Applications of gene sequence polymorphisms in evolutionary genetic studies of Atlantic salmon (Salmo salar) and other teleost fish species

Kansi.FH11 Mon Sep 25 14:57:19 2006 Page 1

C M Y CM MY CY CMY K

Applications of teleost gene sequence

polymorphisms in evolutionary genetic studies of Atlantic salmon (Salmo salar) and other species

H

J

R

H

2006

Ryynänen HJ and Primmer CR. 2006. Varying signals of the effects of natural selection during

Ryynänen HJ and Primmer CR. 2004. Primers for sequence characterization and polymorphisms detection in the Atlantic salmon (Salmo salar) growth hormone 1 (GH1) gene. Molecular Ecology Notes, 4: 664-667.

Ryynänen HJ and Primmer CR. 2004. Distribution of genetic variation in the growth hormone 1 gene in Atlantic salmon (Salmo salar) populations from Europe and North America. Molecular Ecology, 13: 3857-3869.

Ryynänen HJ and Primmer CR. 2006. Single nucleotide polymorphism (SNP) discovery in duplicated genomes: intron-primed exon-crossing (IPEC) as a strategy for avoiding amplification of duplicated loci in Atlantic salmon (Salmo salar) and other salmonid fishes. BMC Genomics, 7: 192.

Ryynänen HJ, Tonteri A, Vasemägi A, and Primmer CR. A comparison of the efficiency of single nucleotide polymorphisms (SNPs) and microsatellites for the estimation of population and conservation genetic parameters in Atlantic salmon (Salmo salar). Submitted

Applications of teleost gene sequence polymorphisms in evolutionary genetic studies of Atlantic salmon (Salmo salar) and other species Heikki J. Ryynänen

Applications of gene sequence polymorphisms in evolutionary genetic studies of Atlantic salmon (Salmo salar) and other teleost fi sh species

H

J. R

Department of Biological and Environmental Sciences Faculty of Biosciences

University of Helsinki Finland

Academic dissertation

To be presented, with the permission of the Faculty of the Biosciences of the University of Helsinki, for public criticism in Auditorium 2041 at the Viikki Biocenter 2

(Viikinkaari 5) on October 27^th 2006, at 12 o’clock noon.

Helsinki, 2006

© Heikki J. Ryynänen (Chapter O)

© NRC Research Press (Chapter I)

© Blackwell Publishing (Chapters II, III)

© Authors (Chapters IV, V) Author’s address:

Department of Biological and Environmental Sciences P.O. Box 65 (Viikinkaari 1)

FIN-00014 University of Helsinki Finland

E-mail: Heikki.J.Ryynanen@helsinki.fi

Cover illustration kindly painted by Kirsi Nevalainen ISBN 952-92-1046-9 (Paperback)

ISBN 952-10-3426-2 (PDF) http://ethesis.helsinki.fi Oy Edita Ab

Helsinki 2006

Applications of gene sequence polymorphisms in evolutionary genetic studies of Atlantic salmon (Salmo salar) and other teleost fi sh species H

J. R

The thesis is based on the following articles:

I Ryynänen HJ and Primmer CR. 2006. Varying signals of the effects of natural selection during Teleost growth hormone gene evolution. Genome, 49, 42-53.

II Ryynänen HJ and Primmer CR. 2004. Primers for sequence characterization and polymorphisms detection in the Atlantic salmon (Salmo salar) growth hormone 1 (GH1) gene. Molecular Ecology Notes, 4, 664-667.

III Ryynänen HJ and Primmer CR. 2004. Distribution of genetic variation in the growth hormone 1 gene in Atlantic salmon (Salmo salar) populations from Europe and North America. Molecular Ecology, 13, 3857-3869.

IV Ryynänen HJ and Primmer CR. 2006. Single nucleotide polymorphism (SNP) discovery in duplicated genomes: intron-primed exon-crossing (IPEC) as a strategy for avoiding amplifi cation of duplicated loci in Atlantic salmon (Salmo salar) and other salmonid fi shes. BMC Genomics, 7, 192.

V Ryynänen HJ, Tonteri A, Vasemägi A, and Primmer CR. A comparison of the effi ciency of single nucleotide polymorphisms (SNPs) and microsatellites for the estimation of population and conservation genetic parameters in Atlantic salmon (Salmo salar). Submitted.

These articles are referred to by their Roman numerals in the text

Contributions

HR: Heikki J. Ryynänen, CP: Craig Primmer^1,2, AV: Anti Vasemägi², AT: Anni Tonteri^1,2 The contributions of laboratory assistants are acknowledged in the relevant part of the thesis.

1 Department of Biological and Environmental Sciences, University of Helsinki, Finland

2 Department of Biology, University of Turku, Finland

Supervised by: Professor Craig Primmer

University of Turku

Finland

Reviewed by: Professor Grant Pogson

University of California

U.S.A

Dr Lisa W. Seeb

Alaska Department of Fish and Game U.S.A

Examined by: Professor Outi Savolainen

University of Oulu

Finland

I II III IV V

Original idea CP, HR CP, HR CP HR, CP CP, AV

Methods HR HR, CP HR, CP HR, CP HR, CP, AV

Data gathering HR HR HR HR HR, AT

Analyses HR HR HR HR HR, AT, AV

Manuscript preparation HR, CP HR, CP HR, CP HR, CP HR, CP, AV

Contents

Summary ... 7

Introduction ... 7

Interpretation of DNA sequence variation in evolutionary genetics ... 7

Molecular tools ... 8

Teleosts, salmonidae (salmonid fi shes) family, and Atlantic salmon (Salmo salar) ... 9

Outline of the thesis ... 10

Summary of the original papers ... 11

I – Molecular evolution of the growth hormone gene in teleost fi sh ... 11

II, III – Investigation of genetic variation in the growth hormone 1 (GH1) gene in Atlantic salmon ... 14

IV – Discovery of single nucleotide polymorphisms (SNPs) in the genome of Atlantic salmon and other salmonids ... 17

V – Comparison of SNP and microsatellite markers in population genetic analysis of Atlantic salmon ... 20

Conclusions ... 22

Acknowledgements ... 24

References ... 26

Introduction

Interpretation of DNA sequence variation in evolutionary genetics Evolutionary genetics is the fi eld that resulted from the integration of genetics and the Darwinian evolution theme and broadly incorporates traditional population genetics and studies of the origins of genetic variation by mutation and recombination, and the molecular evolution of genomes.

