Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Clinal variation at putatively adaptive polymorphisms in mature populations of Sitka spruce (Picea sitchensis… Lobo, Nina L. 2011

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

Item Metadata

Download

Media
24-ubc_2011_fall_lobo_nina.pdf [ 1.83MB ]
Metadata
JSON: 24-1.0103315.json
JSON-LD: 24-1.0103315-ld.json
RDF/XML (Pretty): 24-1.0103315-rdf.xml
RDF/JSON: 24-1.0103315-rdf.json
Turtle: 24-1.0103315-turtle.txt
N-Triples: 24-1.0103315-rdf-ntriples.txt
Original Record: 24-1.0103315-source.json
Full Text
24-1.0103315-fulltext.txt
Citation
24-1.0103315.ris

Full Text

CLINAL VARIATION AT PUTATIVELY ADAPTIVE POLYMORPHISMS IN MATURE POPULATIONS OF SITKA SPRUCE (Picea sitchensis (Bong.) Carr.) by Nina L. Lobo B.A., Mount Holyoke College, 2000  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Forestry)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) May 2011 © Nina L. Lobo, 2011  Abstract Common garden experiments in widely distributed tree species have demonstrated that phenotypic traits timing of bud set exhibit clinal variation across provenance climatic and geographic gradients, emphasizing the importance of these traits in local adaptation. With rapid advances in molecular techniques, spatial patterns of genomic variation underlying these traits can also be studied. Here I assess whether 17 putatively adaptive single nucleotide polymorphisms (SNPs) previously shown to be statistically associated with cold adaptation phenotypes vary clinally along a temperature gradient in natural, mature populations of Sitka spruce (Picea sitchensis). I also test the hypothesis that clinal strength is stronger in mature spruce populations than in seedling populations due to selection. Regressions were run for each of the 17 SNPs with logit-transformed major allele frequency as the dependent variable and provenance mean annual temperature (MAT) as the independent variable. Next, differences in strength of clines between mature and seedling populations were estimated for each SNP separately and for the 17 SNPs as a group. Finally, I ran two alternate analyses – a full analysis that included all seedling populations and a truncated analysis that limited the range of MAT observed in seedling populations to match that of mature populations. My results vary between the full and truncated analyses. In seedlings, the full analysis revealed clines in 11 SNPs (65%) compared to six SNPs (35%) in the truncated analysis. Mature populations had significant clines for five SNPs (29%). For the full analysis, the group test supported the one-sided hypothesis that mature populations have significantly steeper clines than seedlings across SNPs (P=0.027). Parallel clines in seedling and mature populations were observed for a subset of the SNPs, which strengthens their importance for local adaptation. However, low power limited my ability to make conclusive statements about differences in clinal strength between mature and seedling populations. While most SNPs were present in most populations, I also observed that the northern, disjunct population of Kodiak Island, AK was fixed for the highest proportion of SNPs (59%). This suggests that this recently founded population may lack adaptive diversity to respond to rapid climate change in the future.  ii  Table of contents Abstract .................................................................................................................................... ii Table of contents .................................................................................................................... iii List of tables............................................................................................................................. v List of figures ......................................................................................................................... vii Acknowledgements .............................................................................................................. viii 1  Chapter: Introduction ...................................................................................................... 1 1.1 Local adaptation at the molecular level ................................................................................ 2 1.2 Adaptation to cold in trees .................................................................................................... 4 1.3 Climate change and local adaptation..................................................................................... 6 1.4 Ecological genetics of Sitka spruce....................................................................................... 7 1.4.1 Spatial structure of Sitka spruce ....................................................................................... 8 1.4.2 Local adaptation at the genomic level ............................................................................ 10 1.5 Genotype-environment associations and clinal variation at putatively adaptive loci ......... 10 1.5.1 Examples of environmental associations in the literature............................................... 11 1.5.2 Disentangling the effects of population structure from local adaptation ........................ 13 1.6 Research objectives and hypotheses ................................................................................... 14  2 Chapter: Rangewide patterns of variation ................................................................... 18 2.1 Introduction ......................................................................................................................... 18 2.2 Materials and methods ........................................................................................................ 21 2.2.1 Sampling, genotyping and phenotyping of seedling populations ................................... 21 2.2.2 Sampling and genotyping of mature tree populations .................................................... 22 2.2.3 Data analysis ................................................................................................................... 23 2.2.3.1 Linkage disequilibrium .......................................................................................... 23 2.2.3.2 Genetic diversity and F-statistics ........................................................................... 24 2.2.3.3 Genotype-environment associations for putatively adaptive SNPs ....................... 24 2.2.3.4 Analysis of covariance (ANCOVA) ...................................................................... 25 2.2.3.5 Comparing differences in slopes between populations of mature trees and seedlings ... ............................................................................................................................... 26 2.2.3.6 Assessment of plateaus in allele frequency clines in seedlings ............................. 27 2.3 Results ................................................................................................................................. 28 2.3.1 Linkage disequilibrium in mature tree populations ........................................................ 28 2.3.2 Heterozygosity and FST ................................................................................................... 28 2.3.3 Associations between putatively adaptive SNPs and mean annual temperature ............ 29 2.3.3.1 Single-locus clines ................................................................................................. 29 2.3.3.2 Joint analysis of clines in populations of seedlings and mature trees .................... 29 2.3.3.3 Assessment of plateaus in allele frequency clines in seedling populations ........... 31 2.4 Discussion ........................................................................................................................... 31 2.4.1 Results differ between the full mean annual temperature range and the truncated mean annual temperature range for seedling populations ........................................................ 32 2.4.2 Replication of clines in independent populations of mature trees .................................. 33 2.4.3 Mixed results for comparisons of clines between populations of mature trees and seedlings ......................................................................................................................... 35 2.4.4 SNPs without phenotypic associations ........................................................................... 36 2.4.5 Geographic distribution of putatively adaptive standing variation in mature populations . ........................................................................................................................................ 37  iii  2.5  3  Conclusion .......................................................................................................................... 38  Chapter: Conclusions and future directions ................................................................ 49 3.1 3.2 3.3  Overall conclusions ............................................................................................................. 49 Applications ........................................................................................................................ 50 Future directions ................................................................................................................. 50  References .............................................................................................................................. 52 Appendix: Single-locus clines for mature and seedling populations ................................ 61  iv  List of tables Table 1.1. Percent variance explained for SNPs significantly (at the 10% level) associated with budset timing and/or December cold hardiness (from Holliday et al. 2010). ................. 16 Table 2.1. Geographic locations and mean annual temperature (MAT, °C ) of study populations. Sample sizes refer to the number of individuals genotyped per population. Marker-specific sample sizes vary somewhat based on genotyping success rates. ................ 39 Table 2.2. Summary of significant genotype-phenotype associations (data obtained from Holliday et al. 2010a). A1 and A2 refer to the two alternate nucleotide homozygotes, while A1A2 refers to the heterozygote. Cold injury is based on results from artificial freeze tests. ................................................................................................................................................. 40 Table 2.3. (a) Mature population pairwise FST estimates for 17 putatively adaptive SNPs (SNPs with phenotypic associations); (b) Mature population pairwise FST estimates for 14 SNPs without phenotypic associations. .................................................................................. 41 Table 2.4. Observed and expected heterozygosity for 17 putatively adaptive SNPs and 14 SNPs without phenotypic associations. .................................................................................. 42 Table 2.5. Single-locus clines for the 17 putatively adaptive SNPs whose allele frequencies are expected to vary clinally across the range of Sitka spruce. The "full range" includes 13 seedling populations. The "truncated range" includes eight seedling populations; five populations were excluded in order to match the mean annual temperature ranges of mature tree and seedling populations. ................................................................................................. 43 Table 2.6. Single-locus clines for the 14 SNPs without phenotypic associations. The "full range" includes 13 seedling populations. The "truncated range" includes eight seedling populations; five populations were excluded in order to match the temperature ranges of mature tree and seedling populations. ..................................................................................... 44 Table 2.7. ANCOVA of mature and seedling populations for the 17 putatively adaptive SNPs whose allele frequencies are expected to vary clinally. (a) All 13 seedling populations are included, and (b) Only eight seedling populations are included as the mean annual temperature (MAT) range of seedling populations is truncated to match the MAT range of mature tree populations. .......................................................................................................... 45 Table 2.8. Summary of slope differences between mature and seedling populations for putatively adaptive SNPs that showed significant associations with mean annual temperature in the ANCOVA. (a) All 13 seedling populations are included, and (b) Only eight seedling populations are included as the mean annual temperature (MAT) range of seedling populations is truncated to match the MAT range of mature tree populations. ...................... 46  v  Table 2.9. ANCOVA of mature populations and seedlings for the 14 SNPs without phenotypic associations. (a) All 13 seedling populations are included, and (b) Only eight seedling populations are included as the mean annual temperature (MAT) range of seedling populations is truncated to match the MAT range of mature tree populations. ...................... 47  vi  List of figures Figure 1.1. Range of Sitka spruce (Mimura and Aitken 2007) ............................................... 17 Figure 2.1. Range and locations of sampled populations of Sitka spruce. “CC” refers to central, continuous, “CD” to central, disjunct, “PC” to peripheral, continuous and “PD to peripheral, disjunct (Gapare et al., 2005). Numbers indicate the number of putatively adaptive SNPs with clinal effects on allele frequency that are fixed in each population. ...... 48  vii  Acknowledgements First, I thank my supervisor, Sally Aitken whose knowledge, exceptional support and guidance helped make this project possible. I also acknowledge my committee members Rob Guy, Kermit Ritland, Mike Whitlock and Jeannette Whitton for their insights, suggestions and rapid feedback. In particular, I thank Mike Whitlock and Kermit Ritland for their help with data analysis. I also thank Jason Holliday for providing me with data that made my project possible as well as promptly responding to my various requests for advice. Thanks also to Washington Gapare for collecting and saving the tissue samples that allowed me to pursue this project. For their patience and advice as I learned my way around the lab, I would like to express my gratitude to Carol Ritland, Hesther Yueh, Allyson Miscampbell and Jill Hamilton. I am also grateful to have met all members of the Aitken lab whose moral support and good conversations made my time here all the more rewarding. I hope we stay in touch. Finally, I thank my husband, Tamir Moustafa, whose love, support and wisdom help me keep perspective. This project was funded by the Forest Investment Account of British Columbia through the Forest Genetics Council, and by an NSERC Discovery Grant to Sally Aitken.  viii  1 Chapter: Introduction Genecological studies of temperate tree species with broad geographic distributions have demonstrated strong clinal variation corresponding to geographic or climatic gradients in adaptive traits such as height growth and timing of autumn growth cessation (Matyas 1996; Morgenstern 1996; Rehfeldt et al. 1999; Hall et al. 2007). These clines are interpreted as being the outcome of divergent selection pressures in different environments that alter the local optimum value of an adaptive trait. Selection thus results in the adaptation of populations to local biotic and abiotic conditions. While the geographic differentiation of adaptive traits has been repeatedly characterized, less is known about the genetic basis of this adaptive variation (Savolainen and Pyhajarvi 2007). Many phenotypic adaptive traits are polygenic in nature, with each gene having only a small effect on phenotype. Nevertheless, these minor contributions, when aggregated across many genes, have a significant impact on trait values (McKay and Latta 2002). Thus, an understanding of the genetic differentiation underlying adaptive traits is fundamental to a deeper understanding of local adaptation, including the potential for, and limits to, adaptation. Recent advances in molecular biology techniques have enabled the rapid sequencing and genotyping of many individuals, generating volumes of genetic data (Allendorf et al. 2010). As a result, much progress has been made in characterizing the genomic basis of adaptation in models such as Drosophila, Mimulus, Populus, Arabidopsis, as well as humans. With climate change, populations of forest trees are likely to become increasingly mismatched to local conditions. As climate change progresses, selecting the appropriate seed source for reforestation will involve transferring seed from locations that are expected to be optimally adapted to future predicted climates, instead of choosing more local seed sources. Knowledge of the functions of genes that control adaptive traits coupled with patterns of geographic variation at these loci can assist both conservationists and forest managers in assessing the adaptive potential of populations and predicting the performance of different genotypes (Howe et al. 2003).  