DIVERGENCE IS FOCUSED ON FEW GENOMIC REGIONS EARLY IN SPECIATION: INCIPIENT SPECIATION OF SUNFLOWER ECOTYPES (original) (raw)

Journal Article

,

Department of Botany, University of British Columbia, 3529-6270 University Blvd Vancouver British Columbia V6T 1Z4 Canada

Search for other works by this author on:

Biology Department Indiana University 1001 E Third St. Bloomington Indiana 47405

Search for other works by this author on:

Received:

26 September 2012

Accepted:

20 February 2013

Published:

01 September 2013

• PDF
• Split View
• • Article contents
  • Figures & tables
  • Video
  • Audio
  • Supplementary Data
• Cite

Cite

Rose L. Andrew, Loren H. Rieseberg, DIVERGENCE IS FOCUSED ON FEW GENOMIC REGIONS EARLY IN SPECIATION: INCIPIENT SPECIATION OF SUNFLOWER ECOTYPES, Evolution, Volume 67, Issue 9, 1 September 2013, Pages 2468–2482, https://doi.org/10.1111/evo.12106
Close

Navbar Search Filter Mobile Enter search term Search

Abstract

Early in speciation, as populations undergo the transition from local adaptation to incipient species, is when a number of transient, but potentially important, processes appear to be most easily detected. These include signatures of selective sweeps that can point to asymmetry in selection between habitats, divergence hitchhiking, and associations of adaptive genes with environments. In a genomic comparison of ecotypes of the prairie sunflower, Helianthus petiolaris, occurring at Great Sand Dunes National Park and Preserve (Colorado), we found that selective sweeps were mainly restricted to the dune ecotype and that there was variation across the genome in whether proximity to the nondune population constrained or promoted divergence. The major regions of divergence were few and large between ecotypes, in contrast with an interspecific comparison between H. petiolaris and a sympatric congener, Helianthus annuus. In general, the large regions of divergence observed in the ecotypic comparison swamped locus-specific associations with environmental variables. In both comparisons, regions of high divergence occurred in portions of the genetic map with high marker density, probably reflecting regions of low recombination. The difference in genomic distributions of highly divergent regions between ecotypic and interspecific comparisons highlights the value of studies spanning the spectrum of speciation in related taxa.

Speciation is a central component of biological diversification and is increasingly viewed as a continuum or process, rather than an event (Coyne and Orr 2004; Nosil et al. 2009a). The structure of genetic differentiation within the genomes of diverging species receives much attention as a key observable feature of this process. Studies of genomic patterns of divergence between species at various stages on the speciation continuum are accumulating rapidly, owing to advances in sequencing technology, as well as to the development of genomic resources, such as reference genomes and genetic maps (Turner et al. 2005; Hohenlohe et al. 2010; Lawniczak et al. 2010; Neafsey et al. 2010). Reports that genomic regions of high divergence (genomic “islands” of divergence; Turner et al. 2005) may be large in the presence of gene flow have stimulated work attempting to explain how genomes diverge (Via and West 2008; Noor and Bennett 2009; Feder et al. 2012; Hahn et al. 2012; Via 2012). However, there is little evidence for large genomic regions of divergence between species in plants or colocalization of outlier regions with known quantitative trait loci (QTL) underlying reproductive or species differences (Strasburg et al. 2012). This may be because most comparisons to date have been at the species level, where long histories of gene flow, the accumulation of genome-wide divergence and population structure might erode or obscure such patterns (Nadeau et al. 2012; Nosil 2012). Thus, there is an urgent need to study the genomic structure of divergence in the earliest stages of plant speciation. In particular, the number, width, and clustering of divergent regions within the genome of early incipient species, compared with more diverged congeneric pairs, is of interest.

The transition from local adaptation to incipient species is the first step of ecological speciation, but it is subtle and rarely an all-or-nothing change. If ecological speciation occurs in the presence of gene flow, then local adaptation is likely to be the initial barrier limiting gene flow between the populations. Other early-evolving reproductive isolating mechanisms may also be present, but not necessarily fixed, so that gene flow might be affected by geographic structure through endogenous mechanisms as well as exogenous ones (Noor 1999; Lowry et al. 2008; Scopece et al. 2010). The role played by gene flow in constraining or promoting divergence is a major unanswered question (Nosil 2012). It can both limit and be limited by adaptation to local conditions (Lenormand 2002; Räsänen and Hendry 2008), but divergence can be hastened through processes such as reinforcement, which requires gene flow (Noor 1999).

More generally, we need to better understand how variation in biology and geography affect patterns of genomic divergence. In addition to genome architecture and the strength and timing of selection, the impact of divergent selection on the genome is likely to be shaped by the rates of migration between habitats, genetic connectivity within habitat types, and the amount of standing genetic variation (Feder and Nosil 2010; Ralph and Coop 2010; Feder et al. 2012). The relative importance of genome hitchhiking (divergence due to the reduction of effective gene flow genome-wide) and divergence hitchhiking (the effect of local reduction in gene flow close to selected loci) is a challenging question that in any given case probably depends on all of the factors listed earlier.

Young incipient species are useful for observing the signatures of transient, but potentially important processes, such as selective sweeps. As organisms adapt to novel environments, they may experience strong selection on traits and genes that are evolving neutrally in the ancestral environment, but the reduction in genetic diversity at sites linked to swept loci erodes rapidly given sufficient recombination (Kim and Stephan 2000). Young incipient species are therefore more likely to display signatures of selective sweeps that can point to asymmetry in selection between habitats. Early in speciation, when genomic regions under divergent selection stand out most strongly from background variation, may be an ideal time to identify such regions. It is expected that early in speciation with gene flow, divergence will be limited to a few regions, corresponding to loci under selection strong enough to overcome gene flow, whereas older species pairs will have accumulated more genetic differences through both stochastic and deterministic processes (Feder et al. 2012).

Divergence between Helianthus annuus and Helianthus petiolaris, ecologically differentiated sunflower species connected by gene flow, has been shown to involve many different parts of the genome (Yatabe et al. 2007; Strasburg et al. 2009). Heterogeneous genomic divergence or evidence for selection has been documented in several other wild sunflower comparisons as well, most comparing established species (Edelist et al. 2006; Gross et al. 2007; Kane and Rieseberg 2007; Sapir et al. 2007; Scascitelli et al. 2010). In contrast H. petiolaris ecotypes occurring at Great Sand Dunes National Park and Preserve have a recent origin, most likely within the last 10,000 years (Andrew et al. 2013).

Previous studies of this system have documented strong signals of isolation by distance (IBD) and isolation by adaptation (IBA; Nosil et al. 2009a), the association of neutral genetic variation with habitats as a result of selection against immigrants (Andrew et al. 2012). In addition, analyses of restriction-associated DNA (RAD) tags (Baird et al. 2008) indicate that genomic divergence is highly heterogeneous, as predicted by divergence with gene flow models (Andrew et al. 2013); coalescent simulations imply rates of gene migration that are too low to inhibit responses to strong divergent selection, but sufficiently high to impede divergence due to drift (Andrew et al. 2012, 2013). Asymmetry between the ecotypes has been reported in these earlier studies, with smaller effective population sizes and numbers of immigrants in the dune ecotype. However, the distribution of loci under selection across the genome has not been analyzed in this system. Nor have there been attempts to associate outlier loci with environmental variables that may impose divergent selection. Furthermore, comparison of the regions of divergence between this pair of ecotypes with those of the much older species pair, H. annuus and H. petiolaris, offers the opportunity to explore how the timing of divergence may affect the number and size of islands of differentiation. Such a comparison is also of interest because the dune ecotype has evolved certain traits, such as large seeds and flower heads, that are similar to H. annuus (R. Andrew, unpubl. ms.).

