Level and partitioning of genetic diversity is expected to vary between contrasting habitats, reflecting differences in strength of ecological and evolutionary processes. Therefore, it is necessary to consider processes acting on different time scales when trying to explain diversity patterns in different parts of species' distributions. To explore how historical and contemporary factors jointly may influence patterns of genetic diversity and population differentiation, we compared genetic composition in the perennial herb Arabidopsis lyrata ssp. petraea from the northernmost parts of its distribution range on Iceland to that previously documented in Scandinavia. Leaf tissue and soil were sampled from ten Icelandic populations of A. lyrata. Seedlings were grown from soil samples, and tissue from above-ground and seed bank individuals were genotyped with 21 microsatellite markers. Seed bank density in Icelandic populations was low but not significantly different from that observed in Norwegian populations. While within-population genetic diversity was relatively high on Iceland (HE = 0.35), among-population differentiation was low (FST = 0.10) compared to Norwegian and Swedish populations. Population differentiation was positively associated with geographical distance in both Iceland and Scandinavia, but the strength of this relationship varied between regions. Although topography and a larger distribution range may explain the higher differentiation between mountainous Norwegian relative to lowland populations in Sweden, these factors cannot explain the lower differentiation in Icelandic compared to Swedish populations. We propose that low genetic differentiation among Icelandic populations is not caused by differences in connectivity, but is rather due to large historical effective population sizes. Thus, rather than contemporary processes, historical factors such as survival of Icelandic lineages in northern refugia during the last glacial period may have contributed to the observed pattern.
Citation: Falahati-Anbaran M, Lundemo S, Ansell SW, Stenøien HK (2014) Contrasting Patterns of Genetic Structuring in Natural Populations of Arabidopsis lyrata Subsp. petraea across Different Regions in Northern Europe. PLoS ONE 9(9): e107479. https://doi.org/10.1371/journal.pone.0107479
Editor: Nadia Singh, North Carolina State University, United States of America
Received: April 14, 2014; Accepted: August 11, 2014; Published: September 16, 2014
Copyright: © 2014 Falahati-Anbaran et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was financially supported by grants to HKS from the Swedish Research Council (VR 21-2004-3314) and the Research Council of Norway (NFR 10298100). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Genetic diversity and structuring of natural populations are shaped by both historical and contemporary processes within a given region . Much work has focused on the impact of local and regional habitat structure and landscape heterogeneity on levels of genetic diversity, differentiation, patterns of gene flow and adaptability of local populations –. Genetic connectivity between populations often varies across habitats  and different patterns of population differentiation may be expected among conspecific populations inhabiting regions in different parts of the distribution range . Physical barriers and lack of biotic and abiotic dispersal vectors may reduce gene flow by preventing movement of either seed or pollen between populations . Furthermore, limited gene flow over long relative to short distances creates an isolation by distance (IBD) pattern both under an island  and a stepping stone model , as well as increases adaptive differentiation between populations .
Population connectivity and genetic structure will, however, not only be shaped by physical features of landscapes and biological factors affecting the dispersal kernel . Population history, including population age expressed as time since colonization, stability through time and historical effective sizes, may also influence observed levels of structuring , , and thus a diverse range of ecological and evolutionary processes must be examined in order to be able to explain patterns of biodiversity observed within species . Pleistocene glaciations have affected the distribution and diversity of many taxa in central and northern Europe . A number of studies have shown that populations experiencing rapid post-glacial colonization often exhibit different patterns of structuring compared to populations that survived the last glacial maximum in situ –. For instance, old populations may reach drift-gene flow equilibrium because of long time since colonization, resulting in a strong IBD pattern, whereas IBD is not always expected in recently established populations , . Additionally, extent of geographical barriers, dispersal ability and proximity to refugia may influence the association of population age on strength and patterns of IBD , .
Arabidopsis lyrata is frequently used as a model for studies on genetic structuring, evolution of self-incompatibility and distribution in plants –. In addition, the disjunctive distribution of the species in central Europe, Scandinavia, Iceland and North America, with varying ecological and life history characteristics, have promoted studies of natural selection and the genetic basis of local adaptation –. Populations inhabiting sites with different ecological features and differing in life-history characteristics can also be useful in association studies to locate genomic regions associated with adaptive variation , . Gaudeul et al.  recently demonstrated differences in genetic structuring and genetic diversity between contrasting mountainous and coastal populations of A. lyrata in Scandinavia, and hypothesized that variability in landscape structures may explain observed differences. In the Norwegian distribution range, the topography is rugged and dominated by valleys, whereas Swedish populations occur in a more open landscape along the coast. Mountains are expected to restrict dispersal of propagules in space, which can explain the higher between-population differentiation and lower genetic diversity in Norway compared to Sweden. Thus, if landscape structure has a significant impact on processes such as gene flow, a comparable pattern of genetic diversity and structuring might be expected for regions with similar landscape structures.
A. lyrata is distributed evenly throughout Iceland, and as opposed to Norway, there are fewer pronounced topographic barriers such as mountains, fjords and deep valleys to disrupt genetic connectivity among populations. The wide distribution of lava fields on Iceland, where A. lyrata act as a pioneer species , may help explain why the species is more common on Iceland than in most other regions in Europe, since the barren volcanic terrain will not necessarily act as a physical obstacle in restricting connectivity. Several studies of Icelandic A. lyrata using DNA sequence data, possibly subject to strong natural selection, show contrasting patterns of genetic differentiation on various geographical scales , , and overall gene flow levels on Iceland is difficult to infer from these studies. Both natural selection, gene flow, effective population size and age of populations will contribute to contemporary genetic structure, and non-neutral loci or neutral loci linked to selected sites may show contrasting patterns of genetic structure . However, linkage between neutral and non-neutral loci in Icelandic populations has been found to be less likely due to the high frequency of recombination in a large region around the non-neutral loci in A. lyrata , .
It has been shown that life-history characteristics, such as presence of seed banks, may impact effective population sizes in Scandinavian populations of this  and related species . It may therefore be important to take seed bank characteristics into consideration when trying to explain differences in variability patterns among different geographic regions. High levels of within-population variability and low levels of genetic differentiation in central European populations of A. lyrata has been attributed to glacial survival on the continent during the last glacial period . Furthermore, demographic bottlenecks and the presence of a subset of genetic variants in populations from northern Europe and North America compared to populations from central Europe has been attributed to a post glacial colonization pattern . On the other hand, relatively strong divergence between northern and central European populations has also suggested a cryptic Nordic refugium in this species .
In the present study we investigate whether patterns of genetic diversity within and among Icelandic populations of A. lyrata differ from that observed in Scandinavia. We want to quantify the relative impact of historical and contemporary processes in order to understand what factors explain the observed patterns of genetic diversity in northern Europe. In particular, we want to investigate to what extent observed patterns of genetic structuring are attributed to historical factors, like long-term survival and population stability through evolutionary time, and contemporary processes, involving dormancy, dispersal abilities and/or topographic features.
