Population genomics of parallel evolution in gene expression and gene sequence during ecological adaptation


Natural selection often produces parallel phenotypic changes in response to a similar adaptive challenge. However, the extent to which parallel gene expression differences and genomic divergence underlie parallel phenotypic traits and whether they are decoupled or not remains largely unexplored. We performed a population genomic study of parallel ecological adaptation among replicate ecotype pairs of the rough periwinkle (Littorina saxatilis) at a regional geographical scale (NW Spain). We show that genomic changes underlying parallel phenotypic divergence followed a complex pattern of both repeatable differences and of differences unique to specific ecotype pairs, in which parallel changes in expression or sequence are restricted to a limited set of genes. Yet, the majority of divergent genes were divergent either for gene expression or coding sequence, but not for both simultaneously. Overall, our findings suggest that divergent selection significantly contributed to the process of parallel molecular differentiation among ecotype pairs, and that changes in expression and gene sequence underlying phenotypic divergence could, at least to a certain extent, be considered decoupled processes.


The importance of natural selection on population divergence and the genesis of new species remains poorly understood. One reason for this limited knowledge is the stochasticity linked to the somewhat unique history of each population and species, which can overwhelm the fingerprint of adaptive divergence1. Instances of repeated, parallel phenotypic evolution in response to similar environmental pressures provide strong evidence of evolution by natural selection, as genetic drift is unlikely to generate a concerted change in multiple, independent lineages2,3. However, the underlying genetic basis of this process is unclear. In particular, we know very little as to whether selection acts upon the same genetic machineries to generate repeated phenotypes, or if its action follows alternative genetic routes4,5,6. Similarly, it remains unknown to what extent constraints faced by organic evolution might facilitate the repeated use of the same genes during independent phenotypic evolution7,8.

Evidence for parallel evolution has been found in many taxa4,9. However, one limitation of our view that parallel evolution is rather abundant comes from the fact that many studies are based on targeted candidate gene surveys that suffer from an inevitable ascertainment bias, as they do not allow answering whether repeated genetic changes are ubiquitous across the genome or more frequent than the neutral expectation3. Recent studies using a genome-wide approach have provided some unbiased insights into our understanding of the level of genome-wide repeatability linked to parallel evolution. For instance, molecular footprints of selection underlying parallel phenotypic evolution in cichlid fishes10, Australian groundsel11 and lake trout12 involve replicated evolution on a rather restricted subset of genes and, more frequently, divergence events that are unique to each population. Similarly, the early stages of parallel speciation in the stick insect Tinema cristinae involve mostly nonparallel divergence despite evidence of the importance of repeated selection on the same genes13. However, the repeatability of evolution through the reuse of the same genes may be substantial amongst recently diverged lineages9,14. Overall, these and other studies15,16,17,18 suggest that the genomic architecture underlying parallel phenotypic evolution frequently follows complex genetic trajectories, affecting multiple loci that show a mosaic pattern of both repeatable and idiosyncratic divergence, and where ancestral standing variation is frequently an important source of adaptive variation.

Very few studies have attempted to address the extent to which parallel gene expression differences and genomic divergence underlie parallel phenotypic traits19,20,21,22. This question is of central importance, because adaptive variation is likely to be underpinned by changes in both regulatory and coding sequences23. If evolution in coding sequences and regulatory regions are two highly related phenomena, then patterns of differentiation in coding sequence and gene expression should be markedly similar, i.e. they should be coupled. However, previous attempts to test the coupling between coding sequences and gene expression in multicellular organisms have given conflicting results, with markedly similar patterns of differentiation found in some datasets24,25,26,27, but very dissimilar in others17,28,29. To add further uncertainty, the specific mechanism underlying these observations remains elusive. Thus, adaptive selection driving rapid evolution of both gene expression and coding sequence may account for the coupling24, but also variation in functional constraints, in which genes less constrained in coding sequence would also be less constrained in expression26,27,28,29,30. Alternatively, markedly dissimilar patterns of differentiation would point towards the possibility that changes in coding sequence and gene expression underlying phenotypic evolution play different roles during evolution and could, at least to a certain extent, be considered decoupled processes31,32. Unveiling the degree of convergence at different levels of genomic organization will help to establish to what extent natural selection, genetic constraints, or independent modes of evolution, determine whether patterns of genetic differentiation associated with adaptation are predictable.