In this study, we explore the distribution of divergence across the genome between the dune and nondune ecotypes of H. petiolaris growing at Great Sand Dunes National Park and Preserve. RAD sequencing has been used previously to assess gene flow between ecotype of the Great Sand Dunes sunflowers and identify their evolutionary relationships. Here, with the availability of a draft genome assembly and high-density genetic map, we are able to use a larger proportion of the same data to investigate the following questions:

• How are the highly divergent loci between the dune and nondune ecotypes of H. petiolaris distributed across the genome? How do the patterns of divergence between ecotypes differ from interspecific divergence between H. annuus and H. petiolaris? We address these questions using a Gaussian smoothing analysis and resampling test to identify genomic regions with significantly higher than average _F_ST.
• Can outlier SNPs be associated with the environmental factors that may be responsible for directional selection? Addressing this question is challenging because both IBD and IBA can lead to correlations between neutral variation and habitat differences. Therefore, we search for associations that persist after controlling for these background correlations.
• Is geography important? Are the outlier regions the same when only internal subpopulations or those near the edge are compared? In populations undergoing parapatric divergence in response to habitat differences, the adaptation of populations close to the boundary can be constrained by gene flow of maladaptive alleles, whereas IBD protects populations further from the edge. This geographic heterogeneity may be observed in adaptive traits or genes, with the prediction that outlier loci should be more easily detected in tests involving core populations, compared with edge populations.

Methods

RAD SEQUENCE GENERATION, ALIGNMENT, AND SCORING

Great Sand Dunes National Park and Preserve comprises a central dune system, surrounded by a vegetated sand sheet and, beyond that, the Sangre de Cristo Mountains to the north and east or sabka (salt flats) to the west. Intermediate habitats are found in partially stabilized dunes at the margins of the dunefield. This study included five unrelated individuals from each of 20 subpopulations, including dune, nondune, and intermediate habitats (Fig. S1; Andrew et al. 2013). H. annuus and H. petiolaris populations from outside the park were also included, with one individual from each of 10 populations per species. H. petiolaris subsp. canescens appears genetically quite distinct (Andrew et al. 2013), so analyses were performed both including and excluding the three samples of that subspecies, but results did not differ qualitatively. Results obtained omitting subsp. canescens are presented here. RAD sequencing (Baird et al. 2008) for this project was carried out by Floragenex (Portland, OR) using the restriction enzyme _Pst_I and has been described in detail previously (Andrew et al. 2013). All samples were barcoded and sequenced with at least 60 bp reads, with a subset sequenced with 80 bp reads, yielding 55 bp and 75 bp sequences respectively, of which the first 5 bp covered the restriction site.

RAD sequences were aligned to a draft genome reference (April 2011 build) for Helianthus annuus consisting of 1,603,248 contigs. Alignments were performed with Bowtie 0.12.7 (Langmead et al. 2009) using the default settings (i.e., reporting the best alignment for each read) with Illumina quality scores. All sequences were trimmed to 55 bp to avoid biases in the alignment due to sequences of different lengths. SNP scoring employed a custom script that implemented the likelihood method of Hohenlohe et al. (2010), as described previously (Andrew et al. 2013). SNPs scored in fewer than 85% of the individuals and those varying only due to singletons were excluded from the data set.

Samples from the dune and nondune sampling locations at Great Sand Dunes (i.e., excluding the intermediate habitats) were employed for the pairwise analyses described later. In addition, the dune subpopulations were divided into two groups based on their proximity to the edge of the dunefield. The five subpopulations less than 1 km from the margin were classified as edge subpopulations, whereas the remaining five core subpopulations were classified as the core subpopulations. If gene flow from the nondune habitat constrains adaptation, edge populations should exhibit reduced divergence from the nondune ecotype and fewer outlier loci.

Summary statistics were used to quantify genetic divergence and diversity between pairs of ecotypes (dune and nondune H. petiolaris) or species (H. petiolaris and H. annuus). Weir and Cockerham's _F_ST (i.e., θ; Weir 1996), and Nei's allele-based gene diversity were estimated using the “hierfstat” package (version 0.04-6; Goudet 2005) in the R statistical software (R Core Development Team 2011). Reduced diversity within one population relative to the other was detected using the lnRH statistic (Schlötterer and Dieringer 2005). This statistic measures the ratio of genetic diversity (_H_T) in the two populations while accounting for sample size differences and can be mean-variance standardized to control for population size and history. Standardized values less than −2 are evidence for selective sweeps in the focal population (dune or H. annuus), whereas values greater than 2 suggest selective sweeps in the other population (nondune or H. petiolaris).

To avoid bias due to less informative SNPs (Roesti et al. 2012) and those arising from paralogy or sequence error, we filtered out those with unbiased expected heterozygosity (_H_T) less than 0.05 or observed heterozygosity greater than 0.8. Loci scored for less than 60% of individuals in each population or species in the pair were also excluded from further analysis. Of the remaining 19,539 SNPs for the ecotypic comparison and 27,994 SNPs for the interspecific comparison, 13,117 and 16,717, respectively, were placed on the map. The mapped sets for the two comparisons shared 8237 SNPs.

SPATIAL ANALYSIS OF SUMMARY STATISTICS

Major goals of this study were to test whether highly divergent loci are clustered within the genome and to characterize those clusters, which are expected to differ between incipient and established species. For this we used tools from spatial ecology, with the genetic distance along each linkage group treated as a spatial variable in a one-dimensional landscape. Although these methods do not identify specific factors causing correlations among nearby loci, such as physical linkage, they enable us to test for the presence of greater clustering than expected under no spatial structure.

SNPs were placed on a high-density genetic map generated for H. annuus based on 96 eighth generation recombinant inbred lines derived from a cross between two highly homozygous sunflower cultivars: RHA280, a confectionary line, and RHA801, an oil seed line. The map positioned 261,999 contigs from the same draft reference sequence to which the RAD SNPs were aligned. Only the SNPs on mapped contigs were used for spatial analysis of summary statistics. Although the H. annuus chromosomal structure represented by this map differs from that of H. petiolaris, this is unlikely to produce artifactual clusters of highly divergent markers. However, the locations of such clusters in the Great Sand Dunes comparison may be affected by the use of a heterospecific map.

Global and local tests for the clustering of highly divergent regions on the genetic map were performed at multiple scales. First, spatial autocorrelation analysis (Sokal and Oden 1978) of _F_ST via the “correlog” function in the “ncf” R package (R Core Development Team 2011) was conducted over all linkage groups. Because the Moran's I statistic generated by this method summarizes both high and low values of _F_ST, a complementary approach was used to focus on regions of high divergence. Join-count analysis (Fortin et al. 2005) considers whether objects sharing attributes (e.g., at the same level of a factor) are clustered, relative to random expectation. We used this approach to ask whether SNPs in the top 5% of _F_ST values were globally clustered, using the “spdep” R package. To perform both analyses, a composite spatial variable was created such that relative positions in centimorgans were preserved for SNPs on the same linkage group, but those on different linkage groups were outside the maximum lag (distance class), which was 100 cM. Nonoverlapping distance classes of width 5 cM were tested with 1000 permutations of SNPs over positions to evaluate differences from random expectation under no spatial structure. Although these methods are useful for identifying relevant scales and testing overall clustering, they assume that the underlying processes are homogeneous. Because we do not expect the latter to be the case, identifying local regions of high divergence allows their characteristics to be examined more directly.

