LIVING at the EDGE: PERIPHERAL PLANT POPULATIONS and CLIMATE CHANGE by Sabine Dietz B.A., Trent University, 1991 M.É.E, Univers

LIVING AT THE EDGE: PERIPHERAL PLANT POPULATIONS AND CLIMATE CHANGE by

Sabine Dietz

B.A., Trent University, 1991

M.É.E, Université de Moncton, 2003

A Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

in the Graduate Academic Unit of Biology

Supervisors: Kate Frego, Ph.D., Biology, University of New Brunswick Jennifer Baltzer, Ph.D., Biology, Wilfrid Laurier University

Examining Board: Scott Pavey, Ph.D., Biology Lucy Wilson, Ph.D., Geology and Anthropology Joe Nocera, Ph.D., Forestry and Environmental Management

External Examiner: Steve Mockford, Ph.D. Biology, Acadia University

This dissertation is accepted by the Dean of Graduate Studies

THE UNIVERSITY OF NEW BRUNSWICK

July 2017

©Sabine Dietz, 2017

ABSTRACT

Climate change is expected to place peripheral populations of some species at higher risk of extirpation than others, yet these populations may be significant because they may contain important evolutionary potential. Assessing the conservation significance of peripheral populations is complex, and requires an evaluation of their genetic and ecological distinctness, as well as their likely persistence (extirpation risk) in a changing climate.

To address the question of whether rear-edge populations of widespread species are genetically distinct, I used Amplified Fragment Length Polymorphisms (AFLPs) to assess genetic diversity and genetic differentiation of arctic-alpine Anemone parviflora,

Dryas integrifolia and Vaccinium uliginosum in Atlantic Canada. The three species showed different genetic patterns at the rear range edge possibly associated with different dispersal capacities and phylogeographic histories. No strong predictors of genetic patterns (disjunction, distance, disturbance) could be confirmed for rear range- edge populations. Some disjunct populations of all three species maintained considerable genetic diversity.

To evaluate if rear-edge disjunct populations of A. parviflora, and D. integrifolia at

Wilson Brook, NB, are at risk of extirpation from a warming climate, the microclimate of the site was compared to local, regional and northern climates, and short-term responses of population growth to microclimatic variation were evaluated through

survey and experimental approaches (Open Top Chambers). The persistence of rear- edge disjunct populations at Wilson Brook may be attributable to a cooler local microclimate. While the temperature manipulation experiment did not increase temperatures in the experimental plots as intended, there were no biologically significant associations of microclimate variables with population growth.

Species prioritization based on extinction risk assessments may not be useful for peripheral populations of widespread species, because peripheral populations need to be evaluated at the population rather than species level. I propose a framework that is based on genetic and ecological distinctness as well as extinction risk, and apply it to disjunct range-edge populations of the three species. While none of the disjunct populations were identified as of high conservation significance through this process, my research highlights the value of a population-level approach for these rear-edge disjunct populations.

iii

ACKNOWLEDGEMENTS

I would like to thank my supervisors Drs. Kate Frego and Jennifer Baltzer for their guidance, as well as my supervisory committee, Drs. Stephen Clayden, David Kubien,

Chris Martyniuk, Jeff Houlahan, Graham Forbes, and especially Dr. Karen Samis, who was invaluable with her genetics help.

The following individuals helped with fieldwork: Gart Bishop (2010 & 2011), David

Watts (2010), Taryn ONeill (2009), Laura Siegwart (Labrador, 2010), Shirley Alward

(Port au Choix NHS, 2011 & 2012 data logger monitoring), Roland Chiasson, Mira

Dietz Chiasson, Lena Dietz Chiasson, Ulla Von Schroetter (2009-2013). Thanks also to

Claude Isabel for facilitating field work in the Parc de la Gaspésie, Claudia Hamel for aiding with a permit to sample in Newfoundland ecological reserves, and Maryse

Bourgeois of NB Department of Energy and Resource Development for facilitating a permit to work in the Wilson Brook Protected Natural Area.

In addition, I would like to acknowledge Dr. Amanda Cockshutt and Nathalie Donaher from Mount Allison University for letting me use their lab, and helping me with numerous questions that I had about molecular work; Dr. Felix Baerlocher of Mount

Allison University for the use of the Nanodrop; Tim Barton (UNB –data organization);

Michelle Leonard (UNB – germination experiment); Dr. Ed Guerrant, Berry Botanic

Garden who provided advice on seed collecting and conservation; Richard Fournier

(UdeM Edmundston, germination information); Michael Burzynski (Limestone Barrens

history & information); Dr. Luise Hermanutz (Limestone Barrens, NL; assistance with fieldwork in Labrador); Dr. Martin Batterson (MUN, glacial history, sea level change,

Newfoundland); Dr. Alyre Chiasson (UdeM), for his help with statistics (Chapter 3); and

Sean Blaney, Atlantic Conservation Data Centre, for providing records of the target species.

I would also like to acknowledge financial support received from Dr. Baltzer through her Marjorie Young Bell Faculty Fellowship at Mount Allison University and Natural

Science and Engineering Research Council Discovery Grant, the Northern Studies

Travel Program (Polar Knowledge Canada), the NB Wildlife Trust Fund, University of

NB, Saint John (part-time graduate awards), and Dr. Karen Samis, University of PEI.

A special thanks goes to Dr. Kate Frego for the many lunches, and her generous support over the last years. And finally, but definitely not least, thanks to my family, Roland,

Mira and Lena, for having supported me on this very long journey, keeping me safe visiting Wilson Brook, and Mira and Ingrid Tollefsen, for editing various versions.

Table of Contents

ABSTRACT ...... ii

ACKNOWLEDGEMENTS ...... iv

Table of Contents ...... vi

List of Tables ...... xii

List of Figures ...... xvi

List of symbols, nomenclature and abbreviations ...... xxii

Chapter 1 - Introduction ...... 1

1.2 Geographic distribution patterns and range limits ...... 3 1.2.1 Geographic range patterns ...... 4 1.2.2 Determinants of range limits ...... 5 1.2.2.1 Fitness and dispersal ...... 7 1.2.2.2 Genetics and range edges ...... 8 1.2.2.3 Historical factors ...... 9 1.3 Geographic distributions and climate change ...... 11 1.4 Populations at range edges ...... 14 1.5 Study species and locations ...... 18 1.5.1 Arctic-alpine species ...... 18 1.5.1.1 White Mountain Avens ...... 18 1.5.1.2 Northern White Anemone ...... 19 1.5.1.3 Northern Bilberry ...... 20 1.5.2 Study locations ...... 21 1.5.2.1 Central populations ...... 21 1.5.2.2 Range-edge populations ...... 22 vi

1.5.2.3 Disjunct populations ...... 22 1.6 Structure of thesis ...... 26 1.6.1 Chapter 2, “Genetic diversity and differentiation of rear-edge populations of three arctic-alpine species in Eastern Canada”...... 26 1.6.2 Chapter 3, “Microclimate and demography in rear-edge populations of two arctic-alpine species (Dryas integrifolia and Anemone parviflora) at Wilson Brook Protected Natural Area, New Brunswick”...... 27 1.6.3 Chapter 4, “Assessing the conservation significance of peripheral populations – a case study of three arctic-alpine species in Eastern Canada”...... 29 1.6.4 Chapter 5, “Living at the edge - summary” ...... 30 Figures ...... 31 List of references ...... 35

Chapter 2 - Genetic diversity and differentiation of rear-edge populations of three arctic-alpine species in Eastern Canada...... 47

Abstract ...... 47 2.1 Introduction ...... 49 2.1.1 Goals and objectives ...... 55 2.2 Materials and Methods ...... 57 2.2.1 Study region ...... 57 2.2.2 Study species ...... 57 2.2.3 Study site classification and field sampling ...... 60 2.2.4 Analysis of genetic variation ...... 65 2.2.4.1 DNA extraction ...... 65 2.2.4.2 AFLP genotyping ...... 66 2.3 Data Analysis ...... 72 2.3.1 Genetic diversity and differentiation ...... 72 2.3.1.1 Genetic diversity within populations ...... 72 2.3.1.2 Genetic differentiation and structure among populations ...... 73 vii

2.3.1.3 Statistical tests analyzing genetic diversity and differentiation ...... 73 2.3.2 Distance, disjunction and levels of disturbance ...... 75 2.4 Results ...... 78 2.4.1 Genetic diversity and differentiation ...... 78 2.4.1.1 Anemone parviflora ...... 78 2.4.1.2 Dryas integrifolia ...... 80 2.4.1.3 Vaccinium uliginosum ...... 82 2.4.2 Association among genetics, distance, disjunction and levels of disturbance . 83 2.4.2.1 Isolation by distance ...... 84 2.4.2.2 Disjunction ...... 85 2.4.2.3 Disturbance ...... 86 2.5 Discussion ...... 87 2.5.1 Genetic variability patterns at the rear range edge ...... 88 2.5.2 Patterns of genetic variability in Anemone parviflora, Vaccinium uliginosum and Dryas integrifolia ...... 92 2.5.2.1 Anemone parviflora ...... 92 2.5.2.2 Dryas integrifolia ...... 93 2.5.2.3 Vaccinium uliginosum ...... 96 2.5.3 Potential predictors of genetic variability patterns at range edge ...... 97 2.6 Conservation implications and future research ...... 100 Tables ...... 102 Figures ...... 111 List of references ...... 125

Chapter 3 - Microclimate and demography in rear-edge populations of two arctic- alpine species (Dryas integrifolia and Anemone parviflora) at Wilson Brook Protected Natural Area, New Brunswick ...... 140

Abstract ...... 140 3.1 Introduction ...... 142 viii

3.2 Methods ...... 150 3.2.1 Study sites ...... 150 3.2.2 Study species ...... 151 3.2.3 Microclimate variables ...... 153 3.2.4. Warming experiment ...... 154 3.2.5 Data collection ...... 156 3.2.5.1 Regional climate and microclimate data ...... 156 3.3 Data analysis ...... 159 3.3.1 Regional and microclimate at Wilson Brook versus Albert Mines and Port-au- Choix ...... 160 3.3.2 Warming experiment: OTCs versus controls at Wilson Brook ...... 160 3.3.3 Population growth response ...... 161 3.3.3.1 Population growth response to experimental warming: OTCs versus controls ...... 162 3.3.3.2 Population growth response to microclimate variables ...... 162 3.4 Results ...... 164 3.4.1 Regional climates and local microclimates ...... 164 3.4.2 Effectiveness of warming experiment ...... 165 3.4.3 Population growth response to microclimate at Wilson Brook ...... 166 3.5 Discussion ...... 168 3.5.1 Microclimate at Wilson Brook ...... 168 3.5.1.1 Cooling effect of Open Top Chambers at Wilson Brook ...... 171 3.5.2 Population growth and microclimatic variation at Wilson Brook ...... 173 3.6 Conservation implications for Wilson Brook ...... 176 Tables ...... 178 Figures ...... 179 List of references ...... 194

Chapter 4 - Assessing the conservation significance of peripheral populations - a

case study of three arctic-alpine species in Eastern Canada ...... 205

Abstract ...... 205 4.1 Introduction ...... 206 4.1.1 Objectives ...... 210 4.2 Conservation of peripheral populations ...... 212 4.2.1 Potential conservation significance of peripheral populations ...... 212 4.2.2 Challenges managing peripheral populations ...... 213 4.3 Prioritization using extinction risk assessment ...... 218 4.4 Population-based assessment framework for peripheral populations of common species ...... 221 4.4.1 Evaluating population genetic distinctness ...... 223 4.4.1.1 Direct predictors of genetic distinctness ...... 223 4.4.1.2 Indirect predictors of genetic distinctness ...... 224 4.4.2 Assessing risk and final prioritization ...... 226 4.5 Case study: assessment of rear-edge, disjunct populations of three arctic-alpine species for conservation ...... 226 4.5.1 Genetic distinctness (diversity and differentiation) ...... 227 4.5.2 Indirect predictors of genetic distinctness ...... 228 4.5.3 Risk of extirpation with climate warming ...... 230 4.6 Conservation significance of disjunct, rear-edge, populations of three arctic-alpine species ...... 233 4.7 Conclusion ...... 235 Tables ...... 236 Figures ...... 239 List of references ...... 242

Chapter 5 - Living at the edge - summary ...... 249

5.1.1 Genetic distinctness ...... 250 x

5.1.2 Association of ecological and demographic traits with genetic variability ... 251 5.1.3 Extinction risk from climate warming? ...... 252 5.1.4 Assessing the conservation significance of peripheral populations ...... 253 5.2 Implications of my findings ...... 255 5.3 Future directions ...... 257 List of references ...... 260

Appendices ...... 262

Curriculum Vitae ...... 277

List of Tables

Table 2.1. Locations of populations of three species sampled for genetic analysis at each species rear edge in Atlantic Canada...... 102

Table 2.2. Metrics of Anemone parviflora (AP), Dryas integrifolia (DI) and Vaccinium uliginosum (VU) populations sampled for genetic analysis. Range location: populations were grouped as one of three categories: center, range edge, disjunct1; distance from range edge, with edge identified as 0 km; habitat area represents the estimated area occupied by each population; population size is given as approximate number of ramets or plants within the area occupied2; disturbance level refers to recent and historic anthropogenic disturbance3...... 103

Table 2.3. Amplified Fragment Length Polymorphism (AFLP) primer pairs, total number of bands scored, bands retained, error rates and percent replication per primer pair for genetic analysis in Anemone parviflora, Dryas integrifolia and Vaccinium uliginosum...... 105

Table 2.4. Measures of genetic variation calculated within species and populations using binary matrices in the indicated programs. Only %P is presented in the results...... 106

Table 2.5. Mean population genetic diversity (%P) and pairwise differentiation (Φ'PT) for sampled populations of Anemone parviflora, Dryas integrifolia and Vaccinium uliginosum, listed from north to south; n = number of individuals sampled per population; mean %P percent polymorphic loci, within population genetic variation; mean Φ'PT = an analogue of FST, genetic differentiation among population, calculated as mean per population. See text for details on metric calculations. Lines separate central, edge and disjunct populations (see Table 2.2)...... 107

Table 2.6. Partitioning of genetic variation in Anemone parviflora, Dryas integrifolia, and Vaccinium uliginosum determined by analysis of molecular variance (AMOVA). The analysis used a genetic distance (calculated from AFLP genotypes) by geographic distance (km) matrix. Regions were defined as centre, edge and disjunct as described in Table 2.2...... 108

Table 2.7. Pairwise genetic differentiation using GenAlEx v. 6.502 (Peakall & Smouse 2006, 2012) (Φ'PT) for (a) Anemone parviflora, (b) Dryas integrifolia, and (c) Vaccinium uliginosum populations sampled between Labrador and each species' southern range limit in New Brunswick, Canada. Φ'PT, an analog of FST, (shaded) represents the coefficient of population differentiation data (< 0.05 = little differentiation, 0.05–0.15 = moderate genetic differentiation, 0.15–0.25 = great genetic differentiation, > 0.25 = very great genetic differentiation). P-values (above the diagonal, uncorrected) for tests of significance (Ho: Φ'PT = 0 or no differentiation among pairs) were based on 999 xii

permutations; * = significant at P < 0.05. Populations are listed from north to south. . 109

Table 3.1. Summary of micro data collection at 10 min intervals at three sites, as well as regional data used from Environment Canada (Government of Canada 2016)...... 178

Table 4.1. Decision matrix for the assessment of qualitative conservation priority (high, intermediate, low) of disjunct rear-edge populations: xxx = high, xx = intermediate, x = low. Direct evidence of high genetic uniqueness (genetic differentiation) and high genetic diversity within populations using molecular markers should be used as a priority to determine conservation priority (level 1). If genetic information is not available, or incomplete, predictors of distinctiveness could be used (level 2). Either level 1 or level 2 provides information on the populations' priority. To determine the populations' conservation priority, risk of extirpation should be assessed as well (level 3)...... 236

Table 4.2. Qualitative conservation priority (high, intermediate, low) of disjunct rear- edge populations of Dryas integrifolia (Di), Anemone parviflora (Ap), and Vaccinium uliginosum (Vu) based on proposed criteria, as described in text; xxx = high, xx = intermediate, x = low; Pop = population (WB, Wilson Brook; ML, Mount Logan; TL, Tablelands; RS, Restigouche), described in Chapter 1. Direct evidence of high genetic distinctness would place a population in a high conservation priority. See Table 4.1 for details...... 238

Appendix 2.1. Binary matrix file (excerpt) of Dryas integrifolia used in AFLP analysis of within- and among-population genetic diversity, including sample identification, sampling location, latitude and longitude coordinates, and presence or absence of bands for one primer pair...... 262

Appendix 2.2. Population genetic parameters for sampled populations of Anemone parviflora (a), Dryas integrifolia (b) and Vaccinium uliginosum (c) listed from north to south, identified as central, edge or disjunct populations; n = number of individuals sampled per population; # of bands = number of AFLP bands present across all primer combinations; PB = number of private / unique bands; %P = percent polymorphic loci. In addition, the following are reported for comparison purposes: He = expected heterozygosity following Lynch and Milligan (1994) and assuming complete outcrossing; PLP1% = percent polymorphic loci at 1%; I = Shannon Information Index; Pb(2/4/4) = allelic richness calculated using the rarefaction to two/four/four individuals; AE = effective number of alleles; D(2/4/4) Nei's gene diversity (AFLPDat). See Table 2.5 for details on metric calculations and programs used...... 263

Appendix 2.3. Independent sample t-test comparing the %P per population with the mean %P for all populations per species...... 265

Appendix 2.4. Correlation coefficients of mean population genetic metric above diagonal, P-values below diagonal; a) including populations from all three species xiii

Anemone parviflora, Dryas integrifolia and Vaccinium uliginosum (# of populations = 26); b) excluding A. parviflora populations (# of populations = 18); n = number of individuals (all per population), bands = number of AFLP bands present across all primer combinations, PB = number of private / unique bands, %P = percent polymorphic loci, He = expected heterozygosity, PLP1% = percent polymorphic loci at 1%;, I = Shannon information index, Pb = Allelic richness (rarefied), AE = Effective number of alleles, D = Nei's gene diversity (rarefied), Φ'PT = mean pairwise population genetic differentiation. See text and Table 2.4 for details on metrics...... 266

Appendix 3.1. Microclimate variable calculations ...... 269

Appendix 3.2. Post hoc comparisons (Tukey contrasts) of the linear mixed effects model of location vs. year interactions1: mean daily temperatures for (a) Wilson Brook (local data, n = 5), and three regional weather stations each in Newfoundland (NL-regional, Daniels Harbour, Ferrolle Point, Plum Point) and New Brunswick (NB-regional, Alma, Moncton, Nappan) (data downloaded from Environment Canada) (b) over three growing seasons (1 June - 9 November, 2011 & 2013). The least-squared mean estimates for the fixed effect location was obtained using the lsmeans (Lenth 2016) and pbkrtest (Halekoh & Højsgaard 2014) packages. F test is calculated according to the Kenward-Roger method (Kenward & Roger, 1997, Halekoh, 2014)...... 270

Appendix 3.3. Post hoc comparisons (Tukey contrasts) of the linear mixed effects model of location vs. month interactions1: mean daily temperatures for Wilson Brook (local data, n = 5), three regional weather stations each in Newfoundland (NL-regional, Daniels Harbour, Ferrolle Point, Plum Point) and New Brunswick (NB-regional, Alma, Moncton, Nappan) (data downloaded from Environment Canada), and Port-au-Choix (local data n = 3) from 1 June - 23 October 2012. The least-squared mean estimates for the fixed effect location was obtained using the lsmeans (Lenth 2016) and pbkrtest (Halekoh & Højsgaard 2014) packages. F test is calculated according to the Kenward- Roger method (Kenward & Roger, 1997, Halekoh, 2014)...... 271

Appendix 3.4. Coefficients of the linear mixed effects model analysis using the lme4 package (Bates et al. 2015) (lmer in R) of microclimate variables comparing Wilson Brook with Port-au-Choix (treatment, 2012, Table 3.1)...... 273

Appendix 3.5. Coefficients of the linear mixed effects model analysis using the lme4 package (Bates et al. 2015) (lmer in R) of microclimate variables comparing Wilson Brook with Albert Mines (treatment) for 2013 (Table 3.1)...... 274

derived from log likelihood comparison. The least-squared mean estimates for the fixed effect treatment was obtained using the lsmeans (Lenth 2016) and pbkrtest (Halekoh & Højsgaard 2014) packages. F test is calculated according to the Kenward-Roger method (Kenward & Roger, 1997, Halekoh, 2014)...... 275

List of Figures

Figure 1.1 Generalized patterns across the range, from the leading edge to the rear edge (after Hampe & Petit (2005) and Woolbright et al. (2014))...... 31

Figure 1.2 North American distribution of Dryas integrifolia (left) and a reproductive plant in a natural population (arrow points to plumose achene, right). Map created using SimplMapr (Shorthouse 2010) with occurrence records from GBIF (GBIF.org 2016). Photo: S. Dietz ...... 32

Figure 1.3 North American distribution of Anemone parviflora (left) and plant in flower (right). Map created using SimplMapr (Shorthouse 2010) with occurrence records from GBIF (GBIF.org 2016). Photo: S. Dietz ...... 33

Figure 1.4 Canadian distribution of Vaccinium uliginosum (left) and plant with fruit (right). Map created using SimplMapr (Shorthouse 2010) with occurrence records from GBIF (GBIF.org 2016). Photo: S. Dietz ...... 34

Figure 2.1. Sampling locations of Anemone parviflora (AP), Dryas integrifolia (DI) and Vaccinium uliginosum (VU) in Eastern Canada used in this study. Number of samples included in genetic analysis per species are provided in brackets for each site. See Tables 2.1 and 2.2 for descriptions of the populations sampled. All populations occur within either the Eastern Temperate Forest eco-region of the Bay of Fundy or the Arctic Cordillera eco-region (CEC 2009), and were classified according to intraspecific range- edge proximity into (1) central populations, (2) range-edge populations, (3) disjunct populations (Tables 2.1 and 2.2). Central populations are identified as those populations that occur in the range where the species is common based on range maps and occurrence records (pers. com ACCDC 2017)...... 111

Figure 2.2. Sampling of species at Port-au-Choix, Newfoundland with patches of Dryas integrifolia visible in the foreground along the limestone gravel substrate. A transect (white line) was laid for the purpose of estimating population size. See text for details...... 112

Figure 2.3. Sample AFLP results demonstrating the success of the AFLP protocol at each step. (a) DNA extraction: the two first lanes show high molecular weight DNA to be used in AFLP genotyping, and lane 3 represents the 1kb molecular weight ladder. (b) Pre-amplification: lanes 2, 3 and 5 show successful fragment pre-amplification with a wide range of fragment sizes (in contrast to failed amplifications in lanes 1 and 4); lane 6 shows a 1kb ladder for comparison. (c) Final selective amplification: all sample lanes represent successfully amplified fragments with visible band sizes between 50 and 2000 bp; lane 2 shows the DNA molecular weight ladder. Arrows indicate the 1kb fragment on each molecular weight marker...... 113 xvi

Figure 2.4. Genetic population structure of Anemone parviflora. Upper panel (a), principal coordinate analysis (PCoA) of genetic distances (n = 35 A. parviflora individuals from eight populations). Axis one explained 29.40% of variation and axis two explained 18.26% of variation, for a cumulative 47.66% variation explained. Lower panel (b), centroids of mean scores on axes 1 and 2 of the PCoA, with ellipses indicating ± SE based on genetic distances. See Fig. 2.1 for population geographic locations...... 114

Figure 2.5. Genetic population structure of Dryas integrifolia. Upper panel (a), principal coordinate analysis (PCoA) of genetic distances (n = 73 D. integrifolia individuals from nine populations). Axis one explained 37.16% of variation and axis two explained 24.96% of variation, for a cumulative 62.14% variation explained. Lower panel (b), centroids of mean scores on axes 1 and 2 of the PCoA, with ellipses indicating ± SE based on genetic distances. See Fig. 2.1 for population geographic locations...... 115

Figure 2.6. Distribution of individual assignment to putative ancestral clusters (K = 3) determined from STRUCTURE analysis within and among populations sampled for this study (Pritchard et al. 2000, Pritchard et al. 2010). Upper panel (a), proportional cluster memberships within the eight Dryas integrifolia populations. The pie charts superimposed on the map show the average proportions of cluster memberships among the individuals sampled in each population. Lower panel (b), STRUCTURE analysis results for assignment of genotypes to three putative ancestral lineages (indicated by shade) of sampled populations of Dryas integrifolia. Each vertical bar represents one individual (n = 73). Shading indicates the proportion of an individual's genotype that is assigned to each lineage...... 116

Figure 2.7. Genetic population structure of Vaccinium uliginosum. Upper panel (a), principal coordinate analysis (PCoA) of genetic distances (n = 80 V. uliginosum individuals from eight populations). Axis one explained 4.83% of variation and axis two explained 3.97% of variation, for a cumulative 8.80% variation explained. Lower panel (b), centroids of mean scores on axes 1 and 2 of the PCoA, with ellipses indicating ± SE based on genetic distances. See Fig. 2.1 for population geographic locations...... 117

Figure 2.8. Relationship (line) between genetic distance and geographic distance between individuals using the standard Mantel analysis in GenAlEx (Peakall & Smouse 2006, 2012). Anemone parviflora (a), n = 35, P = 0.325), Dryas integrifolia (b), n = 73, P < 0.001), and Vaccinium uliginosum (c), n = 80, P = 0.010), sampled in seven, nine and eight populations, respectively across the species' southern geographic range...... 118

Figure 2.9. Relationship between geographic distance to range edge (km) on the x-axis and panel (a), mean population genetic variation (%P) on the y-axis for seven populations of Anemone parviflora, (F1,6 = 0.22, P = 0.652), dashed line (ns); and panel (b), mean population genetic differentiation (Φ'PT) on the y-axis (F1,6 = 4.64, P = 0.075). Triangles – edge: Burnt Cape, Cooks Harbour, Sandy Cove, Port-au-Choix, Bellburns;

xvii

circles – disjunct: Tablelands, Restigouche, Wilson Brook...... 119

Figure 2.10. Relationship between geographic distance to range edge (km) and panel (a), mean population genetic variation (%P) on the y-axis for eight populations of Dryas integrifolia (excluding the core population at Torngat, F1,6 = 1.55, P = 0.260); dashed line (ns); and panel (b), mean population genetic differentiation (Φ'PT) on the y-axis (excluding the core population at Torngat, F1,6 = 5.58, P = 0.056). Square – central: Torngat; triangles – edge: Burnt Cape, Cooks Harbour, Sandy Cove, Port-au-Choix, Bellburns; circles – disjunct: Tablelands, Mount Logan, Wilson Brook. Because genetic differentiation can be positively associated with maximum geographic distances (Reisch & Bernhardt-Römermann 2014), analyses excluded central populations...... 120

Figure 2.11. Relationship between geographic distance to range edge (km) and panel (a), mean population genetic variation (%P) on the y-axis for 8 populations of Vaccinium uliginosum (excluding the core populations at Torngat and Nain, F1,5 = 3.5, P = 0.120); dashed line (ns); and panel (b), mean population genetic differentiation (Φ'PT) on the y- axis (excluding the core population at Torngat, F1,5 < 0.001, P = 0.978), dashed line (ns). Squares – central: Torngat, Nain; triangles – edge: Burnt Cape, Cooks Harbour, Sandy Cove, Port-au-Choix, Bellburns, Tablelands; circle – disjunct: Mount Logan. Because genetic differentiation can be positively associated with maximum geographic distances (Reisch & Bernhardt-Römermann 2014), analyses excluded central populations...... 121

Figure 2.12. Association of genetic diversity and differentiation with disjunction for three species, Anemone parviflora (AP), Dryas integrifolia (DI) and Vaccinium uliginosum (VU). Panel (a), genetic diversity in %P (mean ± SE, ANOVA and TukeyHSD) for continuous range edge populations (edge & center) (0.49 ± 0.14) versus disjunct populations (0.46 ± 0.20) as defined in text. Genetic diversity did not differ significantly between disjunct, and edge and center populations (F1,21 = 1.21, P = 0.285). Panel (b), genetic differentiation in Φ'PT (mean ± SE) for continuous range edge populations (0.08 ± 0.08) versus disjunct populations (0.17 ± 0.09) as defined in text. Significant differences were detected among species (F2,21 = 25.70, P < 0.001) and disjunction (F1,21 = 6.03, P = 0.023)...... 122

Figure 2.13. Association of genetic diversity and differentiation with disturbance for three species, Anemone parviflora (AP), Dryas integrifolia (DI) and Vaccinium uliginosum (VU). Panel (a), genetic diversity in %P (mean ± SE, ANOVA, TukeyHSD) for all populations of the three species for undisturbed sites (0.53 ± 0.19), recovering sites (0.49 ± 0.13) and disturbed sites (0.39 ± 0.07), disturbance as defined in text. No significant differences were detected among disturbance levels (F2,21 = 3.77, P = 0.066); Panel (b), genetic differentiation in Φ'PT (mean ± SE) for all populations and undisturbed sites (0.14 ± 0.11), recovering sites (0.09 ± 0.07) and disturbed sites (0.08 ± 0.06). Genetic differentiation was significantly associated with disturbance (F2,21 = 6.86, P = 0.016), and differed significantly with species (F2,21 = 25.70, P < 0.001)...... 123

xviii

Figure 2.14. Map showing proposed colonization routes for Dryas integrifolia across Eastern North America after the Last Glacial Maximum (LGM). Grey arrows show colonization routes after LGM hypothesized by Tremblay and Schoen (1999); black arrows show suggested re-colonization routes inferred from results of the present study. Dashed line indicates maximum ice cover during LGM according to Paulen and McClenaghan (2013) and Shaw (2006)...... 124

Figure 3.1. Study sites: Port-au-Choix, Wilson Brook and Albert Mines. The insert shows the locations of plots at Wilson Brook. Wilson Brook: 45.86005, –64.67611, Albert Mines: 45.86443, –64.66719, Port-au-Choix: 50.70065, –57.37788...... 179

Figure 3.2. Wilson Brook Protected Natural Area, New Brunswick, Canada, gypsum cliff habitat. Low vegetation areas are Dryas integrifolia patches. Photo: Sabine Dietz, 2011 ...... 180

Figure 3.3. Open Top Chamber (background) and control plot (foreground), Plot # 10, in 2011. Both plots (diameter 32 cm, 0.08 m2) have data loggers, visible in control plot, normally covered by vegetation. Wilson Brook Protected Natural Area, New Brunswick, Canada. Photo: Sabine Dietz, 2011 ...... 181

Figure 3.4. Open Top Chamber located in unstable location where rocks (white) moved due to erosion during the 3 seasons of an experiment at Wilson Brook Protected Natural Area, New Brunswick, Canada. Photo: Sabine Dietz, 2011 ...... 182

Figure 3.5. Post-hoc comparisons (Tukey contrasts) of the linear mixed effects model of location vs. year interactions: mean daily temperatures (mean ± first and third quartile) for Wilson Brook (local data, n = 5, WB), and three regional weather stations each in Newfoundland (Daniels Harbour, Ferrolle Point, Plum Point, NL-regional) and New Brunswick (Alma, Moncton, Nappan NS, NB-regional, data downloaded from Government of Canada (2016), over three growing seasons (1 June to 9 November, 2011 & 2013). The least-squared mean estimates for the fixed effect location were obtained using the lsmeans (Lenth 2016) and pbkrtest (Halekoh & Højsgaard 2014) packages. There was a significant interaction between location and year (χ2 = 38. 93, df = 4, P < 0.001). There was a significant difference between Wilson Brook and NB region in 2011, with mean daily temperatures in NB region significantly higher than Wilson Brook (P = 0.027); this difference was not apparent in other years. Newfoundland regional mean daily temperatures were significantly lower than both Wilson Brook (except in 2012) and NB region (Appendix 3.4)...... 183

of year and location (F20, 24 = 2.70, P = 0.059). GDD was significantly lower in NL- regional than in NB-regional (by 515 GDD, P < 0.001) or Wilson Brook (by 630 GDD, P < 0.001), over three years...... 184

Figure 3.7. Post-hoc comparisons (Tukey contrasts) of mean daily temperatures in °C (mean ± first and third quartile) by month. New Brunswick weather stations (Alma, Moncton, Nappan NS, NB-regional), Newfoundland weather stations (Daniels Harbour, Ferrolle Point, Plum Point, NL regional), Port-au-Choix (n = 5) and Wilson Brook (n = 5), 1 June to 23 October (2012). Regional data from (Government of Canada 2016). The least-squared mean estimates for the fixed effect location were obtained using the lsmeans (Lenth 2016) and pbkrtest (Halekoh & Højsgaard 2014) packages. There was a significant interaction of month and location (χ2 = 133.88, df = 12, P < 0.001). Mean daily temperatures in NB region did not differ significantly from Wilson Brook (P > 0.05, Appendix 3.5)...... 185

Figure 3.8. Distribution of vapour pressure deficit for Port-au-Choix (n = 5) compared to Wilson Brook (n = 5) for 2012 as recorded by ground level data loggers at five 32 cm diameter (0.08 m2) plots per site. VPD was significantly higher at Wilson Brook by 0.25 kPa ± 0.04 SE than at Port-au-Choix (χ2 = 17.34, df = 1, P < 0.001); neither month nor the interaction between month and location was significant...... 186

Figure 3.9. Comparison of maximum daily temperatures at Wilson Brook (n = 5) and Albert Mines (n = 3) for 2013 as recorded by ground level data loggers at 32 cm diameter (0.08 m2) plots per site. Maximum daily temperature at Wilson Brook was significantly higher by 3.36 °C ± 1.17 SE than at Albert Mines (χ2 = 5.43, df = 1, P < 0.02); neither month nor the interaction between month and location was significant. 187

Figure 3.10. Comparison of vapour pressure deficit at Wilson Brook (n = 5) and Albert Mines (n = 3) for 2013 as recorded by ground level data loggers at 32 cm diameter plots (0.08 m2) per site. VPD at Wilson Brook was significantly lower by 0.12 kPa ± 0.01 SE at Wilson Brook than Albert Mines (χ2 = 17.74, df = 1, P < 0.001); neither month nor the interaction between month and location was significant...... 188

Figure 3.11. Comparison of maximum daily temperatures in OTCs (white bars, n = 10) versus control plots (shaded bars, n = 5) at Wilson Brook from 1 June to 9 November 2011 to 2013, by year (mean ± first and third quartile). There was a significant interaction between year and treatment (χ2 = 19.31, df = 2, P < 0.001). There was a significant difference between treatments in 2013 only (P < 0.001), with lower maximum daily temperatures in OTC's compared to controls, and there were significant differences within treatments among years (Appendix 3.7b) ...... 189

mean ± SD, (b) mean ± first and third quartile). (a) There were fewer > 40°C temperatures in the OTCs (χ2 = 5.16, df = 6, P = 0.023); (b) There was a significant interaction between year and treatment (χ2 = 10.62, df = 1, P = 0.005). There were significantly fewer occurrences of temperatures between 25 and 40 °C in OTCs in 2013 than in controls (P = 0.050), and there were significant differences within treatments among years...... 190

Figure 3.13. Comparison of mean daily temperatures in OTCs (white bars, n = 10) versus control plots (shaded bars, n = 5) at Wilson Brook from 1 June to 9 November 2011 to 2013, by year (mean ± first and third quartile). There was a significant interaction between year and treatment (χ2 = 7.58, df = 2, P < 0.023). There was no significant difference between treatments (P > 0.05), but there was a significant difference within treatments among years, Appendix 3.7a)...... 191

Figure 3.14. Comparison of differences in VPD in OTCs (white bars, n = 10) versus control plots (shaded bars, n = 5) at Wilson Brook from 1 June to 9 November 2011 to 2013, by year (mean ± first and third quartile). There was a significant interaction between year and treatment (χ2 = 58.62, df = 2, P < 0.001). There was no significant difference between treatments (P > 0.05), Appendix 3.7c), but there were significant differences within treatments among years...... 192

Figure 3.15. Change in counts (centered within species) of (a) Anemone parviflora plants and (b) Dryas integrifolia ramets by treatment (OTCs n = 10 plots, controls n = 10 plots) over three years (2012-2014) at Wilson Brook, NB. No significant differences were detected in changes between years or between treatments. Normality of residuals was visually inspected using qqnorm (R Core Team 2015)...... 193

Figure 4.1. Categories of values potentially influencing the setting of conservation priorities for species at risk...... 239

Figure 4.2. Modified species assessment process as used by COSEWIC, following IUCN methodology (COSEWIC 2015a). See COSEWIC (2015a) for detailed descriptions of each category...... 240

Figure 4.3. Proposed process to assess the conservation significance of rear-edge populations of common, wide-ranging species. While the ideal criteria for biological value and hence conservation significance would be genetic characteristics, both historical and ecological distinctness may indirectly indicate genetic distinctness...... 241

xxi

List of symbols, nomenclature and abbreviations

%P Percent polymorphic loci AFLP Amplified Fragment Length Polymorphism AMOVA Analysis of molecular variance ANOVA Analysis of variance ACPF Atlantic coastal plains flora bp Base pairs COSEWIC Committee on the Status of Endangered Wildlife in Canada cpDNA Chloroplast deoxyribonucleic acid D Gene diversity df Degrees of freedom DNA Deoxyribonucleic acid DNR Department of Natural Resources GDD Growing degree days GLM General linear model HE Expected heterozygosity I Shannon’s information index IPCC International Panel on Climate Change ISSR Inter-simple sequence repeat ITEX International Tundra Experiment IUCN International Union for the Conservation of Nature K Number of genetic clusters ka Kilo annum kPa Kilo Pascal LGM Last glacial maximum MCMC Markov Chain Monte Carlo N-fixing Nitrogen-fixing NB New Brunswick Ne Effective number of alleles NHS National Historic Site NL Newfoundland NP National Park OTC Open top chambers PB Total number of private bands Pb Allelic richness PCoA Principal Co-ordinate Analysis

xxii

PHI'PT,Φ'PT Modified version of Wright’s F; refers to the relative contributions of between-population separation to the overall genetic variation in the whole sample PLP 1% Percent polymorphic loci at 1% PNA Protected Natural Area PVP Polyvinylpyrrolidone QC Quebec RAPD Random amplification of polymorphic DNA RCP Representative concentration pathway rfu Relative fluorescence units RH Relative humidity RNase Ribonuclease S1 NatureServe Subnational Conservation Status Rank (SRank), Critically Imperilled S2 NatureServe Subnational Conservation Status Rank (SRank), Imperiled S4 NatureServe Subnational Conservation Status Rank (SRank), Apparently Secure SARA Species at Risk Act (Canada) SD Standard deviation SE Standard error SNP Single-nucleotide polymorphism SSR Simple sequence repeat VPD Vapour pressure deficit

xxiii

Chapter 1 - Introduction

Climate change is expected to increase temperatures in Canada by 1.1 to 5.6°C by 2100 and to increase variability in weather (Government of Canada 2016). In some species, range shifts (range expansions and contractions), local extinctions, and changes in phenology and abundance have already been attributed to climate change (Parmesan &

Yohe 2003, Thomas et al. 2004, Parmesan 2006, 2007, Cahill et al. 2012, Provan &

Maggs 2012, Peñuelas et al. 2013, Field et al. 2014). Populations that are currently at the periphery of a species' range (range edge), and are presumably at their niche limits, will experience stress because they occur beyond their optimal growing conditions, and they might be at higher risk of extinction or extirpation under predicted future climate change

(Lesica & McCune 2004, Thomas et al. 2004, Crawford 2008, Thomas et al. 2008,

Hardie & Hutchings 2010, Pauls et al. 2013).

Predictions of expected biodiversity loss due to climate change are in the range of 1% to over 50% of species, depending on the modeling approach (Field et al. 2014), and are expected to accelerate with warming. Climate change may put additional pressures on rare species that are already under threat for other reasons, but it may also affect common and widespread species. For example, some alien, as well as native species, are already benefitting from warmer conditions (Jia et al. 2016). Some species may be able to respond to the expected changes to some extent (e.g., through migration or adaptation), but if populations are lost, genetic diversity may decline. The loss of populations that harbor unique genetic diversity may have implications for the potential 1

of local populations, and the species, to respond to climate change (Bunnell et al. 2004,

Pfeifer et al. 2010, Hampe & Jump 2011). Given that isolated and / or peripheral populations are expected to harbor unique diversity, they may be particularly important from a conservation perspective (Lesica & Allendorf 1995, Channell 2004, Hampe &

Jump 2011).

In the context of range limit theory on population genetic patterns across ranges, and focusing on rear range-edge populations, I ask whether (1) rear-edge populations of widespread arctic-alpine species in Eastern Canada are genetically distinct from core populations; (2) distance and isolation factors of three species are associated with genetic variability (within and among populations); (3) disjunct populations are likely to be affected by a warming climate; and how (4) conservation significance of rear-edge disjunct populations can be determined based on a species and a population-based assessment process.

1.2 Geographic distribution patterns and range limits

In their simplest form, species' ranges, which form the unit of study in biogeography

(Crawford 2008), are expected to represent the geographic expression of species' ecological tolerances and / or requirements for abiotic (environmental) and biotic factors. Leading edges are characterized by younger populations, which are likely due to recent colonization or expansion events (Fig. 1.1). Populations at leading edges may be critically important for conserving processes that allow species to survive climate change (e.g., by establishing new populations in previously inhospitable areas, and thus permitting range shifts at the expanding range limit; Gibson et al. 2009). Species' populations at rear edges (i.e., those defined as being located at retreating or stable range edges) may contain unique genotypes that can survive and acclimate to changing conditions, and which are not present elsewhere within the range (Channell 2004,

Hampe & Jump 2011). If some rear-edge populations have persisted as climate relicts

(populations that have become isolated due to climate-induced range shifts), or as refugial populations (populations that survived glaciations) for exceptionally long periods of time, these populations may be important for preserving ecological and / or evolutionary histories (Woolbright et al. 2014), and may contribute disproportionally to a species' overall diversity (Channell 2004, Hampe & Petit 2005). They may thus be important for the conservation of a species, especially if local adaptation has occurred

(Lesica & Allendorf 1995, Hampe & Petit 2005). At the range center, within population genetic diversity is expected to be high according to theory, with considerable gene

flow, thereby limiting differentiation among populations. Climate warming could result in shifts at the leading edge (northwards in the northern hemisphere), while increased stress due to temperatures exceeding the tolerable range for the species, or increasing competition, could cause range contractions for some species, including local population extirpations, at the rear edge (Hampe & Petit 2005, Woolbright et al. 2014, Sexton et al.

2016).

1.2.1 Geographic range patterns

There has been considerable research about patterns of demography, fitness and genetic variability across species' geographical ranges (Gapare & Aitken 2005, Sagarin et al.

2006, Samis & Eckert 2007, Eckert et al. 2008, Yakimowski & Eckert 2008, Duffy et al.

2009). The literature provides expectations of a species' range edge such as decreased abundance (Sagarin et al. 2006), decreased fitness (Angert 2006), decreased genetic diversity within (Csergö et al. 2009, Hampe & Jump 2011) and increased genetic differentiation among populations (Bruederle 1999, Hampe & Jump 2011). Existing evidence suggests that patterns across species' ranges are more complex than theoretical models suggest. For example, two Viola species (V. pumila and V. stagina) that occupy floodplain habitat in Europe at the edge of their range showed higher genetic variability in the center of their range as expected, yet in a third related species, V. elatior, there was no difference in genetic variability between central and peripheral populations

(Eckstein et al. 2006). Similar contradictions between theoretically predicted versus observed diversity patterns have been reported for other species. For example, levels of

genetic diversity were not related to populations size (as estimated by census of flowering tussocks) in Stipa pennata (Wagner et al. 2012), and similar levels of genetic diversity compared to the range center were observed in some peripheral populations of

Zostera marina (Diekmann & Serrao 2012). This ambiguity may be attributable to historic factors (i.e., phylogeography, such as glaciation history) and contemporary factors (e.g., population size, density, distance from range center, and habitat fragmentation), which are often ignored in such studies (Eckert et al. 2008). A recent review of the “central-peripheral hypothesis” by Pironon et al. (2016) confirmed that reduced genetic diversity and increased genetic differentiation were demonstrated in <

50% of 248 studies analyzed.

1.2.2 Determinants of range limits

Species' range limits are set by abiotic factors such as light, temperature, water, substrate chemistry, disturbance, climate, topography, and biotic factors including competition, predation, disease and parasitism. Temperature is one of the most important abiotic factors and is an essential variable in plant growth and development (Lambers et al.

2008). For example, metabolic rates of plants are directly affected by temperature through its effect on enzyme function (Mackenzie et al. 2001, Źróbek-Sokolnik 2012).

Abiotic and biotic factors may also interact and therefore play divergent roles at different range edges (Schickhoff 2011). Species are typically adapted to the range of environmental conditions that naturally occur within their habitats (Grime 1979). A species' physiological range of tolerance determines the upper and lower limits of 5

environmental conditions within which members of the species can survive, and beyond which growth and reproduction will not occur. Growth and reproduction are greatest under a narrow range of conditions that are optimal for a species (Mackenzie et al.

2001). The geographic range across which these optimal conditions occur is expected to correspond with the center of the species' distribution (Kormondy 1996), and the area of greatest abundance. In areas beyond this optimum range, populations are predicted to be smaller and individuals more physiologically stressed (Cox & Moore 2005, Pironon et al. 2015).

The relative importance of individual biotic and abiotic factors in setting species' range limits is widely discussed in the literature, as is the contribution of each in determining the response of species to change (Grace 1987, Berg et al. 2010). In general, abiotic factors are thought to be limiting at high latitudes and elevations (i.e., in environmentally stressful areas), and species interactions are thought to be more important in defining range edges at lower elevations and latitudes (MacArthur 1972,

Louthan et al. 2015). For example, temperature is severely limiting at the northern range limit of arctic species (Doak & Morris 2010), whereas competitively superior species

(biotic interactions) are considered more important in determining southern range edges

(MacArthur 1972). However, in a recent review of 125 published empirical studies of

178 species addressing the causes of warm-edge range limits, Cahill et al. (2014) showed that data support the association of abiotic factors with range limits more often than biotic ones, with temperature as one of the primary factors. While temperature may

play a greater role at warm-edge range limits than previously thought, other factors, such as biotic interactions, may modulate species' responses to temperature (Pellissier et al.

2013). As environmental conditions change, species may be more likely to experience the indirect effects of change. For example, climate warming was associated with habitat changes and reduced occurrence of host plants for butterflies in Britain, and thus indirectly led to the local extirpation of some butterfly populations (Franco et al. 2006).