Population genetics itself endeavours to enlighten the basic questions in evolution biology, such as what determines genetic differences within and among individuals and populations, including those that lead to adaptation and speciation. The primary evolutionary forces that affect the genetic variation within and among populations are genetic drift, gene fl ow, mutations and natural selection. Given very long periods of time, these forces among populations will eventually give rise to the evolutionary patterns that characterize the higher taxonomic groups. Thus the central challenge in understanding the genetic basis of evolutionary changes is not only to describe the adaptive forces that drive the selection that causes existent DNA sequence variants but also to identify the nucleotide differences responsible for the observed heritable phenotypic variation.

Molecular evolution describes and analyses variation and evolution in the structure and nucleotide sequences of the genes, and also provides molecular

tools for exploring many questions about the evolution of organisms. The neutral theory of molecular evolution posits that the majority of DNA variation within and between species is neutral with regard to individual fi tness (Kimura, 1983). Natural selection is assumed to rapidly eliminate the deleterious mutations that presumably contribute little to variation within or between species. Furthermore, benefi cial mutations are expected to be extremely rare and thus have a minor role in observed patterns of DNA sequence variation.

However, the force of mutation is the ultimate source of new heritable genetic variation within populations and is thus the raw material for natural selection and the other evolutionary processes. Different types of mutations range from single base- pair substitutions to the duplication of entire chromosome sets and vary in their impact from no effect to single amino acid change or even genome duplication. In addition, the mutations in the coding regions of the genes can be divided into silent-site (synonymous, silent) and replacement (non-synonymous, amino acid replacing) substitutions. To understand the effects of various mutations, interpretation of genetic differentiation has been the principal basis of various genetic studies.

Among the evolutionary forces, natural selection has long been privileged in evolutionary studies because of its crucial role in adaptation. In evolutionary genetics, patterns of polymorphism in the

Summary

HEIKKI J. RYYNÄNEN

University of Helsinki, Department of Biological and Environmental Sciences, PO Box 65, FI-00014 University of Helsinki, Finland

genome, together with several neutrality tests (reviewed by Kreitman, 2000; Ford, 2002), have provided the fi rst means of detecting the types of genes that have been subjected to natural selection based on their level of diversity being greater than expected under neutrality. A more stringent and robust method for detection of adaptive molecular evolution in protein- coding genes is a higher rate of non- synonymous (d_N) relative to synonymous (d_S) substitutions with the ratio being higher than one. Based on this theory, several statistical methods have been developed to reveal the evidence of Darwinian selection at individual gene sites and lineages in various organisms (see Yang &

Bielawski, 2000 for a review). Numerous other research strategies are available to identify functional genetic variation in a wide range of organisms (reviewed by Vasemägi & Primmer, 2005). The growing list of genes with observed signals for positive selection (see Endo et al., 1996;

Ford, 2002) suggests that the adaptive Darwinian selection may be more common than previously estimated. Among teleost fi sh species, however, positive selection at the gene level has only been reported in a few genes, such as the pantophysin gene in Atlantic cod (Pogson, 2001; Pogson &

Mesa, 2004), the immunoglobulin gene in Antarctic icefi sh (Ota et al., 2003) and the transferrin and MHC genes in Salmonids (Ford et al., 1999; Ford, 2000, 2001;

Landry & Bernatchez, 2001; Langefors et al., 2001; Lohm et al., 2002).

The continuous growth of nucleotide sequence information in public databases, such as GenBank (Benson et al., 2004), from a vast range of organisms, including various teleost species (over 4 million sequence entities in GenBank in September 2006), makes possible different levels of investigations of gene sequence

polymorphisms in fi shes. Extensive comparative studies of molecular evolution can be made by harvesting the existing sequence data in the databases, or different applied investigations of the species of interest can be performed by utilizing the sequence data from the target organism or sister taxa. Furthermore, genetic variation within species has mainly been assessed with traditional neutral genetic markers, but this variation is not necessarily a good indication for the adaptive evolution potential or differentiation of populations (Reed & Frankham, 2001). Thus the development and evaluation of new gene- targeted markers is without doubt an important issue in evolutionary genetics as these can been used more directly to investigate quantitative genetic variation (see van Tienderen et al., 2002; Andersen

& Lubberstedt, 2003; Varshney et al., 2005 for reviews).

Molecular tools

DNA sequencing provides the most detailed genetic information from the analyzed region. Sequence determination for a given genomic region from multiple individuals can be easily utilized to investigate variation either within or between species, and several analytical methods are well-established for DNA sequences (Kreitman, 2000). The disadvantages of sequencing are the relatively expensive technique and the fact that a large number of invariant sites are sequenced. Therefore to assess only polymorphisms, variation at a given site has often been determined with a genetic marker. Several different classes of molecular markers have been applied to investigate various biological questions, such as population genetics and genomics, phylogenetic relationships, gene mapping, and individual relatedness (see Luikart et al., 2003; Zhang & Hewitt,

2003; Schlötterer, 2004 for reviews). Until recently, the main tools for molecular studies of ecology and evolution have been microsatellites and mitochondrial DNA, which have, however, some limitations in their use for certain analyses (Morin et al., 2004). Single nucleotide polymorphisms (SNPs) express some benefi cial characteristics compared with other markers and are thus regarded as an alternative tool for evolutionary genetic studies (Vignal et al., 2002; Brumfi eld et al., 2003; Morin et al., 2004). Regarding this thesis, gene sequence polymorphisms between species, together with SNPs and indels (insertion or deletion), were the most investigated and utilized type of molecular variation - microsatellites principally being used in a comparative manner.

A single nucleotide polymorphism (SNP) can be regarded as just a single base change in the DNA sequences of the population. SNPs are the most common type of DNA variation, constituting approximately 90% of genetic variation in humans, and their average abundance in the genome is about 1 SNP per 1,000 base pairs (bp) in non-coding and 1 per 500 bp in coding a DNA stretch (see Brumfi eld et al., 2003 and references therein). It is typically assumed that each SNP is the result of a single mutation event and that different SNPs segregate independently of one another. In practice, SNPs are usually biallelic, with an alternative of two possible nucleotides at a given position. A base position in a DNA stretch that can be considered an SNP usually requires that the least frequent allele should have a frequency greater than 1% in the population (Vignal et al., 2002). One base pair of indels are sometimes considered SNPs, although they, together with longer DNA fragment indels, certainly arise with different mutation mechanisms, such as

due to errors in DNA replication. One promising application of SNPs is their use as gene-targeted markers following the fact that they can occur either in or around a particular gene. These kinds of markers may have several utilizations, such as in the determination of genetic variation in a limited set of genes that might also have an ecologically relevant role for a species (van Tienderen et al., 2002). Thus far, however, the main restriction in the use of SNPs in large-scale evolutionary genetic studies of many wild species has been the paucity of identifi ed markers in the organism of interest.