We incorporated local tests of elevated divergence into a windowed smoothing algorithm. Gaussian smoothing was applied to the SNP data and the resampling test described by Hohenlohe et al. (2010) was used to identify regions of the genome with significantly higher than average _F_ST. The smoothing function was evaluated at 0.25 cM intervals and consisted of a weighted average of the SNPs with Gaussian weights calculated based on the proximity of the SNP to the center of the window and the width parameter (σ, the standard deviation of the normal distribution). The function was evaluated where there was a minimum of three variable SNPs within a distance of σ from the center. We varied this parameter from 0.5 to 4, but present only the results of analysis with sigma set to 1 cM as varying the scale of the analysis had little qualitative effect on the results. For each window, _F_ST values were randomly resampled from the entire dataset and assigned to the actual weights of SNPs observed within the window. When 1000 initial iterations suggested a value close to significance (i.e., if P < 0.1), additional iterations were performed up to 1,000,000 iterations per location and false discovery rate corrections were carried out using the “_q_-value” package of the R statistical software.

When runs of consecutive significant windows were observed, the approximate boundaries of the region of elevated divergence was considered to be σ before and after the centers of the first and last windows, respectively. The width of the region and the number of outlier and nonoutlier SNPs was calculated based on these boundaries.

OUTLIER ANALYSIS

Outlier analysis was used to identify SNPs that were more divergent than would be expected under a purely neutral model, based on the overall background divergence between the dune and nondune ecotypes. The software BayeScan version 2.5 (Foll and Gaggiotti 2008) is a Bayesian method that does not assume that populations are identical in size, which may be important in the Great Sand Dunes study system. It is based on a model in which a common migrant pool exchanges genes with a number of subpopulations, each of which differs from the common gene pool according to a population-specific _F_ST. SNPs on unmapped reference contigs were included in this study in addition to the mapped ones. Loci were considered outliers with _q_-values (FDR-corrected _P_-values) less than 0.05. If selection is driving large regions of elevated divergence between ecotypes, we would predict that the majority of outliers should fall inside these regions.

SNP ASSOCIATIONS WITH ENVIRONMENTAL VARIABLES

Polymorphism at loci subject to selection that varies with habitats is likely to be associated with the environmental variables imposing selection. Because geographic structure and population history can also produce associations between SNPs and environmental variables, the association of the genomic background must be used to control for nonadaptive causes. Previous work identified a set of five environmental variables associated with genetic variation at putatively neutral microsatellite loci from 25 simple and composite measures (Andrew et al. 2012). Here we asked whether specific SNPs were associated with environments in excess of the genomic background in the Great Sand Dunes study system. This analysis was conducted at the subpopulation level to avoid pseudoreplication, as the environmental variables were measured once at each sampling location. Reynolds’ genetic distance, estimating coancestry between pairs of subpopulations, was estimated overall and for single loci. Each biallelic SNP satisfying the conditions also applied to the BayeScan analysis was tested for association with each environmental variable using the multiple regression of matrices procedure in the R package “ecodist.” In each case, the locus-specific distance was the response variable and the Euclidean distance along the environmental axis was the predictor, with overall genetic distance as a covariate. Undefined values where both subpopulations were fixed for the same allele were replaced by zero to avoid missing data. Because relationships between SNPs and habitat variables may be nonlinear, rank correlations were assessed. Tests were based on 1 million permutations and corrected for false discovery rate as described earlier.

Results

Using alignments to a draft reference genome, 675,737 nucleotide sites were scored for at least 85% of Great Sand Dunes individuals (representing approximately 12,286 restriction sites). Of these sites, 45,056 were variable (excluding those only variable due to singletons). Further filtering produced 19,539 high-quality biallelic SNPs on 7067 reference contigs, of which 11,727 were aligned to 4207 contigs that had been placed on the genetic map (in 1468 unique locations).

Bayesian outlier analysis identified 335 loci putatively under divergent selection (q < 0.05), of which 205 were located on mapped contigs. Results of repeated analyses were nearly identical: as a single additional outlier was found in one of the three replicate runs. With a more stringent q cutoff of 0.01, 253 outliers were identified and 173 were also significant at the 0.001 level. At least one outlier was found on 213 of the 7287 mapped contigs and these were located on 14 of the 17 linkage groups; however, most were highly clustered (Fig. 1). The greatest numbers of outliers were found in relatively tight clusters on LG5 and LG9 or in a broader cluster on LG11.

Figure 1.

Weir and Cockerham's _F_ST (θ) between dune and nondune ecotypes, with outliers highlighted. Outliers were identified using BayeScan with a false discovery rate-corrected threshold of 0.05.

The same regions were identified as having significantly higher average _F_ST than the global average (q < 0.05; Fig. 2). Indeed, 82% (168) of mapped outlier SNPs were contained within significant windows and 11.7% of SNPs in significant windows were identified as outliers, compared with 0.3% of those outside significant windows. Out of 5497 overlapping windows (with σ = 1 cM and spaced at 0.25 cM intervals), 196 displayed elevated _F_ST, of which 166 (85%) were on the linkage groups 5, 9, and 11. Most significant windows were found in runs of more than four windows (Table S1). The greatest number of consecutive significant windows (ignoring those with insufficient marker density) was 69 on LG11, representing approximately 22.75 cM. Highly divergent regions on LG5 and LG9 were considerably smaller, ranging from 1 to 33 significant windows. This approach is expected to identify regions of high divergence in the presence of global autocorrelation, for instance due to linkage, because local variance is less than the overall variance. However, large runs of significant windows would not be expected as a result because linkage disequilibrium decays rapidly in large outcrossing populations. Linkage disequilibrium (_r_2) of RAD-derived SNPs was low both within (mean = 0.011, SD = 0.036) and between (mean = 0.011, SD = 0.026) linkage groups. Nevertheless, a few perfectly correlated SNPs (_r_2 = 1.000) could be found on the same linkage group (typically in the same RAD sequence or on the same contig) and moderate LD (_r_2 up to 0.627) occurred in some cases between SNPs on different linkage groups.

Figure 2.

_F_ST between dune (D) and nondune (N) ecotypes of H. petiolaris and comparison of gene diversity between ecotypes. Gaussian smoothing (with σ = 1 cM and windows 0.25 cM apart) was applied to each statistic and resampling was used to test whether _F_ST was greater than the genome-wide average each window. Significant windows are indicated by bars above each plot.

In contrast to the ecotypic comparison, the interspecific comparison between H. annuus and H. petiolaris (AP) identified more “islands of divergence,” but they were more dispersed throughout the genome (Fig. 3). Sixteen such islands occurred in the interspecific comparison on 12 linkage groups, whereas the ecotypic comparison yielded only 10 islands on 7 linkage groups. Although the mean width did not differ significantly between comparisons (Welch's two-sample _t_-test: t = −1.363, df = 9.98, _P_-value = 0.20), the maximum width for AP (7.5 cM on LG13) was considerably smaller than that of the ecotypic comparison (22.75 cM on LG11). Although small islands may be missed due to the marker density and scale of this study, it is clear that large-scale clustering is more prevalent in the dune–nondune comparison than in the interspecific one. Spatial autocorrelation analysis of _F_ST indicated little difference in the scale of positive autocorrelation between comparisons, but the strength was greater in the dune comparison (r = 0.35 within 5 cM) than in the interspecific comparison (r = 0.11 within 5 cM; Fig. 4A). In contrast, the join-count analysis displayed approximately double the extent of clustering of SNPs in the top 5% of _F_ST values in the ecotypic pair, compared with the interspecific one (about 20 cM vs. about 10 cM; Fig. 4B). The lack of repeatability between the ecotypic and interspecific comparisons was striking: even when on the same linkage groups, the divergent regions overlapped little (Fig. S2). In one exception, on LG2, the regions of high divergence largely coincided, but the divergent region between H. annuus and H. petiolaris on LG9 covered less than half of the large peak of divergence found in the ecotypic comparison. In both comparisons, SNP density was higher within islands than the global mean of 8.62 SNP/cM (Table 1; Fig. S3).