1.2.2.1 Fitness and dispersal

Reproduction, dispersal ability, and resulting population dynamics (fluctuation of abundance caused by births, deaths, immigration and emigration) also determine range edges. Lack of genetic diversity required for a species' life history traits to evolve, or negative correlations between required traits, such as mating system, reproductive rate

(births) and dispersal (emigration, immigration), may constrain range expansion. Species are expected to be less fit in locations where conditions fall outside their optimal tolerance, with performance generally declining towards and beyond the range limit

(Hargreaves et al. 2014). Birth rates, such as seed production, are expected to decline towards the range limit, and a shift is expected from sexual reproduction to clonal reproduction, limiting the dispersal capacity, or emigration rates, of populations (both seeds and pollen) (Beatty et al. 2008, Geber 2008, Millar et al. 2010). Pollinator limitations are also linked to range edges and can reduce reproductive fitness (Herlihy &

Eckert 2005). Reduced fitness can limit adaptive capacity by limiting the number of viable offspring, and could thus limit population expansion (Sexton et al. 2009).

Dispersal capacity is an important determinant of species' range edge dynamics (Gaston

2009, Sexton et al. 2016). In a study using elevational and latitudinal gradients and distributions of 155 ruderal species, Halbritter et al. (2013) found that while temperature variability was similar across elevational and latitudinal gradients, species' distributions along these gradients were not; they suggested that average dispersal distance likely determined the extent of a species' distribution, and not temperature. Dispersal can be limited by inherent life history traits such as mating system (e.g., apomixis), dispersal mechanisms of pollen and seeds, physical barriers, and ecological differences driven by local adaptation of the population of a propagule's origin vs the receiving population.

Factors that significantly restrict dispersal are expected to reduce gene flow and hence increase isolation among populations (Westberg & Kadereit 2009, Orsini et al. 2013,

Abeli et al. 2014, Sexton et al. 2014).

1.2.2.2 Genetics and range edges

Genetic variability and gene flow at range edges can contribute to population persistence, and genetic variability can facilitate range expansion and migration. For example, high genetic diversity would suggest a greater pool of available genotypes that could populate new habitats at leading range edges, or a high likelihood of genotypes able to adapt to changes at the rear range edge (Kirk & Freeland 2011). Gene flow along an environmental gradient over which selection varies, and from large populations to small populations, may limit adaptation to local conditions. In those cases, selection can lead to the presence of alleles that may move between populations, but may not be

optimal in all habitats, thus potentially reducing population persistence (Gaston 2009,

Sexton et al. 2009).

While adaptation to local conditions can ensure persistence of a population, it can also reduce gene flow among populations and result in further isolation if immigrants are less successful in getting established in the population (Orsini et al. 2013). Physical isolation, and reproductive isolation as a result of divergent selection, can eventually allow for new species to evolve (ecological speciation; Schluter & Conte 2009). High genetic diversity at loci under selection may indicate local adaptation, but may also limit the capacity for range expansion into novel habitats beyond the range edge. Local adaptation can occur at range edges (Lesica & Allendorf 1995, Diekmann & Serrao 2012) contributing to unique genetic variation (Pfeifer et al. 2010, Provan & Maggs 2012). In contrast, genetic drift and lack of gene flow can results in limited genetic diversity (both neutral and under selection) (Gaston 2009).

1.2.2.3 Historical factors

Post-glacial re-colonization can leave important signatures of historical demography and expansion across species' ranges (Ciotir et al. 2013, Abeli et al. 2014). For example, long-lived species may persist in habitats outside their contemporary main range, and isolated populations may provide evidence of historic range extent. Aside from effects on a species' current distribution, historical colonization patterns may also be revealed in population genetic structure. This might be especially true in areas where glacial relicts have persisted over time in locations isolated from the center of the species' current 9

range (Alsos et al. 2009), or where refugial populations (those that persisted in isolated locations through times of unfavorable conditions) have strongly shaped post glacial distribution and population genetic structure of the species (Shafer et al. 2010). Such refugial and relict populations are often considered of conservation importance, as they may be genetically distinct (Lesica & Allendorf 1995, Leppig & White 2006, Reisch

2008).

1.3 Geographic distributions and climate change

Temperatures have risen globally by an average of 0.13 °C (± 0.03 °C) per decade since

1950 (IPCC 2007), and models of climate change for Canada predict average annual temperatures to increase by 1.1 °C (Representative Concentration Pathway, or RCP2.6) to 5.6 °C (RCP8.5) (Government of Canada 2016) by 2100. Temperature increases will be associated with changes in precipitation and increases in climatic variability. In southeastern New Brunswick, temperatures are predicted to increase by 2.82 (RCP4.5) to 5.22 °C (RCP8.5) by 2080 (Roy & Huard 2016), leading to a longer growing season and increased moisture stress. Precipitation is expected to increase in winter and spring, and increased heat events are expected in the summer. Given that forecasted changes in climate will be faster than those the world has experienced in the past 10,000 years

(Barrow et al. 2004, Jump & Peñuelas 2005), there is concern that many species may be extirpated from parts of their current range or become extinct (Box et al. 1999, Aitken et al. 2008).

Under current models of predicted climate change, species' range limits are expected to shift upward (altitude) and poleward (latitude) (Aitken et al. 2008, Field et al. 2014,

Settele et al. 2014). Range expansions and contractions have already been documented for numerous species. Phenological shifts have been documented in 62% of monitored birds, butterflies and alpine herbs (n = 677), and distribution and abundance shifts in

49% (n = 893) (Parmesan & Yohe 2003); 84% of northwards range shifts of 434 species were consistent with climate change predictions. In Britain, Mason et al. (2015)

documented the range shifts of 1573 species from 21 animal groups and showed a mean rate of northwards movement of the range margin of 23.2 km for the interval from 1966-

1995, and 18 km for the interval from 1986-2010. Climate change is also expected to affect the distribution of genetic variants through extirpation events, and as a result, the evolutionary potential (adaptation and speciation) available in a species, making certain species more vulnerable (Pauls et al. 2013).

Species respond to environmental stress by either avoidance or tolerance (Grime 1979).

In the context of climate change, species may respond to change by acclimating to changing conditions, by changing behavior or phenotype, by shifting their ranges, and / or by adaptation in the longer term (evolutionary adaptation) (Angert et al. 2011). These responses correspond with Bellard et al. (2012), who characterize species' possible responses to climate change along three axes: time, space and self. Species can respond through changes in phenology, such as changes in the timing of flowering or fruiting.

They may shift geographically, such as already observed poleward (latitudinal) and upward (elevational) range shifts. Lastly, there might be phenotypic and / or genetic changes in their physiological tolerance, allowing for adaption to novel environmental conditions. Species may avoid stress temporally by becoming dormant (e.g., hibernating animals or seed banks), or by migrating to avoid environmental conditions occurring outside the range of their active physiology. Others may avoid stress by water storage in their tissues, or by using hairs to reduce transpiration. Species that are stress tolerant use physiological, developmental or molecular adaptations. Such adaptations may include

enzymes that are active within a wide temperature range, anti-nucleation in cells to avoid freezing, or transcriptional regulation of stress pathways. In some species, tolerance is a complex response that involves a variety of pathways and phenotypic alterations. For example, some species are capable of acclimation, a gradual morphological (e.g., deeper root growth in dry conditions) and physiological adjustment by individual plants in response to an environmental stressor (Lambers et al. 2008).

A species' or population's likelihood to persist under climate change is the result of its adaptive capacity. Adaptive capacity is defined as the ability “of a species or constituent populations to cope with climate change by persisting in situ, by shifting to more suitable local microhabitats, or by migrating to more suitable regions” (Dawson et al.

2011). Adaptive capacity includes the ability to respond to change both genetically

(evolutionary potential) and phenotypically (Williams et al. 2008, Eizaguirre &

Baltazar–Soares 2014). Greater genetic diversity within the species allows for greater genotypic variation and increases the likelihood of the presence of tolerant genotypes that will persist in the face of environmental change. Acclimation may be insufficient to enable persistence when conditions change beyond the range of tolerance, e.g., when temperatures become lethal, or if inhospitable conditions extend beyond the time that dormant seeds remain viable. With changes that are sufficiently large and / or sudden, neither acclimation nor adaptation may be possible.

1.4 Populations at range edges

Studies evaluating the impacts of climate change on species' populations and range edges have tended to focus on leading (typically northern latitudinal or upper altitudinal) range edges, where changes have been relatively easy to observe and where, in some cases, extant genotypes can colonize new habitat quickly should conditions change

(Gibson et al. 2009). Studying range expansions or contractions at these limits can help predict community compositional changes due to climate change (Crawford 2008,

Gibson et al. 2009). Factors influencing range expansion at the leading edge can be complex. For example, Brown & Vellend (2014) found that the altitudinal range expansion of Sugar Maple (Acer saccharum) was not limited by temperature as expected, but by seed predation and possibly soil-borne pathogens.

Little is known about range contractions or extirpations at rear range edges, although declines have been observed in rear-edge populations of 28 arctic-alpine species in

Montana, USA, over two decades (Lesica & Crone 2016). It is unclear why only limited contractions and extirpations have been observed, given predictions by theory. There are several possible reasons why the primary literature may provide few examples of distribution changes at rear edges: (1) there may be limited sampling at rear edges resulting in a lower probability of observing changes in those range-edge populations

(Mátyás et al. 2009), (2) there is an extinction lag (e.g., Dawson et al. 2011), because they may occupy habitats that are not suitable for other species, or (3) environmental conditions may not yet have reached critical thresholds for these species or populations

(but see Peñuelas et al. (2013) and Franco et al. (2006)). This knowledge gap is of concern because rear-edge populations are not only expected to be at risk from extirpation and rapid climate change, but may also contribute significantly to the adaptive capacity of species at the rear edge of their range.

Range-edge populations are predicted to display ecological and genetic divergence relative to central populations due to reduced gene flow as a result of isolation (Hardie &

Hutchings 2010, Orsini et al. 2013, Shafer & Wolf 2013), limits to dispersal causing isolation by distance (Orsini et al. 2013, Shafer & Wolf 2013), or local adaptation

(Orsini et al. 2013). Therefore, they may show lower within-population genetic diversity, but also tend to be genetically distinct. Historical range expansions and contractions can affect genetic variation at the range edge (Yakimowski & Eckert 2008).

Persistence of some of these rear-edge, often disjunct, isolated populations, possibly over centuries in some cases (Nantel & Gagnon 1999), indicates that they have been able to tolerate, and may be adapted to, local conditions (Millar et al. 2010).

In spite of their potential repositories of genetic resources (Nantel & Gagnon 1999), the question of whether to conserve range-edge populations is challenging, especially for wide-ranging, common species, and those with ranges spreading across multiple jurisdictions that may have different conservation priorities. With limited conservation resources, identifying those species and populations that are of conservation significance is essential. These challenges are apparent in all Canadian provinces, and British

Columbia and Nova Scotia provide two case studies. British Columbia has more than 15

1300 range-edge species from a variety of taxonomic groups; 900 of these were listed as being of conservation priority in 2004 (Bunnell et al. 2004). In Nova Scotia, four members of the Atlantic Coastal Plain Flora, a disjunct assemblage of species more common further south, are legally protected, yet the species are classified as globally secure (DNR 2015, NatureServe 2016). In Canada, it has been reported that 90% of species listed as endangered or threatened are found at the leading edge of their ranges

(Yakimowski & Eckert 2007). Political boundaries have resulted in the listing of globally widespread and secure species as threatened and endangered in particular jurisdictions. Bunnell et al. (2004) argue that conservation should be focused on relict populations, and populations that may be genetically distinct and may contribute significantly to the future adaptive potential of a species. However, the effective management of peripheral populations in the long term is challenging, especially since data are lacking for most species (Lesica & Allendorf 1995, Nantel & Gagnon 1999,

Eckstein et al. 2006, Leppig & White 2006, Yakimowski & Eckert 2007, Eckert et al.

2008). Assessing range-edge populations in the context of climate change is important, given that these populations may be significant in aiding species to expand their ranges poleward, or because they preserve considerable evolutionary potential (Bunnell et al.

2004, Channell 2004).

In the context of exploring the potential conservation significance of rear-edge populations, the objectives of this research were to evaluate whether rear-edge populations of three wide-ranging arctic-alpine species in Eastern Canada are genetically

distinct and thus possibly of conservation significance, and if they are likely to be at risk from warming. I address these questions by (1) evaluating the genetic variability within and among populations of Dryas integrifolia, Anemone parviflora, and Vaccinium uliginosum from central to rear-edge to disjunct populations, (2) assessing the association of genetic structure at each species' rear range edge with indicators of isolation and disturbance, and (3) experimentally evaluating if warming due to climate change may place two disjunct, rear-edge populations of D. integrifolia and A. parviflora at risk. The thesis includes two data chapters, each providing an introduction to the topic, research objectives and predictions, methodology, data analysis, results and a discussion. Chapter four discusses the challenges associated with assessing the conservation significance of peripheral populations, assesses rear-edge populations of the three species using a risk assessment process, and proposes a population-based framework to better evaluate those populations. The last chapter provides a synthesis of my findings, and suggestions for future research.

1.5 Study species and locations

1.5.1 Arctic-alpine species

Arctic-alpine species are wide-ranging across the northern hemisphere, with typically clear range limits to the south (rear edge) and north (leading edge). Some species extend their ranges far into temperate regions, where they may occur in small, disjunct populations in unique microhabitats that may experience conditions analogous to those in the Arctic, often at higher altitudes. These species may be excellent models for studying range limits at both range edges (leading and rear), especially on a latitudinal gradient. Arctic-alpine species are adapted to cold environments and as such are expected to be sensitive to climate warming (Arft et al. 1999, Doak & Morris 2010).

This research focuses on three species: Dryas integrifolia (White Mountain Avens),

Anemone parviflora (Northern White Anemone) and Vaccinium uliginosum (Northern

Bilberry). These three species were selected according to the following criteria: (i) they follow similar distribution patterns, as they are common in the sub-Arctic and Arctic, and rare in temperate New Brunswick, and (ii) disjunct populations are accessible within the studied region.

1.5.1.1 White Mountain Avens

White Mountain Avens (Dryas integrifolia Vahl., Family Rosaceae, Fig. 1.2) is a long- lived, clonal, mat-forming evergreen dwarf shrub up to 17 cm tall. Plants in the genus

Dryas are self-compatible, insect-pollinated, and produce wind-dispersed, elongated and

plumose achenes (2n (diploid) = 2x = 18 chromosomes). Restricted to the northern hemisphere (North America and Greenland), the species inhabits soil with a low organic content on wet meadows, river terraces, tundra slopes and ridges in the northern portion of its range, and alpine habitats in the Rocky Mountains and the Appalachians at the southern extent of its continuous range (in the Gaspé mountains of Quebec). The species is considered a calciphilic pioneer plant and is a poor competitor (Aiken et al. 2007). A close relative, Dryas drummondii, has been shown to form a symbiotic relationship with the nitrogen-fixing bacteria Frankia (Gardes & Dahlberg 1996), but this relationship has not been shown for D. integrifolia (Markham 2009). In Atlantic Canada, the southernmost extent of its distribution, D. integrifolia occurs at Wilson Brook, New

Brunswick. Here, the single population is disjunct from the nearest population by at least

400 km. The species is listed as globally and nationally secure, as S1 (5 or fewer occurrences or very few remaining individuals) in New Brunswick, and S4 (apparently secure) in Quebec and Newfoundland and Labrador (NatureServe 2016).

1.5.1.2 Northern White Anemone

Northern White Anemone (Anemone parviflora Michx., Family Ranunculaceae, Fig.

1.3) (2n = 2x = 16) is a perennial, clonal herb growing to 35 cm tall with rosettes that die back annually (John & Turkington 1997, Aiken et al. 2007). Achenes are dry, wind- dispersed, and densely woolly (CYSIP 2016). Across its range, the species seems to prefer calcareous soils and grows along streams, in meadows and on rocky slopes

(FNAA 2008). While considerable work has been done on the taxonomic status of

members of the Anemone genus, no genetic studies have assessed A. parviflora phylogeography, or genetic variation across its range. The species is listed as globally and nationally secure in Canada, as S1 in New Brunswick, S4 in Quebec, S5 (secure) in

Newfoundland and S3S4 (80 or fewer occurrences, vulnerable) in Labrador

(NatureServe 2016). In the study region, A. parviflora occurs along the Limestone

Barrens and in the Tablelands in Newfoundland, the Gaspé Peninsula of Quebec, and from a number of locations in New Brunswick (Restigouche region, Wilson Brook), and in Nova Scotia (Cape Breton Highlands National Park (ACCDC 2017)).

1.5.1.3 Northern Bilberry

Northern Bilberry (Vaccinium uliginosum L., Family Ericaceae, Fig. 1.4) has variable ploidy levels across its range (2n = 4x = 48 for the sampled region as identified by

Eidesen et al. (2007)). It is a circumboreal dwarf shrub growing to 20 cm tall. Individual shoot ages of 93 years (Alsos et al. 2002) and patch ages of over 1,800 years have been observed with the age of the oldest genet sampled at 1,880 years old (De Witte et al.

2012). Longevity in plants means that the effects of genetic drift are felt more slowly than in annual species, and genetic divergence is slowed (Loveless & Hamrick 1984). In the Arctic, vegetative reproduction of stems is more important than reproduction by seed

(Jacquemart 1996). The species displays self- and cross-pollination by insects, and is an ornithochore (seeds are dispersed by birds). Considerable research has been undertaken on the species' population genetics and phylogeography throughout its range (Alsos et al. 2002, Alsos et al. 2005, Eidesen et al. 2007, De Witte et al. 2012, Eidesen et al.

2013), as well as of morphological and life history traits (Jacquemart & Thompson 1996,

Alsos et al. 2003, De Witte et al. 2012, Mayer et al. 2012). The species grows primarily in heathlands and tundra on acidic, but also on calcareous soils. The species is listed as globally and nationally secure, as S1 in New Brunswick, S5 in Quebec and

Newfoundland and Labrador (NatureServe 2016). In the study region, V. uliginosum occurs frequently in exposed areas in Newfoundland (along the Limestone Barrens, the

Tablelands, and the Long Range Mountains), in the uplands of the Gaspé Peninsula,

Quebec, as individual clones on Mt. Carleton and Mt. Denis in New Brunswick, and in the Cape Breton Highlands, Nova Scotia. It is known to occur further south in the

Appalachian mountain range. Some historical locations are also known in New

Brunswick, PEI and Nova Scotia (ACCDC 2017), but have not been relocated recently.

1.5.2 Study locations

A number of sampling sites were used in this research, and all of these sites are described below. In summary, analyses in Chapter 2 used all sampled populations for all three species (Chapter 2, Fig. 2.1), while one site (Wilson Brook in New Brunswick) was the focus of Chapter 3 (Chapter 3, Fig. 3.1).

1.5.2.1 Central populations

Two sites (Torngat and Nain, Labrador) were sampled in arctic habitat within a region supporting an abundance of populations of each sampled species. This region represents the only core range sites for which sample collection was possible. Dryas integrifolia

was collected at Torngat, and V. uliginosum at Torngat and Nain.

1.5.2.2 Range-edge populations

Five sites occurred within a continuous 200-km stretch of coastline within the specialized Limestone Barrens habitat in Newfoundland (Burnt Cape, Cooks Harbour,

Sandy Cove, Port-au-Choix and Bellburns). The Limestone Barrens are a coastal, open landscape that supports 114 of Newfoundland's 271 rare plant species (Robinson &

Hermanutz 2015).

1.5.2.3 Disjunct populations

Disjunct populations were defined as being ≥ 100 km distant from the nearest population, and / or based on inferences of seed and pollen dispersal barriers. Gene flow among populations is presumed to be limited due to the heterogeneity of the landscape.

100 km as the threshold was chosen as a distance that may prevent gene flow among populations even in an open landscape.

One site at Tablelands, Newfoundland was designated as disjunct for D. integrifolia and

A. parviflora, but as continuous range edge for V. uliginosum. This site was located in the unique travertine seeps of an area characterized by minerals normally found only in the earth's mantle, and toxic to plants, in an open landscape of an upland valley, 100 km beyond the nearest continuous range-edge population that was sampled (Bellburns).

Other populations to the south are > 50 km distant, and one population of D. integrifolia is known from Bonne Bay (Big Hill) (ACCDC 2017), but I inferred the Tablelands

population to be isolated because of the landscape. The Tablelands population of D. integrifolia occurs on a patch of habitat approximately 150 m long by 50 m wide, whereas A. parviflora occurs in a more extensive habitat along a wet area at the bottom of the slope, approximately 200 m long. Vaccinium uliginosum is more common than the other two species throughout and adjacent to this site.

Mount Logan (sensu lato) on the Gaspé Peninsula within the Parc National de la

Gaspésie (Quebec) was characterized as a disjunct population for D. integrifolia. It is found in an alpine, open habitat of very limited extent that was, until recently, thought to be the lone extant population of D. integrifolia within the region. Vaccinium uliginosum is broadly distributed above the tree line at Mount Logan, while D. integrifolia occurs along a calcareous ridge approximately 20 m wide and 130 m long. A second population of D. integrifolia was recently discovered at Mont de la Table (> 1000 genets in approximately 5 ha) approximately 50 km away (S. Bailey, pers. comm., 2015). Other populations may exist at Mont St.-Anne (> 130 km away), Mont St. Pierre (Rune 1954), near Percé (> 150 km away), in Anticosti Island and in Mingan (> 200 km away)

(GBIF.org 2016). Anemone parviflora was not sampled here due to inaccessibility.

In New Brunswick, two sites were sampled. One is a population of A. parviflora along the Restigouche River in northern New Brunswick, where the species grows along a one km long calcareous ledge of the river along a stretch of approximately 25 km where the species occurs (ACCDC 2017). The second is in the Wilson Brook Protected Natural

Area (Wilson Brook) in Albert County, which harbors four arctic-alpine species on a 23

gypsum cliff site (Clayden 1994). Two of those species (including D. integrifolia) occur nowhere else in the province and have few or no populations further south.

Wilson Brook was classified as an ecological reserve in 1997 (M. Bourgeois, pers. comm.), and established in 2003 as a provincial Protected Natural Area (DNR-NB

2015). The gypsum cliffs along Wilson Brook were mentioned in geological reports in the 19th century (Hind 1865). Dryas integrifolia was discovered at the site by Gorham

(1944), and Roberts (1965) reported three arctic alpine species for the site (Anemone parviflora, Salix myrtillifolia, and Solidago multiradiata). The arctic species at the site include White Mountain Avens (Dryas integrifolia), Mountain Goldenrod (Solidago multiradiata), Myrtle-leaved Willow (Salix myrtillifolia) and Northern Anemone

(Anemone parviflora) (Clayden 1994). These arctic-alpine relicts are classified as S1 or

S2 (6 to 20 occurrences or few remaining individuals) (NatureServe 2016) in New

Brunswick, recognizing their rarity on a provincial scale. The habitat is a crumbling cliff of white gypsum, < 1 km long, extending < 50 m upwards, with a north-facing slope of

45-75°, with some level areas (Chapter 3, Fig. 3.2). The plateau at the top of the cliff is approximately 90 m above sea level. The cliff face is rapidly eroding due to seepage water in the summer and a brook (Wilson Brook) at its base. A cool, moist microclimate prevails at the site (Brown 1983). A narrow band at the lower portion of the slope is dominated by D. integrifolia, with other shrubby species (such as Sheperdia canadensis) from the bottom edge to the forested margin. Trees surround the site on all sides; the location is shaded for much of the day. Anemone parviflora and D. integrifolia occur in

an area approximately 170 m in length, and < 20 m wide. Both species are isolated from the nearest populations by ≥ 300 km. Estimated population size for D. integrifolia is >

50,000 ramets (individuals are difficult to distinguish due to the species' clonal growth form), for A. parviflora approximately 1,000 plants.

1.6 Structure of thesis

1.6.1 Chapter 2, “Genetic diversity and differentiation of rear-edge populations of three arctic-alpine species in Eastern Canada”.

To address whether genetic diversity decreases and differentiation increases at rear range edges, neutral genetic diversity and pairwise genetic differentiation from the center to the rear range edge were assessed for each species. To allow for a generalization of patterns of genetic variability in disjunct, rear-edge populations, the association of indicators of isolation with within-population and among-population genetic diversity for range-edge and disjunct populations was assessed. The aim of this research is to evaluate the conservation significance of disjunct, rear-edge populations of widespread arctic-alpine species in Eastern Canada.

Most studies of range peripheries have focused on leading range edges and have not incorporated historical information on species occurrences during and after the Last

Glacial Maximum. In this chapter, I document genetic diversity and differentiation from near-center to disjunct rear-edge populations of three widespread arctic-alpine species,

Anemone parviflora, Dryas integrifolia and Vaccinium uliginosum in Atlantic Canada. I assess if isolation and disturbance levels of populations are associated with genetic diversity and differentiation, and place my findings into the context of historical processes, life history traits, and landscape configuration to explore how these factors may have influenced the observed genetic patterns of the three species at the range edge.

I used Amplified Fragment Length Polymorphisms (AFLP), a frequency-based, 26

dominant, molecular marker, (1) to evaluate and compare genetic diversity and differentiation across the southern range edge of the three species in Atlantic Canada, and (2) to explore the potential association among proximate ecological factors, such as geographic distance, disjunction and disturbance, that may impinge on genetic variation and the pattern of genetic structure among populations within and across species.

Based on the hypothesis that within population genetic diversity decreases and differentiation among populations increases from the range center to the periphery, I predicted that disjunct populations show lower genetic diversity than range-edge and core populations. Further, I predicted that disjunct populations of all three species show higher levels of mean pairwise genetic differentiation than range-edge and core populations. Based on the hypothesis that isolation increases genetic differentiation, I also predicted that measures of genetic variation are associated with factors affecting isolation among populations, including distance, disjunction, and disturbance, and that these associations would be shared across the three species.

1.6.2 Chapter 3, “Microclimate and demography in rear-edge populations of two arctic-alpine species (Dryas integrifolia and Anemone parviflora) at Wilson Brook

Protected Natural Area, New Brunswick”.

Range-edge populations are presumed to be more vulnerable to change due to their smaller size, isolation from other populations, and because they are often located in specialized habitats (Crawford 2008), often at the limits of their ecological tolerance

(Hampe & Jump 2011). This chapter focuses on temperature because: (1) it may be a 27

primary cause for range limits (Cahill et al. 2014); (2) it is predicted to continue increasing with climate change (Cox & Moore 2005); and (3) it is an essential factor for biological functions such as enzyme activity and membrane integrity, which can be impaired under excessive temperatures (Mackenzie et al. 2001). Climate change could place isolated, disjunct rear-edge populations at risk of extirpation (Hampe & Petit

2005), because rear-edge populations of arctic-alpine species may already be at their limit of tolerance for high temperatures. Assessing species' responses to increasing temperatures has been a long-standing approach by ecologists especially in the arctic, where warming experiments have been conducted for over 20 years (Marion et al. 1997,

Arft et al. 1999, Shaver et al. 2000).

In this chapter I document temperature regimes during three growing seasons and population demography of isolated populations of Dryas integrifolia and Anemone parviflora at Wilson Brook. The aims are (1) to test short-term responses of these species to temperature manipulation using Open Top Chambers (OTCs); and (2) to compare microclimatic conditions of this isolated site with regional temperatures and with an apparently similar site in Newfoundland, but where populations of the two species are more widespread and continuous. Assessing the microclimate for these populations as well as their response to experimental warming can identify potential tolerance limits to specific temperature variables and the level of phenotypic plasticity of the populations. Open Top Chambers (OTCs) are important and effective tools in

assessing the impact of increased temperature (and other variables such as CO2 and

irradiation) on plant responses above and below ground (Aronson & McNulty 2009,

Bokhorst et al. 2011, Elmendorf et al. 2012). OTCs have been used extensively to alter microclimatic conditions in a controlled manner for both short-term and long-term warming experiments (Semenova et al. 2015), to evaluate the impacts of climate warming on arctic flora (Marion et al. 1997, Hollister & Webber 2000, Gugerli & Bauert

2001, Xu et al. 2009).

Based on the hypothesis that the local microclimate supports arctic-alpine species at

Wilson Brook and that warming will negatively affect population growth of two species,

I monitored temperature within and outside of OTCs over three years to address the following research objectives: (1) assess whether the microclimate at Wilson Brook is locally and regionally distinct as might be expected if it represents a climatic refugium; and (2) experimentally evaluate the association between variation in demography and variation in microclimate over three years in two arctic-alpine species populations occurring at Wilson Brook.

I predicted that Wilson Brook’s microclimate would not differ from that found at a more northern site in Newfoundland, that it would be cooler than expected compared to the regional climate in southern New Brunswick, and that abundance of the two species in the warmed plots (OTCs) would be negatively impacted by increased temperatures.

1.6.3 Chapter 4, “Assessing the conservation significance of peripheral populations

– a case study of three arctic-alpine species in Eastern Canada”.

Conservation management decisions are largely based on prioritizing those species that are most at risk (Pullin et al. 2013). This method does not work well for evaluating the conservation significance of rear range-edge populations of widespread species.

Populations of widespread species may only be considered if the species is already rare in a jurisdiction. However, rear-edge populations may have significant ecological and evolutionary potential, and also be at greater risk of extirpation than those in the core. In addition, when species are assessed based on political boundaries, a determination of conservation significance may be based on local or regional rarity, and not necessarily on the contribution of the population to a species' overall biodiversity. In this chapter, I discuss the challenges associated with determining the conservation significance of rear range-edge populations of widespread, common species, and use three arctic-alpine species (Anemone parviflora, Dryas integrifolia, Vaccinium uliginosum) as a case study to show (1) how a general risk assessment process is not useful for peripheral populations of common, widespread species; and (2) how to apply a population-based approach to assess the conservation significance of rear-edge, disjunct populations of the three species.

1.6.4 Chapter 5, “Living at the edge - summary”

Chapter 5 provides a summary of my findings, implications of my research, and suggestions for future work.

Figures

Figure 1.1 Generalized patterns across the range, from the leading edge to the rear edge (after Hampe & Petit (2005) and Woolbright et al. (2014)).

List of references

Abeli T, Gentili R, Mondoni A, Orsenigo S, Rossi G. 2014. Effects of marginality on plant population performance. Journal of Biogeography 41(2):239–249.

ACCDC. 2017. Species Occurrence Records. Sackville: ACCDC. Available from: http://www.accdc.com/.

Aiken SG, Dallwitz MJ, Consaul LL, McJannet CL, Gillespie LJ, Boles RL, Argus GW, Gillett JM, Scott PJ, Elven R, et al. 2007. Flora of the Canadian Arctic Archipelago: Descriptions, Illustrations, Identification, and Information Retrieval. Ottawa: NRC Research Press, National Research Council of Canada. Available from: http://www.mun.ca/biology/delta/arcticf/.

Aitken SN, Yeaman S, Holliday JA, Wang T, Curtis-McLane S. 2008. Adaptation, migration or extirpation: climate change outcomes for tree populations. Evolutionary Applications 1(1):95–111.

Alsos IG, Alm T, Normand S, Brochmann C. 2009. Past and future range shifts and loss of diversity in dwarf willow (Salix herbacea L.) inferred from genetics, fossils and modelling. Global Ecology and Biogeography 18(2):223–239.

Alsos IG, Engelskjon T, Brochman C. 2002. Conservation Genetics and Population History of Betula nana, Vaccinium uliginosum, and Campanula rotundifolia in the Arctic Archipelago of Svalbard. Arctic and Alpine Research 34(4):408–418.

Alsos IG, Engelskjon T, Gielly L, Taberlet P, Brochmann C. 2005. Impact of ice ages on circumpolar molecular diversity: insights from an ecological key species. Molecular Ecology 14(9):2739–2753.

Alsos IG, Spjelkavik S, Engelskjon T. 2003. Seed bank size and composition of Betula nana, Vaccinium uliginosum, and Campanula rotundifolia habitats in Svalbard and northern Norway. Canadian Journal of Botany 81(3):220–232.

Angert AL. 2006. Demography of central and marginal populations of monkeyflowers (Mimulus cardinalis and M. lewisii). Ecology 87(8):2014–2025.

Angert AL, Crozier LG, Rissler LJ, Gilman SE, Tewksbury JJ, Chunco AJ. 2011. Do species’ traits predict recent shifts at expanding range edges? Ecology Letters 14(7):677–689.

Arft AM, Walker MD, Gurevitch J, Alatalo JM, Bret-Harte MS, Dale M, Diemer M, Gugerli F, Henry GHR, Jones MH, et al. 1999. Responses of tundra plants to experimental warming: meta-analysis of the International Tundra Experiment. 35

Ecological Monographs 69(4):491–511.

Aronson EL, McNulty SG. 2009. Appropriate experimental ecosystem warming methods by ecosystem, objective, and practicality. Agricultural and Forest Meteorology 149(11):1791–1799.

Barrow E, Maxwell B, Gachon P. 2004. Climate variability and change in Canada: past, present and future. Ottawa: EC Meteorological Service of Canada.

Beatty GE, McEvoy PE, Sweeney O, Provan J. 2008. Range-edge effects promote clonal growth in peripheral populations of the one-sided wintergreen Orthilia secunda. Diversity and Distribution 14(3):546–555.

Bellard C, Bertelmeier C, Leadley P, Thuiller W, Courchamp F. 2012. Impacts of climate change on the future of biodiversity. Ecology Letters 15(4):365–377.

Berg MP, Kiers T, Driessen G, Heijden Mvd, Kooi BW, Kuenen F, Liefting M, Verhoef HA, Ellers J. 2010. Adapt or disperse: understanding species persistence in a changing world. Global Change Biology 16(2):587–598.

Bokhorst S, Huiskes A, Convey P, Sinclair BJ, Lebouvier M, Vijver BVd, Wall DH. 2011. Microclimate impacts of passive warming methods in Antarctica: implications for climate change studies. Polar Biology 34(10):1421–1435.

Box EO, Crumpacker DW, Hardin ED. 1999. Predicted effects of climatic change on distribution of ecologically important native tree and shrub species in Florida. Climatic Change 41(2):213–248.

Brown CD, Vellend M. 2014. Non-climatic constraints on upper elevational plant range expansion under climate change. Proceedings of the Royal Society B: Biological Sciences 281(1794):20141779.

Brown D. 1983. Report on proposed ecological reserve No: 70, Wilson Brook Gypsum Cliffs and Underground Lake. Fredericton.

Bruederle LP. 1999. Genetic differentiation of geographically marginal populations in Carex mitchelliana (Cyperaceae): implications for conservation. Journal of the Torrey Botanical Society 126(1):1–8.

Bunnell FL, Campbell RW, Squires KA. 2004. Conservation priorities for peripheral species: the example of British Columbia. Canadian Journal of Forest Research 34(11):2240–2247.

Cahill AE, Aiello-Lammens ME, Caitlin Fisher-Reid M, Hua X, Karanewsky CJ, Ryu HY, Sbeglia GC, Spagnolo F, Waldron JB, Wiens JJ. 2014. Causes of warm- 36

edge range limits: systematic review, proximate factors and implications for climate change. Journal of Biogeography 41(3):429–442.

Cahill AE, Aiello-Lammens ME, Fisher-Reid MC, Hua X, Karanewsky CJ, Yeong Ryu H, Sbeglia GC, Spagnolo F, Waldron JB, Warsi O, et al. 2012. How does climate change cause extinction? Proceedings of the Royal Society of London B: Biological Sciences 280(1750):1-9.

Channell R. 2004. The conservation value of peripheral populations: the supporting science. Species at Risk 2004 Pathways to Recovery Conference Victoria BC: Species at Risk 2004 Pathways to Recovery Conference Organizing Committee. pp. 1-17.

Ciotir C, Yesson C, Freeland J. 2013. The evolutionary history and conservation value of disjunct Bartonia paniculata subsp. paniculata (Branched Bartonia) populations in Canada. Botany 91(9):605–613.

Clayden SR. 1994. Rare arctic plants in south-eastern New Brunswick. Saint John, NB: NB Museum.

Cox CB, Moore PD. 2005. Biogeography. 7 ed. UK: Blackwell Publishing.

Crawford RMM. 2008. Plants at the margin – ecological limits and climate change Cambridge, UK: Cambridge University Press. 478pp.

Csergö AM, Schönswetter P, Mara G, Deák T, Boscaiu N, Höhn M. 2009. Genetic structure of peripheral, island-like populations: a case study of Saponaria bellidifolia Sm. (Caryophyllaceae) from the south-eastern Carpathians. Plant Systematics and Evolution 278(1-2):33–41.

CYSIP. 2016: Central Yukon Species Inventory Project Friends of Dempster Country. Available from: http://www.flora.dempstercountry.org/CYSIP.About.html.

Dawson TP, Jackson ST, House JI, Prentice IC, Mace GM. 2011. Beyond predictions: biodiversity conservation in a changing climate. Science 332(6025):53–58.

De Witte L, Armbruster G, Gielly L, Taberlet P, Stoecklin J. 2012. AFLP markers reveal high clonal diversity and extreme longevity in four key arctic-alpine species. Molecular Ecology 21(5):1081–1097.

Diekmann OE, Serrao EA. 2012. Range-edge genetic diversity: locally poor extant southern patches maintain a regionally diverse hotspot in the seagrass Zostera marina. Molecular Ecology 21(7):1647–1657.

DNR: Wildlife and Biodiversity. 2017. Nova Scotia: DNR. Available from http://novascotia.ca/natr/wildlife/biodiversity/species-list.asp

DNR: Protected Natural Areas. 2017. [Website] Fredericton, NB: Department of Natural Resources NB. Available from http://www2.gnb.ca/content/gnb/en/departments/natural_resources/ForestsCrown Lands/content/ProtectedNaturalAreas/Class1.html

Doak DF, Morris WF. 2010. Demographic compensation and tipping points in climate- induced range shifts. Nature 467(7318):959–962.

Duffy KJ, Scopece G, Cozzolino S, Fay MF, Smith RJ, Stout JC. 2009. Ecology and genetic diversity of the dense-flowered orchid, Neotinea maculata, at the centre and edge of its range. Annals of Botany 104(3):507–516.

Eckert CG, Samis KE, Lougheed SC. 2008. Genetic variation across species' geographical ranges: the central-marginal hypothesis and beyond. Molecular Ecology 17(5):1170–1188.

Eckstein RL, O'Neill RA, Danhelka J, Otte A, Köhler W. 2006. Genetic structure among and within peripheral and central populations of three endangered floodplain violets. Molecular Ecology 15(9):2367–2379.

Eidesen PB, Alsos IG, Popp M, Stensrud O, Brochman C. 2007. Nuclear vs. plastid data: complex Pleistocene history of a circumpolar key species. Molecular Ecology 16(18):3902–3925.

Eidesen PB, Ehrich D, Bakkestuen V, Alsos IG, Gilg O, Taberlet P, Brochman C. 2013. Genetic roadmap of the Arctic: plant dispersal highways, traffic barriers and capitals of diversity. New Phytologist 200(3):898–910.

Eizaguirre C, Baltazar–Soares M. 2014. Evolutionary conservation–evaluating the adaptive potential of species. Evolutionary Applications 7(9):963–967.

Elmendorf SC, Henry GHR, Hollister RD, Alatalo J, Björk RG, Bjorkman AD, Callaghan TV, Collier LS, Cooper EJ, Cornelissen JHC, et al. 2012. Global assessment of experimental climate warming on tundra vegetation: heterogeneity over space and time. Ecology Letters 15(2):164–175.

Field CB, Barros VR, Mastrandrea MD, Mach KJ, Abdrabo MA, Adger WN, Anokhin YA, Anisimov OA, Arent DJ, Barnett J, et al. 2014. Summary for Policymakers Climate Change 2014: Impacts, Adaptation, and Vulnerability Working Group II Contribution to the IPCC 5th Assessment Report Cambridge, UK and New York, USA: Cambridge University Press. pp. 1-32. 38

FNAA. 2008. Flora of North America. Flora of North America Association. Available from: http://www.efloras.org/index.aspx.

Franco AM, Hill J, Kitschke C, Collingham YC, Roy DB, Fox R, Huntley B, Thomas CD. 2006. Impacts of climate warming and habitat loss on extinctions at species' low-latitude range boundaries. Global Change Biology 12(8):1545–1553.

Gapare WJ, Aitken SN. 2005. Strong spatial genetic structure in peripheral but not core populations of Sitka spruce [Picea sitchensis (Bong.) Carr.]. Molecular Ecology 14(9):2659–2667.

Gardes M, Dahlberg A. 1996. Mycorrhizal diversity in arctic and alpine tundra: an open question. New Phytologist 133(1):147–157.

Gaston KJ. 2009. Geographic range limits: achieving synthesis. Proceedings of the Royal Society 276(1661):1395–1406.

GBIF.org. 2016. GBIF Occurrence Download. ed.: GBIF.org. Available from: http://www.gbif.org.

Geber MA. 2008. To the edge: studies of species' range limits. New Phytologist 178(2):228–230.

Gibson SY, Van Der Marel RC, Starzomski BM. 2009. Climate change and conservation of leading-edge peripheral populations. Conservation Biology 23(6):1369–1373.

Gorham RP. 1944. Botanical notes. Acadian Naturalist 1(3):183–186.

Government of Canada [Online data]. 2016. Ottawa: Government of Canada. Available from: http://ccds-dscc.ec.gc.ca/?page=main&lang=en.

Grace J. 1987. Climatic tolerance and distribution of plants. New Phytologist 106(1):113–130.

Grime JP. 1979. Plant strategies, vegetation processes, and ecosystem properties. 2nd ed. West Sussex, England: John Wiley & Sons LTD. 456pp.

Gugerli F, Bauert MR. 2001. Growth and reproduction of Polygonum viviparum show weak responses to experimentally increased temperature at a Swiss Alpine site. Botanica Helvetica 111(2):169–180.

Halbritter AH, Alexander JM, Edwards PJ, Billeter R. 2013. How comparable are species distributions along elevational and latitudinal climate gradients? Global

Ecology and Biogeography 22(11):1228–1237.

Hampe A, Jump AS. 2011. Climate Relicts: Past, Present, Future. Annual Review of Ecology, Evolution and Systematics 42(2011):313–333.

Hampe A, Petit RJ. 2005. Conserving biodiversity under climate change: the rear edge matters. Ecology Letters 8(5):461–467.

Hardie DC, Hutchings JA. 2010. Evolutionary ecology at the extremes of species’ ranges. Environmental Reviews 18(2010):1–20.

Hargreaves AL, Samis KE, Eckert CG. 2014. Are species' range limits simply niche limits writ large? A review of transplant experiments beyond the range. American Naturalist 183(2):157–173.

Herlihy CR, Eckert CG. 2005. Evolution of self-fertilization at geographical range margins? A comparison of demographic, floral, and mating system variables in central vs. peripheral populations of Aquilegia canadansis (Ranunculaceae). American Journal of Botany 92(4):744–751.

Hind HY. 1865. A preliminary report on the geology of New Brunswick, together with a special report on the distribution of the "Quebec Group" in the province. In: Observations on the geology of southern New Brunswick. LW Bailey, Matthew GF, Hartt CF, editors. Fredericton: GE Fenety. 293pp.

Hollister RD, Webber PJ. 2000. Biotic validation of small open-top chambers in a tundra ecosystem. Global Change Biology 6(7):835–842p

IPCC. 2007. Climate Change 2007- The Physical Science Basis. Cambridge University Press.

Jacquemart A-L. 1996. Vaccinium uliginosum L. Journal of Ecology 84(5):771–786p

Jacquemart A-L, Thompson JD. 1996. Floral and pollination biology of three sympatric Vaccinium (Ericaceae) species in the Upper Ardennes, Belgium. Canadian Journal of Botany 74(2):210–221.

Jia J, Dai Z, Li F, Liu Y. 2016. How Will Global Environmental Changes Affect the Growth of Alien Plants? Frontiers in Plant Science 7(1623). http://doi.org/10.3389/fpls.2016.01623.

John E, Turkington R. 1997. A 5-year study of the effects of nutrient availability and herbivory on two boreal forest herbs. Journal of Ecology 85(4):419–430.

Jump AS, Peñuelas J. 2005. Running to stand still: adaptation and the response of plants 40

to rapid climate change. Ecology Letters 8(9):1010–1020.

Kirk H, Freeland JR. 2011. Applications and implications of neutral versus non-neutral markers in molecular ecology. International journal of molecular sciences 12(6):3966-3988.

Kormondy EJ. 1996. Concepts of Ecology. 4th ed. USA: Prentice Hall Inc. 209pp.

Lambers H, Chapin FSI, Pons TL. 2008. Plant Physiological Ecology New York: Springer-Verlag. 605pp.

Leppig G, White JW. 2006. Conservation of peripheral plant populations in California. Mardono 53(3):264–274.

Lesica P, Allendorf FW. 1995. When are peripheral populations valuable for conservation? Conservation Biology 9(4):753–760.

Lesica P, Crone EE. 2016. Arctic and boreal plant species decline at their southern range limits in the Rocky Mountains. Ecology Letters 20(2):166-174.

Lesica P, McCune P. 2004. Decline of arctic-alpine plants at the southern margin of their range following a decade of climatic warming. Journal of Vegetation Science 15(5):679-690.

Louthan AM, Doak DF, Angert AL. 2015. Where and when do species interactions set range limits? Trends in Ecology and Evolution 30(12):780–792.

Loveless MD, Hamrick JL. 1984. Ecological determinants of genetic structure in plant populations. Annual Review of Ecology and Systematics 15(1):65–95.

MacArthur RH. 1972. Geographical Ecology: Patterns in the Distribution of Species Princeton, New Jersey: Princeton University Press. 146pp.

Mackenzie A, Ball AS, Virdee SR. 2001. Ecology Oxford, UK: BIOS Scientific Publishers. 339pp.

Marion GM, Henry GHR, Freckman DW, Johnstone J, Jones G, Jones MH, Levesque E, Molau U, Molgaard P, Parsons AN, et al. 1997. Open-top designs for manipulating field temperature in high-latitude ecosystems. Global Change Biology 3(S1):20–32.

Markham JH. 2009. Does Dryas integrifolia fix nitrogen? Botany 87(11):1106–1109.

Mason SC, Palmer G, Fox R, Gillings S, Hill JK, Thomas CD, Oliver TH. 2015. Geographical range margins of many taxonomic groups continue to shift polewards. Biological Journal of the Linnean Society 115(3):586–597.

Mátyás C, Vendramin GG, Fady B. 2009. Forests at the limit: evolutionary–genetic consequences of environmental changes at the receding (xeric) edge of distribution. Annals of Forest Science 66(8):1286–1560.

Mayer C, Michez D, Chyzy A, Brédat E, Jacquemart A-L. 2012. The abundance and pollen foraging behaviour of bumble bees in relation to population size of whortleberry (Vaccinium uliginosum). PLoS ONE 7(11). Epub November 27, 2012.

Millar MA, Byrne M, Coates DJ. 2010. The maintenance of disparate levels of clonality, genetic diversity and genetic differentiation in disjunct subspecies of the rare Banksia ionthocarpa. Molecular Ecology 19(9):4217–4227.

Nantel P, Gagnon D. 1999. Variability in the dynamics of northern peripheral versus southern populations of two clonal plant species, Helianthus divaricatus and Rhus aromatica. Journal of Ecology 87(5):748–760.

