Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Selective breeding of lodgepole pine and interior spruce generates growth gains but maintains phenotypic… MacLachlan, Ian Robert 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_september_maclachlan_ian.pdf [ 6.15MB ]
Metadata
JSON: 24-1.0349088.json
JSON-LD: 24-1.0349088-ld.json
RDF/XML (Pretty): 24-1.0349088-rdf.xml
RDF/JSON: 24-1.0349088-rdf.json
Turtle: 24-1.0349088-turtle.txt
N-Triples: 24-1.0349088-rdf-ntriples.txt
Original Record: 24-1.0349088-source.json
Full Text
24-1.0349088-fulltext.txt
Citation
24-1.0349088.ris

Full Text

 SELECTIVE BREEDING OF LODGEPOLE PINE AND INTERIOR SPRUCE GENERATES GROWTH GAINS BUT MAINTAINS PHENOTYPIC AND GENOMIC ADAPTATION TO CLIMATE  by  Ian Robert MacLachlan  BSc. (Hons.), University of Aberdeen, 2007 MSc., University of Edinburgh, 2008   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Forestry)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  July 2017  © Ian Robert MacLachlan, 2017 ii  Abstract Climate change is disrupting local adaptation in temperate and boreal tree species. As climates shift, tree breeding zones are becoming dissociated from their historical climatic optima and no longer represent optimal seed deployment zones. Assisted gene flow (AGF) policies that match reforestation seedlots with future climates require accurate knowledge of genetic variation in climatically adaptive traits in breeding populations. In this thesis I evaluate the effects of selective breeding on climatic adaptation in the two most planted species in western Canada, lodgepole pine (Pinus contorta) and interior spruce (Picea glauca, P. engelmanii and their hybrids), to inform provincial AGF prescriptions.  I compared natural stand seedlots (n = 105 pine, 154 spruce) with selectively bred seedlots (n = 20 pine, 18 spruce) from across Alberta and British Columbia in common garden experiments. Phenotypic variation among breeding zones was assessed for growth, phenology and cold hardiness in relation to climate. For both species, phenotypic differences between natural and selected seedlings in growth traits were substantial. Height gains resulted from increased growth rate and delayed growth cessation, but autumn cold hardiness was not substantially reduced.  Seedlings were also genotyped for ~30,000 candidate single nucleotide polymorphisms for growth and adaptive traits. Selection for growth has shifted interior spruce hybrid ancestry in some breeding populations, but these effects are not consistent across zones. A genome-wide association study of pine identified many trait-associated SNPs. Positive effect allele frequencies among pine breeding zones were strongly associated with climatic variation. Selection has resulted in small increases in the frequency of positive effect alleles in breeding populations.  Associations among cold hardiness phenotypes, genotypes and climate dominated signals of local adaptation were preserved in breeding populations. Selection, breeding and progeny testing combined have produced taller pine and spruce seedlings without compromising climatic adaptation. Strong phenotype-genotype-climate associations suggest AGF will be necessary to match breeding populations with future climates, but selectively bred and natural seedlots can be safely redeployed using the same AGF prescriptions. Multi-locus genomic profiles of adaptive traits associated with climate provide an accurate, rapid method to assess climatic adaptation that is independent from long-term provenance trials.   iii  Lay Summary Natural selection and evolution have produced tree populations that are adapted to their local climates, but climatic change threatens local adaptation in western Canada’s conifer breeding programs. This effect may compromise future timber yields and can be compounded by trade-offs among climatically adaptive traits in response to breeding. My thesis investigates these trade-offs and their implications for new provincial policies that match reforestation seedlots with the future climates of reforestation sites. I found that selective breeding generates substantial height gains in lodgepole pine and interior spruce seedlings grown in a uniform environment. Climate-related seedling traits and their associated genetic variation both respond to selective breeding, but potential trade-offs between height growth and climatically adaptive traits, particularly cold hardiness, are negligible. Breeding programs maintain the climatic adaptation of reforestation seedlots. Selectively bred seedlots can be deployed to match future climates using the same policy prescriptions as their natural seedlot counterparts.   iv  Preface The research reported in this thesis was conducted as part of AdapTree, a large-scale applied genomics project based at the University of British Columbia Centre for Forest Conservation Genetics jointly lead by Dr. Sally Aitken and Dr. Andreas Hamann. As such, AdapTree was a collaborative project. The molecular genetics laboratory work for DNA extraction and preparation reported in Chapters 2 and 3 was completed by Kristin Nurkowski and her technical staff. The bioinformatics data processing and analyses that led to production of the interior spruce and lodgepole pine SNP arrays used in Chapters 2 and 3 respectively, was led by Dr. Sam Yeaman, Dr. Kay Hodgins, and Dr. Katie Lotterhos, with the assistance and advice from many other AdapTree team members.   Within AdapTree I was lead investigator for the comparisons or natural and selectively bred reforestation seedlots. Under the supervision of Dr. Aitken I refined the experimental questions around natural and selectively bred seedlot comparisons, lead the experimental design and establishment of four common gardens (results from two of these common gardens are reported in this thesis), and data collection assisted by Dr. Pia Smets and a large number of technical staff. I conducted all the analyses reported in the following chapters, and wrote the whole text unless otherwise stated below. Dr. Aitken made intellectual contributions throughout these research chapters, and meticulously edited earlier versions of this thesis including its previously published components.  Chapter 2. A version of this chapter has been published as MacLachlan, I.R., Wang, T., Hamann, A., Smets, P. & Aitken, S.N., (2017). Selective breeding of lodgepole pine increases growth and maintains climatic adaptation. Forest Ecology and Management, 391, pp.404–416. Dr. Tongli Wang and Dr. Hamman contributed to the experimental sampling and manuscript review, Dr. Smets contributed technical advice and reviewed the manuscript. Components of the introduction to this publication have been modified and included in Chapter 1.  Chapter 3. A version of this chapter has been accepted for publication pending minor revisions, by the journal Evolutionary Applications as MacLachlan, I.R., Yeaman, S. & Aitken, S.N., (2017). Growth gains from selective breeding in a spruce hybrid zone do not compromise local adaptation to climate. Dr. Yeaman wrote the description of the SNP array design in Chapter 3.2.3 and reviewed the manuscript. Joane Elleouet contributed Sitka spruce reference genotypes and Jon Degner advised on implementing the hybrid ancestry analysis.  Chapter 4 is unpublished. I conceived the comparisons between natural and selectively bred seedlots and carried out all the analyses for this chapter. Dr. Jeremy Yoder provided intellectual input on genomic data processing and genotype-phenotype association analysis. v  Table of Contents  Abstract ............................................................................................................................................. ii Lay Summary .................................................................................................................................... iii Preface ............................................................................................................................................. iv Table of Contents ............................................................................................................................... v List of Tables ................................................................................................................................... viii List of Figures .................................................................................................................................... ix Acknowledgements........................................................................................................................... xi Dedication ...................................................................................................................................... xiii Chapter 1: Introduction ...................................................................................................................... 1 1.1 Forests and Climate Change ........................................................................................................ 1 1.2 Climatic Adaptation in Temperate and Boreal Conifers .............................................................. 2 1.3 Molecular Genetics of Adaptive Traits in Conifers ...................................................................... 3 1.4 Selective Breeding in Conifers ..................................................................................................... 4 1.4.1 Genetic Management in Conifer Breeding Programs ............................................................. 5 1.4.2 Adaptive Trade-Offs in Conifer Breeding Programs ................................................................ 6 1.4.3 Conifer Breeding Programs in Western Canada ..................................................................... 7 1.5 Climate-Based Seed Transfer and Assisted Gene Flow ............................................................... 8 1.6 Research Rationale and Objectives ............................................................................................. 9 Chapter 2: Selective breeding of lodgepole pine increases growth and maintains climatic adaptation 11 2.1 Introduction ............................................................................................................................... 11 2.3 Results ....................................................................................................................................... 18 2.3.1 Breeding Zone-by-Seedling Type Means ............................................................................... 18 2.3.2 Phenotypic Differentiation among Breeding Zones (VPOP) ................................................... 19 2.3.3 Clinal Analysis ........................................................................................................................ 20 2.3.4 Trait-Trait Correlations .......................................................................................................... 20 2.3.5 Climatic Biases in Breeding Programs ................................................................................... 21 2.4 Discussion .................................................................................................................................. 21 2.4.1 Effects of Selection on Adaptive Traits ................................................................................. 21 2.4.2 Correlated Responses to Selection ....................................................................................... 23 2.4.3 Mechanisms of Growth Responses to Selective Breeding .................................................... 24 vi  2.5 Conclusions ................................................................................................................................ 27 2.6 Tables ......................................................................................................................................... 29 2.7 Figures ....................................................................................................................................... 32 Chapter 3: Growth gains from selective breeding in a spruce hybrid zone do not compromise local adaptation to climate ....................................................................................................................... 36 3.1 Introduction ............................................................................................................................... 36 3.2 Methods .................................................................................................................................... 39 3.2.1 Experimental Sampling and Establishment ........................................................................... 39 3.2.2 Climatic Data ......................................................................................................................... 39 3.2.3 DNA Extraction and SNP Genotyping .................................................................................... 40 3.2.4 Hybrid Index Analyses ........................................................................................................... 41 3.2.5 Phenotypic Data and Analyses .............................................................................................. 42 3.2.6 Climatic Biases ....................................................................................................................... 43 3.3 Results ....................................................................................................................................... 43 3.3.1 Hybrid Index Means .............................................................................................................. 44 3.3.2 Hybrid Index-Trait Relationships ........................................................................................... 45 3.3.3 Hybrid Index Clines ................................................................................................................ 45 3.3.4 Breeding Zone by Seedlot Type Means ................................................................................. 45 3.3.5 Trait-Trait Correlations .......................................................................................................... 46 3.3.6 Phenotypic Clines .................................................................................................................. 47 3.3.7 Climatic Biases ....................................................................................................................... 47 3.4 Discussion .................................................................................................................................. 48 3.4.1 Selective Breeding and Hybrid Ancestry ............................................................................... 48 3.4.2 Effects of Selection on Adaptive Phenotypic Traits .............................................................. 50 3.5 Conclusions ................................................................................................................................ 52 3.6 Tables ......................................................................................................................................... 54 3.7 Figures ....................................................................................................................................... 58 Chapter 4: Subtle shifts in polygenic variation underlying adaptive traits of lodgepole pine confer strong climatic adaptation and modest responses to selective breeding ............................................ 64 4.1 Introduction ............................................................................................................................... 64 4.2 Materials and Methods ............................................................................................................. 67 4.2.1 DNA Extraction and Genotyping ........................................................................................... 67 vii  4.2.2 Genome-Wide Phenotypic Associations (GWAs) .................................................................. 67 4.2.3 Allele Frequency Distribution of Top Candidate SNPs .......................................................... 68 4.2.4 Linkage Disequilibrium Analysis ............................................................................................ 68 4.2.5 Positive Effect Allele-Climate Relationships .......................................................................... 68 4.2.6 Positive Effect Allele-Phenotype Relationships..................................................................... 69 4.3 Results ....................................................................................................................................... 70 4.3.1 Allele Frequency Distribution of Top Candidate SNPs .......................................................... 70 4.3.2 Linkage Disequilibrium Analysis ............................................................................................ 71 4.3.3 Positive Effect Allele-Climate Relationships .......................................................................... 72 4.3.4 Positive Effect Allele-Phenotype Relationships..................................................................... 73 4.4 Discussion .................................................................................................................................. 74 4.4.1 Genome-Wide Signatures of Adaptation and Selective Breeding ........................................ 74 4.4.2 Effects of Selection Among and Within Breeding Zones ....................................................... 76 4.4.3 Prospects for Long-Term Genetic Gain and Assisted Gene Flow .......................................... 78 4.5 Conclusions: A Genomic Basis for Assisted Gene Flow ............................................................. 79 4.6 Tables ......................................................................................................................................... 81 4.7 Figures ....................................................................................................................................... 83 Chapter 5: Conclusions ..................................................................................................................... 90 5.1 Introduction ............................................................................................................................... 90 5.2 Selective Breeding and Climatic Adaptation ............................................................................. 91 5.3 Assisted Gene Flow in Selected Seedlots .................................................................................. 93 5.4 A Domestication Syndrome in Conifers? ................................................................................... 94 5.5 Study Limitations ....................................................................................................................... 95 5.6 Future Research Directions ....................................................................................................... 96 References ....................................................................................................................................... 99 Appendices .................................................................................................................................... 114 Appendix A Supplementary materials for Chapter 2 ............................................................................ 114 Appendix B Supplementary materials for Chapter 3 ............................................................................ 119 Appendix C Supplementary materials for Chapter 4 ............................................................................ 129  viii  List of Tables  Table 2.1 Breeding zones sampled for natural and selected seedlots ....................................................... 29 Table 2.2 Proportions of phenotypic variance among (σp2) and within (σe2) breeding zones ................ 30 Table 2.3 Pairwise correlation coefficients between all six traits for natural and selected seedlings ....... 31 Table 3.1 Breeding zones sampled for selectively bred and natural seedlots ........................................... 54 Table 3.2 Variation in breeding zone hybrid index means explained by breeding zone climate ............... 55 Table 3.3 Variation in breeding zone trait means explained by breeding zone hybrid index means ........ 56 Table 3.4 Pairwise correlation coefficients between all six traits for natural and selected seedlings ....... 57 Table 4.1 Mean number of positive effect alleles for the top 1% (n = 317) of candidate SNPs ................. 81 Table 4.2 Pairwise linkage disequilibrium summarised as the means of squared allelic correlations ....... 82 Table A.1 Climatic variables interpolated from ClimateNA. ..................................................................... 114 Table A.2 PC1–4 effects from PCA of the 22 lodgepole pine seedling climate variables in Table A.1. .... 115 Table A.3 Ranked PC1 and PC2 loadings from the PCA of all 22 climate variables in Table A.1. ............. 116 Table A.4 BLUEs of trait means for each breeding zone and seedlot type combination. ........................ 117 Table A.5 r2 values for clines in eleven climatic variables of the six traits and two seedling types. ........ 118 Table B.1 Mean proportions (ADMIXTURE Q-values) of the three parental spruce species’ .................. 119 Table B.2 PC1–4 effects from PCA of the 16 interior spruce seedling climate variables in Table 3.2. ..... 120 Table B.3 Ranked PC1 and PC2 loadings from the PCA of all 16 climate variables in Table 3.2............... 121 Table B.4 BLUEs of trait means for each breeding zone and seedlot type combination. ........................ 122 Table B.5 r2 values of phenotypic clines with sixteen climatic variables for six traits and two seedling types. ......................................................................................................................................................... 123 Table C.1 Mean frequency of positive effect alleles associated with each trait ...................................... 129 Table C.2 r2 values for clines with climate in the frequency of positive effect alleles by their respective associated trait .......................................................................................................................................... 130   ix  List of Figures  Figure 2.1 Geographic origins of the natural and selected seedling populations sampled from across the range of lodgepole pine in AB and BC ......................................................................................................... 32 Figure 2.2 Bar plots of breeding zone level phenotypic means ................................................................. 33 Figure 2.3 Phenotypic clines with MAT ...................................................................................................... 34 Figure 2.4 Correlations of differences in the values of height between natural and selected seedlings of each breeding zone with equivalent differences in phenology and cold injury traits................................ 35 Figure 3.1 Geographic origins of the natural and selected seedling populations sampled from across the ranges of P. engelmannii, P. glauca, and their hybrid zone in AB and BC .................................................. 58 Figure 3.2 Ternary plots representing the proportion of Engelmann (P. engelmannii), white (P. glauca) and Sitka (P. sitchensis) spruce ancestry .................................................................................................... 59 Figure 3.3 Regressions of a) height and b) cold injury versus hybrid index ............................................... 60 Figure 3.4 Examples of climatic clines in hybrid index ............................................................................... 61 Figure 3.5 Bar plots of breeding zone level trait means (BLUEs) ................................................................ 62 Figure 3.6 Phenotypic clines with MAT ...................................................................................................... 63 Figure 4.1 Probability densities for the frequency of individuals with a given number of positive effect alleles .......................................................................................................................................................... 83 Figure 4.2 Contrasting heat maps of pairwise linkage disequilibrium (r2) ................................................. 84 Figure 4.3 Clines in the frequency of effect alleles for the top 1% candidate SNPs of four traits with four climatic variables......................................................................................................................................... 85 Figure 4.4 Correlations between genotype-climate and phenotype-climate clines .................................. 86 Figure 4.5 Phenotypic variation in each trait and both seedling types explained by the mean frequency of effect alleles from the top 1% of phenotype associated SNPs in each breeding zone .......................... 87 Figure 4.6 Correlations of within-breeding zone phenotypic and genotypic difference between seedling types ............................................................................................................................................................ 88 Figure 4.7 Venn diagram of the overlap between the top 1% of candidate SNPs associated with height, growth cessation and cold injury ................................................................................................................ 89 Figure B.1 Comparison of ADMIXTURE and Structure Q-values ............................................................... 124 Figure B.2 Mean spruce hybrid index (proportion of P. engelmannii ancestry) ...................................... 125 Figure B.3 Bar plots of breeding zone level trait means (BLUEs) ............................................................. 126 x  Figure B.4 Regressions of a) growth rate, b) shoot mass, c) bud break, and d) bud set, versus hybrid index .......................................................................................................................................................... 127 Figure B.5 Regression of height gains in each breeding zone on mean summer temperature differences .................................................................................................................................................................. 128 Figure C.1 Height: heat map of pairwise linkage disequilibrium (r2) for the top 1% of height associated SNPs .......................................................................................................................................................... 130 Figure C.2 Growth rate: heat map of pairwise linkage disequilibrium (r2) for the top 1% of growth rate associated SNPs ........................................................................................................................................ 130 Figure C.3 Shoot mass: heat map of pairwise linkage disequilibrium (r2) for the top 1% of shoot mass associated SNPs ........................................................................................................................................ 130 Figure C.4 Cold injury: heat map of pairwise linkage disequilibrium (r2) for the top 1% of cold injury associated SNPs ........................................................................................................................................ 130 Figure C.5 Growth initiation: heat map of pairwise linkage disequilibrium (r2) for the top 1% of growth initiation associated SNPs ......................................................................................................................... 130 Figure C.6 Growth cessation: heat map of pairwise linkage disequilibrium (r2) for the top 1% of growth cessation associated SNPs ........................................................................................................................ 130  xi  Acknowledgements  I cannot thank Sally Aitken enough as my supervisor, mentor and friend over nearly seven years of collaboration, bringing AdapTree to life and working in the undergraduate teaching environment. Sally has been calm, supportive, and most of all, patient. She has always shared her expertise and experience freely, and encouraged opportunities for my academic development. Sometimes grad school has been personally challenging, but working with Sally has defined my time at UBC as a positive experience. I will always be deeply appreciative of this.  My research would not have been possible without extensive technical assistance, advice and feedback from the Aitken Lab, the AdapTree team, and many undergraduate assistants. I would like to thank Pia Smets for sharing her advice and expertise, but also for working tirelessly with me on the common garden establishment, management and phenotypic data collection. Similarly, Joanne Tuytel was indispensable at all stages of common garden management and data collection. Conor Fitzpatrick, Jon Degner and Sean King also contributed extensively to these activities. Seane Trehearne never failed to be helpful and accommodating at the UBC Totem Field site. The extensive molecular lab work associated with my research was masterminded and led by Kristin Nurkowski with extensive help from Robin Mellway. The AdapTree bioinformatics team led by Sam Yeaman, Kay Hodgins and Katie Lotterhos made the genomic aspects of my research a reality. Tongli Wang has always been generous with his advice on climatic and phenotypic data analysis. Jeremy Yoder has guided me through the perils and pitfalls of genomic data analysis. I am also grateful to my supervisory committee, Yousry El-Kassaby, Loren Rieseberg, and Greg O’Neill for their advice, thoughtful discussions and feedback that have helped to shape my research.  The provincial tree breeders and research scientists who are the end-users of my research in the Alberta and British Columbia provincial governments have been all extremely welcoming, enthusiastic and willing to share their wealth of both scientific and operational expertise. Notable among these people, I thank Nick Ukrainetz and Barry Jaquish as the lodgepole pine and interior spruce breeders for the BC Ministry of Forest Lands and Natural Resource Operations, and Andy Benowicz, Forest Genetics Specialist with AB Sustainable Resource Development. Jack Woods, Program Manager for the Forest Genetics Council of BC has always been supportive, offering his expertise and providing information on provincial forest genetic resources.  The final person I would like to thank is Joane Elleouet. She has provided technical assistance and data analysis advice as a lab mate, but more importantly Joane has given me unwavering support as xii  a friend, provided baked goods, and been a conspirator in occasional outbursts of comical office madness that keep the mentally challenging aspects of graduate school in check.  Seeds for AdapTree were kindly donated by 63 forest companies and agencies in Alberta and British Columbia (listed at http://adaptree.forestry.ubc.ca/seed-contributors/). Seed donation was facilitated by the Alberta Tree Improvement and Seed Centre, and the BCMFLNRO Tree Seed Centre.  As part of AdapTree, my research was funded by Genome Canada, Genome BC, Genome Alberta, Alberta Innovates BioSolutions, the Forest Genetics Council of British Columbia, the British Columbia Ministry of Forests, Lands and Natural Resource Operations (BCMFLNRO), Virginia Polytechnic University, and the University of British Columbia. xiii  Dedication  I dedicate this thesis to my parents, Gillian and Malcolm MacLachlan, whose enthusiasm, expertise and encouragement instilled in me a lifelong passion for trees and plants.  1  Chapter 1: Introduction  1.1 Forests and Climate Change At no point in the history of reforestation has the need to establish the right trees in the right places been more acute or challenging than at present. Post-glacial recolonisation redistributed temperate and boreal trees according to species’ specific ecological niches, while natural selection has shaped the distribution of genetic diversity within species, leading to local adaptation (Savolainen et al. 2007). Shifting climates and greater climatic variation are starting to disrupt historical local adaptation (Gauthier et al. 2015; Millar & Stephenson 2015). As a consequence, tree populations are simultaneously challenged to withstand the consequences of novel climates, and unable to adapt or migrate rapidly enough to remain locally adapted (Aitken et al. 2008). Therefore, to evaluate the risks posed by climate change, knowing the adaptive potential for forest tree populations is crucial (Neale & Kremer 2011). Conifers have a long evolutionary history of adaptation to harsh climatic, edaphic and biological pressures, culminating in their ecological exclusion to unproductive marginal habitats (Bond 1989). In temperate and boreal regions, recent conifer evolution is characterised by post-glacial (Holocene) recolonisation and range expansion.  Most tree species had post-glacial migration rates of 0.05-0.5 km yr-1 estimated from fossil pollen, and post-glacial seed dispersal rates <0.1 km yr-1 from genetic marker evidence (McLachlan et al. 2005; Savolainen et al. 2007), that were sufficient for the biotic velocity of range expansion to track the velocity of climatic warming (Ordonez & Williams 2013). Compared to these tree migration rates, estimates of contemporary climatic change and specifically climate warming are several times greater. Loarie et al. (2009) suggest that during the 21st century the mean global velocity of temperature change will be 0.42 km yr-1. Boreal forests and temperate grasslands slightly exceed this average velocity (0.43 and 0.59 km yr-1 respectively), while mountainous conifer forest biomes experience a substantial topographic buffering effect (velocity = 0.11 km yr-1). On this basis, the velocity of temperature change and the ability of tree populations to track shifting climates may vary substantially in response to the topographic variation of western Canada.  For widely distributed conifer species in unmanaged landscapes, their previous evolutionary history, the potential for adaptive long distance gene flow, and strong selection in large populations suggest that long-term evolutionary adaptation to future climates is realistic (Davis & Shaw 2001; Kremer et al. 2012). Even so, a short-term adaptive lag over a few generations is expected because locally adapted populations will become dissociated from their historical climatic niches by rapid climate 2  change (Davis & Shaw 2001; Aitken et al. 2008). Evidence suggests this as already occurring in western Canada. Using ClimateNA (Wang et al. 2016), estimated mean annual temperature was 0.58oC higher across AB and BC during 2005 to 2014 than during the 1961 to 1990 reference period, although greater increases of 1.5 to 2oC have been observed in northern BC (British Columbia Ministry of Environment 2015). Climate niche models that suggest tree populations in western North America already lag 130 km behind their historic climates (Gray & Hamann 2013). Even a small decrease in forest productivity due to maladaptation will have substantial, long-term economic costs in Alberta (AB) and British Columbia (BC), where the annual timber harvest supports a multi-billion dollar forest industry.  Sustainable future timber yields will depend on establishing well-adapted forest stands that are productive under both the current and future climates expected during their rotation.   1.2 Climatic Adaptation in Temperate and Boreal Conifers Local adaptation to climate is expressed as genotype by environment (G x E) interactions caused by divergent evolution among populations, and characterised by greater relative fitness of genotypes in local versus non-local habitats (Kawecki & Ebert 2004). Across environmental gradients the G x E interactions of continuously distributed adaptive traits results in clinal variation (Endler 1977). Most widely distributed temperate and boreal tree species are locally adapted and exhibit clines in phenotypic traits along climatic gradients (Morgenstern 1996). Across large regional or continental climatic gradients, post-glacial recolonisation has stimulated the differentiation of quantitative variation in adaptive traits among populations (Savolainen et al. 2007; Alberto et al. 2013). However, local scale niche specialisation and adaptation in response to topographic variation overlays elevational clines upon wider climatic trends (Rehfeldt 1988). The earliest comparative studies of intraspecific adaptive variation among populations in relation to their environment (genecology) occurred more than 250 years ago to identify variation among provenances of forest trees that was commercially and strategically useful (Langlet 1971).  Strong associations between growth and provenance mean annual temperature are now known to reflect adaptation to large climatic gradients in a number of temperate species (Aitken & Bemmels 2015), but these associations reflect adaptation to average climatic conditions. It is the climatic extremes experienced by populations that act as selective agents, and are most damaging during the growing season and at growth-dormancy transitions in spring and autumn. During the growing season reduced water availability and increased temperature underlie potential drought stress and mortality. As climates shift and warm, this represents an increasing threat to forest health and ecosystem function 3  (Allen et al. 2010), but substantial variation in the tolerance of low moisture conditions exists among populations of widespread conifer species in response to climatic gradients (Bansal et al. 2016; Schuster & Oberhuber 2013; Montwé et al. 2015; Eckert et al. 2010).  