ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2020

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1899

Past demography and localadaptation in forest trees

Insights from natural populations and breedingprograms of Norway spruce

LILI LI

ISSN 1651-6214ISBN 978-91-513-0865-4urn:nbn:se:uu:diva-403450

Dissertation presented at Uppsala University to be publicly examined in Ekmansalen, EBC,Norbyvägen 18, Uppsala, Thursday, 19 March 2020 at 10:00 for the degree of Doctor ofPhilosophy. The examination will be conducted in English. Faculty examiner: Brunno Fady(INRAE Avignon, France).

AbstractLi, L. 2020. Past demography and local adaptation in forest trees. Insights from naturalpopulations and breeding programs of Norway spruce. Digital Comprehensive Summariesof Uppsala Dissertations from the Faculty of Science and Technology 1899. 60 pp. Uppsala:Acta Universitatis Upsaliensis. ISBN 978-91-513-0865-4.

Spatial changes in natural selection patterns can give rise to local adaptation and geneticdifferentiation between populations. Local adaptation for phenological traits is pronounced inmany forest tree species. The Swedish breeding program was established from ‘plus’ treesselected across the country and can therefore be a very good source of information on localadaptation. In the present thesis, we estimated the genetic basis of local adaptation in twoEurasian spruce species Norway spruce (P. abies) and Siberian species (P. obovata) using large-scale whole-exome data and Sanger sequences from samples taken from the Swedish breedingprogram and from natural populations.

To detect signals of local adaptation in Norway spruce (P. abies), we started by studyingpopulation genetic clustering and inferring the demographic history of the species. In additionto the already known three main domains in Norway spruce, we also found four genetic clusterscreated by admixture events between the aforementioned three main clusters. Demographicinferences indicated two recolonizations directions in Scandinavia: east to west (from centralRussia and Siberia) and south to north (from Alpine and Carpathian), but also revealed repeatedhybridization between P. abies and P. obovata and gene flow among clusters. We next estimatedthe genetic basis of local adaptation of three phenotypic traits (height, diameter and bud-burst) by multivariate analyses and genome-wide association studies. The results showed thatgeographical origin is a strong predictor of growth and phenology and trees of southern originsoutcompeted local provenances. We further revealed that growth traits were highly polygenicand bud-burst somewhat less.

Population genetic structure largely affects the detection of local adaptation. Therefore wefurther visualized the fine-scale map of population genetic structure through dense sampling oftrees from the Swedish breeding program. Trees of Swedish origins were assigned into two mainclusters with an admixture zone in central Sweden and the genetic contribution from P. obovatawas detected in northern Sweden. A large number of SNPs were found to be associated withenvironmental variables and exhibited a stronger pattern of isolation-by-distance than randomSNPs.

Finally we tested for local adaptation in two well-defined candidate genes (FTL2 and GI)of phenology in P. obovata. Clinal variation in FTL2 gene expression, growth cessation, andallele frequency of FLT2 and GI were revealed in populations along Ob River, paralleling theones in Norway spruce populations in Scandinavia and in Siberian spruce populations alongthe Yenisei River.

Keywords: clines, demographic history, local adaptation, , ,population genetic structure, Swedish breeding program

Lili Li, Department of Ecology and Genetics, Plant Ecology and Evolution, Norbyvägen 18 D,Uppsala University, SE-752 36 Uppsala, Sweden.

© Lili Li 2020

ISSN 1651-6214ISBN 978-91-513-0865-4urn:nbn:se:uu:diva-403450 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-403450)

Picea

abies Picea

obovata

“When you reach the end of your rope, tie a knot in it and hang on.”

Franklin D. Roosevelt

To my mom

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Chen J, Li L, Milesi P, Jansson G, Berlin B, Karlsson B, Ale-

ksic J, Vendramin GG, and Lascoux M (2019). Genomic data provides new insights on the demographic history and the ex-tent of recent material transfers in Norway spruce. Evol. Appli-cations 8:1539-1551.

II Milesi P, Berlin M, Chen J, Orsucci M, Li L, Jansson G, Karls-son B, and Lascoux M (2019). Assessing the potential for as-sisted gene flow using past introduction of Norway spruce in Southern Sweden: Local adaptation and genetic basis of quanti-tative traits in trees. Evol. Applications 2: 1946-1959.

III Li L., Milesi P., Chen J., Baison J., Chen Z., Zhou L., Karlsson B., Berlin M., Westin J., Garcia-Gil R., Wu H. and Lascoux M (2020). Population transfers and population movements: a first fine-scale mapping of the population genetic structure of Nor-way spruce across Sweden. Manuscript.

IV Li L., Chen J. and Lascoux M (2019). Clinal variation in growth cessation and FTL2 expression in Siberian spruce. Tree Genet Genomes 15:82.

Reprints were made with permission from the respective publishers.

Contents

Introduction ................................................................................................ 91. What is local adaptation and is it common? ........................................ 92. How does local adaptation work? ......................................................103. Detecting local adaptation at the phenotypic and genetic levels ..........114. Which factors affect population genetic structure? .............................12Effect of demographic history on population genetic structure ...............14

Recovering demographic history .......................................................14Impact of Quaternary climate cycles on population demography of Boreal species ...................................................................................15Population history of Norway spruce in Europe ................................16

Effect of recent translocation on population genetic structure ................17Detection of local adaptation at the molecular level ...............................18

Association mapping studies .............................................................18Genotype-environment associations ..................................................19FST outliers .......................................................................................20Detecting local adaptation from clinal variation ................................20

Gathering information from natural populations and breeding programs in Norway spruce ..................................................................................22

The present thesis ......................................................................................24Paper I: Demographic history and translocation of Norway spruce ........24Paper II: Local adaptation and genetic basic of quantitative traits in spruce ...................................................................................................28Paper III: Fine-scale population genetic structure and local adaptation in Norway spruce across Sweden...........................................................32Paper IV: Inference of the cline in Siberian spruce populations along Ob River ...............................................................................................37

Conclusions and future perspectives ..........................................................42

Summary in Swedish (Sammanfattning) ....................................................45

Acknowledgements ...................................................................................48

Reference ..................................................................................................53

9

Introduction

1. What is local adaptation and is it common? Local individuals tend to have a competitive advantage towards introduced individuals in their native sites if those individuals are locally adapted. If natural selection is spatially variable and strong enough relative to other evolutionary forces, then local adaptation can drive genetic differentiation among populations. Therefore, local adaptation provides important insights into ongoing evolutionary processes along environmental gradients. Local adaptation was initially discovered while trying to understand the relation-ship between different plant phenotypes and their surrounding environment using common garden experiments. Turesson, who coined the term “eco-type” (Turesson 1922), cultivated birch populations sampled from South to North in Sweden, under the same environmental conditions in Uppsala and observed that to avoid frost damage, tree growth period in the north was shorter than in the south (Figure 1). Another milestone in local adaptation research was the introduction of reciprocal transplant experiments by Clausen, Keck and Heisey (Clausen et al. 1940). The foundation experiments by Clausen and a large number of subsequent studies in both plants and ani-mals have demonstrated that local adaptation is widespread among species (Leimu and Fisher 2008; Ågren and Schemske 2012; Savolainen et al. 2013; Lascoux et al. 2016).

10

Figure 1. Birch trees that originated from different latitudes across Sweden and were planted south of Uppsala in Central Sweden show a clear latitudinal pattern in growth cessation in the fall Photo: Jon Ågren. (modified from Lascoux et al. 2016).

2. How does local adaptation work? Local adaptation results from the interaction of multiple evolutionary forces, including natural selection, genetic drift and gene flow. A genetic descrip-tion of local adaptation between two locally adapted demes (ecotypes) fol-lows. For single locus, maintenance of alleles in an infinite population de-pends on a balance between migration and selection. Let us first assume a simple haploid two-deme model with reciprocal migration where m1 is the migration rate from deme 1 to deme 2 and m2 is the migration rate from deme 2 to deme 1. We further assume that the demes are large enough so that genetic drift can be ignored. We have a single locus with two alleles A1 and A2. A1 is favored in deme 1 and is selected against in deme 2. On the opposite, A2 is selected against in deme 1 and favored in deme 2. s1 and s2 are the selection coefficients of A1 in demes 1 and 2, respectively, and are of opposite sign. Bulmer (1972) showed that polymorphism would be main-tained if

If migration is strong enough relative to local selection, one of the two al-leles will disappear. On the other hand, if local selection is stronger than migration, both alleles will be maintained (Bulmer 1972). However, when migration and local selection balance each other in finite populations, genet-ic drift will reduce the domain of values of migration and selection needed

11

for maintaining a polymorphism with locally adapted alleles (Yeaman and Otto 2011).

The situation with quantitative traits that are controlled by many loci is obviously more complicated, even though the basic processes will remain the same. Importantly, quantitative traits can achieve local adaptation through small allele frequency shifts across loci and allelic covariances, even when facing strong gene flow (Le Corre and Kremer 2003; Berg and Coop 2014; Yeaman 2015). This model, initially proposed by Kremer and Le Cor-re (2003) could, in particular, explain the apparently paradoxical situation of forest trees that combine extensive gene flow and strong local adaptation at phenological traits such as bud-set.

3. Detecting local adaptation at the phenotypic and genetic levels Reciprocal transplant experiments, as mentioned above, are the most direct approaches to investigate local adaption at the phenotypic level, however, these experiments are often restricted to a limited number of environments. Common garden experiments are an alternative to reciprocal transplant ex-periments. Basically, a number of populations are planted at common sites and their fitness (or any other traits) measured. If many common gardens are used one can also capture the interaction between genotypes and local envi-ronments on fitness. Extensive common garden experiments have been in-stalled for forest trees of economic interest and are generally referred to as provenance tests. For instance, in Norway spruce a large series of prove-nance tests were installed in the late 1960’s across Europe and were, in part, the basis of the study of Lagercrantz and Ryman (1990). To this day they remain one of the main sources of information on local adaptation in forest trees, both as a source of information on phenotypic local adaptation but also as a convenient source of material for population genetic studies.

Detection of local adaptation at the genomic level is a more recent devel-opment that has bloomed as technologies allowing fine analysis of genetic polymorphisms became readily available. While technology has made these analyses more available, biological complexities still remain and can make interpretation of these studies difficult. One of the main difficulties when trying to detect the genetic bases of local adaptation is to account for the confounding effect of population structure. For example, population struc-ture due to demographic events can correlates with the environmental varia-bles. Correcting for population structure in tests for local adaptation could lead to a high rate of false negatives. On the other hand, in a similar situa-tion, not correcting for population structure can lead to a high rate of false positives. For this reason, it is particularly important to properly account for

12

population structure when estimating the impact of local adaptation at the genetic level.

4. Which factors affect population genetic structure? Population genetic structure describes genetic relatedness within and among populations. Individuals within populations are expected to be more closely related than individuals between populations. Population structure can cause departures from Hardy-Weinberg equilibrium (HWE) because of non-random mating among populations, or because of departure from random mating within a single population.

The Wahlund effect (Wahlund 1928) captures the effect of population structure on allele frequencies. In the presence of population structure there will be a deficit of heterozygotes in subpopulations compared to what would be expected if the subpopulations were mating randomly. Wright fixation indices, FST and FST-like measures have been extensively used to measure population differentiation, although they are in fact measurements of the amount of drift that has contributed to the divergence of populations. Basi-cally, FST estimates the amount of genetic variance in all populations that can be explained by among-population variance as opposed to within-population variance (Wright 1931, 1951; Nei 1972; Slatkin 1995; Weir 1996). It should be noted that FST-like methods are strongly affected by the amount of genetic variance within populations. For instance, assume two populations each with 50 diploid individuals. We study a single locus at which 5 alleles are segre-gating in each population, alleles A1-A5 in population 1 and alleles A6-A10 in population 2. Hence, the two populations do not have any allele in com-mon. Finally we assume that each population is in Hardy-Weinberg equilib-rium at the locus. So, from the point of view of allele content the two popu-lations are completely different. Yet, FST between the two populations will be low. To make things easy, assume that the frequencies of each of the five alleles segregating in either population are p = 20/100 = 1/5. We then have HS1 = HS2 = 1-5(1/5)2 = 4/5 in each population and HT = 1-10(20/200)2 = 9/10 in the total population. Hence, since by definition, FST = 1-HS/HT, where HS = (HS1+HS2)/2, we have FST = 1-(4/5)/(9/10) = 1-8/9 = 1/9≈0.11. This example illustrates what FST actually means and the risk to use it as a measure of population genetic differentiation.

When not considering selection and new mutation events, genetic drift and migration play as two major antagonistic roles in affecting population differentiation. Migration tends to erode differences in allele frequencies between populations, while genetic drift increases the heterogeneity. This is the basic principle behind Wright’s infinite island model (Wright 1931), which assumes that populations are under migration-drift equilibrium and that there is no spatial structure among populations and each individual is as

13

likely to migrate to a nearby population than to a population far away. Clear-ly, both assumptions are not suitable when populations share ancestral poly-morphisms and when migration depends on distance, yet the model captures an essential property of migration: that a few migrants per generation will suffice to offset the effect of random genetic drift. For finite populations with limited dispersal, it is necessary to check if isolation by distance (IBD) can explain population structure patterns across space through regressing FST

or a function of it between pairs of populations against geographical distance (Slatkin 1993; Rousset 1997).

All methods mentioned above assume that we already have identified dis-tinct populations. However, in practice, we often need to first define popula-tions by assigning individuals to a certain number of clusters for downstream population genetic analyses in the hope to identify demographic history, local adaption and migration between populations (Bradburd et al. 2018). Sample assignments based only on their geographic location can be mislead-ing because samples collected from the same place may actually belong to distinct genetic populations. This is particularly the case for species that have been strongly influenced by human-mediated gene flow, such as do-mesticated or semi-domesticated species (Herben and Suzuki 2001; Frank et al. 2017). Population structure can first be inferred through the use of princi-pal component analysis (PCA) (Galinsky et al. 2016). PCA is a very power-ful exploratory method but it is sensitive to sampling schemes across the landscape and interpretation of the results can be difficult (Novembre and Stephens 2008; De Giorgio and Rosenberg 2013; Petkova et al. 2015). Mod-el based genetic clustering, implemented in software such as STRUCTURE (Pritchard et al. 2000) and ADMIXTURE (Alexander et al. 2009), have been used extensively to identify an optimal number of ancestral populations from which current individuals are derived. Both STRUCTURE and ADMIX-TURE group individuals into discrete units. Admixed individuals (individu-als sharing alleles from multiple populations) will then be assigned partly to different clusters. Both STRUCTURE and ADMIXTURE are based on a rather simple pure drift demographic model of population divergence. Popu-lation structure is shaped by many factors, including continuous migration and drift, as well as by more discrete events for example sudden changes in effective population size and separation by environmental or geographical barriers (Wright 1931; Slatkin 1987; Charlesworth 2009). Therefore, it is critical to consider both discrete and continuous processes when characteriz-ing ancestry components across space. The very recently developed method conStruct (Bradburd et al. 2018) based on an isolation-by-distance (IBD) model uses geo-referenced samples to simultaneously characterize continu-ous and discrete patterns of population structure. Finally to visualize popula-tion genetic structure and account for migration patterns across space, Petkova et al. (2015) developed Estimated Effective Migration Surfaces (EEMS). EEMS is based on a stepping stone model (Kimura and Weiss

14

1964), in which individuals migrate locally between subpopulations and migration rates varies by geographical distance, and can capture deviations from exact isolation by distance in order to identify if contact zones and gene flow barriers exist. EEMS is also less sensitive to the sampling scheme than PCA.