NatureServe. 2016. NatureServe Explorer: an online encyclopaedia of life. NatureServe Explorer: an online encyclopaedia of life. 2008 ed., Virginia, U.S.A: NatureServe. Available from: http://explorer.natureserve.org/.

Orsini L, Vanoverbeke J, Swillen I, Mergeay J, Meester L. 2013. Drivers of population genetic differentiation in the wild: isolation by dispersal limitation, isolation by adaptation and isolation by colonization. Molecular Ecology 22(24):5983–5999.

Parmesan C. 2006. Ecological and evolutionary responses to recent climate change. Annual Review of Ecology, Evolution, and Systematics 37(2006):637–669.

Parmesan C. 2007. Influences of species, latitudes and methodologies on estimates of phenological response to global warming. Global Change Biology 13(9):1860– 1872.

Parmesan C, Yohe G. 2003. A globally coherent fingerprint of climate change impacts across natural systems. Nature 421(6918):37–42.

Pauls SU, Nowak C, Balint M, Pfenniger M. 2013. The impact of global climate change on genetic diversity within populations and species. Molecular Ecology 22(4):925–946.

Pellissier L, Bråthen KA, Vittoz P, Yoccoz NG, Dubuis A, Meier ES, Zimmermann NE, Randin CF, Thuiller W, Garraud L, et al. 2013. Thermal niches are more 42

conserved at cold than warm limits in arctic-alpine plant species. Global Ecology and Biogeography 22(8):933–941.

Peñuelas J, Sardans J, Estiarte M, Ogaya R, Carnicer J, Coll M, Barbeta A, Rivas-Ubach A, Llusia J, Garbulsky M, et al. 2013. Evidence of current impact of climate change on life: a walk from genes to the biosphere. Global Change Biology 19(8):2303–2338.

Pfeifer M, Passalacqua NG, Bartrma S, Schatz B, Croce A, Carey PD, Kraudelt H, Jeltsch F. 2010. Conservation priorities differ at opposing species borders of a European orchid. Biological Conservation 143(9):2207–2220.

Pironon S, Papuga G, Villellas J, Angert AL, García MB, Thompson JD. 2016. Geographic variation in genetic and demographic performance: new insights from an old biogeographical paradigm. Biological Reviews: http://onlinelibrary.wiley.com/doi/10.1111/brv.12313/full.

Pironon S, Villellas J, Morris WF, Doak DF, García MB. 2015. Do geographic, climatic or historical ranges differentiate the performance of central versus peripheral populations? Global Ecology and Biogeography 24(6):611–620.

Provan J, Maggs CA. 2012. Unique genetic variation at a species’ rear edge is under threat from global climate change. Proceedings of the Royal Society B 279(1726):39–47.

Pullin AS, Sutherland W, Gardner T, Kapos V, Fa JE. 2013. Conservation priorities: identifying need, taking action and evaluating success. Key Topics in Conservation Biology 2(2013):3-22.

Reisch C. 2008. Glacial history of Saxifraga paniculata (Saxifragaceae): molecular biogeography of a disjunct arctic-alpine species from Europe and North America. Biological Journal of the Linnean Society 93(2):385–398.

Roberts PR. 1965. New Records of arctic species in southeastern New Brunswick. Rhodora 67(769):92–93.

Robinson J, Hermanutz L. 2015. Evaluating human-disturbed habitats for recovery planning of endangered plants. Journal of Environmental Management 150(2015):157–163.

Roy P, Huard D. 2016. Future Climate Scenarios–Province of New-Brunswick. Montreal: Ouranos.

Rune O. 1954. Notes on the flora of the Gaspé Peninsula. Svensk Bot Tidskr

48(1954):117–135.

Sagarin RD, Gaines SD, Gaylord B. 2006. Moving beyond assumptions to understand abundance distributions across the ranges of species. Trends in Ecology and Evolution 21(9):524–530.

Samis KE, Eckert CG. 2007. Testing the abundant center model using range-wide demographic surveys of two coastal dune plants. Ecology 88(7):1747–1758.

Schickhoff U. 2011. Biogeographical distributions: the role of past environments, physical factors, and biotic interactions. In: The SAGE Handbook of Biogeography. Millington A, Blumler M, Schickhoff U, editors. London UK: SAGE Publications. 601pp.

Schluter D, Conte GL. 2009. Genetics and ecological speciation. Proceedings of the National Academy of Sciences 106(Supplement 1):9955-9962.

Semenova TA, Morgado LN, Welker JM, Walker MD, Smets E, Geml J. 2015. Long- term experimental warming alters community composition of ascomycetes in Alaskan moist and dry arctic tundra. Molecular Ecology 24(2):424–437.

Settele J, Scholes R, Betts R, Bunn S, Leadley P, Nepstad D, Overpeck JT, Taboada MA. 2014. Terrestrial and inland water systems. Climate Change 2014: Impacts, Adaptation, and Vulnerability Part A: Global and Sectoral Aspects Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change Cambridge, United Kingdom and New York, NY, USA: Cambridge University Press. pp. 271–359.

Sexton JP, Hangartner SB, Hoffmann AA. 2014. Genetic isolation by environment or distance: which pattern of gene flow is most common? Evolution 68(1):1–15.

Sexton JP, Hufford MB, Bateman CA, Lowry DB, Meimberg H, Strauss SY, Rice KJ. 2016. Climate structures genetic variation across a species' elevation range: a test of range limits hypotheses. Molecular Ecology 25(4):911–928.

Sexton JP, McIntyre PJ, Angerst AL, Rice KJ. 2009. Evolution and ecology of species range limits. Annual Review of Ecology, Evolution and Systematics 40(2009):415–436.

Shafer ABA, Cullingham CI, Côté SD, Coltman AW. 2010. Of glaciers and refugia: a decade of study sheds new light on the phylogeography of northwestern North America. Molecular Ecology 19(21):4589–4621.

Shafer ABA, Wolf JBW. 2013. Widespread evidence for incipient ecological speciation:

a meta-analysis of isolation-by-ecology. Ecology Letters 16(7):940–950.

Shaver GR, Canadell J, Chapin FS, III, Gurevitch J, Harte J, Henry G, Ineson P, Jonasson S, Melillo J, Pitelka L, et al. 2000. Global warming and terrestrial ecosystems: a conceptual framework for analysis. Bioscience 50(10):871–882.

Thomas CD, Bulman CR, Wilson RJ. 2008. Where within a geographical range do species survive best? A matter of scale. Insect Conservation and Diversity 2008(1):2–8.

Thomas CD, Cameron A, Green RE, Bakkenes M, Beaumont L, Vollingham YC, Erasmus BFN, Suqueira MFd, Grainger A, Hannah L, et al. 2004. Extinction risk from climate change. Nature 427(8):145–148.

Wagner V, Treiber J, Danihelka J, Ruprecht E, Wesche K, Hensen I. 2012. Declining genetic diversity and increasing genetic isolation toward the range periphery of Stipa pennata, a Eurasian feather grass. International Journal of Plant Sciences 173(7):802–811.

Westberg E, Kadereit JW. 2009. The influence of sea currents, past disruption of gene flow and species biology on the phylogeographical structure of coastal flowering plants. Journal of Biogeography 36(7):1398–1410.

Williams SE, Shoo LP, Isaac JL, Hoffmann AA, Langham G. 2008. Towards an integrated framework for assessing the vulnerability of species to climate change. PLoS Biology 6(12):2621–2626.

Woolbright SA, Whitham TG, Gehring CA, Allan GJ, Bailey JK. 2014. Climate relicts and their associated communities as natural ecology and evolution laboratories. Trends in Ecology and Evolution 29(7):406–416.

Xu Z-F, Hu T-X, Wang K-Y, Zhang Y-B, Xian J-R. 2009. Short-term responses of phenology, shoot growth and leaf traits of four alpine shrubs in a timberline ecotone to simulated global warming, Eastern Tibetan Plateau, China. Plant Species Biology 24(1):27–34.

Yakimowski SB, Eckert CG. 2007. Threatened peripheral populations in context: geographical variation in population frequency and size and sexual reproduction in a clonal woody shrub. Conservation Biology 21(3):811–822.

Yakimowski SB, Eckert CG. 2008. Populations do not become less genetically diverse or more differentiated towards the northern limit of the geographical range in clonal Vaccinium stamineum (Ericaceae). New Phytologist 180(2):534–544.

Źróbek-Sokolnik A. 2012. Temperature stress and responses of plants. In: Environmental adaptations and stress tolerance of plants in the era of climate change. Parvaiz A., Prasad MNV, editors. New York, USA: Springer Science and Business Media. pp. 113-134.

Chapter 2 - Genetic diversity and differentiation of rear-edge

populations of three arctic-alpine species in Eastern Canada.

Abstract

Arctic-alpine species at the rear edge of their range may face particularly high risks of extirpation with climate change. This is problematic as these populations may contain unique genetic diversity not present in more central populations and may be particularly important for the adaptive capacity of these species in a warming world. Assessing the level of genetic variation within and among populations of wide-ranging species will facilitate in the evaluation of the conservation significance of rear-edge populations. I used Amplified Fragment Length Polymorphisms (AFLPs) on three widespread arctic- alpine species with rear-edge, disjunct populations in Atlantic Canada (Anemone parviflora, Dryas integrifolia, and Vaccinium uliginosum) to assess genetic variation within and among populations at different range-edge locations. All species displayed considerable variation in genetic diversity and differentiation from the near-edge to disjunct populations. Different patterns of diversity and differentiation across the rear edge for three mostly sympatric arctic-alpine species suggest that different abiotic and biotic factors impact gene flow in each species. Based on the findings of this research, disjunct populations of the three species may not warrant special conservation attention, but further investigations may be justified. Future research could include an assessment of genetic diversity based on loci under selection, and the use of different genetic marker

systems or transplant experiments to assess whether locally adapted genotypes are found in the disjunct populations.

2.1 Introduction

Climate change impacts are expected to be most evident at the edges of species' ranges where abiotic and / or biotic conditions will change sufficiently to advance or shrink the range edge (Doak & Morris 2010, Hardie & Hutchings 2010, Reed et al. 2011, Alsos et al. 2012, Pauls et al. 2013). While rapid climate change may outpace the ability of some species to adapt or migrate (Jump & Peñuelas 2005, Crawford 2008), others are expected to shift their distributions. In the short term, responses of populations at range edges will reflect their tolerance to changing conditions (phenotypic plasticity) as well as their ability to disperse to new habitats. In the longer term, species' responses will also depend on their standing genetic variation because it confers adaptive potential (i.e., their ability to respond to climate change via natural selection (Bellard et al. 2012,

Eizaguirre & Baltazar–Soares 2014)).

A commonly held assumption is that genetic diversity declines from the core to the edge of a species' distribution (Garcia et al. 2000, Jump & Woodward 2003, Eckstein et al.

2006, Eckert et al. 2008, Tsaliki & Diekmann 2009). Peripheral or range-edge populations often display decreased genetic diversity and increased genetic differentiation compared to more centrally located populations, with the lowest diversity and highest differentiation often occurring in the most isolated populations (Bruederle

1999, Hamilton & Eckert 2007, Beatty et al. 2008, Eckert et al. 2008, Wagner et al.

2012, Villellas et al. 2014, Hirsch et al. 2015). This has been attributed to the typically small range-edge population sizes, relative to those located at the core of the range

(Gaston 2003). Smaller population size is often associated with small effective population size and reduced genetic diversity due to genetic drift and inbreeding (Hartl

& Clark 1997, Reed & Frankham 2003, Musche et al. 2008). However, this pattern of reduced genetic diversity in peripheral populations is not universally true (Eckstein et al.

2006, Bauert et al. 2007, Eckert et al. 2008), and has recently been shown to apply in fewer than 50% of studied species (Pironon et al. 2016).

If range-edge populations are demographically isolated and genetically differentiated from their neighbors, then we may infer that they may carry unique alleles that are beneficial in edge habitats, potentially contributing to a species' capacity to respond to climate change (Channell 2004, Hewitt 2004, Csergö et al. 2009, Hampe & Jump 2011).

Environmental factors such as topography, substrate type, and surrounding habitat matrix may act to exacerbate the geographic isolation of range-edge populations (Wang

& Bradburd 2014) by creating barriers to dispersal and hence reducing gene flow among populations. Isolation by distance (IBD), presumed to be associated with increased divergence among populations (Wright 1943), is a long-standing measure that has been used to explain significant genetic differentiation among populations. An increase in genetic differentiation based on isolation by distance provides a broad framework that can often explain patterns of population genetic structure, because dispersal ability of a species is inherently constrained by distance (Orsini et al. 2013). Thus populations closer to each other are expected to be genetically more similar (Wright 1943). Habitat fragmentation may also act to exacerbate isolation among populations, can promote

local adaptation, and can have large and negative effects on genetic diversity (Aguilar et al. 2008, Vranckx et al. 2012), causing reduced genetic diversity and increased differentiation as a result of geographic and genetic isolation (Young et al. 1996).

Reduced genetic diversity and increased differentiation is due to increased levels of inbreeding and genetic drift, both also associated with smaller numbers of individuals in fragmented populations (Couvet 2002).

Although population genetic theory provides expectations for patterns of genetic variation associated with population demography and abundance, variation may also be influenced by species-specific factors such as life history traits, which may exacerbate or buffer demographic effects on genetic variability. For example, small populations consisting of long-lived individuals (perennials or clones) may better retain genetic diversity than small populations of annuals (Honnay & Jacquemyn 2008, Abeli et al.

2014, Reisch & Bernhardt-Römermann 2014). Clonality, or vegetative propagation, may increase in frequency and / or importance at the edges of a species' range (Eckert 2002,

Silvertown 2008). Although clonal reproduction increases opportunities for inbreeding, which may reduce genetic diversity (Honnay & Bossuyt 2005), it has also been found to increase allelic diversity and reproductive success in some species (Balloux et al. 2003,

Meloni et al. 2013). Both longevity and clonality can buffer populations from the loss of genetic diversity caused by interrupted gene flow, and slow the effects of genetic drift and inbreeding. By contrast, reduced outcrossing can affect genetic diversity and differentiation by increasing genetic drift and inbreeding (Ellstrand & Elam 1993).

Outcrossing species often show higher levels of genetic diversity (Reisch & Bernhardt-

Römermann 2014) compared to selfing species (Loveless & Hamrick 1984, Hamrick &

Godt 1996, Nybom & Bartish 2000, Nybom 2004). High levels of outcrossing possibly act as buffer against loss of genetic diversity (Jiang et al. 2015) by promoting gene flow among populations. While selfing is expected to increase inbreeding (Ellstrand & Elam

1993), it may also lead to purging of deleterious alleles via selection, and thus reduce inbreeding depression (Levin 2012). High dispersal capacity may increase gene flow among populations even over considerable distances (Cruden 1966, Carlquist 1983,

Webb 1986, Albert et al. 2005, Brochet et al. 2009, Viana et al. 2013), while lack of dispersal can increase genetic differentiation (Westberg & Kadereit 2009, Abeli et al.

2014). Moreover, the impact of dispersal on gene flow is mediated by multiple factors.

In general, species that use wind for pollination and / or seed dispersal tend to show low levels of genetic differentiation among populations but maintain high levels of within- population genetic diversity (Govindaraju 1988, Thiel-Egenter et al. 2009). At the other end of the spectrum, insect- or animal-pollinated species with gravity-dispersed seeds tend to show high levels of genetic differentiation among populations, and genetic diversity is more easily lost from small, isolated populations (Meirmans et al. 2011).

However, seed dispersal and pollen flow are greatly affected by the specific vector in question (Hamilton & Miller 2002, Petit et al. 2005, García et al. 2007). For example, birds can be very effective in dispersing seeds long-distance (Viana et al. 2013).

Rear-edge populations, and in particular refugial or relict populations, may be older

(Hampe & Petit 2005, Woolbright et al. 2014) and may, at one time, have been isolated and acted as a source for range expansion (refugial populations are here referred to as those that survived glaciations in isolated pockets beyond the ice, and relict populations are those that were isolated during the re-colonization process after deglaciation). Like range-edge populations in general, refugial and relict populations may harbor unique diversity due to historical isolation (Lesica & Allendorf 1995, Leppig & White 2006,

Millar et al. 2010, Pfeifer et al. 2010, Dobrowski 2011, Hampe & Jump 2011, Provan &

Maggs 2012, Hampe et al. 2013, Lepais et al. 2013, Woolbright et al. 2014, Kvist et al.

2015). They may also have lost genetic diversity over time due to genetic drift, range contractions resulting in smaller population sizes, and increased levels of inbreeding

(Hampe & Petit 2005, Woolbright et al. 2014). Refugial and relict populations may represent considerable genetic resources (Nantel & Gagnon 1999, Channell 2004,

Hewitt 2004, Beatty et al. 2008, Westergaard et al. 2008, Csergö et al. 2009, Pfeifer et al. 2010, Provan & Maggs 2012, Lepais et al. 2013). They may contribute to a species' adaptive potential, and thus its ability to respond to environmental change, giving these populations significant conservation value (Franco et al. 2006, Millar et al. 2010, Pfeifer et al. 2010, Hampe & Jump 2011, Provan & Maggs 2012).

Predictions of decreasing genetic diversity and increasing genetic differentiation towards the edge of species' ranges may not apply to disjunct populations as they represent an

“extreme manifestation of geographic peripherality” (Hamilton & Eckert 2007). A decrease in genetic diversity with increased genetic differentiation for some disjunct

populations is supported in the literature (Linhart & Premoli 1994, Lammi et al. 1999,

Hamilton & Eckert 2007, Pandey & Geburek 2010, Meeus et al. 2012, Nagamitsu et al.

2014). However, others have found (1) no decline from core to edge populations (Van

Rossum et al. 2003, Van Rossum & Prentice 2004, Ilves et al. 2016); (2) higher genetic diversity in disjunct populations (Kvist et al. 2015); (3) unique alleles in disjunct compared to core populations (Alsos et al. 2005, Skrede et al. 2006, Garcia et al. 2012,

Kvist et al. 2015); and (4) no increase in differentiation in disjunct populations (Ikeda et al. 2015). It appears that especially refugial populations may not follow theoretical predictions because of their phylogeographic history (Hampe & Jump 2011, Garcia et al.

2012, DeChaine et al. 2013). This, in fact, often applies to species that currently occur in high latitudes, which have relict populations in low latitude areas where they were once common (Hewitt 2004, Hampe & Jump 2011).

It is generally accepted that rear-edge populations may be at high risk of disappearing quickly under a warming climate (Franco et al. 2006, Hampe & Jump 2011) and may be more prone to extinction than core and leading-edge populations (Vucetich & Waite

2003). However, they are less frequently considered in research than leading-edge populations, with the exception of refugia, which have received considerable attention.

Given that rear-edge populations may harbor unique genetic variation, further study is needed to assess the status of disjunct, rear-edge populations, to evaluate their conservation significance, and to facilitate management decisions.

2.1.1 Goals and objectives

The overarching goal of this research was to evaluate the conservation significance of rear-edge populations of three wide-ranging arctic-alpine species in Eastern Canada, and to provide information on genetic patterns at the rear range edge to aid in conservation management decisions. To my knowledge, no empirical study has focused on disjunct populations of widespread plant species in North America with similar arctic-alpine ranges, and used data on population structure to infer isolation mechanisms.

My objectives were (1) to evaluate and compare within-population genetic diversity and among-population genetic differentiation at the southern range edge of Anemone parviflora, Dryas integrifolia, and Vaccinium uliginosum in Atlantic Canada; and (2) to explore the potential association between proximate ecological factors, such as geographic distance and habitat fragmentation, that may impinge on genetic variation and the pattern of genetic structure among populations within and across species.

Of eighty arctic-alpine species occurring in New Brunswick, Canada (Hinds 2000,

ACCDC 2008), three (A. parviflora, D. integrifolia, and V. uliginosum) were selected for this study. These species were suitable for intra- and interspecific comparisons associated with range-edge patterns of genetic diversity given that the species share similar geographical distributions. In addition, populations of each occur jointly in some localities as disjunct rear-edge populations at their southern limits (Wilson Brook,

Restigouche, Mount Logan, Tablelands), continuous range-edge populations (Bellburns,

Port-au-Choix, Sandy Cove, Cooks Harbour, Burnt Cape), and core populations (for 55

two of the species). Taken together, these populations provide an opportunity for evaluating genetic patterns at the rear range edge of three locally rare, yet globally common, widespread species.

Based on the hypothesis that ecological and geographic isolation is associated with genetic variation, I predicted that mean population genetic diversity would be lower and mean population genetic differentiation would be greater for populations that are (1) at a greater distance from the rear range edge; (2) disjunct from the continuous range edge; and (3) disturbed (as disturbance can fragment habitats and thus increase isolation among populations). I predicted that these associations would be similar across the three species, given their presumably shared recent histories (post Last Glacial Maximum). In addition, since postglacial re-colonization is presumed to affect contemporary genetic population structure, I predicted that disjunct populations would be clearly distinct from other populations based on genetic metrics.

2.2 Materials and Methods

2.2.1 Study region

The study was carried out in Eastern Canada, from northern Labrador to the west coast of Newfoundland, and south to New Brunswick and Mount Logan in the Gaspé

Peninsula in Quebec (Fig. 2.1). The sampled region contains populations occurring near the core of the species' ranges (except A. parviflora), in the range edge, as well as disjunct populations defined as being ≥ 100 km, and / or distinctly isolated from the range edge (as defined below).

2.2.2 Study species

Each of the three species used here is common throughout most of its range with a large geographic distribution spanning arctic to temperate regions, but with reported population frequencies declining from their range cores to edges and beyond. The species also share some life history and ecological traits, including mating system, dispersal of pollen or seeds, clonal reproduction, and habitat affinities. Vaccinium uliginosum and D. integrifolia are reported to have been much more common after deglaciation >10,000 years ago and, based on data from lake sediment cores, appear to share a similar phylogeographic history (Ritchie 1987, Mayle & Cwynar 1995,

Tremblay & Schoen 1999).

Dryas integrifolia Vahl (White Mountain Avens, Chapter 1, Fig. 1.2) is one of three species in the genus Dryas (Family Rosaceae), and occurs in Alaska, Canada and

Greenland. The species is a diploid (2n = 2x = 18), clonal, evergreen, mat-forming, woody dwarf shrub growing up to 17 cm tall, expanding horizontally with older central stems often dying, while the individual itself is maintained by continued clonal growth

(De Witte et al. 2012). Individuals of Dryas sp. can reach a considerable age; for example, a single individual of D. integrifolia was shown to be 68 years old (Au &

Tardif 2007). However, individuals of D. integrifolia are likely to have a much longer lifespan, since individuals of D. octopetala, a closely related and better-studied species, have been estimated at up to 570 years based on annual horizontal growth (De Witte et al. 2012). Dryas integrifolia is self-compatible, but thought to be primarily an out- crosser (Murray 1997). It is andromonoecious, forming both bi-sexual and male flowers on the same plant (Philipp & Siegismund 2003), and insect pollinated. Fruits are wind- dispersed, elongated, feathery-styled achenes. It prefers soils with a low organic content that support a rather low diversity of other species (Aiken et al. 2007) and occurs in wet meadows, river terraces, tundra slopes and ridges, especially exposed, calcium-rich habitats. In the study region, populations of D. integrifolia occur in northern Labrador, along the Limestone Barrens in Newfoundland, in a unique habitat in an isolated hanging valley in the Tablelands, in the mountain range of the Gaspé Peninsula, and in

New Brunswick. There are no reports of this species from further south.

Anemone parviflora Michx. (Northern White Anemone, Ranunculaceae, Chapter 1, Fig.

1.3) is a diploid (2n = 2x = 16) perennial, clonal herb growing to 35 cm tall with rosettes that die back annually (John & Turkington 1997, Aiken et al. 2007). Pollination is by

insects; fruits are wind-dispersed, densely woolly achenes (CYSIP 2016). Across its range, the species grows along streams, in meadows and on rocky slopes (FNAA 2008).

While considerable work has been done on the taxonomic status of members of the genus Anemone, no genetic studies have assessed the species' phylogeography, or genetic variation across its range. In the study region, A. parviflora occurs along the

Limestone Barrens and the Tablelands of Newfoundland; it is known from a few populations on the Gaspé Peninsula of Quebec, and from at least two locations in New

Brunswick (Table 2.1). The species has also been reported from Nova Scotia (two populations in Cape Breton Highlands National Park (ACCDC 2017)), but resources and time were insufficient to sample these disjunct, rear-edge populations.

Vaccinium uliginosum L. (Alpine Blueberry, Ericaceae, Chapter 1, Fig. 1.4) is a circumboreal, circumpolar dwarf shrub growing up to 20 cm tall. The species is well studied across its range (Jacquemart & Thompson 1996, Alsos et al. 2002, Alsos et al.

2003, Alsos et al. 2005, Eidesen et al. 2007, De Witte et al. 2012, Mayer et al. 2012,

Eidesen et al. 2013). The species shows self- and cross-pollination and is a facultative selfer. Pollination is by insects, and seeds are dispersed by frugivorous birds. The species includes diploids, triploids, and tetraploids across its range (2n = 4x = 48 for the sampled region as identified by Eidesen et al. (2007)). Individual shoot ages of 93 years

(Alsos et al. 2002) have been reported, and the age of the oldest individual sampled at is over 1880 years (De Witte et al. 2012). In the Arctic, vegetative reproduction of stems is more important than reproduction by seed (Jacquemart 1996), but this is not known for

other regions. The species grows primarily on acidic, sometimes calcareous soils, in heathlands and tundra. In the study region, V. uliginosum occurs frequently in Labrador, exposed areas in Newfoundland (along the Limestone Barrens, the Tablelands, and the

Long Range Mountains), in the uplands of the Gaspé Peninsula, as individual clones on

Mt. Carleton and Mt. Denis in New Brunswick, and in the Cape Breton Highlands. It is also known further to the south in the Appalachian mountain range of the USA.

2.2.3 Study site classification and field sampling

Seven to nine populations were selected to represent near-core (termed core in the following for simplicity), range-edge, and disjunct populations in the sampled portion of each species' range (Table 2.1, Fig. 2.1). Samples of A. parviflora were not found at the core population sites, therefore research questions requiring a core population were addressed only for D. integrifolia and V. uliginosum. All populations occur within either the Eastern Temperate Forest eco-region of the Bay of Fundy or the Arctic Cordillera eco-region (CEC 2009), and were classified according to intraspecific range-edge proximity into (1) core populations, (2) range-edge populations, (3) disjunct populations

(Tables 2.1 and 2.2, Fig. 2.1). Core populations are identified as those populations that occur in the range where the species is common based on range maps and occurrence records (ACCDC 2017). Range-edge populations are those limited to specialized habitat, and beyond which populations occur only in isolated locations. Disjunct populations were defined as being ≥ 100 km rom the nearest population, and / or based on inferences of seed and pollen dispersal barriers. Gene flow among populations is presumed to be

limited due to the heterogeneity of the landscape. 100 km as the threshold was chosen as a distance that may prevent gene flow among populations even in an open landscape.

(1) Core: Two sites (Torngat and Nain, Labrador) were within the core range of the species based on range maps. The species are common in this region and occur in arctic habitat within a difficult-to-access region. These are the only core sites for which sample collection was possible for D. integrifolia and V. uliginosum.

(2) Range edge: I defined five sites as located at the warm range edge, where the species are less common, and where calcareous habitat is uncommon compared to the core of the species' ranges. These sites occur over a continuous 200-km stretch of coastline within the specialized Limestone Barrens habitat in Newfoundland (Burnt Cape, Cooks

Harbour, Sandy Cove, Port-au-Choix and Bellburns), an open, coastal landscape that supports 114 of Newfoundland’s 271 rare plant species (Robinson & Hermanutz 2015);

(3) Disjunct: I defined four locations as disjunct where presumed isolation factors

(distance, landscape) would restrict gene flow. These four locations occur beyond the edge of continuous habitat distribution and were located ≥ 100 km from the nearest putative continuous range-edge population, or in isolated habitat patches:

(3a) Tablelands, Newfoundland (D. integrifolia, A. parviflora) – in the unique travertine seeps of an area characterized by exposed mantle rock containing toxic minerals, in an open landscape in an upland valley, 100 km beyond the nearest sampled continuous range-edge population (Bellburns). Other populations to the south are > 50 km distant,

except one population of D. integrifolia approximately 5 km away at Bonne Bay (Big

Hill) (ACCDC 2017), but I inferred the Tablelands population to be isolated because of the landscape. The site was classified as range edge for V. uliginosum because there are other populations within and adjacent to this valley, and the distribution of this species appears continuous with other populations along the coast.

(3b) Mount Logan (sensu lato), Gaspé Peninsula, Québec (D. integrifolia, V. uliginosum) – populations in an alpine, open habitat of < 20 x 10 m. This may be part of a larger metapopulation, disjunct from the continuous range edge; an additional population of D. integrifolia was found in 2015 at Mont de la Table (> 1000 genets in approximately 5 ha), approximately 50 km away (S. Bailey, pers. comm., 2015). Other populations may exist at Mont St.-Anne (> 130 km away), Mont St. Pierre (Rune 1954), near Percé (> 150 km away), on Anticosti Island and on the Mingan Archipelago (> 200 km away) (GBIF.org 2016).

(3c) Wilson Brook, southern New Brunswick (D. integrifolia, A. parviflora) – within the borders of the Wilson Brook Protected Natural Area on a gypsum cliff habitat

(approximately 1 km by 20 m) within an inland forest. Gypsum sites and calcareous areas in the Maritimes are fairly well documented but this is the only gypsum site at which these species are found (ACCDC 2017). No gene flow with these populations is expected because the nearest known population for each species (Mount Logan) is more than 365 km away, and the forested landscape presumably restricts pollen and seed

movement.

(3d) Restigouche, northwestern New Brunswick (A. parviflora) – along the shores of the

Restigouche River, on a calcareous limestone ledge in a linear habitat of approximately

200 m by 2 m. These limestone ledges are uncommon, and few such sites are known in

New Brunswick. The other two study species are not known from this site.

Within each population, 6–31 adult genets were selected; one ramet was sampled per genet. Samples were selected either (1) along three 25 m transects placed strategically across continuously occupied and available habitat when the area was extensive (Burnt

Cape, Cooks Harbour, Sandy Cove and Port-au-Choix, Fig. 2.2); or (2) by walking the entire area without a transect, when the total occupied area was too small to place three transects (Bellburns, Tablelands, Mount Logan, Restigouche, Wilson Brook, Torngat and Nain). Because D. integrifolia is a small, woody perennial with many ramets to each genet, care was taken to sample one ramet per individual by either visual verification

(clearly isolated clumps) or by gently prying around stems to ensure that sampled ramets were not connected to each other. In all cases, sampled genets were > 1 m apart. The growth forms of A. parviflora and V. uliginosum meant that individuals were clearly distinguishable. At least five leaves from each genet were dried in silica gel powder and preserved in separate envelopes until genetic analysis. The latitudinal and longitudinal coordinates of every sampled plant were recorded, except at Torngat and Nain where only population coordinates were recorded.

Low sample sizes are problematic in genetic studies as they can distort the levels of genetic variation found (i.e. allele frequencies, unique alleles). My initial sample sizes were somewhat low (n= 6–31 adults were sampled) but this was exacerbated by failure of some in the DNA extraction process and PCR (polymerase chain reaction) procedure in the AFLP (Amplified Fragment Length Polymorphism) process. Between 2–8 for A. parviflora, and 4–11 samples per population for D. integrifolia and V. uliginosum were retained. While Reisch & Bernhardt-Römermann (2014) found that sample size (n ranging from 2 to 400 of individuals and populations) and genetic diversity were not significantly correlated in 115 AFLP studies, low sample sizes are still expected to affect within and among population genetic diversity measures. In order to determine the effect of the low sample sizes on the data, I used correlations of sample size to various genetic metrics (2.3.1).

At each site, photos were taken during sampling; this allowed for side-by-side comparisons of conditions across sites. A disturbance category was assigned to each site using these photos in combination with information gathered during site visits, documentation on the sites, and personal communication with site managers and / or researchers. Disturbance categories were based on recent (visual) and historic activities that would have caused habitat destruction and / or fragmentation. Disturbed populations showed on-going activities such as active roads, trampling (visible disturbance, trails, crushed plants), and gravel removal. Recovering sites were historically disturbed (gravel removal or road building, 50 to 100 years ago), but now protected (ecological reserves,

National Historic Site, National Park) and show no recent activity such as visible disturbance, trails, or crushed plants. Undisturbed populations showed no signs of recent or on-going disturbances, had no known historic disturbances, and no roads or trails located within 15 m. Natural disturbance, such as erosion or animal activity, was not considered.

2.2.4 Analysis of genetic variation

2.2.4.1 DNA extraction

Approximately 20 mg of dried leaf material was frozen in liquid nitrogen and ground twice (40 and 30 s intervals at 5 m/s) to a fine powder in a FastPrep-24™ bead beater

(MP Biomedicals Inc., Santa Ana, CA) using five mm glass beads (Kimble Chase Life

Science and Research Products, LLC, Rockwood, TN). DNA was extracted using a

NucleoSpin® Plant II kit (Macherey-Nagel Inc., Bethlehem, PA) following the manufacturer's instructions with two exceptions: cell lysis was increased to 60 min, and

60 µl of 1% PVP (polyvinylpyrrolidone) was added to the 400 µl lysis buffer with 10 µl of RNase to eliminate polyphenols and improve DNA yield.

DNA quality was verified on 0.8% agarose gels stained with SybrSafe and visualized on a VersaDocTM imager (Bio-Rad Laboratories (Canada) Ltd., Mississauga, ON; for example, see Fig. 2.3a). DNA concentration and purity were measured on a NanoDrop

1000 spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA). Samples with concentrations lower than recommended in the AFLP protocol were concentrated using

a Savant DNA 120 Speedvac Concentrator (Thermo Scientific Inc., Waltham, MA).

DNA was stored at –20 °C until use.

2.2.4.2 AFLP genotyping

I used Amplified Fragment Length Polymorphism (AFLP) to detect genetic polymorphisms. AFLPs are widely distributed throughout the genome, are comprised largely of non-coding (neutral) DNA, and are dominant (Meudt & Clarke 2007). The

AFLP process entails four steps: (a) DNA restriction digest and fragment ligation to adapters (bands), (b) fragment amplification (in duplicate) using a polymerase chain reaction (PCR), (c) gel analysis of the amplified fragments, and (d) scoring and interpretation (Vos et al. 1995). AFLP analysis requires no prior knowledge of a species' genome and it is highly sensitive to detecting polymorphisms in DNA sequences (Paun

& Schoenswetter 2012). In addition, it can generate a large number of fragments and is likely to produce a unique genetic profile for each individual, with presence or absence of an allele (variant of a gene) identified in a binary matrix (Meudt & Clarke 2007).

However, AFLP analysis is labor intensive and requires high quality DNA. Because the alleles are dominant, it is not possible to separate homozygous and heterozygous genotypes with several resulting weaknesses: information per locus is poor, bands that belong to different loci may co-migrate, fragments of different genomic origin may appear as the same bands (homoplasy), and there can be a lack of reproducibility. Most of those weaknesses can be addressed by using strict criteria for the selection of bands used in the genetic analysis.

In addition to their widespread use in plant population genetic studies (Reisch &

Bernhardt-Römermann 2014), AFLPs are also used for phylogeographical studies

(Stachurska-Swakon et al. 2013), to identify loci (location of a DNA sequence) under selection (Magdy et al. 2015), specifically those of ecological relevance (Manel et al.

2012), and to determine clonal diversity (De Witte et al. 2012) and inbreeding levels

(Chybicki et al. 2011). In this study, I refer to DNA fragments as bands.

I followed the Invitrogen Instruction Manual for AFLP® Analysis System I and AFLP®

Starter Primer Kit (Life Technologies™, Thermo Fisher Scientific Inc., Waltham, MA) protocol to digest and ligate approximately 125 ng of DNA. The following exceptions were applied: reaction volumes were halved, the digestion incubation period was increased to 2.5 hours, the final ligated product was diluted in ultrapure H2O (Gibco), and DNA was re-suspended in ultrapure H2O. Products were stored at –80 °C. Pre- amplification and selective reactions followed the AFLP Plant Mapping Protocol by

Applied Biosystems (Life Technologies™, Thermo Fisher Scientific Inc., Waltham,

MA). Five DNA primer combinations (IDT Integrated DNA Technologies, Coralville,

Iowa; or Life Technologies, Thermo Fisher Scientific Inc., Waltham, MA) were selected from the literature and directly from other researchers (Skrede et al. 2006, Eidesen et al.

2007, De Witte et al. 2012, J. McEwan, pers. comm.).

After testing primers through the selective amplification step, three primer combinations in D. integrifolia and V. uliginosum, and two for A. parviflora were retained based on successful amplification (Table 2.3). Selective amplification products were sent to 67

The Centre for Applied Genomics (TCAG; Toronto, Canada) for fragment analysis

(determination of band presence / absence per individual) on a capillary sequencer

(Applied Biosystems 3730xl, Thermo Fisher Scientific Inc., Waltham, MA). At every step, digestion and PCR products were verified on 1.5% agarose gels stained with

Sybr®Safe (Thermo Fisher Scientific Inc., Waltham, MA; Fig. 2.3b and c). Resulting data included genotype information for each individual by primer pair combination including all replicates. After cleanup of the data (described below), the final datasets consisted of multi-locus genotypes for the following: D. integrifolia: 73 individuals from nine populations (total bands = 254); V. uliginosum: 80 individuals from nine populations (226 bands); A. parviflora: 35 individuals from seven populations (166 bands).

The quality of the amplification products was assessed using Peak Scanner™Software v1.0 (Thermo Fisher Scientific Inc., Waltham, MA). Band presence / absence data were filtered using RawGeno v. 2.0 (Arrigo et al. 2009) to retain only band sizes between 50 and 550 bp with a minimum fluorescence intensity of 50 relative fluorescence units

(rfu), the minimum meaningful peak height (Arrigo et al. 2009). Bins (categories of fragments) that could not be replicated across samples were removed, because they are likely artifacts of the equipment or background noise (Bonin et al. 2004). Based on the number of bands identified across the data set for individuals, the final binary matrix

(band presence / absence x individual samples) was derived from the raw data matrix in a series of steps.

(a) Samples (individual genotypes) were rejected if they showed weak amplification,

(i.e., produced < 10% of the mean number of bands identified as present in that population; N. Arrigo, pers. comm.);

(b) Using only samples that were replicated in a separate matrix (from all steps in the process, including replicated DNA extraction for the same individual, pre-amplification, and amplification; and within plate and among plate replication), bands were rejected if they consistently showed > 10% inconsistencies (mismatches) among replicated samples

(Skrede et al. 2006);

(d) Where replicate samples remained, a single replicate was selected for the final data matrix as follows: if two out of three replicates of a sample presented the band, then the individual replicate that presented with the most consensus bands was selected. If duplicated amplifications were inconsistent (i.e., a band was present in one replicate but absent in the others), then the original fragment analysis files (in Peak Scanner™) with the highest quality of the fragment peak for the present band was selected. A band was assessed as present if the peak was > 100 relative frequency units (rfu). Categories of bands with a high rfu more likely represent a consistent presence signal than those with a lower rfu, which could include background noise (Holland et al. 2008).

(e) From the resulting data matrix containing one sample for each individual, six bands were excluded because they were monomorphic. See Appendix 2.1 for an example

(excerpt) binary matrix used in the genetic analysis.

To determine error rates for each primer pair (Table 2.3), the ratio of observed number of differences in band presence / absence across all replicates and total number of comparisons, after the cleanup process as described above, was calculated as 2.26%,

2.89% and 2.71% for D. integrifolia, A. parviflora and V. uliginosum respectively; < 5% is identified as an acceptable error rate for AFLPs (Bonin et al. 2004, Pompanon et al.

2005, Bonin et al. 2007, Crawford et al. 2012). Seven, 11 and five DNA extractions were repeated for each species, D. integrifolia, A. parviflora and V. uliginosum, respectively, and duplicates within and among plates were also included (47 and 22, respectively for D. integrifolia, 8 and 25 for A. parviflora and 17 and 20 for V. uliginosum) for an overall replication rate of 32%, 47% and 15% for D. integrifolia, A. parviflora and V. uliginosum respectively, which are all above the recommended lower limit of 5 to 10% (Bonin et al. 2004).

Given concern over low and unequal sample sizes for some populations, I calculated two genetic diversity measures using rarefaction (Nei's gene diversity D and Allelic richness

Pb, Appendix 2.2); both were highly correlated with all other measures of genetic diversity. In addition, I calculated the correlation coefficients for all genetic metrics

(genetic diversity and genetic differentiation, PHI'PT (Φ'PT) to determine their sensitivity to sample size (Appendix 2.4). When all three species were included, the number of 70

individuals sampled per population was significantly correlated only with percent polymorphic loci (%P), indicating that %P was not a reliable measure for small sample sizes. However, this appeared to be driven by the inclusion of the extremely small samples sizes for Anemone parviflora; when it was removed, %P was also no longer correlated with the number of individuals sampled per population. I report %P for all species, but Appendix 2.2 provides results for all other metrics.

2.3 Data Analysis

2.3.1 Genetic diversity and differentiation

Although missing data in AFLP datasets are not unusual (e.g., where binary genotype data for an individual do not exist for all primer combinations used in an analysis), they can be problematic for pairwise distance-based analysis (Peakall & Smouse 2006). To deal with missing data (4%, 7% and 0% for A. parviflora, D. integrifolia, and V. uliginosum respectively), which occurred randomly throughout the dataset, I interpolated missing data (GenAlEx v. 6.502) as recommended by Peakall & Smouse (2006), calculating average genetic distances (across all non-missing individual distances) at the locus for the missing pairwise population contrast.

2.3.1.1 Genetic diversity within populations

The percentage of variable bands in a population (%P), one of the most commonly used estimators of genetic diversity within populations (Reisch & Bernhardt-Römermann

2014), was calculated from the final binary matrix of band presence / absence (see Table

2.4). A one-sample t-test was performed to assess significant differences from the mean using the stats package (R Core Team 2015). I also report percent polymorphic loci at

1%, Shannon's information index, effective number of alleles, expected heterozygosity, allelic richness (using rarefaction to account for unequal sample sizes), and Nei's gene diversity (using rarefaction) for comparison with the literature (see Table 2.4 for calculations, and Appendix 2.2). The Pearson correlation coefficients between all pairs

of metrics were calculated using the psych package (Revelle 2016). Analyses were carried out in R v. 3.2.2 (R Core Team 2015).

2.3.1.2 Genetic differentiation and structure among populations

To assess pairwise genetic differentiation, I calculated a dissimilarity matrix of pairwise individual differences in band presence / absence in Euclidean space (GenAlEx v.

6.502). Using the genetic distance matrix, genetic differentiation among all pairs of populations was estimated using PHI'PT (Φ'PT) in GenAlEx v. 6.502 (Peakall & Smouse

2006, 2012). Φ'PT is calculated through Analysis of Molecular Variance (AMOVA) by dividing the estimate of variance among populations by the estimate of variance among populations plus estimate of variance within populations. Φ'PT is an analog of FST for dominant markers. Critical Φ'PT values were categorized (Hartl & Clark 1997): < 0.05

(little genetic differentiation), 0.05–0.15 (moderate genetic differentiation), 0.15–0.25

(high genetic differentiation), and >0.25 (very high genetic differentiation). To test the null hypothesis that populations were not differentiated (i.e., they experience high contemporary gene flow and / or share genotypes such that Φ'PT = 0), permutation analysis was used to generate a distribution of Φ'PT values from 999 random permutations of the data. The null hypothesis was rejected if the observed Φ'PT value was greater than 95% of the Φ'PT values generated from the randomized data.

2.3.1.3 Statistical tests analyzing genetic diversity and differentiation

The spatial structure of genetic variation was evaluated using AMOVA with the default

settings (in the Excel add-in GenAlEx v. 6.502) (Peakall & Smouse 2006, 2012). The analysis partitioned genetic variation into three a priori components. Specifically, the percent of molecular variance allocated to within populations, among populations within regions, and among regions.

Major groupings in genetic variation based on the genetic distance matrix were visualized using Principal Coordinate Analysis (PCoA; also known as multidimensional scaling, MDS, using the genetic distance metric) in GenAlEx v. 6.502 (Peakall &

Smouse 2006, 2012). This eigenvalue technique uses matrices of dissimilarities and is particularly well suited to distance matrices. Successive PCoA axes represent percentages of variation explained using the standardized covariance default setting, which converts the distance matrix into a covariance matrix (Peakall & Smouse 2006,

2012). Only two axes were retained, as subsequent axes explained considerably less variability. Centroid plots were created using mean scores on axes 1 and 2 of the PCoA, with ellipses indicating ± SE based on genetic distances.

The genetic structure of the D. integrifolia data set was further assessed based on the binary matrix of band presences / absence derived from AFLP analysis. Cluster analysis was carried out using the Bayesian approach implemented in STRUCTURE v.2.2.3

(Pritchard et al. 2000, Falush et al. 2003, Falush et al. 2007, Pritchard et al. 2010), assigning individuals to clusters (also termed putative ancestral lineages) according to allele frequency differences by analysis of likelihood (i.e., individuals are placed into groups that share similar patterns of variation; Porras-Hurtado et al. 2013). Because 74

the degree of admixture in the species is unknown, the no-admixture model was used as recommended (Pritchard et al. 2010). STRUCTURE analysis used correlated allele frequencies with a burn-in period of 50,000 with 100,000 Markov Chain Monte Carlo

(MCMC) iterations, with 10 replicates for 1–10 putative clusters (K) (Evanno et al.

2005, Earl & vonHoldt 2012). The most likely number of genetic clusters (K) was selected using the ΔK method (Evanno et al. 2005) implemented in the online program

Structure Harvester (Earl & vonHoldt 2012) assuming prior values of K between 1 and

10. In addition, best K was verified by calculating ln(K) and ΔK using Clumpak

(Kopelman et al. 2015). The results were visualized using Structure Plot (Ramasamy et al. 2014). No STRUCTURE analysis was carried out for Anemone parviflora due to low sample sizes of individuals per population for this species, or for Vaccinium uliginosum, because the PCoA for this species showed no genetic structuring.

2.3.2 Distance, disjunction and levels of disturbance

First, to test the assumption that genetic distance (differentiation) among individuals increases with the geographic distance between them, the correlation between individual pairwise genetic and non-transformed geographic distance matrices was calculated using a Mantel test, based on 999 permutations using the standard permute option in GenAlEx v. 6.502 (Peakall & Smouse 2006, 2012). The genetic distance matrix was calculated using linear genetic distances as recommended by Blyton & Flanagan (2012), and the geographic distances were generated from latitude and longitude coordinates, using the

“haversine” formula (equation that gives great-circle distances between two points on a

sphere), for individual plants (except populations at Nairn and Torngat, where only population coordinates were available).