Synchronisation of the phenological transitions between growth and dormancy with local seasonal changes define adaptation to climate in many plant species, especially conifers (Chuine 2010). This reflects a trade-off between the need to maximise potential growth and reproduction during favourable conditions, with the risk of cold injury from early or late season freezing events (Cooke et al. 2012). Under common garden conditions, variation in growth initiation among conifer populations in response to climate is usually weak (Aitken & Hannerz 2001). By contrast, late summer growth cessation, bud set and the acquisition of winter dormancy are strongly associated with photoperiod and temperature gradients.  These adaptive traits have relatively large amounts of variation among populations, and moderate to high heritabilities (Howe et al. 2003).  1.3 Molecular Genetics of Adaptive Traits in Conifers Several strategies have been used to characterise the genetic architecture of adaptive variation in conifers. Quantitative genetic methods use phenotypic variation to describe differences in traits among genetic groups that are caused by heritable molecular genetic variation, but the molecular basis of this variation is not identified (Falconer & MacKay 1996). Linkage mapping establishes the relative positions of segregating molecular markers on chromosomes based on recombination frequencies. However, linkage mapping on its own does not associate molecular variation with phenotypic variation, and ideally uses crosses with isogenic, inbred lines that are difficult to establish in conifers. Quantitative trait locus mapping is based on co-segregation of phenotypic variation with molecular markers (quantitative trait loci, QTL). It has been used to identify the genetic architecture of adaptive traits in a number of conifer species (Nichols & Neale 2010), but QTL mapping requires pedigree information from controlled crossing, and until recently used relatively few markers which are unlikely to be physically close to causal variants. Therefore, QTL mapping is slow and unsuitable for studying adaptive variation in natural populations of long-lived open-pollinated plants such as conifers (Ingvarsson & Street 2011). Genome-wide association studies (GWAS) overcome this by making statistical associations between phenotypic variation and genetic markers to identify QTL, using mixed linear models that account for population structure and relatedness (Yu et al. 2006; Hall et al. 2010). This allows for identification of markers that are causal, or closely associated with causal loci, but does not map loci to specific genomic regions. High levels of outcrossing, large effective population sizes, low levels of population structure and rapid decay 4  of linkage disequilibrium make natural conifer populations suitable GWAS subjects (Neale & Savolainen 2004). Ideally, GWAS studies are paired with linkage mapping of markers within a pedigree to determine the relative positions of associated markers within the genome. To date, association genetics has been used to identify adaptive genomic variation in a range of ecologically adaptive and commercially important conifer traits including: drought resistance (Eckert et al. 2010; Lind et al. 2017); cone serotiny (Parchman et al. 2012); growth phenology and cold hardiness (Holliday et al. 2010; Eckert et al. 2009); pathogen resistance (Quesada et al. 2010; Liu et al. 2016); and, wood quality traits (Gonzalez-Martinez et al. 2007; Beaulieu et al. 2011). Of these molecular associations, phenology and cold hardiness associations are possibly the most adaptively important for temperate and boreal conifers. They reflect phenotypes that have strong climatic relationships. Variation at loci associated with growth cessation and cold hardiness in several conifer species is climatically partitioned at a range of spatial scales (Lind et al. 2017; Hornoy et al. 2015; Eckert et al. 2009; Eckert et al. 2010), and local adaptation to cold temperatures has driven convergent molecular evolution between distantly related conifer species (Yeaman et al. 2016). Across these association studies, locally adaptive loci usually have small effect sizes (a few percent or less), and reinforce evidence from QTL studies that continuously distributed adaptive traits in conifers have highly polygenic genetic architectures (Neale & Savolainen 2004). However, association analyses do not usually account for the majority of heritable trait variation in quantitative traits (Ingvarsson & Street 2011). This implies that causal loci may be unidentified, heritability may be overestimated due to epistatic interactions; maternal or epigenetic effects; low frequency functional alleles may be important but filtered out by data processing pipelines; too few loci are studied; or loci that are strongly correlated with population structure may not be discovered due to type II errors (Zuk et al. 2012; Myles et al. 2009).  1.4 Selective Breeding in Conifers In forest trees the practice of favouring parents to produce offspring with preferred phenotypes is described as early as the 17th Century (Evelyn 1664). Tree breeders use artificial selection to shift the frequency distribution of desirable quantitative traits in offspring generations (Zobel & Talbert 1984), and the frequency of underlying alleles at associated loci, although these loci are usually anonymous. This results in a response to selection termed genetic gain; the difference in the mean value of a trait under selection between parent and offspring generations (Falconer & MacKay 1996). Breeding value is the genetic gain attributable to an offspring population. Both genetic gain and breeding value are 5  equivalent to the additive genetic variance of a trait; the proportion of quantitative genetic variance in a trait expressed under the prevailing selection regime due to variation at causal loci. Typically, conifer breeding programs start with the selection of wild-stand plus-trees that have been tested in provenance trials. Individuals with favourable phenotypes are then bred using controlled crossing, tested in replicated progeny trials and reselected. Selected genotypes will be used for further breeding cycles, or deployed in seed orchards to supply reforestation seedlots.  The most common objective of conifer breeding is to improve the volume and quality of timber produced and maximise economic returns. Improved resistance to pests and pathogens can also be necessary breeding objectives that contribute to timber production, and support genetic conservation efforts (Neale & Kremer 2011; Yanchuk 2001). The objectives of tree breeding may also be dependent on the silvicultural context of seedling deployment. Breeding programs that produce seedlings for relatively short rotation monoculture plantations focus on genetic gain for growth, at the expense of genetic diversity. Conversely, to deploy seedlings on natural landscapes where long-term ecosystem function is an important reforestation objective, breeding programs aim to achieve growth gains and maintain diverse, adaptive genotypes in forest stands. Genetic diversity is necessary because it enhances a forest stand’s capacity for resilience against environmental disturbances (Ledig & Kitzmiller 1992; Millar et al. 2007), while promoting both ecological and genetic diversity in the local species community (Whitham et al. 2003).   1.4.1 Genetic Management in Conifer Breeding Programs Historically foresters have taken the ‘local is best’ approach to selecting reforestation seed sources, because deploying local, climatically adaptive genotypes maintains natural population structure, and minimises the risk of stand losses during a rotation. Tree breeding programs that produce seed to reforest natural landscapes apply the same philosophy by operating within breeding zones. Breeding zones are defined by genetic differences among populations, as well as geographic and elevational boundaries. This limits the physical and climatic transfer distances of parents sourced for breeding programs, and the deployment transfer distances of seedlings, to minimise the likelihood of negative G x E interactions among adaptive traits. In addition to achieving short-term genetic gain, successful tree improvement programs must conserve the long-term genetic potential initially sampled from a broad base within populations, but avoid incurring the effects of inbreeding depression (Zobel & Talbert 1984). Conifers are biologically well-suited to this objective because phenotypic traits are highly polygenic, while their out-crossing, 6  wind-pollinated mating systems favour low levels of population structure and the maintenance of genetic diversity. Typically, conifer breeding programs select parent trees from spatially separated stands and a number of spatially separated trees within each stand in order to capture the maximum allelic diversity (Kitzmiller 1990). Relatedness among these selections will increase with each successive breeding cycle. Therefore, strategies such as splitting breeding populations into sub-populations or lines, managing co-ancestry using pedigrees, and limiting the number of selections within families, can be used to minimise inbreeding and maximise long-term genetic gain (White et al. 2007). Similarly, the diversity of selectively bred reforestation seedlots needs to be managed to avoid planting inbred trees and promote resilience to disturbance. This can be implemented by using open-pollinated seed orchards that are designed to minimise inbreeding among parent trees, and by establishing standards for the minimum number of parent trees that can contribute to a reforestation seedlot (Lstibuek & El-Kassaby 2010; Stoehr et al. 2004). At present these breeding program management strategies appear to be effective, and should be sustainable over multiple generations (Yanchuk 2001). The effects of conifer breeding on neutral genetic diversity are minimal or non-existent after one or two generations of selection, but low-frequency alleles are often eliminated (Stoehr & El-Kassaby 1997; Schmidtling et al. 1999; Godt et al. 2001; Hansen 2008). Even so, El-Kassaby & Ritland (1996) did find significant genetic divergence between natural populations and second generation seed orchards. Namroud et al. (2012) studied 1134 candidate white spruce (Picea glauca (Monech) Voss) SNPs in expressed genes and their responses to a single generation of selection at different intensities. These authors found that genetic diversity statistics were unaffected by selection, although some rare alleles were lost. Five and seven candidate SNPs in high and low selection intensity sample populations respectively, were associated with 15 year height growth and had small responses to selection. However, this study essentially replicates a genetic bottleneck, without accounting for the effects of advance generation selection, controlled crossing, and progeny testing processes in conifer breeding programs.   1.4.2 Adaptive Trade-Offs in Conifer Breeding Programs By shifting the frequency distribution of the primary trait under selection, selective breeding programs may generate correlated responses to selection in other desirable or adaptive traits. These indirect responses to selection occur when pleiotropy and linkage disequilibrium among underlying loci generate a selection differential in non-target phenotypic traits (Falconer & MacKay 1996). In crop breeding, extreme trade-offs among traits define the domestication syndrome and often confer reduced fitness in natural environments (Doebley et al. 2006). For foresters, antagonistic responses to selection 7  are important because they have the potential to dissociate adaptive traits in selectively bred seedlots from their local climatic optima, causing reduced health, productivity, and possible stand failure. This would cause substantial economic losses from reduced timber yields or expensive restocking (Hotte et al. 2016).  Tree breeders achieve growth gains by selecting faster growing genotypes or genotypes that have a longer growing period. If selection lengthens the growing season, growth gains from selective breeding could cause selectively bred trees to become unsynchronised with local climates and more vulnerable to cold injury. This can be especially damaging at the seedling stage (Howe et al. 2003). A similar effect may also occur if selective breeding reconstitutes adaptive phenotypes that are equivalent to populations present elsewhere on the landscape where milder climates favour increased growth. This could result in phenology and cold hardiness phenotypes that are maladaptive when deployed in local breeding zones (Rehfeldt 1992a; Rehfeldt 1992b). Evidence from O’Neill et al. (2014) suggests that breeding programs achieve increased growth under local climates, but reduce the acceptable transfer distances of selectively bred seedlings, although it is unclear what traits underlie this response.  Conversely, if selective breeding increases stem growth without negative impacts on phenology or cold hardiness, climatically adaptive synchrony to seasonal changes will be unaffected. Previous results from mature Norway spruce (Picea abies (L.) Karst.) (Westin et al. 2000; Hannerz & Westin 2005) and lodgepole pine (Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) (Rehfeldt 1989) indicate this may be possible. However, the extent to which selective breeding actually produces either of these effects remains unclear.  1.4.3 Conifer Breeding Programs in Western Canada Conifers represent the vast majority of western Canada’s harvestable public forest land base. This is a multi-purpose forest resource, where the primary reforestation objective is to establish healthy, productive stands that retain their natural ecological function, and yield a future timber harvest that supports long-term socio-economic needs. Breeding programs that produce improved reforestation seedlots are the foundation for achieving this objective. Conifer breeding programs in western Canada have been underway since the 1960’s (Xie & Yanchuk 2003). Alberta currently has breeding programs for six conifer species (Alberta Forest Genetic Resources Council 2015); British Columbia has 12 species-specific conifer breeding programs (Forest Genetics Council of British Columbia 2015a). Adaptation in these breeding programs is managed using fixed-boundary breeding and deployment zones delimited by genetic variation, geography and 8  elevation. Typically breeding programs in AB and BC represent continuous tree improvement progress and do not have discrete generations. At present most programs are at stages equivalent to second or third cycles of selection, breeding, and progeny testing, but currently produce seedlots from first or second generation seed orchards. By selecting for greater juvenile height growth, these programs primarily focus on producing reforestation seedlots with increased genetic worth for wood volume at rotation age, while maintaining climatic adaptation and genetic diversity. The genetic worth of a seedlot is the average genetic gain expected at the time of timber harvesting, calculated from the breeding values of all seed orchard parent trees weighted by their gametic contribution to a seedlot (Woods et al. 1996). Trees selected for orchard production are forward or backward selections based on progeny tested on three or four climatically representative sites within each breeding program. Differing selection intensities and numbers of breeding cycles mean that genetic worth for growth varies by breeding program and province. Genetic worth at rotation age is up to 10% in AB (A. Benowicz and S. John, personal communications) and up to 35%  (average ~17%) in BC (Forest Genetics Council of British Columbia 2015b). To ensure sustainable future yields of high quality timber, the use of selectively bred seedlots is mandatory in both provinces (British Columbia MFLNRO 2010; Alberta Forest Genetic Resources Council 2015). In western Canada, artificial reforestation of public land with nursery grown seedlings follows timber harvesting on most sites. Annually, 200,000 to 260,000ha of public land is reforested this way across AB and BC, using 250 to 325 million seedlings (Forest Genetics Council of British Columbia 2015a; Alberta Environment and Sustainable Resource Development 2011; National Forest Database 2014). The majority of these seedlings are either lodgepole pine, or interior spruce (Picea glauca (Monech) Voss, Picea engelmannii Parry ex Engelm., and their natural hybrids). Of the total annual planting, approximately 35 million lodgepole pine and 100 million interior spruce seedlings originate from selective breeding programs. Both provinces have objectives to increase the genetic worth of selectively bred reforestation seedlings, and the proportion of selectively bred reforestation seedlings that are deployed on public land, as the basis for sustainable future timber yields (Forest Genetics Council of British Columbia (2015a), A. Benowicz, personal communication).   1.5 Climate-Based Seed Transfer and Assisted Gene Flow Deploying ~300 million seedlings annually requires millions of dollars of initial investment, and represents billions of dollars in future timber revenue. Therefore, accurately matching seedlot deployment with local climates will be essential to optimising timber yields. However, as a management 9  tool, breeding and seedlot deployment zones that match local genotypes with local climates have three important caveats. First, the climatic characteristics of reforestation sites near breeding zone boundaries may overlap with those of adjacent breeding zones, but current rules limit cross-boundary seed transfer. Secondly, current breeding zones may be ecologically and operationally inefficient. The geographic space over which both natural and selectively bred populations are able to perform with an acceptable level of growth is greater than previously thought (O’Neill et al. 2014; Liepe et al. 2016). Thirdly, as global climates shift and become more variable, genotypes in static, geographically defined seed deployment zones will become dissociated from their local climates and maladapted to future climates (Aitken et al. 2008; Gauthier et al. 2015; St Clair & Howe 2007). To increase the efficiency of future seedlot deployment, AB and BC are currently moving away from geographic zones and designing climate-based seed transfer systems that incorporate assisted gene flow to mitigate adaptive lags and improve future forest productivity (O’Neill et al. 2008; Gray & Hamann 2011). This shift in genetic management policy represents one component of a larger forest adaptation strategy aimed at mitigating the negative effects of climate change in Canadian forests (Pedlar et al. 2011; Gauthier et al. 2014).  Climate-based seed transfer (CBST) systems determine the maximum climatic distance that genotypes can be transferred without exceeding an acceptable level of compromise to adaptation and timber production (Ukrainetz et al. 2011). Using climate-based methods means that seedlot deployment zones can be delineated more accurately than static geographic breeding zones, cross-boundary seed transfer limits will be removed, climate-based deployment zones will be more flexible in response to future climates, and assisted gene flow (AGF) can be efficiently implemented (O’Neill et al. 2014; Gray et al. 2016). AGF is a form of assisted migration that aims to redeploy pre-adapted genotypes to match their future climates, within the current species range, and mitigate the maladaptive effects of climate change; whereas, assisted migration implies a species range is extended (Aitken & Whitlock 2013). Changing the deployment of ~300 million seedlings per year to a CBST system, and implementing AGF is predicated on a detailed knowledge of current local adaptation to climate.  1.6 Research Rationale and Objectives As climatic niches shift, the primary threat to the future productivity of reforested stands comes from warming temperatures during the growing season and the effects this has on moisture availability, physiological stress, growth, and new biotic disturbances. Temperate and boreal conifers contain large amounts of phenotypic and genotypic variation, both among and within populations, that is adaptively 10  distributed in relation to climate. AGF within CBST policy is being implemented to pro-actively redistribute this variation and mitigate the effects of maladaptation on forest productivity. Under current climates, an implicit assumption of AGF is that when deployed, seedlings from warmer provenances with longer growing seasons will be shifted towards the colder margins of their adaptive climatic niche. Here is it possible that redeployed seedlings will be phenologically unsynchronised with shorter, cooler growing seasons that will become longer and warmer in the future. This creates an adaptive trade-off between early rotation risks from cold injury, and mid- to late- rotation growth gains under suitable future climates.   Selectively bred seedlings have the potential for adaptive trade-offs among traits under current climates, that would increase their vulnerability to cold injury during the transition between summer growth and winter dormancy. In the context of AGF, this could compound the short-term phenological asynchrony of selectively bred seedlings. It may reduce the climatically safe transfer distance of selected seedlings, relative to natural stand seedlings, limiting their suitability for redeployment using AGF. Accurate AGF prescriptions are required for selectively bred seedlings because they represent an increasing proportion of future reforestation in western Canada. At present, the effects of breeding on adaptive phenotypic variation in selected seedlings relative to current climates is inadequate for the evaluation and implementation of AGF reforestation policies.  To meet the needs of provincial AGF policies for the two most planted conifer species in western Canada, lodgepole pine and interior spruce, I address two primary research objectives; 1) to describe the effects of selective breeding on phenotypic and genomic adaptation to current climates, and 2) to evaluate whether selective breeding alters the risks associated with seed transfer and assisted gene flow, relative to natural stand seedlings. In Chapter 2 I study how selective breeding modifies climatically adaptive lodgepole pine phenotypes, assess the trade-offs among traits and the effects of selection on their relationships with climate. In Chapter 3, I study interior spruce and address similar phenotypic questions to Chapter 1, but combine this with an analysis of hybrid ancestry, to determine the effect hybrid variation has on the response to selection within and among breeding populations. For Chapter 4 I use a genome-wide association study to describe the genomic response to phenotypic selection in conifer breeding programs, and assess genomic versus phenotypic relationships with climate. Lastly, in Chapter 5 I conclude by summarising my findings in relation to AGF reforestation policies, and address areas of relevant future research.   11  Chapter 2: Selective breeding of lodgepole pine increases growth and maintains climatic adaptation  2.1 Introduction Long-term lodgepole pine (Pinus contorta Dougl. ex Loud. var. latifolia Engelm.) trials have been the cornerstone of forest genecology research in western North America. Considerable local adaptation to climate in lodgepole pine has been shown to reflect large amounts of genetic variance, both within and among populations that is mediated by a topographically and climatically heterogeneous landscape.  Pinus contorta var. latifolia is the most widespread and economically important P. contorta variety. Its range extends from high-elevation (~3500 m) populations in southern Colorado (38oN) to the southwest Yukon (64oN) where it is a component of the boreal forest (600 to 800m elevation) (Critchfield 1957). Broad-scale genetic clines in phenotypic traits reflect adaptation to regional climatic patterns, while clines are steep locally along elevational gradients (Rehfeldt 1988). In common garden experiments, height growth is positively related to temperature and negatively related to latitude and elevation (Rehfeldt 1983; Rehfeldt et al. 1999; Chuine et al. 2006; Wang et al. 2006b). Spring growth initiation shows little variation among provenances in common gardens (Rehfeldt & Wykoff 1981), although Chuine et al. (2006) reported that northern provenances have slightly higher threshold temperatures for growth to avoid premature bud break. Growth cessation, terminal bud formation (bud set) and cold acclimation are initiated in response to a genetically-determined critical night length, and occur later in provenances from more southern locations and lower elevations, reflecting the  in situ risk of fall frosts (O’Reilly & Owens 1989; Rehfeldt 1988). Autumn cold hardiness shows strong among-population variation in lodgepole pine (Liepe et al. 2016), which is consistent with studies of adaptation in other temperate and boreal tree species (Alberto et al. 2013). Therefore, local adaptation to climate is conferred by genotypes that optimise trade-offs between growth and cold hardiness (Howe et al. 2003).  Clear trade-offs exist between growth, phenology and cold hardiness among natural lodgepole pine populations. To respect these adaptive trade-offs and minimise the possibility of negative genotype-by-environment interactions among traits in lodgepole pine, AB and BC use relatively conservative seed transfer distances and geographic zone-based breeding programs to maintain local adaptation. However, the correlated responses to artificial selection among stem growth and phenology or cold hardiness traits within breeding populations are less clear. Theory predicts that negative trade-offs among traits are expected within breeding populations as indirect responses to artificial selection 12  on primary traits (Falconer & MacKay 1996). In conifers, evidence for this effect is mixed. Comparisons among wild-stand provenances find adaptive trade-offs equivalent to selecting populations from warmer climates with longer growing seasons (Rehfeldt 1992a; Rehfeldt 1992b), but breeding programs may also maintain climatically adaptive growth phenology and cold hardiness development (Rehfeldt 1989; Westin et al. 2000; Hannerz & Westin 2005). Despite recent climate warming, lodgepole pine breeding zones in BC are predicted to generate a modest (~7%) increase in growth with warming of ~1.5oC until the 2030’s, but the current limits of climatic adaptation within breeding zones will be exceeded and productivity decreased below present levels by the late 2060’s (Wang et al. 2006a). At this point an adaptive lag is expected to have noticeable effects on forest productivity. Accurate redeployment of selectively bred seedlots under current climates is imperative because forest stands must also be adapted to warmer temperatures and longer growing seasons expected later in their rotation. In this case, assisted gene flow prescriptions that deploy warm-adapted provenances to cooler reforestation sites are accompanied by a risk of early rotation damage from unsynchronised growth phenology and cold hardiness development that must be accurately quantified and managed. Lodgepole pine breeding programs produce seedlots with genetic worth values at rotation age in the range of 2.5 to 10%  in AB (A. Benowicz and S. John, personal communications) and 10 to 22% in BC (Forest Genetics Council of British Columbia 2015b). Across AB and BC lodgepole pine constitutes ~45% of 250 to 325 million seedlings planted annually (Alberta Environment and Sustainable Resource Development 2011; Forest Genetics Council of British Columbia 2015a), and currently ~22% (~30 million) of planted lodgepole pine seedlings originate from breeding programs. Therefore, selectively bred lodgepole pine seedlings account for a substantial proportion of reforestation in western Canada and will be redeployed in the future using climate-based seed transfer and assisted gene flow prescriptions.   In this chapter I assess how selective breeding of lodgepole pine modifies adaptive phenotypes, their relationships with climate, and the implications for assisted gene flow. I address three research questions: 1) what are the direct effects of selection on climatically adaptive traits?; 2) how do the indirect effects of selection affect trade-offs among traits?; and 3) how do adaptive phenotype-climate associations respond to selective breeding, and what are the implications for assisted gene flow? Using operational lodgepole pine seedlots sourced from breeding programs across AB and BC, because they are the basis of current and future reforestation, this chapter describes the breadth of standing genetic variation in traits that are relevant to developing AGF polices. My approach compared 105 natural stand and 20 selectively bred lodgepole pine seedlots from Alberta and British Columbia grown in a mild, 13  coastal common garden of ~2200 seedlings, and decomposed the components of phenotypic adaptation to climate that may respond to selective breeding. This approach allowed me to evaluate the effects of selective breeding on seedling traits, and shifts in phenotype-climate associations relative to natural (wild-stand) seedlings within and among breeding zones. From a single common garden test site, the composite fitness of selectively bred and natural populations in field environments cannot be determined, but this research compliments long-term field trials by assessing the relative effects of selection on the growth, phenology and cold hardiness traits of populations from different source climates. Finally, I address whether selective breeding produces phenotypes similar to the climatypes of natural populations adapted to warmer climates that favour faster growth.    2.2 Methods 2.2.1 Experimental Sampling & Establishment Open-pollinated, selectively-bred orchard seedlots were obtained from 12 lodgepole pine breeding zones across AB and BC (Table 2.1). Where available, more than one selectively bred seedlot was included from each breeding zone, for a total of 20 seedlots. Seedlots with the highest available genetic worth from the most recent growing season available were selected. The number of parent tree clones contributing to each selectively bred seedlot ranged from 36 to 117 (Table 2.1). For each geographic breeding zone, 4 to 16 open-pollinated wild stand (natural) seedlots were obtained, for a total of 105 natural seedlots (Table 2.1, Figure 2.1). The number of selectively bred seedlots is less than natural seedlots because the latter are collected from single stands within a breeding zone, while parent trees in each breeding program originate from across the respective zone, and in a few cases also include clones of highly performing parent trees from other breeding zones.  Seed was stratified using a modified version the BC Ministry of Forests, Lands and Natural Resource Operations seed stratification protocol (Kolotelo 1994). Seeds were soaked in distilled water for 24 hours, washed briefly in 2% bleach to reduce pathogens, rinsed in distilled water, surface dried, then chilled at 4oC for 3 weeks. Stratified seeds were sown in early May 2012 into two adjacent outdoor raised beds filled with double-screened topsoil on the UBC campus in Vancouver, Canada. The common garden was split into 12 blocks, with seeds sown in a randomised incomplete block design developed using a custom R script (R Core Team 2016). Each block contained 240 seedlings established as single-seedling plots, surrounded by a row of buffer seedlings. Seeds were initially triple-sown into planting 14  positions at 8 cm spacing, then systematically thinned post-germination by position to leave one healthy seedling. A total of 2176 seedlings (natural n = 976; selected n = 1200) were established. For each breeding zone there was a minimum of 60 and maximum of 112 natural seedlings (Table 2.1), with 12 to 16 seedlings from each of six randomly selected natural seedlots, and at least four seedlings from each remaining natural seedlot. Each selectively bred seedlot was represented by 60 seedlings. The experiments were maintained and measured over three growing seasons. They were well watered and received two or three fertiliser applications per growing season (Peters Excel 15-5-15 NPK water soluble fertiliser applied at a manufacturer recommended N concentration of 200ppm). Some damage to seedlings in the common gardens was caused by an unidentified fungal infection of unlignified new shoots that was treated with an appropriate systemic fungicide, or by shoot boring larvae which were squashed mercilessly between the thumb and forefinger. Seedlings were excluded from analyses (n ≤ 115) if they incurred damage that compromised their data for a given trait.  2.2.2 Phenotypic and Climatic Data During the second and third growing seasons (2013 and 2014 respectively), phenotypic data were collected for six growth, phenology and cold hardiness traits that often show local adaptation to climate in conifers (Savolainen et al. 2007). Height (cm) was measured repeatedly during season 3 and final height measurements were made after growth cessation and bud set. Growth rate (cm day-1) was interpolated from growth curves (section 2.4) fitted to the height growth time series data. Shoots were destructively sampled after the third growing season, and shoot dry mass (g) above the root collar was measured after drying samples at 70oC for a minimum of 48 hours. Growth initiation and cessation in pines are not discrete processes, and these two traits were also interpolated from seedling growth curves (Section 2.4). Autumn cold hardiness testing was performed on needles formed during the preceding summer’s growth. A slightly modified version of the artificial freeze testing  and electrolyte leakage measurement protocol described by Hannerz et al. (1999) was used to estimate damage using the ratio of cellular electrolytes leaked after freezing relative to total electrolyte leakage after heat killing. Cold hardiness testing was performed over three consecutive weeks to accommodate the large number of seedlings. Three samples of five, 5 mm long needle segments were collected from each seedling; two samples were subjected to different freeze test temperatures for a one-hour period, and the third sample served as an unfrozen control. Timing and test temperatures were determined by pre-testing to 15  identify the temperature at which approximately 50% cold injury occurred. Cold hardiness testing commenced on October 14th 2013 (season 2) using -14oC and -18oC test temperatures. Control samples were placed in a fridge at 4oC and test samples were frozen using a Tenney T20C-3 programmable temperature chamber. Electrical conductivity measurements were made on test and control samples after freezing, and again after heat killing at 95oC in a laboratory oven, using Amber Science Inc. Model 2052 Digital Electrical Conductivity meters. The cold injury damage incurred by each seedling at both test temperatures was calculated relative to unfrozen control samples using Flint et al.'s (1967) index of cold injury (I). Lastly, the values of I between test temperatures were averaged, and this mean value was used for analysis. Seedlings with I values of zero were undamaged, while values of 100 indicate maximum freezing damage. I analysed relationships between phenotypic traits and 19 climatic and three geographic variables (collectively referred to as climatic variables) (Table A.1). Climatic variables for natural seedlot provenances were estimated for the 1961 to 1990 climate normal period using ClimateNA version 5.21  available from http://cfcg.forestry.ubc.ca/projects/climate-data/climatebcwna/#ClimateNA, based on the methodology of Wang et al. (2016). This climate normal period is appropriate because it more closely reflects the historical conditions populations are likely to be locally adapted to, preceding climate warming of the last ~25 years.   Selectively bred seedlots used here are the product of open-pollination in seed orchards and bulking of seeds from multiple parent tree clones. To obtain representative climate estimates for selected seedlots, the latitude, longitude and elevation of all parent trees in each seed orchard were obtained, and their climatic variables estimated using ClimateNA. Mean climatic variables for selectively bred seedlots were averages of their respective parent tree climate data, weighted by the maternal contribution of each parent to the seedlot. Maternal contributions to a given selected seedlot are determined from the number of cones collected from each seed orchard clone. Data on paternal contributions of each clone are not consistently available among breeding programs, and were not used to weight the climatic estimates.  Each natural or selected seedling was assigned the average climatic data of its respective seedlot. Climate variables were then summarised as PCA scores for each seedling and these were used as additional climatic variables. Lastly, for both seedling types, breeding zone-specific estimates of every climatic variable, including PC scores, were calculated as the mean of all natural or selectively bred seedlings within a given breeding zone.    16  2.2.3 Data Analyses Pines have compound long-shoot buds that elongate rapidly early in the growing season well before needle fascicles rupture their bud scales (Owens 2006). As a result, bud break and bud set phenology is more difficult to phenotype directly in lodgepole pine than in many other conifers. Instead, I derived pine growth initiation and cessation phenotypes from growth curve analyses (e.g., Chuine et al. 2001).  