Effect of demographic history on population genetic structure Recovering demographic history In order to interpret population structure properly, it is essential to under-stand species history. Genetic data have been widely used to infer the demo-graphic history of populations and species. How genetic variation is distrib-uted within current populations contains information about past population size changes, splits, admixture events and migration. It remains challenging to distinguish between the causes of shared polymorphisms among popula-tions or species because both incomplete lineage sorting (the process result-ing in gene tree differs from the populations or species tree) and migration could result in shared polymorphisms. Many approaches have been devel-oped to discriminate between isolation and migration since Nielsen and Wakeley (2001) introduced the isolation-with-migration (IM) model. For two populations, the basic IM model contains six parameters: three scaled population mutation rates for the ancestral population and two descendant populations (θA, θ1, θ2; θ = 4Neμ, where Ne is the effective population size), two migration rates in both directions (m12, m21) and divergence time since split (t). Migration rates and divergence time are usually scaled by the muta-tion rate in the estimations (m/μ, tμ; (Hey and Nielsen 2004)). The IM model assumes that all loci are free of natural selection, no recombination events within genes, and there is no further hierarchical structure within the de-scendant populations or the ancestral population. There are many approaches to estimate parameters in IM model depending on the type of genetic data (un-linked SNPs, short sequence blocks or chromosome-scale sequences) and computational and statistical methods (likelihood, Bayesian and heuris-tic methods).

Afterwards, more approaches have been designed to i) estimate demo-graphic parameters that allow free recombination within genes, ii) deal with large number of individuals and iii) accommodate more complex demo-graphic models. One of the most widely used methods is Approximate Bayesian Computation (ABC) which bypasses likelihood functions (Beaumont et al. 2002). Summary statistics estimated from the observed data set are compared to the simulated ones under a hypothesized evolutionary scenario, and if the differences between the observed and simulated values

15

of the summary statistics are lower than a certain threshold, we kept the pa-rameter used in the simulated model (Sunnåker et al. 2013). The posterior probabilities of different models can be computed and the best fit model chosen based on Bayes factors. The whole process is repeated for a certain number of times until we get a posterior distribution of parameters. Howev-er, computation time increases rapidly with the increase of the complexity of the model, i.e. the number of parameters to estimate. Recently developed composite-likelihood methods based on the site frequency spectrum are an alternative to ABC approach (Adams and Hudson 2004; Gutenkunst et al. 2009; Lukic and Hey 2012). This method can be applied to very flexible demographic scenarios for two or more populations and runs fast enough with a large parameter space (Gutenkunst et al. 2009; Excoffier et al. 2013).

Impact of Quaternary climate cycles on population demography of Boreal species Species were limited to refugia as the ice sheet increased during glacial peri-ods and they expanded from those refugia as the climate warming during the interglacial periods (Hultén 1937; Dépraz et al. 2008; Pilot et al. 2014). In some species during population contraction phase, geographic barriers in-creased between populations in different refugia, resulting in reduced gene flow and high intraspecific genetic divergence or even speciation events (Petit et al. 2003). During the expansion phase, gene flow increased due to secondary contacts between marginal populations in different refugia leading to increased intraspecies genetic diversity or form hybrids (Hewitt 2000; Arnold 2004) For boreal forest trees, colonization of new habitats arises through seed dispersal. Chloroplast DNA markers that are maternally inher-ited and therefore only dispersed through seeds are well fitted to have a first estimate of range changes. Petit et al. (2003) compared genetic variation of chloroplast DNA for 22 forest trees and shrubs sampled across Europe. Pop-ulations in Western Europe harbored more diverse chloroplast DNA than southern ones, a pattern that likely reflects the mixing of the multiple migra-tion lineages stemming from the southern refugia. This is, for instance, clear for temperate species such as European white oaks. Subsequent studies em-ployed both pollen macrofossils and nuclear and organelle markers, and showed that cold tolerant species, for instance Betula, Salix, and Picea, could have survived during the Last Maximum Glacial beyond southern European peninsulas (Iberian Peninsula, Italy and the Balkans) (Lascoux et al. 2004). In these species it was hard to detect any clear geographical pat-tern, most haplotypes being distributed across the range (Lascoux et al. 2004). It should be noted, however, that the lack of clear population structure observed for cytoplasmic markers may not necessarily be observed at the nuclear level, in particular between species, because introgression are likely

16

more frequent between cytoplasmic markers than nuclear ones (e.g. Tsuda et al. 2016). The comparison between the two types of markers can indeed provide very useful information on past population movements (Currat et al. 2008).

Population history of Norway spruce in Europe Norway spruce is widely and continuously distributed in northern Europe and Western Siberia, while it is restricted to mountains in central Europe (Farjon and Filer 2013). According to morphological differences of cones (Borghetti et al. 1988), as well as population genetic and biogeographical studies based on different data sets, for instance organelle and nuclear DNA marker and genome-wide polymorphisms (Lagercrantz and Ryman 1990; Acheré et al. 2005; Heuertz et al. 2006; Tollefsrud et al. 2008; Chen et al. 2016; Tsuda et al. 2016; Fagernas 2017), the current distribution of P. abies in Europe can be divided into three main domains: a vast northern domain covering almost entirely Fennoscandia and western Russia, an Alpine do-main and a Carpathian domain. The demographic history of P. abies has been widely studied by means of pollen fossils and polymorphism of differ-ent types of genetic markers varying from the early isozymes, cytoplasmic and nuclear loci to next generation sequencing data, e.g. exome sequences. However, it is still challenging to reveal the full picture of the demographic history of P. abies from standing genetic variation because Norway spruce, like many other plant and animal species, went through a series of repeated cycles of contraction and expansion and secondary contacts during gla-cial/interglacial periods. Fossil pollen records and mitochondrial and chloro-plast DNA demonstrated that the current range of P. abies was established from three large refugia (Giesecke and Bennett 2004; Tollefsrud et al. 2008). The northern domain was colonized from a large refugium probably located in Central Russia, through two migration paths: the northwest one from Rus-sia across Finland towards Scandinavia and the southwest path across the Baltics sea to reach southern Scandinavia. Two southern domains came from an Alpine refugium and a Carpathian refugium. Further studies based on nuclear and cytoplasmic (mtDNA) markers demonstrated extensive admix-ture between Siberian spruce and P. abies with a large hybrid zone around the Urals mountains, (Tsuda et al. 2016). The phylogenetic relationships between northern and southern populations of P. abies and P. obovata have sometimes been controversial. Mitochondrial DNA markers grouped P. abies northern populations and P. obovata into the same clade while P. abies southern populations into one single cluster (Lockwood et al. 2013; Tsuda et al. 2016), which is different from the pattern obtained from Simple Sequence Repeats (SSRs) where southern and northern populations of P. abies formed one group and P. obovata formed another (Tsuda et al. 2016). In spruce, mitochondrial DNA is maternally inherited and spread only through seeds

17

whereas nuclear DNA is biparentally inherited and spread through both pol-len and seeds. The divergence between introgression patterns of the two types of markers might explain the conflicting topological relationships. In particular, the model of introgression between a local population and an in-vading one proposed by Currat et al. (2008), where introgression is primarily from the local population to the invading one, could explain the current ge-netic diversity pattern.

Effect of recent translocation on population genetic structure Translocations are defined as intentional or accidental human-mediated movements of living organisms from one place to another (IUCN 1987). To maintain biodiversity and to stabilize ecosystems under the threat of climate change and the habitat fragmentation, translocations have been widely used in species conservation and gene-assisted pre-adaptation (Weeks et al. 2011; Aitken and Whitlock 2013). Translocations are also common for ornamental trees or during forest trees breeding programs where they are used to in-crease forest production (Wagner et al. 2015; Frank et al. 2017; Jansen et al. 2017). Forest tree breeding (for instance pine, spruce, larch or poplar) across Europe have been developed for more than 300 years according to historical records, while modern tree breeding programs are much more recent. The latter were established from seeds collected on ‘plus’ trees (trees with supe-rior phenotypes in quality or quantity) from natural stands. Human-mediated seed transfer tends to change the genetic composition of native populations. European larch is a clear example of the impact of recent translocations im-pact on local genetic diversity, in part because larch has a very limited pollen dispersal. Wagner et al. (2015) showed that local discontinuities in the dis-tribution of genotypes across the Larch species range in Central Europe were the results of past within-range translocations (Wagner et al. 2015). P. abies has high performances under different environmental conditions and it can be found outside of its natural distribution. Therefore seeds from P. abies have been used for afforestation on a large-scale outside of its initial range in Europe since the middle of the 18th century. For instance, large amounts of seeds from Germany were imported for afforestation in Scandinavia, espe-cially in Southern Sweden, Denmark and Western Norway, from the begin-ning to the middle of the 20th century. In 1950s, seeds were also extensively transported from Central Europe (France, Belgium or Hungary) and Eastern Europe (Russia, Estonia, Belarus, Northern Poland, Czech Republic or Ro-mania) into Southern Sweden (Myking et al. 2016). Those frequent translo-cations likely resulted in extensive gene flow between local and introduced P. abies populations. Frank et al. (2017) indicated that P. abies exhibited a

18

high intraspecific adaptive differentiation and strong associations between seedling traits and environmental variables, which can provide vital infor-mation for afforestation and to predict the effects of future climate change on forest growth. Frank et al. (2017) also found that the natural gene pool of P. abies in Central and Eastern Europe had been significantly altered by hu-man-mediated migration.

Detection of local adaptation at the molecular level Many methods have been developed to infer the genetic signature of local adaptation. These methods include FST outliers scans and FST related meth-ods which describe patterns of allele fixation between populations adapted to different conditions, genotype-environment associations which identify links between putatively adaptive loci and the environments they are adapting to, clinal studies that investigate how allele frequencies change across a contin-uous habitat variable and association mapping studies, in particular genome-wide association study (GWAS), which aim to link alleles with known adap-tive traits.

Association mapping studies Association mapping aims to confirm candidate genes or identify new genes that are significantly associated with target phenotypic traits. There are two main categories of association mapping studies, candidate gene based-association approach and genome-wide association study (GWAS) without pre-defined candidate genes. Association mapping is based on the presence of linkage disequilibrium. Forest trees have high genetic diversity, rapid linkage disequilibrium (LD) decay over short genomic distance and weak population structure; they therefore seem ideal species for genetic associa-tion studies (Neale and Savolainen 2004). Early association mapping efforts in forest trees, were primarily candidate gene studies and identified associa-tions of candidate genes with wood quality and growth traits in poplar, pine and spruce (Ingvarsson et al. 2008; Dillon et al. 2010; Beaulieu et al. 2011; Prunier et al. 2013). However candidate gene association studies only can detect a few associations with traits of interests. With the development of high throughput genotyping technologies, GWAS based on a high number of single nucleotide polymorphisms (SNPs) were applied to forest trees and started to unravel the genetic architecture of important growth and adaptive traits. While yielding interesting information these studies have generally explained a limited percentage of the total variance of the traits. This is like-ly due to the fact that, for most quantitative traits, the bulk of the genetic variation segregating in natural populations is due to large numbers of vari-ants with small allele effects, rather than a few loci with alleles of large-

19

effect (Field et al. 2016; Lamara et al. 2016); it is also difficult to capture rare variants contributing significantly to the traits (Manolio et al. 2009). However, the performance on polygenic traits can be improved through combination of GWAS and multivariate analysis. A recent study showed that combination of several association studies is also promising in validating the same SNPs that have a significant impact on complex phenotypes in multiple GWAS (Müller et al. 2019). Müller et al. (2019) found rare and common associations in eight genes involved in cell wall biosynthesis and lignifica-tion by integrating independent results of single-SNP GWAS, regional herit-ability mapping (RHM), Whole-genome regression and Joint-GWAS set based models. It should be clear, however, that "Significantly associated" does not mean that the association identified by the study is necessarily of true significance in biology, medicine, or actual life. In order to identify and prove the causality of the detected association, we need to find the best asso-ciated variations by fine mapping around the GWAS signal, followed by functional studies.

Genotype-environment associations These methods try to associate the variation in allele frequencies at individu-al loci with variation at environmental variables such as temperature, pre-cipitation or altitude. Bayenv2 is based on a Bayesian generalized linear mixed model (Coop et al. 2010; Günther and Coop 2013) and has been used to identify loci strongly associated with one or more environmental varia-bles. To control for demographic history and gene flow among samples, a pairwise covariance matrix of control loci is included in the linear model relating allele frequencies to environmental variables. SNPs that are com-mon to both the tail of Bayes factor (which is the ratio of probability of the alternative and the null models) and the tail of Spearman’s rank correlation coefficient, rho, (calculated based on standard allele frequencies and less sensitive to nonlinear relationships) are often considered as the ones that are significantly associated with environmental variables. Such studies were carried out on large individual-based data sets in humans (Hancock et al. 2008), Arabidopsis thaliana (Hancock et al. 2011) and spruce (Chen et al. 2012, 2014). More recently, Bayenv2 was also applied to Pool-seq based data in non-model organisms (Fischer et al. 2013). The extension of statisti-cal models to incorporate Pool-seq data, in combination with dense popula-tion sampling across a wide range of environmental factors, holds great promise for identifying the genetic basis of adaptation in natural populations (Günther and Coop 2013). However if environment and population genetic structure strongly co-vary, teasing apart the effect of environment and popu-lation history will remain challenging. This often happens when populations colonize new areas and environmental gradients are parallel with the direc-tion of migration. In such cases, it is doubtful that improvement of statistical

20

methods alone will suffice to break the deadlock and a comparative ap-proach, combining species with different histories can be useful (Chen et al. 2014; Hodgins et al. 2016).