Second, to test for a trend of decreasing genetic diversity within populations and increasing genetic differentiation among pairs of populations towards each species' range edge, the association of mean population genetic diversity and mean pairwise genetic differentiation with population distance from edge (in km, edge = 0, disjunct negative, and positive values towards core, Table 2.2) was assessed using analysis of variance (ANOVA, all populations). Predictor variables included distance to edge and species, as well as their interaction term. Species were also analyzed individually, and because genetic differentiation can be positively associated with maximum geographic distances (Reisch & Bernhardt-Römermann (2014), each analysis was run with and without Torngat and Nain (the most distant core populations).

Third, a multi-factor ANOVA was used to test the association of mean population genetic diversity and mean pairwise among population differentiation with population disjunction and habitat disturbance (all populations). Predictor variables included disjunction, disturbance and species, as well as their interaction terms. Population disjunction type was coded as a categorical binary variable (0 = disjunct, 1 = edge / core), with all disjunct populations occurring at a distance of 100 km or more beyond the range edge, or in isolated habitat patches, as defined above. Disturbance type was coded as a categorical variable (0 = undisturbed, 1 = recovering, 2 = disturbed).

The Pearson correlation coefficients between pairs of variables were calculated for several other potential predictor variables that could be associated with isolation

(population size, habitat area, and plant density) using the psych package to assess if there is a linear relationship between variables (Revelle 2016). These variables were not further considered because they were all highly correlated with disjunction (r > 0.69, P <

0.001).

Model residuals were assessed using the Bartlett test for homogeneity of variances, by visually inspecting plots of standard residuals, normal quantile of the residuals (normal distribution), scale-location (identifying heteroscedasticity), and leverage and Cooks

Distance (leverage and influential cases). All analyses were carried out in R v. 3.2.2 (R

Core Team 2015) using the stats package, and agricolae for the TukeyHSD tests (de

Mendiburu 2016).

2.4 Results

2.4.1 Genetic diversity and differentiation

Percent polymorphic loci (%P) were highly correlated with all other measures of within- population genetic diversity for all three species (Pearson correlation coefficients for all pairs of metrics ranged from r = 0.51 to 1.0, with a mean r = 0.82; all P < 0.04;

Appendix 2.2). Overall, all three species showed similar patterns of within population genetic diversity across their respective ranges sampled and in disjunct populations

(Table 2.5). For example, genetic diversity in disjunct populations was not lower than that detected within each species' core and range edge and genetic diversity did not decline beyond the continuous range edge. Highest genetic diversity was detected in core populations, and unique (private) bands were detected in several populations across the sampled range (Appendix 2.2). In contrast, patterns of genetic differentiation, and the PCoA results, varied dramatically among species. In the PCoA for example, two disjunct populations of A. parviflora (Wilson Brook and Restigouche) were clearly separated from all other populations sampled for that species (Fig. 2.4), while D. integrifolia genetic clusters were not associated with the populations and showed considerable mixing among populations (Fig. 2.5). No genetic clustering was detected for V. uliginosum (Fig. 2.6). Detailed results for each species follow.

2.4.1.1 Anemone parviflora

A significant number of A. parviflora samples failed to amplify, such that only 35

individuals from eight populations were successfully genotyped. Results inferred from this low number of individuals / population results in reduced statistical power and should be interpreted cautiously. In addition, one primer combination was entirely removed as too few loci for reliable genotyping were detected. However, some inferences can still be made. In contrast to expectations, partitioning of genetic variation in AMOVA indicated that the greatest variation occurred within populations (84%), with only 1% occurring between the two sampled regions, edge and disjunct (Table 2.6). The highest %P was detected in Port-au-Choix (within the range edge), where 67% of bands

(out of 130) were polymorphic (Table 2.5, t = -4.35, P = 0.003, Appendix 2.3). Diversity in edge populations ranged from 33 to 67 %P (n = 5 populations). Wilson Brook, the most geographically isolated population, at 58 %P (t = -2.87, P = 0.024), was comparable to Port-au-Choix at 34 %P, and higher than average %P across all populations (41 %P). The disjunct population in Restigouche contained the lowest genetic diversity of all populations (10 %P, t = 5.0, P = 0.002).

Few estimates of pairwise differentiation of edge populations were significantly greater than zero. However, those including at least one of the two disjunct populations at

Wilson Brook and Restigouche were frequently significant, with moderate to high levels of Φ'PT from other populations (range from 0.115 to 0.454, Table 2.7a) suggesting these two populations were generally differentiated from other populations more than continuous range-edge populations were differentiated from each other. Mean genetic differentiation among edge populations was Φ'PT = 0.091, and among disjunct

populations was 0.238. Unique bands were observed in Wilson Brook, Tablelands, Port- au-Choix and Burnt Cape (2, 1, 2, 1 unique bands respectively).

STRUCTURE analysis was not carried out for this species due to the very low sample sizes in some populations. The first two axes of the PCoA explained 48% of variation in genetic diversity (Fig. 2.4a). The first axis showed considerable spread and overlap (Fig.

2.4b) and Wilson Brook and Restigouche were clearly separated from other sites on the second axis.

2.4.1.2 Dryas integrifolia

Only 7% of variation was attributable to among-region differences (AMOVA, Table

2.6), however this was almost entirely due to differentiation from Torngat (the most northern core population). When the AMOVA analysis was repeated without the

Torngat population, no among-region variation remained, while among population variation (17%) remained similar to the previous analysis, and within-population variation continued to dominate (83%). As expected, highest %P was detected in

Torngat, where 70% of bands (out of 144) were polymorphic (Table 2.5, t = -4.92, P =

0.001, Appendix 2.3). The five edge populations ranged from 32 to 56 %P. Wilson

Brook, the most geographically isolated population, was comparable at 45 %P (t = 0.25,

P = 0.807) to a number of Newfoundland populations within the range edge, and comparable to the average %P across all populations. Lowest genetic diversity was observed within the range edge at Bellburns (32 %P, t = 2.05, P = 0.019), as well as in

the disjunct population at Tablelands (31 %P, t = 3.15, P = 0.013) in Newfoundland.

Most estimates of pairwise differentiation between populations were significant, with moderate levels of Φ'PT (0.15 to 0.25, Table 2.7b). A number of population comparisons showed high levels of differentiation (> 0.25), but four (Burnt Cape, Sandy Cove, Cooks

Harbour and Port-au-Choix) were not significantly differentiated (Φ'PT ranging between

0.0 and 0.032; all P > 0.21). As expected, Torngat was significantly differentiated from all other populations (Φ'PT = 0.119 to 0.340). Disjunct Wilson Brook was also differentiated from all other populations (0.098 to 0.340) except Cooks Harbour (Φ'PT =

0.076, P = 0.076) and Sandy Cove (Φ'PT = 0.151, P = 0.089). Surprisingly, disjunct

Mount Logan was not significantly differentiated from the edge populations Burnt Cape

(Φ'PT = 0.000, P = 0.369), Cooks Harbour (Φ'PT = 0.131, P = 0.052) or Sandy Cove (Φ'PT

= 0.144, P = 0.067) despite an average distance of 829 km from Mount Logan to the other three. Mean genetic differentiation among core and edge populations was Φ'PT =

0.123, and 0.245 among disjunct populations. Unique bands were observed in Wilson

Brook, Mount Logan, Tablelands, Bellburns, and Torngat (2, 1, 1, 2, 4 unique bands respectively).

The first two axes of the PCoA explained 62% of variation in genetic diversity (Fig.

2.5a). A plot of centroids (mean scores ± SE, Fig. 2.5b) shows genetic separation of

Torngat on the first axis. Port-au-Choix and Wilson Brook were very similar to one another, as was Cooks Harbour to Sandy Cove. Bellburns and Tablelands were also very similar, with low variability, and clearly separated from all other populations on axis 81

2. Wilson Brook was generally closer to all Newfoundland populations than to Mount

Logan or Torngat (Fig. 2.5a).

Mean K values from the Structure analysis using the ΔK method (Evanno et al. 2005) identified one peak at K = 3 (Appendix 2.5). This was confirmed by Best K in Clumpak

(Kopelman et al. 2015) with K = 3 (Prob(K = 3) = 1), while the probability of all other values was < 0.0009. The analysis assigned 69 of 73 individuals to one of three putative ancestral lineages (Fig. 2.6) with 90 – 100% probability, indicating that genetic admixture across the entire sampled region is low for this species. Interestingly, individuals from the disjunct populations at Mount Logan and Wilson Brook were primarily assigned to different ancestral lineages (i.e., > 50%). Individuals from the core

Torngat population and the two northern Newfoundland populations (Burnt Cape, Cooks

Harbour) were assigned to varying degrees to three putative lineages. In the other four

Newfoundland populations (Sandy Cove, Port-au-Choix, Bellburns and Tablelands) individuals were assigned to two lineages each, with all three lineages present, yet with only one lineage common to all four populations.

2.4.1.3 Vaccinium uliginosum

Virtually all variation for this species (99%) was within populations, with no detectable structure among populations (Table 2.6). As with the other species, highest %P was detected in Torngat, where 70% of bands (out of 184) were polymorphic (Table 2.5, t = -

3.77, P = 0.005, Appendix 2.13). Edge populations ranged from 39 to 66 %P (n = 5, number of populations). The disjunct population at Mount Logan was 66 %P (t = - 82

2.25, P = 0.054), which was similar to or higher than the %P detected in all other populations. Lowest genetic diversity was observed in the range-edge population at

Cooks Harbour (39 %P, t = 5.62, P < 0.001).

There was very little pairwise differentiation (Φ'PT) among populations (Table 2.7c), with all Φ'PT < 0.05. However, Bellburns was significantly differentiated from most other populations, and Tablelands was significantly differentiated from Sandy Cove.

Unique bands were observed in Mount Logan, Bellburns, Port-au-Choix and Torngat (1,

1, 2, 1 unique bands respectively).

The first two axes of the PCoA explained only 8.8 % of variation in genetic diversity

(Fig. 2.6a), which is reflected in the plot of centroids (mean scores ± SE) (Fig. 2.7b) showing little separation of populations on either axis. The primary axis of the centroid plot appears to separate Torngat and Nain, but the plot of individual loadings shows a fair amount of overlap in these two populations. The secondary axis separated Cooks

Harbour, Mount Logan and Torngat, but all populations showed considerable overlap

(Fig. 2.7a). STRUCTURE analysis was not carried out, given results of the PCoA analysis.

2.4.2 Association among genetics, distance, disjunction and levels of disturbance

My analysis did not reveal any overall associations between mean population genetic diversity with geographic distance or site disturbance across all species. However, mean population genetic differentiation varied among disjunct and other core and range-edge

populations, with disjunct populations showing higher levels of genetic differentiation.

Population genetic differentiation varied among species and was significantly associated with disturbance for two of the three species, with undisturbed sites showing higher mean population genetic differentiation than both recovering and disturbed sites.

Populations more distant from the edge showed higher levels of genetic differentiation for only two species.

2.4.2.1 Isolation by distance

There was no evidence of isolation by distance in A. parviflora (Mantel test, Fig. 2.8a,

R2 = 0.001, P = 0.325). In contrast, there was evidence of isolation by distance for D. integrifolia (Fig. 2.8b, R2 = 0.187, P < 0.001) and V. uliginosum (R2 = 0.01846, P =

0.010), albeit weak in the latter case (Fig. 2.8c).

Multi-factor ANOVA, including all populations (diversity ~ distance x species), revealed that mean population genetic diversity did not vary among species (F2,20 = 2.16,

P = 0.142), distance from edge (F1,20 = 2.16, P = 0.158), or their interaction (F2,20 = 0.09,

P = 0.915). Genetic differentiation, however, differed significantly among species (F2,20

= 19.13, P < 0.001, Fig. 2.9b), and with distance from range edge (F1,20 = 11.32, P =

0.003). There was no significant interaction of species and distance from edge (F2,20 =

2.57, P = 0.101). When species were analyzed separately, ANOVA of mean population genetic differentiation and distance from range edge including all populations / species showed no significant association between distance from range edge and genetic differentiation for A. parviflora (F1,6 = 4.64, P = 0.075). Populations of D. integrifolia 84

and V. uliginosum showed no association of genetic differentiation with distance from edge (F1,7 = 0.01, P = 0.993, F1,7 = 0.04, P = 0.866, respectively).

When core populations were excluded from the analysis for D. integrifolia and V. uliginosum, ANOVA revealed association between distance from edge and neither genetic diversity (D. integrifolia: F1,6 = 1.55, P = 0.260, V. uliginosum: F1,5 = 3.50, P =

0.120) nor genetic differentiation (F1,5 = 0.001, P = 0.978, F1,6 = 5.58, P = 0.056 respectively). Populations more distant from the edge showed higher levels of genetic differentiation for A. parviflora, and D. integrifolia (Fig. 2.9 and 2.10 respectively) but not for V. uliginosum when only edge and disjunct populations were considered (Fig.

2.11).

2.4.2.2 Disjunction

ANOVA showed that genetic diversity did not vary significantly among species (F2,21 =

2.99, P = 0.072), and there was no interaction of species and disjunction (F2,17 = 0.50, P

= 0.613). Genetic diversity did not vary significantly between population type (disjunct or continuous edge / core) (F1,21 = 1.21, P = 0.285, Fig. 2.12a). Genetic differentiation differed significantly among species (F2,21 = 25.70, P < 0.001) and disjunction (F1,21 =

6.03, P < 0.023, Fig. 2.12b) with no interaction (F2,17 = 2.85, P = 0.086). Disjunct populations showed higher levels of mean population genetic differentiation than populations from the range edge and core combined for A. parviflora and D. integrifolia, but not for V. uliginosum.

2.4.2.3 Disturbance

Analysis of variance (ANOVA) revealed that genetic diversity did not vary significantly with disturbance levels (F2,21 = 3.77, P = 0.066, Fig. 2.13a), or species (F2,21 = 3.0, P =

0.072), and there was no interaction among species and disturbance levels (F2,17 = 0.56,

P = 0.581). Genetic differentiation varied significantly with species (F2,21 = 25.70, P <

0.001) and with disturbance levels (F2,21 = 6.86, P = 0.016, Fig. 2.13b). There was no evidence of an interaction (F2,17 = 2.78, P = 0.091). Undisturbed sites showed higher mean population genetic differentiation than both recovering and disturbed sites for A. parviflora and D. integrifolia, but not for V. uliginosum.

2.5 Discussion

This study is the first to evaluate genetic variability of rear range-edge, disjunct populations of three arctic-alpine species (Anemone parviflora, Dryas integrifolia and

Vaccinium uliginosum) in Atlantic Canada, and to relate genetic patterns to measures of geographic isolation. It was motivated by concerns that small, disjunct, rear-edge populations may be associated with low levels of genetic diversity, making them vulnerable to climate change and stochastic events (Hampe 2005, Hampe & Jump 2011).

Furthermore, if these populations are locally adapted and contain unique genotypes, their extirpation could reduce the overall ability of the species to respond to global climate change (Rehm et al. 2015).

Contrary to some predictions of lower within-, and greater among-population genetic variation (Hamilton & Eckert 2007, Pandey & Geburek 2010, Meeus et al. 2012,

Nagamitsu et al. 2014), I found that genetic diversity in all three species remained high at range edges, and in some disjunct populations, relative to the total sample. Only D. integrifolia and A. parviflora displayed within and among population genetic structuring at the range edge, which supports the hypothesis that disjunct populations may not follow any presumed pattern due to complex interactions of life history traits, historical processes and landscape configuration. The lack of population genetic structuring in V. uliginosum may be attributable to its high dispersal capacity (ornithochory), indicating that patterns in disjunct populations may be species- and population- specific. Based on this multi-species framework, I found that distance, disjunction and disturbance

(indicators of isolation) were weak predictors of genetic variability at the rear range edge for these species.

Below I summarize genetic patterns, first across the landscape and then among species, and speculate on the role of historical factors, life history traits, and landscape configuration as possible drivers for range-edge genetic patterns. I also discuss the utility of three indicators of isolation (distance, disjunction and disturbance) as predictors of genetic variability patterns at the rear range edge in disjunct populations of widespread species.

2.5.1 Genetic variability patterns at the rear range edge

Range theory predicts that many species should display a pattern of decreased genetic diversity and increased differentiation at their rear range edge (Lesica & Allendorf 1995,

Hamilton & Eckert 2007, Eckert et al. 2008, Vakkari et al. 2009, Diekmann & Serrao

2012, Meeus et al. 2012, Wang & Bradburd 2014), which is supported by half of all studies included in a recent review (Pironon et al. 2016). Isolated populations are expected to lose genetic diversity over time due to genetic drift, and subsequently show increased genetic differentiation among populations. My findings show neither loss of diversity, nor increase in differentiation, for two of the three study species, and demonstrate considerable maintenance of genetic diversity in a number of disjunct populations. My results are consistent with those for a number of species in the literature, providing further evidence that many species follow a more complex pattern.

For example, no trend was detected for Saccorhiza polyschides, a large brown 88

seaweed in SW Europe along a latitudinal gradient towards the range edge (Assis et al.

2013), at the rear range edge of Alnus glutinosa, a European tree species (Lepais et al.

2013), nor in disjunct populations of Anacamptis pyramidalis, an orchid occurring in

Europe and North Africa, compared to core and edge populations (Ilves et al. 2016).

Higher genetic diversity and unique markers were found in disjunct populations of the arctic graminoid Puccinellia phryganodes (Kvist et al. 2015), unique genetic diversity was found in disjunct populations of the South Asian tree Shorea robusta (Pandey &

Geburek 2010), and no increase in differentiation was found in disjunct populations of the shrub Vaccinium vitis-idaea in Japan (Ikeda et al. 2015). High levels of genetic diversity have also been reported in relict and refugial populations that have persisted after deglaciation (reviewed by Hampe 2005, Lepais et al. 2013, Pironon et al. 2015). It appears predictions of decreased genetic diversity and increased differentiation do not necessarily apply to range edges that include disjunct, relict or refugial populations.

In addition, and despite differences in dispersal mechanisms, mating system, life form for two species, and longevity, I expected similarities in genetic variability patterns among the three species based on their shared southern geographic range locations and possible shared historical processes (Westberg & Kadereit 2009). All three species are thought to have arrived in New Brunswick through re-colonization after the Last Glacial

Maximum (LGM), given the macro fossil evidence for D. integrifolia and V. uliginosum

(Mayle & Cwynar 1995), and data on local cover of the Laurentide Ice Sheet 23 ka BP to 18 ka BP (Dyke 2005). Despite some similarities, my results indicate that the three

species differ in some important ways in their genetic patterns across the rear range edge. Published comparisons of other closely related or sympatric species have shown similar differences. For example, different patterns of genetic variability have been observed among three violet (Viola) species in similar habitats and range locations in central Europe (Eckstein et al. 2006), in rear-edge populations of the arctic-alpine species Ranunculus pygmaeus and Saxifraga cernua (Bauert et al. 2007) in Svalbard and northern Sweden, in three semi-grassland species (Briza media, Trifolium montanum and

Ranunculus bulbosus) along an altitudinal gradient in the European Alps (Hahn et al.

2012), and in three snow-bed herb species in northern Japan (Peucedanum multivittatum,

Veronica stelleri, and Gentiana nipponica) (Hirao & Kudo 2004). It appears that despite similar life history traits and historical processes, genetic patterns at range edges cannot be presumed to be comparable even across closely related or sympatric species.

Differences among similar species have been attributed to differential changes to habitats (Eckstein et al. 2006), differential reduction and fragmentation during glacial periods and different breeding systems (Bauert et al. 2007), differences in flower morphology (Hahn et al. 2012) and differences in pollination and thus gene flow (Hirao

& Kudo 2004).

An increase in genetic differentiation and a decrease in within-population genetic diversity are expected when gene flow (via seeds or pollen) acts more slowly than genetic drift (Orsini et al. 2013, Sexton et al. 2014, Wang & Bradburd 2014). Data from my research suggest that individual species and populations experience different levels

of genetic isolation not simply as a result of geographic distance, but presumably due to complex and unknown interactions of biotic and abiotic factors, with some of them likely having occurred in the past. While historical processes may have caused an underlying genetic pattern across the landscape of D. integrifolia since the LGM, and which is still detectable today, such was not the case for V. uliginosum. Species' life history traits and those associated with reproduction and fitness can buffer against genetic isolation and the loss of genetic diversity (Balloux et al. 2003, Honnay et al.

2005, Aguilar et al. 2008). Greater gene flow was evident for V. uliginosum than for either D. integrifolia or A. parviflora populations, presumably due to its greater seed dispersal capacity. Abiotic factors such as landscape configuration resulting from habitat features and fragmentation could exacerbate the effects of simple geographic distance by increasing genetic isolation (Aguilar et al. 2008), and different levels of gene flow would be expected as a result of changes in landscape configuration across a species' range.

Landscape heterogeneity can create barriers to gene flow when seeds or pollen cannot disperse beyond a barrier, as may be the case in island situations (Sexton et al. 2014).

Landscape barriers may have affected gene flow among range-edge populations, especially in disjunct populations of D. integrifolia and A. parviflora. On the other hand,

V. uliginosum populations were located in more open landscapes throughout most of the sampling area.

Given the interaction of these biotic and abiotic factors, and the temporal aspect, it is not surprising that gene flow can differ across a species' range, and among populations of a

species located at similar range locations (e.g., edges) (Ellstrand 1992). Patterns of genetic variability may be particularly complex and inconsistent among disjunct populations beyond the continuous rear range edge, since some of these populations may represent an “extreme manifestation of geographic peripherality” (Hamilton & Eckert

2007). They may be occurring in heterogeneous landscapes at considerable distance from other populations, especially if they are refugial. In summary, the occurrence of a population at a rear range edge, even if disjunct or relictual, does not effectively predict its level of genetic diversity, nor its degree of differentiation from other populations.

2.5.2 Patterns of genetic variability in Anemone parviflora, Vaccinium uliginosum and Dryas integrifolia

This study sampled populations on a gradient from species' distributional cores (except

A. parviflora) across the range edge to disjunct populations. Below I summarize the landscape patterns for each species, focusing on the disjunct populations, and relate them to the biotic, abiotic, and historical factors specific to each.

2.5.2.1 Anemone parviflora

While low sample sizes limit inferences concerning genetic diversity of disjunct populations of this species in NB, some discussion is warranted. Anemone parviflora has wind-dispersed seeds and insect-dispersed pollen; although seed and pollen dispersal distance is unknown, we can assume that both may be limited by distance and landscape configuration. While no core populations of A. parviflora were sampled, populations in

the continuous range edge in Newfoundland show mixed levels of genetic diversity and low levels of genetic differentiation. Two disjunct populations ranged from lowest

(Restigouche) to second highest genetic diversity (Wilson Brook). In addition, both populations were significantly differentiated from four and six of the seven other populations. These data support the hypothesis that their very different genetic patterns are attributable to different origins. While the Wilson Brook population may be a relict population that has persisted in this unique habitat after re-colonization, no macrofossils confirm the species' presence after glacial retreat, thus further research would need to include additional populations. In particular, an analysis of uni-parentally inherited haplotypes (i.e., a phylogenetic approach) could help determine if this population is indeed a relict. In contrast, the extremely low level of genetic diversity of A. parviflora in the disjunct Restigouche population suggests establishment by only a few individuals

(i.e., a founder event). Its low diversity could be attributed to a lingering founder effect post-colonization, a lack of gene flow among populations, and / or genetic drift.

2.5.2.2 Dryas integrifolia

Dryas integrifolia is insect pollinated, and fruits are dispersed by wind, which is a moderately good mechanism for maintaining gene flow in open landscapes such as the arctic tundra (Tremblay & Schoen 1999) but may be less effective in heterogeneous landscapes. Unlike the other two species studied, D. integrifolia showed significant population genetic structuring across the landscape in terms of diversity and differentiation. Genetic diversity was highest in Torngat, the most central population.

The range edge showed both low and high levels of genetic diversity. The three disjunct southern populations (Tablelands, Wilson Brook and Mount Logan) differed dramatically: their diversity ranged from lowest at Tablelands to highest at Mount

Logan. Although Tablelands was closest to the range edge, it was significantly differentiated from all other populations.

Fossil records and cpDNA analysis suggest that D. integrifolia presumably expanded northward from refugia along the eastern coastline from the New England states, and southward from those in the Arctic and Beringia (Tremblay & Schoen 1999).

STRUCTURE analysis identified three putative ancestral lineages for D. integrifolia, leading me to propose that two migration fronts met in Newfoundland and Labrador following deglaciation (Fig. 2.14). The incomplete differentiation of populations at

Wilson Brook and Mount Logan from Newfoundland could arise from shared origins followed by subsequently limited gene flow among populations. Additional data from other known populations in Newfoundland, and northward, are required to explore this hypothesis.

Life history traits and landscape patterns are also likely to have influenced the population genetic patterns of the three disjunct populations. Genetic drift is likely to be low in D. integrifolia, a long-lived clonally reproducing woody shrub. De Witte et al.

(2012) determined the maximum age for individuals of clonally reproducing Dryas octopetala (a congener of D. integrifolia) at 500 years. Although the species displays characteristics associated with moderate dispersal (i.e., insect pollinated and wind 94

dispersed), its effective dispersal may be determined relative to the barriers created by landscape heterogeneity (Sexton et al. 2014). The small Wilson Brook population of D. integrifolia exhibited moderate levels of genetic diversity but was significantly differentiated from most other sampled populations, suggesting strong isolation by geography, habitat, or both. It is located on a regionally rare “island” of gypsum substrate within a heterogeneous, hilly mosaic landscape of forests, fields and brooks.

The barriers to seed dispersal in this context are demonstrated by the absence of D. integrifolia from an apparently similar gypsum site located within 15 km of Wilson

Brook (ACCDC 2017).

In contrast to Wilson Brook, the smaller Mount Logan population (in terms of population size and habitat expanse) exhibited a higher level of genetic diversity and a lower degree of differentiation. The Mount Logan population is currently confined to a highly exposed calcareous mountain ridge, but according to the deglaciation history of the area (Dyke et al. 2002, Dyke 2005, Shaw et al. 2006), it may have been part of a much larger and more widespread population soon after glacial retreat, when tundra vegetation was prevalent. Given that an additional population was recently reported and other unreported populations may exist in the mountain range, Mount Logan may be part of a meta-population, disjunct from the continuous range edge. Aided by air movement through the open habitat, gene flow among these sub-populations could be maintained, hence explaining high levels of genetic diversity at Mount Logan.

Disjunct populations of D. integrifolia at the Tablelands site showed very low diversity

(32 %P), with high differentiation from all other populations. The isolation of the site, its unique habitat and its presumed history suggest the relative genetic structure of this population may be the result of a founder event (Allendorf & Luikart 2012). There is evidence neither in the literature nor herbarium records (ACCDC 2017) to suggest that there were additional populations in this valley, and since the species can grow only in weathered substrate, the extent of which is very limited at this location, it appears unlikely that the population was larger in the past. The area is surrounded by toxic mantle rock in an enclosed inland glacial valley. Beyond the valley, one population is known approximately 20 km to the north, and others 50 km to the south (ACCDC 2017).

In contrast, the similarly low diversity and high differentiation of Bellburns (within the continuous range edge) may be attributable to a population bottleneck: it appears to represent the remains of a much larger population, decimated by extensive habitat fragmentation, when substrate from the site was used in road building from 1975 to 1980

(M. Burzynski pers. comm. 2013).

2.5.2.3 Vaccinium uliginosum

Vaccinium uliginosum is insect pollinated and its fleshy multi-seeded fruit are dispersed by animals, notably birds, potentially allowing for long-distance dispersal events and thus gene flow over large distances. As expected given previous studies of this species

(Alsos et al. 2005, Eidesen et al. 2007), genetic diversity within V. uliginosum populations was similar across the sampled range, with very little differentiation

revealed among populations, which. The species' genetic patterns can likely be attributed to its longevity combined with high dispersal capacity, currently and through the past, allowing for continued gene flow among populations even at considerable distances.

Ongoing gene flow may explain the lack of population genetic structure observed in this study, supported by previous findings for this species. For example, Alsos et al. (2005) attributed similar levels of genetic diversity (based on cpDNA) in formerly glaciated and un-glaciated regions throughout the species' entire distribution range to continued gene flow among populations. Low levels of genetic differentiation have been found across the entire geographic range of the species (Eidesen et al. 2007), and PCoA did not identify any clustering. Although long-distance dispersal of fruits by migratory birds, as proposed by Alsos et al. (2005) may be a rare event, it could be more common than modeling exercises have indicated (Higgins et al. 2003, Viana et al. 2013, Ikeda et al.

2015). Given that previous studies have shown low levels of genetic differentiation across the entire species' range, low levels of genetic differentiation in V. uliginosum in my study are not surprising, even among populations at a considerable distance from each other.

2.5.3 Potential predictors of genetic variability patterns at range edge

Given the evidence that patterns of genetic variability differ among species and populations at the rear edges that were sampled in this research, it would be useful to identify correlates that could be used to predict the genetic distinctness of disjunct populations. I predicted that patterns of genetic diversity and differentiation would be

associated with factors such as geographic distance, disjunction, and disturbance as an indicator of habitat fragmentation. However, I found that none of those predictors was associated with genetic diversity and differentiation for my study species.

In this study, I expected isolation by distance to be particularly pronounced for insect- pollinated and wind-dispersed species in a heterogeneous landscape, since both vectors are limited by geographic distance. Based on my research, it appears evident that geographic distance alone is not a reliable indicator of genetic isolation. This conclusion is supported by a recent review of 70 studies that identified geographic distance as a driver of gene flow in only 20% of studies of a variety of taxonomic groups, while 37% of studies found evidence that the environment determined gene flow (Sexton et al.

2014). However, it is simplistic to expect that absolute distance would affect gene flow uniformly. Dispersal vectors of pollen and seeds differ significantly among species.

Even relatively short distances may constitute a significant geographic barrier to species with limited mobility such as invertebrates, while vertebrates, particularly birds, may be able to distribute seeds over much greater distances (Viana et al. 2013).

Based on my findings, genetic diversity was not associated with disjunction, and average pairwise genetic differentiation was greater for disjunct populations of A. parviflora and

D. integrifolia. In addition to the measure of disjunction used here, actual genetic isolation must be assessed through the evaluation of dispersal and connectivity among populations, for example, by assessing the seed shadow of the two wind-dispersed

species.

Natural fragmentation was not assessed in this study, and there was no clear pattern in genetic diversity associated with the relatively recent human disturbance that was evident at some sites. Genetic differentiation was somewhat greater in undisturbed populations of two of the three species (A. parviflora and D. integrifolia). However, the effects of habitat fragmentation may only be detectable in population genetic structure after many generations (Ellstrand 1992), which, in the long-lived species used in this study, may be hundreds of years. However, the study was not designed to test the effects of disturbance and fragmentation, as the undisturbed sites were primarily found in protected areas; this may have confounded the findings.

The absence of simple genetic patterns based on distance and isolation at rear range edges that include disjunct populations suggests complex interactions of physical, biological, and temporal factors that influence gene flow, or the significance of unknown factors not addressed in this study. My findings indicate complex genetic patterns for the three study species, presumably resulting from a history of gene flow among populations. Simple predictors may not exist, because influences of interactions among factors (e.g., distance, dispersal ability, habitat fragmentation, landscape heterogeneity, founder events, and local adaptation) on gene flow are likely to be species-specific

(Durka et al. 2016), causing different levels of isolation even among populations of the same species in similar range-edge locations.

While some patterns suggest an association of disjunction with genetic differentiation, no strong predictor of genetic variability patterns in disjunct populations has emerged.

Two approaches may aid in addressing knowledge gaps. First, further research into connectivity among populations could aid in identifying limits to dispersal of the species within the landscape. Second, further research on the history of the species and whether a disjunct population might be a relict, can help identify potentially genetically unique populations. Low sample size was a weakness of this study. Future studies should include larger sample sizes (of individuals per population, and number of populations), across a wider geographical range to provide a more reliable estimate of genetic diversity, a more accurate assessment of unique alleles (private bands) and a more complete context against which to compare rear-edge disjunct populations.

2.6 Conservation implications and future research

The results of my research contribute considerably to our understanding of the complexity of evaluating the conservation significance of disjunct populations at the rear range edge. My results do not show a trend of declining genetic diversity and increasing genetic differentiation at the edge of the range in disjunct populations of Anemone parviflora, Dryas integrifolia or Vaccinium uliginosum, and provide no indication that these populations warrant special conservation attention at their southern range limits.

Population genetic structure results did not confirm that they are relict populations even though some unique bands were identified. However, these populations have other values that should be considered when making management decisions. Anemone

100

parviflora and D. integrifolia have an extensive range. It is possible that adaptation to local conditions has occurred at these or other disjunct sites. Future research should also address the question of whether disjunct populations are adapted to local conditions as many southerly genotypes have greater tolerance for warmer conditions than genotypes further north. Local adaptation could be assessed through common garden experiments using plants (or seeds) from different locations within the range. Loci that are involved in adaptation could be assessed using AFLPs or other marker systems such as single nucleotide polymorphisms (SNP's) and microsatellites, which are co-dominant and could provide more information than the dominant AFLPs (Schoville et al. 2012). Limiting factors for these disjunct populations should be assessed, including levels of isolation and gene flow among disjunct populations. Further, limits to seedling establishment in these disjunct populations should be evaluated.

101

Tables

Table 2.1. Locations of populations of three species sampled for genetic analysis at each species rear edge in Atlantic Canada.

Population Sampling locality1 Latitude Longitude Habitat description Torngat Arctic Labrador 58.45014 -62.79855 Arctic Nain Arctic Labrador 56.53336 -61.70570 Arctic Burnt Cape Burnt Cape Ecological Area, Newfoundland 51.57807 -55.74773 Coastal barrens Cooks Harbour Newfoundland, Cooks Harbour 51.54683 -55.92718 Coastal barrens Sandy Cove Newfoundland, Sandy Cove 51.34189 -56.68501 Coastal barrens Port-au-Choix Port-au-Choix NHS, Newfoundland 50.70065 -57.37788 Coastal barrens Bellburns Newfoundland, Bellburns 50.38265 -57.52446 Coastal barrens Tablelands Gros Morne NP, Newfoundland 49.47649 -57.97619 Serpentine Tablelands Mount Logan Parc National de la Gaspésie, Quebec 48.89182 -66.63934 Boreal-alpine Wilson Brook Wilson Brook PNA, New Brunswick 45.86005 -64.67611 Boreal-temperate Restigouche Restigouche River, New Brunswick 47.88178 -67.24086 Boreal-temperate

1 NHS = national historic site, NP = national park, PNA= protected natural area

102

Distance from Habitat Population Disturbance Population Species Range location edge (km) area (km2) size level Burnt Cape AP edge 182 5 5000 recovering Cooks Harbour AP edge 171 5 5000 disturbed Sandy Cove AP edge 121 5 5000 disturbed Port-au-Choix AP edge 36 5 5000 recovering Bellburns AP edge 0 5 5000 disturbed Tablelands AP disjunct -100 0.5 1000 undisturbed Wilson Brook AP disjunct -735 0.5 1500 undisturbed Restigouche AP disjunct -755 0.5 2000 undisturbed Torngat DI center 963 100000 undisturbed

Burnt Cape DI edge 182 5 100000 undisturbed Cooks Harbour DI edge 171 5 100000 disturbed Sandy Cove DI edge 121 5 100000 disturbed Port-au-Choix DI edge 36 5 100000 recovering Bellburns DI edge 0 5 100000 disturbed Tablelands DI disjunct -100 0.5 5000 undisturbed Mount Logan DI disjunct -676 0.5 3000 undisturbed Wilson Brook DI disjunct -735 0.5 50000 undisturbed 103

Distance from Habitat Population Disturbance Population Species Range location edge (km) area (km2) size level

VU center 1044 >5 >1000 undisturbed Torngat Nain VU center 821 >5 >1000 undisturbed Burnt Cape VU edge 282 5 1000 recovering Cooks Harbour VU edge 271 5 1000 disturbed Sandy Cove VU edge 221 5 500 disturbed Port-au-Choix VU edge 136 5 500 recovering Bellburns VU edge 100 5 500 disturbed Tablelands VU edge 0 1 500 undisturbed Mount Logan VU disjunct4 -576 5 500 undisturbed

1 Range locations: (1) center - where populations are common and continuous (based on range maps and occurrence records) and gene flow is presumed (between-population distance of 0–40 km, no environmental barriers apparent); (2) range edge - where habitat remains continuously distributed, but is limited, and natural populations are expected to experience limited gene flow (some barriers, and / or 40–100 km apart); (3) disjunct - populations are geographically and presumably environmentally isolated, and over 100 km from a continuous range edge. 2 Population size: is a coarse extrapolation of the # of plants or ramets observed in the sampling area over the habitat available. 3 Disturbance categories, based on visual evidence of recent and historic activities that would have caused habitat fragmentation: undisturbed populations were presumed not to have experienced habitat fragmentation events in the last 50 to 100 years; recovering populations experienced activities, such as surface vegetation removal for road building in Newfoundland, in the last 50-100 years, but the activity was clearly stopped; disturbed populations show evidence of continuing habitat fragmentation impacts, for example gravel. 4 Additional populations are found within 100 km of this site, although they were not sampled. 104

Species Total # of bands scored before Final # of Primer combinations analysis & Total # of bands, (restriction enzyme – selective removal of bands monomorphic % error % nucleotide sequence) unreliable data retained bands removed rate replication Anemone EcoR1-ACA-HEX; Mse1-CAA 170 73 49 2.67% 63% parviflora EcoR1-ACA-HEX; Mse1-CAT 186 132 117 3.11% 31% Totals 356 205 166 2.89% 47% Dryas EcoR1-ACT-FAM; Mse1-CTG 160 72 72 2.66% 21% integrifolia EcoR1-AAG-FAM; Mse1-CTA 145 114 114 1.90% 42% EcoR1-ACT-FAM; Mse1-CAT 122 74 68 2.23% 32% Totals 427 260 254 2.26% 32% Vaccinium EcoR1-ACT-FAM; Mse1-CTG 140 61 52 3.24% 13% uliginosum EcoR1-ACT-FAM; Mse1-CTA 125 71 66 2.63% 15% EcoR1-AAG-FAM; Mse1-CAG 165 121 108 2.16% 19% Totals 430 253 226 2.71% 15%

1 For all primer pairs, the EcoR1 primer was fluorescently labeled. 2 Error rates were calculated according to Bonin et al. (2004, 2007), Pompanon et al. (2005) and Crawford et al. (2012). 3 Replication of AFLP fingerprints according Bonin et al. (2004) and Pompanon et al. (2005).

105

Table 2.4. Measures of genetic variation calculated within species and populations using binary matrices in the indicated programs. Only %P is presented in the results.

Measure Calculation Acronym Software Reference Percent % polymorphic Percentage of variable bands in a population %P GenAlEx v. (Peakall & Smouse bands 6.502 2006, 2012) Percent polymorphic Percentage of variable bands in a population, where PLP 1% AFLPDiv (Petit 2005) loci at 1% the minimum frequency of the band is 1% Total number of private No. of bands unique to a single population PB GenAlEx v. (Peakall & Smouse bands 6.502 2006, 2012) Shannon’s Information I = -1* (p * Ln (p) + q * Ln(q)), p=frequency of “A” I GenAlEx v. (Peakall & Smouse Index allele, q=frequency of the “a” allele 6.502 2006, 2012) Effective number of Number of equally frequent alleles required to reach a AE GenAlEx v. (Peakall & Smouse 2 2 alleles given level of gene diversity, e, with Ne =1 / (p + q ), 6.502 2006, 2012) Expected Proportion of heterozygosity expected under random He GenAlEx v. (Lynch & Milligan heterozygosity mating, HE =2 * p * q 6.502 1994, Peakall & Smouse 2006, 2012) Allelic richness Number of alleles calculated with and without Pb(4)1 AFLPDiv (Coart et al. 2005) rarefaction Gene diversity Average proportion of pairwise differences between D(4) AFLPDat, (Ehrich 2006) individuals with confidence intervals based on bootstrapping across markers 1 Only the rarefied estimate, Pb(4 or 2), is included here. Rarefaction controls for differences in observed allelic richness that are due to differences in sampling effort, i.e., larger sample sizes have more potential to uncover a greater number of alleles (Reisch 2008) 106

Anemone parviflora Dryas integrifolia Vaccinium uliginosum

Population n %P Mean Φ' n %P Mean Φ' n %P Mean Φ' PT PT PT Torngat - - 9 70% 0.254 11 70% 0.009

Nain - - - - 10 58% 0.014

Burnt Cape 4 37% 0.134 6 56% 0.083 7 49% 0.005

Cooks Harbour 6 39% 0.164 6 39% 0.085 4 39% 0.007

Sandy Cove 2 34% 0.087 4 33% 0.093 7 51% 0.016

Port-au-Choix 5 67% 0.098 10 45% 0.131 10 66% 0.005

Bellburns 2 33% 0.183 10 32% 0.191 10 56% 0.028

Tablelands 4 46% 0.127 9 31% 0.203 11 64% 0.016

Mount Logan - - 8 65% 0.179 10 65% 0.009

Restigouche 4 10% 0.323 - - - -

Wilson Brook 8 58% 0.159 11 45% 0.198 - - -

Mean 41% 0.159 46% 0.157 56% 0.012

SD 16% 0.07 14% 0.06 9% 0.001

107

Anemone Dryas Vaccinium parviflora integrifolia uliginosum Source df % df % df % Among regions 1 1% 2 7% 2 0% Among populations within regions 6 15% 6 14% 6 1% Within populations 27 84% 64 79% 71 99%

108

New Brunswick, Canada. Φ'PT, an analog of FST, (shaded) represents the coefficient of population differentiation data (< 0.05 = little differentiation, 0.05–0.15 = moderate genetic differentiation, 0.15–0.25 = great genetic differentiation, > 0.25 = very great genetic differentiation). P-values (above the diagonal, uncorrected) for tests of significance (Ho: Φ'PT = 0 or no differentiation among pairs) were based on 999 permutations; * = significant at P < 0.05. Populations are listed from north to south.

Range Burnt Cooks Sandy Port-au- Wilson Population Bellburns Tablelands Restigouche group Cape Harbour Cove Choix Brook

(a) Anemone parviflora

Edge Burnt Cape 0.096 0.389 0.079 0.072 0.270 0.026 0.002

Cooks Harbour 0.078 0.133 0.012 0.080 0.026 0.007 0.001

Sandy Cove 0.020 0.099 0.435 0.666 0.382 0.067 0.105

Port-au-Choix 0.107 0.129* 0.000 0.565 0.037 0.033 0.009

Bellburns 0.217 0.263 0.000 0.000 0.058 0.067 0.038

Disjunct Tablelands 0.026 0.126* 0.000 0.098* 0.140 0.032 0.002

Restigouche 0.342* 0.314* 0.374 0.230* 0.454 0.329* 0.003

Wilson Brook 0.147* 0.141* 0.115 0.125* 0.205* 0.169* 0.215*

109

Burnt Cooks Sandy Port-au- Mount Wilson (b) Dryas integrifolia Torngat Bellburns Tablelands Cape Harbour Cove Choix Logan Brook Centre Torngat 0.051 0.008 0.025 0.003 0.001 0.001 0.049 0.001

Edge Burnt Cape 0.119* 0.436 0.369 0.235 0.002 0.001 0.386 0.009

Cooks Harbour 0.243* 0.000 0.373 0.319 0.006 0.004 0.052 0.076

Sandy Cove 0.221* 0.000 0.000 0.277 0.004 0.013 0.067 0.089

Port-au-Choix 0.329* 0.032 0.000 0.028 0.002 0.001 0.021 0.034

Bellburns 0.323* 0.177* 0.096* 0.102* 0.181* 0.015 0.001 0.001

Disjunct Tablelands 0.327* 0.197* 0.108* 0.097* 0.213* 0.046* 0.001 0.001

Mount Logan 0.131* 0.000 0.131 0.144 0.152* 0.329* 0.352* 0.009

Wilson Brook 0.340* 0.137* 0.098 0.151 0.114* 0.271* 0.283* 0.190*

(c) Vaccinium Burnt Cooks Sandy Port-au- Mount Torngat Nain Bellburns Tablelands uliginosum Cape Harbour Cove Choix Logan Centre Torngat 0.350 0.452 0.453 0.190 0.473 0.010 0.087 0.370

Nain 0.004 0.243 0.204 0.174 0.416 0.026 0.078 0.312

Edge Burnt Cape 0.001 0.010 0.466 0.469 0.513 0.062 0.289 0.459

Cooks Harbour 0.000 0.020 0.001 0.258 0.496 0.234 0.449 0.476

Sandy Cove 0.012 0.016 0.000 0.015 0.467 0.003 0.014 0.470

Port-au-Choix 0.000 0.003 0.000 0.000 0.000 0.029 0.281 0.240

Bellburns 0.030* 0.027* 0.024 0.014 0.046* 0.024* 0.027 0.001

Tablelands 0.018 0.020 0.007 0.002 0.038* 0.007 0.023* 0.149

Disjunct Mount Logan 0.003 0.008 0.000 0.000 0.000 0.008 0.037* 0.013

110

Figures

111

112

(a) DNA extraction (b) Pre-amplification (c) Selective amplification Lanes 1 2 3 1 2 3 4 5 6 7 1 2 3 4 5 6 7

113

114

115

116

80 70

60 50

30 20 y = 0.0008x + 43.489 R² = 0.00066 10 (a) 0 0 100 200 300 400 500 600 700 800 900 1000 140

) (b) y = 0.0261x + 43.374 120 R² = 0.18701 100

euclidean 80 ( 60

40 distance 20

Genetic Genetic 0 200 400 600 800 1000 1200 1400 1600

70 60

40 30

20 y = 0.0027x + 48.188 10 (c) R² = 0.01846 0 0 200 400 600 800 1000 1200 Geographic distance (km) Figure 2.8. Relationship (line) between genetic distance and geographic distance between individuals using the standard Mantel analysis in GenAlEx (Peakall & Smouse 2006, 2012). Anemone parviflora (a), n = 35, P = 0.325), Dryas integrifolia (b), n = 73, P < 0.001), and Vaccinium uliginosum (c), n = 80, P = 0.010), sampled in seven, nine and eight populations, respectively across the species' southern geographic range. 118

119

120

Significant differences were detected among species (F2,21 = 25.70, P < 0.001) and disjunction (F1,21 = 6.03, P = 0.023). 122

Genetic differentiation was significantly associated with disturbance (F2,21 = 6.86, P =

0.016), and differed significantly with species (F2,21 = 25.70, P < 0.001). 123

124

List of references

Abeli T, Gentili R, Mondoni A, Orsenigo S, Rossi G. 2014. Effects of marginality on plant population performance. Journal of Biogeography 41(2):239–249.

ACCDC. 2008. Sub-national (provincial/state) ranks. Sackville: Atlantic Canada Conservation Data Centre. Available from: http://www.accdc.com/data/ranks.html.

ACCDC. 2017. Species Occurrence Records. Sackville: ACCDC. Available from: http://www.accdc.com/.

Aguilar R, Quesada M, Ashworth L, Herrerias-Diego Y, Lobo J. 2008. Genetic consequences of habitat fragmentation in plant populations: susceptible signals in plant traits and methodological approaches. Molecular Ecology 17(24):5177– 5188.