Materials and Methods
No specific permissions were required for the field activities on Iceland presented in this study. None of the study populations occurred within protected areas, and the study species itself is not endangered or protected. The GPS coordinates for the study populations are provided in Table 1.
Arabidopsis lyrata ssp. petraea (L.) O'Kane and Al-Shebaz (Brassicaceae) is a perennial, self-incompatible herb with a highly disjunctive distribution in Europe, consisting of a narrow central European distribution area in Austria, Germany and the Czech republic, and a wide but fragmented northern European distribution area covering Iceland, Faeroe Islands, northern parts of United Kingdom, southwestern parts of Norway and the eastern coast of central Sweden , . On Iceland, A. lyrata is widespread throughout the island (http://floraislands.is) and found in diverse habitats differing in several aspects from those in central Europe and the Scandinavian Peninsula. The island is dominated by few mountainous areas and extensive lowland lava plains, with the species growing under a range of conditions, from volcanic sands to rich soils in agricultural areas (Table 1). The species grows under a wide range of soil conditions on Iceland, varying from volcanic sands to rich soil of agricultural areas (Table 1).
Icelandic, Norwegian and Swedish localities vary in topography, vegetation cover, altitude and latitude. The Swedish populations are located at low altitude (<5 m a.s.l.) on the coast, covering a smaller geographical area (85×10 km; 62.583–63.183° N) , . This in contrast to the large distribution area of A. lyrata on Iceland, where populations are located over a 300-km range . In Norway, the species occurs from sea level up to 1,700 m a.s.l., and populations are separated by deep valleys and mountains over a large geographical area (385×180 km; 59.451–62.744° N). Differences in landscape topography among regions may lead to overall differences in wind speed among regions, and wind speed at local sites may also be strongly affected by microsite conditions. Lack of strong physical barriers and rather extensive lava fields may suggest wind to be as strong or stronger on Iceland compared to in the mountainous/forested regions of Norway and sandy beach and cliffy landscape of Sweden. To address this, average wind speed for different regions was acquired from meteorological stations nearby the sampling areas. Data for Norwegian, Swedish and Icelandic localities were obtained from the eKlima web portal of the Norwegian meteorological institute, http://www.tutiempo.net and http://en.vedur.is/climatology/data/, respectively. The average wind speed measured at meteorological stations in the study areas in Iceland, Sweden and Norway was 5.0, 3.0 and 2.8 m/s, respectively.
Soil and leaf sampling
A. lyrata populations can form extensive seed banks , and as these may significantly increase contemporary and historical effective population sizes , , we were interested in investigating whether seed bank constituents differed between Iceland and that observed in Norwegian populations. We collected leaf tissue from the above-ground individuals from ten Icelandic populations within a 152×129 km area in the western parts of Iceland in August 2009 (63.881–65.397°N), at altitudes varying from 8 to 231 m a.s.l. (Table 1). Most of the populations included here (n = 6) have previously been studied by Schierup et al.  to investigate the geographical distribution of nucleotide variation at the self-incompatibility (SI) locus. Due to the widespread distribution pattern and the relatively open topography of Icelandic sites, it seems plausible that local sites included in this analysis form metapopulation systems; a study by Lundemo et al.  showed that small levels of genetic differentiation can arise among patches within a local population of A. lyrata.
To examine whether Icelandic populations form a soil seed bank, ten soil samples of 10 cm×10 cm, with a depth of 5 cm, were taken throughout each of the studied populations before the annual seed rain. However, only five samples could be collected from the spatially very restricted population 22. Previous germination trials of field collected seeds from Norway have shown that no stratification is required for successful germination (S. Lundemo, unpublished results). However, to ensure germination of as many seeds as possible, soil samples were stored at 4°C for at least one month. Soil samples were spread out in a thin layer (1 cm) on top of commercial potting soil in the greenhouse and germinated at 16 h day light, 20°C and 65% air humidity. Seedling emergence from the soil was monitored over a period of ten months, and all seedlings germinated within two months after sowing. At each monthly census, new seedlings were counted and removed. Leaf tissue from each seedling was collected and dried at 45°C overnight. Seed bank density was determined by dividing the total number of seedlings emerging from the soil by total volume of soil for each population, following Falahati-Anbaran et al. .
Genomic DNA was extracted from leaf tissue using E-Z 96 Plant DNA Kit (Omega Bio-Tek, GA, USA). Twenty-one primer pairs labeled with different fluorescent dyes were used in 3 separate multiplex PCR reactions (Table S1 in File S1). The total volume of each PCR reaction was 10 µl; containing 5 µl 2x Type-it ™ Microsatellite PCR Kit (Qiagen, Hilden, Germany), 1 µl primer mix (10x) and 1 µl template DNA. A more detailed description of the PCR conditions and fragment analysis protocol can be found in .
Soil seed banks contribute to preservation of genetic variation in natural populations of A. lyrata , and it is therefore of interest to determine the extent and magnitude of seed banks in different regions when comparing patterns of genetic structuring. A Mann-Whitney U test was used to compare the seed bank density in Icelandic populations to that previously reported for 14 Norwegian populations . The number of seeds in the soil may be a function of the above-ground plant density and to examine this, a non-parametric correlation test (Spearman's ρ) was used to evaluate the association between soil and above-ground density. In total, 308 samples were obtained from above-ground and seed-bank cohorts and genotyped using 21 microsatellite markers. Because we did not find significant differences in seed bank density between Iceland and Norway (see below), further analysis of Icelandic samples was limited to the 263 above-ground individuals. Identical multilocus genotypes within populations were removed, i.e., only one individual from each multilocus genotype (genet) was included in the genetic analyses per population in order to avoid estimation bias due to clonality . Following this procedure, 15 individuals were found to be likely the result of clonal propagation, and therefore excluded from the data set, leaving 248 individuals for further analyses. The probability that two individuals with identical multilocus genotypes would be present in a population (PI) was calculated as 2(∑Pi2)2 -∑Pi4 , where pi is the frequency of the ith allele at a given locus and multiplied over all loci. The calculated PI in the study population was less than 1.6×10−8.
