESF provided important feedback on this project as well as emotional support. In particular, I would like to thank Grete Day, Jessica Cuneo, Toby Liss, James Molloy, Alex Petzke, and Justine Weber for sharing their office and botanical/statistical knowledge with me, and always finding time for a kind word. My thanks also to Margaret Roberts for statistical advice. Many other members of the Ecolunch group—professors and students—provided creative ideas and important critique over the years, for which I am grateful. Finally, I’ve discovered that for the PhD process, as with many other challenging endeavors in life, the mental/emotional challenge is far more significant than the academic/physical challenge. My sincere thanks to many wonderful friends who have provided encouragement along the way–some listed above, and others here: Marissa Cardillo, Daniel Cuneo, Maggie Diu, Kiersten Frenchu, Amanda Gordon, Bridget Hegarty, Debby Hepburn, Janice Lau, Sungsu Lee, Hannah MacLean, Christine Madonia, Aditi Naik, Crystal Ngai, Amanda Pachomski, Maureen Rheinheimer, MJ Sun, and the Alibrandi Catholic Community. My thanks also to my aunts, uncles, grandmother, and other family members for their love and continued interest in my progress. My parents-in-law David Beguin and Cindy Bentley have also been wonderfully supportive and I am grateful especially for the helpful and motivating remote dissertation writing sessions with Cindy. Enormous thanks go to my parents Deborah and Russell Haynes, who first inspired my love of the Adirondacks, and mountains in general. Your unending love, support, encouragement, and care have meant everything. I am here today because of you, and because you always believed in me. My thanks also for your field assistance and transcribing support at critical times during this process! Finally, my deep thanks to my husband, Samouel Beguin. He has been my number one laboratory and field assistant, writing reviewer, and scientific sounding board for years. More important than that, though, has been his love and his untiring belief in me. Thank you for being there during this project every step of the way!

“Somewhere between the bottom of the climb and the summit is the answer to the mystery why we climb.” —Greg Child, Australian writer and mountaineer

iv TABLE OF CONTENTS

LIST OF TABLES ...... vi LIST OF FIGURES ...... vii LIST OF APPENDICES ...... ix ABSTRACT ...... xii CHAPTER 1: INTRODUCTION ...... 1 CHAPTER 2: ASSESSING CLIMATE CHANGE TOLERANCE AND THE NICHE BREADTH-RANGE SIZE HYPOTHESIS IN RARE AND WIDESPREAD RATTLESNAKE- ROOTS ...... 15 INTRODUCTION ...... 16 METHODS ...... 20 RESULTS ...... 29 DISCUSSION ...... 33 TABLES & FIGURES ...... 43 CHAPTER 3: GENOMIC INVESTIGATION OF THE HISTORIC AND FUTURE PERSISTENCE OF OBLIGATE AND FACULTATIVE MOUNTAINTOP PLANT SPECIES ...... 54 INTRODUCTION ...... 55 METHODS ...... 61 RESULTS ...... 72 DISCUSSION ...... 79 TABLES & FIGURES ...... 95 CHAPTER 4: DEFINING EVOLUTIONARY SIGNIFICANT UNITS AND CONSERVATION PRIORITIES IN ALPINE PLANTS (NABALUS SPP.) ENDEMIC TO NORTHEASTERN NORTH AMERICA ...... 106 INTRODUCTION ...... 106 METHODS ...... 112 RESULTS ...... 123 DISCUSSION ...... 129 TABLES & FIGURES ...... 144 CHAPTER 5: SYNTHESIS ...... 158 LITERATURE CITED ...... 168 APPENDICES ...... 201 VITA ...... 229

v LIST OF TABLES

Table 2-1. Functional traits examined, their typical trend with elevation, and their potential importance for climate change response...... 43

Table 2-2. Seed collection sites for Nabalus plants propagated in transplant experiments...... 44

Table 2-3. Model summaries for functional trait data from seedling transplant experiment...... 45

Table 2-4. Phenotypic plasticity index values for functional traits of Nabalus plants at high vs. low elevation (1398 vs. 375 m a.s.l.)...... 46

Table 3-1. Sampling information for genomic datasets of Nabalus boottii (obligate mountaintop) and Nabalus trifoliolatus (facultative mountaintop)...... 95

Table 3-2. Diversity statistics for Nabalus boottii (obligate mountaintop) and Nabalus trifoliolatus (facultative mountaintop)...... 97

Table 3-3. Analysis of molecular variance (AMOVA) results for Nabalus boottii (obligate mountaintop) and Nabalus trifioliolatus (facultative mountaintop) as calculated in program Arlequin using the full dataset for each species...... 98

Table 3-4. Migration rate estimates calculated using the private alleles method in genepop for New York populations of our Nabalus boottii (obligate mountaintop) and Nabalus trifoliolatus (facultative mountaintop)...... 99

Table 4-1. Model summaries for functional trait data from the growth chamber experiment for Nabalus boottii (NB), N. trifoliolatus var. nanus (NN) and non-alpine N. trifoliolatus (NT). ..144

Table 4-2. Factors and loadings from the PCA performed on functional trait data from the growth chamber experiment for Nabalus boottii, N. trifoliolatus var. nanus and non-alpine N. trifoliolatus...... 145

Table 4-3. Private alleles per population for Nabalus boottii (left) and N. trifoliolatus (right)...... 146

Table 4-4. Population size and flowering rate for the 15 sampled populations of Nabalus boottii, comprising nearly all known populations (and all of the largest)...... 147

Table 4-5. Summaries of linear models exploring the relationship between population size/flowering rate and diversity statistics for the fifteen sampled populations of Nabalus boottii...... 148

vi LIST OF FIGURES

Figure 1-1. Botanical illustrations and photographs of (A) Nabalus boottii, (B) non-alpine Nabalus trifoliolatus, and (C) Nabalus trifoliolatus var. nanus...... 13

Figure 1-2. Geographic range maps of (A) Nabalus boottii, (B) non-alpine Nabalus trifoliolatus, and (C) Nabalus trifoliolatus var. nanus...... 14

Figure 2-1. Field sites for the reciprocal transplant experiments on Whiteface Mountain in Wilmington, NY...... 47

Figure 2-5. Comparisons of growth-related functional traits for plants at the conclusion of the seedling transplant experiment...... 51

Figure 2-6. Comparisons of allocation-related functional traits for plants at the conclusion of the seedling transplant experiment...... 52

Figure 2-7. Comparisons of leaf coloration for plants at the conclusion of the seedling transplant experiment...... 53

Figure 3-1. Collection sites for Nabalus boottii (obligate mountaintop) and Nabalus trifoliolatus (facultative mountaintop) samples used in the genomic analysis...... 101

Figure 3-4. Estimated migrant/non-migrant proportions for New York populations of Nabalus boottii (obligate mountaintop; top panel) and Nabalus trifoliolatus (facultative mountaintop; bottom panel) as calculated using BayesAss 3.0.4...... 104

vii Figure 4-1. Comparisons of functional trait measurements for Nabalus boottii (NB), N. trifoliolatus var. nanus (NN) and non-alpine N. trifoliolatus (NT) from the growth chamber experiment...... 149

Figure 4-3. Individual-based UPGMA dendrogram of Nabalus boottii individuals using the full dataset and Nei’s genetic distance...... 152

Figure 4-4. Individual-based UPGMA dendrogram of Nabalus trifoliolatus individuals using the full dataset and Nei’s genetic distance...... 153

Figure 4-5. UPGMA dendrogram of Nabalus boottii (A) and Nabalus trifoliolatus (B) populations using full datasets and Nei’s genetic distance...... 154

Figure 4-6. Summary plots showing individual admixture and population structure in Nabalus trifoliolatus (A) and Nabalus boottii (B-E) based on results from program STRUCTURE...... 155

Figure 4-7. Genetic diversity accumulation curve for populations of Nabalus boottii and Nabalus trifoliolatus...... 157

viii LIST OF APPENDICES

APPENDIX 1: Supplementary information for Chapter 2...... 201

Table A1-1. Model selection criteria used for selecting the best-supported zero-inflated GAMLSS model for establishment percent in the seed transplant experiment...... 201

Table A1-2. Model selection criteria used for selecting the best-supported Cox proportional hazards mixed-effects survival model for seedling transplants...... 202

Table A1-3. Model selection criteria used for selecting the best-supported mixed-effects linear model of height in seedling transplants...... 203

Table A1-4. Model selection criteria used for selecting the best-supported mixed-effects linear model of dry mass in seedling transplants...... 204

Table A1-5. Model selection criteria used for selecting the best-supported mixed-effects linear model of total leaf area in seedling transplants...... 205

Table A1-6. Model selection criteria used for selecting the best-supported mixed-effects linear model of leaf number in seedling transplants...... 206

Table A1-7. Model selection criteria used for selecting the best-supported mixed-effects linear model of root to shoot ratio in seedling transplants...... 207

Table A1-8. Model selection criteria used for selecting the best-supported mixed-effects linear model of specific leaf area in seedling transplants...... 208

Table A1-9. Model selection criteria used for selecting the best-supported mixed-effects linear model of specific root length in seedling transplants...... 209

Table A1-10. Model selection criteria used for selecting the best-supported mixed-effects linear model of leaf dry matter content in seedling transplants...... 210

Table A1-11. Model selection criteria used for selecting the best-supported mixed-effects linear model of red coloration in seedling transplants...... 211

Table A1-12. Model selection criteria used for selecting the best-supported mixed-effects linear model of green coloration in seedling transplants...... 212

Table A1-13. Model selection criteria used for selecting the best-supported mixed-effects linear model of blue coloration in seedling transplants...... 213

Table A1-14. Model selection criteria used for selecting the best-supported mixed-effects linear model of circularity in seedling transplants...... 214

Table A1-15. Model selection criteria used for selecting the best-supported mixed-effects linear model of roundness in seedling transplants...... 215

ix Figure A1-1. Thermocron iButton temperature data recorded every two hours at the low (base), mid, and high (summit) experimental sites on Whiteface Mountain...... 216

APPENDIX 2: Supplementary information for Chapter 3...... 217

Table A2-1. Pairwise FST table for Nabalus boottii populations (Weir & Cockerham, 1984) calculated using R package diveRsity (Keenan, McGinnity, Cross, Crozier, & Prodöhl, 2013)...... 217

Table A2-2. Pairwise GST table for Nabalus boottii populations (Nei & Chesser, 1983) calculated using R package diveRsity (Keenan et al., 2013)...... 218

Table A2-3. Pairwise FST table for Nabalus trifoliolatus populations (Weir & Cockerham, 1984) calculated using R package diveRsity (Keenan, McGinnity, Cross, Crozier, & Prodöhl, 2013)...... 219

Table A2-4. Pairwise GST table for Nabalus trifoliolatus populations (Nei & Chesser, 1983) calculated using R package diveRsity (Keenan et al., 2013)...... 220

Figure A2-1. Markov Chain Monte Carlo (MCMC) trace plots for the best BayesAss run (lowest Bayesian deviance) for New York populations of Nabalus boottii (top panel) and Nabalus trifoliolatus (bottom panel), showing chain convergence...... 221

Script A2-1. Example R script for gbs2ploidy analysis...... 222

APPENDIX 3: Supplementary information for Chapter 4...... 223

Table A3-1. Factors and loadings from the PCA performed on functional trait data from the growth chamber experiment for N. trifoliolatus var. nanus and non-alpine N. trifoliolatus (small subset) only...... 223

Table A3-2. Factors and loadings from the PCA performed on functional trait data from the growth chamber experiment for N. trifoliolatus var. nanus and non-alpine N. trifoliolatus (large subset) only...... 224

Figure A3-1. PCA plots of functional trait data for Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus trifoliolatus (NT) only from the growth chamber experiment for the first seven PCs (small subset)...... 225

Figure A3-2. PCA plots of functional trait data for Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus trifoliolatus (NT) only from the growth chamber experiment for the first seven PCs (small subset); data are identical to those presented in figure A3-1 but are colored here according to population of origin instead of taxon...... 226

Figure A3-3. PCA plots of functional trait data for Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus trifoliolatus (NT) only from the growth chamber experiment for

x the first six PCs, based on data from the large subset: 78 individuals and nine functional traits (factors and loadings provided in Table A3-2)...... 227

xi ABSTRACT

K. R. Haynes. Conservation Insights for Endemic Alpine Plants (Nabalus spp.) Facing Global Environmental Change, 231 pages, 14 tables, 20 figures, 2019. APA style guide used.

Amid the current global biodiversity crisis spurred by anthropogenic environmental changes, determining conservation priorities and the extinction vulnerability of rare taxa are tasks of critical importance. Organisms can avoid environmental change-induced extinction through three possible response modes: evolutionary adaptation, migration (range shift), and tolerance through phenotypic plasticity. In this dissertation, I leveraged transplant experiments and population genomics to assess the ability of rare alpine rattlesnake-roots (Nabalus spp.) to adapt, migrate, and/or tolerate environmental change. I also employed these same techniques to define conservation units and priorities within two endemic alpine taxa (Nabalus boottii and Nabalus trifoliolatus var. nanus) and widespread non-alpine Nabalus trifoliolatus. Finally, I used this study system to investigate more fundamental ecological questions: (1) the niche breadth- range size hypothesis; and (2) the factors contributing to historical persistence of Nabalus taxa in small, isolated mountaintop populations. Overall, my results supported probable resilience to environmental change in alpine Nabalus taxa, which harbor moderate to high levels of genetic diversity (especially N. boottii), show evidence of historic and recent migration among summits, and are highly plastic for several functional traits linked to climate change response. However, alpine Nabalus taxa may suffer from reduced seed recruitment under ongoing climate change, and I therefore recommend continued population monitoring. My results further indicated that both N. boottii and broad- sense N. trifoliolatus should be managed at the species level. I did not find evidence for multiple evolutionary significant units or highly distinct individual populations within N. boottii, and morphological and genomic evidence suggested that alpine N. trifoliolatus var. nanus is not distinct from widespread non-alpine N. trifoliolatus. Regarding the more fundamental ecological questions, I found tentative support for the niche breadth-range size hypothesis in the focal Nabalus spp., but not for phenotypic plasticity as the driving mechanism. Finally, the ability of small populations of Nabalus taxa to maintain genetic diversity (likely via tetraploidy for N. boottii) and migrate between summits helps explain their historical persistence on isolated mountaintops of the northeastern United States.

Keywords: alpine, climate change, common garden, conservation genetics, conservation genomics, endemic, functional traits, global environmental change, Nabalus, niche breadth, northeast, plasticity, plants, population genomics, Prenanthes, RADseq, range size, rarity, transplant experiment, United States

K. R. Haynes Candidate for the degree of Doctor of Philosophy, December 2019 Donald J. Leopold, Ph.D. Department of Environmental and Forest Biology State University of New York College of Environmental Science and Forestry, Syracuse, New York

xii CHAPTER 1: INTRODUCTION

As Earth enters a sixth mass extinction, identifying species most at risk and understanding how best to manage them are tasks of paramount importance (Ceballos et al.,

2015). The drivers of this mass extinction event—collectively termed global environmental change (or simply global change)—are all anthropogenic in nature and primarily include land- use change, climate change, nitrogen deposition, pollution, and species invasions (Barnosky et al., 2011; Ceballos et al., 2015; Sala et al., 2000; Tilman et al., 2017). Individually, each one of these drivers poses a serious threat to biodiversity; collectively, their threat is even greater due to synergies among them and the need for species to simultaneously respond to multiple drivers

(Brook, Sodhi, & Bradshaw, 2008; Sala et al., 2000; Vinebrooke et al., 2004). All told, roughly

25% of earth’s species are currently threatened with extinction (International Union for

Conservation of Nature and Natural Resources, 2019).

Rarity

Rare species face a greater extinction risk than common species (Gaston, 1996; Harnik,

Simpson, & Payne, 2012; Soulé, 1983). This greater risk is due to an increased likelihood of entering the “extinction vortex” (Gilpin & Soulé, 1986), a model describing how reduced gene flow, small population size, and environmental, demographic, and genetic stochasticity interact in a positive feedback loop that drives populations and species extinct. Rare species, which often exhibit small local abundances, narrow environmental preferences, and restricted gene flow

(Gitzendanner & Soltis, 2000; Rabinowitz, 1981), are more susceptible to enter the vortex and suffer from consequent reduced fitness and population size to the point of extinction. Of the

1 seven forms of rarity (narrow geographic range, restricted habitat preferences, small local abundances, and their combinations; Rabinowitz, 1981), rare species characterized by a narrow geographic range (endemics) appear to be the most vulnerable to extinction (Harnik et al., 2012).

Among rare species, endemic species are perhaps the most vulnerable because stressors affecting one particular geographic area (e.g., fires, floods, species invasions) are more likely to affect the entirety of the species than for broad-range species (Harnik et al., 2012). Altogether, rare species, especially those with restricted geographic ranges, merit investigation from a global change conservation perspective due to their heightened risk of extinction. Comparative studies of closely related rare versus common species can also help elucidate more fundamental ecological questions, such as the determinants of rarity/range size and the factors that enable rare species to persist. Addressing these questions is particularly important as scientists attempt to forecast species persistence under global environmental change.

Global change response

Understanding species vulnerability to global environmental change requires an understanding of species’ potential responses. Species can respond to stressors in one of three ways if they are to avoid extinction: (1) migration (long-term range shift) to avoid the stressor,

(2) evolutionary adaptation, or (3) tolerance of the stressor through phenotypic plasticity

(Chevin, Lande, & Mace, 2010; Davis, Shaw, & Etterson, 2005; Jump & Peñuelas, 2005;

McCarty, 2001). This framework was developed for climate change response, but is applicable to other modes of global environmental change.

Population genetics (including population genomics) techniques can be leveraged to investigate migration and adaptive potential, which addresses the first two response modes.

2 Population genomics techniques in particular, with their greater precision for estimates of migration and greater representation of whole-genome diversity provide an important tool for understanding environmental change response (Corlett, 2017; McMahon, Teeling, & Höglund,

2014). Conversely, tolerance via phenotypic plasticity is often determined via common garden

(transplant) experiments in which organisms are raised in different climates and monitored for survival and performance. For plants, plasticity of functional traits may be especially critical for environmental change response (Funk et al., 2017; Nicotra et al., 2011; Valladares et al., 2014).

If experimental and genetic/genomic investigation reveals a species to be vulnerable to global environmental change, identifying conservation units and priorities are critical next steps for conservation. In addition to assessing environmental change response, experimental (e.g., transplant) and population genomic techniques can also be leveraged to define conservation units within species, by assessing population structure and identifying genetically and ecologically unique sets of populations (Allendorf, Luikart, & Aitken, 2013; Crandall, Bininda-Emonds,

Mace, & Wayne, 2000; Fraser & Bernatchez, 2001; Waples, 1991; Waples & Lindley, 2018).

Global change in alpine communities of the northeastern United States

The northeastern United States is faced with a number of environmental changes, especially land-use change, species invasions, nitrogen deposition, and climate change (Capers &

Slack, 2016; Fan, Bradley, & Rawlins, 2015; Galloway, Likens, & Hawley, 1984; Reay,

Dentener, Smith, Grace, & Feely, 2008; Thompson, Carpenter, Cogbill, & Foster, 2013). Two of these changes—nitrogen deposition and climate change—may disproportionately affect high elevation communities of the region, which are collectively termed the northeast alpine zone

3 (Baumgardner, Isil, Lavery, Rogers, & Mohnen, 2003; Capers et al., 2013; Freeman, Scholer,

Ruiz-Gutierrez, & Fitzpatrick, 2018; Marris, 2007; Urban, 2018).

For nitrogen deposition, this disproportionate effect is due to both a greater magnitude of change, and a greater susceptibility to change. Within the northeastern United States, a global hotspot for nitrogen deposition (Galloway et al., 1984; Reay et al., 2008), deposition rates are 6–

20 times higher at high elevation sites (≥ 800 m above sea level [a.s.l.]) due to the orographic increase of cloud formation and precipitation (Baumgardner et al., 2003). Northeast alpine plants, adapted to a nutrient-poor environment, are at risk of being out-competed by faster- growing species adapted to higher nitrogen environments (Bobbink et al., 2010). Certain alpine plants, namely bryophytes, are particularly sensitive to high nitrogen inputs, either through direct toxicity, soil acidification with consequent leaching of nutrients and mobilization of toxins, or greater resulting vulnerability to secondary stressors like disease (Bobbink et al., 2010; Vitousek et al., 1997). Unfortunately, despite emissions regulations in North America, anthropogenic inputs of reactive nitrogen to the atmosphere are projected to increase at least through 2030, with consequent increases in nitrogen deposition (Galloway et al., 1984; Reay et al., 2008). Nitrogen deposition therefore continues to pose a significant threat to the northeast alpine zone (Capers et al., 2013).

In terms of climate change, the northeastern United States as a whole is expected to experience 2.5 to 3.2 ºC of warming by the mid-century, with increased variability (but uncertainty in directionality) for precipitation (Fan et al., 2015). While the magnitude of warming at high elevation in the northeast is currently very similar to warming observed at low elevations (Wason, 2016; but see Seidel et al., 2009), geography renders mountaintop communities highly vulnerable to extirpation under climate change (Costion et al., 2015;

4 Dirnböck, Essl, & Rabitsch, 2011; Elsen & Tingley, 2015; Freeman et al., 2018; Marris, 2007;

Urban, 2018). This is because mountaintop species cannot move to higher elevation tracking their current climate, and may be outcompeted by invading lower elevation species.

While land-use change and species invasions may not be disproportionately affecting the northeast alpine zone in comparison to surrounding lower elevation areas, their effects are nevertheless significant. Recreational overuse (i.e., hiker trampling) is a major problem in the northeast alpine zone (Capers et al., 2013). An ever-increasing stream of annual visitors threatens the fragile alpine vegetation, with some mountains currently attracting over 100,000 visitors per year (Eastman, 2018; Region 5 Office of Natural Resources, 1996). Therefore, despite the fact that most areas of the northeast alpine zone are protected, significant habitat loss and degradation still occurs. Finally, disturbance has also likely contributed to the invasion of dandelion

(Taraxacum officinale) in several regional alpine areas, displacing native species (Capers &

Slack, 2016).

All told, the northeast alpine zone is currently faced with a number of global change- related threats, and species must respond simultaneously to all (Vinebrooke et al., 2004).

Identifying which species within the northeast alpine zone are most vulnerable to extirpation or extinction is critically important. A hotspot of regional biodiversity covering only ~30 km2

(0.01% of the land area) across New York and New England, the northeast alpine zone is home to: (1) arctic-alpine species at the southern limit of their range; (2) endemic species that include alpine plants Potentilla robbinsiana (Robbins’ cinquefoil), Geum peckii (mountain avens),

Nabalus boottii (Boott’s rattlesnake-root), Nabalus trifoliolatus var. nanus (alpine three-leaved rattlesnake-root), and butterfly Boloria chariclea montinus (White Mountain fritillary); (3) boreal/subalpine species; and (4) certain lower-elevation species with broad environmental

5 tolerance (Capers et al., 2013). Among these, endemic species are likely the most vulnerable to extinction (Dirnböck, Dullinger, & Grabherr, 2003). Unfortunately, despite abundant scholarship related to the northeast alpine zone (over 550 works represented in the Appalachian Mountain

Club’s northeast alpine bibliography), the relative dearth of population genetic and transplant studies (apart from Berend, Haynes, & McDonough Mackenzie, 2019; Lindwall, 1999;

Robinson, 2012 and sources therein) mean that scientists and managers know relatively little beyond speculation about the response and vulnerability of northeast alpine species to global environmental change (i.e., their ability to migrate, adapt, and/or tolerate change).

Endemic alpine congeners: Nabalus boottii and Nabalus trifoliolatus var. nanus

Studies involving closely related taxa such as congeners can provide insights into fundamental ecological questions and important context for the study of rare species

(Gitzendanner & Soltis, 2000; Grueber, 2015; Jiménez-Alfaro, García-Calvo, García, & Acebes,

2016). Comparative approaches are especially useful when the two taxa display significant differences in their ecology, range size, and rarity, as is the case for the focal Nabalus taxa of this dissertation. By including alpine Nabalus boottii and Nabalus trifoliolatus var. nanus as well as non-alpine Nabalus trifoliolatus in this study, I was able to answer a number of questions of fundamental and conservation importance (these questions are detailed in the next section). Here,

I review the characteristics of each taxon.

The genus Nabalus Cass., within the family Asteraceae, contains 24 species of herbaceous plants native to North America and Asia. The 14 North American Nabalus species were formerly grouped within the genus Prenanthes L. (Bogler, 2006; Haines, Farnsworth, &

Morrison, 2011), but more recent phylogenetic analysis suggested that Prenanthes was

6 polyphyletic and Nabalus was more accurate for North American species (Kim, Crawford, &

Jansen, 1996). Authors have additionally described morphological and chromosomal differences between Eurasian Prenanthes and North American Nabalus (St. Hilaire, 2003): North American species exhibit a greater number of flowers and phyllaries per flower head, glabrous (versus pubescent) corollas, darker pappus, larger basal versus cauline leaves, and a smaller base chromosome number (n = 8 versus 9) than Eurasian Prenanthes. Previous studies of northeastern

North American Nabalus species include Milstead’s (1964) PhD dissertation, which served as a taxonomic revision of North American Prenanthes, and Sayers’ (1989) master’s thesis on cytology, life history, and morphology of six northeastern North American Prenanthes species.

All North American Nabalus species are rosette-forming plants that reproduce both sexually, as monocarpic perennials with wind-dispersed seed, and clonally, via taproot offshoots (Bogler,

2006; Sayers, 1989). Nabalus inflorescences are composed entirely of ray flowers. Leaf morphology is highly variable within Nabalus, and not diagnostic.

Nabalus boottii DC. (Boott’s rattlesnake-root) is an alpine plant native to the highest elevations (1000–1800 m a.s.l.) of New York, Vermont, New Hampshire, and Maine, where it occurs at fewer than 20 sites, and is thus of high conservation concern (Figure 1-1, 1-2; Bogler,

2006). NatureServe (2018) lists N. boottii as globally imperiled (G2) and critically imperiled

(S1) in all four states in which it occurs; N. boottii is also listed as endangered by the same four states (Maine Natural Areas Program, 2015; New Hampshire Natural Heritage Bureau, 2018;

Vermont Natural Heritage Inventory, 2015; Young, 2019). At one point, N. boottii was a candidate for listing under the Endangered Species Act; however, because populations appear to be stable, a full status review was never conducted (Susi vonOettingen, pers. comm.).

Eventually, the United States Fish and Wildlife Service discontinued the candidate list. Limited

7 (and unpublished) field surveys suggest that populations are relatively stable. White Mountain

National Forest staff have identified the following major threats to this species: (1) disturbance from summit roads and hiker trampling, (2) climate change, (3) acid deposition, (4) hydrologic change, and (5) air pollution (Prout, 2005). However, some intermediate degree of disturbance may actually benefit this species, which often occurs along trails, concrete infrastructure, and formerly grazed areas.

Nabalus boottii is characterized by simple triangular, sagittate, or hastate basal leaves and when flowering, a single flowering stem up to 30 cm tall composed of 10–20 white flower heads that each bear 9–20 ray flowers (Figure 1-1, Bogler, 2006; Haines et al., 2011). The ploidy level of N. boottii is uncertain, with one flow cytometry study determining the species to be diploid

(2n = 16; Sayers, 1989) and another concluding that the species is tetraploid (2n = 32; Löve &

Löve, 1966); however, both studies included only a few samples from one geographic location—

Mount Washington. In terms of habitat, N. boottii occurs in drier sites of snowbank, wet meadow, and streamside communities, as well as exposed and disturbed areas along cliffs, ledges, and trails (Prout, 2005). The species flowers in July and August and fruits in August and

September. N. boottii is characterized by a mixed mating system, and is pollinated by a variety of insects including bees, flies, moths, beetles, and true bugs (Tetreault & Burgess, 2019).

Nabalus trifoliolatus Cass. is widespread throughout eastern North America and occurs from sea level to high elevation (Figure 1-1, 1-2; Bogler, 2006). A habitat generalist, N. trifoliolatus inhabits woodland, cliff, sandy, and saline areas. Defining morphological characteristics of N. trifoliolatus include generally palmately lobed or divided basal leaves with three to five lobes/leaflets, two to seven pale yellow nodding flower heads on a single flowering stem, and campanulate involucres (Bogler, 2006). Like N. boottii, N. trifoliolatus exhibits a

8 mixed mating system, flowering in July and August and fruiting in August and September.

Pollinators include bees, syrphid flies, and ants (Tetreault & Burgess, 2019). Unlike N. boottii,

N. trifoliolatus is well established as diploid (Babcock, Stebbins, & Jenkins, 1937; Jones, 1970;

Löve & Löve, 1966; Powell, Kyhos, & Raven, 1974; Sayers, 1989; Tomb, Chambers, Kyhos,

Powell, & Raven, 1978).

Nabalus trifoliolatus var. nanus (Bigelow) Fernald, the alpine variety of N. trifoliolatus, is the only high elevation congener of N. boottii, and can be found from 1100 to 1600 m a.s.l. in

New York, New Hampshire, and Maine, as well as a few potential locations in Canada (Bogler,

2006; Britton & Brown, 1913). Nabalus trifoliolatus var. nanus is distinguished from non-alpine

Nabalus trifoliolatus based on its habitat (above treeline) and morphological differences, which include more deeply divided leaves, darker involucral bracts, and shorter height. Mature flowering plants at low elevation can be as tall as 150 cm (typically at least 90 cm when in flower), while those at high elevation can be as short as 10 cm (typically 30 cm or less) (Bogler,

2006; Haines et al., 2011; Sayers, 1989). Historically, N. trifoliolatus var. nanus was treated as a separate species (Gleason & Cronquist, 1963; Milstead, 1964; Mitchell & Tucker, 1997; Sayers,

1989).Today, most sources recognize it as a variety or simply part of N. trifoliolatus without distinction based on clines in morphology observed along elevation gradients on Mount Katahdin in central Maine (Bogler, 2006; Haines et al., 2011). Because in my experience morphological intermediates rarely occur and populations almost never span elevational gradients, I conservatively refer to high elevation populations as Nabalus trifoliolatus var. nanus in this dissertation. I use “non-alpine Nabalus trifoliolatus” to refer to non-alpine populations, and reserve the unqualified Nabalus trifoliolatus to refer to the species as a whole.

9 Research questions

Although N. boottii and N. trifoliolatus are relatively well studied, there remain several unanswered questions critical for their conservation. First, how will these taxa respond to rapid global environmental change, especially climate change? Will they be able to migrate away from, adapt to, and/or tolerate ongoing changes? Second, is N. trifoliolatus var. nanus ecologically and genetically distinct from non-alpine N. trifoliolatus; in other words, does it qualify as an evolutionary significant unit (ESU) meriting conservation concern? Finally, how do managers prioritize conservation action within each taxon in light of the threats posed by environmental change?

Dissertation objectives

The overarching goal of my dissertation was to inform the conservation of rare alpine rattlesnake-roots (Nabalus spp.) by determining their vulnerability to global environmental change and identifying conservation units and priorities within each species (Nabalus boottii and

Nabalus trifoliolatus). Specific objectives included: (1) determining the direct fitness effects of an imposed change in climate on these taxa (Chapter 2); (2) assessing each species’ potential to respond to an imposed change in climate via phenotypic plasticity (Chapter 2); (3) evaluating potential for migration and/or evolutionary adaptation as a response to environmental change

(Chapter 3); (4) identifying the presence of evolutionary significant units within each species

(Chapter 4); and (5) recommending priority populations for conservation (Chapter 4).

In addition to these practical objectives, my study system also allowed for investigation of several more theoretical/ecological questions. These included an empirical investigation of the niche breadth-range size hypothesis to explain variation in geographic range size in Chapter 2

10 (Brown, 1984; Slatyer, Hirst, & Sexton, 2013), as well as insights into the historical persistence of small, isolated climate relict populations (Chapter 3).

Synopsis of chapters

I have prepared Chapters 2, 3, and 4 as manuscripts for publication, with intended journal and coauthors listed below for each. Chapter 5 summarizes my conclusions and identifies future directions for research.

Chapter 2: Assessing climate change tolerance and the niche breadth-range size hypothesis in rare and widespread rattlesnake-roots. In this chapter, I explored the effects of an imposed change in climate on early life stages (seeds and seedlings) of Nabalus boottii and

Nabalus trifoliolatus, including effects on fitness and functional trait response. I also evaluated niche breadth in early life stages of Nabalus spp. to test the somewhat controversial niche breadth-range size hypothesis (Brown, 1984; Slatyer et al., 2013). I have prepared this manuscript for publication in Journal of Ecology with Jannice Friedman, Donald Leopold, and

John Stella as coauthors.

Chapter 3: Genomic investigation of the historic and future persistence of obligate and facultative mountaintop plant species. In this chapter, I used population genomic techniques to understand how Nabalus boottii and Nabalus trifoliolatus have persisted in small populations on mountaintops of the northeastern United States, and their outlook for the future under rapid global environmental change. I have prepared this manuscript for publication in American

Journal of Botany with Donald Leopold as coauthor.

Chapter 4: Defining evolutionary significant units and conservation priorities in alpine plants (Nabalus spp.) endemic to northeastern North America. In this chapter, I used a variety of

11 techniques (population genomics, field surveys, and a common garden experiment) to identify conservation units and priority populations in Nabalus boottii and Nabalus trifoliolatus in order to guide their management. I have also synthesized findings of previous chapters to inform conservation recommendations. I have prepared this manuscript for publication in Conservation

Genetics with Donald Leopold as coauthor.

12 A B C

Figure 1-1. Botanical illustrations and photographs of (A) Nabalus boottii, (B) non-alpine Nabalus trifoliolatus, and (C) Nabalus trifoliolatus var. nanus. These images represent mature, flowering individuals. As monocarpic perennials, Nabalus plants spend their first year(s) of life as one or more basal leaves arranged in a rosette, with similar morphology to the largest leaves portrayed in the botanical illustrations. Botanical illustrations courtesy of USDA-NRCS PLANTS database (Britton & Brown, 1913).

Figure 1-2. Geographic range maps of (A) Nabalus boottii, (B) non-alpine Nabalus trifoliolatus, and (C) Nabalus trifoliolatus var. nanus. Light green areas lack county-level data but indicate presence in the state/province. Maps courtesy of USDA-NRCS PLANTS database (USDA & NRCS, 2019).

14 CHAPTER 2. ASSESSING CLIMATE CHANGE TOLERANCE AND THE NICHE BREADTH-RANGE SIZE HYPOTHESIS IN RARE AND WIDESPREAD RATTLESNAKE-ROOTS

ABSTRACT

Broad environmental tolerance is a driver of large geographic range size under the niche breadth-range size hypothesis, and it is also a trait that may confer resilience to organisms faced with climate change when tolerance is driven by phenotypic plasticity. In this study, we used transplant experiments to test environmental tolerance and functional trait plasticity in early life stages of two related species in order to assess their climate change vulnerability and the validity of the niche breadth-range size hypothesis. We had two predictions: (1) broad-range N. trifoliolatus (including its alpine and non-alpine varieties) would maintain equal fitness across all sites/climates, while narrow-range N. boottii would suffer from decreased fitness at warmer, lower elevation sites; and (2) broad-range N. trifoliolatus would display greater functional trait plasticity than narrow-range N. boottii. Overall, we did not find strong, consistent support for either prediction. Rather than a species-level response, we instead found that alpine populations of both species exhibited poor seed establishment at warmer sites, and therefore narrower establishment niche breadth. Additionally, we found substantial but uniform plasticity in functional traits across our focal taxa. Altogether, our findings suggest tentative support for the niche breadth-range size hypothesis contributing to range size differences in N. trifoliolatus versus N. boottii, but do not accord with our predicted mechanism: greater phenotypic plasticity in N. trifoliolatus. Furthermore, our findings suggest that environment of origin (alpine versus non-alpine) rather than species identity or range size determines vulnerability of Nabalus populations to climate change. Plasticity will likely confer some resilience to Nabalus taxa faced

15 with ongoing changes, but reduced seed recruitment could lead to population declines in alpine

Nabalus.

INTRODUCTION

Climate change threatens to drive one sixth of the world’s species extinct (Urban, 2015), and has already caused local and global declines and extinctions (Cahill et al., 2013; Freeman et al., 2018; Panetta, Stanton, & Harte, 2018; Wiens, 2016). Identifying the species most at risk is a conservation priority but also a highly complex issue, necessitating an understanding of Earth’s probable future climate and species’ response (Foden et al., 2019; Pacifici et al., 2015).

Theoretically, mountaintop species are among the most vulnerable to climate change due to the

“escalator effect” (Freeman et al., 2018; Marris, 2007; Urban, 2018): as species shift upslope worldwide (the escalator), mountaintop species are at risk of shifting to extinction as their bioclimatic envelopes disappear off mountaintops and lower elevation species invade and out- compete them. Understanding the vulnerability of mountaintop species, particularly those that are rare and/or endemic, is therefore vitally important for biodiversity conservation in the face of climate change.

Transplant experiments using common gardens are a powerful means of testing the climate change response and vulnerability of species. First, they allow researchers to directly test the capability of organisms to survive and maintain fitness after an imposed change in climate, often achieved by moving organisms along an elevational gradient (Nooten & Hughes, 2017).

Second, transplant experiments enable researchers to investigate organisms’ capacity for phenotypic plasticity. Phenotypic plasticity is an important element of climate change response

(Chevin et al., 2010; Matesanz, Gianoli, & Valladares, 2010; Merilä & Hendry, 2014; Nicotra et

16 al., 2010; Valladares et al., 2014), but it is often ignored in climate change vulnerability assessments (Foden et al., 2019; but see Valladares et al., 2014). When adaptive, plasticity enables organisms to maintain fitness when the environment changes via changes in their morphology, physiology, or behavior (Beaman, White, & Seebacher, 2016). Plasticity serves both as a short-term buffer for species responding to change through adaptation and range shift, and as a primary response mechanism for species lacking the genetic diversity and dispersal capabilities necessary for adaptation and range shift (Chevin et al., 2010; Nicotra et al., 2010).

Indeed, plasticity has already been linked to climate change response in a number of organisms

(Anderson & Gezon, 2015; Henn et al., 2018; Liancourt et al., 2015; Matesanz et al., 2010;

Nussey, Postma, Gienapp, & Visser, 2005; Réale, McAdam, Boutin, & Berteaux, 2003; Vedder,

Bouwhuis, & Sheldon, 2013), and in some cases plasticity alone may be sufficient to ensure persistence in light of climate change (Charmantier et al., 2008; Phillimore, Leech, Pearce-

Higgins, & Hadfield, 2016).

In this study, we used transplant experiments to investigate the climate change response

(survival and phenotypic plasticity) of two plants native to mountaintops of the northeastern

United States: one a mountaintop endemic (Nabalus boottii) and the other a widespread congener

(Nabalus trifoliolatus) that lives from sea level to high elevation, where it is represented by an alpine variety (var. nanus) (Bogler, 2006; Haines et al., 2011). This study system was valuable in two ways. From a conservation perspective, it enabled us to investigate the climate change vulnerability of a globally imperiled (G2) species that represents an important element of the natural heritage of the northeastern US (NatureServe, 2018). From a theoretical perspective, the inclusion of congeners differing widely in range size (N. boottii and N. trifoliolatus sensu lato) added a strong comparative element to our study and allowed us to test hypotheses related to

17 environmental tolerance (niche breadth) and range size (Sexton, Montiel, Shay, Stephens, &

Slatyer, 2017).

According to the niche-breadth range-size (NB-RS) hypothesis, species achieve a large range size by tolerating a broad range of environmental conditions (Brown, 1984; Slatyer et al.,

2013). Broader environmental tolerance, in turn, is achieved through phenotypic plasticity and/or local adaptation (Ackerly, 2003). Therefore, we might expect broader-ranging N. trifoliolatus to display not only greater survival in the face of a change in climate (via its greater niche breadth), but also greater plasticity, which could help facilitate that survival. In general, research has supported the NB-RS hypothesis (Sexton, McIntyre, Angert, & Rice, 2009; Slatyer et al., 2013; but see Cardillo, Dinnage, & McAlister, 2018; Moore, Bagchi, Aiello-Lammens, & Schlichting,

2018). However, the degree to which niche breadth (and therefore range size) is determined by phenotypic plasticity (broad environmental tolerance in all individuals) versus local adaptation

(narrow tolerance in individuals/populations, broad tolerance in the species) remains largely unknown and merits further empirical study (Lacher & Schwartz, 2016; Sexton et al., 2017;

Sheth & Angert, 2014; Slatyer et al., 2013). Our study, therefore, contributes not only to understanding the climate change vulnerability of our focal taxa, but also serves as an empirical investigation of the niche breadth-range size hypothesis and its underlying mechanisms.

To investigate the direct fitness effects of a change in climate on N. boottii and N. trifoliolatus, we evaluated the survival of seedlings and the establishment of seeds transplanted along an elevational gradient spanning current to predicted end-of-century climate conditions for northeastern mountaintops (Fan et al., 2015). We focused on early life stages given their demographic importance for the persistence of populations and their vulnerability to environmental change (Angert & Schemske, 2005; Dalgleish, Koons, & Adler, 2010; Donohue,

18 Rubio de Casas, Burghardt, Kovach, & Willis, 2010; Hirst, Griffin, Sexton, & Hoffmann, 2017;

Kim & Donohue, 2013; Walck, Hidayati, Dixon, Thompson, & Poschlod, 2011). Based on the

NB-RS hypothesis and the assumption of phenotypic plasticity as the driving force, we made the following prediction:

(1) Broad-range N. trifoliolatus (including both non-alpine N. trifoliolatus and alpine var.

nanus) will maintain equal fitness (i.e., seedling survival and seed establishment) across

all sites/climates, while narrow-range N. boottii will suffer from decreased fitness at

warmer, lower elevation sites.

To investigate phenotypic plasticity as a mediator of climate change response in N. boottii and N. trifoliolatus, we evaluated plasticity in plant functional traits for seedlings in the aforementioned transplant experiment. Functional traits provide insight into biological and ecological processes from the scale of individual organisms to ecosystems; in light of climate change, functional traits can help elucidate probable responses of species and their cascading effects on ecosystem function (Funk et al., 2017).

Our focal traits for this study aligned with those of high priority from a climate change perspective presented in Nicotra et al. (2010). We selected traits associated with growth, life history strategy, photosynthetic capacity, resource allocation, and resistance to drought/freezing

(Table 2-1). The ability of a plant to maintain growth and photosynthesis in the face of changing weather patterns, retain its competitive position in a changing biotic community, allocate carbon in optimal ways given changing conditions, and survive extreme weather events is crucial for long term persistence in the face of climate change. The selected traits included height, total dry mass, total leaf area, and leaf number (single-measure “growth” traits); root to shoot ratio, specific leaf area, specific root length, leaf dry matter content (derived “allocation” traits); and

19 leaf pigmentation and leaf shape (single-measure leaf form traits) (Nicotra et al., 2010; Pérez-

Harguindeguy et al., 2013). Although we did not specifically test for adaptive plasticity, we assumed that a shift in average trait value toward values typical of a given elevation (i.e., taller plants at lower elevation, Table 2-1) was consistent with an adaptive response and likely to help these plants survive climate change (Enquist et al., 2015; Henn et al., 2018).

Based on the NB-RS hypothesis and the assumption of phenotypic plasticity as the driving force, we made the following prediction:

(2) Broad-range N. trifoliolatus (including both non-alpine N. trifoliolatus and alpine var.

nanus) will display greater functional trait plasticity than narrow-range N. boottii.

If supported, our two predictions would indicate that mountaintop endemic N. boottii is more vulnerable to climate change than broad-range N. trifoliolatus, and may merit conservation action to ensure persistence. Additionally, our predictions would indicate that the NB-RS hypothesis, as mediated by phenotypic plasticity, contributes to the difference in geographic range size and extent between these two species.

METHODS

Study taxa

The genus Nabalus (Syn: Prenanthes), within the family Asteraceae, contains 24 species native to North America and Asia, all rosette-forming monocarpic perennials. Nabalus boottii

(DC.) (Boott’s rattlesnake-root) is an alpine plant endemic to the highest elevations (1000–1800 m a.s.l.) of New York, Vermont, New Hampshire, and Maine, where it occurs in fewer than 20 sites and is thus of high conservation concern (Bogler, 2006). Nabalus trifoliolatus (Cass.) is widespread throughout eastern North America and occurs from sea level to high elevation

20 (Bogler, 2006). A generalist, N. trifoliolatus inhabits woodland, cliff, sandy, and saline habitats.

Nabalus trifoliolatus var. nanus (Bigelow), the alpine variety of N. trifoliolatus, is the only high elevation congener of N. boottii and can be found from 1100 to 1600 m a.s.l. in New York, New

Hampshire, and Maine. Historically, N. trifoliolatus var. nanus was treated as a separate species

(Gleason & Cronquist, 1963; Milstead, 1964; Mitchell & Tucker, 1997; Sayers, 1989), although most sources today recognize it as a variety or simply part of N. trifoliolatus without distinction

(Bogler, 2006; Haines et al., 2011). Because there are very few examples of intermediate subalpine/montane populations of N. trifoliolatus, we conservatively distinguish the two varieties to avoid masking potential differences between them. Here, we refer to the two varieties as

“Nabalus trifoliolatus var. nanus” (abbreviated NN in figures and tables) and “non-alpine

Nabalus trifoliolatus” (abbreviated NT) while we reserve the unqualified “Nabalus trifoliolatus” to refer to the species group as a whole. To distinguish N. trifoliolatus var. nanus, we used the following criteria: height ≤ 30 cm, involucral bracts dark (almost black), basal leaves deeply dissected, and occurrence above treeline.

Seed collection & storage

We collected seed of N. boottii from one site on Whiteface Mountain (Wilmington, New

York), seed of N. trifoliolatus var. nanus from two sites on Mt. Washington (Sargent’s Purchase,

New Hampshire), and seed of non-alpine N. trifoliolatus from two sites in Maine (Canton and

Topsham), all during August and September 2015 (Table 2-2). Our sampling was constrained by permitting and the difficulty of finding non-alpine N. trifoliolatus populations in close proximity to the alpine sites. At each site, we selected 15 to 50 widely spaced individuals and harvested up to 50% of each plant’s single flowering stalk, storing each flowering stalk (i.e., seed family) in

21 an individual envelope. In total, we sampled ≥ 50 seed families per taxon. We stored all seeds at room temperature in silica gel.

In March 2016, we cleaned the seeds and divided them into two batches. One batch we immediately prepared for cold, moist stratification by placing the seeds from each family in sealed plastic bags filled with a moist potting medium of 50% perlite and 50% vermiculite. We covered the bags in black plastic to eliminate light and stored them at 4º C until May 2017, when we used them for the seedling transplant experiment to test predictions one (fitness) and two

(plasticity). We note that stratifying for this length of time is not typically recommended for similar species in Asteraceae (Baskin & Baskin, 2014), but was necessary due to a permitting delay.

In April 2017, we prepared the remaining batch of dry seeds for the seed transplant experiment. We created nine replicate multi-family seed sets for each sampled population (one population for N. boottii and two each for N. trifoliolatus var. nanus and non-alpine N. trifoliolatus). Seed number per set varied by population, from 12 to 29 seeds per set. We stratified these seeds in an identical manner to the first batch, apart from the substitution of peat- based potting mix (Sunshine ® Redi-Earth; Sun Gro Horticulture, Agawam, MA) instead of perlite. We hereafter refer to this 50/50 mix of Redi-Earth and vermiculite as our standard potting mix. In July 2017, we removed these second batch seeds from stratification for use in our seed transplant experiment to test prediction one (fitness).

Experimental site

In June 2017, we placed three raised beds on Whiteface Mountain (Wilmington, NY) at sites of increasing elevation: 375 m a.s.l. (base/low elevation), 887 m a.s.l. (mid elevation) and

22 1398 m a.s.l. (summit/high elevation) (Figure 1). We selected these sites to achieve a ~4 ºC increase in average temperature between the highest and lowest beds, consistent with projected end-of-century warming for this region (Fan et al., 2015). We filled the raised beds with a 2:1 mixture of Redi-Earth potting mix and vermiculite, topped with a layer of mulch. We mixed

~100 g of Soil Moist crystals (JRM Chemical, Cleveland, OH) into each bed to increase water retention, stretched bird netting (1.5 cm mesh) over the top of the beds to deter herbivory, and attached landscape fabric to the bottom of bed frames to prevent soil loss. We attached a rain gauge and two iButton ® thermocrons to each bed to record rainfall and air temperature (Maxim

Integrated, San Jose, CA). We watered each bed with the equivalent of 1.25 cm of water once per week if there was less than 1.25 cm of precipitation the previous week as determined by the rain gauges (average rainfall for the area is 2.5 cm per week).

We were concerned that the sites differed in their shading, so we erected a 1.2 m shading structure at the low elevation site to mimic the southwestern shading provided by a rock face at the high elevation site and a large boulder at the mid elevation site. We included a quantitative indicator of shading/cardinal direction as a covariate in data analyses (see below) to further account for any shading differences plants may have experienced. We removed all materials from Whiteface Mountain at the conclusion of the experiment on 2 September 2017.

Seed transplant

On 15 July 2017, we removed second batch seeds from stratification. We randomly assigned three of the nine replicate seed sets for each population to a site (low, mid, high elevation) and a pot location within each site. We filled each 7.9 cm2 pot with our standard potting mix, added a seed set, and sunk the pot into the raised beds. We stretched mesh nylon

23 fabric over the top of pots to allow light penetration while preventing seeds from moving in or out of pots. We allowed seeds to germinate and grow without interference apart from the aforementioned watering. At the conclusion of the experiment on 2 September 2017, we divided the number of living plants by the number of seeds originally sown in each pot (12 to 29; varied by population) to determine the proportion that successfully established, a proxy for fitness

(prediction one).