Microsatellites are highly polymorp- hic stretches of tandemly repeated 1-5 bp DNA elements (e.g. …TCTCTC…), which are often highly variable in their repeat number (e.g. Ellegren, 2004). They are mostly regarded as neutral genetic markers occurring in the non-coding DNA regions with no further effect on the phenotypic level (but see Li et al., 2002;

Li et al., 2004), but their neutrality can be affected by natural selection in the fl anking region of the target loci, known as ‘genetic hitchhiking’ (Maynard Smith & Haigh, 1974). Numerous reviews have introduced their utilization in various ecological studies (Jarne & Lagoda, 1996; Estoup &

Anglers, 1998; Goldstein & Schlötterer, 1998; Balloux & Lugon-Moulin, 2002), and several statistical analyses have been developed for microsatellites enabling inferences of different evolutionary and population parameters (see Luikart &

England, 1999 for a review).

Teleosts, salmonidae (salmonid fi shes) family, and Atlantic salmon (Salmo salar)

The ray-fi nned fi shes comprise ~23,700 extant species (Nelson, 1994) and are the largest and most diverse group of

vertebrates. They can be further subdivided into the basal ‘non-teleosts’ and the higher teleosts, which consist of about 23,600 species, thus constituting >99% of living fi shes (Venkatesh, 2003). The most ancient teleost fossil is ~235 million years old (Maisey, 1996), and after that the teleosts appear to have undergone a rapid radiation, which is an unparalleled event among vertebrates (Venkatesh, 2003). Several teleost fi sh species have been extensively studied at the genetic and genomic level, and some of these have also been subjected to whole-genome sequencing projects (see Volff, 2005 for a review).

The Salmonidae (including e.g. genera Salmo, Salvelinus and Thymallus) family is a commercially highly valuable and biologically interesting family of fi shes.

The entire family has been attributed to an ancestral genome duplication event (Allendorf & Thorgaard, 1984; Venkatesh, 2003), and the great diversity of species in the Salmonidae family has been proposed as an example of species radiation that followed genome duplications (Taylor et al., 2001). The duplication generates plenty of paralogous sequences, which may have various effects on a species (Venkatesh, 2003), and this topic is stressed several times in this thesis with regard to the salmonid fi shes.

The Atlantic salmon (Salmo salar), together with rainbow trout, is one of the most extensively studied salmonid species due to its importance for fi sheries, aquaculture and recreational fi shing. The natural distribution range of Atlantic salmon spans the Atlantic Ocean from the east coast of North America to Europe, and the species owns both anadromous and non-anadromous (landlocked) migrating behaviours. Based on neutral molecular markers, a clear population division in Atlantic salmon has been reported between

continents (e.g. Ståhl, 1987; Bermingham et al., 1991; McConnell et al., 1995;

Verspoor & McCarthy, 1997; King et al., 2001), within European populations (e.g.

Ståhl, 1987; Bourke et al., 1997; Koljonen et al., 1999; Nilsson et al., 2001; Säisä et al., 2005; Tonteri et al., 2005), and even on a very small geographical scale (e.g. Elo et al., 1994; Garant et al., 2000; Primmer et al., 2006). Previous evolutionary genetic studies with Atlantic salmon have focused on investigating genetic variation related to adaptation (Landry & Bernatchez, 2001) immune defence (Langefors et al., 2001; Lohm et al., 2002), and different behavioural performances (Fontaine

& Dodson, 1999; Landry et al., 2001;

Tiira et al., 2003). In addition to several population and evolutionary genetic studies of this species, recent large-scale sequencing projects of Atlantic salmon (http://web.uvic.ca/cbr/grasp/; http://

www.salmongenome.no/cgi-bin/sgp.cgi) have provided a vast amount of sequence data for various other types of explorations of Atlantic salmon genome, such as identifying potential gene-associated polymorphisms (Vasemägi et al., 2005).

Outline of the thesis

This thesis utilizes gene sequence variation to investigate molecular evolution, population genetics and natural selection in different teleost species with a special focus on Atlantic salmon. The thesis begins with a broad study of the molecular evolution of the growth hormone (GH) gene in teleost (I), investigating the level of Darwinian adaptive evolution affected coding regions of this gene during the teleost evolution. Second, the thesis uses a target gene approach on a species level to identify within-population variation

in the growth hormone 1 (GH1) gene of Atlantic salmon in order to assess the major evolutionary forces affecting genetic variation at this locus (II, III). The thesis then extends the focus on the family level and characterizes single nucleotide polymorphism (SNP) variation in several genomic regions of salmonid fi shes, with a special emphasis on Atlantic salmon, and also introduces a new strategy for SNP detection in species owning potentially duplicated genetic fragments (IV). The last chapter of the thesis further evaluates the usefulness of a limited number of SNP markers as molecular tools in several applications of population genetics in Atlantic salmon by comparing the data obtained from SNP and microsatellite markers (V).

Summary of the original papers

I – Molecular evolution of the growth hormone gene in teleost fi sh

The pituitary growth hormone (GH) is the key protein responsible for the regulation of somatic growth and many aspects of metabolism in all vertebrates (Davidson, 1987; Harvey et al., 1995). Due to this importance, the molecular evolution of GH has been extensively studied in a large number of vertebrate species, including fi sh (Wallis, 1996; Lioupis et al., 1997;

Wallis & Wallis, 2001; Wallis et al., 2001;

Forsyth & Wallis, 2002). In teleost fi sh, a typically slow basal rate of GH evolution has been interrupted by more frequent bursts of rapid evolution, an observation that demonstrates an exception in teleosts compared with other vertebrate groups (Wallis, 1996). This, together with the observed high divergence in the amino acid sequences of the teleost GH gene (Bernardi et al., 1993; Rubin & Dores,

1995; Wallis, 1996), led to a hypothesis to identify signs of Darwinian adaptive evolution in the coding region of the GH gene by utilizing recently introduced statistical methods (see Yang & Bielawski, 2000) that had not previously been used for molecular evolution study of teleost GH.

We investigated the Darwinian adaptive evolution during the molecular evolution of the GH gene within and between different teleost fi sh taxa by utilizing maximum-likelihood models of codon substitutions and two sliding- window-based methods (I). Our analyses were based on complete coding GH sequences of 54 teleost species and the sequences were classifi ed into four taxonomic orders: cypriniformes, perciformes, salmoniformes and siluriformes (I). The rates of molecular evolution were estimated within and between each teleost order (I). The different selective constraints along the GH gene sequences were examined by determining the d_N and d_S rate variation and d_N/d_S (called ‘ω’ ratio) for different lineages and also for individual codons within lineages (I).