The 248 individuals were subsequently used to compare patterns of genetic variability with the Scandinavian populations included by Gaudeul et al. , who also only used above-ground material for genetic analyses. The proportion of genets (G/N) was calculated for each population by dividing the number of unique individuals (G) by total number of genotyped samples (N). For each population, the average observed (HO) and expected heterozygosity (HE) over all loci was estimated using Genepop version 4.1 . To test for Hardy-Weinberg equilibrium, the global exact test for heterozygote deficiency was carried out by estimating the exact P-value using a Markov Chain algorithm . P-value for multiple comparisons was corrected with the sequential Bonferroni correction . Because the presence of null alleles may lead to an increase in the homozygote frequencies compared to that would be expected under Hardy-Weinberg equilibrium, the frequency of null alleles was computed using an expectation maximization algorithm  implemented in the program FreeNA . Inbreeding coefficient (population specific FIS) and the statistical significance of FIS was determined by 10,000 random permutations of alleles between individuals using Arlequin 3.5 . The proportion of polymorphic loci (PPL) was calculated by dividing the number of polymorphic loci by the total number of loci for each population. Allelic richness (RS), the number of alleles adjusted for variation in sample size, and private allelic richness were estimated over all loci for each Icelandic population using a rarefaction method implemented in HP-RARE 1.0 . The squared differences in lengths between two alleles at a locus averaged over all loci (md2) was calculated for each individual  and average md2 across individuals was reported for each population. Neutral alleles at a locus will become progressively more differentiated from one another over evolutionary time due to stepwise mutations in SSR (simple sequence repeat) markers, and a positive relationship between md2 and time since divergence is expected . We also estimated HE, FIS, RS, RP and md2 for Norwegian and Swedish populations based on previously published data  using 17 loci which were common with the Icelandic dataset (Table S1 in File S1). Analysis of variance was used to examine regional differences for HE, FIS, RS, RP and md2. Multiple comparisons between regions were performed using a post hoc Bonferroni test. All parameters were tested for normality and homogeneity of variance prior to analysis, and md2 was square root transformed prior to analysis in order to meet assumptions of the test. The average frequency of null alleles across populations for all loci and between regions was compared using a paired t-test.
To detect recent declines in effective population sizes due to bottlenecks, a one-tailed Wilcoxon signed rank (WSR) test with sequential Bonferroni correction of P-values was performed to assess genetic diversity excess under mutation–drift equilibrium relative to that expected from the number of alleles in a sample. The two-phase mutation model (TPM) was used for probability estimation based on 10,000 replications as implemented in Bottleneck version 1.2 . In the TPM model, the proportion of stepwise and multi-step mutations for TPM was set to 70% and 30%, respectively. The program computes the expected equilibrium heterozygosity (Heq) from the number of alleles for each locus and population, and tests the deviation of Heq from observed gene diversity (HE). The bottlenecked populations should show a decline in both the number of alleles and Heq under mutation drift equilibrium, and thus Heq might be smaller than HE .
Genetic structure of Icelandic populations was assessed by computing pairwise FST among populations. The magnitude of regional FST can reflect the strength of the homogenizing effect of gene flow among populations in each region. To investigate regional differences in population differentiation, pairwise FST between Icelandic populations was compared to that in Scandinavian populations. Due to non-independence of among-population FST values, we applied a randomization test following a one-way analysis of variance (ANOVA) to test the null hypothesis of no difference between regions using EcoSim . The presence of null alleles may bias estimates of population differentiation, and unbiased FST measures accounting for null alleles were computed using FreeNA . These measures were then compared to the uncorrected FST for each region and one-way ANOVA was used to test for differences. For each comparison, a pseudo F ratio similar to a standard F ratio was computed and compared to the simulated F ratio calculated using a randomization test with 1,000 iterations to determine the probability that the observed F ratio was greater or less than that expected by chance. To test whether variation in size of distribution area inflates genetic differentiation levels, we limited the analysis to populations from areas of similar sizes and population numbers across regions. For this, five populations from each area were selected; Iceland (15, 16, 17, 22 and 23), Sweden (S3, S6, S10, S15 and S19) and Norway (N5, N6, N8, N9 and N10). In order to test for isolation by distance, the relationship between log10 [FST/(1-FST)] and geographical (log10 transformed) distances was studied using a Mantel test implemented in GenAlEx 6.4 . To test whether the pattern of isolation by distance differed among regions, i.e., differing slopes of regression lines (b), an analysis of covariance (ANCOVA) was conducted.
Genetic structure of Icelandic populations was also investigated using a Bayesian clustering analysis as implemented in Structure 2.3.3 , and the computations were conducted in https://lifeportal.uio.no/. A model of admixture together with correlated allele frequencies was applied for parameter estimation using 2×105 Markov Chain Monte Carlo (MCMC) iterations, following a burn-in period of 1×105 iterations. The likely number of clusters was obtained by plotting the average ln probability of data [ln P(D)] over 10 independent runs for K = –. We also calculated ΔK based on an ad hoc method to determine the conservative minimum number of clusters . The matrices of membership coefficients across 10 independent runs were used to search for the optimal alignment using a greedy algorithm with 1×105 permutations in CLUMPP version 1.1.2 . Membership coefficients of individuals to the inferred clusters was plotted for each K using DISTRUCT version 1.1 .
To investigate the importance of evolutionary processes on patterns of genetic structuring, we estimated immigration rates and effective population sizes (Ne) for each population. The rate of contemporary immigration into populations was estimated as the probability that individuals belong to their respective populations, and computed with a Monte-Carlo resampling of 1×104 individuals using a Bayesian approach implemented in GeneClass2 . Individuals with assignment probability >0.01 to its respective population were considered to be residents, otherwise individuals were considered immigrants . The mutation-scaled historical effective population size (θ) and immigration rate (M) were simultaneously estimated using a coalescence-based maximum likelihood approach implemented in Migrate 3.2 , . The mutation-scaled migration rate (M = m/u, where m and u are immigration rate and mutation rate per generation, respectively), reflects the average rate of immigration relative to mutation rate through historical time. Migrate assumes constant population size over time, a symmetric migration rate between populations and that all populations exchanging genes have been sampled . Unsampled populations should have a negligible impact on M = m/u, but can bias the effective population size estimates upward . A Brownian motion approximation to the stepwise mutation model was applied and MCMC simulation was conducted with 10 short chains (sampling at 1×104 trees) following one long chain (sampling at 5×105 trees). The simulation was repeated five times to improve the accuracy of results as suggested by Beerli and Felsenstein  and the average over five replicates with similar parameter setting was reported for θ and M. In addition, θ and M were estimated based on the same markers for Scandinavian populations. The historical Ne demonstrates the long term effect of evolutionary processes such as mutations and random genetic changes on genetic variation. Under mutation-drift equilibrium the amount of genetic variation is expected to be θ = 4 Neu in a diploid population, where θ, is four times the product of the effective population size (Ne) and mutation rate per generation (u) for microsatellite loci. Historical Ne can be estimated by calculating θ and dividing by estimates of u, or compared between regions by dividing θ estimates, thereby canceling out the mutation rate, i.e., assuming that neutral mutation rates do not vary across the distribution range. Analysis of variance was conducted to compare θ and M between regions, and multiple post hoc comparisons based on the Bonferroni method was used to test for differences between regions. All statistical tests were conducted using SPSS version 16.0.
The seed banks of the Icelandic populations contained on average (± SE) 26.8±16.6 seedlings/m2 soil sampled (median 7.59, CV = 196%, Table 2). Although this was considerably smaller than in Norwegian populations (average = 67.3 seedlings/m2, median = 23.60, C.V. = 136%; , a Mann-Whitney U test revealed no significant difference between regions (U = 41; ntotal = 24, two-tailed P = 0.088). The percentage of populations either lacking a seed bank entirely or with <10 seedlings/m2, were 21% and 70% for Norway and Iceland, respectively (Table S2 in File S1). The above-ground density of plants on Iceland (average number of individuals/m2 ± SE) was 1.38±0.58, (median = 0.64, C.V. = 133), not significantly different from that observed in Norwegian populations (average ± SE = 3.09±1.19, median = 1.17, C.V. = 125, U = 40, two-tailed P = 0.186, Table S2 in File S1). There was no significant relationship between density of above-ground plants and seed bank density in Icelandic (Spearman's ρ = 0.46, P = 0.18) or Norwegian (Spearman's ρ = 0.36, P = 0.20) populations.