The marine snail Littorina saxatilis provides an excellent opportunity for testing these aspects of evolutionary repeatability. In this ovoviviparous species, pairs of ecotypes adapted to distinct parts of the same shores have repeatedly evolved in different geographical regions from Europe33,34,35. Parallel phenotypic divergence involves many traits, including body size, shell shape, shell thickness, and behavior36. Hybridization occurs in a relatively narrow zone, but gene flow among ecotypes is restricted due to assortative mating, immigrant inviability, and habitat choice37,38,39. Consistent with the prediction of parallel evolution, pairs of sympatric ecotypes cluster in phylogenetic trees by geographic origin but not by ecotype40. Computer simulations assessing the confounding effect of gene flow on phylogenetic inference confirm this result, demonstrating that the time elapsed since the emergence of ecotypes would not be enough to erode the distinctive phylogenetic signal linked to a parallel or a non-parallel (allopatric) origin of ecotypes41. Moreover, the comparison between alternative evolutionary models further supports that data better fit a scenario in which the separation of pairs of ecotypes occurred in parallel at both regional and local scales35. Recent genomic studies comparing populations from three geographically distant regions (Spain, Sweden and United Kingdom) suggest that footprints of selection are frequently region-specific42,43, or even site-specific at a very local scale44. Still, no study in Littorina has so far investigated the extent of parallelism in gene expression nor the relation between variation in gene expression and divergence in coding sequences.

Despite the ongoing development of next-generation sequencing technologies for genome-wide evolutionary analyses, it remains technically and financially unapproachable for many laboratories to sequence whole genomes or transcriptomes. Microarrays remain a simple and inexpensive alternative for genotype-related purposes and gene expression analyses45. Array-based comparative genomic hybridization can be accurately used as a proxy to estimate genome-wide divergence by comparing hybridization intensities of individuals on the microarray46,47. Modifications of this method have been successfully used to identify SNPs or copy number variants without the need of allele-specific probes, thanks to a linear relationship between hybridization signal and sequence divergence47. Similarly, microarrays remain widely used for gene expression profiling, as correlation between microarray data and other platforms such as RNA-seq is usually pretty good48,49. Although microarrays may be problematic for the study of low expressed genes, RNA-seq shows a greater degree of intensity-dependent variation than do microarrays50, which has led some authors to recommend the use of RNA-seq51,52, and others challenging that conclusion53,54.

Here we combine genome-wide evolutionary analyses of coding sequences and gene expression data using microarrays for investigating the molecular basis of adaptive divergence, employing L. saxatilis ecotypes from NW Spain as a model system. In this region, a large “crab ecotype” and a smaller “wave ecotype” have evolved repeatedly in response to crab predation and wave exposure respectively33,35,40,55. The recent origin of these ecotypes (<10,000 years)35 is expected to be associated with high levels of shared genetic constraints and standing variation that would facilitate a rapid and more pervasive repeated evolution. Our objectives were i) assess to what extent expression and sequence differences between ecotypes affect the same genes, ii) determine the level of correspondence between gene expression divergence and coding sequence divergence, and iii) quantify how natural selection may affect repeatability. Our results stress the important contribution that both gene regulation and coding regions can make to rapid phenotypic evolution and adaptation.

Materials and Methods


Adult snails were collected in August 2010 from three Galician (NW Spain) localities: Burela (N 43°40′54.8′′; W 7°22′1.8′′), Roncudo (N 43°16.31′32′′; W 8°59.23′93′′′), and Silleiro (N 42°6′17.20′′; W 8°53′56.59′′) (Fig. 1). At each locality, specimens from the “crab” and the “wave” ecotypes were obtained from the upper and lower shore level respectively to avoid collecting intermediate forms (i.e. hybrids). Snails were immediately transported to our facilities in the ECIMAT Marine Station, and sexed according to the presence or absence of penis. Specimens used for DNA extractions were stored at −80 °C until processed. Specimens targeted for expression analysis were maintained alive in an aquarium under identical environmental conditions for two weeks using a continuous sea water flow (16 °C, 36.1‰ salinity, 6.8 mg/L oxygen level). This step aimed to minimize the impact of environmental variance on gene expression patterns by ensuring that all individuals shared the same environmental conditions prior to expression analysis. After this period, snails were stored at −20 °C in RNAlater solution (Ambion) until RNA extraction.

Figure 1

Experimental design. (A) Geographical situation of the three localities sampled in this study. (B) Transcriptomic scan: 4 pools were assayed for each ecotype and locality (24 pools in total); each pool consisted of 15 females with an equimolecular contribution to the pool. (C) Genomic sequence scan: 12 individuals were assayed for each ecotype and locality (72 individuals in total); each single individual was separately hybridized against the array.

