Does Arctic Fox Change Color

ane. Introduction

Phenotypic variation that causes individual differences in survival or reproductive success may lead to adaptive evolution by natural selection [1]. Contempo advances in molecular analytical methodologies and the increased availability of genomic data allow us to connect phenotypic variation in traits to their causal genes [ii–4]. This enables usa to directly assess the fitness consequences of genotypic variation and improve our understanding of the eco-evolutionary dynamics in wild populations.

A normally used method for mapping genes for phenotypic traits is to conduct a genome-wide clan study (GWAS) [5]. While being widely used to map genes of human diseases [6], the use of GWAS in wild animal populations is notwithstanding somewhat limited [7,8]. Aside from methodological issues (e.g. sample size, density of genetic markers, relatedness and reproducibility of associations [7–9]), about studies that map genes underlying fitness-related traits discover that these traits are polygenic and thus struggle to detect significant associations between single genetic markers and the trait in question [10–14]. Nevertheless, some studies take shown that GWAS is capable of identifying unmarried genes or genomic regions underlying fitness-related traits in wild populations. Johnston et al. [fifteen] establish the factor underlying polymorphism for horn morphology, an important fitness-related trait, in wild Soay sheep Ovis aries. Likewise, Barson et al. [16] discovered a big result locus explaining variation in age at maturity, a highly variable and fitness-related trait in Atlantic salmon Salmo salar. Recently, another study plant that likewise loci on other chromosomes explain some of the phenotypic variances in maturation fourth dimension in Atlantic salmon, thus showing a polygenic basis nonetheless [17]. The adaptive significance of nib morphology in the different Darwin's footing finches Geospiza is well known and a GWAS was used to document a major effect region on chromosome 1A [18].

Coloration is one of the nigh conspicuous phenotypic traits in animals and has been the discipline of research for decades, if not centuries [xix]. Animal coloration tin can have many different purposes (e.thousand. camouflage, communication) [20] and effects of coloration on fitness have been shown in a wide range of fauna species [21–24]. Considering coloration is such a conspicuous trait, information technology is appealing to solely account differences in fitness to the colour phenotype. However, it is important to keep in listen that at that place might be more to a trait than the phenotype itself. Hadfield et al. [25] even showed that color phenotypes practice not always coincide with genetic patterns. Additionally, an association between coloration and other phenotypic traits, such every bit sexual behaviour, aggressiveness, stress response and energy homeostasis, has been shown in different species, suggesting pleiotropic effects of coloration genes [26,27]. Such covariation raises the question of how well nosotros can predict evolutionary consequences of selection on a phenotypic trait when the genes underlying the trait are strongly linked to or touch on (through pleiotropy) other phenotypic traits that themselves could touch fitness. Cognition well-nigh causes of covariation betwixt (potentially) fitness-related traits is, however, challenging to obtain for wild non-model species and demonstrates the importance of more studies aimed at gaining insight into the genetic architecture of adaptive traits.

The Arctic pull a fast one on Vulpes lagopus is a species with interesting coloration features. It occurs in multiple singled-out fur colour morphs and undergoes seasonal moult [28]. The ii common colour morphs are described equally the white and the bluish morph without intermediate morphs. The third morph, called sandy, is extremely rare. White Arctic foxes accept completely white wintertime fur, whereas their summer fur is mostly dark-brown with lighter ventral sides. The blueish morph is uniformly dark chocolate-brown or charcoal year-round, with a slightly lighter coloration during winter. Fur colour in Arctic foxes appears to be inherited equally a simple Mendelian trait with one autosomal locus, where the blueish morph is a effect of the effect of a dominant allele [29,thirty]. The white color morph makes up over 90% of the global Arctic play a joke on population [31]. Importantly though, the relative frequencies of the ii morphs vary across the species distribution [32–35] and between dissimilar environments [34]. For instance, in Iceland, the observed differences in colour morph frequencies are idea to reflect singled-out selection advantages of the 2 color morphs in unlike habitats [34]. The exact mechanisms underlying the global distribution of Chill flim-flam fur color morphs are however not well studied or understood.

Previous molecular analysis suggested that 2 cysteine amino acrid substitutions within the intragenic region of the melanocortin-1-receptor gene (MC1R) co-segregated with the Arctic flim-flam fur colour morphs [29]. MC1R is known to regulate melanin-based coloration in a wide range of animal species [21,36,37], it is thus not surprising that MC1R may be involved in Arctic fox fur coloration. Nevertheless, the written report by Våge et al. [29] was designed every bit a candidate gene analysis and was not able to detect other genes perchance contributing to the color morphs. With very few individuals and/or unknown genetic structures, the candidate gene approaches may likewise have various pitfalls [38]. MC1R is part of a cistron family where five melanocortin receptors (MC1R–MC5R) share the same melanocortin ligands [26]. Pleiotropic covariation between melanin-based coloration and traits governed by MC2R–MC5R can thus be expected and is, in fact, establish in dissimilar species [26].

For increased knowledge on the adaptive importance of variation in fur coloration, we assessed the genetic architecture of fur coloration and analysed fitness consequences of genetic variation related to this trait in a wild population of Arctic foxes. First, we used a whole-genome association analysis to examine the genetic basis and architecture of fur colour. 2d, we quantified selection on fur color genotypes using measures of fitness that link ecological and evolutionary processes. Finally, we investigated the potential for indirect phenotypic effects of fur colour genes through pleiotropy or physical linkage with other genes, and how these furnishings could impact the observed patterns of fitness and genotype frequencies.

2. Methods

(a) Report species and data collection

In the early twentieth century, the Fennoscandian Arctic fox population was close to extinction. Despite protection since the late 1920s, the species did non recover, which led to the implementation of large-scale conservation actions beyond the Scandinavian peninsula, involving supplemental feeding, alternative of red foxes Vulpes vulpes (the most important competitor of the Chill flim-flam) and a captive breeding plan [39,forty]. The captive breeding programme is based on wild-born Arctic foxes held at a convenance station in Oppdal, Central Norway. Breeding pairs are chosen to correspond all extant Scandinavian subpopulations to maintain genetic diversity. Arctic pull a fast one on data used in this study originate from population monitoring [41] and the captive breeding programme, collected in the period 2007–2019. Complete life histories were available for foxes born at the convenance station and subsequently released (n = 371) and for foxes marked as pups during den surveys (n = 810; electronic supplementary material, tabular array S3). For these foxes, Deoxyribonucleic acid analyses, pit-tagging (RFID tags) and/or ear tagging were undertaken for later identification. Additionally, some foxes (north = 206) were simply identified from scat sampling inside the framework of the Norwegian National Arctic Fox Monitoring Plan [42,43].

The monitoring programme uses molecular tracking to certificate population trends annually and is also used to trace the establishment of Arctic foxes released from the captive breeding programme. Sampling of non-invasive material (faeces and hair) is carried out during winter and bound at known Arctic fox den sites across the species distribution [43]. During the study period, approximately 800 samples were nerveless and analysed each year. Individual identification was carried out past comparison of DNA profiles from samples that could be reliably genotyped to a database of known Arctic fox individuals, including released foxes from the captive breeding programme, pit-tagged pups at the dens and non-invasively identified individuals from previous years.

(b) Single nucleotide polymorphism genotyping, information quality control and genome-broad association report

From ear tissue DNA extracts, we successfully genotyped 701 Arctic fox individuals using a custom Affymetrix Precept 702 m SNP array with 507 000 Chill fox specific single nucleotide polymorphisms (SNPs). Merely autosomal SNPs classified as poly high resolution [44] among our genotyped Chill play a trick on individuals were kept for the analyses (361 289 SNPs), and SNP positions were obtained from an Chill play a trick on reference genome assembly comprising 4048 scaffolds with SNP positions given inside every scaffold [45]. See the electronic supplementary material, S2 for more details on the design of the array and quality command (QC) of the SNP data used herein. After QC, our genomic dataset consisted of 681 Chill play a trick on individuals (562 white, 119 blue) genotyped for 359 218 autosomal SNPs (electronic supplementary cloth, table S1).