Seedling height was measured 19 times in the third growing season to characterize rapid early season growth and phenology accurately. Individual seedling height growth time series data were fitted to the sigmoid four-parameter logistic regression model of Chuine et al. (2001) (Equation 2.1) using the nls function of the ‘stats’ package in R. [2.1] 𝐻(𝑡) =  𝑎 +  𝑏1 + 𝑒−𝑐 (𝑡−𝑑)  Where 𝐻(𝑡) = predicted height on day t, the time in days since January 1st, 𝑎 is the previous growing season’s final seedling height, 𝑏 is the current season’s height growth increment, 𝑐 is a component of the maximum growth rate, and 𝑑 is the day since January 1st that half of the current season’s growth increment was attained. Growth initiation and growth cessation timing in each pine seedling were estimated as the day that 5% and 95% of the growing season’s height increment was completed. These 5% and 95% values were chosen as a trade-off between the sensitivity to detect growth initiation or cessation and the possibility of height measurement error. Maximum growth rate was estimated as the tangent of the sigmoidal growth curve at its inflection point. The effects of selective breeding on each trait were tested using a linear mixed effects model (Equation 2.2) that accommodates the unbalanced experimental design, implemented in ASReml-R version 3.0 (Butler 2009).  [2.2] 𝑌𝑖𝑗𝑘𝑙𝑚 =  𝜇 +  𝑆𝑗 + 𝑍𝑘 + (𝑆 ∗ 𝑍)𝑗𝑘 + 𝐵𝑙 + 𝐿(𝐵)𝑙𝑚 + 𝑒𝑖𝑗𝑘𝑙𝑚  Where 𝑌𝑖𝑗𝑘𝑙𝑚 is the phenotypic observation of a trait made on individual i from the jth seedling type (S) and kth breeding zone (Z), grown in the lth block (B), at the mth seedling location (L) nested within block (𝐿(𝐵)𝑙𝑚  ). 𝑆 ∗ 𝑍 denotes the seedling type by breeding zone interaction. 𝜇 is the experimental mean and 𝑒 is the residual error of individual 𝑖. Seedling type (natural stand or selectively bred) and breeding zone were fixed effects in the model; block and location within block were random effects. 17  Residual values from the linear mixed-models of seedling traits were assessed using Shapiro-Wilk normality tests and F-test for homogeneity of variances. All traits except shoot dry mass met normal distribution and homogeneity of variance assumptions for large sample sizes; shoot dry mass data was quarter-root transformed to meet these assumptions. Best linear unbiased estimates (BLUEs) of the fixed effects were extracted using ASReml-R for each seedling type by breeding zone combination as the means of seedlings pooled across seedlots within breeding zones. These means were used to test the effects of selection within specific breeding zones, and for clinal analyses described in section 2.7. The significance of pairwise differences between seedling type BLUEs within breeding zones was tested using two-sample t-tests. My experimental comparisons are designed to use the same bulk seedlots that are used in operational reforestation in western Canada. The lack of family-level population structure prevents the estimation of additive genetic variance within and between seedlots. Instead I estimated VPOP, the ratio of among-breeding zone variance to total phenotypic variance within breeding zones, to estimate how phenotypic variation is partitioned across breeding zones for natural and selectively bred reforestation populations. VPOP is analogous to QST, a quantitative genetic estimate of additive genetic variance among versus within populations (Alberto et al. 2013).   A modified version of Equation 2.2 with seedling type excluded and all factors set as random was used to estimate the among and within breeding zone variance components, whereby each breeding zone is represented by individual phenotypes pooled from across its respective seedlots. Models for each seedling type were run separately and breeding zone differentiation (VPOP) was calculated using Equation 2.3.  [2.3] 𝑉𝑃𝑂𝑃 =  𝜎𝑝2𝜎𝑝2 + 𝜎𝑒2  Where VPOP is the phenotypic differentiation among breeding zones, 𝜎𝑝2 is the variance among breeding zones (populations), and 𝜎𝑒2 is the model’s residual error approximating the variance within breeding zones. The multiple linear regression model (Equation 2.4) was used to compare phenotypic clines along gradients for nine climatic variables (indicated in Table.A1) as well as PC1 and PC2 climate variable scores, between natural and selectively bred seedlings. Clines were estimated using mixed-model BLUEs of each breeding zone and seedling type combination as the dependent variable, and mean climatic 18  values of breeding zones as the independent variable. Seedling type was included in the model as a categorical variable. [2.4] 𝑦𝑖𝑗 = 𝛽0 + 𝛽1(𝑥1) + 𝛽2(𝑥2) + 𝛽3(𝑥1𝑥2) + 𝑒𝑖𝑗   Where 𝑦𝑖𝑗  is the BLUE of seedling type i in breeding zone j, 𝑥1  is a continuous climatic variable, 𝑥2 is the categorical covariate ‘seedling type’, 𝛽0 is the intercept, 𝛽1 and 𝛽2 are the climatic variable and seedling type coefficients respectively, 𝛽3 is the coefficient of the seedling type x climatic variable interaction, and 𝑒𝑖𝑗  is the residual error of  𝑦𝑖𝑗. The fit and significance of clines for each seedling were tested independently, and also for significant differences between slopes of seedling type clines. To identify differences in trade-offs among adaptive traits in the selected versus natural seedlings, seedling type-specific correlations between the mixed-model BLUEs for seedling height and the other five traits were calculated. For each trait I also calculated the difference between natural and selected seedling BLUEs of every breeding zone. Difference values for height were then correlated with difference values of each remaining trait, to identify differences between selected and natural seedlings which co-varied between traits. It is possible that climatic biases exist within breeding programs. For each breeding zone I calculated the temperature difference between mean source MAT of natural seedlings and the weighted mean MAT of selected seedlings. Similarly, I also calculated height gains from natural to selected seedlots in each zone, and then tested how much variation in height gains among breeding zones was explained by within-zone MAT differences. Lodgepole pine embryo development occurs between June and late August (Owens 2006). To test for potential epigenetic effects of seed orchard environments during seed development on seedling traits, the differences were calculated between mean summer temperature (MST) (June to August) of breeding zones and their respective seed orchard locations in the years the seedlots were produced. Height gains of each zone were then regressed upon these MST differences.   2.3 Results 2.3.1 Breeding Zone-by-Seedling Type Means After germination failure, subsequent mortality, and damage were accounted for, 89% of seedlings remained for analysis after three growing seasons. Individual seedling growth curves were 19  successfully modelled for the height time series data of all but two seedlings. The average height growth curves for both seedling types (analysed separately) had an R2 value of 0.98. BLUEs of seedling height were greater, in some cases up to ~50% greater, for selected seedlings than for wild stand seedlings in all but one breeding zone (Figure 2.2a, Table A.4). The differences between seedling types, reflecting genetic gain from selective breeding for faster growth, were significant in 10 of 12 pine breeding zones in pairwise t-tests. Growth rate and transformed shoot dry mass exhibited very similar results to seedling height (Figures 2.2b and c, Table A.4), and gains in growth traits were consistently greater in BC breeding zones that have older, more advanced breeding programs than in AB. Unlike growth traits, selective breeding had only minor effects on growth initiation timing. The day of growth initiation varied by no more than 1.4 days between seedling types in any breeding zone, and varied by only 3.5 days across all breeding zones x seedling type combinations (Figure 2.2d, Table A.4). Among breeding zones, the direction of change was inconsistent. Differences between seedling types were statistically significant only in BV low and CP low, but were in opposite directions in these two zones. In contrast to growth initiation, differences between seedling types for growth cessation timing were relatively consistent across breeding zones. On average across breeding zones, growth cessation occurred 4 days later in selected seedlings. Growth cessation was delayed in 11 of 12 breeding zones, and significantly so in nine of these cases (Figure 2.2e, Table A.4), with a maximum growth cessation delay of 10 days in the TO low breeding zone.  On average across breeding zones, selectively bred seedlings had 2.5% greater cold injury. Selectively bred seedlings exhibited slightly greater cold injury in 9 of 12 breeding zones (Figure 2.2f, Table A.4), but the only statistically significant difference between seedling types was 11% in the CP low breeding zone. All differences between seedling types in the remaining breeding zones were <7%.  2.3.2 Phenotypic Differentiation among Breeding Zones (VPOP) Differentiation among breeding zones varied substantially among traits, but was always stronger in selected than in natural seedlings (Table 2.2). VPOP for growth rate was 2.4x greater in selected than in natural seedlings (VPOP = 0.124 and 0.052 respectively), while seedling height VPOP was ~2x and shoot mass 1.5x greater. VPOP was slightly greater in selected versus natural seedlings for growth initiation (selected VPOP = 0.086, natural VPOP = 0.052), and 2x greater for growth cessation (selected VPOP = 0.345, natural VPOP = 0.178). Cold injury showed the strongest breeding zone differentiation, with VPOP values 20  of 0.545 and 0.415 in selected and natural seedlings respectively, but the smallest difference between seedlot types of any trait.   2.3.3 Clinal Analysis PCA of seedling climate variables found PCs 1, 2, 3 & 4 cumulatively accounted for 54, 71, 85 and 93% of climatic variation (Table A.2). PC1 loadings were dominated by temperature and frost-related variables, while PC2 loadings were dominated by precipitation as well as derived variables combining temperature and precipitation (Table A.3).  Most traits showed significant clinal variation along gradients of latitude, mean annual temperature, summer heat-moisture index, degree days above 5oC, number of frost-free days, extreme minimum temperature and climate PC1, after Bonferroni adjustment for multiple climatic comparisons (Table A.5). Clines in longitude, elevation, log mean annual precipitation and climate PC2 were moderate to weak for all trait-by-seedling type combinations. Here clines in mean annual temperature (MAT) are used to illustrate differences between selected and natural stand seedlings. Second only to climatic PC1 scores, clines in MAT are consistently the strongest and show the greatest differences between natural and selected seedlings. Consistent with estimates of VPOP, patterns of clinal variation for the three growth traits were markedly stronger for selected than for natural seedlings (Figure 2.3a, b and c). However, after correcting for multiple comparisons (α = 0.0045), slope differences between seedling type clines were not statistically significant for any growth trait. Growth initiation had a significant cline in MAT (Figure 2.3d) only for selected seedlings, while growth cessation clines with MAT (Figure 2.3e) are strong but not significantly different between seedling types. Consistent with the high VPOP values, clines in cold injury for MAT are the strongest of any trait and highly significant in both seedling types, but differences between seedlot type slopes are small and not significant (Figure 2.3f).   2.3.4 Trait-Trait Correlations Correlations among traits were uniformly strong (r ≥ 0. 74) and statistically significant (p ≤ α = 0.0033) in all but one case: height versus growth initiation (Table 2.3). Correlations with height were also slightly stronger for selected seedlings in all traits. Corresponding with my clinal analyses, seedling height was positively associated with growth initiation and growth cessation, indicating that taller seedlings broke and set bud later (although the actual magnitude of this delay is greater for growth cessation, described in section 3.1). 21  Difference correlations of height versus growth rate (Figure 2.4a), and shoot dry mass were moderate to strong and statistically significant (p ≤ 0.01). Differences in height were not strongly correlated with differences in growth initiation (Figure 2.4b), but height-growth cessation differences were strongly and significantly correlated (r = 0.94) (Figure 2.4c). Lastly, height differences between seedling types were neither strongly nor significantly correlated with cold injury differences (Figure 2.4d).  2.3.5 Climatic Biases in Breeding Programs Mean source MATs of natural seedlings differ from weighted mean MATs of selected seedlings within breeding zones by -0.09 oC to +1.6oC, but this explained little variation in height gains among breeding zones (r2 = ~0.12; p = 0.27). Likewise, differences in mean summer temperature (MST) (June to August) between breeding zones and seed orchards had a range of 0.81 to 5.95oC, but the variation in height gains these differences explain is negligible (r2 ≤ 0.14, p = 0.11). Equivalent regressions for the remaining five traits were also weak; cold injury had the strongest relationship (r2=0.22; p = 0.035).  2.4 Discussion Lodgepole pine breeding programs in western Canada have achieved substantial growth gains over wild-stand seedlings that are reflected in this common garden study. Breeding zone-level analyses showed that in addition to growing up to 50% taller, selectively bred seedlings expressed greater phenotypic differentiation among breeding zones and stronger climatic clines for all six traits. The increased height of selected seedlings was underlain by stronger correlations across breeding zones with faster growth rate and slightly later growth cessation. Crucially, autumn cold hardiness remained similar between natural and selected seedlings and was not significantly correlated with increased height growth resulting from selection. This indicates that selection and breeding for greater growth within local populations has not compromised the cold hardiness of seedlings used in operational reforestation compared to natural seedlings.  2.4.1 Effects of Selection on Adaptive Traits Seedling growth trait differences between natural and selected seedlings from the same geographic areas in this study reflect height gains achieved by selective breeding programs. Broadly, these gains are greater in BC where breeding programs have applied a greater selection intensity, and have undergone an additional breeding cycle relative to AB. Growth traits have relatively low phenotypic differentiation 22  among breeding zones for both seedling types. Estimates of  VPOP (Table 2.2) fall into the bottom third of QST  ranges for height and growth traits summarised by Savolainen et al. (2007) and Alberto et al. (2013), but they are congruent with those of Liepe et al. (2016) who estimated VPOP from >250 natural populations, including all those in this study, in multiple growth chambers rather than outdoor common garden experiments. Selected seedlings have greater VPOP values for growth traits and steeper slopes of clines with MAT (Figure 2.3a, b and c), albeit not significantly. This indicates that population differentiation and climatic associations are stronger in selected seedlings, which corresponds to similar findings from selectively bred progeny of other conifers in BC (O’Neill et al. 2014). Natural seedlings showed modest relationships between MAT and growth initiation timing among breeding zones (Rehfeldt & Wykoff 1981; Chuine et al. 2006). Selected seedlings had slightly stronger relationships between MAT and growth initiation, but across all breeding zones and both seedling types, timing of mean breeding zone growth initiation spanned only 3.5 days (Figures 2.2d and 2.3d). This suggests, consistently with a number of other temperate-boreal conifers, that timing of bud break and growth initiation is under strong genetic control in response to heat sum accumulation but varies little among populations (Bigras et al. 2001; Cooke et al. 2012). In contrast, growth cessation had the second largest breeding zone differentiation (VPOP) out of all traits, as expressed by VPOP values, and had the greatest VPOP difference between seedling types (Table 2.2). Clines in temperature-related climatic variables for growth cessation (Figure 2.3e and Table A.5) were also stronger for selected seedlings. In natural populations, the growth cessation results are consistent with bud set VPOP values of Liepe et al. (2016), and previously reported clines for lodgepole pine in temperature-related climatic variables (Rehfeldt 1988; O’Reilly & Owens 1989). However, for growth cessation in both seedling types there were only weak to moderate clines associated with latitude, and therefore photoperiod, as the primary cue for growth cessation in woody plants (Bigras et al. 2001; Petterle et al. 2013). In line with other conifers from this region (Bansal et al. 2015; Rehfeldt 1983; Hannerz et al. 1999), cold hardiness showed the strongest association of any trait to local climate (Figure 2.3f, Table A.5),  reflecting the importance of cold hardiness as an adaptive trait (Howe et al. 2003).  Both VPOP and clines were slightly stronger for selected seedlings, but the differences are relatively small compared to the effects of selection on growth traits and growth cessation. For cold hardiness, there were no clines with elevation identified by other studies of natural populations, because seedlings in this study are selected from within breeding zone elevational limits (Table 2.1) that are narrow relative to the species range in elevation. Therefore, clines in cold hardiness and the remaining traits are more likely to reflect 23  broad regional gradients, principally in latitude, rather than steep local gradients mediated by topography.  2.4.2 Correlated Responses to Selection Strong correlations between average population seedling height and the remaining five traits conform to expectations of how traits co-vary across natural environments in lodgepole pine and other temperate conifers (Rehfeldt & Wykoff 1981; Rehfeldt 1988; Howe et al. 2003; Savolainen et al. 2007). Differences in the strength of correlations between seedling types are small (Table 2.3), suggesting that negative trade-offs between height gains and phenology or cold hardiness within populations are weak or non-existent. As a result, selection within populations for growth has not compromised local adaptation to low temperatures. Although changes in the strength of trait-trait correlations resulting from selective breeding are small, among-population height differences between seedling types correlate strongly with differences in both growth rate (r = 0.98, p < 0.0001) (Figure 2.4a) and growth cessation timing (r = 0.94, p < 0.0001) (Figure 2.4c). Therefore, selected seedlings attained greater height by growing both faster and longer. The respective contributions of these effects cannot be determined, although our results support those of Chuine et al. (2001) who found more of the among-provenance height differences are associated with growth rate than with phenology in lodgepole pine. Correlations of the differences between seedling types for height vs growth initiation and height vs cold injury were weak and not significant (Figure 2.4b and d). This implies that height gains achieved by extending the growing season are mostly derived from delayed growth cessation, rather than earlier growth initiation, while negative trade-offs in phenology and cold hardiness attributable to breeding for increased height are weak, conforming to the predictions of Rehfeldt (1989).  Cold hardening and shoot dormancy follow growth cessation and bud set in what typically is thought of as a sequential process with two or three steps. Growth cessation is triggered when night length exceeds a genotype’s critical value (Petterle et al. 2013). This stimulates bud formation and initial cold hardening (Cooke et al. 2012). Critical night length is a cue for climatically adaptive bud set and autumn cold acclimation that reflects the local timing and severity of low temperatures which are the selective agents. Relationships between growth cessation and latitude (a proxy for photoperiod) are relatively weak across the 9.5 degrees of latitude sampled in this study (natural r2 = 0.39, selected r2 = 0.35) (Table A.5). Growth cessation has much stronger relationships with extreme minimum temperature (EMT) (natural r2 = 0.62, selected r2 = 0.85) (Table A.5). Similarly, cold hardiness has strong 24  relationships with latitude, MAT and EMT that differ little between natural and selected seedlings, and the correlation between growth cessation and cold hardiness is almost identical between seedling types (Δr = 0.01) (Table 2.3). Therefore, selectively bred seedlings have greater growth and similar cold hardiness to natural seedlings, despite slightly delayed growth cessation, and breeding does not appear to compromise adaptive relationships with low temperatures.  The finding that selective breeding programs increase the strength of climatic clines across breeding zones in growth traits might initially seem like a paradox; gains from selective breeding in growth might be expected to result from trade-offs with phenology or cold hardiness traits. However, this is not the case. Breeding programs sampled here test progeny over three or more typical field sites within a breeding zone to estimate the genetic value of parent trees and make selections. To perform successfully, families must be well adapted to climatic conditions on all of the test sites. Thus, selection and breeding within populations produces faster-growing phenotypes, and progeny testing in natural environments within breeding zones constrains phenotypes within the limits of local climates. Selected genotypes grow taller by growing faster and for slightly longer, while refining trade-offs between growth, growth cessation and dormancy to maintain cold hardiness and climate adaptation.   2.4.3 Mechanisms of Growth Responses to Selective Breeding The increased height growth and stronger height-climate associations in selected seedlings could be achieved by several indirect or direct mechanisms, including the following. 1) Selection may favour genotypes equivalent to those from warmer adapted populations. 2) Parent trees in breeding programs may be selected from parts of the breeding zone where climates are most favourable to growth. 3) Pollen contamination may originate from warmer-adapted populations local to seed orchards. 4) Seed size and conditions during seed development could result in epigenetic effects on seedling phenotypes. 5) Selected genotypes may have more phenotypic plasticity than unselected genotypes. Here I address each of these possibilities.   Artificial selection has not resulted in growth cessation and cold hardiness phenotypes equivalent to those of faster growing but less cold-adapted natural populations (Rehfeldt 1992a; Rehfeldt 1992b). On a trait-by-trait basis, mean phenotypes may be equivalent to those of other breeding zones. However, considering all traits in combination and the correlations among them, it appears that provenance testing, selection, breeding and subsequent progeny testing have not reproduced pre-existing phenotypes similar to those from warmer provenances, but produced phenotypes well suited to their progeny test environments. 25   Microgeographic phenotypic and genetic variation contributes within-stand genetic structure of forest trees (reviewed in Scotti et al. (2016)). Genetic gain could result from simply selecting productive parent trees from warmer sites that are more favourable for growth within topographically and climatically heterogeneous breeding zones. This hypothesis is not supported by my results, because differences between mean source MATs of natural seedlings and weighted mean MATs of selected seedlings explain a negligible proportion of the variation in observed height gains.   Seed orchards in AB and BC are typically located on warmer, drier sites that are optimal for intensive seed production, but may be geographically distant from their breeding zone of origin. Using neutral molecular markers, pollen contamination in two of the interior BC lodgepole pine seed orchards that I obtained seed from was previously estimated to be 8% and 14.5% (Funda et al. 2014; Stoehr & Newton 2002). It is possible that pollen contamination from nearby warm-adapted populations or other seed orchards could contribute to greater height growth in selected seedlings, but this effect would be greatest in seedlings from cooler breeding zones and would reduce rather than increase the slopes of clines in height with MAT for selected seedlings.  Conditions during seed development are also known to contribute to differences in growth among provenances. Larger seeds are expected to  confer greater seedling growth and survival (Castro 1999 and references therein). Cultural practices in lodgepole pine seed orchards favour the production of relatively large seeds that germinate and grow rapidly. Larger average seed size could contribute to observed differences between natural and selected seedlings, particularly for growth traits. However, weighting the linear mixed models (Equation 2.3) by average seedlot seed mass had no tangible effect on BLUEs of trait means, and correlations between seed weight and means of all six phenotypic traits were weak (r < 0.25, except growth initiation r =-0.39, p = 0.06).  Temperature effects on seed development have also been associated with epigenetic changes in some conifers (Bräutigam et al. 2013). Extensive studies on Norway spruce show that progeny from cold-adapted provenances grown using seed produced in warmer environments grow taller, cease growth later and are less cold hardy than progeny from seeds with the same genetic background collected in the respective cold adapted wild stands (Johnsen et al. 2005; Skrøppa et al. 2007). This is because temperature regulates differential gene expression during embryo development and seed maturation, leaving a long-term epigenetic memory that affects the progeny’s phenotype. Similar effects have been reported in Pinus species, although they only lasted for a few years (Dormling & Johnsen 1992; Schmidtling & Hipkins 2004). The epigenetic effects of temperature on seed maturation and selected seedling phenotypes are not directly quantifiable in this study, but the evidence here 26  suggests they are likely to be weak by the third year of growth. Differences between breeding zone and seed orchard MSTs explained little variation in height gains, or in the remaining traits. By using a single common garden site in a mild climate it was possible to make many repeated growth, phenology and cold hardiness measurements, as the basis to quantify and test differences among populations and between seedling types in the absence of stress-related seedling mortality (Campbell 1986). With this experimental design, is it not possible to rule out variation in phenotypic plasticity among breeding zones or between seedlot types. A plastic response to the mild climate may have contributed to height gains of selected seedlings, especially for seedlings from warmer breeding zones. This could contribute to the stronger relationships observed between growth and climate in selected seedlings. However, if this effect occurs it is likely to be small, because observed seedling height gains are consistent with estimates of genetic gain for growth from field progeny tests in provincial breeding programs. The primary reason for the stronger climatic associations with height and stronger phenotypic differentiation among breeding zones in selected versus natural seedlings is the greater genetic gain that has been achieved in the warmer breeding zones. Even if breeding values for height were equal among breeding zones, zones with taller seedlings would have greater absolute height gains because breeding values are estimated relative to the base population mean. Climatically favourable breeding zones have the tallest seedlings, and the highest field-based breeding value estimates for growth, while the need for adaptation to extreme low temperatures constrains genetic gains in growth in the coldest zones. Breeding program history also varies among zones, and the oldest most advanced programs have achieved the greatest gains. Similarly, greater population differentiation (VPOP) of selected seedlots reflects stronger growth climate relationships. For all traits, variance among breeding zones (𝜎𝑝2) increased with selection relative to variance within breeding zones (𝜎𝑒2). Greater population differentiation corresponds to stronger clinal variation and a narrowing of the climatic niche within breeding zones. These effects mean that the greater realised growth gains in climatically favourable breeding zones account for most of the greater phenotypic differentiation among breeding zones and stronger growth clines in selected versus natural seedlings. The strong correlations between growth and other adaptive traits in selected seedlings result in stronger clines in phenology and cold hardiness traits of selected seedlings. However, clinal responses are variable among traits because trait-trait correlations are imperfect and do not change consistently between seedling types.   27  2.5 Conclusions Assisted gene flow (AGF) is a promising, proactive strategy to mitigate the negative impacts of climate change on the health and productivity of planted temperate and boreal forests (Gauthier et al. 2014; Gray et al. 2016). Strong phenotypic clines among populations of both natural and selected seedlings support adopting AGF policies to accurately redeploy seedlots for future climates. Selective breeding of lodgepole pine should be compatible with AGF because it produces seedlings that grow vigorously and are well adapted to recent local climatic conditions. Within breeding zones, the faster growth of selected seedlots should buffer some of the short-term negative impacts of climatic change on forest productivity for as long as their growth exceeds that of natural populations. This runs contrary to concerns that selective breeding and the increased deployment of selected seedlings might negatively impact climatic adaptation and future AGF.  Artificial selection for greater height growth has strengthened associations between adaptive traits and climate in lodgepole pine. Cold hardiness has far stronger relationships with climate than any other trait, but artificial selection for greater growth has not substantially decreased cold hardiness. While selectively bred seedlings grow faster than natural populations from the same geographic areas, they are not the adaptive equivalents of natural populations from warmer climates because their tolerance of cold injury is largely maintained. By extension, differences in growth observed among natural populations in provenance trials should not be considered a proxy for the effects of selective breeding on other adaptive traits. In future, the suitability of selectively bred populations for reforestation must be assessed in relation to the complete testing, selection and breeding process, rather than just the isolated effects of selection. Growth differences among breeding zones in this seedling common garden reflect similar patterns of variation among natural populations identified from long-term lodgepole pine field trials. By decomposing the relationships between growth, phenology and cold hardiness, I found that selective breeding within zones has balanced these phenotypic components of adaptive variation. The growth potential of selected seedlots is increased under favourable conditions, yet adequate phenological synchronisation and autumn cold hardiness is retained. Through replicated long-term provenance and progeny tests in a range of field test site conditions, lodgepole pine breeding programs in AB and BC have effectively avoided negative trade-offs between growth and climatically adaptive phenology and cold hardiness traits. Climatic adaptation of selected lodgepole pine seedlings has not been compromised in terms of phenology or cold hardiness relative to natural seedlings, and on this basis different AGF prescriptions for natural stand and selectively bred seedlots are not warranted. Assisted 28  gene flow of selectively bred seedlots is a valid mechanism for increasing the productivity of lodgepole pine under future climates in western Canada.   29  2.6 Tables    Province Breeding     zone Elevation           range (m) Natural   Selected Seedlots Seedlings Seedlots Clones per Seedlot Seedlings AB A 1050 - 1350 6 72 1 36 (36) 60 AB B1 800 - 1200 5 60 2 117 (118), 117 (118) 120 AB B2 1200 - 1600 4 60 1 108 (111) 60 AB C 800 - 1200 5 60 1 113 (113) 60 AB J 600 - 1000 8 80 2 50 (60), 38 (60) 120 AB K1 1100 - 1500 8 80 1 60 (60) 60 BC BV low 700 - 1200 11 92 2 67 (71), 73 (88) 120 BC CP low 700 - 1100 7 76 3 55 (61), 72 (72), 65 (67) 180 BC EK low 800 - 1500 13 100 1 48 (49) 60 BC NE low 700 - 1400 11 92 2 38 (42), 40 (46) 120 BC PG low 600 - 1200 11 92 2 45 (64), 84 (86) 120 BC TO low 700 - 1400 16 112 2 72 (77), 48 (65) 120  Table 2.1 Breeding zones sampled for natural and selected seedlots, their elevational range, number of seedlots and seedlings per breeding zone. The number of clones contributing cones to selected seedlots is given, with the total number of clones in their respective seed orchards in brackets. Alberta’s breeding zones are formally identified as A, B1, B2, C, J and K1. British Columbia’s breeding zone abbreviations are BV = Bulkley Valley, CP = Central Plateau, EK = East Kootenay, NE = Nelson, PG = Prince George, and TO = Thompson - Okanagan. 30       Natural Selected             𝜎𝑝2         𝜎𝑒2  VPOP              𝜎𝑝2  𝜎𝑒2   VPOP   Seedling Height 27.519 (12.945) 205.89 (9.958) 0.118 (0.056) 70.004 (31.223) 248.08 (10.828) 0.220 (0.101)   Growth Rate 0.002 (0.001) 0.031 (0.001) 0.052 (0.028) 0.005 (0.002) 0.037 (0.002) 0.124 (0.058)   Shoot Mass 0.018 (0.009) 0.186 (0.009) 0.088 (0.043) 0.035 (0.016) 0.222 (0.01) 0.135 (0.063)   Growth Initiation 1.071 (0.573) 19.465 (0.944) 0.052 (0.028) 1.403 (0.676) 14.904 (0.652) 0.086 (0.042)   Growth Cessation 14.068 (6.393) 65.012 (3.172) 0.178 (0.082) 30.116 (13.175) 57.131 (2.501) 0.345 (0.16)   Cold Injury 169.05 (73.381) 238.53 (11.218) 0.415 (0.195) 234.62 (100.99) 196.24 (8.254) 0.545 (0.267)  Table 2.2 Proportions of phenotypic variance among (𝜎𝑝2) and within (𝜎𝑒2) breeding zones  (i.e. populations), and population differentiation among breeding zones (VPOP) for the six phenotypic traits of both natural and selectively bred seedling types. Standard errors of all estimates are given in brackets. 31    Comparison Seedling Type r p-value Height - Growth Rate Natural 0.92 < 0.0001  Selected 0.96 < 0.0001 Height - Shoot Mass Natural 0.94 < 0.0001  Selected 0.99 < 0.0001 Height - Growth Initiation Natural 0.52    0.0827  Selected 0.80    0.0019 Height - Growth Cessation Natural 0.95 < 0.0001  Selected 0.97 < 0.0001 Height - Cold Injury Natural 0.83    0.0009  Selected 0.86    0.0004 Growth Rate - Shoot Mass Natural 0.86    0.0003  Selected 0.96 < 0.0001 Growth Rate - Growth Initiation Natural 0.49    0.1021  Selected 0.85    0.0005 Growth Rate - Growth Cessation Natural 0.87    0.0002  Selected 0.91 < 0.0001 Growth Rate - Cold Injury Natural 0.66    0.0191  Selected 0.81    0.0014 Shoot Mass - Growth Initiation Natural 0.33    0.2985  Selected 0.82    0.0012 Shoot Mass - Growth Cessation Natural 0.85    0.0005  Selected 0.96 < 0.0001 Shoot Mass - Cold Injury Natural 0.78    0.003  Selected 0.87    0.0002 Growth Initiation - Growth Cessation Natural 0.71    0.0092  Selected 0.84    0.0007 Growth Initiation - Cold Injury Natural 0.67    0.0177  Selected 0.81    0.0016 Growth Cessation - Cold Injury Natural 0.85    0.0005   Selected 0.84    0.0007  Table 2.3 Pairwise correlation coefficients between all six traits for natural and selected seedlings. Correlations are calculated using breeding zone means of each trait. p-values are statistically significant at an adjusted α = 0.0033 cut-off value for 15 correlations per seedling type.  32  2.7 Figures    Figure 2.1 Geographic origins of the natural and selected seedling populations sampled from across the range of lodgepole pine in AB and BC. Natural populations are represented by filled circles, selected seedling breeding zones are represented by filled polygons.  33    Figure 2.2 Bar plots of breeding zone level phenotypic means (BLUEs) including standard error bars, for seedling a) height; b) growth rate; c) shoot dry mass (quarter-root transformed); d) growth initiation; e) growth cessation; and f) cold injury. 34    Figure 2.3 Phenotypic clines with MAT for seedling a) height; b) growth rate; c) shoot dry mass (quarter-root transformed); d) growth initiation; e) growth cessation; and f) cold injury. Points represent trait means displayed in Figure 2 and source MAT means for each of 12 breeding zones as the dependent and independent variables respectively. Cline r2 values are significant at an α = 0.0045 cut-off value after correction for multiple comparisons across 11 climate variables per seedling type. 35    Figure 2.4 Correlations of differences in the values of height between natural and selected seedlings of each breeding zone with equivalent differences in phenology and cold injury traits. Difference correlations are plotted for height versus a) growth rate; b) growth initiation day; c) growth cessation day; and d) cold injury. Correlations are significant at an α = 0.0125 cut-off value after Bonferroni correction for four comparisons.  