The average amino acid divergence in the mature GH protein was 29% for the entire teleost data and the variation in divergence times of different teleost orders was about twofold, salmoniformes showing the youngest age of divergence (Table 1, I). The overall evolution rate for GH protein was 1.15 ± 0.01 x 10^-9 substitutions/aa site/year and the variation within different orders was about fourfold (Table 1, I).

No signals of positive selection were observed for the main branches of the teleost with lineage-specifi c ML analysis but the ω values varied between 0.11 - 0.16, supporting the existence of purifying

Table 1. Divergence time estimates, pair-wise amino acid (aa) divergences and rates of evolution for GH gene in different teleost orders (I). Average amino acid differenceb Substitutions/aa site/year (x 109 ) (±S.E.M) OrderEstimated divergence time (My)a Mature protein Signal peptide Mature protein Signal peptide Cypriniformes 108 16.45 (0.088) 2.15 (0.102) 0.81 ± 0.05 0.91 ± 0.10 Salmoniformes 57 8.00 (0.043) 0.83 (0.038) 0.75 ± 0.07 0.66 ± 0.09 Siluriformes 103 5.82 (0.033) 2.14 (0.097) 0.32 ± 0.03 0.95 ± 0.08 Perciformes 89 21.71 (0.117) 3.21 (0.189) 1.30 ± 0.05 1.90 ± 0.06 All 235c 52.74 (0.293) 8.54 (0.467) 1.15 ± 0.01 1.58 ± 0.02 a) The estimates refer to the divergence times for a particular teleost lineage from its most recent common ancestor based on cytochrome b sequence data (I). b) The numbers in brackets refer to the proportion of aa changes averaged across all pair-wise comparisons. c) The oldest fossil record of a teleost fish (Maisey, 1996).

selection in these lineages (Figure 1, I).

However, the statistical sliding-window- based method indicated some positively selected regions in the salmoniformes branch, suggesting the possible occurrence of adaptive evolution in this lineage (I).

When the test for positive selection was performed at the codon level within different orders, one codon site in salmoniformes and two sites in siluriformes were indicated as putative positively selected sites (I). The same regions also exhibited elevated rates

of non-synonymous substitutions in the sliding window analysis, weakly suggesting the existence of the Darwinian adaptive evolution in these codons (I).

The majority of the amino acid sites were characterized by very low ω ratios, indicating the predominance of purifying selection as a major selection force in GH gene (I).

The rate of molecular evolution of the GH gene varied between orders but was, however, within the range of previous estimates (Wallis, 1996). This study

Figure 1. A phylogeny of the growth hormone gene for the three teleost orders used to analyse Darwinian adaptive evolution between lineages (I). Perciformes was excluded from the lineage- specifi c analysis due to a high level of saturation observed in synonymous substitutions (see publication I for further information on the species analyzed). The topology is the 60% majority- rule consensus tree based on the mRNA sequences and maximum parsimony method with 100 bootstrap implemented in PAUP* (Swofford, 1998). The scale represents nucleotide substitutions.

The numbers next to the branches refer to the values of ω for the particular lineage (I).

could not show unambiguous evidence of positive selection acting on the teleost GH gene during teleost evolution. This supports earlier fi ndings that most genes typically evolve under stringent purifying selection (Endo et al., 1996), and the observed variation in ω values across different codons of the GH gene is probably due to variable levels of functional importance in the amino acid sites. The observed lack of clear Darwinian positive selection acting on the GH gene of teleost fi shes may be explained by the vital structure of this protein (I).

While the majority of the amino acid changes in the teleost GH may be a consequence of relaxed purifying selection, we also speculated on the existence of potential positive selection at some codon sites of the salmoniformes lineage. Two functional isozyme loci of growth hormone genes have been found in salmonid fi sh (McKay et al., 2004), indicating that the duplication event in this teleost order has enabled diverse selection pressures for these paralogs.

Positive selection after gene duplication has been studied in several species (e.g.

Zhang et al., 1998; Hughes et al., 2000;

Conant & Wagner, 2003; Blanc & Wolfe, 2004) but only rarely reported in teleosts (Merritt & Quattro, 2001). However, putative positively selected regions in the salmoniformes lineage weakly support some extent of Darwinian evolution at the GH gene after the divergence of this teleost order (I).

II, III – Investigation of genetic variation in the growth hormone 1 (GH1) gene in Atlantic salmon

The Atlantic salmon, Salmo salar, is a species for which local adaptation has been suggested as maintaining and promoting genetic variation (reviewed by Taylor,

1991; Allendorf & Waples, 1996; Verspoor, 1997). In addition, inter-population variation in numerous growth-related life history characteristics have been reported in salmon (Nicieza et al., 1994; Hutchings

& Jones, 1998; Jonsson et al., 2001).

Previously, associations between growth and sequence polymorphism in growth- related genes have been observed in some hatchery stocks of salmonid fi shes (Gross

& Nilsson, 1999; Tao & Boulding, 2003), and a human study has also demonstrated an association between the GH haplotype and GH expression level (Horan et al., 2003). Given that Atlantic salmon exhibits local adaptation and some evidence of geographical variation in growth parameters has been demonstrated among populations, a growth-associated gene such as growth hormone 1 (GH1) was therefore hypothesized as a good candidate for investigating the intra-specifi c level of polymorphisms at the target gene potentially promoting local adaptation in Atlantic salmon.

The principal purposes of these two studies were to identify intra-specifi c DNA sequence variation in the Atlantic salmon growth hormone 1 gene (GH1), and to investigate the level and geographical distribution of this variation in Atlantic salmon populations from Europe and North America with a special focus on inferring the major evolutionary forces affecting genetic variation at this locus (II, III). We fi rst describe primers for sequence characterization and polymorphism detection in the Atlantic salmon GH1 gene and also introduce secondary assays for ten of the polymorphic sites identifi ed in the GH1 gene (II). The entire GH1 gene was sequenced from nine salmon populations, and about 30 individuals from each population were genotyped by using the polymorphisms detection assays

outlined above to investigate the variation between populations (III). Haplotypes of each individual were inferred from the genotype data and a haplotype network of GH1 haplotypes was constructed (III).

Nucleotide diversities (π or θ_W) were conducted for the entire sequenced GH1 gene, and also for each population (III). In addition, neutrality tests were used to test the null hypothesis that mutational patterns in the GH1 gene within populations were as expected under a model of neutral evolution (Kimura, 1983) (III).