A GWAS was used to investigate associations between autosomal genetic markers (SNPs) and the Arctic fox fur colour morphs. The analysis was performed using the GenABEL bundle in R [46] with fur color equally the response variable. A genomic relatedness matrix (GRM) was included in the model to account for relatedness. The indep role of PLINK was used with recommended parameters (50, 5, 2) to create a subset of twoscore 539 unlinked SNPs prior to the GRM calculation, to obtain most accurate relatedness estimates [47]. In the GWAS, a polygenic model including the full GRM was fitted, and a mixed model was used to test for association between Arctic fox fur colour and the genetic markers included in the study. Attributable to genomic inflation (λ = 1.92), p-values were corrected for lambda (electronic supplementary material, effigy S2a). To investigate whether boosted correction for population structure was necessary, we reran the analysis including the first three principal components (PCs) achieved through classical multidimensional scaling. The genomic aggrandizement factor (λ = i.902) and the according quantile-quantile (Q–Q) plot (electronic supplementary material, figure S2b) were virtually unchanged. The three outset PCs explained only ca 6, 5 and three.5% of the total variation in the data. Removing the scaffolds with significant SNPs also removed the skew in the Q–Q plot (electronic supplementary material, effigy S2c, λ = 0.91), indicating that the unusually large number of highly significant SNPs could be generating the big skew and genomic inflation. Additionally, a cluster plot of the first two PCs did not reveal whatsoever structure concerning Arctic fox fur colour or origin (i.e. captive versus wild) (electronic supplementary fabric, figure S3). To business relationship for multiple testing [5], we practical the Bonferroni correction, where the significance level (α = 0.05) was divided by the number of SNPs included in the analysis [48].

To increment the number of individuals genotyped at the fur color factor for pick analyses, we used a recently developed Fluidigm SNP array (I.J. Hagen, O. Kleven, 50.K. Arntsen, J. von Seth, 50. Dalen, N.E. Eide, Ø. Flagstad, H. Jensen 2018, unpublished data). This SNP array included 87 autosomal markers, including the SNP called to stand for the Arctic fox fur color genotype (AX-176934441; see Results). Deoxyribonucleic acid genotyped on this platform was extracted from hair, scat and tissue. 9 hundred and twelve Arctic fox individuals were genotyped using the Fluidigm platform (electronic supplementary material, tabular array S1). Of these, 109 were also genotyped using the Affymetrix SNP array. The AX-176934441 genotype was identical across the ii SNP arrays in all these individuals. Of the remaining 803 individuals merely genotyped using the Fluidigm platform, fur colour phenotype was known for 444 individuals (329 white, 115 blueish). These individuals were used as a relatively contained dataset to verify the association betwixt the top GWAS SNP and fur colour because they were not included in the dataset used for the GWAS.

(c) Selection analyses

Complete life-history information (annual survival and fecundity) were available for 1181 individuals from 2007 to 2018 in the Norwegian subpopulations (electronic supplementary material, effigy S1 and table S3). Individuals were assigned to 1 of v age classes (10 = 1–v). Thirty-v individuals older than 5 years were assigned to historic period class five to ensure sufficient sample size in each historic period class. Annual survival and fecundity were based on a range of sources: (i) ascertainment and trapping during den surveys, (ii) DNA from faeces and hair samples, (iii) Biomark (Biomark, Inc., ID, USA) and Trovan Systems (Trovan Ltd, UK) RFID tag readers at feeding stations, and (iv) records from wildlife cameras. These sources allowed for a dataset with loftier resolution at an individual level. Arctic foxes suffer high mortality during wintertime (Oct–Apr) [49]. Thus, we used pre-breeding demography, with each census covering the period from 1 Apr to 31 March the following year. The beginning of April coincides with the end of the mating flavour. Individual almanac survival in demography year t was recorded as 1 for individuals that were inferred to be alive after 1 Apr in twelvemonth t + 1 (otherwise 0).

Parentage was determined for 1497 individuals with known nativity year and genotype, based on 85 autosomal SNPs, using the Sequoia R package [50] (electronic supplementary material, S4). The final pedigree was used to decide the number of pups that emerged from the den (and were genotyped) for each adult present in a subpopulation in a given year t. Annual fecundity was so adamant as the number of pups that survived to recruit into the next year'due south population (i.e. were live after 1 April next year t + 1). In addition, a dichotomous variable was made which was set to 1 if an individual had been institute to brood in a given year t (otherwise 0). Adults not recorded in the pedigree as parents of any pups in a given year t were assumed not to have produced pups or bred that yr. Because we used recruits as the base for the fecundity mensurate in this pre-convenance demography framework, undetected pups (i.eastward. those that die quickly) do not touch the fecundity analysis. Despite extensive monitoring, it is expected that some observations are not recorded, given that the written report population is a wild population spanning a large area. Nonetheless, recapture rates are loftier with but 10% of the study individuals being missed in one census twelvemonth but reappearing after. The missing information are very likely random and not associated with our measurements. Thus, despite imperfect sampling, we do non expect systematic bias in our results.

(i) Individual fitness

Selection on the fur colour genotype was estimated using a demographic model framework that uses reproductive value weighting to account for age structure and fluctuations in the age distribution [51–53]; see also the electronic supplementary fabric, S7. Using this framework, annual private fettle $(Λ_{i})$ in a given twelvemonth for individual i in age class x was defined as $Λ_{i} = W_{i} / v_{10} = (B_{i} 5_{1} / ii + J_{i} v_{10 + ane}) / {five}_{x}$ [54], where West_i is the individual reproductive value, B_i is the number of recruits produced, J_i is the indicator of survival (i if the individual survived, otherwise 0) and the v'due south are historic period-specific reproductive values estimated from the mean projection matrices for males and females separately (electronic supplementary material, S7 and tabular array S5). The reproductive value weighting ensures that $Λ_{i}$ is an age-contained measure out of individual fitness, such that $E (Λ_{i}) = λ$ , where λ is the multiplicative growth rate of the population [52].

The relationships between fur colour genotype and annual individual fitness were modelled using generalized linear mixed effect models (GLMMs) with Poisson distribution, log link part and random intercepts for subpopulation and year, fitted with the lme4 packet in R [55]. Models were fitted for females and males separately (see the electronic supplementary material, S7 for details). Likelihood ratio tests (LRTs) between models containing but the intercept and models containing genotype equally predictor variable were performed to assess the effect of the genotype on fitness.

(ii) Fettle components

To further investigate causes for any differences in fettle, the relationships betwixt fur colour genotype and fecundity (i.e. number of recruits) and adult annual survival were analysed in divide models. Fecundity was modelled using a zero-inflated Poisson GLMM with log link role (electronic supplementary material, figure S7) and survival was modelled using a binomial GLMM and logit link office. In addition, the relationships between genotype and convenance probability of adults and the recruitment probability of juveniles (i.e. juvenile survival until at least 1 Apr the year following nascence) were modelled using binomial GLMMs with logit link function. The analysis on juvenile recruitment probability was performed on a restricted dataset including only juveniles marked at the dens or released from the breeding station in guild to be certain most their birth yr (n = 597). As a starting point, all fitness component models included genotype and sex every bit stock-still factors and a random intercept for subpopulation. Models with adult fitness components (fecundity, developed survival, convenance probability) included in addition age and age² equally continuous covariates and random intercept for year, while models with recruitment probability included random intercepts for birth yr and den. Interactions between genotype and sexual practice or age were included to test whether the upshot of genotype differed between sexes or changed with age. Statistical significance of the different variables was assessed using LRTs betwixt models with and without the term of interest. In the case of non-significance, these terms were excluded (electronic supplementary textile, table S6). The models were fitted using the glmmTMB R package [56] for zero-inflated models and the lme4 R bundle for the remaining models [55].