Table 1.

Summary of genomic islands or regions of elevated divergence between ecotypes and between H. petiolaris and H. annuus. Islands of high divergence were identified from local mean _F_ST (Gaussian smoothed in 1 cM windows) as contiguous runs of significant windows, based on random resampling. Mean and standard deviation are shown

Table 1.

Summary of genomic islands or regions of elevated divergence between ecotypes and between H. petiolaris and H. annuus. Islands of high divergence were identified from local mean _F_ST (Gaussian smoothed in 1 cM windows) as contiguous runs of significant windows, based on random resampling. Mean and standard deviation are shown

Figure 3.

_F_ST between H. annuus (A) and H. petiolaris (P) and comparison of gene diversity between ecotypes. Gaussian smoothing (with σ = 1 cM and windows 0.25 cM apart) was applied to each statistic and resampling was used to test whether _F_ST was greater than the genome-wide average each window. Significant windows are indicated by bars above each plot.

Figure 4.

Spatial autocorrelation analysis of _F_ST (A) and join-count analysis of SNPs in top 5% of _F_ST distribution (B), with 5 cM distance classes. Dependent axes show Moran's I and a _z_-score, calculated as (observed counts – expected counts)/standard deviation. Ecotypic (DN) and interspecific (AP) comparisons are shown and filled symbols indicate autocorrelation coefficients significantly different from zero (Two-sided P < 0.05, estimated from 1000 permutations).

Evidence for selective sweeps was observed in the relative genetic diversity of _F_ST outliers, which was skewed toward the nondune ecotype (Fig. 5). In regions of the genome that were highly diverged between ecotypes, gene diversity was consistently lower in the dune population than in the nondune population (Fig. 2). By comparison, the genetic diversity in islands of divergence was only slightly skewed toward H. annuus (Fig. 3; Table 1). The standardized logarithm of the gene diversity ratio indicated, on average, lower diversity in dune sunflowers (one-sample _t_-test: t = −2.83, df = 9, _P_-value = 0.020) and marginally lower diversity in H. petiolaris (one-sample _t_-test: t = 1.82, df = 15, _P_-value = 0.088) in the ecotypic and interspecific analyses, respectively. However, the trend was reversed when H. petiolaris subsp. canescens was included (mean = −0.178; one-sample _t_-test: t = −5.22, df = 16, _P_-value = 8.3 × 10−05).

Figure 5.

Ratio of gene diversity between dune and nondune ecotypes of H. petiolaris, with FST outliers highlighted. ln RH has been mean–variance standardized.

Significant associations between SNPs and environmental variables were mainly limited to cover variables, especially total vegetation cover (Table 2). The SNPs that were associated at the q < 0.05 level (256 or 1.3% of the total number assessed) displayed a similar distribution to that of the _F_ST outliers, albeit more diffuse (Fig. 6). Linkage groups 5, 9, and 11 were strongly represented, but some strong associations also occurred on linkage groups that appeared less important in the outlier analysis, such as LG8 (Fig. 1). In addition, several SNPs on linkage group 10 were strongly associated with grass cover and significant at the q < 0.1 level (Fig. S4).

Table 2.

SNP associations with environmental variables, controlling for background coancestry, and conducted at the subpopulation level. Standardized partial regression slopes were estimated from multiple regression of distance matrix models and significant associations were based on FDR-corrected _P_-values, q < 0.05. Variables identified by Andrew et al. (2012) as associated with microsatellite genetic distance while controlling for landscape resistance are shown in bold

Table 2.

SNP associations with environmental variables, controlling for background coancestry, and conducted at the subpopulation level. Standardized partial regression slopes were estimated from multiple regression of distance matrix models and significant associations were based on FDR-corrected _P_-values, q < 0.05. Variables identified by Andrew et al. (2012) as associated with microsatellite genetic distance while controlling for landscape resistance are shown in bold

Figure 6.

Associations of SNPs with total vegetation cover, controlling for the association of the genomic background. Highlighted SNPs were significant with a false discovery rate-corrected threshold of 0.05.

The subpopulations at the edge of the dunefield displayed different patterns of divergence from the nondune ecotype compared to those in the core (Fig. 7). The same regions of elevated divergence appeared in both comparisons, but they differed in strength. The core dune subpopulations were slightly less divergent from the nondune ecotype than were the edge subpopulations, with 145 and 192 BayeScan outliers detected at the q < 0.01 level, respectively. Although linkage groups 5 and 7 showed the expected pattern of greater divergence in the core, substantially more outliers were detected on linkage groups 9 and 11 in the edge comparison. The difference in the proportion of outliers was statistically significant using Fisher's exact test with 10,000 permutations (P = 0.0048). Jackknifing subpopulations (i.e., repeating the analysis while omitting one subpopulation at a time) yielded similar results (not shown), suggesting that stochastic sampling was not the cause of this significant difference.

Figure 7.

Comparison of divergence between core dune and nondune (CN) subpopulations with that between edge and nondune (EN) subpopulations. The number of outliers on each linkage group is shown compared with those for the overall dune–nondune analysis (DN) in panel (A) and the smoothed local _F_ST estimates for the core (gold) and edge (blue) are shown in panel (B).

Discussion

The results presented here add to the growing body of data on the genomic structure of divergence among species. High-throughput sequencing is enabling genome-wide scans of divergence, and via reduced representation methods such as RAD sequencing this approach now extends to nonmodel organisms (Hohenlohe et al. 2010; Nadeau et al. 2013; Stölting et al. 2013). Contrasting patterns of genomic divergence have characterized results in different study systems. For example, Nadeau et al. (2013) and Renaut et al. (2012) found that the scale of highly divergent regions of the genome increased with increasing divergence between pairs of taxa. However, there is evidence that large regions of divergence may be prevalent in incipient species (Turner et al. 2005; Via and West 2008; Michel et al. 2010).

Our study asked how divergent loci are distributed across the genome in the earliest stage of ecological speciation in a dune-adapted sunflower. Much of the divergence was restricted to three linkage groups, suggesting that the whole genome is not yet involved in differentiating these ecotypes. However, we cannot conclude that few genes are involved: with divergent regions up to 22.75 cM in width, it is likely that multiple selected loci are responsible for each “island of divergence.” Three linkage groups are strongly represented in outlier scans and regions of high mean _F_ST. A narrow peak containing many outliers occurred on linkage group 5 and broader regions of high divergence occurred on LG9 and LG11. That these regions, especially on LG5, occur where marker density is high on the H. annuus genetic map suggests that genomic architecture may play a role in shaping these patterns, via regions of either low recombination (Noor and Bennett 2009) or high gene density. Low recombination within the Eda region of the stickleback (Gasterosteus aculeatus) genome has made it difficult to resolve gene–trait associations in that species and may have contributed to the repeated evolution of similar morphological forms in many geographic locations (Hohenlohe et al. 2012; Jones et al. 2012). Chromosomal inversions, which reduce local rates of recombination, have been shown to be important in local adaptation and speciation (Noor et al. 2001; Lowry and Willis 2010; Neafsey et al. 2010). Whether variation in marker density in this data set is a feature of the genetic map stemming from low recombination between H. annuus genotypes in certain regions, which may or may not correspond to similar features in H. petiolaris, or the targeting of gene-rich regions by the _Pst_I restriction enzyme is not clear; a full reference genome assembly is required to resolve this issue.