Among the analyzed populations, seventeen polymorphic sites (1/277 bp) were detected in the complete GH1 sequences, all of them being located in non-coding regions (Figure 2, III). The estimated nucleotide diversity (π) for the entire GH1 gene was 14.8 x 10^-4 and varied about twentyfold among coding and non-coding regions (III). At the population level, nucleotide diversities varied sevenfold between populations, and neutrality tests indicated that the null hypothesis of neutral evolution could be rejected in the three Atlantic salmon populations (Table 2, III). Based on the maximum parsimony network, GH haplotypes generally clustered into two main groups, one including the majority

of haplotypes from European and another containing haplotypes mainly from North American populations (Figure 3, III).

However, a small number of haplotypes occurring in the two European populations were more closely related to the North American than the European haplotypes (Figure 3, III).

We observed no polymorphisms in exonic regions of the GH1 gene among Atlantic salmon populations spanning the geographic range of the species, indicating that intra-specifi c GH1 coding sequence variation is very low. This suggests that purifying selection against non- synonymous mutations is the predominant contemporary force controlling the molecular evolution of the GH1 gene (III). The lack of observed association between growth rate and GH1 sequence variation highlights the challenges and importance of the choice of candidate gene when studying complex traits such as growth in wild species. The observed signifi cant association between growth and polymorphis sites in the growth hormone receptor gene of another salmonid fi sh (Tao & Boulding, 2003) suggests that genes other than GH1 can either directly or indirectly be involved in regulating growth in salmonids and may thus also be

Figure 2. Schematic diagram indicating the structure of the Atlantic salmon growth hormone 1 gene (GH1) and positions of polymorphic sites detected in the analyzed nine populations (III).

Black and grey boxes represent exons and introns respectively, and solid black lines indicate 5’- and 3’- untranslated regions. Solid arrows indicate SNPs and dashed arrows represent indels, and those ten sites for which secondary assays were developed to investigate within-population variation are marked with asterisks (II, III).

Table 2. Summaries of diversity indices and neutrality tests for the ten polymorphic sites analyzed for Atlantic salmon GH1 variation at the population level. Abbreviations in parenthesis next to the population name refer to the continent of origin: Europe (Eu), North America (NA). See paper IIIfor a map of location of the analyzed populations. Diversity indicesa Neutrality test statisticsb Population ncNo. of SNPsNo. of IndelsNo. of HaplotypesπθW/ L Rm Tajima’s DFu and Li’s D*Fu and Li’s F* Teno (Eu) 62 5 5 13 0.00093 0.00045 8 2.875 *** 2.233 *** 1.392 ns Pechora (Eu) 64 5 5 14 0.00086 0.00046 7 2.428 ** 2.055 *** 1.393 ns Dee (Eu) 62 5 2 7 0.00057 0.00032 5 1.971 * 1.730 *** 1.230 ns Tornio (Eu) 60 5 3 7 0.00033 0.00038 4 -0.357 ns 0.896 ns 1.298 ns Saimaa (Eu) 62 2 2 4 0.00013 0.00019 0 -0.609 ns -1.227 ns -1.248 ns Sella (Eu) 62 4 3 11 0.00052 0.00032 4 1.531 ns 0.290 ns -0.369 ns Esva (Eu) 60 4 3 12 0.00045 0.00034 4 0.839 ns 0.704 ns 0.491 ns Penobscot (NA) 62 4 4 5 0.00028 0.00036 4 -0.575 ns 0.207 ns 0.544 ns St.Jean (NA) 62 5 4 11 0.00059 0.00041 5 1.183 ns 0.983 ns 0.660 ns a-π - the average number of nucleotide differences per site; θW/ L- the proportion of segregating sites in a sample corrected for sample size per site; Rm – the minimum number of recombination events. b- ns, not significant; * P < 0.05; ** P < 0.01; *** P < 0.001. c- number of chromosomes surveyed.

involved in adaptive divergence of Atlantic salmon at the population level (III).

The haplotype network indicated a certain level of differentiation between haplotypes observed in European and North American populations, but an unexpected low level of molecular variation existed between the continents. One potential reason for this decrease in inter-continental differences is explained by the fact that one haplotype commonly observed in the two European populations only differed by one mutation from the North American cluster (Figure 3, III). This fi nding adds further support to the growing evidence that inter-continental gene fl ow has either taken place at some time in the past or is currently taking place (Bermingham et al., 1991; King et al., 2000; Nilsson et al., 2001; Asplund et al., 2004). The gene

fl ow is also supported by neutrality tests as the positive test statistics in the European populations could be due to invasion by one of the two allelic lineages from the North American populations (III).

IV – Discovery of single nucleotide polymorphisms (SNPs) in the genome of Atlantic salmon and other salmonids SNPs represent the most abundant type of DNA variation in all vertebrates’ genomes, and special interest is being taken in their application as genetic markers in numerous studies of the molecular ecology of natural populations (Brumfi eld et al., 2003;

Morin et al., 2004). While numerous SNP identifi cation studies have been established in several model organisms, like humans (e.g. Wang et al., 1998), Drosophila (Hoskins et al., 2001) and

Figure 3. Statistical parsimony network of 43 different haplotypes inferred from Atlantic salmon GH1 sequences of nine populations. The size of each circle is proportional to the relative frequency of the haplotype in the total sample and solid lines connecting each haplotype represent a single mutational event. Black dots represent missing or theoretical haplotypes. Due to the potential occurrence of recombination (Table 2), this network cannot be assumed to represent a gene genealogy, but rather refl ects the approximate patterns of relatedness between haplotypes (III).

Arabidopsis (Schmid et al., 2003), similar studies with wild species such as Atlantic salmon have been scarce. Usually, SNP discovery has been performed in a new species by a targeted gene approach, where several individuals have been sequenced by designing primers in conserved regions of orthologous gene sequences from sister taxa, generally termed the ‘exon-primed intron-crossing’ (‘EPIC’) method (Palumbi

& Baker, 1994). However, the suggested duplicated nature of the salmonid genome (Allendorf & Thorgaard, 1984) may hamper SNP characterization if the primers designed in conserved gene regions amplify multiple loci (see e.g. Smith et al., 2005).

Thus far, to our knowledge, no study has been utilized non-coding segments of the genes, such as introns, to design primers in more variable regions in attempts to avoid amplifying both of the duplicated loci.

The principal aim of this study was to identify potential SNPs in the Atlantic salmon genome using the gene sequence data for salmonids and other teleost species existing in GenBank (Benson et al., 2004). PCR primers for SNP identifi cation were fi rst designed by using

this ‘EPIC’ approach but after observing that numerous duplicated genes had likely been amplifi ed, a new method - termed intron-primed exon-crossing (IPEC) - was developed (IV). SNPs were identifi ed by sequencing 69 genetic fragments with different priming strategies from 15 salmon populations, and a subset of primers was tested also in cross-species applications (IV). Locus-specifi c and overall nucleotide diversities (θ) were estimated based on the successfully sequenced loci (IV).