(iii) Environmental variables

Arctic foxes in this study were of two origins (wild- or convict-born). In the wild, the reproductive performance (number of litters and litter size) of the Arctic fox is to a large extent driven by food availability, varying strongly through the rodent cycle [57]. Although the Arctic play tricks is well adapted to winter severity and prey scarcity, the duration of snow encompass could possibly explicate geographical variation in the frequency of the two colour morphs [34]. Hence, for private fettle and each of the fitness components, we tested whether the effect of colour genotype depended on rodent phase, duration of snow cover (i.e. outset snowfall and last snowfall) or origin past plumbing fixtures models with an interaction betwixt an ecology variable and genotype, with dissever models for each environmental variable (come across the electronic supplementary material, S7 for further details). Models with individual fitness were fitted for females and males separately. Statistical significance was assessed using LRTs between models with and without the term of interest.

Heterozygosity advantage [58,59] could potentially exist a reason for any differences in the fettle of fur colour genotypes. Hence, genome-wide heterozygosity was calculated for the 689 individuals genotyped on the Affymetrix platform using the GenABEL R packet [46]. Differences in genome-wide heterozygosity were tested using a linear mixed-furnishings model with a Gaussian fault distribution. Fur colour genotype and origin (i.east. convict- or wild-born) were included as fixed factor predictor variables. See the electronic supplementary material, S7 for further details.

To further investigate whether the observed differences in individual fettle coincide with the SNPs establish to exist significantly associated with Arctic play a joke on fur colour, nosotros performed a candidate region GWAS for individual fitness that included SNPs on Chill play a joke on scaffold 11 where significant SNPs were found in the fur color GWAS (details in the electronic supplementary textile, S9).

(d) Gene analyses

BLAST searches [60], using BLAST+ 2.9.0 software [61], were performed to investigate genes located in the vicinity of SNPs that, based on the GWAS, were significantly associated with Chill flim-flam fur colour. An annotated Chill fob genome is nonetheless to be published, just there appears to be high synteny between domestic dog and Arctic fox for large parts of their genomes [62]. Thus, the annotated dog genome CanFam 3.1 [63] was used as the reference genome. Encounter the electronic supplementary fabric, S3 for details.

Genes within 10 kb of pregnant SNPs were analysed for gene ontology (Become) term enrichment using the GOstat tool [64]. The distance of 10 kb was chosen to ensure strong linkage between the SNP and the gene. Owing to the lack of a dog-specific GO-database, the goa_human database was used. p-values for over-representation significance were corrected based on false discovery charge per unit. Furthermore, for genes inside ten kb of a significant SNP that likewise was in high linkage disequilibrium (LD; r ^two ≥ 0.v) with the top SNP, gene functions were investigated using the UniProt knowledgebase [65] and primary literature. These genes were also included in a GeneMANIA network assay [66]. GeneMANIA uses a large dataset of functional association data to analyse relations and known co-expression between genes. GeneMANIA does not include a database for canines, thus the homo database was used.

Unless otherwise stated, all analyses were performed in statistical software R 5. 3.6.1 [67].

three. Results

(a) Cistron mapping

The GWAS revealed a total of 495 SNPs significantly associated with Arctic flim-flam fur color at a Bonferroni-adapted significance level (p < 1.39 × 10^−vii, electronic supplementary material, effigy S4). The significant SNPs, that were located on four different scaffolds of the Arctic fox genome (electronic supplementary material, table S14), were BLASTed against the annotated dog genome CanFam three.ane. We obtained a match in the dog genome for 489 SNPs (486 on chromosome 5 (figure 1a), ii on chromosome 27 and i on chromosome 17, electronic supplementary fabric, table S14). The Smash results also show that the four scaffolds which mapped to chromosome 5 and contain significant SNPs assemble next to each other (effigy onea). A full of 438 SNPs were intragenic in the dog genome, whereas the remaining 51 SNPs were located in intergenic regions. The intragenic SNPs were distributed beyond 97 different genes (electronic supplementary fabric, tabular array S15). An additional 57 genes were found less than 20 kb abroad from a meaning SNP, with 34 of these being closer than ten kb from a pregnant SNP (electronic supplementary fabric, table S15). The positions of significant SNPs on chromosome 5 stretched from 52 617 594 to 76 592 936 bp (figure 1a), a distance that appears to be longer than that of stiff LD in the Arctic fox genome (electronic supplementary material, effigy S12). A total of 379 genes are located in this region of the domestic dog genome.

Figure 1. (a) Plot showing BLAST determined dog chromosome 5 locations of 486 SNPs significant in GWAS of fur colour in Arctic play a trick on. The horizontal lines to a higher place the x-axis and the corresponding numbers show how the different Arctic fox scaffolds Nail to dog chromosome 5. On the y-axis, significance levels of the SNPs in the GWAS are shown on a negative log calibration. Pairwise LD (r ^two) between top SNP AX177333963 and the other significant SNPs is shown by the blue color gradient. All domestic dog genes in the region are shown every bit grey lines at the top. The position of putative causal gene *MC1R* is shown with an orange dot (annotation that the y-axis values do not apply for genes). The dashed horizontal line shows the significance threshold after Bonferroni correction of the GWAS. (b) Predicted fitness (lambda) of the Arctic fox fur color morph genotypes CC (white) and TC (bluish). Whiskers represent 95% confidence intervals of predicted values. Predictions are based on additive GLMMs with genotype as predictor variable and year and subpopulation equally random factors. (Online version in colour.)

The Affymetrix SNP assortment used in this report did not include any SNP located in the intragenic region of the candidate gene MC1R. SNP AX-176934441 was the closest significant SNP (5961 bp upstream; p = 6.7 × 10⁻⁶¹) and was chosen as the diagnostic SNP for the culling genotypes at the MC1R gene in farther analyses. Indeed, at that place was a near-perfect Mendelian relationship between genotypes at MC1R and fur colour phenotypes, where the C allele represented a recessive white fur colour allele, and T a dominant blue fur color allele (878 of 882 CC individuals were white, 221 of 234 TC individuals were blue and nine of nine TT individuals were blueish; electronic supplementary material, figure S6 and table S4). The MC1R genotypes concord with unproblematic Mendelian inheritance of fur color phenotype for 98.four% of the 681 Arctic foxes that were genotyped at the Affymetrix SNP array (electronic supplementary material, table S4). Furthermore, genotyping of 444 Chill foxes with fur colour phenotype on MC1R using a Fluidigm SNP array confirmed this result: genotypes of 98.6% individuals were concordant with a simple Mendelian style of inheritance (electronic supplementary textile, table S4). Analysis of 12 whole-genome sequenced Arctic play a joke on individuals (11 white and 1 bluish) constitute the same base-pair mutations in MC1R, that were plant previously [29], in the one bluish private. The 11 white individuals did not evidence these mutations. All other SNPs found in the MC1R sequence had the same genotype in one or more than white foxes and the blueish trick. Come across the electronic supplementary material, S12 for detailed data.

(b) Selection analyses

Owing to depression sample size, TT individuals had to be excluded from the analyses of individual fitness (north _females = 2, n _males = five). Annual private fitness appeared higher for heterozygous (TC) females than females homozygous (CC) for the white allele, although not statistically significant at the 0.05 level (b_TC = 0.173 ± 0.102, 95% confidence interval (CI) (−0.030, 0.370), $χ_{1}^{two} = 2.79,$ p = 0.095, effigy aneb; electronic supplementary fabric, tabular array S8). The same pattern was nowadays in males (b _TC = 0.123 ± 0.105, 95% CI (−0.086, 0.325), $χ_{ane}^{2} = one.35,$ p = 0.245, effigy 1b; electronic supplementary material, table S8). The effects of genotype on private fettle were constitute to exist independent of origin, rodent phase and snow fall (electronic supplementary material, table S12).