FST outliers As noted above, Wright’s fixation index, FST is a relative measure of the genetic differentiation among populations at specific locus caused by genetic drift. Loci under divergent selection are expected to have higher FST values among populations than neutral ones. Lewontin and Krakauer (1973) were first to come up with an FST outlier approach to test whether significant het-erogeneity between loci could reflect the effect of natural selection at given loci. Lewontin–Krakauer assumed that all populations were independent, but that’s not always the case in nature. Robertson (1975a,b) and Nei and Maryuyama (1975) showed that if populations are hierarchically structured the Lewontin–Krakauer method would lead to false positives (Excoffier et al. 2009). Later on, more FST outlier related approaches have been proposed using different demographic models to identify loci with extreme FST values for each allele frequency category (Beaumont and Nichols 1996; Beaumont and Balding 2004; Whitlock and Lotterhos 2015). One of the most widely used FST outlier programs is BayeScan v.2 (Foll and Gaggiotti 2008; Foll et al. 2010; Fischer et al. 2011). Two alternative models are defined for each locus, one including the effect of selection and another one considering only genetic drift. BayeScan calculates a posterior probability for each locus un-der each one of these two models and compare those to determine whether selection acts on the locus or whether it is more likely to have evolved neu-trally. BayeScan generally has a low power when neutral markers are highly differentiated in populations (De Kovel 2006; Butlin 2010) and BayeScan does not consider demographic history, which could increase the rate of false positives. In all cases, outlier detection approaches should be viewed as merely identifying candidate genes worthy of further, more in-depth studies.

Detecting local adaptation from clinal variation Clines can be generated by selection, gene flow (Mayr 1963; Ingvarsson et al. 2006) and genetic drift (Polechová and Barton 2011). Clines can reflect the degradation of gene flow due to adaptation to the local environment or genetic background, and detection of clines could help to estimate the strength of selection experienced by continuous natural populations (Savolainen et al. 2013). Forest trees are widely distributed and are exposed to a broad range of environmental conditions, and natural populations of those trees are often locally adapted and display pronounced geographic clines in phenotypic traits related to climatic adaptation even in the face of substantial gene flow (Le Corre and Kremer 2003; Neale and Kremer 2011;

21

Savolainen et al. 2013). Clinal variation in phenological traits, such as bud burst, bud set and growing days, is well documented at the phenotypic level in many long-lived perennial species. For example, latitudinal clines in growth cessation were identified in Picea (Chen et al. 2012, 2014), Populus (Ma et al. 2010; Wang et al. 2018) and Pinus (Notivol et al. 2007), with the number of growing days gradually decreasing toward higher latitude. In contrast, detecting clinal variation at the gene level turns out to be hard since pollen flow in forest trees is extensive, and has given rise in many cases to uniform neutral allele frequencies and limited linkage disequilibrium (LD). With respect to those issues, it is necessary to integrate comprehensive sam-pling, the right choice of statistical models and multilocus genetic markers to identify clines in forest trees (Le Corre and Kremer 2012).

Parallel evolution can provide even more noticeable evidence of local ad-aptation if the same genes are adapted to similar climatic variation in closely related species (Stern 2013). In spruce, Chen et al. (2012, 2014) studied two parallel latitudinal clines in Norway spruce in Scandinavia and in Siberian spruce along the Yenisei River. Along both clines allele frequency in photo-periodic (FLOWERING LOCUS T/TERMINAL FLOWER1-Like2: FTL2) and circadian clock (Gigantea: GI) genes, and expression of FTL2 varied gradu-ally with latitude. Siberian spruce (Picea obovata) is mainly distributed in Siberia, as well as Kazakhstan, Mongolia and Xinjiang province, Northwest China. Western Siberia was never fully glaciated and forests survived in both Western and in Central Siberia during Pleistocene and Holocene. Dur-ing the Late Pleistocene, western Siberia was a vast cold desert with relicts of arboreal vegetation in large river valleys. Microfossils indicate that coni-fer trees survived in sand dunes, mound valleys in western Siberia through the last glacial maximum (LGM) to the Holocene (Binney et al. 2009; Väliranta et al. 2011). Possibly as a consequence of colonization from local refugia, P. obovata has relatively weak population genetic structure over large areas in Siberia, which makes P. obovata an ideal model to study clinal variation. A South-North cline was identified in P. obovata populations along the Yenisei River, in which allele frequencies of two candidate genes FLOWERING LOCUS T/TERMINAL FLOWER1-Like2 (FTL2) and GIGAN-TEA (GI) varied with latitude (Chen et al. 2014). FTL2 is associated with growth cessation and its expression increase from south to north (Gyllenstrand et al. 2007; Chen et al. 2012, 2014; Karlgren et al. 2013). A parallel clinal pattern in the same candidate genes had already been reported in Norway spruce Scandinavian populations (Chen et al. 2012) that went through a very different history. FT-like genes integrate signals from differ-ent pathways controlling flowering time in plants (Fornara et al. 2010; Pin et al. 2010; Pin and Nilsson 2012) and FT promoter length changes through insertions and deletions to adapt to certain light length and temperature (Liu et al. 2014). Interestingly, Wang et al. (2018) recently found that expression of FLOWERING LOCUS T2 (FT2) also mediates local adaptation in the

22

timing of bud set in European aspen. Finally, Yeaman et al. (2016) compared signatures of local adaptation at the genome scale in two distantly related species, lodgepole pine and interior spruce, and found 47 genes that exhibit-ed signatures of clinal variation in temperature or cold hardiness in both lodgepole pine and interior spruce.

Gathering information from natural populations and breeding programs in Norway spruce

Figure 2. Illustration of Swedish Norway spruce breeding program.

Weather changes caused by climate warming and frequent fluctuations in temperature and rainfall can have strong effects on plant and animal distribu-tions and survival, especially at high latitudes (Walther et al. 2002; Root et al. 2003). In order to maintain biodiversity and keep ecosystem stability in the face of rapid climate change, Aitken and Whitlock (2013) suggested to use assisted gene flow. Assisted gene flow basically consists in transferring southern populations towards the north in order to facilitate more rapid adap-tation to climate change. As mentioned before, Sweden imported Norway spruce seeds from other European countries and planted them in Southern and central Sweden. Many of these recently transferred trees were included in the plus trees used to establish the current Norway spruce breeding popu-lation. The current Swedish breeding program was established around mid-dle-1970’s by the Swedish forestry institute (Skogforsk). It consists of 22 breeding populations across whole Sweden. For each sub-population, 50 parental/plus trees were selected from forest stands or plantations and each parent was crossed twice on average to obtain offspring (Figure 2). About 50 well-developed seedling progenies were selected from each of the 50 full-sib families and about 10-20 rooted cuttings per seedling ortet were then planted in four field-test sites. Therefore the total field-test size for one sub-

23

population is 50 full-sib families × 50 progenies = 2500 clones × 16 ramets = 40,000 test plants per cycle (Karlsson and Rosvall, 1993; Chen 2016). So the combination of extensive past translocations and progeny tests from the breeding program, where environmental variation is controlled for, offer us a great opportunity to study the genetic basis of phenotypic traits and to assess the capacity for adaptation to new climates. As some trees in the breeding population do not have reliable information on their origin, it is necessary to first identify it. This can be done by genotyping individuals from the breed-ing population simultaneously with samples from natural populations. Fur-thermore, we can combine information from natural populations and breed-ing populations to disentangle whether genetic factors or plasticity results in statistical differences in phenotypes.

24

The present thesis

The four articles in the present thesis form part of a long-term series of stud-ies on population genetics and local adaptation of the two main Eurasian Boreal spruce species, P. abies and P. obovata. The first three papers used a large exome capture dataset of trees from natural populations and the breed-ing program to understand population genetics structure and local adaptation in Norway spruce. In Paper I, we defined the genetic clusters and inferred the demographic history of Norway spruce. In Paper II, we used information from the breeding population to detect signatures of local adaptation for the seven Norway spruce genetic clusters defined in Paper I. In Paper III, we inferred fine-scale population structure, validated the migration paths of trees into Scandinavia and tested for local adaptation in trees with Swedish genetic origin. Finally in Paper IV, we studied latitudinal clines in Siberian spruce based on candidate genes identified in previous studies (Chen et al. 2012, 2014).

Paper I: Demographic history and translocation of Norway spruce The current distribution of Norway spruce comprises three main domains: a northern domain ranging from western Russia to Fennoscandia; an Alpine domain and a Carpathian one. The exact location of refugia during the Last Maximum Glacial and the routes of recolonization of spruce in Europe after glaciation have always been hard to pinpoint. Early studies based on iso-zymes and cytoplasmic markers identified two main refugia during Last Glacial Maximum (LGM): one in Russia and another in the Alps (Lagercrantz and Ryman, 1990; Tollefsrud et al. 2008, 2009). Later studies based on nuclear and cytoplasmic markers inferred more complex admix-tures in the northern domain (Chen et al. 2012; Tsuda et al. 2016). Phyloge-netic relations between northern and southern populations of P. abies, and P. obovata, however, are inconsistent according to different DNA markers (e.g.: Zou et al. 2013: mtDNA, Lockwood et al. 2013; Tsuda et al. 2016: nuclear DNA and SSR). Chen et al. (2016) used nuclear markers to estimate the speciation events between three main domains of Picea abies and indi-cated that the Fennoscandian domain diverged from Alpine and Carpathian

25

domains around 5 million years ago. All above-mentioned studies assumed that the sampled populations are of true local origin. However, seeds of P. abies were widely imported from central Europe, Belarus and Romania to southern Sweden during the 20th century which certainly would affect demo-graphic inferences if not properly accounted for (Myking et al. 2016).

Figure 3. The map of samples coordinates. Size of cycles corresponds to the number of individuals collected from the same location. Cycle shapes represent different spruce species.

In order to investigate the joint roles of demographic history and artificial gene flow in P. abies, we sampled 1,499 individuals that had been sampled to establish the Swedish spruce breeding program. 575 of these trees were lacking reliable records of geographic origin. We also sampled material from 173 individuals from natural populations of the three spruce species found in Europe, (P. abies: 74; P. obovata: 53; Hybrid individuals: 40; P. omorika: 6) (Figure 3). These trees from natural populations were used as references for genetic cluster definition and genotype assignment. 40,018 20-bp long probes were designed to cover 26,219 exons. After genomic DNA extrac-tion, library preparation and exome-capture sequencing were performed by RAPiD Genomics, USA. We mapped the raw reads to the P. abies genome reference v1.0 (Nystedt et al. 2013) using BWA-men (Li and Durbin 2010), performed SNP calling on GATK (McKenna et al. 2010) and recalibrated

40

45

50

55

60

65

700 10 20 30 40 50 60 70

P.omorikaP.abiesP.obovataHybrid-zone

26

variants according to Baison et al. (2018). 1,004,742 SNPs were retained in the end after quality control. We calculated population genetic diversity π0/π4, Tajima’s D (Tajima 1989) and pairwise population fixation index, FST, between species and between the three main P. abies domains (Fennoscandi-an, Alpine, and Carpathian). 399,801 intron and intergenic SNPs excluding significantly linked sites were used for inference of population structure and history. We first clustered all samples by Principle Component Analysis (PCA) using EIGENSOFT v6.1.4 (Galinsky et al. 2016). To understand the history of the clusters detected in the PCA, we then used unsupervised popu-lation clustering using ADMIXTURE v1.3 (Alexander et al. 2009). Samples without geographic origin records were clustered into seven ascertained ma-jor clusters based on their genotype similarity using “Random Forest” re-gression model. Finally, TreeMix v1.13 (Pickrell and Pritchard 2012) was used to assess the strength and direction of admixture events between spruce genetic clusters in western Europe. We inferred the demographic history of spruce species and of the three main domains of P. abies using multidimen-sional joint site frequency spectra (SFS) in Fastsimcoal2 v2.6.02 (Excoffier et al. 2013). Only fixed polymorphic sites between target populations and outgroup P. omorika were applied. To limit the number of parameters of already complex demographic models, a constant migration rate m= 1 × 10−6 was assumed, but we also tested m= 0.

Figure 4. Population genetic structure of Picea abies, Picea obovata and Picea omorika from ADMIXTURE analysis. Colors represent the inferred ancestral popu-lations when K=5. Black vertical lines delineate different populations and text colors correspond to genetic clusters except for the Fennoscandia. Text letters are abbrevia-tions of countries. RU: Russian; BY: Belarus; EE: Estonia; LV: Latvia; LT: Lithua-nia; DE: Germany; CH: Switzerland; DK: Denmark; SK: Slovakia; CZ: Cze-republic; SPL: Southern Poland; NPL: Northern Poland; RO: Romania; SE: Sweden and FI: Finland.

27

Besides the three major genetic clusters (Alpine, Carpathian, and Fen-noscandian) in P. abies, extensive gene flow from P. obovata was identified as in previous studies (Tsuda et al. 2016; Fagernas 2017). More frequent admixture events were also detected, especially between the Alpine and Car-pathian domains and across the Baltic-Scandinavian regions (Figure 4). We found that spruce recolonized glacial areas after the LGM not only from the east as shown in early studies, but also from south to north, in agreement with paleoecological studies on the vegetation in Poland (Kupryjanowicz et al. 2018). “Unknown” trees could be assigned to the Baltic and central Euro-pean clusters that were defined based on PCA. Most likely, genotypes with an important Alpine or Carpathian component reflect recent translocations from central Europe that took place during the twentieth century massive reforestation efforts. This is consistent with historical records (Myking 2016). Hence genomic polymorphisms could serve as an efficient way to trace the origins of translocations and our study also provides the basic ground for investigating the genetic adaptive effects of translocation, which is of great importance for tree breeding, conservation, and forest manage-ment under climate change.

Figure 5. Demographic history of three spruce species: Picea abies, Picea obovata and Picea omorika. Black dotted lines indicate migrations and blue lines represent admixture events. Numbers on the bars are estimated effective population sizes.

Demographic inferences retrieved the same main clusters within P. abies than previous studies. A northern domain where the species is progressively replaced by Siberian spruce (P. obovata) and two smaller domains, an Al-pine domain and a Carpathian one, but also revealed further subdivision, bottleneck during LGM and gene flow among clusters (Figure 5). The di-vergence of the three main domains happened around 15 mya which is much

12,850 ya

15.3 mya

17.6 mya

22.9 mya

P.omorika P.obovata Hybrid P.abies (FEN) P.abies (ALP) P.abies (CARP)

78

250,000

35,500

1,455,500

390 7,500

750,000

6,000

600,000

8,000

800,000

400,000

380,000

168,000

700,000

127,500

200 190

m=1e-6

28

older than 5-6 mya from previous studies (Zou et al. 2013; Tsuda et al. 2016). One of the possible explanations could be that previous studies used admixed samples in demographic inferences and relative recent admixture events between the three major domains could significantly reduce split time. P. omorika diverged around 23 mya and P. abies and P. obovata split-ted around 18 mya, which is consistent with a divergence of the Eurasian spruce clade around the early to middle Miocene (13 ~ 23 mya), as inferred from molecular clock dating methods (Leslie et al., 2012; Lockwood et al., 2013). For P. omorika, a positive Tajima’s D and relatively high π0/π4 indi-cates that a strong bottleneck happened in the recent past and demographic analysis showed a recent and sudden contraction of its population size which corresponds to the large decrease of forests resources during the Iron Age in southwestern Bulgaria (Marinova et al. 2012) and could also be caused by contamination of pathogen due to global warming (Ivetić et al. 2016).