A total of 47 EPIC loci were designed for Atlantic salmon and of these only 10 (21%) were successfully sequenced, whereas the success rate of the new IPEC approach, 77% (17 out of 22 loci), was signifi cantly higher (P = 0.005) (Table 3, IV). In addition, the proportion of loci in which polymorphism was identifi ed was considerably higher in the IPEC- derived sequences (Table 3, IV). A total of 12,911 bp of high-quality sequences were obtained from 27 loci, and a total of 19 polymorphic sites were observed, resulting in an average of one SNP per 680 in the Atlantic salmon genome (IV). The overall nucleotide diversity for Atlantic Table 3. Summary of the success of candidate SNP loci identification with different priming

approaches in Atlantic salmon (IV).

PCR primer design strategy

Process description EPIC I ^a EPIC II ^b IPEC ^c Total S. salar

No. loci tested 22 25 22 69

No. loci producing clear PCR product 13 1 ^d 21 36

No. loci successfully sequenced 4 6 17 27

No. polymorphic loci 0 3 8 11

a – primers in exons of salmonid genes.

b – primers in exons of other teleost genes.

c – at least one primer in intron regions of salmonid genes.

d – most of the primers produced several PCR bands and thus re-amplifications were needed (see IV).

salmon was 3.99 x 10^-4, but the distribution of polymorphisms among the sequenced loci was highly uneven with no variation observed in about 60% of loci (Figure 4, VI).

Exploitation of the continuously increasing amount of gene sequence data in public databases is a very useful approach to characterize new polymorphic loci from the genomes of non-model organisms. However, as observed in this study, polymorphisms can be biased towards a relatively small portion of loci (Figure 4) and this, together with the lack of information per locus (e.g. Glaubitz et al., 2003), requires increased screening effort in order to identify a suffi cient number of novel SNPs. The results of this study suggest an effective alternative for SNP discovery in species with duplicated genome fragments as the proportion of loci to screen in order to identify polymorphic loci was around six times higher with the new IPEC than the EPIC method (Table 3, IV). The reduced success of the EPIC

Figure 4. Frequency distribution of nucleotide diversities (θ) observed in the sequenced loci investigated in Atlantic salmon (see paper IV for further information on these loci).

approach compared with the IPEC method in SNP discovery is most likely due to the duplicated nature of the salmonid genome (Allendorf & Thorgaard, 1984) (IV).

The observed nucleotide diversity estimate over all loci in Atlantic salmon was in congruence with the study on European humans (Wang et al., 1998), but the estimations are about one-tenth of that reported in birds (Primmer et al., 2002) or plants (Dvornyk et al., 2002) and about three times less than reported in the study of the GH1 gene of S. salar (Ryynänen & Primmer, 2004, III). Low nucleotide diversity in Atlantic salmon is further supported by the observation that about 60% of all analyzed loci showed no variation (Figure 4, IV). One potential explanation for the lower level of sequence variation in Atlantic salmon could be a consequence of the species’

relatively young population history in its present habitats in the northern hemisphere following the last period of glaciation (Ståhl, 1987) (IV).

0.6

Proportion of loci

θ x 10⁴

0.5

0.4

0.3

0.2

0.1

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8

0

V – Comparison of SNP and

microsatellite markers in population genetic analysis of Atlantic salmon Single nucleotide polymorphisms (SNPs) have potential as an alternative molecular tool in population genetics due to their high abundance and broad coverage in the genomes together with effective genotyping procedures (Morin et al., 2004). SNPs also make available a new type of gene-targeted markers that can subsequently contribute to various ecological studies of the diversity and conservation of wild species (van Tienderen et al., 2002). However, especially in non-model organisms, the effort required to develop a large number of SNPs is still substantial (see e.g. IV), and the paucity of information per SNP needs to be compensated by increasing the number of loci analyzed. A simulation study showed that at least a fi vefold number of SNPs are needed compared with microsatellites to reliably determine

genetic relationships between human populations (Glaubitz et al., 2003), but no empirical studies have assessed the utility of SNPs in various population genetic aspects of wild species with a special focus on investigating the required number of SNP for obtaining useful estimates.

In order to investigate whether a limited number of SNPs can provide valuable information for population genetic analyses compared with microsatellites we performed a comparative study by genotyping the identical Atlantic salmon individuals from 21 populations around Northern Europe with seven previously identifi ed SNP loci (from papers II-IV) and 14 microsatellites (V). Based on various population genetic parameters for both datasets, several analyses were performed, including phylogenetic, isolation-by- distance and conservation value analyses, to compare the congruency between the signals obtained from these different

Figure 5. Pair-wise F_ST estimates measured over seven SNPs and 14 microsatellites (Mantel’s r_xy

= 0.652, P = 0.001, r² = 0.425) for 21 Atlantic salmon populations (V).

markers. Furthermore, a statistical method was applied to test for potential signatures of selection acting on these SNP and microsatellite loci (V).

Overall, a signifi cant correlation between the genetic variability (H_e) estimates of different marker types was observed, and population pair-wise F_ST estimates calculated over seven SNP and 14 microsatellite markers also showed a highly signifi cant correlation (Figure 5, V). Furthermore, an association between genetic divergence (F_ST) and geographical distances revealed a strikingly similar isolation-by-distance (IBD) signal from both SNP and microsatellite data (V).

We also observed that the SNP data

alone was not suffi cient for phylogenetic analysis and only some minor alterations occurred when combined together with microsatellite data (V). The estimated conservation priorities for seven different geographical groupings showed highly correlated estimates between the SNP and microsatellite datasets (Figure 6, V), and similar correlations were observed when population contributions to the total gene diversity were assessed separately (V).

Furthermore, some putative candidate outlier loci were identifi ed with a multi- locus simulation test for neutrality (see paper V for further information on these analyses).

Figure 6. Conservation priority values estimated for seven different geographical groupings using seven SNPs and 14 microsatellites (Pearson’s r = 0.964, P = 0.0005) (V). The error bars refer to 95% confi dence intervals. The conservation value is estimated as GD - the probability that the set of taxa preserves more than one allele per site (see Crozier & Kusmierski, 1994) – and the estimates from different population groupings are compared with the information obtained from all populations. See paper V for information on the different groupings of Atlantic salmon populations.