Albert T, Raspé O, Jacquemart A-L. 2005. Diversity and spatial structure of clones in Vaccinium uliginosum populations. Canadian Journal of Botany 83(2):211–218.

Allendorf FW, Luikart G. 2012. Conservation and the genetics of populations: John Wiley & Sons. 626pp.

Alsos IG, Ehrich D, Thuiller W, Eidesen PB, Tribsch A, Schoenswetter P, Lagaye C, Taberlet P, Brochman C. 2012. Genetic consequences of climate change for northern plants. Proceedings of the Royal Society B 279(1735):2042–2051.

Alsos IG, Engelskjon T, Gielly L, Taberlet P, Brochmann C. 2005. Impact of ice ages on circumpolar molecular diversity: insights from an ecological key species. Molecular Ecology 14(9):2739–2753.

Alsos IG, Spjelkavik S, Engelskjon T. 2003. Seed bank size and composition of Betula nana, Vaccinium uliginosum, and Campanula rotundifolia habitats in Svalbard

125

and northern Norway. Canadian Journal of Botany 81(3):220–232.

Arrigo N, Tuszynski JW, Ehrich D, Gerdes T, Alvarez N. 2009. Evaluating the impact of scoring parameters on the structure of intra-specific genetic variation using RawGeno, an R package for automating AFLP scoring. BMC Bioinformatics 10(33). Epub 2009: http://www.biomedcentral.com/1471-2105/10/33.

Assis J, Coelho NC, Alberto F, Valero M, Raimondi P, Reed D, Serrao EA. 2013. High and distinct range-edge genetic diversity despite local bottlenecks. PLoS ONE 8(7).10.1371/journal.pone.0068646.

Au R, Tardif JC. 2007. Allometric relationships and dendroecology of the dwarf shrub Dryas integrifolia near Churchill, subarctic Manitoba. Canadian Journal of Botany 85(6):585–597.

Balloux F, Lehmann L, de Meeus T. 2003. The population genetics of clonal and partially clonal diploids. Genetics 164(4):1635–1644.

Bauert MR, Kaelin M, Edwards PJ, Baltisberger M. 2007. Genetic structure and phylogeography of alpine relict populations of Ranunculus pygmaeus and Saxifraga cernua. Botanica Helvetica 117(2):181–196.

Bellard C, Bertelmeier C, Leadley P, Thuiller W, Courchamp F. 2012. Impacts of climate change on the future of biodiversity. Ecology Letters 15(4):365–377.

Blyton MDJ, Flanagan NS. 2012. A comprehensive guide to: GenAlEx 6.5. Australia: Australian National University. 131 pp. Available from: http://docplayer.net/15262142-Http-biology-anu-edu-au-genalex.html.

Bonin A, Bellemain E, Eidesen PB, Pompanon F, Brochman C, Taberlet P. 2004. How to track and assess genotyping errors in population genetics studies. Molecular Ecology 13(11):3261–3273.

Bonin A, Ehrich D, Manel S. 2007. Statistical analysis of Amplified Fragment Length Polymorphism data: a toolbox for molecular ecologists and evolutionists. Molecular Ecology 16(18):3737–3758.

Brochet A-L, M. Guillemain, H. Fritz, M. Gauthier-Clerc, Green AJ. 2009. The role of migratory ducks in the long-distance dispersal of native plants and the spread of

126

exotic plants in Europe. Ecography 32(6):919–928.

Carlquist S. 1983. Intercontinental Dispersal. Dispersal and Distribution 737–47.

CEC: Terrestrial Ecoregions, 2009: Commission for Environmental Cooperation. Available from: http://www.cec.org/Page.asp?PageID=122&ContentID=1329.

Chybicki IJ, Oleska A, Burczyk J. 2011. Increased inbreeding and strong kinship structure in Taxus baccata estimated from both AFLP and SSR data. Heredity 107(6):589–600.

Couvet D. 2002. Deleterious effects of restricted gene flow in fragmented populations. Conservation Biology 16(2):369–376.

Crawford LA, Koscinski D, Keyghobadi N. 2012. A call for more transparent reporting of error rates: the quality of AFLP data in ecological and evolutionary research. Molecular Ecology 21(24):5911–5917.

Crawford RMM. 2008. Plants at the margin – ecological limits and climate change Cambridge, UK: Cambridge University Press. 478 pp.

Cruden RW. 1966. Birds as agents of long-distance dispersal for disjunct plant groups of the temperate western hemisphere. Evolution 20(4):517–532.

CYSIP. CSYP: Central Yukon Species Inventory Project Friends of Dempster Country. Available from: http://www.flora.dempstercountry.org/CYSIP.About.html. de Mendiburu F. 2016. agricolae: statistical procedures for agricultural research. R package version 1.2-4 ed. Available from: http://CRAN.R- project.org/package=agricolae.

127

DeChaine EG, Forester BR, Schaefer H, Davis CC. 2013. Deep Genetic Divergence between Disjunct Refugia in the Arctic-Alpine King’s Crown, Rhodiola integrifolia (Crassulaceae). PLoS ONE 8(11).10.1371/journal.pone.0079451.

Doak DF, Morris WF. 2010. Demographic compensation and tipping points in climate- induced range shifts. Nature 467(7318):959–962.

Dobrowski SZ. 2011. A climatic basis for microrefugia: the influence of terrain on climate. Global Change Biology 17(2):1022–1035.

Durka W, Michalski SG, Berendzen KW, Bossdorf O, Bucharova A, Hermann JM, Hölzel N, Kollmann J. 2016. Genetic differentiation within multiple common grassland plants supports seed transfer zones for ecological restoration. Journal of Applied Ecology 54(4):116-126.

Dyke AS. 2005. Late Quaternary vegetation history of northern North America based on pollen, macrofossil, and faunal remains. Géographie Physique et Quaternaire 59(2-3):211–262.

Dyke AS, Andrews JT, Clark PU, England JH, Miller GH, Shaw J, Veillette JJ. 2002. The Laurentide and Innuitian ice sheets during the Last Glacial Maximum. Quaternary Science Reviews 21(1–3):9–31.

Earl DA, vonHoldt BM. 2012. STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method Conservation Genetics Resources 4(2):359–361.

Eckert CG. 2002. The loss of sex in clonal plants. In: Ecology and Evolutionary Biology of Clonal Plants. JF Stuefer, Erschbamer B, Huber H, Suzuki JI, editors. Dordrecht: Springer Science and Business Media. pp. 279–298.

Eckert CG, Samis KE, Lougheed SC. 2008. Genetic variation across species' geographical ranges: the central-marginal hypothesis and beyond. Molecular Ecology 17(5):1170–1188.

Eckstein RL, O'Neill RA, Danhelka J, Otte A, Köhler W. 2006. Genetic structure among

128

and within peripheral and central populations of three endangered floodplain violets. Molecular Ecology 15(9):2367–2379.

Eidesen PB, Alsos IG, Popp M, Stensrud O, Brochman C. 2007. Nuclear vs. plastid data: complex Pleistocene history of a circumpolar key species. Molecular Ecology 16(18):3902–3925.

Eizaguirre C, Baltazar–Soares M. 2014. Evolutionary conservation–evaluating the adaptive potential of species. Evolutionary Applications 7(9):963–967.

Ellstrand NC. 1992. Gene flow among seed plant populations. New Forests 6(1-4):241– 256.

Ellstrand NC, Elam DR. 1993. Population genetic consequences of small population size: Implications for plant conservation. Annual Review of Ecology and Systematics 24(1):217–242.

Evanno G, Regnault S, Goudet J. 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular Ecology 14(8):2611–2620.

Falush D, Stephens M, Pritchard JK. 2003. Inference of population structure using multilocus genotype data: Linked loci and correlated allele frequencies. Genetics 164(4):1567–1587.

Falush D, Stephens M, Pritchard JK. 2007. Inference of population structure using multilocus genotype data: dominant markers and null alleles. Molecular Ecology Notes 7(4):574–578.

FNAA. 2008. Flora of North America. Flora of North America Association. Available from: http://www.efloras.org/index.aspx.

García C, Jordano P, Godoy JA. 2007. Contemporary pollen and seed dispersal in a Prunus mahaleb population: patterns in distance and direction. Molecular Ecology 16(9):1947–1955.

129

Garcia D, Zamora R, Gomez JM, Jordano P, Hodar J. 2000. Geographical variation in seed production, predation and abortion in Juniperus communis throughout its range in Europe. Journal of Ecology 88(3):436–446.

Garcia PE, Winkler M, Flatscher R, Sonnleitner M, Krejcikova J, Suda J, Hulber K, Schneeweiss GM, Schonswetter P. 2012. Extensive range persistence in peripheral and interior refugia characterizes Pleistocene range dynamics in a widespread Alpine plant species (Senecio carniolicus, Asteraceae). Molecular Ecology 21(5):1255–1270.

Gaston KJ. 2003. The structure and dynamics of geographic ranges Oxford: Oxford University Press. 265 pp.

GBIF.org. 2016. GBIF Occurrence Download. 2016 ed.: GBIF.org. Available from: http://www.gbif.org.

Govindaraju DR. 1988. Relationship between dispersal ability and levels of gene flow in plants. Oikos (1988):31–35.

Hahn T, Kettle CJ, Ghazoul J, Frei ER, Matter P, Pluess AR. 2012. Patterns of genetic variation across altitude in three plant species of semi-dry grasslands. PLoS ONE 7(8): http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.00416 08.

Hamilton JA, Eckert CG. 2007. Population genetic consequences of geographic disjunction: a prairie plant isolated on Great Lakes alvars. Molecular Ecology 16(8):1649–1660.

Hamilton MB, Miller JR. 2002. Comparing relative rates of pollen and seed gene flow in the island model using nuclear and organelle measures of population structure. Genetics 162(4):1897–1909

Hampe A. 2005. Fecundity limits in Frangula alnus (Rhamnaceae) relict populations at the species’ southern range margin. Oecologia 143(3):377–386

Hampe A, Jump AS. 2011. Climate Relicts: Past, Present, Future. Annual Review of Ecology, Evolution and Systematics 42(2011):313–333.

Hampe A, Petit RJ. 2005. Conserving biodiversity under climate change: the rear edge matters. Ecology Letters 8(5):461–467.

Hampe A, Rodriguez-Sanchez F, Dobrowski S, Hu SS, Gavin DG. 2013. Climate refugia: from the Last Glacial Maximum to the twenty-first century. New

130

Phytologist 197(1):16–18.

Hamrick JL, Godt MJW. 1996. Effects of life history traits on genetic diversity in plant species. Philosophical Transactions of the Royal Society B: Biological Sciences 351(1345):1291–1298.

Hardie DC, Hutchings JA. 2010. Evolutionary ecology at the extremes of species’ ranges. Environmental Reviews 18(2010):1–20.

Hartl DL, Clark AG. 1997. Principles of population genetics. Third edition ed. Sunderland, Massachusetts: Sinauer Associates, Inc. 542 pp.

Hewitt GM. 2004. Genetic consequences of climatic oscillations in the Quaternary. Philosophical Transactions of the Royal Society of London B 359(1442):183– 195.

Higgins SI, Nathan R, Cain ML. 2003. Are long-distance dispersal events in plants usually caused by nonstandard means of dispersal? Ecology 84(8):1945–1956.

Hinds H. 2000. The Flora of New Brunswick Fredericton, NB: Department of Biology, University of New Brunswick. 699 pp.

Hirao A, Kudo G. 2004. Landscape genetics of alpine-snowbed plants: comparisons along geographic and snowmelt gradients. Heredity 93(3):290–298.

Hirsch H, Wagner V, Danihelka J, Ruprecht E, Sánchez-Gómez P, Seifert M, Hensen I. 2015. High genetic diversity declines towards the geographic range periphery of Adonis vernalis, a Eurasian dry grassland plant. Plant Biology 17(6):1233–1241.

Holland BR, Clarke AC, Meudt HM. 2008. Optimizing automated AFLP scoring parameters to improve phylogenetic resolution. Systematic Biology 57(3):347– 366.

Honnay O, Bossuyt B. 2005. Prolonged clonal growth: escape route or route to extinction? Oikos 108(2):427–432.

Honnay O, Jacquemyn H. 2008. A meta-analysis of the relation between mating system, growth form and genotypic diversity in clonal plant species. Evolutionary Ecology 22(3):299–312.

Honnay O, Jacquemyn H, Bossuyt B, Hermy M. 2005. Forest fragmentation effects on patch occupancy and population viability of herbaceous plant species. New Phytologist 166(3):723–736

Ikeda H, Yoneta Y, Higashi H, Eidesen PB, Barkalov V, Yakubov V, Brochmann C, 131

Setoguchi H. 2015. Persistent history of the bird–dispersed arctic–alpine plant Vaccinium vitis-idaea L.(Ericaceae) in Japan. Journal of Plant Research 128(3):437–444.

Ilves A, Metsare M, Seliškar A, García MB, Vassiliou L, Pierce S, Tatarenko I, Tali K, Kull T. 2016. Genetic diversity patterns of the orchid Anacamptis pyramidalis at the edges of its distribution range. Plant Systematics and Evolution 302(9):1227– 1238.

Jacquemart A-L. 1996. Vaccinium uliginosum L. Journal of Ecology 84(5):771–786.

Jacquemart A-L, Thompson JD. 1996. Floral and pollination biology of three sympatric Vaccinium (Ericaceae) species in the Upper Ardennes, Belgium. Canadian Journal of Botany 74(2):210–221.

Janes JK, Miller JM, Dupuis JR, Malenfant RM, Gorrell JC, Cullingham CI, Andrew RL. 2017. The K= 2 conundrum. Molecular Ecology: http://onlinelibrary.wiley.com/doi/10.1111/mec.14187/full.

Jiang D, Wu G, Mao K, Feng J. 2015. Structure of genetic diversity in marginal populations of black poplar (Populus nigra L.). Biochemical Systematics and Ecology 61(2015):297–302.

John E, Turkington R. 1997. A 5-year study of the effects of nutrient availability and herbivory on two boreal forest herbs. Journal of Ecology 85(4):419–430.

Jump AS, Peñuelas J. 2005. Running to stand still: adaptation and the response of plants to rapid climate change. Ecology Letters 8(9):1010–1020.

Jump AS, Woodward FI. 2003. Seed production and population density decline approaching the range-edge of Cirsium species. New Phytologist 160(2):349– 358.

Kopelman NM, Mayzel J, Jakobsson M, Rosenberg NA, Mayrose I. 2015. Clumpak: a program for identifying clustering modes and packaging population structure inferences across K. Molecular Ecology Resources 15(5):1179-1191.

Kvist L, Aminian L, Rouger R, Kreivi M, Laurila M, Hyvärinen M, Aspi J, Markkola A. 2015. A climatic relict or a long distance disperser: conservation genetics of an Arctic disjunct polyploid plant. Conservation Genetics 16(6):1–11.

Lammi A, P. Silkamaki, Mustajarvi K. 1999. Genetic diversity, population size, and fitness in central and peripheral populations of a rare plant Lychnis viscaria.

132

Conservation Biology 13(5):1069–1078.

Lepais O, Muller SD, Ben Saad-Limam S, Benslama M, Rhazi L, Belouahem-Abed D, Daoud-Bouattour A, Gammar AM, Ghrabi-Gammar Z, Bacles CFE. 2013. High genetic diversity and distinctiveness of rear-edge climate relicts maintained by ancient tetraploidisation for Alnus glutinosa. PLoS ONE 8(9).10.1371/journal.pone.0075029.

Leppig G, White JW. 2006. Conservation of peripheral plant populations in California. Mardono 53(3):264–274.

Lesica P, Allendorf FW. 1995. When are peripheral populations valuable for conservation? Conservation Biology 9(4):753–760.

Levin DA. 2012. Mating system shifts on the trailing edge. Annals of Botany 109(613):613–620.

Linhart YB, Premoli AC. 1994. Genetic variation in central and disjunct populations of Lilium parryi. Canadian Journal of Botany 72(1):79–85.

Loveless MD, Hamrick JL. 1984. Ecological determinants of genetic structure in plant populations. Annual review of Ecology and Systematics 15(1):65–95.

Magdy M, Werner O, McDaniel S, Goffinet B, Ros R. 2015. Genomic scanning using AFLP to detect loci under selection in the moss Funaria hygrometrica along a climate gradient in the Sierra Nevada Mountains, Spain. Plant Biology 18(2):280–288.

Manel S, Gugerli F, Thuiller W, Alvarez N, Legendre P, Holderegger R, Gielly L, Taberlet P. 2012. Broad-scale adaptive genetic variation in alpine plants is driven by temperature and precipitation. Molecular Ecology 21(15):3729–3738.

Mayle FE, Cwynar LC. 1995. Impact of the younger Dryas cooling event upon lowland vegetation of Maritime Canada. Ecological Monographs 65(2):129–154

Meeus S, Honnay O, Jacquemyn H. 2012. Strong differences in genetic structure across disjunct, edge, and core populations of the distylous forest herb Pulmonaria officinalis (Boraginaceae). American Journal of Botany 99(11):1809–1818.

133

Meirmans PG, Goudet J, Gaggiotti OE. 2011. Ecology and life history affect different aspects of the population structure of 27 high-alpine plants. Molecular Ecology 20(15):3144–3155.

Meloni M, Reid A, Caujape-Castells J, Marrero A, Fernandez-Palacios JM, Mesa-Coelo RA, Conti E. 2013. Effects of clonality on the genetic variability of rare, insular species: the case of Ruta microcarpa from the Canary Islands. Ecology and Evolution 3(6):1569–1579.

Meudt HM, Clarke AC. 2007. Almost forgotten or latest practice? AFLP applications, analyses and advances. Trends in Plant Science 12(3):106–117.

Murray DF. 1997. Systematics of the ITEX species. Global Change Biology 3(S1):10– 19.

Musche M, Settele J, Durka W. 2008. Genetic population structure and reproductive fitness in the plant Sanguisorba officinalis in populations supporting colonies of an endangered maculinea butterfly. International Journal of Plant Sciences 169(2):253–262.

Nagamitsu T, Hoshikawa T, Kawahara T, Barkalov VY, Sabirov RN. 2014. Phylogeography and genetic structure of disjunct Salix arbutifolia populations in Japan. Population Ecology 56(3):539–549.

Nybom H. 2004. Comparison of different nuclear DNA markers for estimating intraspecific genetic diversity in plants. Molecular Ecology 13(5):1143–1155.

Nybom H, Bartish IV. 2000. Effects of life history traits and sampling strategies on genetic diversity estimates obtained with RAPD markers in plants. Perspectives in Plant Ecology, Evolution and Systematics 3(2):93–114.

Pandey M, Geburek T. 2010. Genetic differences between continuous and disjunct

134

populations: some insights from sal (Shorea robusta Roxb.) in Nepal. Conservation Genetics 11(3):977–984.

Paulen RC, McClenaghan MB. 2013. New frontiers for exploration in glaciated terrain. Ottawa: Geological Survey of Canada.

Pauls SU, Nowak C, Balint M, Pfenniger M. 2013. The impact of global climate change on genetic diversity within populations and species. Molecular Ecology 22(4):925–946.

Paun O, Schoenswetter P. 2012. Amplified Fragment Length Polymorphism: an invaluable fingerprinting technique for genomic, transcriptomic, and epigenetic studies. Methods in Molecular Biology (Clifton, NJ) 862(2012):75–87.

Peakall R, Smouse P. 2006. GenAlEx 6: Genetic Analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes 6(1):288–295.

Peakall R, Smouse P. 2012. GenAlEx 6.5: Genetic analysis in Excel. Population genetic software for teaching and research – an update. Bioinformatics: 10.1093/bioinformatics/bts460.10.1093/bioinformatics/bts460.

Petit RJ, Duminil J, Fineschi S, Hampe A, Salvini D, Vendramin GG. 2005. Invited review: comparative organization of chloroplast, mitochondrial and nuclear diversity in plant populations. Molecular Ecology 14(3):689–701.

Philipp M, Siegismund HR. 2003. What can morphology and isozymes tell us about the history of the Dryas integrifolia-octopetala complex? Molecular Ecology 12(8):2231–2242.

Pompanon F, Bonin A, Bellemain E, Taberlet P. 2005. Genotyping errors: causes,

135

consequences and solutions. Nature Reviews Genetics 6(11):847–859.

Porras-Hurtado L, Ruiz Y, Santos C, Phillips C, Carracedo A, Lareu MV. 2013. An overview of STRUCTURE: applications, parameter settings, and supporting software. Frontiers in Genetics 4(2013):98.

Pritchard JK, Stephens M, Donnelly P. 2000. Inference of population structure using multilocus genotype data. Genetics 155(2):945–959.

Pritchard JK, Wen X, Falush D. 2010. Documentation for structure software. 2.3 ed. University of Chicago: Pritchard et al. Available from: http://pritch.bsd.uchicago.edu/structure.html.

Provan J, Maggs CA. 2012. Unique genetic variation at a species’ rear edge is under threat from global climate change. Proceedings of the Royal Society B 279(1726):39–47.

R Core Team. 2015. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Available from: http://www.R- project.org.

Ramasamy RK, Ramasamy S, Bindroo BB, Naik VG. 2014. STRUCTURE PLOT: a program for drawing elegant STRUCTURE bar plots in user friendly interface. SpringerPlus 3(1):431.

Reed DH, Frankham R. 2003. Correlation between fitness and genetic diversity. Conservation Biology 17(1):230–237.

Reed TE, Schindler DE, Waples RS. 2011. Interacting effects of phenotypic plasticity and evolution on population persistence in a changing climate. Conservation Biology 25(1):56–63.

Rehm EM, Olivas P, Stroud J, Feeley KJ. 2015. Losing your edge: climate change and the conservation value of range-edge populations. Ecology and Evolution 5(19):4315–4326.

Reisch C, Bernhardt-Römermann M. 2014. The impact of study design and life history traits on genetic variation of plants determined with AFLPs. Plant Ecology 215(12):1493–1511.

136

Revelle W. 2016. psych: Procedures for Personality and Psychological Research. 1.6.9 ed. Evanston, Illinois: Northwestern University. Available from: https://CRAN.R-project.org/package=psych.

Ritchie JC. 1987. Postglacial Vegetation of Canada Cambridge, UK: Cambridge University Press. 178 pp.

Robinson J, Hermanutz L. 2015. Evaluating human-disturbed habitats for recovery planning of endangered plants. Journal of Environmental Management 150(2015):157–163.

Rune O. 1954. Notes on the flora of the Gaspé Peninsula. Svensk Bot Tidskr 48(1954):117–135.

Schoville SD, Bonin A, Francois O, Lobreaux S, Melodelima C, Manel S. 2012. Adaptive genetic variation on the landscape: methods and cases. In: Annual Review of Ecology, Evolution, and Systematics. DJ Futuyma, editor. pp. 23–43.

Sexton JP, Hangartner SB, Hoffmann AA. 2014. Genetic isolation by environment or distance: which pattern of gene flow is most common? Evolution 68(1):1–15.

Shaw J, Piper DJW, Fader GBJ, King EL, Todd BJ, Bell T, Batterson MJ, Liverman DGE. 2006. A conceptual model of the deglaciation of Atlantic Canada. Quaternary Science Review 25(17-18):2059–2081.

Silvertown J. 2008. The evolutionary maintenance of sexual reproduction: evidence from the ecological distribution of asexual reproduction in clonal plants. International Journal of Plant Sciences 169(1):157–168.

Skrede I, Eidesen PB, Portela RP, Brochmann C. 2006. Refugia, differentiation and postglacial migration in arctic- alpine Eurasia, exemplified by the mountain avens (Dryas octopetala L.). Molecular Ecology 15(7):1827–1840.

Stachurska-Swakon A, Cieslak E, Ronikier M. 2013. Phylogeography of a subalpine tall-herb Ranunculus platanifolius (Ranunculaceae) reveals two main genetic lineages in the European mountains. Botanical Journal of the Linnean Society 171(2):413–428.

Thiel-Egenter C, Gugerli F, Alvarez N, Brodbeck S, Cieslak E, Colli L, Englisch T, Gaudeul M, Gielly L, Korbecka G, et al. 2009. Effects of species traits on the genetic diversity of high-mountain plants: a multi-species study across the Alps and the Carpathians. Global Ecology and Biogeography 18(1):78–87.

Tremblay NO, Schoen DJ. 1999. Molecular phylogeography of Dryas integrifolia: a

137

glacial refugia and postglacial recolinization. Molecular Ecology 8(7):1187– 1198.

Tsaliki M, Diekmann M. 2009. Fitness and survival in fragmented populations of Narthecium ossifragum at the species’ range margin. Acta Oecologica 35(3):415–421.

Vakkari P, Rusanen M, Karkkainen K. 2009. High genetic differentiation in marginal populations of European White elm (Ulmus laevis). Silva Fennica 43(2):185– 196.

Van Rossum F, Prentice HC. 2004. Structure of allozyme variation in Nordic Silene nutans (Caryophyllaceae): population size, geographical position and immigration history. Biological Journal of the Linnean Society 81(3):357–371.

Van Rossum F, Vekemans X, Gratia E, Meerts P. 2003. A comparative study of allozyme variation of peripheral and central populations of Silene nutans L. (Caryophyllaceae) from Western Europe: implications for conservation. Plant Systematics and Evolution 242(1-4):49–61.

Viana DS, Santamaria L, Michot TC, Figuerola J. 2013. Allometric scaling of long- distance seed dispersal by migratory birds. The American Naturalist 181(5):649– 662.

Villellas J, Berjano R, Terrab A, García MB. 2014. Divergence between phenotypic and genetic variation within populations of a common herb across Europe. Ecosphere 5(5):1–14.

Vos P, Hogers R, Bleeker M, Reijans M, Lee Tvd, Hornes M, Frijters A, Pot J, Peleman J, Kuiper M, et al. 1995. AFLP: a new technique for DNA fingerprinting. Nucleic Acids Research 23(21):4407–4414.

Vranckx G, Jacquemyn H, Muys B, Honnay O. 2012. Meta-analysis of susceptibility of woody plants to loss of genetic diversity through habitat fragmentation. Conservation Biology 26(2):228–237.

Vucetich JA, Waite TA. 2003. Spatial patterns of demography and genetic processes across the species' range: Null hypothesis for landscape conservation genetics. Conservation Genetics 4(5):639–645.

138

173(7):802–811.

Wang IJ, Bradburd GS. 2014. Isolation by environment. Molecular Ecology 23(23):5649–5662.

Webb SL. 1986. Potential role of passenger pigeons and other vertebrates in the rapid Holocene migrations of nut trees. Quaternary Research 26(3):367–375.

Westergaard KB, Alsos IG, Ehrich D, Eidesen PB, Hollingsworth PM, Brochman C. 2008. Genetic diversity and distinctiveness in Scottish alpine plants. Plant Ecology and Diversity 1(2):329–338.

Wright S. 1943. Isolation by distance. Genetics 28(2):114–138.

Young A, Boyle T, Brown T. 1996. The population genetic consequences of habitat fragmentation for plants. Trends in Ecology and Evolution 11(10):413–418.

139

Chapter 3 - Microclimate and demography in rear-edge populations of

two arctic-alpine species (Dryas integrifolia and Anemone parviflora) at

Wilson Brook Protected Natural Area, New Brunswick

Abstract

If North American mean temperatures change by 1.1 to 5.6 °C over the next 80 years as predicted, populations at the edges of species' ranges, particularly the rear edges, may be at risk of extirpation. I experimentally evaluated the response of rear-edge disjunct populations of two arctic-alpine species (Dryas integrifolia and Anemone parviflora) to altered temperature regimes. The microclimate at Wilson Brook Protected Natural Area was compared to (1) regional climate in New Brunswick and Newfoundland; (2) a nearby site without the target species; and (3) a more distant Newfoundland site further north within the range. Population dynamics were assessed under ambient and experimental conditions over three years (2011–2013). The microclimate at Wilson

Brook was similar to the New Brunswick regional climate when compared over three years, and was similar to the microclimate at a site further north, providing indirect evidence that the site has a local cooling effect, since local microclimate is expected to be warmer than the regional climate. In contrast to expectations, Open Top Chambers

(OTCs) significantly lowered microclimate maximum daily temperature in one year, and fewer high temperature events were recorded in OTCs over the three years. This variation in extreme temperature had no impact on population growth. These data

140

suggest that these populations may not be sensitive to temperature variability within the ranges tested, and in the timeframes observed. Other factors that influence the local habitat, such as the instability and type of substrate (gypsum) or biotic interactions, may play important roles in the persistence of these populations. For conservation purposes, these populations appear to be secure.

141

3.1 Introduction

Geographical range limits (also termed range edges or peripheries) are areas in which a species' dispersal may be limited, and where the species may be at the limit of its range of ecological tolerance (Hampe & Jump 2011). At range limits, species may experience greater stress (e.g., increased drought, competition, or temperature), making them more vulnerable to change than central populations. In addition, edge populations are generally small, isolated and located in specialized habitats (Crawford 2008). Range- edge populations are often characterized by decreased seed production and seedling recruitment (Jump & Woodward 2003, Tomback et al. 2005, Tsaliki & Diekmann 2009,

Moeller et al. 2012), by lower growth rates and reduced fitness (Angert 2006, Salzer &

Gugerli 2012, Vaupel & Matthies 2012), higher inbreeding (Michalski & Durka 2007) and selfing (Mimura & Aitken 2007, Levin 2012). However, there is some ambiguity in these patterns. In some range-edge populations, seed dispersal and plant performance have been found to be similar to, or higher than those of central populations (Darling et al. 2007, Wagner et al. 2011, Pironon et al. 2015).

At leading range edges, individuals are recruited, and the range can advance. In contrast, at rear range edges (also termed warm-edge limits), recruitment is limited, and the range is stable or contracting (Woolbright et al. 2014) because populations are generally more isolated at the rear edge than at the leading edge. Because rear range-edge populations are at greatest risk of extirpation (Lesica & Allendorf 1995, Hampe & Petit 2005), they are of significant conservation concern. The particular concern is that rear-edge

142

populations may be older, possibly refugia (i.e., populations that persisted through glaciations) or relicts (i.e., populations that persisted in formerly glaciated areas after deglaciation) that have retained ancient genotypes (Alsos et al. 2009, Hampe & Jump

2011). Edge populations may contain locally adapted genotypes that do not occur elsewhere in the range (Hampe & Petit 2005, Woolbright et al. 2014) due to the contemporary uniqueness of the habitat at a rear-edge; they may therefore be valuable for biodiversity conservation (Santamaria et al. 2003, Busch 2005, Vergeer & Kunin

2013). For that reason, evaluating the possible responses of such populations to climate change is essential.

The effects of climate change are already evident in many plant species across the northern hemisphere. Assessments of species' range responses to climate change fall into two broad approaches, each with its strengths and weaknesses. Those focused on modeling species' distributions include ecological niche models projecting future range shifts (Thuiller et al. 2008, Valladares et al. 2014), bioclimatic models (Beaumont et al.

2007), and dynamic global vegetation models (Neilson et al. 2005). A review of 29 studies (Buckley & Kingsolver 2012) suggests that such models are frequently poor predictors of individual range shifts as highlighted in. Empirical studies include assessments of phenological responses to temperature changes using data collected over

10 to 20 years (Parmesan 2007). For example, phenological changes and shifts in some species' ranges have occurred, with range contractions at the rear edge and expansions at the leading edge (Parmesan & Yohe 2003, Thomas et al. 2004, Parmesan 2006, 2007,

143

Sparks et al. 2011, Provan & Maggs 2012, Field et al. 2014, Settele et al. 2014, Mason et al. 2015). Short and long-term warming experiments have assessed changes in demography and life history over time (Chapin & Shaver 1996, Hudson et al. 2011,

Klady et al. 2011, Elmendorf et al. 2012). However, generalization across taxa is limited because many experiments are small-scale and short-term (Buckley & Kingsolver 2012,

Elmendorf et al. 2012). Populations at leading edges in higher altitudes and particularly at high latitudes have received considerable attention because they may play a critical role in facilitating range shifts (Rehm et al. 2015). While detrimental impacts of warming temperatures have been predicted for rear-edge populations (Hampe & Petit

2005), direct tests are limited.

Temperature and water balance are hypothesized to be the most significant drivers of changes to population demography at the rear range edge (Hampe & Jump 2011, Cahill et al. 2014). Temperature may arguably be of greatest concern because it impacts the rates of all biological and chemical processes (Shaver et al. 2000), including water uptake and loss. Previous research on the effect of climate change on rear-edge populations has focused on well-studied regions such as the Mediterranean (Hampe

2005), or on woody species (Vicente-Serrano et al. 2015), but is generally limited. For example, higher mortality rate in trees in temperate forests at their southern limits in

California have been attributed to temperature and drought (Van Mantgem &

Stephenson 2007). Decline in populations of Eucalyptus wandoo in Australia was shown to be in response to increased drought (Brouwers et al. 2013). Decline in abundance of

144

Abies alba at its southern limits in Europe has been attributed to reduced water availability (Gazol et al. 2015). Long-term growth decline of Pinus sylvestris in the

Mediterranean was correlated with increased May-July diurnal temperatures (Büntgen et al. 2013) in a study of growth rings, and changes in regional warming and water balance have led to mortality at the southern range edge of Aloe dichotoma (Foden et al. 2007).

However, some range-edge populations may also be better adapted to drought than central populations (Lázaro-Nogal et al. 2016). In addition to local adaptations in rear- edge populations, climate refugia at the rear edge of a species' range may allow persistence through periods of unfavorable regional climates (Dobrowski 2011). It has long been known that microclimates vary considerably from regional climates (usually standardized measurements 2 m above the ground), and that many factors, such as topography, slope, aspect, cover, presence of water and cold air pooling, will influence those microclimates (Geiger et al. 2009). Thus, local site characteristics may decouple the local microclimate from regional trends (Rull 2009, Hannah et al. 2014, Hylander et al. 2015). Such sites may act as buffers to climate changes (Hampe & Jump 2011,

Růžička et al. 2012) and allow populations to persist, at least until regional climate changes result in a modification of the microclimate.

Arctic-alpine species may provide good models for assessing the effects of climate warming on rear-edge populations, because they are adapted to low temperature regimes, desiccation during the dormant season, and to low rates of precipitation

(Billings & Mooney 1968). Populations at warm-range limits can thus be expected to

145

respond negatively to climate warming.

Arctic-alpine species tend to have broad ecological niches (Crawford 2008), which could buffer them from some climate change impacts in most parts of their ranges

(Breen et al. 2014). In addition, ecological studies have demonstrated the occurrence of local adaptation along latitudinal and altitudinal gradients inhabited by many arctic- alpine species, suggesting a strong potential for differentiation across their ranges that may make some populations more tolerant to future conditions (Mooney & Billings

1961, Leimu & Fischer 2008, Pellissier et al. 2013, Read et al. 2014, Souther et al.

2014). Arctic-alpine species have also persisted through historical climate change (e.g., during the Holocene (Birks 2008)), indicating high tolerance and effective adaptations to climate variability. This tolerance suggests that they may be resilient to climate change

(Breen et al. 2014). However, indirect impacts may affect species in unexpected ways.

For example, vegetative growth of two arctic-alpine species (Silene acaulis, Polygonum viviparum) at their southern range limits revealed high resilience and tolerance to temperature variability, but their lower survival and recruitment may cause population declines over the long term (Doak & Morris 2010). In an example of an indirect effect, high temperatures increased growth of tall species which negatively affected some disjunct, southern arctic-alpine populations of Arnica angustifolia in Europe

(Jäkäläniemi 2011). A number of ecological studies have evaluated characteristics of plant life history and growth in populations of arctic-alpine species at their southern range limit at Mt. Washington (New Hampshire). Those studies assessed phenology,

146

plant growth, microenvironment (including temperature) and photosynthesis (Hadley &

Bliss 1964), plant productivity (Bliss 1966), carbohydrate cycle (Fonda & Bliss 1966), and plant reproduction (Bell & Bliss 1980). Only one warming experiment included a rear range-edge alpine population of Dryas octopetala in western USA, and showed a positive response of plant performance to warming (Welker et al. 1997). In summary, while some populations of arctic-alpine species at their southern range limit may be adapted to current conditions, rapid climate warming may put them at risk in the longer term.

The focus of the present study is a protected area at Wilson Brook, New Brunswick,

Canada. Disjunct populations of several arctic-alpine species, including Anemone parviflora and Dryas integrifolia, occur at the rear edge of their geographic ranges at this site. Conditions here are thus presumably near the limit of their ecological tolerance.

It has been hypothesized that Wilson Brook was a climate refugium during the re- colonization process northwards after the Last Glacial Maximum (LGM), providing a microclimate and habitat inhospitable to other species, while supporting arctic-alpine species as the surrounding vegetation changed (Clayden 1994). The site is a north-facing gypsum cliff along a brook, surrounded by forest. Arctic-alpine species here are thought to have been “left behind” during postglacial re-colonization, making them contemporary relict populations. It is unknown whether the Wilson Brook site can continue to support the populations of arctic-alpine species in the face of climate warming. My research objectives were to: (1) assess whether the microclimate at Wilson

147

Brook is locally and regionally distinct as might be expected if it was a climate refugium; and (2) experimentally evaluate the association between variation in demography and variation in microclimate over three years in three arctic-alpine species occurring at Wilson Brook.

Based on the long-term persistence of the arctic-alpine species, I postulated that the microclimate at Wilson Brook would be (a) decoupled from the regional climate

(specifically, not warmer than the regional climate), (b) similar to sites further north at which the same arctic-alpine species occur, and (c) cooler than sites that occur in close geographic proximity but which do not host arctic-alpine species. Given that microclimate is expected to differ significantly from standardized observations due to the influence of boundary layer, slope, aspect, soil, and other factors that influence temperature at ground level (Geiger et al. 2009), I expect significantly higher ground- level temperatures than regional temperatures. In fact, the standardized measurements can be 20 °C lower than ground level measurements (Ashcroft & Gollan 2013). To address the second objective, I assessed temporal variation in population size as a measure of vegetative growth for the target species at Wilson Brook under ambient and open top chamber (OTC) conditions. OTCs are an effective and inexpensive experimental method to assess the effect of warming on plant growth and phenology

(Marion et al. 1997, Arft et al. 1999, Shaver et al. 2000, Aronson & McNulty 2009). The side panels of these chambers affect wind speed, resulting in a reduction in heat dissipation; they permit shortwave energy to enter while preventing longwave energy

148

from leaving. I predicted that population growth would be negatively affected by higher temperatures. The OTC experiment was monitored for three years, and similar experiments of one to four years have resulted in significant responses in phenology and vegetative growth (Arft et al. 1999). The herbaceous perennial A. parviflora was expected to show a greater response than the woody perennial D. integrifolia, as would be expected according to a meta-analysis of warming experiments carried out in the

Arctic by (Arft et al. 1999), where herbaceous species showed stronger growth response to warming than woody species.

149

3.2 Methods

3.2.1 Study sites

Wilson Brook Protected Natural Area (Albert County NB; 45.86005, –64.67611; Fig.

3.1; hereafter Wilson Brook) is accessible by road and legally protected, permitting only limited access to the site. No conservation management strategies are in place to address possible effects of future climate change. Wilson Brook (76 ha) includes a crumbling north-facing cliff of white gypsum, less than 1 km long and less than 20 m high (to the top of the cliff, Fig. 3.2). Gypsum underlies much of the site, but forest is the dominant vegetation. The slope of the gypsum cliff (45–75º with some level areas) provides a maximum of 10 m wide habitat along mostly the lower edge (Fig. 3.2). The cliff face below the plateau is eroding due to seepage water in the summer and the brook at its base. Dryas integrifolia and A. parviflora occur on the cliff face, with other shrubby species (e.g., Sheperdia canadensis) at the bottom edge extending to the forested margin. Trees surround the site on all sides and the location is shaded for much of the day. Four arctic-alpine species occur here, within an approximately 150 m x 25 m area:

White Mountain Avens (Dryas integrifolia), Mountain Goldenrod (Solidago multiradiata), Myrtle-leaved Willow (Salix myrtillifolia) and Northern Anemone

(Anemone parviflora) (Clayden 1994, Clayden 2000). Three of those species (D. integrifolia, S. myrtillifolia and S. multiradiata) occur nowhere else in the province, and no populations of the four species are known to occur further south in Eastern North

America (GBIF.org 2016). Dryas integrifolia and A. parviflora were selected based on

150

their availability at a wider range of sites in the Maritimes and were used to evaluate whether future warming is likely to affect disjunct rear-edge populations at this site.

Two additional sites, Port-au-Choix, Newfoundland and Albert Mines, New Brunswick were chosen for comparison with Wilson Brook (Fig. 3.1). Port-au-Choix (50.70065, –

57.37788) in northwestern Newfoundland is more than 750 km, or 5 ° north of Wilson

Brook, and is located on the Limestone Barrens within the southern range edge for the two study species. The site is in a protected area accessible to a local field assistant to verify data loggers during the season. In contrast, Albert Mines in Albert County, NB

(45.86443, –64.66719) is approximately 1.5 km from Wilson Brook. The substrate at

Albert Mines is similar to Wilson Brook but the site is not as shaded, lacks the steep, north-facing slopes of Wilson Brook, and the focal species, A. parviflora and D. integrifolia have never been reported to occur there. The Wilson Brook – Albert Mines comparison was used to assess whether microclimate differs between the two sites, since the species are absent at nearby Albert Mines despite seemingly suitable substrate. The

Wilson Brook – Port-au-Choix comparison was used to assess whether microclimate at

Wilson Brook is similar to that of western Newfoundland despite its relatively southern location, since the study species are present at both sites.

3.2.2 Study species

White Mountain Avens (Dryas integrifolia Vahl., Rosaceae, Fig. 1.2) is an evergreen, dwarf shrub forming mats up to 17 cm tall in calcareous wet meadows, river terraces,

151

tundra slopes and ridges in the Arctic; it prefers soil with a low organic content. It is considered a pioneer species and a poor competitor (Aiken et al. 2007). Dryas integrifolia is common and wide-ranging across the northern hemisphere, and in the

West, extends south into alpine habitats in the Rocky Mountains. In Atlantic Canada, the southernmost extent of the species’ distribution occurs at Wilson Brook, wherein the single population is disjunct from the nearest continuous range edge by at least 300 km.

The species is listed as S1 (5 or fewer occurrences or very few remaining individuals) in

New Brunswick, and S4 (apparently secure) in Quebec and Newfoundland and Labrador

(NatureServe 2016).

Northern White Anemone (Anemone parviflora Michx., Ranunculaceae, Fig. 1.3) is a perennial herb, which grows to 15 cm tall (2n = 2x = 16). The species is common in the lower Arctic with a circumpolar distribution; it occurs on both calcareous and non- calcareous soils but seems to prefer calcareous habitats (Aiken et al. 2007). The species is listed as S2 (20 or fewer occurrences or very few remaining individuals) in New

Brunswick, S1 in Nova Scotia, but not ranked elsewhere in Canada (NatureServe 2016).

The Wilson Brook populations of both species are isolated from the nearest populations by at least 300 km. Based on scaling up plant densities in sample plots, the Wilson

Brook population of D. integrifolia is estimated at > 50,000 ramets (individuals are difficult to distinguish due to the species' clonal growth form), and at approximately

1,000 individuals for A. parviflora. North and south of Port-au-Choix, Newfoundland,

152

populations of both species are more or less continuous, occurring in large patches north toward Burnt Cape, and south toward Bellburns (approximately 200 km in total).

3.2.3 Microclimate variables

Growing degree days (GDD, derived from temperature data) represents the duration during which temperature regimes allow growth to occur (McMaster & Wilhelm 1997).

For the calculation of GDD, see Appendix 3.1. GDD has been shown to be correlated with the northern limits of many woody species (Sykes et al. 1996) and is used widely in bioclimatic models of species distributions (Normand et al. 2009). Plant development occurs only if GDD exceeds a threshold temperature for a critical number of days specific to the species (Grigorieva et al. 2010). In the Arctic, the presumed range center of the two species, annual GDD is 198.6 in Iqualuit and 582.2 in Kuujjuaq for the period between 1981 to 2010 (Government of Canada 2016).

Moisture stress increases as the rate of transpiration increases relative to water uptake.

Water loss from leaves increases as water vapor content of the air declines (Lambers et al. 2008), which has been well studied in horticulturally and agriculturally important crops. Water vapor pressure deficit (VPD) is therefore a potential measure of moisture stress in plants if soil moisture is not sufficient to meet plant water needs. While ranges of tolerance for moisture stress are likely species- and site-specific (Leuschner 2002), optimal VPD in greenhouse settings for vegetables and cultivars is suggested to be from at 0.3 to 0.7 kPa (Alberta Agriculture and Forestry 2016). However, in the field VPD is

153

habitat specific and research by Leuschner (2002) showed that the woodland herbs in moist environments grew better in lower VPD (i.e., between 0.12 kPa to 0.58 kPa). VPD is derived as the difference between the saturation vapor pressure (es) and the actual vapor pressure (ea). For the calculation of VPD see Appendix 3.1.

3.2.4. Warming experiment

Open Top Chambers (OTCs) are an ideal experimental technique to assess species' responses to warming for several reasons: first, the Wilson Brook site has no long-term data series to facilitate predictive modeling; second, despite the limitations of such short- term experiments, all experimental warming methods have yielded information about ecological changes associated with climate warming (Shaver et al. 2000). For example,

OTCs have been effective in comparing impacts of warming on plant growth and phenology for over 20 years (Marion et al. 1997, Arft et al. 1999, Shaver et al. 2000).

Furthermore, because of the passive warming nature of OTCs, they are an inexpensive method (Aronson & McNulty 2009) that has been used in a variety of locations worldwide to predict population and ecosystem responses to future warming for a number of different taxa (Bokhorst et al. 2011, Scherrer & Koerner 2011, Semenova et al. 2015).

Following the standardized protocol provided in the International Tundra Experiment

(ITEX) Manual (Molau & Molgaard 1996, Marion et al. 1997), 10 hexagonal OTCs were placed to represent the range of microhabitats and slopes at Wilson Brook, each

154

paired with a control plot within a three to seven meter distance on a similar slope, substrate and aspect (Fig. 3.3). All plots were marked with five-sided metal frames anchored with light steel pegs. OTCs were 0.08 m2 (32 cm) in diameter at ground level, narrowing to 22 cm at the top), five-sided, with 15 cm high walls, and made of Sun-Lite

HP (Solar Components Corporation; www.solar- components.com/sun.htm#solargardening). The chambers were placed in the same positions each year (May 2011, and April 2012 and 2013) as soon as each site was accessible and snow-free. The OTCs were removed at the end of the growing season

(November to December) each year. Data loggers recorded temperature and relative humidity for both OTCs and controls during the periods from June 1 – November 9,

2011 to 2013.