Differences in within-population genetic diversity between regions
In total, 134 alleles were detected across 21 microsatellite loci in Icelandic populations, with the average (± SD) number of alleles per locus being 6.38±5.92 (range 1–26). Four loci (ELF3, ICE3, MSAT2.22 and nga112; Table S1 in File S1) were excluded from estimation of genetic diversity parameters in order to enable comparisons with previous estimates from Norway and Sweden . Because of the lack of significant differences in seed bank parameters between Iceland and Norway, we excluded genotypic data from seed banks from subsequent genetic analyses, enabling a direct comparison with results presented in . In total 77 alleles were scored in the 17 loci from above-ground Icelandic plants, and the average number of alleles was found to be 4.53±3.34 (range 1–14, Table S1 in File S1). The average proportion of genets (± SE), G/N was 0.94±0.02. All loci except nga162 were polymorphic in at least one population, and the average proportion of polymorphic loci (PPL) over all Icelandic populations was 0.76±0.02 (± SE). Estimates of HO, HE, FIS, PPL, RS and RP for each population are presented in Table 2. Genetic diversity (HE) varied considerably among loci (range 0.008–0.643), but less between Icelandic populations, with an average of 0.35 (range 0.32–0.39), with no populations showed significant heterozygote deficiency after Bonferroni corrections (P>0.005). None of the FIS values were significantly different from zero (P>0.05, Table 2). Three loci (ICE6, ICE7 and F19K23-483) showed heterozygote deficiency after sequential Bonferroni correction (P<0.0035), probably due to presence of null alleles. Additionally, the average (± SE) frequency of null alleles across populations over all loci was 0.02±0.004, and only three loci (ICE6, ICE7 and F19K23-483) exhibited a null allele frequency higher than 0.05. Average (± SE) allelic richness and private allelic richness over all populations was 2.28±0.05 (range 2.03–2.55) and 0.13±0.03 (range 0.01–0.26), respectively. The mean d (md2) (± SE) over all Icelandic populations was 32±1.7. Bottleneck tests revealed a signature of recent reduction in effective size in three populations (3, 17 and 21) after sequential Bonferroni correction on the P-value, with significant heterozygosity excess compared to expectations under mutation-drift equilibrium based on the two-phased model (TPM).
Analysis of variance revealed a significant difference among regions for genetic variation (HE; F2,33 = 20.46, P<0.001), allelic richness (RS; F2,33 = 4.7, P = 0.016), private allelic richness (RP; F2,33 = 5.06, P = 0.012), and md2 (F2,33 = 32.65, P<0.001), but no difference in FIS (F2,33 = 0.10, P = 0.41; Table S2 and Figure S1 in File S1). The mean (± SE) standing genetic variation (HE) in Icelandic (n = 10, 0.35±0.007) was similar to Swedish (n = 12, 0.33±0.015) populations and both were significantly higher than Norwegian populations based on post hoc Bonferroni test (n = 14, 0.25±0.011; P<0.01). Pairwise comparison revealed that allelic richness, RS, was significantly higher in Icelandic populations (average ± SE, 2.28±0.05) than in Norwegian (average ± SE, 2.03±0.07; P = 0.027) populations, and similar to Swedish populations (average ± SE, 2.23±0.06). Private allelic richness was significantly higher in Icelandic (average 0.13±0.030) compared to Norwegian (average 0.05±0.01; P = 0.032) and Swedish (average 0.05±0.02; P = 0.020) populations, whereas Norwegian and Swedish populations did not differ from each other (P>0.1). Md2 in Icelandic populations was significantly higher than that in Norwegian (average 13.58±1.66; P<0.01) and Swedish (average 19.42±1.53; P<0.01) populations based on post hoc Bonferroni tests. The average (± SE) frequency of null alleles for all loci across Icelandic populations (0.02±0.004) was similar to that for Norwegian (0.02±0.005, t = 1.35, two-tailed P = 0.19) and Swedish populations (0.03±0.004, t = 1.02, two-tailed P = 0.32). No significant difference was observed between Norway and Sweden (t = 0.07, two-tailed P = 0.95)
Spatial genetic structure across regions
The average FST (± SE) among Icelandic populations was 0.10±0.01 (range 0.010–0.23). One-way ANOVA revealed a significant difference among regions in levels of genetic differentiation. The observed F ratio for Norway-Sweden, Norway-Iceland and Sweden-Iceland comparisons were 46.60, 133.80, and 44.28, respectively (all P<0.001). The unbiased FST corrected for null alleles was similar to uncorrected FST for Icelandic (observed F ratio = 0.03, P = 0.87), Norwegian (observed F ratio = 0.13, P = 0.71) and Swedish (observed F ratio = 0.05, P = 0.83) populations. The average FST (± SE) was higher in Norwegian (0.30±0.12, range 0.01–0.52) than Swedish populations (0.19±0.01, range 0.05–0.45, P<0.001), which in turn was higher than in Icelandic populations (P<0.001). Geographical and genetic distances were significantly correlated in the samples from Iceland (r = 0.55, n = 45, P = 0.001), Norway (r = 0.57, n = 91, P<0.001), and Sweden (r = 0.27, n = 66, P = 0.008; Figure 1). Analysis of covariance revealed a significant country × geographical distance interaction, indicating dissimilar regression slopes between countries (F2,196 = 5.37, P = 0.005). The regression slope of Swedish populations (b = 0.15) differed significantly from Norwegian (b = 0.44; F1,153 = 9.00, P = 0.003) and Icelandic (b = 0.45; F2,107 = 5.70, P = 0.019) populations, whereas the two latter regions did not differ (F1,132 = 0.007, P = 0.92). We also investigated genetic differentiation by including five populations at similar geographic scales in Icelandic (mean between-population distance = 56.0 km, range 8.6–106), Norwegian (mean = 60.7 km, range 2.6–82.6), and Swedish populations (mean = 46.3 km, range 9.1–85.3). This approach resulted in average FST estimates being reduced to 0.07, 0.17 and 0.20 for Icelandic, Swedish and Norwegian populations, respectively, but regions were still significantly differentiated (F1,27 = 7.23, P<0.001). Accordingly, a post hoc Bonferroni test revealed that the population differentiation in Iceland was significantly lower than that in Norway (P = 0.003) and Sweden (P = 0.019), but the latter regions did not differ significantly from each other (P = 0.932).
Blue, red and green represent Swedish (r = 0.27, n = 66, P = 0.008), Norwegian (r = 0.57, n = 91, P = 0.001) and Icelandic ( = 0.55, n = 45, P = 0.001) populations, respectively.