In the analysis of fitness components, heterozygous individuals were found to accept college fecundity than homozygote CC individuals (b _TC = 0.497 ± 0.162, $χ_{1}^{ii} = 4.54,$ p = 0.033). This effect tended to be more than pronounced in females than in males (genotype 10 sex interaction: $χ_{2}^{2} = 5.47,$ p = 0.065; figure 2a; electronic supplementary textile, table S9). In addition, the difference between the two fur colour genotypes in fecundity was more pronounced in years of low (i) and increasing rodent phase (ii), where adult TC individuals produced more recruits than CC individuals ( $χ_{3}^{2} = nine.32,$ p = 0.025; electronic supplementary material, figure S9b). Differences in the effects of genotype on fecundity did non depend on origin and snowfall (electronic supplementary material, table S12).

Survival tended to be higher for heterozygous individuals compared to homozygous CC individuals (b_TC = 0.296 ± 0.157, $χ_{one}^{2} = three.632,$ p = 0.057, figure 2b; electronic supplementary cloth, tabular array S10), with no deviation betwixt sexes (genotype × sex interaction: $χ_{2}^{ii} = 0.4321,$ p = 0.8057). There was also a tendency for the difference in survival betwixt genotypes to depend on rodent phase ( $χ_{3}^{2} = 7.36,$ p = 0.061; electronic supplementary material, figure S9a), where heterozygous individuals had higher survival than homozygous individuals in low (i) and increasing (ii) rodent stage. Differences in the effects of genotype on survival did non depend on origin and snowfall (electronic supplementary cloth, table S12).

An individual'southward probability of breeding was found to be significantly higher for heterozygous than homozygous females ( $χ_{one}^{two} = five.949,$ p = 0.015), simply there was no deviation between the two genotypes in males (genotype × sexual activity interaction: $χ_{2}^{two} = vii.487,$ p = 0.024; figure iic; electronic supplementary material, table S11).

All three developed fettle components (fecundity, survival probability and breeding probability) first increased with historic period and then decreased at older ages (electronic supplementary material, figure S8), but at that place were no differences in the effects of genotypes between age classes (i.east. no pregnant genotype × age interactions; electronic supplementary material, S7).

The recruitment probability (juvenile survival) was found to exist contained of genotype ( $χ_{1}^{2} = 0.012,$ p = 0.914, figure 2d), did not depend on sex activity ( $χ_{one}^{2} = 0.004,$ p = 0.948), and the lack of any relationship between genotype and recruitment probability was similar in both sexes (genotype × sex activity interaction: $χ_{1}^{2} = 2.403,$ p = 0.121). Hence, the higher fecundity of heterozygous individuals originated from the probability of breeding and/or the number of pups produced.

Genome-broad heterozygosity was 0.020 ± 0.007 lower in wild-built-in foxes (n = 312) compared to foxes built-in at the breeding station ( $χ_{1}^{2} = five.34,$ p = 0.021, n = 374), and foxes heterozygous at MC1R (n = 123) had 0.006 ± 0.002 higher genome-wide heterozygosity than foxes with the CC genotype ( $χ_{1}^{2} = 5.96,$ p = 0.015, due north = 563). The higher genome-wide heterozygosity of TC foxes was like for individuals of unlike origin (genotype × origin interaction: $χ_{one}^{2} = 0.fourteen,$ p = 0.710). Individual fitness and the fitness components investigated did nonetheless non depend on genome-wide heterozygosity (electronic supplementary material, tabular array S13).

The candidate region GWAS for individual fitness revealed one SNP significantly associated with individual fitness at the Bonferroni-corrected significance level (p < 1.24 × 10⁻⁵, electronic supplementary material, figure S10). However, this SNP (AX-177107035, p = 3.56 × 10⁻⁶) was non significantly associated with Arctic fox fur colour.

(c) Gene analyses

Many genes are located close to and/or are in strong LD with MC1R (figure 1a). Consequently, changes in MC1R allele- or genotype frequencies would atomic number 82 to changes in frequencies of variants at other genes as well. To gather insight on what functions these genes have and how they might touch Arctic foxes, nosotros conducted some preliminary Get investigations. For 132 genes that were institute to be less than 10 kb away from an SNP significantly associated with Arctic fox fur colour, a GO term enrichment analysis showed over-representation of 33 Become terms (electronic supplementary material, tabular array S16). Many of these GO terms represent fundamental biological functions (e.yard. cytoplasm, intracellular or organelle). Eight of the 33 over-represented Become terms are involved in metabolic processes, six of them in lipid metabolism (electronic supplementary material, tabular array S16). Other GO terms are involved in developmental processes (developmental processes, regulation of Wnt signalling pathway).

To limit the analysis to genes that probably are highly associated with MC1R genotype, we looked for genes closer than 10 kb to an SNP that is (i) significantly associated with Arctic fox fur colour and (2) in loftier LD (r ⁱⁱ ≥ 0.five) with the SNP most associated with Arctic fox fur colour. Here, 41 genes were found, and their functions according to UniProtKB are summarized in the electronic supplementary textile, table S17. Simply iii of these genes were Swiss-Prot reviewed for dogs, MC1R being one of them. For several of the genes listed here, important functions are known. These include regulation of the Wnt signalling pathway (CTNNBIP1), Deoxyribonucleic acid reparation (FANCA), glucose metabolism (H6PD), development (RERE) and immune response (PIK3CD, BANP). These 41 genes were included in the GeneMANIA analysis, which showed co-expression of MC1R with four genes: CTNNBIP1, GSE1, PIEZO2, TCF25 (electronic supplementary fabric, figure S11). One gene (HSBP1) was located closer than xx kb to the SNP that was significantly associated with private Arctic flim-flam fitness (electronic supplementary textile, effigy S10). HSBP1 plays a function in stress resistance and actin system.

4. Word

In this study, we investigated the genetic basis and architecture of Arctic pull a fast one on fur colour. Our results demonstrate that MC1R is the but causal gene underlying the white and blue fur colour morphs in the Chill play a joke on. Quantification of pick on the color morphs showed signs of a fitness reward of heterozygous individuals at the fur colour locus that appeared to be similar across nigh ecology weather. This fettle advantage was stronger in females than in males, and dissimilar fur colour genotypes were to some extent affected differently by food access (rodent wheel). The MC1R gene is located in a cistron-rich region in the Arctic trick genome, and gene analyses showed that SNPs in several genes involved in developmental and metabolic processes are in potent LD with the diagnostic Arctic fox fur colour SNP.

Our GWAS identified many SNPs with significant clan with Arctic fox fur colour (effigy 1a; electronic supplementary material, effigy S4). Boom results showed that all but three of the pregnant SNPs were near MC1R in the region from 52 to 77 Mb on canis familiaris chromosome 5 (effigy anea). Our results also showed that the different scaffolds that contain significant SNPs assemble next to each other (effigy ia; electronic supplementary material, table S14) and thus do not stand for independent peaks of significance. Together with MC1R sequence information from 12 whole-genome sequenced Chill foxes (electronic supplementary material, S12) and the near-perfect association betwixt Arctic fox fur colour and the height SNP genotype, these GWAS results back up the hypothesis past Adalsteinsson et al. [30] that fur color morphs in Arctic fox is determined by a single Mendelian cistron and the results of Våge et al. [29] which suggested MC1R equally the sole causal factor underlying the distinct Arctic fox fur color morphs. The few cases of mismatch betwixt recoded fur colour phenotype and expected genotype are likely to be caused past wrong phenotyping in the field equally field information are collected under sometimes demanding conditions. The genome-wide calibration of this study confirms that no other areas in the genome explained variation in fur color and provides firm evidence of MC1R's role based on much larger sample size than previously applied. While the candidate gene approach has worked in this case, large-scale genome scans should be the preferred method to verify causal genes owing to their unbiased approach [38].

The quantification of selection showed that foxes heterozygous at the fur colour locus tend to have higher individual fettle than individuals that are homozygous for the white allele C (figure 1b). Our analyses revealed a larger departure in fitness in female Arctic foxes than in male individuals (figure ib). Decomposition of fitness into different components corroborated the results based on individual fitness and showed that TC individuals (blue) scored ameliorate in fecundity, breeding probability and developed survival probability than CC individuals (white), with the furnishings on the fecundity measures existence strongest in female foxes (figure 2). The fitness differences between the two fur colour genotypes were more pronounced at depression and increasing rodent affluence (electronic supplementary material, figure S9). These results suggest that blue foxes take a higher probability than white foxes to survive under poor food weather and show a stronger ability to use favourable weather for reproduction (i.e. during years of increasing rodent abundance where juvenile survival and subsequent recruitment is high) [68].