The microclimate database was somewhat smaller than planned due to unanticipated events. At Wilson Brook, one OTC sensor did not function in 2012 and 2013, and a control sensor was damaged by rodents late in 2011. On 11 occasions (3 % of plot visits), some of the OTCs or frames had been disturbed, either by animals (one side panel of an OTC had to be replaced early in 2011), or by falling rocks (Fig. 3.4). Five

OTCs and three control plots were affected, with frequency distributed evenly over the years (three in 2011, four in 2012 and 2013 each). These issues are considered to have had negligible impact on the study, as they occurred randomly, affected a very small proportion (2%) of the dataset, and were corrected within one to two months, as soon as

155

the disturbance was observed.

3.2.5 Data collection

Population response was measured as number of ramets (D. integrifolia) and plants (A. parviflora) observed between May 31 and June 11 each year from 2012 to 2014, in ten

OTCs and ten control plots. Data from 2011 were excluded due to possible counting inconsistencies.

3.2.5.1 Regional climate and microclimate data

Climate was compared between pairs of sites using two levels of data collection: (1) data from weather stations were used to represent regional climate, and (2) data from ground level data loggers were used to represent the microclimate experienced by individual plants at each site, inside and outside warming plots. Below, I explain the data collected at each level, and in Section 3.3.1, the analyses used to assess microclimate are described. Regional weather station data were downloaded from Environment Canada for Nappan (NS), Moncton and Alma (NB), Daniels Harbor, Ferolle Point, and Plum

Point (NL) (Government of Canada 2016) within the timeframe of interest, when available (June 1 – November 9, 2011 – 2013 for NB except May 23 to October 19,

2012 for Port-au-Choix, Table 3.1). Although the regional data for southeastern New

Brunswick and Nova Scotia originated from different eco-regions than Wilson Brook, they were the closest stations and representative of southeastern NB climate.

Second, to characterize the local microclimate, I recorded temperature and relative

156

humidity with data loggers. Microclimate variables were calculated from logged data:

GDD and VPD per year, and the frequency of occurrences per year of six temperature categories (see below) for the three years of the study at Wilson Brook (Table 3.1).

Temperature and relative humidity (RH) were monitored using five data loggers each at

Wilson Brook and Port-au-Choix (HOBO U23 Pro, Onset,) and three at Albert Mines

(HOBO Pro Series, H08-032-08, Onset). For the warming experiment, an additional 10 sensors were placed inside the OTCs. Sensors were placed at ground level, under rocks or leaves that were present in the plots, to avoid direct radiation (Molau & Molgaard

1996). Temperature and relative humidity were recorded at 10-minute intervals during the same periods each year, when possible (Table 3.1). Microclimate data based on loggers near the ground are expected to show higher temperatures than the standardized regional measurements 2 m above ground.

The first five variables used to assess microclimate in this research at Wilson Brook,

Albert Mines and Port-au-Choix were mean, maximum, and minimum daily temperatures, growing degree days (GDD), and vapor pressure deficit. Warmer temperatures affect moisture content in soils, and vapor pressure deficit (VPD) can be used as a measure of moisture stress for plants depending on soil moisture. Soil moisture was not monitored during this study, nor was precipitation. Temperature data from the data loggers were further divided into an additional variable based on biologically relevant temperature grades (six classes; calculated over 10 min intervals), and based on the expectation that plants are susceptible to broad thresholds of temperature rather than

157

specific values, as per (Źróbek-Sokolnik 2012):

(1) below 0 ºC – negligible photosynthetic activity expected (Źróbek-Sokolnik 2012),

(2) 0 to 10 ºC – brackets the lower limit of photosynthesis for arctic plants (5ºC; (Molau

& Molgaard 1996, Wipf 2010), although photosynthesis can be maintained around

0ºC (Chapin III 1983),

(3) 10 to 15 ºC – optimal temperature range for maximum photosynthetic rate in arctic-

alpine plants (Chapin III 1983, Larkindale et al. 2005, Lambers et al. 2008),

(4) 15 to 25 ºC – mild / moderate heat stress expected in arctic species (Źróbek-Sokolnik

2012),

(5) 25 to 40 ºC – serious heat stress expected, and

(6) above 40 ºC – heat stress expected to cause damage. The lower limit of lethal

temperature observations is 42 ºC (Dahl 1951). Temperatures above 47.4 ºC have

been found to be lethal to 50% of D. octopetala plants after 30 minutes of exposure,

and the range of lethal temperatures for other arctic-alpine species ranged between

43.5 and 47.5 ºC (Gauslaa 1984).

158

3.3 Data analysis

All analyses, unless otherwise indicated, were carried out with the linear mixed effects package lme4 (Bates et al. 2015) in R version 3.2.2 (R Core Team 2015). A linear mixed effects model was chosen for its suitability to handle unbalanced sample size, and the use of a random factor (plots), since variability in individual plots (slope, shading, etc.) had to be accounted for in the models. Data were visually inspected for normality and homogeneity with plots of residuals against fitted values; outliers and leverage were identified (Cooks distance) and plotted with the available functions in R, and with influence.ME (Nieuwenhuis et al. 2012). Normality of residuals was visually inspected using qqnorm. Outliers were excluded from analyses only if clear equipment error could be identified (i.e., broken logger cord), logger readings were above 60 °C (presumed to be due to logger error), and when individual readings (once in the 10-min interval) were an order of magnitude higher than an earlier or later readings. P-values are based on likelihood ratio tests, unless otherwise stated. Individual plot location (insert, Fig. 3.1) was the random effect in all models unless otherwise stated, and temperature variables, year, location, and species were fixed effects in the models as stated. A linear model

(using lm from the stats package) was used for examining differences in growing degree days as a function of year and treatment (predictors). Post-hoc analyses were carried out for factors with significant interactions with the R package lsmeans (Lenth 2016) for linear mixed effects models, and ghlt from the multcomp package for linear fixed effects models using lm (Hothorn et al. 2008). Graphs were created with the graphics (R Core

159

Team 2015) and ggplot2 (Wickham 2016) packages.

3.3.1 Regional and microclimate at Wilson Brook versus Albert Mines and Port-au-

Choix

Separate linear mixed effects models were used to compare temporal and spatial variation in climate. First, each climate variable (mean, maximum and minimum daily temperature, and GDD) available from both weather stations (regional) and data loggers

(local) was fitted against location, month (5 months per year from June to November), and year (2011-2013) as fixed effects, and plot or station location as random effects.

Location included Wilson Brook (data loggers at n = 5 control plots), southeastern New

Brunswick and northwestern Newfoundland (n = 3 weather stations each). These models assessed the degree of climatic variation among sites regardless of the level at which climate data were recorded.

Second, variation in microclimate conditions from the same four variables but recorded only with data loggers was fit against location (Wilson Brook [data loggers at n = 5 control plots], Port-au-Choix [data loggers at n = 5], Albert Mines [data loggers at n =

3]) as the fixed effect, and plot as the random effect. These models assessed the degree of climatic variation occurring at ground level among sites, and ignored trends in climate recorded by weather stations. Each climate response variable was assessed separately.

3.3.2 Warming experiment: OTCs versus controls at Wilson Brook

To assess the effect of experimental OTC plots on microclimate, variation in each of the

160

six microclimatic variables was assessed using a linear mixed effects model with year

(2011-2013) and treatment (OTC [data loggers at n = 10 plots] or control [data loggers at n = 5]) and their interaction as fixed effects, and plot as the random effect. For all analyses, the full factorial model including the interaction between time and treatment was compared with reduced models that (1) did not include the interaction term, and (2) did not include the term of interest (treatment). Models were compared using log likelihood, and the best-fit model selected based on Chi-square. Where the full model was selected (i.e., where a significant interaction was present), Tukey Post-Hoc comparisons were undertaken to determine where significant differences were present.

The least-squared mean estimates for the fixed effect treatment was obtained using the lsmeans (Lenth 2016) and pbkrtest (Halekoh & Højsgaard 2014) packages. F-tests were calculated according to the Kenward-Roger method (Kenward & Roger, 1997, Halekoh,

2014). In these models, a significant treatment effect would indicate that the OTCs effectively changed the microclimate relative to ambient conditions, while a significant interaction between treatment and year would indicate that the effect of OTC impacts on microclimate varied temporally with other unmeasured factors varying with year.

3.3.3 Population growth response

Because growth in a given year is indirectly the result of the effects of the previous year's environmental conditions on root biomass, reproduction and storage organs, all analyses related environmental data from year t to ramet and plant counts in year t + 1.

Population growth for D. integrifolia and A. parviflora were calculated separately as the

161

difference in the number of ramets and plants within plots between years (2013 vs. 2012,

2014 vs. 2013). Species were combined for this analysis, and since counts between species varied strongly, they were centered within species by subtracting the overall mean from the raw count per species.

3.3.3.1 Population growth response to experimental warming: OTCs versus controls

To assess if the OTCs affected population growth, the change in the number of ramets

(D. integrifolia) and individual plants (A. parviflora) was compared between OTCs (n =

10) and controls (n = 10) at Wilson Brook (2013 vs. 2012, 2014 vs. 2013). The change in the number of ramets or plants was used as a response variable; treatment and year were the predictor variables, and individual plot was a random variable in linear mixed effects models. The full factorial model including the interaction between time and treatment was compared with reduced models that did not include either (1) the interaction term, and (2) the term of interest (treatment). Models were compared using log likelihood, and the best-fit model selected based on Chi-square. All interactions included in the model (treatment and year) were non-significant, were not included in the final model, and are not reported below.

3.3.3.2 Population growth response to microclimate variables

Results of the above models revealed no significant effect of treatment on ramet / plant number. For that reason, treatments were combined for further analysis. Population

162

growth response was assessed using ramet / plant counts as the response variable, and

GDD, annual temperature frequency classes as fixed effects, with plot as the random term. The full factorial model including the interaction between time and treatment was compared with reduced models that did not include either (1) the interaction term, or (2) the term of interest (treatment). Models were compared using log likelihood, and the best-fit model selected based on Chi-square. Temperature frequencies were centered to fit the models since the numbers were highly variable.

163

3.4 Results

3.4.1 Regional climates and local microclimates

Regional climate comparisons revealed a significant interaction between year and location for mean daily temperatures (χ2 = 38.93, df = 4, P < 0.001). Mean daily temperatures over the three years were significantly higher at Wilson Brook and in southeastern New Brunswick region in general than in northwestern Newfoundland (Fig.

3.5, Appendix 3.2). The regional mean daily temperatures for northwestern

Newfoundland were significantly higher in 2012 than in either 2011 or 2013, with no among-year differences within either of the other locations (Appendix 3.2). Growing degree days were significantly lower in northwestern Newfoundland than in southeastern New Brunswick and Wilson Brook in all three years (Fig. 3.6, P < 0.001); there was no effect of year. Evidence from linear models showed that month-to-month variation in mean daily temperatures differed among locations (analysis including all four sites for 2012 only, χ2 = 133.88, df = 12, P < 0.001). Mean daily temperatures at

Wilson Brook were not significantly different from southeastern New Brunswick, but local Port-au-Choix mean daily temperatures were significantly higher than northwestern Newfoundland, except in October (Fig. 3.7, Appendix 3.3). Southeastern

New Brunswick mean daily temperatures were significantly higher than northwestern

Newfoundland, except in September and October; northwestern Newfoundland mean daily temperatures were significantly lower than Wilson Brook only in June and July.

164

At the site level, there was no significant variation in mean, minimum or maximum daily temperature, or GDD between Wilson Brook and the more distant Port-au-Choix site

(Appendix 3.4). VPD was significantly higher for Wilson Brook than for Port-au-Choix in 2012 (0.25 kPa ± 0.04 SE, χ2 = 17.34, df = 1, P < 0.001, Fig. 3.8). As was expected based on geographic proximity, Wilson Brook and Albert Mines did not differ in mean or minimum daily temperatures or GDD (Appendix 3.5). However, maximum daily temperatures were significantly higher 3.36 ± 1.17 °C (mean ± SE) at Wilson Brook than at Albert Mines (χ2 = 5.43, df = 1, P = 0.02, Fig. 3.9). In addition, VPD was significantly lower by 0.12 kPa ± 0.01 SE for Wilson Brook than for Albert Mines (χ2 =

17.74, df = 1, P < 0.001, Fig. 3.10).

Since I expected temperatures at ground level to be higher than at the level of standardized measurements, these data suggest that conditions at Wilson Brook were cooler than regional New Brunswick climate for most variables, and similar to Port-au-

Choix in Newfoundland, as well as Albert Mines, the local control site.

3.4.2 Effectiveness of warming experiment

It was expected that OTCs would raise temperatures in experimental plots by at least 1

°C, resulting in higher mean and maximum daily temperatures as well as higher effective GDD and VPD than in control plots. In contrast, OTCs showed some cooling.

Maximum daily temperatures were lower in OTCs compared to controls in 2013 (P <

0.001, Fig 3.11), the frequency of temperatures in the 25–40°C range was greater in

165

controls 2013 (P = 0.030, Fig. 3.12 b), and OTCs reduced the frequency of occurrences of the temperature range >40 °C all three years (χ2 = 5.16, df = 6, P = 0.023, Fig. 3.12 a). No effect of OTCs on GDD (F41,40 = 1.08, P = 0.31) was detected.

Treatment affected temperatures differently depending on the year: there were significant interactions of year with treatment for mean daily temperatures (χ2 = 7.58, df

= 2, P = 0.02), maximum daily temperatures (χ2 = 19.31, df = 2, P < 0.001), VPD (χ2 =

58.62, df = 2, P < 0.001), and temperature ranges 25 to 40 °C (χ2 = 10.62, df = 2, P =

0.005) (see Appendix 3.6).

There were also significant differences in temperature between OTCs and ambient plots among years. Mean daily temperatures were highest in 2012 in the OTC plots compared to all 2011 and 2013 (Fig. 3.13, Appendix 3.6b), and maximum daily temperatures were significantly different between years for OTCs. OTCs only lowered maximum daily temperatures significantly compared to controls in 2013 (Fig 3.11, Appendix 3.6b). VPD were significantly different among years for controls and OTCs (Fig. 3.14, Appendix

3.6c). The frequency of temperatures in the 25–40°C was greater in controls with a significant difference in 2013 (Fig. 3.12 b, Appendix 3.6d).

3.4.3 Population growth response to microclimate at Wilson Brook

Population growth did not differ between OTCs and controls for either species, and there was no interaction between year and treatment (χ2 = 0.003, df = 1, P = 0.959, Fig. 3.15).

166

OTCs and controls were therefore pooled for further analyses, but abundance did not differ among years or as a function of GDD or temperature using pooled data.

167

3.5 Discussion

To assess the sensitivity of rear-edge disjunct populations to predicted future warming, this study used two complementary approaches: documenting current microclimate at sites with and without arctic-alpine species, and experimental manipulation of microclimate to evaluate population growth response of two species. The microclimate at Wilson Brook was expected to be warmer than the regional climate simply due to differences in instrument height and associated boundary layer effects. However, since

Wilson Brook was similar to southeastern New Brunswick over the three years, cooling seems to occur at Wilson Brook. This suggests that the persistence of rear-edge, disjunct populations of two arctic-alpine species at Wilson Brook may be attributable to a cooler local microclimate compared to regional climate. Further, the warming experiment did not increase temperatures in the experimental plots as intended; instead, OTCs reduced the frequency of high temperatures, and may have cooled the plots overall. While population sizes of D. integrifolia and A. parviflora varied among years, I found no biologically significant associations of microclimate variables with population growth.

Below I discuss the microclimate at Wilson Brook, the unexpected cooling effect of the

OTCs and the apparent absence of an association between population and microclimate variables.

3.5.1 Microclimate at Wilson Brook

Based on the occurrence of arctic-alpine species, the Wilson Brook site was expected to

168

display a cooler ground-level microclimate than that of the broader region in which it occurs, and microclimate characteristics similar to those further north in Newfoundland.

For example, occurrences of arctic-alpine vascular plant elements far south of their current range on Lake Superior have been attributed to a harsh microclimate resembling arctic conditions (Given & Soper 1981).

Comparison of the microclimate at the study sites must be interpreted with caution, given the limited data (three years for Wilson Brook, and one for Albert Mines and Port- au-Choix). However, several generalizations are worth noting. Although the regional temperature data lack precision, mean daily temperatures and GDD during the growing season at Wilson Brook did not differ significantly from the regional average for New

Brunswick. Given that microclimate is expected to differ significantly from standardized observations (Geiger et al. 2009), I would have expected significantly higher ground- level temperatures than regional temperatures. This principle is reflected in the fact that local microclimate at Port-au-Choix was significantly higher (by over 1.6 °C) than weather station data from northwestern Newfoundland in 2012, during the summer months. In contrast, particular landforms such as rock outcrops have been shown to reduce temperature and increase moisture , while also reducing the variability in temperature (Locosselli et al. 2016). Wilson Brook did not differ from Port-au-Choix in any of the temperature metrics examined: daily mean, minimum, or maximum, or GDD;

I expected Wilson Brook temperatures to be significantly higher than Port-au-Choix due to the more northerly location of the latter. The lack of difference between Wilson Brook

169

and the regional climate as well as Port-au-Choix provides indirect evidence that Wilson

Brook may induce local cooling, which my OTC experiment suggests may be due to substrate conditions as described in section 3.5.1.1.

Unexpectedly, Wilson Brook had higher maximum daily temperatures than the nearby

Albert Mines site in 2013, although no other variables differed between the sites. It should be noted that Albert Mines measurements were limited to one growing season, and conditions at which the sensors were placed (different slopes at the two sites, slightly different aspects) may have influenced those measurements. If Albert Mines received maximum irradiation at a different time of day than Wilson Brook, this could explain differences in daily maximum temperatures.

Significant differences in VPD at Wilson Brook compared to both Albert Mines and Port au Choix were detected, but are unlikely to be biologically significant because (a) the differences were small (< 0.25 kPa), and (b) all mean daily VPD values during the three years over all plots (0.12 ± 0.00 SE) were well below the putatively ideal range (0.45 –

1.25 kPa) for commercial plants grown under greenhouse conditions (Leuschner 2002), which is likely a conservative estimate for wild plants. The ideal range of VPD for the study species is unknown, but it seems unlikely that it constitutes a critical stress at these sites.

It is possible that there are additional subtle critical microclimatic differences that were not considered. For example, cooler microclimate may be more pronounced during some

170

months, and may influence key processes such as flowering regardless of the length of the growing season.

It was interesting to note that Albert Mines, a proximal site with similar substrate and lower maximum daily temperatures than Wilson Brook, contains none of the study species. The absence of arctic-alpine species at Albert Mines is not likely due to substrate or microclimate conditions alone. Albert Mines is a former gypsum mine site, and experiences considerable disturbance (S. Dietz, pers. obs.). Further studies could evaluate limitations to the establishment of arctic-alpine species at Albert Mines that may be imposed by competition from other species, dispersal limitations, and human disturbance, or assess the potential to introduce arctic-alpine species to the site.

3.5.1.1 Cooling effect of Open Top Chambers at Wilson Brook

Rather than the expected warming, the OTCs at Wilson Brook lowered the maximum daily temperatures relative to control plots in one year out of three. Furthermore, OTCs had significantly fewer high temperature events (above 40 ºC) than control plots, and fewer occurrences of temperatures between 25 and 40 ºC in 2013. Such unexpected results have been occasionally noted in the literature, and several explanations were considered, but seem unlikely. The period of recording (122 days) was consistent across years, and occurred during the snow free season, avoiding potential impacts of snow load inside the OTC (noted by Bokhorst et al. 2011). Shading by increasing canopy was implicated in a decrease in soil surface temperature by Kudo & Suzuki (2003), while

171

Jonsdottir et al. (2005) were not able to explain a decrease in soil temperature in treated plots. Hollister et al. (2006) associated dense plant cover with higher ground temperature as well as high albedo with lower ground temperature. An increase in shading was not observed during the study duration, and OTCs presumably received the same shading as the control plots given the paired study design. Marion et al. (1997) suggested that both chamber size and shading by the chamber itself could reduce the intended heating effect.

While my OTCs (32 cm diameter, 15 cm high) were somewhat smaller than OTCs most frequently used in the Arctic (104 cm basal diameter), OTCs as small as 0.5 m2 are not uncommon and have effectively increased temperature (Arft et al. 1999, Day et al. 1999,

Walker et al. 2006, Aronson & McNulty 2009, Elmendorf et al. 2012). As such, size alone should not have caused the effect observed. The material used in this study has been effectively used in previous studies and should have achieved good irradiation in the OTC plots. Sun-Lite HP was chosen for its high solar transmittance (86%) and low long-wave transmittance, which would minimize shading and maximize warming. Since the OTCs were not placed level but at varying aspects, this might have caused some internal shading. However, the angles of OTCs along the slope were selected randomly, and each was consistent across years. Future studies should assess the possibility of decreased solar radiation caused by non-horizontal OTCs.

Another explanation for the failure of the OTCs to warm the plots may also provide an explanation for the conditions allowing the arctic-alpine species to persist at Wilson

Brook. The highly porous gypsum substrate of the cliff has cavities underneath and in its

172

face, which are filled with water and / or ice. Indeed, visible ice persisted in early June

2013 in a shaded crevice in the cliff face (S. Dietz, pers. obs.). Water moves up easily through gypsum soils; under conditions of adequate water supply this can maintain a continuously moist and cool surface environment (Meyer et al. 1992, Escudero et al.

2015). OTCs might well have caused a localized cooling effect by slowing air movement and retaining the substrate-cooled moisture. It is not clear why this effect would be more pronounced in some years than in others. It has to be noted that precipitation, which might have played a role, was not recorded in this study. In the broader sense, the gypsum substrate may provide the mechanism for these species to persist through effects on local air movement resulting in a cooler ground level microclimate. Elsewhere, alpine species occur below their usual elevation range in micro habitats associated with unstable glacial debris and subsurface ice (Fickert et al. 2007).

In evaluating microrefugia in mountainous landscapes, Dobrowski (2011) argued that local terrain conditions (e.g., cool air pooling, temperature inversions, slope and aspect, elevation (Gentili et al. 2015)) can decouple microclimate from regional climate.

However, further study is required to understand the influence of this substrate, including comparison of above- vs. belowground temperature and moisture content, and measures of establishment success of competing species at the site.

3.5.2 Population growth and microclimatic variation at Wilson Brook

Two species of the genus Dryas (D. integrifolia and D. octopetala) have shown both short- and long-term responses to warming in International Tundra Experiment (ITEX)

173

studies (Murray 1997). All of the sites monitored through ITEX are in arctic / alpine locations, and the same increased growth response to warming would not have been expected at southern sites, since temperatures here are presumed to be at the upper tolerance limit. If D. integrifolia was at the upper limit of its range of temperature tolerance at Wilson Brook, detectable changes would be expected in a three-year experiment. For example, D. integrifolia responded positively after one year to additional snow pack (Legault & Cusa 2015), and to OTC warming (with increased flower production) (Robinson 2014). Tolvanen & Henry (2001) documented positive responses to warming and nutrient addition in five growing seasons. In longer-term warming studies of D. integrifolia, plant height and total cover increased over 9 years

(Welker et al. 2004); flower and seed biomass, and seed germination increased over 12 years (Klady et al. 2011); and leaf size and plant height increased over 16 years (Hudson

& Henry 2010, Hudson et al. 2011). Together, these studies indicate that D. integrifolia can display important biological responses to increased temperatures, in both the short- and long-term. In addition, I would have expected a positive response to cooling if the microclimate at the southern range edge had reached temperatures that exceeded either species' limits of tolerance.

This study does not provide evidence to support microclimate as a limiting factor for the study species in southeastern New Brunswick. The fact that I found neither a response of the two species to local variability in microclimate, nor a positive or negative response to cooling, suggests that the species are relatively robust to the variation in temperature

174

experienced at this most southerly location. Although the OTCs did not function as intended, the study provides useful information on population changes in the two species at a disjunct rear-edge site, and provides insights into knowledge gaps to be addressed by future studies.

Future research should explore the soil water content to determine if the substrate may play a role in moderating temperatures at the site. To my knowledge, no ecotypes of D. integrifolia and A. parviflora have been identified across the range edge. Reciprocal transplant experiments, including individuals from different locations across the range, should be undertaken to elucidate if the Wilson Brook populations show specific local adaptation to temperature variability, providing information on their capacity to deal with temperature changes at this southern range edge. In addition, transplant experiments with seeds, seedlings and adult plants beyond the current range limit could help identify the existence of limitations to the establishment of the two species beyond their ranges.

175

3.6 Conservation implications for Wilson Brook

The presence of populations of Dryas integrifolia and Anemone parviflora at the Wilson

Brook site is clear evidence that the historic and current microclimates were / are within their ecological ranges of tolerance. Further, the absence of a population growth response to temperatures above the assumed optimum suggests that, in the short term, these populations are unlikely to succumb to warming. Seedling establishment success is unknown for the site. Because the OTCs did not increase temperatures, it is impossible to determine what range of increasing temperatures associated with climate change these arctic-alpine species will tolerate. It is possible that site conditions, such as buffering of temperatures caused by the substrate, as well as the topography of the site (a north- facing, shaded slope) could play a role in maintaining arctic-alpine species at Wilson

Brook in the face of a changing climate.

Given that the site is already protected as a provincial Protected Natural Area with minimal human disturbance, there does not appear to be a need to change its management. It is possible that the slope might destabilize in the future due to erosion, or that surrounding species might encroach on these populations. If that becomes a concern, monitoring should be implemented to detect changes in population sizes, level of cliff erosion, and intrusion by other species, especially by weedy, non-native species that occur in the area. I would argue that the arctic-alpine populations and their habitats should be conserved because they are an important component of New Brunswick's

176

natural heritage, representing a unique assemblage of plants, in a unique habitat.

177

Tables

Table 3.1. Summary of micro data collection at 10 min intervals at three sites, as well as regional data used from Environment Canada (Government of Canada 2016).

Comparison Location Years Sampling with local period Wilson Brook a) Regional Alma (EC), 2011 to June 1 to climate Moncton (EC) 2013 November 9 Nappan (EC) Daniels Harbour (EC) Ferolle Point (EC) Plum Point (EC) b) Regional Alma (EC), 2012 23 May to 19 climate NB, Moncton (EC) October for micro & regional Nappan (EC) Port-au-Choix, NL Port-au-Choix to 9 November Daniels Harbour (EC) for all others Ferolle Point (EC) Plum Point (EC) c) Micro NL Port-au-Choix 2012 23 May to 19 October d) Micro local Albert Mines 2013 22 April to (control) November 9 e) Micro local Wilson Brook 2011 to June 1 to 2013 November 9 2012 23 May to November 9 2013 22 April to November 9

178

Figures

179

Figure 3.2. Wilson Brook Protected Natural Area, New Brunswick, Canada, gypsum cliff habitat. Low vegetation areas are Dryas integrifolia patches. Photo: Sabine Dietz, 2011

180

181

182

NB-regional Figure 3.5. Post-hoc comparisons (Tukey contrasts) of the linear mixed effects model of location vs. year interactions: mean daily temperatures (mean ± first and third quartile) for Wilson Brook (local data, n = 5, WB), and three regional weather stations each in Newfoundland (Daniels Harbour, Ferrolle Point, Plum Point, NL-regional) and New Brunswick (Alma, Moncton, Nappan NS, NB-regional, data downloaded from Government of Canada (2016), over three growing seasons (1 June to 9 November, 2011 & 2013). The least-squared mean estimates for the fixed effect location were obtained using the lsmeans (Lenth 2016) and pbkrtest (Halekoh & Højsgaard 2014) packages. There was a significant interaction between location and year (χ2 = 38. 93, df = 4, P < 0.001). There was a significant difference between Wilson Brook and NB region in 2011, with mean daily temperatures in NB region significantly higher than Wilson Brook (P = 0.027); this difference was not apparent in other years. Newfoundland regional mean daily temperatures were significantly lower than both Wilson Brook (except in 2012) and NB region (Appendix 3.4).

183

Figure 3.6. Post-hoc comparisons (Tukey contrasts) of growing degree days (mean ± first and third quartile) by year from weather stations in New Brunswick (NB-regional, Alma, Moncton, Nappan NS), Newfoundland (NL-regional, Daniels Harbour, Ferrolle Point, Plum Point), and Wilson Brook (n = 5), 1 June - 9 November (2011 to 2013). Regional data from (Government of Canada 2016). There was no significant interaction of year and location (F20, 24 = 2.70, P = 0.059). GDD was significantly lower in NL- regional than in NB-regional (by 515 GDD, P < 0.001) or Wilson Brook (by 630 GDD, P < 0.001), over three years.

184

185

186

Maximum daily temperatures (ºC)

187

188

189

Figure 3.12. Comparison of frequency of the occurrence of temperatures between (a) > 40°C and (b) 25 and 40 °C in OTCs (white bars, n = 10) versus control plots (shaded bars, n = 5) at Wilson Brook from 1 June to 9 November 2011 to 2013, by year ((a) mean ± SD, (b) mean ± first and third quartile). (a) There were fewer > 40°C temperatures in the OTCs (χ2 = 5.16, df = 6, P = 0.023); (b) There was a significant interaction between year and treatment (χ2 = 10.62, df = 1, P = 0.005). There were significantly fewer occurrences of temperatures between 25 and 40 °C in OTCs in 2013 than in controls (P = 0.050), and there were significant differences within treatments among years.

190

191

192

193

List of references

Alberta Agriculture and Forestry. 2016. Management of the greenhouse environment. Alberta. Available from: http://www1.agric.gov.ab.ca/$department/deptdocs.nsf/all/opp2902.

Angert AL. 2006. Demography of central and marginal populations of monkeyflowers (Mimulus cardinalis and M. lewisii). Ecology 87(8):2014–2025.

Aronson EL, McNulty SG. 2009. Appropriate experimental ecosystem warming methods by ecosystem, objective, and practicality. Agricultural and Forest Meteorology 149(11):1791–1799.

Ashcroft MB, Gollan JR. 2013. Moisture, thermal inertia, and the spatial distributions of near-surface soil and air temperatures: Understanding factors that promote microrefugia. Agricultural and Forest Meteorology 17(2013):677–89.

Bates D, Maechler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67(1):1–48.

Beaumont LJ, Pitman AJ, Poulsen M, Hughes L. 2007. Where will species go? Incorporating new advances in climate modelling into projections of species distributions. Global Change Biology 13(7):1368–1385.

Bell KL, Bliss LC. 1980. Plant reproduction in a high arctic environment. Arctic and Alpine Research 12(1):1–10.

Billings WD, Mooney HA. 1968. The ecology of arctic and alpine plants. Biological

194

Review 43(4):481–529.

Birks HH. 2008. The Late-Quaternary history of arctic and alpine plants. Plant Ecology and Diversity 1(2):135–146.

Bliss LC. 1966. Plant productivity in alpine microenvironments on Mt. Washington, New Hampshire. Ecological Monographs 36(2):125–155.

Breen AL, Murray DF, Raynolds MK, Timling I, Walker DA. 2014. Ecology and evolution of plants in arctic and alpine environments. In: Plant Ecology and Evolution in Harsh Environments. N Rajakaruna, Boyd RS, Harris TB, editors. USA: Nova Science Publishers, Inc. pp. 149–177.

Brouwers NC, Mercer J, Lyons T, Poot P, Veneklaas E, Hardy G. 2013. Climate and landscape drivers of tree decline in a Mediterranean ecoregion. Ecology and Evolution 3(1):67–79.

Buckley LB, Kingsolver JG. 2012. Functional and phylogenetic approaches to forecasting species' responses to climate change. In: Annual Review of Ecology, Evolution, and Systematics. DJ Futuyma, editor. pp. 205–205.

Büntgen U, Martínez-Peña F, Aldea J, Rigling A, Fischer EM, Camarero JJ, Hayes MJ, Fatton V, Egli S. 2013. Declining pine growth in central Spain coincides with increasing diurnal temperature range since the 1970s. Global and Planetary Change 107(2013):177–185.

Busch JW. 2005. The evolution of self-compatibility in geographically peripheral populations of Leavenworthia alabamica (Brassicaceae). American Journal of Botany 92(9):1503–1512.

Cahill AE, Aiello-Lammens ME, Caitlin Fisher-Reid M, Hua X, Karanewsky CJ, Ryu HY, Sbeglia GC, Spagnolo F, Waldron JB, Wiens JJ. 2014. Causes of warm- edge range limits: systematic review, proximate factors and implications for climate change. Journal of Biogeography 41(3):429–442.

Chapin FSI, Shaver GR. 1996. Physiological and growth responses of arctic plants to a field experiment simulating climatic change. Ecology 77(3):822–840.

Chapin III FS. 1983. Direct and indirect effects of temperature on arctic plants. Polar

195

Biology 2(1):47–52.

Clayden SR. 1994. Rare arctic plants in south-eastern New Brunswick. Saint John, NB: NB Museum.

Clayden SR. 2000. History, physical setting, and regional variation of the flora of New Brunswick. In: Flora of New Brunswick. HR Hinds, editor. Fredericton, NB: Biology Department, University of New Brunswick. pp. 35–73.

Crawford RMM. 2008. Plants at the margin – ecological limits and climate change Cambridge, UK: Cambridge University Press. 478pp.

Dahl E. 1951. On the relation between summer temperature and the distribution of alpine vascular plants in the lowlands of Fennoscandia. Oikos 3(1):22–52.

Darling E, Samis KE, Eckert CG. 2007. Increased seed dispersal potential towards geographic range limits in a Pacific coast dune plant. New Phytologist 178(2):424–435.

Day TA, Ruhland CT, Grobe CW, Xiong F. 1999. Growth and reproduction of Antarctic vascular plants in response to warming and UV radiation reductions in the field. Oecologia 119(1):24–35.

Doak DF, Morris WF. 2010. Demographic compensation and tipping points in climate- induced range shifts. Nature 467(7318):959–962.

Dobrowski SZ. 2011. A climatic basis for microrefugia: the influence of terrain on climate. Global Change Biology 17(2):1022–1035.

Escudero A, Palacio S, Maestre FT, Luzuriaga AL. 2015. Plant life on gypsum: a review of its multiple facets. Biological Reviews (Cambridge) 90(1):1–18.

Fickert T, Friend D, Grüninger F, Molnia B, Richter M. 2007. Did debris-covered glaciers serve as Pleistocene refugia for plants? A new hypothesis derived from observations of recent plant growth on glacier surfaces. Arctic and Alpine Research 39(2):245–257.

Field CB, Barros VR, Mastrandrea MD, Mach KJ, Abdrabo MA, Adger WN, Anokhin YA, Anisimov OA, Arent DJ, Barnett J, et al. 2014. Summary for Policymakers

196

Climate Change 2014: Impacts, Adaptation, and Vulnerability Working Group II Contribution to the IPCC 5th Assessment Report Cambridge, UK and New York, USA: Cambridge University Press. 32pp.

Foden W, Midgley GF, Hughes G, Bond WJ, Thuiller W, Hoffman MT, Kaleme P, Underhill LG, Rebelo A, Hannah L. 2007. A changing climate is eroding the geographical range of the Namib Desert tree Aloe through population declines and dispersal lags. Diversity and Distributions 13(5):645–653.

Fonda RW, Bliss LC. 1966. Annual carbohydrate cycle of alpine plants on Mt. Washington, New Hampshire. Bulletin Torrey Botanical Club 93(4):268–277.

Gauslaa Y. 1984. Heat-resistance and energy budget in different Scandinavian plants. Holarctic Ecology 7(1):1–78.

Gazol A, Camarero JJ, Gutiérrez E, Popa I, Andreu-Hayles L, Motta R, Nola P, Ribas M, SangüesaBarreda G, Urbinati C. 2015. Distinct effects of climate warming on populations of silver fir (Abies alba) across Europe. Journal of Biogeography 42(6):1150–1162.

GBIF.org. 2016. GBIF Occurrence Download. 2016 ed.: GBIF.org. Available from: http://www.gbif.org.

Geiger R, Aron RH, Todhunter P. 2009. The climate near the ground Cambridge: Rowman & Littlefield. 611pp.

Gentili R, Baroni C, Caccianiga M, Armiraglio S, Ghiani A, Citterio S. 2015. Potential warm-stage microrefugia for alpine plants: feedback between geomorphological and biological processes. Ecological Complexity 21(2015):87–99.

Given DR, Soper JH. 1981. The arctic-alpine element of the vascular flora at Lake Superior. Publications in Botany 10 Ottawa.

Government of Canada. 2016. Past weather and climate - historical data. Ottawa: Environment and Climate Change Canada. Available from: http://climate.weather.gc.ca/historical_data/search_historic_data_e.html.

Grigorieva EA, Matzarakis A, Freitas CRd. 2010. Analysis of growing degree-days as a climate impact indicator in a region with extreme annual air temperature amplitude. Climate Research 42(2):143–154.

Hadley EB, Bliss LC. 1964. Energy relationships of alpine plants on Mt. Washington, New Hampshire. Ecological Monographs 34(4):331–357.

197

Halekoh U, Højsgaard S. 2014. A Kenward-Roger approximation and parametric bootstrap methods for tests in linear mixed models–the R package pbkrtest. Journal of Statistical Software 59(9):1–30.

Hampe A. 2005. Fecundity limits in Frangula alnus (Rhamnaceae) relict populations at the species’ southern range margin. Oecologia 143(3):377–386.

Hampe A, Jump AS. 2011. Climate Relicts: Past, Present, Future. Annual Review of Ecology, Evolution and Systematics 42(2011):313–333.

Hampe A, Petit RJ. 2005. Conserving biodiversity under climate change: the rear edge matters. Ecology Letters 8(5):461–467.

Hannah L, Flint L, Syphard AD, Moritz MA, Buckley LB, McCullough IM. 2014. Fine- grain modeling of species’ response to climate change: holdouts, stepping-stones, and microrefugia. Trends in Ecology and Evolution 29(7): 390–397.

Hollister RD, Webber PJ, Nelson FE, Tweedie CE. 2006. Soil thaw and temperature response to air warming varies by plant community: Results from an open-top chamber experiment in northern Alaska. Arctic, Antarctic and Alpine Research 38(2):206–215.

Hothorn T, Bretz F, Westfall P. 2008. Simultaneous Inference in General Parametric Models. Biometrical Journal 50(3):346–363.

Hudson JM, Henry GH. 2010. High Arctic plant community resists 15 years of experimental warming. Journal of Ecology 98(5):1035–1041.

Hudson JMG, Henry GHR, Cornwell WK. 2011. Taller and larger: shifts in Arctic tundra leaf traits after 16 years of experimental warming. Global Change Biology 17(2):1013–1021.

Hylander K, Ehrlen J, Luoto M, Meineri E. 2015. Microrefugia: Not for everyone. AMBIO 44(1):60-68.

Jäkäläniemi A. 2011. Narrow climate and habitat envelope affect the survival of relict populations of a northern Arnica angustifolia. Environmental and Experimental Botany 72(3):415–421.

Jonsdottir IS, Magnusson B, Gudmundsson J, Elmarsdottir A, Hjartarson H. 2005. Variable sensitivity of plant communities in Iceland to experimental warming. Global Change Biology 11(4):553–563.

Jump AS, Woodward FI. 2003. Seed production and population density decline

198

approaching the range-edge of Cirsium species. New Phytologist 160(2):349– 358.

Klady RA, Henry GHR, Lemay V. 2011. Changes in high arctic tundra plant reproduction in response to long-term experimental warming. Global Change Biology 17(4):1611–1624.

Kudo G, Suzuki S. 2003. Warming effects on growth, production, and vegetation structure of alpine shrubs: a five-year experiment in northern Japan. Oecologia (Berlin) 135(2):280–287.

Lambers H, Chapin FSI, Pons TL. 2008. Plant Physiological Ecology New York: Springer-Verlag. 605pp.

Larkindale J, Mishkind M, Vierling E. 2005. Plant responses to high temperature. In: Plant Abiotic Stress. MA Jenks, Hasegawa P, editors. Oxford, UK: Blackwell Publishing. pp. 100–144.

Lázaro-Nogal A, Matesanz S, Hallik L, Krasnova A, Traveset A, Valladares F. 2016. Population differentiation in a Mediterranean relict shrub: the potential role of local adaptation for coping with climate change. Oecologia 180(4):1075-1090.

Legault G, Cusa M. 2015. Temperature and delayed snowmelt jointly affect the vegetative and reproductive phenologies of four sub-Arctic plants. Polar Biology 38(10):1701–1711.

Leimu R, Fischer M. 2008. A meta-analysis of local adaptation in plants. PLoS ONE 3(12):e4010 .

Lenth RV. 2016. Least-Squares Means: the R package lsmeans. Journal of Statistical Software 69(1):1–33.

Lesica P, Allendorf FW. 1995. When are peripheral populations valuable for conservation? Conservation Biology 9(4):753–760.

Leuschner C. 2002. Air humidity as an ecological factor for woodland herbs: leaf water status, nutrient uptake, leaf anatomy, and productivity of eight species grown at low or high vpd levels. Flora (Jena) 197(4):262–274.

Levin DA. 2012. Mating system shifts on the trailing edge. Annals of Botany 109(613):613–620.

Locosselli GM, Cardim RH, Ceccantini G. 2016. Rock outcrops reduce temperature- induced stress for tropical conifer by decoupling regional climate in the semiarid

199

environment. International Journal of Biometeorology 60(5):639-649.

McMaster GS, Wilhelm WW. 1997. Growing degree-days: one equation, two interpretations. Agricultural and Forest Meteorology 87(4):291–300.

Meyer SE, Garcia-Moya E, Lagunes-Espinoza LDC. 1992. Topographic and soil surface effects on gypsophile plant community patterns in central Mexico. Journal of Vegetation Science 3(4):429–438.

Michalski SG, Durka W. 2007. High selfing and high inbreeding depression in peripheral populaions of Juncus atratus. Molecular Ecology 16(22):4715–4727.

Mimura M, Aitken SN. 2007. Increased selfing and decreased effective pollen donor number in peripheral relative to central populations in Picea sitchensis (Pinaceae). American Journal of Botany 94(6):991–998.

Moeller DA, Geber MA, Eckhart VM, Tiffin P. 2012. Reduced pollinator service and elevated pollen limitation at the geographic range limit of an annual plant. Ecology 93(5):1036–1048.

Molau U, Molgaard P. 1996. ITEX Manual. Second edition ed. Copenhagen: Danish Polar Centre. 21pp.

Mooney HA, Billings WD. 1961. Comparative physiological ecology of arctic and alpine populations of Oxyria digyna. Ecological Monographs 31(1):1–29.

Murray DF. 1997. Systematics of the ITEX species. Global Change Biology 3(S1):10– 19.

NatureServe. 2016. NatureServe Explorer: an online encyclopaedia of life. NatureServe Explorer: an online encyclopaedia of life. 2008 ed. Virginia, U.S.A: NatureServe. Available from: http://explorer.natureserve.org/.

Neilson RO, Pitelka LF, Solomon AM, Nathan R. 2005. Forecasting regional to global plant migration in response to climate change. Bioscience 55(9):749–759.

200

Nieuwenhuis R, Grotenhuis Mt, Pelzer B. 2012. Influence.ME: tools for detecting influential data in mixed effects models. R Journal 4(2):38–47.

Normand S, Treier UA, Randin C, Vittoz P, Guisan A, Svenning J-C. 2009. Importance of abiotic stress as a range-limit determinant for European plants: insights from species responses to climatic gradients. Global Ecology and Biogeography 18(4):437–449.

Parmesan C. 2006. Ecological and evolutionary responses to recent climate change. Annual Review of Ecology, Evolution, and Systematics 37(2006):637–669.

Parmesan C. 2007. Influences of species, latitudes and methodologies on estimates of phenological response to global warming. Global Change Biology 13(9):1860– 1872.

Parmesan C, Yohe G. 2003. A globally coherent fingerprint of climate change impacts across natural systems. Nature 421(6918):37–42.

Pellissier L, Bråthen KA, Vittoz P, Yoccoz NG, Dubuis A, Meier ES, Zimmermann NE, Randin CF, Thuiller W, Garraud L, et al. 2013. Thermal niches are more conserved at cold than warm limits in arctic-alpine plant species. Global Ecology and Biogeography 22(8):933–941.

Provan J, Maggs CA. 2012. Unique genetic variation at a species’ rear edge is under threat from global climate change. Proceedings of the Royal Society B 279(1726):39–47.

R Core Team. 2015. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. Available from: http://www.R- project.org.

Read QD, Moorhead LC, Swenson NG, Bailey JK, Sanders NJ. 2014. Convergent effects of elevation on functional leaf traits within and among species. Functional Ecology 28(1):37-45.

Rehm EM, Olivas P, Stroud J, Feeley KJ. 2015. Losing your edge: climate change and the conservation value of range-edge populations. Ecology and Evolution 5(19):4315–4326.

Robinson SVJ. 2014. Insect pollination and experimental warming in the High Arctic.

201

Vancouver: University of British Columbia.

Rull V. 2009. Microrefugia. Journal of Biogeography 36(3):481–484.

Růžička V, Zacharda M, Němcová L, Šmilauer P, Nekola JC. 2012. Periglacial microclimate in low-altitude scree slopes supports relict biodiversity. Journal of Natural History 46(35–36):2145–2157.

Salzer K, Gugerli F. 2012. Reduced fitness at early life stages in peripheral versus core populations of Swiss stone pine (Pinus cembra) is not reflected by levels of inbreeding in seed families. Alpine Botany 122(2):75–85.

Santamaria L, Figuerola J, Pilon JJ, Mjelde M, Green AJ, Boer Td, King RA, Gornall RJ. 2003. Plant performance across latitude: the role of plasticity and local adaptation in an aquatic plant. Ecology 84(9):2454–2461.

Scherrer D, Koerner C. 2011. Topographically controlled thermal-habitat differentiation buffers alpine plant diversity against climate warming. Journal of Biogeography 38(2):406–416.

Souther S, Fetcher N, Fowler Z, Shaver GR, McGraw JB. 2014. Ecotypic differentiation in photosynthesis and growth of Eriophorum vaginatum along a latitudinal gradient in the Arctic tundra. Botany 92(8):551–561.

Sparks TH, Menzel A, Penuelas J, Tryjanowski P. 2011. Species response to contemporary climate change. In: The SAGE Handbook of Biogeography. A Millington, Blumler M, Schickhoff U, editors. London UK: Sage Publishing. pp. 231–242.

202

Sykes MT, Prentice C, Cramer W. 1996. A bioclimatic model for the potential distributions of north european tree species under present and future climates. Journal of Biogeography 23(2):202–233.

Thomas CD, Cameron A, Green RE, Bakkenes M, Beaumont L, Vollingham YC, Erasmus BFN, Suqueira MFd, Grainger A, Hannah L, et al. 2004. Extinction risk from climate change. Nature 427(8):145–148.

Thuiller W, Albert C, Araujo MB, Berry PM, Cabeza M, Guisan A, Hickler T, Midgely GF, Paterson J, Schurr FM, et al. 2008. Predicting global change impacts on plant species' distributions: future challenges. Perspectives in Plant Ecology Evolution and Systematics 9(3-4):137–152.

Tolvanen A, Henry GHR. 2001. Responses of carbon and nitrogen concentrations in high arctic plants to experimental warming. Canadian Journal of Botany 79(6):711–718.

Tomback DF, Schoettle AW, Chevalier KE, Jones CA. 2005. Life on the edge for limber pine: seed dispersal within a peripheral population. Ecoscience 12(4):519–529.