The optimal number of clusters in Icelandic populations using Bayesian clustering revealed a peak for ΔK at K = 2 based on the method  (Figure 2), splitting the northern and southern populations (Figure 3a, b). However, based on the mean ln P(D), the highest peak, representing the optimal number of clusters, was observed for K = 7 (Figure 2). The results based on K = 7 showed that populations are geographically structured into several ancestral genetic clusters (Figure S2a, b in File S1).
Vertical lines indicate the standard deviation for the mean probability of data.
Immigration rate and historical effective population size
Bayesian analyses showed that none of the populations from the different regions had experienced a recent immigration event (Figure S3 in File S1). The average historical immigration rate (mutation-scaled migration rate; M ± SE) was 1.06±0.03, 1.05±0.05 and 0.95±0.03 per generation for populations from Iceland, Sweden and Norway, respectively, implying that migration and mutation rates are similar. Furthermore, no significant difference was found among the regions for historical migration rates (F2,33 = 2.56, P = 0.093). The average (± SD) historical effective population size (estimated as θ) was 0.46±0.03 in Icelandic populations, 0.29±0.05 in Norway and 0.32±0.07 in Sweden (Table S2 in File S1). Historical Ne differed significantly between regions (F2,33 = 19.42, P<0.01, Figure S1 in File S1), and a Bonferroni test indicated significantly higher historical Ne in Icelandic than Norwegian and Swedish populations (P<0.01).
Variation in ecological characteristics
Old and stable populations may contain more allelic variants than short-lived, unstable populations. Furthermore, ecological features like soil seed banks may contribute to maintaining genetic diversity by preserving alleles descended from multiple generations in the past, resulting in increased effective sizes of plant populations , , . Despite a similar sampling regime, the percentage of populations lacking a seed bank in Iceland (30%) is seemingly twice as high compared to a previous study of Norwegian populations (14%) . Our results thus indicate that Icelandic populations maintain small seed banks relative to the Norwegian populations of A. lyrata, although the difference is not statistically significant. Seed bank densities observed in our study are similar to those documented for A. lyrata by Marteinsdóttir et al.  in an outwash plain in southern Iceland (average = 27 seeds/m2). Moreover, above-ground plant density do not differ between the countries . In our study, it seems that above-ground plant density does not affect density of ungerminated seeds in soil seed banks, and other factors such as soil substrate, dormancy, percentage of adult plants and/or reproductive output can be more important in this respect (cf. , ). Since our results show no significant differences between Iceland and Scandinavia regarding seed bank characteristics, we have compared the above-ground statistics for various regions assuming that the differences observed for genetic parameters are likely not due to differences in seed dormancy patterns.
Within-population genetic diversity and spatial genetic structure
The average genetic diversity in Icelandic populations (HE = 0.35) is similar to what is found in Swedish populations (HE = 0.33; , which in turn is higher than that observed in Norwegian populations (HE = 0.25). This is consistent with Muller et al. , who also found a higher level of genetic diversity in Icelandic than Norwegian populations, based on results from one and two populations from Iceland and Norway, respectively. Most of the genetic variation in Icelandic A. lyrata is partitioned within populations, with a low but significant level of among-population differentiation (average FST = 0.10). The level of genetic differentiation among Icelandic populations (mean distance = 93 km) is ∼2- and 3-fold lower than that in Swedish (mean distance = 40 km) and Norwegian (mean distance = 165 km) populations, respectively. Furthermore, the level of differentiation in Icelandic populations remains significantly lower than Norwegian and Swedish populations after adjusting for geographical scale and population number across regions. This suggests that the wider distribution area on Iceland cannot explain differences in population differentiation between the studied regions.
Our sampling regime has covered only the western parts of Iceland, but this has probably not had much or any impact on our estimates of genetic structure in Icelandic populations. It is worth noting that the Icelandic populations used for between region comparisons accounting for similar sized distribution area (see material and methods), exhibit similar levels of genetic diversity within populations compared to all Icelandic populations sampled in our study. Thus, the range of within-population diversity is narrow (HE = 0.32–0.39) on Iceland compared to Norwegian (HE = 0.–0.34) and Swedish (HE = 0.21–0.38) populations. This suggests that increasing the sampling efforts to more than ten studied populations will likely not change our estimates.
Homogeneous and continuous habitats may facilitate gene flow and reduce genetic differentiation between populations , though it is not straightforward to assess to what degree differences in dispersal vector efficiency vary among studied regions. The sparse tree cover and lack of strong topographic barriers over large geographical areas on Iceland can potentially cause wind to act as a dispersal agent and facilitate seed dispersal over longer distances compared to that in more closed vegetation types. However, several factors make it difficult to conclude on this. First, even though wind speed measured at meteorological stations on average is stronger on Iceland compared to in Swedish and Norwegian study areas, it is not given that efficiency of wind at canopy level is any different among regions. Microsite wind patterns can be different from average measures on local or regional levels, and e.g. localities in Norwegian mountain are expected to be strongly influenced by wind. Second, and perhaps more importantly, as A. lyrata is insect pollinated , , high wind speed could influence pollen flow negatively by restricting pollinator activity –.
The wider distribution area and rugged mountainous landscape in the Norwegian compared to the Swedish distribution area could explain why population differentiation is higher among Norwegian relative to Swedish populations . The increased population differentiation as a function of geographical distance suggests that dispersal via pollen or seed is distance dependent, i.e., with a higher level of gene flow at shorter relative to longer distances . The IBD pattern for different regions shows that increased differentiation as a function of distance is stronger in Norwegian and Icelandic relative to Swedish populations (Figure 1). These results indicate that pollen and/or seed dispersal is more distance-restricted in Icelandic and Norwegian populations relative to Swedish populations . A. lyrata seeds are primarily gravity-dispersed without specialized mechanisms for long-distance dispersal in open habitats, and large distribution areas on Iceland and perhaps Norway during or after the last glacial period may explain the difference in IBD patterns to that observed in Swedish populations. Additionally, multiple colonization routes into the Norwegian and Icelandic distribution ranges may account for the observed differences in the IBD pattern , .
It should be noted that homoplasy and presence of null alleles in microsatellite loci may influence population genetic estimates . Allele size homoplasy has been reported for some microsatellite loci in A. lyrata, and Muller et al.  suggest that size homoplasy may bias estimates of between region divergence. We find frequency of null alleles to be consistently low in populations from all three regions, and these alleles have no detectable effect on estimates of population differentiation. It has been argued that the presence of size homoplasy will be problematic while reconstructing phylogenetic history, but has been found to have no significant effect on population parameters because the effect is compensated by high variability at microsatellite loci .
Historical and contemporary processes in different regions
We observe that the historical effective population size (Ne) in the Icelandic region is considerably larger than in Swedish populations while levels of genetic diversity are similar. On the other hand, Icelandic populations exhibit higher allelic richness and genetic diversity, and has larger effective population size compared to the Norwegian region. The larger effective population size of Icelandic than Swedish populations may be related to differences in private allelic richness between regions, possibly explained by older age of the Icelandic compared to the Swedish metapopulation. Interestingly, we find that the average mutation-scaled migration rate (M) do not differ between regions, suggesting similar rates of immigration over time in populations of different regions. In addition, we find no evidence of contemporary migration into populations in different regions using an assignment test, suggesting limited gene flow on both ecological and evolutionary time-scales.