Unfortunately, blue homozygotes (TT) were rare in the study area. Thus, the differences between blue homozygotes and the other genotypes could not be reliably quantified. The frequency of blue homozygotes is increasing in the Scandinavian Arctic flim-flam population, hence such analyses may exist possible in the future.

Evolutionary mechanisms underlying Arctic fox fur coloration are not well studied and the chief difference between the colour morphs is thought to be their camouflage value in different habitats [34]. Recently, Di Bernardi et al. [69] also showed fitness advantages in Norwegian blueish foxes. However, the operation of the ii colour morphs was non differentially afflicted past the tested climatic variables (snow cover and winter temperature), except for a weak indication of thermal advantage of bluish juveniles, with a tendency of higher survival in colder winter temperatures compared to white juveniles [69]. Overall, they did not observe consistent prove that these advantages are attributable to differences in camouflage or thermoregulation [69]. Likewise, our results are not plausibly explained past differences in camouflage (i.e. white morph is expected to have better camouflage values in mountain habitats [34]) or thermoregulation (i.e. start and end of snowfall flavour did non touch fitness of colour genotypes differently (electronic supplementary material, table S12)), indicating that that the adaptations the Arctic flim-flam has to withstand the Arctic winter are to a big extent independent of color genotype.

Because support for the two almost likely routes of directly selection on Arctic fox fur color is weak or missing in our study, it seems reasonable to explore potential routes of indirect selection. Pleiotropic interactions in the melanocortin complex, which MC1R is role of, are well known and reviewed [26]. Both experimental and observational studies accept shown a large variety of traits that are affected by the melanocortin system and thus covary with melanin-based coloration [26]. These traits, e.g. resistance to stressors and enhanced allowed response, have the potential to play vital roles for a wild species living in a harsh climate [26]. Behavioural traits such equally aggressiveness are affected by pleiotropy in the melanocortin system and could impact foxes with genotypes for the bluish colour morph positively in terms of getting admission to expert den sites and chasing abroad both conspecifics as well as competitors (e.g. red foxes). The last grouping of traits affected by pleiotropy in the melanocortin organisation is sexual traits, where both sexes can exist affected positively in terms of sexual receptivity and fertility [26]. One could also expect higher fertility in male blue foxes based on findings that male person blue foxes had higher concentrations of spermatozoa in their ejaculates compared to white foxes [70]. Yet, our findings practice non concur every bit we did not detect a difference in fecundity or convenance probability between male person foxes with the CC (white) and TC (blue) genotype (effigy 2a,c), indicating that whatever deviation in spermatozoa concentration does non translate into higher reproduction in wild Arctic foxes in the Scandinavian population.

MC1R is located in a region with numerous other genes and nosotros found several genes close to SNPs that were significantly associated with Arctic fox fur color (figure 1a). Based on an analysis of LD (figure 1a; electronic supplementary fabric, figure S12), some of these genes certainly covary with fur colour genotype in the Scandinavian Chill fox. Both the GO term analysis (electronic supplementary cloth, table S16) and the analysis of genes close to SNPs significantly associated with Arctic fox fur color (electronic supplementary material, table S17) testify that genes covarying with MC1R genotypes may be involved in important processes. Equally for all species enduring harsh winter conditions, the ability to control metabolism is relevant and potentially vital for Arctic foxes in enduring cold climate and food scarcity. Eight of the over-represented Get terms were related to lipid and steroid metabolism, making this an interesting pathway to investigate for future studies. Regulation of the Wnt signalling pathway showed up in our results as an enriched GO term (electronic supplementary material, tabular array S16), as well as a single cistron in form of CTNNBIP1. This pathway plays meaning roles in organism evolution [71] and inhibition tin can lead to severe and potentially fatal effects [72]. Several other over-represented GO terms were also office of developmental processes (electronic supplementary material, table S16). In improver, the gene RERE that plays a part in developmental processes was found among the genes likely to covary with MC1R. Another two of these genes are involved in immune responses (BANP and PIK3CD), a trait that likewise is office of the pleiotropic melanocortin system. HSBP1 was found shut to the SNP significantly associated with individual Chill play a joke on fettle and is involved in stress resistance. Precisely how these genotypes are expressed phenotypically and whether these phenotypes can affect fecundity and/or viability in the Arctic fox remains to exist seen. However, although there is a adventure of 'storytelling' in this kind of analysis [73], these genes provide examples of covarying genes that may potentially have implications for individual Chill fox fitness and should be investigated in more item in hereafter studies that aim to understand the molecular basis for fettle differences between Arctic play tricks fur colour genotypes and phenotypes.

Another possible caption of college fitness in blueish heterozygotes compared to white homozygotes is heterozygosity advantage [58,59]. All the same, despite having found significantly higher genome-wide heterozygosity in individuals heterozygous at the fur colour locus compared to those homozygous for the C allele, variation in individual fitness did not seem to exist driven by this difference (electronic supplementary material, table S13). We also showed that foxes born at the captive breeding station had higher genome-broad heterozygosity than individuals born in the wild. This could indicate a lower degree of inbreeding in convict-born foxes, which may seem counterintuitive at outset glance. However, this is expected as convenance pairs in the convenance station represent all extant subpopulations in Scandinavia and are chosen to maintain genetic diversity [40]. Hasselgren et al. [74] presented a adept example of the genetic rescue event where blue Chill foxes showed high reproductive success in an inbred population in Sweden. Information technology is possible that we see a weak genetic rescue effect in this study besides, and that the observed reproductive advantages of heterozygous individuals (figures 1b and 2) might be the event of genetic rescue by the release of TC individuals from the breeding station. The importance of such furnishings, and whether they contribute to explaining the observed growth of the Scandinavian Arctic flim-flam population, will exist explored in future studies.

Our study adds to the body of research that has identified major genes underlying traits with fettle implications for a wild animate being species through genetic mapping [fifteen–18]. However, our results as well reveal the large potential for interesting genetic interactions that are hidden behind the seemingly uncomplicated trait, such as Arctic fob fur colour. Covariation between colour and other phenotypic traits is well documented [26,27] and our overall results may suggest that such covariation, owing to LD between MC1R and other genes or pleiotropic effects of MC1R, may be the driver of option on fur colour also in the Scandinavian Arctic play a joke on population. This emphasizes the need to wait further than the most credible phenotype when attempting to understand the mechanisms of selection in wild populations. More specifically, it is clear that gene mapping tin can provide valuable insight into the genetic architecture of adaptive traits and other linked traits. Also, when the linked cistron that actually affects individual fettle cannot be identified, the molecular genetic data generated (due east.m. on linked genes, pleiotropy, genome-wide heterozygosity) can be used to determine knowledge gaps and areas of interest for hereafter research. In our study species, 1 major effect is the lack of data on other phenotypic differences between the colour morphs, such as behaviour, metabolism, energy expenditure or immune response. Future research may employ such traits as a starting signal for gaining more insight into pick processes that occur in the Arctic fob.

Ideals

Necessary permits for research on a wild species in Norway were in place and all research was conducted according to national rules and regulations. Permits include approvals of animal care protocols for captive breeding and live capture of wild animals, as well as permits for conducting research and handling a species of conservation concern.

Information accessibility

Because the Arctic fox is an endangered species in the written report area, sensitive information (i.e. den locations) will non be released. Owing to ongoing enquiry projects and the conservation condition of the Scandinavian Arctic play a trick on we ask for a 2 yr embargo earlier making data available.