1  1.1  Local adaptation at the molecular level Provenance trials in tree species have repeatedly demonstrated high levels of among-  population differentiation for quantitative traits (QST) despite low levels of genetic differentiation (FST) at neutral loci (Jaramillo-Correa et al. 2001; Savolainen et al. 2007). FST for a locus is estimated from allele frequency variation among populations; hence, populations that differ widely in allele frequencies have stronger FST values. High QST compared to low FST for presumably selectively neutral genetic markers suggests that the quantitative traits in question are under selection. Importantly, high QST in combination with low FST means that divergent selection is possible even in the face of high gene flow (Leinonen et al. 2008). Such comparisons, however, do not capture spatial patterns of molecular variation underlying QST (Reed and Frankham 2001; McKay and Latta 2002; Knopp et al. 2007). For example, a meta-analysis by Reed and Frankham (2001) revealed that molecular measures of genetic differentiation based on neutral markers explained a meagre 4% of quantitative trait variation, indicating that neutral markers are not a reasonable proxy for loci underlying quantitative traits. A more recent meta-analysis by Leinonen et al. (2008) found that neutral FST predicts QST, but the correlation between the two measures was low (r2 = 0.16). Hence, there is a need to independently detail the molecular variation underlying quantitative traits that appear to be under selection. Comparing FST patterns between putatively adaptive and neutral markers is a popular method for capturing loci under selection in natural populations (Akey et al. 2002; Luikart et al. 2003; Nielsen 2005). FST for markers under divergent selection (or for markers that hitchhiked with the selected loci) is expected to exceed genetic differentiation at neutral loci, while markers under balancing selection should have lower FST than expected under neutrality (Hamblin et al. 2002; Kohn et al. 2003; Hahn et al. 2004). This rationale has been used to develop FST outlier tests, which detect locus-specific FST values that are significantly higher or lower than expected against a null FST distribution. The null distribution reflects the range of FST values expected under neutrality, including specific demographic scenarios. An inaccurate model of population structure when simulating the null distribution of population differentiation can influence the detection of FST outliers. For example, Excoffier et al. (2009) developed a hierarchical island model to account for varying levels of gene flow 2  within and between island groups. In species with a hierarchical structure, this model helped reduce the proportion of false positives detected by FST outlier tests. An underlying problem with outlier methods for quantitative trait loci (QTL), which characterize many ecologically important traits, is that genetic differentiation at QTL may not exceed differentiation at neutral loci (Latta 1998; Le Corre and Kremer 2003). Allele frequency shifts in response to divergent selection on a trait create between-population linkage disequilibrium among loci. Linkage disequilibrium generates covariances in allele frequencies at the underlying QTL, which have a stronger effect on quantitative trait variation (QST) than the sum of locus-specific changes in allele frequency (Kremer 2000). For example, positive covariances increase the variance of a trait (QST) to a greater degree than the additive effects of individual loci. Therefore, testing individual loci for selection may fail to reveal outliers. Low levels of genetic differentiation at QTL are more likely to occur for highly polygenic traits. In addition, evolutionary forces other than divergent (or balancing) selection can create genomic patterns that resemble selection. For example, purifying selection or other forces that reduce local effective population size (Ne) across the genome can lead to high FST values (Storz 2005). One way to distinguish between divergent selection and other factors that leave a similar mark on the genome is to combine outlier tests with site frequency spectrum statistics that assess population-specific patterns of genetic diversity. A site frequency spectrum refers to the proportion of sites in a sample of DNA sequences with different numbers of copies of an allele (Hartl and Clark 2007). Two major statistics that aim to detect selective patterns from site frequency spectra are Tajima’s D (Tajima 1989) and Fay and Wu’s H (Fay and Wu 2000). Tajima’s D uses two measures of genetic variation, the number of segregating sites and average nucleotide heterozygosity in a sample to detect an excess of low-frequency variants, which is expected under positive selection (Tajima 1989). However, an excess of low-frequency variants may also be present due to factors other than spatially varying selection, such as background selection or population growth. To tease apart the effect of these confounding factors from divergent selection, Fay and Wu’s H is typically estimated with Tajima’s D. Fay and Wu’s H is able to detect a selective sweep as it is sensitive to an excess of high-frequency allelic variants, which occurs following a  3  hitchhiking event but is absent under purifying selection or population growth (Fay and Wu 2000). Outlier detection methods and frequency spectrum statistics are thus a first step towards detecting signatures of selection in the genome. Markers that appear to be under selection can be further analyzed by assessing their putative functional role (for e.g, see Namroud et al. 2008) as well as testing their spatial patterns of variation against adaptive expectations. A method that is increasingly being employed to detect the presence of genomic adaptation, often in conjunction with outlier and frequency spectrum methods, is association mapping. This method involves statistically testing the relationship between individual genetic markers such as single nucleotide polymorphisms (SNPs) and the phenotype of individuals for a particular trait via regression analysis or other related techniques (Neale and Savolainen 2004; Gonzalez-Martinez et al. 2006). Since underlying neutral population structure has the potential to create spurious associations between markers and traits (Zhao et al. 2007), population structure is explicitly taken into account in association studies using techniques such as Price et al.’s (2006) principal component analysis or the mixed linear model developed by Yu et al. (2006), which allows for the removal of both population structure (e.g., STRUCTURE’s Q matrix (Pritchard et al. 2000)), and co-ancestry among individuals.  1.2  Adaptation to cold in trees Trees have developed an exceptional ability to withstand the wide range of climatic  conditions they experience seasonally, particularly below-freezing temperatures in the case of temperate and boreal species. While actively growing, trees are susceptible to cold damage and death (Howe et al. 2003). However, when acclimated to cold, species can survive a wide range of below-freezing temperatures. A number of environmental cues operate in concert to induce cold acclimation. In the fall, long nights induce trees to cease active growth, set bud, enter winter dormancy and initiate cold acclimation. Next, in response to low temperatures, trees enter a deeper state of cold hardiness, breaking bud in the spring only after specific chilling requirements  4  (continuous exposure for a sufficient period to temperatures slightly above freezing) and heat sum requirements have been met (Sakai and Larcher 1987). Populations of the same species and individuals within the same population show genetic variation in their degree of cold adaptation (Campbell and Sugano 1979; Howe et al. 1995; Chuine et al. 2001). This is particularly true of fall cold hardiness and timing of budset, which have been found to be strongly correlated with climatic and geographic gradients across species (reviewed in Morgenstern 1996). A typical pattern is that trees from colder geographic regions such as higher latitudes or elevations set bud earlier in the fall and flush later in the spring. Timing of spring flush, however, does not vary to as great a degree among populations as timing of bud set (Howe et al. 2003). While classical common garden and reciprocal transplant experiments have been the standard tools for shedding light on spatial variation in cold adaptation, genomic data is increasingly being used to understand the genetic underpinnings of climatic adaptation in tree populations (Neale and Kremer 2011). For species like forest trees that are characterized by high levels of genetic diversity and low linkage disequilibrium, association mapping has become the technique of choice for detecting markers associated with phenotype (Neale and Savolainen, 2004; Gonzalez-Martinez et al. 2006; Neale and Kremer 2011). In conjunction with association studies, the significance of candidate genes in cold adaptation can be made stronger by undertaking genotype-environment correlation studies, where environmental variables of provenance environments such as annual temperature or frost-free period are correlated to frequencies of genetic variants (Ingvarsson et al. 2006; Samis et al. 2008; Holliday et al. 2010a). Determining the putative functional role of genes associated with cold adaptation may also help reveal their importance in adaptation. For example, in their genotype-phenotype association study on Sitka spruce (Picea sitchensis (Bong.) Carr.), Holliday et al. (2010a) found that SNPs with the greatest effects on phenotype lay within a xyloglucan:xyloglucosyl transferase gene (xth) and a peroxidise gene (per) that is involved in reducing oxidative stress. xth may play a role in altering cell wall flexibility, a trait that is expected to enhance cold tolerance although annotations of this gene have not previously included this function. Cellular dehydration caused by osmotic stress from freezing temperatures reduces cell volume as water is removed from the cell (Moore et al. 2006). An elastic cell wall may  5  therefore prevent dehydrated cells from collapsing as tension increases within the cells. Holliday et al. (2010a) noted significant associations for five additional genes that are also likely involved in reducing oxidative stress. Other major functional groups included light signalling (for example, a homolog to phytochrome A), auxin regulation and cryptochromelike genes. A similar association mapping study in Douglas-fir (Pseudotsuga menziesii) revealed candidate genes with putative functions involving lignin biosynthesis, cell wall structure, transcription regulation and signal transduction (Eckert et al. 2009).  1.3  Climate change and local adaptation The IPCC (2007) projects that the future holds higher mean temperatures and changes  in precipitation regimes as well as greater climatic variability with more frequent extreme climatic events. With rapid climate change, populations can track their climate by migrating to more favourable locales or persist in their current environments by adapting to climate (or in the shorter term, by being plastic) (Aitken et al. 2008). The alternative is the extirpation of populations. This is especially relevant for tree species as the fossil record indicates that while trees have tracked their climate during the Quaternary (Davis and Shaw 2001), slower migration rates than necessary under current climate change scenarios make adaptation all the more important for the survival of certain populations and species (Malcolm et al. 2002; Iverson et al. 2004). For example, based on the uniqueness and diversity of chloroplast DNA (cpDNA) haplotypes, Anderson et al. (2006) found evidence for white spruce (Picea glauca) glacial refugia in Alaska. Their findings argue against a rapid migration of the species into Alaska following the last glaciation as was previously hypothesized, and instead provide support for the species’ ability to survive climate change in situ in locally favourable habitats. These findings are corroborated by McLachlan et al. (2005) who found evidence from cpDNA for glacial refugia in American beech (Fagus grandifolia) and red maple (Acer rubrum) within close proximity of the Laurentide ice sheet. Their results revealed substantial discrepancy between migration estimates obtained from molecular data and the fossil pollen record. Molecular data estimated a slow recolonization rate during the early Holocene of approximately 80 – 90 m2/yr while previously the fossil pollen record had been used to estimate a more rapid migration rate of approximately 150 – 350 m2/yr. The presence of cryptic northern refugia suggests that the migration ability of populations is more limited 6  than was previously estimated via fossil pollen reconstruction. Therefore, two major questions are whether tree species have sufficient genetic diversity to adapt to climate change, and whether they can respond rapidly enough given long generation times coupled with rapidly changing environments (Aitken et al. 2008). Even if genetic diversity appears to be sufficient for rapid adaptation, ecological factors such as dispersal, density and interspecific relationships will largely determine how well species track their climatic niche. By simulating the impact of ecological factors on local density, Atkins and Travis (2010) showed that extirpation can occur even if there is significant overlap between current and future predicted ranges. Their simulations suggest that critical to the survival of a species, aside from the dispersal ability of pollen, is the ability of individuals from other portions of the range to establish in areas that are already occupied by individuals of the same species that may currently be maladapted to local conditions. Thus, species with broad niches are not immune from extinction as densitydependent issues and degree of pre-existing maladaptation influence extinction potential. Long-term common garden experiments have provided valuable information on the relative ability of populations to survive a range of climatic conditions. In conjunction with these studies, or when long-term data on adaptive phenotypic traits is unavailable, characterizing the distribution of adaptive molecular diversity within and among populations will provide insights into the adaptive potential of populations.  1.4  Ecological genetics of Sitka spruce Sitka spruce is an ecologically and economically important conifer that spans a wide  latitudinal and climatic range from northern California to Alaska (Figure 1.1). Prior research on this species revealed strong geographic clines for traits associated with growth and cold adaptation such as height growth, growth rate, budset timing and cold hardiness. Burley (1966) demonstrated a clinal relationship between budset timing and provenance latitude in a greenhouse experiment. He also found that the relationship between height growth, budset timing and provenance latitude was modified by experimental variation of temperature and photoperiod; higher temperatures and longer days in the greenhouse delayed bud formation and increased height. These results were supported by Xu et al. (2000) and Mimura and Aitken (2007a, 2010). For example, the role of provenance photoperiod on height growth 7  was noted by Xu et al. (2000), who used multivariate statistics to demonstrate that differences in photoperiod created a north-south cline in height. In an outdoor common garden, Mimura and Aitken (2007a, 2010) showed that southern populations of seedlings grow taller than northern ones, which have adapted to utilize the longer growing season in the south. In contrast, northern populations have a faster growth rate, allowing them to maximize growth within the shorter available growth period. Higher-latitude populations set bud earlier in the fall and were cold hardier (suffered less cold injury during freezing tests), thereby avoiding frost injury from early fall frosts. On the other hand, populations from lower latitudes, which experience frosts later in the fall, maximized growth by setting bud at a later date and being less cold hardy than more northern populations. Estimates of population differentiation for quantitative traits were high and similar for budset timing and cold hardiness (QST = 0.89) (Mimura and Aitken 2007a).  