Seedling transplant

On 27 May 2017, we removed first batch seeds from stratification and germinated each sample indoors under humidity domes. Within one to two days of germination, we transplanted seedlings into individual 6.4 cm2 pots containing our standard potting mix, and moved them to a shaded outdoor location, randomizing tray position every few days. As first true leaves appeared, we provided seedlings greater exposure to direct sun and rain by moving them to a less sheltered location. At the point when seedlings’ first true leaves were expanding, we transported them to

Whiteface Mountain for transplantation, which occurred on 3 July 2017. Altogether, we transplanted 178 seedlings representing all taxa: 40 of N. boottii, 48 of N. trifoliolatus var. nanus, and 90 of non-alpine N. trifoliolatus. We randomized seedlings into transplantation site (low versus high elevation) and pot position within each site. We excluded the mid elevation site in this experiment to maintain adequate sample sizes for each treatment. In cases where we had multiple surviving seedlings from the same seed family, we randomly divided these half-siblings between the two sites. To allow for unrestricted root growth and water movement, we scored the sides of pots and removed bottoms before sinking pots into the raised beds.

24 We recorded survival (a proxy for fitness; prediction one) five times over the two-month growing season, scoring plants as alive or dead. Several plants lost their aboveground foliage after transplant and were erroneously scored as dead; we subsequently amended these to “alive” if the plants sent out new foliage on subsequent visits. On 2 September 2017, we removed all pots from Whiteface and transported them to the laboratory. We carefully uprooted plants from pots, removed excess soil, and allowed them to imbibe distilled water overnight in dark, refrigerated conditions to achieve maximal leaf expansion (Pérez-Harguindeguy et al., 2013). We then used a flatbed scanner to take whole-plant images at 300 dpi, with a color and size standard included in each image (see Figure 2-2 for example images). Afterward, we divided plants into their above and belowground portions and measured their wet mass (g) with an electronic balance. We also recorded the number of leaves for each individual, scoring young unexpanded leaves (< 2 mm long) as 0.5. We then placed plant specimens in paper envelopes and dried them for 72 hours at 65 ºC before recording dry mass (mg).

We used the software ImageJ to measure additional traits from whole-plant scans

(Schneider, Rasband, & Eliceiri, 2012). These traits included height (length of longest leaf in the rosette; mm), taproot length (mm), whole-rosette leaf area (mm2), leaf shape, and leaf coloration.

We used the “shape descriptors” option in ImageJ to take leaf shape measurements for each individual leaf. Roundness, one shape descriptor, is a standardized value inversely related to the aspect ratio; it is calculated according to the formula: 4(#$%&)()*(+&,-$ &/01))2. Elongated leaves have lower roundness values than circular leaves. “Circularity” is a standardized value that compares area to perimeter according to the formula 4((#$%&)(3%$04%5%$))2. Leaves with a higher perimeter value for their size, perhaps due to a sinuous, toothed, or otherwise irregular margin, have lower circularity. After recording roundness and circularity for each fully expanded

25 leaf present in our images, we averaged values across the leaves of each individual plant. We measured leaf coloration using the ImageJ plugin RGBmeasure, which divides and quantifies the red, green and blue color channels of all or a portion of an image (Rasband, 2004). For each whole-plant image, we delimited the boundaries of all leaves, and from this highlighted area we calculated an average value for red, green and blue coloration. Coloration values ranged from 0 to 255 for each color channel.

We calculated certain functional traits (“allocation” traits) using combinations of length, area and mass measurements. We calculated root to shoot ratio by dividing root dry mass (mg) by shoot dry mass (mg) (Pérez-Harguindeguy et al., 2013). We calculated specific root length by dividing taproot length (mm) by taproot dry weight (mg). To calculate leaf dry matter content

(whole rosette), we divided rosette dry mass (mg) by rosette wet mass (g). We calculated specific leaf area (whole rosette) by dividing rosette leaf area (mm2) by rosette dry mass (mg). We note that rosette dry mass included petioles, whereas rosette leaf area did not. These measures therefore correlate with true specific leaf area, but should not be used for comparison with other studies.

In sum, we recorded measurement for the following traits: height, dry mass, total leaf area and leaf number (growth-related traits); root to shoot ratio, specific leaf area, specific root length and leaf dry matter content (allocation-related traits); and red/green/blue leaf coloration, leaf roundness, and leaf circularity (leaf form-related traits) (Table 2-1).

Statistical analysis

We compared establishment success from seed (as proxy for fitness, prediction one) across sites and taxa using likelihood-based model selection criteria (i.e., AICc) of GAMLSS

26 models (generalized additive models for location scale and shape), a method which extends generalized linear models (GLMs) and generalized additive models (GAMs) to include additional distributions of the response variable. Location (mean), scale, and shape refer to parameters of the response variable distribution. Using package “gamlss” in program R version

3.4.3 (R core team, 2017; Rigby & Stasinopoulos, 2005), we built GAMLSS models with a zero- inflated beta distribution to account for the high number of zeros in the dataset, which were generated through two processes: failure to germinate, or failure to survive once germinated. Our competing GAMLSS models included a global interaction model of site and taxon, an additive model including site and taxon, site- and taxon-only models, and a null model. Because shading was uniform across the seed transplant experiment, we did not include a shading covariate in these models. We calculated pairwise differences in estimated marginal means among taxa and sites using the Tukey method for P-value adjustment in package emmeans (Lenth, 2018).

To analyze differences in survival over time for the seedling transplant experiment (as another fitness proxy; prediction one), we built Cox proportional hazards mixed models using package “coxme” in program R version 3.4.3 (R core team, 2017; Therneau, 2018). We included seed family as a random effect in all models to account for maternal relatedness among individuals in the experiment. Competing models included the additive and interactive effects of site, taxon, and shading as well as a null model. We used AIC model selection to choose the best supported model within the candidate model set (Bolker & R Development Core Team, 2017).

We assessed plasticity (prediction two) for each functional trait by building general linear mixed models to compare values among sites and taxa using the R package lme4 (Bates,

Mächler, Bolker, & Walker, 2015). We used package lmerTest to calculate P-values using

Satterthwaite’s approximation for denominator degrees of freedom (Kuznetsova, Brockhoff, &

27 Christensen, 2017); this method provides a reduction in Type I error rates as compared to likelihood ratio test methods (Luke, 2017). Based on inspection of diagnostic plots (normal Q-Q plots and residual versus fitted plots), we log-transformed data for some traits prior to analysis to improve normality of residuals and homogeneity of variance. We included seed family as a random effect in all models to account for maternal relatedness among individuals in the experiment. For each trait, models included the additive and interactive effects of site, taxon, and shading as well as a null model. We used AICc model selection to choose the best supported model (Bolker & R Development Core Team, 2017), and package emmeans to calculate pairwise differences in estimated marginal means among taxa and sites using the Tukey method for P- value adjustment (Lenth, 2018).

To facilitate the comparison of plasticity across different traits, we calculated phenotypic plasticity index values. Plasticity index values, in contrast to reaction norms, provide a standardized measure of phenotypic difference across environments (Valladares, Sanchez-

Gomez, & Zavala, 2006). For each taxon, we calculated the median plasticity index (PImd) as the absolute difference in median trait value at high versus low elevation divided by the larger median trait value (Valladares et al., 2006). We used median (versus mean) trait values due to the non-normality exhibited by data for several traits. However, for one trait (leaf number), we used mean values instead as they provided a better representation of the central tendency of the data, which were comprised of small whole and half-integer values. We compared PImd values among taxa using a paired Wilcoxon signed-rank test (Gugger, Kesselring, Stöcklin, & Hamann,

2015; Hamann, Kesselring, & Stöcklin, 2018).

We calculated one additional plasticity index—an environmentally standardized index— in order to facilitate future cross-study comparison. Scaling plasticity index values according to

28 the magnitude of the study’s environmental gradient allows researchers to compare plasticity across studies that employ different gradients. For example, this would allow comparison of plasticity among plants in one study subjected to a 600 m elevational gradient and those in another with a 1000 m gradient. The index we used is an environmentally standardized version of PImd, similar to ESPI from Valladares et al., 2006 (termed here ESPImd). We calculated

ESPImd values by dividing our PImd values by 4.9º C, the difference in mean growing season temperature between the high and low elevation sites.

RESULTS

Site environmental conditions

From 3 July 2017 until 2 September 2017, average site temperature was 19.1 ºC (SE

0.25) at the base/low elevation site, 16.4 ºC (SE 0.24) at the mid elevation site, and 14.1 ºC (SE

0.22) at the summit/high elevation site. The average temperature gradient between the low and high elevation sites was 4.9 ºC. Sites received equal rainfall; however, surface-level cloud cover

(i.e., fog) increased with elevation.

Seed transplant

We measured establishment from seed as one proxy for fitness in these species

(prediction one). Broad-range non-alpine Nabalus trifoliolatus successfully established at the base, mid, and summit sites on Whiteface Mountain (1.2%, 1.8%, and 22.3% establishment respectively), maintaining some degree of fitness across all sites/climates. Narrow-range Nabalus boottii and N. trifoliolatus var. nanus (the alpine variety of broad-range N. trifoliolatus), conversely, only established at the summit location (3.4% and 0.8% establishment respectively),

29 suffering from decreased fitness (0% establishment) at warmer lower elevation sites.

Unexpectedly, all three taxa established most successfully at the high elevation/summit location

(OR = 0.23, SE = 0.11, P < 0.01) based on our best-supported GAMLSS model (an additive model including site and taxon; N = 45, d.f. = 7), although this difference was not significant for the low elevation/base location. The GAMLSS model also indicated higher establishment in non- alpine N. trifoliolatus versus N. boottii (OR = 0.21, SE = 0.11, P < 0.01; Figure 2-3); however, we suggest caution in cross-taxon comparisons of absolute establishment rates due to potential viability differences between taxa we discovered in laboratory germination trials. We note the null model as a competing model in this analysis with ΔAIC =1.0 (appendix 1).

Seedling transplant

Survival

We monitored survival of transplanted seedlings as an additional measure of fitness

(prediction one). Overall, we found that all taxa maintained some degree of fitness across all sites/climates, although survival was generally low by the end of the growing season (~30% on average, Figure 2-4). Survival did not differ significantly among taxa according to our best- supported Cox proportional hazards model (an interaction model including site and shading with

N = 178 and d.f. = 3.05 and no competing models ΔAIC < 2). Unexpectedly, all seedlings survived best at the warmer low elevation site, exhibiting an 84% increased chance of survival over time as compared with plants at the high elevation site (OR = 0.16, β = -1.81, SE = 0.59, P

< 0.01). Altogether, 24% of seedlings (22 of 90) survived the full extent of the growing season at high elevation, while 35% (31 of 88) survived at low elevation. At low elevation, shading also

30 affected survival: for each unit of distance away from the most shaded corner of the bed, survival decreased 16% across taxa (OR = 1.16, β = 0.14, SE = 0.05, P < 0.01).

Functional trait plasticity

We compared plasticity in functional traits of transplanted seedlings to investigate plasticity as a climate change response mechanism in narrow-range Nabalus boottii and broad- range N. trifoliolatus var. nanus and non-alpine N. trifoliolatus (prediction two). Overall, we found substantial but equal plasticity in our three focal taxa (P > 0.05, V > 23.5) for paired

Wilcoxon signed-rank tests), with average PImd values across traits for each taxon ranging from

0.36 to 0.42, indicating a 36 to 42% change in median trait value between plants growing at low versus high elevation (Table 2-4). These differences translate into a 7–9% change in median trait value per ºC difference in average growing season temperature over our 4.9 ºC temperature gradient. Among traits, however, plasticity varied substantially, with average PImd values across taxa ranging from 0.03 for roundness to 0.80 for total leaf area. On average, single measure growth-related traits showed the greatest degree of plasticity with an average PImd value of 0.66 across taxa, derived allocation-related traits showing an intermediate degree of plasticity with an average PImd of 0.41, and single measure leaf form-related traits showing little plasticity with an average PImd of 0.16 (Table 2-4).

On a trait by trait basis, we found significant plasticity (indicated by a significant effect of site in best models following AICc selection) in the following traits: height, dry mass, total leaf area, leaf number (growth; Figure 2-5); specific leaf area, specific root length, leaf dry matter content (allocation; Figure 2-6); and green coloration (leaf form; Figure 2-7) (Table 2-3).

We investigated the direction of significant trait differences between low and high elevation in

31 order to see if plastic changes were plausibly adaptive (i.e., corresponded with naturally observed trends in morphology with elevation, Table 2-1). We found greater growth at low elevation across all growth-related traits, using log-transformed values (Figure 2-5; height [β = 1.26, SE =

0.17, P < 0.001], dry mass [β = 1.42, SE = 0.24, P < 0.001], total leaf area [β = 2.12, SE = 0.34,

P < 0.001], leaf number [β = 0.34, SE = 0.14, P = 0.017]). For our allocation-related traits, plants at low elevation exhibited higher specific leaf area (β = 12.71, SE = 3.0, P < 0.001), lower log specific root length (β = -1.50, SE = 0.19, P < 0.001) and lower log leaf dry matter content (β = -

0.30, SE = 0.09, P = 0.003) than plants at high elevation (Figure 2-6). Finally, for our leaf form traits, only green coloration showed significant plasticity, with low elevation plants exhibiting significantly greater green coloration than high elevation plants (β = 43.53, SE = 7.41, P < 0.001;

Figure 2-7). These significant differences were consistent with naturally observed trends in morphology along elevational gradients (Table 2-1), which we address further in the discussion.

The remaining traits (root to shoot ratio, roundness, circularity, red coloration, and blue coloration) did not vary between high and low elevation transplants.

Unrelated to our prediction, some traits differed significantly (height, leaf number, root to shoot ratio, specific root length) or marginally (blue coloration) among taxa. Most often traits differed significantly between N. boottii and one of the varieties of N. trifoliolatus; however, we did find a significant difference in specific root length between N. trifoliolatus var. nanus and non-alpine N. trifoliolatus. The remainder of traits were uniform across taxa (Table 2-3).

Shading was not significant in competing models for any trait and was thus excluded from Table 2-3. Interaction terms were also not significant for any traits. For all traits in which a non-null model was best supported, ΔAICc between the best and null models was ≥ 6, apart from blue coloration, for which ΔAICc = 0.1 (appendix 1).

32 DISCUSSION

We used seed and seedling transplant experiments to investigate the climate change vulnerability of rare and widespread mountaintop plants in the northeastern United States: narrow-range Nabalus boottii and broad-range Nabalus trifoliolatus. We investigated both the direct fitness effects of an imposed change in climate (prediction one) and the potential for these taxa to respond to change via phenotypic plasticity in functional traits (prediction two). Overall, we did not find strong, consistent support for either prediction. An imposed change in climate did influence seed establishment and seedling survival, but not always negatively and not according to range size, as we predicted (prediction one). With regard to our second prediction, we found substantial but uniform plasticity in functional traits across our focal taxa. Below, we discuss these results further, including their implications for climate change vulnerability and the niche-breadth range-size hypothesis.

Climate change vulnerability

Fitness

The results of our seed and seedling transplant experiments revealed that environment of origin (i.e., alpine versus non-alpine) was a stronger predictor of fitness (prediction one) across sites of varying climate than species identity or range size. Alpine Nabalus trifoliolatus var. nanus’s seed establishment and seedling survival rates closely mirrored those of alpine N. boottii, rather than those of conspecific non-alpine N. trifoliolatus. Seeds of both alpine taxa suffered from reduced fitness when transplanted to warmer sites: they exhibited 0% establishment success at mid and low elevation. Even accounting for zero-inflation in our dataset

(due to possible low seed viability), we still found estimates of establishment success

33 approaching 0% in both alpine taxa at lower elevation sites, although only mid- and high- elevation establishment rates were significantly different. In our seedling transplant experiment, survival of alpine N. trifoliolatus var. nanus again closely followed that of N. boottii: the survival rate of both taxa was ~9% higher than in non-alpine N. trifoliolatus, although this difference was not significant. For seedlings, however, transplantation to a warmer site did not negatively affect fitness: survival of all taxa was actually higher at low elevation.

Our results have several implications for the climate change response of these taxa. First, they indicate a probable direct negative effect of climate change on alpine Nabalus populations, via decreased recruitment from seeds. This finding is significant, as studies have emphasized that indirect rather than direct effects of climate change will pose a greater threat for most species

(Cahill et al., 2013; Ockendon et al., 2014). Germination and early establishment, however, are life stages highly dependent on climate: most seeds require high and consistent humidity and relatively warm temperatures to germinate and establish, and are sensitive to change or variability of environmental factors (Baskin & Baskin, 2014; Finch et al., 2019; Walck et al.,

2011). These early stages are also vital to the demographic persistence of plant populations

(Baskin & Baskin, 2014; Donohue, 2005; Walck et al., 2011). The increasing volatility of our global climate, including greater variability in rainfall and temperature, may render seeds (and therefore plants) highly vulnerable to the direct effects of climate change (Walck et al., 2011).

Indeed, theoretical and empirical work suggest that germination is the most vulnerable life stage of plants to changes in climate (Dalgleish et al., 2010; Fay & Schultz, 2009; Lloret, Penuelas, &

Estiarte, 2004; Walck et al., 2011), affecting both the timing and success of germination––either of which could negatively impact plant populations and lead to declines.

34 Conversely, post-germination, seedlings appear resilient to climate change in alpine

Nabalus taxa, as evidenced the increased survival of seedling transplants under end-of-century predicted conditions at the low elevation (4.9 ºC warmer) site versus their current climate conditions at the high elevation site. Hirst et al. (2017) found similar performance differences across life stages in alpine daisies (Brachyscome spp.), with seedlings resilient to transplantation to warmer sites while seeds experienced declines in germination. Hirst et al.’s (2017) and our findings are in accordance with others’ assertions that germination is the most vulnerable life stage of plants to climate change, and exhibits the narrowest niche breadth (Dalgleish et al.,

2010; Finch et al., 2019; Walck et al., 2011).

Finch et al. (2019) argue that the vulnerability of a plant species is linked to the most vulnerable life stage (i.e., with the narrowest niche); however, the flexible reproductive strategy of alpine Nabalus plants, which includes clonal reproduction (Sayers, 1989), may buffer them somewhat from the probable negative effect of climate change on recruitment from seed.

Nevertheless, while alpine plant populations may be able to persist solely via clonal reproduction for many years (even a thousand), sexual reproduction is still important for virtually all alpine plants (Körner, 2003). Declines in successful sexual reproduction could mean reduced evolutionary potential for alpine Nabalus plant populations, a reduced ability to colonize or re- colonize sites, and probable demographic declines in existing populations.

With regard to non-alpine Nabalus trifoliolatus, it is more difficult to predict the direct effects of forecasted regional warming for the coming decades (Fan et al., 2015) because seeds and seedlings were transplanted to sites of colder, rather than warmer climate. Nevertheless, non- alpine N. trifoliolatus maintained fitness across a range of climates, which suggests it is generally tolerant of a range of temperatures. This suggestion of broad tolerance is further evidenced by its

35 geographic range, which spans from Newfoundland to Georgia and from sea level to treeline

(~1000 m a.s.l.) (Bogler, 2006), where it is replaced by N. trifoliolatus var. nanus. Given its tolerance for a variety of environments and climates, we suggest that non-alpine N. trifoliolatus will be resilient to climate change.

Finally, beyond tolerance breadth, we consider the temperature optima for seeds and seedlings of our three focal Nabalus taxa. Unexpectedly, we found that seeds and seedlings did not always exhibit greatest fitness at sites of their native climate/elevation. Also unexpectedly, the site of highest fitness differed by life stage. Seeds of all three taxa established best at high elevation, while seedlings survived best at low elevation. These results suggest that while environmental tolerance breadth varied among taxa (i.e., alpine Nabalus exhibited narrower tolerance breadth for seed establishment), optima did not. This finding is not altogether surprising: Baskin and Baskin (2014) note very similar germination temperature optima across many alpine and non-alpine species. Similarly, many alpine plants grow quite successfully at low elevation in gardens, and appear to be limited only by competition at their lower elevation limit

(Choler, Michalet, & Callaway, 2001; Ettinger, Ford, & HilleRisLambers, 2011; Pellissier et al.,

2013). We suspect that our seedling transplants were responding universally to environmental factors that either promoted or hindered fitness at sites. For example, the cooler temperatures and consequent reduced evaporative loss of our high elevation sites may have provided ideal germination and establishment conditions for all three taxa. Conversely, the high elevation site appeared to universally challenge our seedling transplants, which spent their first 4–5 weeks growing at low elevation and promptly lost their first true leaves after transplant at high elevation, perhaps due to high wind. In different ways, the high and low elevation sites may have provided universally harsh or beneficial environments for seeds and seedlings, yielding our

36 unexpected results––a conclusion also reached by Kim and Donohue (2013) to explain similar results in their seed and seedling transplant experiments.

Plasticity

The results for phenotypic trait plasticity also suggested resilience to climate change in our focal Nabalus taxa (prediction two). Rather than discovering greater plasticity in the broad- range N. trifoliolatus taxa as we predicted, we instead found significant and uniform plasticity across all three taxa for almost all the functional traits we evaluated. Nabalus as a whole, then, appears to be moderately to highly plastic for a number of traits of probable importance for climate change response (Table 2-1; Nicotra et al., 2010). While we cannot be certain which traits will be important for these particular taxa, their overall strong plasticity across a variety of traits is a good indicator of their general resilience (Nicotra et al., 2010; Valladares et al., 2014).

Additionally, we can conclude that the plasticity exhibited by our three focal taxa is consistent with an adaptive response to a change in climate, as the direction of phenotypic changes mirrored those naturally observed in plants along climatic (i.e., elevational or latitudinal) clines (Table 2-1). Growth traits exhibited the greatest plasticity in our focal taxa

(average PImd = 0.66; Table 2-3). Plants at high elevation were shorter, less massive, had a smaller total leaf area, and had fewer leaves than plants at low elevation. These differences are in accordance with the general observation of declines in plant size as elevation increases (Billings

& Mooney, 1968; Körner et al., 2016). Trends in leaf number are more variable and actually trend toward equal or greater leaf number at high elevation (Körner, 2003); however, examples of the opposite trend (like ours) do exist (e.g., Kim & Donohue, 2013). Our allocation traits, which measured plants’ relative investment into different structures, also followed naturally

37 observed trends. As expected, specific leaf area declined with elevation (Körner, 2003), reflecting higher relative growth rate and photosynthetic activity at lower elevation (Pérez-

Harguindeguy et al., 2013). Similarly, leaf dry matter content also declined with elevation, probably resulting from declines in cell wall thickness and leaf density, as expected (Körner,

2003). Specific root length also declined with elevation in our study taxa, again echoing naturally observed trends in root investment over elevational gradients (Bliss, 1956; Körner, 2003; Körner

& Renhardt, 1987). Finally, in our leaf form traits, we found declines in green pigmentation at high elevation. Green coloration of leaves often (but not always) reflects foliar chlorophyll concentration (Do Amaral et al., 2019; Hu et al., 2010; Murakami, Turner, van den Berg, &

Schaberg, 2005; Vollmann, Walter, Sato, & Schweiger, 2011), for which elevational trends are mixed (Covington, 1975; Filella & Peñuelas, 1999; Young et al., 2018).

Certain traits did not exhibit plasticity although they do vary naturally along elevational gradients: root to shoot ratio, red and blue coloration (i.e., anthocyanins), and leaf shape (Körner,

2003; Körner & Renhardt, 1987; Peppe et al., 2011; Royer, Meyerson, Robertson, & Adams,

2009). At least with regard to root to shoot ratio, this finding is not surprising: plants are generally less able to modulate above versus belowground investment than other traits (Poorter et al., 2012). Nevertheless, although our focal taxa may not be able to respond to changes in climate via plasticity in all functional traits, the strong response they exhibit in a number of important functional traits will likely confer resilience in the face of change (Chevin et al., 2010;

Jump & Peñuelas, 2005; Nicotra et al., 2010; Valladares et al., 2014). This finding echoes our seedling survival results, in suggesting less vulnerability to climate change in older, established plants of N. boottii, N. trifoliolatus var. nanus, and non-alpine N. trifoliolatus.

38 What explains the uniformity plasticity across our focal taxa? An abundance of theoretical and empirical work exists relating strong plasticity responses to environmental heterogeneity, elevation, and range size (Bradshaw, 1965; Gugger et al., 2015; Hamann et al.,

2018; Levins, 1963; Scherrer & Körner, 2011; Schmid, Stöcklin, Hamann, & Kesselring, 2017;

Slatyer et al., 2013; Sultan & Spencer, 2002; Via & Lande, 1985; Vitasse et al., 2014, 2013).

However, much less evidence exists to explain uniformity of plasticity across taxa. Here, we suggest that the uniform plasticity we found in Nabalus taxa indicates evolutionary conservatism of plasticity for these traits. Given the morphological variability noted within these and other

Nabalus species, this explanation seems plausible (Bogler, 2006; Sayers, 1989).

In sum, our results provide evidence for a strong plastic response to climate in functional traits of Nabalus taxa, indicating some degree of resilience to climate change (Chevin et al.,

2010; Matesanz et al., 2010; Merilä & Hendry, 2014; Nicotra et al., 2010; Valladares et al.,

2014). Many organisms are already responding to climate change via plasticity, including montane/alpine (Anderson & Gezon, 2015; Henn et al., 2018) and other plants (Liancourt et al.,

2015; Matesanz et al., 2010), birds (Charmantier et al., 2008; Nussey et al., 2005; Phillimore et al., 2016; Vedder et al., 2013), and small mammals (Réale et al., 2003). In some cases, plasticity alone may ensure population persistence despite rapid climate change (Charmantier et al., 2008;

Phillimore et al., 2016). While plasticity alone may not be enough to ensure the persistence of our focal species (this merits further investigation), it will certainly buffer populations from change and give them time to respond more completely through evolutionary adaptation or range shift (Chevin et al., 2010; Jump & Peñuelas, 2005; Matesanz et al., 2010; Merilä & Hendry,

2014; Nicotra et al., 2010).

39 Niche breadth-range size hypothesis

Our results somewhat support the niche-breadth range-size hypothesis for these species: at least at the seed establishment stage, we found greater fitness in broad-range N. trifoliolatus sensu lato across a wider range of climates/elevations than in the narrow-range mountaintop endemic N. boottii, suggesting greater environmental tolerance/niche breadth in N. trifoliolatus.

N. boottii’s failure to recruit in warmer environments could therefore at least partially explain its narrower geographic range. These results echo those of Hirst et al. (2017) for Australian alpine daisies (Brachyscome spp.): they also discovered support for the NB-RS hypothesis in seed germination but not in seedling performance.

While our results were consistent in some aspects with the niche breadth-range size hypothesis, the mechanism underlying N. trifoliolatus’s broader establishment niche breadth differed from our prediction. We assumed individual phenotypic plasticity to be the principal driver of greater niche breadth in broad-range N. trifoliolatus, based on speculation of significant plasticity in this species (Bogler, 2006; Haines et al., 2011). We therefore expected that individuals/populations of N. trifoliolatus would exhibit broad and uniform niche breadth, like

Wasof et al. (2015) discovered for European alpine plants. Instead, we found that N. trifoliolatus populations differed in their niche breadth for seed establishment: non-alpine N. trifoliolatus populations exhibited broad niche breadth but alpine N. trifoliolatus var. nanus populations exhibited narrow niche breadth. N. trifoliolatus, therefore, appears to be a generalist species composed of both specialist and generalist populations, at least with regard to establishment niche breadth (Sexton et al., 2017; Slatyer et al., 2013).

This finding of population-level niche breadth differences suggests a different mechanism driving greater species-level niche breadth in N. trifoliolatus than uniform individual plasticity.

40 For instance, local adaptation to environmental conditions in N. trifoliolatus var. nanus but not in non-alpine N. trifoliolatus could explain our observed pattern. Griffith and Sultan (2012), although working with two different species, found evidence for the adaptation of specialization in one species but not its generalist congener; similar evolutionary forces (e.g., performance trade-offs) could have driven local adaptation of specialization in N. trifoliolatus var. nanus’s stable but extreme alpine environments but not in non-alpine N. trifoliolatus’s variable non- alpine environments. Second, high elevation var. nanus populations could have lost plasticity through canalization (evolution of an invariant phenotype), as has been found in other high elevation plant populations (Schmid et al., 2017). However, our finding of uniform plasticity for functional traits across all of our taxa does not support this explanation. Third, non-alpine N. trifoliolatus populations could harbor greater variability for genetically-determined climatic optima and tolerance breadth than var. nanus populations, as discovered in a number of other species (Angert, Sheth, & Paul, 2011; Bolnick et al., 2003, 2010). Finally, some combination of the above factors could be acting in concert to determine niche breadth and range size in N. trifoliolatus, as Liu et al. (2015) discovered for a Mongolian steppe grass species (Liu et al.,

2015).

In general, the relative contributions of plasticity versus local adaptation and variability among individuals versus populations to species-level niche breadth and range size remain understudied and merit further investigation (Slatyer et al., 2013). Furthermore, although our findings, like many others, support the niche breadth-range size hypothesis for at least some life stages (Sexton et al., 2017; Slatyer et al., 2013), the hypothesis is still controversial (Cardillo et al., 2019; Moore et al., 2018) and itself needs further investigation.

41 Regardless of mechanism, climatic niche breadth in populations and species has important implications for climate change vulnerability. Generalist taxa with generalist populations should be less vulnerable to climate change than generalist taxa composed of specialist populations, or specialist taxa (Sexton et al., 2017; Slatyer et al., 2013). Therefore, we expect populations of N. boottii and N. trifoliolatus var. nanus, which exhibit narrower seed establishment niche breadth, to be more vulnerable to climate change than non-alpine N. trifoliolatus. However, at the species level, N. trifoliolatus is likely secure, as non-alpine populations exhibit broad niche breadth.

Conclusion

Altogether, our findings suggest tentative support for the niche breadth-range size hypothesis in driving range size differences in broad range N. trifoliolatus versus narrow-range

N. boottii, but do not accord with our predicted mechanism: universally greater niche breadth in

N. trifoliolatus individuals via phenotypic plasticity. Our findings suggest instead that individuals and populations of N. trifoliolatus differ in their niche breadth, and that environment of origin rather than species identity or range size determines vulnerability of Nabalus populations to climate change. Although older life stages of the mountaintop endemic N. boottii

(like the other Nabalus taxa) appear resilient to climate change, both in terms of survival of direct effects and phenotypic plasticity, populations are likely to suffer from reduced seed recruitment as the climate continues to warm. Given its status as a globally rare species, we recommend monitoring of N. boottii populations to assess population stability over the coming years and decades.

42 Table 2-1. Functional traits examined, their typical trend with elevation, and their potential importance for climate change response.

Trait Elevation trend Climate change importance Growth traits (single traits) Declines with elevation Competitive ability Height (Billings & Mooney, 1968; Körner et al., 2016) (Nicotra et al., 2010; Westoby, 1998) Declines with elevation (as net primary productivity) Indicator of net primary productivity, competitive ability Dry mass (Girardin et al., 2010; Körner, 2003; Luo et al., 2004) (Nicotra et al., 2010) Declines with elevation Photosynthetic rate, growth, & carbon balance Total leaf area (Pérez-Harguindeguy et al., 2013) (Nicotra et al., 2010; Pérez-Harguindeguy et al., 2013) Stable or slightly increases with elevation Photosynthetic rate, growth, & carbon balance Leaf number (Körner, 2003) (Nicotra et al., 2010) Allocation traits (derived traits) Investment in above- vs. belowground structures. Important for Increases with elevation Root to shoot ratio competitive ability & as precipitation patterns change (Körner, 2003; Körner & Renhardt, 1987) (Nicotra et al., 2010; Pérez-Harguindeguy et al., 2013) Leaf investment; correlate of relative growth rate, photosynthetic rate, leaf longevity, etc. Important for life history strategy & Declines with elevation Specific leaf area competitive ability (Körner, 2003) (Nicotra et al., 2010; Poorter, Niinemets, Poorter, Wright, & Villar, 2009; Westoby, 1998; Wright et al., 2004) Increases with elevation Root investment; important esp. as precipitation patterns change Specific root length (Bliss, 1956; Körner, 2003; Körner & Renhardt, 1987) (Nicotra et al., 2010; Pérez-Harguindeguy et al., 2013) Product of leaf thickness & density; trend is variable for thickness, density increases with elevation Growth and carbon balance Leaf dry matter content (Choler, 2005; Körner, 2003; Poorter et al., 2009; Scheepens, (Nicotra et al., 2010; Pérez-Harguindeguy et al., 2013) Frei, & Stöcklin, 2010) Leaf form traits (single traits) Photosynthetic rate (chlorophyll concentration); freezing- or Foliar anthocyanins (red) increase with elevation, chlorophyll Leaf coloration drought-resistance, nutrient levels (anthocyanin concentration) (green) variable (red, green, blue) (Chalker-Scott, 1999; Close & Beadle, 2003; Do Amaral et al., 2019; (Covington, 1975; Filella & Peñuelas, 1999; Riebesell, 1981) Nicotra et al., 2010; Steyn, Wand, Holcroft, & Jacobs, 2002) Degree of dissection, number of leaf teeth increase with Leaf shape Photosynthetic rate, growth, & carbon balance elevation (roundness, circularity) (Nicotra et al., 2010; Pérez-Harguindeguy et al., 2013) (Peppe et al., 2011; Royer et al., 2009)

43 Table 2-2. Seed collection sites for Nabalus plants propagated in transplant experiments. Latitude and longitude values indicate the approximate central point of sampling, while elevation values indicate the approximate average elevation of sampling locations within each site. N designates the number of partial inflorescences sampled at each site, totaling 50 to 51 seed families for each taxon and 151 across all three taxa.

N per Taxon Population State Latitude Longitude Elev (m) N taxon N. boottii Whiteface NY 44.365758 -73.902878 1437 50 50 N. trifoliolatus Lakes of the Clouds NH 44.258907 -71.317071 1543 15 50 var. nanus Washington Auto Road NH 44.282153 -71.277077 1470 35 N. trifoliolatus Canton ME 44.463887 -70.333863 154 25 51 (non-alpine) Topsham ME 43.938874 -69.965660 38 26

44 Table 2-3. Model summaries for functional trait data from seedling transplant experiment. We grew seedlings of Nabalus boottii (NB), Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus boottii (NT) at low elevation (“Lo”, 375 m a.s.l.) and high elevation (“Hi,” 1398 m a.s.l.) on Whiteface Mountain in Wilmington, NY for 61 days before recording functional trait measurements. We log-transformed some trait data prior to analysis to improve normality of residuals and homogeneity of variance. P-values in bold are significant at α = 0.05. The estimate (β) and its associated standard error are given below each P-value. Under the “Best model” column, S stands for site, T for taxon, and N for null; + indicates the model is additive. None of the best models included an interaction of site and taxon.

Model attributes Significant differences Trait N Best Log d.f. Site Taxon model Low - Hi NN - NB NT - NB NN - NT

Growth P < 0.001 P = 0.009 P = 0.002 P = 0.634 Height 53 S+T 6 1.26 (0.17) 0.61 (0.22) 0.70 (0.21) -0.10 (0.20) P < 0.001 Dry mass 52 S 4 – – – 1.42 (0.24) P < 0.001 Total leaf area 54 S 4 – – – 2.12 (0.34) P = 0.017 P = 0.001 P = 0.002 P = 0.621 Leaf number 53 S+T 6 0.34 (0.14) -0.62 (0.18) -0.54 (0.17) -0.08 (0.16) Allocation Root to shoot P = 0.133 P = 0.001 P = 0.051 52 T 5 – ratio -0.42 (0.28) -0.92 (0.26) 0.50 (0.25) Specific leaf P < 0.001 52 S 4 – – – area 12.71 (3.0) Specific root P < 0.001 P = 0.108 P < 0.001 P = 0.035 52 S+T 6 length -1.50 (0.19) 0.40 (0.24) 0.87 (0.23) -0.47 (0.22) Leaf dry P = 0.003 matter 52 S 4 – – – -0.30 (0.09) content Leaf form Red 54 N 3 – – – – coloration Green P < 0.001 54 S 4 – – – coloration 43.53 (7.41) Blue P = 0.754 P = 0.051 P = 0.083 54 T 5 – coloration -2.10 (6.66) -12.51 (6.25) 10.41 (5.90) Shape: 52 N 3 – – – – roundness Shape: 52 N 3 – – – – circularity

45 Table 2-4. Phenotypic plasticity index values for functional traits of Nabalus plants at high versus low elevation (1398 vs. 375 m a.s.l.). PImd (phenotypic plasticity index based on median values) is calculated as the absolute difference in median trait values among the two environments divided by the larger median value (Valladares et al., 2006). ESPImd, the environmentally standardized median plasticity index value, accounts for the environmental gradient among the two environments. In this case, we divided PImd by 4.9º C (the difference in average growing season temperature between sites) to yield an index indicating plasticity per ºC difference in growing season temperature. Taxa are Nabalus boottii (NB), Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus boottii (NT).

PImd ESPImd Trait NB NN NT Avg PImd NB NN NT Avg ESPImd Height 0.76 0.69 0.73 0.16 0.14 0.15 Dry mass 0.63 0.77 0.76 0.13 0.16 0.16 Total leaf area 0.85 0.72 0.83 0.17 0.15 0.17 Leaf number* 0.30 0.43 0.40 0.06 0.09 0.08 0.1 0.1 0.1 Overall growth average 0.63 0.65 0.68 0.66 3 3 4 0.14

Root to shoot ratio 0.25 0.02 0.36 0.05 0.00 0.07 Specific leaf area 0.63 0.40 0.29 0.13 0.08 0.06 Specific root length 0.44 0.86 0.87 0.09 0.18 0.18 Leaf dry matter content 0.39 0.28 0.18 0.08 0.06 0.04 0.0 0.0 0.0 Overall allocation average 0.43 0.39 0.42 0.41 9 8 9 0.08

Red coloration 0.26 0.07 0.05 0.05 0.02 0.01 Green coloration 0.44 0.28 0.39 0.09 0.06 0.08 Blue coloration 0.27 0.14 0.29 0.06 0.03 0.06 Shape: roundness 0.03 0.04 0.00 0.01 0.01 0.00 Shape: circularity 0.10 0.04 0.04 0.02 0.01 0.01 0.0 0.0 0.0 Overall leaf form average 0.22 0.12 0.15 0.16 4 2 3 0.03

0.0 0.0 0.0 Overall averages 0.42 0.36 0.40 0.39 9 7 8 0.08

*Leaf number plasticity index values were calculated from mean, not median values; they therefore correspond with PIv from Valladares et al. (2006).

Mid

(887 m)

High/Summit

(1398 m)

Low/Base (375 m)

Figure 2-1. Field sites for the reciprocal transplant experiments on Whiteface Mountain in Wilmington, NY. We constructed raised beds with identical soil mixtures at each site (see inset photo for high elevation/summit bed). All three sites were used for the seed transplant experiment. Only the low/base and high/summit sites were used for the seedling transplant experiment.

47 NB NN NT Low High

Figure 2-2. Example whole-plant scans of alpine Nabalus boottii (NB), alpine Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus trifoliolatus (NT) at the conclusion of the seedling transplant experiment (day 61). We transplanted seedlings of each taxon into raised beds at high (1398 m a.s.l.) or low (375 m a.s.l.) elevation. Plants in each column are maternally related: their seeds were derived from the same mother plant. These images depict some of the most extreme examples of phenotypic plasticity among relatives in the seedling transplant experiment.

48 NB NN NT 25% Mid vs Summit: ** NB vs NT: ** 20%

15%

10% Establishment 5%

0% Base Mid Summit Base Mid Summit Base Mid Summit Site

Figure 2-3. Predicted establishment of Nabalus boottii (NB), Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus trifoliolatus (NT) from seed at the base (375 m a.s.l.), mid elevation (887 m a.s.l.) and summit (1398 m a.s.l.) sites. Predictions were based on the best zero- inflated GAMLSS model for establishment percent, measured at the conclusion of the 50-day seed transplant experiment. Significant differences are indicated in the upper left corner of the plot (* P <0.05, ** P <0.01, *** P <0.001).

AB Site BA Taxon 1.00 1.00

+ NB + Lo + NN 0.75 + Hi 0.75 + NT

0.50 0.50 + + + 0.25 + 0.25 + Survival probability Survival probability

0.00 Lo vs Hi: ** 0.00 N.S. 0 20 40 60 0 20 40 60 Time (days) Time (days)

Figure 2-4. Predicted survival functions based on Cox proportional hazards models for seedlings of Nabalus boottii (NB), Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus trifoliolatus (NT). Panel A (“Site”) includes survival functions averaged across taxa at each elevation: low (“Lo”, 375 m a.s.l.) and high (“Hi”, 1398 m a.s.l.) elevation. Panel B (“Taxon”) includes survival functions averaged across sites for each taxon. Significant differences are indicated in the lower left corner of the plots (* P <0.05, ** P <0.01, *** P <0.001).

50 A Height B Dry mass

NB NN NT NB NN NT 100 Hi vs Lo: *** 100 Hi vs Lo: *** NB vs NN: ** NB vs NT: **

10 10

Dry mass (mg) 1 Plant height (mm) Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo

C Total leaf area D Leaf number NB NN NT NB NN NT ) 2 Hi vs Lo: *** 8 Hi vs Lo: * m NB vs NN: ** m 1000 ( 6 NB vs NT: ** a e r 100 a

f 4 a e l

l 10 Leaf number 2 a t o T Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo Site

Figure 2-5. Comparisons of growth-related functional traits for plants at the conclusion of the seedling transplant experiment. Seedlings of Nabalus boottii (NB), Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus trifoliolatus (NT) were planted in low (“Lo”, 375 m a.s.l.) and high (“Hi”, 1398 m a.s.l.) elevation raised beds and allowed to grow for 61 days. Dry mass is whole-plant dry mass. Total leaf area is equal to whole rosette leaf area for each individual. Height is the length of the longest leaf in the rosette. Significant differences at α = 0.05 are indicated in the plots (* P <0.05, ** P <0.01, *** P <0.001).

51 A Root:shoot B Specific leaf area

NB NN NT NB NN NT

) Hi vs Lo: ***

10.0 g 50 m 40 3.0 2 m 30 m ( 1.0 20 A L S Root:shoot ratio 0.3 NB vs NT: ** 10 Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo

C Specific root length D Leaf dry matter content

NB NN NT NB NN NT 300 Hi vs Lo: *** NB vs NT: *** 700 Hi vs Lo: *** ) NN vs NT: * ) g 100 g m g m m 30 300 (

m C (

10 L M

R 3 D L S 100 Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo

Figure 2-6. Comparisons of allocation-related functional traits for plants at the conclusion of the seedling transplant experiment. Seedlings of Nabalus boottii (NB), Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus trifoliolatus (NT) were planted in low (“Lo”, 375 m a.s.l.) and high (“Hi”, 1398 m a.s.l.) elevation raised beds and allowed to grow for 61 days. Specific leaf area and leaf dry matter content were calculated for whole rosettes. Significant differences at α = 0.05 are indicated in the plots (* P <0.05, ** P <0.01, *** P <0.001).

52 A Red B Green C Blue

NB NN NT NB NN NT NB NN NT N.S. Hi vs Lo 100 N.S. *** 160 150 75

120 100 50 Red value Blue value 80 Green value 50 25

Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo Hi Lo Site Site Site

Figure 2-7. Comparisons of leaf coloration for plants at the conclusion of the seedling transplant experiment. Seedlings of Nabalus boottii (NB), Nabalus trifoliolatus var. nanus (NN) and non- alpine Nabalus trifoliolatus (NT) were planted in low (“Lo”, 375 m a.s.l.) and high (“Hi”, 1398 m a.s.l.) elevation raised beds and allowed to grow for 61 days. We used the RGB Measure plugin for ImageJ to divide foliage coloration into separate red, green and blue channels and quantify the color value for each channel. For each color, values can range from 0 to 255. Significant differences at α = 0.05 are indicated in the upper-right corner of the plots (* P <0.05, ** P <0.01, *** P <0.001).

53 CHAPTER 3. GENOMIC INVESTIGATION OF THE HISTORIC AND FUTURE PERSISTENCE OF OBLIGATE AND FACULTATIVE MOUNTAINTOP PLANT SPECIES

ABSTRACT

In post-glaciated areas, relict populations of tundra species often persist in isolated refugia on mountaintops. Certain traits of these species have therefore enabled their persistence in small, isolated populations over thousands of years. Understanding the factors enabling mountaintop species’ persistence and whether these factors will promote their future survival are topics of paramount importance as global environmental change pushes Earth into a sixth mass extinction. Here, we used population genomic techniques to understand the past and future persistence of obligate and facultative mountaintop plant species native to the northeastern

United States. We assessed factors related to these species’ adaptive and migration potential as indicators of their ability to persist under global environmental change. We made the following hypotheses: (1) Nabalus trifoliolatus (facultative mountaintop species) will exhibit greater genetic diversity and equal ploidy as Nabalus boottii (obligate mountaintop species), indicating overall greater adaptive potential; and (2) Nabalus trifoliolatus (facultative mountaintop species) will exhibit greater migration potential than Nabalus boottii (obligate mountaintop species) on both historical and recent time scales. Contrary to our hypotheses, we found greater genetic diversity, higher ploidy, and equal to higher migration potential in the mountaintop obligate N. boottii versus N. trifoliolatus. High genetic diversity (likely maintained through tetraploidy) and migration potential have likely enabled the historical persistence of N. boottii populations on northeastern mountaintops, and should contribute to resilience of this species in the face of global environmental change.

INTRODUCTION

Mountaintops provide natural laboratories for island biogeography, harboring pockets of colder and/or wetter habitat in a warmer, drier, lower elevation landscape (McCormack, Huang,

& Knowles, 2009). These “sky islands” often support relict populations of species adapted to cold or moist conditions that have become restricted to high elevation sites during the Holocene

(Hampe & Jump, 2011; Migliore et al., 2013). In temperate zones, mountaintops frequently host endemic alpine species or relict populations of arctic tundra species that have persisted in small, isolated populations since the last glacial retreat (Körner, 2003; Martin & Germain, 2016;

Schmitt & Schönswetter, 2010; Varga & Schmitt, 2008; Woolbright, Whitham, Gehring, Allan,

& Bailey, 2014). Based on population genetic theory, these populations should be highly vulnerable to extinction (Allendorf et al., 2013; Gilpin & Soulé, 1986). How, then, are small mountaintop populations able to persist? And can they persist in a changing world?

The process of extinction elucidates the foundation for species persistence. Under the extinction vortex model (Gilpin & Soulé, 1986), populations go extinct when their small, isolated nature causes inbreeding and loss of genetic diversity, which in turn causes a decline in fitness and adaptability, resulting in a reduced population size. This positive feedback cycle repeats until the population is driven extinct or the cycle is interrupted by demographic and/or genetic rescue through migration (natural or assisted). Persistence, then, depends on the opposite factors as extinction: typically, a large population size (at least several hundred individuals; Frankham,

Bradshaw, & Brook, 2014), moderate to high genetic diversity, and migration/gene flow, which helps maintain population size and genetic diversity across interacting populations (a

55 metapopulation). Migration also enables species to re-colonize sites following local extirpation, again adding stability to the overall metapopulation.

These same factors that generally enable persistence (population size, genetic diversity, migration) are also important for persistence in the face of global environmental change, a phenomenon that encompasses climate change, land-use change, nitrogen deposition, species invasions, and more, and threatens to send Earth’s biota into a sixth mass extinction (Barnosky et al., 2011; Ceballos et al., 2015; Sala et al., 2000; Tilman et al., 2017). Although mountaintop species are experiencing a number of environmental changes (all those aforementioned), they are especially vulnerable to climate change (Hampe & Jump, 2011; Jiménez-Alfaro et al., 2016), as upwardly shifting species and environmental conditions threaten to push mountaintop species off summits in an “escalator to extinction” (Costion et al., 2015; Dirnböck et al., 2011; Elsen &

Tingley, 2015; Freeman et al., 2018; Marris, 2007; Urban, 2018). Survival of any environmental change depends on organisms’ abilities to tolerate change, adapt evolutionarily to change, or migrate to more suitable locations (Chevin et al., 2010; Davis et al., 2005; Jump & Peñuelas,

2005; McCarty, 2001).

Traits of mountaintop plant species that have enabled their historical persistence could threaten their persistence in the face of global environmental change. For instance, many mountaintop/alpine species are classified as stress tolerators under Grime’s CSR model (Grime,

1977). As stress tolerators (S), mountaintop plants succeed in extreme environments by growing slowly, investing in specialized structures, and devoting more resources to survival than to reproduction (Grime, 1977; Hampe & Jump, 2011). Many mountaintop plant species delay reproduction for years while accumulating the necessary resources; for example, Körner suggests

5–10 years for alpine monocarpic perennials versus two years at lower elevation (Körner, 2003).

56 Some may only successfully reproduce sexually once every millennium, but their slow growth and extreme longevity enable their persistence at a site nonetheless (Körner, 2003). These stress tolerator strategies function well in relatively stable environments (Grime, 1977; Hampe &

Jump, 2011), but render many mountaintop plant species less capable of swift migration

(smaller/delayed reproductive output) or evolutionary adaptation (longer generation times) when faced with environmental change.

Mountaintop plant species also exhibit certain reproductive strategies that may have enabled their long-term persistence in small populations: many replace or supplement outcrossing sexual reproduction with selfing, vegetative (clonal) reproduction, or apomixis, including the asexual reproduction of seeds (agamospermy) (Hampe & Jump, 2011; Körner,

2003; Tuxill & Nabhan, 2001). These reproductive strategies help small, isolated plant populations maintain their demographic population size (Hampe & Jump, 2011), but they

(especially selfing) can result in increased inbreeding and loss of genetic diversity, increasing the chance of inbreeding depression and reducing adaptive potential in the face of global change.

Asexual reproduction can actually help populations maintain genetic diversity (Gorelick & Heng,

2011; Hamrick, Godt, & Sherman-Broyles, 1992), but relict populations consisting of a single clone certainly experience reduced adaptive potential compared to populations with higher genetic and genotypic diversity.

Other traits characteristic of mountaintop plant species, such as polyploidy, may actually benefit their odds of surviving environmental change. Although higher rates of polyploidy are not ubiquitous in alpine/mountaintop floras (Körner, 2003), there are several regions where rates of polyploidy do increase with elevation (Packer, 1974), including parts of Africa and eastern

57 and western North America (Löve & Löve, 1967; Morton, 1993; Weiss-Schneeweiss, Emadzade,

Jang, & Schneeweiss, 2013).