36  Chapter 3: Growth gains from selective breeding in a spruce hybrid zone do not compromise local adaptation to climate  3.1 Introduction Natural hybrid zones contain large amounts of standing genetic variation that provides the raw material for  rapid selection responses and transgressive adaptive phenotypes, even if the selection pressure is relatively weak (Rieseberg et al. 1999; Barrett & Schluter 2008). Where gene flow from progenitor populations is ongoing, admixture facilitates adaptive introgression among hybrids and parents (Buerkle & Lexer 2008). Admixture can also be sufficient to generate divergent phenotypes that allow rapid ecological transitions into novel environments (Rieseberg et al. 2003). From a genetic conservation and forest management perspective, the ability for hybrids to adapt rapidly to changing environmental conditions could be useful for managing and promoting adaptation in populations that are threatened by anthropogenic environmental change (Hamilton & Miller 2016; Aitken et al. 2008).  Hybridization is a source of genetic variation and adaptive introgression that is being increasingly documented in natural populations of forest trees (Hamilton et al. 2015; Cullingham et al. 2012; De Carvalho et al. 2010; Suarez-Gonzalez et al. 2016). Many forest trees, particularly outcrossing wind-pollinated species, already harbour large amounts of additive genetic variation that facilitate adaptive responses to novel environments (Kremer et al. 2012). It is a reasonable expectation that where species hybridize, adaptive responses to selection will be enhanced further, in the absence of hybrid incompatibility or inferiority. Species within some of the world’s most important commercial timber genera readily hybridize to produce fit, fertile offspring, e.g. Eucalyptus, Pinus, Picea and Populus, and tree breeders take advantage of this. Artificially crossing parental taxa generates heterosis in some cases, and a greater range of additive genetic variation for selection and accelerated genetic gain, but F1 progeny from artificial hybrid breeding programs are usually limited to high-gain plantation forestry systems. Alternatively, breeding programs may select on genetic variation in populations within natural hybrid zones that are adapted to local environmental conditions and are suitable for reforestation objectives that include maintaining long-term forest productivity, health and ecosystem function. In the western Canadian ‘interior spruce’ hybrid zone, these objectives of increased timber yields from selective breeding and natural forest management coincide. ‘Interior spruce’ is used to describe hybrid populations composed of white spruce (Picea glauca (Monech) Voss) and Engelmann spruce (P. engelmannii Parry ex Engelm.) ancestry, although some 37  western populations also contain small amounts of ancestry from sub-maritime Sitka spruce (P. sitchensis (Bong.) Carr.) (Roche 1969; Hamilton et al. 2015). British Columbia (BC) was the convergence point of these three species’ ranges following post-glacial recolonization (De La Torre et al. 2014b; Daubenmire 1974), but low levels of P. glauca ancestry are also found in P. engelmannii dominated hybrids that extend south into the USA (Haselhorst & Buerkle 2013). In Alberta (AB), interior spruce is largely represented by pure P. glauca genotypes, except in the Rocky Mountain foothills where low levels of P. engelmannii ancestry are present. P. engelmannii is a mid to high-elevation species, but is more or less continuously distributed throughout sub-alpine forests in the southern half of BC between 1000 and 2200m elevation (Alexander 1987; Coates et al. 1994). P. glauca has a continuous transcontinental, boreal and sub-boreal distribution at low to mid elevations (< 1500m) from eastern Canada to BC and western Alaska (Nienstaedt & Zasada 1990).   P. glauca x P. engelmannii hybrids have formed from extensive multigenerational crossing and backcrossing resulting in asymmetric introgression that favours P. engelmannii in a contact zone between these two species that pre-dates the last glacial maximum (Haselhorst & Buerkle 2013; De La Torre et al. 2014b). Adaptive ecological divergence preserves P. glauca and P. engelmannii as distinct species, but the interior spruce hybrid zone is maintained by greater fitness of hybrid individuals in intermediate local environments (De La Torre et al. 2014a). P. engelmannii is adapted to long montane winters with high snowfall, and short, cool, humid summers (Alexander 1987). By contrast, P. glauca is adapted to extreme boreal and sub-boreal winter cold, warm summers, and low annual precipitation (Nienstaedt & Zasada 1990). Proportions of parental ancestry in hybrids are driven by adaptation to summer aridity and winter precipitation as snow. Hybrid fitness in intermediate habitats follows regional gradients in latitude and temperature, as well as local climatic gradients associated with elevation (De La Torre et al. 2014a). Across AB and BC interior spruce populations, potential height growth is greatest in populations from the valleys of southern BC when grown in local common gardens, and decreases as tolerance to extreme cold increases with latitude. Growth also decreases locally as tolerance of deep snow pack and short growing seasons increases with elevation (Liepe et al. 2016; De La Torre et al. 2014a; Gray et al. 2016; Rehfeldt 1994).   Effective reforestation using interior spruce progeny requires planting genotypes with hybrid ancestry composition that matches fine-scale climatic variation in topographically heterogeneous landscapes. This has been managed using breeding and seedlot deployment zones to maintain adaptive variation that reflect both regional and local elevational gradients in ancestry, while increasing the genetic worth for timber volume of reforestation seedlings.  At present, interior spruce seed orchards 38  produce open-pollinated seedlots with genetic worth values for growth of 10 to 30% in BC (Forest Genetics Council of British Columbia 2015b) and ~2.5% in AB (A. Benowicz and S. John, personal communications).  Where genetic gains for growth are high, undesirable correlated responses in adaptive traits such as drought tolerance, seasonally synchronized growth phenology and cold hardiness may be adaptively important, even if correlations between these traits and growth are relatively weak (Pita et al. 2005; Howe et al. 2003). In Chapter 2 I found that trade-offs among adaptive traits in lodgepole pine were negligible and climatic adaptation was maintained in selectively bred seedlings, but in interior spruce expectations for correlated responses to selection among traits are also confounded by hybrid ancestry. Among breeding zones hybrid ancestry will vary substantially, and within breeding zones hybrid variation  has the potential to affect selection responses,  even though levels of variation in height growth appear to be maintained in breeding populations (Stoehr et al. 2005). In southern BC, open-pollinated wild-stand families with greater P. glauca ancestry had greater breeding values for height, cold hardiness and tolerance of low moisture conditions, rather than negative height gain-cold injury trade-offs (De La Torre et al. 2014a). Across BC, breeding programs generate growth gains under local climates, and although the safe climatic transfer distances of breeding populations were diminished relative to natural populations, they were still greater than permitted in current deployment zones (O’Neill et al. 2014). Annually, 85 to 90 million seedlings planted in AB and BC originate from these spruce selective breeding programs (Alberta Environment and Sustainable Resource Development 2011; Forest Genetics Council of British Columbia 2015b). This hybrid complex represents a large proportion of current and future reforestation and as climates both warm and become more variable in western Canada, interior spruce will also become subject to assisted gene flow. Implementing AGF requires a nuanced understanding of how selective breeding of interior spruce modifies adaptive traits, their trade-offs with growth within and among breeding populations, and their climatic associations across a hybrid zone that contains high levels of adaptive genetic variation. In Chapter 2 I address four primary questions. 1) How does selection modify hybrid ancestry composition within and among breeding zones? 2) What are the effects of selective breeding on adaptive phenotypic traits and the relationships among traits? 3) How does selection change the relationships between growth gains, adaptive traits and climate? 4) How do changes in the hybrid index of selected seedlots modify relationships with adaptive traits, or with climate? By answering these questions I provide results to support the integration of selectively bred interior spruce reforestation seedlings into AGF policies. My approach combines detailed phenotypic 39  data for growth, phenology and cold hardiness traits, with hybrid index estimates from ~6500 SNP markers, and climate estimates for 16 variables. Decomposing the phenotypic, climatic and hybrid ancestry components of adaptive variation allows me to describe the changes that underlie differences in ancestry within and among interior spruce breeding zones, then relate these findings to future AGF needs. I achieve this by evaluating seedlings from selectively bred seedlots relative to their natural stand counterparts using a large common garden established in benign conditions that compliments long-term replicated field trials established across AB and BC.    3.2 Methods 3.2.1 Experimental Sampling and Establishment Seedlots were sampled from 14 breeding zones across the range of P. engelmannii, P. glauca and their hybrids in AB and BC. Eighteen open-pollinated, selectively-bred orchard seedlots with the highest genetic worth for growth were obtained for the most recent year from each breeding zone (Table 3.1). 140 open-pollinated natural (wild-stand) spruce seedlots were sourced from the same 14 breeding zones (Figure 3.1). Seeds from these 158 seedlots were established in a common garden on the UBC campus in Vancouver, BC, Canada. Seeds were triple-sown at 8cm spacing into 12 randomized incomplete blocks across two adjacent raised nursery beds filled with double-screened sandy loam topsoil. Post germination, seedlings were systematically thinned to leave one healthy individual, kept well-watered, and fertilized two or three times per growing season (Peters Excel® 15-5-15 NPK water soluble fertilizer applied at a manufacturer recommended N concentration of 200ppm).  One or two selectively bred seedlots were sampled per breeding zone, and each was represented by 66 seedlings. In each breeding zone, six randomly selected natural seedlots had 12 to 16 seedlings, and any additional natural seedlots had a minimum of four seedlings. A total of 2424 (natural n = 1244, selected n = 1180) seedlings were established.  3.2.2 Climatic Data To characterize climate-phenotype and climate-hybrid index relationships, 16 climatic variables (Table 3.2) were interpolated for the 1961 to 1990 climate normal period using ClimateNA version 5.21 (available at http://cfcg.forestry.ubc.ca/projects/climate-data/climatebcwna/#ClimateNA; Wang et al. (2016)). This climate normal period better represents historical conditions that are likely to have shaped local adaptation of interior spruce populations, without much influence of recent climatic warming 40  trends. Three geographic variables, latitude, longitude, and elevation that reflect climatic gradients were also included in the analyses (Table 3.2).  Climatic data were estimated for the source population of each natural seedlot. The selectively bred seedlots sampled here are composed of bulked seed produced in open-pollinated seed orchards from clones of 19 to 118 parent trees. Climatic variables for every parent tree contributing to a selected seedlot were estimated using ClimateNA, and averages of each parent tree’s climate variables were weighted by their estimated maternal contribution to the seedlot. Maternal contributions are defined as the proportion of cones contributed to the seedlot by each seed orchard clone. This gave us the most representative climate estimate possible for each selected seedlot, because paternal contributions were inconsistently available among seed orchards.  Every seedling from the same seedlot was assigned the same provenance climatic data. Principal components analysis scores were also used to summarize each seedlot’s climatic variables and included as additional climatic variables. Finally, for each climatic variable the mean of all natural or selectively bred seedlings within a given breeding zone was calculated.   3.2.3 DNA Extraction and SNP Genotyping Newly emerged needle tissue was collected from each seedling in spring 2013 and DNA was extracted using a Macherey-Nagel Nucleospin 96 Plant II Core™ kit, automated on an Eppendorf EpMotion 5075™ liquid handling platform. These samples were genotyped using the AdapTree interior spruce Affymetrix Axiom 50K interior spruce SNP array (Yeaman et al., unpublished). A more detailed description of this array’s design is forthcoming, but briefly, it was constructed as follows. SNP discovery was conducted using the sequence capture dataset for interior spruce described in Yeaman et al. (2016) and Suren et al. (2016), which included successful probes for the exons of >23,000 genes and a large number of intergenic regions. Intergenic regions were identified by using GMAP (Wu & Watanabe 2005) to map the draft transcriptome for interior spruce (Yeaman et al. 2014) to the white spruce draft genome (Birol et al. 2013), with intergenic regions identified as any genomic contigs that did not have a good hit to a gene based on this mapping. The entire array design included 51029 SNPs, of which 5144 were designed based on SNPs from intergenic regions. Intergenic SNPs were chosen for inclusion by randomly picking one diallelic SNP per identified intergenic genomic contig, and using the Affymetrix design algorithm recommendations to prioritize SNPs with a rating of either “recommended” (preferred) or “neutral” (when the contig contained no “recommended" SNPs). The Affymetrix design algorithm attempts to minimize the inclusion of regions that are repetitive in the genome or that have SNPs in the regions 41  flanking the target SNP. Samples were genotyped using the Affymetrix array by Neogen GeneSeek (Lincoln, Nebraska).  3.2.4 Hybrid Index Analyses After genotyping 2424 individuals and filtering out SNPs that had GeneSeek quality values > 0.15 (0 = high quality, 1 = low quality), and call rates < 0.85, 6482 SNPs called in 2072 genotypes remained for hybrid index estimation. This represents a large reduction from the initial 51029 SNPs because SNPs on the array were selected based on P. glauca and P. engelmannii, resulting of relatively low SNP quality scores and call rates for P. sitchensis reference genotypes.  Previously, Hamilton et al. (2015) identified a small number of natural hybrids with shared P. engelmannii, P. glauca and P. sitchensis ancestry within the Bulkley Valley and Prince George breeding zones. The majority of the ancestry in these zones originated from P. engelmannii and P. glauca, with P. sitchensis representing a minor ancestral component (~7% in Prince George). To accommodate this unbalanced hybrid ancestry, I estimated hybrid indices for each individual seedling’s genotype using the projection analysis function in the software ADMIXTURE (Alexander & Novembre 2009). The ADMIXTURE projection analysis reference panel consisted of 500 genotypes. From the initial sample of 2424 individuals, 45 P. engelmannii genotypes were selected from higher elevation (1350 to 2000m) populations in south central BC that had >90% P. engelmannii ancestry (J. Degner, unpublished data, University of British Columbia, Centre for Forest Conservation Genetics), and 31 pure P. glauca genotypes were selected from populations sampled in north eastern Alberta. An additional 31 pure P. sitchensis from southern Alaska were included (genotyped by J. Elleouet, unpublished data, University of British Columbia, Centre for Forest Conservation Genetics) using the same SNP array. The remaining 393 reference panel genotypes were sampled randomly from the 2348 individuals not already sampled for the reference panel. Hybrid ancestry of the reference panel was estimated in ADMIXTURE using K = 3 species (Hamilton et al. 2015), and then hybrid ancestry estimates for the remaining 1572 genotypes were projected from population allele frequencies of the reference panel. Hybrid ancestry proportions of the reference panel were validated using Structure analysis (Pritchard et al. 2000) of 817 putatively neutral SNPs that were not adaptive candidates in the analyses of Yeaman et al. (2016). Unbalanced ancestry proportions from the three parental taxa meant the full sample of 2424 individuals could not be validated in this way. ADMIXTURE Q-values of P. engelmannii, P. glauca and P. sitchensis ancestry proportions for each genotype were averaged for each breeding zone by seedling type combination (natural or selected), in 42  line with the phenotypic and climatic data. Because mean P. sitchensis ancestry proportions in most zones were minimal, the proportion of P. engelmannii ancestry was used to characterize hybrid index for the purpose of comparing seedling types: values of zero indicate complete P. glauca ancestry; values of one indicate complete P. engelmannii ancestry. Pairwise significant differences in hybrid index between seedling types within breeding zones were preceded by tests for the normal distribution of each seedling type’s hybrid index, and equal variances between seedling types. Equation 2.2 was used to estimate the amount of variance hybrid ancestry explained in the phenotypic data, using hybrid index means as the independent variable. Similarly, the variance in hybrid index explained by each climate variable was also estimated using Equation 2.2, with 𝑦𝑖𝑗  being replaced by the hybrid index value of seedling type i in breeding zone j. These analyses, and all subsequent analyses were performed in R (R Core Team 2016) unless otherwise stated.  3.2.5 Phenotypic Data and Analyses Seedlings were phenotyped for six growth, phenology and cold injury traits during the second (2013) or third (2014) growing seasons. Seedling height (cm) and shoot dry mass (g), were measured during season three as in Chapter 1, and the same growth curve analysis was used to estimate growth rate (cm day-1) from 14 season two height measurements. Bud break was recorded weekly from late March to late May of the second growing season, and classified as visible needle emergence from within the apical bud scales. Bud set was recorded weekly from mid-June to mid-September of the second growing season, and defined as the presence of brown apical bud scales. Timing of bud break and bud set were translated to day of year starting January 1st. Autumn cold injury testing of needles followed the artificial freeze testing protocol described in Chapter 1. Cold injury testing was completed over a three-week period, commencing on September 16th 2014 (season three), using -15oC and -22oC test temperatures. Flint et al.'s (1967) index of cold injury (I) was calculated for individual seedlings at each test temperature, relative to the unfrozen control samples. Individual seedling’s I values from both test temperatures were then averaged and the mean I values were used for analysis. I represents the percent cold injury incurred; zero = undamaged, 100 = maximum freezing damage.  Phenotypic data analyses were similar to those of lodgepole pine in Chapter 2. Breeding zone by seedling type means were estimated as best linear unbiased estimates (BLUEs) of the fixed effects using the linear mixed model in Equation 2.1 implemented in the software package ASReml-R version 3.0 (Butler 2009). Significant pairwise differences between BLUEs of seedling type means within breeding 43  zones were tested using two-sample t-tests. Using the breeding zone by seedling type means, the pairwise trait-trait correlations were calculated for each seedling type to identify possible trade-offs among adaptive traits resulting from selection for growth.  To identify differences in trade-offs among adaptive traits in the selected versus natural seedlings, seedling type-specific correlations between the mixed-model BLUEs for seedling height and the other five traits were calculated. Phenotypic clines associated with 16 climatic variables (Table 3.2), as well as climate PC1 and PC2 scores, were estimated for each trait using the model in Equation 2.2. The fit and significance of clines for each seedling were tested independently, and also for significant differences between slopes of seedling type clines using the model’s interaction term.  3.2.6 Climatic Biases Within each breeding zone I calculated the difference between mean natural seedling MAT and the mean MAT of selected seedlings weighted by maternal contribution to seedlots, then I calculated the percent height gain from natural to selected seedlings in each zone. Height gains were regressed upon MAT differences to test the hypothesis that growth gains occur because parent trees in breeding programs tend to be sourced from warmer, more productive sites within breeding zones. The embryos of interior spruce seeds develop between July and late August (Owens & Molder 1984). To test for the epigenetic effects on seedlings traits due to seed orchard temperature during embryo development differences were calculated between the mean summer temperatures (MST) (June to August) of mean seed orchard parent tree climates (calculated in Chapter 3.2.2) and their respective seed orchard locations in the years selected seedlots were produced. Height gains among breeding zones were regressed upon these MST differences. The common garden test site for this study is located on a mild, moist coastal site, outside the natural range of interior spruce. To test possible effects of common garden climate on seedling growth, I calculated the difference between each breeding zone’s mean MAT of selected seedlings and the test site MAT, then regressed seedling height gains upon these MAT differences.   3.3 Results Ninety-two percent of seedlings germinated and survived for three growing seasons without incurring damage that compromised their genotypic or phenotypic data. Of these, individual height growth curves 44  were successfully modelled for 2058 of 2249 seedlings. Growth curves of natural and selected seedlings had R2 estimates of 0.879 and 0.904 respectively.   3.3.1 Hybrid Index Means Estimates of hybrid index from ADMIXTURE for the projection analysis reference population were strongly correlated with Structure hybrid index estimates for the same samples (r = 0.99, p < 0.0001) (Figure B.1). Interior spruce populations from AB were dominated by P. glauca, where the maximum hybrid index (proportion of P. engelmannii ancestry) of natural seedlings was 0.10 in breeding zone G1, immediately proximal to the central AB-BC border. In BC, hybrid index of natural seedlings varied among breeding zones from 0.19 (PR mid) to 0.89 (NE mid), averaging 0.53 (Table B.1). The difference in hybrid index between natural and selected seedlings in AB breeding zones was 2% or less, and so here I focus only on BC breeding zones.  The effects of selective breeding on hybrid ancestry varied among BC breeding zones. Ternary plots show several trends within breeding zones (Figure 3.2). 1) The majority of interior spruce ancestry in BC derives from P. engelmannii and P. glauca. Ancestry from P. sitchensis is most prominent in BV low, but the P. sitchensis components of hybrid indices were low and similar between seedling types (both ~4%), except for a few outliers. Elsewhere, average P. sitchensis contributions to ancestry are negligible (≤0.02%). 2) For breeding zones with primarily P. engelmannii x P. glauca ancestry, variation in ancestry among trees within zones is less in selected seedlings than in natural seedlings, with the exception of NE low. 3) Selection and breeding have shifted the distribution of ancestry proportions towards P. glauca in several breeding zones including NE low, NE mid, PG high and PR mid. The greatest difference between seedling types within a zone was in NE low, where hybrid index was 24% lower in selected than in natural seedlings, while hybrid indices of selected seedlings in NE mid and PG high were 18% and 14% lower respectively (Table B.1). In contrast, PG low and TO low, both had hybrid index values that were 4% greater in selected than in natural seedlings. Significant differences between seedling types were not tested because within all breeding zones the distributions of natural seedling hybrid index values were disparate, inconsistent and would invalidate assumptions of equal variances for pairwise statistical tests. However, small standard errors of hybrid index means (Table B.1, Figure B.2), suggest these differences between seedlot types are substantive.   45  3.3.2 Hybrid Index-Trait Relationships For all traits, hybrid index explains more of the phenotypic variance in selected than in natural seedlings, although differences in r2 values between seedling types vary greatly among traits (Table 3.3). In general, growth trait-hybrid index relationships were weaker than growth trait-climate relationships (Table B.5). Hybrid index did not explain significant variation among breeding zones in final seedling height (Figure 3.3a), but growth rate in selected seedlings was significantly related to hybrid index (r2 = 0.59). Bud break timing had a significant relationship with hybrid index only in selected seedlings (r2 = 0.54), while bud set timing had no relationship with hybrid index for either seedlot type (Table 3.3). In contrast, cold injury-hybrid index relationships were strong and statistically significant, but differed very little between seedling types (natural r2 = 0.79, selected r2 = 0.83) (Figure 3.3b), reflecting the strong clines in cold injury with low temperature-related climatic variables (Table B.5).  3.3.3 Hybrid Index Clines Large amounts of variation in interior spruce hybrid index were explained by latitude, mean annual temperature, and cold-related variables (mean coldest month temperature, degree days below zero, and extreme minimum temperature) (Table 3.2). The most striking result from clines in hybrid index along these climatic gradients is their very similar r2 values between seedlot types (Δ r2 ≤ 0.05), indicating that selective breeding is not significantly altering these relationships (e.g. Figure 3.4a). However, the clines in hybrid index with log mean annual precipitation and precipitation as snow show substantial differences between seedlot types with r2 estimates for hybrid index of natural seedlings twice as large as those for selected seedlings (Table 3.2, Figure 3.4b). Conversely, mean summer precipitation, summer heat-moisture index and climatic moisture deficit explain a somewhat greater proportion of variation in hybrid index for selected seedlings (Δr2 = 0.2 to 0.28), but these clines were not significant (Table 3.2).  3.3.4 Breeding Zone by Seedlot Type Means Of the six phenotypic traits analysed, four had normal distributions and met homogeneity of variance assumptions without transformation. Shoot mass values were quarter-root transformed and subsequently met normal distribution and homogeneity of variance assumptions. Bud set had a bimodal frequency distribution and was not transformed. Mean seedling height was greater for selected seedlings than natural seedlings in all breeding zones (Table B.4, Figure 3.5a), and significantly greater in 10 out of 14 zones. Height gains in selected 46  seedlings from BC breeding zones ranged from 28% to 86% (average 51%), while AB breeding zones had more modest gains ranging from 12% to 30% (average 19%). In 11 of 14 breeding zones, growth rate was greater in selected seedlings, and significantly so in seven zones (Figure B.3a). Selected seedlings had greater shoot dry mass in all breeding zones, and differences between seedlot types were statistically significant in 12 of 14 zones (Figure B.3b).  Selective breeding had small effects on mean bud break timing. Mean bud break varied by 14 days across all breeding zones and seedlot types, and differed between seedlot types within zones by a maximum of 5.1 days (Table B.4, Figure B.3c). In AB, the direction of bud break differences between seedlot types was inconsistent, but in BC, selected seedlings broke bud 0.5 to 5 days earlier (average 1.5 days) in every BC breeding zone except NE low (2.5 days later). Even so, significantly earlier bud break only occurred in selected seedlings from PG low and TO low.  Bud set timing showed far greater variation among breeding zones than bud break timing, with mean bud set date varying 55 days across all breeding zone by seedling type combinations. Bud set was delayed by 3.5 days to 29 days (average 13 days) in selected seedlings across all breeding zones, significantly so in nine zones (Table B.4, Figure B.3d), and generally delays associated with selection were greater in BC than AB. BV low and PR mid had the greatest difference in bud set timing between selected seedlings and natural seedlings; bud set was 29 days later in selected seedlings of both zones. Cold injury exhibited no consistent differences between natural and selected seedlings among breeding zones (Table B.4, Figure 3.5b), despite the generally delayed bud set timing of selected seedlings. The range of average cold injury differences between seedling types within breeding zones was -7% to +7%. Mean cold injury of selected seedlings was less than natural seedlings in the EK and NE low breeding zones, even though relatively large height gains were observed in these zones. By contrast, BV low had the greatest increase in cold injury associated with selection (7%), as well as the greatest relative (86%) and absolute height gains of any zone compared to the respective natural seedling mean.  3.3.5 Trait-Trait Correlations Correlations among growth traits, as well as between growth traits and bud break, bud set and cold injury, were uniformly strong (r ≥ 0. 73), and statistically significant (α = 0.0033) (Table 3.4). Selected seedlings tended to have slightly stronger correlations among traits than natural seedlings, although differences were small (maximum Δr = 0.14) with the exception of growth rate versus bud set.  The correlation coefficient between bud break and bud set decreased slightly from natural to selected seedlings (Δr = 0.18), while it increased slightly between bud set and cold injury (Δr = 0.17). 47   3.3.6 Phenotypic Clines The first principal component of seedling climate variables explained 41% of climatic variation, while PCs 2, 3 and 4 cumulatively accounted for 72%, 87% and 93% of climatic variation (Table B.2). PC1 loadings were dominated by low and average temperature variables, while PC2 loadings were harder to summarize (Table B.3). Breeding zone means of PC1 and PC2 scores were used in clinal analyses to summarize phenotypic variation associated with climate in addition to the individual climate variables.  After Bonferroni adjustment for multiple comparisons within each seedling type using an α = 0.0031 cut-off value, most traits showed significant clinal variation along gradients of latitude, mean annual temperature , mean coldest month temperature, degree days below zero, extreme minimum temperature and climate PC1 (Table B.5). In most cases these variables explained greater proportions of variation in selected than in natural seedlings. For growth traits, the derived temperature-precipitation variables climatic moisture deficit and summer heat-moisture index also explain greater proportion of variation in selected seedlings. Clines with longitude, elevation, extreme maximum temperature, precipitation variables and climatic PC2 were mostly moderate to weak and not statistically significant (except in one case, cold injury of natural seedlings versus log mean annual precipitation). Phenotypic trait means consistently had strong relationships with mean annual temperature (MAT), and they are used here to illustrate clinal differences between natural and selected seedlings.  Clines in the three growth traits associated with MAT were moderate to strong, and individually significant, but all had non-significant differences in cline slope between seedling types (Figure 3.6a, b and c). Bud break has a moderate to strong relationship with MAT (r2 = ~0.6), but this relationship did not differ between seedling types (Figure 3.6d). Bud set varied significantly only with MAT for selected seedlings (r2 = 0.54) (Figure 3.6e), and mean bud set day is similarly delayed in selected seedlings in all breeding zone MATs.  These clines in phenology also show that seedlings from warmer breeding zones both break bud earlier and set bud later, thus growing season duration is extended at both ends compared to colder breeding zones. Growing season duration varied among zones from ~83 days to ~118 days. Cold injury had the strongest clines in MAT of any trait, but these clines differed very little between natural and selected seedlots (Figure 3.6f).   3.3.7 Climatic Biases Within breeding zones, the average MATs of selected seedlings differed from natural seedlings by -0.76oC to +0.78oC (average 0.14oC), and among breeding zones height gains were weakly associated with 48  MAT differences between seedling types (r2 = 0.14, p = 0.173). On average, mean summer temperature (MST) was greater for seed orchards than mean parent tree origins by 3.2oC (range +0.51 to +7.02oC), but this varied provincially. Seed orchard MST’s were greater by an average of 1.15oC (range +0.51oC to +1.5oc) in AB and 4.7oC (range 1.93oC to 7.02oC) in BC.  Seed orchard MST differences among breeding zones has a modest relationship with height gains (r2 = 0.41, p = 0.014) (Figure B.5). Differences between mean breeding zone MAT’s for selected seedlings and test site MAT’s were large (average 9.3oC, range 5.2oC to 12.8oC), but among breeding zones these differences explained little variation in height gains (r2 = 0.15, p = 0.173).   3.4 Discussion Hybrid ancestry and the corresponding adaptive phenotypes of selectively bred interior spruce seedlings appear to have been refined for growth within the constraints of local breeding zone climates. Growth gains result from both increased growth rate and delayed bud set of selected seedlings, but have not compromised the adaptive synchrony of autumn cold hardiness development relative to their natural counterparts. The P. glauca–P. engelmannii hybrid index has weak to moderate relationships with growth traits, but strong relationships with cold injury. Cold temperature-related climate variables explain the greatest proportions of variance among populations in both hybrid index and phenotypic traits, although the strength of relationships among hybrid index, cold injury and cold temperature-related climate variables differ little between natural and selected seedling types. This suggests that adaptive cold hardiness is closely tied to hybrid ancestry and not compromised in breeding populations. Strong relationships among hybrid index, phenotypic traits, and climate among breeding zones mean that assisted gene flow (AGF) prescriptions will be necessary to match the adaptation of selectively bred interior spruce seedlots to future climates. Even so, within acceptable transfer limits these prescriptions are unlikely to be compromised by breeding because there appear to be no major antagonistic trade-offs with adaptive phenology and cold hardiness.  3.4.1 Selective Breeding and Hybrid Ancestry Coarse-scale patterns of hybrid ancestry reflect large differences between AB and BC in the extent of hybridization. AB is dominated by P. glauca genotypes and hybrid ancestry changes in selected seedlings are small (Figures 3.2 and B2), implying either P. engelmannii genotypes are maladaptive in those environments and don’t offer growth gains in unsuitably cold, dry climates, or adaptive alleles are at too 49  low a frequency to respond to selection. In contrast, hybrid ancestry of natural populations is heterogeneous in BC, spanning the range of pure P. engelmannii to pure P. glauca genotypes. Changes in the distribution of individual seedling hybrid index values and mean values within BC breeding zones (Figure 3.2, Table B.1) indicate the effects of selective breeding on hybrid ancestry composition varies among breeding zones. Shifts in hybrid index means towards P. glauca parental genotypes were observed in selected seedlings from NE and EK breeding programs of south eastern BC that are consistent with De La Torre et al. (2014). In the EK, PG high and PR mid zones, increased P. glauca ancestry reflects selection on standing genetic variation within breeding zones. However, the NE low and NE mid breeding populations contain 10 pure P. glauca genotypes sourced from Ontario in Eastern Canada and tested for growth under local conditions in BC.  In NE low and NE mid these parent trees provide ~16% and ~6% of the respective maternal contributions to sampled seedlots based on cone production, and probably underlie substantial shifts toward P. glauca ancestry in these naturally P. engelmannii dominated zones. Within those BC breeding populations that are selected entirely from local standing genetic variation (all zones except NE), decreased variation in seedling hybrid index values (Figures 3.2 and B2) suggests that selection for growth is favouring a narrower range of genotypes.  