The large regions of high divergence between H. petiolaris ecotypes may result from a reduction in effective gene flow via mechanisms that impede recombination around strongly selected genes, such as chromosomal inversions or divergence hitchhiking (Via and West 2008; Via 2012). With a full reference genome, it will be possible to assess the patterns of linkage disequilibrium within and between divergent regions and ask whether the influence of selection on linked sites extends beyond what is expected due to adaptive hitchhiking alone. A high-density genetic map specific to this study system will also be necessary to investigate the potential role of chromosomal inversions in facilitating islands of divergence (Rieseberg 2001; Feder and Nosil 2009).

The background divergence was extremely low on all linkage groups. Although this may reflect high amounts of migration, which appear sufficient to hinder divergence by genetic drift, it is also possible that the ecotypes have not existed for long enough for drift to drive much divergence (Andrew et al. 2012, 2013). In contrast to the Great Sand Dunes ecotypes, H. annuus and H. petiolaris are estimated to have been separate for 1.8 ± 0.3 million years but exchange genes at high rates in regions of sympatry (Sambatti et al. 2012). These species, which were also more strongly diverged overall, were highly divergent in a larger number of relatively small regions, although we should remember that small regions of high divergence (< 1 cM) are likely to have been missed due to map and marker density in both pairs. The greater number and dispersion of divergent regions between species is consistent with expectations that adaptive ecological and genetic differences accumulate over time and that ongoing gene flow may erode differentiation between selected genes. The presence of substantial population structure within both species could also obscure the influence of selection on linked sites when sampled at this scale (Ralph and Coop 2010). Because this study only considers two points in the speciation process, further replication with a larger number of comparisons is necessary to identify general patterns of divergence in sunflower speciation and to understand the causes of variation in those patterns.

From a different perspective, how far taxa will progress along the speciation continuum may be determined by the dimensionality of selection (the alternative being the strength of selection at few loci), which also predicts a greater number of divergent regions in more reproductively isolated taxa (Nosil et al. 2009b). Although the divergence of the Great Sand Dunes sunflowers appears to be the result of extremely strong edaphic selection, the dune habitat could challenge sunflowers in numerous ways. Several traits have diverged, including seed size, flowering time, and branching, although a common genetic basis may affect such disparate traits (Melzer et al. 2008; He et al. 2010). Clusters of QTL for disparate traits have been found within H. annuus and in interspecific mapping populations in Helianthus (Rieseberg et al. 2003; Lexer et al. 2005; Tang et al. 2006; Wills and Burke 2007). A broader sample of dune-adapted sunflowers would help to assess the impact of age on the number of divergent regions, as it is likely that the environmental factors acting on the H. petiolaris ecotypes are different from those driving divergence between H. annuus and H. petiolaris divergence, which probably influences the number of loci under divergent selection.

Strong evidence for the asymmetry of divergence has accumulated in the Great Sand Dunes sunflowers. Earlier work has found asymmetry in the effective population sizes as well as in the numbers of migrants per generation (Andrew et al. 2012, 2013) and here we have shown that in strongly diverged regions of the genome, gene diversity is substantially reduced in the dunes with respect to the nondune ecotype. These asymmetries are entirely consistent in direction and those specific to the genomic regions of high divergence support selective sweeps as a cause, whereas the more general patterns observed in putatively neutral regions reflect demography more strongly, The persistence of this signature could reflect the recent divergence of the ecotypes and this pattern may be general to the early stage of ecological speciation (Hohenlohe et al. 2010; Neafsey et al. 2010). The effect of selective sweeps on genetic diversity at neutral sites that are linked to selected loci are thought to decay rapidly (Kim and Stephan 2000). The Great Sand Dunes sunflower may be particularly representative of narrow endemics with widespread relatives, because adaptation to extreme environments is likely to involve stronger selection pressures that those acting in more mesic populations and because species with large and environmentally variable habitats are likely to harbor high levels of genetic diversity.

Instead of identifying which outlier SNPs (or clusters of SNPs) permitted adaptation to particular environmental variables (Eckert et al. 2010; Hancock et al. 2011), the association of SNPs with environmental factors generally reiterated the distribution of _F_ST outliers across the genome (Fig. 6). The highest number of significant associations occurred with total vegetation cover, which was one of the key indicators of IBA with microsatellites (Andrew et al. 2012), but very few occurred with soil variables (Table 2). Vegetation cover may control the erosion and deposition of sand, and hence the burial of plants or the exposure of their roots. However, the association could result from the amount of organic matter in the soil, or even the amount of competition with other plants for water. The similarity of cover associations to the outlier scans could result from the coupling of endogenous reproductive isolating mechanisms to exogenous ones, which may be the cause of many associations between genes and environmental variables (Bierne et al. 2011). However, the fact that microsatellite-based genetic distance was strongly correlated with environmental dissimilarity even within ecotypes suggested otherwise (Andrew et al. 2012). It is also possible that statistical issues contribute: using a coancestry matrix may not adequately capture the background divergence, leading to the signal of IBA biasing locus-specific effects. Under both explanations, we would expect similar numbers of associations with soil variables as with vegetation cover, which was not the case. Nevertheless, grass cover displayed a greater number of strong associations with SNPs on LG10 and LG12 than would be expected based on the proportion of _F_ST outliers located on those linkage groups. Those on LG10 are close to the B locus, a region containing QTL for numerous traits, including branching and seed traits in H. annuus (Tang et al. 2006; Wills and Burke 2007).

We detected differences between edge and core subpopulations of the dune ecotype in the genomic structure of their divergence from the nondune form (Fig. 7). The overall difference was not as striking as the decrease of dune ancestry in the nondune subpopulations with increasing distance from the dunes in putatively neutral microsatellites and RAD SNPs (Andrew et al. 2012, 2013). This further example of the asymmetry of gene flow and selection in the Great Sand Dunes system may be due to the fact that strong selection against migrants in the dunes squeezes the effective genetic exchange to a narrower zone on the dune side of the boundary between the ecotypes (Barton and Bengtsson 1986). If that is the case, sampling at a finer scale would be necessary to investigate the penetration of maladaptive alleles into the dune ecotype.

The significant effect of including only the core or the edge of the dune population on the distribution of _F_ST outliers among linkage groups implies that regions of the genome differ in the extent to which IBD protects the core of the ecotype from gene flow more than the edge. These results suggest that gene flow variously constrains divergence and promotes it. Gene flow is expected to constrain intraspecific diversification due to local adaptation (reviewed by Lenormand 2002; Räsänen and Hendry 2008), but the promotion of divergence by the proximity of nonadapted populations is a signature of ecological character displacement (Schluter 2000), or the evolution of reproductive isolating mechanisms (Noor 1999) in response to a cost of immigration. In a compelling example, the reinforcement of a premating reproductive barrier, flower color, has recently been dissected to the level of the gene and mutation (Hopkins and Rausher 2011). Several studies of such phenomena focus on similar geographic scales to that of our study (Silvertown et al. 2005; Schwartz et al. 2010; Nosil and Hohenlohe 2012), and there is little genetic differentiation between the core and edge of the Great Sand Dunes sunflowers. However, these ecotypes diverged relatively recently (Andrew et al. 2013) and, although we may not expect to see the signature of gene flow promoting divergence in such a restricted study system at equilibrium, it may occur transiently during divergence. In the sunflowers of Great Sand Dunes, it appears that both local adaptation to an extreme habitat and the evolution of reproductive barriers contribute to the continued persistence of large phenotypic differences despite the potential for ample pollen migration between ecotypes (Andrew et al. 2012). The divergent regions displaying higher divergence at the edge of the dunefield are too large to offer strong candidate genes for reproductive isolating mechanisms, but further examination of these regions in mapping studies may yield novel results. Linkage group 9 has not been implicated in the reproductive barriers previously, but a QTL for pollen viability in crosses between H. annuus and H. petiolaris is located in the middle of linkage group 11 (Lai et al. 2005).