Although polyploidy is not without costs (Comai, 2005; Madlung, 2013), it confers many potential benefits to mountaintop plants. First, polyploidy increases tolerance of environmental extremes, meaning that polyploid species may be better equipped to tolerate environmental changes (Dar & Rehman, 2017; Kawecki, 2008; Levin, 1983; Van De Peer, Mizrachi, &

Marchal, 2017). Indeed, whole genome duplication events appear to have facilitated species’ survival of past climate change (Cai et al., 2019; Fawcett, Maere, & Van de Peer, 2009; Sessa,

2019; Vanneste, Baele, Maere, & Van De Peer, 2014). Additionally, polyploidy confers greater colonization ability (Brochmann et al., 2004; Stebbins, 1940; Te Beest et al., 2012; Weiss-

Schneeweiss et al., 2013), making polyploid species potentially more capable of range shift, re- colonization of extirpated sites, and inter-population migration to maintain population size and genetic diversity. Finally, and perhaps most importantly, polyploidy increases species’ adaptive potential in at least two ways (Levin, 1983; Mable, 2013; Sessa, 2019; Soltis, Visger, & Soltis,

2014; Weiss-Schneeweiss et al., 2013). First, polyploid species typically maintain high levels of genetic diversity despite their tendency toward apomictic or clonal reproduction through fixed genome heterozygosity following whole genome duplication (Brochmann et al., 2004; Kawecki,

2008; Van De Peer et al., 2017; Weiss-Schneeweiss et al., 2013). Second, the redundancy present in polyploid genomes enables evolution at redundant sites (Levin, 1983; Mable, 2013; Soltis et al., 2014; Van De Peer et al., 2017; Weiss-Schneeweiss et al., 2013). Soltis et al. (2014) review additional genomic changes that take place after whole genome duplication which increase adaptive potential, while Sessa (2019) notes: “There are infinite possible ways by which an extra genome can generate new raw materials for selection to shape into adaptation.” All told,

58 polyploidy has likely not only contributed to the historical persistence of mountaintop plant populations, but will also contribute to their persistence in a changing world.

Given that mountaintop plant species may exhibit traits that constrain or enhance their ability to persist under environmental change, diagnosing their response is non-trivial yet important, especially for rare, endemic mountaintop species that may be particularly vulnerable to extinction (Costion et al., 2015; Dirnböck et al., 2011; Elsen & Tingley, 2015; Freeman et al.,

2018; Marris, 2007; Urban, 2018). For studies of environmental change response in rare/endemic species, comparison with a non-rare/non-endemic congener can provide important context for understanding the vulnerability of the rare species (Gitzendanner & Soltis, 2000; Jiménez-Alfaro et al., 2016).

In this study, we use population genomics to investigate the evolutionary and migration potential of one obligate and one facultative mountaintop species in order to understand how these species have historically persisted on mountaintops and how vulnerable they are to extinction under environmental change. Our focal species include Nabalus boottii (Boott’s rattlesnake-root), a globally rare alpine plant species endemic to mountaintops of the northeastern United States (Bogler, 2006; Haines et al., 2011; NatureServe, 2018), and N. trifoliolatus, a facultative mountaintop species in the northeastern United States (Bogler, 2006;

Haines et al., 2011). The range of the alpine variety of N. trifoliolatus (N. trifoliolatus var. nanus) completely overlaps that of N. boottii, while non-alpine N. trifoliolatus is widespread in eastern North America. Little is known about the genetics of either species, apart from contrasting findings of N. boottii being diploid or tetraploid (Löve & Löve, 1966; Sayers, 1989), and consistent findings of N. trifoliolatus being diploid (Babcock et al., 1937; Jones, 1970; Löve

& Löve, 1966; Powell et al., 1974; Sayers, 1989; Tomb et al., 1978).

59 Nabalus boottii and Nabalus trifoliolatus have likely inhabited mountaintops of the northeastern United States since the last glacial retreat 10,000–15,000 years ago, colonizing from refugia in Greenland/eastern Canada or south of the Laurentide ice sheet (Bierman, Davis,

Corbett, Lifton, & Finkel, 2015; Brochmann, Gabrielsen, Nordal, Landvik, & Elven, 2003;

Martin & Germain, 2016). These species survived the Holocene Climate Optimum, but now face rates of warming unprecedented in the last 50 million years (Jansen et al., 2007), along with a suite of other environmental changes. These changes include high levels of nitrogen deposition, recreational overuse of habitat, and even invasive species (Baumgardner et al., 2003; Capers et al., 2013; Capers & Slack, 2016; Galloway et al., 1984; Reay et al., 2008). Although limited surveys suggest that populations of N. boottii and N. trifoliolatus are relatively stable, these species may reach a tipping point as environmental changes intensify over the coming decades

(Doak & Morris, 2010).

We used molecular data obtained using double digest restriction-site associated DNA sequencing (ddRADseq) (Baird et al., 2008; Peterson, Weber, Kay, Fisher, & Hoekstra, 2012), to investigated the adaptive potential of these species by assessing their within-population genetic diversity/inbreeding, species-level genetic diversity, and ploidy (objective 1). We also investigated their migration potential by assessing rates of historic and recent gene flow

(objective 2).

Genomic techniques like ddRADseq in particular are useful for diagnosing adaptability/evolutionary potential, as they reflect overall genome diversity at hundreds to many thousands of sites (Corlett, 2017; Harrisson, Pavlova, Telonis-Scott, & Sunnucks, 2014;

McMahon et al., 2014). We cannot know in many cases which genes/loci will be important for environmental change response, so a genomic-level survey of genetic diversity is the best option

60 for inferring adaptive potential. For investigating migration, molecular methods in general are useful, as they are often easier to employ than direct methods (i.e., observing dispersal), reveal only effective migration (that which results in survival/reproduction), and capture rare long- distance dispersal events that are unlikely to be directly observed (Meirmans, 2014; Whitlock &

Mccauley, 1999). Genomics techniques offer an advantage over other molecular methods in the precision of their estimates (Corlett, 2017).

Given the broader habitat preferences and greater potential for connectivity among populations of N. trifoliolatus, we make the following hypotheses:

(1) Nabalus trifoliolatus (facultative mountaintop species) will exhibit greater genetic

diversity and equal ploidy as Nabalus boottii (obligate mountaintop species), indicating

overall greater adaptive potential.

(2) Nabalus trifoliolatus (facultative mountaintop species) will exhibit greater migration

potential than Nabalus boottii (obligate mountaintop species) on both historical and

recent scales.

If our hypotheses are supported, our results will indicate greater vulnerability to environmental change in the rare mountaintop endemic N. boottii, suggesting that conservation efforts may be needed to ensure species’ persistence in light of environmental change.

METHODS

Study taxa

The genus Nabalus, within the Asteraceae, contains 24 species native to North America and Asia. The 14 North American Nabalus species were formerly grouped within the genus

Prenanthes (Bogler, 2006; Haines et al., 2011). All North American Nabalus species reproduce

61 both sexually, as monocarpic perennials with wind-dispersed seed, and clonally, via taproot offshoots (Bogler, 2006; Sayers, 1989).

Nabalus boottii DC. (Boott’s rattlesnake-root) is an alpine plant endemic to the highest elevations (1000–1800 m a.s.l.) of New York, Vermont, New Hampshire, and Maine, where it occurs in fewer than 20 sites and is thus of high conservation concern (Bogler, 2006). The ploidy level of N. boottii is uncertain, with one flow cytometry study determining the species as diploid

(Sayers, 1989) and one as tetraploid (Löve & Löve, 1966); both studies included only a few samples from one geographic location.

Nabalus trifoliolatus Cass. is widespread throughout eastern North America and occurs from sea level to high elevation (Bogler, 2006). A habitat generalist, N. trifoliolatus inhabits woodland, cliff, sandy, and saline areas. N. trifoliolatus, unlike N. boottii, is well established as diploid (Babcock et al., 1937; Jones, 1970; Löve & Löve, 1966; Powell et al., 1974; Sayers,

1989; Tomb et al., 1978). Nabalus trifoliolatus var. nanus (Bigelow) Fernald, the alpine variety of N. trifoliolatus, is the only high elevation congener of N. boottii and can be found from 1100 to 1600 m a.s.l. in New York, New Hampshire, and Maine. Historically, N. trifoliolatus var. nanus was treated as a separate species (Gleason & Cronquist, 1963; Milstead, 1964; Mitchell &

Tucker, 1997; Sayers, 1989), although most sources today recognize it as a variety or simply part of N. trifoliolatus without distinction (Bogler, 2006; Haines et al., 2011). Here, we refer to the two varieties as “Nabalus trifoliolatus var. nanus” (abbreviated NN in figures and tables) and

“non-alpine Nabalus trifoliolatus” (abbreviated NT) while we reserve the unqualified “Nabalus trifoliolatus” to refer to the species as a whole.

62 Field collection, sample treatment, and storage

We collected leaf tissue for genomic analyses in 2014 and 2015 from a total of 30 populations (15 of N. boottii, nine of N. trifoliolatus var. nanus, six of non-alpine N. trifoliolatus) across New York, Vermont, New Hampshire, and Maine. In each population, we selected fifteen widely spaced individuals and removed 2–4 cm2 of leaf tissue, storing each sample in a separate paper envelope placed in silica gel. To extract DNA, we ground ≤ 20 mg tissue in a FastPrep

FP120 Cell Disruptor (Thermo Savant, Qbiogene, Carlsbad, CA) using 2 mL XXTuff vials and

2.3 mm chrome steel beads (BioSpec, Bartlesville, OK). We then used QIAGEN’s DNeasy Plant

Mini Kit (Hilden, Germany) to extract and purify DNA samples following the manufacturer’s protocol. We assessed DNA quantity and purity using a Thermo Scientific NanoDrop Lite UV spectrophotometer (Wilmington, DE) and gel electrophoresis. We stored the DNA samples in a -

20 ºC freezer before transferring them to a -80 ºC freezer for long-term storage.

Genomic library preparation and sequencing

We randomly chose 10 samples per population (30 from the N. boottii Whiteface population) yielding a total of 320 samples for genomic analysis, and contracted the Genomics

Core Lab at Texas A&M University Corpus Christi (GCL) to perform double digest restriction- site associated DNA sequencing (ddRADseq, Peterson, Weber, Kay, Fisher, & Hoekstra, 2012).

Through enzyme testing, GCL determined SbfI and MluCI as the optimal enzyme combination for our Nabalus taxa, targeting fragments of 550 to 625 base pairs. The combination of a rare cutter enzyme (SbfI) and a frequent cutter enzyme (MluCI) is useful for reducing the number of targeted fragments and improving sequencing coverage for species with large genomes (e.g., Qiu et al., 2016), like Nabalus spp. and their relatives (Fernandez et al., 2018; Garnatje et al., 2011).

63 GCL used SPRI based size selection to eliminate low molecular weight DNA from 100 samples, then prepared libraries by digesting samples with both restriction enzymes, adding unique barcodes and dual-indexed adaptors, and pooling samples for sequencing. GCL sequenced the pooled libraries on one lane of an Illumina NovaSeq 6000 (San Diego, CA) run and provided us with both raw and demultiplexed sequences. To demultiplex sequences, GCL used the process_radtags function in program STACKS version 1.44 (Catchen, Hohenlohe, Bassham,

Amores, & Cresko, 2013), without including the clean (-c) and quality check (-q) options, as required by program dDocent (Puritz, 2019b; Puritz, Hollenbeck, & Gold, 2014).

SNP calling and filtering

We divided demultiplexed sequences according to species (170 N. boottii samples and

151 N. trifoliolatus samples) and used dDocent version 2.7.7 to perform quality filtering, de novo paired-end assembly, read mapping, and SNP calling for each set of files (Puritz et al., 2014).

For our data reduction steps, we eliminated unique sequences that appeared fewer than three times in the whole dataset (parameter K1), or were present in fewer than four individuals

(parameter K2) based on visual inspection of output graphs. Because the Nabalus genome is large and likely repetitive (Fernandez et al., 2018; Garnatje et al., 2011), we adjusted the -c parameter for CD-HIT (percent similarity to cluster by) to 0.99, the upper end of the range recommended by dDocent creator Jonathan Puritz (0.8 – 0.99) (Puritz, 2019b). We also adjusted the BWA mapping parameter –B (mismatch score) from four to five, thereby increasing the penalty for a mapping mismatch. Through these two parameter adjustments, we attempted to decrease the prevalence of paralogs in our dataset. Because certain elements of the dDocent pipeline (like most genetic/genomic software) cannot handle polyploid data, we performed

64 diploid SNP calling both species while retaining information about the number of calls per allele in our Variant Call Format (VCF) files for downstream ploidy analysis.

We performed filtering of the total raw SNP files (one file for each species) generated by dDocent according to a slightly modified version of filtering strategy five from O’Leary et al.

(2018). We performed steps one, two, and four as described in O’Leary et al.’s (2018) filtering strategy five using VCFtools (Danecek et al., 2011) with two adjustments: we set the minor allele count (--mac) filter to two instead of three (as supported by Linck & Battey, 2019) and reduced the mean depth of coverage (--meanDP) filter from 15 to 3. In our final datasets, mean depth of coverage for all loci was greater than 15 despite relaxing this early filter. We performed step three using Jonathan Puritz’s dDocent_filters script (available at https://github.com/jpuritz/dDocent/raw/master/scripts/dDocent_filters), with an adjustment of the allele balance (AB) cutoffs from 0.2–0.8 to 0.15–0.85. The default values assume diploidy and a typical allele balance of 0.5 (equal number of calls for each allele at heterozygous sites); given the possible tetraploidy of N. boottii, we slightly relaxed these parameters. Because we wished to perform the same filtering strategy on both species in order to create comparable datasets, we also used these relaxed cutoffs for N. trifoliolatus.

After applying this modified version of O’Leary et al.’s (2018) filtering strategy 5, we used the vcflib (Garrison, 2012) function vcfallelicprimitives and the VCFtools function -- remove-indels to decompose our complex variant datasets into SNPs and indels, and then filter out indels. We then applied Christopher Hollenbeck’s Hardy-Weinberg filter (available at https://github.com/jpuritz/dDocent/raw/master/scripts/filter_hwe_by_pop.pl) to additionally eliminate erroneous SNPs (Puritz, 2019a). We specified a P-value cutoff of 0.05 in at minimum

10% of the populations (stricter than program defaults). Finally, we filtered our datasets for

65 linkage disequilibrium. For sets of SNPs that were highly correlated (r2 > 0.40), we retained only one site for further analysis, following the approximate cutoff used by Thrasher et al. (2018). We used the resulting files as our final full datasets for N. boottii and N. trifoliolatus.

Given the needs of downstream analyses, we created several additional files that represented subsets of these full datasets: one of equal sample size at the population level for each species, and two of reduced locus number for N. trifoliolatus (reduced locus full and equal sample size datasets) to achieve a locus number equal to that of the N. boottii datasets. We chose n = 4 as the population size for the equal sample size datasets, and eliminated populations with n

< 4. For populations with n > 4, we first eliminated individuals with the highest percent of missing data, and then randomly eliminated individuals once only individuals with 0% missing data remained. For the reduced locus number N. trifoliolatus datasets, we randomly eliminated loci until we achieved a number equal to that of the N. boottii datasets. We used PGDSpider version 2.1.1.5 to convert our final datasets (full, equal population size, reduced locus number) from VCF to data formats used in downstream programs, including genepop, Arlequin, and

Immanc (BayesAss) (Lischer & Excoffier, 2012).

Adaptive potential

Genetic diversity

We used the basicStats function of R package diveRsity to calculate several population- level diversity statistics for both species using the full datasets (Keenan, McGinnity, Cross,

Crozier, & Prodöhl, 2013). These statistics included allelic richness (AR, the average number of alleles per locus), expected heterozygosity (HE), observed heterozygosity (HO), and the inbreeding coefficient (FIS) (Keenan et al., 2013; Rousset, 2008). Because allelic richness is

66 affected by sample size, we used the rarefaction-based estimates calculated by the basicStats function to account for uneven population-level sample sizes in our full datasets.

We calculated an additional measure of population-level genetic diversity using program

Arlequin version 3.5.2.2 (Excoffier & Lischer, 2010): percent polymorphic loci (PPL). For this analysis, we used the equal population size/equal locus number datasets. We calculated PPL by dividing the number of polymorphic loci per population by the total number of loci, ensuring that all loci were deemed “usable” and retained in each population.

We additionally performed Analysis of Molecular Variance (AMOVA) to determine species-level genetic variation and its partitioning within and among populations in each species.

We specified all populations for each species as part of the same “group” and did not include variance within individuals as part of the AMOVA. For this analysis, we used program Arelquin version 3.5.2.2 and our full datasets as input (Excoffier & Lischer, 2010).

Finally, we investigated genotype diversity (i.e., clonality) in these taxa. Although our sampling method was intended to maximize genetic diversity by sampling widely-spaced individuals, we nonetheless investigated clonality given that some alpine populations and even groups of populations exhibit 100% clonality (Bauert, Kälin, Baltisberger, & Edwards, 1998;

Robinson, 2012). To investigate clonality, we calculated the number of multilocus genotypes in our full datasets using the function poppr from the package poppr version 2.8.2 (Kamvar,

Brooks, & Grünwald, 2015; Kamvar, Tabima, & Grünwald, 2014) in program R version 3.6.0 (R core team, 2017).

67 Ploidy

We used R package gbs2ploidy to estimate ploidy levels in N. boottii (Gompert & Mock,

2017). We repeated the analysis for N. trifoliolatus, a well-established diploid (Babcock et al.,

1937; Jones, 1970; Löve & Löve, 1966; Powell et al., 1974; Sayers, 1989; Tomb et al., 1978), for comparative purposes. Package gbs2ploidy leverages information present in GBS/RAD-type data to infer cytotype (ploidy level) for individuals in a sample composed of diploids, triploids, and/or tetraploids (Gompert & Mock, 2017). The information used by gbs2ploidy includes observed heterozygosity (which we expect to be higher in polyploids versus diploids) and allelic proportions (the ratio of the number of calls for the reference to the alternate allele at each heterozygous locus) for each individual. We typically expect diploids to exhibit 1:1 allelic proportions at heterozygous sites (equal calls for each allele), triploids to exhibit 1:2 or 2:1 proportions, and tetraploids 3:1, 1:1, and 1:3 proportions. The first step of the gbs2ploidy pipeline uses a Bayesian MCMC approach to estimate allelic proportions and genome-level heterozygosity for each individual. This step is followed by Principal Component Analysis

(PCA) of the estimated heterozygosity and allelic proportions and discriminant analysis (DA) to assign individuals to groups (i.e., cytotypes). Gompert and Mock suggest that for datasets of sufficient coverage (> 15x), cytotypes can be assigned directly from the allelic proportions calculated during the first step.

For our analysis, we used VCFtools to extract genotype and allele counts per locus for all individuals in our full datasets using the function --extract-FORMAT-info for the GT (genotype),

RO (reference allele) and AO (alternate allele) fields. From these outputs, we created our input matrices of reference and alternate allele counts, scoring homozygous sites as “NA” in both matrices. We also eliminated all loci with more than one alternate allele. We ran function

68 estprops specifying 10,000 MCMC steps with a 5000 step burn-in, with props set to 0.25, 0.5, and 0.75 (specifying allelic proportions to investigate—1:3, 1:1, and 3:1). We then ran estploidy twice with nclasses set to two and then one (specifying the number of cytotypes we expected in our samples—mixed or all diploid/tetraploid) to perform PCA and DA.

Because our coverage was sufficient (> 15x) for inferring cytotypes directly from estimated allelic proportions, we output posterior probability matrices for each individual detailing the posterior probability distribution for the three possible allelic proportions (1:3, 1:1,

3:1). Most individuals scored the highest posterior probabilities for the 3:1 allelic proportion

(even N. trifoliolatus individuals); therefore, we determined that absolute posterior probabilities for allelic proportions alone were not sufficient for diagnosing ploidy in our datasets. We decided instead to compare relative probabilities for 1:3, 1:1, and 3:1 allelic proportions, assuming that diploids individuals would exhibit higher posterior probabilities of a 1:1 allelic proportion relative to 1:3 or 3:1 proportions compared to tetraploid individuals. After computing the ratio of posterior probabilities for 1:1 to 3:1 allelic proportions and 1:3 to 1:1 allelic proportions for each individual (using the median posterior probability value), we compared ratios across the two species using linear models in R. We log-transformed ratios prior to analysis to improve the normality of residuals and homogeneity of variance.

Migration potential

Historic migration

We used two model-based methods to estimate rates of historic migration (over the past

10–100 generations) among populations for both N. boottii and N. trifoliolatus. First, we used

Barton and Slatkin’s private allele method for estimating the number of migrants per generation

69 (Nm) as implemented in the R package genepop (Barton & Slatkin, 1986; Rousset, 2008). This method utilizes the inverse relationship between the average frequency of rare/private alleles and

Nm to estimate migration, utilizing Wright’s island model with relaxed assumptions (Barton &

Slatkin, 1986; Wright, 1931). We used the equal population size datasets to compute an overall estimate for each species. To compute overall and pairwise estimates for the New York populations of each species, we used the subset of New York populations from the equal population size datasets, to which we added back the Algonquin N. boottii population and the

Wright N. trifoliolatus population (N = 3 for both). We note that although package genepop accounts for the sample size of populations in estimating Nm, estimates increase in precision for

N ≥ 10 per population (Rousset, 2008).

We chose the New York subset here and in later migration rate estimates for several reasons: (1) our sampling included almost all New York populations of N. boottii and N. trifoliolatus var. nanus, so virtually all likely sources of migration were covered, (2) New York represents the westernmost part of the geographical distribution of N. boottii and N. trifoliolatus var. nanus, making long-distance migration from another state unlikely given prevailing wind directions for dispersal (again, ensuring coverage of all likely migration sources), and (3) the

New York populations of each species are clustered in a relatively small geographic area, unlike the populations in other states (excepting New Hampshire for N. boottii).

For our second model-based method of Nm estimation, we used Barton and Slatkin’s

1 FST/GST (fixation index) method, based on the formula !" = '4 × ( ( 1 ÷ +,- ) −1 ), with

GST substitutable for FST (Barton & Slatkin, 1986). This is the method for Nm estimation implemented in the program POPGENE (Yeh, Yang, Boyle, Ye, & Mao, 1997), and is also based upon Wright’s island model with relaxed assumptions (Barton & Slatkin, 1986; Wright,

70 1931). According to Barton and Slatkin (1986), the method of FST/GST calculation (including whether FST or GST is used) should not substantially affect results. Because our datasets are composed almost entirely of biallelic SNPs, FST is reasonably appropriate. However, because several sites in each dataset include more than two alleles, we also performed a GST-based analysis for each species. For each full dataset, we computed pairwise (for New York populations) and global FST and GST (for each species and all new York populations) values using the diffCalc function in package diveRsity (Keenan et al., 2013). The diveRsity package calculates five different differentiation statistics; we used Weir and Cockerham’s FST and Nei and Chesser’s GST (Nei & Chesser, 1983; Weir & Cockerham, 1984). From these statistics, we calculated Nm using Barton and Slatkin’s formula in Microsoft Excel.

Recent migration

We used the program BayesAss to estimate recent migration among our sampled populations for each species. BayesAss is a non-model (population genetic model) based

Bayesian estimator of recent migration, defined as the current and previous two generations

(Rannala, 2007; Wilson & Rannala, 2003). BayesAss estimates the migrant proportion of each sampled population, the source population for migrants, and migrant ancestries for each individual. Because BayesAss is unbiased by uneven sample sizes (Rannala, 2013), we used the full datasets for our analyses.

BayesAss estimates five parameters: migration rates, individual migrant ancestries, allele frequencies, inbreeding coefficients, and missing genotypes. For three of these parameters

(migration rates, allele frequencies, and inbreeding coefficients), the user can adjust “mixing parameters” to improve the acceptance rate of proposed changes to these parameters (optimal is

71 between 40 and 60%), thereby improving the mixing of the Markov chain and maximizing the exploration of the parameter space (Rannala, 2007). After experimentation, we set mixing parameters -a, -m, and -f equal to 1.0 to optimize chain mixing for these datasets. We specified a chain length of 30,000,000 steps (-i parameter) with a 10,000,000 step burn-in, a sampling interval of 2000 (-n parameter), and used option -s to set a different random number seed for each of the 10 replicate runs we performed for each species. Finally, we repeated this analysis with just the New York populations of each species. We kept all parameters equal apart from the mixing parameters, which we adjusted to 0.8 (for -m), 0.5 (for -f), and 1.0 (for -a).

According to Meirmans et al. (2014), Bayesian deviance is the best criterion to use when selecting BayesAss results for reporting, versus averaging estimates among runs, as is commonly performed. We used Meirmans’s (2013) R code (available at https://onlinelibrary.wiley.com/doi/abs/10.1111/1755-0998.12216) to calculate Bayesian deviance based on the trace file outputs for each of the 40 runs we performed (four sets of 10 replicate runs). We retained the results of the single run in each replicate set with the lowest

Bayesian deviance. We ensured that the MCMC chain had successfully converged in each case through visual inspect of the trace plot in Tracer version 1.7.1 (appendix 2; Rambaut,

Drummond, Xie, Baele, & Suchard, 2018). As suggested by Rannala (2007), we also compared results among replicate runs to assess successful convergence.

RESULTS

DNA quality and quantity

Most (~90%) DNA samples had decent purity, with A260/A280 ratios ranging between

1.7 and 2.0. The ~10% of samples with A260/280 ratios falling outside this range generally had a

72 low concentration of DNA (< 10 ng/μl), which may have caused inaccuracy in the purity measurement. Gel electrophoresis revealed some degree of DNA degradation in most samples, which was not resolved through re-extraction using a modified procedure; this indicated that degradation likely occurred during storage of plant tissue samples at room temperature in silica gel. GCL used SPRI based size selection to eliminate low molecular weight DNA from 100 samples showing the most degradation, and standardized DNA concentrations across samples, which typically ranged from 2–100 ng/μl.

SNP calling and dataset filtering

Out of 1,103,595 sites and 170 individuals present in our total raw SNP file output by dDocent for N. boottii, we retained 388 sites (0.04%) and 74 individuals (44%) after performing all filtering steps (Table 3-1). For N. trifoliolatus, we retained 466 of 1,031,944 sites (0.05%) and

72 of 151 individuals (48%) after performing all filtering steps. Most sites and individuals were eliminated due to low coverage, likely the result of large genome size and small starting DNA quantities. The Hardy-Weinberg filter eliminated zero loci from the N. boottii dataset and six from the N. trifoliolatus dataset. We present details of our final datasets (full, equal population size, equal locus number) in Table 3-1.

Adaptive potential

Genetic diversity

We found greater genetic diversity in populations of the mountaintop obligate N. boottii than in populations of N. trifoliolatus. Expected heterozygosity (HE; gene diversity (Nei, 1973)) ranged from 0.187 to 0.237 and averaged 0.219 (SD = 0.015) for N. boottii, while for N.

73 trifoliolatus, HE ranged from 0.101 to 0.143 with an average of 0.124 (SD = 0.012) (Table 3-2).

The difference in observed heterozygosity (HO) was similarly pronounced: HO averaged 0.335

(SD = 0.025) for N. boottii and only 0.159 (SD = 0.019) for N. trifoliolatus (Table 3-2). HO was greater than HE for both species. Allelic richness, the average number of alleles per locus, averaged 1.449 (SD = 0.019) for N. boottii and 1.292 (SD = 0.022) for N. trifoliolatus, again revealing higher diversity within N. boottii. Finally, the percent of polymorphic loci (PPL, Table

3-2), calculated using the equal sample size/locus number subsets, ranged from 53–61% (average

57%, SD 3%) in N. boottii populations but only 29–40% (average 35%, SD 4%) in N. trifoliolatus populations. We did not find evidence of inbreeding in either species, with FIS ranging from -0.329 to -0.661 for N. boottii and from -0.063 to -0.410 in N. trifoliolatus (Table

3-2).

AMOVA revealed greater overall genetic variation in N. boottii versus N. trifoliolatus

(Table 3-3). The total sum of squares for N. boottii was 20% higher than that for N. trifoliolatus

(5839.0 vs. 4880.0) and the total variance was 15% higher (39.7 vs. 34.3). For both species, we found most genetic variance within populations, although the percentage was higher in N. boottii

(102.1; functionally 100%) than in N. trifoliolatus (92.1%). The AMOVA-generated FST values ranged from -0.021 (functionally 0) in N. boottii to 0.079 in N. trifoliolatus.

Finally, the number of multilocus genotypes exactly equaled the number of samples for both Nabalus boottii and Nabalus trifoliolatus (Table 3-2). Therefore, we failed to detect clones in our dataset, even for the smallest populations (e.g., NN-NY-GI with estimated 20 individuals,

Table 3-2).

74 Ploidy

Averaged across individuals for both N. boottii and N. trifoliolatus (a well-established diploid), absolute posterior probabilities of allelic proportions were highest for a 3:1 proportion of reference allele: alternate allele calls, followed by 1:1 and then 1:3. In relative terms, the ratio of posterior probabilities for 1:1 to 3:1 allelic proportions was much higher in N. trifoliolatus versus N. boottii (means before log-transformation 0.62 and 0.37, respectively) (P < 0.001, β =

0.52, R2 = 0.40, d.f. = 144; Figure 3-2), while the ratio of posterior probabilities for 1:3 to 1:1 allelic proportions was higher in N. boottii versus N. trifoliolatus (means before log- transformation 0.43 and 0.31, respectively) (P < 0.001, β = -0.55, R2 = 0.18, d.f. = 144; Figure 3-

2).

Under the assumption of one cytotype, the principal component analysis and discriminant analysis performed using gbs2ploidy grouped all individuals into one cluster for each species.

Under the assumption of two cytotypes, the PCA/DA grouped a handful of individuals from each species into a separate cluster. For N. boottii, seven of 74 individuals clustered into a separate cytotype grouping. Our examination of posterior probabilities of allelic proportions revealed that these seven individuals scored higher 1:1 probabilities relative to 3:1 probabilities than other N. boottii individuals. The seven individuals in question originated in six populations: NB-ME-HA,

NB-NH-AP, NB-NH-LC, NB-NH-NE, NB-NY-AL, and NB-NY-WF (described in Table 3-1).

For N. trifoliolatus, the 12 individuals that formed a separate cytotype cluster scored higher 1:3 posterior probabilities relative to 1:1 posterior probabilities versus the 60 other N. trifoliolatus individuals.

75 Migration potential

Historic migration

We found higher historic rates of migration in N. boottii as opposed to N. trifoliolatus.

Using genepop’s private alleles method, we calculated the number of migrants per generation

(Nm) as 0.92 across all populations of N. boottii and 0.81 across New York populations (Table

3-4). Conversely, for N. trifoliolatus, we calculated Nm as 0.75 across all populations and 0.58 for New York populations. Pairwise migration rates among New York populations of N. boottii varied from 0.44 (NB-NY-AR and NB-NY-WR) to 0.81 (NB-NY-GO and NB-NY-WF). The

Whiteface population of N. boottii (NB-NY-WF) exhibited the highest average pairwise migration rate within New York populations (average pairwise Nm = 0.75) while the Armstrong population of N. boottii exhibited the lowest (average pairwise Nm = 0.58). For New York N. trifoliolatus populations, pairwise migration rates varied from 0.34 (NN-NY-GI and NN-NY-

MA) to 0.53 (NN-NY-GI and NT-NY-RP). We found the highest average pairwise migration rate within New York populations of N. trifoliolatus in the Whiteface population (NN-NY-WF, average pairwise Nm = 0.48), while the Marcy population (NN-NY-MA) exhibited the lowest

(average pairwise Nm = 0.42).

Using Barton and Slatkin’s (1986) FST/GST method of migration rate calculation, we similarly found higher estimates of migration rate for N. boottii versus N. trifoliolatus populations (Table 3-5). At the species level, we calculated Nm using FST as 8.52 migrants per generation for N. boottii and only 2.34 migrants per generation for N. trifoliolatus (Table 3-5), with similar estimates for all New York populations (8.06 and 1.74, respectively). Pairwise Nm estimates using FST ranged from 4.08 (NB-NY-AL and NB-NY-WR) to 16.64 (NB-NY-GO and

NB-NY-WF) for N. boottii and from 1.16 (NN-NY-WR and NN-NY-RP) to 2.82 (NN-NY-MA

76 and NN-NY-WR) for N. trifoliolatus. For N. boottii, the Gothics population (NB-NY-GO) displayed the highest average pairwise Nm based on FST (10.69) while the Wright population

(NB-NY-WR) displayed the lowest (6.00). For N. trifoliolatus on the other hand, the Marcy population (NN-NY-MA) displayed the highest average Nm based on FST (2.17) while the Giant population (NN-NY-GI) displayed the lowest (1.45). Global FST was 0.029 for N. boottii and

0.097 for N. trifoliolatus (Weir & Cockerham, 1984).

Our calculations of Nm based on GST for N. trifoliolatus yielded higher pairwise estimates than those based on FST (Table 3-5), with the smallest calculated value 2.89 (versus

1.16; NT-NY-RP and NN-NY-WR) and the largest 7.81 (versus 2.82; NN-NY-MA and NN-NY-

WR). This pattern held for our calculation using global FST/GST across New York populations

(1.74/2.48 respectively) and at the species level (2.48/2.59 respectively); global GST for N. trifoliolatus was 0.088. We do not report Nm values calculated from GST for N. boottii as our GST values were slightly negative (e.g., global value for N. boottii = -0.026); GST values ≤ 0 yield undefined or nonsensical negative migration rate estimates.

Recent migration

Our BayesAss analysis for New York populations provided evidence of recent migration in all populations of N. boottii and three of five populations of N. trifoliolatus. For N. boottii, the estimated migrant proportion of populations (considering migration from the current and previous two generations) ranged from a low of 0.17 (95% CI approx. 0.06–0.28) for the

Algonquin population (NB-NY-AL) to a high of 0.23 (95% CI approx. 0.16–0.30) for the

Whiteface population (NB-NY-WF) (Figure 3-3A). These values derive from the best run out of

77 10 replicate BayesAss runs, as determined through Bayesian deviance (Meirmans, 2014); the second and third best runs yielded nearly identical statistics.

For New York populations of N. trifoliolatus, we detected no evidence of recent migration in two populations (NN-NY-GI and NN-NY-MA), and comparable migrant proportions to those of N. boottii populations in the remaining three N. trifoliolatus populations

(average 0.19 migrant proportion) (Figure 3-3B). The Whiteface population (NN-NY-WF) again exhibited the highest proportion of migrants at 0.22 (95% CI approx. 0.14–0.31). In contrast to our very consistent findings across BayesAss runs for N. boottii, we did find a few differences among the results of the top three BayesAss runs for N. trifoliolatus. Our second and third best runs estimated significant migration from the Marcy population (NN-NY-MA) into the Giant population (NN-NY-GI), whereas the best run estimated no significant migration into the Giant population. Additionally, our best run indicated Giant (NN-NY-GI) as the source of the migrant proportion of the Rocky Peak population (NT-NY-RP), while our second and third best runs indicated Marcy (NN-NY-MA) as the source. All other estimates remained consistent among the top BayesAss runs for N. trifoliolatus New York populations. We note that across both species, only two of our forty runs (and none of the best runs; appendix 2) showed evidence of potential convergence problems.

Figure 3-4 provides the geographic arrangement and origin of migrants detected in each

New York population of N. boottii and N. trifoliolatus. For N. boottii, we detected significant migration from Whiteface (NB-NY-WF) in the other four populations, and migration from

Gothics (NB-NY-GO) into Whiteface. For N. trifoliolatus, we detected Marcy (NN-NY-MA) as the source of migrants in the Whiteface (NN-NY-WF) and Wright (NN-NY-WR) populations, and Giant (NN-NY-GI) as the source of migrants in the Rocky Peak population (NT-NY-RP).

78 The gray sections of the pie charts represent uncertainty in the estimates, as each population had non-zero estimates of migration derived from other populations that we determined to be non- significant using the approximate 95% confidence intervals (two standard deviations, after

Rannala, 2007).

For individual migrant ancestries, we did not detect any current (generation 0) migrants in our New York datasets, but we did detect individuals with migrant ancestry from one or two generations ago. All N. trifoliolatus individuals sampled from the Whiteface (NN-NY-WF) and

Wright (NN-NY-WR) populations had migrant ancestry from Marcy (NN-NY-MA) one generation ago. Similarly, all individuals from the Rocky Peak population (NT-NY-RP) had migrant ancestry from Giant one generation ago. We also detected one second-generation migrant (migration occurred two generations ago) from the Whiteface population (NN-NY-WF) into Giant (NN-NY-GI) and one individual of 75% probability of being a second generation migrant from Giant (NN-NY-GI) in the Marcy population (NN-NY-MA), but neither of these estimates translated into significant migration at the population level. For the sampled New York individuals of N. boottii, BayesAss estimated that 100% were second-generation migrants.

Although we calculated species-wide migration in addition to our New York-only analysis for each species, we do not report the statistics here because we did not find consistency among replicate BayesAss runs. Overall, these runs yielded a smaller average migrant proportion for each population (0.09 for N. boottii; 0.08 for N. trifoliolatus).

DISCUSSION

We used population genomics to investigate the adaptive and migration potential of obligate and facultative mountaintop congeners in order to understand their historic and future

79 persistence in isolated “sky island” populations. Because facultative Nabalus trifoliolatus is a widespread species with potential connectivity between high and low elevation populations in at least some sites (Bogler, 2006; Haines et al., 2011), we expected to find greater migration potential and adaptive potential (genotypic/genetic diversity) in this species versus mountaintop obligate N. boottii. Our results instead revealed greater adaptive and migration potential in N. boottii, likely driven by tetraploidy in N. boottii. We discuss these findings and their implications for the historic and future persistence of N. boottii and N. trifoliolatus below.

Adaptive potential

Genetic diversity

Across all genetic diversity statistics, we found greater diversity in obligate mountaintop

N. boottii versus facultative mountaintop N. trifoliolatus (Table 3-2). Generally, studies demonstrate the opposite trend: greater genetic diversity in populations of widespread species versus narrow endemics or isolated relict populations (Gitzendanner & Soltis, 2000; Hampe &

Jump, 2011; Hampe & Petit, 2005; Hamrick & Godt, 1990, 1996). For example, in their recent ddRADseq study of endemic and widespread mountain buckwheats in the American west,

Lemon and Wolf (2018) found significantly greater heterozygosity in the widespread species.

Similarly, Zlonis and Gross (2018) discovered greater diversity (HE and HO) in relict populations of the arctic plant Euphrasia hudsoniana versus populations of invasive Euphrasia stricta.

Nevertheless, it is not unusual to find counterfactual results such as in our study; in their review, Gitzendanner and Soltis (2000) found that ~25% of rare species had higher population- level genetic diversity (PPL, AR, HE) than their widespread congeners. Recent examples of this pattern for relict/montane taxa include Lanes et al.'s (2018) finding of higher genetic diversity in

80 narrow-endemic versus widespread morning glories in Amazonian montane savannas, and

Martín-Hernanz et al.'s (2019) finding of higher diversity in two Helianthemum plants endemic to dolomite outcrops in the Spanish Sierra Nevada versus two widespread congeners. These studies highlight the importance of factors apart from geography and demography in determining a population’s genetic diversity (Ellegren & Galtier, 2016; Martín-Hernanz et al., 2019). Lanes et al. (2018) and Martín-Hernanz et al. (2019) attribute their findings in part to differences in mating system, including a higher rate of selfing in low-diversity populations. While we did indeed find slightly more outbreeding in N. boottii versus N. trifoliolatus, we instead attribute our unexpected finding of greater population-level genetic diversity in the mountaintop endemic

N. boottii to its probable tetraploidy (discussed further below).

Direct comparisons of genetic diversity statistics are best made among studies involving congeners and the same type of molecular marker (Allendorf et al., 2013; Gitzendanner & Soltis,

2000; Grueber, 2015). Genetic diversity is highly correlated among related species and can be highly variable among non-related species (Gitzendanner & Soltis, 2000), while different molecular markers yield vastly different estimates of genetic diversity (Allendorf et al., 2013).

No other population genetic studies exist for other Nabalus species (including both

RADseq/GBS and traditional techniques), limiting our ability to make direct comparisons.

Nevertheless, RADseq/GBS studies involving mountaintop/relict and widespread populations of other plant genera have generally recovered similar within-population diversity statistics as ours.

Zlonis and Gross (2018) reported average HE and HO as 0.129/0.189 (respectively) for relict populations of Euphrasia hudsoniana (lower than ours for N. boottii) versus 0.159/0.239 for the widespread invasive E. stricta (slightly higher than ours for N. trifoliolatus). Similarly, Lemon and Wolf (2018) found lower allelic richness in an endemic buckwheat (Eriogonum soredium)

81 than we did in N. boottii (1.27 vs. 1.449), but higher allelic richness in a common congener

(Eriogonum shockleyi) than we found in N. trifoliolatus (1.94 vs. 1.292). The average HE values reported by Lanes et al. (2018) for endemic and widespread morning glories (0.18 and 0.19 respectively) fall in between our values for N. trifoliolatus (0.124) and N. boottii (0.219). In contrast, Ren et al. (Ren et al., 2017) found PPL and HO values one to two orders of magnitude smaller than ours in the widespread Himalayan alpine perennial Primula tibetica, while Torres-

Martínez and Emery (2016), found slightly higher average within-population HE in an endemic

Asteraceae species (0.28, SE = 0.01) than we did for either species. However, like Harrison et al.

(2019) and Zlonis and Gross (2018), we found no evidence for reduced observed heterozygosity or increased inbreeding in the endemic/relict species. It appears, therefore, that populations of N. boottii and N. trifoliolatus harbor moderate levels of genetic diversity when compared with similar mountaintop/relict and widespread plant congeners analyzed via RADseq. Altogether, our findings suggest at least a moderate reservoir of genetic diversity within populations to support adaptation.

Regarding AMOVA-derived species-level genetic variance, we again found greater diversity in the obligate mountaintop plant N. boottii versus facultative mountaintop N. trifoliolatus (Table 3-3). Our estimates suggest that N. boottii harbors greater species-level genetic diversity (and likely, therefore, adaptive potential) than N. trifoliolatus. In both species, genetic variation was carried mostly within populations, although this percentage was higher in

N. boottii (functionally 100%) versus N. trifoliolatus (92%).

Gitzendanner and Soltis (2000) and Nybom (2004) report no difference in the partitioning of genetic variation within versus among populations in rare and widespread species. Differences are instead driven by breeding system, life form, successional status, and mode of seed/pollen

82 dispersal (Nybom, 2004). The variance estimates for our focal species are similar to those of

European Ash, which is long-lived and capable of long-distance dispersed via wind dispersed pollen and seed (Beatty et al., 2015). Conversely, AMOVA for mountaintop/relict species in the northeastern United States has often revealed a smaller proportion of genetic variation carried within populations than we found for our species. For example, Weber-Townsend (2017) found just 24% of genetic variation within populations of the rare fern Asplenium scolopendrium var. americanum; Bouchard et al. (2017) likewise found 22% of genetic variation carried within relict southern populations of the arctic fern Dryopteris fragrans. Robinson (2012), however, found proportions of within-population genetic variation similar to ours in mountaintop populations of the widespread mosses Sphagnum tenellum and S. pylaesii: 92% and 100%, respectively. For S. pylaesii, this high proportion was the result of complete clonality in high elevation populations;

S. tenellum, however, did display high overall genetic variance.

In theory, we would expect species that harbor a high proportion of their genetic variance within populations to exhibit greater adaptive potential, since diversity within populations is the raw material for natural selection; an exception is cases like S. pylaesii where the high proportion is caused by complete clonality and a total lack of genetic diversity. Fortunately, in the case of

Nabalus boottii and Nabalus trifoliolatus, we find a high proportion of within-population genetic variance and substantial genetic diversity within populations, indicating good adaptive potential.

With regard to clonality, we discovered equal and maximal genotype diversity in the obligate mountaintop plant Nabalus boottii and the facultative mountaintop plant Nabalus trifoliolatus. Although our strategy of sampling widely spaced individuals in each population was intended to maximize genetic and genotypic diversity in our datasets, we expected to discover at least some clones given the high frequency of clonal reproduction in alpine species,

83 including Nabalus spp. (Dullinger et al., 2012; Körner, 2003; Sayers, 1989), and the very small population sizes/areas for some of our sampled populations. The smallest populations of our focal species (NN-NY-GI and NB-NY-WR) each consisted of one clump, covering a ~0.25 m2 area and including just 20–67 basal leaves; even in these small, dense populations in which sampling widely spaced individuals was impossible, we failed to find a single duplicate genotype

(i.e., clone).

In contrast, other studies have found greater clonality in mountaintop/alpine plants, even when their methods were likewise not designed to capture clones. Pleuss and Stöcklin (2004), for example, actively attempted to avoid sampling clones of alpine Geum reptans by sampling individuals > 4 m apart, and yet they still recovered a few clones in their dataset. In two extreme examples, Bauert et al. (1998) and Robinson (2012) found 100% clonality in relict populations of the alpine plants Saxifraga cernua and Sphagnum pylaesii, respectively. N. boottii and N. trifoliolatus, therefore, maintain higher genotypic diversity (and therefore, adaptive potential) than some mountaintop plants, even in their smallest populations (but see Lushai, Loxdale, &

Allen, 2003 for discussion of adaptive potential in clonal organims).

Ploidy

Our results revealed a significant difference in ratios of allelic proportions between

Nabalus boottii and Nabalus trifoliolatus, with N. boottii’s proportions skewed more toward those typical of tetraploids and N. trifoliolatus’s skewed more toward diploid proportions (Figure

3-2). This result, coupled with the group assignment of gbs2ploidy’s PCA/DA, suggest that N. boottii is tetraploid, or majority tetraploid with mixed cytotype. This finding of probable tetraploidy in N. boottii corroborates the ploidy level given by Flora of North America for this

84 species (Bogler, 2006) and the early cytological study on which it was based (Löve & Löve,

1966). However, our results refute the more recent cytological work of Sayers (1989), who reported the species as diploid, unless N. boottii is indeed of mixed cytotype.

Nabalus boottii, if truly tetraploid, would have a significant adaptive advantage compared with diploid N. trifoliolatus (Levin, 1983; Mable, 2013; Sessa, 2019; Soltis et al., 2014; Weiss-

Schneeweiss et al., 2013). First, the fixed heterozygosity that results from whole genome duplication provides high species-level genetic diversity and helps small, isolated populations retain genetic diversity despite inbreeding (Brochmann et al., 2004; Kawecki, 2008; Van De Peer et al., 2017; Weiss-Schneeweiss et al., 2013); we see this greater diversity in N. boottii.

Furthermore, each migration event has greater impact in polyploids in terms of maintaining genetic diversity among populations, as each individual carries more alleles (Meirmans, Liu, &

Van Tienderen, 2018). Finally, the genomic flexibility afforded by a duplicate genome offers greater opportunities for adaptation, including evolution at redundant genes (Levin, 1983; Mable,

2013; Sessa, 2019; Soltis et al., 2014; Van De Peer et al., 2017; Weiss-Schneeweiss et al., 2013).

Altogether, there is a multitude of ways by which tetraploidy likely confers N. boottii greater adaptive potential than N. trifoliolatus.

Our finding of probable tetraploidy in N. boottii should be interpreted cautiously for a few reasons. First, package gbs2ploidy is relatively new and has only been used in a few published studies (Burns, Hedin, & Tsurusaki, 2018; Gompert & Mock, 2017; Siadjeu, Mayland-

Quellhorst, & Albach, 2018). It performs well but not perfectly in assigning cytotype (> 90% accuracy in most cases) (Gompert & Mock, 2017). Additionally, our method of comparing relative allelic proportions is novel, and should be validated in other studies. Normally, researchers could infer cytotype from absolute rather than relative allelic proportions; however,

85 in our case, both N. boottii and known diploid N. trifoliolatus exhibited slightly higher posterior probabilities for 3:1 allelic proportions (typical of tetraploids) versus 1:1 allelic proportions

(typical of diploids). We suspect that in our study, the DNA degradation present in many of our samples may have led to biases or stochasticity in amplification and sequencing of alleles, resulting in a slight bias toward the reference allele in both species (Graham et al., 2015).

Finally, high heterozygosity (one of the ways gbs2ploidy determines cytotype) has a few other potential causes than polyploidy: it could be caused by paralogs (e.g., due to ancient genome duplication in Asteraceae) or recent population bottlenecks (McKinney, Waples, Pascal, Seeb, &

Seeb, 2018; McKinney, Waples, Seeb, & Seeb, 2017; Piry, Luikart, & Cornuet, 1999). However, because (1) we adjusted parameters and used filters to avoid paralogs in our datasets, (2) we observed lower heterozygosity in N. trifoliolatus despite its shared history of ancient genome duplication in Asteraceae, and (3) because populations of N. boottii have been relatively stable or even increasing over the last several generations (Prout, 2005), these causes are less likely than tetraploidy to have generated higher heterozygosity in N. boottii.

Our finding of tetraploidy in N. boottii is supported by Löve and Löve (1966), who determined N. boottii to be tetraploid using flow cytometry of a few specimens from Mount

Washington. Furthermore, Löve and Löve (1967) discovered polyploidy in 64% of mountaintop vascular plants in northeastern North America, versus only 45% in plants from corresponding low elevation sites. Given that N. boottii’s range is restricted to mountaintops of the Northeast while the majority of N. trifoliolatus’s range is in low elevation areas, our finding of probable tetraploidy for the former and diploidy for the latter are in accordance with community trends in ploidy. Further support comes from a recent study which demonstrated that ploidy can be determined in some species merely by comparing heterozygosity among individuals; although

86 our samples constitute two different species, the strong difference in heterozygosity between our two congeners also supports tetraploidy in N. boottii (Larsen et al., 2018).

Migration potential

Contrary to our hypothesis, we found equal to higher rates of historic and current migration in the obligate mountaintop species Nabalus boottii than in facultative mountaintop N. trifoliolatus. However, estimates of gene flow—particularly those based on FST—may be upwardly biased in polyploids, which typically exhibit lower equilibrium values of FST, are less prone to drift, and experience a greater impact with each migration event (Meirmans et al.,

2018). Because N. boottii is likely tetraploid, we should consider our migration estimates based on Barton and Slatkin's (1986) FST method as potentially upwardly inflated compared with those for N. trifoliolatus. Our migration estimates based on the private alleles method may also be upwardly biased for N. boottii given the method’s dependence on allele frequencies, which exhibit less divergence among populations for polyploids (Meirmans, 2014); however, this method tends to be less biased in general as compared with the FST method (Allen, Amos,

Pomeroy, & Twiss, 1995; Allendorf et al., 2013).