Authors' contributions

L.T.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, software, validation, visualization, writing—original draft, writing—review and editing; I.J.H.: conceptualization, data curation, formal assay, methodology, supervision, writing—original typhoon, writing—review and editing; O.K.: data curation, supervision, writing—original typhoon; C.D.B.: data curation, investigation, methodology, writing—original draft, writing—review and editing; T.K.: formal analysis, methodology, supervision, writing—original draft, writing—review and editing; K.North.: data curation; Grand.H.: information curation; J.F.Westward.: data curation; A.A.: data curation; A.L.: conceptualization, data curation, funding conquering, project administration, supervision, writing—original typhoon, writing—review and editing; Due north.Due east.E.: conceptualization, data curation, funding acquisition, investigation, methodology, project administration, writing—original typhoon, writing—review and editing; Ø.F.: conceptualization, data curation, funding acquisition, investigation, project administration, supervision, writing—original draft, writing—review and editing; H.J.: conceptualization, funding acquisition, investigation, project administration, supervision, writing—original typhoon, writing—review and editing. All authors gave concluding approval for publication and agreed to be held accountable for the work performed therein.

Competing interests

We declare nosotros take no competing interests.

Funding

The Norwegian Captive Breeding Programme (contract 19087015) and the Arctic Trick Monitoring Programme in Norway (contract 18087019) contributed a significant corporeality of data to this study, all funded by the Norwegian Ecology Agency. This report was part of ECOFUNC funded by the Research Council of Norway (grant 244557). The work was also partly supported by the Research Council of Norway through its Centres of Excellence funding scheme (grant 22357).

Acknowledgements

We like to thank numerous park rangers and summer field assistants involved in collecting field data and genetic samples. Cheers to many people at the Genlab at NINA for performing genetic analysis and to Roger Meås and Roy Anderson for operating chip readers and all technical equipment. Further nosotros would similar to thank Kristine Ulvund who extracted and merged recovery and reproduction information, Stefan Blumentrath for extracting snow data, Alina Niskanen for aid with genetic parentage analyses and Sarah Lundregan and Martin Kuiper for fruitful discussions almost GWAS, BLAST and GO analyses. Genotyping on the custom play a joke on Affymetrix Axiom 702 Thousand SNP array was carried out at CIGENE, Norwegian University of Life Sciences, Norway.

Footnotes

Special Characteristic: Wild quantitative genomics: the genomic footing of fitness variation in natural populations edited by Susan Johnston, Nancy Chen, Emily Josephs.

Electronic supplementary textile is bachelor online at https://doi.org/10.6084/m9.figshare.c.5628908.

Published past the Royal Society nether the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/iv.0/, which permits unrestricted use, provided the original writer and source are credited.

References

one.
Fisher RA
. 1930 The genetical theory of natural pick . Oxford, UK: Clarendon Press. Crossref, Google Scholar
2.
Stinchcombe JR, Hoekstra HE
. 2008 Combining population genomics and quantitative genetics: finding the genes underlying ecologically important traits. Heredity (Edinb.) 100 , 158-170. (doi:10.1038/sj.hdy.6800937) Crossref, PubMed, ISI, Google Scholar
3.
Andrew RLet al.
2013 A route map for molecular ecology. Mol. Ecol. 22 , 2605-2626. (doi:10.1111/mec.12319) Crossref, PubMed, ISI, Google Scholar
4.
Ellegren H
. 2014 Genome sequencing and population genomics in not-model organisms. Trends Ecol. Evol. 29 , 51-63. (doi:10.1016/j.tree.2013.09.008) Crossref, PubMed, ISI, Google Scholar
5.
Bush WS, Moore JH
. 2012 Genome-wide clan studies. PLoS Comput. Biol. 8 , e1002822. (doi:10.1371/journal.pcbi.1002822) Crossref, PubMed, ISI, Google Scholar
6.
Visscher PM, Wray NR, Zhang Q, Sklar P, McCarthy MI, Brown MA, Yang J
. 2017 x years of GWAS discovery: biological science, office, and translation. Am. J. Hum. Genet. 101 , 5-22. (doi:10.1016/j.ajhg.2017.06.005) Crossref, PubMed, ISI, Google Scholar
vii.
Jensen H, Szulkin One thousand, Slate J
. 2014 Molecular quantitative genetics. In Quantitative genetics in the wild (eds
Charmantier A, Garant D, Kruuk LEB
), pp. 209-227. Oxford, United kingdom: Oxford University Press. Crossref, Google Scholar
eight.
Santure AW, Garant D
. 2018 Wild GWAS—association mapping in natural populations. Mol. Ecol. Resour. 18 , 729-738. (doi:10.1111/1755-0998.12901) Crossref, PubMed, ISI, Google Scholar
nine.
Korte A, Farlow A
. 2013 The advantages and limitations of trait analysis with GWAS: a review. Found Methods 9 , 29. (doi:x.1186/1746-4811-9-29) Crossref, PubMed, ISI, Google Scholar
10.
Santure AWet al.
2015 Replicated analysis of the genetic architecture of quantitative traits in two wild great tit populations. Mol. Ecol. 24 , 6148-6162. (doi:x.1111/mec.13452) Crossref, PubMed, ISI, Google Scholar
11.
Santure AW, De Cauwer I, Robinson MR, Poissant J, Sheldon BC, Slate J
. 2013 Genomic autopsy of variation in clutch size and egg mass in a wild great tit (Parus major) population. Mol. Ecol. 22 , 3949-3962. (doi:10.1111/mec.12376) Crossref, PubMed, ISI, Google Scholar
12.
Wenzel MA, James MC, Douglas A, Piertney SB
. 2015 Genome-broad clan and genome division reveal novel genomic regions underlying variation in gastrointestinal nematode burden in a wild bird. Mol. Ecol. 24 , 4175-4192. (doi:10.1111/mec.13313) Crossref, PubMed, ISI, Google Scholar
13.
Kardos 1000, Husby A, McFarlane SE, Qvarnström A, Ellegren H
. 2016 Whole-genome resequencing of farthermost phenotypes in collared flycatchers highlights the difficulty of detecting quantitative trait loci in natural populations. Mol. Ecol. Resour. 16 , 727-741. (doi:ten.1111/1755-0998.12498) Crossref, PubMed, ISI, Google Scholar
xiv.
Lundregan SLet al.
2018 Inferences of genetic architecture of nib morphology in house sparrow using a high-density SNP array betoken to a polygenic footing. Mol. Ecol. 27 , 3498-3514. (doi:10.1111/mec.14811) Crossref, PubMed, ISI, Google Scholar
15.
Johnston SE, Mcewan JC, Pickering NK, Kijas JW, Beraldi D, Pilkington JG, Pemberton JM, Slate J
. 2011 Genome-wide clan mapping identifies the genetic basis of discrete and quantitative variation in sexual weaponry in a wild sheep population. Mol. Ecol. xx , 2555-2566. (doi:x.1111/j.1365-294X.2011.05076.x) Crossref, PubMed, ISI, Google Scholar
16.
Barson NJet al.
2015 Sex activity-dependent dominance at a single locus maintains variation in historic period at maturity in salmon. Nature 528 , 405-408. (doi:10.1038/nature16062) Crossref, PubMed, ISI, Google Scholar
17.
Sinclair-Waters M, Ødegård J, Korsvoll SA, Moen T, Lien South, Primmer CR, Barson NJ
. 2020 Beyond large-consequence loci: large-scale GWAS reveals a mixed big-effect and polygenic architecture for historic period at maturity of Atlantic salmon. Genet. Sel. Evol. 52 , 9. (doi:ten.1186/s12711-020-0529-8) Crossref, PubMed, ISI, Google Scholar
18.
Lawson LP, Petren M
. 2017 The adaptive genomic landscape of nib morphology in Darwin'southward finches. Mol. Ecol. 26 , 4978-4989. (doi:10.1111/mec.14166) Crossref, PubMed, ISI, Google Scholar
19.
Beddard FE
. 1892 Beast coloration: an account of the chief facts and theories relating to the colours and markings of animals . London, U.k.: South. Sonnenschein & Visitor. Google Scholar
20.
Caro T
. 2005 The adaptive significance of coloration in mammals. Bioscience 55 , 125-136. (doi:10.1641/0006-3568(2005)055[0125:TASOCI]2.0.CO;2) Crossref, ISI, Google Scholar
21.
Nachman MW, Hoekstra HE, D'Agostino SL
. 2003 The genetic ground of adaptive melanism in pocket mice. Proc. Natl Acad. Sci. USA 100 , 5268-5273. (doi:x.1073/pnas.0431157100) Crossref, PubMed, ISI, Google Scholar
22.
Brommer JE, Ahola One thousand, Karstinen T
. 2005 The colour of fitness: plume coloration and lifetime reproductive success in the tawny owl. Proc. R. Soc. B 272 , 935-940. (doi:10.1098/rspb.2005.3052) Link, ISI, Google Scholar
23.
Gratten J, Wilson A, McRae A, Beraldi D, Visscher P, Pemberton J, Slate J
. 2008 A localized negative genetic correlation constrains microevolution of coat color in wild sheep. Science 319 , 318-320. (doi:x.1126/scientific discipline.1151182) Crossref, PubMed, ISI, Google Scholar
24.
Roff D, Fairbairn D
. 2013 The costs of being dark: the genetic ground of melanism and its association with fitness-related traits in the sand cricket. J. Evol. Biol. 26 , 1406-1416. (doi:10.1111/jeb.12150) Crossref, PubMed, ISI, Google Scholar
25.
Hadfield JD, Nutall A, Osorio D, Owens IPF
. 2007 Testing the phenotypic gambit: phenotypic, genetic and environmental correlations of colour. J. Evol. Biol. twenty , 549-557. (doi:ten.1111/j.1420-9101.2006.01262.x) Crossref, PubMed, ISI, Google Scholar
26.
Ducrest A-L, Keller L, Roulin A
. 2008 Pleiotropy in the melanocortin system, coloration and behavioural syndromes. Trends Ecol. Evol. 23 , 502-510. (doi:x.1016/j.tree.2008.06.001) Crossref, PubMed, ISI, Google Scholar
27.
San-Jose LM, Roulin A
. 2017 Genomics of coloration in natural creature populations. Phil. Trans. R. Soc. B 372 , 20160337. (doi:10.1098/rstb.2016.0337) Link, ISI, Google Scholar
28.
Chesemore DL
. 1970 Notes on the pelage and priming sequence of arctic foxes in northern Alaska. J. Mammal. 51 , 156-159. (doi:10.2307/1378547) Crossref, Google Scholar
29.
Våge DI, Fuglei East, Snipstad Thousand, Beheim J, Landsem VM, Klungland H
. 2005 Two cysteine substitutions in the MC1R generate the blue variant of the chill play tricks (Alopex lagopus) and prevent expression of the white wintertime glaze. Peptides 26 , 1814-1817. (doi:10.1016/j.peptides.2004.eleven.040) Crossref, PubMed, ISI, Google Scholar
xxx.
Adalsteinsson Southward, Hersteinsson P, Gunnarsson E
. 1987 Fox colors in relation to colors in mice and sheep. J. Hered. 78 , 235-237. (doi:10.1093/oxfordjournals.jhered.a110373) Crossref, PubMed, ISI, Google Scholar
31.