Because of the apparent importance of certain highly divergent regions on linkage groups 5, 9, and 11, it is worth revisiting the QTL and other features that have been detected in these locations in sunflowers. LG5 is fused with part of LG14 of H. annuus in H. petiolaris (Lai et al. 2005), so considering them separately in the Great Sand Dunes system may lead to some inaccuracy. Seed weight and floral morphology QTL have been detected on LG5 and LG9 in both inter- and intraspecific studies (Kim and Rieseberg 1999; Rieseberg et al. 2003; Tang et al. 2006; Wills and Burke 2007). In contrast, few or no QTL have been detected on LG11, with the exception of the large interspecific study of Rieseberg et al. (2003). The lack of overlap between peaks of divergence between Great Sand Dunes ecotypes with those of H. annuus is striking. Although significantly elevated _F_ST occurs on LG5 in both comparisons, the islands of divergence do not overlap.

Groups of taxa differ dramatically in the amount of phenotypic, genetic, and ecological variation among species. The progression of genetic divergence throughout the speciation process may be a driver of this diversity and analysis of the structure of genetic differentiation within the genomes of diverging species is one way we have gained insight into this process. In this study we observed greater numbers of divergent regions in an interspecific comparison than in one between ecotypes, as predicted by models of ecological speciation. We also found highly asymmetrical evidence of selective sweeps in the dune habitat, rather than the nondune one and a marked difference between the core and edge dune samples in the genomic structure of divergence. Our results are consistent with these ecotypes undergoing a shift in the migration-selection balance that marks the transition from locally adapted morphs to incipient species.

ACKNOWLEDGMENTS

The authors thank N. Kane, S. Renaut, S. Yeaman, C. Fitzpatrick, M. King, K. Ostevik, and D. Ebert for assistance with the project and discussions about analyses. The authors also thank Great Sand Dunes National Park and Preserve and The Nature Conservancy for access and support for our work. The authors also thank the editors and anonymous reviewers for their comments and suggestions. This work was supported by a Killam Postdoctoral Fellowship to RLA and Natural Sciences and Engineering Research Council grant (327475) to LHR.

Associate Editor: J. Feder-Guest

LITERATURE CITED

Andrew

,

R. L.

,

N. C.

Kane

,

G. J.

Baute

,

C. J.

Grassa

, and

L. H.

Rieseberg

.

2013

.

Recent nonhybrid origin of sunflower ecotypes in a novel habitat

.

Mol. Ecol.

22

:

799

–

813

.

Andrew

,

R. L.

,

K. L.

Ostevik

,

D. P.

Ebert

, and

L. H.

Rieseberg

.

2012

.

Adaptation with gene flow across the landscape in a dune sunflower

.

Mol. Ecol.

21

:

2078

–

2091

.

Baird

,

N. A.

,

P. D.

Etter

,

T. S.

Atwood

,

M. C.

Currey

,

A. L.

Shiver

,

Z. A.

Lewis

,

E. U.

Selker

,

W. A.

Cresko

, and

E. A.

Johnson

.

2008

.

Rapid SNP discovery and genetic mapping using sequenced RAD markers

.

PLoS ONE

3

:

e3376

.

Barton

,

N.

, and

B. O.

Bengtsson

.

1986

.

The barrier to genetic exchange between hybridising populations

.

Heredity

57

:

357

–

376

.

Bierne

,

N.

,

J.

Welch

,

E.

Loire

,

F.

Bonhomme

, and

P.

David

.

2011

.

The coupling hypothesis: why genome scans may fail to map local adaptation genes

.

Mol. Ecol.

20

:

2044

–

2072

.

Coyne

,

J. A.

, and

H. A.

Orr

.

2004

.

Speciation

.

Sinauer Associates

,

Sunderland, MA

.

Eckert

,

A. J.

,

J.

van Heerwaarden

,

J. L.

Wegrzyn

,

C. D.

Nelson

,

J.

Ross-Ibarra

,

S. C.

González-Martínez

, and

D. B.

Neale

.

2010

.

Patterns of population structure and environmental associations to aridity across the range of loblolly pine (Pinus taeda L., Pinaceae)

.

Genetics

185

:

969

–

982

.

Edelist

,

C.

,

C.

Lexer

,

C.

Dillmann

,

D.

Sicard

, and

L. H.

Rieseberg

.

2006

.

Microsatellite signature of ecological selection for salt tolerance in a wild sunflower hybrid species, Helianthus paradoxus

.

Mol. Ecol.

15

:

4623

–

4634

.

Feder

,

J. L.

, and

P.

Nosil

.

2009

.

Chromosomal inversions and species differences: when are genes affecting adaptive divergence and reproductive isolation expected to reside within inversions?

Evolution

63

:

3061

–

3075

.

Feder

,

J. L.

, and

P.

Nosil

.

2010

.

The efficacy of divergence hitchhiking in generating genomic islands during ecological speciation

.

Evolution

64

:

1729

–

1747

.

Feder

,

J. L.

,

R.

Gejji

,

S.

Yeaman

, and

P.

Nosil

.

2012

.

Establishment of new mutations under divergence and genome hitchhiking

.

Philos. Trans. R. Soc. B-Biol. Sci.

367

:

461

–

474

.

Foll

,

M.

, and

O.

Gaggiotti

.

2008

.

A genome-scan method to identify selected loci appropriate for both dominant and codominant markers: a Bayesian perspective

.

Genetics

180

:

977

–

993

.

Fortin

,

M.-J.

,

M.

Randall

, and

T.

Dale

.

2005

.

Spatial analysis: a guide for ecologists

.

Cambridge Univ. Press

,

Cambridge

, U.K.

Goudet

,

J.

2005

.

HIERFSTAT, a package for R to compute and test hierarchical F-statistics

.

Mol. Ecol. Notes

5

:

184

–

186

.

Gross

,

B. L.

,

K. G.

Turner

, and

L. H.

Rieseberg

.

2007

.

Selective sweeps in the homoploid hybrid species Helianthus deserticola: evolution in concert across populations and across origins

.

Mol. Ecol.

16

:

5246

–

5258

.

Hahn

,

M. W.

,

B. J.

White

,

C. D.

Muir

, and

N. J.

Besansky

.

2012

.

No evidence for biased co-transmission of speciation islands in Anopheles gambiae

.

Philos. Trans. R. Soc. B-Biol. Sci.

367

:

374

–

384

.

Hancock

,

A. M.

,

B.

Brachi

,

N.

Faure

,

M. W.

Horton

,

L. B.

Jarymowycz

,

F. G.

Sperone

,

C.

Toomajian

,

F.

Roux

, and

J.

Bergelson

.

2011

.

Adaptation to climate across the Arabidopsis thaliana genome

.

Science

334

:

83

–

86

.