Because our estimation of recent migration using BayesAss was non-model based (with regard to population genetic models; Wilson & Rannala, 2003), our results are likely less biased by N. boottii’s probable tetraploidy. Nevertheless, results from BayesAss analyses can be upwardly biased for a different reason: convergence issues (Meirmans, 2014). Visual inspection of MCMC trace files (appendix 2) indicated convergence in all but two of our 40 runs (neither being a best run based on BIC). Furthermore, the general consistency of our results across the three best replicate runs for New York populations of each species also indicated convergence

87 (Rannala, 2007). However, our consistent estimation of non-migrant proportions around 67%, especially for our full datasets, indicated possible convergence problems: the MCMC chain may have been stuck at the lower bound of the prior (Meirmans, 2014). Our New York-only analyses

(especially for N. trifoliolatus) were less affected by this issue, probably because BayesAss performs better when datasets include fewer populations (Meirmans, 2014). Additionally, we followed Meirmans et al.’s (2014) suggested practice of presenting results only from the best run of each set in order to increase the accuracy of our results. Nonetheless, we conservatively suggest interpreting our estimates of recent migration in terms of their approximate 95% confidence intervals rather than their exact estimates (Figure 3-3), recognizing that the true value may be closer to the lower end of the 95% confidence intervals. Even so, our findings still conservatively indicate a 5% or greater migrant proportion in all New York N. boottii populations and three of five N. trifoliolatus populations. Because a greater number of N. boottii populations show evidence of recent migration than N. trifoliolatus populations, we conclude that N. boottii, at least in New York, may have a slightly higher overall rate of migration versus

N. trifoliolatus. Altogether, we conclude that the migration potential of N. boottii, based on historic and recent migration, is equal to or slightly higher than that of N. trifoliolatus.

For Nabalus boottii and Nabalus trifoliolatus, migration of propagules and genetic material is probably achieved both through insect-dispersed pollen and wind-dispersed seed.

Radio telemetry has revealed foraging pollinators (bees) travelling up to six km per day

(Kissling, Pattemore, & Hagen, 2014), which would certainly allow for more localized gene flow within mountain ranges for N. boottii and N. trifoliolatus var. nanus, or between high and low elevation populations N. trifoliolatus. N. boottii in particular attracts a wide array of pollinators,

88 including bees (Tetreault & Burgess, 2019), many of which would be capable of long-distance flight.

In terms of seed dispersal, updrafts created by mountain topography coupled with the high winds common at high elevation likely enable the long-distance dispersal of Nabalus’s seed, via its feathery pappus (Tackenberg, Poschlod, & Kahmen, 2003). Our data support migration between separate mountain peaks within New York, most of which are < 10 km apart

(maximum distance is 28 km). Regional studies support migration at this distance as entirely possible: an empirical study demonstrated on Whiteface Mountain from at least 5 km away

(Miller & McDaniel, 2004), while a theoretical study based on island biogeography theory strongly supported ongoing immigration to explain plant distributions among 13 Adirondack mountain peaks (Riebesell, 1982). Extreme weather events, birds, and humans may also contribute to long-distance dispersal in Nabalus spp. (Nathan, 2006; Nathan et al., 2008); these particular dispersal forces operate over scales of 10-1 to 103 kilometers. As recreational use reaches all-time highs in the northeastern mountains, the role of humans in dispersing alpine seed between mountain peaks could be particularly significant (Eastman, 2018).

The typical benchmark for migration rate in conservation-focused studies is one migrant per generation, a rate high enough to prevent the loss of genetic diversity but low enough to allow divergence of allele frequencies in populations (Spieth, 1974). However, Mills and

Allendorf (1996) argued that the one migrant per generation rule is based on a number of simplifying assumptions, and suggested one as an acceptable minimum, with the ideal range being one to 10 migrants per generation. Our rates of historical migration straddle the one migrant per generation line; thus, it is difficult to say if N. boottii and N. trifoliolatus are experiencing a theoretically “ideal” migration rate, especially considering the uncertainty

89 inherent in model-based approaches to estimating migration (Hutchison & Templeton, 1999;

Whitlock & Mccauley, 1999); they appear, however, to at least approach the ideal range. What we can say with high certainty is that populations of both species have experienced some degree of historic and recent migration (excepting current migration in NN-NY-GI and NN-NY-MA). In other words, they are not reproductively isolated, as one might presume from their highly isolated geographies. Thus migration potential likely plays an important role in enabling mountaintop populations of N. boottii and N. trifoliolatus to maintain genetic diversity and (re-

)colonize sites.

Understanding historic persistence and dynamics

Our results shed light on the historic persistence and dynamics of N. boottii and N. trifoliolatus on mountaintops of the northeastern United States. The moderate to high migration estimates in these species explain their ability to shift to lower latitudes and elevations during glacial periods and re-colonize mountaintop sites during interglacial periods. At a minimum, this would have required migration (over generations) of at least ~400 km and perhaps 1000 km or more following the Last Glacial Maximum (LGM), as the Northeast mountains were completely ice-covered during the LGM and few to no refugia existed within the extent of the ice sheet

(Bierman et al., 2015; Dyke et al., 2002). As noted by Brochmann et al. (2003), endemic arctic- alpine species of northeastern North America must have greater dispersal abilities than generally thought in order to have accomplished these migrations. Dispersal over open, barren habitats during re-colonization, hitchhiking on animals, and storms may help explain long-distance postglacial dispersal (Nathan, 2006; Nathan et al., 2008).

90 Although dispersal over distances of up to 103 kilometers is not impossible for N. boottii and N. trifoliolatus according to some estimates (Nathan, 2006; Nathan et al., 2008), others put the maximum figure for most plants under 40 km (Cain, Milligan, & Strand, 2000). Since N. boottii and N. trifoliolatus var. nanus are absent or exceedingly rare (respectively) in Canadian alpine areas, we presume that the maximum dispersal distance for Nabalus spp. is closer to the

40 km figure. Because gaps greater than 40 km separate many mountaintop populations and population groups of N. boottii and N. trifoliolatus (especially N. boottii), we suspect that populations of these species used to have greater connectivity following the last glacial retreat, and that vicariance helps explain their current distribution. N. trifoliolatus has mostly maintained a widespread and connected range, but N. boottii appears to have become restricted to high elevations sites as the climate warmed, with at least some populations becoming extirpated. A combination, therefore, of regional dispersal and range-wide vicariance likely explains the geographic distribution of these species.

Altogether, migration has likely assisted the historic persistence of Nabalus boottii and

Nabalus trifoliolatus. Range wide, migration has enabled these species to shift between mountaintops and refugia during glacial cycles. Within mountain ranges, migration has likely helped mountaintop populations maintain genetic diversity and (re-)colonize sites. In fact, the individual migrant ancestries estimated by BayesAss indicated that migration may have enabled the persistence of these species in the recent past as well. Our estimates indicate that migrants from the Marcy population (NN-NY-MA) one generation ago may have colonized the Whiteface

(NN-NY-WF) and Wright (NN-NY-WR) populations of Nabalus trifoliolatus var. nanus, while the Rocky Peak non-alpine population (NT-NY-RP) may have been colonized from the alpine

Giant (NN-NY-GI) population one generation ago.

91 Additionally, the moderate (N. trifoliolatus) to high (N. boottii) adaptive potential of our focal species also helps explain their historic persistence in mountaintop populations. The relatively high genetic diversity we find in these species today is likely comparable to historic levels, suggesting that these species historically harbored the diversity necessary to avoid inbreeding and enable adaptation during times of environmental change. The probable tetraploidy of N. boottii in particular has likely helped this species historically to maintain genetic diversity. This maintenance of diversity, together with stress-tolerator life history strategies (e.g., long life spans), likely explain how the narrow mountaintop endemic Nabalus boottii and mountaintop N. trifoliolatus var. nanus populations have avoided the extinction vortex (Gilpin & Soulé, 1986; Grime, 1977).

Future persistence in a changing world

Just as our results provide insight into the historic persistence of mountaintop populations of Nabalus boottii and Nabalus trifoliolatus, so too do they inform these species’ outlook for the future. Our finding of moderate to high rates of historic and recent migration in these species offers several benefits in terms of future persistence. First, it will likely enable these species to re-colonize sites that have been locally extirpated due to an acute environmental stress (e.g., trampling, drought, etc.), thus maintaining a more stable metapopulation (Allendorf et al., 2013;

Thrall, Richards, Mccauley, & Antonovics, 1998). Second, migration may allow these species to shift their range away from an environmental stress (Brook et al., 2008; Tilman et al., 2017). In the case of climate change, however, more northerly mountain ranges are likely beyond the dispersal capacity of these species, barring inadvertent human dispersal or an extreme weather event (Nathan et al., 2008). Nevertheless, if managers decided to undertake assisted migration of

92 N. boottii to a more northern mountain range, its intrinsic migration ability would likely help the species to spread within its new range. Finally, migration will likely allow mountaintop populations of N. boottii and N. trifoliolatus to maintain genetic diversity within populations, and enable further evolutionary adaptation.

With regard to adaptation, our finding of moderate to high genotypic/genomic diversity in these species suggests strong adaptive potential in the face of global environmental changes

(Harrisson et al., 2014). The rare mountaintop endemic N. boottii in particular should harbor very good adaptive potential due to its probable tetraploidy (Levin, 1983; Mable, 2013; Sessa, 2019;

Soltis et al., 2014; Weiss-Schneeweiss et al., 2013). Indeed, as a polyploid, N. boottii may generally be better equipped to tolerate environmental extremes and survive environmental change (Cai et al., 2019; Dar & Rehman, 2017; Fawcett et al., 2009; Kawecki, 2008; Levin,

1983; Sessa, 2019; Van De Peer et al., 2017; Vanneste et al., 2014). Although the generation time of N. boottii is relatively long (likely 5–10 years or more; Körner, 2003), its ability to maintain genetic diversity ensures that its populations retain the raw materials for adaptation, even if it occurs more slowly than in faster-reproducing species. Additionally, the substantial phenotypic plasticity for functional traits we discovered for both N. boottii and N. trifoliolatus

(Chapter 2) will likely help buffer populations of these species faced with change, buying them time for a long-term response through evolutionary adaptation including the acquisition of beneficial mutations (Chevin et al., 2010; Jump & Peñuelas, 2005; Nicotra et al., 2010; Price,

Qvarnström, & Irwin, 2003). And in fact, although longevity may slow the rate of adaptation in these species, it may still be beneficial for surviving environmental change (Morris et al., 2008).

Altogether, our findings indicate good migration and adaptive potential for our two focal species, and especially for the obligate mountaintop Nabalus boottii, which will likely help

93 mountaintop populations of N. boottii and N. trifoliolatus persist in the face of rapid global environmental change (Chevin et al., 2010; Jump & Peñuelas, 2005). Given their geographic isolation, we would expect these species to be highly vulnerable to environmental change, especially climate change (Elsen & Tingley, 2015; Marris, 2007; Urban, 2018); indeed, many mountaintop species are quite vulnerable (Costion et al., 2015; Dirnböck et al., 2011; Dullinger et al., 2012; Freeman et al., 2018; Marris, 2007). However, our study, like others (e.g., De Witte,

Armbruster, Gielly, Taberlet, & Stöcklin, 2012), demonstrates probable resilience to environmental change in mountaintop species and emphasizes the importance of evaluating factors beyond geography in determining vulnerability.

Nevertheless, we still recommend caution to managers of these mountaintop taxa, particularly globally rare Nabalus boottii. Although these taxa appear to harbor good migration and adaptive potential, the necessity of simultaneously responding to multiple environmental changes and their synergistic effects could overwhelm their ability to respond (Brook et al.,

2008; De Boeck, Bassin, Verlinden, Zeiter, & Hiltbrunner, 2016). We expect many species will reach tipping points in the coming decades, and that “extinction debts” will start being paid off

(Alexander et al., 2018; Dullinger et al., 2012; Jiménez-Alfaro et al., 2016). While N. trifoliolatus sensu lato is not in immediate danger given its very broad distribution, we recommend continued monitoring of the globally rare species N. boottii as a precaution.

94 Table 3-1. Sampling information for genomic datasets of Nabalus boottii (obligate mountaintop species) and Nabalus trifoliolatus (facultative mountaintop species). Latitude and longitude values indicate the approximate central point of sampling, while elevation values indicate the approximate average elevation of sampling locations within each site (in meters). N designates the number of samples entering the ddRADseq pipeline. Nf designates the number of samples retained in the final full dataset for each species after filtering, while Ns gives the number of samples retained in the even sample size datasets. We created datasets for N. trifoliolatus which included the full number of loci retained after filtering (466; set A) as well as a set with the same number of loci as the N. boottii sets (338, set B) in order to facilitate comparison among the two species for certain statistics that could be affected by locus number. We note that we have classified the NT-NY-RP population as non-alpine despite its fairly high elevation because it occurred below treeline in mixed forest; all Nabalus boottii and Nabalus trifoliolatus var. nanus populations occurred above treeline. Note: Lat/Long information redacted from the open access version of this dissertation to comply with rare species permitting requirement. Table appears on the following page.

95 Table 3-1. (Continued from previous page.)

Site abbrev Site name Lat Long Elev N Nf Ns

Nabalus boottii NB-ME-BB Boundary Bald 1081 10 4 4 NB-ME-BX Baxter (Katahdin) 1409 10 5 4 NB-ME-HA Hamlin (Katahdin) Lat/Long 1421 10 4 4 NB-NH-AP Alpine Garden information 1616 10 5 4 NB-NH-CP Cow Pasture redacted from 1735 10 5 4 NB-NH-ED Edmands Col Cutoff 1508 10 4 4 open access NB-NH-LC Lakes of the Clouds version to 1524 10 6 4 NB-NH-MO Monroe comply with 1579 10 4 4 NB-NH-NE Eisenhower rare species 1360 10 5 4 NB-NY-AL Algonquin permitting 1554 10 3 NB-NY-AR Armstrong requirements 1352 10 4 4 NB-NY-GO Gothics 1413 10 5 4 NB-NY-WF Whiteface 1437 30 14 4 NB-NY-WR Wright 1371 10 4 4 NB-VT-CH Camel's Hump 1208 10 2 Total indiv 170 74 52 Total pops 15 15 13 Total loci 338 338 Nabalus trifoliolatus var. nanus NN-ME-GE Goose Eye 1178 10 5 4 NN-ME-WE West Peak 1251 10 3 NN-NH-LC Lakes of the Clouds 1543 10 6 4 NN-NH-TU Tuckerman Ravine 1490 10 3 NN-NH-WA Washington Auto Road 1470 10 4 4 NN-NY-GI Giant 1386 10 5 4 NN-NY-MA Marcy 1610 10 6 4 NN-NY-WF Whiteface 1463 10 7 4 NN-NY-WR Wright 1384 10 3

Nabalus trifoliolatus (non-alpine) NT-ME-BR Brunswick 25 10 4 4 NT-ME-CA Canton 154 10 3 NT-ME-TO Topsham 38 10 4 4 NT-NY-RP Rocky Peak Ridge 1160 10 4 4 NT-VT-BU Burlington 76 10 8 4 NT-VT-RI Ripton 378 10 7 4 Total indiv 150 72 44 Total pops 15 15 11 Total loci set A 466 466 Total loci set B 338 338

96 Table 3-2. Diversity statistics for Nabalus boottii (obligate mountaintop) and Nabalus trifoliolatus (facultative mountaintop). MLG gives the number of detected multi-locus genotypes in our datasets, while N indicates the sample size for each population. Other abbreviations include PPL (percent polymorphic loci), AR (allelic richness), HE (expected heterozygosity), HO (observed heterozygosity), and FIS (inbreeding coefficient). We calculated PPL using the equal sample size/equal locus number subsets; for all other calculations, we used the full datasets.

Pop N MLG PPL AR HE HO FIS Nabalus boottii NB-ME-BB 4 4 54% 1.428 0.208 0.329 -0.493 NB-ME-BX 5 5 59% 1.454 0.227 0.347 -0.436 NB-ME-HA 4 4 61% 1.461 0.219 0.305 -0.336 NB-NH-AP 5 5 58% 1.468 0.231 0.324 -0.329 NB-NH-CP 5 5 57% 1.458 0.228 0.352 -0.436 NB-NH-ED 4 4 57% 1.437 0.212 0.323 -0.444 NB-NH-LC 6 6 59% 1.446 0.224 0.315 -0.316 NB-NH-MO 4 4 53% 1.439 0.216 0.354 -0.559 NB-NH-NE 5 5 57% 1.460 0.228 0.339 -0.408 NB-NY-AL 3 3 – 1.417 0.190 0.281 -0.426 NB-NY-AR 4 4 55% 1.454 0.223 0.362 -0.551 NB-NY-GO 5 5 61% 1.471 0.235 0.363 -0.445 NB-NY-WF 14 14 54% 1.425 0.226 0.335 -0.331 NB-NY-WR 4 4 60% 1.486 0.237 0.379 -0.513 NB-VT-CH 2 2 – 1.432 0.187 0.320 -0.661

Mean 4.9 4.9 57% 1.449 0.219 0.335 -0.446 SD 2.7 2.7 3% 0.019 0.015 0.025 0.098

Nabalus trifoliolatus NN-ME-GE 5 5 33% 1.299 0.129 0.174 -0.270 NN-ME-WE 3 3 – 1.274 0.116 0.171 -0.410 NN-NH-LC 6 6 40% 1.312 0.134 0.162 -0.151 NN-NH-TU 3 3 – 1.252 0.101 0.109 -0.103 NN-NH-WA 4 4 33% 1.291 0.124 0.166 -0.288 NN-NY-GI 5 5 36% 1.303 0.132 0.169 -0.220 NN-NY-MA 6 6 29% 1.288 0.128 0.168 -0.242 NN-NY-WF 7 7 36% 1.305 0.136 0.158 -0.120 NN-NY-WR 3 3 – 1.261 0.106 0.135 -0.261 NT-ME-BR 4 4 38% 1.320 0.134 0.177 -0.270 NT-ME-CA 3 3 – 1.294 0.117 0.158 -0.319 NT-ME-TO 4 4 36% 1.298 0.127 0.179 -0.335 NT-NY-RP 4 4 31% 1.258 0.110 0.157 -0.358 NT-VT-BU 8 8 41% 1.325 0.143 0.164 -0.098 NT-VT-RI 7 7 35% 1.300 0.129 0.140 -0.063

Mean 4.8 4.8 35% 1.292 0.124 0.159 -0.234 SD 1.7 1.7 4% 0.022 0.012 0.019 0.105

97 Table 3-3. Analysis of molecular variance (AMOVA) results for Nabalus boottii (obligate mountaintop) and Nabalus trifoliolatus (facultative mountaintop) as calculated in program Arlequin using the full dataset for each species.

Nabalus boottii Source of Sum of Variance Percentage d.f. variation squares components of variation Among 14 452.3 Va -0.8 -2.1 populations Within 133 5386.8 Vb 40.5 102.1 populations

Total 147 5839.0 39.7 100.0

Nabalus trifoliolatus Source of Sum of Variance Percentage d.f. variation squares components of variation Among 14 801.9 Va 2.7 7.9 populations Within 129 4078.1 Vb 31.6 92.1 populations

Total 143 4880.0 34.3 100.0

98 Table 3-4. Migration rate estimates calculated using the private alleles method in genepop for New York populations of our Nabalus boottii (obligate mountaintop) and Nabalus trifoliolatus (facultative mountaintop). Table values represent estimates of Nm, the number of migrants per generation. We give averages for each population in bold on the diagonal. The overall migration rate for New York populations and for each species is given in the right-most columns. We calculated all values using the even population size datasets (N = 4) to which we added NB-NY- AL and NN-NY-WR (N = 3) for the New York-only subset. Table 3-1 provides full population names and information for each population abbreviation.

Nabalus boottii NB-NY-AL NB-NY-AR NB-NY-GO NB-NY-WF NB-NY-WR Overall NY Overall sp.

NB-NY-AL 0.66

NB-NY-AR 0.55 0.58

NB-NY-GO 0.75 0.73 0.71 0.81 0.92

NB-NY-WF 0.78 0.61 0.81 0.75

NB-NY-WR 0.55 0.44 0.57 0.79 0.59

Nabalus trifoliolatus NN-NY-GI NN-NY-MA NN-NY-WF NN-NY-WR NT-NY-RP Overall NY Overall sp.

NN-NY-GI 0.44

NN-NY-MA 0.34 0.42

NN-NY-WF 0.45 0.49 0.48 0.58 0.75

NN-NY-WR 0.46 0.48 0.50 0.46

NT-NY-RP 0.53 0.37 0.48 0.40 0.44

Table 3-5. Migration rate estimates for New York populations of Nabalus boottii (obligate mountaintop) and Nabalus trifoliolatus (facultative mountaintop) as calculated using the FST/GST method of Barton and Slatkin (1986). Table values represent estimates of Nm, the number of migrants per generation. We used Weir and Cockerham’s FST values to calculate values under the diagonal and Nei and Chesser’s GST for values above the diagonal (Nei & Chesser, 1983; Weir & Cockerham, 1984). The “Avg lower” row gives population averages based on FST while the “Avg upper” column gives population averages based on GST. The two right-most columns provide the overall values for New York populations and each species. We calculated all values using the full dataset (uneven sample sizes). We omitted GST-based Nm values for Nabalus boottii as they were negative. Table 3-1 provides full population names and information for each population abbreviation.

Nabalus boottii

Overall Overall NB-NY-AL NB-NY-AR NB-NY-GO NB-NY-WF NB-NY-WR NY sp. NB-NY-AL –

NB-NY-AR 4.96 –

F F NB-NY-GO 10.12 9.75 – ST ST based: based: NB-NY-WF 8.00 7.47 16.64 – 8.06 8.52

NB-NY-WR 4.08 4.37 6.26 9.18 –

Avg lower 6.79 6.64 10.69 10.32 6.00

Nabalus trifoliolatus

Overall Overall NN-NY-GI NN-NY-MA NN-NY-WF NN-NY-WR NT-NY-RP Avg upper NY sp. NN-NY-GI – 3.25 3.07 3.12 4.02 3.36

F F NN-NY-MA 1.40 – 6.64 7.81 4.22 5.48 ST ST based: based: NN-NY-WF 1.42 2.81 – 4.92 3.78 4.60 1.74 2.34 G G NN-NY-WR 1.38 2.82 2.35 – 2.89 4.68 ST ST based: based: NT-NY-RP 1.61 1.67 1.70 1.16 – 3.73 2.48 2.59

Avg lower 1.45 2.17 2.07 1.93 1.53 –

100 N. boottii 15 sites

N. trifoliolatus var. nanus 9 sites

N. trifoliolatus (non-alpine) 6 sites

Figure 3-1. Collection sites for Nabalus boottii (obligate mountaintop) and Nabalus trifoliolatus (facultative mountaintop) samples used in the genomic analysis. Sites for N. boottii (top panel) comprise most of the known populations for the species apart from a few small populations. Nabalus trifoliolatus var. nanus collection sites span the geographic range of the variety. Non- alpine Nabalus trifoliolatus collection sites represent a similar geographic range as the other taxa, although the range of N. trifoliolatus extends throughout much of eastern North America.

101 0.7 0.7 *** A *** B

0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2 Ra#o of posterior probabili#es for Ra#o of posterior probabili#es for

es#mated allelic propor#ons (1:1 vs. 3:1) 0.1 es#mated allelic propor#ons (1:3 vs. 1:1) 0.1

0.37 0.62 0.43 0.31 0 0 NB NT NB NT Taxon Taxon

Figure 3-2. Ratios of posterior probabilities for 1:1 versus 3:1 allelic proportions (A) and 1:3 versus 1:1 allelic proportions (B) at heterozygous loci for Nabalus boottii and Nabalus trifoliolatus. We note that here the NT abbreviation stands for the whole species, instead of the non-alpine variety, as in other figures and tables. Asterisks (***) denote significant differences (P < 0.001) in log-scaled ratios of allelic proportions based on linear models in program R. Tetraploids are expected to exhibit a higher probability for 1:3 and 3:1 allelic proportions at heterozygous loci versus diploids, which typically show highest probability for 1:1 proportions.

102 A N. boottii – New York

NB-NY-AL WF

NB-NY-AR WF

NB-NY-GO WF

NB-NY-WF GO

NB-NY-WR WF

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Migrant proportion

B N. trifoliolatus – New York

NN-NY-GI

NN-NY-MA

NN-NY-WF MA

NN-NY-WR MA

NT-NY-RP GI

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 Migrant proportion

Figure 3-3. Estimates for the proportion of recent migrants (zero to two generations from present) for New York populations of Nabalus boottii (obligate mountaintop) and Nabalus trifoliolatus (facultative mountaintop) as calculated using BayesAss 3.0.4. We present estimated proportions for the best run (out of ten) for each species based on Bayesian deviance. We did not detect statistically significant migration for the NN-NY-GI or NN-NY-MA populations for the best N. trifoliolatus run. Gray bars represent the estimate while error bars give +/- two standard deviations, which approximate 95% confidence intervals (Rannala, 2007). Two-letter codes to the right of each bar provide the population of origin for migrants, corresponding to the last two letters of the full population abbreviations given on y axis. Table 3-1 provides full population names and information for each population abbreviation. We used the New York data from our full datasets in this analysis.

103 AL AR GO WF WR O

NN-NY-GI

GI MA WF WR RP O

Figure 3-4. (Figure caption appears on the following page.)

104 Figure 3-4. (Figure appears on the previous page.) Estimated migrant/non-migrant proportions for New York populations of Nabalus boottii (obligate mountaintop; top panel) and Nabalus trifoliolatus (facultative mountaintop; bottom panel) as calculated using BayesAss 3.0.4. We present estimated proportions for the best run (out of ten) for each species based on Bayesian deviance. NN-NY-MA and NN-NY-GI are comprised entirely of non-migrants, while the remaining populations are comprised of a mix of migrants and non-migrants. We lumped non- significant (but non-zero) migrant estimates into the “O” (other) category for each population. The sampling locations above represent almost all known high elevation occurrences of each species within New York. The small yellow polygon in the inset map shows the geographic extent of these 10 sampling locations, which were all located in the high peaks region of the Adirondack Mountains. Table 3-1 provides full population names and information for each population abbreviation.

105 CHAPTER 4: DEFINING EVOLUTIONARY SIGNIFICANT UNITS AND CONSERVATION PRIORITIES IN NORTHEASTERN NABALUS PLANT SPECIES

ABSTRACT

Amid Earth’s current biodiversity crisis, identifying conservation priorities is a task of paramount importance. Endemic taxa are potentially more vulnerable to extinction, and their evaluation is therefore of primary concern. In this study, we used morphometrics and population genomics to identify evolutionary significant units and populations of high conservation priority in two endemic alpine plants (Nabalus boottii and Nabalus trifoliolatus var. nanus) and their non-alpine counterpart (non-alpine N. trifoliolatus). Morphological and genomic evidence did not support alpine N. trifoliolatus var. nanus as an evolutionary significant unit distinct from non-alpine N. trifoliolatus, indicating low conservation priority for populations of this overall widespread species. Likewise, we failed to find evidence for multiple evolutionary significant units within alpine N. boottii based on its almost non-existent population genetic structure.

Populations of N. boottii also harbored few private alleles when compared with N. trifoliolatus populations, again revealing their genetic indistinctiveness. Our findings suggest that N. boottii can be appropriately conserved at the species level. Conservation actions should target a minimum of two populations, which is sufficient to capture 99.9% of species-level genetic diversity in this endemic species.

INTRODUCTION

Earth is in the midst of a biodiversity crisis (Barnosky et al., 2011; Ceballos et al., 2015;

Pimm et al., 2014). Even conservative estimates indicate that current extinction rates are 8–100 times higher than background extinction rates, signaling the arrival of a sixth mass extinction

106 (Ceballos et al., 2015). Unlike previous mass extinction events, the major forces driving this crisis are all anthropogenic in nature; among others, these include land-use change, climate change, invasive species, overharvest, and pollution (Sala et al., 2000; Tilman et al., 2017).

These elements of global change act separately and synergistically, and currently threaten roughly 25% of species with extinction (Brook et al., 2008; International Union for Conservation of Nature and Natural Resources, 2019). With so many species at risk, defining appropriate conservation units and identifying individual populations of highest priority for at-risk species are tasks of paramount importance.

Conservation units are groups of organisms that are viewed as distinct for the purposes of conservation (Allendorf et al., 2013; Funk, McKay, Hohenlohe, & Allendorf, 2012). Usually conservation units are defined below the species level (but see Ahrens et al., 2017). Commonly used conservation units include evolutionary significant units (Ryder, 1986; Waples, 1991) and management units (MUs), (Moritz, 1994); other examples include relevant genetic units for conservation (Caujapé-Castells & Pedrola-Monfort, 2004; Pérez-Collazos, Segarra-Moragues, &

Catalán, 2008), operational conservation units (Doadrio, Perdices, & Machordom, 1996), fundamental geographic and evolutionary units (Riddle & Hafner, 1999), and functional conservation units (Maes, Vanreusel, Talloen, & Van Dyck, 2004).

Among defined conservation units, evolutionary significant units (ESUs) are the most widely recognized, and even serve as the basis for protected vertebrate “discrete population segments” under the US Endangered Species Act (USFWS & NOAA, 1996). ESUs represent evolutionarily and ecologically important lineages of species that have arisen through long-term reproductive isolation and/or adaptive differentiation (Waples, 1991; Waples & Lindley, 2018).

Authors have emphasized different criteria for the establishment of ESUs. Most definitions have

107 emphasized both genetic and ecological distinctiveness or non-exchangeability (reviewed in

Allendorf, Luikart, & Aitken, 2013 and Fraser & Bernatchez, 2001); however, ESUs can also be defined solely using genetic or non-genetic characteristics to determine distinctiveness (Moritz,

1994; Vogler & Desalle, 1994). The recommended best practice is to embrace a flexible definition for ESUs, using one or more criteria as they are available and appropriate to each situation (Allendorf et al., 2013; Fraser & Bernatchez, 2001; Peñas, Barrios, Bobo-Pinilla,

Lorite, & Martínez-Ortega, 2016).

Genetic distinctiveness is typically determined using molecular analyses of population structure, phylogeny, or gene flow (Fraser & Bernatchez, 2001; Moritz, 1994). Ecological distinctiveness is determined through a variety of means, including the comparisons of morphology, life history, and behavior among populations and environmental conditions among their respective sites (Allendorf et al., 2013; Crandall et al., 2000; Moritz, 1994). Common garden and reciprocal transplant studies are particularly useful for determining ecological distinctiveness by standardizing environmental conditions and revealing genetically determined differences among populations (Allendorf et al., 2013).

Increasingly, conservation researchers are using genomics versus traditional genetic methods for defining genetic distinctiveness in conservation unit determination (Allendorf,

Hohenlohe, & Luikart, 2010; Corlett, 2017; Funk et al., 2012; McMahon et al., 2014). Reduced- representation genomic sequencing methods such as genotyping by sequencing (GBS; Elshire et al., 2011) and restriction-site associated DNA sequencing (RADseq; Baird et al., 2008) have paved the way for genomic analysis of non-model species. Additionally, as the costs of next- generation sequencing decline and analysis pipelines become more robust, genomic techniques are becoming more tenable for conservation researchers (Corlett, 2017). These techniques offer

108 many benefits, including eliminating the time-intensive process of developing reliable molecular markers for each species, and greatly increasing the accuracy and precision of results through the generation of hundreds to hundreds of thousands of markers (Allendorf et al., 2010; Corlett,

2017; Funk et al., 2012). This greater resolution allows researchers to detect more subtle population structure than is often possible with traditional molecular markers (McCartney-

Melstad, Vu, & Shaffer, 2018), and correct past interpretations that were based on fewer markers

(Zhang, Li, & Li, 2018). Finally, by targeting both neutral and non-neutral loci, genomic techniques like RADseq and GBS offer researchers the opportunity to investigate both neutral differentiation driven by demography, and adaptive differentiation driven by selection, both of which are relevant for the designation of ESUs (Allendorf et al., 2010; Funk et al., 2012).

In this study, we used genomic and morphological data to investigate the presence of

ESUs and identify conservation priorities in two sympatric plant species: Nabalus trifoliolatus

(three-leaved rattlesnake-root) and Nabalus boottii (Boott’s rattlesnake-root). These species occur entirely (Nabalus boottii) or partially (Nabalus trifoliolatus) within the alpine zone of the northeastern United States, where N. trifoliolatus is represented by an alpine variety (var nanus).

The northeast alpine zone is an area that covers just 0.01% of the land area of New York,

Vermont, New Hampshire, and Maine (30 km2), but contributes enormously to the region’s biodiversity (Capers et al., 2013). Northeast alpine plants are threatened by several elements of global change, including nitrogen deposition, recreational overuse, and invasive species, but are perhaps most threatened by climate change given their occurrence on isolated mountaintops

(Baumgardner et al., 2003; Capers et al., 2013; Capers & Slack, 2016; Freeman et al., 2018;

Galloway et al., 1984; Marris, 2007; Reay et al., 2008; Urban, 2018). As environmental conditions and lower-elevation species shift to higher elevations under climate change, they may

109 drive mountaintop species extinct (Marris, 2007; Urban, 2018). Mountaintop endemics are particularly vulnerable to extinction, with local extinctions already reported for some species including American pika, Quino checkerspot butterfly, and tropical birds (Beever et al., 2016;

Freeman et al., 2018; Marris, 2007; Parmesan, Williams-Anderson, Moskwik, Mikheyev, &

Singer, 2015; Urban, 2018). The northeast alpine zone is home to several mountaintop endemic plants: Potentilla robbinsiana (Robbins’ cinquefoil), Geum peckii (mountain avens), Nabalus boottii (Boott’s rattlesnake-root), and Nabalus trifoliolatus var. nanus (Capers et al., 2013; Jones

& Willey, 2012).

Nabalus boottii is a species of high conservation concern, known from fewer than 20 sites and ranked as globally imperiled (G2) by NatureServe (NatureServe, 2018). This species is state listed as endangered in all states in which it occurs: New York, Vermont, New Hampshire, and Maine (Maine Natural Areas Program, 2015; New Hampshire Natural Heritage Bureau,

2018; Vermont Natural Heritage Inventory, 2015; Young, 2019). N. boottii was considered for federal listing under the Endangered Species Act, but was removed from the federal candidate list in the 1990s due to lack of evidence that the species was declining or in danger of extinction

(Susi von Oettingen, pers. comm.). Current conservation efforts for this species include intermittent surveys by conservation/land managing organizations and at least one seed banking effort by the New England Wildflower Society (B. Brumback, pers. comm.). Based on our field observations, populations of N. boottii appear similar in their morphology, phenology, and environmental preferences, suggesting that N. boottii is likely comprised of a single ESU.

However, an investigation of genetic distinctiveness is needed to help confirm the absence of multiple ESUs within N. boottii. Additionally, below the species/ESU level, identifying

110 individual priority populations for conservation efforts is needed to help guide current and future management of this globally rare species.

Nabalus trifoliolatus var. nanus, the alpine variety of Nabalus trifoliolatus, was historically treated as a distinct species (Prenanthes nana; Syn: Nabalus nanus) by some authors due to its morphological differences versus non-alpine N. trifoliolatus (Gleason & Cronquist,

1963; Milstead, 1964; Mitchell & Tucker, 1997; Sayers, 1989). Based on early discoveries of fertile morphological intermediates (Löve & Löve, 1966) and the more recent discovery of morphological gradients over elevational gradients on Mount Katahdin, most authors now recognize the high elevation form as a variety of N. trifoliolatus or simply part of N. trifoliolatus without distinction (Bogler, 2006; Haines et al., 2011). Following this reclassification, natural heritage and conservation agencies halted monitoring of N. trifoliolatus var. nanus (J. Goren, pers. comm.), because the species as a whole is broadly distributed and globally secure (G5)

(NatureServe, 2018). However, the distinctiveness (or lack thereof) of N. trifoliolatus var. nanus has not been rigorously tested using experimental or genetic techniques. Furthermore, in most areas, significant geographic separation exists between high and low elevation populations, with no clines in morphology. This geographic separation may have resulted in genetic divergence between high and low elevation populations or local adaptation to alpine versus non-alpine sites.

Given the vastly different habitat and climate preferences of alpine versus non-alpine N. trifoliolatus as well as their morphological differences, it is possible that N. trifoliolatus var. nanus constitutes a distinct evolutionary legacy within N. trifoliolatus (i.e., an ESU), with genetic and morphological distinctiveness. If determined to be an ESU, N. trifoliolatus var. nanus would merit conservation consideration: like N. boottii, it is a narrowly distributed mountaintop endemic that may be vulnerable to extinction.

111 We took a comparative approach in this study, performing almost all analyses with both

N. boottii and N. trifoliolatus despite our somewhat divergent objectives for each species

(detailed below), because comparison with a congener can provide important context for results

(Gitzendanner & Soltis, 2000; Grueber, 2015; Jiménez-Alfaro et al., 2016). Comparative approaches are especially useful for genetic/genomic analyses when the two species, as in this case, display significant differences in their ecology, range size, rarity, etc.

Our primary goal in this study was to inform the conservation of Nabalus trifoliolatus

(including its non-alpine and alpine var nanus varieties) and Nabalus boottii using genomic and morphological data. Specifically, we aimed to address the following questions:

1. Does Nabalus trifoliolatus var. nanus, the alpine variety of N. trifoliolatus, qualify as an

ESU based on genomic and morphological evidence?

2. Do we find genomic evidence for multiple ESUs within N. boottii?

3. How many and which individual populations should be our priorities for conservation in

N. boottii based on genomic distinctiveness and diversity?

METHODS

Study taxa

Nabalus boottii DC. (Syn: Prenanthes boottii (DC.) Gray) is a monocarpic perennial plant of the family Asteraceae native to the highest elevations (1000–1800 m a.s.l.) of New

York, Vermont, New Hampshire, and Maine (Bogler, 2006). N. boottii is likely a tetraploid or mixed diploid/tetraploid species, based on our analysis (Chapter 3) and previous cytological work (Löve & Löve, 1966; Sayers, 1989). Nabalus trifoliolatus Cass. (Syn: Prenanthes trifoliolata (Cass.) Fern.), also a monocarpic perennial (Sayers, 1989), is widespread throughout

112 eastern North America and occurs from sea level to high elevation. A habitat generalist, N. trifoliolatus inhabits woodland, cliff, sandy, and saline areas. Nabalus trifoliolatus Cass. var. nanus Bigelow (Syn: Prenanthes trifoliolata (Cass.) Fern. var. nana Bigelow), the alpine variety of N. trifoliolatus, is the only high elevation congener of N. boottii and can be found from 1100 to 1600 m a.s.l. in New York, New Hampshire, and Maine. Historically, N. trifoliolatus var. nanus was treated as a separate species (Nabalus nanus (Bigelow) DC., Prenanthes nana

(Bigelow) Torr. ex DC.; Milstead, 1964; Mitchell & Tucker, 1997; Sayers, 1989) based on its morphological and habitat differences. Morphological differences include more deeply divided leaves, darker involucral bracts, and shorter height: mature flowering plants at low elevation can be as tall as 150 cm (typically at least 90 cm when in flower), while those at high elevation can be as short as 10 cm (typically 30 cm or less) (Bogler, 2006; Haines et al., 2011; Sayers, 1989).

Here, we refer to the two varieties as “Nabalus trifoliolatus var. nanus” (the alpine variety, abbreviated NN in figures and tables) and “non-alpine Nabalus trifoliolatus” (abbreviated NT) while we reserve the unqualified “Nabalus trifoliolatus” to refer to the species as a whole.

Morphological distinctiveness

We investigated the morphological distinctiveness of N. trifoliolatus var. nanus and non- alpine N. trifoliolatus using a controlled common garden experiment in a growth chamber (based in Syracuse, NY) to search for genetically-based morphological differences between the two varieties (question 1). We included N. boottii in the experiment to contextualize our findings given the morphological variability within and similarity among Nabalus species (Sayers, 1989).

As starting material for the experiment, we used the same seed source as that described in

Chapter 2, stratified at the same time and in the same manner. Altogether, our seed represented

113 25+ seed families for two populations each of N. trifoliolatus var. nanus and non-alpine N. trifoliolatus, and 50 seed families from one population of N. boottii. We included only a single population of N. boottii due to permitting constraints, which eliminated the possibility of searching for morphological evidence of multiple ESUs within N. boottii but did allow for contextualization of our N. trifoliolatus findings.

On 7–8 March 2017, we removed seeds from cold stratification and prepared them for germination by adding 30 mL of a 500 ppm gibberellic acid (GA) solution to each bagged seed family, which contained seeds plus moist potting medium (50/50 mixture of vermiculite and perlite). After adding the GA solution, we spread each seed family mixture evenly over pots filled with a 50/50 mixture of vermiculite and peat-based potting mix (Sunshine ® Redi-Earth;

Sun Gro Horticulture, Agawam, MA). We placed pots in 1020 trays, covered trays with clear humidity domes, and placed trays in a greenhouse set to 20º C during the day/10º C at night with

15-hour days achieved through supplemental lighting. These conditions represented the average preferred germination conditions of alpine Asteraceae (Baskin & Baskin, 2014). We randomized tray position at the start of germination and every two days thereafter until the end of germination.

Seeds germinated from 10–16 March 2017. Within 48 hours of germinating, we transplanted seedlings into individual 6.4 cm pots (retaining up to two seedlings per seed family), and moved pots into trays in a growth chamber (Puffer-Hubbard Environmental Chamber model

CEC 36-10, Thermo Fisher Scientific, Asheville, NC). We set growth chamber conditions at

15.6ºC during the day and 10ºC at night, with lighting for the 15-hour days supplied by both fluorescent and incandescent bulbs at 238 μmol s-1 m-2 of photosynthetically active radiation

(photosynthetic photon flux density) at plant height, as measured using an AccuPAR PAR/LAI

114 Ceptometer (Model LP-80, METER Group, Inc., Pullman, WA). On 16 March, when all 105 seedlings had been moved to the growth chamber, we randomized seedling position within trays and tray position within the chamber. We kept plants covered by humidity domes for the first few days, and then removed domes. Once and later twice per week, we soak-watered plants to saturation with distilled water; twice per week we rotated tray position within the chamber.

On 8 May 2017 (53–59 days after germination; similar in length to the alpine growing season), we retrieved the 99 surviving plants from the growth chamber, removed soil from the plants’ taproots, and allowed plants to imbibe distilled water overnight in dark, refrigerated conditions. We scanned, weighed, dried, and re-weighed plants using the same methods described in Chapter 2. In the end, we obtained measurements for 14 functional traits: height, root length, dry mass (shoot, root, total), total leaf area, leaf number, root to shoot ratio, specific leaf area, specific root length, leaf coloration (red, green, blue), and leaf shape. We selected these traits based on their ecological relevance and their suitability for measurement in young plants

(Funk et al., 2017; Pérez-Harguindeguy et al., 2013; Wright et al., 2004).

Many of these functional traits were ones we also investigated in Chapter 2, using identical measurement methodology. A few traits were new or were measured using different methodology. To measure root length, we used ImageJ (Schneider et al., 2012) to draw and measure the length of a segmented line following the midpoint of the taproot for each scanned plant image. To measure total leaf area (also used in specific leaf area), we used ImageJ to convert scanned plant images to 8-bit grayscale and used automatic thresholding to convert grayscale images to black-and-white (plants appearing black against a white background). Then, we selected all rosette leaves and measured leaf area (including petioles) in cm2. We also measured shape in a different manner than in Chapter 2, by measuring straight-line leaf length

115 and width (cm) of each unfolded leaf in ImageJ, measuring length at the longest point and width at the widest point. We divided length by width to obtain a metric of leaf shape for each leaf, and averaged shape values across all leaves of each individual plant.

We compared functional trait measurements among N. trifoliolatus var. nanus, non- alpine N. trifoliolatus, and N. boottii individuals using linear mixed models and principal component analysis (PCA) in program R (R core team, 2017). For linear models, we used package lme4 to build models and package lmerTest to provide P-values according to

Satterthwaite’s approximation (Bates et al., 2015; Kuznetsova et al., 2017). We created one model per trait, with taxon as the explanatory variable and seed family as a random effect. For certain traits, we log-transformed data to improve the normality of residuals and heterogeneity of variance. We used package emmeans to calculate pairwise differences in estimated marginal means among taxa using the Tukey method for P-value adjustment (Lenth, 2018).

We performed a principal component analysis using function prcomp in program R, centering and scaling data for all 14 traits (R core team, 2017). For this analysis, we used a subset of our full trait dataset, to avoid issues with missing data: the subset included data for 63 individuals. We used function autoplot from R package ggfortify to plot our PCA results over the first seven principal components (Tang, Horikoshi, & Li, 2016). We repeated this analysis using just N. trifoliolatus individuals, without N. boottii.

Genetic distinctiveness

We assessed population structure in N. trifoliolatus and N. boottii to investigate the presence of ESUs within these species (questions one and two). For comparative purposes, we employed three different methods to assess structure, which we discuss in further detail below:

116 Unweighted Pair-Group Method with Arithmetic mean (UPGMA) clustering, Discriminate

Analysis of Principal Components (DAPC), and program STRUCTURE. Our molecular data included genomic datasets produced through double-digest restriction-site associated DNA sequencing (ddRADseq) of each species (Baird et al., 2008; Peterson et al., 2012). We provide full methods for sampling, ddRADseq, and data processing in Chapter 3. In brief, our sampling included ≥ 10 individuals from 15 populations of each species; it spanned the geographic extent of N. boottii and N. trifoliolatus var. nanus, with sampling of non-alpine N. trifoliolatus covering a similar geographic area. For N. boottii, we sampled almost all known populations, apart from a few small populations where we were not permitted to collect samples.

After data processing and filtering, we produced several final datasets. Our full datasets included 72–74 total individuals representing all 15 populations of each species, with 338 retained single nucleotide polymorphisms (SNPs) for N. boottii and 466 for N. trifoliolatus

(Table 3-1). We also produced equal sample size datasets for use in analyses in which uneven sample sizes for populations would yield biased results, including program STRUCTURE

(Puechmaille, 2016). Our equal sample size datasets included 13 populations and 52 individuals for N. boottii and 11 populations and 44 individuals for N. trifoliolatus, after dropping populations with fewer than four individuals. Finally, for our full and even population size datasets for N. trifoliolatus, we also created datasets of equal locus number as N. boottii (338 loci) to enable comparisons across species for certain analyses.

UPGMA clustering

UPGMA is a model-based clustering method based on genetic distance that assumes equal rates of evolution, and therefore produces dendrograms with equal branch lengths. We

117 produced individual- and population-based UPGMA dendrograms using package poppr version

2.8.2 (Kamvar et al., 2015, 2014) in program R version 3.6.0 (R core team, 2017). We used functions aboot and plot.phylo to generate trees using Nei’s genetic distance (Nei, 1972, 1978) and 1000 bootstraps.

DAPC and k-means clustering

Discriminate Analysis of Principal Components (DAPC) is a non-model based (in terms of evolution) multivariate approach to genetic clustering for the investigation of population structure (Jombart & Collins, 2015). Being non-model based, DAPC is free from assumptions about Hardy-Weinberg equilibrium or linkage disequilibrium, and is therefore appropriate for a wider variety of studies including those of polyploids (such as probable tetraploid N. boottii).

Based on simulation data, DAPC often performs better than the leading model-based approach,

STRUCTURE (Jombart, Devillard, & Balloux, 2010). We performed DAPC using package adegenet in program R (Jombart, 2008; Jombart & Ahmed, 2011; Jombart & Collins, 2015; R core team, 2017). We produced the genlight input files from our full dataset raw VCF files using program poppr version 2.8.2 (Kamvar et al., 2015, 2014). We transformed our data to principle components using principal component analysis (PCA); we then performed k-means clustering using the function find.clusters to determine the optimal number of clusters based on Bayesian

Information Criterion (BIC). Using the group assignments from the k-means clustering, we performed DAPC using function dapc, choosing to retain 35 principal components (the inflection point of the curves) and one linear discriminant function.

118 STRUCTURE

STRUCTURE, a Bayesian model-based clustering program, is the most widely used method for detecting population genetic structure (Pritchard, Stephens, & Donnelly, 2000;

Puechmaille, 2016). STRUCTURE can be biased or perform non-optimally in certain circumstances, including when sample sizes are uneven, samples include close relatives, or isolation by distance exists (Pritchard, Wen, & Falush, 2010; Puechmaille, 2016). By using our equal population size datasets and sampling widely spaced individuals in each population, we avoided biases associated with uneven sample sizes and the inclusion of close relatives. To ensure isolation by distance did not affect our results, we performed a Mantel test in program R version 3.6.0 (R core team, 2017).

To perform the Mantel test, we used package adegenet version 2.1.1 (Jombart, 2008;

Jombart & Ahmed, 2011) to create genetic distance matrices using three methods: Nei’s distance

(non-Euclidean; Nei, 1972), Reynold’s distance/coancestrality coefficient (Euclidean; Reynolds,

Weir, & Cockerham, 1983), and Prevosti’s distance (non-Euclidean; Prevosti, 1974). We created geographic distance matrices for populations of each species using the program Geographic

Distance Matrix Generator version 1.2.3 (Ersts, n.d.). To perform the Mantel tests for each pair of genetic and geographic distance matrices, we used function mantel.randtest in package ade4 and specified 10,000 permutations (Chessel, Dufour, & Thioulouse, 2004).

For our analysis in program STRUCTURE, we converted our raw VCF files into the correct input format using program PGDSpider version 2.1.1.5 (Lischer & Excoffier, 2012). We specified 250,000 MCMC steps with a 100,000 step burn-in for 10 replicate runs at each level of k (1–13 for N. boottii, 1–11 for N. trifoliolatus; based on number of included populations), each with a different random number seed. Given the almost non-existent population structure we

119 detected in early trials, we specified correlated allele frequencies among populations (Falush,

Stephens, & Pritchard, 2003) and used sampling locations as an informative prior (Hubisz,

Falush, Stephens, & Pritchard, 2009). Due to the possible tetraploidy of N. boottii, we analyzed this species both as a diploid and as a tetraploid, following the instructions in Pritchard et al.