Paper II: Local adaptation and genetic basic of quantitative traits in spruce Climate warming could threaten ecosystems and economic benefits of for-ests through direct impacts on forest trees or damage caused by insect and fungus. Human assisted gene flow is a potential strategy to pre-adapt to fu-ture extreme climate and has been proposed as a key forest management strategy to maintain forests health and productivity (Aitken and Whitlock 2013; Aitken and Bemmels 2016). Since the 1950s Sweden has imported vast amount of Norway spruce seeds from across the natural range of the species (Myking et al. 2016; Jansen et al. 2017) for reforestation. The Swe-dish breeding program was established gradually by ‘plus’ trees selection from forest stands across Sweden. Offsprings of these plus trees were later used in progeny trials installed across south and central Sweden in order to estimate breeding values of parent trees for interesting phenotypic traits. For each trial, the number of tested parents varied from 11 to 1395 and the num-ber of replicates from 11 to 40. We recently showed that the current South-ern Swedish breeding populations includes individuals originating from the seven P. abies genetic clusters (Fennoscandia, Alpine, Carpathian; Central Sweden, Central Europe, Russian-Baltic and Northern Poland) (Chen et al. 2019). In this study, we took advantage of past transfers to infer local adap-tation of trees currently growing in southern Sweden and investigate the genetic architecture of quantitative traits.

29

Figure 6. Trees original locations. Black squares are Norway spruce progeny test belonging to the Swedish breeding program. ‘Plus’ and ‘multiply’ signs are tree sampling locations, and the latter indicating a wrong assignation in available rec-ords. Discs are the centroid of the geographical coordinates of trees belonging to a same genetic cluster and having a known origin. The grey arrow (right panel) indi-cates the location of the trial ‘Ekebo’, where bud burst was characterized. Colors correspond to genetic clusters (Carpathian, dark blue, ROM; Alpine, light blue, ALP; Central Europe, green, CEU; Northern Poland, yellow, NPL; Russia-Baltic, orange, Rus_Bal; central and southern Sweden, red, CSE; Fennoscandian, pink, NFE).

In order to test for local adaptation and genetic architecture of quantitative traits, we used 1545 Norway spruce trees from Chen et al. (Chen et al. 2019), of which 70 individuals were collected from natural populations in three main domains: Fennoscandia, Alpine and Carpathian and the rest are plus trees with superior phenotypes from breeding populations (Figure 6). 560 of plus trees lacked geographic origins records. To estimate breeding values of parent trees and perform backward selection, two growth traits (height, diameter) and bud-burst were observed from 763, 808 and 834 plus trees respectively across progeny trials established between 1978 and 1998 by Skogforsk (Figure 6). 712 of those plus trees have all three-phenotype data, but only 279 of them had records of genetic origin. Height was collect-ed from 6 to 15 years old trees and diameter from 9 to 15 years old trees. Bud-burst was measured using Krutzsch scale (Krutzsch 1973), varying from 0 (budset) to 9 (full development of the needles). The genotypic dataset was defined by Chen et al. (Chen et al. 2019). In total, 917,107 SNPs were identified for P. abies trees, 399,801 of which were unlinked noncoding SNPs. EIGENSOFT V6.1.4 (Galinsky et al. 2016) was used to perform prin-cipal component analysis (PCA) to define genetic clusters for P. abies based on unlinked noncoding SNPs. The top five principle components were used to predict the genetic origins of the trees lacking records of geographical origin using machine learning ‘random forest’ implemented in R (‘random-Forest’ v4.6-14 package, Liaw and Wiener 2002, R software v.3.3.1, R Core Team 2019).

30

Parent breeding values (BVs) of two growth traits were computed by mixed linear model and BLUPs (best linear unbiased predictors). BVs were reported as relative percentage of the average of all observations. Bud-burst observations were transformed to normal scores based on the method in Danell (1991). Missing phenotypes were predicted by ‘genomic selection’ implemented in TASSEL v.5.2.38 (Bradbury et al. 2007; Zhang et al. 2010). To estimate the level of local adaptation, we combined information from phenotype, ancestral environment and genotype. The relationship between phenotype and ancestral environment relationship was estimated through generalized linear model (GLM). 19 ancestral environmental variables were downloaded from WorldClim v.2.0 with 10-minute resolution (http://worldclim.org/, Fick and Hijmans 2017) and three more variables were derived from these 19 climate variables. For trees without origin coor-dinates, bioclimatic data were taken from the centroid of genetic cluster to which they belong (Figure 6). The association between genotype and envi-ronment was performed by Bayenv2 (Coop et al. 2010; Günther and Coop 2013) for 48 populations and both Bayes factor (BF) and Spearman’s rho were used to filter SNPs strongly correlated with environmental variables. Genotype-phenotype association was computed by compressed mixed linear model (Zhang et al. 2010) implemented in the R package GAPIT. The first three principal components of genotypes were used to correct for population structure and kinship (individuals’ relatedness) using TASSEL software v.5.2.38 (Bradbury et al. 2007). An empirical Bayes approach for adaptive shrinkage developed by Stephens (2017) was used to investigate statistical significance of the SNPs associated to the corresponding phenotypes. Final-ly, gene function and enrichment test for statistically significant SNPs were inferred using the ‘top GO’ R package (Alexa and Rahnenfuhrer 2010).

Figure 7. Influence of trees origin on phenotype. Diameter, height (breeding values, BVs) and bud-burst (normal scores) values are represented for the different genetic clusters (Carpathian, dark blue, ROM; Alpine, light blue, ALP; Central Europe, green, CEU; Northern Poland, yellow, NPL; Russia Baltic, orange, Rus_Bal; central and southern Sweden, red, CSE; Fennoscandian, pink, NFE). The genetic clusters are ordered regarding latitude for height and diameter and given longitude for budburst. The number of trees belonging to each genetic cluster is given within paren-theses. Letters represent the levels of significance.

31

Comparisons of height, diameter and bud-burst for trees with different genet-ic origins indicated that all three traits differ among the seven P. abies genet-ic clusters (Figure 7). Height and diameter increased from high to low lati-tude (Person’s r = -0.47 and -0.62, respectively, all p < 0.001) and bud-burst varies along both latitude (South to North, r = 0.39, p < 0.001) and longitude (West to East, r = −0.47, p < 0 .001). These results indicate that phenotypic performance of Norway spruce can be well predicted by ancestral environ-ment, with southern genotypes outcompeting local trees for growth traits in southern Sweden. PCA for phenotypes, genotypes and climate data at tree origins presented similar clustering patterns. Moreover, the principle com-ponent 1 (PC1) or principle component 2 (PC2) of the three PCAs were highly correlated (Figure 8). As expected, the strongest correlation was found between genotype and climate data since P. abies tends to present a relatively clear population structure for populations under heterogeneous habitats. The association between genotype and phenotype indicated that growth traits and bud-burst are genetically controlled, a result consistent with Lagercrantz and Ryman (1990). Demographic history had a strong im-pact on the divergence of phenotypes. The association between phenotypic traits and environment data reflected that trees are locally adapted to their home environment. Overall our result agrees with previous suggestions that local adaptation is common in forest trees despite extensive gene flow (Chen et al. 2012, 2014; Avia et al. 2014; Yeaman et al. 2016; Li et al. 2019).

Figure 8. Spearman's correlation coefficient, r, between principle component 1 (Comp 1) or principle component 2 (Comp 2) of principle component analysis (PCA) for genotypes, phenotypes and climate variables of each genetic clusters. The size of the circles corresponds to correlation coefficient.

Bayenv2 identified many statistically significant SNPs associated with cli-matic variables even though population structure was obviously over-corrected due to the strong correlation between genotypes and environment variables. Briefly speaking, more transcripts were associated with precipita-tion-related variables than with temperature-related variables or moisture.

32

The largest group consists of transcripts associated to photoperiod. We also found that climate variation produced a widespread selective pressure across the Norway spruce genome because candidate genes displayed very low overlap. To characterize the genetic architecture of phenotypic traits, we performed GWAS and found 180, 175 and 32 SNPs with significant allelic effects on diameter, height and bud-burst, respectively. Furthermore, we found that height and diameter were highly correlated (Spearman’s = 0.86) which may be due to pleiotropic effects at the genotypic level (ρ = 0.91 and 0.93, when considering SNPs significant for height or SNPs significant for diameter, respectively). GWAS without population structure correction iden-tified a much larger number of statistically significant SNPs for growth traits than after correction, which means that our estimates of ~180 SNPs affecting growth traits is likely to be conservative and that the two growth traits are highly polygenic. Similar results were reported for other quantitative traits in model species (Daub et al. 2013; Berg et al. 2017; Boyle et al. 2017).

Overall, our study identified a strong pattern of local adaptation in Nor-way spruce, tree origins strongly affect phenotypic traits and growth traits are high polygenic traits. Our study therefore demonstrates that breeding programs are valuable for large-scale genomic studies.

Paper III: Fine-scale population genetic structure and local adaptation in Norway spruce across Sweden Population genetic structure is ubiquitous in natural populations (Barton et al. 2019). Accurate estimation of current population genetic structure is criti-cal to properly infer speciation processes, local adaptation and historic events. Population structure also limits the power of association studies, where it can lead to a large number of false positives if not accounted for. Numerous approaches have been designed to capture fine-scaled population genetic structure and the availability of genome-wide polymorphisms defi-nitely increases our ability to detect population genetic structure. However, it is still challenging to detect fine-scale population genetic structure especially when genetic diversity changes both continuously and discretely (Bradburd et al. 2018; Diaz-Papkovich et al. 2019).

For Norway spruce (Picea abies), previous studies indicated that current populations are structured into three main genetic groups: Alps, Carpathians and Fennoscandia-Russia (Bucci and Vendramin 2000; Vendramin et al. 2000; Heuertz et al. 2006; Tollefsrud et al. 2008, 2009, 2015; Chen et al. 2012; Tsuda et al. 2016; Fagernas 2017). To these three basic groups recent studies added four additional groups that correspond to admixture between the three main lineages: Central Sweden, Central Europe, Russian-Baltic and Northern Poland (Chen et al. 2019). These studies also identified important

33

gene flow from Siberian spruce, extending as far as Scandinavia (Tsuda et al. 2016; Fagernas 2017; Chen et al. 2019). Moreover, introgression from Siberian spruce into Norway spruce was more pronounced at high latitudes (> 65°N) than at intermediate ones (around 60°N) (Chen et al. 2019). Recent translocation also had a strong impact on current population genetic struc-ture. For instance, Norway spruce populations used to establish the Swedish breeding programs are highly admixed (Chen et al. 2019; Milesi et al. 2019).

In the current study, we first focused on analyzing population genetic structure of Norway spruce populations across Sweden using exome capture data from 4769 trees sampled along a latitudinal gradient extending from southern Sweden to its very north. In particular, we aimed to identify barri-ers to gene flow and tested whether those reflect physical barriers or simply historical contingencies. Finally, we inferred association between genetic data and environment variables.

Figure 9. The map of samples location. Points are sampling coordinates and their shapes represent different species. Point colors correspond to genetic clusters infer-ring from PCA analysis. P. obo: Picea obovata; Kirov and Indigo are two popula-tions collected from hybrid zone; seven Picea abies genetic clusters: ALP: Alpine; CEU: Central Europe; CSE: Central and Southern Sweden; NFE: Fennoscandian; NPL: Northern Poland; CAR: Carpathian; Rus_Bal: Russian-Baltic. The distribution range of Picea abies and Picea obovata was downloaded from EUFORGEN (www.euforgen.org) and Lockwood et al. (2013) respectively.

P. oboIndigoKirovALPCEUCSENFENPLCARRus_Bal

0 20 40 60 80

4045

5055

6065

70

N

34

We used 4769 spruce trees in total, 1523 of which were also used in Chen et al. (2019) containing Picea abies (P. abies), Picea obovata (P. obovata) and including 162 samples collected from natural distribution, and 3246 addi-tional P. abies that were only used in this study and were collected from the breeding program across Sweden (Figure 9). We extracted DNA from either buds or needles and then sent DNA to RAPiD Genomics who prepared li-braries and captured 26,219 P. abies genes using 40,018 diploid probes (Vidalis et al. 2018). After SNP calling and stringent quality control, 4508 individuals were kept and a total of 504,110 SNPs were identified. Of these SNPs, 205,337 fell in introns and 108,300 in intergenic regions. Of the SNPs in exons, 120,808 are synonymous variants and 69,655 are nonsynonymous variants. After removing highly linked SNPs (r2 > 0.2) in introns and inter-genic regions by PLINK v.1.9 (Chang et al. 2015), 155,211 SNPs were kept for further genetic clustering analysis. To visualize the population genetic structure in Norway spruce, we started with principal component analysis (PCA) implemented in EIGENSOFT v7.2.0 (Galinsky et al. 2016) and then performed ADMIXTURE v1.3 (Alexander et al. 2009) with 10-fold cross-validation and 200 bootstraps. Trees without clearly identified geographical origin were assigned one by applying a “Random Forest” algorithm (Liaw and Wiener 2002; R Core Team 2019) to the top five principle components of the PCA. Both PCA and ADMIXTURE divide populations into discrete units, however, genetic diversity of Swedish P. abies is continuously distrib-uted across Sweden. To account for the mixture of continuous and discrete distribution of genetic variation we ran the recently developed software con-Struct to estimate the population genetic structure. conStruct (Bradburd et al. 2018) combines model-based clustering algorithms with an isolation by dis-tance (IBD) model. 1758 geo-referenced trees originating from Sweden were used and assigned to 47 populations. 113,748 unlinked SNPs from introns and intergenic regions were used after deleting singletons. Cross-validation and layer contributions were used to choose the best fitting model between spatial and non-spatial models and to choose the ‘best’ number of layers, respectively. To identify gene flow barriers or corridors, Effective Migration Surfaces (EEMS) was performed (Petkova et al. 2015) based on the same data set as the conStruct. EEMS is a tool to visualizing spatial population structure. It assumes an isolation by distance model with isotropic migration and plots the extent of departure from the model. For example, barriers to gene flow will be characterized by a negative effective migration surface. EEMS was run for four different deme numbers (30, 50, 80 and 100) and the combination over four different deme numbers was taken to plot the migra-tion surface by “rEEMSplots”. We also tested if the IBD model can explain the genetic differential by regressing FST on geographic distance.

To estimate associations between genotypes and environmental variables, we ran Bayenv2 (Coop et al. 2010; Günther and Coop 2013) on the same data set as the one used for the conStruct analysis excluding loci with minor

35

allele frequency (MAF) < 0.05. Twenty-four environmental variables were used, of which 19 were downloaded from WorldClim v.2 (http://worldclim.org/version2) with 10-minutes resolution and 3 were de-rived from those 19 variables, one is the latitude and the last one is the longi-tude. SNPs with BF > 150 or with BF > 20 but still within the 0.1% highest BF, and also within the 1% highest absolute Spearman’s rho, were consid-ered as candidate SNPs (Milesi et al. 2019) and we analyzed related gene function for those candidate SNPs using the ‘top GO’ R package (Alexa and Rahnenfuhrer 2010).