Across the 14 breeding zones, hybrid index explains low to moderate proportions of variation in the growth traits of natural seedlings (Table 3.3). These relationships are much stronger than similar latitude versus growth comparisons of Roche (1969), whose P. glauca x P. engelmannii hybrid index estimated from cone morphology broadly corresponds to latitude. Hybrid index explains a greater proportion of variation in the growth traits of selected seedlings, and slope differences between seedling types appear to be driven by a combination of hybrid index changes favouring P. glauca and height gains in southern BC breeding zones (Figure 3.3a). By contrast, cold injury in artificial freezing tests is closely associated with ancestry. Greater P. engelmannii ancestry corresponds to greater cold injury, but the differences between seedling types are negligible (Figure 3.3b). Strong clines in hybrid index with cold-related climatic variables suggest low temperatures are the strongest climatic selective agents, but again, negligible differences occur between seedling types. These relationships with hybrid index reflect both the importance of cold hardiness to adaptation in the parental species, and the strongly conserved nature of cold adaptation within breeding populations. Like Hamilton et al. (2015), hybrid index clines in log mean annual precipitation (log MAP) and precipitation as snow (PAS) are important adaptive gradients in natural populations, but these relationships with precipitation are disrupted in selected seedlings (Figure 3.4b). Conversely, the slightly stronger relationships of hybrid index with climatic moisture deficit (CMD) and summer heat moisture 50  index (SHM) for selected versus natural seedlings corresponds to a shift towards P. glauca ancestry and greater tolerance for low moisture conditions over P. engelmannii (Table 3.2).  3.4.2 Effects of Selection on Adaptive Phenotypic Traits Differences between natural and selected seedlings (Figure 3.5a) correspond to genetic gains for growth achieved by selective breeding programs. The modest height gains in AB compared with larger gains in BC reflect greater selection intensity and an additional breeding cycle in older BC breeding programs. Large height gains are also feasible because BC breeding zones encompass most of the interior spruce hybrid zone, allowing rapid responses to selection from large pools of standing hybrid genetic variation within breeding zones. Height gains are not related to microgeographic variation in mean annual temperature (MAT) within breeding zones, or to differences between breeding zones and test site MAT. A modest relationship was present between height gains and the differences between breeding zone and seed orchard mean summer temperatures during embryo maturation, which is largely driven by height gains in BC breeding zones (Figure B.5). This relationship could be an epigenetic consequence of warmer than native seed maturation environments of seed orchards (Bräutigam et al. 2013), but I suspect this effect weak or non-existent. Seed orchards are located at opposite ends of the temperature gradient that populations used here sample (Figure 3.1), with low genetic gain Alberta seed orchards located in north-central  Alberta and the high-gain BC orchards mostly in the warm Okanagan Valley of southern BC.  It is impossible to separate genetic from potential epigenetic effects with this data set because genetic gain and seed orchard climate are correlated.  The effects of selection on cold injury within breeding zones were small (Figure 3.5b).  Some breeding zones had increased height associated with slight increases in cold injury, but selected seedlings from southern BC breeding zones had increased height and decreased cold injury, corresponding with shifts toward fast growing, cold resistant P. glauca ancestry, consistent with the findings of De La Torre et al. (2014) from this region of the hybrid zone. Even so, changes in ancestry do not necessarily underlie phenotypic responses to selection. For example, BV low has the greatest height gains of any breeding zone and slightly greater cold injury in selected seedlings. Ancestry proportions in BV low remain identical (Δ~1%) between seedling types (Figures 3.2 and B2), and selection does not appear to have favoured introgression from faster growing trees with more P. sitchensis ancestry previously identified in this breeding zone (Hamilton et al. 2015). Pairwise correlations among traits were uniformly strong, but differed little between seedling types (Δr ≤ 0.17) for 13 of 15 pairwise trait combinations (Table 3.4). This suggests any negative trade-51  offs among height growth, adaptive phenology and cold injury that might constrain gains are weak within breeding populations. Even though bud set-growth rate correlations increased modestly (Δr ~0.3) with selection, seedlings that grew faster and set bud later had negligible increases in cold injury.  Clines in growth traits along gradients in mean annual temperature (MAT) explained moderate to large proportions of variation among breeding zones, but differences in slopes between seedling types were small (Figure 3.3a, b and c). Greater cline intercepts but similar slopes of selected relative to natural seedlings suggest that gains in growth are relatively similar across the range of MAT variation in interior spruce. Among breeding zones, MAT reflects variation in mean climate, but does not necessarily capture climatic extremes that cause physiological stress and drive local adaptation. CMD and SHM, representing moisture stress, and the cold-related variable extreme minimum temperature (EMT) all have strong relationships with growth. Selected seedlings have stronger clines in growth traits with CMD and SHM than natural seedlings, corresponding to the stronger relationships observed in selected seedlings between these two variables and hybrid index (section 4.1). For selected seedling growth traits, the greater proportion of variation among breeding zones explained by CMD and SHM for selected seedling growth traits suggests that within breeding populations, hybrid seedlings may be more able to tolerate reduced local moisture conditions expected in the future. Growth trait clines with EMT for selected seedlings explained 1.5 to 2 times as much variance as natural seedlings. Mean coldest month temperature shows a similar pattern, as does degree days below 0oC, although this latter variable reflects average rather than extreme cold. Bud break clines with MAT were identical between seedling types and unexpectedly strong given that Roche (1969) only found weak  bud break associations with latitude across a wide range of BC interior spruce provenances, and in many other tree species this trait varies little among populations (Howe et al. 2003). This result probably originates from an expanded range of standing genetic variation for bud break phenology across these three hybridizing species.  Even though selective breeding changes the strength of some pairwise climate- trait relationships, these results suggest there is little overall effect of selection on adaptive trait-climate relationships. Height gains appear to be derived from both delayed bud set and faster growth rate, although their specific effects are inseparable. Height-growth rate correlations increased with selection by 0.14 (Table 3.4), while clines in selected seedling growth rate and bud set with climate PC 1 were also slightly stronger (Δr2 = +0.08 and +0.17 respectively). Bud set precedes development of autumn cold hardiness, but the adaptive synchrony of these two processes appears to be temperature mediated and flexible in response to selection (Tanino et al. 2010; Hamilton et al. 2016). This is congruent with the lodgepole pine findings of Chapter 1; delayed bud set did not result in adaptive compromises to cold 52  hardiness of selected seedlings that were able to acquire hardiness more rapidly than their natural equivalents. In interior spruce I also find that bud set-cold injury correlations are weak (Table 3.4), and climatic associations with bud set are weak to moderate in both seedling types, unlike those for cold injury (Table B.5).  Adaptation of interior spruce to cold dominates signals of local adaptation in this study. Strong associations among hybrid index, cold injury and cold-related climatic variables are congruent with patterns of adaptive variation in autumn cold hardiness for widespread conifer species (Bansal et al. 2016; Gray et al. 2016; Hurme et al. 1997; Liepe et al. 2016; Rehfeldt 1994), and genotype-environment  associations for adaptive markers associated with cold tolerance in interior spruce (Yeaman et al. 2016). Adequate autumn cold hardiness developed prior to potentially damaging temperatures is essential to seasonal climatic synchrony, survival and growth. The need for locally adaptive autumn cold hardiness in natural populations is strong enough to be highly conserved by both interior spruce and lodgepole pine breeding programs in western Canada, even though growth traits and bud set phenology experience strong directional selection in selected seedlings of both species.   3.5 Conclusions Deploying the correct interior spruce genotypes on reforestation sites to match future rather than historic climates requires a thorough understanding of the relationships among ancestry, phenotypes and climate that underpin adaptation in this hybrid zone (Aitken et al. 2008). Selection and breeding resulted in decreased variation in hybrid ancestry within most of the selected BC breeding zone populations. Among breeding zones, relationships of growth traits with both hybrid ancestry and climate were stronger in selected than in natural seedlings, and adaptive associations between hybrid ancestry, cold hardiness and cold-related climate variables were always strong. This means relatively large proportions of adaptive variation are explained by differences among breeding zones, and that growth and cold hardiness of breeding populations are closely associated with breeding zone climates. Therefore AGF will be required for accurate deployment of selectively bred genotypes to match future climates.  Similarly-adapted conifer populations in western Canada appear to be distributed over wider geographic areas than previously thought (O’Neill et al. 2014; Liepe et al. 2016). My findings of stronger growth-hybrid index-climate relationships in selected interior spruce seedlings correspond to O’Neill et al. (2014) who found that safe transfer distances of breeding populations exceed those of current seed 53  deployment zones, but are somewhat shorter than safe transfer distances of equivalent natural stand populations. By decomposing the relationships among hybrid ancestry, phenotypic traits, and climate, it seems that negative trade-offs between growth and the climatically adaptive traits I studied in selected seedlings are small or absent relative to their natural stand counterparts. Within each breeding zone, adaptation to local climates has been effectively maintained by breeding programs that select, breed and test genotypes on several representative local sites, where low temperatures in particular limit the growth gains that can be achieved.  This finding in interior spruce is consistent with the lodgepole pine results in Chapter 1. It implies that within the critical seed transfer distance (see Ukrainetz et al. 2011) of breeding populations under current climates, AGF prescriptions for natural and selected seedlots do not need to differ. AB and BC should be able to implement AGF that optimizes growth gains from selection within breeding zones without heightened risk of cold injury, because adaptive climatic relationships are maintained in breeding populations. Additionally, the hybrid ancestry and phenotypes of selected interior spruce seedlings may be somewhat pre-adapted to tolerate reduced moisture conditions expected as the climate changes. It is also likely that in the future, seedlots will be safely deployable over more extensive areas than are presently allowed in reforestation policies. This is partly because the spatial scale of similarly adapted populations is greater than previously thought, and partially because seed movements based on climatic distance rather than geographic distance better matches genotypes with environments (O’Neill et al. 2014; Liepe et al. 2016). Furthermore, where breeding populations have growth gains and decreased risk of cold injury that are associated with an increased proportion of P. glauca ancestry, it may be safe to deploy their seedlots even more extensively to achieve future gains in timber supply.   54  3.6 Tables           Natural Selected Province Breeding     zone Elevation           range (m) Seedlots Seedlings Seedlots Clones per seedlot Seedlings AB D1 501 - 800 15 108 2 81 & 72 112 AB E 301 - 650 6 72 1 87 66 AB G1 651 - 1050 8 80 1 146 66 AB G2 501 - 900 5 70 1 87 66 AB H 251 - 550 9 84 1 50 66 AB I 701 - 1200 11 92 1 153 66 BC BV low 0 - 1200 8 80 2 118 & 19 112 BC EK all 750 -1700 8 88 1 26 66 BC NE low 0 - 800 5 56 1 34 66 BC NE mid 800 - 1500 11 110 1 63 66 BC PG low 600 -1200 26 152 2 19 & 53 112 BC PG high 1200 - 1550 4 64 1 40 66 BC PR mid 650 - 1200 10 88 1 40 66 BC TO low 700 - 1300 11 92 2 30 & 41 112  Table 3.1 Breeding zones sampled for selectively bred and natural seedlots, their elevational range, number of seedlots sampled, number of seedlings established, and for selected seedlots the number of parent tree clones in each seed orchard contributing cones to those seed lots. AB breeding zones are formally identified as D1, E, G1, G2, H and I. BC breeding zone abbreviations are BV (Bulkley Valley), EK (East Kootenay), NE (Nelson), PG (Prince George), PR (Peace River), and TO (Thompson – Okanagan). 55     Geographic or Climatic Variable Acronym Natural Selected  r2    p-value  r2    p-value Latitude (oN) LAT 0.75 < 0.0001 0.73 < 0.0001 Longitude (oW) LONG 0.13 0.2141 0.05 0.4617 Elevation (m) ELEV 0.41 0.0131 0.35 0.027 Mean annual temperature (oC) MAT  0.68 0.0003 0.62 0.0008 Mean warmest month temperature (oC) MWMT 0.02 0.6104 0.00 0.9782 Mean coldest month temperature (oC) MCMT 0.75 < 0.0001 0.75 < 0.0001 Log mean annual precipitation (mm)  log MAP  0.76 < 0.0001 0.34 0.0284 Mean annual summer precipitation (mm) MSP 0.03 0.5421 0.23 0.0816 Summer heat moisture index  SHM  0.03 0.5558 0.26 0.063 Degree days below zero (oC) DD < 0 0.74 < 0.0001 0.71 0.0002 Growing degree days above five degrees (oC) DD > 5 0.01 0.8154 0.04 0.4757 End of frost free period (day) eFFP 0.58 0.0017 0.48 0.0062 Precipitation as snow (mm) PAS  0.84 < 0.0001 0.47 0.0071 Extreme minimum temperature (oC)  EMT  0.88 < 0.0001 0.83 < 0.0001 Extreme maximum temperature (oC) EXT 0.00 0.9591 0.09 0.3132 Hargreaves climatic moisture deficit (mm) CMD 0.10 0.2737 0.38 0.0188 Climatic principle component 1 scores PC1 0.51 0.0039 0.54 0.0027 Climatic principle component 2 scores   PC2 0.52 0.0038 0.32 0.0347  Table 3.2 Variation in breeding zone hybrid index means explained by breeding zone climate means for  natural and selected seedlots. p-values are statistically significant at an adjusted α = 0.0028 cut-off (bold font). 56     Trait Natural Selected r2 p-value r2 p-value Height (cm) 0.25 0.0686 0.45 0.0085 Growth Rate (cm/Day) 0.36 0.0231 0.59 0.0014 Shoot Mass (^0.25)(g) 0.36 0.0241 0.41 0.0141 Bud Break (Day) 0.27 0.058 0.54 0.0027 Bud Set (Day) 0.03 0.5871 0.12 0.233 Cold Injury 0.79 < 0.0001 0.83 < 0.0001  Table 3.3 Variation in breeding zone trait means explained by breeding zone hybrid index means, for natural and selected seedlots.  p-values are statistically significant at an adjusted α = 0.0083 cut-off (bold font).  57    Comparison Seedlot Type r p-value Height - Growth Rate Natural 0.79 0.0007  Selected 0.93 < 0.0001 Height - Shoot Dry Mass Natural 0.99 < 0.0001  Selected 0.99 < 0.0001 Height - Bud Break Natural -0.92 < 0.0001  Selected -0.86 < 0.0001 Height - Bud Set Natural 0.84 < 0.0001  Selected 0.85 < 0.0001 Height - Cold Injury Natural 0.74 0.0023  Selected 0.79 0.0007 Growth Rate - Shoot Dry Mass Natural 0.80 0.0006  Selected 0.93 < 0.0001 Growth Rate - Bud Break Natural -0.81 0.0004  Selected -0.86 < 0.0001 Growth Rate - Bud Set Natural 0.43 0.1235  Selected 0.73 0.0027 Growth Rate - Cold Injury Natural 0.73 0.0029  Selected 0.82 0.0003 Shoot Dry Mass - Bud Break Natural -0.91 < 0.0001  Selected -0.86 < 0.0001 Shoot Dry Mass - Bud Set Natural 0.83 0.0003  Selected 0.83 0.0002 Shoot Dry Mass - Cold Injury Natural 0.76 0.0015  Selected 0.81 0.0005 Bud Break - Bud Set Natural -0.75 0.0021  Selected -0.57 0.0349 Bud Break - Cold Injury Natural -0.75 0.0019  Selected -0.76 0.0018 Bud Set - Cold Injury Natural 0.44 0.1111   Selected 0.61 0.0202  Table 3.4 Pairwise correlation coefficients between all six traits for natural and selected seedlings. For each seedling type p-values are statistically significant at an adjusted α = 0.0033 cut-off (bold font). 58  3.7 Figures    Figure 3.1 Geographic origins of the natural and selected seedling populations sampled from across the ranges of P. engelmannii, P. glauca, and their hybrid zone in AB and BC. Natural populations are represented by filled circles, selected seedling breeding zones are represented by filled polygons. Diamonds indicate seed orchard locations. 59   Figure 3.2 Ternary plots representing the proportion of Engelmann (P. engelmannii), white (P. glauca) and Sitka (P. sitchensis) spruce ancestry of individual seedlings within the BC breeding zones. All AB breeding zones have >90% white spruce ancestry and are omitted from this figure. 60     Figure 3.3 Regressions of a) height and b) cold injury versus hybrid index. Points represent BLUEs of trait means (Table B.4, Figure 3.5), and the mean spruce hybrid index (P. engelmannii proportion) for each breeding zone by seedling type combination (Table B.1). p-values are statistically significant at the adjusted α = 0.0083 cut-off used in Table 3.3. 61     Figure 3.4 Examples of climatic clines in hybrid index. EMT (a) has the strongest climatic cline, while log MAP (b) has is the cline in hybrid index with the greatest difference between seedling types. Points represent mean hybrid index (proportion of P. engelmannii ancestry, Table B.4) and mean climatic values for each breeding zone x seedling type combination. p-values are statistically significant at the same adjusted α = 0.0028 cut-off Table 3.4. 62     Figure 3.5 Bar plots of breeding zone level trait means (BLUEs) including standard error bars, for seedling a) height and b) cold injury.  63    Figure 3.6 Phenotypic clines with MAT for a) height, b) cold injury, c) growth rate, d) shoot dry mass, e) bud break and f) bud set. None of the interactions between seed lot types are significant. Points represent trait means (BLUEs, Table B.4) and climatic means for each of 14 breeding zones. p-values are statistically significant at the same adjusted α = 0.0028 cut-off used in Table B.5. 64  Chapter 4: Subtle shifts in polygenic variation underlying adaptive traits of lodgepole pine confer strong climatic adaptation and modest responses to selective breeding  4.1 Introduction In forest trees, and especially widespread conifers, genetic variation in climate-related traits has led to local adaptation that is both phenotypically and genotypically detectable. Foresters have long been aware of adaptive variation among natural tree populations that is partitioned in response to climate. They have taken advantage of genetic variation both among and within populations to achieve economic and strategic objectives. For more than half a century tree improvement programs that favour timber production have utilised additive genetic variation to achieve substantial growth gains within breeding populations. Genetic components of additive variation for economic and climatically adaptive traits have been estimated and genetic gain from selection predicted using quantitative models that assume many anonymous loci of small effect underlie continuously distributed phenotypes and differentiation among families. The genomic architecture of traits under selection in terms of type, number and effect size of loci was not understood until the development of quantitative trait loci (QTL) mapping and genome-wide association studies (GWAS). Within and among populations, these analytical tools have become a lens to examine the genome-wide architecture of adaptive and economically important traits.  In conifers, genomic variation associated with adaptive phenotypes has been described for a number of species (Chapter 1.3). These studies aim to identify adaptive alleles with detectable phenotypic effects on a locus-by-locus basis using QTL mapping or GWAS (Neale & Savolainen 2004). Similarly, genotype-environment associations (GEA) have been also used to identify adaptive loci in conifers (Eckert et al. 2010; Lind et al. 2017), but environment-associated loci are often anonymous with respect to which phenotypic traits they affect and what the size of those effects are. Both these approaches are limited by their bias towards detecting loci with large phenotypic effects (Hoban et al. 2016). From a quantitative genetics perspective the expectation that genome-scans will detect individual adaptive loci with large frequency differences among populations appears to be biologically and statistically unrealistic (Yeaman 2015). As the number of loci contributing to variation in a trait increases, individual loci will have decreasing effect sizes (Kremer & Le Corre 2012), and adaptive loci of small effect will go undetected as the number corrections for multiple comparisons increases. 65  Therefore, allele frequency shifts in response to selection are likely to constitute subtle changes across many loci (Stephan 2016). Tests to identify polygenic signatures of selection underlying adaptive traits have been developed, e.g., Berg & Coop (2014), but they are not yet widely applied. At present, GWAS in conifers that have typically detected relatively few statistically significant loci, or GEA that detect phenotypically anonymous loci may be inadequate to describe the genome-wide effects of selective breeding programs on adaptation and diversity. Adaptive traits often have intercorrelated phenotypes, either due to pleiotropy or linkage disequilibrium, and may also be composites of multiple adaptive traits, each with smaller effect (Riska 1989). Pleiotropy means that allelic variants associated with adaptive phenotypes cannot truly be considered in isolation, and that selection on one trait may result in correlated responses in another (Ingvarsson & Street 2011). Equally, linkage disequilibrium can cause this effect, due to the physical linkage of loci on chromosomes, or population structure. Genome-wide estimates of linkage disequilibrium in conifers appear to be low on average (Neale & Savolainen 2004; Chhatre et al. 2013), but may be more substantial (r2 ≤ 0.4) within genes even though LD decay is still rapid (Pavy et al. 2012). The accumulation of linkage disequilibrium and increased correlations among loci is an important component of selection responses in outcrossing organisms with high gene flow such as conifers. Theoretical expectations are that phenotypic responses to divergent selection in initial generations (n ~ 20) will be driven by covariance among loci, followed by slow increases in the frequency of adaptive alleles (Kremer & Le Corre 2012). This suggests correlations among loci will dominate responses to selection in natural environments that have continuously varying spatial and temporal selection regimes, while individual allelic responses will remain small and hard to detect in an association analysis context (Yeaman 2015). Correlations among loci and allele frequency shifts will both contribute to phenotypic selection responses in artificially selected populations. The genome-wide effects of domestication in agricultural crop species are characterised by population genetic bottlenecks, selective sweeps in genomic regions under selection and a prevalence of single-gene traits associated with high selection intensities over hundreds of generations (Doebley et al. 2006; Wright et al. 2005; Gross & Olsen 2010). Over two or three generations, conifer breeding programs will certainly not have the same genomic consequences, but if the effects of artificial selection for increased productivity are detectable within conifer genomes, their correlated effects on adaptive phenotypic traits may shed light on genetic relationships and potential trade-offs between growth and adaptation. The expectation that forest trees will become maladapted as climates shift comes from a long history of local adaptation identified in common garden studies (Aitken et al. 2008). With the application 66  of genomic tools to forest trees, this maladaptation can now also be estimated in terms of molecular genetic variation using loci associated with climate (Jaramillo-Correa et al. 2015). Conserving, managing and redeploying genetic variation associated with adaptive phenotypes will be essential to mitigating the effects of global change on forest resources (Fady et al. 2015). Selection has produced detectable genetic gains in desirable phenotypes in lodgepole pine breeding populations in Alberta and British Columbia, and in Chapter 2 I found these populations remain locally adapted to historic climates. Yet, outside the realm of high-gain, short-rotation clonal forestry (e.g., Eucalyptus (Resende et al. 2012)), the effects of artificial selection on genomic variation associated with adaptive phenotypes are not well understood, and are inadequate to understand the full effects of proactive forest genetics policies that mitigate negative climate change impacts. In this chapter, I combine genomic data obtained from a ~50K lodgepole pine single nucleotide polymorphism (SNP) array designed with GWAS and GEA-identified candidate SNPs with the phenotypic and climatic lodgepole pine data from Chapter 2. These data allow me to address four primary research questions. 1) Are there genomic signatures of selection associated with selective lodgepole pine breeding in terms of allele frequency shifts and correlations among loci associated with height or climatically adaptive traits? 2) How strong are the relationships between SNPs associated with adaptive genotypes and climate, and do these relationships respond to selection? 3) Are genotype-climate and phenotype-climate relationships similar? 4) How much of the total phenotypic variation for individual traits do adaptive SNPs explain, and do genomic differences between populations of natural and selected seedlings reflect phenotypic shifts? To address these questions I use genome-wide association analyses to identify candidate adaptive loci for each seedling trait. Using the approach of Turchin et al. (2012), alleles with a positive effect on adaptive phenotypes were summed across candidate loci and their frequencies calculated, both for individual seedlings, and across seedlings within breeding zones. This simple method summarises polygenic allele frequencies and avoids reductions in genomic variation that occur when calculating allele frequency averages or principle components across loci. Integrating genomic, climatic and phenotypic data gives me a robust basis to detect the effects of selection in relation to climatically adaptive phenotypes that are relevant to breeding strategies and assisted gene flow policies. It also allows me to consider the respective merits of using genotypic versus phenotypic variation to inform these policies.   67  4.2 Materials and Methods 4.2.1 DNA Extraction and Genotyping From 2176 seedlings in the lodgepole pine common garden described in Chapter 2, fresh needle tissue was collected in spring 2013.  DNA extraction, SNP array design and genotyping followed the same protocol as spruce in Chapter 3.2.2, with the following exceptions. Samples were genotyped using the AdapTree lodgepole pine Affymetrix Axiom 50K lodgepole pine SNP array (Yeaman et al., unpublished). SNP discovery for this array was based on the lodgepole pine sequence capture dataset described by Yeaman et al. (2016) and (Suren et al. (2016). It included probes for the exons of 24,388 genes. Intergenic regions were identified by mapping the lodgepole pine transcriptome to the loblolly pine (Pinus taeda L.) draft genome (Neale et al. 2014). The SNP array included 50298 SNPs. After successful genotyping of 2084 individuals and filtering out SNPs that had GeneSeek quality values > 0.15 (0 = high quality, 1 = low quality), and call rates < 0.85, 40163 SNPs remained in the SNP table.  4.2.2 Genome-Wide Phenotypic Associations (GWAs) Candidate SNPs were identified from GWAs of the natural seedlings only, before evaluating the effects of selection on these SNPs in the selected seedlings. This approach is consistent with Chapters 1 & 2, whereby natural seedling genotypes form the comparative basis to assess the genomic effects of selection on loci that are adaptive under natural conditions, although it will overlook SNPs that have significant GWAs in selected but not natural seedlings.  Separate GWA analyses were run using the phenotypic residual values for each of the six traits analysed in Chapter 2: seedling height; growth rate; shoot dry mass; growth initiation day; growth cessation day; and, cold hardiness. Residuals were calculated for each trait from Equation 2.2 using ASreml-R (Butler 2009), with the fixed effects removed; block and location in block remained as random effects. Residuals were standardised to a mean of zero and standard deviation of 1 so that GWA effects would be comparable among traits.  Natural seedling genotypes (n = 929) from the SNP table were filtered further to retain 36384 SNPs with a minor allele frequency ≥ 0.01. Of these, 4750 selectively neutral SNPs that were both included in the filtered SNP table and had no significant genotype-phenotype or genotype-environment associations in the analyses of Yeaman et al. (2016) were selected randomly to be used for population structure correction in GWA analyses (J. Yoder, unpublished data), and removed from the SNP table. GWA analysis of natural seedlings was implemented using the mlma function in GCTA (Yang et al. 2011) 68  and 31634 SNPs for each of the six traits. Population structure was corrected for using the grm option of mlma to reconstruct relatedness from the 4750 selectively neutral SNPs. SNPs in the bottom 0.01 quantile of the GWA p-values for each trait were identified as candidate SNPs (n = 317 SNPs per trait). The remaining analyses in this chapter are based on these candidate SNPs.  4.2.3 Allele Frequency Distribution of Top Candidate SNPs To determine if there is a detectable, directional response in the frequency of adaptive alleles among loci, alleles with a positive effect on the phenotype (positive effect alleles) were identified from the ‘b’ (regression slope) value in the GCTA mlma output. For individual seedlings, the number of positive effect alleles was counted across all top candidate SNPs associated with each trait. Probability distributions of these counts were plotted and means were compared between natural and selected seedlings using the Wilcoxon rank sum test function (wilcox.test) in R (R Core Team 2016), with a Bonferroni adjusted cut-off value of α = 0.0083 for six comparisons.   4.2.4 Linkage Disequilibrium Analysis Linkage disequilibrium (LD) was used to determine the extent to which genomic responses to selection resulted from correlations among top candidate loci, and test the expectation that selection should increase the LD among loci under selection. For natural and selected seedlings, pairwise LD was calculated as the squared correlation (r2) among all pairs (n = 50,086 pairwise comparisons for 317 SNPs) of SNPs associated with each trait using the genetics package LD function in R. P-values for each pairwise comparison were obtained from the Chi-squared test of linkage disequilibrium (Ho: LD = 0) included in the LD function, and the proportion of significant (p ≤ 0.05) pairwise comparisons was calculated. Linkage blocks were visually assessed from hierarchically clustered heat maps of pairwise r2 values for SNPs associated with each trait and seedling type. In addition, 317 SNPs were randomly sampled 250 times from the 31634 genotyped SNPs on the array.  Pairwise LD (r2) was averaged from each set of random SNPs, and these mean values were used to estimate background pairwise LD among SNPs on the array.   4.2.5 Positive Effect Allele-Climate Relationships For each breeding zone-by-seedling type combination (n = 24; the 12 lodgepole pine breeding zones (Table 2.1) x two seedling types) the frequency of positive effect alleles across all seedlings was 69  calculated for every top candidate SNP. These allele frequencies were then averaged across SNPs to give the mean positive effect allele frequency for each breeding zone-by-seedling type combination. This process was repeated using the top candidate SNPs for each of the six phenotypic traits. Clinal regressions were used to determine the amount of genotypic variation among breeding zones for each trait’s top candidate SNPs that was explained by climate. The mean frequency of positive effect alleles in each breeding zone was regressed upon the mean breeding zone climate values calculated in Chapter 2.2.2. I used a version of equation 2.4 (repeated below) to compare clines in the frequency of positive effect alleles along gradients for 11 climatic variables (indicated in Table A.1), between natural and selectively bred seedlings. [2.4] 𝑦𝑖𝑗 = 𝛽0 + 𝛽1(𝑥1) + 𝛽2(𝑥2) + 𝛽3(𝑥1𝑥2) + 𝑒𝑖𝑗   Where 𝑦𝑖𝑗  represents the mean frequency of positive effect alleles for seedling type i in breeding zone j, 𝑥1  is a climatic variable, 𝑥2 is the categorical covariate ‘seedling type’, 𝛽0 is the intercept, 𝛽1 and 𝛽2 are the climatic variable and seedling type coefficients respectively, 𝛽3 is the coefficient of the seedling type-by-climatic variable interaction, and 𝑒𝑖𝑗  is the residual error of 𝑦𝑖𝑗. Fit and significance of the seedling type clines were tested independently. Significant differences between cline slopes that correspond to climatic variation were tested using the climate-seedling type interaction term of the model. Cline r2 values for each seedling type are significant at an α = 0.0045 cut-off value after Bonferroni adjustment for comparisons across 11 climate variables. To determine whether climate explains more variance in genotypes or phenotypes, for each seedling type r2 values from the genotype - climate clines were correlated with r2 values from the equivalent phenotype-climate clines in Chapter 2.3.3 (Table A.5).  4.2.6 Positive Effect Allele-Phenotype Relationships The SNP array is highly unlikely to capture all causal or tightly linked loci associated with an adaptive trait. Furthermore, after population structure correction in the GWAS analysis followed by reduction to the top 1% of candidate SNPs, genotyped SNPs are unlikely to explain all of the phenotypic variance in a trait. Therefore, I summarised the relationships between top candidate SNPs and phenotypes by regressing the best linear unbiased estimates of breeding zone-by-seedling type phenotypic means from Chapter 2.3.1, upon the respective frequencies of positive effect alleles calculated above. The model in equation 2.4 was used again, where 𝑦𝑖𝑗  is the best linear unbiased estimate of the mean phenotype for 70  seedling type i in breeding zone j, but instead 𝑥1 and 𝛽1 were the respective variable and coefficient for the frequency of positive effect alleles. The interaction term was included to determine whether phenotype had a significantly different relationship with genotype between natural and selected seedlings.   Lastly, I tested whether differences in the frequency of positive effect alleles co-vary with phenotypic differences between seedling types. Within each breeding zone, two values were calculated for each trait: 1) The difference between mean natural and selected phenotypes; and 2) The difference between natural and selected seedlings mean frequencies of positive effect alleles from the top 1% of SNPs. Among breeding zones, the phenotypic differences in each trait were correlated with the respective positive effect allele differences. Significance of both the genotype-phenotype regressions, and genotype difference-phenotype difference correlations were both tested at an α = 0.00833 cut-off value adjusted for six comparisons.   4.3 Results 4.3.1 Allele Frequency Distribution of Top Candidate SNPs The mean count of positive effect alleles varied considerably among traits across all populations. Out of a maximum possible count of 634 alleles, SNPs associated with the three growth traits and cold injury had mean counts of ~260 to 290 positive effect alleles per population in both seedling types (Table 4.1). Frequency distributions of effect allele counts were slightly negatively skewed for these four traits, and were similar in both seedling types (Figure 4.1a, b, c and f). Even so, there was a detectable, positive directional shift in the frequency of effect alleles in selected versus natural seedlings, and Wilcoxon tests of differences between seedling type frequency distributions were significant after correction for six multiple comparisons.  Growth initiation and growth cessation associated SNPs had marked differences in the mean count of positive effect alleles compared to the growth traits and cold injury. Mean counts of positive effect alleles associated with these two phenology traits were in the range of 131 to 172 (Table 4.1). Conspicuously, growth initiation and cessation SNPs also had bi or tri-modal effect allele frequency distributions for both seedling types (Figure 4.1d and e), but a shift towards higher counts of positive effect alleles in selected seedlings.  The frequency of effect alleles increased slightly in selected seedlings for the SNPs associated with all traits. These increases were in the range of 0.01 to 0.02 (Table 4.1), and were an order of 71  magnitude greater than the LD Δr2 values between seedling types measured on the same scale (Table 4.2).  