Mean d2 (md2) has been used as measure of divergence time between lineages within populations over time . A higher md2 in Icelandic populations compared to Norwegian and Swedish populations suggests longer coalescence times for Icelandic allelic variants, and hence possibly older populations on Iceland than in Scandinavia. Even though there is a significant difference between regions for md2, the presence of null alleles may negatively bias md2 by increasing the observed frequency of homozygotes. However, since the frequency of null alleles is low and similar between regions, this may not be influencing the patterns observed in the present study. Additionally, md2 is sensitive to inbreeding patterns , but this is also likely not an issue here since all populations exhibit FIS equal to zero. Finally, the extent of allele length differences could be due to admixture of divergent lineages, for instance central European and Icelandic lineages in relatively recent time (cf. ), but several observations argue against this possibility. First, recent admixture should be reflected in subsequent heterozygote deficiency over studied loci, and despite indications of heterozygote deficiency at three loci when analyzing all samples together (probably due to null alleles), we find no evidence of heterozygote deficiency at the population level. Second, immigrant individuals carrying other genotypes than in residents might be expected to have a low probability of assignment to the population to which it has immigrated . Our assignment analysis shows no overall differences in the pattern of assignment probability between regions, and seemingly no indication of differences in recent admixture patterns. Thus, contrasting patterns of population differentiation between regions seems more likely due to glacial survival of Icelandic populations and differences in the demographic histories of populations, including larger historical Ne in Icelandic than Norwegian and Swedish populations, than migration rates per se . Several studies using genetic and paleobotanical data have suggested glacial survival of plant species in Icelandic and Scandinavian refugia –. Interestingly, the presence of seeds of A. lyrata ssp. petraea has also been recorded in the Ballybetagh bogs near Dublin in Ireland and dated to the end of last glacial period , and a recent study using nuclear genes have suggested that northern European populations of A. lyrata diverged before the last glacial maximum .
Our results show that there is large variation in genetic diversity, population structure and effective sizes in natural populations of A. lyrata inhabiting three topographically contrasting regions in its northernmost European range. Population differentiation in Icelandic populations is unexpectedly low, even though immigration rates do not differ between regions. The observed low population differentiation can be due to large effective population size rather than differences in historical or ongoing gene flow between populations. Interestingly, Ansell et al.  and Vigueira et al.  found indications that the species may have survived last glacial maximum in a cryptic northern refugium. One may thus hypothesize whether the higher effective size of Icelandic populations reflects old ages of lineages residing in stable refugial populations. The strong isolation by distance pattern observed in our study is in contrast to results reported by Schierup et al. , suggesting a recent colonization of Icelandic populations because of absence of IBD and low population differentiation at the self-incompatibility locus. However, it seems possible that this could be due to frequency-dependent selection and rapid homogenization of rare S-alleles , as opposed to likely neutral or near-neutral evolution at microsatellite markers. To conclude, we find that Icelandic populations express different patterns of genetic structuring compared with Scandinavian populations, probably representing an older colonization history than that previously assumed and possibly reflecting relatively more stable population sizes through time.
File contains Tables S1 and S2 and Figures S1, S2, and S3. Table S1. Characteristics of microsatellite loci used to describe the genetic structure in natural populations of Arabidopsis lyrata in Iceland. Table S2. Above-ground plant and seed bank density, and estimates of genetic parameters based on 17 microsatellite markers in Icelandic, Norwegian and Swedish populations of Arabidopsis lyrata. Figure S1. Heterozygosity (HE), effective population size (Ne), allelic richness (RS) and private allelic richness (RP) in Icelandic, Norwegian and Swedish populations. Figure S2. Genetic structure of Icelandic populations of Arabidopsis lyrata detected by Structure, with optimal number of clusters determined by the highest peak for mean ln P(D). Figure S3. Probability of assignment of individuals to their respective population for different regions.
We wish to thank Jon Ågren and three anonymous reviewers for comments on the manuscript, Hanna Maja Tunset for help with fragment analysis and Jenny Hagenblad for providing geographic coordinates of the Icelandic populations.
Conceived and designed the experiments: MFA SL HKS. Performed the experiments: MFA SL. Analyzed the data: MFA SL HKS. Contributed reagents/materials/analysis tools: MFA SL HKS. Wrote the paper: MFA SL HKS SA.
- 1. Sork VL, Nason J, Campbell DR, Fernandez JF (1999) Landscape approaches to historical and contemporary gene flow in plants. Trends Ecol Evol 14: 219–224.
- 2. Manicacci D, Olivieri I, Perrot V, Atlan A, Gouyon P-H, et al. (1992) Landscape ecology: population genetics at the metapopulation level. Landsc Ecol 6: 147–159.
- 3. Manel S, Schwartz MK, Luikart G, Taberlet P (2003) Landscape genetics: combining landscape ecology and population genetics. Trends Ecol Evol 18: 189–197.
- 4. Storfer A, Murphy MA, Spear SF, Holderegger R, Waits LP (2010) Landscape genetics: where are we now? Mol Ecol 19: 3496–3514.
- 5. Holderegger R, Buehler D, Gugerli F, Manel S (2010) Landscape genetics of plants. Trends Plant Sci 15: 675–683.
- 6. He T, Lamont BB, Krauss SL, Enright NJ (2010) Genetic connectivity and inter-population seed dispersal of Banksia hookeriana at the landscape scale. Ann Bot 106: 457–466.
- 7. Hutchison DW, Templeton AR (1999) Correlation of pairwise genetic and geographic distance measures: inferring the relative influences of gene flow and drift on the distribution of genetic variability. Evolution 53: 1898–1914.
- 8. Wright S (1943) Isolation by distance. Genetics 28: 114–138.
- 9. Kimura M, Weiss GH (1964) The stepping stone model of population structure and the secrease of genetic correlation with distance. Genetics 49: 561–576.
- 10. Lenormand T (2002) Gene flow and the limits to natural selection. Trends Ecol Evol 17: 183–189.
- 11. Moyle LC (2006) Correlates of genetic differentiation and isolation by distance in 17 congeneric Silene species. Mol Ecol 15: 1067–1081.
- 12. Brown JH, Stevens GC, Kaufman DW (1996) The geographic range: size, shape, boundaries and internal structure. Annu Rev Ecol Syst 27: 597–623.
- 13. Hewitt G (2000) The genetic legacy of the quaternary ice ages. Nature 405: 907–913.
- 14. Crispo E, Hendry A (2005) Does time since colonization influence isolation by distance? A meta-analysis. Conserv Genet 6: 665–682.
- 15. Hewitt GM (2004) Genetic consequences of climatic oscillations in the Quaternary. Phil Trans R Soc B 359: 183–195.
- 16. Schmitt T (2007) Molecular biogeography of Europe: pleistocene cycles and postglacial trends. Front Zool 4: 1–13.