The PCA grouped individuals into seven genetic clusters corresponding to Alpine, Carpathians and Fennoscandia, Central Sweden, Central Europe, Russian-Baltic and Northern Poland domains. This is consistent with previ-ous studies (Chen et al. 2019; Milesi et al. 2019). For Fennoscandia and Central Sweden clusters, all top three principle components presented a broad-spectrum of variation and all of them were associated with latitude between 60°N and 63°N. ADMIXTURE indicated that the ‘best’ number of ancestral populations was K = 6. P. abies trees from Sweden were divided into, two different genetic clusters, basically one in the north and one in the south, separated by a contact zone in Central Sweden. It was interesting that there was also a small contribution from both P. obovata and the Carpathi-ans. For conStruct performance, the spatial model fits the data set better and three layers could explain most of the variation. As for Admixture Northern trees with Swedish origin were distinguished from southern ones (Figure 10a) and an admixture zone was observed between 60ºN and 63ºN. Further-more, we also discovered a small portion of unique admixture component in the northernmost trees. EEMS analysis identified two regions with low ef-fective migration, one in northern Sweden around 65°N and the other in central Sweden between 58.5°N and 63°N (Figure 10b). The contact zone identified in our case was also observed in many other species, for instance in Populus tremula L. (De Carvalho et al. 2010); brown bears (Bray et al. 2013) and rodents (Jaarola et al. 1999). It is generally explained as the con-tact zone between northern and southern recolonizing routes after the last ice ages.

36

Figure 10. a): The estimation of admixture proportions for trees of Swedish origins based on spatial model in conStruct for three layers K = 3. The colors correspond to different layers. b): Estimated effective migration surfaces (EEMS) analysis of Swe-dish originating trees. The plot shows the posterior mean effective migration rates across the deme sizes 30, 50, 80 and 100. Blue area indicates effective migration rate between populations are higher than expected ones under isolation by distance (IBD) model while orange area means lower migration rate than average. If the effective migration rate closes to the expected one, it is indicated by white color.

For Norway spruce in Scandinavia, both genetic and fossil pollen studies have shown two re-colonized paths after the last glacial maximum, one from northern Russia across Finland (Giesecke and Bennett 2004) and the other from central Russia across the Baltic sea (Tollefsrud et al. 2008, 2009, 2015;

37

Chen et al. 2019). Small contribution from P. obovota presented in northern Norway spruce trees may also reflect the northern migration from P. obovata and also indicated that the contribution of P. obovata to the northern Swe-dish P. abies trees is larger than assumed so far. The IBD tests through re-gressing FST against logarithm geographic distance indicated that genetic variation follows the IBD patter and also found less dispersal individuals along latitudinal gradient (dispersalLat = 228 ± 33) than along longitudinal gradient (dispersalLong = 680 ± 187), which suggested that genetic barriers identified by EEMS were likely artefacts.

Bayenv2 showed that many SNPs were strongly associated with climate variables even after over-correcting population genetic structure because the population genetic structure correlated with environmental variables. The pattern of isolation-by-distance at the SNP identified by Bayence2 was much more pronounced than for random SNPs suggesting the presence of local adaptation.

Paper IV: Inference of the cline in Siberian spruce populations along Ob River It should, a priori, be relatively difficult to detect local adaptation in forest trees due to large effective population size, long generation time and exten-sive gene flow (González-Martínez et al. 2002; Chambel et al. 2007; Leimu and Fischer 2008; Hancock et al. 2011; Ågren and Schemske 2012). Recent studies have indicated clinal variation in forest trees at both phenotypic and genotypic levels at different geographic scales (Ma et al. 2010; Chen et al. 2012, 2014; Yeaman et al. 2016). Furthermore, the existence of population structure could complicate the detection of local adaptation. Not accounting properly for population genetic structure can increase significantly the false discovery rate. Although many methods were developed to correct popula-tion structure, for instance Bayenv2, population genetic structure may still lead to false negatives if population structure co-varies with the phenotypic cline. In such cases, to validate the signatures of local adaptation, it is essen-tial to replicate studies and use parallel clines in the same species that expe-rienced different historic events (Chen et al. 2012, 2014), or clines in differ-ent species (Yeaman et al. 2016). Previous studies found latitudinal clines in growth cessation, allele frequency at photoperiodic (FLOWERING LOCUS T/TERMINAL FLOWER1-Like2: FTL2) and circadian clock (GIGANTEA: GI) genes and the expression of FTL2 in both Norway spruce (Picea abies) populations in Scandinavia (Chen et al. 2012) and in Siberian spruce (Picea obovata) populations along the Yenisei River (Chen et al. 2014). Since the steepest change in relative growing length in Yenisei River populations was observed between the northernmost populations (Chen et al. 2014) one of the

38

aims of the present study was to test for the existence of a cline in growth cessation in populations along the Ob River considering a denser set of pop-ulations at high latitude (> 60°N).We also checked whether the clinal varia-tion in allele frequency at FTL2 and GI, as well as in gene expression of FTL2 also existed in the Ob River populations that are located in the hybrid zone between Picea abies (P. abies) and Picea obovata (P. obovata). We also tested for selection signals in GI by using four times longer GI frag-ments than in our previous studies (Chen et al. 2012, 2014) which failed to detect direct evidence of selection due to limited number of synonymous and nonsynonymous sites along the part of GI that was sequenced.

Table 1. Population locations of Picea obovata along the Ob River. The collection sites span a gradient of c. 9º.

Location Abbreviation Latitude [° N] Longitude [° E] N1 N2

Tobolsk TOB-58 58.17 68.37 12 49 Kanty-Mansyik XM-61 61.05 69.25 18 51 Oktoberskaya OKT-62 62.45 66.09 20 71 Berezovo BER-64 63.93 65.03 5 17 Kamys Mys KAZ-65 64.70 65.61 9 6 Pytier PUT-66 65.91 65.91 11 19 Krasnij Kamenj KK-67 66.91 65.76 12 7 Total 87 220 Note: N1 is the number of individuals sequenced in a population; N2 is the number of seed-lings used to seedling height measurement.

We collected P. obovata seeds from seven populations along the Ob River, between 58°N and 67°N (Table 1), and germinated healthy seeds from dif-ferent mother trees of each population. For each population, 6–72 seedlings at 20-needle stage from 3 to 15 maternal trees were transplanted in a growth chamber under a temperature of 18°C, 54% humidity, and continuous light for 8 weeks. Thereafter those seedlings were exposed to 22-h light for one week, followed by periods where daylength was shortened by 1.5-h per week until the daylength reached 14.5-h. The last photoperiod lasted for two weeks and the whole photoperiod treatments continued for 8 weeks, in total. For seedling height measurement, we started with twice a week before the different photoperiod treatments and once a week afterwards. Growth cessa-tion was defined when the weekly height changes are smaller than 5% of the total seedling height. Relative growing days, the total number of growing days divided by the total number of days of the whole photoperiod process, were used in the clinal variation analysis in growth cessation and related comparisons between the Ob River populations and the Yenisei River popu-lations. FTL2 expression levels were measured at 9:00 and 17:00 during the last 24 hours of each photoperiod and the average was taken at the two time points. Furthermore, gene expression measurement was repeated once at each time point. To compare the expression levels, for each population, the

39

relative expression level was defined as the average FTL2 expression value of two replicates under each photoperiod minus the value of the southern-most population (TOB-58).

We extracted DNA from 87 samples of seven P. obovata populations and sequenced two candidate genes FTL2, GI and 14 control loci (Can8a, Can12, Can14, Can28, Can31, Can32, Can33, Can37, Can49, Can56, Can58, Can59, Can60, and Can62). Control loci were randomly selected from the genes with irrelevant functions (Pavy et al. 2012). For each gene, the pairwise nucleotide diversity (π), Tajima’s D and Watterson’s mutation rate (θw) were estimated at both coding and noncoding loci. Linkage disequi-librium (LD) within a gene were estimated and SNPs were considered as strongly linked if r2 ≥ 0.25 and Bonferroni-corrected p value ≤0.05. The overall decay of LD with physical distance within genes was regressed by nonlinear model (Remington et al. 2001). Population structure was per-formed by STRUCTURE V2.3.4 (Pritchard et al. 2000). To identify SNPs associated with latitude, we performed linear regression and Bayenv2. Con-trol SNPs were used to build an empirical distribution in the tail enrichment test of candidate SNPs. Bayenv2 was also run to test the association between FTL2 expression and allele frequencies. FTL2 expression in each light peri-od and population were standardized. To test whether the observed latitudi-nal variation was caused by diversifying selection, we performed an FST outlier test using the program BayeScan v. 2.1 (Foll and Gaggiotti 2008; Fischer et al. 2011). We also applied a McDonald-Kreitman neutrality test (McDonald and Kreitman 1991) to GI and Picea breweriana was used as an outgroup. Neutrality index (NI) was compared with 1, as well as with the NI for control loci.

As in previous studies (Chen et al. 2012, 2014), relative growing days of the Ob River populations decreased as latitude increased and expression of FTL2 under each photoperiod increased from south to north as well as the expression levels increased when the light-length shorten from 24h to 14.5h (Figure 11). These results, together with transformation experiments (Karlgren et al. 2013), once again confirm that FTL2 expression affects growth cessation in trees (Wang et al. 2018). Clinal variation at SNPs in both GI and FTL2 was much weaker than that in the case of the Scandinavian or Yenisei River clines. In the current study, allele frequencies at twenty SNPs at two candidate genes were significantly correlated with latitude and the enrichment of candidate SNPs was detected at both 10% and 5% tails, but the selection signal disappeared in Bayenv2 analysis after correcting popula-tion structure and no outliers were identified in the FST outlier test. Weaker clinal variation at candidate genes along the Ob River populations may be caused by the more limited number of candidate genes and samples than in the two previous studies (Chen et al. 2012, 2014). Alternatively, natural se-lection could affect different paths to establish local adaptation since growth cessation is a quantitative trait controlled by a large number of loci (Milesi et

40

al. 2019). In addition, a large role of other photoperiodic candidate genes related to growth cessation could weaken the effect of selection on FTL2 and GI in Ob River populations. The weak clinal variation was unlikely to have been caused by the inclusion of higher latitude populations since the growth cessation pattern was actually more pronounced than along the Yenisei Riv-er. Similarly, the hybrid zone should be more ancient than the cline in Scan-dinavia, so the age of the Ob River cline might not be either the cause of the weak cline. The low level of linkage disequilibrium, similar to the ones in other studies (Beaulieu et al. 2011; Chen et al. 2012, 2014) further suggest that the hybrid zone is rather ancient (Tsuda et al. 2016).

Figure 11. Relative number of growing days for Siberian spruce populations along the Yenisei River (Chen et al. 2014) and Ob River, estimated in growth chamber experiments. Populations along the Ob River are represented by blue dots, and popu-lations along the Yenisei River are by red ones. Vertical bars represent the standard deviation for each measurement. The dark gray regions show the 95% confidence interval for the relative growing days of each population.

The neutrality index (NI) in GI was 2.33, higher than 1 and than the one in the concatenated fourteen control loci, 2.03, and the polymorphism ratio contributed more than the divergence ratio to the increased NI in GI. There-fore, our study suggests the presence of selection in GI although we cannot be more specific on the type of selection acting on GI. Recurrent hitchhiking selective sweeps are a possible explanation according to the evidence found

41

in GIGANTEA GIA and GIB in Populus tremula by Hall et al. (2011). Se-quencing the whole GI gene across a number of species and considering both polymorphism and divergence, would be necessary to better understand the pattern of selection on GI.

42

Conclusions and future perspectives

With the release of the Norway spruce (P. abies) genome in 2013 and the development of re-sequencing technologies, it is today affordable to capture whole-exome sequences for thousands of trees. The combination of exten-sive genomic polymorphism data with data from the Norway spruce breed-ing program provided a unique opportunity to investigate demographic histo-ry and local adaptation in Norway spruce. The Swedish spruce breeding program is particularly interesting to study local adaptation since a lot of the individuals used to establish it originated from other parts of the natural range. More generally, the present thesis illustrates how much information is available within forest tree breeding programs and how this information can be used to address both applied and fundamental questions on the evolution of forest trees, especially when combined with studies in natural populations. Below I will briefly summarize the main findings and outline how we could extend the study.

In paper I, the analysis of joint site frequency spectra across populations allowed us to test alternative demographic scenarios and unravel the subtle genetic admixture between Picea obovata and the three main P. abies do-mains. The current P. abies populations are the result of complex population movements (from East to West and from South to North) and admixture events with at least three major ancestral groups (Alpine, Carpathian and Fennoscandia, the latter being strongly influenced by P. obovata) and four derived ones (Central Europe, Central Sweden, Russian-Baltic and Northern Poland). Furthermore, we confirmed that breeding populations in southern Sweden are highly admixed and correspond to recent translocations, mainly introductions from Central Europe. From a practical point of view an im-portant result is that genomic polymorphisms can be an efficient way to pre-dict the origins of translocations based on genetic similarities. Regarding the demographic history of P. abies there are two aspects that would warrant further studies. First, in all studies carried out hitherto, sampling in North-west Russia has been limited or absent. This geographical area likely played a crucial role in the recolonization of Scandinavia and additional sampling would be crucial, especially if combined with ancient DNA studies. Second, the evidence of an eastward expansion of P. abies from the Carpathians needs to be strengthened.

After assigning individuals used to establish the Norway spruce breeding program to different genetic clusters, we focused on local adaptation by tak-

43

ing advantage of the phenotypic data available in the breeding program (pa-per II). PCAs from (i) environmental data from the geographical areas of origin of the trees, (ii) genomic data, and (iii) phenotypic traits (bud-burst, height and diameter) were highly congruent indicating a strong pattern of local adaptation. Genome-wide association studies (GWAS) showed that both growth traits (height and diameter) are more polygenic than bud-burst. Trees from the Carpathians grew faster than local ones in southern Sweden suggesting that assisted gene flow could be a viable strategy to alleviate the effects of climate change. An obvious limitation of that conclusion is that it is based on only three phenotypic traits. It would be important in the future to add additional traits, in particular traits that might be more directly associ-ated to adaptation to climate change and to fitness. Adding traits could also help to limit the amount of “false negatives” when correcting for population genetic structure in GWAS. In the present case, two highly correlated traits, height and diameter, varied along a latitudinal cline and one, bud burst, along a longitudinal cline. The former was consequently much more strongly affected by correction for population genetic structure than the latter. Having more traits showing different geographical variation pattern would help to quantify the impact of the correction for population structure and develop approaches to limit the number of false negatives.