Norwegian Polar Institute. 2021 Arctic fox (Vulpes lagopus) . Norwegian Polar Institute. Run into https://world wide web.npolar.no/en/species/chill-fob/. Google Scholar
32.
Braestrup FW
. 1941 A report on the Arctic pull a fast one on in Greenland: immigrations, fluctuations in numbers, based mainly on trading statistics . Copenhagen, Denmark: Reitzel. Google Scholar
33.
Fetherston K
. 1947 Geographic variation in the incidence of occurrence of the blue phase of the arctic pull a fast one on in Canada. Can. Field-Nat. 61 , 15-eighteen. Google Scholar
34.
Hersteinsson P
. 1989 Population genetics and ecology of dissimilar colour morphs of chill foxes Alopex lagopus in Iceland. Finn. Game Res. 46 , 64-78. Google Scholar
35.
Chesemore DL
. 1968 Distribution and movements of white foxes in northern and western Alaska. Tin can. J. Zool. 46 , 849-854. (doi:10.1139/z68-121) Crossref, Google Scholar
36.
Schmutz SM, Berryere TG
. 2007 Genes affecting glaze colour and blueprint in domestic dogs: a review. Anim. Genet. 38 , 539-549. (doi:10.1111/j.1365-2052.2007.01664.x) Crossref, PubMed, ISI, Google Scholar
37.
Roulin A, Ducrest A-L
. 2013 Genetics of colouration in birds. Semin. Prison cell Dev. Biol. 24 , 594-608. (doi:ten.1016/j.semcdb.2013.05.005) Crossref, PubMed, ISI, Google Scholar
38.
Brownish EAet al.
2013 Detecting genes for variation in parasite burden and immunological traits in a wild population: testing the candidate cistron approach. Mol. Ecol. 22 , 757-773. (doi:x.1111/j.1365-294X.2012.05757.x) Crossref, PubMed, ISI, Google Scholar
39.
Angerbjörn Aet al.
2013 Carnivore conservation in practice: replicated management actions on a large spatial scale. J. Appl. Ecol. 50 , 59-67. (doi:10.1111/1365-2664.12033) Crossref, ISI, Google Scholar
xl.
Landa A, Flagstad Ø, Areskoug V, Linnell JDC, Strand O, Ulvund KR, Thierry A-Grand, Rød-Eriksen L, Eide NE
. 2017 The endangered Arctic fox in Kingdom of norway—the failure and success of captive breeding and reintroduction. Polar Res. 36 , nine. (doi:10.1080/17518369.2017.1325139) Crossref, ISI, Google Scholar
41.
Ulvund 1000, Flagstad Ø, Sandercock B, Kleven O, Landa A, Eide NE
. 2019Chill trick in Norway 2019. Results from the National Monitoring Programme for chill play tricks. NINA report 1737. Norwegian Plant for Nature Enquiry, Trondheim, Kingdom of norway. Google Scholar
42.
Norén K, Dalén L, Flagstad Ø, Berteaux D, Wallén J, Angerbjörn A
. 2017 Evolution, environmental and conservation—revisiting iii decades of Arctic fox population genetic research. Polar Res. 36 , 4. (doi:ten.1080/17518369.2017.1325135) Crossref, ISI, Google Scholar
43.
Eide NE, Ulvund Chiliad, Kleven O, Landa A, Flagstad Ø
. 2020Arctic pull a fast one on in Norway 2020. Results from the national monitoring programme for arctic fox. NINA Study 1913. Norwegian Institute for Nature Enquiry, Trondheim, Norway. Google Scholar
44.
Bassil NVet al.
2015 Development and preliminary evaluation of a 90 K Axiom® SNP assortment for the allo-octoploid cultivated strawberry Fragaria × ananassa . BMC Genom. 16 , 155. (doi:10.1186/s12864-015-1310-one) Crossref, PubMed, ISI, Google Scholar
45.
Hasselgren One thousand, Dussex N, von Seth J, Angerbjörn A, Olsen RA, Dalén Fifty, Norén Grand
. 2021 Genomic and fitness consequences of inbreeding in an endangered carnivore. Mol. Ecol. xxx, 2790-2799. (doi:x.1111/mec.15943) Crossref, PubMed, ISI, Google Scholar
46.
Aulchenko YS, Ripke S, Isaacs A, Van Duijn CM
. 2007 GenABEL: an R library for genome-wide association analysis. Bioinformatics 23 , 1294-1296. (doi:ten.1093/bioinformatics/btm108) Crossref, PubMed, ISI, Google Scholar
47.
Santure AW, Stapley J, Ball AD, Birkhead TR, Burke T, Slate J
. 2010 On the use of large marking panels to approximate inbreeding and relatedness: empirical and simulation studies of a pedigreed zebra finch population typed at 771 SNPs. Mol. Ecol. 19 , 1439-1451. (doi:10.1111/j.1365-294X.2010.04554.10) Crossref, PubMed, ISI, Google Scholar
48.
Johnson RC, Nelson GW, Troyer JL, Lautenberger JA, Kessing BD, Winkler CA, O'Brien SJ
. 2010 Accounting for multiple comparisons in a genome-wide clan report (GWAS). BMC Genom. xi , 724. (doi:10.1186/1471-2164-11-724) Crossref, PubMed, ISI, Google Scholar
49.
Meijer T, Norén K, Hellström P, Dalén L, Angerbjörn A
. 2008 Estimating population parameters in a threatened arctic fox population using molecular tracking and traditional field methods. Anim. Conserv. 11 , 330-338. (doi:10.1111/j.1469-1795.2008.00188.x) Crossref, ISI, Google Scholar
50.
Huisman J
. 2017 Pedigree reconstruction from SNP data: parentage assignment, sibship clustering and across. Mol. Ecol. Resour. 17 , 1009-1024. (doi:x.1111/1755-0998.12665) Crossref, PubMed, ISI, Google Scholar
51.
Engen Due south, Sæther B-E, Kvalnes T, Jensen H
. 2012 Estimating fluctuating option in historic period-structured populations. J. Evol. Biol. 25 , 1487-1499. (doi:x.1111/j.1420-9101.2012.02530.x) Crossref, PubMed, ISI, Google Scholar
52.
Engen S, Kvalnes T, Sæther Exist
. 2014 Estimating phenotypic selection in historic period-structured populations by removing transient fluctuations. Evolution 68 , 2509-2523. (doi:10.1111/evo.12456) Crossref, PubMed, ISI, Google Scholar