RNA and DNA extraction

Total RNA was isolated from the foot muscle tissue of single females using TRIZOL reagent (Invitrogen) according to the manufacturer’s instructions. Females have the advantage of providing larger RNA yields than males given their bigger size, while displaying expression patterns similar to those from males across the different ontogenetic stages of each ecotype56. The Turbo DNA-freeTM kit (Ambion) was used to remove any remaining DNA from RNA extractions. Genomic DNA was isolated from the foot muscle tissue of single males and females using a CTAB extraction method57 modified to include RNAse treatment. DNA samples were further cleaned with NucleoSpin columns (Macherey-Nagel) following manufacturer’s instructions. RNA and DNA purity was assessed using a NanoDrop spectrophotometer (NanoDrop Tech. Inc., Wilmington, DE). All extractions were standardized to 100 ng/µL after checking their integrity in agarose gels.

NimbleGen L. saxatilis microarray

The L. saxatilis oligonucleotide microarray58 was developed by NimbleGen Roche (090824_L_saxatilis_expr_HX12, 12 × 135K array format) on the basis of draft or versioned assemblies from the Littorina saxatilis EST database59 and the GenBank database. The microarray contained sequence information based on 25,205 partial transcripts, hereafter referred to as “genes” for simplicity, and which represent the coding part of the genome. These transcripts were obtained mainly by 454 sequencing of cDNA libraries from both the “crab” and “wave” ecotypes59. Each gene was usually represented on the array by five non-overlapping 60-nt probes. Each NimbleGen slide contained 12 identical subarrays. We used this microarray to assess variation in gene expression and also in genomic sequence using, for the latter, a comparative genomic hybridization (CGH) approach, which is based on hybridization of labeled DNA fragments to a microarray46. This allowed us to compare variation in expression and nucleotide sequence for the same subset of the L. saxatilis genome.

Gene expression profiling

For this analysis, pools of total RNAs were retrotranscribed to cDNAs representing the coding part of the transcriptome, which were then compared to establish patterns of over- and under-expressed genes. We prepared four replicate samples from each ecotype and locality (24 samples in total), each including 15 pooled female specimens with an equimolar RNA contribution of each specimen to the pool. The RNA from each pool was retrotranscribed with the SuperScriptTM Double-Stranded cDNA Synthesis Kit (Invitrogen) following the conditions recommended by the manufacturer. Cy3 labeling was performed with the NimbleGen One-Color DNA labeling kit (Roche) using a starting amount of 1 µg of cDNA per pool. Then, for each pool, 4 µg of Cy3 labeled cDNA was resuspended in 12 µL of hybridization solution, of which 6 µL was applied onto a subarray. Pools were randomly distributed in the subarrays. Hybridization was carried out at 42 °C for 19 h on a NimbleGen Hybridization System with continuous mixing. After hybridization, the arrays were washed in buffers of various stringencies using the NimbleGen Washing kit. Arrays were scanned using an Agilent G2565AA microarray scanner (Agilent Technologies) with a resolution of 2 µm. The quality of the images was assessed using the NimbleScan v.2.5. software (NimbleGen/Roche), discarding those images with signal intensity or other metrics outside the range recommended by the manufacturer.

Data were extracted using NimbleScan v.2.5 and analyzed in the R/Bioconductor statistical environment. The pdfInfoBuilder and oligo60 packages were used for data handling and pre-processing, with the robust multichip average (RMA) method61 used for background correction, quantile normalization and probe-level summarization of the microarray samples. A single value was obtained for each gene, resulting from each summarization of probe-level data. To decide whether a gene was expressed, a threshold level representing the “background signal” was calculated based on the average hybridization signal of the empty spots present in the array. Genes for which more than 20% of the probes had an average hybridization signal lower than the “background signal” were disregarded62.

Genomic divergence profiling

For the analysis of variation in genomic sequence, each subarray hosted the genomic DNA of one single individual and the genomic DNA of a common reference sample. Thus, in this experiment, genomic DNA was hybridized against the coding portion of the L. saxatilis genome represented in the microarray. The reference sample was composed of a DNA pool of 100 “crab” and “wave” snails from two British L. saxatilis locations (Dunvar and Thornwick, the latter used in the array design58), to ensure consistent and non-zero hybridization signals for the reference sample in all the probes from the array. Therefore, “crab” and “wave” Galician ecotypes should be equally diverged from both the array and the reference, as the latter included mostly (90%) individuals from the same location used in the array design.