Papers I and II focused on the southern part of the Swedish spruce breed-ing populations. In paper III we analyzed the population genetic structure of 4769 trees covering the whole country. To characterize population genetic structure and capture the different processes contributing to it, i.e. population movements, contact zone and isolation-by-distance, we used multiple ap-proaches (PCA, ADMIXTURE, conStruct and EEMS). Swedish trees were clustered into two main genetic groups with an introgression zone in central Sweden. A third, smaller cluster was also detected in the far north that re-flects a more important contribution from Siberian spruce (P. obovata). The introgression zone between the two main groups reflects the meeting point of the two main waves of recolonization of Scandinavia after the Last Glacial Maximum. The variation at a large number of SNPs was associated with environmental variables and exhibited a stronger pattern of Isolation-by-distance than average SNPs suggesting that natural selection played an im-portant role during the establishment of the population. A natural extension would be to assess the association with phenotypic traits. As in paper II, one limitation will be the low number of traits measured in the breeding program and the fact that, while these traits are certainly relevant for breeding, they may be less so for recolonization processes or adaptation to climate change. Also, we did not test for signature of selection along the genome and, now that we have a fairly good grasp of the population genetic structure of the Swedish populations this would also be a very natural extension.

44

Finally, identification of parallel clines in the same species or in different species can help to detect or confirm local adaptation in quantitative traits and underlying genes. In trees two major candidate genes for phenology are FTL2 and Gigantea (GI). In paper IV, we detected a latitudinal cline in bud set and FTL2 expression in Siberian spruce populations along the Ob River. This corroborates results previously obtained in Norway and Siberian spruc-es along other latitudinal clines. It was, however, difficult to detect evidence of natural selection in FTL2 and GI, especially in the latter. Given the key roles played by these two genes in plant development and phenology it would certainly be worth understanding their response to selection better. In the case of Gigantea progress has been hindered by the lack of complete sequences of the gene across species and it would likely be very rewarding to obtain such data.

45

Summary in Swedish (Sammanfattning)

I kärna för alla studier inom populationsgenetik finns DNA. DNA är livets kod som finns inom alla levande organismer. Det är cellens ritning. Det som definierar cellens egenskaper och slutligen bestämmer organismens form och funktion.

I min avhandling tittar jag närmare på granens DNA. Närmare bestämt hur variationen i DNA mellan granpopulationer relaterar till deras anpass-ningsförmåga för kommande klimatförändringar. Vad skiljer en gran i Ki-runa från en gran i norra Italien? Är granen i Italien tåligare mot torka? Vilka gener ligger bakom dessa skillnader?

Ett område nära kopplat till dessa frågor är fenologi, vilket är forskningen av hur årstider interagera med ekosystemen och arterna därinom. Om vi plantera en italiensk gran i en svenska skog kommer den slå skott samtidigt som de svenska granarna? Kan vi korsa granar från olika klimatzoner för att få fram träd som är bättre lämpade för att växa i ett framtida klimat? Vilka gener är avgörande och vilka populationer borde vi plantera tillsammans? Frågor som dessa är ligger till grunden för mina studier och är fokuset i min avhandling.

En viktig förutsättning för avhandlingen var publiceringen av granens ge-nom under 2013 och utvecklingen av sekvenseringsteknologier vilket gjorde det ekonomiskt möjligt att mäta hela exomsekvenser för tusentals träd.

Kombinationen av genetisk information från naturliga populationer med data från granens förädlingsprogram skapade en unik möjlighet att under-söka den historiska diverseringen av lokal adaption hos gran. Den svenska granens förädlingsprogram är särskilt intressant att studera i avseende för lokaladaption då många av individerna som användes vid dess etablering har sitt ursprung i andra regioner av Europa.

Avhandlingen illustrerar hur mycket information som är tillgängligt inom skogsförädlingsprogrammen och hur denna information kan användas för att besvara applicerad och fundamentala frågor gällande evolutionen hos skogs-träd. Nedanför kommer jag kortfattat sammanställa de huvudsakliga upp-täckterna, samt redogöra hur framtida studier inom området kan vidareut-vecklas.

I artikel fokuserad på analysen av områdesfrekvensspektra mellan flera populationer (artikel i) av Siberiskgran och tre andra huvuddomänerna inom gran. Den nuvarande granpopulationen är ett resultat av en komplex populat-ionsrörelse där genutbyte mellan minst tre urspungspopulationer (Alpinsk,

46

Karpatisk och Fennoskandinavisk gran) bildat Europas nuvarande populat-ioner (centraleuropeisk, centralsvensk, rysk-baltisk och nord polsk).

Vi observerade att förädlingsprogrammen i södra Sverige är stark mixad och sammanfaller med omflyttningar i närtid, huvudsakligen med introdukt-ioner från Central Europa. En viktigt upptäckt var att observation av gene-tiska likheter var ett effektivt sätt att förutse ursprunget av tidigare granin-troduktioner. Gällande den demografiska utvecklingen hos gran finns det aspekter som erfordrar ytterligare efterforskning. Få studier har provtagit i nordvästra Ryssland vilket är av vikt då detta geografiska område troligen spelade en avgörande roll i koloniseringen av Skandinavien. Vidare bör be-visen för den antagna östliga expansion av granen från Karpaterna stärkas.

Efter att vi fördelat ursprungsindividerna för granens förädlingsprogram till olika genetiska kluster, fokuserad vi på lokal adaptation genom att nyttja den fenotypiskdata som är tillgänglig via förädlingsprogrammet. PCAs från miljödata av trädets ursprung, genomisk data samt fenotypiska egenskaper (lövsprickning, höjd och diameter) var enhetligt, vilket indikerar ett starkt mönster av lokal adaption. Egenskaper höjd och diameter var starkt korrele-rade och varierade längs en latitudsgradient medan lövsprickning varierade inom en longitudsgradient. Den tidigare var konsekvent starkare påverkad av korrektion för populationsgenetiska strukturer än den senare.

Vi fann också att träd från Karpaterna växte snabbare än lokala träd i södra Sverige, vilket indikerar att assisterat genflöde skulle kunna vara en fungerande strategi för att underlätta effekterna av klimatförändringen.

Genome-wide association studies (GWAS) visade också att både tillväxt-egenskaper (höjd och diameter) är mer polygenetiska än lövsprickning. En begränsning av den slutsatsen är att den baseras på endast tre fenotypiska egenskaper. I framtida studier är det viktigt att inkludera ytterligare egen-skaper, särskilt de som kan vara direkt kopplade till anpassning för kom-mande klimatförändringar. Tillförande av ytterligare egenskaper kan också reducera mängden ”false-negatives” vid korrigering för populationsgenetiska strukturer inom GWAS. Inkludering av fler egenskaper med geografisk vari-ation skulle underlätta kvantifieringen av den negativa påverkan av korrige-ring av populationsstrukturer och utveckla metoder för att begränsa antalet ”false negatives”.

De första två studierna fokuserar på den södra delen av den svenska för-ädlingspopulationen medan den tredje analyserade den genetiska strukturen av 4769 träd spridd över hela landet. För att karaktärisera populationens genetiskastruktur och fånga de olika processerna som bidragit till denna an-vändes ett flertal olika tillvägagångssätt.

Svenska träd gruppera sig i två genetiska grupper med en kontaktzon i central Sverige. Ett tredje mindre kluster upptäckes också i norr vilket indi-kera en mer betydande påverkan från Sibiriskgran. Kontaktzonen mellan de två huvudgrupperna reflekterar mötespunkten för Skandinaviens två koloni-seringsvågor som följde den senaste istiden. Variationen inom en stor andel

47

av alla SNP:s var kopplade till miljövariabler och uppvisade ett starkare mönster av Isolering via distans än övriga SNP förändringarna. Detta indike-rar att naturligturval spelade en viktig roll under populationens etablering. En lämplig vidareutveckling vore att fastställa associationen med fenoty-piska egenskaper. Det låga antalet av fenotypiska egenskaper som mättes i förädlingsprogrammet utgör dock en begränsning. Vidare är många av de inkluderade egenskaper viktiga ur förädlingssynpunkt, men mindre bety-dande ur rekoloniseringssynpunkt samt i avseende till granens anpassnings-förmåga till klimatförändringar.

Slutligen i den fjärde artikeln titta vi på fenologi hos gran och dess kopp-ling till generna FTL2 och GI. Vi såg här en latitudgradient i lövsprickning som sammanföll med förändringar i FTL2 uttryck i Sibiriska granpopulat-ioner längs Ob floden. Detta överensstämmer med tidigare resultat observe-rade i gran och sibiriskgran i andra latitudsgradienter. Det var dock svårt att hitta bevis för naturligturval i FTL2 and GI. Givet nyckelpositionen som dessa gener har för plantans utveckling och fenologi vore det önskvärt att bättre förstå deras utveckling i relation till selektion.

48

Acknowledgements

It has taken a lot of efforts to reach this point. No doubt that I have gained a lot from my PhD studies. To be honest it is not easy to overcome all these difficulties without help from many of you.

I would like to thank my supervisor Martin Lascoux. You are the great supervisor. You always give me good advice in both my PhD projects and life and always make sure I am doing well on my journey to the PhD disser-tation. You are always so patient to explain my questions, even very basic ones. You always gave me freedom and trust to solve my personal issues especially the beginning of the PhD study. I also want to thank you for your unconditional support during my sick leave. I am really lucky to work in your golden team.

I also would like to thank Jun Chen for giving me a lot of clear explana-tions and good suggestions including how to propose interesting scientific questions and how to look at the principles behind statistics as well as how to properly analyze data. I also would like to thank you for sharing your scripts. You are really good at not only population genetics theory but also bioinformatic analysis. I also would like to give my special thanks to Pascal Milesi for helping the last chapter. It is amazing that you can always point out the exact problems and give very interesting solutions and suggestions. I always think very hard when I discuss with you, which helps me find the right path of the last chapter. I also would like to thanks Per Unneberg and Douglas Scofield for helping me work with huge spruce exome data set efficiently on UPPMAX. Many thanks to my second supervisor Ulf Lagercrantz, to my officemate Sylvain Glemin for explaining academic questions, to Kerstin Jeppsson for helping me in the lab and to people who helped me with sampling.

My special thanks to Jenny Lundh and Linn Eriksson for preparing all kinds of documents and talking with me when I am stressed, to Magnus Eklund for encouraging me be more confident, to Niclas Backström for organizing us to watch ice hockey and floor ball matches, to Sophie Karrenberg for opening the door to programming for me and practicing Swedish with me, to Yvonne Meyer-Lucht for making me relax during the thesis writhing stage and to all other PIs.

I also would like to give many thanks to my collogues and friends. My PhD life would have rather boring without you. Anja Billhardt: Thanks a lot for all help from you and all the talks we had. My PhD life has been

49

much more fun after I met you! You know you are super sweet. I am really touched every time you came to my office and checked if everything was OK. I really like to talk to you because you always can make me feel easy and comfortable. You definitely are the one that makes Plant Ecology and Evolution people more social. It was the great idea to make waffles in the lunch room! Aiying Wang: It is always nice to talk to you and thanks for having lunch together. Alessandro Nobile: Thanks for great lunch time. It was fun to extract DNA together. Anna-Malin Linde: Thanks a lot for shar-ing your lectures notes and courses information with me. It was very nice to take course together with you. You are always so modest. Do you still re-member we did the exam of Population Genetic Analysis by mistakes? You always said you didn’t know so much, but it turned out you got full score! Asheley Dai: Thanks for cooking for us. I already miss your food. Wish you and your family a great Chinese New Year! Charles Campbell: Thanks for having great lunch time. Baosheng Wang: Thanks for having nice talks and great meals together. Chen Chen: I am happy you were my roommate. I was glad to share my food with you. Do you still remember we Swished to wrong person? But luckily, we got it back, it was fun. Dmytro Kryvokhyzha: It was so nice to work with you in the same group. Eva Kondrysová: We are so similar in a lot of ways. It was a pity that we knew each other so late and didn’t have so much time to talk before you went back Czech Republic. Fernando Chaguaceda: It was fun to have Swedish course together. Fia Bengtsson: Thanks a lot for painting my thesis cover. You are super crea-tive. It is unbelievable, how could you come up with a lot of great ideas im-mediately? Froukje Postma: Thanks for having nice talks with me at the beginning of my PhD. Elodie Chapurlat: It was fun to Spex with you! Giu-lia Zacchello: It is always so nice to hear you asking for lunch every time. Thanks for giving me free zucchini and pumpkin plants. They did grow well at the beginning, but didn’t make it in the end because I was away for a while. Han Lu: Thanks a lot for happy office time for two years. You are always so positive. Hope everything goes well in China. Judith Trunschke: You are so brave to do the post-doc in China. I am so glad that you enjoy the research and life there. Karin Näsvall: It was nice to read the book “Molec-ular Population Genetics” together. Kevin Nota: You changed my opinion of sandwiches. Sandwiches can actually be very tasty and also can save a lot of time! Leyi Su: Thanks for being my officemate for a while and having lunch together. I had great badminton time with you. Linus Söderquist: Thanks for having lunch together. Luis Leal: Thanks for all the great time. I always enjoin playing badminton with you, discovering Asian restaurants, having pancakes with you, your daughter even brought liquor to me, talking about gender equality, Japanese amines, movies, sharing great traveling pho-to, watching ice hockey and floor balls and so on… Wow, there are a lot of great moments. You know it is really fun to start with “itadakimasu!” before meal. Thanks for encourage me be more confident. Nina Joffard: Thanks

50

for great lunch time. Maria Guerrina: It was fun to have lunch with you. Thanks for sharing your pesto recipe which is the best one. I am so glad that you chose to visit fantastic geographic landscape in my home region in your honeymoon. Marion Orsucci: Thanks for being my officemate for three years. It was great to play badminton with you. I like the tea flavor which you gave me. Mathieu Tiret: Thanks a lot for helping me photoshop figures and helping me correct my KAPPA. It was fun to discovery food with you and Luis and fun to learn Japanese. Mattias Vass: Thanks for sharing your super nature photos with me. Hope everyone can protect nature like you. Salim Hossain Reza: Thanks for bringing fantastic Fika. It is always nice to talk with you. I really hope everything is going well with you and your fami-ly. Sophia Renes: You are the great friend for us. You are so smart and con-siderable. It is always nice to talk with you. Susanne Gramlich: I had a lot of fun having Swedish course with you. You are the real environmentalist. I was shocked when I knew that you took trains between Uppsala and South-ern Germany. Theresa Lumpi: It is always pleasurable to talk with you! Thomas Richards: Thanks a lot for helping correct my KAPPA. I got a lot of precious comments. It is very nice to have lunch together. Tianlin Duan: Thanks for helping me in research and life. I am always amazed by your passion of biology and your professions on all things. Even the suggestions for Xian travelling was so complete. In my mind, you should be the Nobel prize winner in the future. Xiaodong Liu: Thanks for all your great helps in research and my life. You are the competitive person, know well about mod-els, formulas of population genetics and really good at coding. It is great you find your dreaming job now. Thanks for taking care of my parents when they were in Sweden. You are always there whenever I need you. Thanks for having a lot of jokes and fun talks, cooking for me and playing basketball and ping-pang.