1.4.1  Spatial structure of Sitka spruce Using six microsatellite markers, Mimura and Aitken (2007a) showed that Sitka  spruce displays a stepping-stone pattern of migration and moderate isolation by distance (IBD), where genetic correlations among populations decrease with geographic distance. A stepping-stone migration pattern was further corroborated by Holliday et al. (2010b) who found evidence supporting a south to north post-glacial recolonization pattern in the site frequency spectrum (SFS) of six Sitka spruce populations from across the species range for 153 nuclear genes. Coalescent simulations revealed that a bottleneck model best explained the SFS for Holliday et al.’s northernmost populations, Kodiak Island, Valdez and Prince Rupert. A strong relationship (R2 = 0.89) was also detected between bottleneck timing and distance from the southern limit of the species range, with populations in the south possessing molecular signatures of older bottlenecks compared to northern populations. Sitka spruce also shows a relatively strong central-peripheral structure (Mimura and Aitken 2007b), which suggests that the species has fairly rapidly developed an “abundantcentre distribution” (Brown 1984) following the colonization of previously glaciated areas. A comparison of pairs of continuous and disjunct populations from the northern periphery, southern periphery and center of the range demonstrated diminished microsatellite allelic richness and changes in mating system from the range centre to the peripheries (Mimura and 8  Aitken 2007b). For example, populations at the core of the range had a higher number of effective pollen donors and lower rates of selfing compared to peripheral and particularly peripheral disjunct populations. A reduction in effective pollen donors and an increase in inbreeding suggest that populations at the range peripheries have lower densities or effective population sizes than core populations (Loveless and Hamrick 1984). In addition, Sitka spruce’s range width narrows towards the peripheries, which reduces its habitable geographic area and absolute density in the north and south. Density differences between the core and range edges make it likely that the distribution fits the Garcia-Ramos and Kirkpatrick (1997) model of adaptation and gene flow, where gene flow is more abundant from the higher-density core to the lowerdensity peripheries. Asymmetric gene flow has the potential to cause maladaptation at the range peripheries by homogenizing peripheral populations and reducing their ability to track their phenotypic optima (Garcia-Ramos and Kirkpatrick 1997; Kirkpatrick and Barton 1997). The homogenizing effects of gene flow are likely to be most apparent at the seedling stage, which has not undergone a cycle of selection. Neither strong IBD nor a core-periphery structure were supported by Gapare and Aitken’s (2005) analysis of genetic structure in eight natural populations of mature trees using sequence-tagged-site (STS) loci, or by Yeh and El-Kassaby’s (1980) study of isozyme variation in ten rangewide populations. This discrepancy likely results from the use of different types of markers as microsatellites have higher mutation rates and levels of polymorphism than STS loci and isozymes. Nevertheless, Gapare and Aitken’s (2005) research found that peripheral populations have stronger fine-scale spatial genetic structure than central populations due to stronger within-population kinship relationships. Their findings provide further support that peripheral populations of Sitka spruce likely have lower population densities than central populations. Populations with a high density of mature trees tend to have overlapping seed shadows, which reduces the amount of genetic relatedness among neighbours as seedlings of different maternal parents become established next to each other (Parker et al. 2001). Overlapping seed shadows are therefore expected to reduce kinship among neighbours. Population density also influences fine-scale structure by altering pollen dispersal patterns. Leptokurtic pollen dispersal, which enhances pollen diversity and reduces correlated paternity among maternal siblings, is thought to be more pronounced in  9  larger populations (Robledo-Arnuncio et al. 2004). This source of variation may be lost in low-density populations where matings likely occur primarily among neighbouring individuals.  1.4.2  Local adaptation at the genomic level To achieve a more comprehensive understanding of local adaptation at the molecular  level, Holliday et al. (2010a) recently conducted an association study that related bud set timing and cold hardiness of Sitka spruce seedlings to their underlying genotype. They genotyped seedlings for 339 single nucleotide polymorphisms (SNPs) in 153 candidate genes for adaptation to cold. Holliday et al. found 45 significant genotype-phenotype associations for 35 SNPs within 28 genes (Table 1.1). Percent of phenotypic variance of quantitative traits explained by individual SNPs (PVE) ranged from 1.0% to 4.3% for budset timing and 0.7% to 5.4% for December cold hardiness. Collectively, the SNPs explained 28% and 34% of phenotypic variance in cold hardiness and budset timing respectively (Holliday et al. 2010). Cumulative SNP effects are close to heritability estimates for cold hardiness (0.30) and budset (0.32) in an outdoor seedling common garden in Vancouver (Mimura and Aitken, 2007a). Population sample sizes in Holliday et al’s. (2010) study were small and may thus overestimate effect sizes. Nevertheless, this result is surprising given the issue of missing heritability in many association studies, particularly large-scale genomic studies in humans (Manolio et al. 2009).  1.5  Genotype-environment associations and clinal variation at putatively adaptive loci In this section, I review prior research on adaptive patterns of molecular variation  along environmental gradients. I also discuss the complexities of interpreting environmental associations at the molecular level. Population demographic history, gene flow, genetic drift, mutation and natural selection all leave their mark on the genome. Gene flow promotes the homogenization of genetic variation among individuals and populations. While its homogenizing effect may hinder adaptation to local conditions, gene flow can also be an important source of standing variation for adaptation (Przeworski et al. 2005). Factors such as changes in population size and random genetic drift can create genomic patterns of  10  polymorphism that mimic natural selection (Kawecki and Ebert 2004; Neale and Ingvarsson 2008); however, this is fairly unlikely to occur in a predictable pattern across loci. While natural selection only targets specific adaptive regions of the genome, other evolutionary factors such as genetic drift and gene flow are expected to have genome-wide effects (Andolfatto 2001; Nielsen 2001; Charlesworth et al. 2003). Therefore, genomic patterns of variation that diverge from the range of patterns expected from neutral factors alone are interpreted as being the product of natural selection. For markers underlying a strongly divergent phenotypic trait, clinal variation is often cited as evidence that the marker is a likely target of selection. However, clinal variation may not be detected for a number of reasons. For a SNP that is locally adaptive, i.e., confers higher fitness in some environments but not others, allele frequency differences may be difficult to detect because only minor frequency shifts at individual loci may interact to determine the phenotypic value of polygenic traits. As a result, the strength of selection for a trait that is determined by multiple, interacting loci may be weak when each locus is considered independently (McKay and Latta 2002; Coop et al. 2009). In addition, selection acts on phenotypes and the same phenotype can result from different allelic combinations across multiple loci. Gene flow may also reduce among-population variation in allele frequencies, if selection at the allelic level is weak relative to gene flow (Savolainen et al. 2007). Nevertheless, if selection is sufficiently strong, correlations between allele frequencies and environmental gradients can still be detected although not detecting a significant association does not rule out the possibility of the SNP being locally adaptive (Sezgin et al. 2004).  1.5.1  Examples of environmental associations in the literature Some of the classic studies on putatively adaptive clines stem from research on  Drosophila species. Mettler et al. (1977) studied D. melanogaster populations along the eastern U.S. seaboard to assess the frequency of inversion polymorphisms in these populations. They discovered the presence of latitudinal clines across populations and suggested that natural selection may be maintaining these clines. Knibb et al. (1981) found that latitudinal clines present in Australasian populations at four cosmopolitan inversions mirrored the clines detected in northern hemisphere populations, including those found by 11  Mettler et al. (1977). Such parallel clines strengthen the likelihood that the inversions confer a selective advantage. Similar inter-continental, parallel clines across latitude were also found for alcohol dehydrogenase (ADH) allozymes (Oakeshott et al. 1982). However, some studies in the same species have failed to detect signatures of selection, even when the geographic area under study is virtually identical to previous studies (for e.g., see Sgro et al. 2006). With the advent of rapid molecular sequencing, the literature on single-locus clines and environmental correlations has grown to encompass a diversity of genera and species such as Arabidopsis (Samis et al. 2008), tree species such as European aspen (Populus tremula) (Ingvarsson et al. 2006, Ingvarsson et al. 2008; De Carvalho et al. 2010; Ma et al. 2010), loblolly pine (Pinus taeda) (Eckert et al. 2010) and humans (Hancock et al. 2008; Balloux et al. 2010). Here I summarize a few examples. Samis et al. (2008) conducted a detailed study on phenotypic and genetic clines in 138 Eurasian populations of Arabidopsis thaliana. Their study encompassed 1) common garden experiments to assess latitudinal and longitudinal clines in flowering response (photoperiod sensitivity), 2) an association analysis to study the relationship between flowering time and the phytochrome gene, PHYC, which is known to influence flowering time, and 3) tests to determine the presence of latitudinal or longitudinal clines in PHYC that resemble the clines in flowering phenotype. They found significant longitudinal, but not latitudinal clines in photoperiod sensitivity. Similarly, only longitudinal clines were observed for PHYC haplotypes. Counter-intuitively, the longitudinal distribution of PHYC haplotypes was in the opposite direction of the phenotypic clines. The authors cite a number of factors that may have contributed to their results such as the importance of genes other than PHYC on photoperiodic sensitivity and the possibility of pleiotropic effects for PHYC, which complicate the interpretation of the clines that they observed. Ma et al. (2010) provide examples of single-locus clines from the P. tremula model system. Again, genes involved in photoperiod sensitivity were assessed for clinal variation and genetic differentiation beyond the amount expected for neutral markers. They discovered clinal variation at four of 113 SNPs, which may have arisen from admixture between different lineages in Sweden and not solely from selection; admixture among lineages was found to promote a strong step cline in a photoperiod-sensitive trait, timing of  12  budset (De Carvalho et al. 2010). However, Ma et al. (2010) consider it unlikely that admixture was the primary driver of the molecular clines as the SNPs in their study did not show significant IBD; if gene flow between lineages were creating clines, a larger number of SNPs would be expected to vary clinally. In another study on the same species, Ingvarsson et al. (2008) regressed population allele frequencies on latitude and detected clinal variation at three of 42 SNPs in the phytochrome B2 (phyB2) locus. Again, given that clinal variation is not a common observation in P. tremula, these clines are expected to be the product of balance between migration and directional selection. In humans, climate-related clines have been detected for SNPs from candidate genes for common metabolic disorders such as type 2 diabetes, obesity, dyslipidemia and hypertension (Hancock et al. 2008). Another study by Balloux et al. (2010) used Mantel and partial Mantel tests to establish that temperature had influenced the distribution of mitochondrial DNA (mtDNA) sequence variation in human populations distributed worldwide, even after accounting for population structure.  1.5.2  Disentangling the effects of population structure from local adaptation A common theme underlying a number of these diverse studies is the importance of  disentangling neutral population structure such as IBD from the effects of divergent selection to ensure that spurious patterns of differentiation are not interpreted as the outcome of adaptation. Sitka spruce, the focal species in this study, has a relatively linear distribution along the Pacific coast, which causes climatic and geographic distances separating populations to be highly correlated; for example, the correlation coefficient (r) between mean annual temperature and distance from the southern edge of the range is 96%. A strong correlation implies that single-locus clines observed for putatively adaptive SNPs may not necessarily be the product of selection; instead, clines may represent an IBD pattern of gene flow where genetic similarity among populations decreases as the distances separating populations increase. To determine the potential for putatively adaptive single-locus clines to be the outcome of gene flow and drift, Vasemagi (2006) simulated a neutral population genetic model comprising 11 populations and varying levels of migration and drift. His simulations 13  revealed that the proportion of clinal loci increased with strength of IBD. When Vasemagi compared his results with those from an actual Drosophila melanogaster data set (Gockel et al. 2001) that also showed IBD (Mantel test r = 0.90), he found that the proportion of clinal loci in this dataset fell within the range expected for neutral loci based on his simulations. Furthermore, all except two loci from the Drosophila study had similar regression slopes and R2 estimates as the neutral, simulated loci. In a landscape genomic study in loblolly pine, Eckert et al. (2010) used a Bayesian generalized linear mixed model, which takes into account baseline neutral population structure. Incorporating population structure into their model increased the likelihood that SNPs depicting a selection signal were truly adaptive and not false positives arising due to gene flow and genetic drift. Such studies highlight the difficulty and complexity of separating neutral forces from selection in studies on clinal variation. The genomic effects of demography can be separated from environmental effects by including a set of control markers that are not considered adaptive a priori or by adopting tools such as the partial Mantel test (Balloux et al. 2010). However, Mantel tests for nondistance-based variables such as temperature and precipitation may suffer from a lack of statistical power (Legendre and Fortin 2010). Alternatively, population structure can be incorporated into statistical models using ordination techniques such as principal component analysis (PCA) or one of its variants (Patterson et al. 2006; Jombart et al. 2008; Jombart et al. 2009; Grivet et al. 2010), and Bayesian assignment methods such as STRUCTURE (Pritchard et al. 2000) and Geneland (Guillot et al. 2005). Inferences regarding the adaptive importance of alleles are strengthened if climate has a significant relationship with allele frequencies after controlling for neutral patterns of population structure (Samis et al. 2008; Gaggiotti et al. 2009). 1.6  Research objectives and hypotheses My overarching goals were to characterize geographic patterns of variation of  putatively adaptive SNPs in natural populations of mature Sitka spruce trees, and to compare these patterns to those found by Holliday et al. (2010) in seedlings of the same species. To this end, in chapter 2, I address the following questions and hypotheses:  14  (1) In mature, natural populations of Sitka spruce, do SNPs associated with cold adaptation phenotypes vary clinally along north-south gradients in temperature? Temperature has been identified as one of the most important variables in local adaptation (Howe et al. 2003). Thus, local adaptation from divergent selection pressures is expected to cause putatively adaptive SNPs to vary clinally along temperature gradients. (2) Do clinal patterns of variation in mature tree populations correspond to those found in populations of seedlings? If so, has selection strengthened correlations in mature compared to seedling populations? Similar geographic patterns of variation are observed between populations of mature trees and seedlings, with mature populations showing a higher proportion of, and stronger, environmental associations, having undergone at least one cycle of selection following mating and establishment.  15  Table 1.1. Percent variance explained for SNPs significantly (at the 10% level) associated with budset timing and/or December cold hardiness (from Holliday et al. 2010).  