(2010) and Meirmans et al. (2018).

For our three main analyses (N. trifoliolatus as diploid, N. boottii as diploid, N. boottii as tetraploid), we used CLUMPAK (Kopelman, Mayzel, Jakobsson, Rosenberg, & Mayrose, 2015) to determine the optimal number of clusters (k) according to the Evanno method (Evanno,

Regnaut, & Goudet, 2005) and Pritchard’s maximum log probability method (Pritchard et al.,

2000). We considered all best estimates of K as potentially informative.

Conservation priorities for N. boottii

Establishing conservation priorities below the species or ESU level––at the management unit or individual population level––is important for guiding management decisions and conservation actions of rare or threatened species (Moritz, 1994). Because Nabalus boottii is a globally rare species with fewer than 20 existing populations, we aimed to determine priority populations for conservation by investigating how many populations were needed in order to preserve N. boottii’s genetic diversity, and which individual populations were most valuable for conservation based on their genetic diversity or distinctiveness (question 3). Determining the number and identity of priority populations will help guide conservation efforts, such as the seed banking undertaken by the New England Wildflower Society (B. Brumback, pers. comm.).

120 How many populations to conserve

To determine the number of populations needed in order to conserve 99% of N. boottii’s genetic diversity (a threshold used by Caujapé-Castells & Pedrola-Monfort, 2004 and Ceska,

N Affolter, & Hamrick, 1997), we used the formula P = 1 – FST , where P is the proportion of genetic diversity preserved in N number of populations based on global FST (fixation index, a measure of population differentiation) (Ceska et al., 1997; Hamrick, Godt, Murawski, &

Loveless, 1991; Pérez-Collazos et al., 2008). Ceska et al.’s (1997) original equation used GST

(fixation index corrected for multiple alleles), but we chose to use Pérez-Collazos et al.’s (2008)

FST-based version given the slightly negative GST results we obtained for N. boottii (discussed in

Chapter 3) which would have interfered with the formula. We included N. trifoliolatus in this analysis for comparative purposes with N. boottii so we could cross-validate our FST-based results with GST-based results. We used function basicStats from R package diveRsity to calculate global FST (and GST, for N. trifoliolatus) based on our full datasets (Keenan et al., 2013;

Nei & Chesser, 1983; R core team, 2017; Weir & Cockerham, 1984).

Which populations to conserve

A. Private alleles

Populations that harbor rare or unique (private) alleles are particularly valuable for conservation, as they represent reservoirs of potentially important genetic diversity for the future evolution of the species (Bengtsson, Weibull, & Ghatnekar, 1995; Caujapé-Castells & Pedrola-

Monfort, 2004; Ceska et al., 1997; Peñas et al., 2016). We used the function private_alleles in package poppr version 2.8.2 (Kamvar et al., 2015, 2014) to compute the number of private alleles per population using program R version 3.6.0 (R core team, 2017), after first converting

121 our raw VCF files to genid format. We performed this analysis four times for each species, using our full and equal population size datasets, and specifying to include or exclude dosage from allele counts. For these analyses, we used the equal locus number N. trifoliolatus datasets (Table

3-1) so that private allele numbers would be comparable across the two species.

B. Correlates of diversity

To further determine which N. boottii populations were most valuable for conservation, we investigated the relationship between genetic diversity and population size/flowering rate.

Such a relationship would not only shed light on the demographic factors that may be important for the maintenance of genetic diversity in N. boottii populations, but would also help guide conservation efforts in populations we were not able to include in our genetic sampling. We used the number of basal leaves plus the number of flowering individuals as a proxy for true population size; this method is standard for the species (J. Goren, pers. comm.). For most populations, we performed a direct count to determine population size and percent flowering. For five of the six largest populations (each with 3800+ individuals), we determined population size indirectly: we counted basal leaves and flowering individuals in a 2 m by 2 m test patch, counted all flowering individuals in the population, and used the flowering rate of the test patch to estimate the overall population size.

We compared population size and flowering rate to the genetic diversity statistics we calculated in Chapter 3 (PPL, AR, HE, HO, FIS) using linear models in program R (R core team,

2017). We used package lme4 to build the models and lmerTest to generate P-values according to Satterthwaite’s approximation (Bates et al., 2015; Kuznetsova et al., 2017; R core team, 2017).

To improve the normality of residuals and homogeneity of variance for our models, we log-

122 transformed population size. Each of our 10 models consisted of logged population size or percent flowering as an explanatory variable and one of the five diversity statistics as a response variable.

RESULTS

Morphological distinctiveness

Only one of the 14 traits assessed, red coloration of leaves, showed significant differentiation (α = 0.05) between N. trifoliolatus var. nanus and non-alpine N. trifoliolatus

(Table 4-1). Non-alpine N. trifoliolatus leaves were redder than N. trifoliolatus var. nanus leaves by 16.92 units (SE = 6.77, P = 0.040) or 6.6% on a 255-unit scale. In contrast, we found significant species-level differences between N. boottii and N. trifoliolatus var. nanus and/or non-alpine N. trifoliolatus in six of 14 traits: height, total leaf area, leaf number, root to shoot ratio, specific root length, and leaf shape. We discovered marginally significant differences between N. boottii and the N. trifoliolatus taxa in an additional three traits (root length, red and green coloration). We failed to detect significant differences between any taxa in dry mass

(shoots, roots, total) and specific leaf area. We found certain traits (e.g., height and root to shoot ratio, Figure 4-1A,B) for which the median value for N. trifoliolatus var. nanus was intermediate to that of N. boottii and non-alpine N. trifoliolatus (although not significantly so); we found others (e.g., leaf shape and root length, Figure 4-1C,D) for which the median value for N. trifoliolatus var. nanus was instead quite similar to non-alpine N. trifoliolatus.

Our principal component analysis (PCA) revealed similar results, showing some species- level differences between N. boottii and N. trifoliolatus but no discernable differences between the N. trifoliolatus varieties along the first seven principal components, which together explained

123 > 90% of the variance (Table 4-2 for factors and loadings, Figure 4-2 for PCA plots). We found species-level separation along PC3 between N. boottii and both varieties of N. trifoliolatus, with almost no overlap in data points between the two species. The factors most strongly associated with PC3 included specific leaf area, specific root length, and red coloration, closely followed by root length and height (Table 4-2, Figure 4-1E,F,G,D,A). PC3 explained 12.3% of the variance in our data, with the remainder of variance explained by principal components that showed little to no separation between any taxa. Even when N. boottii was excluded from the PCA, we still found no discernable differentiation between the two varieties of N. trifoliolatus (appendix 3).

Genetic distinctiveness

UPGMA clustering

UPGMA clustering revealed little structuring for N. boottii individuals, with limited bootstrap support for most nodes in our individual-based UPGMA dendrogram (Figure 4-3).

Most nodes showing > 50% bootstrap support grouped two to three individuals sampled from the same population, although in one case there was > 50% support for a grouping that included individuals derived from two sampling locations (NB-ME-HA and NB-NH-ED). There was

100% bootstrap support for an individual from the Alpine Garden population (NB-NH-AP) as the most basal individual in the tree. In general, individuals did not cluster strongly according to population or geographic region.

For N. trifoliolatus as a whole, UPGMA clustering revealed slightly greater genetic structure among populations and smaller genetic distances among individuals versus N. boottii.

We found > 50% bootstrap support for more nodes of the individual-based UPGMA dendrogram and stronger clustering by population versus N. boottii, but the same lack of significant clustering

124 by geographic region (Figure 4-4). Additionally, we failed to find significant clustering by variety (non-alpine vs. alpine var nanus). We found 100% bootstrap support for two individuals from the Tuckerman population (NB-NH-TU) as the most basal individuals in the tree.

In our population-based UPGMA dendrograms, we again found relatively little genetic structuring among populations, as evidenced by low bootstrap support for most nodes (Figure 4-

5). For both trees, the node splitting the most basal group from the remainder of the tree was the best supported, at 100% for each. The Algonquin population (NB-NY-AL) was most basal in the

N. boottii tree, while the Tuckerman population (NN-NH-TU) was most basal in the N. trifoliolatus tree. Each tree contained one node with over 50% bootstrap support that linked two geographically close populations (NB-NY-WF and NB-NY-GO for N. boottii; NT-NY-RP and

NN-NY-GI for N. trifoliolatus); however, both also contained at least one node with > 50% support uniting populations from three to four different states. As above, populations did not cluster according to variety for N. trifoliolatus. For these population-based dendrograms, we found overall greater genetic distance among populations of N. trifoliolatus versus N. boottii.

DAPC and k-means clustering

K-means clustering with function find.clusters indicated k = 1 as optimal based on BIC for both N. boottii and N. trifoliolatus; however, as DAPC cannot be performed with k < 2 and given the somewhat arbitrary nature of choosing the optimal k (Jombart & Collins, 2015), we proceeded with k = 2, the next best value. The two clusters identified in N. boottii contained 49 and 25 individuals. Of the 15 N. boottii populations, 13 included individuals from both clusters.

Only Camel’s Hump (NB-VT-CH) and Monroe (NB-NH-MO) had individuals all belonging to the same cluster. For N. trifoliolatus, our results revealed slightly more structure. One cluster

125 contained the majority of individuals (62) and populations. The other contained 10 individuals, including all five individuals from Giant (NN-NY-GI), all four individuals from Rocky Peak

(NT-NY-RP), and one individual from Marcy (NN-NY-MA-12).

STRUCTURE

The Mantel tests did not reveal significant isolation by distance in either species, validating our use of program STRUCTURE (P = 0.06–0.24 for N. trifoliolatus and P = 0.08–

0.23 for N. boottii, depending on distance method). Using CLUMPAK, we determined k = 4 as the optimal number of clusters for the 11 populations of N. trifoliolatus making up our equal population size dataset (Figure 4-6A). Both the Evanno method and Pritchard’s method identified k = 4 as optimal (Evanno et al., 2005; Pritchard et al., 2000). Apart from the adjacent

Giant and Rocky Peak populations (NN-NY-GI and NN-NY-RP), populations did not cluster according to geographic proximity. They also did not cluster according to variety, with all clusters composed of both alpine (NN-) and non-alpine (NT-) populations, apart from the cluster formed solely by the Marcy population (NN-NY-MA). We found little to no between-cluster admixture in certain populations (NN-NY-GI, NN-NY-WF, NT-ME-TO, NT-NY-RP), and a higher degree of admixture in others (NN-ME-GE, NT-VT-RI).

For the diploid analysis of N. boottii, both methods of estimation determined k = 2 as the optimal number of clusters (Figure 4-6B). All individuals showed admixed ancestry and belonged to the same majority cluster. Therefore, STRUCTURE failed to detect population structure in the diploid analysis of N. boottii. For the tetraploid analysis of N. boottii, we report results from the three best estimates of k, as Prichard’s method and the Evanno method yielded different estimates. The Evanno method determined k = 3 as optimal, closely followed by k = 5

126 (ΔK = 4.076 vs. ΔK = 4.008, Figure 4-6C,D). Pritchard’s method identified k = 8 as optimal

(Figure 4-6E). Similar to our diploid analysis, at all three levels of k, we found between-cluster admixture in all populations, and found that all populations belonged to the same majority cluster. Under the tetraploid analysis, we found certain individuals with majority ancestry different from the dominant cluster, especially one individual in the Eisenhower population (NB-

NH-NE). Overall, we found greater but still weak population structure with the tetraploid analysis versus the diploid analysis of N. boottii.

Conservation priorities for N. boottii

How many units to conserve

We found that just one population was necessary to preserve 97% of species-level genetic diversity in N. boottii, while just two were necessary to reach the target of 99% of species-level diversity (Figure 4-7). Even to achieve the higher 99.9% threshold targeted by some researchers

(Peñas et al., 2016; Pérez-Collazos et al., 2008; Segarra-Moragues & Catalán, 2010), two populations were sufficient. For N. trifoliolatus, we found that one population would preserve slightly less genetic diversity (90%); however, like N. boottii, just two populations were necessary to reach the 99% target. Three populations were needed to achieve the higher 99.9% threshold. Our GST-based cross-validation revealed very similar results for N. trifoliolatus as our

FST-based results described above, with one population preserving 91% of species-level genetic diversity, two populations 99%, and three populations 99.9%.

127 Which units to conserve

A. Private alleles

For N. boottii, the Baxter population (NB-ME-BX) was the most unique in terms of private alleles, ranking among the top three populations for private allele number under every analysis scheme (Table 4-3). The Hamlin (NB-ME-HA), Alpine Garden (NB-NH-AP),

Armstrong (NB-NY-AR) and Wright (NB-NY-WR) populations followed Baxter in uniqueness, each ranking among the top three populations for private allele number under two of the four analysis schemes.

Comparing the two species, N. trifoliolatus harbored more private alleles than N. boottii across a variety of metrics (Table 4-3). We found three to five times more total private alleles in

N. trifoliolatus versus N. boottii datasets (e.g., 66 versus 21 for the no-dosage, full datasets).

Similarly, we found on average three to five times more private alleles per population in N. trifoliolatus versus N. boottii (e.g., 4.4 versus 1.4 for the no-dosage, full datasets). We also found a greater proportion of populations harboring private alleles for N. trifoliolatus: 93–100% depending on dataset, versus 77–80% for N. boottii.

B. Correlates of diversity

Our overall population size estimate for N. boottii was 129,517, representing almost all known occurrences of the species and all of the largest populations (Table 4-4). On average, individual populations contained 8,634 individuals, 1.94% of which were flowering. Population size ranged from 67 (Wright, NB-NY-WR) to 53,625 (Cow Pasture, NB-NH-CP), while flowering rate ranged from 0.03% (Boundary Bald, NB-ME-BB) to 4.74% (Algonquin, NB-NY-

AL). New Hampshire contained approximately 88% of the individuals in our sampled

128 populations, which represent almost all known populations for the species. New York, in contrast, contained only 4.7% of the individuals in our sampled populations. Based on our linear models, we found that neither population size nor flowering rate were significant predictors (α =

0.05) of genetic diversity (PPL, AR, HE, HO) or inbreeding (FIS) in N. boottii populations (Table

4-5). We also found no relationship between population size and flowering rate (P = 0.875).

DISCUSSION

We undertook a joint morphometric/genomic investigation of rare and widespread rattlesnake-roots (Nabalus spp.) to determine their conservation units and priorities. Overall, we discovered little genomic or morphological differentiation among populations and varieties of our focal Nabalus species, indicating that both Nabalus boottii and Nabalus trifoliolatus can be managed at the species level. Below, we discuss our results with regard to our specific research questions.

1. Is Nabalus trifoliolatus var. nanus an ESU?

Our morphological and genomic evidence suggest that Nabalus trifoliolatus var. nanus does not qualify as an evolutionary significant unit distinct from non-alpine N. trifoliolatus.

ESUs are defined in terms of ecological and genetic distinctiveness/non-exchangeability

(Crandall et al., 2000; Fraser & Bernatchez, 2001; Waples, 1991). With regard to morphological distinctiveness (an indicator of ecological distinctiveness), our growth chamber common garden experiment revealed a significant difference between alpine Nabalus trifoliolatus var. nanus and non-alpine Nabalus trifoliolatus in only red coloration of the leaves, with no distinction in 13 other traits evaluated. Even this difference was fairly inconsequential from a biological

129 perspective: 16.92 units (SE 6.77) on a 255-unit color scale, or 6.6%. Two of the traits that vary most among wild individuals of the two varieties — height and leaf shape (Bogler, 2006; Haines et al., 2011; Sayers, 1989) — did not differ significantly among growth chamber individuals. Our

PCA corroborated the results of our linear models: the polygons bounding the points for non- alpine Nabalus trifoliolatus almost completely overlapped the polygons bounding Nabalus trifoliolatus var. nanus for the first seven PCs (Figure 4-2). Instead of harboring distinct functional trait values, Nabalus trifoliolatus var. nanus appears to harbor a subset of the trait variation present in non-alpine Nabalus trifoliolatus.

Our finding of morphological indistinctiveness between the two varieties of Nabalus trifoliolatus is likely not a mere artifact of the life stages we investigated, or the morphological variability within/overlap among species that characterizes Nabalus (Bogler, 2006; Sayers,

1989), as validated by our findings for N. boottii. We included N. boottii in the experiment as a control, ensuring that the traits we chose varied sufficiently among taxa that we could recover significant differences even at early life stages. Although we failed to find significant differences between N. boottii and N. trifoliolatus sensu lato for five of 14 traits, we did indeed discover significant species-level differences in six of 14 traits and marginally significant differences in an additional three traits (Table 4-1). Furthermore, despite the overlap in trait values typical for

Nabalus species visible in our PCA plots, we see the morphological distinctiveness between the two species in PC3 (correlated with specific leaf area, specific root length, red coloration, height, and root length; Table 4-2), for which there is almost no overlap between N. boottii and N. trifoliolatus. N. boottii and N. trifoliolatus, then, show morphological distinctiveness, but alpine

N. trifoliolatus var. nanus and non-alpine N. trifoliolatus do not. The two varieties of N. trifoliolatus appear morphologically indistinct and exchangeable.

130 Several additional lines of evidence support our finding of no morphological distinctiveness in N. trifoliolatus var. nanus versus non-alpine N. trifoliolatus. First, the transplant experiment we describe in Chapter 2 similarly revealed just one significant functional trait difference between N. trifoliolatus var. nanus and non-alpine N. trifoliolatus for the 13 traits we examined, while it recovered species-level differences between N. trifoliolatus and N. boottii in four traits (Table 2-2). Second, in at least one location (Mount Katahdin, Maine), wild populations show clines in morphology along elevational gradients (Bogler, 2006; Haines et al.,

2011), further suggesting that the differences often noted between high and low elevation populations are merely due to phenotypic plasticity. Indeed, we discovered a high degree of morphological plasticity in both varieties of N. trifoliolatus in Chapter 2 (Table 2-3). Finally, although continuous populations do not exist along elevation gradients in most locations, researchers have also reported the presence of morphologically intermediate populations in subalpine areas (Löve & Löve, 1966; Mount Washington). Indeed, the population we discovered on Rocky Peak below treeline (NT-NY-RP) could also be considered a morphological intermediate between dwarfed alpine N. trifoliolatus var. nanus and taller non-alpine N. trifoliolatus. All of these lines of evidence support morphological indistinctiveness among alpine and non-alpine N. trifoliolatus populations.

Two lines of evidence somewhat call into question our conclusion of morphological exchangeability between N. trifoliolatus var. nanus and non-alpine N. trifoliolatus. First, our seed transplant experiment in Chapter 2 revealed that alpine N. trifoliolatus var. nanus did not successfully establish from seed at mid or low elevation, only at high elevation; this finding suggests that N. trifoliolatus var. nanus may not be able to recruit outside the alpine zone (Figure

2-3). However, N. trifoliolatus var. nanus established poorly even at high elevation, and we did

131 not find significant differences in its establishment versus non-alpine N. trifoliolatus. We therefore do not find this to be strong evidence of non-exchangeability. Second, in our transplant experiment (Chapter 2) and the current study, we found some traits for which N. trifoliolatus var. nanus’s average trait value appeared intermediate between that of non-alpine N. trifoliolatus and

N. boottii (see Figures 2-5, 2-6, 4-1), suggesting possible morphological differentiation between

N. trifoliolatus varieties; however, the difference was non-significant in all cases. Indeed, we found just as many traits for which the average value of N. trifoliolatus var. nanus was nearly identical to that of non-alpine N. trifoliolatus. Nevertheless, we cannot rule out the possibility that real differences do exist between the two N. trifoliolatus varieties for some traits, and we simply lacked the power (or experimental length) to detect them. Until a larger-scale experiment is conducted, involving a greater number of populations and individuals and carried out over at least two generations to eliminate maternal effects and enable the comparison of mature, flowering individuals (Allendorf et al., 2013; Conner & Hartl, 2004), the best we can conclude based on current evidence is that N. trifoliolatus var. nanus and non-alpine N. trifoliolatus appear morphologically indistinct.

With regard to genetic distinctiveness, we did not find evidence that high elevation N. trifoliolatus var. nanus populations were distinct from low elevation N. trifoliolatus using any of our three clustering methods. Using UPGMA, we did not find individuals and populations clustering according to variety. Of population-level clusters with > 50% support, all included populations from both varieties (Figure 4-5). Similarly, k-means clustering initially revealed k =

1 as the best-supported number of clusters in N. trifoliolatus, suggesting that all populations

(regardless of variety) comprise a single ESU; when forced to proceed with k = 2, DAPC did not group individuals by variety. Our STRUCTURE result was similar. STRUCTURE identified

132 greater population structure (k = 4) in N. trifoliolatus than the other two methods, likely because we used options in our analysis that assist in the detection of weak population structure

(correlated allele frequencies and sampling locations as prior; Falush et al., 2003; Hubisz et al.,

2009). Nevertheless, STRUCTURE too failed to identify N. trifoliolatus var. nanus as genetically distinct: all clusters with more than one population contained both N. trifoliolatus var. nanus and non-alpine N. trifoliolatus.

Although we failed to find consistent clustering among N. trifoliolatus var. nanus individuals using all three methods of population structure analysis, all three methods did indicate a relationship between the Giant (NN-NY-GI) and Rocky Peak (NT-NY-RP) populations. Population-based UPGMA identified a cluster with > 50% bootstrap support that included these two populations. DAPC identified one cluster composed almost entirely of the

Rocky Peak and Giant populations (plus one individual from Marcy) at k = 2. Finally,

STRUCTURE identified one cluster that contained only the Giant and Rocky Peak populations.

Our results from Chapter 3 further support a relationship between these two populations, indicating that Rocky Peak may have been colonized from the Giant population; this is quite possible, as the sites are only ~630 m apart. The close genetic relationship (and possible history of colonization) between the alpine Giant population and the non-alpine Rocky Peak population strongly suggest that alpine N. trifoliolatus var. nanus is not genetically distinct from non-alpine populations; instead, they appear genetically exchangeable (Crandall et al., 2000).

Apart from the Giant and Rocky Peak populations, we found little structuring according to geography in N. trifoliolatus. In part, this result was due to the overall weak population structure we discovered in the species (especially with UPGMA and k-means clustering/DAPC).

However, even the results of our STRUCTURE analysis, which indicated greater population

133 structure in the species, did not reveal a consistent pattern in geographic structure (Figure 4-6).

Two clusters were comprised of populations from three states. We attribute this finding to a few possible causes. First, there may have been a history of colonization uniting the clustered populations (e.g., NN-ME-GE, NN-NY-WF, NT-VT-RI) that explains their relationship—either with one population colonizing the others or all three derived from the same ancestral lineage

(Montero et al., 2019; Prunier et al., 2017; Reynolds, Gerber, & Fitzpatrick, 2011). Alternatively, clustered populations may have diverged from other populations through selection or drift in a pattern that that did not correspond to geography (“isolation by environment”) (Sexton,

Hangartner, & Hoffmann, 2014). However, as the populations in these clusters include high and low elevation populations with quite different environments and phonologies, this explanation seems unlikely. Overall, population structure that does not correspond with geography is difficult to explain (Bossart & Pashley, 1998). Perhaps the non-geographic clustering we find is an artifact of over-clustering (a risk when using correlated allele frequencies; Falush et al., 2003), as we do not find the same non-geographic pattern at lower levels of k.

Altogether, morphological and population genomic evidence suggest that N. trifoliolatus var. nanus does not qualify as an ESU distinct from non-alpine N. trifoliolatus. Because N. trifoliolatus as a whole is widespread and globally secure (NatureServe, 2018), the species appears to merit little conservation concern. Our results validate certain taxonomic and conservation decisions regarding N. trifoliolatus. First, they validate managers’ decisions to stop monitoring high elevation populations of N. trifoliolatus: their time and effort are better directed to other taxa. Second, our results support dropping the varietal epithet (nanus) from descriptions of high elevation populations, a change which has already been advocated by several authors

(Bogler, 2006; Haines et al., 2011). Although some authors have suggested that taxonomic

134 change hinders conservation (Garnett & Christidis, 2017), we believe here, like others (Pillon &

Chase, 2007; Thomson et al., 2018), that recognition of a rare variety of a common species as indistinct is important for focusing conservation efforts instead on truly rare taxa.

2. Does genomic evidence support multiple ESUs within N. boottii?

Our genomic analyses of population structure do not support multiple ESUs within N. boottii. All three methods we employed revealed very little population structure in the species.

Few nodes of our UPGMA dendrograms received >50% bootstrap support in our individual- based and population-based dendrograms (Figure 4-3, 4-5A), and in neither dendrogram did individuals or populations cluster strongly by geography. K-means clustering identified one as the optimal number of clusters (suggestive of one ESU within N. boottii); when forced to proceed with k = 2, DAPC did not group individuals according to geography or even population of origin, with all but two populations split among clusters. Finally, using both diploid and tetraploid analysis procedures, STRUCTURE detected little population structure in N. boottii; all populations belonged to the same majority cluster (Figure 4-6). Altogether, our population genomic analyses suggest that N. boottii is comprised of a single ESU.

Several additional lines of evidence support the conclusion of a single ESU within N. boottii. We did not test for morphological distinctiveness among populations of N. boottii due to permitting constraints; however, researchers often use proxies to determine ecological/adaptive differences for ESU identification, including habitat and life history characteristics (Dizon,

Lockyer, Perrin, Demaster, & Sisson, 1992; Fraser & Bernatchez, 2001; Waples, 1991). With regard to habitat characteristics, N. boottii has narrow and consistent preferences. The restricted range of the species spans just 800 m in elevation (1000–1800 m a.s.l.) and less than 2º of

135 latitude. Within this range, N. boottii inhabits drier areas of wet meadow and heath shrub communities (Prout, 2005). Due to the consistency of its environmental preferences, the New

York Natural Heritage Program’s Element Distribution Model can predict with high accuracy the presence of N. boottii based on characteristics such as elevation, average temperature, and summer precipitation (New York Natural Heritage Program, 2012). Botanical accounts of the species make no mention of wide variability in life history or morphological features among N. boottii populations (Bogler, 2006; Haines et al., 2011), nor do our own observations suggest this.

Therefore, we do not find evidence of adaptive differentiation in N. boottii or plausible habitat differences that could give rise to adaptive differentiation. In short, our findings support considering N. boottii as a single ESU.

Other species for which a single ESU has been identified include widespread species such as the humpback whale (Baker et al., 1993; Moritz, 1994), as well as narrow endemics, including Canary Island Lotus spp. (Oliva-Tejera et al., 2006), the Turks Island boa (Reynolds et al., 2011), the maritime shrew (Dawe, Shafer, Herman, & Stewart, 2009), and the New Zealand long-tailed bat (Dool, O’Donnell, Monks, Puechmaille, & Kerth, 2016). In one notable example, researchers lumped four putative orchid species (including both endemic and widespread species) into a single ESU based on genomics and morphology (Ahrens et al., 2017).

In all cases, the identification of a single ESU for the species was largely based upon weak population structure. For widespread, highly mobile species like the humpback whale, gray wolf, and North American coyote, weak to non-existent population structure results from long- distance dispersal capabilities and frequent migration (Baker et al., 1993; Wayne, Lehman,

Allard, & Honeycutt, 1992). These same forces also generate weak population structure in plant species with wind-dispersed pollen and seeds (Beatty et al., 2015).

136 For narrow endemics that exist in isolated populations (like N. boottii), researchers have often invoked another explanation for weak population structure: glacial cycles. In geographically isolated populations of the Canadian maritime shrew (Dawe et al., 2009) and

Turks island boa (Reynolds et al., 2011), researchers explained weak population structure as resulting from land bridges that enabled panmixia during the last glacial period. Because populations of these species experienced connectivity until recently (i.e., 8,000–10,000 years

BP), they have not diverged sufficiently through selection or drift to yield strong population structure. Recent gene flow may also have contributed to weak structure (Dawe et al., 2009). For the New Zealand long-tailed bat, glacial advance and ecosystem change in the South Island forced populations into North Island refugia, where they too experienced panmixia (Dool et al.,

2016). Rapid range expansion following glacial retreat and some degree of current gene flow between the North and South Islands explains the maintenance of weak population structure in the species. We find this same pattern of weak genetic divergence in many northern hemisphere alpine plant species, which experienced connectivity in glacial refugia during the Last Glacial

Maximum, re-colonized mountains after the last glacial retreat, and subsequently (and relatively recently) became geographically isolated (Bell, Griffin, Hoffmann, & Miller, 2018; Schmitt,

2007). Alpine plant species that have not experienced recent connectivity due to glaciation, such as those in southwestern Australia, instead show much greater population structure and genetic divergence (Bell et al., 2018), supporting the role of connectivity during glacial maxima for giving rise to weak population structure in currently isolated populations of many species.

As an alpine species inhabiting previously glaciated mountaintops, Nabalus boottii likely also experienced population connectivity in glacial refugia during the Last Glacial Maximum

(Bierman et al., 2015; Brochmann et al., 2003; Martin & Germain, 2016), which resulted in the

137 weak population structure we find today (Bubac & Spellman, 2016; Schmitt, 2007). Moderate to high rates of recent migration (at least within mountain ranges; Chapter 3), coupled with N. bootti’s probable tetraploidy, have likely also prevented divergence due to drift in N. boottii’s populations (Allendorf et al., 2013; Meirmans et al., 2018). Furthermore, the similar abiotic and biotic conditions characteristic of sites where N. boottii occurs may have forestalled adaptive divergence in its populations. Together, these forces may have given rise to the weak population we discovered in N. boottii.

The specific population structure we found in N. boottii (Figure 4-6) is that of one metapopulation with admixed ancestry (Montero et al., 2019). This structure arose through mixing (likely historical) of two or more distinct genetic lineages, perhaps in glacial refugia, or through admixture following two or more independent polyploid origins of N. boottii (Welles &

Ellstrand, 2016). The subtle geographic structuring we discovered at higher levels of k (Figure 4-

6) could indicate slight genetic divergence of populations through drift or selection.

Nevertheless, the dominance of one genetic lineage we found in all or almost all N. boottii individuals (depending on level of k) and all N. boottii populations (regardless of k) strongly supports this species as comprised of a single ESU.

N. boottii’s status as a single ESU simplifies conservation planning, eliminating the necessity of preserving multiple genetic lineages within the species. Instead, conservation actions can simply target the most singular, diverse, and/or convenient populations that capture a significant amount of species-level diversity. We have identified priority populations and conservation strategies for N. boottii below.

138 3. How many and which populations of N. boottii should we target for conservation?

Based on our calculations using Ceska et al.'s (1997) formula (modified to use FST after

Pérez-Collazos et al., 2008 and cross-validated using N. trifoliolatus), we found that just two populations are necessary to preserve 99.9% of species-level genetic diversity in N. boottii. This finding strongly contrasts with that of other rare species, for which five to six populations are commonly necessary to reach the 99.9% target (Peñas et al., 2016; Pérez-Collazos et al., 2008;

Segarra-Moragues & Catalán, 2010). For species with especially isolated populations, preserving all populations may be necessary to reach the 99.9% target (e.g., 18 of 18 populations in Weber-

Townsend, 2017); in these cases, the conservation target is normally lowered for practical reasons. Instead, the very shallow population structure we find in N. boottii places the number of populations to target for conservation within very reasonable bounds for managers.

As one means of determining priority populations for conservation, we assessed the number of private alleles per population. Populations with a high number of unique alleles may harbor important genetic diversity for the future evolution of the species (Bengtsson et al., 1995;

Caujapé-Castells & Pedrola-Monfort, 2004; Ceska et al., 1997; Peñas et al., 2016). Across all N. boottii datasets, the Baxter population (NB-ME-BX), also one of the most geographically isolated populations, harbored a high number of private alleles (Table 4-3). The Hamlin (NB-

ME-HA), Alpine Garden (NB-NH-AP), Armstrong (NB-NY-AR) and Wright (NB-NY-WR) populations followed Baxter in uniqueness. However, as compared with N. trifoliolatus, we found very few private alleles in N. boottii populations; given the very weak population structure in N. boottii, this result makes sense. Baxter, the most unique population, harbored just three to six private alleles (~0.2% of the population’s total alleles), depending on dataset (full or subset) and analysis type (dosage or no dosage); thus, we did not find any populations harboring a high

139 number of private alleles (compared to N. trifoliolatus) or a significant spread in private allele numbers among populations. We therefore question whether the uniqueness of Baxter and the other populations is biologically meaningful. Conservatively, assuming our results are biologically meaningful, they suggest that the Baxter, Hamlin, Alpine Garden, Armstrong and

Wright populations should have slightly higher conservation priority than populations displaying few to no unique alleles, such as Boundary Bald (NB-ME-BB).

As an additional means of determining priority populations for conservation in N. boottii, we completed population surveys for almost all extant populations and correlated population size and flowering rate with genetic diversity statistics (statistics presented in Table 3-2). Based on our linear models, we found that neither population size nor flowering rate significantly predicted genetic diversity in N. boottii populations. Generally, population size is positively correlated with genetic diversity (PPL, AR, heterozygosity), because larger populations are less susceptible to loss of diversity through drift (in plants: Ellstrand & Elam, 1993; Nybom, 2004).

Additionally, we often expect flowering rate to correlate positively with genetic diversity and negatively with inbreeding, given the role outcrossing sexual reproduction plays in maintaining population heterozygosity (e.g., Jeong, Lee, Yoo, Jang, & Kim, 2012; Larkin et al., 2006; Ruiz et al., 2018). Conversely, some empirical and theoretical evidence suggests that populations that are asexually reproducing can actually maintain higher levels of genetic diversity (Gorelick & Heng,

2011; Hamrick et al., 1992). In cases where relationships do not exist between populations size or flowering rate and genetic diversity, authors have attributed this result to historical processes

(i.e., the populations are not at equilibrium; Ellstrand & Elam, 1993).

In this case, N. boottii’s probable polyploidy is a likely explanation of the lack of relationship between genetic diversity and population size or flowering rate. Small polyploid

140 populations are less prone to lose genetic diversity through drift because of their greater effective population size (Meirmans et al., 2018), weakening the relationship between population size and genetic diversity. Furthermore, polyploid organisms often display fixed heterozygosity as a result of whole genome duplication, helping populations maintain diversity (heterozygosity) and avoid inbreeding even if they have low rates of sexual reproduction (Brochmann et al., 2004; Kawecki,

2008; Van De Peer et al., 2017; Weiss-Schneeweiss et al., 2013). Apart from polyploidy, the longevity of N. boottii (estimated to be 5–10+ years for ramets and greater for genets; Körner,

2003), coupled with a possible persistent seed bank, could also help this species maintain genetic diversity even in small populations (Ellstrand & Elam, 1993). Whatever the cause, our findings indicate that large populations or those with higher flowering rates do not necessarily represent conservation priorities for N. boottii based on genetic diversity.

Currently, populations of Nabalus boottii appear relatively stable; however, given the high conservation priority of this globally rare species, it is prudent to plan for possible future conservation action (NatureServe, 2018; Prout, 2005). Our results provide several important insights for guiding conservation. First, our results suggest that certain populations merit slightly greater conservation concern than others based on their genetic uniqueness: Baxter (NB-ME-

BX), Hamlin (NB-ME-HA), Alpine Garden (NB-NH-AP), Armstrong (NB-NY-AR) and Wright

(NB-NY-WR). We encourage managers of these populations in particular to engage in routine demographic monitoring to identify any declines or threats. Second, we have identified two populations as a sufficient minimum target for ex-situ conservation efforts like seed banking, which has been undertaken in the past for this species (B. Brumback, pers. comm.). Regarding potential source populations, we do not recommend many of the populations we identified as unique due to their highly protected nature (Baxter and Hamlin) or small size (Armstrong and

141 Wright). We instead suggest targeting the Alpine Garden (NB-NH-AP), Lakes of the Clouds

(NB-NH-LC), and/or Cow Pasture (NB-NH-CP) populations for ex-situ conservation. These three are by far the largest populations, ensuring that seed banking efforts would have minimal demographic impact and researchers would find suitable numbers of flowering individuals to make representative collections (Vitt, Havens, Kramer, Sollenberger, & Yates, 2010).

Additionally, these populations are relatively easy to access via short hikes from the Mount

Washington Auto Road. Finally, the Alpine Garden population (and, to a lesser extent, the Cow

Pasture populations) harbor a high number of private alleles (Table 4-3), and all three display average to high levels of genetic diversity for the species (Table 3-2). Seed banking efforts that followed published guidelines for maximizing genetic diversity in collections (e.g., Vitt et al.

(2010) and Basey et al. (2015)) and that targeted two or all three of these populations should capture most of the species-level genetic diversity in N. boottii.

The shallow structure and lack of genetic distinctiveness we find in N. boottii suggests that populations are genetically exchangeable, which also has implications for conservation.

Often, managers are hesitant to undertake efforts such as population augmentation, assisted gene flow, or the establishment of ex-situ populations of mixed origin for fear of outbreeding depression (Frankham, 2015; Frankham et al., 2011). However, because N. boottii populations are minimally differentiated, have likely been isolated for a relatively short time span, and inhabit very similar environments, outbreeding depression is very unlikely (Frankham et al.,

2011). While N. boottii populations may not need genetic rescue as they appear to maintain high genetic diversity even in small populations (< 500 individuals), they may need demographic rescue to avoid extinction if factors like hiker trampling, rock slides, or environmental stochasticity cause declines (Gilpin & Soulé, 1986). Should managers identify populations

142 demographically at risk of extinction, we recommend population augmentation through transplanted individuals or seeds from nearby populations without excessive fear of outbreeding depression; in fact, almost all such efforts are successful for species like N. boottii that are good candidates (Frankham, 2015).

4. Conclusion

In an era characterized by mounting threats to biodiversity and limited funds for conservation, identifying conservation priorities is of paramount importance (Barnosky et al.,

2011; Ceballos et al., 2015; Pimm et al., 2014). This task is especially important for plants, which comprise over half of federally listed endangered species in the United States but receive under 3% of federal funding (Balding & Williams, 2016; Havens, Kramer, & Guerrant, 2013).

Here, we have used a combination of population genomic, experimental, and demographic analyses to inform conservation of two species native to the northeastern United States: N. boottii and N. trifoliolatus, including the latter’s non-alpine and alpine (var nanus) varieties. We did not find support for alpine populations of N. trifoliolatus constituting an Evolutionary

Significant Unit, and therefore conclude that they are most likely not of particular conservation value; managers are justified in diverting resources elsewhere. For the rare alpine congener N. boottii, we found that genetic evidence supported a single evolutionary significant unit within the species, a result corroborated by our anecdotal observations of similar morphology and habitat characteristics among populations. This finding suggests that N. boottii can be managed at the species level. Finally, we determined that just two populations are sufficient to capture almost

100% of species-level genetic diversity in N. boottii, and that managers can target populations for ex-situ conservation based on demographic and practical rather than genetic considerations.

143 Table 4-1. Model summaries for functional trait data from the growth chamber experiment for Nabalus boottii (NB), N. trifoliolatus var. nanus (NN) and non-alpine N. trifoliolatus (NT). We log-transformed some trait data prior to analysis to improve normality of residuals and heterogeneity of variance (indicated with a dot in the “Log” column). Significant differences at α = 0.05 are shown in bold. Differences that were significant before but not after Tukey adjustment were considered marginal, and indicated with an asterisk. Columns titled β (SE) give the difference in estimated marginal means and corresponding standard error for significant and marginally significant pairwise differences.

Model Significant differences Trait attributes N d.f. Log NB - NN NB - NT NN - NT P β (SE) P β (SE) P β (SE) Height 97 92 0.402 <0.001 -3.53 0.104 (0.77) Root length 64 59 0.241 0.071* 0.33 0.986 (0.15) Dry mass: shoot 93 88 0.697 0.273 0.979

Dry mass: root 63 58 0.796 0.172 0.786

Dry mass: total 63 58 0.997 0.934 0.929

Total leaf area 97 92 0.597 0.021 -2.66 0.587 (0.97) Leaf number 96 91 0.114* 1.08 <0.001 1.51 0.604 (0.54) (0.36) Root to shoot ratio 63 58 0.386 0.015 -0.54 0.752 (0.19) Specific leaf area 93 88 0.508 0.375 0.947

Specific root length 63 58 0.259 0.003 0.82 0.717 (0.24) Red coloration 95 90 0.057* 18.69 0.938 0.040 -16.92 (7.99) (6.77) Green coloration 95 90 0.085* 22.97 0.546 0.200 (10.6) Blue coloration 95 90 0.501 0.927 0.551

Leaf shape (L/W) 96 91 0.009 0.58 0.020 0.34 0.302 (0.19) (0.12)

144 Table 4-2. Factors and loadings from the PCA performed on functional trait data from the growth chamber experiment for Nabalus boottii, N. trifoliolatus var. nanus and non-alpine N. trifoliolatus. Our dataset consisted of measurements of 14 traits for 63 individuals. We have bolded the three loadings with the greatest absolute value under each PC.

Trait PC1 PC2 PC3 PC4 PC5 Height -0.205 -0.328 -0.334 -0.007 0.443 Root length -0.171 0.054 0.341 0.546 0.252 Dry mass: shoot -0.393 -0.243 0.023 -0.117 -0.035 Dry mass: root -0.380 0.133 -0.157 0.259 -0.135 Dry mass: total -0.426 -0.106 -0.052 0.032 -0.081 Total leaf area -0.391 -0.192 -0.146 -0.169 0.061 Leaf number -0.058 -0.386 0.255 -0.057 -0.371

Root to shoot ratio -0.083 0.422 -0.221 0.512 -0.205

Specific leaf area 0.148 0.146 -0.438 -0.096 0.370

Specific root length 0.277 -0.129 0.412 0.041 0.165 Red coloration 0.255 -0.322 -0.365 0.271 -0.160 Green coloration 0.206 -0.270 -0.280 0.119 -0.500 Blue coloration 0.279 -0.264 -0.133 0.204 0.249 Leaf shape (length/width) -0.037 -0.390 0.138 0.433 0.170

145 Table 4-3. Private alleles per population for Nabalus boottii (left) and N. trifoliolatus (right). “ND” columns do not include dosage: each private allele is counted once per population. “D” columns include dosage. “Full” corresponds to the full datasets for each species, while “Sub” corresponds to the even-sample sized subsets, all with 338 loci. The numbers in bold give the highest three (or four, given a tie) private allele counts per column. Table 3-1 provides full population names and information for each population abbreviation.

Nabalus boottii Nabalus trifoliolatus ND D ND D ND D ND D Pop Pop Full Sub Full Sub NB-ME-BB 0 0 0 0 NN-ME-GE 3 5 3 4 NB-ME-BX 3 4 5 6 NN-ME-WE 3 8 – – NB-ME-HA 1 1 5 6 NN-NH-LC 4 12 11 18 NB-NH-AP 3 3 3 3 NN-NH-TU 12 31 – – NB-NH-CP 3 3 0 0 NN-NH-WA 1 3 5 7 NB-NH-ED 1 1 1 1 NN-NY-GI 6 14 6 11 NB-NH-LC 0 0 1 1 NN-NY-MA 2 7 8 19 NB-NH-MO 1 3 1 3 NN-NY-WF 9 18 11 19 NB-NH-NE 2 6 2 4 NN-NY-WR 2 3 – – NB-NY-AL 2 2 – – NT-ME-BR 3 3 4 4 NB-NY-AR 1 1 3 5 NT-ME-CA 3 4 – – NB-NY-GO 0 0 1 1 NT-ME-TO 2 7 4 10 NB-NY-WF 2 2 0 0 NT-NY-RP 0 0 2 2 NB-NY-WR 1 4 2 5 NT-VT-BU 10 24 10 12 NB-VT-CH 1 1 – – NT-VT-RI 6 12 8 10 Total 21 31 24 35 Total 66 151 72 116 % of pops 80% 80% 77% 77% % of pops 93% 93% 100% 100% Avg per pop 1.4 2.1 1.8 2.7 Avg per pop 4.4 10.1 6.5 10.5

146 Table 4-4. Population size and flowering rate for the 15 sampled populations of Nabalus boottii, comprising nearly all known populations (and including all of the largest). Populations are arranged by size. We determined population size and flowering rate for most populations (no asterisk) by a direct count of the number of basal leaves and flowering individuals. For populations with an asterisk, we directly counted all flowering individuals and estimated population size (i.e., basal leaves plus flowering individuals) based on a test patch. We note that although it is standard to count numbers for this species using basal leaves plus flowering individuals, the number of individual plants is likely less. Table 3-1 provides full population names and information for each population abbreviation.

Pop Size Flowering

NB-NY-WR 67 4.48% NB-NH-NE 167 2.40% NB-NY-AR 250 0.40% NB-NY-GO 267 0.38% NB-NY-AL 464 4.74% NB-ME-HA 588 0.34% NB-ME-BX 1,204 0.50% NB-VT-CH 1,337 0.45% NB-NH-MO 1,500 2.13% NB-NH-ED* 3,842 1.90% NB-NY-WF* 4,998 4.40% NB-ME-BB 6,015 0.03% NB-NH-LC* 26,833 3.60% NB-NH-AP* 28,360 2.50% NB-NH-CP* 53,625 0.80%

ME average 2,602 0.29% ME total 7,807 NH average 19,055 2.22% NH total 114,327 NY average 1,209 2.88% NY total 6,046 VT average 1,337 0.45% VT total 1,337 Average 8,634 1.94% Total 129,517

147 Table 4-5. Summaries of linear models exploring the relationship between population size/flowering rate and diversity statistics for the fifteen sampled populations of Nabalus boottii. We also examined the relationship between population size and flowering rate. We log- transformed population size prior to analysis to improve normality of residuals and homogeneity of variance. We provide population size and flowering rate estimates in Table 4-4.

Model N d.f. R2 β P PPL ~ log(pop_size) 13 11 0.048 -0.271 0.470 AR ~ log(pop_size) 15 13 0.075 -0.003 0.323 HE ~ log(pop_size) 15 13 0.000 0.000 0.954 HO ~ log(pop_size) 15 13 0.079 -0.004 0.311 FIS ~ log(pop_size) 15 13 0.140 0.018 0.169

PPL ~ PercFlower 13 11 0.000 0.015 0.976 AR ~ PercFlower 15 13 0.016 -0.001 0.649 HE ~ PercFlower 15 13 0.002 0.000 0.869 HO ~ PercFlower 15 13 0.035 -0.003 0.503 FIS ~ PercFlower 15 13 0.122 0.020 0.201

PercFlower ~ log(pop_size) 15 13 0.002 -0.053 0.875

148 A Height B Root to shoot 16 NB vs NT: *** NB vs NT: *

12 1.0

8 0.5 Height (cm) 4 Root: shoot ratio

NB NN NT NB NN NT

C Leaf shape D Root length NB vs NN: ** NB vs NT: P=0.071 NB vs NT: * 10.0 3.5

3.0 7.5

2.5 5.0

2.0 Root length (cm) Leaf length/width 2.5

NB NN NT NB NN NT

E Specific leaf area F Specific root length

50 N.S. 15

) NB vs NT: ** ) g 40 g m m 10 2 30 m m m m ( (

L 5 A R L

10 S S

0 0 NB NN NT NB NN NT

G Red coloration 150 NN vs NT: *

100

Red value 50

NB NN NT

Figure 4-1. (Figure caption appears on the following page.)

149 Figure 4-1. (Figure appears on the previous page.) Comparisons of functional trait measurements for Nabalus boottii (NB), N. trifoliolatus var. nanus (NN) and non-alpine N. trifoliolatus (NT) from the growth chamber experiment. We indicate significant differences in the upper-left corner of the plots (* P <0.05, ** P <0.01, *** P <0.001) based on the results of linear models. A and B show traits for which N. trifoliolatus var. nanus appears to achieve a trait score intermediate between the alpine obligate species N. boottii and non-alpine N. trifoliolatus, although these differences were not significant. C and D show traits for which N. trifoliolatus var. nanus’s values appear very similar to those of non-alpine N. trifoliolatus. Plots E, F, G, D, and A show data for the traits most correlated with PC3, listed here from most to least correlated. PC3 separated N. boottii from the N. trifoliolatus taxa.

150 ● 0.2

● ● 0.2 ● Taxon Taxon ● ● ● 0.0 ● ● NB ● NB ● NN 0.0 NN ● NT NT PC4 (9.05%) PC2 (17.86%) ● ● −0.2 ●

● −0.2

● ●

−0.4 −0.2 0.0 0.2 −0.2 −0.1 0.0 0.1 0.2 0.3 PC1 (36.39%) PC3 (12.32%) 0.50 ●

●

0.2 ● 0.25 ● Taxon Taxon

● NB ● ● NB ● ● 0.00 ● 0.0 ● NN NN ● ● NT NT PC7 (4.35%) PC6 (4.96%) ● ●

● −0.25 −0.2

● ●

−0.4 −0.2 0.0 0.2 −0.2 0.0 0.2 PC5 (7.74%) PC6 (4.96%)

Figure 4-2. PCA plots of functional trait data for Nabalus boottii (NB), Nabalus trifoliolatus var. nanus (NN), and non-alpine Nabalus trifoliolatus (NT) from the growth chamber experiment for the first seven PCs. We performed PCA on data from 63 individuals for 14 functional traits (factors and loadings provided in Table 4-2). Although many PC combinations of these seven are possible for plotting, we have chosen these four plots as sufficient to visualize separation along each PC. Envelopes enclose 100% of points for each taxon.