4.3.2 Linkage Disequilibrium Analysis Background pairwise LD among filtered SNPs was low (mean r2 = 0.0016, se = 9.5 x10-5) from the 250 random SNP samples. Mean LD for the top 1% of SNPs associated with height, growth rate and shoot mass, had a range of r2 values from 0.015 to 0.022, which were 10 to 15 times greater than the background value (Table 4.2). Although these LD increases appear to be small, based on their standard errors, they were statistically significant, and some pairs of candidate SNPs had r2 values approaching unity. The number of significant pairwise LD comparisons increased by ~10% from natural to selected seedlings for SNPs associated with the three growth traits. Corresponding heat maps for height growth associated SNPs (Figure 4.2a and b, and Figure C.1), growth rate (Figure C.2) and shoot mass (Figure C.3) SNPs, suggest only ~1/4 of the 317 SNPs are in strong linkage with other SNPs, and these are grouped into two or three hierarchical clusters. Most of the growth trait SNPs appear to be unlinked. LD for cold injury-associated SNPs had similar mean r2 values to those of growth traits, but the number of significant pairwise comparisons increases by 3.6% in selected relative to natural seedlings (Table 4.2). Heat maps of cold injury SNPs (Figure C.4) show that 1/3 of all 317 SNPs are strongly linked to other SNPs across four or five small to modest sized clusters, while at least 1/2 of the cold injury-associated SNPs are unlinked with any other SNPs, suggesting wide dispersion across the genome. SNPs associated with growth initiation and growth cessation had mean LD values that were strong for both seedling types and much greater on average (r2 ~0.25) than the background value. Mean LD differences between growth initiation and cessation SNPs, and between seedling types for each trait were small. The number of significant pairwise comparisons decreased by ~5% for the selected versus natural growth initiation SNPs, but increased by ~2% for growth cessation SNPs.  High mean LD values for phenology are reflected in heat maps of pairwise r2 values for growth initiation (Figure 4.2c and d, and Figure C.5) and growth cessation (Figure C.6). Although these also differ little between seedling types, each is dominated by a single large cluster of 1/3 to 1/2 of the 317 associated SNPs which are in strong LD. A further ~1/3 of the remaining SNPs form three smaller clusters of strong LD, each containing ~20 to ~35 SNPs. High LD among phenology SNPs and large clusters of pairwise comparisons nearing unity are consistent with multimodal positive effect allele counts among individuals (Figure 4.1d and e), suggesting LD may be driving this pattern.   72  4.3.3 Positive Effect Allele-Climate Relationships Among breeding zones there was substantial variation in the frequency of positive effect alleles (Table C.1). The range of positive effect allele frequencies among breeding zones and seedling types was 0.13 to 0.18 for SNPs associated with seedling height, growth rate, shoot mass and cold injury, but growth initiation and cessation associated SNPs have greater frequency ranges of 0.23 and 0.29 respectively. For all six traits, this variation in the frequency of positive effect alleles among breeding zones is strongly structured in relation to climate.  Clines in frequencies of positive effect alleles for trait-associated SNPs are very strong for both seedling types with temperature related variables (mean annual temperature, extreme minimum temperature, and degree days above 5oC) as well as latitude (Table C.2, Figure 4.3). Similarly, variation in climatic PC 1 is dominated by temperature-related variables (Table A.3) and r2 for positive effect allele-PC 1 clines was > 0.9 for all but two trait-by-seedling type combinations (growth initiation and cessation in natural seedlings). The strongest positive effect allele-climate relationships are between frequencies of shoot mass alleles and PC1 (r2 = 0.95 (natural) and 0.96 (selected)). Relationships of cold injury alleles with both extreme minimum temperature and PC 1 were almost identical regardless of seedling type. Selected seedlings have steeper slopes than natural seedlings in all of these positive effect allele-climate clines (Figure 4.3), but none of the differences between slopes is statistically significant using a Bonferroni α = 0.0045 cut-off value for eleven comparisons. Longitude, elevation, log mean annual precipitation and PC 2 have no strong or significant relationships with the frequency of effect alleles among breeding zones for any trait.  Positive effect allele-climate and phenotype-climate r2 values are strongly correlated (Figure 4.4), and co-vary almost identically between seedling types (natural r = 0.85; selected r = 0.87). Across 66 comparisons per seedling type, the mean difference between natural and selected phenotype - climate r2 values is 0.157, but the mean difference between natural and selected seedling positive effect allele frequency-climate r2 values was 0.094. The parallel but shifted correlations of selected versus natural r2 values in Figure 4.4 occur because on average, the r2 values of selected seedling phenotype-climate clines differ from natural clines by r2 = 0.063 more than selected seedling effect allele-climate clines. Across multiple climatic clines, phenotypes experience a somewhat stronger but similar response to artificial selection than phenotypically associated SNPs.  73  4.3.4 Positive Effect Allele-Phenotype Relationships For growth traits, positive effect allele-phenotype associations are moderate in natural seedlings (Figure 4.5a, b and c), and stronger in selected seedlings, although seedling type slope interactions are not significant. Among breeding zones, an increase in the frequency of positive effect height alleles by 10% corresponds to ~35% (~12 cm) greater growth in selected seedlings (Figure 4.5a). Between seedling types, phenological traits have little difference in the variation among breeding zones explained by their respective effect alleles (Figure 4.5d and e). However, a 10% increase in the frequency of positive effect alleles generates a delay in growth initiation of ~1.5 days, but a delay in growth cessation of ~5 days. Cold injury has strong relationships with variation in positive effect allele frequencies among breeding zones, but these frequencies differ very little between natural and selected seedlings within zones (Figure 4.5f). An increase in the frequency of cold injury effect alleles by 10% generates 20% greater cold injury/less cold hardiness.  Similar patterns of phenotypic covariance with the frequency of positive effect alleles are also evident within breeding zones in response to selective breeding (Figure 4.6). Among breeding zones, the within breeding zone differences between natural and selected seedlots in the frequency of positive effect alleles are moderately correlated with phenotypic differences. None of these difference correlations is significant after correction for multiple comparisons, but small changes in the frequency of effect alleles by 2 - 3 % are correlated with substantial phenotypic differences. Growth rate is the most prominent example; an increase in frequency of positive effect alleles by 3% is correlated with a ~12% (1.2mm) increase in daily growth (Figure 4.6b). Phenology and cold injury difference correlations are somewhat weaker, but for cold injury an increase in frequency of effect alleles of 3% is associated with 4% greater cold injury in selected seedlings (Figure 4.6f). These correlations may have a slight downward bias because each contains a one breeding zone with a low outlying phenotypic value, although among traits these outliers originate from different breeding zones.        74  4.4 Discussion 4.4.1 Genome-Wide Signatures of Adaptation and Selective Breeding My results from this chapter reflect quantitative genetic expectations and empirical results that suggest continuously distributed phenotypic variation in growth and climatically adaptive traits has a highly polygenic genomic basis (Hornoy et al. 2015; Yeaman et al. 2016; Kujala et al. 2017), but the genomic architecture of these traits varies. Among individual seedlings, continuously distributed counts of positive effect alleles for height, growth rate, shoot mass and cold injury correspond to low average linkage disequilibrium and moderate to low proportions of significant pairwise LD comparisons (Table 4.2). It appears adaptive SNPs associated with these four traits are mostly unlinked (Figures 4.2, and C1 to C6), and are widely dispersed across the lodgepole pine genome. This is consistent with other conifer species in which pairwise r2 among the majority of candidate adaptive SNPs decays rapidly over a few hundred base pairs (Chhatre et al. 2013; Lu et al. 2016; Namroud et al. 2010). The genomic architecture of phenology in lodgepole pine is dominated by a cluster of loci in high LD that results in large variation in positive effect allele counts among individuals. High LD among phenology associated candidate SNPs could be due to physical linkage or population structure. It also may be inflated due to the design of the SNP array. Candidate SNPs were chosen for the array based on two main criteria: 1) SNP associations with phenotypic traits including phenology in 280 natural populations from analysis of sequence capture data for exons in ~23,000 genes (Yeaman et al. 2016); and 2)  SNP-environment associations in the same dataset. The loblolly pine reference genome was used for alignment of contigs from the sequence capture data. Because it (and other conifer genomes) is not well-assembled into few scaffolds, the physical linkage of contigs from sequence capture is not known. However, SNPs on the array were selected in the same manner based on GWAS analysis for all phenotypic traits, so highly contrasting LD patterns for growth, cold injury and phenology traits are likely to reflect biological differences in genomic architecture.    Counts of positive effect alleles for growth initiation and cessation SNPs are highly variable (~30 to ~500 positive effect alleles per genotype: maximum n = 634) among individual seedlings, but multi-modal frequency distributions suggest variation is also partitioned among individuals or populations (Figure 4.1d and e). These multi-modal patterns correspond to clusters of SNPs that have much greater LD than background levels, or the levels of LD for growth trait- and cold injury-associated SNPs. Loss or gain of one allele in these clusters has approximately the same effect across ~150 SNPs (Figure 4.2c and d, and Figures C5 and C6), resulting in multi-modal distributions (Figure 4.1d and e) that have evenly spaced peaks separated by ~150 SNPs. Given that LD between and within genes decays rapidly, on 75  average in huge conifer genomes (Pavy et al. 2012; Jaramillo-Correa et al. 2010), these results suggest large clusters of phenology-associated SNPs in high LD for each trait originate from the same genomic region that has experienced strong directional selection and relatively low levels of recombination. Mean r2 values of these two phenology traits are consistent with estimates of within-gene LD for several conifer species (Pavy et al. 2012). They also reflect my genecology results from chapter 2 (Figure 2.2d). Traits with greater proportions of tightly linked SNPs in the top 1% of GWA SNPs have less phenotypic variation and stronger genetic control. Growth initiation in particular has low phenotypic variation among breeding populations (Table 2.2), and modest to high heritability (Howe et al. 2003). These results also show that although bud set necessarily precedes the development of cold hardiness, growth cessation and cold injury associated SNPs have sharply contrasting genomic architectures with only 23% overlap in associated SNP identities (Figure 4.7). This limits the possible extent of pleiotropy among genes associated with growth cessation and cold injury, supporting suggestions that even though these two traits are temporally related, they may also exhibit considerable functional independence (Tanino et al. 2010; Rohde et al. 2011). Using the approach of Turchin et al. (2012) to count positive effect alleles and calculate the frequency of these alleles within individuals, I detected small but systematic shifts in the frequency of individuals with more positive effect alleles from across the top 1% of candidate SNPs associated with each trait. These allele frequency shifts among individuals were subtle and highly polygenic, but similar genomic architectures estimated from height, growth rate, shoot mass and cold injury candidate SNPs identified in independent GWAS analyses, suggest this method is reliable and sensitive. These responses to selection may have gone undetected using single-locus analyses when among individuals, different suites of loci could be responding to selection.  Systematic shifts in the frequencies of positive effect alleles across many loci fit quantitative genetic expectations that if responses to recent selection are detectable, they will be associated with small allele frequency shifts at many loci of small effect (Stephan 2016). However, shifts in the mean frequency of positive effect alleles from natural to selected seedlings (Table 4.1) were an order of magnitude greater than equivalent shifts in mean LD values for SNPs associated with each trait (Table 4.2). This may contradict expectations that associations among loci, including LD, will drive initial responses to selection more than allele frequency differences (Le Corre & Kremer 2012), but it is consistent with Namroud et al. (2012) who assessed the effects of artificial selection in Picea glauca. Assuming a small number of different loci respond to selection among individuals, then LD could be insensitive to selection if responsive loci are not consistently linked; frequencies of positive effect alleles 76  would give a more sensitive response to polygenic selection. It may also be possible that lodgepole pine breeding programs which intentionally minimise inbreeding and loss of variation while increasing the genetic worth of seedlots through selection, maintain sufficient population sizes and outcrossing levels to keep LD low.   4.4.2 Effects of Selection Among and Within Breeding Zones Among breeding zones, small to modest variation in the mean frequency of positive effect alleles is associated with climatically adaptive genomic differentiation. This variation is polygenic and associated with many loci of small effect, yet adaptive genomic variation among breeding zones is strong and can be accurately detected from the frequency of positive effect alleles (Figure 4.3). The frequency of positive effect alleles is more sensitive to climatically-adaptive genomic variation than single-locus approaches that have often used much larger climatic gradients to detect variation among populations in the frequency of adaptive alleles (e.g., Holliday et al. 2010; Eckert et al. 2009; Chen et al. 2012). The effects of selective breeding on clines in the frequency of positive effect alleles with climate are weak and non-significant. Even so, where pairs of natural and selected clines are individually significant for each combination of trait-associated SNPs and climate, selected seedling clines are always stronger (Figure 4.3, Table C.2). It appears that selection, breeding and testing within lodgepole pine breeding zones generate genomic responses which confer stronger adaptive genomic relationships with climate among breeding zones, particularly for temperature-related climate variables. This finding broadly reflects the responses of phenotype-climate clines to selective breeding described in Chapter 2 (Table A.5), which are strongly correlated with genotype-climate clines in both seedlings types (Figure 4.4), but there are exceptions. For growth traits, slopes of selected versus natural seedling phenotypic clines with mean annual temperature (Figure 2.3a, b and c), are much greater than the equivalent positive effect allele frequency-climate slope increases. Similarly, when climate variables are summarised as principle components, phenotypic clines for height with climate PC 1 increase in strength from r2 = 0.67 to 0.87 (Table 4.3), but clines in height-associated positive effect allele frequency with PC 1 are virtually identical between seedling types (Figure 4.3). This comparison suggests that a portion of the phenotypic height growth response in relation to climate that I observed for selected seedlings in Chapter 2 is not detected in this genomic analysis. Relative to the whole conifer genome, this study examined a small proportion of SNPs associated with polygenic growth, phenology and cold injury traits. Of approximately one million SNPs analysed (Yeaman et al. 2016), ~50,000 were included on the SNP array, and ~37,000 passed genotyping 77  quality filters. Unexplained phenotypic variation, analogous to the missing heritability problem, was expected. The initial sequence capture and SNP array design processes that included a 5% minor allele  frequency requirement, statistical limitations of GWAS, call rate and minor allele filtering of the genotyped candidate SNP, as well as the 1% cut-off for candidate SNPs that I analysed for positive effect allele frequencies, means that many adaptive SNPs may remain undetected. Using the 1% cut-off for associated SNPs may be somewhat arbitrary, but for traits that are highly polygenic, varying this 1% cut-off value should have little effect on the ability to detect positive effect allele frequencies. My results suggest the 1% cut-off value is not overly conservative. Variation in the frequency of positive effect alleles associated with the top 1% of SNPs for each trait explains a moderate to large proportion of the respective trait’s phenotypic variation among breeding zones (Figure 4.5).  Positive effect allele frequency-phenotype relationships were affected by selection differently for the six traits. The amount of phenotypic variation in cold injury (Figure 4.5f) explained by positive effect alleles was almost identical between seedling types. By contrast, positive effect allele regression slopes for growth trait phenotypes were all stronger in selected seedlings. Three possible explanations for stronger slopes in selected seedling growth trait clines are 1) Positive effect alleles with greater effect sizes constitute more of the phenotypic selection response in breeding zones that also have greater shifts in positive effect allele frequencies; 2) Correlations among loci have a greater effect in breeding populations with higher frequencies of positive effect alleles; 3) Non-genetic mechanisms underlie a portion of the phenotypic growth response to selective breeding. Of these explanations, the first seems most likely, while evidence to reject the second comes from the observation that growth traits and cold injury have extremely similar genetic architectures in natural and selected seedlings, but divergent phenotypic selection responses. In Chapter 2.4.3 I indirectly evaluated the possible reasons for growth gains in selected lodgepole pine seedlings and concluded phenotypic plasticity and epigenetic responses to selection (explanation three) were likely to be small. Genomic results suggest phenotypic plasticity and epigenetic responses are possible, but do not confer a large proportion of the selection response.  Within breeding populations, selection has generated small but detectable shifts in the frequencies of positive effect alleles that are consistent enough to be correlated with phenotypic responses to selection among breeding zones (Figure 4.6). This has two important implications: 1) The frequency of positive effect alleles is sensitive enough to detect polygenic selection within populations, even after just one or two generations of selection and breeding; and 2) sufficient detectable genomic variation underlies phenotypic variation for their fine scale responses to be correlated when many 78  candidate loci from across the genome are genotyped. Point two is further evidence that non-genetic components of the phenotypic response to selection must be relatively small. The genomic basis of adaptive variation within breeding zones is also revealed by correlations of phenotypic versus positive effect allele frequency differences. A 3% increase in positive effect allele frequency is associated with a phenotypic increase in cold injury of ~4%, but growth traits experience much larger increases of ~120% (Figure 4.6a, b and c). This implies that the cold injury alleles with a correlated response to selective breeding for growth are much smaller in effect size and cumulative effect relative to increases in growth trait alleles that respond to direct selection on the trait.  4.4.3 Prospects for Long-Term Genetic Gain and Assisted Gene Flow Conifer genomes, including those of lodgepole pine, are characterised by their very large size, owing to extensive intergenic regions, and an abundance of very long introns and repeat-rich repetitive elements (De La Torre et al. 2014c). While recombination rates and nucleotide diversity in coding regions of the conifer genome are low, they have low linkage disequilibrium and high levels of heterozygosity attributable to large effective populations sizes and high levels of outcrossing (Jaramillo-Correa et al. 2010; Savolainen & Pyhajarvi 2007; Pavy et al. 2012).  Preliminary analyses using population genetic measures of diversity (He and Ho) and inbreeding (FIS) comparing selected and natural seedlots show no effects of one or two generations of selection (results not shown), which is consistent with the findings of Namroud et al. (2012). Using genome-wide samples of candidate SNPs from across the lodgepole pine genome that are most strongly associated with climatically adaptive phenotypic traits, I have found selective breeding has small effects distributed across many loci. It appears that there is no detectable loss of genomic variation across loci that would compromise future gains in growth, or the climatically adaptive synchrony of phenology and cold injury traits. Small polygenic effects of selection in the large conifer genome are negligible compared to the highly inbred lines of domesticated crop species which still retain sufficient genomic variation to yield large, long-term selection responses in desirable polygenic traits (Hamblin et al. 2011), e.g., the University of Illinois long term maize selection trial for kernel oil and protein content (Moose et al. 2004).  Cold hardiness in temperate and boreal conifers has repeatedly been identified as the strongest locally adaptive trait in relation to climate (Howe et al. 2003). In this chapter I have found that cold injury has polygenic genomic variation among individuals (Figure 4.1f) and among breeding zones (Figure 4.5f) that is equivalent to growth traits, but the genotype-phenotype-climate relationships of 79  cold injury remain largely unchanged in selectively bred seedlings because of this trait’s adaptive importance. Evidence from Chapters two and three suggests that adaptive trade-offs between growth gains and both growth phenology and cold hardiness will to some extent constrain future growth gains that can be achieved within breeding populations.  4.5 Conclusions: A Genomic Basis for Assisted Gene Flow Strong clines in the frequency of positive effect alleles with climate suggest small genotypic changes among populations confer adaptive divergence among populations. For example, both height and cold injury associated allele frequencies vary by ~0.15 among breeding zones and between seedlings types. This genotypic variation is associated with 6oC of variation in mean annual temperature (Figure 4.3), 20cm variation in height (~50% of the minimum height), and ~ 40% variation in cold injury. The implications of this are two-fold. Firstly, the frequency of positive effect alleles within breeding zones may be a suitable summary statistic to characterise climatic adaptation, track adaptive lags of populations and associate populations with future climates, and secondly, assisted gene flow will be necessary to match genotypes with future climates. Previously observed and predicted future climate changes are equivalent to differences in breeding zone mean annual temperatures that produce detectable genotypic divergence for both natural and selected seedling types. However, the effects of selective breeding for growth on the genotypic associations with both phenotypic traits and climatic variables are negligible for phenology and cold injury. This accurately reflects findings from phenotypic analysis of the same traits in Chapter 2. At a genomic level, the climatically adaptive synchrony of growth phenology, dormancy and cold hardiness development is maintained in selectively bred lodgepole pine seedlings, providing further evidence that the same assisted gene flow prescriptions can be applied equally to natural and selectively bred seedlots. Although the strength of correlations between genotype-climate and phenotype-climate associations is strong and almost identical between seedling types, in selected seedlings, phenotype-climate associations explain slightly more variation in the associations between traits and climate than genotype-climate associations (Figure 4.4). For assisted gene flow of selected seedlots, this intuitively means phenotype-climate associations are a more accurate basis to characterise climatically associated variation in adaptive traits than genotype-climate associations. However, climatic PC 1 explains large proportions of variation in the frequencies of positive effect alleles among breeding zones for both seedling types (Table C.2). These genotypic r2 values are always greater than r2 values of the equivalent phenotypic trait associations with PC 1 (Table A.5). This suggests genotype-climate PC 1 associations 80  may be the most suitable way to characterise the climatic adaptation of breeding populations or even individual seedlots, independently from long-term provenance trials and non-genetic effects that can cause phenotypic variability. Given the increasing simplicity and decreasing costs of genotyping, and the extraordinary effort required to phenotype climate-related traits in large numbers of seedlings, genotyping may offer a rapid means of profiling climate adaptation in breeding populations. Lastly, Figure 4.4 suggests there is a distinct cluster of strong genotype-climate associations with temperature-related variables (identified from Table C.2), that would be an informative basis for assisted gene flow prescriptions.    81  4.6 Tables     Associated SNP Phenotype Mean Number of Positive Effect Alleles Natural Selected Δ Mean  Δ Frequency  W-statistic p-value Height 259.6 (1.45) 267.4 (1.27) 7.8 (1.92) 0.0123 476210 < 0.0001 Growth Rate 280 (1.31) 289 (1.2) 9.1 (1.78) 0.0144 466705.5 < 0.0001 Shoot Mass 279.2 (1.29) 286.1 (1.09) 6.9 (1.67) 0.0109 474610 < 0.0001 Growth Initiation 131.8 (3.19) 142.1 (2.9) 10.3 (4.32) 0.0162 486297 0.0002 Growth Cessation 159.9 (3.34) 172.1 (3.1) 12.2 (4.58) 0.0192 492539 0.0013 Cold Injury 282.3 (1.67) 291.5 (1.46) 9.2 (2.21) 0.0145 476101 < 0.0001  Table 4.1 Mean number of positive effect alleles for the top 1% (n = 317) of candidate SNPs associated with each trait in both seeding types. The differences between seedling types means are given with their associated Wilcoxon rank sum test statistics (W-statistic). p-values are significant at a Bonferroni adjusted cut-off of α = 0.0083 for six comparisons. Δ frequency is the difference in frequency of positive effect alleles between seedling types calculated as Δ mean/(317 x 2). Standard errors are given in brackets.  82    SNP Phenotype Mean Pairwise Linkage Disequilibrium Natural Selected Δ r2  r2 p ≤ 0.05 ∝ n r2 p ≤ 0.05 ∝ n Height 0.0209 (0.0005) 0.312 0.0225 (0.0005) 0.415 0.0015 (0.0007) Growth Rate 0.0202 (0.0005) 0.275 0.0217 (0.0005) 0.369 0.0015 (0.0007) Shoot Mass 0.0151 (0.0004) 0.274 0.0154 (0.0004) 0.351 0.0003 (0.0006) Growth Initiation 0.2552 (0.0017) 0.657 0.2579 (0.0017) 0.603 0.0026 (0.0024) Growth Cessation 0.2429 (0.0017) 0.736 0.2504 (0.0017) 0.754 0.0074 (0.0024) Cold Injury 0.0226 (0.0005) 0.460 0.023 (0.0004) 0.496 0.0004 (0.0006)  Table 4.2 Pairwise linkage disequilibrium summarised as the means of squared allelic correlations among the top 1% (n = 317) of candidate SNPs associated with each trait for both seedling types. The proportion of significant (p ≤ 0.05) r2 values is given for n = 50,086 pairwise comparisons per set of trait associated SNPs and seedling type. The background pairwise LD of all SNPs on the array was r2 = 0.0016 (se = 9.5 x10-5), calculated from 250 random samples of 317 SNPs and 1000 individuals. Standard errors of means are given in brackets. 83  4.7 Figures Figure 4.1 Probability densities for the frequency of individuals with a given number of positive effect alleles from the top 1% (n = 317) of candidate SNPs associated with each trait in both seeding types. Wilcoxon rank sum tests for significant differences between the mean number of positive effect alleles between seedling  type are significant at a Bonferroni adjusted cut-off value of α = 0.0083 for six comparisons.  84    Figure 4.2 Contrasting heat maps of pairwise linkage disequilibrium (r2) for the top 1% SNPs associated with two traits, height and growth initiation. Panels represent SNPs for a)height in natural seedlings, b) height in selected seedlings, c) growth initiation in natural seedlings, and d) growth initiation in selected seedlings For both traits, LD heat maps differ little between natural seedling and selected seedling SNPs. 317 SNPs are hierarchically clustered on both axes of each plot. Larger versions of these heat maps are given in Figures C1 and C5. 85    Figure 4.3 Clines in the frequency of effect alleles for the top 1% candidate SNPs of four traits with four climatic variables. None of the slopes differ significantly between seedling types. r2 and p-values for all six traits and 11 climate variables are given in Table (C1). Points represent the mean climatic and allele frequency values for each breeding zone.  86    Figure 4.4 Correlations between genotype-climate and phenotype-climate clines. Correlations are calculated by seedling type between frequency of positive effect alleles (genotype)-climate cline r2 values (Figure 4.2 and table C1), and phenotype-climate cline r2 values from Chapter 2 (Figure 2.3 and Table A.5). Each point represents the genotype-climate and the phenotype-climate relationship for one climate variable and one phenotype. 87    Figure 4.5 Phenotypic variation in each trait and both seedling types explained by the mean frequency of effect alleles from the top 1% of phenotype associated SNPs in each breeding zone. Phenotypic values are best linear unbiased estimates of breeding zone means calculated in Chapter 2.3.3 (Table A.4). p-values are significant at a Bonferroni adjusted cut-off value of α = 0.0083 for six comparisons. 88    Figure 4.6 Correlations of within-breeding zone phenotypic and genotypic difference between seedling types. Phenotypic differences are between natural and selected seedlings’ best linear unbiased estimates of breeding zone means calculated in Chapter 2.3.3. Genotypic differences between seedlings types refer to the frequency of positive effect alleles from top 1% SNPs associated with each trait. None of these correlations are significant at a Bonferroni adjusted cut-off value of α = 0.0083 for six comparisons. Each of these correlations also has a low, outlying phenotypic value, but this is not a systematic difference  or error. Among correlations these outliers come from three different breeding zones. 89    Figure 4.7 Venn diagram of the overlap between the top 1% of candidate SNPs associated with height, growth cessation and cold injury, separately identified from GWA analyses of natural seedlings. 90  Chapter 5: Conclusions  5.1 Introduction Shifting climates are dissociating forest tree populations from historical climatic optima and threaten mid- to long-term forest productivity. This is predicted to accelerate in coming decades. To mitigate these effects, assisted gene flow policies are being developed to redeploy forest genetic material and match locally adapted populations with new climates.   Within populations, selective breeding has the potential to modify trade-offs among correlated, climatically adaptive traits. In addition to climate change, selective breeding adds an extra layer of complexity and uncertainty to our understanding of climatic adaptation within and among populations of forest trees. However, implementing accurate assisted gene flow policies to redeploy populations of selectively bred forest trees requires an accurate knowledge of the strength and nature of local adaptation to climate.  The primary objectives of my thesis were to describe how selective breeding within conifer populations modifies climatically adaptive variation, relative to their natural counterparts, and evaluate this in the context of assisted gene flow. Focussing on lodgepole pine and interior spruce, the two most planted conifers in western Canada, I decomposed the relationships among climatically adaptive phenotypic traits, their associations with hybrid ancestry in spruce, and the genomic architectures associated with these traits in pine. I then compared phenotype-genotype-climate relationships between natural and selected seedlings to identify adaptive trade-offs associated with selection and breeding for increased height growth. These analyses were based on phenotypic and genotypic data collected from two parallel pine and spruce seedling common garden experiments, in addition to high-resolution climate data for population origins, interpolated from ClimateNA (Wang et al. 2016). The phenotypic data in this thesis summarised > 300,000 individual observations or measurements (pine and spruce combined) collected over two growing seasons from ~2200 seedlings of each species. These seedlings  were also genotyped on the AdapTree lodgepole pine and interior spruce ~50K SNP arrays containing candidate SNPs for growth and climate adaptation selected from over 1 million SNPs. Integrating precision phenotyping, genome-wide markers, and multivariate climatic data provided a powerful basis to both detect adaptive variation and assess the effects of selective breeding.  91  5.2 Selective Breeding and Climatic Adaptation Selective breeding has generated height growth increases in both species that reflect the estimated genetic gains for growth from their respective breeding programs. Among breeding zones, the strength of associations between growth and temperature-related climate variables increased with selection effort. This effect was much stronger in pine than spruce, but in both species it corresponds to the partitioning of greater phenotypic variance among versus within breeding zones for growth traits of selected seedlings.  Stronger associations between climate and growth in selected than in natural pine populations occur primarily because climatically favourable breeding zones have the tallest seedlings in natural populations and have achieved the greatest genetic gains in selected populations. As expected, the genomic architecture of pine growth traits is highly polygenic. I detected subtle genomic responses to selective breeding across many loci in both the frequencies of positive effect alleles and the average strength of LD, similar to Namroud et al. (2012). Variation in positive effect allele frequencies of growth-associated SNPs among breeding zones corresponds to the strong adaptive divergence among these populations in relation to temperature-related climate variables. These genotype-climate relationships are stronger for selected than natural seedlings, but differ less between seedling types than the corresponding phenotype-climate associations, suggesting that not all of the phenotypic response to selective breeding has a genetic basis, or that I did not identify all loci underlying these traits.  In interior spruce, unlike pine, growth gains are relatively similar among breeding zones. This is evidenced by the small differences in slope between seedling types in clines of growth traits along temperature gradients. Some spruce breeding zones have both substantial height growth gains and shifts in hybrid ancestry towards faster growing Picea glauca genotypes in selected seedlings, but these effects are not common to all breeding zones. If subtle shifts in the positive effect allele frequencies associated with spruce growth traits occur following selective breeding, as they do in pine, it is unclear whether they will have strong relationships with hybrid ancestry, even though this ancestry broadly characterises adaptation to climate across the hybrid zone.  