16  Figure 1.1. Range of Sitka spruce (Mimura and Aitken 2007)  17  2 Chapter: Rangewide patterns of variation 2.1  Introduction For sessile organisms such as trees, synchronizing phenological processes such as  timing of growth initiation and cessation to environmental cues is essential for survival and reproduction. Therefore, traits involved in cold adaptation are expected to be under strong divergent selection in temperate and boreal species. Common garden experiments have repeatedly demonstrated that ecologically relevant traits such as height, timing of growth cessation (budset) and degree of cold injury exhibit clinal variation across environmental and geographic gradients in temperate tree species and are therefore expected to be locally adaptive (Matyas 1996; Morgenstern 1996; Rehfeldt et al. 1999; Hall et al. 2007). Prior research has also highlighted that a number of these traits are under moderate to strong genetic control (Howe et al. 2003). This suggests that similar spatial patterns may be observed between adaptive phenotypes and segregating molecular variation for these traits. However, many adaptive traits are polygenic in nature, with variation at each locus making only a small contribution to phenotypic variation (Latta 1998). As a result, patterns of variation at the molecular level are unlikely to be as strong as patterns at the phenotypic level, and may even resemble neutral loci (Latta 1998; Le Corre and Kremer 2003). Numerous techniques have been developed to identify loci involved in divergent selection. One method that is particularly well suited for species such as trees that typically possess large effective population sizes and low levels of linkage disequilibrium is association mapping (Neale and Savolainen 2004; Gonzalez-Martinez et al. 2006; Neale and Kremer 2011). In an association study, an individual’s phenotype is statistically related to its genotype at a set of pre-selected loci. Pre-selected loci typically represent candidate genes for the trait of interest (for e.g., see Gonzalez-Martinez 2007). In conjunction with or following association studies, the significance of candidate genes in cold adaptation can be made stronger by undertaking genotype-environment correlation studies to assess if the geographic distribution of markers follows patterns predicted from the effect of each marker on phenotype (Samis et al. 2008; Holliday et al. 2010a). While patterns that support predictions increase the likelihood of the marker or a linked marker being adaptive, a failure to observe the predicted pattern of variation does not  18  imply that the marker is not involved in local adaptation (Sezgin et al. 2004). Statistical issues such as low power can significantly impact interpretations of the adaptive significance of a marker. In addition, statistically significant allele frequency differences may be difficult to detect because only minor frequency shifts at individual loci may interact to determine the expression of a quantitative trait and multiple genotypes can produce the same phenotype. Therefore, the strength of selection for a trait, when spread across multiple, interacting loci may be weak for each, individual locus (McKay and Latta 2002; Coop et al. 2009). Gene flow may also reduce among-population variation in allele frequencies if selection at the allelic level is weak relative to gene flow (Savolainen et al. 2007). Nevertheless, if selection is sufficiently strong, correlations between marker frequency and environmental variables are likely to be detected. Similarly, observing a predicted pattern such as a cline does not confirm that natural selection is instrumental in structuring the spatial distribution of the marker (Vasemagi 2006). Neutral forces such as the interplay between gene flow and genetic drift, and demographic factors such as changes in population size, bottlenecks and secondary introgression can create patterns of polymorphism that mimic natural selection (Neale and Ingvarsson 2008). Therefore, care has to be taken when interpreting geographic patterns depicted by putatively adaptive markers. To minimize false positives arising from neutral population structure, background levels of neutral structure can be incorporated into statistical models (for e.g., see Samis et al. 2008; Eckert et al. 2010). In forestry, knowledge of the functions of genes that control adaptive traits coupled with an understanding of their patterns of geographic variation can assist both conservationists and forest managers in estimating the adaptive potential of different populations and predicting the performance of genotypes, particularly in the face of rapid climate change (Howe et al. 2003). The importance of characterizing the adaptive potential of populations at the genomic level is only increasing as climate change is expected to alter the ecological niche of many tree species (Davis and Shaw 2001). A tree species of both economic and ecological importance in British Columbia is Sitka spruce (Picea sitchensis (Bong.) Carr.). This species has a long, linear distribution along the Pacific coast of North America, spanning a wide latitudinal range from northern California to Alaska (Figure 1.1). Prior research on Sitka spruce revealed steep geographic  19  clines for traits associated with growth and cold adaptation such as height growth, budset timing and cold hardiness, with populations showing a trade-off between height and cold hardiness (Burley 1966; Mimura and Aitken 2007a). The geographic distribution of putatively neutral genetic variation (microsatellites and single nucleotide polymorphisms) in this species also supports a stepping-stone pattern of migration (Mimura and Aitken 2007a; Holliday et al. 2010b) and a strong core-periphery structure, with populations at the centre of the range showing evidence of a higher number of effective pollen donors, greater allelic richness and lower rates of selfing compared to peripheral and particularly peripheral disjunct populations (Mimura and Aitken 2007b). Recently, to achieve a more comprehensive understanding of local adaptation at the molecular level, Holliday et al. (2010a) conducted an association study on cold adaptation. Budset timing and cold hardiness of Sitka spruce seedlings grown in a common garden were statistically related to a total of 339 single nucleotide polymorphisms (SNPs) in 153 candidate genes. Holliday et al. found 45 significant genotype-phenotype associations for 35 SNPs within 28 genes (Table 1.1). Percent of variance explained (PVE) by individual SNPs ranged from 1.0% to 4.3% for budset timing and 0.7% to 5.4% for cold hardiness. Collectively, the SNPs explained 28% and 34% of phenotypic variance in cold hardiness and budset timing respectively (Holliday et al. 2010a). In this chapter, I assess whether the 35 SNPs associated with budset timing or cold hardiness in Sitka spruce show evidence of being structured by divergent selection in independent, natural populations of mature trees. I focus on allele frequency patterns along a temperature gradient as temperature exerts a strong effect on local adaptation in temperate tree species (Rehfeldt et al. 1999; Howe et al. 2003). My primary objective is to test the hypothesis that the magnitude of associations between putatively adaptive, segregating variation and temperature is stronger in natural populations of mature trees than in seedling populations grown from seed collected in situ. Unlike mature trees, seedlings have not undergone a cycle of exogenous selection since reproducing. Therefore, at the seedling stage, homogenizing gene flow is likely to be stronger than selection, which is expected to reduce allele frequency differences among populations (Garcia-Ramos and Kirkpatrick 1997).  20  2.2 2.2.1  Materials and methods Sampling, genotyping and phenotyping of seedling populations Phenotypic and genotypic data for Sitka spruce seedlings were provided by J.A.  Holliday (Holliday et al. 2010a). Open-pollinated seed from 10-13 seed parents per population were collected from 17 rangewide populations and grown in a common garden in Vancouver established in 2003 (Figure 1.1) (Mimura and Aitken 2007a). Fourteen of the 17 populations with known within-population family structure were utilized by Holliday et al. (2010a). In 2006, Holliday et al. (2010a) set up a second common garden consisting of rooted cuttings of 410 individuals from the 14 populations. In May 2006, foliage from newly flushed lateral buds was collected for DNA extraction using a standard cetyl trimethyl ammonium bromide (CTAB) procedure. SNPs were genotyped using the Illumina GoldenGate platform. SNPs to be genotyped were selected from a panel of 202 cold-adaptation candidate genes resequenced in 24 individuals from across the species range using three criteria: (1) a minimum minor allele frequency threshold of 5%, (2) spacing between SNPs of at least 60 base-pairs as required by the GoldenGate system to avoid overlapping primers, and (3) preferential selection of nonsynonymous over synonymous SNPs (Holliday et al. 2010a). Subsequent analyses by Holliday et al. (2010a) were carried out for 339 SNPs distributed in 153 candidate genes. The vast majority of the 202 candidate genes were selected based on their differential expression during fall cold acclimation (Holliday et al. 2008, 2010a). A few candidate genes were included because their annotations suggested a functional role in either timing of budset or cold hardiness. Cold hardiness was assessed on October 1, 2007 and December 1, 2007 in artificial freeze tests by exposing detached needle segments to a range of test temperatures and measuring electrolytic leakage from the segments relative to a control treatment. A cold injury index was used to estimate the degree of cold injury of each individual (Holliday et al. 2010a). Only cold hardiness estimates from December 1, 2007 are used in my study. Budset timing was estimated in the fall of 2008 by recording the day of the year when bud scales were observed with a naked eye on a seedling`s apical bud.  21  Data on cold hardiness and bud set timing were used to conduct an association study, where the phenotype of each of the 410 seedlings was statistically related to the seedlings` genotype at the 339 SNPs (Holliday et al. 2010a). Yu et al.'s (2006) mixed linear model was adopted for the association analysis. The model, implemented in TASSEL (http://www.maizegenetics.net) allows for the removal of the effects of population structure and co-ancestry among individuals, thus avoiding false-positives due to this structure. Holliday et al. accounted for population structure by running the Bayesian program, STRUCTURE (Pritchard et al. 2000) using a separate set of 98 SNPs from candidate genes not related to adaptation to cold. Holliday et al.’s (2010) association study revealed 45 significant genotype-phenotype associations for 35 SNPs within 28 genes (Table 1.1). The 45 SNPs had call rates exceeding 90 percent on the GoldenGate assay (Holliday et al. 2010a). Population sizes for the 14 seedling populations range from three (Fort Bragg, CA) to 48 individuals (Ocean Falls, BC) (Table 2.1). Since including only three individuals in a population introduces sampling noise in estimating population allele frequencies, I excluded Fort Bragg from all analyses. The smallest sample size in my study is 13 seedlings from Columbia River, OR, and the average population size is 31 individuals. 2.2.2  Sampling and genotyping of mature tree populations In an earlier study on range-wide neutral genetic diversity and population structure in  Sitka spruce, 200 mature trees were sampled from each of eight populations spread across the species’ range (Figure 2.1, Table 2.1) (Gapare et al. 2005). Population samples were collected following a fine-scale sampling strategy. Fresh foliage from trees separated by at least 30 m was collected along 100 m wide east-west transects (Gapare and Aitken 2005). Elevation did not exceed 50m for any sample location (D. Piggott, pers. communication). I used a subset of these samples based on the availability of either foliar tissue or previously extracted stock DNA. When foliage was available, I extracted DNA from frozen needles using a CTAB extraction protocol modified from Doyle and Doyle (1990). For each sample, approximately 500 mg of tissue was freeze-dried in liquid nitrogen and ground, followed by DNA extraction. If tissue samples were unavailable for DNA extraction, I used previously extracted and archived stock DNA from Gapare et al. (2005). Stock DNA for five  22  populations was shipped to McGill University’s Genome Quebec for genotyping using the Sequenom MassARRAY® iPLEX® Gold platform. The remaining three populations (Prince Rupert, BC, Seward, AK, and Kodiak Island, AK) were genotyped using the Illumina platform. Over 100 individuals were genotyped per population (Table 2.1). In all mature populations, we genotyped 54 SNPs used in Holliday et al.'s (2010a) seedling study. 35 SNPs were significantly associated with budset timing, cold hardiness or both phenotypes (henceforth, "putatively adaptive SNPs"). The remaining 19 SNPs are from candidate genes for cold adaptation, but were not significantly associated with phenotypes in Holliday et al.'s (2010a) study (henceforth, "SNPs without phenotypic associations"). Fifteen of the 19 SNPs were selected based on the Gentrain score of the three populations (Prince Rupert, Seward and Kodiak Island) genotyped via Illumina; Illumina uses the Gentrain score to measure the quality and reliability of genotyping based on the distribution of homozygotes and heterozygotes. Fourteen SNPs had Gentrain scores above 0.80, and one SNP scored above 0.70. The remaining 4 SNPs (113-189-S, 27-73-NS, 60-358-NS, and 70-168-NS) were selected for their high FST values in an FST outlier test that Holliday et al. (2010a) ran using “BayesFst” (Beaumont and Balding 2004). For populations genotyped using the Illumina platform, all 54 SNPs were successfully genotyped. For the remaining populations genotyped using the Sequenom platform, 45 of 54 SNPs were successfully genotyped. Thirty-six of the 45 SNPs had call rates exceeding 90%, two SNPs had call rates between 80 – 90%, five SNPs had call rates between 60% and 80%, and two SNPs were below 60%. The two markers with the lowest call rates (166-468-S and 87-256-S) were excluded from all subsequent analyses. The final dataset comprised 27 putatively adaptive SNPs that are predicted to vary clinally, either in terms of allele frequency or frequency of heterozygotes, and 16 SNPs without phenotypic associations.  2.2.3 2.2.3.1  Data analysis Linkage disequilibrium Linkage disequilibrium (LD) was estimated separately in mature populations for  putatively adaptive SNPs and SNPs without phenotypic associations. To assess if pairs of  23  SNPs were in LD, genotypic LD (composite LD in Weir 1996) within and among genes was calculated in Genepop 4.0 (Raymond and Rousset 1995). A global exact test was performed to estimate P-values for the correlation between pairs of loci across all populations. I used Benjamini and Hochberg’s (1995) false discovery rate (FDR) method in R (http://www.rproject.org/) to estimate Q-values that adjust for multiple comparisons. For pairs of SNPs in linkage disequilibrium, the SNP with higher percent of phenotypic variance explained in Holliday et al.’s (2010a) association study was retained for further analysis while the linked SNP was excluded. 2.2.3.2  Genetic diversity and F-statistics To detect differences in heterozygosity between populations of mature trees and  seedlings as well as between different groups of SNPs, I calculated observed heterozygosity (Ho) and unbiased expected heterozygosity (He) for each population in Arlequin ver 3.5 (Excoffier and Lischer 2010). Unbiased He is estimated as:  A t-test was run to test for 1) differences in observed heterozygosity between putatively adaptive SNPs and SNPs without phenotypic associations, and 2) differences between mature trees and seedlings in observed heterozygosity pooled across loci and populations. Pairwise FST (Weir and Cockerham 1984) in mature populations was estimated separately for putatively adaptive SNPs and SNPs without phenotypic associations in Arlequin ver 3.5. Genotypes were permuted 10,098 times among populations to obtain the null distribution of pairwise FST values. The P value for each pairwise estimate is calculated as the proportion of permutations that results in an FST value equal to or larger than the observed statistic (Excoffier and Lischer 2010). A significance level of P ≤ 0.05 was used for each test. 2.2.3.3  Genotype-environment associations for putatively adaptive SNPs Based on the effect of each SNP on bud set timing or cold hardiness, predictions can  be made about whether the SNP is expected to vary clinally across the range of Sitka spruce  24  as well as its direction of clinality (Table 2.2). A reasonable prediction is that the allele frequencies of putatively adaptive SNPs with additive or partially additive effects should vary clinally. The rationale is that the nucleotide associated with increased adaptation to cold should be most frequent in populations with low mean annual temperatures, while the alternate nucleotide should be most frequent at the opposite end of the range, where mean annual temperatures are relatively higher. In contrast, for SNPs where the heterozygote is associated with greater or lesser adaptation to cold than either of the homozygous genotypes, clines in heterozygosity are expected. Given the small number of SNPs with expected clines in heterozygosity (five SNPs in the final dataset), we chose not to analyze clines in heterozygosity. In addition, when it was unclear from Holliday et al.’s (2010a) association test results whether a SNP was additive or partially additive, the SNP was treated as if it were expected to vary clinally. For the remainder of the chapter, my focus is on the SNPs that are expected to show clinal variation in allele frequency. Genotype-environment associations were tested for two alternative nested sets of data. The analysis was originally carried out with the full dataset consisting of eight mature tree populations and 13 seedling populations. In the alternative analysis, the range of mean annual temperatures for populations of seedlings was truncated to span only the mean annual temperature range of mature populations. The difference in temperature ranges stems from my having data for only two mature populations at the north of the range in Alaska while Holliday et al.’s (2010a) study featured six populations of seedlings in Alaska (Table 2.1). The alternative analysis was performed to remove the effects of possible non-linearity in the relationship between temperature and allele frequency at the northern end of the species range, which may have been captured by the more numerous seedling populations but not the few mature populations in Alaska (for details, see section 2.2.3.6). 2.2.3.4  Analysis of covariance (ANCOVA) Analyses of covariance (ANCOVA) were performed separately for each SNP to  determine whether the SNP was significantly associated with provenance mean annual temperature (°C) across populations of both mature trees and seedlings. I included provenance mean annual temperature as a covariate, and a dummy variable for age of a population as a predictor variable (1 for mature populations and 0 for seedlings). SNPs  25  significantly associated with mean annual temperature were then further analyzed for differences in strength of clines between the two life stages (see section 2.2.3.5). My dependent variable was logit-transformed major allele frequency. Population allele frequencies were logit-transformed (ln (p / (1 - p)) prior to analysis of covariance using the logit function in the car package in R (Fox and Weisberg 2011). If the dataset includes proportions 0 and 1, which are undefined, logit automatically adjusts the proportions to the interval (0.025, 0.975) before transforming the data. Climate data for mature and seedling populations were obtained from ClimateWNA (Wang et al. 2006). My first step was to run the full model (with interaction term): Y = Age + Mean annual temperature + Age * Mean annual temperature If the interaction term was not significant at the 0.05 level (without adjusting for multiple comparisons), the model was re-run excluding the interaction term. A multiple comparison adjustment was made using Benjamini and Hochberg’s (1995) false discovery rate (FDR) method. Following the adjustment, SNPs that showed a significant relationship with mean annual temperature (Q ≤ 0.05) were retained for further analysis. 