I also would like to thank my Chinese friends. Haiqing Sun: It was great to travel together in Helsinki, Tallinn and Saint Petersburg. Thanks for teaching me all kinds of things. Even a lot of times, I didn’t agree with, but time proves you are right. Jingyi Liu: It was great to know you before we came to Sweden and it was even greater to share my apartment with you. You have strong wills because you always know what you want. Hope eve-rything goes well in China. Liao Zhen: You know you are very brave and smart. I should definitely learn that from you. You really know how to cook. I will definitely try your eggplant Baozi recipe once I am not afraid of mak-ing doughs. It is always right to ask you for restaurant recommendations. Ruixue Sun: You are always so kind to me. Thanks for all your helps and for sharing your food, especially YouZhaGao. Thanks for taking care my mom. I hope everything will go well. Qinghua Liao: Thanks for playing ping-pong with me. It’s great to smash most of time. Wang Xi: We have known each other for almost 8 years. Wow, time goes fast, does it? We had a lot of great lunch time in the town. In some way, we kind of keeping the

51

traditions we did with Nizi and others in Lanzhou, except SanGuoSha. Thanks for taking care of my parents in the airplane. Yan Ping: I always feel we are so similar. It’s great to hang out with you and go shopping together. It’s great you come back Sweden. Yule: Thanks for having great lunch time in Asian restaurants. ‘Liu Laoban’ group is getting bigger in Uppsala now.

Special thanks to doctors and nurses. Tomas Nilsson: Thank you so much for trying best when treating me. To me, you are not only the doctor, but also the friend. You are so experienced, knowledgeable and kind. I am unlucky to be sick, but I am lucky to be your patient. Ander Nygren and Maria Berg-ström: Thank you for taking care of me for one year. You always treated me like your own kid, which made me feel so relax every time when I visited you or asked help from you. Mia Andersson: Thank you very much for solving all kinds of problems and taking care of my parents. Thank you for helping me dealing with bad emotions. You are like an old friend to me. Carina Nilsson: Thank you so much for taking care of my mom and I. You are always so kind. Do you still remember the cake you brought? That’s the best cake. You are one of the best nurses in the world. They should reward you a Florence Nightingale Medal. Pernilla Blomquist: It is always fun to talk to you. Thank you for giving me chances to practice Swedish and taking care of me.

Many thanks to the lovely Andersson family. Lena and Sture You are two of most kind people in the world. I always feel easy with you and Här-lunda is the heaven to me. I still remember how fantastic it was in the 2016 Christmas. You knew I couldn’t travel that time, but you just made that hap-pen. Jämtland hiking and Säfsen skiing, were two great choices. Hope we can travel once a year in the future. Jag har alltid en trevlig tid med er, tack så mycket! Jonas You are a challenger. It was fun to climb the rock moun-tain in STF Sylarna Fjällstation, except we lost lilla Sara. Thanks for bring coffee in the hiking, that’s the tastiest coffee in the world. Sara it is always easy to talk to you and I had a lot of fun. We all know you are always so kind to me and solve all kinds of my troubles without any hesitations. You are the super sister. I am so lucky to be your friend. Tusen Tack! I would like to thank my lively sambo, Martin: I cannot imagine how my life would have been without you. Meeting you four year ago was definitely an im-portant turning point in my life. Thanks for dragging me out of the desperate situation. Thanks for correcting my English insistently. You knew I didn’t even know how to communicate with others before I met you. You have always been with me no matter how hard from the beginning and how bad of things. You can always distract my focus on sad things, believe in me and encourage me to be more confident. I am really lucky to have you in my life. You are to me like the sun to nature, let we dance beneath the sunshine. I always have great time with you my angel, love you!

52

502

53

Reference

Ågren J, Schemske DW (2012) Reciprocal transplants demonstrate strong adaptive differentiation of the model organism Arabidopsis thaliana in its native range. New Phytol 4:1112–1122

Acheré V, Favre JM, Besnard G, Jeandroz S (2005) Genomic organization of molecular differentiation in Norway spruce (Picea abies). Mol Ecol 14:3191–3201

Adams AM, Hudson RR (2004) Maximum-likelihood estimation of demographic parameters using the frequency spectrum of unlinked single-nucleotide polymorphisms. Genetics 168:1699–1712

Aitken SN, Whitlock MC (2013) Assisted gene flow to facilitate local adaptation to climate change. Annu Rev Ecol Evol Syst 44:367–388

Aitken SN, Bemmels JB (2016) Time to get moving: Assisted gene flow of forest trees. Evol Appl 9:271–290

Alexa A, Rahnenfuhrer J (2010) topGO: Enrichment analysis for gene ontology. R Package Version 2

Alexander DH, Novembre J, Lange K (2009) Fast model-based estimation of ancestry in unrelated individuals. Genome Res 19:1655–1664

Arnold ML (2004) Natural hybridization and the evolution of domesticated, pest and disease organisms. Mol Ecol 13:997–1007

Avia K, Kärkkäinen K, Lagercrantz U, Savolainen O (2014) Association of FLOWERING LOCUS T/TERMINAL FLOWER 1-like gene FTL2 expression with growth rhythm in Scots pine (Pinus sylvestris). New Phytol 204:159–170

Baison J, Vidalis A, Zhou L, et al (2018) Association mapping identified novel candidate loci affecting wood formation in Norway spruce. bioRxiv

Barton N, Hermisson J, Nordborg M (2019) Why structure matters. Elife 8:e45380 Beaulieu J, Doerksen T, Boyle B, et al (2011) Association genetics of wood physical

traits in the conifer white spruce and relationships with gene expression. Genetics 188:197–214

Beaumont MA, Nichols RA (1996) Evaluating loci for use in the genetic analysis of population structure. Proc R Soc B Biol Sci 263:1619–1626

Beaumont MA, Zhang W, Balding DJ (2002) Approximate Bayesian computation in population genetics. Genetics 162:2025–35

Berg JJ, Co Beaumont MA, Balding DJ (2004) Identifying adaptive genetic divergence among populations from genome scans. Mol Ecol 13:969–980op

Berg JJ, Coop G (2014) A population genetic signal of polygenic adaptation. PLoS Genet 10:e1004412

Berg JJ, Zhang X, Coop G (2017) Polygenic Adaptation has Impacted Multiple Anthropometric Traits. bioRxiv

Binney HA, Willis KJ, Edwards ME, et al (2009) The distribution of late-Quaternary woody taxa in northern Eurasia: evidence from a new macrofossil database. Quat Sci Rev 28:2445–2464

54

Borghetti M, Glannini R, Menozzi P (1988) Geographic variation in cones of Norway spruce (Picea abies (L.) Karst.). Silvae Genet (Germany, FR) 37:178–184

Boyle EA, Li YI, Pritchard JK (2017) An expanded view of complex traits: from polygenic to omnigenic. Cell 15:7

Bradburd GS, Coop GM, Ralph PL (2018) Inferring continuous and discrete population genetic. Genetics 210:33–52

Bradbury PJ, Zhang Z, Kroon DE, et al (2007) TASSEL: Software for association mapping of complex traits in diverse samples. Bioinformatics 23:2633–2635

Bray SCE, Austin JJ, Metcalf JL, et al (2013) Ancient DNA identifies post-glacial recolonisation, not recent bottlenecks, as the primary driver of contemporary mtDNA phylogeography and diversity in Scandinavian brown bears. Divers Distrib 19:245–256

Bucci G, Vendramin GG (2000) Delineation of genetic zones in the European Norway spruce natural range: Preliminary evidence. Mol Ecol 9:923–934

Bulmer MG (1972) Multiple Niche Polymorphism. Am Nat 106:254–257 Butlin RK (2010) Population genomics and speciation. Genetica 138:409–418 Chambel MR, Climent J, Alía R (2007) Divergence among species and populations

of Mediterranean pines in biomass allocation of seedlings grown under two watering regimes. Ann For Sci 64:87–97

Chang CC, Chow CC, Tellier LCAM, et al (2015) Second-generation PLINK: Rising to the challenge of larger and richer datasets. Gigascience 4:7

Charlesworth B (2009) Fundamental concepts in genetics: Effective population size and patterns of molecular evolution and variation. Nat Rev Genet 10:195–205

Chen J, Källman T, Ma X, et al (2012) Disentangling the roles of history and local selection in shaping clinal variation of allele frequencies and gene expression in Norway spruce (Picea abies). Genetics 191:865–881

Chen J, Tsuda Y, Stocks M, et al (2014) Clinal variation at phenology-related genes in spruce: Parallel evolution in FTL2 and Gigantea? Genetics 197:1025–1038

Chen J, Källman T, Ma XF, et al (2016) Identifying Genetic Signatures of Natural Selection Using Pooled Population Sequencing in Picea abies. G3 Genes, Ge-nomes, Genet 6:1979–1989

Chen J, Li L, Milesi P, et al (2019) Genomic data provide new insights on the de-mographic history and the extent of recent material transfers in Norway spruce. Evol Appl 12:1539–1551

Chen Z (2016) Quantitative genetics of Norway spruce in Sweden. Swedish Univer-sity of Agricultural Sciences, Umeå

Clausen J, Keck DD, Hiesey WM (1940) Experimental studies on the nature of species. I. Effect of varied environments on western North American plants. Publication 520, Washington, D.C.: Carnegie Institution of Washington

Coop G, Witonsky D, Di Rienzo A, Pritchard JK (2010) Using environmental correlations to identify loci underlying local adaptation. Genetics 185:1411–1423

Currat M, Ruedi M, Petit RJ, Excoffier L (2008) The hidden side of invasions: Massive introgression by local genes. Evolution 62:1908–1920

Dépraz A, Cordellier M, Hausser J, Pfenninger M (2008) Postglacial recolonization at a snail’s pace (Trochulus villosus): Confronting competing refugia hypotheses using model selection. Mol Ecol 17:2449–2462

De Carvalho D, Ingvarsson PK, Joseph J, et al (2010) Admixture facilitates adaptation from standing variation in the European aspen (Populus tremula L.), a widespread forest tree. Mol Ecol 19:1638–1650

55

Excoffier L, Hofer T, Foll M (2009) Detecting loci under selection in a hierarchically structured population. Heredity 103:285–298

Excoffier L, Dupanloup I, Huerta-Sánchez E, et al (2013) Robust demographic inference from genomic and SNP Data. PLoS Genet 9:p.e1003905

Danell O (1991) Transformering av avels och genotypvärden från underliggande skala till sannolikhetsskala. Uppsala, Sweden: Institute for Forest Improvement

Daub JT, Hofer T, Cutivet E, et al (2013) Evidence for polygenic adaptation to pathogens in the human genome. Mol Biol Evol 30:1544–1558

De Kovel CGF (2006) The power of allele frequency comparisons to detect the footprint of selection in natural and experimental situations. Genet Sel Evol 38:3–23

De Giorgio M, Rosenberg NA (2013) Geographic sampling scheine as a determinant of the major axis of genetic variation in principal components analysis. Mol Biol Evol 30:480–488

Diaz-Papkovich A, Anderson-Trocmé L, Ben-Eghan C, Gravel S (2019) UMAP reveals cryptic population structure and phenotype heterogeneity in large genomic cohorts. PLoS Genet 15:e1008432

Dillon SK, Nolan M, Li W, et al (2010) Allelic variation in cell wall candidate genes affecting solid wood properties in natural populations and land races of Pinus radiata. Genetics 185:1477–1487

Fagernas Z (2017) Biogeography of Norway spruce (Picea abies (L.) Karst.): Insights from a genome-wide study. University of Umeå, Sweden

Farjon A, Filer D (2013) Europe and the Mediterranean. In: an Atlas of the World’s conifers: An analysis of their distribution, biogeography, diversity and conservation status. Brill, Leiden

Fedorov VB, Fredga K, Jarrell GH (1999) Mitochondrial DNA variation and the evolutionary history of chromosome races of collared lemmings (Dicrostonyx) in the Eurasian Arctic. J Evol Biol 12:134–145

Fick SE, Hijmans RJ (2017) WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. Int J Climatol 37:4302–4315

Field Y, Boyle EA, Telis N, et al (2016) Detection of human adaptation during the past 2000 years. Science 354:760–764

Fischer MC, Foll M, Excoffier L, Heckel G (2011) Enhanced AFLP genome scans detect local adaptation in high-altitude populations of a small rodent (Microtus arvalis). Mol Ecol 20:1450 1462

Fischer MC, Rellstab C, Tedder A, et al (2013) Population genomic footprints of selection and associations with climate in natural populations of Arabidopsis halleri from the Alps. Mol Ecol 22:5594–5607

Foll M, Gaggiotti O (2008) A genome-scan method to iden- tify selected loci appropriate for both dominant and codominant markers: a Bayesian perspective. Genetics 180:977–993

Foll M, Fischer MC, Heckel G, Excoffier L (2010) Estimating population structure from AFLP amplification intensity. Mol Ecol 19:4638–4647

Fornara F, de Montaigu A, Coupland G (2010) SnapShot: Control of flowering in arabidopsis. Cell 141:3–5

Frank A, Sperisen C, Howe GT, et al (2017) Distinct genecological patterns in seedlings of Norway spruce and silver fir from a mountainous landscape. Ecology 98:211–227

Galinsky KJ, Bhatia G, Loh PR, et al (2016) Fast principal component analysis reveals convergent evolution of ADH1B in Europe and East Asia. Am J Hum Genet 98:456–472

56

Giesecke T, Bennett KD (2004) The Holocene spread of Picea abies (L.) Karst. in Fennoscandia and adjacent areas. J Biogeogr 31:1523–1548

González-Martínez SC, Alía R, Gil L (2002) Population genetic structure in a Mediterranean pine (Pinus pinaster Ait.): A comparison of allozyme markers and quantitative traits. Heredity 3:199–206

Günther T, Coop G (2013) Robust identification of local adaptation from allele frequencies. Genetics 195:205–220

Gutenkunst RN, Hernandez RD, Williamson SH, Bustamante CD (2009) Inferring the joint demographic history of multiple populations from multidimensional SNP frequency data. PLoS Genet 5:e1000695

Gyllenstrand N, Clapham D, Källman T, Lagercrantz U (2007) A Norway Spruce FLOWERING LOCUS T homolog is implicated in control of growth rhythm in conifers. Plant Physiol 144:248–257

Hall D, Ma XF, Ingvarsson PK (2011) Adaptive evolution of the Populus tremula photoperiod pathway. Mol Ecol 20:1463–1474

Hancock AM, Witonsky DB, Gordon AS, et al (2008) Adaptations to climate in candidate genes for common metabolic disorders. PLoS Genet 4:

Hancock AM, Brachi B, Faure N, et al (2011) Adaptation to climate across the Arabidopsis thaliana genome. Science (80- ) 334:83–86

Herben T, Suzuki JI (2001) A simulation study of the effects of architectural constraints and resource translocation on population structure and competition in clonal plants. Evol Ecol 15:403–423

Heuertz M, De Paoli E, Källman T, et al (2006) Multilocus patterns of nucleotide diversity, linkage disequilibrium and demographic history of Norway spruce [Picea abies (L.) Karst]. Genetics 174:2095–2105

Hewitt G (2000) The genetic legacy of the Quaternary ice ages. Nature 405:907–13 Hey J, Nielsen R (2004) Multilocus methods for estimating population sizes,

migration rates and divergence time, with applications to the divergence of Drosophila pseudoobscura and D. persimilis. Genetics 167:747–760

Hodgins KA, Yeaman S, Nurkowski KA, et al (2016) Expression divergence is correlated with sequence evolution but not positive selection in conifers. Mol Biol Evol 33:1502–1516

Hultén E (1937) Outline of the history of Arctic and boreal biota during the Quaternary period. In: Foundations of biogeography: classic papers with commentaries. Stockholm, Bokförlags Aktiebolaget Thule, p 168

Ingvarsson PK, García MV, Hall D, et al (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 PK, Garcia MV, Luquez V, et al (2008) Nucleotide polymorphism and phenotypic associations within and around the phytochrome B2 locus in European aspen (Populus tremula, Salicaceae). Genetics 178:2217–2226

IUCN (1987) The IUCN position statement on translocation of living organisms. In: IUCN, Gland. Switz.