53.
Kvalnes T
. 2013Lmf: functions for interpretation and inference of selection in historic period-structured populations. R package version 1.two. See https://CRAN.R-project.org/packet=lmf. Google Scholar
54.
Engen South, Lande R, Sæther B-E, Dobson FS
. 2009 Reproductive value and the stochastic demography of age-structured populations. Am. Nat. 174 , 795-804. (doi:10.1086/647930) Crossref, PubMed, ISI, Google Scholar
55.
Bates D, Maechler Thou, Bolker B, Walker S
. 2015 Plumbing fixtures linear mixed-effects models using lme4. J. Stat. Softw. 67 , one-48. (doi:x.18637/jss.v067.i01) Crossref, ISI, Google Scholar
56.
Brooks ME, Kristensen K, van Benthem KJ, Magnusson A, Berg CW, Nielsen A, Skaug HJ, Machler M, Bolker BM
. 2017 glmmTMB balances speed and flexibility among packages for aught-inflated generalized linear mixed modeling. R J. 9 , 378-400. (doi:10.3929/ethz-b-000240890) Crossref, ISI, Google Scholar
57.
Angerbjörn A, Tannerfeldt Thou, Erlinge S
. 1999 Predator-prey relationships: Arctic foxes and lemmings. J. Anim. Ecol. 68 , 34-49. (doi:10.1046/j.1365-2656.1999.00258.x) Crossref, ISI, Google Scholar
58.
Sellis D, Callahan BJ, Petrov DA, Messer PW
. 2011 Heterozygote reward as a natural consequence of adaptation in diploids. Proc. Natl Acad. Sci. The states 108 , 20 666-20 671. (doi:10.1073/pnas.1114573108) Crossref, ISI, Google Scholar
59.
Szulkin Thousand, Bierne Northward, David P
. 2010 Heterozygosity-fitness correlations: a time for reappraisal. Evolution 64 , 1202-1217. (doi:x.1111/j.1558-5646.2010.00966.x) PubMed, ISI, Google Scholar
60.
Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ
. 1990 Basic local alignment search tool. J. Mol. Biol. 215 , 403-410. (doi:10.1016/S0022-2836(05)80360-ii) Crossref, PubMed, ISI, Google Scholar
61.
Camacho C, Coulouris G, Avagyan V, Ma North, Papadopoulos J, Bealer K, Madden TL
. 2009 Boom+: architecture and applications. BMC Bioinf. 10 , 421. (doi:10.1186/1471-2105-10-421) Crossref, PubMed, ISI, Google Scholar
62.
Graphodatsky AS, Yang F, O'Brien PCM, Serdukova North, Milne BS, Trifonov V, Ferguson-Smith MA
. 2000 A comparative chromosome map of the Chill Play a joke on, red play a trick on and dog defined by chromosome painting and high resolution G-banding. Chromosome Res. 8 , 253-263. (doi:10.1023/A:1009217400140) Crossref, PubMed, ISI, Google Scholar
63.
Hoeppner MPet al.
2014 An improved canine genome and a comprehensive catalogue of coding genes and non-coding transcripts. PLoS I nine , e91172. (doi:10.1371/journal.pone.0091172) Crossref, PubMed, ISI, Google Scholar
64.
Beißbarth T, Speed TP
. 2004 GOstat: find statistically overrepresented cistron ontologies within a group of genes. Bioinformatics 20 , 1464-1465. (doi:10.1093/bioinformatics/bth088) Crossref, PubMed, ISI, Google Scholar
65.
Consortium TU
. 2018 UniProt: the universal protein knowledgebase. Nucleic Acids Res. 46 , 2699. (doi:10.1093/nar/gky092) Crossref, PubMed, ISI, Google Scholar
66.
Warde-Farley Det al.
2010 The GeneMANIA prediction server: biological network integration for factor prioritization and predicting gene function. Nucleic Acids Res. 38 , W214-W220. (doi:10.1093/nar/gkq537) Crossref, PubMed, ISI, Google Scholar
67.
Core Team R
. 2019 R: a language and environment for statistical computing, 3.half-dozen.1 edn . Vienna, Austria: R Foundation for Statistical Calculating. Google Scholar
68.
Meijer T, Elmhagen B, Eide NE, Angerbjörn A
. 2013 Life history traits in a cyclic ecosystem: a field experiment on the chill pull a fast one on. Oecologia 173 , 439-447. (doi:10.1007/s00442-013-2641-eight) Crossref, PubMed, ISI, Google Scholar
69.
Di Bernardi Cet al.
2021 Fitness and fur colouration: testing the cover-up and thermoregulation hypotheses in an Chill mammal. J. Anim. Ecol. xc , 1328-1340. (doi:10.1111/1365-2656.13457) Crossref, PubMed, ISI, Google Scholar
70.
Stasiak Yard, Kondracki S, Iwanina M
. 2019 Assessment of chosen semen characteristics of ii colour morphs of the Arctic flim-flam Alopex lagopus Fifty. Anim. Sci. J. 90 , 1120-1126. (doi:10.1111/asj.13257) Crossref, PubMed, ISI, Google Scholar
71.
Komiya Y, Habas R
. 2008 Wnt bespeak transduction pathways. Organogenesis 4 , 68-75. (doi:10.4161/org.4.2.5851) Crossref, PubMed, Google Scholar
72.
Satoh Ket al.
2004 Anteriorization of neural fate by inhibitor of β-catenin and T cell factor (ICAT), a negative regulator of Wnt signaling. Proc. Natl Acad. Sci. USA 101 , 8017-8021. (doi:ten.1073/pnas.0401733101) Crossref, PubMed, ISI, Google Scholar
73.
Pavlidis P, Jensen JD, Stephan W, Stamatakis A
. 2012 A critical assessment of storytelling: gene ontology categories and the importance of validating genomic scans. Mol. Biol. Evol. 29 , 3237-3248. (doi:ten.1093/molbev/mss136) Crossref, PubMed, ISI, Google Scholar
74.
Hasselgren Thousand, Angerbjörn A, Eide NE, Erlandsson R, Flagstad Ø, Landa A, Wallén J, Norén Thou
. 2018 Genetic rescue in an inbred Arctic fox Vulpes lagopus population. Proc. R. Soc. B 285 , 20172814. (doi:10.1098/rspb.2017.2814) Link, ISI, Google Scholar