Growth gains could compromise adaptation in selected seedlings if the synchronisation of growth phenology with local seasonal climatic transitions is disrupted by breeding. I have found that growth initiation and growth cessation have different responses to selection. Growth initiation (bud break in spruce) differs very little between seedling types within pine and spruce breeding zones, while the adaptive phenotypic and genotypic associations with climate for this trait do not appear compromised by selective breeding. By contrast, growth cessation (bud set in spruce) is delayed in 92  selected seedlings for all but one breeding zone across both species, suggesting longer growing season duration contributes a portion of observed growth gains. However, delayed growth cessation doesn’t appear to compromise the development of cold hardiness in either species. The long-held view is that cold hardiness development in woody plants occurs sequentially in three stages: 1) growth cessation triggered by increasing night length; 2) dormancy induction by exposure to cool temperatures and mild freezing; and 3) deep cold hardening induced by exposure to extreme low temperature below ~-30oC (Weiser 1970). Several recent studies have challenged this view, suggesting that while increasing late-summer night length drives growth cessation, the development of cold hardiness can be driven by temperature more than photoperiod and may be somewhat decoupled from growth cessation (Tanino et al. 2010; Hamilton et al. 2016; Rohde et al. 2011). My phenotypic results from both species indicate the adaptive synchrony of these two processes respond flexibly to selection. In pine, contrasting patterns of LD and relatively low overlap in associated SNPs between growth cessation and cold injury also suggest the extent of pleiotropy among their respective genes is limited. It is possible that selective breeding shifts some control of growth cessation and cold hardening towards temperature- mediated climatic cues, rather than photoperiod. Locally adaptive autumn cold hardiness reflects the need for seedlings to acquire dormancy and minimise the risk of cold injury prior to damaging freezing events. As the strongest locally adaptive trait in temperate and boreal conifers, relationships among cold hardiness phenotypes, genotypes and temperature-related climate variables were strong in the natural seedlings of both species. These relationships were always maintained in selectively bred seedlots and often slightly stronger than in natural seedlings. Cold hardiness is highly conserved in populations within the pine and spruce breeding programs, and antagonistic effects of selective breeding for growth on cold injury were small or absent. However, the pine CP low breeding zone may be an exception and suggests that cold hardiness needs to be carefully managed in this breeding population. From CP low I sampled three selectively bred seedlots. Cold injury in selected pine seedlings from CP low was 11% higher, on average, than in natural seedlings. This corresponds to a 4% increase in the frequency of positive effect alleles associated with cold injury. It is equivalent to populations from source climates with mean annual temperatures ~1.5oC warmer, and extreme minimum temperatures that are ~3oC warmer, and a shift of ~25% of the range in extreme minimum temperature across all pine breeding zones. In the other 25 (pine and spruce combined) breeding zones, selective breeding appears to have optimised growth gains within the constraints of local climates and adaptive cold hardiness. My findings suggests future height gains are 93  more likely to be limited by the the need to maintain adaptive cold hardiness in selected seedlings than by a lack of genomic variation for growth. At both the phenotypic and genomic levels, selective breeding has generated phenotypic combinations that will increase productivity over natural populations and remain adapted to local climates. The phenotypes of selected seedlings are not simply equivalent to those of populations from warmer climates with similar growth, because phenological synchrony and cold hardiness are both largely maintained. Breeding programs have achieved this by selecting, breeding and testing genotypes within the limits of local climates.  5.3 Assisted Gene Flow in Selected Seedlots Assisted gene flow (AGF) represents one component of a broader strategy to mitigate the negative effects of climate change on Canadian forests (Pedlar et al. 2011; Gauthier et al. 2014). In this context, assisted migration is implemented to maintain the climatic adaptation of forest stands that are part of the harvestable timber resource, rather than promoting the long-term evolutionary adaptation of natural forests. My results show that in pine and spruce, selective breeding programs are compatible with AGF policies.   Strong phenotypic and genotypic clines with temperature-related climate variables indicate relatively large proportions of adaptive variation in pine and spruce are partitioned among versus with populations in relation to climate, leading to adaptive differentiation among populations. Close phenotype-genotype-climate associations within breeding zones mean that AGF will be necessary to redeploy selectively bred populations and match them with anticipated future climates. Selective breeding is compatible with the objectives of AGF because it produces seedlings that grow vigorously, but avoids antagonistic trade-offs among climatically adaptive traits under current climates. This means that the same or similar AGF prescriptions can be applied to natural stand and selectively bred seedlots.  AGF prescriptions are often determined using phenotypic data, primarily for height growth, collected from long-term provenance trials. These trials are a valuable resource, but limited by the economic investment and time required for establishment and data collection. My pine genomic results show that the frequencies of positive effect alleles associated with all traits have strong relationships with temperature-related climate variables of the source populations, and particularly climate PC 1. The frequency of positive effect alleles is sensitive to adaptive climatic divergence among populations, as well as divergence between natural and selected seedling types within breeding zones. Using this method to summarise polygenic genotypes and their responses to selection, it appears that genotypes 94  associated with multivariate climate distance metrics such as the one proposed in the BC Ministry of Forests, Lands and Natural Resources Operations climate-based seed transfer framework, may be a more suitable basis for tracking adaptive climatic lags and determining AGF prescriptions, than current phenotype-based methods. This genotype-based method could offer rapid, accurate assessment of climatic adaptation in alleles associated with specific climate-risk phenotypes and avoid sources of error due to non-genetic variance within and among field trials. It might also be directly applicable to wild-stand trees to provide an accurate profile of climatic adaptation and growth potential that is unbiased by local environmental variation.  5.4 A Domestication Syndrome in Conifers? Intense direct and indirect selection has modified traits and generated trade-offs among desirable or adaptive phenotypes that define the domestication syndrome in crop plants (Doebley et al. 2006). The genome-wide effects of domestication in crop species are characterised by population genetic bottlenecks, selective sweeps in genomic regions under selection and a prevalence of single-gene traits, associated with high selection intensities applied over hundreds of generations (Wright et al. 2005; Gross & Olsen 2010).  Conifers are selectively bred for improved timber yields and wood quality traits in forest plantations. Unlike domesticated agricultural crops, conifers have long generation times, which mean they must be adapted and resilient to local climatic conditions over the rotation of a forest stand. It also means breeding cycles are very slow. Conifers are relatively undomesticated with only a few cycles of breeding and testing, even in the most advanced breeding programs (Neale & Kremer 2011). Population genetic diversity seems to be unaffected by selective breeding, even though rare alleles are lost (Schmidtling et al. 1999; Godt et al. 2001; Hansen 2008), putatively adaptive alleles associated with highly polygenic traits have slight frequency increases (Namroud et al. 2012), and levels of LD within genes are comparable to natural populations (Chapter 4).  Despite suggestions that conifers can exhibit a domestication syndrome after a single generation of selection (Santos-del-Blanco et al. 2015), the results from my thesis suggest this is not the case. The phenotypic effects of selection in lodgepole pine and interior spruce do not produce trade-offs among traits that compromise timber yields or adaptation within breeding populations. Small allele frequency shifts within breeding zones are detectable, but attributable to only a few alleles per individual, on average, in the candidate SNPs associated with each of the traits I studied. This suggests adaptive diversity is not being compromised at the molecular level by selective breeding. Strong shifts in LD for 95  loci under selection are also absent. Based on my results and the criteria of Meyer et al. (2012), selectively bred conifers do not exhibit a crop-like domestication syndrome. Growth gains are observed in pine and spruce breeding programs, but they are not outside either species range of natural variation; obvious phenotypic differentiation and trade-offs that distinguish domesticates or varieties are absent; and selectively bred trees are still reproductively integral with wild stand counterparts.  5.5 Study Limitations Combining a large amount of phenotypic, genomic and climatic data has given me a robust basis to evaluate adaptive divergence among and within populations of lodgepole pine and interior spruce in response to selective breeding. However, there are imitations to my approach associated with experimental design and sampling, as well as genome-wide analyses of molecular markers in large conifer genomes.  Growing seedlings in mild coastal common gardens with ample water availability allows comparisons among populations from a wide range of source climates at a vulnerable life stage, minimising climate related seedling damage and mortality. It also allows high-intensity phenotyping of traits that would not be possible and difficult to assess in field trials. However, a single site beyond the climatic range of any sampled populations means that it is impossible to quantify genotype-by-environment interactions that could be a source of within and among population variation. Relaxed selection under favourable conditions may result in reduced phenotypic variation among populations, and phenotypes associated with extreme source climates may be poorly represented (Campbell 1986). Short-term seedling trials may escape exposure to climatic extremes at the test site, and long-term trials offer better indications of lifetime growth performance and fitness. Although the phenotypic results of my thesis correspond to expectations from provincial field trials and gains within breeding programs, the phenotypic results from my thesis require field validation before they can become more broadly applicable. Using a partial (~70%) replicate of my pine and spruce experiments, I established this validation trial in 2013 on a representative site in the central interior of BC at the UBC Alex Fraser Research Forest, but phenotypic data are yet to be collected.   From an experimental sampling perspective, it would be preferable to have a balanced sampling of natural stand seedlots from within each breeding zone. Zones where few natural seedlots were sampled may have phenotypes and genotypes that poorly represent the whole breeding zone. Three breeding zones of each species had less than six natural seedlots sampled. Similarly, only one or two selected seedlots were sampled per breeding zone.  Patterns of genetic gains in growth corresponded to 96  expectations, but within breeding zones my sampling was insufficient to characterise phenotypic or genotypic variability among selected seedlots produced in different years or from different seed orchards. Quantifying this variation may be valuable, to help determine the AGF prescriptions that should be applied to breeding populations.  The genomic data used in my thesis are comprehensive, summarising SNPs with the strongest phenotypic and climatic associations from an initial pool of more than 1 million candidate adaptive SNPs from the respective species sequence captures (Suren et al. 2016). This has offered a powerful lens to detect fine-scale genomic responses to artificial selection. Even so, at all steps of the sequence capture, SNP array design and association analyses there was potential for SNPs associated with adaptive but highly polygenic traits to go undetected. The effects of many adaptive SNPs that underlie the growth, phenology and cold injury phenotypes that I studied will not be accounted for. Those SNPs that were associated with adaptive phenotypes had substantial differences in the patterns of LD among traits, but I was unable to dissect this variation. Construction of a pine linkage map is currently underway. It will be the future basis to dissect relationships among strongly linked loci that are associated with highly conserved phenology traits which determine the synchronisation of growth and dormancy with local climates.   5.6 Future Research Directions This thesis provides a detailed appraisal of the relationships among phenotypes, genotypes and climate for selectively bred lodgepole pine and interior spruce seedlings. Relative to natural stand seedlings, these relationships strengthen with breeding for growth traits, but growth phenology and cold injury phenotypes are not compromised as a result of faster growth because provincial tree improvement programs select, breed and test genotypes of sites that are representative of local climates. The testing phase of this process is critical because it constrains gains for growth within the limits of local climates and adaptive cold hardiness.  Essential adaptive phenotypes associated with drought stress and pathogen resistance were not evaluated in my thesis. It is feasible that direct selection for growth will indirectly select genotypes that confer drought or pathogen resistance, but this cannot be assumed. Height growth is a composite of many other traits including drought and pathogen resistance. If the negative effects of drought and pathogen infection cause mortality or reduced height gains in susceptible selected genotypes during progeny testing, they will be culled from breeding programs. Assuming progeny test sites are climatically representative and provide the opportunity for pathogen exposure, selective breeding programs may 97  confer a degree of drought or pathogen resistance. Quantifying this resistance would be a valuable next step from my research, to understand whether such resistance buffers drought and pathogen stress in selected genotypes, and evaluate the implications for assisted gene flow prescriptions.  The weighted mean parent tree climate data I used to characterise selected seedlings are limited by incomplete data on paternal contributions and do not account for possible pollen contamination from local sources or nearby seed orchards. Repeating phenotype-climate and genotype-climate analyses using mean progeny test site climates of seed orchard parent trees may better represent local conditions that selected genotypes are adapted to. It would also help evaluate biases in test site selection relative to local climates across zones, and test the possibility that differences between test site and local climates drive gains in growth traits.  The genomic results in my thesis raise many further questions and analysis possibilities. A direct continuation of my analyses that use positive effect allele frequencies to characterise polygenic selection responses would be to assess variability in the identities of SNPs responding to selection. Only a few of the top 1% of phenotype-associated SNPs I analysed for each trait responded to selection. Knowing if the same SNPs respond to selection would indicate whether adaptive genomic responses to selection are predictable among individuals or populations, and if specific loci can be targeted to characterise the effects of selective breeding. Identifying and characterising genes associated with the top 1% candidate SNPs would be a further extension of this. Firstly, to determine the function of genes represented by large proportions of phenology SNPs that have strong pairwise LD, and secondly, to determine if the phenotype-associated SNPs responding to selection have a similar functional basis.   Strong associations between positive effect allele frequencies and temperature-related climate variables suggest that adaptation to climate may be better characterised at the genomic level across a reasonably large number of adaptive loci, than by using phenotypes. This finding has importance for the implementation of AGF, although validation using phenotypes assessed in a climatically representative test site will be necessary. It creates the possibility that positive effect allele frequencies which are tightly linked to current climates can be accurately mapped on the landscape and used to redeploy selectively bred seedlots to match future climates. Identifying independent suites of SNPs associated with each trait suggests that genotypes could be matched with future climates on a trait-by-trait basis, then used to quantify trade-offs between the cold injury risks to seedlings transferred under current climates and the potential growth gains as climate warms over the rotation of a forest stand. Lastly, because positive effect allele frequencies provide a sensitive method to accurately characterise climatic associations independently from phenotypic variation, this method may be used in natural forest stands 98  to assess the portfolio of climatically adaptive standing genetic variation. This represents a solid basis to evaluate the capacity of natural forest for rapid evolutionary adaptation to new climates.  99  References Aitken, S.N. & Bemmels, J.B., (2015). Time to get moving: assisted gene flow of forest trees. Evolutionary Applications, 9, pp.271–290. Aitken, S.N. & Hannerz, M., (2001). Genecology and gene resource management strategies for conifer cold hardiness. In F. J. Bigras & S. J. Colombo, eds. Conifer Cold Hardiness. Klewer Academic Publishers, pp. 23–53. Aitken, S.N. & Whitlock, M.C., (2013). Assisted gene flow to facilitate local adaptation to climate change. Annual Review of Ecology, Evolution, and Systematics, 44(1), pp.367–388. Aitken, S.N., Yeaman, S., Holliday, J.A., Wang, T.L. & Curtis-McLane, S., (2008). Adaptation, migration or extirpation: climate change outcomes for tree populations. Evolutionary Applications, 1(1), pp.95–111. Alberta Environment and Sustainable Resource Development, (2011). Sustainable Forest Management: Current Facts and Statistics 2011. Alberta Environment and Sustainable Resource Development. Alberta, Canada. Alberta Forest Genetic Resources Council, (2015). Alberta Forest Genetic Resources Council 2014-2015 Annual Report. Alberta Forest Genetic Resources Council, Alberta, Canada. Alberto, F.J., Aitken, S.N., Alía, R., González-Martínez, S.C., Hänninen, H., Kremer, A., Lefèvre, F., Lenormand, T., Yeaman, S., Whetten, R. & Savolainen, O., (2013). Potential for evolutionary responses to climate change - evidence from tree populations. Global Change Biology, 19(6), pp.1645–1661. Alexander, D.H. & Novembre, J., (2009). Fast model-based estimation of ancestry in unrelated individuals. Genome Research, 19, pp.1655–1664. Alexander, R.R., (1987). Ecology, silviculture, and management of the Engelmann spruce - subalpine fir type in the central and southern Rocky Mountains. USDA Forest Service, Agriculture Handbook, No. 659. Allen, C.D., Macalady, A.K., Chenchouni, H., Bachelet, D., McDowell, N., Vennetier, M., Kitzberger, T., Rigling, A., Breshears, D.D., Hogg, E.H. (Ted), Gonzalez, P., Fensham, R., Zhang, Z., Castro, J., Demidova, N., Lim, J.H., Allard, G., Running, S.W., Semerci, A. & Cobb, N., (2010). A global overview of drought and heat-induced tree mortality reveals emerging climate change risks for forests. Forest Ecology and Management, 259(4), pp.660–684. Bansal, S., Harrington, C.A. & St. Clair, J.B., (2016). Tolerance to multiple climate stressors: A case study of Douglas-fir drought and cold hardiness. Ecology and Evolution, 6(7), pp.2074–2083. 100  Bansal, S., St Clair, J.B., Harrington, C.A. & Gould, P.J., (2015). Impact of climate change on cold hardiness of Douglas-fir (Pseudotsuga menziesii): Environmental and genetic considerations. Global Change Biology, 21, pp.3814–3826. Barrett, R.D.H. & Schluter, D., (2008). Adaptation from standing genetic variation. Trends in Ecology & Evolution, 23(1), pp.38–44. Beaulieu, J., Doerksen, T., Boyle, B., Clément, S., Deslauriers, M., Beauseigle, S., Blais, S., Poulin, P., Lenz, P., Caron, S., Rigault, P., Bicho, P., Bousquet, J. & Mackay, J., (2011). Association genetics of wood physical traits in the conifer white spruce and relationships with gene expression. Genetics, 188, pp.197–214. Berg, J.J. & Coop, G., (2014). A population genetic signal of polygenic adaptation. PLoS genetics, 10(8), p.10.1371/journal.pgen.1004412. Bigras, F.J., Ryyppö, A., Lindström, A. & Stattin, E., (2001). Cold acclimation and deacclimation of shoots and roots of conifer seedlings. In F. J. Bigras & S. J. Colombo, eds. Conifer cold hardiness. Klewer Academic Publishers, pp. 57–88. Birol, I., Raymond, A., Jackman, S.D., Pleasance, S., Coope, R., Taylor, G.A., Yuen, M.M. Saint, Keeling, C.I., Brand, D., Vandervalk, B.P., Kirk, H., Pandoh, P., Moore, R.A., Zhao, Y., Mungall, A.J., Jaquish, B., Yanchuk, A., Ritland, C., Boyle, B., Bousquet, J., Ritland, K., MacKay, J., Bohlmann, J. & Jones, S.J.M., (2013). Assembling the 20 Gb white spruce (Picea glauca) genome from whole-genome shotgun sequencing data. Bioinformatics, 29(12), pp.1492–1497. Bond, W.J., (1989). The tortoise and the hare - ecology of angiosperm dominance and gymnosperm persistence. Biological Journal of the Linnean Society, 36(3), pp.227–249. Bräutigam, K., Vining, K.J., Lafon-Placette, C., Fossdal, C.G., Mirouze, M., Marcos, J.G., Fluch, S., Fraga, M.F., Guevara, M.Á., Abarca, D., Johnsen, Ø., Maury, S., Strauss, S.H., Campbell, M.M., Rohde, A., Díaz-Sala, C. & Cervera, M.T., (2013). Epigenetic regulation of adaptive responses of forest tree species to the environment. Ecology and Evolution, 3(2), pp.399–415. British Columbia MFLNRO, (2010). Chief Forester’s Standards for Seed Use. British Columbia Ministry of Forests, Lands and Natural Resource Operations, British Columbia, Canada. British Columbia Ministry of Environment, (2015). Indicators of Climate Change for British Columbia: 2015 Update. Victoria, BC. Buerkle, C.A. & Lexer, C., (2008). Admixture as the basis for genetic mapping. Trends in Ecology & Evolution, 23(12), pp.686–694. Butler, D., (2009). ASReml: ASReml() fits the linear mixed model. R package version 3.0. 101  Campbell, R.K., (1986). Mapped genetic variation of Douglas-fir to guide seed transfer in southwest Oregon. Silvae Genetica, 35, pp.85–96. De Carvalho, D., Ingvarsson, P.K., Joseph, J., Suter, L., Sedivy, C., Macaya-Sanz, D., Cottrell, J., Heinze, B., Schanzer, I. & Lexer, C., (2010). Admixture facilitates adaptation from standing variation in the European aspen (Populus tremula L.), a widespread forest tree. Molecular Ecology, 19(8), pp.1638–1650. Castro, J., (1999). Seed mass versus seedling performance in Scots pine: A maternally dependent trait. New Phytologist, 144, pp.153–161. Chen, J., Gyllenstrand, N., Zaina, G., Morgante, M., Bousquet, J., Eckert, A., Wegrzyn, J., Neale, D., Lagercrantz, U. & Lascoux, M., (2012). Disentangling the roles of history and local selection in shaping clinal variation of allele frequencies and gene expression in Norway spruce (Picea abies). Genetics, 191(July), pp.1–17. Chhatre, V.E., Byram, T.D., Neale, D.B., Wegrzyn, J.L. & Krutovsky, K. V, (2013). Genetic structure and association mapping of adaptive and selective traits in the east Texas loblolly pine (Pinus taeda L .) breeding populations. Tree Genetics & Genomes, 9, pp.1161–1178. Chuine, I., (2010). Why does phenology drive species distribution? Philosophical Transactions of the Royal Society B-Biological Sciences, 365, pp.3149–3160. Chuine, I., Aitken, S.N. & Ying, C.C., (2001). Temperature thresholds of shoot elongation in provenances of Pinus contorta. Canadian Journal of Forest Research, 31(8), pp.1444–1455. Chuine, I., Rehfeldt, G.E. & Aitken, S.N., (2006). Height growth determinants and adaptation to temperature in pines: a case study of Pinus contorta and Pinus monticola. Canadian Journal of Forest Research, 36(5), pp.1059–1066. Coates, K.D., Haeussler, S., Lindeburgh, S., Pojar, R. & Stock, A.J., (1994). Ecology and silviculture of interior spruce in British Columbia. Canadian - British Columbia Partnership Agreement on Forest Resource Development, publication number 220. Canadian Forest Service and BC Ministry of Forests. Cooke, J.E.K., Eriksson, M.E. & Junttila, O., (2012). The dynamic nature of bud dormancy in trees: Environmental control and molecular mechanisms. Plant, Cell and Environment, 35(10), pp.1707–1728. Le Corre, V. & Kremer, A., (2012). The genetic differentiation at quantitative trait loci under local adaptation. Molecular Ecology, 21(7), pp.1548–1566. Critchfield, W.B., (1957). Geographic Variation in Pinus contorta. Maria Moors Cabot Foundation, 102  publication No.3. Harvard University, Cambridge, MA. Cullingham, C.I., James, P.M.A., Cooke, J.E.K. & Coltman, D.W., (2012). Characterizing the physical and genetic structure of the lodgepole pine x jack pine hybrid zone: Mosaic structure and differential introgression. Evolutionary Applications, 5(8), pp.879–891. Daubenmire, R., (1974). Taxonomic and ecologic relationships between Picea glauca and Picea engelmannii. Canadian Journal of Botany, 52(7), pp.1545–1560. Davis, M.B. & Shaw, R.G., (2001). Range shifts and adaptive responses to Quaternary climate change. Science, 292(5517), pp.673–679. Doebley, J.F., Gaut, B.S. & Smith, B.D., (2006). The molecular genetics of crop domestication. Cell, 127, pp.1309–1321. Dormling, I. & Johnsen, Ø., (1992). Effects of parental environment on full-sib families of Pinus sylvestris. Canadian Journal of Forest Research, 22, pp.88–100. Eckert, A.J., Bower, A.D., Wegrzyn, J.L., Pande, B., Jermstad, K.D., Krutovsky, K. V, Clair, J.B.S. & Neale, D.B., (2009). Asssociation Genetics of Coastal Douglas Fir (Pseudotsuga menziesu var. menziesii, Pinaceae). I. Cold-Hardiness Related Traits. Genetics, 182(4), pp.1289–1302. Eckert, A.J., Van Heerwaarden, J., Wegrzyn, J.L., Nelson, C.D., Ross-Ibarra, J., González-Martínez, S.C. & Neale, D.B., (2010). Patterns of population structure and environmental associations to aridity across the range of loblolly pine (Pinus taeda L., Pinaceae). Genetics, 185(3), pp.969–982. El-Kassaby, Y.A. & Ritland, K., (1996). Impact of selection and breeding on the genetic diversity in Douglas-fir. Biodiversity and Conservation, 5(6), pp.795–813. Endler, J.A., (1977). Geographic Variation, Speciation and Clines. Princeton University Press, Princeton. Evelyn, J., (1664). Sylva, or, A discourse of forest-trees, and the propagation of timber in His Majesties dominions. Royal Society of London, UK. Fady, B., Cottrell, J., Ackzell, L., Alía, R., Muys, B., Prada, A. & Gonza, S.C., (2015). Forests and global change: what can genetics contribute to the major forest management and policy challenges of the twenty-first century? Regional Environmental Change, 16(4), pp.927–939. Falconer, D.S. & MacKay, T.F.C., (1996). Introduction to Quantitative Genetics. 4th Edition. Essex, England. Longman Scientific & Technical. Flint, H.L., Boyce, B.R. & Beattie, D.J., (1967). Index of injury – a useful expression of freezing injury to plant tissues as determined by the electrolytic method. Canadian Journal of Plant Science, 47(2), pp.229–230. Forest Genetics Council of British Columbia, (2015a). Annual Report 2014 - 2015 J. H. Woods, ed. Forest 103  Genetic Council of British Columbia, British Columbia, Canada. Forest Genetics Council of British Columbia, (2015b). Business Plan 2015 - 2016 J. H. Woods, ed. Forest Genetics Council of British Columbia, British Columbia, Canada. Funda, T., Liewlaksaneeyanawin, C. & El-Kassaby, Y.A., (2014). Determination of paternal and maternal parentage in lodgepole pine seed: full versus partial pedigree reconstruction. Canadian Journal of Forest Research, 44(9), pp.1122–1127. Gauthier, S., Bernier, P., Burton, P.J., Edwards, J., Isaac, K., Isabel, N., Jayen, K., Goff, H. Le & Nelson, E.A., (2014). Climate change vulnerability and adaptation in the managed Canadian boreal forest. Environmental Review, 30, pp.1–30. Gauthier, S., Bernier, P., Kuuluvainen, T., Shvidenko, A.Z. & Schepaschenko, D.G., (2015). Boreal forest health and global change. Science, 349, pp.819–822. Godt, M.J.W., Hamrick, J.L., Edwards-Burke, M.A. & Williams, J.H., (2001). Comparisons of genetic diversity in white spruce (Picea glauca) and jack pine (Pinus banksiana) seed orchards with natural populations. Canadian Journal of Forest Research, 31(6), pp.943–949. Gonzalez-Martinez, S.C., Wheeler, N.C., Ersoz, E., Nelson, C.D. & Neale, D.B., (2007). Association genetics in Pinus taeda L. I. Wood property traits. Genetics, 409, pp.399–409. Gray, L.K. & Hamann, A., (2011). Strategies for reforestation under uncertain future climates: guidelines for Alberta, Canada. PLoS ONE, 6, p.doi:10.1371/journal.pone.0022977. Gray, L.K. & Hamann, A., (2013). Tracking suitable habitat for tree populations under climate change in western North America. Climatic Change, 117(1–2), pp.289–303. Gray, L.K., Hamann, A., John, S., Rweyongeza, D., Barnhardt, L. & Thomas, B.R., (2016). Climate change risk management in tree improvement programs: selection and movement of genotypes. Tree Genetics & Genomes, 12(2), p.23. Gross, B.L. & Olsen, K.M., (2010). Genetic perspectives on crop domestication. Trends in Plant Science, 15(9), pp.529–37. Hall, D., Tegström, C. & Ingvarsson, P.K., (2010). Using association mapping to dissect the genetic basis of complex traits in plants. Briefings in Functional Genomics and Proteomics, 9(2), pp.157–165. Hamblin, M.T., Buckler, E.S. & Jannink, J.L., (2011). Population genetics of genomics-based crop improvement methods. Trends in Genetics, 27(3), pp.98–106. Hamilton, J.A., El Kayal, W., Hart, A.T., Runcie, D.E., Arango-Velez, A. & Cooke, J.E.K., (2016). The joint influence of photoperiod and temperature during growth cessation and development of dormancy in white spruce (Picea glauca). Tree Physiology, 36, pp.1432–1448. 104  Hamilton, J.A., De la Torre, A.R. & Aitken, S.N., (2015). Fine-scale environmental variation contributes to introgression in a three-species spruce hybrid complex. Tree Genetics & Genomes, 11(1), p.817. Hamilton, J.A. & Miller, J.M., (2016). Adaptive introgression as a resource for management and genetic conservation in a changing climate. Conservation Biology, 30(1), pp.33–41. Hannerz, M., Aitken, S.N., King, J.N. & Budge, S., (1999). Effects of genetic selection for growth on frost hardensess in western hemlock. Canadian Journal of Forest Research, 29(4), pp.509–516. Hannerz, M. & Westin, J., (2005). Autumn frost hardiness in Norway spruce plus tree progeny and trees of the local and transferred provenances in central Sweden. Tree physiology, 25(9), pp.1181–1186. Hansen, O.K., (2008). Mating patterns, genetic composition and diversity levels in two seed orchards with few clones - impact on planting crop. Forest Ecology and Management, 256(5), pp.1167–1177. Haselhorst, M.S.H. & Buerkle, C.A., (2013). Population genetic structure of Picea engelmannii, P. glauca and their previously unrecognized hybrids in the central Rocky Mountains. Tree Genetics and Genomes, 9(3), pp.669–681. Hoban, S., Kelley, J.L., Lotterhos, K.E., Antolin, M.F., Bradburd, G., Lowry, D.B., Poss, M.L., Reed, L.K., Storfer, A. & Whitlock, M.C., (2016). Finding the genomic basis of local adaptation: pitfalls, practical solutions, and future directions. American Naturalist, 188(4), pp.379–397. Holliday, J.A., Ritland, K. & Aitken, S.N., (2010). Widespread, ecologically relevant genetic markers developed from association mapping of climate-related traits in Sitka spruce (Picea sitchensis). New Phytologist, 188(2), pp.501–514. Hornoy, B., Pavy, N., Gerardi, S., Beaulieu, J. & Bousquet, J., (2015). Genetic adaptation to climate in white spruce involved small to moderate allele frequency shifts in functionally diverse genes. Genome Biology and Evolution, 7, pp.3269–3285. Hotte, N., Mahony, C. & Nelson, H., (2016). The principal-agent problem and climate change adaptation on public lands. Global Environmental Change, 36, pp.163–174. Howe, G.T., Aitken, S.N., Neale, D.B., Jermstad, K.D., Wheeler, N.C. & Chen, T.H.H., (2003). From genotype to phenotype: unraveling the complexities of cold adaptation in forest trees. Canadian Journal of Botany, 81(12), pp.1247–1266. Hurme, P., Repo, T., Savolainen, O. & Pääkkönen, T., (1997). Climatic adaptation of bud set and frost hardiness in Scots pine (Pinus sylvestris). Canadian Journal of Forest Research, 27, pp.716–723. Ingvarsson, P.K. & Street, N.R., (2011). Association genetics of complex traits in plants. New Phytologist, 189(4), pp.909–922. Jaramillo-Correa, J., Rodriguez-Quilon, I., Grivet, D., Lepoittevin, C., Sebastiani, F., Heuertz, M., Garnier-105  Gere, P., Plomion, C. & Vendramin, G., (2015). Molecular proxies for climate maladaptation in a long-lived tree (Pinus pinaster, Pinaceae). Genetics, 199, pp.793–807. Jaramillo-Correa, J.P., Verdú, M. & González-martínez, S.C., (2010). The contribution of recombination to heterozygosity differs among plant evolutionary lineages and life-forms. BMC Evolutionary Biology, 10:22. Johnsen, Ø., Fossdal, C.G., Nagy, N., Mølmann, J., Dæhlen, O.G. & Skrøppa, T., (2005). Climatic adaptation in Picea abies progenies is affected by the temperature during zygotic embryogenesis and seed maturation. Plant, Cell and Environment, 28(9), pp.1090–1102. Kawecki, T.J. & Ebert, D., (2004). Conceptual issues in local adaptation. Ecology Letters, 7(12), pp.1225–1241. Kitzmiller, J.H., (1990). Managing Genetic Diversity in a Tree Improvement Program. Forest Ecology and Management, 35, pp.131–149. Kolotelo, D., (1994). Response of interior spruce to extended stratification. British Columbia Ministry of Forests, Lands and Natural Resource Operations Tree Improvement Branch. Victoria, British Columbia. Kremer, A. & Le Corre, V., (2012). Decoupling of differentiation between traits and their underlying genes in response to divergent selection. Heredity, 108(4), pp.375–385. Kremer, A., Ronce, O., Robledo-Arnuncio, J.J., Guillaume, F., Bohrer, G., Nathan, R., Bridle, J.R., Gomulkiewicz, R., Klein, E.K., Ritland, K., Kuparinen, A., Gerber, S. & Schueler, S., (2012). Long-distance gene flow and adaptation of forest trees to rapid climate change. Ecology Letters, 15(4), pp.378–392. Kujala, S.T., Knürr, T., Kärkkäinen, K., Neale, D.B., Sillanpää, M.J. & Savolainen, O., (2017). Genetic heterogeneity underlying variation in a locally adaptive clinal trait in Pinus sylvestris revealed by a Bayesian multipopulation analysis. Heredity, 118, pp.413–423. De La Torre, A., Jaquish, B. & Aitken, S.N., (2014a). Adaptation and exogenous selection in a Picea glauca x Picea engelmannii hybrid zone: implications for forest management under climate change. New Phytologist, (201), pp.687–699. De La Torre, A., Roberts, D. & Aitken, S., (2014b). Genome-wide admixture and ecological niche modelling reveal the maintenance of species boundaries despite long history of interspecific gene flow. Molecular Ecology, 23(8), pp.2046–2059. De La Torre, A.R., Birol, I., Bousquet, J., Ingvarsson, P.K., Jansson, S., Jones, S.J.M., Keeling, C.I., Mackay, J., Nilsson, O., Ritland, K., Street, N. & Yanchuk, A., (2014c). Insights into conifer giga-genomes. 106  Plant Physiology, 166, pp.1724–1732. Langlet, O., (1971). Two hundred years of genecology. Taxon, 20, pp.653–721. Ledig, F.T. & Kitzmiller, J.H., (1992). Genetic strategies for reforestation in the face of global climate change. Forest Ecology and Management, 50, pp.153–169. Liepe, K.J., Hamann, A., Smets, P., Fitzpatrick, C.R. & Aitken, S.N., (2016). Adaptation of lodgepole pine and interior spruce to climate: implications for reforestation in a warming world. Evolutionary Applications, 9, pp.409–419. Lind, B.M., Friedline, C.J., Wegrzyn, J.L., Maloney, P.E., Detlev, R., Neale, D.B. & Eckert, A.J., (2017). Water availability drives signatures of local adaptation in whitebark pine (Pinus albicaulis) across fine spatial scales in the Lake Tahoe Basin, USA. Molecular Ecology, (in press). Liu, J., Schoettle, A.W., Sniezko, R.A., Sturrock, R.N., Zamany, A., Williams, H., Ha, A., Chan, D., Danchok, B., Savin, D.P. & Kegley, A., (2016). Genetic mapping of Pinus flexilis major gene (Cr4) for resistance to white pine blister rust using transcriptome-based SNP genotyping. BMC Genomics, 17:753, p.DOI: 10.1186/s12864-016-3079-2. Loarie, S.R., Duffy, P.B., Hamilton, H., Asner, G.P., Field, C.B. & Ackerly, D.D., (2009). The velocity of climate change. Nature, 462(7276), pp.1052–1055. Lstibuek, M. & El-Kassaby, Y.A., (2010). Minimum-inbreeding seed orchard design. Forest Science, 56, pp.603–608. Lu, M., Krutovsky, K. V, Nelson, C.D., Koralewski, T.E., Byram, T.D. & Loopstra, C.A., (2016). Exome genotyping, linkage disequilibrium and population structure in loblolly pine (Pinus taeda L.). BMC Genomics, 17:730, p.DOI: 10.1186/s12864-016-3081-8. McLachlan, J.S., Clark, J.S. & Manos, P.S., (2005). Molecular indicators of tree migration capacity under rapid climate change. Ecology, 86(8), pp.2088–2098. Meyer, R.S., DuVal, A.E. & Jensen, H.R., (2012). Patterns and processes in crop domestication: an historical review and quantitative analysis of 203 global food crops. The New Phytologist, 196(1), pp.29–48. Millar, C.I. & Stephenson, N.L., (2015). Temperate forest health in an era of emerging megadisturbance. Science, 349(6250), pp.823–826. Millar, C.I., Stephenson, N.L. & Stephens, S.L., (2007). Climate change and the forests of the future. Ecological Applications, 17, pp.2145–2151. Montwé, D., Spiecker, H. & Hamann, A., (2015). Five decades of growth in a genetic field trial of Douglas-fir reveal trade-offs between productivity and drought tolerance. Tree Genetics & Genomes, 11:29. 