The NimbleGen/Roche Dual label kit was used to label the reference sample (Cy5 dye) and the DNA of each specimen (Cy3 dye) following manufacturer´s instructions. Hybridization procedures, array scanning and quality control of the images obtained were performed as in the expression arrays, but increasing the amount of labeled DNA (20 µg) and incubation time (72 h). The whole experiment included 72 Galician snails (12 per ecotype and locality) for which genomic DNA extracts were individually hybridized to the array. The data from scanning pictures generated by NimbleScan were parsed using ringo63, an R/Bioconductor package. The signal intensity data for each channel was corrected for the local background signal using the normexp + offset method64, log2-transformed, and quantile normalized using the method proposed for two channels65, as implemented in the package limma for R/Bioconductor66. We performed a probe-level data analysis to test DNA sequence differences between the distinct gene fragments included in a probe set and the hybridized DNA.

Low hybridization signals (below 10.35) in the L. saxatilis microarray may correspond in some instances to probes spanning exon boundaries and/or untranslated regions58. To account for this possible source of noise in our data, and also to exclude probes that were not accurately detected in the array, we have filtered these sequences by removing probes with an average hybridization signal lower than the “background signal” (i.e. 10.7) of each channel. Untranslated regions would similarly generate low hybridization signals in the expression study, and these were also removed from the data (see above). Only genes containing probes that simultaneously passed genome and expression profiling filters were used in the subsequent analyses, to ensure that all the probes/genes only span coding sequences.

Statistical analysis

Differential expression (genes) and genomic divergence (probes) were determined using the linear modeling analysis for microarrays implemented in the limma package66 with empirical Bayes adjustment to the variance. For both expression and sequence divergence records, three different linear models were fitted to the data and contrast matrices were created to identify (i) differences between ecotypes, localities and their interaction, (ii) differences between ecotypes within localities, and (iii) differences between localities for each ecotype. The resulting p-values were corrected for multiple tests using the binomial sequential goodness of fit procedure (SGoF)67 at α = 0.05.

For genes/probes showing significant differences between ecotype pairs in the three localities examined, we computed the p-value that the observed parallelism could be due to chance alone using both a randomization test68 and the algorithm developed by Derome et al.69 modified to include three localities (P. Duchesne, personal communication). Randomization tests were also used to estimate the random expectation of parallel and nonparallel changes, and of directional and nondirectional changes. For each randomization test, data were sorted 200,000 times and the corresponding outcome was obtained after multitest correction. An unpaired t test70 was used to test that intrapopulation variance differs between genes/probes showing directional versus nondirectional parallel changes. The program Blast2GO71 was used to identify which GO terms were significantly over-represented in those genes or probes showing significant differences for each analysis. General patterns of gene expression and sequence divergence were visualized with heat maps using R/Bioconductor.


We examined transcriptomes from pools including snails from the “crab” or “wave” ecotypes, and variation in the coding sequences of single snails. Snails were collected from three isolated, independently evolved population pairs of sympatric “crab” and “wave” ecotypes (Fig. 1) that previously showed a repeatable morphological divergence by parallel evolution33,35,40. Variation in expression and genomic sequence was determined for the same genes using a microarray specifically developed for L. saxatilis. After quality control of the hybridized arrays, we retained 22 out of 24 pools for gene expression, 69 out of 72 individuals for coding sequence divergence, and 17,431 genes.

Table 1 shows the proportion of genes displaying expression and genomic sequence differences between pairs of ecotypes for the three localities examined after using SGoF multitest correction (α = 0.05). Pairs of ecotypes living in the same site displayed significant differences in expression and genomic sequence, respectively, for up to 17.2 and 25.5% of all assayed genes. Transcriptomic differences were more prevalent than genomic differences in only one of the three localities assayed. The majority of divergent genes (93–94%) were divergent either for gene expression or genomic sequence, but not for both simultaneously. The number of differences between ecotype pairs varied among localities (P < 0.05; G test), and many of these differences (40.6–79% for gene expression; 68–71% for genomic divergence) occurred only in a single locality. Therefore, we tested whether differences between ecotype pairs frequently involved the same genes in the three localities (i.e. parallel changes). The observed numbers of genes with parallel changes in expression and genomic sequence were, respectively, 146 (0.8% of all assayed genes) and 216 (1.2%). Such levels of parallelism are highly unlikely just by chance (p < 10−5 for both expression and genomic data using a permutation test, or the algorithm by Derome et al.69), and therefore consistent with repeatable genetic differentiation by natural selection. Remarkably, as few as 15 genes displayed simultaneous parallel changes in expression and genomic sequence, representing 4% of all genes with parallel changes.