Tsaliki M, Diekmann M. 2009. Fitness and survival in fragmented populations of Narthecium ossifragum at the species’ range margin. Acta Oecologica 35(3):415–421.

Valladares F, Matesanz S, Guilhaumon F, Araújo MB, Balaguer L, Benito-Garzón M, Cornwell W, Zavala MA. 2014. The effects of phenotypic plasticity and local adaptation on forecasts of species range shifts under climate change. Ecology Letters 17(11):1351–1364.

Van Mantgem PJ, Stephenson NL. 2007. Apparent climatically induced increase of tree mortality rates in a temperate forest. Ecology Letters 10(10):909–916.

Vaupel A, Matthies D. 2012. Abundance, reproduction, and seed predation of an alpine plant decrease from the center toward the range limit. Ecology 93(10):2253– 2262.

Vergeer P, Kunin WE. 2013. Adaptation at range margins: common garden trials and the performance of Arabidopsis lyrata across its north-western European range. New Phytologist 197(3):989–1001.

Vicente-Serrano SM, Camarero JJ, Zabalza J, Sangüesa-Barreda G, López-Moreno JI, Tague CL. 2015. Evapotranspiration deficit controls net primary production and growth of silver fir: Implications for Circum-Mediterranean forests under forecasted warmer and drier conditions. Agricultural and Forest Meteorology 203

206(2015):45–54.

Wagner V, Wehrden Hv, Wesche K, Fedulin A, Sidorova T, Hensen I. 2011. Similar performance in central and range-edge populations of a Eurasian steppe grass under different climate and soil pH regimes. Ecography 34(3):498–506.

Walker MD, Wahren CH, Hollister RD, Henry GHR, Ahlquist LE, Alatalo JM, Bret- Harte MS, Calef MP, Callaghan TV, Carroll AB, et al. 2006. Plant community responses to experimental warming across the tundra biome. Proceedings of the National Academy of Sciences of the United States of America 103(5):1342– 1346.

Welker JM, Fahnestock JT, Henry GHR, O'Dea KW, Chimner RA. 2004. CO2 exchange in three Canadian High Arctic ecosystems: response to long-term experimental warming. Global Change Biology 10(12):1981–1995.

Welker JM, Molau U, Parsons AN, Robinson CH, Wookey PA. 1997. Responses of Dryas octopetala to ITEX environmental manipulations: a synthesis with circumpolar comparisons. Global Change Biology 3(S1):1365–2486.

Wickham H. 2016. ggplot2: Elegant Graphics for Data Analysis. New York: Springer- Verlag. Available from: http://had.co.nz/ggplot2/book.

Wipf S. 2010. Phenology, growth, and fecundity of eight subarctic tundra species in response to snowmelt manipulations. Plant Ecology 207(1):53–66.

204

Chapter 4 - Assessing the conservation significance of peripheral

populations - a case study of three arctic-alpine species in Eastern

Canada

Abstract

Conservation prioritization is usually based on a species' risk of extinction. This approach results in prioritizing conservation action for rare species, and often results in common and wide-ranging species receiving less consideration in conservation decisions. However, peripheral populations of these species may be of evolutionary and ecological significance and hold unique genetic variability that could aid in a species' ability to adapt to ongoing climate change. Despite the considerable challenges, the conservation significance of peripheral populations should be assessed based on their genetic distinctness, ecological uniqueness, and extinction risk. I assessed the conservation significance of rear-edge populations of three widespread and common arctic-alpine species based on an investigation of their levels of genetic diversity and differentiation, as well as extinction risk for two populations. The assessment indicated that some disjunct populations may be of intermediate conservation significance, and that it is important to consider rear-edge disjunct populations in conservation considerations. Future research should focus on establishing predictors that can facilitate the evaluation of genetic distinctness.

205

4.1 Introduction

The conservation of biodiversity (species diversity, genetic diversity, community and ecosystem diversity) is generally viewed as desirable, and all biodiversity may be seen as worthy of conservation for ethical (intrinsic) values. However, a wide range of potentially contradictory values (e.g., economic costs vs. social benefits) and considerations drive conservation management decisions (Fig. 4.1; Primack 2002).

These values may be direct or indirect; practical or intrinsic; current or future. For example, some would argue that decisions to conserve salmon (Salmo spp.) arise primarily from its monetary value as an economically important, consumable species

(Richardson & Loomis 2009). Narwhal (Monodon monoceros) is given high priority because it has high traditional indigenous and spiritual value, even though it is not a species at risk (Government of Canada 2005). Other species may be conserved for indirect contributions, for their role in ecosystem functioning (e.g., keystone species such as wolves, Canis spp.) and ecosystem services (e.g., vegetation, flood control)

(Primack 2002). At a more general level, the conservation of ecological communities and ecosystems such as parks and greenspaces has historically been driven by cultural and aesthetic values.

The current loss of biodiversity worldwide attributable to human activities is dramatic in its speed and magnitude, and could become the sixth mass extinction in the Earth’s history (Barnosky et al. 2011); global climate change is expected to increase this extinction risk (Thomas et al. 2004, Williams et al. 2008, Urban 2015). The concern

206

over biodiversity loss worldwide is well illustrated by the Convention on Biological

Diversity (United Nations 1992), a broad, international framework for conservation, signed and ratified by 196 countries. While the need to conserve biological diversity is widely recognized, given limited resources (time, human and financial), prioritization is required to identify those species and populations most in need of attention (Pullin et al.

2013). At the highest level, the International Convention on Biological Diversity commits signatories to identify “components of biological diversity important for its conservation...” (United Nations 1992). These components, identified in Annex 1 of the

Convention, include species that are threatened (i.e., at risk)], and of “medicinal, agricultural or other economic value; or social, scientific or cultural importance [i.e., to humans]; or importance for research into the conservation and sustainable use of biological diversity, such as indicator species [i.e., importance to the biosphere]”(United

Nations 1992). It is important to note that risk of extinction (“threatened”, Red Lists, and other terms) is the primary criterion for setting conservation priorities, even though the need to consider other criteria is often highlighted in the literature (Martín-López et al.

2011).

Many templates to prioritize species based on biological criteria have been developed over the years, all fitting within the framework of prioritizing species based on

“irreplaceability” (e.g., endemic species) relative to “vulnerability” (e.g., habitat loss)

(Brooks et al. 2006), “biological consequences of loss” (e.g., keystone species), and

“threat to persistence” (e.g., likelihood of extinction) (Taylor et al. 2011, Taylor et al.

207

2013). The most widely accepted species prioritization process is based on risk of decline or extinction, using a system of triage (Bottrill et al. 2008). This is the primary tenet of the species risk assessment of the International Union for the Conservation of

Nature (IUCN 2017), and 99 countries used the red listing of the IUCN in 2008 (50% of countries in the world) (Rodríguez 2008). In a survey of 47 countries, 82% based their conservation priority setting on the IUCN lists, or use their criteria to establish their priority lists (Miller et al. 2006). IUCN assesses species using criteria that are measures of threat (or risk of loss), for example, population size, rate of decline, and size of the distribution range (Mace et al. 2008, Pullin et al. 2013). Martín-López et al. (2011) argue that not only may this assessment process itself be biased in favor of listing species that are of scientific interest. Furthermore, it was also never meant as a primary tool to set conservation priorities, but rather intended for the identification of conservation actions.

The Committee on the Status of Wildlife in Canada (COSEWIC) bases its assessment process on the IUCN methodology as well. This includes status assessments based primarily on the abundance, range, population size, and trend of decline, as well as a threat assessment (COSEWIC 2015a). While this approach is primarily applied to species or subspecies, COSEWIC may consider other designatable units (DUs, elsewhere termed Ecologically Significant Units (Crandall et al. 2000)) of species potentially at risk, if they are disjunct, or genetically or biogeographically distinct

(COSEWIC 2015a). DUs under COSEWIC can be assessed using two criteria: distinctness, which is based on discreteness (genetic, disjunction, ecological), and

208

significance (phylogenetic divergence based on slow-evolving markers, unique ecological setting, disjunct occurrence; COSEWIC 2015b).

Conservation resources (time, funding, human effort) are primarily allocated to species that have been designated “at risk” under legal schemes such as the Species at Risk Act

(Government of Canada 2002). If a species is not categorized as legally “at risk”, priority may be determined by other conservation values such as cultural and social, but most commonly, economic / commercial (Mee et al. 2015). However, species that are widespread and common, and have no broad commercial, social or cultural value, essentially fall through the cracks of commonly applied prioritization processes and hence receive little to no conservation attention, a challenge that I will address here.

Intraspecific units (i.e., populations) of any species are even less likely to be considered if the species are not deemed to be “at risk”. As a result, populations at the range periphery of common, widespread species may be overlooked. In some instances, however, they may be placed in an “at risk” category, if they are assessed based on one jurisdiction alone. While individual jurisdictions have sole control over the protection of species, there is no ecological reason to conserve edge populations solely on the basis of their location in the range or proximity to the range edge (Hardie & Hutchings 2010), or because political units in which species are managed are misaligned with their geographic ranges (Hunter & Hutchinson 1994).

However, these edge populations may contain biologically unique variability. They may

209

be ecologically and / or genetically distinct, and contain significant ecological and evolutionary potential (Lesica & Allendorf 1995, Nantel & Gagnon 1999, Leppig &

White 2006, Reisch 2008, Hampe & Jump 2011). Thus, certain range-edge populations may be of conservation significance and their loss could increase a species' extinction risk (Sgro et al. 2011) or limit future evolutionary potential. For the purpose of this thesis, conservation significance can be defined for a population as the population’s contribution to the species' genetic biodiversity, its adaptive potential (i.e., its locally adapted genotypes, or high levels of genetic diversity), its evolutionary potential

(speciation potential), its resulting ecological distinctness, and its extirpation risk.

4.1.1 Objectives

Commonly used prioritization processes focusing on extinction risk, typically at the species level, do not capture the potential importance of peripheral populations, especially in widespread species. In the following, I argue for the consideration of peripheral populations in conservation decision-making, apply the COSEWIC risk assessment to peripheral populations of three common, widespread species, and propose a different, population-based, assessment for those species. My objectives are to (1) highlight the rationale and the challenges associated with considering peripheral populations in conservation, specifically of common, widespread species; (2) evaluate the usefulness of a commonly used risk assessment process (COSEWIC) to the conservation significance of three common, widespread arctic-alpine species (Anemone parviflora, Dryas integrifolia, Vaccinium uliginosum); and (3) propose a population-

210

based approach to assess the conservation significance of such rear-edge populations.

211

4.2 Conservation of peripheral populations

4.2.1 Potential conservation significance of peripheral populations

Peripheral populations play an important role in the ability of a species to respond to environmental changes, by allowing range expansions into newly available habitats under a warming climate, and by their ability to respond to environmental change via natural selection (Bellard et al. 2012, Eizaguirre & Baltazar–Soares 2014). Peripheral populations may contain unique genotypes not present elsewhere within the range

(Channell 2004, Hampe & Jump 2011). If gene flow between peripheral and other populations has been reduced for some time, due to dispersal limitations or local adaptation (Hardie & Hutchings 2010, Orsini et al. 2013, Shafer & Wolf 2013), peripheral populations will likely display some degree of genetic divergence. This would be expected to be greatest for refugial or relict populations, since the elapsed time for genetic divergence to have occurred is particularly long (Lesica & Allendorf 1995,

Hampe & Petit 2005, Leppig & White 2006, Millar et al. 2010, Pfeifer et al. 2010,

Dobrowski 2011, Hampe & Jump 2011, Provan & Maggs 2012, Hampe et al. 2013,

Lepais et al. 2013, Woolbright et al. 2014, Kvist et al. 2015).

In particular, rear-edge populations may be at high risk of disappearing quickly under a warming climate (Franco et al. 2006, Hampe & Jump 2011) and may be more prone to extinction than central and leading-edge populations (Vucetich & Waite 2003).

However, they may contribute significantly to a species' adaptive potential or its ability

212

to respond to environmental changes (Franco et al. 2006, Millar et al. 2010, Pfeifer et al.

2010, Hampe & Jump 2011, Provan & Maggs 2012).

4.2.2 Challenges managing peripheral populations

Peripheral populations of widespread species are rarely given conservation priority for several practical reasons. First, wide-ranging species are less frequently assessed than rare species because they are not expected to be “at risk”. Second, these species often cross multiple jurisdictions; range-edge populations are usually only assessed only if they are rare in one jurisdiction (e.g., rare in Canada). For example, of 280 terrestrial species classified as “at risk” by COSEWIC, 75% (224) were at their northern range limit (or periphery) in Canada (Gibson et al. 2009), but may not be threatened at the center of their ranges. Third, managing species across jurisdictional borders is administratively difficult. An inconsistency in including the entire range of species in assessments has resulted in peripheral populations either being automatically placed on a lower priority level (Fraser 1999), or categorized as rare at various administrative levels

(NatureServe 2016). In some cases they are given high local threat status (Abeli et al.

2009), and receive protection in individual jurisdictions without consideration at the species level. I will provide two case studies that illustrate the specific challenges surrounding management decisions for range-edge populations.

The classification of range-edge populations in British Columbia has changed over time.

Before 1992, peripheral populations of species were automatically excluded from the

213

Red List (highest conservation priority), regardless of the status of the species overall, and thus were excluded from becoming candidates for threatened or endangered status

(Fraser 1999). This policy was based on the assumptions that populations at the center of the species' distribution were sufficient to conserve the species globally, and that populations occurring at the range edge were more likely to disappear, thus were not worth the investment of limited conservation resources (Fraser 1999). In an effort to recognize the potential genetic and ecological significance of edge populations, the policy was changed in 1992, using the precautionary approach; peripheral populations of species were now allowed on the Red List. In 2004, 912 out of 1327 range-edge taxa were on the province's Red and Blue Lists, and were considered priorities for conservation (Bunnell et al. 2004). Bunnell & Squires (2004) argue that this puts a strain on scarce conservation resources, and that conservation efforts for populations of a species at the continuous range edge, such as many of those in British Columbia, are mostly doomed to failure if they are also ecologically marginal (i.e., at the limits of the species' range of tolerance). The authors suggest that efforts should focus only on disjunct range-edge populations, because they have been isolated from the range center for longer, and are thus more likely to be genetically distinct and / or locally adapted, and more likely to survive (Bunnell & Squires 2004). Applying the precautionary principle to range-edge populations may be ideal, and an appropriate first strategy, but strategic investments of limited resources require further research to assess the actual conservation significance of such populations.

214

The Atlantic Coastal Plain Flora (ACPF) is an example where the protection of an entire assemblage of range-edge populations of species has been biologically justified. The

ACPF is composed of approximately 90 species with central distributions along the

Eastern USA, and disjunct populations along the Great Lakes and in Nova Scotia

(Environment Canada & Parks Canada Agency 2010, Ciotir et al. 2013). Fifty of the

ACPF species occur as disjuncts in southwestern Nova Scotia, over 700 km from the continuous range in southern New England and New Jersey (Clayden et al. 2010). In

Nova Scotia, some ACPF species are legally protected under the provincial species at risk legislation (DNR 2015), as well as through federal legislation (Environment Canada

& Parks Canada Agency 2010), even though they are classified as globally secure (e.g.,

Hydrocotyle umbellata, Potamogeton pulcher, Eleocharis tuberculosa, Clethra alnifolia). Other ACPF species that are protected in Nova Scotia are classified as globally vulnerable (e.g., Sabatia kennedyana (NatureServe 2016)). Considerable conservation resources have been invested in managing these Nova Scotian populations over many years, and the recovery plan for the ACPF outlines numerous actions required to maintain the ACPF in the province (Environment Canada & Parks Canada

Agency 2010), but genetic and ecological significance of this flora remains to be confirmed. For example, if ACPF species originated only from refugia to the south, colonizing Nova Scotia through step-wise migration after deglaciation via long-distance dispersal events from points in the USA (Clayden et al. 2010), Nova Scotia ACPF populations may not be genetically distinct from the Eastern USA populations, and

215

would potentially be genetically impoverished as a result of founder events (Barrett et al.

1991).

Recently, Suarez-Gonzalez et al. (2015) proposed that populations of Sabatia kennedyana in Nova Scotia resulted from multiple founder events originating from offshore glacial refugia along the Atlantic Coast. While the putative refugium is now underwater (Dyke 2005, Shaw et al. 2006), genetic analyses suggest that Nova Scotia populations of S. kennedyana are genetically distinct, more diverse, and older than the

USA populations (Suarez-Gonzalez et al. 2015). Thus, populations of S. kennedyana in

Nova Scotia may be relict populations “left behind” in suitable habitat in Nova Scotia during re-colonization after deglaciation. Consequently, their conservation significance is higher, and the population may be worthy of conservation investment. However, the history and conservation significance (i.e., genetic distinctness) for other species among the ACPF flora have yet to be investigated.

The two examples presented above highlight the practical challenges in considering peripheral populations of widespread species. In basing conservation decisions on jurisdictional rarity, conservation resources could be wasted on populations of common species that are neither rare, nor biologically unique (BC example). Alternately, conserving peripheral populations may conserve distinct genotypes not present elsewhere in a species' range (ACPF example). There is currently no framework in place to assess populations of widespread species to ensure that resources are strategically

216

invested on genetically distinct populations, or populations of conservation significance.

217

4.3 Prioritization using extinction risk assessment

The three species of interest, Dryas integrifolia, Anemone parviflora, and Vaccinium uliginosum, are common in Canada, and have widespread distributions (NatureServe

2016). White Mountain Avens (Dryas integrifolia Vahl., Family Rosaceae, Fig. 1.2) is a long-lived, clonal, mat-forming evergreen dwarf shrub. In Atlantic Canada, the southernmost extent of the species' distribution occurs at Wilson Brook, New

Brunswick, with a single population. The species is listed as globally and nationally secure, as S1 (five or fewer occurrences or very few remaining individuals) in New

Brunswick, and S4 (uncommon but not rare) in Quebec and Newfoundland and

Labrador (NatureServe 2016). Northern White Anemone (Anemone parviflora Michx.,

Family Ranunculaceae, Fig. 1.3) is a perennial clonal herb, and in the study region, it occurs along the Limestone Barrens in Newfoundland, and in the Tablelands. It is known from the Gaspé Peninsula, and from more than two locations in New Brunswick

(Restigouche region, Wilson Brook), as well as from Nova Scotia. The species is listed as globally and nationally secure, as S1 in New Brunswick, S4 in Quebec, S5 (common and abundant) in Newfoundland and S3/S4 (S3 vulnerable to S4 apparently secure, fewer than 80 populations) in Labrador (NatureServe 2016). Northern Bilberry

(Vaccinium uliginosum L., Family Ericaceae, Fig. 1.4) is a circumboreal dwarf shrub and occurs in the study region frequently in exposed areas in Newfoundland (along the

Limestone Barrens, the Tablelands, and the Long Range Mountains), in the uplands of the Gaspé Peninsula, as individual clones on Mt. Carleton and Mt. Denis in New

218

Brunswick, and in the Cape Breton Highlands. It is also known to occur further south in the Appalachian mountain range. Some historical locations are also known in New

Brunswick, PEI and Nova Scotia (ACCDC 2017). The species is listed as globally and nationally secure, as S1 in New Brunswick, and S5 in Quebec and Newfoundland and

Labrador (NatureServe 2016). None of the species have been assessed by COSEWIC

(COSEWIC 2015a).

In the Canadian species' prioritization framework, which follows the IUCN Redlist guidelines (IUCN 2017), COSEWIC uses criteria that include (1) distribution, (2) population size, and (3) trends in either distribution or population size (Fig. 4.2).

Common and widespread species are unlikely to be considered for any further assessment since COSEWIC usually assesses individual populations only of species potentially at risk as DUs. Based on these criteria, the three arctic-alpine species considered in this research (Dryas integrifolia, Anemone parviflora, and Vaccinium uliginosum) would not receive consideration. Their distribution is extensive and no declines in populations have been noted except in areas where their habitat has been affected, such as along the Limestone Barrens in Newfoundland, where the species occur at the continuous edge of their ranges. Based on these criteria, there does not appear to be any reason to invest in the effort required for a formal risk assessment.

Further, the other values invoked to justify conservation (Fig. 4.1) raise few concerns, except that these species occur in unique habitats in New Brunswick, which are only represented in very few locations in the province (upland barren, calcareous ledges,

219

gypsum cliff). Since those are rare habitats in the jurisdiction and region, they, and their characteristic flora, are in and of themselves of conservation value (jurisdictional, biological) within the regions. No economic uses are known for D. integrifolia and A. parviflora; V. uliginosum berries and stems are used by indigenous peoples in the Arctic.

While V. uliginosum may be economically and culturally valuable in the Arctic, since the species is highly uncommon at its rear range edge, there is no evidence for its value here.

220

4.4 Population-based assessment framework for peripheral populations of common species

Because extinction risk assessments at the species level are rarely applied to common, widespread species, a population-based approach is needed to assess populations that may be important for the conservation of diversity within a species. This approach focuses on a population’s genetic or ecological contribution to a species' biodiversity first, and extirpation risk second. For range-edge populations, two concepts often used in this context are based on ecological distinctness where populations occur in unique habitats not found elsewhere, and genetic distinctness resulting from isolation and locally adaptation (Lesica & Allendorf 1995, Leppig & White 2006). Genetic isolation may occur by means of distance, ecology and environment, genetic drift, and adaptation

(as defined by Shafer & Wolf (2013) and Wang & Bradburd (2014)). These factors may display complex interactions. For example, dispersal of both seeds and pollen can significantly affect gene flow, and thus degree of isolation (Duminil et al. 2007, Aguilar et al. 2008, Meirmans et al. 2011, Orsini et al. 2013, Abeli et al. 2014). Seeds and pollen often differ in vectors, and among species; dispersal vectors in turn are influenced by habitat fragmentation (Aguilar et al. 2008), or by landscape heterogeneity.

The concept of DUs as defined by COSEWIC (2015b) uses discreteness (genetic distinctness, disjunction, local adaptation) and significance (genetically different, persistence in unusual ecological setting, sole disjunct surviving population, or one whose loss would cause disjunction). It has been used in Canada to assess populations of

221

Lake Whitefish (Coregonus spp.) across their entire range by Mee et al. (2015) based on taxonomic units, genetic lineages, traits (i.e., local adaptation) and biogeographic region; by Taylor et al. (2013) in setting conservation priorities for Lake Chub (Couesius plumbeus) in North America based on mtDNA lineages and physiological variation; and by Taylor et al. (2011) to prioritize Rainbow Trout (Oncorhynchus mykiss) populations in British Columbia based on ecological and evolutionary legacy. The concept of DUs has also been used by COSEWIC to recommend protection under SARA (Species at

Risk Act) for populations of, e.g., Sockeye Salmon (Oncorhynchus nerka) in Sackinaw and Cultus Lakes (Government of Canada 2017).

Such assessments that include range-wide genetics of widespread species may be feasible for species whose economic value is considerable (Taylor et al. 2011, Taylor et al. 2013, Mee et al. 2015), or for species at risk where conservation significance has already been identified (Pouget et al. 2016). However, assessments of DUs for widespread species that have no direct human or economic value, and / or that are not species at risk, may be much more difficult to justify. They may thus have to take a more targeted approach to data collection, or in some cases rely on existing data, which may be limited. These species may be considered at the regional level for at-risk populations.

I propose the following process to guide strategic decision-making (Fig. 4.3). The first step is an evaluation of the population's biological or ecological value. Ideally, conservation priority would be based on genetic data. However, when (as is often the

222

case) genetic data are incomplete and only a portion of each species' range has been sampled, indirect factors affecting genetic distinctness should also be evaluated. Such indirect factors include a population's phylogeographic history (relict, refugial populations) and ecological distinctness (level of isolation). The second step is to assess the risk of extirpation. Together these two steps can be used to prioritize DUs for conservation management.

4.4.1 Evaluating population genetic distinctness

Populations are genetically distinct if they are significantly differentiated (i.e., high intra-population genetic divergence compared to other populations of the species). Since highly differentiated populations could be the result of genetic drift and / or inbreeding, differentiation alone would not confer high conservation priority.

4.4.1.1 Direct predictors of genetic distinctness

Genetic distinctness. As a first level of assessment, highest conservation priority should be given to populations that are known to be both genetically distinct and have high genetic diversity relative to other populations within the species' range. Their genetic distinctness infers value in terms of either evolutionary legacy such that a population represents a distinct lineage, or adaptive potential such that the populations' genetics contain the ability to respond to environmental change in the future (Moritz & Potter

2013).

Direct measures to assess genetic distinctness include differentiation (genetic distance as

223

an indicator of genetic divergence) among populations (Peakall & Smouse 2006, 2012).

This can be expressed as FST or Φ'PT (depending on the analysis and molecular marker used). The classification proposed by Hartl & Clark (1997) would give highest differentiation to populations with mean FST values > 0.15.

Collection of genetic data of suitable scope and rigor is both labor and resource intensive. Furthermore, genetic distinctness based on neutral markers does not necessarily infer adaptive potential, and genetic assessments using those markers need to be carefully approached.

4.4.1.2 Indirect predictors of genetic distinctness

In the absence of genetic data, genetic distinctness can be cautiously inferred from indirect evidence (Table 4.1). Indirect predictors can be used to assign intermediate conservation priority at a precautionary level, but also serve to guide further study, including increased sampling, targeted genetic analysis, and experimental manipulations such as reciprocal transplants. The following summary begins with the most practical and least costly forms of indirect evidence of distinctness, but also the most tentative.

Glacial history. Phylogenetic data, if available, may be able to be used to infer the genetic variability and distinctness. The literature contains significant evidence that refugial and, in some cases, relict populations contain unique presence of unique genotypes and lineages (Petit 2003, Lepais et al. 2013). Determination of where species may have survived during the Last Glacial Maximum (LGM), or been left behind in the

224

re-colonization process after deglaciation (Meirmans et al. 2011), may also help to identify the range limit and verify its type (i.e., leading vs. rear edge). Refugial populations would be classified as high priority given their high likelihood of containing unique genetic variation both due to local adaptation, and / or because they contain unique lineages. Relict populations could be classified as intermediate given their age and possible local adaptation. However, further experimental or genetic analyses would be required to test the generalization that refugial and relict populations possess unique genetic variability. Small refugial or relict populations with a long history of isolation may contain low levels of genetic diversity as a result of genetic drift and inbreeding; these would be less significant to be considered for conservation.

Ecological distinctness. Populations occurring in environments distinctly different from other locations may be adapted to local conditions, and thus contain unique genetic diversity not found elsewhere within the range (Mee et al. 2015). Confirmation of local adaptation in plant species would require further study such as reciprocal transplant experiments to to determine whether there is a home site advantage. High genetic differentiation would be assigned to populations that show high local adaptation compared to other populations. This could be determined for example through experimentally determined fitness-related traits, such as seed production and viability, that are associated with high adaptive potential.

Geographic isolation. Genetic distinctness may be predicted by levels of geographic

225

isolation (Orsini et al. 2013), which can further be inferred from heterogeneity of the landscape (Wang & Bradburd 2014). Disjunct populations are generally defined by some level of geographical isolation. Restricted inter-population gene flow combined with local selection pressures are expected to result in local adaption (Hampe & Jump

2011). However, like refugial or relict populations, geographically isolated small populations could also be prone to genetic drift and inbreeding depending on their population size and duration of their isolation.

4.4.2 Assessing risk and final prioritization

With either direct or indirect evidence that a population is of conservation significance, the second level of assessment (Table 4.1) is evaluating extirpation risk. The existing

COSEWIC threat assessment process, based on population size and trends, provides a strong structure for this assessment (COSEWIC 2015a). Highest priority should be given to populations (1) that have been assigned high conservation priority on the basis of their genetic and ecological distinctness, and (2) for which there is clear evidence of imminent risk of extirpation. Intermediate priority would be given to those populations are inferred to be at risk but without supporting evidence.

4.5 Case study: assessment of rear-edge, disjunct populations of three arctic-alpine species for conservation

My research, summarized below, applies my proposed process for assigning

226

conservation priority (Fig. 4.3) to three common and globally widespread arctic-alpine species, Anemone parviflora, Dryas integrifolia and Vaccinium uliginosum. Although no reliable thresholds for genetic distinctness exist for the three species, my data provide sufficient understanding to rank the populations relative to other populations within the species' ranges (Table 4.2).

4.5.1 Genetic distinctness (diversity and differentiation)

Regardless of their glacial history, range-edge populations with high genetic diversity and high genetic differentiation may contain the bulk of a species' genetic diversity, and

/ or unique local adaptations. Amplified Fragment Length Polymorphisms (AFLP) analysis (Table 2.5) showed that some, but not all, disjunct populations of A. parviflora and D. integrifolia contain high levels of genetic diversity. Populations of V. uliginosum showed variable levels of genetic diversity throughout the studied populations (Table

2.5), but no evidence that any rear-edge population was genetically distinct. All disjunct populations of A. parviflora and D. integrifolia contained unique markers; however, because some but not all were significantly differentiated from most other populations, the results are inconclusive. Further, because neutral genetic markers were used and data on levels of local adaptation (i.e., presence of locally-adapted genotypes) are lacking for these populations, genetic distinctness can only be inferred for the populations studied.

Relative genetic diversity (i.e., compared to all other populations) of disjunct populations of the three species is complex. For example, disjunct populations of D.

227

integrifolia showed a range of genetic diversities: high at Mount Logan (Table 2.5), medium at Wilson Brook, and low at the Tablelands. Genetic diversity of A. parviflora was high at both Wilson Brook and Tablelands populations, and low at the Restigouche population. The Mount Logan population of V. uliginosum showed high levels of genetic diversity, which was comparable to the edge and central populations.

Genetic differentiation of the disjunct populations was also complex, with no consistent patterns among the species (Table 4.2). All disjunct populations of D. integrifolia, showed high levels of mean population differentiation from all other populations, while two of the A. parviflora populations were highly differentiated from almost all other populations. However, the disjunct population of V. uliginosum was not significantly differentiated from any other population.

Overall, based on these measures, I suggest that the Wilson Brook and Tablelands populations of A. parviflora, and the Mount Logan and Wilson Brook populations of D. integrifolia are of high genetic significance compared to other populations sampled in this study (Table 4.2). However, this is evaluation must be interpreted with caution because all measures were based on neutral molecular markers, which evaluate only a small subset of the potential genetic diversity.

4.5.2 Indirect predictors of genetic distinctness

Glacial history

228

In North America, likely refugia have been identified for the LGM 23,000 to 18,000 years ago (Shaw et al. 2006, Alsos et al. 2007). While none of my study populations are refugial populations, genetic analyses using Amplified Fragment Length Polymorphisms

(AFLP, Chapter 2) suggest that the Wilson Brook and Mount Logan populations of D. integrifolia may be postglacial relicts; this is also supported by fossil evidence from presumed refugia south of the ice sheet during the LGM. The Wilson Brook population of A. parviflora may also be a relict, possibly having shared the same postglacial history as D. integrifolia; however, no fossil evidence of A. parviflora has been identified in areas in which D. integrifolia occurred during the LGM. The populations of V. uliginosum are unlikely to be relicts because there is evidence to suggest that the species re-colonized repeatedly after the LGM (Alsos et al. 2007, Eidesen et al. 2007). Further,

V. uliginosum appears to maintain considerable gene flow across its range (Alsos et al.

2007, Eidesen et al. 2007). On the basis of their glacial history (i.e., as probable relicts),

I suggest that the Wilson Brook populations of A. parviflora and D. integrifolia be given intermediate conservation priority, higher than any of the other populations studied

(Table 4.2).

Geographic isolation

Other predictors such as geographic isolation may be useful, as both distance and environment have been associated with increased divergence among populations (Orsini et al. 2013, Sexton et al. 2014). I tested distance, disjunction and human disturbance, all

229

potentially causes of genetic isolation, for their association with genetic variability using regression analysis (Chapter 2). I did not find that these measures of isolation could be used as indicators of genetic diversity and differentiation in rear-edge disjunct populations for A. parviflora, D. integrifolia and V. uliginosum. Only D. integrifolia showed an increase in genetic differentiation with distance to range edge, disjunction, and disturbance levels.

The consistently high levels of genetic diversity with low levels of differentiation among edge populations of V. uliginosum, compared to those of D. integrifolia and A. parviflora populations, suggest consistently lower isolation that could be related to pollen or seed vectors. Fruit of V.uliginosum are adapted for animal dispersion, such as birds, whereas fruits of the other two are wind-dispersed; birds have been known to carry seeds over long distances (Viana et al. 2013) despite physical barriers among ecosystems, while wind as a dispersal vector is most effective in open landscapes such as the Arctic.

Overall, the level of geographic isolation provides some evidence or potential genetic divergence, but since genetic data are available, this category does not change the assessment. The relationship between genetic patterns and these elements of isolation, especially dispersal ability, should be explored.

4.5.3 Risk of extirpation with climate warming

Detrimental impacts predicted for rear-edge populations of species under a warming

230

climate (Hampe & Petit 2005) are based on the assumption that environmental conditions at the rear edge are more severe than in other areas within a species' range, and may be at the species' limits of ecological tolerance (Gibson et al. 2009, Hampe &

Jump 2011). In Chapter 3, I assessed risk of population extirpation in terms of the microclimate tolerance for individuals from disjunct populations of A. parviflora and D. integrifolia at Wilson Brook. I compared Wilson Brook to sites in their continuous range, and used an OTC experiment to understand the potential growth response of these populations to changes in temperature. Populations of both species appear to be relatively stable at Wilson Brook, with no detectable changes over three years.

Microclimate at the study site did not differ significantly from the regional climate.

Because temperatures at ground level are expected to be higher than standard regional temperature measurements (Ashcroft & Gollan 2013), this means that the Wilson Brook site is in effect cooler than typical at ground level during the growing season which may partially explain the persistence of the arctic-alpine populations there. The absence of significant changes in population growth in response to variable growing season temperatures over three years, along with the wide range of temperatures that they currently experience without obvious detriment, indicate that these populations tolerate considerable temperature variability, including events > 40 °C. Together, the relatively cooler microclimate coupled with this tolerance may buffer them against immediate risk from a warming climate. Further assessments of trends in population changes over a longer time period would be needed to confirm the long-term risk. At this time,

231

extirpation risk for all populations is unknown but extirpation is not believed to be imminent (Table 4.2), thus none of the populations was given high conservation priority.

232

4.6 Conservation significance of disjunct, rear-edge, populations of three arctic- alpine species

While conservation significance of disjunct populations requires consideration of many factors, including some beyond the scope of this study (such as uniqueness of habitat in a regional context), my research and assessment, focused on genetic information and extinction risk, indicate that the disjunct populations of D. integrifolia, A. parviflora and

V. uliginosum may not have high priority for conservation action (Table 4.2). Based on their likely history, and / or relatively high levels of genetic diversity and differentiation, and with the acknowledged limitations of my data (i.e., relatively low sample sizes and extent of range sampled), the Wilson Brook and Mount Logan populations of D. integrifolia, and the Wilson Brook and Tablelands populations of A. parviflora should be given intermediate conservation priority. Neither direct nor indirect evidence of genetic distinctness is compelling. There is no conclusive evidence that the disjunct populations of the three species are genetically distinct based on neutral genetic markers.

Although the postglacial re-colonization histories of D. integrifolia and V. uliginosum are presumed to be similar, and Wilson Brook and Mount Logan populations may be relicts, this is not associated with genetic distinctness for V. uliginosum. In spite of their apparent isolation from all other populations, the Wilson Brook populations of A. parviflora and D. integrifolia are not significantly differentiated from them. Based on my results and the literature, it is clear that assumptions about using geographic isolation and phylogeographic history (i.e., relicts or refugial populations) to determine

233

conservation priority must be applied with care; they will require revision as new data become available.

In terms of extirpation risk assessment of two of the disjunct populations at Wilson

Brook, the temperature regimes experienced at this disjunct site are extremely variable and apparently not different from those at sites within the continuous range edge, suggesting that at least two of the species are tolerant of a broad range of microclimatic conditions. Within the limits of my results, neither of the disjunct rear-edge populations of D. integrifolia and A. parviflora at Wilson Brook appears to be at immediate risk from climate warming, however factors other than simple temperature tolerance must be considered. For example, sudden and heavy rainfall events predicted under climate change (Roy & Huard 2016) may accelerate the process of cliff erosion, causing a reduction in population size on this already strongly delimited slope.

234

4.7 Conclusion

In summary, while rear-edge populations of common species are generally overlooked by assessments that focus on species' extinction risk, these peripheral populations may provide essential genetic diversity that could influence the response of the species to climate warming. Conservation priority assessments of rear-edge populations of wide- ranging species should be based on, first, genetic distinctness and second, risk of extirpation. In the absence of direct evidence and using a precautionary approach, genetic distinctness may be inferred from indirect evidence, such as a species' phylogeographic or glaciation history, the level of geographic isolation of individual edge populations, as well as the population's occurrence in unique habitat compared to elsewhere within a species' range. Other predictors may be useful, but until such are identified and calibrated, evaluating phylogeography, and levels of isolation in relation to dispersal and population connectivity of rear range-edge populations will contribute to broadening understanding of their likelihood of being genetically or / and ecologically distinct. Future research should focus on establishing predictors of genetic distinctness, and should include an evaluation of my proposed assessment process using other peripheral populations, which will allow the process to be refined.

235

Tables

Level 1 Level 3 Level 2 Priority Conservation Direct evidence of Risk of Predictors of distinctiveness2 population3 priority5 distinctiveness1 extirpation4 Genetic Genetic Overall Refugial8 Relict9 Isolation10 diversity6 differentiation7 xxx xxx xxx xxx yes & no high xxx xx xx xx yes intermediate xx no low xxx x xx xx yes & no low xx xxx xx xx yes high xx no low xx xx xx xx yes intermediate no low xx x x x yes & no low x xxx x x yes low x x no low x xx x x yes & no low x x x x yes & no low

236

Level 3 Level 1 Level 2 Priority Conservation Risk of Direct evidence of distinctiveness1 Predictors of distinctiveness2 population3 priority5 extirpation4 Genetic Genetic Overall Refugial8 Relict9 Isolation10 diversity6 differentiation7 no low no yes xxx xx yes intermediate no low no no xxx - x x yes & no low

1 The first level of assessment is provided by direct evidence of genetic variability via molecular markers (genetic diversity & genetic differentiation). 2 If direct genetic information derived from molecular markers is not available, populations can be assessed as to their phylogeographic history (refugial or relict populations), and / or their level of isolation. Those inferences, without direct evidence, should receive only intermediate conservation priority status. 3 A populations' significance to the overall diversity of the species. 4 A populations' likelihood of extirpation, based on an assessment of risk. 5 Conservation priority based on the populations' overall significance, as well as its risk of extirpation. If the risk of extirpation is unknown, conservation significance will equal the populations' overall status. 6 Level of genetic diversity as determined by molecular markers. This measure is relative to all populations sampled. 7 Level of genetic differentiation, as measured by molecular markers, to determine genetic divergence from all other populations sampled. 8 Refugial populations are those that survived glaciations in isolated locations and may hold unique genetic diversity (levels of diversity and unique markers) 9 Relict populations have been isolation after deglaciation in the re-colonization process, and may retain unique genetic variation. 10 A high level of isolation could indicate a population that does not receive gene flow, and / or where local adaptation has occurred, resulting in unique genetic variation. 237

Table 4.2. Qualitative conservation priority (high, intermediate, low) of disjunct rear-edge populations of Dryas integrifolia (Di), Anemone parviflora (Ap), and Vaccinium uliginosum (Vu) based on proposed criteria, as described in text; xxx = high, xx = intermediate, x = low; Pop = population (WB, Wilson Brook; ML, Mount Logan; TL, Tablelands; RS, Restigouche), described in Chapter 1. Direct evidence of high genetic distinctness would place a population in a high conservation priority. See Table 4.1 for details.

Level 1 Level 2 Conservation Sp. Pop.1 Direct evidence of Predictors of distinctiveness priority3 distinctiveness Genetic Genetic Overall Refugial Relict Isolation diversity differentiation2 genetic Di WB xx xx xx no possible xxx interm ediate ML xxx xx xxx no possible xx intermediate TL x xxx x no no xx low Ap WB xxx xx xxx no possible xxx intermediate RS x xxx x no no xx low TL xx xx xx no no xx intermediate VU ML xxx x x no no x low

1 All populations are currently located within legally protected areas (Wilson Brook Protected Natural Area, Parc national de la Gaspésie (Mount Logan), and Gros Morne National Park (Tablelands). 2 Although my results suggest high priority due to high levels of genetic differentiation, since neutral genetic markers were used, genetic uniqueness is inferred, and thus the priority category was lowered to intermediate. 3 My research showed no immediate high risk to any populations, thus extirpation risk was not included in this table, and populations were not placed in a high conservation priority category as a result. 238

Figures

Figure 4.1. Categories of values potentially influencing the setting of conservation priorities for species at risk.

239

Figure 4.2. Modified species assessment process as used by COSEWIC, following IUCN methodology (COSEWIC 2015a). See COSEWIC (2015a) for detailed descriptions of each category.

240

241

List of references

Abeli T, Gentili R, Mondoni A, Orsenigo S, Rossi G. 2014. Effects of marginality on plant population performance. Journal of Biogeography 41(2):239–249.

Abeli T, Gentili R, Rossi G, Bedini G, Foggi B. 2009. Can the IUCN criteria be effectively applied to peripheral isolated plant populations? Biodiversity and Conservation 18(14):3877–3890.

ACCDC. 2017. Species Occurrence Records. Sackville: ACCDC. Available from: http://www.accdc.com/.

Alsos IG, P. B. Eidesen, D. Ehrich, I. Skrede, K. Westergaard, G. H. Jacobsen, J. Y. Landvik, P. Taberlet, Brochmann C. 2007. Frequent long-distance plant colonization in the changing Arctic. Science 316(5831):1606-1609.

Barnosky AD, Matzke N, Tomiya S, Wogan GO, Swartz B, Quental TB, Marshall C, McGuire JL, Lindsey EL, Maguire KC. 2011. Has the Earth's sixth mass extinction already arrived? Nature 471(7336):51-57.

Barrett SC, Kohn JR, Falk D, Holsinger K. 1991. Genetic and evolutionary consequences of small population size in plants: implications for conservation. In: Genetics and conservation of rare plants. DA Falk, Holsinger KE, editors. New York: Oxford University Press. 283pp.

Bellard C, Bertelmeier C, Leadley P, Thuiller W, Courchamp F. 2012. Impacts of climate change on the future of biodiversity. Ecology Letters 15(4):365–377.

Bottrill MC, Joseph LN, Carwardine J, Bode M, Cook C, Game ET, Grantham H, Kark S, Linke S, McDonald-Madden E. 2008. Is conservation triage just smart decision making? Trends in Ecology and Evolution 23(12):649-654.

Brooks TM, Mittermeier RA, da Fonseca GA, Gerlach J, Hoffmann M, Lamoreux JF, Mittermeier CG, Pilgrim JD, Rodrigues AS. 2006. Global biodiversity conservation priorities. Science 313(5783):58-61.

242

Bunnell FL, Campbell RW, Squires KA. 2004. Conservation priorities for peripheral species: the example of British Columbia. Canadian Journal of Forest Research 34(11):2240–2247.

Bunnell FL, Squires KA. Plagued by a plethora of peripherals: Refining guidelines for peripheral taxa. In TD Hooper, editor. Proceedings of the Species at Risk 2004 Pathways to Recovery Conference; 2004 March 2-6, 2004: Species at Risk 2004 Pathways to Recovery Conference Organizing Committee.

Clayden SR, Munro MC, Blaney CS, Kloet SPV. 2010. Vascular flora of the Atlantic Maritime Ecozone: some new perspectives. In: Assessment of species diversity in the Atlantic Maritime Ecozone. DF McAlpine, Smith IM, editors. Ottawa, Canada: NRC Research Press. pp. 197–213.

COSEWIC. 2015a. Wildlife species assessment. 2015 ed. Ottawa, ON: COSEWIC. Available from: http://www.cosewic.gc.ca/default.asp?lang=en&n=ED199D3B- 1.

COSEWIC. 2015b. Guidelines for recognizing designatable units. Ottawa, ON: COSEWIC. Available from: http://www.cosewic.gc.ca/default.asp?lang=en&n=DD31EAEE-1.

Crandall KA, Bininda-Emonds OR, Mace GM, Wayne RK. 2000. Considering evolutionary processes in conservation biology. Trends in Ecology and Evolution 15(7):290-295.

DNR: Wildlife and Biodiversity Nova Scotia: DNR. Available from: http://novascotia.ca/natr/wildlife/biodiversity/species-list.asp.

Dobrowski SZ. 2011. A climatic basis for microrefugia: the influence of terrain on climate. Global Change Biology 17(2):1022–1035.

Duminil J, Fineschi S, Hampe A, Jordano P, Salvini D, Vendramin GG, Petit RJ. 2007. Can population genetic structure be predicted from life-history traits? The American Naturalist 169(5):662–672.

243

Dyke AS. 2005. Late Quaternary vegetation history of northern North America based on pollen, macrofossil, and faunal remains. Géographie Physique et Quaternaire 59(2-3):211–262.

Eidesen PB, Alsos IG, Popp M, Stensrud O, Brochman C. 2007. Nuclear vs. plastid data: complex Pleistocene history of a circumpolar key species. Molecular Ecology 16(18):3902–3925.

Eizaguirre C, Baltazar–Soares M. 2014. Evolutionary conservation–evaluating the adaptive potential of species. Evolutionary Applications 7(9):963–967.

Environment Canada, Parks Canada Agency. 2010. Recovery Strategy and Management Plan for Multiple Species of Atlantic Coastal Plain Flora in Canada. Ottawa: Environment Canada and Parks Canada Agency. 96pp. + appendices.

Fraser DF. Species at the Edge: The Case for Listing of “Peripheral” Species. In LM Darling, editor. Proceedings of the Conference on the Biology and Management of Species and Habitats at Risk; 1999: B.C. Ministry of Environment, Lands and Parks, Victoria, B.C. and University College of the Cariboo, Kamloops, B.C.

Gibson SY, Van Der Marel RC, Starzomski BM. 2009. Climate change and conservation of leading-edge peripheral populations. Conservation Biology 23(6):1369–1373.

Government of Canada. 2002. Bill C-5, An act respecting the protection of wildlife species at risk in Canada. Ottawa: Government of Canada. Available from: http://laws.justice.gc.ca/PDF/Statute/S/S-15.3.pdf.

Government of Canada. 2005. COSEWIC Assessment and Update Status Report on the Narwhal (Monodon Monoceros) in Canada. Ottawa: Government of Canada. Available from: http://www.sararegistry.gc.ca/default.asp?lang=En&n=51A4BCD1- 1&offset=2&toc=show.

Government of Canada. 2017. Species at Risk Public Registry. Ottawa: Government of Canada. Available from: http://www.registrelep- sararegistry.gc.ca/species/speciesDetails_e.cfm?sid=730.