- 17. González-Wevar C, Hüne M, Cañete J, Mansilla A, Nakano T, et al. (2012) Towards a model of postglacial biogeography in shallow marine species along the Patagonian Province: lessons from the limpet Nacella magellanica (Gmelin, 1791). BMC Evol Biol 12: 1–17.
- 18. Marko PB (2004) What's larvae got to do with it? disparate patterns of post-glacial population structure in two benthic marine gastropods with identical dispersal potential. Mol Ecol 13: 597–611.
- 19. Slatkin M (1993) Isolation by distance in equilibrium and non-equilibrium populations. Evolution 47: 264–279.
- 20. Peterson M, Denno R (1998) The influence of dispersal and diet breadth on patterns of genetic isolation by distance in phytophagous insects. Am Nat 152: 428–446.
- 21. Schierup MH, Bechsgaard JS, Christiansen FB (2008) Selection at work in self-incompatible Arabidopsis lyrata. II. spatial distribution of S haplotypes in Iceland. Genetics 180: 1051–1059.
- 22. Schierup MH, Bechsgaard JS, Nielsen LH, Christiansen FB (2006) Selection at work in self-incompatible Arabidopsis lyrata: mating patterns in a natural population. Genetics 172: 477–484.
- 23. Muller MH, Leppala J, Savolainen O (2008) Genome-wide effects of postglacial colonization in Arabidopsis lyrata. Heredity 100: 47–58.
- 24. Clauss MJ, Mitchell-Olds T (2006) Population genetic structure of Arabidopsis lyrata in Europe. Mol Ecol 15: 2753–2766.
- 25. Mable BK, Robertson AV, Dart S, Berardo CD, Witham L (2005) Breakdown of self-incompatibility in the perennial Arabidopsis Lyrata (Brassicaceae) and its genetic consequences. Evolution 59: 1437–1448.
- 26. Ansell SW, Stenoien HK, Grundmann M, Schneider H, Hemp A, et al. (2010) Population structure and historical biogeography of European Arabidopsis lyrata. Heredity 105: 543–553.
- 27. Davey M, Ian Woodward F, Paul Quick W (2009) Intraspecfic variation in cold-temperature metabolic phenotypes of Arabidopsis lyrata ssp. petraea. Metabolomics 5: 138–149.
- 28. Davey MP, Burrell MM, Woodward FI, Quick WP (2008) Population-specific metabolic phenotypes of Arabidopsis lyrata ssp. petraea. New Phytol 177: 380–388.
- 29. Karkkäinen K, Løe G, ÅGren J (2004) Population structure in Arabidopsis lyrata: evidence for divergent selection on trichome production. Evolution 58: 2831–2836.
- 30. Leinonen PH, Remington DL, Savolainen O (2011) Local adaptation, phenotypic differentiation and hybrid fitness in diverged natural populations of Arabidopsis lyrata. Evolution 65: 90–107.
- 31. Leinonen PH, Sandring S, Quilot B, Clauss MJ, Mitchell-Olds T, et al. (2009) Local adaptation in European populations of Arabidopsis lyrata (Brassicaceae). Am J Bot 96: 1129–1137.
- 32. Ross-Ibarra J, Wright SI, Foxe JP, Kawabe A, DeRose-Wilson L, et al. (2008) Patterns of polymorphism and demographic history in natural populations of Arabidopsis lyrata. PLoS ONE 3: e2411.
- 33. Schmickl R, Jorgensen M, Brysting A, Koch M (2010) The evolutionary history of the Arabidopsis lyrata complex: a hybrid in the amphi-Beringian area closes a large distribution gap and builds up a genetic barrier. BMC Evol Biol 10: 98.
- 34. Turner TL, Bourne EC, Von Wettberg EJ, Hu TT, Nuzhdin SV (2010) Population resequencing reveals local adaptation of Arabidopsis lyrata to serpentine soils. Nat Genet 42: 260–263.
- 35. Turner TL, von Wettberg EJ, Nuzhdin SV (2008) Genomic analysis of differentiation between soil types reveals candidate genes for local adaptation in Arabidopsis lyrata. PLoS ONE 3: e3183.
- 36. Wright SI, Lauga B, Charlesworth D (2003) Subdivision and haplotype structure in natural populations of Arabidopsis lyrata. Mol Ecol 12: 1247–1263.
- 37. Vergeer P, Kunin WE (2013) Adaptation at range margins: common garden trials and the performance of Arabidopsis lyrata across its northwestern European range. New Phytol 197: 989–1001.
- 38. Toivainen T, Pyhäjärvi T, Niittyvuopio A, Savolainen O (2014) A recent local sweep at the PHYA locus in the Northern European Spiterstulen population of Arabidopsis lyrata. Mol Ecol 23: 1040–1052.
- 39. Bergelson J, Roux F (2010) Towards identifying genes underlying ecologically relevant traits in Arabidopsis thaliana. Nat Rev Genet 11: 867–879.
- 40. Gaudeul M, Stenøien HK, Ågren J (2007) Landscape structure, clonal propagation, and genetic diversity in Scandinavian populations of Arabidopsis lyrata (Brassicaceae). Am J Bot 94: 1146–1155.
- 41. Schierup MH (1998) The effect of enzyme heterozygosity on growth in a strictly outcrossing species, the self-incompatible Arabis Petraea (Brassicaceae). Hereditas 128: 21–31.
- 42. Sousa VC, Carneiro M, Ferrand N, Hey J (2013) Identifying loci under selection against gene flow in isolation-with-migration models. Genetics 194: 211–233.
- 43. Kamau E, Charlesworth B, Charlesworth D (2007) Linkage disequilibrium and recombination rate estimates in the self-incompatibility region of Arabidopsis lyrata. Genetics 176: 2357–2369.
- 44. Roux C, Pauwels M, Ruggiero M-V, Charlesworth D, Castric V, et al. (2013) Recent and cncient signature of balancing selection around the S-Locus in Arabidopsis halleri and A. lyrata. Mol Biol Evol 30: 435–447.
- 45. Falahati-Anbaran M, Lundemo S, Ågren J, Stenøien HK (2011) Genetic consequences of seed banks in the perennial herb Arabidopsis lyrata subsp. petraea (Brassicaceae). Am J Bot 98: 1475–1485.
- 46. Lundemo S, Falahati-Anbaran M, Stenøien HK (2009) Seed banks cause elevated generation times and effective population sizes of Arabidopsis thaliana in northern Europe. Mol Ecol 18: 2798–2811.
- 47. Jonsell B, Kustås K, Nordal I (1995) Genetic variation in Arabis petraea, a disjunct species in northern Europe. Ecography 18: 321–332.
- 48. Jalas J, Suominen J (1994) Atlas florae Europaeae: distribution of vascular plants in Europe. Vol. 10. Cruciferae (Sisymbrium to Aubrieta). Helsinki: The Committee for Mapping the Flora of Europe & Societas Biologica Fennica Vanamo. 224 p.