Table 1 Number of genes (percentage and absolute values) showing transcriptomic or genomic sequence differences between pairs of ecotypes in each single locality, and across the three localities simultaneously (parallel changes), after using SGoF multitest correction (α = 0.05).

We examined the directionality of observed parallel differences. To increase precision, genomic divergence is referred to in the subsequent analyses in terms of the 354 probes rather than the 216 genes that showed parallel changes (see methods). The majority of parallel differences between ecotype pairs were due to changes in the same direction (directional changes), whereas only a few were due to differences in opposite directions (non-directional changes) (Fig. 2). Specifically, up to 132 (90%) of all genes displaying parallel differences in expression showed directional changes (54% of which were up-regulated in the “crab ecotype”). Similarly, 294 (83%) of all probes with parallel variation in genomic sequence also showed directional changes (75% of which displayed a higher hybridization signal in “crab” than “wave” snails). The large over-representation of directional parallel differences for both expression and sequence divergence data is highly unlikely just by chance (each p < 0.0001), as determined by randomization tests resorting expression or genomic data sets (Fig. 2).

Figure 2

Directionality of observed parallel differences for (A) the transcriptomic and (B) genomic sequence scans. Top pannels: heat maps of hybridization signals across localities for the 146 genes (left) and 354 probes (right) that displayed significant parallel changes after multitest correction. Yellow indicates a greater hybridization signal for the “wave ecotype”, and grey a higher hybridization signal for the “crab ecotype”. Directional parallel changes appear as straight horizontal lines of one single color (yellow or grey). Nondirectional parallel changes display horizontal lines scattered of yellow and grey streches. Botton pannels: barplots indicating the observed number of directional (D) and nondirectional (ND) changes versus the random expectation, obtained using a randomization test (see methods). Vertical error lines indicate the 95% confidence intervals across all replicates of the randomization test.

We also determined whether the mean intrapopulation variance differs between genes/probes showing directional versus nondirectional parallel changes. We found that variance in expression and sequence divergence for directional changes was twice less than that observed for nondirectional changes (Fig. 3). These differences were statistically significant for both variation in expression (p = 0.0007) and genomic sequence (p = 0.0185) using a randomization test, and also using 2-tailed t tests (all p < 0.027).

Figure 3

Mean intrapopulational variance in hybridization signal for directional parallel changes and nondirectional parallel changes. Variance for directional changes was significantly lower than that observed for nondirectional changes for both transcriptomic and genomic sequence scans.

To further assess the nature of evolutionary forces underlying parallel variation, we determined which proportion of genes/probes showing parallel and nonparallel differences among ecotype pairs also showed a significant geographic differentiation among the three localities for the “crab” or “wave” ecotypes. We found that, independently of the ecotype considered, genes/probes with parallel changes showed more frequently geographic differentiation than genes/probes with nonparallel changes after SGoF multitest correction (α = 0.05), and that this pattern was more pronounced for genomic than transcriptomic differences (Fig. 4). The proportion of genes/probes with parallel changes that displayed geographic differentiation deviated more strongly (p < 0.001) from the random expectation than the proportion observed for nonparallel changes.

Figure 4

Geographical differentiation for genes (left) and probes (right) that displayed parallelism in gene expression or genomic sequence, and for those not showing parallelism. The y axis indicates the percentage of cases for which these genes/probes showed statistically significant differences among the three localities. Vertical error lines indicate the 95% confidence intervals for the random expectation (almost not appreciable in the figure), obtained resampling 1000 times at random the same number of genes/probes showing either parallel or nonparallel changes from the total set of genes/probes.