He

,

C.

,

Y.

Tian

,

R.

Saedler

,

N.

Efremova

,

S.

Riss

,

M.

Khan

,

A.

Yephremov

, and

H.

Saedler

.

2010

.

The MADS-domain protein MPF1 of Physalis floridana controls plant architecture, seed development and flowering time

.

Planta

231

:

767

–

777

.

Hohenlohe

,

P. A.

,

S.

Bassham

,

P. D.

Etter

,

N.

Stiffler

,

E. A.

Johnson

, and

W. A.

Cresko

.

2010

.

Population genomics of parallel adaptation in threespine stickleback using sequenced RAD tags

.

PLoS Genet.

6

:

e1000862

.

Hohenlohe

,

P. A.

,

S.

Bassham

,

M.

Currey

, and

W. A.

Cresko

.

2012

.

Extensive linkage disequilibrium and parallel adaptive divergence across threespine stickleback genomes

.

Philos. Trans. R. Soc. B-Biol. Sci.

367

:

395

–

408

.

Hopkins

,

R.

, and

M. D.

Rausher

.

2011

.

Identification of two genes causing reinforcement in the Texas wildflower Phlox drummondii

.

Nature

469

:

411

–

414

.

Jones

,

F. C.

,

M. G.

Grabherr

,

Y. F.

Chan

,

P.

Russell

,

E.

Mauceli

,

J.

Johnson

,

R.

Swofford

,

M.

Pirun

,

M. C.

Zody

,

S.

White

, et al.

2012

.

The genomic basis of adaptive evolution in threespine sticklebacks

.

Nature

484

:

55

–

61

.

Kane

,

N. C.

, and

L. H.

Rieseberg

.

2007

.

Selective sweeps reveal candidate genes for adaptation to drought and salt tolerance in common sunflower, Helianthus annuus

.

Genetics

175

:

1823

–

1834

.

Kim

,

S. C.

, and

L. H.

Rieseberg

.

1999

.

Genetic architecture of species differences in annual sunflowers: implications for adaptive trait introgression

.

Genetics

153

:

965

–

977

.

Kim

,

Y.

, and

W.

Stephan

.

2000

.

Joint effects of genetic hitchhiking and background selection on neutral variation

.

Genetics

155

:

1415

–

1427

.

Lai

,

Z.

,

T.

Nakazato

,

M.

Salmaso

,

J. M.

Burke

,

S.

Tang

,

S. J.

Knapp

, and

L. H.

Rieseberg

.

2005

.

Extensive chromosomal repatterning and the evolution of sterility barriers in hybrid sunflower species

.

Genetics

171

:

291

–

303

.

Langmead

,

B.

,

C.

Trapnell

,

M.

Pop

, and

S.

Salzberg

.

2009

.

Ultrafast and memory-efficient alignment of short DNA sequences to the human genome

.

Genome Biol.

10

:

R25

. doi:10.1186/gb-2009-10-3-r25.

Lawniczak

,

M. K. N.

,

S. J.

Emrich

,

A. K.

Holloway

,

A. P.

Regier

,

M.

Olson

,

B.

White

,

S.

Redmond

,

L.

Fulton

,

E.

Appelbaum

,

J.

Godfrey

, et al.

2010

.

Widespread divergence between incipient Anopheles gambiae species revealed by whole genome sequences

.

Science

330

:

512

–

514

.

Lenormand

,

T.

2002

.

Gene flow and the limits to natural selection

.

Trends Ecol. Evol.

17

:

183

–

189

.

Lexer

,

C.

,

D. M.

Rosenthal

,

O.

Raymond

,

L. A.

Donovan

, and

L. H.

Rieseberg

.

2005

.

Genetics of species differences in the wild annual sunflowers, Helianthus annuus and H. petiolaris

.

Genetics

169

:

2225

–

2239

.

Lowry

,

D. B.

, and

J. H.

Willis

.

2010

.

A widespread chromosomal Inversion polymorphism contributes to a major life-history transition, local adaptation, and reproductive isolation

.

PLoS Biol

8

:

e1000500

.

Lowry

,

D. B.

,

J. L.

Modliszewski

,

K. M.

Wright

,

C. A.

Wu

, and

J. H.

Willis

.

2008

.

The strength and genetic basis of reproductive isolating barriers in flowering plants

.

Philos. Trans. R. Soc. B-Biol. Sci.

363

:

3009

–

3021

.

Melzer

,

S.

,

F.

Lens

,

J.

Gennen

,

S.

Vanneste

,

A.

Rohde

, and

T.

Beeckman

.

2008

.

Flowering-time genes modulate meristem determinacy and growth form in Arabidopsis thaliana

.

Nat. Genet.

40

:

1489

–

1492

.

Michel

,

A. P.

,

S.

Sim

,

T. H. Q.

Powell

,

M. S.

Taylor

,

P.

Nosil

, and

J. L.

Feder

.

2010

.

Widespread genomic divergence during sympatric speciation

.

Proc. Natl. Acad. Sci

.

107

:

9724

–

9729

.

Nadeau

,

N. J.

,

A.

Whibley

,

R. T.

Jones

,

J. W.

Davey

,

K. K.

Dasmahapatra

,

S. W.

Baxter

,

M. A.

Quail

,

M.

Joron

,

R. H.

ffrench-Constant

,

M. L.

Blaxter

, et al.

2012

.

Genomic islands of divergence in hybridizing Heliconius butterflies identified by large-scale targeted sequencing

.

Philos. Trans. R. Soc. B-Biol. Sci.

367

:

343

–

353

.

Nadeau

,

N. J.

,

S. H.

Martin

,

K. M.

Kozak

,

C.

Salazar

,

K. K.

Dasmahapatra

,

J. W.

Davey

,

S. W.

Baxter

,

M. L.

Blaxter

,

J.

Mallet

, and

C. D.

Jiggins

.

2013

.

Genome-wide patterns of divergence and gene flow across a butterfly radiation

.

Mol. Ecol.

22

:

814

–

826

.

Neafsey

,

D. E.

,

M. K. N.

Lawniczak

,

D. J.

Park

,

S. N.

Redmond

,

M. B.

Coulibaly

,

S. F.

Traoré

,

N.

Sagnon

,

C.

Costantini

,

C.

Johnson

, et al.

2010

.

SNP genotyping defines complex gene-flow boundaries among African malaria vector mosquitoes

.

Science

330

:

514

–

517

.

Noor

,

M. A. F.

1999

.

Reinforcement and other consequences of sympatry

.

Heredity

83

:

503

–

508

.

Noor

,

M. A. F.

, and

S. M.

Bennett

.

2009

.

Islands of speciation or mirages in the desert? Examining the role of restricted recombination in maintaining species

.

Heredity

103

:

439

–

444

.

Noor

,

M. A. F.

,

K. L.

Grams

,

L. A.

Bertucci

, and

J.

Reiland

.

2001

.

Chromosomal inversions and the reproductive isolation of species

.

Proc. Natl. Acad. Sci.

98

:

12084

–

12088

.

Nosil

,

P.

2012

.

Ecological speciation

.

Oxford Univ. Press

,

Oxford, U.K

.

Nosil

,

P.

, and

P. A.

Hohenlohe

.

2012

.

Dimensionality of sexual isolation during reinforcement and ecological speciation in Timema cristinae stick insects

.

Evol. Ecol. Res.

14

:

467

–

485

.

Nosil

,

P.

,

D. J.

Funk

, and

D.

Ortiz-Barrientos

.

2009a

.

Divergent selection and heterogeneous genomic divergence

.