2.2.3.5 Comparing differences in slopes between populations of mature trees and seedlings To test my hypothesis that mature populations have stronger clines than seedling populations, I compared clinal strength between the two life stages for (i) each SNP separately, and (ii) for the set of SNPs as a whole. When considering each SNP separately, the ANCOVA interaction term (Age * Mean annual temperature), following an FDR adjustment, indicated whether there was a difference in slope between mature trees and seedlings. Only SNPs that showed a significant relationship with mean annual temperature after adjusting for multiple comparisons in the analyses of covariance were considered. To analyze the SNPs as a group, I used the Z-transform test (or "Stouffer's method") described in Whitlock (2005). This statistic combines the probabilities of multiple independent tests that have the same hypothesis (Whitlock 2005). In my analysis, the difference in slope between mature trees and seedlings for each SNP constitutes an independent test, and my hypothesis is that mature tree populations have steeper clines than  26  seedling populations (one-sided test). One-tailed P-values for each test are converted to Zscores, Zi. The Z-transform statistic is then calculated as follows: Zs =  , where k is the number of tests performed.  Zs is compared to a standard normal distribution to assess the significance of the hypothesis that mature populations have steeper slopes than seedlings. 2.2.3.6  Assessment of plateaus in allele frequency clines in seedlings As discussed earlier, I ran an alternative analysis that truncated the mean annual  temperature range of seedling populations to span only the range of provenance temperatures experienced by mature tree populations. My full analysis included four additional seedling populations in the Alaska portion of the species’ range, while my alternative analysis excluded these populations. My rationale for including an alternative, truncated analysis was to ensure that differences in slope between mature trees and seedlings were not driven by possible non-linearity in the relationship between allele frequency and temperature. Nonlinearity may be captured by the four additional seedling populations from a portion of Alaska not sampled for mature populations. The potential for a non-linear relationship arises because, in seedling populations, alleles for a number of SNPs are fixed (or close to fixation) north of Prince Rupert, BC. I refer to this fixation (or near fixation) of SNP frequencies at the northern edge of the species range as the plateau in allele frequency clines. This plateau effect in seedling populations flattened the estimated linear clines of many SNPs. Because the mean annual temperature range of mature populations did not cover as wide a range as seedling populations, it is unclear if a similar effect would have been observed in mature populations had we had access to a larger number of Alaskan populations. Including a truncated analysis is an attempt to ensure that comparisons of clines between mature and seedling populations are not biased by this portion of the climatic range in seedlings. The downside of using a truncated range with fewer seedling populations is that valuable data that may shed light on potentially relevant trends are ignored. It is likely that the full dataset (13 seedling populations) that does not truncate the mean annual temperature range of seedling populations is still valid and meaningful for at least a few SNPs.  27  Therefore, to create a bridge between the full and truncated analysis, I assessed each putatively adaptive SNP for a potential plateau in allele frequency clines in the seedling populations from colder provenances. SNPs that plateaued were analyzed with the truncated temperature range that included only eight seedling populations. In contrast, all 13 seedling populations that spanned the full range of mean annual temperatures were used to analyze the SNPs that did not plateau. My criterion for a plateau effect was the stabilization of allele frequencies at or above 0.95 (or conversely, at or below 0.05) for mean annual temperatures below 5 °C. This is the lowest mean annual temperature observed among my eight populations of mature trees (in Seward, AK).  2.3 2.3.1  Results Linkage disequilibrium in mature tree populations Out of 990 pairwise comparisons, 69 pairs of SNPs showed significant LD (P ≤ 0.05).  After adjusting for multiple comparisons, nine pairs were retained. To avoid pseudoreplication from analyzing SNPs in LD for clinal variation, five putatively adaptive SNPs and two SNPs without phenotypic associations were excluded from the analysis. After adjusting for LD, the dataset consisted of 17 putatively adaptive SNPs whose allele frequencies are predicted to vary clinally, and 14 SNPs without phenotypic associations. All subsequent analyses are based on this dataset.  2.3.2  Heterozygosity and FST The t-test for populations of mature trees revealed a significant difference in observed  heterozygosity between putatively adaptive SNPs and SNPs without phenotypic associations, with putatively adaptive SNPs being less heterozygous than the SNPs without phenotypic associations, as expected under directional selection (t = -3.76, P < 0.01). The relationship was reversed in seedling populations; putatively adaptive SNPs were more heterozygous than SNPs without phenotypic associations (t = 3.27, P < 0.01). Observed heterozygosity was also significantly higher in seedling than mature tree populations for the set of putatively adaptive SNPs (t = -7.11, P < 0.01) (Table 2.4). This relationship did not hold for the set of SNPs without phenotypic associations (t = -0.95, P =  28  0.36). FST values were generally higher for putatively adaptive SNPs than for SNPs without phenotypic associations (Table 2.3).  2.3.3 2.3.3.1  Associations between putatively adaptive SNPs and mean annual temperature Single-locus clines Provenance mean annual temperatures for populations of mature trees ranged from  5 °C to 12.1 °C, while populations of seedlings spanned a wider range of temperatures, from 1.4 °C to 12.1 °C (Table 2.1). Five seedling populations from the coldest provenances in Alaska (Icy Bay (IB), Valdez (VD), Montague (MI), Rocky Bay (RB) and Iniskin (IN)) were excluded in the truncated analysis to match the temperature range of seedling populations with that of mature populations. When considering the full range of provenance mean annual temperatures for seedling populations, 11 out of 17 (65%) putatively adaptive SNPs showed significant clinal variation (Q ≤ 0.05) after adjusting for multiple comparisons (Table 2.5, Appendix A). For the SNPs without phenotypic associations, eight out of 14 SNPs (57%) varied clinally (Table 2.6). For the truncated mean annual temperature range of seedling populations, six out of 17 (35%) putatively adaptive SNPs showed significant clinal variation after adjusting for multiple comparisons (Q ≤ 0.05) (Table 2.5, Appendix A), while none of the SNPs without phenotypic associations varied clinally (Table 2.6). For mature tree populations, after adjusting for multiple comparisons, five out of 17 (29%) putatively adaptive SNPs were significantly associated with mean annual temperature (Q ≤ 0.05) (Table 2.5, Appendix A), while none of the 14 SNPs without phenotypic associations varied clinally (Table 2.6). 2.3.3.2  Joint analysis of clines in populations of seedlings and mature trees  Full comparison: includes all seedling populations The analysis of covariance (ANCOVA), which combined seedling and mature populations, revealed that for putatively adaptive SNPs, 13 out of 17 SNPs (76%) were significantly associated with mean annual temperature (P ≤ 0.01) (Table 2.7a). Temperature  29  clines remained significant for all 13 SNPs after adjusting for multiple comparisons (Q ≤ 0.05). Qualitatively, the majority of SNPs had steeper clines for mature trees than seedlings; however, these differences were generally not significant. Populations of mature trees and seedlings showed significantly different slopes for three of the 13 SNPs with significant clines (P ≤ 0.05) (Table 2.8a). Two of the three SNPs (209-523-S and 162-350-S) showed significantly flatter clines in seedling populations, as per our hypothesis. After adjusting for multiple comparisons, differences in slope for all three SNPs were no longer significant (Table 2.8a). However, the Z-transform method, which tested the hypothesis that, overall, seedlings have flatter slopes than mature trees, depicted a one-tailed P-value of 0.027, supporting our hypothesis that the signal of local adaptation is stronger in mature populations than in seedling populations. For SNPs without phenotypic associations, ANCOVA revealed eight of 14 (57%) significant clines after adjusting for multiple comparisons (Q ≤ 0.05) (Table 2.8a). However, clinal strength did not differ significantly between mature and seedling populations (data not shown). Truncated range of seedling populations When the mean annual temperature range of seedling populations was truncated to span the temperature range of mature trees, 11 of the 17 (65%) putatively adaptive SNPs showed significant clinal variation both before and after adjusting for multiple comparisons (P, Q ≤ 0.05) (Table 2.7b). Of the 11 SNPs, only one (266-573-S) exhibited a significant difference in slope between populations of mature trees and seedlings (Table 2.8b). This difference was in the opposite direction of our hypothesis, with seedling populations displaying a steeper cline than mature trees. The difference in slope for SNP 266-573-S was no longer significant after the FDR adjustment. In addition, the Z-transform method did not support our hypothesis of flatter slopes in populations of seedlings compared to mature trees (one-tailed P-value of 0.538). The number of significant clines for SNPs without phenotypic associations dropped to three (21%) when considering the truncated mean annual temperature range (Q ≤ 0.05)  30  (Table 2.8b). As earlier, strength of clines did not vary significantly between mature and seedling populations. 2.3.3.3  Assessment of plateaus in allele frequency clines in seedling populations Among populations of seedlings, of the 17 putative adaptive SNPs with clinal effects  on phenotype (Table 2.2), the frequencies of 10 were found to plateau at fixation (or close to fixation) below a mean annual temperature of 5 °C (Appendix I). This observation suggests that differences in slope between mature and seedling populations for these 10 SNPs may be driven by a flattening of the estimated linear clines in the northern portion of the range when all 13 seedling populations are included in the analysis. SNPs that depict this plateau in frequencies are best analyzed only within the truncated mean annual temperature range, which excludes five northern seedling populations (see section 2.2.3.6 for details on the rationale behind this analysis). In contrast, it seems reasonable to include all 13 seedling populations when analyzing clines for the five SNPs that do not plateau across the full mean annual temperature range of seedlings; these SNPs are 209-523-S, 99-395-NS, 103-455-NS, 19-567-NS, and 73-182-NS. Of these five SNPs, ANCOVA revealed that three (209-523-S, 103-455-NS and 73-182-NS) were significantly associated with mean annual temperature (Q ≤ 0.05) (Table 2.7a). In addition, our hypothesis that mature populations show stronger clines than seedlings was supported when we re-calculated the Z-transform statistic for these three SNPs (one-tailed Pvalue of 0.0007).  2.4  Discussion Within the forest genetics literature, our study is unique in that we searched for clinal  variation among genome-wide cold adaptation candidate genes in independent samples of cohorts of different ages. Rather than casting the net broadly, a number of studies on forest trees have focused on well-studied gene families of known function such as the phytochrome gene family in Populus (Ingvarsson et al. 2006; Ingvarsson et al. 2008; Ma et al. 2010). Phytochromes, which are photoreceptors that aid plants in detecting light, play a major role in synchronizing plant phenology to local environmental conditions (Bae and Choi 2008).  31  In our study, we found clines in both expected and unexpected gene families (Table 2.2). The strongest clines (based on R2) occurred in a histone-like gene (hta-3), a xyloglucan:xyloglucosyl transferase (xth-1), cryptochrome-like genes (cry), an abscisic acid (ABA)-like gene (chlh-2), a CONSTANS-like gene (co), and a peroxidase gene (per) (Tables 1.1 and 2.5) (Holliday et al. 2010a). xth may play a role in altering cell wall flexibility, a trait that is likely to enhance cold tolerance although its current annotations do not include this function. Cellular dehydration caused by osmotic stress from freezing temperatures reduces cell volume as water is removed from the cell (Moore et al. 2006). An elastic cell wall would therefore prevent dehydrated cells from collapsing as tension increases within the cells. As expected, light-signalling genes showed clinal variation, such as the cryptochrome-like and CONSTANS-like genes. Cryptochromes are a class of blue light receptors that regulate circadian rhythm in plants and animals (Holliday et al. 2010) while the CONSTANS gene, which is a member of the photoperiodic pathway, has been found to help regulate flowering time (Griffiths et al. 2003). The peroxidase and ABA-like genes respond to stress; ABA is a plant stress hormone that is directly linked to the initiation of growth cessation and bud dormancy (Pallardy 2008) while peroxidases are related to oxidative stress (Holliday et al. 2010a).  2.4.1 Results differ between the full mean annual temperature range and the truncated mean annual temperature range for seedling populations My results depend on the range of mean annual temperatures used to analyze the seedling populations. There is no clear choice when selecting between the full and truncated temperature ranges for populations of seedlings. On the one hand, solely analyzing the truncated mean annual temperature range makes comparisons between mature and seedling populations simpler as degrees of freedom are identical and possible non-linearity in the northern portion of the range is controlled for. However, by eliminating five seedling populations, valuable data are also lost. To establish a bridge between these alternate analyses, I identified SNPs that did not plateau in allele frequency in the Alaskan portion of the species range and emphasized the full analysis that includes all 13 seedling populations when interpreting these SNPs (for details, see section 2.2.3.6 and 2.3.3.3).  32  An important point that the plateau in allele frequency clines in seedling populations raises is the possibility of temperature thresholds in the northern periphery of the range. It is plausible that after a threshold temperature is attained, optimal phenotypes and thus adaptive allele frequencies do not vary in a linear fashion. Instead, frequencies may remain relatively constant, implying that divergent selection is no longer acting over that portion of the range. For example, after a particular budset date has been reached, the degree of local adaptation may not vary as trees are already well adapted to a wide range of low temperatures. Another explanation is simply that once a locus is fixed, additional selection on phenotypes cannot further shift SNP frequencies at that locus.  