151 NB−WF 9 NB−WF 3 NB−WF 4 NB−WF 1 90.7 NB−WF 29 NB−WF 23 NB−WF 10 NB−WF 16 NB−WF 17 NB−WR 5 NB−BX 1 NB−NE 14 NB−CP 7 NB−CH 3 NB−AL 2 NB−BB 2 NB−BB 15 NB−GO 14 NB−ED 2 NB−LC 3 NB−CP 12 NB−MO 11 NB−CP 8 NB−AP 14 NB−AP 9 NB−LC 9 54.4 NB−HA 5 NB−ED 15 NB−WF 25 NB−WF 22 53.7 NB−GO 9 NB−GO 8 NB−CH 13 NB−AR 5 NB−AR 1 NB−AR 2 NB−AL 3 NB−BB 13 NB−CP 10 NB−BX 7 NB−AP 13 71.2 63.3 NB−MO 4 NB−MO 10 NB−MO 6 NB−LC 2 NB−BX 6 NB−NE 4 NB−NE 13 NB−WR 13 NB−BB 4 NB−WR 4 NB−WR 7 NB−BX 9 NB−NE 3 NB−ED 6 NB−WF 19 NB−AR 8 NB−GO 4 NB−BX 2 NB−AP 15 NB−GO 12 NB−LC 5 NB−HA 8 NB−HA 10 NB−LC 10 NB−HA 4 NB−ED 1 NB−CP 15 NB−WF 7 NB−AL 15 NB−NE 12 100 NB−WF 15 NB−LC 8 NB−AP 1

0.00 0.02 0.04 0.06 0.08

Nei's genetic distance for Nabalus boottii individuals

Figure 4-3. Individual-based UPGMA dendrogram of Nabalus boottii individuals using the full dataset and Nei’s genetic distance. Numbers on the dendrogram branches give bootstrap support (%) for nodes with > 50% support based on 1000 bootstraps. Colors correspond to population of origin. Table 3-1 provides full population names and information for each population abbreviation (two letter state code dropped for this figure).

152 NT−RI 10 NT−BR 9 NN−LC 3 NN−LC 2 NT−TO 14 NT−RI 2 NN−WA 8 NN−WA 7 NN−WA 1 NN−WA 10 NN−GE 4 NN−GE 2 NN−GE 9 NN−GE 3 NN−GE 5 NT−BU 8 NT−BU 10 NT−BU 6 NT−BU 7 NT−BU 1 90 NT−TO 8 NT−TO 11 NT−TO 9 NN−LC 5 NN−LC 9 NN−LC 6 NT−BR 5 NT−BR 13 NT−BR 11 NN−WF 8 NN−WF 7 NN−WF 14 NN−WF 3 NN−WF 15 77.1 NN−WR 7 NN−WR 6 62.7 NN−MA 4 100 NN−MA 3 NN−MA 1 NN−MA 8 74.9 99.8 NT−RP 4 70.5 NT−RP 1 NT−RP 5 NT−RP 7 NN−MA 12 NN−MA 11 54.4 NT−RI 6 NT−RI 3 NT−RI 8 NT−RI 7 57 NT−CA 6 NT−CA 4 NT−CA 3 64 86.2 NN−WE 5 NN−WE 4 NN−WE 7 NN−WF 6 58 NT−BU 9 NT−BU 2 NN−WR 5 100 89.5 NN−GI 7 NN−GI 6 NN−GI 1 NN−GI 5 NN−GI 2 NN−TU 9 NN−LC 8 NN−WF 1 100 NT−RI 1 NT−BU 5 99.9 NN−TU 30 NN−TU 27

0.00 0.02 0.04 0.06

Nei's genetic distance for Nabalus trifoliolatus individuals

Figure 4-4. Individual-based UPGMA dendrogram of Nabalus trifoliolatus individuals using the full dataset and Nei’s genetic distance. Numbers give bootstrap support (%) for nodes with > 50% support based on 1000 bootstraps. Colors correspond to population of origin. Table 3-1 provides full population names and information for each population abbreviation (two letter state code dropped for this figure).

153

NB−NY−WF 59.5 A NB−NY−GO

NB−NH−ED

NB−NH−CP 50 NB−NH−AP

NB−NH−LC

NB−ME−BX

NB−ME−HA

NB−ME−BB

NB−NH−NE

NB−NY−AR NB−NY−WR NB−VT−CH 100 NB−NH−MO NB−NY−AL

0.000 0.005 0.010 0.015 0.020 0.025

Nei's genetic distance for Nabalus boottii NT−VT−RI B NN−NH−WA NT−VT−BU NN−NH−LC NT−ME−BR NN−ME−GE NT−ME−TO 73 NN−NY−WF NN−NY−MA NN−NY−WR NT−ME−CA 99.1 NN−ME−WE

100 52.9 NT−NY−RP NN−NY−GI NN−NH−TU

0.00 0.01 0.02 0.03 0.04

Nei's genetic distance for Nabalus trifoliolatus

Figure 4-5. UPGMA dendrogram of Nabalus boottii (A) and Nabalus trifoliolatus (B) populations using full datasets and Nei’s genetic distance. Numbers give bootstrap support (%) for nodes with > 50% support based on 1000 bootstraps. Table 3-1 provides full population names and information for each population abbreviation.

154 Nabalus trifoliolatus (diploid analysis; k = 4) A

Nabalus boottii (diploid analysis; k = 2) B

Nabalus boottii (tetraploid analysis; k = 3) C

Nabalus boottii (tetraploid analysis; k = 5) D

Nabalus boottii (tetraploid analysis; k = 8) E

Figure 4-6. (Figure caption appears on the following page.)

155 Figure 4-6. (Figure appears on the previous page.) Summary plots showing individual admixture and population structure in Nabalus trifoliolatus (A) and Nabalus boottii (B-E) based on results from program STRUCTURE. CLUMPAK determined k = 4 as the optimal number of clusters for Nabalus trifoliolatus based on the Evanno method and Pritchard’s maximum log probability method (A). We performed both a diploid and tetraploid analysis for Nabalus boottii. B shows the result of the diploid analysis, with k = 2 being the optimal number of clusters for both Evanno’s and Pritchard’s methods. C–E give the results of the tetraploid analysis. According to the Evanno method, k = 3 is the optimal number of clusters (C), very closely followed by k = 5 (D) (ΔK = 4.076 vs. ΔK = 4.008). For Pritchard’s method, k = 8 (E) is the optimal cluster number. Pritchard et al. (2010) warn that determinations of optimal k may be unreliable for polyploid analyses. Table 3-1 provides full population names and information for each population abbreviation.

156 1

0.98

0.96

0.94

0.92

NB 0.9 NT Proportion of genetic diversity captured

0.88 0 2 4 6 8 10 Number of populations

Figure 4-7. Genetic diversity accumulation curve for populations of Nabalus boottii and Nabalus trifoliolatus. Proportion of genetic diversity captured by N populations was calculated using N global FST and the formula P = 1 – FST . Here the label “NT” represents N. trifoliolatus sensu lato; “NB” represents Nabalus boottii.

157 CHAPTER 5: SYNTHESIS

Summary of conclusions

My goal for this dissertation was to inform the conservation of two alpine plants endemic to the northeastern United States—Nabalus boottii and Nabalus trifoliolatus var. nanus—by assessing their ability to respond to ongoing environmental change (especially climate change) and defining conservation units and priorities for each taxon. Additionally, the genomic and experimental techniques I used to address my goal allowed me to investigate two fundamental ecological topics: the niche breadth-range size hypothesis and factors contributing to the historical persistence of small, isolated mountaintop Nabalus populations.

Conservation and environmental change

My findings suggest some degree of resilience to environmental change in alpine

Nabalus populations. The three modes of response that allow organisms to avoid extinction in the face of change—adaptation, migration, and tolerance through phenotypic plasticity—all appear to be viable options to a greater or lesser extent for alpine Nabalus. Nabalus boottii and

Nabalus trifoliolatus both exhibited substantial phenotypic plasticity in a variety of ecologically important traits (Chapter 2), suggesting that they likely will be able to maintain growth, photosynthesis, competitive position, and optimal carbon allocation despite changing environmental conditions. Furthermore, the moderate to high genetic diversity I found within even small populations of alpine Nabalus spp. indicates that populations likely harbor the raw materials needed to adapt to their changing environment (Chapter 3). Finally, the evidence of recent and historical migration I discovered between mountaintop populations of Nabalus spp. suggests that while these taxa may not be able to successfully migrate to southern Canadian

158 alpine areas as climate change intensifies, they at least appear able to re-colonize extirpated locations within mountain ranges, adding to the overall stability of their metapopulations

(Chapter 3).

Although alpine Nabalus taxa appear to have viable response mechanisms in the face of ongoing environmental change, some caution is still warranted: I found that no alpine seeds transplanted to warmer (low elevation) successfully established (Chapter 2). Therefore, alpine

Nabalus populations may suffer from reduced seedling recruitment as climate change intensifies.

Although older plants appear more resilient to changes in temperature, and although asexual reproduction likely dominates in most alpine Nabalus populations based on low flowering rates, sexual reproduction is indeed important for most alpine species (Körner, 2003). As such, I recommend regular population monitoring to assess population trends and thereby determine if conservation action is necessary for globally rare N. boottii.

Given my results in Chapter 4 regarding conservation units and priorities within alpine

Nabalus taxa, conservation action should be relatively straightforward for N. boottii if future monitoring indicates that it is merited. I did not find genomic evidence of multiple evolutionary significant units (ESUs) within N. boottii, which displayed very little population structure. Nor did I find strong evidence suggesting that certain populations were more valuable than others from a conservation perspective: a few populations (such as Baxter in Maine) contained slightly more private alleles than others, but overall I found few private alleles in N. boottii populations as compared with N. trifoliolatus. By targeting just two populations of N. boottii for conservation efforts, managers can effectively conserve upwards of 99.9% of species-level genetic diversity in

N. boottii thanks to its weak population structure. I therefore recommend that future conservation efforts target at least two of the largest known N. boottii populations as source populations, such

159 as Cow Pasture, Lakes of the Clouds, or Alpine Garden (all in New Hampshire). N. boottii also appears to be a good candidate for conservation actions such as population augmentation or assisted migration without fear of outbreeding depression (Frankham, 2015; Frankham et al.,

2011).

Alpine N. trifoliolatus var. nanus, conversely, does not appear to merit conservation concern. I did not find evidence that high elevation populations (var. nanus) were morphologically or genetically differentiated from non-alpine populations. Given that N. trifoliolatus as a whole is widespread and globally secure, northeastern alpine managers are validated in their decisions to discontinue monitoring of high elevation populations and focus their efforts on truly rare taxa. My finding additionally supports discontinuation of the varietal epithet for high elevation populations—a practice already advocated by some authors (e.g.,

Haines, Farnsworth, & Morrison, 2011).

Fundamental questions

My results from Chapter 2 provided tentative support for the niche breadth-range size hypothesis in explaining range size differences between N. boottii and N. trifoliolatus. Although my findings for individual populations of N. trifoliolatus varied, at the species level, N. trifoliolatus displayed greater seed establishment niche breadth than the narrow endemic N. boottii. At the population level, environment of origin (alpine vs. non-alpine) was a better predictor of establishment niche breadth than species identity. Coupled with my finding of uniform phenotypic trait plasticity across both N. boottii and N. trifoliolatus, this finding suggested that a population-level factor (such as local adaptation) explains differences in niche breadth, rather than greater phenotypic plasticity in N. trifoliolatus, as I predicted. Overall, more

160 research into the mechanisms driving the niche breadth differences across species is sorely needed for understanding how niche breadth contributes to range size differences across taxa

(Slatyer et al., 2013).

With regard to the historical persistence of alpine Nabalus spp. in isolated populations, my findings from Chapter 3 suggested that the ability of these taxa to maintain genetic diversity in small populations (likely through tetraploidy for N. boottii) and gene flow among separate populations (likely through insect pollination and wind-dispersed seed) were both important factors in the historical persistence of alpine Nabalus populations. The probable tetraploidy of N. boottii may have also conferred increased tolerance of environmental extremes, higher baseline genetic diversity, and/or greater impact for each migration event, further explaining its historical persistence in small populations (Levin, 1983; Sessa, 2019; Weiss-Schneeweiss et al., 2013). In terms of future persistence, polyploidy may confer N. boottii a higher likelihood of survival in our era of accelerating climate change (Cai et al., 2019; Fawcett et al., 2009; Sessa, 2019;

Vanneste et al., 2014).

Broader impacts and future directions

Study methodology

In this dissertation, I presented a method for diagnosing species’ responses to environmental change by assessing their ability to adapt, migrate, and tolerate change through phenotypic plasticity. This evaluation was performed using both genomics and transplant experiments. Although the adapt–migrate–tolerate framework is well-established in global environmental change literature (Chevin et al., 2010; Davis & Shaw, 2001; Jump & Peñuelas,

2005; Merilä & Hendry, 2014; Nicotra et al., 2010), no studies to my knowledge have

161 systematically examined all three response mechanisms in species using both common garden experiments and population genomics.

As sequencing becomes increasingly economical, using genetics or genomics to investigate migration and adaptive potential should become tenable for a greater variety of species. Within the northeast alpine zone, I suggest prioritizing globally rare taxa such as

Potentilla robbinsiana and Geum peckii. For future transplant experiments, the incorporation of additional treatments apart from elevation, such as precipitation variability or biotic community composition, would help shed greater light on the response of organisms to other types of environmental changes (beyond temperature via elevation). Finally, I suggest researchers attempt to secure permission to carry out transplant experiments over multiple years to enable comparison of flowering traits and flowering phenology in addition to vegetative traits (ex.

McDonough Mackenzie, Primack, & Miller-Rushing, 2018). Permitting for longer term transplant experiments may be difficult in the northeast alpine zone, but a growing awareness of the utility of these experiments may encourage improvement of this process (Berend et al.,

2019).

Environmental change in the northeastern United States

This dissertation contributes to a growing understanding of the effects of global change— especially climate change—on northeastern mountain biota. A steadily increasing body of research has documented shifts in species/community distributions and phenology in both the northeast alpine zone and high elevation spruce-fir forests. These changes have been attributed to climate change, land-use change, and nitrogen deposition. Scholarship documenting change in the northeast alpine zone includes Robinson et al.’s (2010) study of historical transects in the

162 Adirondacks, Goren and Monz's photopoint monitoring on Adirondack alpine summits (2011),

Kimball et al.'s (2014) assessment of alpine climate warming and phenology on Mount

Washington (New Hampshire), and recent vegetation monitoring work by Tim Howard et al. on

Adirondack summits (Howard, White, & Goren, 2019). In addition, Kimball and Weihrauch

(2000) studied the factors related to the alpine-treeline ecotone, and mapped alpine communities for future monitoring efforts in the White Mountains of New Hampshire and at Mount Katahdin in Maine. McDonough Mackenzie (2018) examined factors controlling phenology in three species that occur from low elevation to the edaphic alpine zone of Cadillac Mountain in Maine.

Below treeline, several studies have examined species performance, phenology, and range shifts in high elevation spruce-fir forests, as well as their lower-elevation ecotone with the northern hardwood forest (Beckage et al., 2008; Berdugo & Dovciak, 2019; DeLuca & King, 2017; Foster

& D’Amato, 2015; McLaughlin, Downing, Blasing, Cook, & Adams, 1987; Seidel et al., 2009;

Wason, Beier, Battles, & Dovciak, 2019; Wason & Dovciak, 2017).

Together, these studies paint a complex portrait of change in northeastern mountain species and communities. Some authors have found evidence of resilience in mountain biota, such as documented recoveries in alpine vegetation following hiker trampling (Goren & Monz,

2011) and in red spruce growth as rainwater pH rises (Wason et al., 2019). The northeast alpine also appears to exhibit a smaller magnitude of climate-related abiotic and biotic change than other montane areas of the world (Kimball et al., 2014; Kimball & Weihrauch, 2000; Seidel et al., 2009). On the other hand, other authors have discovered significant change in regional biota that can be tied to climate or other environmental changes (Beckage et al., 2008; Berdugo &

Dovciak, 2019; Robinson et al., 2010). Still others have discovered unexpected downslope shifts

163 in species distributions despite an overall warming regional climate (DeLuca & King, 2017;

Foster & D’Amato, 2015).

All told, the impact of ongoing global environmental change on mountain systems of the northeastern United States is multifaceted, and at times, paradoxical. Continued research will be particularly essential if we are to effectively predict the future effects of global environmental change in these globally significant areas. The simultaneous interplay of various environmental factors complicates this line of research considerably; for instance, it can be challenging to distinguish vegetation shifts caused primarily by recreation use change from those driven more by nitrogen/acid deposition, or climate change (Robinson et al., 2010; Wason et al., 2019;

Wason, Dovciak, Beier, & Battles, 2017; Wason & Dovciak, 2017). Nevertheless, the relative stability documented in northeast alpine communities (Kimball et al., 2014; Kimball &

Weihrauch, 2000) and the probable resilience I discovered in alpine Nabalus taxa together suggest that northeastern alpine biota may be at less risk than species in other mountaintop areas of the world (Marris, 2007; Urban, 2018).

For future research, I recommend further studies evaluating probable response and vulnerability for other globally rare taxa (as this study did), as well as continued community- level monitoring and analysis of long-term datasets to document shifts in distributions and phenology. Studies covering the range of the northeast alpine zone or associated mountains will be most useful, as individual mountains/mountain ranges have shown anomalous trends (e.g.,

Seidel et al., 2009 [single mountain] vs. Wason, 2016 [multiple mountains]).

164 Northeastern alpine dynamics

This dissertation also makes an important contribution to our understanding of genetic diversity and inter-population dynamics in the alpine zone of the northeastern United States and southeastern Canada: it is the first region-wide population genetic study of alpine vascular plants.

Robinson has previously studied region-wide genetic diversity and gene flow in northeast alpine

Sphagnum species (Robinson, 2012), and is currently conducting a similar study with Diapensia lapponica (Martinez Munoz, Robinson, Vollmer, & Popp, 2019). The only other population genetic study conducted in the northeast alpine ecosystem is Lindwall's (1999) dissertation, which examined diversity and gene flow in populations of three alpine species (Carex bigelowii,

Diapensia lapponica, and Minuartia groenlandica) within a relatively small geographic area surrounding Mount Washington. From these studies, it appears that weak population structure is characteristic of many northeastern alpine plants, although Lindwall (1999) found little to no gene flow among populations of Carex bigelowii and Diapensia lapponica. Ultimately, additional studies involving a greater variety of taxa, including plants with different life histories, as well as alpine dwelling animals, will need to be conducted before a more holistic understanding of diversity and dynamics among alpine populations can emerge. Such studies contribute not only to our understanding of general diversity/dynamics of alpine species, but would also inform the conservation of their focal species and also potentially contribute to our understanding of highly disputed glacial refugia for northeastern alpine taxa.

Notes and caveats

(1) Migration rate is notoriously difficult to estimate, directly or indirectly. One difficult aspect is that a reduction in gene flow is not detectable by some genetic methods until many

165 generations have passed; divergence through drift is especially slow in large populations and may result in weak population structure and/or high migration rate estimates that do not reflect current isolation (Allendorf et al., 2013). Indeed, some authors studying alpine species in post- glaciated areas have attributed weak population genetic structure to population mixing in refugia during glacial maxima rather than to ongoing gene flow (Bell et al., 2018 and sources therein). In this study, I interpreted the estimates of recent migration as truly reflecting ongoing gene flow among populations. This interpretation was based on the theoretical ability of these species to accomplish long distance migration via insect-dispersed pollen and wind-dispersed seed, as well as their typically small population size (many under 1000 individuals). Small population size increases the rate of divergence through drift in isolated populations––divergence I did not detect. Nevertheless, as BayesAss (like all methods) is not a perfect estimator, my estimates of recent migration should be interpreted cautiously. The low rates of flowering, low germination rates, and restricted geographic distribution exhibited by alpine Nabalus further suggest a cautious interpretation. Future studies should examine gene flow at different molecular markers and experimentally test dispersal ability in these species to confirm my findings of recent migration.

(2) Table 3-2 shows negative FIS values for all populations. For species with a mixed mating strategy, this finding seems unusual, as negative FIS values indicate a heterozygote excess and a bias toward outbreeding. However, there are several other potential causes: small effective population size, heterosis, asexual reproduction (discussed in Stoeckel et al., 2006), or paralogs in the dataset despite stringent filtering. Finally, it is also possible that some element of the biology of Nabalus spp. or their pollinators favors outbreeding over inbreeding.

166 (3) All morphological (functional trait) data presented in chapters 2 and 4 came from relatively young plants (2–3 months old, approximate length of the alpine growing season) sourced from just a few populations. Future studies should examine plasticity and morphological distinctiveness in and among these taxa in a greater number of populations and over a longer time span—ideally at least two years or generations—to confirm my findings.

Concluding thoughts

In a world faced with an acute biodiversity crisis, a finding that supports some degree of resilience in endemic, globally rare taxa is encouraging for conservation biologists. Nevertheless, future change and response is difficult to predict, as are tipping points that species may or may not reach in the coming decades. Additionally, not all mountaintop taxa (or similarly vulnerable groups) will harbor the same genetic diversity, migration potential, or plasticity as our focal

Nabalus spp. All told, even taxa with some ability to adapt, migrate, and/or tolerate change are likely to face significant challenges as the rate of global environmental change accelerates.

Although determining the extinction vulnerability of rare species and identifying conservation priorities within those species is vitally important for biodiversity conservation, ultimately, we desperately need policy change, individual behavioral change, creative solutions, and global cooperation to curb the sixth mass extinction.

167 LITERATURE CITED

Ackerly, D. D. (2003). Community assembly, niche conservatism, and adaptive evolution in changing environments. International Journal of Plant Sciences, 164(S3), S165–S184. https://doi.org/10.1086/368401

Ahrens, C. W., Supple, M. A., Aitken, N. C., Cantrill, D. J., Borevitz, J. O., & James, E. A. (2017). Genomic diversity guides conservation strategies among rare terrestrial orchid species when taxonomy remains uncertain. Annals of Botany, 119(8), 1267–1277. https://doi.org/10.1093/aob/mcx022

Alexander, J. M., Chalmandrier, L., Lenoir, J., Burgess, T. I., Essl, F., Haider, S., … Pellissier, L. (2018). Lags in the response of mountain plant communities to climate change. Global Change Biology, 24(2), 563–579. https://doi.org/10.1111/gcb.13976

Allen, P. J., Amos, W., Pomeroy, P. P., & Twiss, S. D. (1995). Microsatellite variation in grey seals (Halichoerus grypus) shows evidence of genetic differentiation between two British breeding colonies. Molecular Ecology, 4(6), 653–662. https://doi.org/10.1111/j.1365- 294X.1995.tb00266.x

Allendorf, F. W., Hohenlohe, P. A., & Luikart, G. (2010). Genomics and the future of conservation genetics. Nature Reviews Genetics, 11(10), 697–709. https://doi.org/10.1038/nrg2844

Allendorf, F. W., Luikart, G., & Aitken, S. N. (2013). Conservation and the genetics of populations. Chichester, UK: Wiley-Blackwell.

Anderson, J. T., & Gezon, Z. J. (2015). Plasticity in functional traits in the context of climate change: A case study of the subalpine forb Boechera stricta (Brassicaceae). Global Change Biology, 21(4), 1689–1703. https://doi.org/10.1111/gcb.12770

Angert, A. L., & Schemske, D. W. (2005). The evolution of species’ distributions: Reciprocal transplants across the elevation ranges of Mimulus cardinalis and M. lewisii. Evolution, 59(8), 1671–1684. https://doi.org/10.1554/05-107.1

Angert, A. L., Sheth, S. N., & Paul, J. R. (2011). Incorporating population-level variation in thermal performance into predictions of geographic range shifts. Integrative and Comparative Biology, 51(5), 733–750. https://doi.org/10.1093/icb/icr048

Babcock, E. B., Stebbins, G. L., & Jenkins, J. A. (1937). Chromosomes and phylogeny in some genera of the Crepidinae. Cytologia, Fujii Jubi, 188–210. https://doi.org/10.1508/cytologia.fujiijubilaei.188

Baird, N. A., Etter, P. D., Atwood, T. S., Currey, M. C., Shiver, A. L., Lewis, Z. A., … Johnson, E. A. (2008). Rapid SNP discovery and genetic mapping using sequenced RAD markers. PloS One, 3(10), e3376. https://doi.org/10.1371/journal.pone.0003376

168 Baker, C. S., Perry, A., Bannister, J. L., Weinrich, M. T., Abernethy, R. B., Calambokidis, J., … Vasquez, O. (1993). Abundant mitochondrial DNA variation and world-wide population structure in humpback whales. Proceedings of the National Academy of Sciences, 90(17), 8239–8243. https://doi.org/10.1073/pnas.90.17.8239

Balding, M., & Williams, K. J. H. (2016). Plant blindness and the implications for plant conservation. Conservation Biology, 30(6), 1192–1199. https://doi.org/10.1111/cobi.12738

Barnosky, A. D., Matzke, N., Tomiya, S., Wogan, G. O. U., Swartz, B., Quental, T. B., … Ferrer, E. A. (2011). Has the Earth’s sixth mass extinction already arrived? Nature, 471(7336), 51–57. https://doi.org/10.1038/nature09678

Barton, N. H., & Slatkin, M. (1986). A Quasi-equilibrium theory of the distribution of rare alleles in a subdivided population. Heredity, 56, 409–415. Retrieved from https://www.nature.com/articles/hdy198663.pdf

Basey, A. B., Fant, J. B., & Kramer, A. T. (2015). Producing native plant materials for restoration: 10 rules to collect and maintain genetic diversity. Native Plants, 16(1), 37–52.

Baskin, C. C., & Baskin, J. M. (2014). Seeds: Ecology, biogeography, and evolution of dormancy and germination (Second Edi). Amsterdam: Elsevier Academic Press.

Bates, D., Mächler, M., Bolker, B., & Walker, S. (2015). Fitting linear mixed-effects models using lme4. Journal of Statistical Software, 67(1), 1–48. https://doi.org/10.18637/jss.v067.i01

Bauert, M. R., Kälin, M., Baltisberger, M., & Edwards, P. J. (1998). No genetic variation detected within isolated relict populations of Saxifraga cernua in the Alps using RAPD markers. Molecular Ecology, 7, 1519–1527.

Baumgardner, R. E., Isil, S. S., Lavery, T. F., Rogers, C. M., & Mohnen, V. A. (2003). Estimates of cloud water deposition at mountain acid deposition program sites in the Appalachian Mountains. Journal of the Air and Waste Management Association, 53(3), 291–308. https://doi.org/10.1080/10473289.2003.10466153

Beaman, J. E., White, C. R., & Seebacher, F. (2016). Evolution of plasticity: Mechanistic link between development and reversible acclimation. Trends in Ecology and Evolution, 31(3), 237–249. https://doi.org/10.1016/j.tree.2016.01.004

Beatty, G. E., Brown, J. A., Cassidy, E. M., Finlay, C. M. V., McKendrick, L., Montgomery, W. I., … Provan, J. (2015). Lack of genetic structure and evidence for long-distance dispersal in ash (Fraxinus excelsior) populations under threat from an emergent fungal pathogen: Implications for restorative planting. Tree Genetics and Genomes, 11(3), 53. https://doi.org/10.1007/s11295-015-0879-5

169 Beckage, B., Osborne, B., Gavin, D. G., Pucko, C., Siccama, T., & Perkins, T. (2008). A rapid upward shift of a forest ecotone during 40 years of warming in the Green Mountains of Vermont. Proceedings of the National Academy of Sciences, 105(11), 4197–4202. https://doi.org/10.1073/pnas.0708921105

Beever, E. A., Perrine, J. D., Rickman, T., Flores, M., Clark, J. P., Waters, C., … Collins, G. H. (2016). Pika (Ochotona princeps) losses from two isolated regions reflect temperature and water balance, but reflect habitat area in a mainland region. Journal of Mammalogy, 97(6), 1495–1511. https://doi.org/10.1093/jmammal/gyw128

Bell, N., Griffin, P. C., Hoffmann, A. A., & Miller, A. D. (2018). Spatial patterns of genetic diversity among Australian alpine flora communities revealed by comparative phylogenomics. Journal of Biogeography, 45(1), 177–189. https://doi.org/10.1111/jbi.13120

Bengtsson, B. O., Weibull, P., & Ghatnekar, L. (1995). The loss of alleles by sampling: A study of the common outbreeding grass Festuca ovina over three geographic scales. Hereditas, 122, 221–238. https://doi.org/10.1111/j.1601-5223.1995.00221.x

Berdugo, M. B., & Dovciak, M. (2019). Bryophytes in fir waves: Forest canopy indicator species and functional diversity decline in canopy gaps. Journal of Vegetation Science, 30(2), 235– 246. https://doi.org/10.1111/jvs.12718

Berend, K., Haynes, K., & McDonough Mackenzie, C. (2019). Common garden experiments as a dynamic tool for ecological studies of alpine plants and communities in northeastern North America. Rhodora, 121(987), 174–212. https://doi.org/10.3119/18-16

Bierman, P. R., Davis, P. T., Corbett, L. B., Lifton, N. A., & Finkel, R. C. (2015). Cold-based Laurentide ice covered New England’s highest summits during the Last Glacial Maximum. Geology, 43(12), 1059–1062. https://doi.org/10.1130/G37225.1

Billings, W. D., & Mooney, H. A. (1968). The ecology of arctic and alpine plants. Biological Reviews, 43(4), 481–529. https://doi.org/10.1111/j.1469-185X.1968.tb00968.x

Bliss, L. C. (1956). A comparison of plant development in microenvironments of arctic and alpine tundras. Ecological Monographs, 26(4), 303–337. Retrieved from https://www.jstor.org

Bobbink, R. B., Hicks, K., Galloway, J., Pranger, T., Alkemade, R., Ashmore, M., … De Vries, W. (2010). Global assessment of nitrogen deposition effects on terrestrial plant diversity: A synthesis. Ecological Applications, 20(1), 30–59.

Bogler, D. J. (2006). Prenanthes. In Flora of North America Association (Ed.), Flora of North America north of Mexico (pp. 264–271). New York and Oxford: Oxford University Press.

Bolker, B., & R Development Core Team. (2017). bbmle: Tools for General Maximum Likelihood Estimation. Retrieved from https://cran.r-project.org/package=bbmle

170 Bolnick, D. I., Fordyce, J. A., Yang, L. H., Davis, J. M., Hulsey, C. D., Forister, M. L., & Svanbäck, R. (2003). The ecology of individuals: Incidence and implications of individual specialization. American Naturalist, 161(1), 1–28. https://doi.org/10.1086/343878

Bolnick, D. I., Ingram, T., Stutz, W. E., Snowberg, L. K., Lau, O. L., & Pauli, J. S. (2010). Ecological release from interspecific competition leads to decoupled changes in population and individual niche width. Proceedings of the Royal Society B: Biological Sciences, 277(1689), 1789–1797. https://doi.org/10.1098/rspb.2010.0018

Bossart, J. L., & Pashley, D. (1998). Genetic estimates of population structure and gene flow: Limitations, lessons and new directions. Trends in Ecology and Evolution, 13(5), 202–206. https://doi.org/10.1016/S0169-5347(97)01284-6

Bouchard, J. R., Fernando, D. D., Bailey, S. W., Weber-Townsend, J., & Leopold, D. J. (2017). Contrasting patterns of genetic variation in central and peripheral populations of Dryopteris fragrans (fragrant wood fern) and implications for colonization dynamics and conservation. International Journal of Plant Sciences, 178(8), 607–617. https://doi.org/10.1086/693109

Bradshaw, A. D. (1965). Evolutionary significance of phenotypic plasticity in plants. Advances in Genetics, 13, 115–155.

Britton, N. L., & Brown, A. (1913). An illustrated flora of the northern United States, Canada and the British Possessions, vol 3. (2nd ed.). Retrieved from USDA-NRCS PLANTS Database

Brochmann, C., Brysting, A. K., Alsos, I. G., Borgen, L., Grundt, H. H., Scheen, A. C., & Elven, R. (2004). Polyploidy in arctic plants. Biological Journal of the Linnean Society, 82(4), 521–536. https://doi.org/10.1111/j.1095-8312.2004.00337.x

Brochmann, C., Gabrielsen, T. M., Nordal, I., Landvik, J. Y., & Elven, R. (2003). Glacial survival or tabula rasa? The history of North Atlantic biota revisited. Taxon, 52(3), 417– 450.

Brook, B. W., Sodhi, N. S., & Bradshaw, C. J. A. (2008). Synergies among extinction drivers under global change. Trends in Ecology and Evolution, 23(8), 453–460. https://doi.org/10.1016/j.tree.2008.03.011

Brown, J. H. (1984). On the relationship between abundance and distribution of species. The American Naturalist, 124(2), 255–279. https://doi.org/10.1086/284267

Bubac, C. M., & Spellman, G. M. (2016). How connectivity shapes genetic structure during range expansion: Insights from the Virginia’s Warbler. The Auk, 133(2), 213–230. https://doi.org/10.1642/auk-15-124.1

Burns, M., Hedin, M., & Tsurusaki, N. (2018). Population genomics and geographical parthenogenesis in Japanese harvestmen (Opiliones, Sclerosomatidae, Leiobunum). Ecology and Evolution, 8(1), 36–52. https://doi.org/10.1002/ece3.3605

171 Cahill, A. E., Aiello-Lammens, M. E., Fisher-Reid, M. C., Hua, X., Karanewsky, C. J., Yeong Ryu, H., … Wiens, J. J. (2013). How does climate change cause extinction? Proceedings of the Royal Society B: Biological Sciences, 280, 20121890. https://doi.org/10.1098/rspb.2012.1890

Cai, L., Xi, Z., Amorim, A. M., Sugumaran, M., Rest, J. S., Liu, L., & Davis, C. C. (2019). Widespread ancient whole-genome duplications in Malpighiales coincide with Eocene global climatic upheaval. New Phytologist, 221, 565–576. https://doi.org/10.1111/nph.15357

Cain, M. L., Milligan, B. G., & Strand, A. E. (2000). Long-distance seed dispersal in plant populations. American Journal of Botany, 87(9), 1217–1227. Retrieved from https://pdfs.semanticscholar.org

Capers, R. S., Kimball, K. D., McFarland, K. P., Jones, M. T., Lloyd, A. H., Munroe, J. S., … Paradis, R. (2013). Establishing alpine research priorities in northeastern North America. Northeastern Naturalist, 20(4), 559–577. https://doi.org/10.1656/045.020.0406

Capers, R. S., & Slack, N. G. (2016). A baseline study of alpine snowbed and rill communities on Mount Washington, NH. Rhodora, 118(976), 345–381. https://doi.org/10.3119/16-07

Cardillo, M., Dinnage, R., & McAlister, W. (2019). The relationship between environmental niche breadth and geographic range size across plant species. Journal of Biogeography, 46(1), 97–109. https://doi.org/10.1111/jbi.13477

Catchen, J., Hohenlohe, P. A., Bassham, S., Amores, A., & Cresko, W. A. (2013). Stacks: An analysis tool set for population genomics. Molecular Ecology, 22(11), 3124–3140. https://doi.org/10.1111/mec.12354

Caujapé-Castells, J., & Pedrola-Monfort, J. (2004). Designing ex-situ conservation strategies through the assessment of neutral genetic markers: Application to the endangered Androcymbium gramineum. Conservation Genetics, 5(2), 131–144. https://doi.org/10.1023/B:COGE.0000029997.59502.88

Ceballos, G., Ehrlich, P. R., Barnosky, A. D., García, A., Pringle, R. M., & Palmer, T. M. (2015). Accelerated modern human-induced species losses: Entering the sixth mass extinction. Science Advances, 1(5). https://doi.org/10.1126/sciadv.1400253

Ceska, J. F., Affolter, J. M., & Hamrick, J. L. (1997). Developing a sampling strategy for Baptisia arachnifera based on allozyme diversity. Conservation Biology, 11(5), 1133–1139. https://doi.org/10.1046/j.1523-1739.1997.95527.x

Chalker-Scott, L. (1999). Environmental significance of anthocyanins in plant stress responses. Photochemistry and Photobiology, 70(1), 1–9. https://doi.org/10.1111/j.1751- 1097.1999.tb01944.x

172 Charmantier, A., McCleery, R. H., Cole, L. R., Perrins, C., Kruuk, L. E. B., & Sheldon, B. C. (2008). Adaptive phenotypic plasticity in response to climate change in a wild bird population. Science, 320(5877), 800–803. https://doi.org/10.1126/science.1157174

Chessel, D., Dufour, A. B., & Thioulouse, J. (2004). The ade4 package - I: One-table methods. R News, 4(1), 5–10. Retrieved from http://pbil.univ-lyon1.fr/

Chevin, L. M., Lande, R., & Mace, G. M. (2010). Adaptation, plasticity, and extinction in a changing environment: Towards a predictive theory. PLoS Biology, 8(4). https://doi.org/10.1371/journal.pbio.1000357

Choler, P. (2005). Consistent shifts in alpine plant traits along a mesotopographical gradient. Arctic, Antarctic, and Alpine Research, 37(4), 444–453. https://doi.org/10.1657/1523- 0430(2005)037[0444:CSIAPT]2.0.CO;2

Choler, P., Michalet, R., & Callaway, R. M. (2001). Facilitation and competition on gradients in alpine plant communities. Ecology, 82(12), 3295–3308. https://doi.org/10.1890/0012- 9658(2001)082[3295:FACOGI]2.0.CO;2

Close, D. C., & Beadle, C. L. (2003). The ecophysiology of foliar anthocyanin. Botanical Review, 69(2), 149–161. https://doi.org/Doi 10.1663/0006- 8101(2003)069[0149:Teofa]2.0.Co;2

Comai, L. (2005). The advantages and disadvantages of being polyploid. Nature Reviews Genetics, 6(11), 836–846. https://doi.org/10.1038/nrg1711

Conner, J. K., & Hartl, D. L. (2004). A primer of ecological genetics (1st ed.). Sunderland, MA: Sinauer Associates.

Corlett, R. T. (2017). A bigger toolbox: Biotechnology in biodiversity conservation. Trends in Biotechnology, 35(1), 55–65. https://doi.org/10.1016/j.tibtech.2016.06.009

Costion, C. M., Simpson, L., Pert, P. L., Carlsen, M. M., John Kress, W., & Crayn, D. (2015). Will tropical mountaintop plant species survive climate change? Identifying key knowledge gaps using species distribution modelling in Australia. Biological Conservation, 191, 322– 330. https://doi.org/10.1016/j.biocon.2015.07.022

Covington, W. W. (1975). Altitudinal variation of chlorophyll concentration and reflectance of the bark of Populus tremuloides. Ecology, 56(3), 715–720. https://doi.org/10.2307/1935507

Crandall, K. A., Bininda-Emonds, O. R. R., Mace, G. M., & Wayne, R. K. (2000). Considering evolutionary processes in conservation biology. Trends in Ecology and Evolution, 15(7), 290–295. https://doi.org/10.1016/S0169-5347(00)01876-0

Dalgleish, H. J., Koons, D. N., & Adler, P. B. (2010). Can life-history traits predict the response of forb populations to changes in climate variability? Journal of Ecology, 98(1), 209–217. https://doi.org/10.1111/j.1365-2745.2009.01585.x

173 Danecek, P., Auton, A., Abecasis, G., Albers, C. A., Banks, E., DePristo, M. A., … Durbin, R. (2011). The variant call format and VCFtools. Bioinformatics, 27(15), 2156–2158. https://doi.org/10.1093/bioinformatics/btr330

Dar, T.-U.-H., & Rehman, R.-U. (2017). Polyploidy in changing environment. In Polyploidy: Recent trends and future perspectives (pp. 88–99). New Delhi: Springer.

Davis, M. B., & Shaw, R. G. (2001). Range shifts and adaptive responses to quaternary climate change. Science, 292(5517), 673–679. https://doi.org/10.1126/science.292.5517.673

Davis, M. B., Shaw, R. G., & Etterson, J. R. (2005). Evolutionary responses to changing climate. Ecology, 86(7), 1704–1714.

Dawe, K. L., Shafer, A. B. A., Herman, T. B., & Stewart, D. T. (2009). Diffusion of nuclear and mitochondrial genes across a zone of secondary contact in the maritime shrew, Sorex maritimensis: Implications for the conservation of a Canadian endemic mammal. Conservation Genetics, 10(4), 851–857. https://doi.org/10.1007/s10592-008-9645-7

De Boeck, H. J., Bassin, S., Verlinden, M., Zeiter, M., & Hiltbrunner, E. (2016). Simulated heat waves affected alpine grassland only in combination with drought. New Phytologist, 209(2), 531–541. https://doi.org/10.1111/nph.13601

De Witte, L. C., Armbruster, G. F. J., Gielly, L., Taberlet, P., & Stöcklin, J. (2012). AFLP markers reveal high clonal diversity and extreme longevity in four key arctic-alpine species. Molecular Ecology, 21(5), 1081–1097. https://doi.org/10.1111/j.1365-294X.2011.05326.x

DeLuca, W. V., & King, D. I. (2017). Montane birds shift downslope despite recent warming in the northern Appalachian Mountains. Journal of Ornithology, 158(2), 493–505. https://doi.org/10.1007/s10336-016-1414-7

Dirnböck, T., Dullinger, S., & Grabherr, G. (2003). A regional impact assessment of climate and land use change on alpine vegetation. Journal of Biogeography, 30, 1–17. https://doi.org/10.1046/j.1365-2699.2003.00839.x

Dirnböck, T., Essl, F., & Rabitsch, W. (2011). Disproportional risk for habitat loss of high- altitude endemic species under climate change. Global Change Biology, 17(2), 990–996. https://doi.org/10.1111/j.1365-2486.2010.02266.x

Dizon, A. E., Lockyer, C., Perrin, W. F., Demaster, D. P., & Sisson, J. (1992). Rethinking the stock concept: A phylogeographic approach. Conservation Biology, 6(1), 24–36. Retrieved from https://www.jstor.org

Do Amaral, E. S., Vieira Silva, D., Dos Anjos, L., Schilling, A. C., Dalmolin, Â. C., & Mielke, M. S. (2019). Relationships between reflectance and absorbance chlorophyll indices with RGB (red, green, blue) image components in seedlings of tropical tree species at nursery stage. New Forests, 50(3), 377–388. https://doi.org/10.1007/s11056-018-9662-4

174 Doadrio, I., Perdices, A., & Machordom, A. (1996). Allozymic variation of the endangered killifish Aphanius iberus and its application to conservation. Environmental Biology of Fishes, 45(3), 259–271. https://doi.org/10.1007/BF00003094

Doak, D. F., & Morris, W. F. (2010). Demographic compensation and tipping points in climate- induced range shifts. Nature, 467(7318), 959–962. https://doi.org/10.1038/nature09439

Donohue, K. (2005). Niche construction through phenological plasticity: Life history dynamics and ecological consequences. New Phytologist, 166(1), 83–92. https://doi.org/10.1111/j.1469-8137.2005.01357.x

Donohue, K., Rubio de Casas, R., Burghardt, L., Kovach, K., & Willis, C. G. (2010). Germination, postgermination adaptation, and species ecological ranges. Annual Review of Ecology, Evolution, and Systematics, 41(1), 293–319. https://doi.org/10.1146/annurev- ecolsys-102209-144715

Dool, S. E., O’Donnell, C. F. J., Monks, J. M., Puechmaille, S. J., & Kerth, G. (2016). Phylogeographic-based conservation implications for the New Zealand long-tailed bat, (Chalinolobus tuberculatus): Identification of a single ESU and a candidate population for genetic rescue. Conservation Genetics, 17(5), 1067–1079. https://doi.org/10.1007/s10592- 016-0844-3

Dullinger, S., Gattringer, A., Thuiller, W., Moser, D., Zimmermann, N. E., Guisan, A., … Hülber, K. (2012). Extinction debt of high-mountain plants under twenty-first-century climate change. Nature Climate Change, 2(8), 619–622. https://doi.org/10.1038/nclimate1514

Dyke, A. S., Andrews, J. T., Clark, P. U., England, J. H., Miller, G. H., Shaw, J., & Veillette, J. J. (2002). The Laurentide and Innuitian ice sheets during the Last Glacial Maximum. Quaternary Science Reviews, 21, 9–31. https://doi.org/10.1016/S0277-3791(01)00095-6

Eastman, T. (2018, July 20). State to study Mount Washington’s summit capacity. The Conway Daily Sun. Retrieved from https://www.conwaydailysun.com/news/local/state-to-study- mount-washington-s-summit-capacity/article_15ec12d4-8b6f-11e8-92e2- 3bf94ddcf750.html

Ellegren, H., & Galtier, N. (2016). Determinants of genetic diversity. Nature Reviews Genetics, 17(7), 422–433. https://doi.org/10.1038/nrg.2016.58

Ellstrand, N. C., & Elam, D. R. (1993). Population genetic consequences of small population size: Implications for plant conservation. Annual Review of Ecology and Systematics, 24(1), 217–242. https://doi.org/10.1146/annurev.es.24.110193.001245

Elsen, P. R., & Tingley, M. W. (2015). Global mountain topography and the fate of montane species under climate change. Nature Climate Change, 5(8), 772–776. https://doi.org/10.1038/nclimate2656

175 Elshire, R. J., Glaubitz, J. C., Sun, Q., Poland, J. A., Kawamoto, K., Buckler, E. S., & Mitchell, S. E. (2011). A robust, simple genotyping-by-sequencing (GBS) approach for high diversity species. PloS One, 6(5), e19379. https://doi.org/10.1371/journal.pone.0019379

Enquist, B. J., Norberg, J., Bonser, S. P., Violle, C., Webb, C. T., Henderson, A., … Savage, V. M. (2015). Scaling from traits to ecosystems: Developing a general trait driver theory via integrating trait-based and metabolic scaling theories. Advances in Ecological Research, 52, 249–318. https://doi.org/10.1016/bs.aecr.2015.02.001

Ersts, P. J. (n.d.). Geographic distance matrix generator version 1.2.3. American Museum of Natural History, Center for Biodiversity and Conservation.

Ettinger, A. K., Ford, K. R., & HilleRisLambers, J. (2011). Climate determines upper, but not lower, altitudinal range limits of Pacific Northwest conifers. Ecology, 92(6), 1323–1331. https://doi.org/10.1890/10-1639.1

Evanno, G., Regnaut, S., & Goudet, J. (2005). Detecting the number of clusters of individuals using the software STRUCTURE: A simulation study. Molecular Ecology, 14(8), 2611– 2620. https://doi.org/10.1111/j.1365-294X.2005.02553.x

Excoffier, L., & Lischer, H. E. L. (2010). Arlequin suite ver 3.5: A new series of programs to perform population genetics analyses under Linux and Windows. Molecular Ecology Resources, 10(3), 564–567. https://doi.org/10.1111/j.1755-0998.2010.02847.x

Falush, D., Stephens, M., & Pritchard, J. K. (2003). Inference of population structure using multilocus genotype data: Linked loci and correlated allele frequencies. Genetics, 164, 1567–1587. Retrieved from http://pritch.

Fan, F., Bradley, R. S., & Rawlins, M. A. (2015). Climate change in the northeast United States: An analysis of the NARCCAP multimodel simulations. Journal of Geophysical Research: Atmospheres, 120, 10,569-10,592. https://doi.org/10.1002/2015JD023073.Received

Fawcett, J. A., Maere, S., & Van de Peer, Y. (2009). Plants with double genomes might have had a better chance to survive the Cretaceous-Tertiary extinction event. Proceedings of the National Academy of Sciences, 106(14), 5737–5742. https://doi.org/10.1073/pnas.0900906106

Fay, P. A., & Schultz, M. J. (2009). Germination, survival, and growth of grass and forb seedlings: Effects of soil moisture variability. Acta Oecologica, 35(5), 679–684. https://doi.org/10.1016/j.actao.2009.06.007

Fernandez, P., Gálvez, F., Garcia, S., Garnatje, T., Gras, A., Hidalgo, O., … Vallès, J. (2018). GSAD: Genome size in Asteraceae database. Release 3.0. Retrieved July 9, 2019, from http://www.etnobiofic.cat/gsad_v2/

176 Filella, I., & Peñuelas, J. (1999). Altitudinal differences in UV absorbance, UV reflectance and related morphological traits of Quercus ilex and Rhododendron ferrugineum in the Mediterranean region. Plant Ecology, 145(1), 157–165. https://doi.org/10.1023/A:1009826803540

Finch, J., Walck, J. L., Hidayati, S. N., Kramer, A. T., Lason, V., & Havens, K. (2019). Germination niche breadth varies inconsistently among three Asclepias congeners along a latitudinal gradient. Plant Biology, 21(3), 425–438. https://doi.org/10.1111/plb.12843

Foden, W. B., Young, B. E., Akçakaya, H. R., Garcia, R. A., Hoffmann, A. A., Stein, B. A., … Huntley, B. (2019). Climate change vulnerability assessment of species. Wiley Interdisciplinary Reviews: Climate Change, 1–36. https://doi.org/10.1002/wcc.551

Foster, J. R., & D’Amato, A. W. (2015). Montane forest ecotones moved downslope in northeastern USA in spite of warming between 1984 and 2011. Global Change Biology, 21(12), 4497–4507. https://doi.org/10.1111/gcb.13046

Frankham, R. (2015). Genetic rescue of small inbred populations: Meta-analysis reveals large and consistent benefits of gene flow. Molecular Ecology, 24(11), 2610–2618. https://doi.org/10.1111/mec.13139

Frankham, R., Ballou, J. D., Eldridge, M. D. B., Lacy, R. C., Ralls, K., Dudash, M. R., & Fenster, C. B. (2011). Predicting the probability of outbreeding depression. Conservation Biology, 25(3), 465–475. https://doi.org/10.1111/j.1523-1739.2011.01662.x

Frankham, R., Bradshaw, C. J. A., & Brook, B. W. (2014). Genetics in conservation management: Revised recommendations for the 50/500 rules, Red List criteria and population viability analyses. Biological Conservation, 170, 56–63. https://doi.org/10.1016/j.biocon.2013.12.036

Fraser, D. J., & Bernatchez, L. (2001). Adaptive evolutionary conservation: Towards a unified concept for defining conservation units. Molecular Ecology, 10(12), 2741–2752. https://doi.org/10.1046/j.1365-294X.2001.t01-1-01411.x

Freeman, B. G., Scholer, M. N., Ruiz-Gutierrez, V., & Fitzpatrick, J. W. (2018). Climate change causes upslope shifts and mountaintop extirpations in a tropical bird community. Proceedings of the National Academy of Sciences, 115(47), 201804224. https://doi.org/10.1073/pnas.1804224115

Funk, J. L., Larson, J. E., Ames, G. M., Butterfield, B. J., Cavender-Bares, J., Firn, J., … Wright, J. (2017). Revisiting the Holy Grail: Using plant functional traits to understand ecological processes. Biological Reviews, 92(2), 1156–1173. https://doi.org/10.1111/brv.12275

Funk, W. C., McKay, J. K., Hohenlohe, P. A., & Allendorf, F. W. (2012). Harnessing genomics for delineating conservation units. Trends in Ecology and Evolution, 27(9), 489–496. https://doi.org/10.1016/j.tree.2012.05.012

177 Galloway, J. N., Likens, G. E., & Hawley, M. E. (1984). Acid precipitation: Natural versus anthropogenic components. Science, 226(4676), 829–831.