Mol. Ecol.

18

:

375

–

402

.

Nosil

,

P.

,

L. J.

Harmon

, and

O.

Seehausen

.

2009b

.

Ecological explanations for (incomplete) speciation

.

Trends Ecol. Evol.

24

:

145

–

156

.

R Core Development Team

.

2011

.

R: A Language and Environment for Statistical Computing

.

R Foundation for Statistical Computing

,

Vienna, Austria

.

Ralph

,

P.

, and

G.

Coop

.

2010

.

Parallel adaptation: one or many waves of advance of an advantageous allele?

Genetics

186

:

647

–

668

.

Räsänen

,

K.

, and

A. P.

Hendry

.

2008

.

Disentangling interactions between adaptive divergence and gene flow when ecology drives diversification

.

Ecol. Lett.

11

:

624

–

636

.

Renaut

,

S.

,

N.

Maillet

,

E.

Normandeau

,

C.

Sauvage

,

N.

Derome

,

S. M.

Rogers

, and

L.

Bernatchez

.

2012

.

Genome-wide patterns of divergence during speciation: the lake whitefish case study

.

Philos. Trans. R. Soc. B-Biol. Sci.

367

:

354

–

363

.

Rieseberg

,

L. H.

2001

.

Chromosomal rearrangements and speciation

.

Trends Ecol. Evol.

16

:

351

–

358

.

Rieseberg

,

L. H.

,

O.

Raymond

,

D. M.

Rosenthal

,

Z.

Lai

,

K.

Livingstone

,

T.

Nakazato

,

J. L.

Durphy

,

A. E.

Schwarzbach

,

L. A.

Donovan

, and

C.

Lexer

.

2003

.

Major ecological transitions in wild sunflowers facilitated by hybridization

.

Science

301

:

1211

–

1216

.

Roesti

,

M.

,

W.

Salzburger

, and

D.

Berner

.

2012

.

Uninformative polymorphisms bias genome scans for signatures of selection

.

BMC Evol. Biol.

12

:

doi:10.1186/1471-2148-12-94.

Sambatti

,

J. B. M.

,

J. L.

Strasburg

,

D.

Ortiz-Barrientos

,

E. J.

Baack

, and

L. H.

Rieseberg

.

2012

.

Reconciling extremely strong barriers with high levels of gene exchange in annual sunflowers

.

Evolution

66

:

1459

–

1473

.

Sapir

,

Y.

,

M. L.

Moody

,

L. C.

Brouillette

,

L. A.

Donovan

, and

L. H.

Rieseberg

.

2007

.

Patterns of genetic diversity and candidate genes for ecological divergence in a homoploid hybrid sunflower, Helianthus anomalus

.

Mol Ecol

16

:

5017

–

5029

.

Scascitelli

,

M.

,

K. D.

Whitney

,

R. A.

Randell

,

M.

King

,

C. A.

Buerkle

, and

L. H.

Rieseberg

.

2010

.

Genome scan of hybridizing sunflowers from Texas (Helianthus annuus and H. debilis) reveals asymmetric patterns of introgression and small islands of genomic differentiation

.

Mol. Ecol.

19

:

521

–

541

.

Schlötterer

,

C.

, and

D.

Dieringer

.

2005

. A novel test statistic for the identification of local selective sweeps based on microsatellite gene diversity. Pp.

55

–

64

in

D.

Nurminsky

, ed.

Selective sweep

.

Kluwer Academic/Plenum

,

New York, NY

.

Schluter

,

D.

2000

.

Ecological character displacement in adaptive radiation

.

Am. Nat.

156

:

S4

–

S16

.

Schwartz

,

A. K.

,

D. J.

Weese

,

P.

Bentzen

,

M. T.

Kinnison

, and

A. P.

Hendry

.

2010

.

Both geography and ecology contribute to mating isolation in guppies

.

PLoS ONE

5

:

e15659

.

Scopece

,

G.

,

C.

Lexer

,

A.

Widmer

, and

S.

Cozzolino

.

2010

.

Polymorphism of postmating reproductive isolation within plant species

.

Taxon

59

:

1367

–

1374

.

Silvertown

,

J.

,

C.

Servaes

,

P.

Biss

, and

D.

Macleod

.

2005

.

Reinforcement of reproductive isolation between adjacent populations in the Park Grass Experiment

.

Heredity

95

:

198

–

205

.

Sokal

,

R. R.

, and

N. L.

Oden

.

1978

.

Spatial autocorrelation in biology 1. Methodology

.

Biol. J. Linnean Soc.

10

:

199

–

228

.

Stölting

,

K. N.

,

R.

Nipper

,

D.

Lindtke

,

C.

Caseys

,

S.

Waeber

,

S.

Castiglione

, and

C.

Lexer

.

2013

.

Genomic scan for single nucleotide polymorphisms reveals patterns of divergence and gene flow between ecologically divergent species

.

Mol. Ecol.

22

:

842

–

855

.

Strasburg

,

J. L.

,

C.

Scotti-Saintagne

,

I.

Scotti

,

Z.

Lai

, and

L. H.

Rieseberg

.

2009

.

Genomic patterns of adaptive divergence between chromosomally differentiated sunflower species

.

Mol. Biol. Evol

26

:

1341

–

1355

.

Strasburg

,

J. L.

,

N. A.

Sherman

,

K. M.

Wright

,

L. C.

Moyle

,

J. H.

Willis

, and

L. H.

Rieseberg

.

2012

.

What can patterns of differentiation across plant genomes tell us about adaptation and speciation?

Philos. Trans. R. Soc. B-Biol. Sci.

367

:

364

–

373

.

Tang

,

S.

,

A.

Leon

,

W. C.

Bridges

, and

S. J.

Knapp

.

2006

.

Quantitative trait loci for genetically correlated seed traits are tightly linked to branching and pericarp pigment loci in sunflower

.

Crop Sci.

46

:

721

–

734

.

Turner

,

T. L.

,

M. W.

Hahn

, and

S. V.

Nuzhdin

.

2005

.

Genomic islands of speciation in Anopheles gambiae

.

PLoS Biol.

3

:

e285

.

Via

,

S.

2012

.

Divergence hitchhiking and the spread of genomic isolation during ecological speciation-with-gene-flow

.

Philos. Trans. R. Soc. B-Biol. Sci.

367

:

451

–

460

.

Via

,

S.

, and

J.

West

.

2008

.

The genetic mosaic suggests a new role for hitchhiking in ecological speciation

.

Mol. Ecol.

17

:

4334

–

4345

.

Weir

,

B. S.

1996

.

Genetic data analysis II

.

Sinauer Associates

,

Sunderland, MA

.

Wills

,

D. M.

, and

J. M.

Burke

.

2007

.

Quantitative trait locus analysis of the early domestication of sunflower

.

Genetics

176

:

2589

–

2599

.

Yatabe

,

Y.

,

N. C.

Kane

,

C.

Scotti-Saintagne

, and

L. H.

Rieseberg

.

2007

.

Rampant gene exchange across a strong reproductive barrier between the annual sunflowers, Helianthus annuus and H. petiolaris

.

Genetics

175

:

1883

–

1893

.

Author notes

Data Archived: Dryad doi: 10.5061/dryad.fd31n

© 2013, Society for the Study of Evolution

Supplementary data

Citations

Views

Altmetric

Metrics

Total Views 414

317 Pageviews

97 PDF Downloads

Since 1/1/2023

Citations

80 Web of Science

×

Email alerts

Citing articles via

• Latest

• Most Read

• Most Cited

More from Oxford Academic