Hampe A, Jump AS. 2011. Climate Relicts: Past, Present, Future. Annual Review of Ecology, Evolution and Systematics 42(2011):313–333.

244

Hampe A, Petit RJ. 2005. Conserving biodiversity under climate change: the rear edge matters. Ecology Letters 8(5):461–467.

Hampe A, Rodriguez-Sanchez F, Dobrowski S, Hu SS, Gavin DG. 2013. Climate refugia: from the Last Glacial Maximum to the twenty-first century. New Phytologist 197(1):16–18.

Hardie DC, Hutchings JA. 2010. Evolutionary ecology at the extremes of species’ ranges. Environmental Reviews 18(2010):1–20.

Hartl DL, Clark AG. 1997. Principles of population genetics. Third edition ed. Sunderland, Massachusetts: Sinauer Associates, Inc. 542pp.

Hunter ML, Hutchinson A. 1994. The virtues and shortcomings of parochialism: Conserving species that are locally rare, but globally common. Conservation Biology 8(4):1163–1165.

IUCN. 2017. International Union for the Conservation of Nature. IUCN. Available from: http://iucn.org/iyb/about/.

Leppig G, White JW. 2006. Conservation of peripheral plant populations in California. Mardono 53(3):264–274.

Lesica P, Allendorf FW. 1995. When are peripheral populations valuable for conservation? Conservation Biology 9(4):753–760.

Mace GM, Collar NJ, Gaston KJ, Hilton‐Taylor C, Akçakaya HR, Leader-Williams N, Milner-Gulland EJ, Stuart SN. 2008. Quantification of extinction risk: IUCN's system for classifying threatened species. Conservation Biology 22(6):1424- 1442.

Martín-López B, González JA, Montes C. 2011. The pitfall-trap of species conservation priority setting. Biodiversity and Conservation 20(3):663-682.

245

Mee JA, Bernatchez L, Reist JD, Rogers SM, Taylor EB. 2015. Identifying designatable units for intraspecific conservation prioritization: a hierarchical approach applied to the lake whitefish species complex (Coregonus spp.). Evolutionary Applications 8(5):423-441.

Meirmans PG, Goudet J, Gaggiotti OE. 2011. Ecology and life history affect different aspects of the population structure of 27 high-alpine plants. Molecular Ecology 20(15):3144–3155.

Miller RM, Rodríguez JP, Aniskowicz-Fowler T, Bambaradeniya C, Boles R, Eaton MA, Gärdenfors U, Keller V, Molur S, Walker S. 2006. Extinction risk and conservation priorities. Science 313(5786):441-441.

Moritz CC, Potter S. 2013. The importance of an evolutionary perspective in conservation policy planning. Molecular Ecology 22(24):5969-5971.

Peakall R, Smouse P. 2006. GenAlEx 6: Genetic Analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes 6(1):288–295.

Peakall R, Smouse P. 2012. GenAlEx 6.5: Genetic analysis in Excel. Population genetic software for teaching and research – an update. Bioinformatics 10.1093/bioinformatics/bts460.10.1093/bioinformatics/bts460.

Petit RJ. 2003. Glacial Refugia: Hotspots but not melting pots of genetic diversity. Science 300(5625):1563-1565.

Pouget M, Baumel A, Diadema K, Médail F. 2016. Conservation unit allows assessing vulnerability and setting conservation priorities for a Mediterranean endemic plant within the context of extreme urbanization. Biodiversity and Conservation 26(2):1-15.

Primack RB. 2002. Essentials of conservation biology. 3rd ed.: Sinauer Associates Sunderland, Massachusetts. 698pp.

Provan J, Maggs CA. 2012. Unique genetic variation at a species’ rear edge is under threat from global climate change. Proceedings of the Royal Society B 279(1726):39–47.

Pullin AS, Sutherland W, Gardner T, Kapos V, Fa JE. 2013. Conservation priorities: identifying need, taking action and evaluating success. Key Topics in Conservation Biology 2(2013):3-22.

Richardson L, Loomis J. 2009. The total economic value of threatened, endangered and rare species: an updated meta-analysis. Ecological Economics 68(5):1535-1548.

Rodríguez JP. 2008. National Red Lists: the largest global market for IUCN Red List categories and criteria. Endangered Species Research 6(2):193-198.

Roy P, Huard D. 2016. Future Climate Scenarios–Province of New-Brunswick. Montreal: Ouranos. 80 pp.

Sexton JP, Hangartner SB, Hoffmann AA. 2014. Genetic isolation by environment or distance: which pattern of gene flow is most common? Evolution 68(1):1–15

Sgro CM, Lowe AJ, Hoffmann AA. 2011. Building evolutionary resilience for conserving biodiversity under climate change. Evolutionary Applications 4(2):326–337.

Shafer ABA, Wolf JBW. 2013. Widespread evidence for incipient ecological speciation: a meta-analysis of isolation-by-ecology. Ecology Letters 16(7):940–950.

Shaw J, Piper DJW, Fader GBJ, King EL, Todd BJ, Bell T, Batterson MJ, Liverman DGE. 2006. A conceptual model of the deglaciation of Atlantic Canada. Quaternary Science Review 25(17-18):2059–2081.

Suarez-Gonzalez A, Sutton JT, Trant AJ, Zamlynny E, Good SV. 2015. Rethinking refugia: tree topology, divergence dates, and demographic history trace the 247

distribution of the endangered Plymouth gentian (Sabatia kennedyana) from the Pleistocene glaciation to present day. American Journal of Botany 102(4):609– 620.

Taylor EB, Darveau C-A, Schulte PM. 2013. Setting conservation priorities in a widespread species: phylogeographic and physiological variation in the lake chub, Couesius plumbeus (Pisces: Cyprinidae). Diversity 5(2):149-165.

Taylor EB, Tamkee P, Keeley ER, Parkinson EA. 2011. Conservation prioritization in widespread species: the use of genetic and morphological data to assess population distinctiveness in rainbow trout (Oncorhynchus mykiss) from British Columbia, Canada. Evolutionary Applications 4(1):100-115.

Thomas CD, Cameron A, Green RE, Bakkenes M, Beaumont L, Vollingham YC, Erasmus BFN, Suqueira MFd, Grainger A, Hannah L, et al. 2004. Extinction risk from climate change. Nature 427(8):145–148.

United Nations. 1992. Convention on biological diversity. United Nations. Available from: http://www.cbd.int.

Urban MC. 2015. Accelerating extinction risk from climate change. Science 348(6234):571-573.

Viana DS, Santamaria L, Michot TC, Figuerola J. 2013. Allometric scaling of long- distance seed dispersal by migratory birds. The American Naturalist 181(5):649– 662.

Vucetich JA, Waite TA. 2003. Spatial patterns of demography and genetic processes across the species' range: Null hypothesis for landscape conservation genetics. Conservation Genetics 4(5):639–645.

Wang IJ, Bradburd GS. 2014. Isolation by environment. Molecular Ecology 23(23):5649–5662.

Williams SE, Shoo LP, Isaac JL, Hoffmann AA, Langham G. 2008. Towards an integrated framework for assessing the vulnerability of species to climate change. PLoS Biology 6(12):2621–2626.

248

Chapter 5 - Living at the edge - summary

This thesis is built on the premise that peripheral populations, specifically those at the rear range edge, may contain adaptive potential in the form of genetic variability that could allow species to respond more successfully to climate change (Lesica & Allendorf

1995, Channell 2004, Hampe & Jump 2011). Although most research linking peripheral populations to species' vulnerabilities to climate change has focused on leading range edges, there is evidence of the evolutionary significance of disjunct and refugial populations (Lesica & Allendorf 1995, Nantel & Gagnon 1999, Leppig & White 2006).

Assessing conservation significance is complex, partly because criteria have not been clearly established to assess peripheral populations. Furthermore, such assessments are rarely carried out for common, wide-ranging species, mainly because common species are not typically at risk of extinction – the primary criterion of current systems for setting conservation priorities (Bottrill et al. 2008, Martín-López et al. 2011). My work contributes to a better understanding of how such an assessment could be carried out for peripheral plant populations.

This research was motivated by concern over the risk to rear-edge, disjunct New

Brunswick populations under climate change. I wanted to determine whether rear-edge populations of three wide-ranging, arctic-alpine species (Anemone parviflora, Dryas integrifolia, Vaccinium uliginosum) were “worth conserving” based on their genetic distinctness. I evaluated their conservation significance in terms of genetic distinctness, and potential predictors of genetic distinctness as to their usefulness. Working

249

specifically with two disjunct populations of A. parviflora and D. integrifolia at Wilson

Brook, NB, I evaluated their extirpation risk. I considered the challenges in assessing peripheral populations' conservation significance by first applying a standard approach

(extinction risk assessment process (COSEWIC 2015, IUCN 2017). I then applied a proposed population-based approach based on criteria that would be meaningful for peripheral populations of wide-ranging, common species: genetic distinctness, ecological distinctness, and extinction risk.

5.1.1 Genetic distinctness

Genetic distinctness has been proposed as an important criterion for identifying the conservation significance of range-edge populations (Leppig & White 2006). High levels of genetic diversity presumably confer high evolutionary and adaptive potential because a larger gene pool increases the likelihood that adaptive variation is present

(Kirk & Freeland 2011). High levels of among-population genetic variability suggest that local adaptation, or unique genotypes, may be present. Range-edge populations that have both high within-population genetic diversity and show high levels of population genetic differentiation presumably have high conservation significance.

In Chapter 2, I documented genetic diversity in populations of D. integrifolia, Anemone parviflora and Vaccinium uliginosum, using Amplified Fragment Length

Polymorphisms (AFLP) as molecular marker, across the range edge of the species in

Atlantic Canada, including central populations (except for A. parviflora), and disjunct populations. All species displayed considerable genetic diversity in most populations.

250

Some disjunct populations of D. integrifolia and A. parviflora showed significant differentiation from all other populations, while there was negligible among-population genetic diversity among all populations of V. uliginosum.

Assessing genetic distinctness in the context of conservation significance of rear-edge disjunct populations may be difficult even when some genetic information is available, because neutral genetic variation (e.g., AFLP) does not necessarily reflect adaptive genetic variation (Reed & Frankham 2001). It appears that disjunct populations of these three species are not genetically distinct based on the AFLP analysis. However, the significance of their genetic variation requires determining whether locally adapted genotypes are present within these populations.

5.1.2 Association of ecological and demographic traits with genetic variability

In the absence of reliable genetic data, and to simplify assessments of rear-edge, disjunct populations, it would be useful to identify predictors of genetic distinctness. Such predictors, if they exist, would be expected to differ with life form, history, or overall distribution patterns. In Chapter 2, I tested for an association of genetic variability with various ecological and demographic factors for A. parviflora, D. integrifolia and V. uliginosum, at the rear range edge. These three species share a similar distribution range, rear-edge distribution, phylogeographic history (D. integrifolia and V. uliginosum), and some life history traits. Patterns of genetic diversity were not associated with factors such as geographic distance, disjunction, and disturbance for all three species, although disjunct populations of A. parviflora and D. integrifolia showed higher levels of mean

251

among-population genetic diversity than populations from the range edge and center.

Undisturbed populations of these two species also showed higher pairwise mean genetic differentiation than both recovering and disturbed sites. Despite some similarities among the species, I did not find a general characteristic that could easily be used as an indicator for genetic variability among and within rear-edge populations. While isolation factors presumably affect gene flow (Sexton et al. 2014) and thus genetic distinctness, the mechanisms may be more complex than simple geographical distance. For example, geographic or environmental isolation can vary with pollen or seed dispersal vectors.

Further research into any putative predictors associated with isolation should be a priority.

5.1.3 Extinction risk from climate warming?

Populations of species at range edges, particularly the rear edges, may be at risk of extirpation under a warming climate (Lesica & Allendorf 1995, Hampe & Petit 2005). In

Chapter 3, I assessed the microclimate of Wilson Brook, where A. parviflora and D. integrifolia occur in disjunct populations, and undertook a warming experiment to understand the potential response of these populations to warming associated with climate change. I measured microclimatic conditions over three years, and compared them to a site within 15 km, and another 750 km further to north (believed to be more similar to the center of distribution). These local microclimates were also compared to coarser regional climate data for context. Microclimate did not differ significantly from the regional climate suggesting that the site has a local cooling effect, given that microclimate is expected to differ significantly from standardized observations due to 252

the influence of slope, aspect, soil, and other factors that influence temperature at ground level (Geiger et al. 2009).

While the warming experiment did not raise temperatures as expected, no significant changes in population growth were detected over three years, inside or outside the OTCs in response to naturally or experimentally varied temperatures. Overall, the Wilson

Brook populations of both species appear to be relatively stable. Other factors that influence the local habitat, such as the instability and type of substrate (gypsum) or biotic interactions, may play important roles in the persistence of these populations. For conservation purposes, these populations appear to be stable in the short term and a warming climate may not pose an immediate risk. Monitoring may aid in evaluating population changes, and factors such as erosion at the site, over time.

5.1.4 Assessing the conservation significance of peripheral populations

Given limited resources, prioritization is necessary, and is mostly undertaken as a triage, identifying species most urgently in need of attention (Bottrill et al. 2008, Pullin et al.

2013). Range-edge populations of wide-ranging species are not commonly assessed as to their conservation significance. While jurisdictions in which a species may be regionally rare sometimes include peripheral populations in their priority lists (Bunnell et al. 2004,

Channell 2004), jurisdictional rarity in itself should not be a reason to prioritize species for conservation (Hardie & Hutchings 2010). In Chapter 4, I make the case for considering peripheral populations of common species in conservation decision-making, because they may provide important genetic diversity to increase a species' capacity to

253

respond to climate change (Hampe & Jump 2011, Bellard et al. 2012, Eizaguirre &

Baltazar–Soares 2014). I show how the commonly used prioritization process based on a species' extinction risk, used by the IUCN and COSEWIC in Canada (Mace et al. 2008,

COSEWIC 2015, IUCN 2017), has limited usefulness for assessing range-edge, disjunct populations of widespread species, and I propose a population-based assessment process using genetic distinctiveness, ecological distinctness and extinction risk as criteria.

While the proposed type of assessment may be appropriate, in the absence of sufficient data, using a precautionary approach in managing rear-edge disjunct populations would be a first step.

254

5.2 Implications of my findings

My research contributes new information on patterns of genetic diversity in rear-edge, disjunct populations of arctic-alpine species in Eastern Canada, and provides some new insights into their complex genetic patterns. Through the identification of putative lineages, I speculate on the mixing of lineages from the north and south in

Newfoundland, adding new, fine-scale information for D. integrifolia, V. uliginosum and

A. parviflora. Supporting previous findings, I have not found reduced genetic diversity in most of the rear-edge disjunct populations, and have shown that even for species that are presumed to have followed the same re-colonization patterns after deglaciation (e.g.,

D. integrifolia and V. uliginosum), genetic diversity patterns can differ considerably, possibly associated with life history traits. Specifically, my genetic analysis has revealed some new information about disjunct rear-edge populations of D. integrifolia and A. parviflora. The relative high genetic diversity of D. integrifolia at Mount Logan might make this population important to monitor. Low diversity of an A. parviflora population at Restigouche suggests a founder event in that location, as did the low genetic diversity of D. integrifolia at Tablelands. The populations of D. integrifolia and A. parviflora at

Wilson Brook are not genetically impoverished as expected due to their isolation. This information is new, and can be used to focus future research, and support management decisions for those populations.

I confirmed that the microclimate at Wilson Brook is cooler than the regional climate, partially explaining the persistence of D. integrifolia and A. parviflora (and other arctic species) at this site 300 km away from the nearest population of either species, and over 255

400 km south of their continuous range edge. Although based on only three years' data, my results indicate these two species are demographically stable at the site. While there does not appear to be an immediate extirpation risk for those populations, this research provides a baseline for site managers to continue monitoring Wilson Brook as the climate warms to understand if the population size is indeed stable.

In Chapter 4, I argued that disjunct populations of common widespread species are worth assessing, because they may have genetic variability that potentially aids the species to respond to climate change. I demonstrated that the commonly used prioritization based on a species' extinction risk assessment is not useful in assessing disjunct populations of common, widespread species, and I proposed a new population- based approach, using genetic distinctness and ecological distinctness, as well as extirpation risk, as criteria, which will require further testing and refining. The template developed will be helpful in compiling direct and indirect evidence of genetic distinctness, and thus aid in making decisions about conservation and research needs for specific disjunct populations.

256

5.3 Future directions

In order to refine the proposed process for assessing rear-edge disjunct populations of widespread species, and because critical processes are likely to vary among species and populations, studies must consider a larger number of species and more populations in assessments of patterns, and the relationship of factors such as isolation (genetic, geographic, ecological) with genetic variability patterns at the rear edge. Research focused on predictors of genetic distinctiveness of peripheral (particularly disjunct) populations may be costly in the short-term, but allow for quicker assessments in the long-term. This may help to identify patterns that can refine my proposed approach. In addition, assessing the adaptive capacity of disjunct, rear-edge populations is essential.

To assess adaptive genetic diversity, which is an active field of research (Pauls et al.

2013), two approaches can be used. Genetic markers that provide information about adaptive genetic variation (or where loci under selection, or quantitative trait loci (QTL) can be easily identified), and thus levels of local adaptation (e.g., SSRs, microsatellites and next-generation sequencing of single nucleotide polymorphisms) could be used.

Alternately, common garden or reciprocal transplant experiments could be used to identify genotypic differences and levels of local adaptation by testing for fitness-related trait differences among populations, a methodology that has been used successfully for many years. My proposed assessment process could help identify and prioritize populations for transplant experiments that may contain local adaptation. As more research is carried out on disjunct rear-edge populations of common species, predictors of genetic and ecological distinctness should be assessed. Research should focus on

257

exploring the contribution of isolation, resulting, for example, from dispersal limitations, as possible predictor variables.

Specifically, the D. integrifolia and A. parviflora (and possibly other arctic-alpine species) populations at Wilson Brook should be evaluated as to the presence of local adaptations. The site is ecologically unique, and if locally adapted genotypes are present, it would support the case for continued monitoring of the site, to better evaluate immediate risks posed by the eroding cliff face. The population of D. integrifolia at

Mount Logan should also receive further attention: is the population's level of genetic diversity indicative of a meta-population in these mountains? Is the population at risk as warming may cause more temperate species to move upwards in altitude and potentially out-compete the arctic species found at the location? If the Mount Logan population is part of a meta-population, extirpation of one of these populations may not be as critical.

On the other hand, if the population is at risk from encroaching vegetation, then a careful consideration of all populations could identify those most likely to persist in a warming climate. The arctic-alpine species that are part of the Limestone Barrens ecosystem should also receive attention: is there a similar pattern in putative lineages in these species as I suggested for D. integrifolia? If the re-colonization scenario I proposed for that region is correct, we would expect that other species recolonized the area from the refugia to the south and north, and would therefore show similar lineage patterns.

Peripheral populations have long been of interest to conservation biology, especially as bellwethers of climate change, which is already impacting species' range edges. While

258

conservation resources may remain limited and focus will likely remain on species' risk of extinction, in the context of climate change, rear-edge disjunct populations of widespread, common species, should not be ignored and require more attention, given their potential role in allowing species to respond to the coming changes. Range-edge populations will likely receive more attention as the climate warms, when we start seeing the first losses of such populations. In the meantime, identifying potentially important range-edge populations is feasible for many species as data on regional occurrences exist through the Conservation Data Centres across Canada (NatureServe

2016). While insufficient data will always be an issue, those edge populations that are classified as rare (S1 or S2) could be identified to conservation authorities, as well as university researchers. A protocol could help identify the critical information needed

(DNA samples, extirpation risk, population trends) to start building a bank of samples and information that at least could be available if funding and interest increase.

259

List of references

Bellard C, Bertelmeier C, Leadley P, Thuiller W, Courchamp F. 2012. Impacts of climate change on the future of biodiversity. Ecology Letters 15(4):365–377.

Bunnell FL, Campbell RW, Squires KA. 2004. Conservation priorities for peripheral species: the example of British Columbia. Canadian Journal of Forest Research 34(11):2240–2247.

COSEWIC. 2015. Wildlife species assessment. 2015 ed. Ottawa, ON: COSEWIC. Available from: http://www.cosewic.gc.ca/default.asp?lang=en&n=ED199D3B- 1.

Eizaguirre C, Baltazar–Soares M. 2014. Evolutionary conservation–evaluating the adaptive potential of species. Evolutionary Applications 7(9):963–967.

Geiger R, Aron RH, Todhunter P. 2009. The climate near the ground Cambridge: Rowman & Littlefield. 611 pp.

Hampe A, Jump AS. 2011. Climate Relicts: Past, Present, Future. Annual Review of Ecology, Evolution and Systematics 42(2011):313–333.

Hampe A, Petit RJ. 2005. Conserving biodiversity under climate change: the rear edge matters. Ecology Letters 8(5):461–467.

Hardie DC, Hutchings JA. 2010. Evolutionary ecology at the extremes of species’ ranges. Environmental Reviews 18(2010):1–20.

IUCN. 2017. International Union for the Conservation of Nature. IUCN. Available from: http://iucn.org/iyb/about/.

Kirk H, Freeland JR. 2011. Applications and implications of neutral versus non-neutral markers in molecular ecology. International Journal of Molecular Sciences 12(6):3966-3988.

260

Leppig G, White JW. 2006. Conservation of peripheral plant populations in California. Mardono 53(3):264–274.

Lesica P, Allendorf FW. 1995. When are peripheral populations valuable for conservation? Conservation Biology 9(4):753–760.

Martín-López B, González JA, Montes C. 2011. The pitfall-trap of species conservation priority setting. Biodiversity and Conservation 20(3):663-682.

Pullin AS, Sutherland W, Gardner T, Kapos V, Fa JE. 2013. Conservation priorities: identifying need, taking action and evaluating success. Key Topics in Conservation Biology 2(2013):3-22.

Reed DH, Frankham R. 2001. How closely correlated are molecular and quantitative measures of genetic variation? A meta-analysis. Evolution 55(6):1095-1103.

Sexton JP, Hangartner SB, Hoffmann AA. 2014. Genetic isolation by environment or distance: which pattern of gene flow is most common? Evolution 68(1):1–15.

261

Appendices

location Latitude Sampling Sampling Longitude Sample ID Sample PrimA509 PrimA501 PrimA439 PrimA427 PrimA403 PrimA390 PrimA382 PrimA374 PrimA359 DIL2.02 Torngat 58.45683 -62.80534 0 1 0 0 0 0 1 0 1

DIL2.04 Torngat 58.45683 -62.80534 1 1 0 0 1 0 1 0 0

DIL2.06 Torngat 58.45683 -62.80534 1 1 0 0 1 0 1 0 0

DIL2.13 Torngat 58.45683 -62.80534 1 1 1 0 0 0 1 0 0

DIL2.14 Torngat 58.45683 -62.80534 1 1 0 0 0 0 1 0 0

DIL2.15 Torngat 58.45683 -62.80534 1 1 1 0 1 0 1 0 0

DIL2.16 Torngat 58.45683 -62.80534 0 1 0 0 0 0 1 0 0

DIL2.20 Torngat 58.45683 -62.80534 1 1 1 0 0 0 1 0 0

DIL2.24 Torngat 58.45683 -62.80534 1 1 0 0 0 0 1 0 0

Burnt DI201 Cape 51.57838 -55.74734 0 1 0 0 0 0 1 0 0

262

Index; Pb(2/4/4) = allelic richness calculated using the rarefaction to two/four/four individuals; AE = effective number of alleles; D(2/4/4) Nei's gene diversity (AFLPDat). See Table 2.5 for details on metric calculations and programs used.

Range # PLP Pb(2/ Population n PB %P He (±SE) I (±SE) A (±SE) D(2/4/ group bands 1% 4/4) E 4) (a) Anemone parviflora Edge Burnt Cape 4 121 1 37.35 0.135 (0.015) 41.55 0.202 (0.021) 1.209 1.231 (0.027) 0.187 Cooks Harbour 6 109 0 39.16 0.134 (0.015) 42.80 0.202 (0.021) 1.171 1.228 (0.027) 0.259 Sandy Cove 2 105 0 33.73 0.140 (0.015) 31.05 0.204 (0.022) 1.311 1.239 (0.026) 0.337 Port-au-Choix 5 130 2 66.87 0.239 (0.015) 71.80 0.357 (0.022) 1.344 1.408 (0.029) 0.392 Bellburns 2 94 0 33.13 0.137 (0.015) 39.20 0.200 (0.022) 1.288 1.234 (0.026) 0.331 Disjunct Tablelands 4 123 1 46.39 0.181 (0.016) 47.50 0.265 (0.023) 1.235 1.321 (0.031) 0.139 Restigouche 4 95 0 10.24 0.037 (0.011) 10.80 0.056 (0.013) 1.080 1.060 (0.015) 0.096 Wilson Brook 8 126 2 57.83 0.205 (0.016) 56.10 0.306 (0.023) 1.221 1.356 (0.030) 0.271 Mean 4 113 40.15 0.151 42.60 0.224 1.232 1.260 0.252 SE 0.7 5 6.06 0.005 16.64 0.008 0.03 0.010 0.097

263

(b) Dryas integrifolia Centre Torngat 9 193 4 69.69 0.205 (0.012) 69.70 0.317 (±0.016) 1.463 1.339 (0.022) 0.180 Edge Burnt Cape 6 161 0 55.51 0.173 (0.011) 55.50 0.267 (±0.017) 1.471 1.279 (0.02) 0.257 Cooks Harbour 6 130 0 38.98 0.124 (0.011) 38.20 0.190 (±0.016) 1.36 1.205 (0.02) 0.210 Sandy Cove 4 120 0 33.07 0.116 (0.011) 33.10 0.175 (±0.016) 1.327 1.197 (0.021) 0.177 Port-au-Choix 10 144 0 45.28 0.125 (0.011) 45.30 0.196 (±0.015) 1.314 1.198 (0.018) 0.166 Bellburns 10 121 2 31.89 0.09 (0.01) 24.80 0.141 (±0.015) 1.202 1.149 (0.018) 0.096 Disjunct Tablelands 9 122 1 30.71 0.088 (0.01) 25.60 0.137 (±0.014) 1.194 1.143 (0.018) 0.088 Mount Logan 8 179 1 64.57 0.196 (0.011) 63.00 0.304 (±0.016) 1.499 1.312 (0.02) 0.315 Wilson Brook 11 144 2 44.88 0.13 (0.011) 41.70 0.203 (±0.016) 1.313 1.208 (0.019) 0.186 Mean 8 146 46.06 0.139 44.10 0.214 1.345 1.226 0.186

SD 0.8 9 4.78 0.014 5.30 0.022 0.04 0.023 0.071

(c) Vaccinium uliginosum Centre Torngat 11 184 1 70.35 0.181 (0.011) 58.20 0.288 (0.016) 1.441 1.243 (0.022) 0.429 Nain 10 149 0 58.41 0.169 (0.012) 51.00 0.261 (0.018) 1.369 1.25 (0.025) 0.376 Edge Burnt Cape 7 144 0 49.12 0.147 (0.012) 42.30 0.227 (0.017) 1.369 1.207 (0.023) 0.363 Cooks Harbour 4 126 0 38.94 0.136 (0.012) 32.50 0.205 (0.018) 1.389 1.192 (0.023) 0.389 Sandy Cove 7 140 0 51.33 0.16 (0.013) 42.30 0.245 (0.018) 1.381 1.238 (0.025) 0.327 Port-au-Choix 10 171 2 65.93 0.18 (0.012) 53.60 0.282 (0.017) 1.414 1.256 (0.025) 0.407 Bellburns 10 156 1 56.19 0.155 (0.012) 47.40 0.243 (0.017) 1.354 1.213 (0.023) 0.358 Tablelands 11 173 0 64.16 0.176 (0.012) 53.10 0.276 (0.017) 1.41 1.239 (0.023) 0.425 Disjunct Mount Logan 10 166 1 65.49 0.187 (0.012) 56.20 0.29 (0.017) 1.429 1.278 (0.025) 0.398 Mean 9 157 57.77 0.166 48.51 0.257 1.395 1.235 0.386 SE 0.8 6 3.33 0.006 2.74 0.010 0.010 0.009 0.033 Mean correlation coefficient 0.82 0.84 0.82 0.86 0.65 0.73 0.57 264

Appendix 2.3. Independent sample t-test comparing the %P per population with the mean %P for all populations per species.

Anemone parviflora Dryas integrifolia Vaccinium uliginosum Population %P t-value p-value %P t-value p-value %P t-value p-value Torngat - - - 0.70 -4.923 0.001 0.70 -3.766 0.005 Nain ------0.58 -0.135 0.896 Burnt Cape 0.37 0.574 0.584 0.56 -2.026 0.077 0.49 2.589 0.032 Cooks Harbour 0.39 0.246 0.813 0.39 1.496 0.173 0.39 5.615 < 0.001 Sandy Cove 0.34 1.066 0.322 0.33 2.739 0.026 0.51 1.984 0.083 Port-au-Choix 0.67 -4.346 0.003 0.45 0.253 0.807 0.66 -2.555 0.034 Bellburns 0.33 1.230 0.258 0.32 2.946 0.019 0.56 0.471 0.65 Tablelands 0.46 -0.902 0.397 0.31 3.154 0.013 0.64 -1.950 0.087 Mount Logan - - - 0.65 -3.890 0.005 0.65 -2.253 0.054 Restigouche 0.1 5.002 0.002 ------Wilson Brook 0.58 -2.870 0.024 0.45 0.253 0.807 - - - df 7 8 8 Mean 0.41 0.46 0.58 SD 0.17 0.14 0.10

265

Appendix 2.4. Correlation coefficients of mean population genetic metric above diagonal, P-values below diagonal; a) including populations from all three species Anemone parviflora, Dryas integrifolia and Vaccinium uliginosum (# of populations = 26); b) excluding A. parviflora populations (# of populations = 18); n = number of individuals (all per population), bands = number of AFLP bands present across all primer combinations, PB = number of private / unique bands, %P = percent polymorphic loci, He = expected heterozygosity, PLP1% = percent polymorphic loci at 1%;, I = Shannon information index, Pb = Allelic richness (rarefied), AE = Effective number of alleles, D = Nei's gene diversity (rarefied), Φ'PT = mean pairwise population genetic differentiation. See text and Table 2.4 for details on metrics.

a) n bands PB %P He PLP1% I Pb AE D Φ'PT n 0.71 0.38 0.56 0.20 0.35 0.28 0.34 0.02 0.15 -0.10 bands 0.000 0.40 0.86 0.60 0.72 0.66 0.82 0.40 0.42 -0.29 PB 0.057 0.042 0.43 0.41 0.46 0.42 0.11 0.45 -0.16 0.48 %P 0.003 0.000 0.028 0.90 0.94 0.94 0.77 0.75 0.64 -0.26 He 0.321 0.001 0.040 0.000 0.95 1.00 0.65 0.95 0.61 -0.12 PLP1% 0.080 0.000 0.017 0.000 0.000 0.96 0.68 0.89 0.51 -0.06 I 0.169 0.000 0.034 0.000 0.000 0.000 0.69 0.93 0.62 -0.15 Pb 0.094 0.000 0.599 0.000 0.000 0.000 0.000 0.47 0.63 -0.47 AE 0.932 0.044 0.020 0.000 0.000 0.000 0.000 0.016 0.40 0.11 D 0.464 0.035 0.446 0.001 0.001 0.007 0.001 0.001 0.043 -0.66 Φ'PT 0.626 0.157 0.014 0.197 0.557 0.774 0.464 0.015 0.587 0.000

266

b) n bands PB %P He PLP1% I Pb AE D Φ'PT n 0.47 0.43 0.46 0.21 0.33 0.26 -0.10 0.10 0.12 0.11 bands 0.050 0.43 0.96 0.91 0.96 0.93 0.79 0.86 0.50 -0.06 PB 0.075 0.078 0.29 0.22 0.32 0.24 0.03 0.33 -0.28 0.60 %P 0.057 0.000 0.243 0.96 0.94 0.97 0.82 0.85 0.67 -0.28 He 0.400 0.000 0.371 0.000 0.95 1.00 0.92 0.94 0.67 -0.28 PLP1% 0.186 0.000 0.194 0.000 0.000 0.96 0.87 0.94 0.48 -0.07 I 0.302 0.000 0.341 0.000 0.000 0.000 0.91 0.93 0.67 -0.27 Pb 0.703 0.000 0.909 0.000 0.000 0.000 0.000 0.90 0.64 -0.31 AE 0.696 0.000 0.179 0.000 0.000 0.000 0.000 0.000 0.40 -0.01 D 0.646 0.035 0.263 0.002 0.002 0.044 0.002 0.004 0.099 -0.84 Φ'PT 0.672 0.824 0.009 0.259 0.263 0.775 0.269 0.208 0.976 0.000

267

Appendix 2.5. STRUCTURE analysis results determining the optimal number of K using the ∆K method by Evanno et al. (2005). # K = number of K analyzed; Reps = # of replicates; LnP(K) = Variation in mean (± SD); Variations in LnP(D) and ΔK over the estimates of K were calculated using Structure Harvester v.0.6.92 for Dryas integrifolia. The STRUCTURE analysis was performed using 50,000 iterations as burn-in followed by 100,000 iterations, running 10 replicates for a set of genetic clusters (K) one to 10. The no-admixture model with correlated allele frequencies was used.

Mean Stdev # K Reps LnP(K) LnP(K) Ln'(K) |Ln''(K)| Delta K a) Dryas integrifolia 1 10 -9051.58 7.07 NA NA NA 2 10 -7548.39 119.17 1503.19 170.30 1.43 3 10 -6215.50 11.13 1332.89 1564.85 140.65 4 10 -6447.46 184.70 -231.96 0.13 0.00 5 10 -6679.29 192.42 -231.83 4.17 0.02 6 10 -6906.95 345.14 -227.66 75.67 0.22 7 10 -7058.94 531.54 -151.99 58.13 0.11 8 10 -7152.80 331.34 -93.86 376.26 1.14 9 10 -6870.40 344.56 282.40 650.75 1.89 10 10 -7238.75 407.43 -368.35 NA NA

268

Appendix 3.1. Microclimate variable calculations

a) Growing degree days (GGDD)

GDD= ((Tmax+Tmin)/2)-Tbase

where Tmax equals the daily maximum temperature, Tmin equals the daily minimum temperature, and Tbase is the base or threshold temperature for metabolism

(here 5 ºC is used). If the mean daily temperature (Tmax–Tmin/2) is lower than the base temperature (Tbase) then GDD = 0. b) Vapour pressure deficit (VPD)

This is calculated using the following equations (Walter et al. 2005):

es = 0.6108exp(17.27T/(T+237.3))

Saturation vapour pressure (es) represents the capacity of the air to hold water; VPD units are kPa and temperature is in ºC.

ea = es * (RH/100)

Actual vapour pressure (ea, in kPa) represents amount of water vapour in air; RH is relative humidity in %.

269

Comparison Mean (°C) SE df t ratio p value (a) Locations within year 2011 NB-regional - NL-regional 3.89 0.54 37.42 7.19 <.0001 NB-regional - Wilson Brook 1.32 0.49 38.34 2.70 0.027 NL-regional - Wilson Brook -2.57 0.47 34.92 -5.51 <.0001 2012 NB-regional - NL-regional 2.18 0.54 38.13 4.01 0.001 NB-regional - Wilson Brook 1.11 0.49 38.90 2.25 0.075 NL-regional - Wilson Brook -1.08 0.47 34.97 -2.31 0.068 2013 NB-regional - NL-regional 4.08 0.57 44.28 7.13 <.0001 NB-regional - Wilson Brook 0.45 0.52 44.26 0.87 0.663 NL-regional - Wilson Brook -3.63 0.47 37.25 -7.68 <.0001 (b) Years within location NB-regional 2011 - 2012 -0.51 0.38 4906.15 -1.36 0.364 2011 - 2013 0.18 0.40 4596.93 0.44 0.899 2012 - 2013 0.69 0.40 4759.97 1.72 0.196 NL-regional 2011 - 2012 -2.22 0.34 4902.89 -6.63 <.0001 2011 - 2013 0.37 0.34 4908.11 1.07 0.536 2012 - 2013 2.59 0.34 4908.47 7.52 <.0001 Wilson Brook-local 2011 - 2012 -0.73 0.26 4903.00 -2.81 0.014 2011 - 2013 -0.70 0.26 4903.00 -2.68 0.020 2012 - 2013 0.03 0.26 4902.82 0.12 0.992

1 There was a significant interaction of year and site (χ2 = 38.93, df = 4, P < 0.001).

270

Comparison: month within Mean SE df t ratio p value region (°C) June NB-regional - NL-regional 4.413 0.538 134.9 8.198 <.0001 NB-regional - Port-au-Choix 1.167 0.487 138.22 2.397 0.662 NB-regional - Wilson Brook 0.303 0.487 138.22 0.622 1.000 NL-regional - Port-au-Choix - 3.246 0.456 118.46 -7.119 <.0001 NL-regional - Wilson Brook - 4.110 0.456 118.46 -9.014 <.0001 Port-au-Choix - Wilson Brook - 0.864 0.394 117.55 -2.193 0.798 July NB-regional - NL-regional 3.030 0.536 130.58 5.65 <.0001 NB-regional - Port-au-Choix 0.765 0.487 135.31 1.571 0.990 NB-regional - Wilson Brook 0.643 0.487 135.31 1.321 0.999 NL-regional - Port-au-Choix - 2.265 0.450 111.96 -5.032 0.000 NL-regional - Wilson Brook - 2.387 0.450 111.96 -5.302 0.000 Port-au-Choix - Wilson Brook - 0.122 0.390 112.21 -0.312 1.000 August NB-regional - NL-regional 2.512 0.528 125.65 4.753 0.001 NB-regional - Port-au-Choix 0.731 0.478 129.18 1.53 0.993 NB-regional - Wilson Brook 1.606 0.478 129.18 3.359 0.104 NL-regional - Port-au-Choix - 1.780 0.450 111.96 -3.954 0.018 NL-regional - Wilson Brook - 0.906 0.450 111.96 -2.012 0.891 Port-au-Choix - Wilson Brook 0.875 0.390 112.21 2.243 0.767 September NB-regional - NL-regional 0.764 0.533 130.85 1.435 0.997 NB-regional - Port-au-Choix 0.295 0.481 133.18 0.613 1.000 NB-regional - Wilson Brook 1.671 0.481 133.18 3.477 0.075 NL-regional - Port-au-Choix - 0.470 0.456 118.47 -1.03 1.000 NL-regional - Wilson Brook 0.907 0.456 118.47 1.989 0.901 Port-au-Choix - Wilson Brook 1.376 0.394 117.55 3.493 0.073 October 271

NB-regional - NL-regional 1.780 0.638 273.3 2.788 0.370 NB-regional - Port-au-Choix 2.447 0.576 276.36 4.249 0.005 NB-regional - Wilson Brook 1.801 0.576 276.36 3.127 0.178 NL-regional - Wilson Brook 0.667 0.537 237.86 1.243 1.000 NL-regional - Port-au-Choix 0.021 0.537 237.86 0.04 1.000 Port-au-Choix - Wilson Brook - 0.646 0.461 230.46 -1.401 0.998

1 There was a significant interaction of month and location (χ2 = 133.88.46, df = 12, P < 0.001).

272

Location Estimate SE t value a) Model: mean daily temperature= treatment, random=plot Port-au-Choix (Intercept) 15.096 0.165 91.425 Wilson Brook -0.017 0.234 -0.075 b) Model: maximum daily temperature= treatment, random=plot Port-au-Choix (Intercept) 25.082 0.737 34.030 Wilson Brook 0.626 1.042 0.600 c) Model: minimum daily temperature= treatment, random=plot Port-au-Choix (Intercept) 8.984 0.283 31.747 Wilson Brook 0.579 0.400 1.446 d) Model: vapour pressure deficit (VPD)= treatment, random=plot Port-au-Choix (Intercept) 0.158 0.026 6.106 Wilson Brook1 0.250 0.037 6.830 e) Model: growing degree days (GDD)= location Port-au-Choix (Intercept) 1765.6 25.94 68.067 Wilson Brook -2.6 36.68 -0.071

1 Significantly different: χ2 = 17.34, df = 1, P < 0.001

273

Location Estimate SE t value a) Model: mean daily temperature= treatment, random=plot Albert Mines (Intercept) 12.676 0.610 20.785 Wilson Brook 0.353 0.722 0.489 b) Model: maximum daily temperature= treatment, random=plot Albert Mines (Intercept) 19.750 0.992 19.908 Wilson Brook1 3.363 1.174 2.865 c) Model: minimum daily temperature= treatment, random=plot Albert Mines (Intercept) 6.954 0.652 10.663 Wilson Brook 0.612 0.772 0.793 d) Model: vapour pressure deficit (VPD) = treatment, random=plot Albert Mines (Intercept) 0.209 0.011 19.414 Wilson Brook2 -0.123 0.013 -9.327 e) Model: growing degree days (GDD)= location Albert Mines (Intercept) 1851.5 126.2 14.672 Wilson Brook 227.1 149.3 1.521

1 Significantly different: χ2 = 5.43, df = 1, P = 0.02 2 Significantly different: χ2 = 17.74, df = 1, P < 0.001

274

Appendix 3.6. Post-hoc multiple comparisons (Tukey contrasts) of linear mixed effects model analysis (lmer in R) of microclimate variables using least-squares means to compare Open Top Chambers (OTC) and Control (Cont) plots over three years, 1 June to 9 November, 2011 to 2013. Values for variables were derived from 22,896 readings (taken at 10 min. intervals) over 162 days in each year; significance of interactions were derived from log likelihood comparison. The least-squared mean estimates for the fixed effect treatment was obtained using the lsmeans (Lenth 2016) and pbkrtest (Halekoh & Højsgaard 2014) packages. F test is calculated according to the Kenward-Roger method (Kenward & Roger, 1997, Halekoh, 2014).

Comparison Mean (°C) SE df t ratio p value a) Mean daily temperatures (significance of interaction year and treatment: χ 2 = 7.58, df = 2, P = 0.023) Treatments within years 2011 Cont - OTC -0.01 0.39 31.89 -0.01 >0.999 2012 Cont - OTC -0.27 0.39 32.97 -0.68 0.983 2013 Cont - OTC 0.64 0.39 32.97 1.63 0.586 Years within treatments 2011 Cont - 2012 Cont -0.73 0.27 6947.24 -2.66 0.083 2011 Cont - 2013 Cont -0.70 0.27 6947.24 -2.55 0.111 2012 Cont - 2013 Cont 0.03 0.27 6947.06 0.12 >0.999 2011 OTC - 2012 OTC -0.99 0.20 6449.10 -4.91 <0.001 2011 OTC - 2013 OTC -0.05 0.20 6449.10 -0.25 >0.999 2012 OTC - 2013OTC 0.94 0.20 6947.06 4.63 <0.001 b) Maximum daily temperatures (significance of interaction year and treatment: χ 2 = 19.31, df = 2, P < 0.001) Treatments within years 2011 Cont - OTC 1.59 0.67 32.35 2.38 0.195 2012 Cont - OTC 0.89 0.67 32.15 1.33 0.764 2013 Cont - OTC 3.38 0.67 32.15 5.07 <0.001 Years within treatments 2011 Cont - 2012 Cont -0.39 0.47 6785.18 -0.82 0.963 2011 Cont - 2013 Cont -0.31 0.47 6785.18 -0.66 0.986 2012 Cont - 2013 Cont 0.08 0.47 6785.00 0.17 >0.999 2011 OTC - 2012 OTC -1.08 0.35 6785.00 -3.10 0.024 2011 OTC - 2013 OTC 1.49 0.35 6785.00 4.24 <0.001 2012 OTC - 2013OTC 2.57 0.35 6785.00 7.34 <0.001

275

Appendix 3.6. continued Comparison Mean °(C) SE df t ratio p value c) Vapour pressure deficit (significance of interaction year and treatment: χ 2 = 58.62, df = 2, P < 0.001) Treatments within years 2011 Cont - OTC 0.04 0.03 20.05 1.22 0.824 2012 Cont - OTC -0.01 0.03 19.70 -0.22 >0.999 2013 Cont - OTC -0.07 0.03 20.95 -2.28 0.248 Years within treatments 2011 Cont - 2012 Cont -0.03 0.01 6256.00 -2.86 0.049 2011 Cont - 2013 Cont 0.02 0.01 6273.24 2.22 0.230 2012 Cont - 2013 Cont 0.05 0.01 6273.32 4.84 <0.001 2011 OTC - 2012 OTC -0.07 0.01 5960.66 -8.56 <0.001 2011 OTC - 2013 OTC -0.08 0.01 6187.08 -9.65 <0.001 2012 OTC - 2013 OTC -0.01 0.01 6259.09 -1.18 0.847 d) Frequency of temperature 25–40°C (significance of interaction year and treatment: χ 2 = 10.62, df = 1, P = 0.005) Treatments within years 2011 Cont - OTC 85.400 201.402 35.21 0.424 0.998 2012 Cont - OTC -53.039 204.535 36.00 -0.259 1.000 2013 Cont - OTC 616.317 204.535 36.00 3.013 0.050 Years within treatments 2011 Cont - 2012 Cont -223.600 170.749 32.70 -1.310 0.778 2011 Cont - 2013 Cont -287.400 170.749 32.70 -1.683 0.552 2012 Cont - 2013 Cont -63.800 170.749 32.70 -0.374 0.999 2011 OTC - 2012 OTC -362.039 125.894 33.88 -2.876 0.069 2011 OTC - 2013 OTC 243.517 125.894 33.88 1.934 0.400 2012 OTC - 2013 OTC 605.556 127.269 32.70 4.758 <0.001

276

Curriculum Vitae

Candidate’s full name: Sabine Barbara Dietz

Universities attended:

Degree University Field Year

Masters Université de Moncton Environmental Studies 2003

Bachelor of Arts Trent University Environmental and Resource 1990 Studies Publications:

Dietz, S. 2004. Big Flat and Grand Falls Furbish’s lousewort Sites, New Brunswick, Site Specific Stewardship Measures and Actions. The Nature Trust of New Brunswick & NB Power, Fredericton, NB.

Bishop, G. and S. Dietz. 2001. Champ de Tir de Tracadie. Tracadie River Vascular Plant Inventory and Habitat Evaluation. Prepared for the Department of Natural Resources and Energy. New Brunswick Federation of Naturalists/ Piper Project.

Dietz, S. and R. Chiasson. 2001. Gulf of St. Lawrence Aster (Symphyotrichum laurentianum). Management and Monitoring Plan. Kouchibouguac National Park. Kent County, NB.

Chiasson, R. and S. Dietz. 2000. The Brothers Important Bird Area. Conservation Concerns and Measures. Can. Nature Fed., Bird Studies Can., NB Federation of Naturalists, Natural History Soc. of PEI., Federation of NS Naturalists.

Conference Presentations:

May 29, 2014, “The genetic structure and phylogeography of White Mountain Avens (Dryas integrifolia) in Eastern Canada, from New Brunswick to Labrador”, Genomes to / aux Biomes, Montreal, Canada.

277