Garnatje, T., Canela, M. Á., Garcia, S., Hidalgo, O., Pellicer, J., Sánchez-Jiménez, I., … Vallès, J. (2011). GSAD: A genome size in the Asteraceae database. Cytometry Part A, 79A(6), 401–404. https://doi.org/10.1002/cyto.a.21056

Garnett, S. T., & Christidis, L. (2017). Taxonomy anarchy hampers conservation. Nature, 546(7656), 25–27. https://doi.org/10.1038/546025a

Garrison, E. (2012). vcflib: A C++ library for parsing and manipulating VCF files. Retrieved from https://github.com/vcflib/vcflib

Gaston, K. (1996). Rarity (M. B. Usher, D. L. DeAngelis, & R. L. Kitching, Eds.). London: Chapman & Hall.

Gilpin, M. E., & Soulé, M. E. (1986). Minimum viable populations: Processes of species extinction. In M. E. Soulé (Ed.), Conservation biology: The science of scarcity and diversity (pp. 19–34). Sunderland, MA: Sinauer.

Girardin, C. A. J., Malhi, Y., Aragão, L. E. O. C., Mamani, M., Huaraca Huasco, W., Durand, L., … Whittaker, R. J. (2010). Net primary productivity allocation and cycling of carbon along a tropical forest elevational transect in the Peruvian Andes. Global Change Biology, 16(12), 3176–3192. https://doi.org/10.1111/j.1365-2486.2010.02235.x

Gitzendanner, M. A., & Soltis, P. S. (2000). Patterns of genetic variation in rare and widespread plant congeners. American Journal of Botany, 87(6), 783–792. https://doi.org/10.2307/2656886

Gleason, H. A., & Cronquist, A. (1963). Manual of vascular plants of northeastern United States and adjacent Canada. Retrieved from http://www.sciencemag.org/cgi/doi/10.1126/science.140.3567.637

Gompert, Z., & Mock, K. E. (2017). Detection of individual ploidy levels with genotyping-by- sequencing (GBS) analysis. Molecular Ecology Resources, 17(6), 1156–1167. https://doi.org/10.1111/1755-0998.12657

Gorelick, R., & Heng, H. H. Q. (2011). Sex reduces genetic variation: A multidisciplinary review. Evolution, 65, 1088–1098. https://doi.org/10.1111/j.1558-5646.2010.01173.x

Goren, J., & Monz, C. (2011). Monitoring vegetation changes in historical photos over a 45+ year period in the Adirondack alpine zone. Adirondack Journal of Environmental Studies, 17, 29–35. Retrieved from https://digitalcommons.usu.edu/envs_facpub/922

Graham, C. F., Glenn, T. C., Mcarthur, A. G., Boreham, D. R., Kieran, T., Lance, S., … Somers, C. M. (2015). Impacts of degraded DNA on restriction enzyme associated DNA sequencing (RADSeq). Molecular Ecology Resources, 15(6), 1304–1315. https://doi.org/10.1111/1755- 0998.12404

178 Griffith, T., & Sultan, S. E. (2012). Field-based insights to the evolution of specialization: Plasticity and fitness across habitats in a specialist/generalist species pair. Ecology and Evolution, 2(4), 778–791. https://doi.org/10.1002/ece3.202

Grime, J. P. (1977). Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. The American Naturalist, 111(982), 1169– 1194. https://doi.org/10.1086/283244

Grueber, C. E. (2015). Comparative genomics for biodiversity conservation. Computational and Structural Biotechnology Journal, 13, 370–375. https://doi.org/10.1016/j.csbj.2015.05.003

Gugger, S., Kesselring, H., Stöcklin, J., & Hamann, E. (2015). Lower plasticity exhibited by high- versus mid-elevation species in their phenological responses to manipulated temperature and drought. Annals of Botany, 116(6), 953–962. https://doi.org/10.1093/aob/mcv155

Haines, A., Farnsworth, E., & Morrison, G. (2011). New England Wildflower Society’s flora Novae Angliae: A manual for the identification of native and naturalized vascular plants of New England. New Haven, CT: Yale University Press.

Hamann, E., Kesselring, H., & Stöcklin, J. (2018). Plant responses to simulated warming and drought: A comparative study of functional plasticity between congeneric mid and high elevation species. Journal of Plant Ecology, 11(3), 364–374. https://doi.org/10.1093/jpe/rtx023

Hampe, A., & Jump, A. S. (2011). Climate relicts: Past, present, future. Annual Review of Ecology, Evolution, and Systematics, 42(1), 313–333. https://doi.org/10.1146/annurev- ecolsys-102710-145015

Hampe, A., & Petit, R. J. (2005). Conserving biodiversity under climate change: The rear edge matters. Ecology Letters, 8(5), 461–467. https://doi.org/10.1111/j.1461-0248.2005.00739.x

Hamrick, J. L., & Godt, M. J. W. (1990). Allozyme diversity in plant species. In A. H. D. Brown, M. T. Clegg, A. L. Kahler, & B. S. Weir (Eds.), Plant population genetics, breeding, and genetic resources (pp. 43–63). Sunderland, MA: Sinauer.

Hamrick, J. L., & Godt, M. J. W. (1996). Conservation genetics of endemic plants. In J. C. Avise & J. L. Hamrick (Eds.), Conservation genetics: Case histories from nature (pp. 281–304). New York: Chapman & Hall.

Hamrick, J. L., Godt, M. J. W., Murawski, D. A., & Loveless, M. D. (1991). Correlations between species traits and allozyme diversity: Implications for conservation biology. In D. A. Falk & K. Holsinger (Eds.), Genetics and conservation of rare plants (pp. 75–86). New York: Oxford University Press.

Hamrick, J. L., Godt, M. J. W., & Sherman-Broyles, S. L. (1992). Factors influencing levels of genetic diversity in woody plant species. New Forests, 6, 95–124. https://doi.org/10.1007/978-94-011-2815-5

179 Harnik, P. G., Simpson, C., & Payne, J. L. (2012). Long-term differences in extinction risk among the seven forms of rarity. Proceedings of the Royal Society B: Biological Sciences, 279(1749), 4969–4976. https://doi.org/10.1098/rspb.2012.1902

Harrison, J. G., Forister, M. L., Mcknight, S. R., Nordin, E., & Parchman, T. L. (2019). Rarity does not limit genetic variation or preclude subpopulation structure in the geographically restricted desert forb Astragalus lentiginosus var. piscinensis. American Journal of Botany, 106(2), 260–269. https://doi.org/10.1002/ajb2.1235

Harrisson, K. A., Pavlova, A., Telonis-Scott, M., & Sunnucks, P. (2014). Using genomics to characterize evolutionary potential for conservation of wild populations. Evolutionary Applications, 7(9), 1008–1025. https://doi.org/10.1111/eva.12149

Havens, K., Kramer, A. T., & Guerrant, E. O. (2013). Getting plant conservation right (or not): The case of the United States. International Journal of Plant Sciences, 175(1), 3–10. https://doi.org/10.1086/674103

Henn, J. J., Buzzard, V., Enquist, B. J., Halbritter, A. H., Klanderud, K., Maitner, B. S., … Vandvik, V. (2018). Intraspecific trait variation and phenotypic plasticity mediate alpine plant species response to climate change. Frontiers in Plant Science, 9, 1548. https://doi.org/10.3389/fpls.2018.01548

Hirst, M. J., Griffin, P. C., Sexton, J. P., & Hoffmann, A. A. (2017). Testing the niche-breadth– range-size hypothesis: Habitat specialization vs. performance in Australian alpine daisies. Ecology, 98(10), 2708–2724. https://doi.org/10.1002/ecy.1964

Howard, T., White, K., & Goren, J. (2019). Monitoring plant populations in the Adirondack alpine. Lake Placid, NY: Presentation at the 11th Northeastern Alpine Stewardship Gathering.

Hu, H., Liu, H. Q., Zhang, H., Zhu, J. H., Yao, X. G., Zhang, X. Bin, & Zheng, K. F. (2010). Assessment of chlorophyll content based on image color analysis, comparison with SPAD- 502. 2nd International Conference on Information Engineering and Computer Science - Proceedings, ICIECS 2010, 476–478. https://doi.org/10.1109/ICIECS.2010.5678413

Hubisz, M. J., Falush, D., Stephens, M., & Pritchard, J. K. (2009). Inferring weak population structure with the assistance of sample group information. Molecular Ecology Resources, 9(5), 1322–1332. https://doi.org/10.1111/j.1755-0998.2009.02591.x

Hutchison, D. W., & Templeton, A. R. (1999). Correlation of pairwise genetic and geographic distance measures: Inferring the relative influences of gene flow and drift on the distribution of genetic variability. Evolution, 53(6), 1898–1914. https://doi.org/10.2307/2640449

International Union for Conservation of Nature and Natural Resources. (2019). The IUCN red list of threatened species. Retrieved July 24, 2019, from Version 2019-2 website: https://www.iucnredlist.org/

180 Jansen, E., Overpeck, J., Briffa, K. R., Duplessy, J., Joos, F., Masson-Delmotte, V., … Zhang, D. (2007). Palaeoclimate. In S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K. B. Averyt, … H. L. Miller (Eds.), Climate change 2007: The physical science basis. Contribution of Working Group I to the fourth assessment report of the Intergovernmental Panel on Climate Change (pp. 433–497). https://doi.org/10.2753/JES1097-203X330403

Jeong, J. H., Lee, B. C., Yoo, K. O., Jang, S. K., & Kim, Z. S. (2012). Influence of small-scale habitat patchiness on the genetic diversity of the Korean endemic species Saussurea chabyoungsanica (Asteraceae). Biochemical Systematics and Ecology, 43, 14–24. https://doi.org/10.1016/j.bse.2012.02.015

Jiménez-Alfaro, B., García-Calvo, L., García, P., & Acebes, J. L. (2016). Anticipating extinctions of glacial relict populations in mountain refugia. Biological Conservation, 201, 243–251. https://doi.org/10.1016/j.biocon.2016.07.015

Jombart, T. (2008). adegenet: A R package for the multivariate analysis of genetic markers. Bioinformatics, 24(11), 1403–1405. https://doi.org/10.1093/bioinformatics/btn129

Jombart, T., & Ahmed, I. (2011). adegenet 1.3-1: New tools for the analysis of genome-wide SNP data. Bioinformatics, 27(21), 3070–3071. https://doi.org/10.1093/bioinformatics/btr521

Jombart, T., & Collins, C. (2015). A tutorial for discriminant analysis of principal components (DAPC) using adegenet 2.0.0. R Vignette, pp. 1–37. https://doi.org/10.1038/72708

Jombart, T., Devillard, S., & Balloux, F. (2010). Discriminant analysis of principal components: A new method for the analysis of genetically structured populations. BMC Genetics, 11(1), 94. https://doi.org/10.1186/1471-2156-11-94

Jones, M. T., & Willey, L. L. (2012). Eastern alpine guide. New Salem, MA: Beyond Ktaadn and Boghaunter Books.

Jones, S. B. (1970). Chromosome numbers in Compositae. Bulletin of the Torrey Botanical Club, 97(3), 168–171. Retrieved from https://www.jstor.org

Jump, A. S., & Peñuelas, J. (2005). Running to stand still: Adaptation and the response of plants to rapid climate change. Ecology Letters, 8(9), 1010–1020. https://doi.org/10.1111/j.1461- 0248.2005.00796.x

Kamvar, Z. N., Brooks, J. C., & Grünwald, N. J. (2015). Novel R tools for analysis of genome- wide population genetic data with emphasis on clonality. Frontiers in Genetics, 6, 208. https://doi.org/10.3389/fgene.2015.00208

Kamvar, Z. N., Tabima, J. F., & Grünwald, N. J. (2014). Poppr: An R package for genetic analysis of populations with clonal, partially clonal, and/or sexual reproduction. PeerJ, 2, e281. https://doi.org/10.7717/peerj.281

181 Kawecki, T. J. (2008). Adaptation to marginal habitats. Annual Review of Ecology, Evolution, and Systematics, 39(1), 321–342. https://doi.org/10.1146/annurev.ecolsys.38.091206.095622

Keenan, K., McGinnity, P., Cross, T. F., Crozier, W. W., & Prodöhl, P. A. (2013). diveRsity: An R package for the estimation and exploration of population genetics parameters and their associated errors. Methods in Ecology and Evolution, 4(8), 782–788. https://doi.org/10.1111/2041-210X.12067

Kim, E., & Donohue, K. (2013). Local adaptation and plasticity of Erysimum capitatum to altitude: Its implications for responses to climate change. Journal of Ecology, 101(3), 796– 805. https://doi.org/10.1111/1365-2745.12077

Kim, S.-C., Crawford, D. J., & Jansen, R. K. (1996). Phylogenetic relationships among the genera of the subtribe Sonchinae (Asteraceae): Evidence from ITS sequences. Systematic Botany, 21(3), 417. https://doi.org/10.2307/2419668

Kimball, K. D., Davis, M. L., Davis, M. L., Weihrauch, D. M., Murray, G. L. D., & Rancourt, K. (2014). Limited alpine climatic warming and modeled phenology advancement for three alpine species in the northeast United States. American Journal of Botany, 101(9), 1437– 1446. https://doi.org/10.3732/ajb.1400214

Kimball, K. D., & Weihrauch, D. M. (2000). Alpine vegetation communities and the alpine- treeline ecotone boundary in New England as biomonitors for climate change. In USDA Forest Service Proceedings RMRS-P-15-VOL-3 (Vol. 3). USDA Forest Service.

Kissling, D. W., Pattemore, D. E., & Hagen, M. (2014). Challenges and prospects in the telemetry of insects. Biological Reviews, 89(3), 511–530. https://doi.org/10.1111/brv.12065

Kopelman, N. M., Mayzel, J., Jakobsson, M., Rosenberg, N. A., & Mayrose, I. (2015). Clumpak: A program for identifying clustering modes and packaging population structure inferences across K. Molecular Ecology Resources, 15(5), 1179–1191. https://doi.org/10.1111/1755- 0998.12387

Körner, C. (2003). Alpine plant life: Functional plant ecology of high mountain ecosystems (2nd ed.). https://doi.org/10.1659/0276-4741(2001)021[0202:APLFPE]2.0.CO;2

Körner, C., Basler, D., Hoch, G., Kollas, C., Lenz, A., Randin, C. F., … Zimmermann, N. E. (2016). Where, why and how? Explaining the low-temperature range limits of temperate tree species. Journal of Ecology, Vol. 104, pp. 1076–1088. https://doi.org/10.1111/1365- 2745.12574

Körner, C., & Renhardt, U. (1987). Dry matter partitioning and root length/leaf area ratios in herbaceous perennial plants with diverse altitudinal distribution. Oecologia, 74(3), 411– 418. Retrieved from https://www.jstor.org

182 Kuznetsova, A., Brockhoff, P. B., & Christensen, R. H. B. (2017). lmerTest package: Tests in linear mixed effects models. Journal of Statistical Software, 82(13), 1–26. https://doi.org/10.18637/jss.v082.i13

Lacher, I., & Schwartz, M. W. (2016). Empirical test on the relative climatic sensitivity between individuals of narrowly and broadly distributed species. Ecosphere, 7(3), 1–12. https://doi.org/10.1002/ecs2.1227

Lanes, É. C., Pope, N. S., Alves, R., Carvalho Filho, N. M., Giannini, T. C., Giulietti, A. M., … Jaffé, R. (2018). Landscape genomic conservation assessment of a narrow-endemic and a widespread morning glory from Amazonian savannas. Frontiers in Plant Science, 9(Article 532), 1–13. https://doi.org/10.3389/fpls.2018.00532

Larkin, P., Quevedo, E., Salinas, S., Parker, J., Storey, K., & Hardegree, B. (2006). Genetic structure of two Thalassia testudinum populations from the south Texas Gulf coast. Aquatic Botany, 85(3), 198–202. https://doi.org/10.1016/j.aquabot.2006.03.012

Larsen, B., Gardner, K., Pedersen, C., Ørgaard, M., Migicovsky, Z., Myles, S., & Toldam- Andersen, T. B. (2018). Population structure, relatedness and ploidy levels in an apple gene bank revealed through genotyping-by-sequencing. PLoS ONE, 13(8). https://doi.org/10.1371/journal.pone.0201889

Lemon, J. B., & Wolf, P. G. (2018). Genetic differentiation between endemic Eriogonum soredium and its common relative E. shockleyi (Polygonaceae). Systematic Botany, 43(4), 901–909. https://doi.org/10.1600/036364418X697797

Lenth, R. (2018). emmeans: Estimated marginal means, aka least-squares means. https://doi.org/10.1080/00031305.1980.10483031

Levin, D. A. (1983). Polyploidy and novelty in flowering plants. The American Naturalist, 122(1), 1–25. https://doi.org/10.1086/284115

Levins, R. (1963). Theory of fitness in a hetergeneous environment. II. Developmental flexibility and niche selection. The American Naturalist, 97(893), 75–90. Retrieved from http://max2.ese.u-psud.fr/epc/conservation/PDFs/HIPE/Levins1963.pdf

Liancourt, P., Boldgiv, B., Song, D. S., Spence, L. A., Helliker, B. R., Petraitis, P. S., & Casper, B. B. (2015). Leaf-trait plasticity and species vulnerability to climate change in a Mongolian steppe. Global Change Biology, 21(9), 3489–3498. https://doi.org/10.1111/gcb.12934

Linck, E., & Battey, C. J. (2019). Minor allele frequency thresholds strongly affect population structure inference with genomic data sets. Molecular Ecology Resources, 19(3), 639–647. https://doi.org/10.1111/1755-0998.12995

Lindwall, B. H. (1999). Long-term genetic consequences of habitat fragmentation: A study of isozyme variation in the alpine plants; Carex bigelowii Torr., Diapensia lapponica L., and Minuartia groenlandica Retz. University of Massachusetts at Amherst, doctoral dissertation.

183 Lischer, H. E. L., & Excoffier, L. (2012). PGDSpider: An automated data conversion tool for connecting population genetics and genomics programs. Bioinformatics, 28(2), 298–299. https://doi.org/10.1093/bioinformatics/btr642

Liu, Y., Zhang, L., Xu, X., & Niu, H. (2015). Understanding the wide geographic range of a clonal perennial grass: Plasticity versus local adaptation. AoB PLANTS, 8, 1–7. https://doi.org/10.1093/aobpla/plv141

Lloret, F., Penuelas, J., & Estiarte, M. (2004). Experimental evidence of reduced diversity of seedlings due to climate modification in a Mediterranean-type community. Global Change Biology, 10(2), 248–258. https://doi.org/10.1111/j.1365-2486.2004.00725.x

Löve, A., & Löve, D. (1966). Cytotaxonomy of the alpine vascular plants of Mount Washington. University of Colorado Studies. Series in Biology, 24, 1–74.

Löve, A., & Löve, D. (1967). Polyploidy and altitude: Mt. Washington. Biologisches Zentralblatt, Suppl. Vol, 307–312.

Luke, S. G. (2017). Evaluating significance in linear mixed-effects models in R. Behavior Research Methods, 49(4), 1494–1502. https://doi.org/10.3758/s13428-016-0809-y

Luo, T., Pan, Y., Ouyang, H., Shi, P., Luo, J., Yu, Z., & Lu, Q. (2004). Leaf area index and net primary productivity along subtropical to alpine gradients in the Tibetan Plateau. Source: Global Ecology and Biogeography, 13(4), 345–358. Retrieved from https://www-jstor-org

Lushai, G., Loxdale, H. D., & Allen, J. A. (2003). The dynamic clonal genome and its adaptive potential. Biological Journal of the Linnean Society, 79(1), 193–208. https://doi.org/10.1046/j.1095-8312.2003.00189.x

Mable, B. K. (2013). Polyploids and hybrids in changing environments: Winners or losers in the struggle for adaptation. Heredity, 110(2), 95–96. https://doi.org/10.1038/hdy.2012.105

Madlung, A. (2013). Polyploidy and its effect on evolutionary success: Old questions revisited with new tools. Heredity, 110(2), 99–104. https://doi.org/10.1038/hdy.2012.79

Maes, D., Vanreusel, W., Talloen, W., & Van Dyck, H. (2004). Functional conservation units for the endangered Alcon Blue butterfly Maculinea alcon in Belgium (Lepidoptera: Lycaenidae). Biological Conservation, 120(2), 229–241. https://doi.org/10.1016/j.biocon.2004.02.018

Maine Natural Areas Program. (2015). Rare, threatened, and endangered plant taxa. Retrieved from Maine Department of Agriculture, Conservation and Forestry website: https://www.maine.gov/dacf/mnap/features/rare_plants/2015_tracking_list.pdf

Marris, E. (2007). The escalator effect. Nature Reports Climate Change, 1, 94–96. https://doi.org/10.1038/climate.2007.70

184 Martín-Hernanz, S., Martínez-Sánchez, S., Albaladejo, R. G., Lorite, J., Arroyo, J., & Aparicio, A. (2019). Genetic diversity and differentiation in narrow versus widespread taxa of Helianthemum (Cistaceae) in a hotspot: The role of geographic range, habitat, and reproductive traits. Ecology and Evolution, 9(6), 3016–3029. https://doi.org/10.1002/ece3.4481

Martin, J.-P., & Germain, D. (2016). Late-glacial and Holocene evolution as a driver of diversity and complexity of the northeastern North American alpine landscapes: A synthesis. Canadian Journal of Earth Sciences, 53(5), 494–505. https://doi.org/10.1139/cjes-2016- 0004

Martinez Munoz, K., Robinson, S. C., Vollmer, H., & Popp, B. (2019). Population genetic structure of the flowering plant Diapensia lapponica (Diapensiaceae) in the northeastern alpine zone. Lake Placid, NY.

Matesanz, S., Gianoli, E., & Valladares, F. (2010). Global change and the evolution of phenotypic plasticity in plants. Annals of the New York Academy of Sciences, 1206, 35–55. https://doi.org/10.1111/j.1749-6632.2010.05704.x

McCartney-Melstad, E., Vu, J. K., & Shaffer, H. B. (2018). Genomic data recover previously undetectable fragmentation effects in an endangered amphibian. Molecular Ecology, 27(22), 4430–4443. https://doi.org/10.1111/mec.14892

McCarty, J. P. (2001). Ecological consequences of recent climate change. Conservation Biology, 15(2), 320–331. https://doi.org/10.1046/j.1523-1739.2001.015002320.x

McCormack, J. E., Huang, H., & Knowles, L. L. (2009). Sky islands. In R. G. Gillespie & D. A. Clague (Eds.), Encyclopoedia of islands (pp. 839–843). Berkeley, CA: University of California Press.

McDonough Mackenzie, C., Primack, R. B., & Miller-Rushing, A. J. (2018). Local environment, not local adaptation, drives leaf-out phenology in common gardens along an elevational gradient in Acadia National Park, Maine. 105(6), 986–995. https://doi.org/10.1002/ajb2.1108

McKinney, G. J., Waples, R. K., Pascal, C. E., Seeb, L. W., & Seeb, J. E. (2018). Resolving allele dosage in duplicated loci using genotyping-by-sequencing data: A path forward for population genetic analysis. Molecular Ecology Resources, 18(3), 570–579. https://doi.org/10.1111/1755-0998.12763

McKinney, G. J., Waples, R. K., Seeb, L. W., & Seeb, J. E. (2017). Paralogs are revealed by proportion of heterozygotes and deviations in read ratios in genotyping-by-sequencing data from natural populations. Molecular Ecology Resources, 17(4), 656–669. https://doi.org/10.1111/1755-0998.12613

185 McLaughlin, S. B., Downing, D. J., Blasing, T. J., Cook, E. R., & Adams, H. S. (1987). An analysis of climate and competition as contributors to decline of red spruce in high elevation Appalachian forests of the eastern United states. Oecologia, 72(4), 487–501. https://doi.org/10.1007/BF00378973

McMahon, B. J., Teeling, E. C., & Höglund, J. (2014). How and why should we implement genomics into conservation? Evolutionary Applications, 7(9), 999–1007. https://doi.org/10.1111/eva.12193

Meirmans, P. G. (2014). Nonconvergence in Bayesian estimation of migration rates. Molecular Ecology Resources, 14(4), 726–733. https://doi.org/10.1111/1755-0998.12216

Meirmans, P. G., Liu, S., & Van Tienderen, P. H. (2018). The analysis of polyploid genetic data (F. W. Allendorf, Ed.). Journal of Heredity, Vol. 109, pp. 283–296. https://doi.org/10.1093/jhered/esy006

Merilä, J., & Hendry, A. P. (2014). Climate change, adaptation, and phenotypic plasticity: The problem and the evidence. Evolutionary Applications, 7(1), 1–14. https://doi.org/10.1111/eva.12137

Migliore, J., Baumel, A., Juin, M., Fady, B., Roig, A., Duong, N., & Médail, F. (2013). Surviving in mountain climate refugia: New insights from the genetic diversity and structure of the relict shrub Myrtus nivellei (Myrtaceae) in the Sahara Desert. PLoS ONE, 8(9), e73795. https://doi.org/10.1371/journal.pone.0073795

Miller, N. G., & McDaniel, S. F. (2004). Bryophyte dispersal inferred from colonization of an introduced substratum on Whiteface Mountain, New York. American Journal of Botany, 91(8), 1173–1182. https://doi.org/10.3732/ajb.91.8.1173

Mills, L. S., & Allendorf, F. W. (1996). One-migrant-per-generation rule in conservation management. Conservation Biology, 10(6), 1509–1518.

Milstead, W. L. (1964). A revision of the North American species of Prenanthes. Purdue University, Department of Botany, doctoral dissertation.

Mitchell, R. S., & Tucker, G. C. (1997). Revised checklist of New York State plants, Bulletin 490. Albany, New York: New York State Museum.

Montero, B. K., Refaly, E., Ramanamanjato, J. B., Randriatafika, F., Rakotondranary, S. J., Wilhelm, K., … Sommer, S. (2019). Challenges of next-generation sequencing in conservation management: Insights from long-term monitoring of corridor effects on the genetic diversity of mouse lemurs in a fragmented landscape. Evolutionary Applications, 12(3), 425–442. https://doi.org/10.1111/eva.12723

Moore, T. E., Bagchi, R., Aiello-Lammens, M. E., & Schlichting, C. D. (2018). Spatial autocorrelation inflates niche breadth–range size relationships. Global Ecology and Biogeography, 27, 1426–1436. https://doi.org/10.1111/geb.12818

186 Moritz, C. (1994). Defining “Evolutionarily Significant Units” for conservation. Trends in Ecology and Evolution, 9(10), 373–375. https://doi.org/10.1016/0169-5347(94)90057-4

Morris, W. S., Pfister, C. A., Tuliapurkar, S., Chirrakal, H. V., Boggs, C. L., Boyce, M. S., … Menges, E. S. (2008). Longevity can buffer plant and animal populations against changing climatic variability. Ecology, 89(1), 19–25. Retrieved from http://dx.doi.org/10.1890/07- 0774.1

Morton, J. K. (1993). Chromosome numbers and polyploidy in the flora of Cameroons Mountain. Opera Botanica, 121, 159–172.

Murakami, P. F., Turner, M. R., van den Berg, A. K., & Schaberg, P. G. (2005). An instructional guide for leaf color analysis using digital imaging software. In General Technical Report NE-327. Retrieved from USDA Forest Service Northeastern Research Station website: www.scioncorp.com

Nathan, R. (2006). Long-distance dispersal of plants. Science, 313, 786–788. https://doi.org/10.1126/science.1124975

Nathan, R., Schurr, F. M., Spiegel, O., Steinitz, O., Trakhtenbrot, A., & Tsoar, A. (2008). Mechanisms of long-distance seed dispersal. Trends in Ecology and Evolution, 23(11), 638– 647. https://doi.org/10.1016/j.tree.2008.08.003

NatureServe. (2018). NatureServe Explorer: An online encyclopedia of life. Version 7.1. Retrieved February 13, 2019, from NatureServe, Arlington, Virginia website: http://explorer.natureserve.org

Nei, M. (1972). Genetic distance between populations. The American Naturalist, 106(949), 283– 292. https://doi.org/10.1086/282771

Nei, M. (1973). Analysis of gene diversity in subdivided populations. Proceedings of the National Academy of Sciences, 70(12), 3321–3323. Retrieved from https://www.pnas.org/content/pnas/70/12/3321.full.pdf

Nei, M. (1978). Estimation of average heterozygosity and genetic distance from a small number of individuals. Genetics, 89, 583–590. https://doi.org/10.3390/ijms15010277

Nei, M., & Chesser, R. K. (1983). Estimation of fixation indices and gene diversities. Annals of Human Genetics, 47(3), 253–259. https://doi.org/10.1111/j.1469-1809.1983.tb00993.x

New Hampshire Natural Heritage Bureau. (2018). Rare plant list for New Hampshire: Common names. Retrieved from New Hampshire Department of Natural and Cultural Resources: Division of Forests & Lands website: https://www.nh.gov/nhdfl/documents/tracking-list- plant-general.pdf

New York Natural Heritage Program. (2012). Element distribution model, model validation, and environmental variable importance for Prenanthes boottii. Albany, NY.

187 Nicotra, A. B., Atkin, O. K., Bonser, S. P., Davidson, A. M., Finnegan, E. J., Mathesius, U., … van Kleunen, M. (2010). Plant phenotypic plasticity in a changing climate. Trends in Plant Science, 15(12), 684–692. https://doi.org/10.1016/j.tplants.2010.09.008

Nicotra, A. B., Leigh, A., Boyce, C. K., Jones, C. S., Niklas, K. J., Royer, D. L., & Tsukaya, H. (2011). The evolution and functional significance of leaf shape in the angiosperms. Functional Plant Biology, 38, 535–552. https://doi.org/10.1071/FP11057

Nooten, S. S., & Hughes, L. (2017). The power of the transplant: Direct assessment of climate change impacts. Climatic Change, 144, 237–255. https://doi.org/10.1007/s10584-017-2037- 6

Nussey, D. H., Postma, E., Gienapp, P., & Visser, M. E. (2005). Selection on heritable phenotypic plasticity in a wild bird population. Science, 310(5746), 304–306. https://doi.org/10.1126/science.1117004

Nybom, H. (2004). Comparison of different nuclear DNA markers for estimating intraspecific genetic diversity in plants. Molecular Ecology, 13(5), 1143–1155. https://doi.org/10.1111/j.1365-294X.2004.02141.x

O’Leary, S. J., Puritz, J. B., Willis, S. C., Hollenbeck, C. M., & Portnoy, D. S. (2018). These aren’t the loci you’re looking for: Principles of effective SNP filtering for molecular ecologists. Molecular Ecology, 27, 3193–3206. https://doi.org/10.1111/mec.14792

Ockendon, N., Baker, D. J., Carr, J. A., White, E. C., Almond, R. E. A., Amano, T., … Pearce- Higgins, J. W. (2014). Mechanisms underpinning climatic impacts on natural populations: Altered species interactions are more important than direct effects. Global Change Biology, 20(7), 2221–2229. https://doi.org/10.1111/gcb.12559

Oliva-Tejera, F., Caujape-Castells, J., Navarro-Deniz, J., Reyes-Betancort, A., Scholz, S., Baccarani-Rosas, M., & Cabrera-Garcia, N. (2006). Patterns of genetic divergence of three Canarian endemic Lotus (Fabaceae): Implications for the conservation of the endangered L. kunkelii. American Journal of Botany, 93(8), 1116–1124. https://doi.org/10.3732/ajb.93.8.1116

Pacifici, M., Foden, W. B., Visconti, P., Watson, J. E. M., Butchart, S. H. M., Kovacs, K. M., … Rondinini, C. (2015). Assessing species vulnerability to climate change. Nature Climate Change, 5(3), 215–225. https://doi.org/10.1038/nclimate2448

Packer, J. G. (1974). Differentiation and dispersal in alpine floras. Arctic and Alpine Research, 6(2), 117–128. https://doi.org/10.1080/00040851.1974.12003768

Panetta, A. M., Stanton, M. L., & Harte, J. (2018). Climate warming drives local extinction: Evidence from observation and experimentation. Science Advances, 4(2). https://doi.org/10.1126/sciadv.aaq1819

188 Parmesan, C., Williams-Anderson, A., Moskwik, M., Mikheyev, A. S., & Singer, M. C. (2015). Endangered Quino checkerspot butterfly and climate change: Short-term success but long- term vulnerability? Journal of Insect Conservation, 19(2), 185–204. https://doi.org/10.1007/s10841-014-9743-4

Pellissier, L., Bråthen, K. A., Vittoz, P., Yoccoz, N. G., Dubuis, A., Meier, E. S., … Guisan, A. (2013). Thermal niches are more conserved at cold than warm limits in arctic-alpine plant species. Global Ecology and Biogeography, 22(8), 933–941. https://doi.org/10.1111/geb.12057

Peñas, J., Barrios, S., Bobo-Pinilla, J., Lorite, J., & Martínez-Ortega, M. M. (2016). Designing conservation strategies to preserve the genetic diversity of Astragalus edulis Bunge, an endangered species from western Mediterranean region. PeerJ, 4, e1474. https://doi.org/10.7717/peerj.1474

Peppe, D. J., Royer, D. L., Cariglino, B., Oliver, S. Y., Newman, S., Leight, E., … Wright, I. J. (2011). Sensitivity of leaf size and shape to climate: Global patterns and paleoclimatic applications. New Phytologist, 190(3), 724–739. https://doi.org/10.1111/j.1469- 8137.2010.03615.x

Pérez-Collazos, E., Segarra-Moragues, J. G., & Catalán, P. (2008). Two approaches for the selection of Relevant Genetic Units for Conservation in the narrow European endemic steppe plant Boleum asperum (Brassicaceae). Biological Journal of the Linnean Society, 94(2), 341–354. https://doi.org/10.1111/j.1095-8312.2008.00961.x

Pérez-Harguindeguy, N., Díaz, S., Garnier, E., Lavorel, S., Poorter, H., Jaureguiberry, P., … Cornellssen, J. H. C. (2013). New handbook for standardized measurement of plant functional traits worldwide. Australian Journal of Botany, 61, 167–234. https://doi.org/http://dx.doi.org/10.1071/BT12225

Peterson, B. K., Weber, J. N., Kay, E. H., Fisher, H. S., & Hoekstra, H. E. (2012). Double digest RADseq: An inexpensive method for de novo SNP discovery and genotyping in model and non-model species. PloS One, 7(5), e37135. https://doi.org/10.1371/journal.pone.0037135

Phillimore, A. B., Leech, D. I., Pearce-Higgins, J. W., & Hadfield, J. D. (2016). Passerines may be sufficiently plastic to track temperature-mediated shifts in optimum lay date. Global Change Biology, 22(10), 3259–3272. https://doi.org/10.1111/gcb.13302

Pillon, Y., & Chase, M. W. (2007). Taxonomic exaggeration and its effects on orchid conservation. Conservation Biology, 21(1), 263–265. https://doi.org/10.1111/j.1523- 1739.2006.00573.x

Pimm, S. L., Jenkins, C. N., Abell, R., Brooks, T. M., Gittleman, J. L., Joppa, L. N., … Sexton, J. O. (2014). The biodiversity of species and their rates of extinction, distribution, and protection. Science, 344(6187), 987,1246752-1–10. https://doi.org/10.1126/science.1246752

189 Piry, S., Luikart, G., & Cornuet, J.-M. (1999). BOTTLENECK: A computer program for detecting recent reductions in the effective population size using allele frequency data. Heredity, 90(4), 502–503. Retrieved from https://pdfs.semanticscholar.org

Pluess, A. R., & Stöcklin, J. (2004). Population genetic diversity of the clonal plant Geum reptans (Rosaceae) in the Swiss Alps. American Journal of Botany, 91(12), 2013–2021. https://doi.org/10.3732/ajb.91.12.2013

Poorter, H., Niinemets, Ü., Poorter, L., Wright, I. J., & Villar, R. (2009). Causes and consequences of variation in leaf mass per area (LMA): A meta-analysis. New Phytologist, 182(3), 565–588. https://doi.org/10.1111/j.1469-8137.2009.02830.x

Poorter, H., Niklas, K. J., Reich, P. B., Oleksyn, J., Poot, P., & Mommer, L. (2012). Biomass allocation to leaves, stems and roots: Meta-analysis of interspecific variation and environmental control. New Phytologist, 193(1), 30–50. https://doi.org/10.1111/j.1469- 8137.2011.03952.x

Powell, A. M., Kyhos, D. W., & Raven, P. H. (1974). Chromosome numbers in Compositae. X. American Journal of Botany, 61(8), 909–913. Retrieved from https://www.jstor.org

Prevosti, A. (1974). La distancia genetica entre poblaciones. Miscellanea Alcobe, 68, 109–118.

Price, T. D., Qvarnström, A., & Irwin, D. E. (2003, July 22). The role of phenotypic plasticity in driving genetic evolution. Proceedings of the Royal Society B: Biological Sciences, Vol. 270, pp. 1433–1440. https://doi.org/10.1098/rspb.2003.2372

Pritchard, J. K., Stephens, M., & Donnelly, P. (2000). Inference of population structure using multilocus genotype data. Genetics, 155, 945–959. Retrieved from http://www.stats.ox.ac.uk/pritch/home.html.

Pritchard, J. K., Wen, X., & Falush, D. (2010). Documentation for structure software: Version 2.3 (p. 39). p. 39.

Prout, L. (2005). Biological evaluation of the White Mountain National Forest land and resource management plan revision on federal endangered, threatened, and proposed species and regional forester sensitive species. Retrieved from USDA Forest Service website: https://www.fs.usda.gov/Internet/FSE_DOCUMENTS/stelprdb5199898.pdf

Prunier, R., Akman, M., Kremer, C. T., Aitken, N., Chuah, A., Borevitz, J., & Holsinger, K. E. (2017). Isolation by distance and isolation by environment contribute to population differentiation in Protea repens (Proteaceae L.), a widespread South African species. American Journal of Botany, 104(5), 674–684. https://doi.org/10.3732/ajb.1600232

Puechmaille, S. J. (2016). The program STRUCTURE does not reliably recover the correct population structure when sampling is uneven: Subsampling and new estimators alleviate the problem. Molecular Ecology Resources, 16, 608–627. https://doi.org/10.1111/1755- 0998.12512

190 Puritz, J. B. (2019a). dDocent SNP filtering tutorial. Retrieved July 9, 2019, from https://www.ddocent.com/filtering/

Puritz, J. B. (2019b). dDocent user guide. Retrieved July 9, 2019, from https://www.ddocent.com/UserGuide/

Puritz, J. B., Hollenbeck, C. M., & Gold, J. R. (2014). dDocent: A RADseq, variant-calling pipeline designed for population genomics of non-model organisms. PeerJ, 2(e431). https://doi.org/10.7717/peerj.431

Qiu, S., Bergero, R., Guirao-Rico, S., Campos, J. L., Cezard, T., Gharbi, K., & Charlesworth, D. (2016). RAD mapping reveals an evolving, polymorphic and fuzzy boundary of a plant pseudoautosomal region. Molecular Ecology, 25(1), 414–430. https://doi.org/10.1111/mec.13297

R core team. (2017). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://doi.org/http://www.R-project.org/

Rabinowitz, D. (1981). Seven forms of rarity. In H. Synge (Ed.), The biological aspects of rare plant conservation (pp. 205–217). New York: Wiley.

Rambaut, A., Drummond, A. J., Xie, D., Baele, G., & Suchard, M. A. (2018). Posterior summarization in Bayesian phylogenetics using Tracer 1.7. Systematic Biology, 67(5), 901– 904. https://doi.org/10.1093/sysbio/syy032

Rannala, B. (2007). BayesAss edition 3.0 user’s manual (p. 14). p. 14. University of California Davis.

Rannala, B. (2013). BayesAss Support: BayesAss Q&A. Post on 2013-06-27. Retrieved from SourceForge.net website: https://sourceforge.net/p/bayesass/discussion/bayesass/thread/c88c457a/?limit=25#de33

Rasband, W. S. (2004). RGB Measure. Retrieved from https://imagej.nih.gov/ij/plugins/rgbmeasure.html

Réale, D., McAdam, A. G., Boutin, S., & Berteaux, D. (2003). Genetic and plastic responses of a northern mammal to climate change. Proceedings of the Royal Society B: Biological Sciences, 270(1515), 591–596. https://doi.org/10.1098/rspb.2002.2224

Reay, D. S., Dentener, F., Smith, P., Grace, J., & Feely, R. A. (2008). Global nitrogen deposition and carbon sinks. Nature Geoscience, 1, 430–437. https://doi.org/10.1023/A:1005270101276

Region 5 Office of Natural Resources. (1996). High Peaks unit management plan final draft. New York State Department of Environmental Conservation.

191 Ren, G., Mateo, R. G., Liu, J., Suchan, T., Alvarez, N., Guisan, A., … Salamin, N. (2017). Genetic consequences of Quaternary climatic oscillations in the Himalayas: Primula tibetica as a case study based on restriction site-associated DNA sequencing. New Phytologist, 213(3), 1500–1512. https://doi.org/10.1111/nph.14221

Reynolds, J., Weir, B. S., & Cockerham, C. C. (1983). Estimation of the coancestry coefficient: Basis for a short-term genetic distance. Genetics, 105, 767–779. Retrieved from https://www.genetics.org/content/genetics/105/3/767.full.pdf

Reynolds, R. G., Gerber, G. P., & Fitzpatrick, B. M. (2011). Unexpected shallow genetic divergence in Turks Island boas (Epicrates c. chrysogaster) reveals single evolutionarily significant unit for conservation. Herpetologica, 67(4), 477–486. https://doi.org/10.1655/herpetologica-d-11-00014.1

Riddle, B. R., & Hafner, D. J. (1999). Species as units of analysis in ecology and biogeography: Time to take the blinders off. Global Ecology and Biogeography, 8(6), 433–441. https://doi.org/10.1046/j.1365-2699.1999.00170.x

Riebesell, J. F. (1981). Photosynthetic adaptations in bog and alpine populations of Ledum groenlandicum. Ecology, 62(3), 579–586. https://doi.org/10.2307/1937724

Riebesell, J. F. (1982). Arctic-alpine plants on mountaintops: Agreement with island biogeography theory. American Naturalist, 119(5), 657–674. https://doi.org/10.1086/283941

Rigby, R. A., & Stasinopoulos, D. M. (2005). Generalized additive models for location, scale and shape (with discussion). Journal of the Royal Statistical Society: Series C (Applied Statistics), 54(3), 507–554. https://doi.org/10.1111/j.1467-9876.2005.00510.x

Robinson, S. C. (2012). Experimental and molecular studies of bryophyte dispersal on alpine summits. SUNY University at Albany, Department of Biological Sciences, doctoral dissertation.

Robinson, S. C., Ketchledge, E. H., Fitzgerald, B. T., Raynal, D. J., & Kimmerer, R. W. (2010). A 23-year assessment of vegetation composition and change in the Adirondack alpine zone, New York State. Rhodora, 112(952), 355–377. https://doi.org/10.3119/09-03.1

Rousset, F. (2008). GENEPOP’007: A complete re-implementation of the GENEPOP software for Windows and Linux. Molecular Ecology Resources, 8(1), 103–106. https://doi.org/10.1111/j.1471-8286.2007.01931.x

Royer, D. L., Meyerson, L. A., Robertson, K. M., & Adams, J. M. (2009). Phenotypic plasticity of leaf shape along a temperature gradient in Acer rubrum. PLoS ONE, 4(10), 7653. https://doi.org/10.1371/journal.pone.0007653

192 Ruiz, J. M., Marín-Guirao, L., García-Muñoz, R., Ramos-Segura, A., Bernardeau-Esteller, J., Pérez, M., … Procaccini, G. (2018). Experimental evidence of warming-induced flowering in the Mediterranean seagrass Posidonia oceanica. Marine Pollution Bulletin, 134, 49–54. https://doi.org/10.1016/j.marpolbul.2017.10.037

Ryder, O. A. (1986). Species conservation and systematics: The dilemma of subspecies. Trends in Ecology & Evolution, 1(1), 9–10. https://doi.org/10.1016/0169-5347(86)90059-5

Sala, O. E., Chapin, F. S., Armesto, J. J., Berlow, E., Bloomfield, J., Dirzo, R., … Wall, D. H. (2000). Global biodiversity scenarios for the year 2100. Science, 287(5459), 1770–1774. https://doi.org/10.1126/science.287.5459.1770

Sayers, K. F. (1989). A study of six taxa of Prenanthes L. (Asteraceae; Lactuceae) in northeastern North America. University of Guelph, master’s thesis.

Scheepens, J. F., Frei, E. S., & Stöcklin, J. (2010). Genotypic and environmental variation in specific leaf area in a widespread Alpine plant after transplantation to different altitudes. Oecologia, 164(1), 141–150. https://doi.org/10.1007/s00442-010-1650-0

Scherrer, D., & Körner, C. (2011). Topographically controlled thermal-habitat differentiation buffers alpine plant diversity against climate warming. Journal of Biogeography, 38(2), 406–416. https://doi.org/10.1111/j.1365-2699.2010.02407.x

Schmid, S. F., Stöcklin, J., Hamann, E., & Kesselring, H. (2017). High-elevation plants have reduced plasticity in flowering time in response to warming compared to low-elevation congeners. Basic and Applied Ecology, 21, 1–12. https://doi.org/10.1016/j.baae.2017.05.003

Schmitt, T. (2007). Molecular biogeography of Europe: Pleistocene cycles and postglacial trends. Frontiers in Zoology, 4(1), 11. https://doi.org/10.1186/1742-9994-4-11

Schmitt, T., & Schönswetter, P. (2010). Are disjunct alpine and arctic-alpine animal and plant species in the western palearctic really “relics of a cold past”? In J. C. Habel & T. Assmann (Eds.), Relict species: Phylogeography and conservation biology (pp. 239–252). https://doi.org/10.1007/978-3-540-92160-8_13

Schneider, C. A., Rasband, W. S., & Eliceiri, K. W. (2012). NIH Image to ImageJ: 25 years of image analysis. Nature Methods, 9(7), 671–675. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/22930834

Segarra-Moragues, J. G., & Catalán, P. (2010). The fewer and the better: Prioritization of populations for conservation under limited resources, a genetic study with Borderea pyrenaica (Dioscoreaceae) in the Pyrenean National Park. Genetica, 138(3), 363–376. https://doi.org/10.1007/s10709-009-9427-2

193 Seidel, T. M., Weihrauch, D. M., Kimball, K. D., Pszenny, A. A. P., Soboleski, R., Crete, E., & Murray, G. L. D. (2009). Evidence of climate change declines with elevation based on temperature and snow records from 1930s to 2006 on Mount Washington, New Hampshire, U.S.A. Arctic, Antarctic, and Alpine Research, 41(3), 362–372. https://doi.org/10.1657/1938-4246-41.3.362

Sessa, E. B. (2019). Polyploidy as a mechanism for surviving global change. New Phytologist, 221(1), 5–6. https://doi.org/10.1111/nph.15513

Sexton, J. P., Hangartner, S. B., & Hoffmann, A. A. (2014). Genetic isolation by environment or distance: Which pattern of gene flow is most common? Evolution, 68(1), 1–15. https://doi.org/10.1111/evo.12258

Sexton, J. P., McIntyre, P. J., Angert, A. L., & Rice, K. J. (2009). Evolution and ecology of species range limits. Annual Review of Ecology, Evolution, and Systematics, 40(1), 415– 436. https://doi.org/10.1146/annurev.ecolsys.110308.120317

Sexton, J. P., Montiel, J., Shay, J. E., Stephens, M. R., & Slatyer, R. A. (2017). Evolution of ecological niche breadth. Annual Review of Ecology, Evolution, and Systematics, 48, 183– 206. https://doi.org/10.1146/annurev-ecolsys-110316

Sheth, S. N., & Angert, A. L. (2014). The evolution of environmental tolerance and range size: A comparison of geographically restricted and widespread Mimulus. Evolution, 68(10), 2917– 2931. https://doi.org/10.1111/evo.12494

Siadjeu, C., Mayland-Quellhorst, E., & Albach, D. C. (2018). Genetic diversity and population structure of trifoliate yam (Dioscorea dumetorum Kunth) in Cameroon revealed by genotyping-by-sequencing (GBS). BMC Plant Biology, 18(1), 359. https://doi.org/10.1186/s12870-018-1593-x

Slatyer, R. A., Hirst, M. J., & Sexton, J. P. (2013). Niche breadth predicts geographical range size: A general ecological pattern. Ecology Letters, 16(8), 1104–1114. https://doi.org/10.1111/ele.12140

Soltis, D. E., Visger, C. J., & Soltis, P. S. (2014). The polyploidy revolution then...and now: Stebbins revisited. American Journal of Botany, 101(7), 1057–1078. https://doi.org/10.3732/ajb.1400178

Soulé, M. E. (1983). What do we really know about extinction? In C. M. Schonewald-Cox, S. M. Chambers, B. MacBryde, & L. Thomas (Eds.), Genetics and conservation: A reference for managing wild animal and plant populations (pp. 111–124). Meulo Park, CA: Benjamin/Cummings.

Spieth, P. T. (1974). Gene flow and genetic differentiation. Genetics, 78(3), 961–965. Retrieved from https://www.genetics.org/content/genetics/78/3/961.full.pdf

194 St. Hilaire, L. (2003). Nabalus racemosus (Michx.) Hook. glaucous white lettuce: Conservation and research plan for New England. Framingham, MA: New England Plant Conservation Program and New England Wild Flower Society.