Ivetić V, Aleksić JM (2016) Response of rare and endangered species Picea omorika to climate change - The need for speed. Reforesta 2:81–99

Jaarola M, Tegelström H, Fredga K (1999) Colonization history in Fennoscandian rodents. Biol J Linn Soc 68:113–127

Jansen S, Konrad H, Geburek T (2017) The extent of historic translocation of Norway spruce forest reproductive material in Europe. Ann For Sci 74:56

57

Karlgren A, Gyllenstrand N, Clapham D, Lagercrantz U (2013) FLOWERING LOCUS T/TERMINAL FLOWER1-like genes affect growth rhythm and bud set in Norway spruce. Plant Physiol 163:792–803

Karlsson B, Rosvall O (1993) Progeny testing and breeding strategies. In: Proceedings of the Nordic group for tree breeding. Edinburgh

Kimura M, Weiss GH (1964) The stepping stone model of population structure and the decrease of genetic correlation with distance. Genetics 49:561–576

Krutzsch (1973) Norway spruce development of buds. Vienna Kupryjanowicz M, Nalepka D, Pidek IA, et al (2018) The east-west migration of

trees during the Eemian Interglacial registered on isopollen maps of Poland. Quat Int 467:178–191

Lagercrantz U, Ryman N (1990) Genetic structure of Norway Spruce (Picea abies): concordance of morphological and allozymic variation. Evolution 44:38

Lamara M, Raherison E, Lenz P, et al (2016) Genetic architecture of wood properties based on association analysis and co-expression networks in white spruce. New Phytol 210:240–255

Lascoux M, Palmé AE, Cheddadi R, Latta RG (2004) Impact of ice ages on the genetic structure of trees and shrubs. Philos Trans R Soc B Biol Sci 359:197–207

Lascoux M, Glémin S, Savolainen O (2016) Local adaptation in plants. eLS 1–7 Le Corre V, Kremer A (2003) Genetic variability at neutral markers, quantitative

trait loci and trait in a subdivided population under selection. Genetics 164:1205–1219

Le Corre V, Kremer A (2012) The genetic differentiation at quantitative trait loci under local adaptation. Mol Ecol 21:1548–1566

Leimu R, Fischer M (2008) A meta-analysis of local adaptation in plants. PLoS One 3:e4010

Leslie AB, Beaulieu JM, Rai HS, et al (2012) Hemisphere-scale differences in conifer evolutionary dynamics. Proc Natl Acad Sci U S A 109:16217–16221

Lewontin RC, J. K (1973) Distribution of gene frequency as a test of the theory of the selective neutrality of polymorphisms. Genetics 74:175–195

Li H, Durbin R (2010) Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics 26:589–595

Li L, Chen J, Lascoux M (2019) Clinal variation in growth cessation and FTL2 expression in Siberian spruce. Tree Genet Genomes 15:82

Liaw A, Wiener M (2002) Classification and regression by random-Forest. R News 2:18 22

Liu L, Adrian J, Pankin A, et al (2014) Induced and natural variation of promoter length modulates the photoperiodic response of FLOWERING LOCUS T. Nat Commun 5:4558

Lockwood JD, Aleksić JM, Zou J, et al (2013) A new phylogeny for the genus Picea from plastid, mitochondrial, and nuclear sequences. Mol Phylogenet Evol 69:717–727

Lukic S, Hey J (2012) Demographic inference using spectral methods on SNP data, with an analysis of the human out-of-Africa expansion. Genetics 192:619–639

Ma XF, Hall D, St. Onge KR, et al (2010) Genetic differentiation, clinal variation and phenotypic associations with growth cessation across the Populus tremula photoperiodic pathway. Genetics 186:1033–1044

Manolio TA, Collins FS, Cox NJ, et al (2009) Finding the missing heritability of complex diseases. Nature 461:747–753

58

Marinova E, Tonkov S, Bozilova E, Vajsov I (2012) Holocene anthropogenic landscapes in the Balkans: the palaeobotanical evidence from southwestern Bulgaria. Veg Hist Archaeobot 21:413–427

Mayr E (1963) Animal species and evolution. Cambridge, MA: harvard University Press

McDonald JH, Kreitman M (1991) Adaptive protein evolution at the Adh locus in Drosophila. Nature 351:652–654

McKenna A, Hanna M, Banks E, et al (2010) The genome analysis toolkit: A MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res 20:1297–1303

Milesi P, Berlin M, Chen J, et al (2019) Assessing the potential for assisted gene flow using past introduction of Norway spruce in southern Sweden: Local adaptation and genetic basis of quantitative traits in trees. Evol Appl 12:1946–1959

Moritz C (1994) Defining “Evolutionarily Significant Units” for conservation. Trends Ecol Evol 9:373–375

Moritz C (2002) Strategies to protect biological diversity and the evolutionary processes that sustain it. Syst Biol 5:238–254

Müller BSF, de Almeida Filho JE, Lima BM, et al (2019) Independent and Joint-GWAS for growth traits in Eucalyptus by assembling genome-wide data for 3373 individuals across four breeding populations. New Phytol 221:818–833

Myking T, Rusanen M, Steffenrem A, et al (2016) Historic transfer of forest reproductive material in the Nordic region: Drivers, scale and implications. Forestry 89:325–337

Neale DB, Kremer A (2011) Forest tree genomics: Growing resources and applications. Nat Rev Genet 12:111–122

Nei M (1972) Genetic Distance between Populations. Am Nat 106:283–292 Nei M, Maryuyama T (1975) Letters to the editors: Lewontin–Krakauer test for

neutral genes. Genetics 80:395 Nielsen R, Wakeley J (2001) Distinguishing migration from isolation: a Markov

chain Monte Carlo approach. Genetics 158:885–896 Notivol E, García-Gil MR, Alía R, Savolainen O (2007) Genetic variation of growth

rhythm traits in the limits of a latitudinal cline in Scots pine. Can J For Res 37:540–551

Novembre J, Stephens M (2008) Interpreting principal component analyses of spatial population genetic variation. Nat Genet 40:646–649

Nystedt B, Street NR, Wetterbom A, et al (2013) The Norway spruce genome sequence and conifer genome evolution. Nature 497:579–584

Pavy N, Namroud MC, Gagnon F, et al (2012) The heterogeneous levels of linkage disequilibrium in white spruce genes and comparative analysis with other conifers. Heredity 108:273–284

Petit RJ, Petit J, Aguinagalde I, et al (2003) Genetic diversity glacial refugia: hotspots but not melting pots of diversity. Science 1563:1563–1565

Petkova D, Novembre J, Stephens M (2015) Visualizing spatial population structure with estimated effective migration surfaces. Nat Genet 48:94–100

Pickrell JK, Pritchard JK (2012) Inference of population splits and mixtures from genome-wide allele frequency data. PLoS Genet 8:e1002967

Pilot M, Greco C, Vonholdt BM, et al (2014) Genome-wide signatures of population bottlenecks and diversifying selection in European wolves. Heredity 112:428–442

Pin PA, Benlloch R, Bonnet D, et al (2010) An antagonistic pair of FT homologs mediates the control of flowering time in sugar beet. Science 330:1397–1400

59

Pin PA, Nilsson O (2012) The multifaceted roles of FLOWERING LOCUS T in plant development. Plant, Cell Environ 35:1742–1755

Polechová J, Barton N (2011) Genetic drift widens the expected cline but narrows the expected cline width. Genetics 189:227–235

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

Prunier J, Pelgas B, Gagnon F, et al (2013) The genomic architecture and association genetics of adaptive characters using a candidate SNP approach in boreal black spruce. BMC Genomics 14:368

R Core Team (2019) R: A language and environment for statistical computing. Vienna,Austria: R foundation for Statistical Computing. https://www.r-project.org/

Remington DL, Thornsberry JM, Matsuoka Y, et al (2001) Structure of linkage disequilibrium and phenotypic associations in the maize genome. Proc Natl Acad Sci U S A 98:11479–11484

Robertson A (1975a) Letters to the editors: Remarks on the Lewontin–Krakauer test. Genetics 80:396

Robertson A (1975b) Gene frequency distributions as a test of selective neutrality. Genetics 81:775–785

Root TL, Price JT, Hall KR, et al (2003) Fingerprints of global warming on wild animals and plants. Nature 421:57–60

Rousset F (1997) Genetic differentiation and estimation of gene flow from F-statistics under isolation by distance. Genetics 145:1219–1228

Savolainen O, Lascoux M, Merilä J (2013) Ecological genomics of local adaptation. Nat Rev Genet 14:807–820

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

Slatkin M (1993) Isolation by distance in equilibrium and non-equilibrium populations. Evolution 47:264–279

Slatkin M (1995) A measure of population subdivision based on microsatellite allele frequencies. Genetics 139:457–462

Stephens M (2017) False discovery rates: A new deal. Biostatistics 18:275–294 Stern DL (2013) The genetic causes of convergent evolution. Nat Rev Genet

14:751–764 Sunnåker M, Busetto AG, Numminen E, et al (2013) Approximate Bayesian

Computation. PLoS Comput Biol 9:e1002803 Tajima F (1989) Statistical method for testing the neutral mutation hypothesis by

DNA polymorphism. Genetics 123:585–595 Tollefsrud MM, Kissling R, Gugerli F, et al (2008) Genetic consequences of glacial

survival and postglacial colonization in Norway spruce: Combined analysis of mitochondrial DNA and fossil pollen. Mol Ecol 17:4134–4150

Tollefsrud MM, Sønstebø JH, Brochmann C, et al (2009) Combined analysis of nuclear and mitochondrial markers provide new insight into the genetic structure of North European Picea abies. Heredity 102:549–562

Tollefsrud MM, Latałowa M, van der Knaap WO, et al (2015) Late Quaternary history of North Eurasian Norway spruce (Picea abies) and Siberian spruce (Picea obovata) inferred from macrofossils, pollen and cytoplasmic DNA variation. J Biogeogr 42:1431–1442

Tsuda Y, Chen J, Stocks M, et al (2016) The extent and meaning of hybridization and introgression between Siberian spruce (Picea obovata) and Norway spruce (Picea abies): cryptic refugia as stepping stones to the west? Mol Ecol 25:2773–2789

60

Turesson G (1922) The genotypical response of the plant species to the habitat. Hereditas 3:211–350

Väliranta M, Kaakinen A, Kuhry P, et al (2011) Scattered late-glacial and early Holocene tree populations as dispersal nuclei for forest development in north-eastern European Russia. J Biogeogr 38:922–932

Vendramin GG, Anzidei M, Madaghiele A, et al (2000) Chloroplast microsatellite analysis reveals the presence of population subdivision in Norway spruce (Picea abies K.). Genome 43:68–78

Vidalis A, Scofield DG, Neves LG, et al (2018) Design and evaluation of a large sequence-capture probe set and associated SNPs for diploid and haploid samples of Norway spruce (Picea abies). bioRxiv

Wagner S, Liepelt S, Gerber S, Petit RJ (2015) Within-range translocations and their consequences in European larch. PLoS One 10:1–17

Wahlund S. (1928) Zuzammensetzung von populationen und korrelationserscheinungen vom standpunkt der vererbungslehre aus betrachtet. Hereditas 11:65–106

Walther GR, Post E, Convey P, et al (2002) Ecological responses to recent climate change. Nature 416:389–395

Wang J, Ding J, Tan B, et al (2018) A major locus controls local adaptation and adaptive life history variation in a perennial plant. Genome Biol 19:1–17

Waples RS (1998) Separating the wheat from the chaff: Patterns of genetic differentiation in high gene flow species. J Hered 89:438–450

Weeks AR, Sgro CM, Young AG, et al (2011) Assessing the benefits and risks of translocations in changing environments: A genetic perspective. Evol Appl 4:709–725

Weir BS (1996) Genetic data analysis II. Sunderland, MA: Sinauer Associate Whitlock MC, Lotterhos KE (2015) Reliable detection of loci responsible for local

adaptation: Inference of a null model through trimming the distribution of FST. Arm Nat 186:S24–S36

Wright S. (1931) Evolution in Mendelian populations. Genetics 16:97–159 Wright S. (1951) The genetical structure of populations. Ann Eugen 15:323–354 Yeaman S, Otto SP (2011) Establishment and maintenance of adaptive genetic

divergence under migration, selection, and drift. Evolution 65:2123–2129 Yeaman S (2015) Local adaptation by alleles of small effect. Am Nat 186:74–89 Yeaman S, Hodgins KA, Lotterhos KE, et al (2016) Climate in distantly related

conifers. Science 353:23–26 Zhang Z, Ersoz E, Lai CQ, et al (2010) Mixed linear model approach adapted for

genome-wide association studies. Nat Genet 42:355–360 Zou J, Wang J, Liu J, et al (2013) Molecular Phylogenetics and Evolution A new

phylogeny for the genus Picea from plastid, mitochondrial, and nuclear se-quences. Mol Phylogenet Evol 69:717–727

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 1899

Editor: The Dean of the Faculty of Science and Technology

A doctoral dissertation from the Faculty of Science andTechnology, Uppsala University, is usually a summary of anumber of papers. A few copies of the complete dissertationare kept at major Swedish research libraries, while thesummary alone is distributed internationally throughthe series Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Science and Technology.(Prior to January, 2005, the series was published under thetitle “Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology”.)

Distribution: publications.uu.seurn:nbn:se:uu:diva-403450

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2020