2.4.2  Replication of clines in independent populations of mature trees In this section I focus on the17 putatively adaptive SNPs whose allele frequencies are  expected to vary in a clinal fashion. When considering the full mean annual temperature range for seedling populations, at least five significant clines observed in seedling populations were replicated in mature trees (Table 2.5). However, differences in degrees of freedom arising from analyzing a larger number of seedling populations (13) than mature populations (eight) make comparisons of P-values complicated. Instead, R2 values are helpful in assessing the extent of replication of clines. SNPs with strong clines (defined as R2 values above 0.40) at both life stages are 209-523-S, 162-350-S, 257-188-NS, 260-504-S, 257-264-S, and 73-182-S. Overall, mature and seedling populations had a similar number of strong clines (10 for mature trees and eight for seedlings). With the truncated mean annual temperature range that excludes five seedling populations, degrees of freedom are identical between populations of mature trees and seedlings. Three SNPs (209-523-S, 162-350-S and 260-504-S) showed comparable, significant clines between mature and seedling populations (Table 2.5). Again, mature and seedling populations possessed a similar number of significant clines (five versus six respectively). Overall, there appears to be some matching of clines corresponding to mean annual temperature between populations of seedlings sampled as seed and independent, natural populations of mature trees. Such replication strengthens the likelihood that these SNPs are  33  locally adaptive and important in cold adaptation, as observing similar patterns of variation in independent datasets are unlikely to be the outcome of chance alone. As our understanding of the genomics of local adaptation in tree species increases, studies focusing on the distribution of putatively adaptive molecular variation along steep environmental gradients are also being emphasized. Such studies strengthen the adaptive significance of the markers under study. The recent literature provides numerous examples of single-locus putatively adaptive clines in plant and tree species such as Arabidopsis (Samis et al. 2008), Zea mays (Lia et al. 2007) and Populus tremula (Ingvarsson et al. 2006; De Carvalho et al. 2010; Ma et al. 2010). However, explanations for clines are often far from straightforward, which makes generalizations difficult. For example, in Arabidopsis, longitudinal clines were observed in haplotypes of the phytochrome gene, PHYC (Samis et al. 2008). Counter-intuitively, the longitudinal distribution of PHYC haplotypes was in the opposite direction predicted from the phenotypic clines. The authors cite both sampling and biological factors such as pleiotropic effects and the role of genes other than PHYC in photoperiod sensitivity that may have contributed to their contrary results. Ma et al. (2010) provide examples of single-locus clines from the P. tremula model system. Again, genes involved in photoperiod sensitivity were assessed for clinal variation and genetic differentiation beyond the amount expected for neutral markers. Ma et al. discovered clinal variation at four of 113 SNPs, which may have arisen from admixture between different lineages in Sweden and not solely from selection; admixture among lineages was found to promote a strong step cline in a photoperiod-sensitive trait, timing of budset (De Carvalho et al. 2010). However, Ma et al. (2010) consider it unlikely that admixture was the primary driver of the molecular clines as the SNPs in their study did not show significant isolation by distance (IBD). A number of the studies on single-locus clines also highlight the importance of disentangling underlying population structure from the effects of divergent selection to ensure that spurious patterns of differentiation are not interpreted as the product of natural selection. Simulations of clinal variation under neutral population genetic models have demonstrated the potential for putatively adaptive single-locus clines to be the outcome of  34  gene flow and drift (Vasemagi 2006). Such simulations highlight the difficulty of separating neutral forces from selection in studies on clinal variation. Despite the complications introduced by neutral population structure, clines for loci significantly associated with strongly adaptive traits such as timing of budset are unlikely to result solely from the interplay between gene flow and drift. In our study, we were unable to adjust for population structure as we lacked an appropriate set of neutral SNPs for comparative purposes. Nevertheless, the parallel clines that we observed between populations of mature trees and seedlings suggest that natural selection is shaping the geographic distribution of at least a few markers. It is also important to note that the absence of control SNPs does not affect our comparison of strength of clinal variation between mature trees and seedlings (section 2.4.3).  2.4.3 Mixed results for comparisons of clines between populations of mature trees and seedlings Differences in the strength of clines varied widely between the full analysis that included all 13 seedling populations and the analysis that truncated the mean annual temperature range of seedlings to include eight populations. In the full analysis, the Ztransform statistic supported our hypothesis that seedlings have flatter slopes than mature trees. This result held true even when we retained only five SNPs whose allele frequencies in seedling populations did not plateau in the northern portion of the range (see section 2.3.3.3). Again, when restricting our analysis to SNPs whose allele frequencies did not plateau in the northern portion of the range, slopes for mature trees and seedlings significantly differed at only one SNP (209-523-S) (Table 2.8a). While this difference did not remain significant following the FDR adjustment, it is still possible that variation for this SNP is structured by directional selection that we were unable to detect given the limited power of our analysis (discussed below). The direction of the difference in clinal strength also follows our expected pattern, with seedling populations showing a weaker cline than mature trees. With the truncated mean annual temperature range, only SNP 266-573-S differed significantly between populations of mature trees and seedlings prior to adjusting for multiple comparisons (Table 2.8b). The difference was in the opposite direction of our  35  hypothesis, with seedlings depicting a steeper cline than mature trees (Appendix A). Like SNP 209-523-S, this SNP may be involved in local adaptation, but there does not appear to be a biological reason for observing a stronger cline among populations of seedlings than among mature populations. A statistical difficulty that we encountered in our analysis was limited power to detect clines, particularly differences in the strength of clines between mature and seedling populations. Low power stems from the limited number of populations in our analysis, particularly for mature trees (eight mature populations versus 13 seedling populations). Including a larger number of mature populations would likely have strengthened our results and helped us reach more concrete conclusions on the adaptive significance of the SNPs. In particular, a larger number of mature populations at the northern end of the range would have permitted us to detect whether the fixation of allele frequencies observed in seedlings are also replicated in mature populations. These additional data would help us better detect the presence of potential temperature thresholds below a specific minimum temperature. Low power also restricted our ability to implement alternate regression techniques such as nonlinear regressions, which may have lent additional insights into patterns observed in the north of the range.  2.4.4  SNPs without phenotypic associations In this study, the SNPs without phenotypic associations were not statistically  correlated with timing of budset or degree of cold hardiness. However, this does not imply that the SNPs are not associated with other ecologically important traits, including those involved in adaptation to temperature. An encouraging result is that a considerably greater proportion of the putatively adaptive SNPs compared to the SNPs without phenotypic associations varied clinally (Tables 2.7 and 2.9), which increases the likelihood of the former set of clines being adaptive. The clines we observed for the SNPs without phenotypic associations may result from IBD. Given the linear range of Sitka spruce along the Pacific coast coupled with prior research that highlighted a strong IBD pattern in microsatellites (Mimura and Aitken 2007a), we would expect a proportion of all SNPs to show clinal variation due to IBD or some combination of  36  IBD and natural selection. This logic also holds true for our set of putatively adaptive SNPs as we did not control for neutral population structure. However, the large number of clines that we detected among the putatively adaptive SNPs, coupled with the replication of clines between mature trees and seedlings make it unlikely that a majority of these clines arose from neutral evolutionary factors alone.  2.4.5 Geographic distribution of putatively adaptive standing variation in mature populations We found that peripheral populations had the least standing variation across the 17 putatively adaptive SNPs analyzed in this study (Figure 2.1). Kodiak Island, AK was fixed for seven of the 17 SNPs (41%) and close to fixation (minor allele frequency of less than 1%) for an additional three SNPs (18%). Fort Bragg, CA and Seward, AK, were both fixed for three SNPs (18%), and close to fixation for one and two SNPs respectively. For Kodiak Island and Seward, all except one of the fixed alleles are associated with earlier budset or greater cold hardiness. At the southern margin in Fort Bragg, three of the four alleles that are fixed or close to fixation are associated with earlier budset or greater cold hardiness. As climate warming progresses, rear-edge populations, likely at the limit of their heat and drought tolerance face potential habitat loss and extirpation. Leading-edge populations, which generally inhabit colder climates than optimal for their genotypes and phenotypes (Garcia-Ramos and Kirkpatrick 1997), are expected to initially benefit from warming, and to possess greater potential to adapt to climate change in the longer term because of gene flow from the center of the range (Davis and Shaw 2001). This implies that the southern, isolated population of Fort Bragg, CA may be at greater risk of extirpation than the northern, isolated population of Kodiak Island, AK. My results suggest that Kodiak Island has less adaptive diversity than other populations of Sitka spruce including Fort Bragg as it is fixed (or close to fixation) for a relatively high proportion of “cold” alleles. This suggests that in the future, Kodiak Island may also be prone to maladaptation when compared to populations in the continuous portion of the species range, which harbour higher levels of standing variation. However, the Kodiak genotypes may be capable of colonizing colder, unoccupied habitat to the north as climates warm. Analyzing a larger set of adaptive SNPs, and testing their 37  associations with phenotypes under extreme conditions, will help establish the adaptive capacity of peripheral populations and provide insights into the extent of possible maladaptation under different climate change scenarios.  2.5  Conclusion My results strengthen the importance of a subset of putatively adaptive SNPs in local  adaptation to cold. Parallel clines observed in independent populations of mature trees and seedlings increase the probability that these SNPs are selective. However, low power coupled with fewer mature populations in the Alaskan portion of the species range limited my ability to detect differences in strength of clines between mature and seedling populations. Nevertheless, analyzing the markers as a group provided support for my prediction that mature populations have steeper clines than seedlings. My analyses also revealed a higher proportion of fixed SNPs in the peripheral, disjunct population of Kodiak Island, AK, which lies at the leading edge of the species range. Less adaptive diversity suggests that this population may be susceptible to maladaptation in the future compared to populations in the continuous portion of the species range.  38  Table 2.1. Geographic locations and mean annual temperature (MAT, °C ) of study populations. Sample sizes refer to the number of individuals genotyped per population. Marker-specific sample sizes vary somewhat based on genotyping success rates.  Population  State / Province  Sample size Latitude (°N)  Longitude (°W)  MAT (°C)  Mature tree populations Fort Bragg (F) Brookings (B) Qualicum (L) Port McNeill (P) Prince Rupert (R) Queen Charlotte Islands (Q) Seward (S) Kodiak Island (K)  CA OR BC BC BC BC AK AK  164 174 122 129 134 141 150 168  39°26'37.69" 42°05'20.44" 49°28'43.51" 50°32'17.44" 54°12' 53°60' 60°08'57.47" 57°49'33.69"  123°48'21.78" 124°20'01.29" 124°48'13.28" 127°34'10.72" 130°9' 132°10.5' 149°25'20.73" 152°21'19.40"  11.9 12.1 8.9 8.6 7.9 7.6 5.0 6.1  CA OR BC BC BC BC BC AK AK AK AK AK AK  38 13 39 46 48 37 33 15 38 12 35 15 38  42°00' 47° 49°03'42.54'' 49°30'09.97'' 52°30'10.56'' 53°00'03.31'' 53°29'58.40'' 60° 3'15.24" 61°08'19.91" 61° 59°12'18.32" 59°54'36.38" 57°00'07.20"  124°12'43.09" 124° 123°00'44.91'' 124°51'03.08" 128°14'15.16'' 131°56'50.92'' 130° 141°17'23.02" 146°20'58.36" 147° 151°14'30.57" 152°46'43.02" 153°18'26.35"  12.1 10.8 9.7 8.8 8.1 7.9 8.0 3.3 1.4 3.6 3.8 1.7 5.9  Seedling populations Redwood (RW) Columbia River (CR) Vancouver (VA) Vancouver Island (VI) Ocean Falls (OF) Queen Charlotte Islands (QC) Prince Rupert (PR) Icy Bay (IB) Valdez (VL) Montague (MI) Rocky Bay (RB) Iniskin (IN) Kodiak Island (KD)  39  Table 2.2. Summary of significant genotype-phenotype associations (data obtained from Holliday et al. 2010a). A1 and A2 refer to the two alternate nucleotide homozygotes, while A1A2 refers to the heterozygote. Cold injury is based on results from artificial freeze tests. Budset (Day of the year)  Mean cold injury (percent)  SNP  Gene  Phenotypic association  A1  A1A2  A2  A1  A1A2  A2  If Clinal, "Y"  103-455-NS  erf1  Both  201.6  206.5  219.9  14.0  18.5  26.7  Y  162-350-S  xth1  Both  257.4  215.7  200.7  45.3  27.5  13.4  Y  19-567-S  aux1  Both  217.8  205.9  200.2  22.1  17.1  17.1  Y  209-585-S  hta3  Both  211.4  223.6  204.4  20.5  27.8  16.9  N  44.1  33.1  16.7  209-523-S  Bud set  191.5  197.8  215.6  266-573-S  per6  Both  282.5  228.8  203.4  195-441-S  phya  Bud set  189.0  222.0  204.2  N  207-363-S  pal3  Bud set  196.0  207.3  207.1  ?  20-374-NS  gi  Bud set  207.6  201.2  241.5  N  257-188-NS  cry2  Bud set  206.3  213.8  284.7  Y  257-264-S 54-570-NS 56-206-S  Cold injury  Y  17.3  39.1  52.3  Y  Y  per3  Bud set  227.5  199.3  209.0  N  co  Bud set  247.8  221.3  204.6  Y  89-475-NS  aec3  Bud set  197.8  208.0  212.3  Y  99-395-NS  gst4  Bud set  225.7  205.4  --  ?  62-424-S  lrr3  Bud set  259.0  233.8  205.1  Y  73-182-NS  skip2  Bud set  204.5  214.6  259.0  Y  87-319-S  pip5k  Bud set  206.2  232.4  259.0  101-828-S  cbl3  Cold injury  18.0  32.9  14-438-S  drm2  Cold injury  41.1  18.2  --  ?  258-207-S  cry3  Cold injury  42.3  31.0  16.8  Y  260-504-S  chlh2  Cold injury  32.0  34.0  16.3  N  Y 10.3  N  Note: SNPs with "?" included in clinal analysis  40  Table 2.3. (a) Mature population pairwise FST estimates for 17 putatively adaptive SNPs (SNPs with phenotypic associations); (b) Mature population pairwise FST estimates for 14 SNPs without phenotypic associations.  (a) Putatively adaptive SNPs  Fort Bragg Brookings Qualicum Port McNeill Prince Rupert QCI Seward Kodiak island  Fort Bragg  Brookings  Qualicum  Port McNeill  Prince Rupert  QCI  Seward  Kodiak island  -0.079 0.229 0.187 0.169 0.157 0.242 0.289  -0.145 0.058 0.056 0.048 0.137 0.170  -0.133 0.133 0.124 0.149 0.231  -0.008 0.006 0.086 0.101  -0.004 0.089 0.104  -0.075 0.088  -0.084  --  All estimates are significant at the 5% level based on 10,098 permutations.  (b) SNPs without phenotypic associations  Fort Bragg Brookings Qualicum Port McNeill Prince Rupert QCI Seward Kodiak Island  Fort Bragg x  Brookings  Qualicum  Port McNeill  Prince Rupert  QCI  Seward  Kodiak Island  -0.026 0.017 0.038 0.032 0.034 0.085 0.064  -0.029 0.040 0.041 0.057 0.136 0.088  -0.008 0.005 0.008 0.080 0.108  -0.002 0.026 0.099 0.138  -0.011 0.075 0.128  -0.064 0.126  -0.116  --  All estimates, except the Prince Rupert - QCI pair are significant at the 5% level based on 10,098 permutations.  41  Table 2.4. Observed and expected heterozygosity for 17 putatively adaptive SNPs and 14 SNPs without phenotypic associations.  MATURE TREE POPULATIONS Observed Heterozygosity Expected Heterozygosity Putatively adaptive SNPs  SNPs w/o associations  Putatively adaptive SNPs  SNPs w/o associations  Fort Bragg, CA Brookings, CA Qualicum, BC Port McNeill, BC Prince Rupert, BC QCI, BC Seward, AK Kodiak Island, AK  0.205 0.231 0.148 0.174 0.149 0.162 0.157 0.085  0.210 0.202 0.219 0.227 0.252 0.233 0.249 0.216  0.219 0.228 0.153 0.176 0.154 0.155 0.148 0.111  0.211 0.196 0.222 0.237 0.244 0.233 0.277 0.218  Average  0.164  0.226  0.168  0.230  Population  SEEDLING POPULATIONS Observed Heterozygosity Expected Heterozygosity Putatively adaptive SNPs  SNPs w/o associations  Putatively adaptive SNPs  SNPs w/o associations  Redwood, CA Columbia River, OR Vancouver, BC Vancouver Island, BC Ocean Falls, BC QCI, BC Prince Rupert, BC Icy Bay, AK Valdez, AK Montague, AK Rocky Bay, AK Iniskin, AK Kodiak Island, AK  0.348 0.398 0.301 0.310 0.305 0.284 0.280 0.263 0.272 0.265 0.272 0.289 0.270  0.199 0.165 0.233 0.186 0.223 0.224 0.299 0.252 0.265 0.202 0.294 0.338 0.242  0.319 0.293 0.261 0.236 0.229 0.223 0.203 0.201 0.190 0.177 0.196 0.204 0.178  0.215 0.208 0.251 0.238 0.253 0.239 0.289 0.256 0.265 0.296 0.308 0.329 0.258  Average  0.297  0.240  0.224  0.262  Population  42  Table 2.5. Single-locus clines for the 17 putatively adaptive SNPs whose allele frequencies are expected to vary clinally across the range of Sitka spruce. The "full range" includes 13 seedling populations. The "truncated range" includes eight seedling populations; five populations were excluded in order to match the mean annual temperature ranges of mature tree and seedling populations.  Mature tree populations SNP 209-523-S 162-350-S  Phenotypic association  R2  Q value1  Bud set  0.953  0.000  Seedling populations Full range Truncated range R2  Q value1  ***  0.806  0.000  R2  Q value1  ***  0.657  0.036  ** **  Both  0.831  0.009  ***  0.705  0.001  ***  0.669  0.036  Bud set  0.731  0.024  **  0.485  0.012  **  0.340  0.107  260-504-S  Cold injury  0.638  0.039  **  0.563  0.008  ***  0.618  0.036  56-206-S  Bud set  0.629  0.039  **  0.341  0.040  **  0.486  0.070  257-264-S  Cold injury  0.547  0.062  0.485  0.012  **  0.340  0.107  73-182-NS  Bud set  0.445  0.080  0.429  0.019  **  258-207-S  Cold injury  0.516  0.066  0.228  0.080  266-573-S  Both  0.473  0.076  0.202  0.091  103-455-NS  Both  0.409  0.088  0.305  0.045  99-395-NS  Bud set  0.262  0.172  -0.076  89-475-NS  Bud set  0.211  0.199  --  14-438-S  Cold injury  0.066  0.339  --  --  87-319-S  Bud set  0.056  0.339  0.320  0.044  **  207-363-S  Bud set  --  0.850  0.538  0.009  ***  19-567-S  Both  --  0.948  --  --  62-424-S  Bud set  --  0.948  0.777  0.000  257-188-NS  **  0.283  0.122  0.790  0.017  **  0.870  0.008  ***  0.061  0.107  --  0.487  0.070  --  -0.164  0.907  0.355  0.107  0.118  0.242  **  ***  0.310  0.115  -0.130  0.714  0.631  0.036  **  Note: 1  The Q value measures the false discovery rate using Benjamini and Hochberg's (1995) method.  