In this study we observed that seven SNPs performed highly similar signals in most of the population genetic analyses when compared with the data obtained using 14 microsatellite markers (Tonteri et al., 2005). Thus far, the lack of identifi ed SNP markers has been the main restriction on their use in the ecology and population genetics of non-model species (Morin et al., 2004), but the results here shed new light on the discussion on the minimum number of SNP markers required for certain analyses. Even if the need has been shown to be at least fi vefold for genetic assignment analysis in humans (Glaubitz et al., 2003), the results here indicate that a lesser number seems to be adequate for many other analyses (V).

The correlation in pair-wise F_ST estimates between markers showed the usefulness of SNPs for estimating the genetic differentiation in salmon populations, and the potential bias in the level of estimates could be standardized over the loci owning different mutation rates (see Hedrick, 2005). Another application comes from the fi nding of high congruency in the isolation-by-distance analysis. Our data also showed that seven SNP loci alone were not suffi cient for phylogenetic analysis, presumably due to the low number of alleles per SNP loci together with a limited number of independent loci (Kalinowski, 2005) (V).

We also assessed the genetic diversity content of SNP and microsatellite loci from a conservation point of view and the congruence between the results obtained from different molecular markers was high (Figure 6, V). This may have useful applications when estimating genetic diversity in genes having a potential role in ecological and conservation issues (van Tienderen et al., 2002). A further advantage of SNPs over microsatellites,

e.g. in forensic studies, is that they can be amplifi ed from highly degraded DNA (Budowle, 2004). Finally, we detected some putative outliers in the analyzed SNP and microsatellite loci among Atlantic salmon populations. Subsequent tests are, however, required to confi rm these results as the number of gene-targeted SNP loci here was inadequate for performing other large-scale SNP outlier tests (V).

Conclusions

To the best of our knowledge, this thesis provides the fi rst comprehensive study of the molecular evolution of the teleost growth hormone (GH) gene and reveals that even if the nucleotide variation is considerably high in the coding sequences of GH between and within different teleost orders, most of these variations are presumably due to a relaxation of purifying selection rather than direct evidence of positive Darwinian selection (I). A similar evolutionary study has been conducted with the growth hormone in primates, which presented some evidence of positive selection but speculated that the relaxation of purifying selection may have played a role in the rapid evolution of the GH gene in simians, probably as a result of multiple gene duplications (Liu et al., 2001). Interestingly, this thesis also suggests some potential positively selected sites in the teleost lineage salmonidae, which own two functional isozymes of growth hormone genes (McKay et al., 2004 and references therein). Hence a subsequent adaptive radiation investigation of these GH paralogs would be interesting to clarify the role of positive Darwinian selection in the evolution of duplicated genes in salmonids (I).

Papers II-III described the targeted- gene approach in an evolutionary genetic study of Atlantic salmon, and mainly reported that polymorphism in the growth hormone 1 gene is not responsible for the local adaptation of different populations.

Furthermore, the intra-specifi c variation in the coding regions of the GH1 gene was very low, most probably due to strong purifying selection against deleterious mutations as an alternative explanation of the generally low mutation level in the GH1 gene being eliminated by the observed higher nucleotide diversity in the non-coding regions. Overall, the results of these studies underline the importance of the choice of candidate gene when studying complex and potential multi- gene traits in the target organisms (III).

Haplotype clustering of the GH1 gene from European and North American populations revealed a certain level of differentiation between continents, with some discrepancies in the overall haplotype distribution pattern. These results, together with the positive neutrality test statistics in some populations caused by potentially invasive allelic lineages, were regarded as support for the growing suggestion that an historical or contemporary intercontinental gene fl ow exists in Atlantic salmon. The inclusion of Icelandic populations in future investigations would therefore be an interesting topic to clarify the rate and origin of this gene fl ow in the different continents (III).

The main fi nding of Paper IV should provide an applicable method for the identifi cation of novel SNP markers, especially in target species presumably owning a duplicated genome or gene fragments. The new IPEC (intron-primed exon-crossing) approach suffered signifi cantly less from the duplicated nature of the salmonid genome (Allendorf

& Thorgaard, 1984), most likely due to the improved discriminative priming strategy between putative paralogs. The reason for this improvement is probably the fact that the divergence in introns is presumed to be higher as a consequence of less stringent selection in non-coding regions (IV). The new IPEC approach would also be a useful tool to shed new light on the number of duplicated genes retained within the salmon genome; the current estimates are that approximately 50% of protein loci exhibit duplication (Allendorf

& Thorgaard, 1984; Smith et al., 2005).

The SNP frequencies and the nucleotide diversities of Atlantic salmon were within range of the estimates from a number of organisms (see Brumfi eld et al., 2003 and references therein), but the variation between loci was high. A more extensive survey of the nucleotide diversities among different genome regions would be benefi cial as this study revealed a highly uneven distribution of polymorphisms among gene fragments of Atlantic salmon (IV).

The last chapter, V, reported an interesting finding in that a limited number of SNPs can provide reliable information for some population genetic analyses in Atlantic salmon. While the present study indicated there are still noticeable limitations on the use of SNPs for phylogenetic analysis, seven SNPs performed surprisingly well in other analyses compared with 14 microsatellites in Atlantic salmon. A recent comparative study in salmon showed the usefulness of 26 SNPs in parental testing and genetic assignment analysis (Rengmark et al., 2006), but here more than three times fewer loci were adequate to give comparable signals to microsatellites when analyzing genetic divergence, isolation-by-distance and conservation values of different

populations. The results are especially worth noting when analyzing species with a limited number or no available genetic markers as a set of SNPs can be identifi ed using the gene sequence information from closely related species (e.g. IV). A further use of SNPs as a marker of choice might be in applications with diffi cult biological samples due to their more successful performance with highly degraded DNA (Budowle, 2004). Hopefully, these fi ndings will encourage the application of SNP markers in addressing diverse questions related to various aspects of population genetics of wild species (V).

Acknowledgements

First and foremost I would like to thank my supervisor, Craig Primmer, for his excellent guidance thorough the whole of my PhD work. I am especially grateful that you always had time for me, despite my having small technical questions or needing more thorough discussion of my PhD project. Your incredible ability to always fi nd positive sites in all the setbacks I faced during this journey taught me how to make science, and several times you made me see the wood for the trees.

Thanks!

My warmest thanks go to former and present members of the PnP group: Mikko Koskinen, Leena Laaksonen, J-P Vähä, Anni Tonteri, Laura Buggiotti, Paula Lehtonen and Anti Vasemägi. It has always been fruitful to discuss my projects with you and receive many suggestions and new ideas, and outside work I have spent numerous nice times with you, which has further strengthened our friendship.