- 49. Lundemo S, Stenøien HK, Savolainen O (2010) Investigating the effects of topography and clonality on genetic structuring within a large Norwegian population of Arabidopsis lyrata. Ann Bot 106: 243–254.
- 50. Peakall R, Smouse P (2012) GenAlEx 6.5: Genetic analysis in Excel. Population genetic software for teaching and research – an update. Bioinformatics.
- 51. Rousset F (2008) genepop'007: a complete re-implementation of the genepop software for Windows and Linux. Mol Ecol Resour 8: 103–106.
- 52. Raymond M, Rousset F (1995) GENEPOP (Version 1.2): Population genetics software for exact rests and ecumenicism. J Hered 86: 248–249.
- 53. Rice WR (1989) Analyzing tables of statistical tests. Evolution 43: 223–225.
- 54. Dempster AP, Laird NM, Rubin DB (1977) Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society Series B (Methodological) 39: 1–38.
- 55. Chapuis M-P, Estoup A (2007) Microsatellite null alleles and estimation of population differentiation. Mol Biol Evol 24: 621–631.
- 56. Excoffier L, Lischer HEL (2010) Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Mol Ecol Resour 10: 564–567.
- 57. Kalinowski ST (2005) hp-rare 1.0: a computer program for performing rarefaction on measures of allelic richness. Mol Ecol Notes 5: 187–189.
- 58. Neff BD (2004) Mean d2 and divergence time: transformations and standardizations. J Hered 95: 165–171.
- 59. Piry S, Luikart G, Cornuet J-M (1999) Computer note. BOTTLENECK: a computer program for detecting recent reductions in the effective size using allele frequency data. J Hered 90: 502–503.
- 60. Cornuet JM, Luikart G (1996) Description and power analysis of two tests for detecting recent population bottlenecks from allele frequency data. Genetics 144: 2001–2014.
- 61. Gotelli NJ, Entsminger GL (2001) Garyentsminger website. EcoSim: null models software for ecology. Version 7.0. Acquired Intelligence Inc. & Kesey-Bear. Jericho, VT 05465. Available: http://garyentsminger.com/ecosim/index.htm. Accessed 2013 Jun 12.
- 62. Peakall R, Smouse PE (2006) Genalex 6: genetic analysis in Excel. population genetic software for teaching and research. Mol Ecol Notes 6: 288–295.
- 63. Falush D, Stephens M, Pritchard JK (2003) Inference of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics 164: 1567–1587.
- 64. Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using the software structure: a simulation study. Mol Ecol 14: 2611–2620.
- 65. Jakobsson M, Rosenberg NA (2007) CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23: 1801–1806.
- 66. Rosenberg NA (2004) distruct: a program for the graphical display of population structure. Mol Ecol Notes 4: 137–138.
- 67. Piry S, Alapetite A, Cornuet J-M, Paetkau D, Baudouin L, et al. (2004) GENECLASS2: a software for genetic assignment and first-generation migrant detection. J Hered 95: 536–539.
- 68. Cornuet J-M, Piry S, Luikart G, Estoup A, Solignac M (1999) New methods employing multilocus genotypes to select or exclude populations as origins of individuals. Genetics 153: 1989–2000.
- 69. Beerli P, Felsenstein J (1999) Maximum-likelihood estimation of migration rates and effective population numbers in two populations using a coalescent approach. Genetics 152: 763–773.
- 70. Beerli P, Felsenstein J (2001) Maximum likelihood estimation of a migration matrix and effective population sizes in n subpopulations by using a coalescent approach. Proc Natl Acad Sci U S A 98: 4563–4568.
- 71. Beerli P (2004) Effect of unsampled populations on the estimation of population sizes and migration rates between sampled populations. Mol Ecol 13: 827–836.
- 72. Honnay O, Bossuyt B, Jacquemyn H, Shimono A, Uchiyama K (2008) Can a seed bank maintain the genetic variation in the above ground plant population? Oikos 117: 1–5.
- 73. Marteinsdóttir B, Svavarsdóttir K, Thórhallsdóttir TE (2010) Development of vegetation patterns in early primary succession. J Veg Sci 21: 531–540.
- 74. Vergeer P, Kunin WE (2011) Life history variation in Arabidopsis lyrata across its range: effects of climate, population size and herbivory. Oikos 120: 979–990.
- 75. Sandring S, Ågren J (2009) Pollinator-mediated selection on floral display and flowering time in the perennial herb Arabidopsis lyrata. Evolution 63: 1292–1300.
- 76. Brown AO, McNeil JN (2009) Pollination ecology of the high latitude, dioecious cloudberry (Rubus chamaemorus; Rosaceae). Am J Bot 96: 1096–1107.
- 77. Totland O (1994) Influence of climate, time of day and season, and flower density in insect flower visitation in alpine Norway. Arct Alp Res 26: 66–71.
- 78. Brantjes NBM (1981) Wind as a factor influencing flower-visiting by Hadena bicruris (Noctuidae) and Deilephila elpenor (Sphingidae). Ecol Entomol 6: 361–363.
- 79. Castric V, Bernatchez L (2003) The rise and fall of isolation by distance in the Anadromous Brook Charr (Salvelinus fontinalis Mitchill). Genetics 163: 983–996.
- 80. Garnier S, Alibert P, Audiot P, Prieur B, Rasplus JY (2004) Isolation by distance and sharp discontinuities in gene frequencies: implications for the phylogeography of an alpine insect species, Carabus solieri. Mol Ecol 13: 1883–1897.
- 81. Selkoe KA, Toonen RJ (2006) Microsatellites for ecologists: a practical guide to using and evaluating microsatellite markers. Ecol Lett 9: 615–629.
- 82. Estoup A, Jarne P, Cornuet JM (2002) Homoplasy and mutation model at microsatellite loci and their consequences for population genetics analysis. Mol Ecol 11: 1591–1604.
- 83. Coulson T, Albon S, Jon S, Pemberton J (1999) Microsatellite loci reveal sex-dependent responses to inbreeding and outbreeding in red deer calves. Evolution 53: 1951–1960.
- 84. Rundgren M, Ingólfsson Ó (1999) Plant survival in Iceland during periods of glaciation? J Biogeogr 26: 387–396.
- 85. Westergaard KB, Alsos IG, Popp M, EngelskjØN T, Flatberg KI, et al. (2011) Glacial survival may matter after all: nunatak signatures in the rare European populations of two west-arctic species. Mol Ecol 20: 376–393.
- 86. Parducci L, Jørgensen T, Tollefsrud MM, Elverland E, Alm T, et al. (2012) Glacial survival of boreal trees in northern Scandinavia. Science 335: 1083–1086.
- 87. Jessen K, Farrington A (1938) The bogs at Ballybetagh, near Dublin, with remarks on late-glacial conditions in Ireland. Proc R Ir Acad 60: 1–77.
- 88. Vigueira CC, Rauh B, Mitchell-Olds T, Lawton-Rauh AL (2013) Signatures of demography and recombination at coding genes in naturally-distributed populations of Arabidopsis lyrata subsp. Petraea. PLoS ONE 8: e58916.