Significance levels: "***" 0.01, and "**": 0.05.  43  Table 2.6. Single-locus clines for the 14 SNPs without phenotypic associations. The "full range" includes 13 seedling populations. The "truncated range" includes eight seedling populations; five populations were excluded in order to match the temperature ranges of mature tree and seedling populations.  Mature tree populations SNP 100-316-NS  2  R  Q value  1  Seedling populations Full range Truncated range 2  R  Q value  1  2  R  Q value  --  0.790  --  0.790  --  0.831  0.013  0.387  164-465-S  0.005  0.609  0.403  0.024  214-655-S  0.039  0.609  --  0.841  0.416  0.221  257-366-S  0.596  0.212  0.667  0.002  ***  0.215  0.376  265-363-S  0.054  0.609  0.535  0.008  ***  0.232  0.256  --  0.831  --  0.574  --  0.980  68-545-S  0.310  0.410  0.751  0.001  ***  0.713  0.072  69-330-S  0.080  0.609  0.383  0.025  **  0.379  0.221  112-180-S  --  0.935  --  0.506  --  0.980  113-189-S  0.607  0.772  0.672  0.002  0.349  0.221  13-496-NS  --  0.690  0.057  0.301  --  0.980  161-576-S  --  0.612  0.188  0.121  --  0.790  60-358-NS  0.073  0.609  0.618  0.003  ***  0.295  0.221  70-168-NS  0.431  0.322  0.502  0.009  ***  0.306  0.221  288-628-NS  **  ***  1  Notes: 1. The Q value measures the false discovery rate using Benjamini and Hochberg's (1995) method. Significance levels: "***" 0.01, and "**": 0.05.  44  Table 2.7. ANCOVA of mature and seedling populations for the 17 putatively adaptive SNPs whose allele frequencies are expected to vary clinally. (a) All 13 seedling populations are included, and (b) Only eight seedling populations are included as the mean annual temperature (MAT) range of seedling populations is truncated to match the MAT range of mature tree populations.  (a) All seedling populations P - value  Slope coefficient SNP  1  14-438-S 207-363-S 209-523-S 257-188-NS 56-206-S 89-475-NS 103-455-NS 162-350-S 19-567-S 257-264-S 258-207-S 266-573-S 62-424-S 73-182-NS 87-319-S 260-504-S 99-395-NS  Age  MAT  Age  2.000 0.200 -1.741 -0.294 -0.092 3.898 0.324 -2.775 0.262 -0.201 0.022 0.716 -1.308 0.156 0.434 0.269 0.366  -0.029 -0.165 0.317 -0.294 -0.163 0.006 0.136 0.341 -0.013 -0.180 -0.129 -0.200 -0.192 -0.127 -0.080 -0.219 0.005  0.000 0.530 0.021 0.094 0.744 0.008 0.246 0.023 0.449 0.493 0.938 0.035 0.112 0.560 0.029 0.350 0.545  Q -value MAT  ***  0.387 0.003 0.000 0.000 0.001 0.834 0.004 0.000 0.803 0.001 0.008 0.001 0.000 0.005 0.010 0.000 0.871  **  *** **  **  **  Age  MAT  0.000 *** 0.634 0.097 0.229 0.790 0.069 0.464 0.097 0.634 0.634 0.938 0.101 0.238 0.634 0.099 0.595 0.634  *** *** *** *** *** *** *** *** *** *** *** ** ***  0.470 0.005 0.000 0.001 0.003 0.871 0.007 0.000 0.871 0.002 0.011 0.002 0.001 0.008 0.014 0.000 0.871  *** *** *** *** *** *** *** ** *** *** *** ** ***  Note: 1  Data for SNPs 209_523_S, 89_475_NS, 162_350_S, 62_424_S, and 99_395_NS are for the full model as these SNPs had significant interaction term P -values.  (b) Truncated range of seedling populations P - value  Slope coefficient SNP1 14_438_S 207_363_S 209_523_S 257_188_NS 56_206_S 89_475_NS 103_455_NS 162_350_S 19_567_S 257_264_S 258_207_S 266_573_S 62_424_S 73_182_NS 87_319_S 260_504_S 99_395_NS 1  Age  MAT  Age  1.916 0.289 0.096 -0.417 -0.209 2.580 0.439 0.332 0.238 -0.422 -0.212 -2.440 0.402 0.046 0.458 0.036 1.946  -0.075 -0.127 0.474 -0.194 -0.242 -0.089 0.213 0.642 -0.018 -0.311 -0.275 -0.558 -0.101 -0.206 -0.072 -0.345 0.138  0.000 0.431 0.667 0.034 0.428 0.000 0.111 0.386 0.574 0.188 0.386 0.050 0.252 0.850 0.065 0.898 0.000  Q -value MAT  ***  ** ***  ***  0.240 0.159 0.000 0.000 0.002 0.107 0.004 0.000 0.857 0.001 0.000 0.000 0.228 0.003 0.209 0.000 0.026  Age  *** *** *** *** *** *** *** *** *** *** **  0.000 *** 0.564 0.756 0.143 0.564 0.000 *** 0.270 0.564 0.697 0.399 0.564 0.170 0.475 0.898 0.185 0.898 0.000 ***  MAT 0.255 0.207 0.000 0.001 0.003 0.151 0.007 0.000 0.857 0.002 0.001 0.001 0.255 0.005 0.254 0.001 0.041  *** *** *** *** *** *** *** *** *** *** **  Data for SNP 266_573_S is for the full model, including the interaction term, which is significant.  45  Table 2.8. Summary of slope differences between mature and seedling populations for putatively adaptive SNPs that showed significant associations with mean annual temperature in the ANCOVA. (a) All 13 seedling populations are included, and (b) Only eight seedling populations are included as the mean annual temperature (MAT) range of seedling populations is truncated to match the MAT range of mature tree populations.  (a) All seedling populations SNP  P- value  Q- value  Interpretation  209-523-S  0.019 **  0.125  Seedling slope flatter  162-350-S  0.019 **  0.125  Seedling slope flatter  62-424-S  0.043 **  0.170  Seedling slope steeper  257-264-S  0.052  0.170  Seedling slope flatter  257-188-NS  0.080  0.209  Seedling slope flatter  207-363-S  0.131  0.263  Seedling slope steeper  260-504-S  0.142  0.263  Seedling slope flatter  73-182-NS  0.189  0.296  Seedling slope flatter  87-319-S  0.214  0.296  Seedling slope steeper  103-455-NS  0.231  0.296  Seedling slope flatter  258-207-S  0.250  0.296  Seedling slope flatter  56-206-S  0.758  0.783  Seedling slope flatter  266-573-S  0.783  0.783  Seedling slope flatter  (b) Truncated range of seedling populations SNP  P- value  Q- value  Interpretation  266-573-S  0.020 **  0.224  Seedling slope steeper  209-523-S  0.200  0.716  Seedling slope flatter  258-207-S  0.233  0.716  Seedling slope steeper  56-206-S  0.261  0.716  Seedling slope steeper  99-395-NS  0.337  0.741  Seedling slope flatter  162-350-S  0.612  0.826  Seedling slope flatter  73-182-NS  0.672  0.826  Seedling slope flatter  257-264-S  0.690  0.826  Seedling slope flatter  257-188-NS  0.726  0.826  Seedling slope flatter  103-455-NS  0.751  0.826  Seedling slope flatter  260-504-S  0.868  0.868  Seedling slope steeper  46  Table 2.9. ANCOVA of mature populations and seedlings for the 14 SNPs without phenotypic associations. (a) All 13 seedling populations are included, and (b) Only eight seedling populations are included as the mean annual temperature (MAT) range of seedling populations is truncated to match the MAT range of mature tree populations.  (a) All seedling populations SNP 100-316-NS 164-465-S 214-655-S 257-366-S 265-363-S 288-628-NS 68-545-S 69-330-S 112-180-S 113-189-S 13-496-NS 161-576-S 60-358-NS 70-168-NS  Slope coefficient Age MAT 0.428 0.284 -0.256 0.098 0.044 -0.179 -0.169 -0.168 -0.126 -0.379 -0.039 0.318 0.108 0.057  -0.058 0.104 0.043 0.216 0.153 0.031 0.212 0.116 -0.040 0.250 0.073 -0.100 -0.232 0.206  P - value  Q -value  Age  MAT  0.193 0.230 0.499 0.680 0.866 0.668 0.457 0.471 0.686 0.331 0.899 0.279 0.755 0.852  0.244 0.008 0.456 0.000 0.001 0.626 0.000 0.004 0.404 0.000 0.129 0.035 0.000 0.000  *** *** *** *** *** *** ** *** ***  Age  MAT  0.899 0.899 0.899 0.899 0.899 0.899 0.899 0.899 0.899 0.899 0.899 0.899 0.899 0.899  0.311 0.014 0.491 0.000 0.002 0.626 0.000 0.008 0.471 0.000 0.181 0.054 0.000 0.000  ** *** *** *** *** ***  *** ***  (b) Truncated range of seedling populations SNP 100-316-NS 164-465-S 214-655-S 257-366-S 265-363-S 288-628-NS 68-545-S 69-330-S 112-180-S 113-189-S 13-496-NS 161-576-S 60-358-NS 70-168-NS  Slope coefficient Age MAT 0.482 0.194 0.034 0.085 -0.005 -0.186 -0.270 -0.145 -0.079 -0.637 -0.070 0.374 0.229 -0.086  -0.038 0.042 0.210 0.220 0.138 -0.009 0.179 0.140 -0.005 0.145 0.026 -0.064 -0.196 0.171  P - value  Q -value  Age  MAT  Age  0.063 0.448 0.930 0.750 0.986 0.655 0.197 0.567 0.792 0.109 0.763 0.206 0.547 0.723  0.513 0.488 0.038 0.004 0.060 0.927 0.002 0.033 0.943 0.125 0.634 0.352 0.044 0.010  0.721 0.924 0.986 0.924 0.986 0.924 0.721 0.924 0.924 0.721 0.924 0.721 0.924 0.924  ** ***  *** **  ** **  MAT 0.653 0.653 0.103 0.028 ** 0.120 0.943 0.028 ** 0.103 0.943 0.219 0.740 0.548 0.103 0.047 **  47  3 7  2 0 2 1  0  3  Figure 2.1. Range and locations of sampled populations of Sitka spruce. “CC” refers to central, continuous, “CD” to central, disjunct, “PC” to peripheral, continuous and “PD to peripheral, disjunct (Gapare et al., 2005). Numbers indicate the number of putatively adaptive SNPs with clinal effects on allele frequency that are fixed in each population.  48  3 Chapter: Conclusions and future directions 3.1  Overall conclusions My results strengthen the evidence for the involvement in local adaptation of a set of  putatively adaptive SNPs, and add to a growing field of research focused on detecting signatures of selection on a genome-wide basis. In this thesis, I looked for evidence of directional selection along a temperature gradient in mature, natural populations of Sitka spruce, with the dual goals of assessing if populations of mature trees and seedlings show parallel clines, and whether the steepness of clines vary between these two life stages. Seedlings and mature trees from independent samples of populations across the species range demonstrated clines for approximately the same markers, which increases the likelihood that these markers are selective. In addition, comparing clines for putatively adaptive SNPs against clines for a set of SNPs not associated with phenotype but present in candidate genes clearly highlighted a larger proportion of clines for the set of putatively adaptive markers. My results were mixed when comparing differences in the strength of clines, which made drawing a clear conclusion difficult. At a minimum, when I analyze the SNPs as a group, there appears to be evidence that populations of mature trees show stronger clines than populations of seedlings; however, this analysis needs to be extended to include a larger number of populations before more concrete conclusions can be reached. Another drawback that my study faced is the absence of a set of control markers that are not from candidate genes for local adaptation. Despite this drawback, it is important to note that an absence of control SNPs does not affect comparisons of clines between populations of mature trees and seedlings. In addition to assessing clinal variation, we observed that the peripheral populations of Kodiak Island, AK and Seward, AK in the north, and Fort Bragg, CA in the south harbour less putatively adaptive genetic variation than populations at the centre of the range. Kodiak Island was fixed for seven of 17 putatively adaptive SNPs (41%) while Seward and Fort Bragg were each fixed for three markers (18%). While rear-edge populations such as Fort Bragg are expected to face a greater risk of extirpation than leading-edge populations in the north of a species range, reduced standing variation in Kodiak Island suggests that this recently founded population may also lack adaptive diversity to respond to future climate  49  change. Nevertheless, peripheral populations may possess additional standing variation for adaptive markers that were not captured in Holliday et al.’s (2010a) study, particularly since a threshold minor allele frequency of 5% across the species range was used for SNP selection. Furthermore, as different combinations of markers can result in the same phenotype, it is possible that the peripheral populations still harbour sufficient adaptive variation for future climatic variation. To comprehensively capture the adaptive potential of the peripheral populations, a broader set of adaptive markers needs to be analyzed.  3.2  Applications Our results are an initial step towards matching the genotypes of trees to future,  predicted climates. With climate change, populations are likely to become increasingly mismatched to local conditions. As climate change progresses, selecting the optimal seed source for reforestation efforts will involve transferring seed between more distant locations that are expected to be optimally adapted to future predicted climate instead of choosing more local seed sources. Thus, knowledge of an individual’s genotype at markers associated with climate will assist in selecting the appropriate seed source for maintaining the productivity and health of our forests. Establishing a set of markers that are expected to be locally adaptive is an essential step towards determining the adaptive capacity of individual seed sources as well as populations.  3.3  Future directions Future studies can expand on our research by analyzing a larger number of  populations, particularly at the edges of the range. Instead of sampling specific populations, a more intensive sampling scheme can be adopted where individual trees are sampled in a diffuse approach across multiple environmental gradients. Numerous statistical techniques have been developed to cluster individuals into populations based on their genetic similarity, instead of using geographic proximity as the primary mechanism for delineating populations. By shedding light on potentially complex population structure in Sitka spruce, this approach would help facilitate a more complete understanding of gene flow and genetic differentiation in the species. A comprehensive understanding of the dynamics of neutral evolution is also  50  valuable when assessing the strength of selection at adaptive loci as population structure can be explicitly taken into account. Such an analysis also requires the development of a large set of neutral markers, preferably SNPs from non-coding regions or those randomly selected from across the genome rather than from candidate genes. Currently, such a dataset does not exist for Sitka spruce. Another avenue of future research is to develop a more extensive list of putatively adaptive SNPs so that the adaptive potential of Sitka spruce can be characterized more completely. Next generation sequencing has facilitated the rapid sequencing of thousands of genes for non-model organisms (Hudson 2008) and has enabled the adoption of genomewide approaches to characterizing the genetics of local adaptation by bringing together genomic and ecological data (Stapley et al. 2010). In addition, large amounts of sequence data are providing researchers with the tools to determine the history of adaptive loci. These insights help reveal the relative importance of new mutations and standing genetic variation in adaptive responses to environmental stimuli (Stapley et al. 2010). Finally, the validity of putatively adaptive genotypes can be further tested in independent association studies that cover a wide range of ecologically important phenotypic traits such as frost hardiness, drought hardiness, growth rates, and insect and disease resistance. Once a strong link is established between phenotype and genotype, management decisions such as selecting an appropriate seed source can be based on predictions of phenotypic values derived from an individual’s underlying genotype.  51  References Aitken, S. N., S. Yeaman, J. A. Holliday, T. Wang, and S. Curtis-McLane. 2008. Adaptation, migration or extirpation: climate outcomes for tree populations. Evolutionary Applications 1: 95-111. Akey, J., G. Zhang, K. Zhang, L. Jin, and M. Shriver. 2002. Interrogating a high-density SNP map for signatures of natural selection. Genome Research 12: 1805-1814. Allendorf, F. W., P. A. Hohenlohe, and G. Luikart. 2010. Genomics and the future of conservation genetics. Nature Reviews Genetics 11: 697-709. Anderson, L. L., F. S. Hu, D. M. Nelson, R. J. Petit, and K. N. Paige. 2006. Ice-age endurance: DNA evidence of a white spruce refugium in Alaska. Proceedings of the National Academy of Sciences of the United States of America 103: 12447-12450. Andolfatto, P. 2001. Adaptive hitchhiking effects on genome variability. Current Opinion in Genetics and Development 11: 635-641. Atkins, K. E., and J. M. J. Travis. 2010. Local adaptation and the evolution of species’ ranges under climate change. Journal of Theoretical Biology 266: 449-457. Bae, G., and G. Choi. 2008. Decoding of light signals by plant phytochromes and their interacting proteins. Annual Review of Plant Biology 59: 281-311. Balloux, F., L-J. L. Handley, T. Jombart, H. Liu and A. Manica. 2010. Climate shaped the worldwide distribution of human mitochondrial DNA sequence variation. Proceedings of the Royal Society B 276: 3447-3455. Beaumont, M. A., and D. J. Balding. 2004. Identifying adaptive genetic divergence among populations from genome scans. Molecular Ecology 13: 969-980. Benjamini, Y., and Y. Hochberg. 1995. Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, Series B 57: 289-300. Brown, J. H. 1984. On the relationship between abundance and distribution of species. The American Naturalist 124: 255-279. Burley, J. 1966. Genetic variation in seedling development of Sitka spruce, Picea sitchensis (Bong). Carr. Forestry 39: 68-94. Campbell, R.K. and A.I. Sugano. 1979. Genecology of bud-burst phenology in Douglas-fir: response to flushing temperature and chilling. Botanical Gazette 140: 223–231.  52  Charlesworth, B., D. Charlesworth, and N. H. Barton. 2003. The effects of genetic and geographic structure on neutral variation. Annual Review of Ecology, Evolution, and Systematics 34: 99-125. Chuine, I., S. N. Aitken, and C. C. Ying. 2001. Variation in threshold temperature produces genotype by environment interaction for growing season length in provenances of Pinus contorta. Canadian Journal of Forest Research 31: 1444-1455. Coop, G., J. K. Pickrell, J. Novembre, S. Kudaravalli, J. Li, D. Absher, R. M. Myers, L. L. Cavalli-Sforza, M. W. Feldman, and J. K. Pritchard. 2009. The role of geography in human adaptation. PLoS Genetics 5: 1-16. Davis, M. B., and R. G. Shaw. 2001. Range shifts and adaptive responses to Quaternary climate change. Science 292: 673-679. De Carvahlo, D., P. K. Ingvarsson, J. Joseph, L. Suter, C. Sedivy, D. Macaya-Sanz, J. Cottrell, B. Heinze, I. Schanzer, and C. Lexer. 2010. Admixture facilitates adaptation from standing variation in the European aspen (Populus tremula L.), a widespread forest tree. Molecular Ecology 19: 1638-1650. Doyle, J. J., and J. L Doyle. 