Stebbins, G. L. (1940). The significance of polyploidy in plant evolution. The American Naturalist, 74(750), 54–66. https://doi.org/10.1086/280872

Steyn, W. J., Wand, S. J. E., Holcroft, D. M., & Jacobs, G. (2002). Anthocyanins in vegetative tissues: A proposed unified function in photoprotection. New Phytologist, 155(3), 349–361. https://doi.org/10.1046/j.1469-8137.2002.00482.x

Stoeckel, S., Grange, J., Fernández-Manjarres, J. F., Bilger, I., Frascaria-Lacoste, N., & Mariette, S. (2006). Heterozygote excess in a self-incompatible and partially clonal forest tree species - Prunus avium L. Molecular Ecology, 15(8), 2109–2118. https://doi.org/10.1111/j.1365- 294X.2006.02926.x

Sultan, S. E., & Spencer, H. G. (2002). Metapopulation structure favors plasticity over local adaptation. The American Naturalist, 160(2), 271–283. https://doi.org/10.1086/341015

Tackenberg, O., Poschlod, P., & Kahmen, S. (2003). Dandelion seed dispersal: The horizontal wind speed does not matter for long-distance dispersal—it is updraft! Plant Biology, 5(5), 451–454. https://doi.org/10.1055/s-2003-44789

Tang, Y., Horikoshi, M., & Li, W. (2016). ggfortify: Unified interface to visualize statistical results of popular R packages. The R Journal, 8(2), 478–489. Retrieved from http://adv- r.had.co.nz/S3.html

Te Beest, M., Le Roux, J. J., Richardson, D. M., Brysting, A. K., Suda, J., Kubešová, M., & Pyšek, P. (2012). The more the better? The role of polyploidy in facilitating plant invasions. Annals of Botany, 109(1), 19–45. https://doi.org/10.1093/aob/mcr277

Tetreault, T., & Burgess, M. B. (2019). Alpine pollinator guilds of the Adirondack High Peaks. Lake Placid, NY: Poster presentation at the 11th Northeastern Alpine Stewardship Gathering.

Therneau, T. M. (2018). coxme: Mixed effects Cox models. Retrieved from https://cran.r- project.org/package=coxme

Thompson, J. R., Carpenter, D. N., Cogbill, C. V., & Foster, D. R. (2013). Four centuries of change in northeastern United States forests. PLoS ONE, 8(9). https://doi.org/10.1371/journal.pone.0072540

Thomson, S. A., Pyle, R. L., Ahyong, S. T., Alonso-Zarazaga, M., Ammirati, J., Araya, J. F., … Zhou, H. Z. (2018). Taxonomy based on science is necessary for global conservation. PLoS Biology, 16(3), e2005075. https://doi.org/10.1371/journal.pbio.2005075

195 Thrall, P. H., Richards, C. M., Mccauley, D. E., & Antonovics, J. (1998). Metapopulation collapse: The consequences of limited gene-flow in spatially structured populations. In J. Bascompte & R. V Solé (Eds.), Modeling spatiotemporal dynamics in ecology (pp. 83– 104). Berlin, New York: Springer.

Thrasher, D. J., Butcher, B. G., Campagna, L., Webster, M. S., & Lovette, I. J. (2018). Double- digest RAD sequencing outperforms microsatellite loci at assigning paternity and estimating relatedness: A proof of concept in a highly promiscuous bird. Molecular Ecology Resources, 18(5), 953–965. https://doi.org/10.1111/1755-0998.12771

Tilman, D., Clark, M., Williams, D. R., Kimmel, K., Polasky, S., & Packer, C. (2017). Future threats to biodiversity and pathways to their prevention. Nature, 546, 73–81.

Tomb, S. A., Chambers, K. L., Kyhos, D. W., Powell, A. M., & Raven, P. H. (1978). Chromosome numbers in the Compositae. XIV. Lactuceae. American Journal of Botany, 65(7), 717–721. Retrieved from https://www.jstor.org

Torres-Martínez, L., & Emery, N. C. (2016). Genome-wide SNP discovery in the annual herb, Lasthenia fremontii (Asteraceae): Genetic resources for the conservation and restoration of a California vernal pool endemic. Conservation Genetics Resources, 8(2), 145–158. https://doi.org/10.1007/s12686-016-0524-0

Tuxill, J., & Nabhan, G. P. (2001). People, plants and protected areas. A guide to in situ management. London: Earthscan Publications.

Urban, M. C. (2015). Accelerating extinction risk from climate change. Science, 348(6234), 571–573. Retrieved from http://ci.nii.ac.jp/naid/40015386704/

Urban, M. C. (2018). Escalator to extinction. Proceedings of the National Academy of Sciences, 115(47), 11871–11873. https://doi.org/10.1073/pnas.1817416115

USDA, & NRCS. (2019). The PLANTS Database. Retrieved from http://plants.usda.gov

USFWS, & NOAA. (1996). Interagency policy regarding the recognition of distinct vertebrate population segments under the ESA. US Department of the Interior and Department of Commerce.

Valladares, F., Matesanz, S., Guilhaumon, F., Araújo, M. B., Balaguer, L., Benito-Garzón, M., … Zavala, M. A. (2014). The effects of phenotypic plasticity and local adaptation on forecasts of species range shifts under climate change. Ecology Letters, 17(11), 1351–1364. https://doi.org/10.1111/ele.12348

Valladares, F., Sanchez-Gomez, D., & Zavala, M. A. (2006). Quantitative estimation of phenotypic plasticity: Bridging the gap between the evolutionary concept and its ecological applications. Journal of Ecology, Vol. 94, pp. 1103–1116. https://doi.org/10.1111/j.1365- 2745.2006.01176.x

196 Van De Peer, Y., Mizrachi, E., & Marchal, K. (2017). The evolutionary significance of polyploidy. Nature Reviews Genetics, 18(7), 411–424. https://doi.org/10.1038/nrg.2017.26

Vanneste, K., Baele, G., Maere, S., & Van De Peer, Y. (2014). Analysis of 41 plant genomes supports a wave of successful genome duplications in association with the Cretaceous- Paleogene boundary. Genome Research, 24(8), 1334–1347. https://doi.org/10.1101/gr.168997.113

Varga, Z. S., & Schmitt, T. (2008). Types of oreal and oreotundral disjunctions in the western Palearctic. Biological Journal of the Linnean Society, 93, 415–430. https://doi.org/10.1111/j.1095-8312.2007.00934.x

Vedder, O., Bouwhuis, S., & Sheldon, B. C. (2013). Quantitative assessment of the importance of phenotypic plasticity in adaptation to climate change in wild bird populations. PLoS Biology, 11(7), 1001605. https://doi.org/10.1371/journal.pbio.1001605

Vermont Natural Heritage Inventory. (2015). Endangered and threatened plants of Vermont. Retrieved from Vermont Fish & Wildlife Department website: https://vtfishandwildlife.com

Via, S., & Lande, R. (1985). Genotype-environment interaction and the evolution of phenotypic plasticity. Evolution, 39(3), 505–522.

Vinebrooke, R. D., Cottingham, K. L., Norberg, J., Scheffer, M., Dodson, S. I., Maberly, S. C., & Sommer, U. (2004). Impacts of multiple stressors on biodiversity and ecosystem functioning: The role of species co-tolerance. Oikos, 104(3), 451–457. https://doi.org/10.1111/j.0030-1299.2004.13255.x

Vitasse, Y., Hoch, G., Randin, C. F., Lenz, A., Kollas, C., Scheepens, J. F., & Körner, C. (2013). Elevational adaptation and plasticity in seedling phenology of temperate deciduous tree species. Oecologia, 171(3), 663–678. https://doi.org/10.1007/s00442-012-2580-9

Vitasse, Y., Lenz, A., Kollas, C., Randin, C. F., Hoch, G., & Körner, C. (2014). Genetic vs. non- genetic responses of leaf morphology and growth to elevation in temperate tree species. Functional Ecology, 28(1), 243–252. https://doi.org/10.1111/1365-2435.12161

Vitousek, P. M., Aber, J. D., Howarth, R. W., Likens, G. E., Matson, P. A., Schindler, D. W., … Tilman, D. (1997). Human alteration of the global nitrogen cycle: Sources and consequences. Ecological Applications, 7(3), 737–750.

Vitt, P., Havens, K., Kramer, A. T., Sollenberger, D., & Yates, E. (2010). Assisted migration of plants: Changes in latitudes, changes in attitudes. Biological Conservation, 143(1), 18–27. https://doi.org/10.1016/j.biocon.2009.08.015

Vogler, A. P., & Desalle, R. (1994). Diagnosing units of conservation management. Conservation Biology, 8(2), 354–363. https://doi.org/10.1046/j.1523-1739.1994.08020354.x

197 Vollmann, J., Walter, H., Sato, T., & Schweiger, P. (2011). Digital image analysis and chlorophyll metering for phenotyping the effects of nodulation in soybean. Computers and Electronics in Agriculture, 75(1), 190–195. https://doi.org/10.1016/j.compag.2010.11.003

Walck, J. L., Hidayati, S. N., Dixon, K. W., Thompson, K., & Poschlod, P. (2011). Climate change and plant regeneration from seed. Global Change Biology, 17(6), 2145–2161. https://doi.org/10.1111/j.1365-2486.2010.02368.x

Waples, R. S. (1991). Pacific salmon, Oncorhynchus spp., and the definition of “species” under the Endangered Species Act. Marine Fisheries Review, 53(3), 11–22. Retrieved from https://core.ac.uk/download/pdf/11024429.pdf

Waples, R. S., & Lindley, S. T. (2018). Genomics and conservation units: The genetic basis of adult migration timing in Pacific salmonids. Evolutionary Applications, 11(9), 1518–1526. https://doi.org/10.1111/eva.12687

Wasof, S., Lenoir, J., Aarrestad, P. A., Alsos, I. G., Armbruster, W. S., Austrheim, G., … Decocq, G. (2015). Disjunct populations of European vascular plant species keep the same climatic niches. Global Ecology and Biogeography, 24(12), 1401–1412. https://doi.org/10.1111/geb.12375

Wason, J. W. (2016). Environmental controls on forest tree species growth and distributions along elevation gradients in the northeastern United States. SUNY College of Environmental Science and Forestry, Department of Environmental and Forest Biology, doctoral dissertation.

Wason, J. W., Beier, C. M., Battles, J. J., & Dovciak, M. (2019). Acidic deposition and climate warming as drivers of tree growth in high-elevation spruce-fir forests of the northeastern US. Frontiers in Forests and Global Change, 2, 1–9. https://doi.org/10.3389/ffgc.2019.00063

Wason, J. W., & Dovciak, M. (2017). Tree demography suggests multiple directions and drivers for species range shifts in mountains of northeastern United States. Global Change Biology, 23(8), 3335–3347. https://doi.org/10.1111/gcb.13584

Wason, J. W., Dovciak, M., Beier, C. M., & Battles, J. J. (2017). Tree growth is more sensitive than species distributions to recent changes in climate and acidic deposition in the northeastern United States. Journal of Applied Ecology, 54(6), 1648–1657. https://doi.org/10.1111/1365-2664.12899

Wayne, R. K., Lehman, N., Allard, M. W., & Honeycutt, R. L. (1992). Mitochondrial DNA variability of the gray wolf: Genetic consequences of population decline and habitat fragmentation. Conservation Biology, 6(4), 559–569. https://doi.org/10.1046/j.1523- 1739.1992.06040559.x

Weber-Townsend, J. R. (2017). Contributions of genetic data to the conservation and management of the threatened American hart’s-tongue fern (Asplenium scolopendrium var. americanum). SUNY College of Environmental Science and Forestry, master’s thesis.

198 Weir, B. S., & Cockerham, C. C. (1984). Estimating F-statistics for the analysis of population structure. Evolution, 38(6), 1358–1370. Retrieved from https://onlinelibrary.wiley.com/doi/pdf/10.1111/j.1558-5646.1984.tb05657.x

Weiss-Schneeweiss, H., Emadzade, K., Jang, T.-S., & Schneeweiss, G. M. (2013). Evolutionary consequences, constraints and potential of polyploidy in plants. Cytogenetic and Genome Research, 140, 137–150. https://doi.org/10.1159/000351727

Welles, S. R., & Ellstrand, N. C. (2016). Genetic structure reveals a history of multiple independent origins followed by admixture in the allopolyploid weed Salsola ryanii. Evolutionary Applications, 9(7), 871–878. https://doi.org/10.1111/eva.12399

Westoby, M. (1998). A leaf-height-seed (LHS) plant ecology strategy scheme. Plant and Soil, 199(2), 213–227. https://doi.org/10.1023/A:1004327224729

Whitlock, M. C., & Mccauley, D. E. (1999). Indirect measures of gene flow and migration: Fst is not equal to 1/(4Nm+1). Heredity, 82, 117–125. Retrieved from https://www.icts.res.in/sites/default/files/Heredity Whitlock 1999.pdf

Wiens, J. J. (2016). Climate-related local extinctions are already widespread among plant and animal species. PLoS Biology, 14(12). https://doi.org/10.1371/journal.pbio.2001104

Wilson, G. A., & Rannala, B. (2003). Bayesian inference of recent migration rates using multilocus genotypes. Genetics, 163(3), 1177–1191. Retrieved from https://www.genetics.org/content/genetics/163/3/1177.full.pdf

Woolbright, S. A., Whitham, T. G., Gehring, C. A., Allan, G. J., & Bailey, J. K. (2014). Climate relicts and their associated communities as natural ecology and evolution laboratories. Trends in Ecology and Evolution, 29(7), 406–416. https://doi.org/10.1016/j.tree.2014.05.003

Wright, I. J., Reich, P. B., Westoby, M., Ackerly, D. D., Baruch, Z., Bongers, F., … Villar, R. (2004). The worldwide leaf economics spectrum. Nature, 428(6985), 821–827. https://doi.org/10.1038/nature02403

Wright, S. (1931). Evolution in Mendelian populations. Genetics, 16, 97–159.

Yeh, F. C., Yang, R. C., Boyle, T. B. J., Ye, Z. H., & Mao, J. X. (1997). POPGENE, the user- friendly shareware for population genetic analysis. Molecular Biology and Biotechnology Centre, University of Alberta, Edmonton, Canada 10.

Young, G., Andrew, I., Lee, K., Li, X., Robb, R., Robinson, I., … Sinclair, B. (2018). Analysing phenotypic variation in Eucalyptus pauciflora across an elevation gradient in the Australian Alps. Field Studies in Ecology, Vol. 1. Retrieved from http://studentjournals.anu.edu.au/index.php/fse/article/view/190

Young, S. M. (2019). New York rare plant status lists March 2019. Retrieved from New York Natural Heritage Program website: www.nynhp.org

199 Zhang, H. X., Li, H. Y., & Li, Y. X. (2018). Identifying evolutionarily significant units for conservation of the endangered Malus sieversii using genome-wide RADseq data. Nordic Journal of Botany, 36(7), 1–10. https://doi.org/10.1111/njb.01733

Zlonis, K. J., & Gross, B. L. (2018). Genetic structure, diversity, and hybridization in populations of the rare arctic relict Euphrasia hudsoniana (Orobanchaceae) and its invasive congener Euphrasia stricta. Conservation Genetics, 19, 43–55. https://doi.org/10.1007/s10592-017-0995-x

200 APPENDIX 1: SUPPLEMENTARY MATERIAL FOR CHAPTER 2

Table A1-1. Model selection criteria used for selecting the best-supported zero-inflated GAMLSS model for establishment percent in the seed transplant experiment. The first column identifies each model according to its fixed effects: site (S), taxon (T), and their additive (+) and interactive (*) effects. N is the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight S+T 30.8 0.0 7 0.524 N 31.8 1.0 3 0.317 T 34.0 3.3 5 0.102 S 35.2 4.5 5 0.057 S*T 43.7 13.0 11 <0.001

201 Table A1-2. Model selection criteria used for selecting the best-supported Cox proportional hazards mixed-effects survival model for seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AIC gives the Akaike’s Information Criterion score, column dAIC gives the change in AIC score (ΔAIC) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AIC dAIC df weight S*Sh 1163.7 0.0 3.0 0.6853 S*Sh+T 1165.8 2.1 5.0 0.2421 Sh 1169.5 5.8 2.5 0.0378 S+Sh 1171.0 7.3 4.4 0.0181 S+Sh+T 1172.7 9.0 6.2 0.0077 S 1174.7 11.0 2.7 0.0028 N 1175.0 11.3 1.1 0.0024 S+T 1176.3 12.5 3.7 0.0013 Global 1176.3 12.5 11.0 0.0013 T 1177.0 13.2 2.0 <0.001 S*T 1179.9 16.1 5.7 <0.001

202 Table A1-3. Model selection criteria used for selecting the best-supported mixed-effects linear model of height in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight S+T 113.2 0.0 6 0.3598 S+Sh+T 113.4 0.1 7 0.3339 S*Sh+T 114.1 0.9 8 0.2329 S*T 118.3 5.1 8 0.0285 S 119.0 5.8 4 0.0196 S+Sh 119.3 6.1 5 0.0173 S*Sh 120.9 7.7 6 0.0078 Global 129.0 15.8 14 <0.001 Sh 137.4 24.2 4 <0.001 T 147.6 34.4 5 <0.001 N 148.9 35.7 3 <0.001

203 Table A1-4. Model selection criteria used for selecting the best-supported mixed-effects linear model of dry mass in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight S*Sh 139.1 0.0 6 0.3083 S 139.9 0.8 4 0.2118 S+Sh 140.3 1.2 5 0.1705 S*Sh+T 141.2 2.0 8 0.1110 S+T 141.4 2.2 6 0.1006 S+Sh+T 141.9 2.7 7 0.0787 S*T 145.1 6.0 8 0.0154 Global 148.1 9.0 14 0.0034 Sh 154.1 15.0 4 <0.001 N 164.3 25.1 3 <0.001 T 167.5 28.3 5 <0.001

204 Table A1-5. Model selection criteria used for selecting the best-supported mixed-effects linear model of total leaf area in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight S 185.2 0.0 4 0.512 S+Sh 186.5 1.3 5 0.271 S*Sh 188.7 3.4 6 0.091 S+T 189.1 3.8 6 0.075 S+Sh+T 190.6 5.4 7 0.035 S*Sh+T 192.9 7.6 8 0.011 S*T 194.5 9.3 8 0.005 Sh 203.4 18.1 4 <0.001 Global 208.0 22.8 14 <0.001 N 212.1 26.8 3 <0.001 T 216.2 31.0 5 <0.001

205 Table A1-6. Model selection criteria used for selecting the best-supported mixed-effects linear model of leaf number in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight S+T 89.4 0.0 6 0.5992 S+Sh+T 92.1 2.7 7 0.1585 T 92.7 3.2 5 0.1200 S*Sh+T 94.5 5.1 8 0.0479 S*T 94.5 5.1 8 0.0471 S 97.0 7.5 4 0.0138 N 99.0 9.6 3 0.0049 S+Sh 99.4 10.0 5 0.0041 Sh 100.2 10.8 4 0.0027 S*Sh 101.2 11.7 6 0.0017 Global 109.4 19.9 14 <0.001

206 Table A1-7. Model selection criteria used for selecting the best-supported mixed-effects linear model of root to shoot ratio in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight T 131.4 0.0 5 0.4147 S+T 131.6 0.3 6 0.3618 S+Sh+T 134.2 2.8 7 0.1013 S 136.5 5.2 4 0.0314 S*Sh+T 136.9 5.5 8 0.0260 S*T 137.1 5.7 8 0.0235 N 137.4 6.0 3 0.0203 S+Sh 139.0 7.6 5 0.0094 Sh 139.3 7.9 4 0.0079 S*Sh 140.8 9.5 6 0.0037 Global 151.7 20.4 14 <0.001

207 Table A1-8. Model selection criteria used for selecting the best-supported mixed-effects linear model of specific leaf area in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight S 402.1 0.0 4 0.5617 S+Sh 404.5 2.4 5 0.1679 S+T 405.0 2.9 6 0.1286 S*Sh 406.7 4.6 6 0.0559 S*T 407.4 5.4 8 0.0383 S+Sh+T 407.6 5.6 7 0.0345 S*Sh+T 410.2 8.2 8 0.0095 Global 414.2 12.1 14 0.0013 Sh 414.2 12.2 4 0.0013 N 415.1 13.1 3 <0.001 T 418.1 16.0 5 <0.001

208 Table A1-9. Model selection criteria used for selecting the best-supported mixed-effects linear model of specific root length in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight S+T 119.9 0.0 6 0.5656 S*T 121.9 2.1 8 0.2021 S+Sh+T 122.4 2.6 7 0.1578 S*Sh+T 124.5 4.6 8 0.0554 S 127.7 7.8 4 0.0115 S+Sh 130.0 10.1 5 0.0035 Global 130.7 10.8 14 0.0025 S*Sh 131.7 11.8 6 0.0015 Sh 152.5 32.6 4 <0.001 T 158.3 38.4 5 <0.001 N 158.7 38.8 3 <0.001

209 Table A1-10. Model selection criteria used for selecting the best-supported mixed-effects linear model of leaf dry matter content in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight S 48.9 0.0 4 0.5321 S+Sh 51.1 2.2 5 0.1805 S+T 52.0 3.0 6 0.1161 S*Sh 53.6 4.7 6 0.0516 S*T 53.9 5.0 8 0.0433 S+Sh+T 54.4 5.5 7 0.0337 N 55.7 6.8 3 0.0176 Sh 56.5 7.6 4 0.0121 S*Sh+T 57.2 8.3 8 0.0084 T 58.5 9.5 5 0.0045 Global 67.7 18.7 14 <0.001

210 Table A1-11. Model selection criteria used for selecting the best-supported mixed-effects linear model of red coloration in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight N 534.4 0.0 3 0.3485 S 534.8 0.4 4 0.2893 Sh 536.5 2.1 4 0.1216 S+Sh 537.1 2.7 5 0.0892 T 537.9 3.5 5 0.0599 S+T 538.5 4.1 6 0.0448 S*Sh 539.6 5.2 6 0.0262 S+Sh+T 541.1 6.7 7 0.0124 S*T 543.0 8.5 8 0.0049 S*Sh+T 543.8 9.4 8 0.0032 Global 558.7 24.3 14 <0.001

211 Table A1-12. Model selection criteria used for selecting the best-supported mixed-effects linear model of green coloration in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight S 517.8 0.0 4 0.6372 S+Sh 520.0 2.2 5 0.2124 S*Sh 522.4 4.6 6 0.0632 S+T 522.6 4.8 6 0.0568 S+Sh+T 525.0 7.3 7 0.0170 S*T 526.3 8.5 8 0.0090 S*Sh+T 527.7 9.9 8 0.0045 Sh 536.6 18.8 4 <0.001 Global 541.9 24.1 14 <0.001 N 542.1 24.3 3 <0.001 T 546.7 29.0 5 <0.001

212 Table A1-13. Model selection criteria used for selecting the best-supported mixed-effects linear model of blue coloration in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight T 479.3 0.0 5 0.3138 N 479.4 0.1 3 0.2974 Sh 481.7 2.4 4 0.0926 S 481.8 2.4 4 0.0925 S+T 481.8 2.5 6 0.0883 S*T 483.1 3.8 8 0.0473 S+Sh 484.2 4.9 5 0.0274 S+Sh+T 484.5 5.2 7 0.0236 S*Sh 486.2 6.9 6 0.0098 S*Sh+T 486.9 7.6 8 0.0072 Global 498.8 19.5 14 <0.001

213 Table A1-14. Model selection criteria used for selecting the best-supported mixed-effects linear model of circularity in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

Model AICc dAICc df weight N -114.2 0.0 3 0.4324 S -112.2 2.0 4 0.1570 Sh -111.9 2.3 4 0.1341 T -111.3 2.9 5 0.1022 S*T -110.1 4.1 8 0.0554 S+Sh -109.8 4.4 5 0.0470 S+T -109.2 5.0 6 0.0348 S*Sh -108.3 5.9 6 0.0231 S+Sh+T -106.5 7.7 7 0.0092 S*Sh+T -105.3 8.9 8 0.0049 Global -95.3 18.9 14 <0.001

214 Table A1-15. Model selection criteria used for selecting the best-supported mixed-effects linear model of roundness in seedling transplants. The first column identifies each model according to its fixed effects: site (S), taxon (T), shading (Sh), and their additive (+) and interactive (*) effects. Global gives the model for S*Sh*T while N gives the null model. Column AICc gives the sample-size corrected Akaike’s Information Criterion score, column dAICc gives the change in AICc score (ΔAICc) for subsequent models, column df provides the degrees of freedom in the model, and weight gives the Akaike weight for each model (all sum to 1).

AICc dAICc df weight Sh -98.8 0.0 4 0.2943 N -98.7 0.1 3 0.2736 S -98.1 0.8 4 0.2007 S+Sh -96.8 2.0 5 0.1066 S*T -94.6 4.3 8 0.0346 S*Sh -94.3 4.5 6 0.0306 T -94.2 4.6 5 0.0290 S+T -93.4 5.5 6 0.0192 S+Sh+T -91.9 7.0 7 0.0090 S*Sh+T -89.1 9.7 8 0.0023 Global -84.6 14.2 14 <0.001

215

Figure A1-1. Thermocron iButton temperature data recorded every two hours at the low (base), mid, and high (summit) experimental sites on Whiteface Mountain.

216 APPENDIX 2: SUPPLEMENTARY INFORMATION FOR CHAPTER 3

Table A2-1. Pairwise FST table for Nabalus boottii populations (Weir & Cockerham, 1984) calculated using R package diveRsity (Keenan, McGinnity, Cross, Crozier, & Prodöhl, 2013). Negative values are usually interpreted as functionally zero.

FST pops NB-AL NB-AP NB-AR NB-BB NB-BX NB-CH NB-CP NB-ED NB-GO NB-HA NB-LC NB-MO NB-NE NB-WF NB-WR NB-AL NB-AP 0.014 NB-AR 0.048 0.026 NB-BB 0.050 0.019 0.054 NB-BX 0.036 0.011 0.035 0.027 NB-CH 0.009 -0.013 0.044 0.026 0.014 NB-CP 0.021 -0.001 0.042 0.025 0.015 0.001 NB-ED 0.030 -0.002 0.040 0.023 0.016 0.015 0.007 NB-GO 0.024 0.004 0.025 0.036 0.023 0.019 0.016 0.012 NB-HA 0.006 0.007 0.040 0.014 0.009 -0.014 0.007 0.003 0.013 NB-LC 0.033 0.009 0.023 0.037 0.008 0.017 0.016 0.014 0.023 0.013 NB-MO 0.084 0.038 0.063 0.058 0.062 0.052 0.030 0.043 0.046 0.036 0.052 NB-NE 0.044 0.027 0.051 0.033 0.040 0.025 0.027 0.019 0.029 0.031 0.034 0.057 NB-WF 0.030 0.032 0.032 0.021 0.023 0.021 0.022 0.019 0.015 0.023 0.028 0.057 0.040 NB-WR 0.058 0.045 0.054 0.041 0.034 0.034 0.034 0.043 0.038 0.029 0.045 0.066 0.051 0.027

217 Table A2-2. Pairwise GST table for Nabalus boottii populations (Nei & Chesser, 1983) calculated using R package diveRsity (Keenan et al., 2013). Negative values are usually interpreted as functionally zero.

GST pops NB-AL NB-AP NB-AR NB-BB NB-BX NB-CH NB-CP NB-ED NB-GO NB-HA NB-LC NB-MO NB-NE NB-WF NB-WR NB-AL NB-AP -0.016 NB-AR -0.014 -0.012 NB-BB -0.010 -0.014 -0.008 NB-BX -0.012 -0.015 -0.012 -0.014 NB-CH -0.044 -0.041 -0.033 -0.039 -0.037 NB-CP -0.021 -0.022 -0.009 -0.016 -0.017 -0.045 NB-ED -0.018 -0.023 -0.013 -0.019 -0.018 -0.040 -0.024 NB-GO -0.019 -0.019 -0.017 -0.011 -0.013 -0.036 -0.017 -0.021 NB-HA -0.021 -0.016 -0.008 -0.020 -0.019 -0.042 -0.021 -0.023 -0.018 NB-LC -0.006 -0.012 -0.012 -0.003 -0.015 -0.025 -0.012 -0.014 -0.008 -0.012 NB-MO 0.005 -0.006 -0.005 -0.006 0.002 -0.030 -0.015 -0.012 -0.007 -0.011 0.003 NB-NE -0.006 -0.007 -0.002 -0.010 -0.003 -0.029 -0.010 -0.016 -0.009 -0.007 -0.002 0.000 NB-WF -0.009 0.000 -0.004 -0.009 -0.005 -0.025 -0.006 -0.010 -0.009 -0.008 0.000 0.009 0.004 NB-WR -0.007 -0.002 -0.008 -0.014 -0.012 -0.035 -0.013 -0.010 -0.010 -0.013 -0.001 -0.002 -0.003 -0.008

218 Table A2-3. Pairwise FST table for Nabalus trifoliolatus populations (Weir & Cockerham, 1984) calculated using R package diveRsity (Keenan, McGinnity, Cross, Crozier, & Prodöhl, 2013).

FST pops NN-GE NN-GI NN-LC NN-MA NN-TU NN-WA NN-WE NN-WF NN-WR NT-BR NT-BU NT-CA NT-RI NT-RP NT-TO NN-GE NN-GI 0.131 NN-LC 0.055 0.129 NN-MA 0.083 0.151 0.088 NN-TU 0.207 0.251 0.173 0.212 NN-WA 0.052 0.148 0.025 0.080 0.190 NN-WE 0.111 0.217 0.087 0.127 0.243 0.090 NN-WF 0.066 0.150 0.064 0.082 0.195 0.061 0.100 NN-WR 0.102 0.153 0.086 0.082 0.217 0.101 0.173 0.096 NT-BR 0.050 0.125 0.033 0.079 0.176 0.034 0.102 0.057 0.068 NT-BU 0.051 0.118 0.038 0.073 0.167 0.031 0.092 0.066 0.066 0.032 NT-CA 0.099 0.173 0.069 0.135 0.199 0.087 0.163 0.079 0.121 0.067 0.076 NT-RI 0.053 0.141 0.043 0.087 0.175 0.026 0.085 0.061 0.073 0.023 0.042 0.069 NT-RP 0.132 0.135 0.113 0.130 0.252 0.120 0.214 0.128 0.177 0.140 0.116 0.181 0.135 NT-TO 0.071 0.149 0.052 0.105 0.183 0.048 0.120 0.081 0.108 0.047 0.036 0.096 0.060 0.140

219 Table A2-4. Pairwise GST table for Nabalus trifoliolatus populations (Nei & Chesser, 1983) calculated using R package diveRsity (Keenan et al., 2013).

GST pops NN-GE NN-GI NN-LC NN-MA NN-TU NN-WA NN-WE NN-WF NN-WR NT-BR NT-BU NT-CA NT-RI NT-RP NT-TO NN-GE NN-GI 0.059 NN-LC 0.019 0.062 NN-MA 0.031 0.071 0.038 NN-TU 0.110 0.143 0.098 0.114 NN-WA 0.011 0.068 0.004 0.028 0.102 NN-WE 0.039 0.107 0.034 0.051 0.135 0.026 NN-WF 0.028 0.075 0.028 0.036 0.113 0.025 0.044 NN-WR 0.041 0.074 0.040 0.031 0.129 0.040 0.075 0.048 NT-BR 0.012 0.055 0.008 0.028 0.095 0.003 0.035 0.023 0.023 NT-BU 0.021 0.058 0.016 0.033 0.097 0.010 0.042 0.031 0.033 0.010 NT-CA 0.036 0.082 0.027 0.057 0.111 0.028 0.066 0.034 0.052 0.021 0.035 NT-RI 0.022 0.072 0.019 0.041 0.104 0.009 0.039 0.030 0.039 0.007 0.020 0.032 NT-RP 0.054 0.059 0.051 0.056 0.135 0.046 0.095 0.062 0.080 0.059 0.057 0.078 0.068 NT-TO 0.019 0.067 0.016 0.041 0.094 0.007 0.041 0.034 0.041 0.008 0.012 0.030 0.025 0.055

220 Figure A2-1. Markov Chain Monte Carlo (MCMC) trace plots for the best BayesAss run (lowest Bayesian deviance) for New York populations of Nabalus boottii (top panel) and Nabalus trifoliolatus (bottom panel), showing chain convergence. The red lines show log probability at 2,000-step intervals of the 30,000,000-step chain. The burn-in interval was set to 10,000,000 steps for both analyses, indicated by the dotted gray lines. Estimates from the burn-in interval were discarded.

221 Script A2-1. Example R script for gbs2ploidy analysis.

# Here I will make the input files for gbs2ploidy for Nabalus boottii # Starting text files included a table of alternate allele counts (AO), reference allele counts (RO), # a true/false table (detailed more below), and an ID table that included a list of all # individuals included in the analysis in the first column with the header “ID”. install.packages(“gbs2ploidy”) library(“gbs2ploidy”) setwd("~/Dropbox/A Dissertation/NB.NT_final/gbs2ploidy") logic <- read.table("NB_FINAL_logic.txt", row.names = 1, header = TRUE) AO <- read.table("NB_FINAL_AO.txt", row.names = 1, header = TRUE) RO <- read.table("NB_FINAL_RO.txt", row.names = 1, header = TRUE) ID <- read.table("NB_FINAL_ID.txt", header = TRUE) mat.logic <- as.matrix(logic) mat.ao <- as.matrix(AO) mat.ro <- as.matrix(RO) mat.ao.ed <- ifelse(mat.logic, mat.ao, NA) mat.ro.ed <- ifelse(mat.logic, mat.ro, NA)

# Above, I created a table with TRUE for heterozygotes and FALSE for homozygotes # Then I applied this to the allele count matrices to change values to NA if # the individual was a homozygote library(gbs2ploidy) props <- estprops(cov1 = mat.ro.ed, cov2 = mat.ao.ed, mcmc.steps = 10000, mcmc.burnin = 5000, mcmc.thin = 2, props = c(0.25, 0.5, 0.75))

H <- apply(is.na(mat.ro.ed)==FALSE,2,mean) D <- apply(mat.ro.ed+mat.ao.ed,2,mean,na.rm=TRUE) pl3<-estploidy(alphas=props,het=H,depth=D,train=FALSE,pl=NA,set=NA,nclasses=2, ids=ID) pl4<-estploidy(alphas=props,het=H,depth=D,train=FALSE,pl=NA,set=NA,nclasses=1, ids=ID) props[[74]]

# I changed the value after props in the above line from 1 to 74 to see the proportions for each # individual; I cut and pasted results into Excel # Looks like NB is tetraploid; verify with comparison to NT.

222 APPENDIX 3: SUPPLEMENTARY MATERIAL FOR CHAPTER 4

Table A3-1. Factors and loadings from the Principal Component Analysis (PCA) performed on functional trait data from the growth chamber experiment for N. trifoliolatus var. nanus and non- alpine N. trifoliolatus (small subset) only. The small subset dataset for broad-sense N. trifoliolatus consisted of measurements of 14 traits for 55 individuals. The three (or four, given a tie) loadings with the greatest absolute value under each PC are presented in bold text. Together, PCs one through seven cumulatively explained 93.4% of variance.

Trait PC1 PC2 PC3 PC4 PC5 PC6 PC7 Root length -0.173 -0.223 0.065 0.637 0.311 0.116 0.319 Dry mass: root -0.363 -0.141 0.285 -0.038 0.057 -0.037 0.166 Dry mass: shoot -0.384 0.190 -0.089 -0.062 0.035 0.086 0.088 Dry mass: total -0.409 0.067 0.061 -0.057 0.047 0.041 0.129 Root to shoot ratio -0.076 -0.414 0.547 0.044 0.058 -0.121 0.137 Specific root length 0.302 0.004 -0.352 0.255 0.311 -0.031 0.310 Height -0.212 0.384 0.080 0.257 -0.328 0.328 -0.003 Red coloration 0.239 0.351 0.448 -0.062 -0.029 0.114 0.229 Green coloration 0.174 0.317 0.417 -0.257 0.413 -0.145 0.005 Blue coloration 0.277 0.217 0.092 0.193 -0.423 0.096 0.471 Leaf shape (l/w) -0.048 0.294 0.164 0.559 -0.060 -0.530 -0.473 Specific leaf area 0.256 -0.045 0.165 0.176 0.308 0.652 -0.430 Total leaf area -0.375 0.196 -0.036 -0.039 0.093 0.288 -0.102 Leaf number -0.096 0.417 -0.188 -0.040 0.482 -0.144 0.195

223 Table A3-2. Factors and loadings from the PCA performed on functional trait data from the growth chamber experiment for N. trifoliolatus var. nanus and non-alpine N. trifoliolatus (large subset) only. The large subset dataset for broad-sense N. trifoliolatus consisted of measurements of nine traits for 78 individuals (fewer traits but more individuals than used in chapter 4 and Table A3-1). The three loadings with the greatest absolute value under each PC are presented in bold. Together, PCs one through six cumulatively explained 95.5% of variance.

Trait PC1 PC2 PC3 PC4 PC5 PC6 Dry mass: shoot -0.543 0.105 0.194 -0.109 0.108 0.113 Height -0.402 0.260 -0.361 0.062 0.237 0.196 Red coloration 0.196 0.615 0.057 -0.042 0.194 -0.038 Green coloration 0.135 0.570 0.151 -0.063 0.187 -0.473 Blue coloration 0.247 0.378 0.052 -0.008 -0.311 0.781 Leaf shape (l/w) -0.087 0.111 -0.632 -0.626 -0.361 -0.155 Specific leaf area 0.241 0.018 -0.623 0.598 0.151 -0.034 Total leaf area -0.536 0.120 -0.072 0.175 0.195 0.114 Leaf number -0.270 0.221 0.103 0.445 -0.756 -0.274

224 0.3 ●

0.2 0.2 ●

0.1 ● Taxon Taxon ● 0.0 0.0 ● NN ● NN ● ● NT NT

−0.1 PC4 (8.37%) PC2 (18.47%) ● ● ● −0.2 ● −0.2

−0.3 −0.4 −0.2 0.0 0.2 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 PC1 (40.48%) PC3 (10.1%)

0.3

0.2 0.2 ● ● 0.1 ● ● Taxon Taxon

0.0 ● NN 0.0 ● NN

● ● ● NT NT −0.1 PC6 (5.12%) ● PC7 (4.49%)

● ● −0.2 −0.2

−0.3 −0.2 0.0 0.2 −0.2 0.0 0.2 PC5 (6.33%) PC6 (5.12%)

Figure A3-1. PCA plots of functional trait data (small subset) for Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus trifoliolatus (NT) (excluding N. boottii) from the growth chamber experiment for the first seven PCs. We performed PCA on data from 55 individuals for 14 functional traits (factors and loadings provided in Table A3-1). Although many PC combinations of these seven are possible for plotting, we have chosen these four plots as sufficient to visualize separation along each PC. Envelopes enclose 100% of points for each respective taxon.

225 0.3 ● ●

● 0.2 ● 0.2 ●

● ● ● 0.1 ● Pop Pop ● ● ● ● ● ● ● ● ● ● CA ● CA ● ● 0.0 0.0 ● ● ● ● ● ●● ● TO TO ● ● ● WA ● WA −0.1 ● PC4 (8.37%) PC2 (18.47%) ● ● ●

● −0.2 ● −0.2 ● ●

● ● −0.3 −0.4 −0.2 0.0 0.2 −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 PC1 (40.48%) PC3 (10.1%)

0.3

● 0.2 0.2 ● ● ● ● ● ● Pop 0.1 ● Pop ● ● ● ● ● ● ● ● CA ● CA 0.0 ● ● 0.0 ● ● ● TO ● TO ● ● ● ●● WA −0.1 ● ● WA PC6 (5.12%) ● ● PC7 (4.49%) ● ● ● ● ● ● ● −0.2 −0.2 ●

● ● ● −0.3 ● −0.2 0.0 0.2 −0.2 0.0 0.2 PC5 (6.33%) PC6 (5.12%)

Figure A3-2. PCA plots of functional trait data (small subset) for Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus trifoliolatus (NT) (excluding N. boottii) from the growth chamber experiment for the first seven PCs; data are identical to those presented in figure A3-1 but are colored here according to population of origin instead of taxon. Only one population of Nabalus trifoliolatus var. nanus (WA, Washington Auto Road) had surviving individuals at the end of the experiment. Non-alpine Nabalus trifoliolatus was represented by two populations: Canton, Maine (CA) and Topsham, Maine (TO). We performed PCA on data from 55 individuals for 14 functional traits (factors and loadings provided in Table A3-1). Although many PC combinations of these seven are possible for plotting, we have chosen these four plots as sufficient to visualize separation along each PC. Envelopes enclose 100% of points for each respective taxon.

226 ●

0.2 ● 0.2

● ● ● ● ● ● ● Pop ● ● ● Taxon ●● 0.0 0.0 ● CA ● ● ● NN ● ●● ● ● ●● ● TO ● ● ● ●● NT ● ● WA ● ● ● PC2 (23.84%) PC2 (23.84%) ● −0.2 −0.2 ●

● −0.4 −0.4 −0.4 −0.2 0.0 0.2 −0.4 −0.2 0.0 0.2 PC1 (32.22%) PC1 (32.22%)

●

0.2 ●● 0.2 ● ● ● ●

● ● ●● ● ● ●●● ● 0.0 ● 0.0 ● Pop Taxon ●● ● ● ● ● ● ● CA ● ● ● ● ● NN ● TO NT ● −0.2 −0.2 ● WA PC4 (9.63%) PC4 (9.63%) ● ●

−0.4 −0.4 ●

−0.2 0.0 0.2 0.4 0.6 −0.2 0.0 0.2 0.4 0.6 PC3 (13.49%) PC3 (13.49%)

0.4 0.4

●

0.2 0.2 ● Pop ● Taxon ● ● ● ● ● CA ● NN ● ●● ● ● ● ● TO 0.0 NT 0.0 ● ● ● ● ● ● ● ● WA PC6 (7.75%) ● PC6 (7.75%) ● ●● ● ● ● ● ● ● ●

●

−0.2 ● −0.2 ●

●

−0.2 −0.1 0.0 0.1 0.2 −0.2 −0.1 0.0 0.1 0.2 PC5 (8.57%) PC5 (8.57%)

Figure A3-3. PCA plots of functional trait data for Nabalus trifoliolatus var. nanus (NN) and non-alpine Nabalus trifoliolatus (NT) (excluding N. boottii) from the growth chamber experiment for the first six PCs, based on data from the large subset: 78 individuals and nine functional traits (factors and loadings provided in Table A3-2). Plots in the left column are

227 (Figure A3-3 caption continued) colored according to taxon; those in the right column are colored according to population. Only one population of Nabalus trifoliolatus var. nanus (WA, Washington Auto Road) had surviving individuals at the end of the experiment. Non-alpine Nabalus trifoliolatus was represented by two populations: Canton, Maine (CA) and Topsham, Maine (TO). Although many PC combinations of these six are possible for plotting, we have chosen these plots as sufficient to visualize separation along each PC. Envelopes enclose 100% of points for each respective taxon.

228 KRISTEN HAYNES 13 N Division Street, Oswego, New York 13126 [email protected] (315) 794-5034

EDUCATION Ph.D. Ecology | Department of Environmental and Forest Biology Expected December 2019 SUNY College of Environmental Science and Forestry (SUNY-ESF) | Syracuse, NY GPA: 4.0

B.S. Natural Resources | Department of Natural Resources | Cornell University | Ithaca, NY May 2013 GPA: 4.18, Summa Cum Laude

HONORS & AWARDS Merrill Presidential Scholar, Cornell University May 2013 Academic Excellence award in Natural Resources, Cornell University May 2013 Dean’s List – all semesters, Cornell University 2009 – 2013 Ho-Nun-De-Kah Honor Society, Cornell University 2011 – 2013 High Academic Achievement Merit-based Scholarship, IES Abroad May 2011 Holland B. Brumsted Scholarship in Natural Resources, Cornell University October 2010

FIELD STATION MANAGEMENT Assistant Director August 2019 – present Rice Creek Field Station | SUNY Oswego | Oswego, NY Business Manager January 2015 – May 2018 Cranberry Lake Biological Station | SUNY-ESF | Cranberry Lake, NY

TEACHING EXPERIENCE Visiting Instructor Fall semester 2018 General Ecology Laboratory | SUNY-ESF | Syracuse, NY Invited Guest Lecturer November – December 2018 General Ecology | SUNY Onondaga Community College | Syracuse, NY Teaching Assistant (Laboratory Instructor) Spring semester 2016, 2017 Plant Evolution, Diversification and Conservation | SUNY-ESF | Syracuse, NY Teaching Assistant Fall semester 2017 General Biology I | SUNY-ESF | Syracuse, NY Teaching Assistant Fall semester 2017 Freshman Orientation Seminars | SUNY-ESF | Syracuse, NY Teaching Assistant (Recitation Instructor) Spring semester 2014, 2015 Principles of Animal Behavior | SUNY-ESF | Syracuse, NY Teaching Assistant (Coordinator and Laboratory Instructor) Fall semester 2014, 2015, 2013 Principles of Genetics Laboratory | SUNY-ESF | Syracuse, NY

UNDERGRADUATE RESEARCH MENTORSHIP Research for credit Jani Liu| SUNY-ESF | Syracuse, NY September 2017 – May 2018 Hannah Kowalsky| SUNY-ESF | Syracuse, NY September – December 2017

Volunteer research Robert Coady| SUNY-ESF | Syracuse, NY January – May 2016

Short-term lab/field research | SUNY-ESF | NY, VT, NH, ME July 2014 – May 2018 • Charlotte Bernhard • Sean Cromwell • Austin Miller • Jared Carpentier (ME) • Aaron Goodell • Siobhan Rubsam (VT) • Amanda Christiano • Alexandra Grove • Kyle Turchick (NH, ME)

229 RESEARCH EXPERIENCE Ecology and Conservation Biology August 2013 – Present Lab of Donald J. Leopold | SUNY-ESF | Syracuse, NY Chemical Ecology August 2012 – May 2013 Lab of Robert A. Raguso | Cornell University | Ithaca, NY Soil Respiration and Climate Change May 2012 Advised by Dr. Martin Maier | Albert-Ludwigs-Universität Freiburg | Freiburg, Germany Conservation Biology January – May 2011 Lab of Joseph Bernardo | Cornell University | Ithaca, NY

PUBLICATIONS & PRESENTATIONS Peer-reviewed publications: Berend, K., Haynes, K., & McDonough Mackenzie, C. (2019). Common garden experiments as a dynamic tool for ecological studies of alpine plants and communities in northeastern North America. Rhodora, 121(987), 174–212. https://doi.org/10.3119/18-16

Non-peer-reviewed publications: K. Haynes. 2014. Conservation of a rare alpine plant (Prenanthes boottii) in the face of rapid environmental change. Final report to the Edna Bailey Sussman Foundation. Available at: https://www.esf.edu/sussman/past.htm K. Haynes. 2011. Recreation use study: Fourth Lake state boat launch. In Watershed Stewardship Program Summary of Programs and Research 2011, pages 38-47. Available at: http://www.adkwatershed.org/sites/default/files/wsp_program_summary_2011.pdf

Presentations: K. Haynes. 2019. Genomic insights into the conservation of northeast alpine rattlesnake-roots (Nabalus spp.). Contributed talk at the 11th Northeast Alpine Stewardship Gathering, Lake Placid, NY. K. Haynes. 2018. Conservation of alpine rattlesnake-roots (Nabalus spp.) under climate change. Grober Graduate Research Fellowship final presentation, Cranberry Lake Biological Station, Cranberry Lake, NY. K. Haynes. 2018. Field experiments with alpine Nabalus (rattlesnake-root) taxa: Implications for conservation and climate change response. Contributed talk at the 10th Northeast Alpine Stewardship Gathering, Fairlee, VT. K. Haynes. 2015. Assessing the climate change vulnerability of the northeast alpine zone using genetics and experimental warming. Poster presented at the 9th Northeast Alpine Stewardship Gathering, Millinocket, ME. K. Haynes. 2013. Scent and color as mediators of hawkmoth (Hyles lineata) foraging behavior on Ipomopsis flowers. Poster presented at the Cornell Undergraduate Research Board spring symposium, Ithaca, NY.

GRANTS Over $30,000 secured through competitive grants to support graduate research: • Grober Graduate Research Fellowship, SUNY-ESF ($13,500) May 2017 • Edwin H. Ketchledge Scholarship, SUNY-ESF ($6,300 total) May 2014, 2016, 2017 • New York Flora Association Research Award ($500) May 2016 • Graduate Student Travel Grant, SUNY-ESF ($250) October 2015 • ADKHighpeaks Foundation Grant ($5,000) June 2014 • Edna Bailey Sussman Foundation Graduate Internship ($6,370) March 2014

OUTREACH EXPERIENCE Graduate Assistant Fall semester 2016 Adirondack Ecological Center | SUNY-ESF | Newcomb, NY Volunteer Exhibit Guide July – August 2013 The Wild Center | Tupper Lake, NY Outreach Speaker Fall semester 2012 Naturalist Outreach Practicum | Cornell University | Ithaca, NY Outreach Speaker (“Citizen Diplomat”) March – May 2012 Rent an American Program | German-American Institute Tübingen | Freiburg, Germany Watershed Steward May – September 2011 Paul Smith’s Watershed Stewardship Program | Paul Smith’s, NY

230

UNIVERSITY SERVICE & VOLUNTEERING Committees: • Animal Physiologist Search Committee | SUNY-ESF | Syracuse, NY Spring semester 2019 • Graduate Program Advisory Committee | SUNY-ESF |Syracuse, NY September 2018 – May 2019 • Cranberry Lake Biological Station Advisory Committee | SUNY-ESF |Syracuse, NY Spring semester 2015

University Outreach Events: • Judge, Environmental Summit Research Symposium June 2018 ESF in the High School | SUNY-ESF | Syracuse, NY • Judge, Environmental Challenge | SUNY-ESF | Syracuse, NY May 2015, 2018 • Buddy/Guide, Expanding Your Horizons | Cornell University | Ithaca, NY April 2013 • Volunteer Interpreter | Insectapalooza | Cornell University | Ithaca, NY October 2012 • Guide, Society of Women Engineers Brownie/Girl Scout Day March 2011 Cornell University | Ithaca, NY

PROFESSIONAL DEVELOPMENT IN EDUCATION Certificate in University Teaching Completed April 2019 Future Professoriate Program | Syracuse University | Syracuse, NY Graduate Assistant Colloquium on Teaching and Learning August 2013 SUNY-ESF | Syracuse, NY

PROFESSIONAL SOCIETIES • New York Flora Association January 2017 – Present • American Genetics Association 2016 – 2018

231