Paula and Leena, you have provided me with tremendous laboratory assistance in the data gathering. One important lesson

I have learned during this project is what it means to be a member of a group and how much support you can obtain from it. Sometimes you only realize what you have had when you have already lost it.

I’ve been missing you ‘PnP’ since you moved to Turku!

The Department of Ecology and Systematics/Biological and Environmental Sciences has been an extremely nice place to work in. I would like to thank all PhD students for many memorable moments together, and especially Jukka Palo and Hannu Mäkinen, who have helped me a lot with different analytical problems. I also want to thank all those people in the MES- lab for providing an enthusiastic working atmosphere, and give special thanks to Anne Aronta for her excellent laboratory assistance. I thank Ilkka Teräs and Veijo Kaitala for ensuring my studies remained on schedule. I am grateful for my post in the Finnish Graduate School of Population Genetics and I want to acknowledge its head, Pekka Pamilo, for organizing all those interesting courses and other events.

For helping me to forget work, I want to thank all my friends, especially Tuomo, Mika L. and the Kukko-Klubi guys, for friendship over years and all those social and cultural activities together. Big thanks to the Copa Callio and BIKO teams for enjoyable company in sport. Fishing and hunting have always been very important leisure activities for me and with you, Mikko, J-P, Jukka, Heikki, Tero, Timo, Hannu, Sampsa and the Nevalainen brothers, it has been a pleasure to share all those unforgettable experiences in the great outdoors.

Haluan kiittää vanhempiani Aini ja Lauri Ryynästä sekä siskoni perhettä Kirsi, Jukka, Ville ja Eetu Nevalaista kaikesta siitä näkyvästä ja näkymättömästä tuesta, minkä olette minulle suoneet näiden

vuosien aikana. Läheisillä suhteillamme on ollut minulle todella suuri merkitys työni aikana ja olette aina jaksaneet kannustaa minua väitöskirjatyössäni ilman että olen välttämättä osannut selittää sen sisältöä teille selkokielellä. Kiitos myös Kaarina ja Atte Maunolle ymmärryksestänne työtäni kohtaan.

Finally, my warmest gratitude goes to Marika and Ella, the two most important people in my life. Marika, you have always supported me in my decisions and were ready to share my good and bad moments during this journey. Without your love, the completion of this thesis would have

been even harder. Ella, my sunshine, I’ll never forget the day you were born as you immediately changed my values on life. The real delight in your smiley face I always saw in the evenings after coming home kept me going, especially in the last oppressive stages of this work. Thanks ladies!

This thesis was funded by the University of Helsinki, the Finnish Graduate School of Population Genetics, the Finnish Academy, the Finnish Cultural Foundation, and the Emil Aaltonen Foundation.

References

Allendorf FW, Thorgaard GH (1984) Tetraploidy and the evolution of salmonid fi shes. In: The Evolutionary Genetics of Fishes (ed. Turner BJ), pp. 1-53. Plenum Press, New York.

Allendorf FW, Waples RS (1996) Conservation and genetics of salmonid fi shes. In:

Conservation genetics: case histories from nature (eds. Avise JC, Hamrick JL), pp.

238-280. Chapman & Hall, New York.

Andersen JR, Lubberstedt T (2003) Functional markers in plants. Trends in Plant Science 8, 554-560.

Asplund T, Veselov A, Primmer CR, et al.

(2004) Origin of maternal lineages of anadromous Atlantic salmon (Salmo salar L.) in the Barents Sea and White Sea rivers.

Annales Zoologici Fennici 41, 465-475.

Balloux F, Lugon-Moulin N (2002) The estimation of population differentiation with microsatellite markers. Molecular Ecology 11, 155-165.

Benson DA, Karsch-Mizrachi I, Lipman DJ, Ostell J, Wheeler DL (2004) GenBank:

update. Nucleic Acids Research 32, D23- 26.

Bermingham E, Forbes SH, Friedland K, Pla C (1991) Discrimination between Atlantic Salmon (Salmo salar) of North American and European Origin using Restriction Analyses of Mitochondrial DNA. Canadian Journal of Fisheries and Aquatic Science 48, 884-893.

Bernardi G, D’Onofrio G, Caccio S, Bernardi G (1993) Molecular phylogeny of bony fi shes, based on the amino acid sequence of the growth hormone. Journal of Molecular Evolution 37, 644-649.

Blanc G, Wolfe KH (2004) Functional divergence of duplicated genes formed by polyploidy during Arabidopsis evolution.

Plant Cell 16, 1679-1691.

Bourke EA, Coughlan J, Jansson H, Galvin P, Cross TF (1997) Allozyme variation in populations of Atlantic salmon located throughout Europe: diversity that could be compromised by introductions of reared

fi sh. ICES Journal of Marine Science 54, 974-985.

Brumfield RT, Beerli P, Nickerson DA, Edwards SV (2003) The utility of single nucleotide polymorphisms in inferences of population history. Trends in Ecology &

Evolution 18, 249-256.

Budowle B (2004) SNP Typing Strategies.

Forensic Science International 146s, S139–S142.

Conant GC, Wagner A (2003) Asymmetric sequence divergence of duplicate genes.

Genome Research 13, 2052-2058.

Crozier RH, Kusmierski RM (1994) Genetic distances and the settings of conservation priorities. In: Conservation genetics (eds.

Loeschcke V, Tomiuk J, Jain SK), pp. 227- 237. Birkhäuser Verlag, Basel.

Davidson MB (1987) Effect of growth hormone on carbohydrate and lipid metabolism.

Endocrine Reviews 8, 115-131.

Dvornyk V, Sirvio A, Mikkonen M, Savolainen O (2002) Low nucleotide diversity at the pal1 locus in the widely distributed Pinus sylvestris. Molecular Biology and Evolution 19, 179-188.

Ellegren H (2004) Microsatellites: Simple sequences with complex evolution. Nature Reviews Genetics 5, 435-445.

Elo K, Vuorinen JA, Niemelä E (1994) Genetic resources of Atlantic salmon (Salmo Salar L.) in Teno and Näätämö rivers, Northernmost Europe. Hereditas 120, 19- 28.

Endo T, Ikeo K, Gojobori T (1996) Large- scale search for genes on which positive selection may operate. Molecular Biology and Evolution 13, 685-690.

Estoup A, Anglers B (1998) Microsatellites and minisatellites for molecular ecology:

theoretical and empirical considerations.

In: Advances in Molecular Ecology (ed.

Carvalho GR). IOS Press, Amsterdam, Netherlands.

Fontaine P-M, Dodson JJ (1999) An analysis of the distribution of juvenile Atlantic salmon (Salmo salar) in nature as a function of