1990. Isolation of plant DNA from fresh tissue. Focus 12: 13-15. Eckert, A. J., A. D. Bower, J. L. Wegrzyn, B. Pande, K. D. Jermstad, K. V. Krutovsky, J. B. St. Clair, and D. B. Neale. 2009. Association genetics of coast Douglas fir (Pseudotsuga menziesii var. menziesii, Pinaceae). I. cold-hardiness related traits. Genetics 182: 1289-1302. Eckert, A. J., A. D. Bower, S. C. Gonzalez-Martinez, J. L. Wegrzyn, G. Coop, and D. B. Neale. 2010. Back to nature: ecological genomics of loblolly pine (Pinus taeda, Pineaceae). Molecular Ecology 19: 3789-3805. Excoffier, L., T. Hofer, and M. Foll. 2009. Detecting loci under selection in a hierarchically structured population. Heredity 103: 285-298. Excoffier, L., and H. E. L. Lischer. 2010. Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Molecular Ecology Resources 10: 564-567. Fay, J. C., and C. Wu. 2000. Hitchhiking under positive Darwinian selection. Genetics 155: 1405-1413. Fox, J., and S. Weisberg. 2011. An R companion to applied regression, 2nd edition. SAGE Publications, CA.  53  Gaggiotti, O. E., D. Bekkevold, H. B. H. Jorgensen, F. Matthieu, G. R. Carvalho, C. Andre, and D. E. Ruzzante. 2009. Disentangling the effects of evolutionary, demographic, and environmental factors influencing genetic structure of natural populations: Atlantic herring as a case study. Evolution 63: 2939-2951. Gapare, W. J., and S. N. Aitken. 2005. Strong spatial genetic structure in peripheral but not core populations of Sitka spruce (Picea sitchensis (Bong.) Carr.). Molecular Ecology 14: 2659-2667. Gapare, W. J., S. N. Aitken, and C. E. Ritland. 2005. Genetic diversity of core and peripheral Sitka spruce (Picea sitchensis (Bong.) Carr) populations: implications for conservation of widespread species. Biological Conservation 123: 113-123. Garcia-Ramos, G., and M. Kirkpatrick. 1997. Genetic models of adaptation and gene flow in peripheral populations. Evolution 51: 21-28. Gockel , J., W. J. Kennington, A. A. Hoffmann, D. B. Goldstein, and L. Partridge. 2001. Nonclinality of molecular variation implicates selection in maintaining a morphological cline of Drosophila melanogaster. Genetics 158: 319-323. Gonzalez-Martinez, S. C., K. V. Krutovsky, and D. B. Neale. 2006. Forest-tree population genomics and adaptive evolution. New Phytologist 170: 227-238. Gonzalez-Martinez, S. C., N. C. Wheeler, E. Ersoz, C. D. Nelson, D. B. Neale. 2007. Association genetics in Pinus taeda L. I. Wood property traits. Genetics 175: 399409. Griffiths S., R. P. Dunford, G. Coupland, and D. A. Laurie. 2003. The evolution of CONSTANS-like gene families in barley, rice and Arabidopsis. Plant Physiology 131: 1855-1867. Grivet, D., F. Sebastiani, R. Alia, T. Bataillon, S. Torre, M. Zabal-Aguirre, G. G. Vendramin, and S. C. Gonzalez-Martinez. 2010. Molecular footprints of local adaptation in two Mediterranean conifers. Molecular Biology and Evolution 28: 101-116. Guillot, G., A. Estoup, F. Mortier, and J. F. Cosson. 2005. A spatial statistical model for landscape genetics. Genetics 170: 1261-1280. Hahn, M. W., M. V. Rockman, N. Soranzo, D. B. Goldstein, and G. A. Wray. 2004. Population genetic and phylogenetic evidence for positive selection on regulatory mutations at the factor VII locus in humans. Genetics 167: 867-877. Hall, D., V. Luquez, M. V. Garcia, K. R. St. Ongle, S. Jansson, and P. K. Ingvarsson. 2007. Adaptive population differentiation in phenology across a latitudinal gradient in European aspen (Populus tremula, L.): a comparison of neutral markers, candidate genes and phenotypic traits. Evolution 61: 2849-60.  54  Hamblin, M. T., E. E. Thompson, and A. Di Rienzo. 2002. Complex signatures of natural selection at the Duffy blood group locus. American Journal of Human Genetics 70: 369-383. Hancock, A. M., D. B. Witonsky, A. S. Gordon, G. Eshel, J. K. Pritchard, G. Coop, A. Di Rienzo. 2008. Adaptations to climate in candidate genes for common metabolic disorders. PLoS Genetics 4. Hartl, D. L., and A. G. Clark. 2007. Principles of population genetics, 4th edition. Pp. 525. Sinauer Associates, Inc., MA. Holliday, J. A., S. G. Ralph, R. White, J. Bohlmann, and S. N. Aitken. 2008. Global monitoring of autumn gene expression within and among phenotypically divergent populations of Sitka spruce (Picea sitchensis). New Phytologist 178: 103-122. Holliday, J. A., K. Ritland, and S. N. Aitken. 2010a. Widespread, ecologically relevant genetic markers developed from association mapping of climate-related traits in Sitka spruce (Picea sitchensis). New Phytologist 188: 501-514. Holliday, J. A., M. Yuen, K. Ritland, and S. N. Aitken. 2010b. Postglacial history of a widespread conifer produces inverse clines in selective neutrality tests. Molecular Ecology 19: 3857-3864. Howe, G. T., W. P. Hackett, G. R. Furnier and R. E. Klevorn. 1995. Photoperiodic responses of a northern and southern ecotype of black cottonwood. Physiologia Plantarium 93: 695-708. Howe, G. T., S. N. Aitken, D. B. Neale, K. D. Jermstad, N. C. Wheeler, and T. H. H. Chen. 2003. From genotype to phenotype: unravelling the complexities of cold adaptation in forest trees. Canadian Journal of Botany 81: 1247-1266. Hudson, M. E. 2008. Sequencing breakthroughs for genomic ecology and evolutionary biology. Molecular Ecology Resources 8: 3-17. Ingvarsson, P. K., M. V. Garcia, D. Hall, V. Luquez, and S. Jansson. 2006. Clinal variation in phyB2 , a candidate gene for day-length-induced growth cessation and bud set, across a latitudinal gradient in European aspen (Populus tremula). Genetics 172: 1845-1853. Ingvarsson, P. K., M. V. Garcia, V. Luquez, D. Hall, and S. Jansson. 2008. Nucleotide polymorphism and phenotypic associations within and around the phytochrome B2 locus in European aspen (Populus tremula, Salicaceae). Genetics 178: 2217-2226. Iverson, L. R., M. W. Schwartz, and A. Prasad. 2004. How fast and far might tree species migrate in the eastern United States due to climate change? Global Ecology and Biogeography 13: 209-219.  55  IPCC. 2007. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K. B. Averyt, M. Tignor and H.L. Miller, eds. Cambridge University Press, Cambridge, United Kingdom and New York, USA. Jaramillo-Correa, J., P. J. Beaulieu, and J. Bousquet. 2001. Contrasting evolutionary forces driving population structure at expressed sequence tag polymorphisms, allozymes and quantitative traits in white spruce. Molecular Ecology 10: 2729–2740. Jombart, T., S. Devillard, A-B.Dufour, and D. Pontier. 2008. Revealing cryptic spatial patterns in genetic variability by a new multivariate method. Heredity 101: 92-103. Jombart, T., D. Pontier, and A-B.Dufour. 2009. Genetic markers in the playground of multivariate analysis. Heredity 102: 330-341. Kawecki, T. J., and D. Ebert. 2004. Conceptual issues in local adaptation. Ecology Letters 7: 1225-1241. Kirkpatrick, M., and Barton N. H. 1997. Evolution of a species' range. The American Naturalist 150: 1-23. Knibb, W. R., J. G. Oakeshott, and J. B. Gibson. 1981. Chromosome inversion polymorphisms in Drosophila melanogaster. I. Latitudinal clines and associations between inversions in Australasian populations. Genetics 98: 833-847. Kohn, M. H., H. J. Pelz, and R. K. Wayne. 2003. Locus-specific genetic differentiation at Rw among warfarin-resistant rats (Rattus norvegicus). Genetics 164: 1055-1070. Knopp, T., J. M. Cano, P. A. Crochet, and J. Merila. 2007. Contrasting levels of variation in neutral and quantitative genetic loci on island populations of moor frogs (Rana arvalis). Conservation genetics 8: 45-56. Kremer, A., V. Le Corre, and S. Mariette. 2000. Population differentiation for adaptive traits and their underlying loci in forest trees: theoretical predictions and experimental results. Pp. 59-74 in C. Matyas, ed. Forest Genetics and Sustainability, vol.63. Kluwer Academic Publishers, The Netherlands. Latta, R. G. 1998. Differentiation of allelic frequencies at quantitative trait loci affecting locally adaptive traits. American Naturalist 151: 283-292. Le Corre, V., and A. Kremer. 2003. Genetic variability at neutral markers, quantitative trait loci and trait in a subdivided population under selection. Genetics 164: 1205-1219.  56  Legendre, P., and M. J. Fortin. 2010. Comparison of the Mantel test and alternative approaches for detecting complex multivariate relationships in the spatial analysis of genetic data. Molecular Ecology Resources 10: 831-844. Leinonen, T., R. O' Hara, J. M. Cano, and J. Merila. 2008. Comparative studies of quantitative trait and neutral marker divergence: a meta-analysis. Journal of Evolutionary Biology 21: 1-17. Lia, V. V., V. A. Confalonieri, and L. Poggio. 2007. B chromosome polymorphism in maize landraces: adaptive vs. demographic hypothesis of clinal variation. Genetics 177: 895-904. Loveless, M. D., and J. L. Hamrick. 1984. Ecological determinants of genetic structure in plant populations. Annual Review of Ecology and Systematics 15: 65-95. Luikart, G., P. England, D. Tallmon, S. Jordan, and P. Taberlet. 2003. The power and promise of population genomics: from genotyping to genome typing. Nature Review Genetics 4: 981-994. Ma, X. F., D. Hall, K. R. St. Onge, S. Jansson, and P. K. Ingvarsson. 2010. Genetic differentiation, clinal variation and phenotypic associations with growth cessation across the Populus tremula photoperiodic pathway. Genetics 186: 1033-1044. Malcolm, J.R., A. Markham, R. P. Neilson, and M. Garaci. 2002. Estimated migration rates under scenarios of global climate change. Journal of Biogeography 29: 835-849. Manolio, T. A., F. S. Collins, N. J. Cox, D. B. Goldstein, L. A. Hindorff, D. J. Hunter, M. I. McCarthy, E. M. Ramos, L. R. Cardon, A. Chakravarti, J. H. Cho, A. E. Guttmacher, A. Kong, L. Kruglyak, E. Mardis, C. N. Rotimi, M. Slatkin, D. Valle, A. S. Whittemore, M. Boehnke, A. G. Clark, E. E. Eichler, G. Gibson, J. L. Haines, T. F. Mackay, S. A. McCarroll, P. M. Visscher. 2009. Finding the missing heritability of complex diseases. Nature 461: 747-753. Matyas, C. 1996. Climatic adaptation of trees: rediscovering provenance tests. Euphytica 92: 45-54. McKay, J. K., and R. G. Latta. 2002. Adaptive population divergence: markers, QTL and traits. Trends in Ecology and Evolution 17: 285-291. McLachlan, J. S., J. S. Clark, and P. S. Manos. Molecular indicators of tree migration capacity under rapid climate change. Ecology 86: 2088-2098. Mettler, L. E., R. A. Voelker, and T. Mukai. 1977. Inversion clines in populations of Drosophila melanogaster 87: 169-176.  57  Mimura, M., and S. N. Aitken. 2007a. Adaptive gradients and isolation-by-distance with postglacial migration in Picea sitchensis. Heredity 99: 224-232. Mimura, M., and S. N. Aitken. 2007b. Increased selfing and decreased effective pollen donor number in peripheral relative to central populations in Picea sitchensis (Pinaceae). American Journal of Botany 94: 991-998. Mimura, M., and S. N. Aitken. 2010. Local adaptation at the range peripheries of Sitka spruce. Journal of Evolutionary Biology 23: 249-258. Moore, J. P., E. Nguema-Ona, L. Chevalier, G. G. Lindsey, W. F. Brandt, P. Lerouge, J. M. Farrant, A. Driouich. 2006. Response of the leaf cell wall to desiccation in the resurrection plant Myrothamnus flabellifolius. Plant Physiology 141: 651-662. Morgenstern, E. K. 1996. Geographic variation in forest trees. University of British Columbia Press, Vancouver. Namroud, M. C., J. Beaulieu, N. Juge, J. Laroche, and J. Bousquet. 2008. Scanning the genome for gene single nucleotide polymorphisms involved in adaptive population differentiation in white spruce. Molecular Ecology 17: 3599-3613. Neale, D. B., and O. Savolainen. 2004. Association genetics of complex traits in conifers. Trends in Plant Science 9: 325-330. Neale, D. B., and P. K. Ingvarsson. 2008. Population, quantitative and comparative genomics of adaptation in forest trees. Current Opinion in Plant Biology 11: 149-155. Neale, D. B., and A. Kremer. 2011. Forest tree genomics: growing resources and applications. Nature Reviews Genetics 12: 111-122. Nielsen, R. 2001. Statistical tests of selective neutrality in the age of genomics. Heredity 86: 641-647. Nielsen, R. 2005. Molecular signatures of natural selection. Annual Review of Genetics 39: 197-218. Oakeshott, J. G., J. B. Gibson, P. R. Anderson, W. R. Knibb, D. G. Anderson, and G. K. Chambers. 1982. Alcohol dehydrogenase and glycerol-3-phosphate dehydrogenase clines in Drosophila melanogaster on different continents. Evolution 36: 86-96. Pallardy, S. G. 2008. Physiology of woody plants. 3rd ed. Pp. 44. Elsevier, USA. Patterson, N., A. L. Price, and D. Reich. 2006. Population structure and eigenanalysis. PLoS Genetics 2: 2074-2093.  58  Parker, K. C., J. L. Hamrick, A. J. Parker, and J. D. Nason. 2001. Fine-scale genetic structure in Pinus clausa (Pinaceae) populations: effects of disturbance history. Heredity 87: 99-113. Price, A. L., N. J. Patterson, R. M. Plenge, M. E. Weinblatt, N.A. Shadick, and D. Reich. 2006. Principal component analysis corrects for stratification in genome-wide association studies. Nature Genetics 38: 904-909. Pritchard, J. K., M. Stephens, and P. Donnelly. 2000. Inference of population structure using multilocus genotype data. Genetics 155: 945-959. Przeworski, M., G. Coop, and J. D. Wall. 2005. The signature of positive selection on standing genetic variation. Evolution 59: 2312-2323. Raymond, M., and F. Rousset. 1995. GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. Journal of Heredity 86: 248-249. Reed D. H., and R. Frankham. 2001. How closely correlated are molecular and quantitative measures of genetic variation? A meta-analysis. Evolution 55: 1095-1103. Rehfeldt, G. E., C. C. Ying, D. L. Spittlehouse, and D. A. Hamilton Jr. 1999. Genetic responses to climate in Pinus contorta: Niche breadth, climate change and reforestation. Ecological Monographs 69: 375-407. Robledo-Arnuncio, J. J., R. Alia, L. Gil. 2004. Increased selfing and correlated paternity in a small population of a predominantly outcrossing conifer, Pinus sylvestris. Molecular Ecology 13: 2567-2577. Sakai, A., and W. Larcher. 1987. Frost Survival of Plants. Springer-Verlag, Berlin. Samis, K. E., K. D. Heath, and J. R. Stinchcombe. 2008. Discordant longitudinal clines in flowering time and Phytochrome C in Arabidopsis thaliana. Evolution 62: 29712983. Savolainen, O., T. Pyhajarvi, and T. Knurr. 2007. Gene flow and local adaptation in trees. Annual Review of Ecology, Evolution and Systematics 38: 595-619. Savolainen, O., and T. Pyhajarvi. 2007. Genomic diversity in forest trees. Current Opinion in Plant Biology 10: 162-167. Sezgin, E., D. D. Duvernell, L. M. Matzkin, Y. Duan, C. T. Zhu, B. C. Verrelli, and W. F. Eanes. 2004. Single-locus latitudinal clines and their relationship to temperate adaptation in metabolic genes and derived alleles in Drosophila melanogaster. Genetics 168: 923-931.  59  Sgro, C. M., A. Magiafoglou, L. Faine, and A. A. Hoffmann. 2006. Absence of clinal variation in virgin retention capacity in Australian Drosophila melanogaster. Evolutionary Ecology 20: 407-413. Stapley, J., J. Reger, P. G. D. Feulner, C. Smadja, J. Galindo, R. Ekblom, C. Bennison, A. D. Ball, A. P. Beckerman, and J. Slate. 2010. Adaptation genomics: the next generation. Trends in Ecology and Evolution 25: 705-712. Storz, J. F. 2005. Using genome scans of DNA polymorphism to infer adaptive population divergence. Molecular Ecology 14: 671-688. Tajima, F. 1989. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123: 585-595. Vasemagi, A. The adaptive hypothesis of clinal variation revisited: single-locus clines as a result of spatially restricted gene flow. Genetics 173: 2411-2414. Wang, T., A. Hamann, D. Spittlehouse, and S. N. Aitken. 2006. Development of scale-free climate data for western Canada for use in resource management. International Journal of Climatology 26: 383-397. Weir, B. S. 1996. Genetic data analysis 2: Methods for discrete population genetic data. Sinauer Associates, MA. Weir, B. S., and C. C. Cockerham. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358-1370. Whitlock, M. C. 2005. Combining probability from independent tests: the weighted Zmethod is superior to Fisher's approach. Journal of Evolutionary Biology 18: 13681373. Xu, P., C. C. Ying, and Y. A. El-Kassaby. 2000. Multivariate analyses of causal correlation between growth and climate in Sitka spruce. Silvae Genetica 49: 257-263. Yeh, F.C., Y.A. El-Kassaby. 1980. Enzyme variation in natural populations of Sitka spruce (Picea sitchensis). 1. Genetic variation patterns among trees from 10 IUFRO provenances. Canadian Journal of Forest Research 10: 415-422. Yu, J. M., G. Pressoir, W. H. Briggs, I. V. Bi, M. Yamasaki, J. F. Doebley, M. D. McMullen, B. S. Gaut, D. M. Nielsen, J. B. Holland, S. Kresovich, E. S. Buckler. 2006. A unified mixed-model method for association mapping that accounts for multiple levels of relatedness. Nature Genetics 38: 203-208. Zhao, K., M. J. Aranzana, S. Kim, C. Lister, C. Shindo, C. Tang, C. Toomajian, H. Zheng, C. Dean, P. Majoram, and M. Nordborg. 2007. An Arabidopsis example of association mapping in structured samples. PLoS Genetics 3: e4.  60  Appendix: Single-locus clines for mature and seedling populations SNP  Full MAT range for seedlings  Truncated MAT range for seedlings  209-523-S  162-350-S  257-188-NS  61  SNP  Full MAT range for seedlings  Truncated MAT range for seedlings  260-504-S  56-206-S  257-264-S  62  SNP  Full MAT range for seedlings  Truncated MAT range for seedlings  73-182-NS  258-207-S  266-573-S  63  SNP  Full MAT range for seedlings  Truncated MAT range for seedlings  103-455-NS  99-395-NS  89-475-NS  64  SNP  Full MAT range for seedlings  Truncated MAT range for seedlings  14-438-S  87-319-S  207-363-S  65  SNP  Full MAT range for seedlings  Truncated MAT range for seedlings  19-567-S  62-424-S  66  

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.24.1-0103315/manifest

Comment

Related Items