107  Moose, S.P., Dudley, J.W. & Rocheford, T.R., (2004). Maize selection passes the century mark: a unique resource for 21st century genomics. Trends in Plant Science, 9(7), pp.358–364. Morgenstern, E.K., (1996). Geographic Variation in Forest Trees: Genetic Basis and of Knowledge in Silviculture. University of British Columbia Press, Vancouver, Canada. Myles, S., Peiffer, J., Brown, P.J., Ersoz, E.S., Zhang, Z., Costich, D.E. & Buckler, E.S., (2009). Association mapping: critical considerations shift from genotyping to experimental design. The Plant cell, 21(8), pp.2194–2202. Namroud, M., Guillet-Claude, C., Mackay, J., Isabel, N. & Bousquet, J., (2010). Molecular evolution of regulatory genes in spruces from different species and continents: heterogeneous patterns of linkage disequilibrium and selection but correlated recent demographic changes. Journal of Molecular Evolution, 70, pp.371–386. Namroud, M.C., Bousquet, J., Doerksen, T. & Beaulieu, J., (2012). Scanning SNPs from a large set of expressed genes to assess the impact of artificial selection on the undomesticated genetic diversity of white spruce. Evolutionary Applications, 5(6), pp.641–656. National Forest Database, (2014). The Canadian National Forest Database. Available at: http://nfdp.ccfm.org/index_e.php [Accessed April 14, 2016]. Neale, D.B. & Kremer, A., (2011). Forest tree genomics: growing resources and applications. Nature Reviews Genetics, 12(2), pp.111–122. Neale, D.B. & Savolainen, O., (2004). Association genetics of complex traits in conifers. Trends in Plant Science, 9(7), pp.325–330. Neale, D.B., Wegrzyn, J.L., Stevens, K.A., Zimin, A. V, Puiu, D., Crepeau, M.W., Cardeno, C., Koriabine, M., Holtz-morris, A.E., Liechty, J.D., Martínez-garcía, P.J., Vasquez-gross, H.A., Lin, B.Y., Zieve, J.J., Dougherty, W.M., Fuentes-soriano, S., Wu, L., Gilbert, D., Marçais, G., Roberts, M., Holt, C., Yandell, M., Davis, J.M., Smith, K.E., Dean, J.F.D., Lorenz, W.W., Whetten, R.W., Sederoff, R., Wheeler, N., Mcguire, P.E., Main, D., Loopstra, C.A., Mockaitis, K., Pieter, J., Yorke, J.A., Salzberg, S.L. & Langley, C.H., (2014). Decoding the massive genome of loblolly pine using haploid DNA and novel assembly strategies. Genome Biology, 15:R59, pp.1–13. Nichols, K. & Neale, D., (2010). Association genetics, population genomics, and conservation: revealing the genes underlying adaptation in natural populations of plants and animals. In J. A. Dewoody, J. W. Bickham, C. H. Michler, K. M. Nichols, O. E. Rhodes, & K. E. Woeste, eds. Molecular Approaches in Natural Resource Conservation and Management. Cambridge University Press, Cambridge, UK. Nienstaedt, H. & Zasada, J., (1990). Picea glauca (Monech) Voss, white spruce. In R. M. Burns & B. H. 108  Honkala, eds. Silvics of North America Vol. 1, Conifers. U.S. Department of Agriculture Forest Service, Handbook 654, pp. 165–185. O’Neill, G.A., Stoehr, M. & Jaquish, B., (2014). Quantifying safe seed transfer distance and impacts of tree breeding on adaptation. Forest Ecology and Management, 328, pp.122–130. O’Neill, G.A., Ukrainetz, N.K., Carlson, M.R., Cartwright, C.V., Jaquish, B.C., King, J.N., Krakowski, J., Russell, J.H., Stoehr, M.U., Xie, C. & Yanchuk, A.D., (2008). Assisted migration to address climate change in British Columbia: recommendations for interim seed transfer standards. British Columbia Ministry of Forests and Range Research Branch, Victoria, BC, Canada. O’Reilly, C. & Owens, J.N., (1989). Shoot, needle, and cambial growth phenology and branch tracheid dimensions in provenances of lodgepole pine. Canadian Journal of Forest Research, 19, pp.599–605. Ordonez, A.O. & Williams, J.W., (2013). Climatic and biotic velocities for woody taxa distributions over the last 16 000 years in eastern North America. Ecology Letters, 16, pp.773–781. Owens, J.N., (2006). The Reproductive Biology of Lodgepole Pine. British Columbia Forest Genetics Council, Extension Note 7. Forest Genetics Council of British Columbia, British Columbia, Canada. Owens, J.N. & Molder, M., (1984). The reproductive cycle of interior spruce. British Columbia Ministry of Forests, Victoria, BC, Canada. Parchman, T.L., Gompert, Z., Mudge, J., Schilkey, F.D., Benkman, C.W. & Buerkle, C.A., (2012). Genome-wide association genetics of an adaptive trait in lodgepole pine. Molecular Ecology, 21(12), pp.2991–3005. Pavy, N., Namroud, M., Gagnon, F., Isabel, N. & Bousquet, J., (2012). The heterogeneous levels of linkage disequilibrium in white spruce genes and comparative analysis with other conifers. Heredity, 108, pp.273–284. Pedlar, J.H., McKenney, D.W., Beaulieu, J., Colombo, S.J., McLachlan, J.S. & O’Neill, G.A., (2011). The implementation of assisted migration in Canadian forests. Forestry Chronicle, 87(6), pp.766–770. Petterle, A., Karlberg, A. & Bhalerao, R.P., (2013). Daylength mediated control of seasonal growth patterns in perennial trees. Current Opinion in Plant Biology, 16(3), pp.301–306. Pita, P., Cañas, I., Soria, F., Ruiz, F. & Toval, G., (2005). Use of physiological traits in tree breeding for improved yield in drought-prone environments. The case of Eucalyptus globulus. Investigación Agraria: Sistemas y Recursos Forestales, 14(3), pp.383–393. Pritchard, J.K., Stephens, M. & Donnelly, P., (2000). Inference of population structure using multilocus genotype data. Genetics, 155(2), pp.945–959. 109  Quesada, T., Gopal, V., Cumbie, W.P., Eckert, A.J., Wegrzyn, J.L., Neale, D.B., Goldfarb, B., Huber, D.A., Casella, G. & Davis, J.M., (2010). Association mapping of quantitative disease resistance in a natural population of loblolly pine (Pinus taeda L.). Genetics, 186, pp.677–686. R Core Team, (2016). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/. Rehfeldt, G.E., (1983). Adaptation of Pinus contorta populations to heterogeneous environments in northern Idaho. Canadian Journal of Forest Research, 13, pp.405–411. Rehfeldt, G.E., (1988). Ecological Genetics of Pinus contorta from the Rocky Mountains (USA): A synthesis. Silvae Genetica, 37, pp.131–135. Rehfeldt, G.E., (1989). Genetic variances and covariances in freezing tolerance of lodgepole pine during early winter acclimation. Silvae Genetica, 38(3–4), pp.133–137. Rehfeldt, G.E., (1992a). Breeding strategies for Larix occidentalis: adaptation to the biotic and abiotic environment in relation to improving growth. Canadian Journal of Forest Research, 22, pp.5–13. Rehfeldt, G.E., (1992b). Early selection in Pinus ponderosa: compromises between growth potential and growth rhythm in developing breeding strategies. Forest Science, 38(3), pp.661–677. Rehfeldt, G.E., (1994). Adaptation of Picea engelmannii populations to the heterogeneous environments of the intermountain west. Canadian Journal of Botany, 72(8), pp.1197–1208. Rehfeldt, G.E. & Wykoff, W.R., (1981). Periodicity in shoot elongation among populations of Pinus contorta from the northern Rocky Mountains. Annals of Botany, 48, pp.371–377. Rehfeldt, G.E., Ying, C.C., Spittlehouse, D.L. & Hamilton, Jr., D.A., (1999). Genetic responses to climate in Pinus contorta: niche breadth, climate change, and reforestation. Ecological Monographs, 69(3), pp.375–407. Resende, M.D. V, Resende, M.F.R., Sansaloni, C.P., Petroli, C.D., Missiaggia, A. a., Aguiar, A.M., Abad, J.M., Takahashi, E.K., Rosado, A.M., Faria, D. a., Pappas, G.J., Kilian, A. & Grattapaglia, D., (2012). Genomic selection for growth and wood quality in Eucalyptus: Capturing the missing heritability and accelerating breeding for complex traits in forest trees. New Phytologist, 194(1), pp.116–128. Rieseberg, L.H., Archer, M. a & Wayne, R.K., (1999). Transgressive segregation, adaptation and speciation. Heredity, 83, pp.363–372. Rieseberg, L.H., Raymond, O., Rosenthal, D.M., Lai, Z., Livingstone, K., Nakazato, T., Durphy, J.L., Schwarzbach, A.E., Donovan, L.A. & Lexer, C., (2003). Major ecological transitions in wild sunflowers facilitated by hybridization. Science, 301(5637), pp.1211–1216. Riska, B., (1989). Composite traits, selection response and evolution. Evolution, 43, pp.1172–1191. 110  Roche, L., (1969). A genecological study of the genus Picea in British Columbia. New Phytologist, 68, pp.505–554. Rohde, A., Bastien, C. & Boerjan, W., (2011). Temperature signals contribute to the timing of photoperiodic growth cessation and bud set in poplar. Tree Physiology, 31(5), pp.472–482. Santos-del-Blanco, L., Alía, R., González-Martínez, S.C., Sampedro, L., Lario, F. & Climent, J., (2015). Correlated genetic effects on reproduction define a domestication syndrome in a forest tree. Evolutionary Applications, 8(4), pp.403–410. Savolainen, O. & Pyhajarvi, T., (2007). Genomic diversity in forest trees. Current Opinion in Plant Biology, 10(2), pp.162–167. Savolainen, O., Pyhajarvi, T. & Knurr, T., (2007). Gene flow and local adaptation in trees. Annual Review of Ecology Evolution and Systematics, 38, pp.595–619. Schmidtling, R.C., Carroll, E.C. & LaFarge, T., (1999). Allozyme diversity of selected and natural loblolly pine populations. Silvae Genetica, 48, pp.35–45. Schmidtling, R.C. & Hipkins, V., (2004). The after-effects of reproductive environment in shortleaf pine. Forestry, 77(4), pp.287–295. Schuster, R. & Oberhuber, W., (2013). Drought sensitivity of three co-occurring conifers within a dry inner Alpine environment. Tree, 27, pp.61–69. Scotti, I., González-Martínez, S.C., Budde, K.B. & Lalagüe, H., (2016). Fifty years of genetic studies: what to make of the large amounts of variation found within populations? Annals of Forest Science, 73(1), pp.69–75. Skrøppa, T., Kohmann, K., Johnsen, Ø., Steffenrem, A. & Edvardsen, Ø.M., (2007). Field performance and early test results of offspring from two Norway spruce seed orchards containing clones transferred to warmer climates. Canadian Journal of Forest Research, 37(3), pp.515–522. St Clair, J.B. & Howe, G.T., (2007). Genetic maladaptation of coastal Douglas-fir seedlings to future climates. Global Change Biology, 13(7), pp.1441–1454. Stephan, W., (2016). Signatures of positive selection: from selective sweeps at individual loci to subtle allele frequency changes in polygenic adaptation. Molecular Ecology, 25, pp.79–88. Stoehr, M., O’Neill, G., Hollefreund, C. & Yanchuk, A., (2005). Within and among family variation of orchard and wild-stand progeny of interior spruce in British Columbia. Tree Genetics & Genomes, 1(2), pp.64–68. Stoehr, M., Webber, J.O.E. & Woods, J., (2004). Protocol for rating seed orchard seedlots in British Columbia: quantifying genetic gain and diversity. Forestry, 77, pp.297–303. 111  Stoehr, M.U. & El-Kassaby, Y.A., (1997). Levels of genetic diversity at different stages of the domestication cycle of interior spruce in British Columbia. Theoretical and Applied Genetics, 94(1), pp.83–90. Stoehr, M.U. & Newton, C.H., (2002). Evaluation of mating dynamics in a lodgepole pine seed orchard using chloroplast DNA markers. Canadian Journal of Forest Research, 32(3), pp.469–476. Suarez-Gonzalez, A., Hefer, C.A., Christe, C., Corea, O., Lexer, C., Cronk, Q.C.B. & Douglas, C.J., (2016). Genomic and functional approaches reveal a case of adaptive introgression from Populus balsamifera (balsam poplar) in P. trichocarpa (black cottonwood). Molecular Ecology, 25(11), pp.2427–2442. Suren, H., Hodgins, K.A., Yeaman, S., Nurkowski, K.A., Smets, P., Rieseberg, L.H., Aitken, S.N. & Holliday, J.A., (2016). Exome capture from the spruce and pine giga-genomes. Molecular Ecology Resources, 16(5), pp.1136–1146. Tanino, K.K., Kalcsits, L., Silim, S., Kendall, E. & Gray, G.R., (2010). Temperature-driven plasticity in growth cessation and dormancy development in deciduous woody plants: A working hypothesis suggesting how molecular and cellular function is affected by temperature during dormancy induction. Plant Molecular Biology, 73(1–2), pp.49–65. Turchin, M.C., Chiang, C.W.K., Palmer, C.D., Sankararaman, S., Reich, D., Investigation, G., Giant, T. & Hirschhorn, J.N., (2012). Evidence of widespread selection on standing variation in Europe at height-associated SNPs. Nature Genetics, 44, pp.1015–1019. Ukrainetz, N.K., O’Neill, G.A. & Jaquish, B., (2011). Comparison of fixed and focal point seed transfer systems for reforestation and assisted migration: a case study for interior spruce in British Columbia. Canadian Journal of Forest Research, 41(7), pp.1452–1464. Wang, T., Hamann, A., Spittlehouse, D. & Carroll, C., (2016). Locally downscaled and spatially customizable climate data for historical and future periods for North America. Plos One, 11(6), p.e0156720 10.1371/journal.pone.0156720. Wang, T., Hamann, A., Spittlehouse, D.L. & Aitken, S.N., (2006a). Development of scale-free climate data for western Canada for use in resource management. International Journal of Climatology, 26(3), pp.383–397. Wang, T., Hamann, A., Yanchuk, A.D., O’Neill, G.A. & Aitken, S.N., (2006b). Use of response functions in selecting lodgepole pine populations for future climates. Global Change Biology, 12, pp.2004–2016. Weiser, C.J., (1970). Cold resistance and injury in woody plants. Science, 169(3952), pp.1269–1278. 112  Westin, J., Sundblad, L., Strand, M. & Hallgren, J., (2000). Phenotypic differences between natural and selected populations of Picea abies. I . Frost hardiness. Canadian Journal of Forest Research, 15, pp.489–499. White, T.L., Adams, W.T. & Neale, D.B., (2007). Forest Genetics. Wallingford, United Kingdom: CABI Publishing. Whitham, T.G., Young, W.P., Martinsen, G.D., Gehring, C.A., Schweitzer, J.A., Shuster, S.M., Wimp, G.M., Fischer, D.G., Bailey, J.K., Lindroth, R.L., Woolbright, S. & Kuske, C.R., (2003). Community and ecosystem genetics: A consequence of the extended phenotype. Ecology, 84(3), pp.559–573. Woods, J., Stoehr, M. & Webber, J., (1996). Protocols for rating seed orchard seedlots in British Columbia. Research Report 06. Victoria, BC: British Columbia Ministry of Forests. Wright, S.I., Bi, I.V., Schroeder, S.G., Yamasaki, M., Doebley, J.F., McMullen, M.D. & Gaut, B.S., (2005). The effects of artificial selection on the maize genome. Science, 308(5726), pp.1310–4. Wu, T.D. & Watanabe, C.K., (2005). GMAP: A genomic mapping and alignment program for mRNA and EST sequences. Bioinformatics, 21(9), pp.1859–1875. Xie, C. & Yanchuk, A.D., (2003). Breeding values of parental trees, genetic worth of seed orchard seedlots, and yields of improved stocks in British Columbia. Western Journal of Applied Forestry, 18(2), pp.88–100. Yanchuk, A.D., (2001). A quantitative framework for breeding and conservation of forest tree genetic resources in British Columbia. Canadian Journal of Forest Research-Revue Canadienne De Recherche Forestiere, 31(4), pp.566–576. Yang, J., Lee, S.H., Goddard, M.E. & Visscher, P.M., (2011). GCTA: A tool for genome-wide complex trait analysis. The American Journal of Human Genetics, 88, pp.76–82. Yeaman, S., (2015). Local adaptation by alleles of small effect. American Naturalist, 186, pp.S74–S89. Yeaman, S., Hodgins, K.A., Lotterhos, K.E., Suren, H., Nadeau, S., Degner, J.C., Nurkowski, K.A., Smets, P., Wang, T., Gray, L.K., Liepe, K.J., Hamann, A., Holliday, J.A., Whitlock, M.C., Rieseberg, L.H. & Aitken, S.N., (2016). Convergent local adaptation to climate in distantly related conifers. Science, 353(6306), pp.23–26. Yeaman, S., Hodgins, K.A., Suren, H., Nurkowski, K.A., Rieseberg, L.H., Holliday, J.A. & Aitken, S.N., (2014). Conservation and divergence of gene expression plasticity following c. 140 million years of evolution in lodgepole pine (Pinus contorta) and interior spruce (Picea glauca × Picea engelmannii). The New phytologist, 203(2), pp.578–91. Yu, J., Pressoir, G., Briggs, W.H., Vroh Bi, I., Yamasaki, M., Doebley, J.F., McMullen, M.D., Gaut, B.S., 113  Nielsen, D.M., Holland, J.B., Kresovich, S. & Buckler, E.S., (2006). A unified mixed-model method for association mapping that accounts for multiple levels of relatedness. Nature Genetics, 38(2), pp.203–208. Zobel, B. & Talbert, J., (1984). Applied forest tree improvement. John Wiley & Sons. Zuk, O., Hechter, E., Sunyaev, S.R. & Lander, E.S., (2012). The mystery of missing heritability: Genetic interactions create phantom heritability. Proceedings of the National Academy of Sciences, 109(4), pp.1193–1198.  114  Appendices  Appendix A  Supplementary materials for Chapter 2    Geographic or Climatic Variable Abbreviation Latitude (oN)* LAT Longitude (oW)* LONG Elevation (m)* ELEV Mean annual temperature (oC)* MAT  Mean warmest month temperature (oC) MWMT Mean coldest month temperature (oC) MCMT Temperature difference (oC) TD Mean annual precipitation (mm) * MAP  Mean annual summer precipitation (mm) MSP Annual heat moisture index AHM  Summer heat moisture index * SHM  Chilling degree days below zero degrees (oC) DD < 0 Growing degree days above five degrees (oC) * DD > 5 Frost free period (days) FFP Number of frost free days (days) * NFFD Beginning of frost free period (day) bFFP End of frost free period (day) eFFP Precipitation as snow (mm) PAS  Extreme minimum temperature (oC) * EMT  Extreme maximum temperature (oC) EXT Hargreaves reference evaporation (mm) Eref Hargreaves climatic moisture deficit (mm) CMD  Table A.1 Climatic variables interpolated from ClimateNA. Variables marked by an asterisk were used for clinal analysis; all 22 variables were used to generate climatic PCA scores. 115      PC1 PC2 PC3 PC4 Standard deviation 3.441 1.935 1.746 1.342 Proportion of Variance 0.538 0.170 0.139 0.082 Cumulative Proportion 0.538 0.708 0.847 0.929  Table A.2 PC1–4 effects from PCA of the 22 lodgepole pine seedling climate variables in Table A.1. 116    PC1 PC2 Rank Variable Loading Variable Loading 1 EMT  0.279 Log MAP 0.392 2 MAT  0.278 ELEV 0.376 3 NFFD 0.271 MSP 0.320 4 DD > 5 0.265 TD 0.303 5 Eref 0.264 SHM  0.281 6 eFFP 0.250 AHM  0.275 7 MWMT 0.249 CMD 0.268 8 EXT 0.247 MCMT 0.234 9 DD < 0 0.246 LONG 0.232 10 MCMT 0.238 DD < 0 0.190 11 LAT 0.235 LAT 0.185 12 CMD 0.225 EXT 0.180 13 SHM  0.221 eFFP 0.135 14 FFP 0.221 PAS  0.124 15 bFFP 0.172 MAT  0.109 16 TD 0.169 MWMT 0.102 17 MSP 0.167 FFP 0.089 18 AHM  0.165 EMT  0.055 19 ELEV 0.067 DD > 5 0.048 20 PAS  0.066 Eref 0.047 21 Log MAP 0.063 bFFP 0.035 22 LONG 0.004 NFFD 0.028  Table A.3 Ranked PC1 and PC2 loadings from the PCA of all 22 climate variables in Table A.1. 117    Province Breeding Zone Seedlot Type Seedling Height (cm) Growth Rate (cm day-1) Shoot Dry Mass ^1/4 (g) Growth Initiation (day) Growth Cessation (day) Cold Injury (%) AB A Natural 38.8 (1.89) 0.57 (0.023) 1.8 (0.056) 87 (0.51) 144 (1.02) 51.8 (2.74) AB A Selected 44.1 (2.07) 0.63 (0.025) 1.92 (0.062) 87 (0.56) 146 (1.11) 54.3 (2.84) AB B1 Natural 40.1 (2.04) 0.61 (0.025) 1.78 (0.061) 88 (0.55) 145 (1.1) 50.4 (2.84) AB B1 Selected 46.5 (1.51) 0.67 (0.018) 1.98 (0.045) 89 (0.4) 148 (0.84) 50.4 (2.51) AB B2 Natural 33.1 (2.11) 0.53 (0.026) 1.56 (0.063) 87 (0.57) 140 (1.13) 50.7 (2.85) AB B2 Selected 36.9 (2.09) 0.56 (0.025) 1.72 (0.062) 87 (0.56) 142 (1.12) 53.8 (2.85) AB C Natural 34.5 (2.04) 0.53 (0.025) 1.63 (0.061) 88 (0.55) 143 (1.1) 52 (2.86) AB C Selected 41.8 (2.07) 0.59 (0.025) 1.85 (0.062) 88 (0.56) 147 (1.12) 52.2 (2.88) AB J Natural 32 (1.76) 0.51 (0.021) 1.48 (0.052) 87 (0.47) 141 (0.96) 37.7 (2.67) AB J Selected 37.9 (1.48) 0.54 (0.018) 1.69 (0.044) 86 (0.4) 145 (0.83) 31.1 (2.5) AB K1 Natural 34.6 (1.86) 0.54 (0.023) 1.68 (0.055) 86 (0.51) 142 (1.03) 52.6 (2.67) AB K1 Selected 34.4 (2.01) 0.54 (0.024) 1.69 (0.06) 87 (0.54) 141 (1.08) 54 (2.82) BC BV low Natural 34.4 (1.71) 0.53 (0.021) 1.58 (0.051) 90 (0.46) 145 (0.94) 59.7 (2.62) BC BV low Selected 52.1 (1.49) 0.7 (0.018) 2.09 (0.044) 89 (0.4) 153 (0.83) 64.7 (2.5) BC CP low Natural 39.6 (1.94) 0.59 (0.024) 1.68 (0.058) 89 (0.52) 147 (1.05) 52.8 (2.75) BC CP low Selected 48.8 (1.24) 0.7 (0.015) 2 (0.036) 90 (0.33) 151 (0.71) 63.8 (2.37) BC EK low Natural 48 (1.67) 0.63 (0.02) 1.93 (0.049) 89 (0.45) 151 (0.92) 77.3 (2.59) BC EK low Selected 54.1 (2.09) 0.67 (0.025) 2.13 (0.062) 90 (0.56) 157 (1.12) 80.7 (2.83) BC NE low Natural 47.7 (1.75) 0.62 (0.021) 1.93 (0.052) 89 (0.47) 152 (0.96) 79.7 (2.62) BC NE low Selected 59.1 (1.56) 0.74 (0.019) 2.22 (0.046) 90 (0.41) 157 (0.86) 82.9 (2.51) BC PG low Natural 44.6 (1.68) 0.65 (0.02) 1.83 (0.05) 90 (0.45) 150 (0.93) 69.6 (2.63) BC PG low Selected 51.2 (1.54) 0.71 (0.019) 2.03 (0.045) 90 (0.41) 152 (0.85) 69.1 (2.5) BC TO low Natural 39.8 (1.61) 0.56 (0.019) 1.74 (0.048) 89 (0.43) 146 (0.89) 74.6 (2.54) BC TO low Selected 60.6 (1.55) 0.75 (0.019) 2.25 (0.046) 89 (0.41) 156 (0.86) 81.3 (2.51)  Table A.4 BLUEs of trait means for each breeding zone and seedlot type combination. Standard errors are in brackets.  118   Climatic Variable Seedling Height Growth Rate Shoot Mass Growth Initiation Growth Cessation Cold Injury Natural Selected Natural Selected Natural Selected Natural Selected Natural Selected Natural Selected LAT 0.46 0.39 0.22 0.29 0.55 0.42 0.11 0.31 0.39 0.35 0.81 0.79  (0.015) (0.03) (0.126) (0.07) (0.006) (0.022) (0.301) (0.062) (0.031) (0.044) (0.0001) (0.0001) LONG 0.00 0.23 0.01 0.32 0.07 0.19 0.44 0.26 0.02 0.27 0.00 0.07  (0.839) (0.118) (0.82) (0.055) (0.404) (0.159) (0.018) (0.091) (0.635) (0.085) (0.913) (0.408) ELEV 0.01 0.32 0.03 0.36 0.02 0.25 0.27 0.25 0.07 0.37 0.01 0.03  (0.786) (0.058) (0.563) (0.04) (0.668) (0.1) (0.085) (0.098) (0.4) (0.035) (0.744) (0.58) MAT  0.62 0.80 0.41 0.71 0.68 0.83 0.27 0.61 0.58 0.73 0.88 0.95  (0.003) (0.0001) (0.026) (0.0006) (0.001) (<0.0001) (0.084) (0.003) (0.004) (0.0004) (<0.0001) (<0.0001) log MAP 0.29 0.23 0.22 0.36 0.37 0.28 0.00 0.50 0.20 0.18 0.26 0.42  (0.073) (0.116) (0.123) (0.039) (0.035) (0.078) (0.832) (0.01) (0.146) (0.164) (0.091) (0.022) SHM  0.24 0.72 0.10 0.56 0.08 0.63 0.57 0.44 0.38 0.79 0.46 0.55  (0.11) (0.0005) (0.31) (0.005) (0.373) (0.002) (0.005) (0.018) (0.032) (0.0001) (0.016) (0.006) DD > 5 0.66 0.77 0.40 0.59 0.63 0.72 0.10 0.41 0.56 0.73 0.57 0.61  (0.001) (0.0002) (0.027) (0.004) (0.002) (0.0005) (0.305) (0.025) (0.005) (0.0004) (0.004) (0.003) NFFD 0.56 0.83 0.34 0.64 0.51 0.79 0.19 0.46 0.52 0.81 0.60 0.64  (0.005) (<0.0001) (0.046) (0.002) (0.009) (0.0001) (0.16) (0.016) (0.008) (0.0001) (0.003) (0.002) EMT  0.56 0.86 0.32 0.69 0.47 0.83 0.43 0.60 0.62 0.85 0.89 0.89  (0.005) (<0.0001) (0.058) (0.0009) (0.014) (<0.0001) (0.021) (0.003) (0.002) (<0.0001) (<0.0001) (<0.0001) PC1 0.67 0.87 0.39 0.70 0.60 0.83 0.35 0.57 0.68 0.85 0.92 0.85  (0.001) (<0.0001) (0.03) (0.0007) (0.003) (<0.0001) (0.041) (0.005) (0.001) (<0.0001) (<0.0001) (<0.0001) PC2 0.02 0.04 0.01 0.03 0.14 0.01 0.11 0.01 0.00 0.09 0.04 0.01   (0.65) (0.51) (0.736) (0.583) (0.234) (0.708) (0.292) (0.821) (0.964) (0.355) (0.524) (0.736)  Table A.5 r2 values for clines in eleven climatic variables of the six traits and two seedling types. p-values (in brackets) are statistically significant at an adjusted α = 0.0045 cut-off value. 119  Province Breeding Zone Seedlot Type Parental Spruce Species Engelmann White Sitka AB D1 Natural 0.05 (0.00) 0.95 (0.00) 0.00 (0.000) AB D1 Orchard 0.04 (0.00) 0.95 (0.00) 0.00 (0.000) AB E Natural 0.02 (0.00) 0.98 (0.00) 0.00 (0.000) AB E Orchard 0.02 (0.00) 0.98 (0.00) 0.00 (0.000) AB G1 Natural 0.10 (0.01) 0.90 (0.01) 0.00 (0.000) AB G1 Orchard 0.12 (0.00) 0.88 (0.00) 0.00 (0.000) AB G2 Natural 0.06 (0.00) 0.94 (0.00) 0.00 (0.001) AB G2 Orchard 0.06 (0.00) 0.94 (0.00) 0.00 (0.001) AB H Natural 0.02 (0.00) 0.98 (0.00) 0.00 (0.000) AB H Orchard 0.02 (0.00) 0.98 (0.00) 0.00 (0.000) AB I Natural 0.07 (0.00) 0.92 (0.00) 0.00 (0.001) AB I Orchard 0.09 (0.00) 0.91 (0.00) 0.00 (0.000) BC BV low Natural 0.34 (0.00) 0.62 (0.01) 0.04 (0.003) BC BV low Orchard 0.33 (0.01) 0.62 (0.01) 0.05 (0.004) BC EK all Natural 0.55 (0.02) 0.45 (0.02) 0.00 (0.000) BC EK all Orchard 0.47 (0.01) 0.53 (0.01) 0.00 (0.000) BC NE low Natural 0.72 (0.01) 0.28 (0.01) 0.00 (0.000) BC NE low Orchard 0.47 (0.02) 0.53 (0.02) 0.00 (0.001) BC NE mid Natural 0.89 (0.01) 0.11 (0.01) 0.00 (0.000) BC NE mid Orchard 0.75 (0.01) 0.25 (0.01) 0.00 (0.000) BC PG high Natural 0.54 (0.02) 0.46 (0.02) 0.00 (0.001) BC PG high Orchard 0.36 (0.01) 0.61 (0.01) 0.02 (0.002) BC PG low Natural 0.40 (0.01) 0.60 (0.01) 0.01 (0.001) BC PG low Orchard 0.44 (0.01) 0.55 (0.01) 0.01 (0.001) BC PR mid Natural 0.19 (0.02) 0.80 (0.02) 0.00 (0.001) BC PR mid Orchard 0.16 (0.01) 0.83 (0.01) 0.01 (0.001) BC TO low Natural 0.60 (0.03) 0.39 (0.03) 0.01 (0.002) BC TO low Orchard 0.64 (0.01) 0.35 (0.01) 0.01 (0.001)  Table B.1 Mean proportions (ADMIXTURE Q-values) of the three parental spruce species’ genetic contributions to each breeding zone and seedling type. Standard errors are in brackets. Appendix B  Supplementary materials for Chapter 3     120      PC1 PC2 PC3 PC4 Standard deviation 3.019 2.601 1.898 1.028 Proportion of Variance 0.414 0.307 0.164 0.048 Cumulative Proportion 0.414 0.722 0.886 0.934  Table B.2 PC1–4 effects from PCA of the 16 interior spruce seedling climate variables in Table 3.2. 121  PC1 PC2 Variable Loading Variable Loading MAT  0.304 ELEV 0.318 eFFP 0.286 MWMT 0.298 EMT  0.284 PAS  0.295 LAT 0.269 log MAP 0.284 DD > 5 0.247 MCMT 0.263 DD < 0 0.243 EXT 0.253 MCMT 0.232 DD < 0 0.246 EXT 0.224 LONG 0.245 CMD 0.219 DD > 5 0.232 SHM  0.195 LAT 0.172 MWMT 0.187 EMT  0.168 MSP 0.105 MAT  0.130 log MAP 0.074 SHM  0.121 LONG 0.048 CMD 0.080 ELEV 0.038 eFFP 0.052 PAS  0.032 MSP 0.027  Table B.3 Ranked PC1 and PC2 loadings from the PCA of all 16 climate variables in Table 3.2. 122    Province Breeding Zone Seedlot Type Height                    (cm) Growth Rate (cm/Day) Shoot Dry             Mass (g) Bud Break           (Day) Bud Set                  (Day) Cold Injury              (%) AB D1 Natural 12.99 (0.88) 0.1 (0.008) 2.24 (0.76) 105 (1.1) 195.6 (3.0) 25.7  (2.2) AB D1 Orchard 14.82 (0.75) 0.13 (0.007) 3.28 (0.65) 106.6 (0.9) 200.6 (2.6) 31.5  (2.1) AB E Natural 12.89 (1.15) 0.1 (0.011) 2.49 (1.01) 104.7 (1.4) 194.3 (3.9) 21.7  (2.5) AB E Orchard 16.77 (1.04) 0.13 (0.009) 3.83 (0.90) 108.2 (1.3) 207.1 (3.5) 26  (2.4) AB G1 Natural 14.39 (1.06) 0.13 (0.01) 2.95 (0.92) 104.2 (1.3) 188.8 (3.6) 25.1  (2.4) AB G1 Orchard 16.72 (1.04) 0.13 (0.01) 4.03 (0.9) 102.6 (1.3) 199.1 (3.5) 28.8  (2.4) AB G2 Natural 12.19 (1.07) 0.1 (0.01) 1.86 (0.93) 105.2 (1.3) 187.8 (3.6) 23.9  (2.4) AB G2 Orchard 15.43 (1.04) 0.11 (0.009) 2.97 (0.90) 103.1 (1.3) 195.4 (3.5) 22.8  (2.4) AB H Natural 10.47 (0.97) 0.11 (0.01) 1.53 (0.84) 105.3 (1.2) 183.6 (3.3) 18.3  (2.3) AB H Orchard 11.81 (1.03) 0.11 (0.01) 1.55 (0.90) 105.6 (1.2) 187.5 (3.5) 18  (2.4) AB I Natural 14.17 (0.98) 0.13 (0.009) 2.93 (0.85) 101.7 (1.2) 191.7 (3.4) 35.5  (2.3) AB I Orchard 15.9 (1.03) 0.11 (0.009) 3.87 (0.90) 104.6 (1.2) 196.3 (3.5) 32.1  (2.3) BC BV low Natural 11.21 (0.96) 0.12 (0.009) 1.81 (0.84) 105.4 (1.2) 177.9 (3.3) 34.5  (2.3) BC BV low Orchard 20.88 (0.76) 0.16 (0.007) 7.15 (0.65) 103.4 (0.9) 207.2 (2.6) 41.1  (2.1) BC EK all Natural 21.35 (0.91) 0.15 (0.008) 7.02 (0.79) 97.4 (1.1) 203.2 (3.1) 50  (2.2) BC EK all Orchard 28.31 (1.03) 0.19 (0.009) 15.84 (0.90) 96.9 (1.2) 206.8 (3.5) 42.7  (2.3) BC NE low Natural 23.08 (1.12) 0.16 (0.01) 8.31 (0.97) 94.4 (1.4) 208.8 (3.8) 58.5  (2.4) BC NE low Orchard 30.94 (1.05) 0.18 (0.009) 17.45 (0.91) 97 (1.3) 224.2 (3.6) 52.7  (2.4) BC NE mid Natural 12.73 (0.85) 0.13 (0.008) 2.74 (0.73) 102.3 (1.0) 183.7 (2.9) 46.5  (2.2) BC NE mid Orchard 20.53 (1.04) 0.17 (0.01) 7.97 (0.90) 99.5 (1.3) 198.9 (3.5) 51.8  (2.4) BC PG high Natural 10.29 (1.08) 0.1 (0.011) 1.63 (0.93) 106.4 (1.3) 176.6 (3.6) 39  (2.4) BC PG high Orchard 15.43 (1.06) 0.11 (0.01) 3.67 (0.92) 106 (1.3) 189.6 (3.6) 37.4  (2.4) BC PG low Natural 14.81 (0.71) 0.13 (0.007) 3.88 (0.61) 104.5 (0.9) 185.4 (2.4) 38.9  (2.0) BC PG low Orchard 18.89 (0.75) 0.14 (0.007) 5.39 (0.65) 101 (0.9) 202.3 (2.6) 44.4  (2.1) BC PR mid Natural 10.7 (0.92) 0.13 (0.009) 1.96 (0.80) 104.2 (1.1) 169.5 (3.1) 24.9  (2.2) BC PR mid Orchard 17.34 (1.04) 0.12 (0.01) 4.56 (0.90) 103.3 (1.3) 199 (3.5) 30  (2.4) BC TO low Natural 15.8 (0.94) 0.14 (0.009) 5.03 (0.81) 103.9 (1.1) 188 (3.2) 43.9  (2.2) BC TO low Orchard 23.92 (0.75) 0.17 (0.007) 9.16 (0.64) 98.8 (0.9) 206 (2.6) 49.5  (2.1)  Table B.4 BLUEs of trait means for each breeding zone and seedlot type combination. Standard errors are in brackets.  123   Climatic Variable Seedling Height Growth Rate Shoot Dry Mass Bud Break Bud Set Cold Injury Natural Selected Natural Selected Natural Selected Natural Selected Natural Selected Natural Selected LAT 0.42 0.75 0.43 0.77 0.45 0.78 0.37 0.66 0.15 0.43 0.86 0.85  (0.012) (0.0001) (0.01) (< 0.0001) (0.009) (< 0.0001) (0.02) (0.0005) (0.167) (0.011) (< 0.0001) (< 0.0001) LONG 0.05 0.03 0.02 0.02 0.04 0.03 0.06 0.00 0.37 0.12 0.04 0.01  (0.444) (0.531) (0.591) (0.656) (0.497) (0.58) (0.387) (0.89) (0.022) (0.218) (0.486) (0.714) ELEV 0.00 0.07 0.07 0.15 0.00 0.12 0.00 0.17 0.08 0.01 0.30 0.23  (0.98) (0.362) (0.379) (0.171) (0.884) (0.217) (0.921) (0.149) (0.318) (0.772) (0.043) (0.079) MAT  0.58 0.68 0.61 0.59 0.64 0.69 0.56 0.59 0.21 0.54 0.91 0.87  (0.0015) (0.0003) (0.001) (0.001) (0.0006) (0.0002) (0.002) (0.001) (0.096) (0.003) (< 0.0001) (< 0.0001) MWMT 0.24 0.13 0.04 0.07 0.21 0.08 0.27 0.08 0.51 0.29 0.00 0.01  (0.075) (0.21) (0.486) (0.37) (0.1) (0.32) (0.057) (0.32) (0.0043) (0.047) (0.99) (0.6814) MCMT 0.26 0.55 0.45 0.54 0.31 0.60 0.26 0.53 0.01 0.29 0.82 0.90  (0.061) (0.0023) (0.0082) (0.0029) (0.039) (0.0012) (0.063) (0.0031) (0.68) (0.046) (< 0.0001) (< 0.0001) log MAP 0.15 0.03 0.19 0.04 0.18 0.04 0.25 0.06 0.01 0.01 0.69 0.41  (0.18) (0.58) (0.16) (0.51) (0.13) (0.48) (0.068) (0.41) (0.75) (0.76) (0.0002) (0.014) MSP 0.03 0.21 0.05 0.27 0.01 0.17 0.00 0.19 0.00 0.04 0.01 0.11  (0.55) (0.01) (0.42) (0.056) (0.717) (0.146) (0.98) (0.11) (0.89) (0.506) (0.807) (0.258) SHM  0.13 0.44 0.12 0.45 0.09 0.36 0.05 0.38 0.09 0.21 0.02 0.19  (0.2) (0.009) (0.22) (0.009) (0.308) (0.025) (0.456) (0.019) (0.29) (0.098) (0.615) (0.117) DD < 0 0.32 0.57 0.46 0.53 0.37 0.62 0.30 0.51 0.04 0.35 0.85 0.90  (0.035) (0.0017) (0.0073) (0.003) (0.0202) (0.0008) (0.044) (0.004) (0.497) (0.025) (< 0.0001) (< 0.0001) DD > 5 0.46 0.29 0.16 0.18 0.44 0.23 0.49 0.21 0.65 0.46 0.07 0.13  (0.008) (0.045) (0.153) (0.13) (0.01) (0.082) (0.005) (0.096) (0.0005) (0.007) (0.356) (0.206) EFFP 0.39 0.37 0.39 0.37 0.43 0.34 0.52 0.39 0.21 0.38 0.64 0.64  (0.017) (0.0206) (0.0174) (0.021) (0.0113) (0.028) (0.0035) (0.0173) (0.1002) (0.0193) (0.0006) (0.0006) PAS  0.05 0.02 0.11 0.06 0.10 0.01 0.10 0.06 0.01 0.02 0.56 0.32  (0.44) (0.66) (0.247) (0.404) (0.269) (0.72) (0.26) (0.385) (0.76) (0.66) (0.002) (0.034) EMT  0.36 0.63 0.38 0.64 0.37 0.61 0.38 0.59 0.09 0.41 0.87 0.94  (0.024) (0.0007) (0.02) (0.0006) (0.02) (0.0009) (0.019) (0.001) (0.29) (0.014) (< 0.0001) (< 0.0001) EXT 0.31 0.34 0.07 0.28 0.26 0.25 0.27 0.27 0.52 0.38 0.02 0.11  (0.037) (0.028) (0.346) (0.053) (0.063) (0.066) (0.057) (0.055) (0.0037) (0.019) (0.669) (0.24) CMD 0.30 0.62 0.21 0.59 0.24 0.53 0.13 0.50 0.19 0.35 0.12 0.34  (0.04) (0.0009) (0.1003) (0.0014) (0.076) (0.0032) (0.2) (0.0048) (0.116) (0.026) (0.23) (0.028) PC1 0.70 0.72 0.56 0.64 0.71 0.67 0.68 0.64 0.43 0.60 0.71 0.71  (0.0002) (< 0.0001) (0.002) (0.0006) (0.0002) (0.0003) (0.0003) (0.0006) (0.011) (0.001) (< 0.0001) (< 0.0001) PC2 0.00 0.02 0.08 0.04 0.01 0.04 0.00 0.04 0.11 0.01 0.38 0.27   (0.91) (0.66) (0.33) (0.47) (0.79) (0.476) (0.83) (0.49) (0.24) (0.7) (0.019) (0.059)  Table B.5 r2 values of phenotypic clines with sixteen climatic variables for six traits and two seedling types.  p-values. (in brackets) are statistically significant at an adjusted α = 0.0031 cut-off value.  123 124    Figure B.1 Comparison of ADMIXTURE and Structure Q-values for 500 seedlings used in the ADMIXTURE projection analysis reference panel. ADMIXTURE Q-values are estimated from 6482 neutral or candidate adaptive SNPs. Structure Q-values are estimated from 817 neutral SNPs. 125    Figure B.2 Mean spruce hybrid index (proportion of P. engelmannii ancestry) including standard errors, for each breeding zone by seedling type combination. 126    Figure B.3 Bar plots of breeding zone level trait means (BLUEs) including standard error bars for a) growth rate, b) shoot dry mass, c) bud break, and d) bud set. 127    Figure B.4 Regressions of a) growth rate, b) shoot mass, c) bud break, and d) bud set, versus hybrid index  complimenting Figure 3.4. Points represent BLUEs of trait means (Table B.5), and the mean spruce hybrid index (P. engelmannii proportion) for each breeding zone by seedling type combination (Table B.1). p-values are statistically significant at the adjusted α = 0.0083 cut-off used in Table 3.3. 128   Figure B.5 Regression of height gains in each breeding zone on mean summer temperature differences (June to August) within each breeding zone for selected seedlings and their respective seed orchard sites. 129  Appendix C  Supplementary materials for Chapter 4 Province Breeding Zone Seedlot Type Mean Frequency of Positive Effect Alleles by Associated Phenotype Height Growth Rate Shoot Dry Mass Growth Initiation Growth Cessation Cold Injury AB A Natural 0.38 (0.018) 0.41 (0.019) 0.42 (0.018) 0.15 (0.01) 0.18 (0.012) 0.4 (0.019) AB A Orchard 0.37 (0.019) 0.4 (0.02) 0.41 (0.019) 0.13 (0.011) 0.16 (0.013) 0.4 (0.019) AB B1 Natural 0.38 (0.018) 0.42 (0.019) 0.42 (0.018) 0.16 (0.01) 0.2 (0.012) 0.39 (0.018) AB B1 Orchard 0.38 (0.018) 0.42 (0.019) 0.42 (0.018) 0.15 (0.01) 0.19 (0.012) 0.4 (0.018) AB B2 Natural 0.37 (0.018) 0.41 (0.02) 0.4 (0.018) 0.14 (0.01) 0.18 (0.012) 0.39 (0.019) AB B2 Orchard 0.37 (0.019) 0.41 (0.02) 0.41 (0.019) 0.15 (0.01) 0.17 (0.013) 0.4 (0.019) AB C Natural 0.36 (0.018) 0.4 (0.02) 0.41 (0.018) 0.12 (0.011) 0.15 (0.013) 0.39 (0.019) AB C Orchard 0.37 (0.018) 0.4 (0.019) 0.41 (0.018) 0.12 (0.011) 0.15 (0.013) 0.39 (0.018) AB J Natural 0.36 (0.017) 0.4 (0.019) 0.39 (0.018) 0.16 (0.01) 0.18 (0.012) 0.37 (0.017) AB J Orchard 0.36 (0.018) 0.4 (0.019) 0.4 (0.018) 0.14 (0.011) 0.17 (0.013) 0.39 (0.018) AB K1 Natural 0.37 (0.019) 0.4 (0.02) 0.42 (0.019) 0.12 (0.011) 0.16 (0.013) 0.39 (0.019) AB K1 Orchard 0.39 (0.018) 0.42 (0.019) 0.43 (0.018) 0.17 (0.01) 0.21 (0.012) 0.42 (0.019) BC BV low Natural 0.39 (0.017) 0.43 (0.018) 0.41 (0.017) 0.21 (0.011) 0.23 (0.013) 0.43 (0.018) BC BV low Orchard 0.43 (0.017) 0.47 (0.018) 0.45 (0.017) 0.24 (0.011) 0.28 (0.012) 0.47 (0.018) BC CP low Natural 0.39 (0.017) 0.44 (0.018) 0.41 (0.018) 0.22 (0.01) 0.24 (0.012) 0.42 (0.018) BC CP low Orchard 0.42 (0.017) 0.46 (0.018) 0.44 (0.017) 0.22 (0.011) 0.25 (0.013) 0.46 (0.018) BC EK low Natural 0.46 (0.016) 0.48 (0.017) 0.49 (0.016) 0.26 (0.01) 0.34 (0.011) 0.52 (0.017) BC EK low Orchard 0.48 (0.016) 0.49 (0.017) 0.51 (0.017) 0.34 (0.011) 0.42 (0.011) 0.54 (0.017) BC NE low Natural 0.49 (0.016) 0.5 (0.017) 0.52 (0.016) 0.29 (0.011) 0.38 (0.012) 0.55 (0.017) BC NE low Orchard 0.5 (0.016) 0.53 (0.017) 0.52 (0.017) 0.35 (0.012) 0.43 (0.011) 0.55 (0.017) BC PG low Natural 0.43 (0.016) 0.48 (0.018) 0.45 (0.017) 0.26 (0.011) 0.3 (0.012) 0.47 (0.018) BC PG low Orchard 0.44 (0.016) 0.47 (0.017) 0.46 (0.017) 0.25 (0.01) 0.29 (0.012) 0.47 (0.017) BC TO low Natural 0.46 (0.016) 0.48 (0.017) 0.48 (0.017) 0.3 (0.01) 0.36 (0.01) 0.52 (0.017) BC TO low Orchard 0.49 (0.016) 0.51 (0.017) 0.51 (0.017) 0.33 (0.011) 0.42 (0.011) 0.55 (0.017)  Table C.1 Mean frequency of positive effect alleles associated with each trait by breeding zone and seedling type. Standard errors are in brackets. 130   Climatic Variable Seedling Height Growth Rate Shoot Dry Mass Growth Initiation Growth Cessation Cold Injury Natural Selected Natural Selected Natural Selected Natural Selected Natural Selected Natural Selected LAT 0.69 0.71 0.55 0.59 0.77 0.74 0.42 0.67 0.56 0.70 0.71 0.70  (0.001) (0.001) (0.006) (0.004) (0.0002) (0.0003) (0.024) (0.0012) (0.005) (0.001) (0.001) (0.001) LONG 0.00 0.11 0.05 0.22 0.02 0.08 0.14 0.15 0.04 0.10 0.01 0.12  (0.93) (0.29) (0.49) (0.13) (0.69) (0.38) (0.24) (0.21) (0.55) (0.32) (0.81) (0.28) ELEV 0.00 0.12 0.02 0.22 0.01 0.10 0.07 0.13 0.02 0.11 0.00 0.11  (0.95) (0.28) (0.66) (0.13) (0.76) (0.32) (0.41) (0.25) (0.67) (0.30) (0.99) (0.29) MAT  0.75 0.89 0.68 0.86 0.81 0.91 0.50 0.83 0.62 0.84 0.75 0.86  (0.0003) (<0.0001) (0.0009) (<0.0001) (0.0001) (<0.0001) (0.01) (<0.0001) (0.0024) (<0.0001) (0.0003) (<0.0001) log MAP 0.21 0.16 0.17 0.19 0.28 0.16 0.05 0.14 0.11 0.11 0.20 0.13  (0.135) (0.20) (0.185) (0.155) (0.075) (0.19) (0.51) (0.24) (0.31) (0.29) (0.15) (0.24) SHM  0.56 0.82 0.62 0.86 0.42 0.78 0.81 0.86 0.72 0.84 0.61 0.84  (0.005) (<0.0001) (0.002) (<0.0001) (0.023) (0.0001) (0.0001) (<0.0001) (0.0004) (<0.0001) (0.003) (<0.0001) DD > 5 0.71 0.80 0.57 0.76 0.79 0.82 0.41 0.73 0.57 0.79 0.62 0.77  (0.0006) (0.0001) (0.0044) (0.0002) (0.0001) (<0.0001) (0.025) (0.0004) (0.005) (0.0001) (0.002) (0.0002) NFFD 0.72 0.77 0.64 0.74 0.77 0.79 0.47 0.71 0.59 0.75 0.66 0.75  (0.0005) (0.0002) (0.002) (0.0003) (0.0002) (0.0001) (0.014) (0.0006) (0.003) (0.0003) (0.0013) (0.0003) EMT  0.91 0.93 0.88 0.91 0.89 0.94 0.77 0.90 0.85 0.91 0.94 0.92  (<0.0001) (<0.0001) (<0.0001) (<0.0001) (<0.0001) (<0.0001) (0.0002) (<0.0001) (<0.0001) (<0.0001) (<0.0001) (<0.0001) PC1 0.94 0.95 0.86 0.92 0.95 0.96 0.73 0.91 0.85 0.93 0.93 0.94  (<0.0001) (<0.0001) (<0.0001) (<0.0001) (<0.0001) (<0.0001) (0.0004) (<0.0001) (<0.0001) (<0.0001) (<0.0001) (<0.0001) PC2 0.00 0.04 0.00 0.07 0.04 0.02 0.06 0.06 0.01 0.05 0.00 0.05   (0.86) (0.55) (0.91) (0.4) (0.56) (0.65) (0.45) (0.43) (0.73) (0.48) (0.88) (0.50)  Table C.2 r2 values for clines with climate in the frequency of positive effect alleles by their respective associated trait. p-values (in brackets) are statistically significant at an adjusted α = 0.0045 cut-off value for 11 comparisons. Climate e variable abbreviations are defined in Table A.1 131    Figure C.1 Height: heat map of pairwise linkage disequilibrium (r2) for the top 1% of height associated SNPs for a) natural seedlings, and b) selected seedlings. 317 SNPs are hierarchically clustered on both axes. 132    Figure C.2 Growth rate: heat map of pairwise linkage disequilibrium (r2) for the top 1% of growth rate associated SNPs for a) natural seedlings, and b) selected seedlings. 317 SNPs are hierarchically clustered on both axes. 133    Figure C.3 Shoot mass: heat map of pairwise linkage disequilibrium (r2) for the top 1% of shoot mass associated SNPs for a) natural seedlings, and b) selected seedlings. 317 SNPs are hierarchically clustered on both axes. 134    Figure C.4 Cold injury: heat map of pairwise linkage disequilibrium (r2) for the top 1% of cold injury associated SNPs for a) natural seedlings, and b) selected seedlings. 317 SNPs are hierarchically clustered on both axes. 135    Figure C.5 Growth initiation: heat map of pairwise linkage disequilibrium (r2) for the top 1% of growth initiation associated SNPs for a) natural seedlings, and b) selected seedlings. 317 SNPs are hierarchically clustered on both axes. 136   Figure C.6 Growth cessation: heat map of pairwise linkage disequilibrium (r2) for the top 1% of growth cessation associated SNPs for a) natural seedlings, and b) selected seedlings. 317 SNPs are hierarchically clustered on both axes. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0349088/manifest

Comment

Related Items