We used an enrichment analysis with BLAST2GO to test whether parallel differentiation is linked with specific functional groups. This analysis did not identify enriched gene/probe sets after correction for multiple testing when the whole data set or only intra-site GO enrichment tests were considered. Nevertheless, some genes/probes with the most extreme parallel directional changes in hybridization signal included annotations expected to be involved in adaptation. For example, the Dermatopontin 2 (for gene expression profiling) and the Keratin-associated protein 4–3 (for sequence divergence profiling) are involved, respectively, in the formation of the shell72 and the operculum73, key features defining differences between ecotype pairs (Supplementary Tables S1 and S2). We also tested whether the differences between ecotype pairs that are unique to each locality are linked with specific functional groups. For this analysis, significant enrichment GO terms were observed only for gene expression profiling after correction for multiple testing. We observed an important enrichment in energetic metabolism GO terms for Burela, but almost no GO terms were shared among pairs of localities, and none between the three localities simultaneously, either for the categories of molecular function, biological process, or cellular component (Supplementary Figs S1 and S2).


Parallelism in gene expression and coding sequences

We report evidence that parallel differences in expression and sequence divergence of a limited set of genes underlay the repeated phenotypic divergence of replicate pairs of L. saxatilis ecotypes. The observed numbers of parallel differences in gene expression and sequence divergence largely exceeded the random expectation. Several results suggest that adaptive selection played a role, direct or indirect, in the process of molecular divergence among ecotypes.

We expect that genes repeatedly recruited by strong natural selection would show striking habitat-associated differences74, would display less variation than those under weaker selection69, and would show a higher geographical differentiation75. Overall, our results fit these expectations and are consistent with a scenario in which the same subset of genes, or regulatory regions, were repeatedly recruited by natural selection in populations adapted to similar habitats. Parallel changes in hybridization signal were nearly restricted to directional changes, denoting a repeated and significant habitat-association among independently evolving populations of similar phenotype that cannot be explained by chance. Therefore, directional parallel changes showed a lower intrapopulation variance than nondirectional parallel changes, as expected from a stronger impact of selection in the former69,76. Also, the distinctive higher geographic differentiation in expression and coding sequence for genes displaying parallelism did not match the random expectation. This in turn suggest that geographic differentiation for genes showing parallelism is determined by the joint action of divergent selection and stochastic forces, whereas geographic differentiation at nonparallel genes is mostly driven by stochastic forces. Last, we examined the function of genes with parallel divergence. Although annotation was very incomplete due to the poor representation of mollusk sequences in public databases77, some of the genes that could be annotated exhibited functions related with well know adaptive phenotypic characters, such as the formation of the snail shell and the operculum.

Despite the observed parallelism, the majority of differences in gene expression and coding sequence were not shared among localities. Low sharing of divergent genes contrasts with the expectation of high gene reuse for the parallel evolution of individual phenotypic traits among closely related taxa and populations9. Several non-mutually exclusive factors may account for this discrepancy. First, we might have underestimated the parallelism existing in natural populations. For example, parallelism owing to low diverged alleles, or to alleles equally diverged from the reference but carrying mutations at different sequence positions, could remain somewhat undetected using microarrays. Sequence mismatches due to sequence polymorphisms could also affect the ability to detect parallelism in gene expression. Therefore, the number of genes showing parallelism in our study should be viewed as conservative. Yet, the impact of these challenges on our patterns of parallelism seems to be modest since we detected many differences between ecotype pairs of a very recent origin within each locality, and still only a minor fraction of these differences were repeatable among localities. Moreover, expression measurements in different species did not reveal a consistent variation in signal intensity due to sequence mismatches24,78, since the expression of each gene is calculated as the average intensity for each probe set. Second, if divergent traits in Littorina (e.g. shell size and shell shape) are highly polygenic, then they may show greater genetic redundancy than traits determined by a single gene or molecular pathway. As such, changes in different pathways of a complex polygenic trait could lead to similar phenotypes and show less repeatable genetic signatures of adaptation3,22. Third, patterns of parallel evolution could be more common at higher levels of biological organization79. Thus, sharing of physiological processes, biochemical pathways, or organismal functions may therefore be more prevalent than observed at the gene or regulatory level7,80,81,82,83. Last, a number of biases could have inflated the very high expectation of gene reuse, such as publication bias against non-sharing genetic patterns, or an emphasis on genes of large effect that may not be illustrative of the true spectrum of phenotypes3,9,84.

Other studies in a number of different organisms have similarly demonstrated little sharing of sequence divergence10,13,85,86 and gene expression patterns69,87 linked to recent parallel evolution. Our findings are consistent with recent genome scan studies in Littorina indicating a low sharing of genomic divergence among ecotypes that arose in parallel in different parts of Europe but also, as shown in Sweden, in geographically close localities42,44. Moreover, parallelism between ecotype pairs mostly involved genomic regions under strong selection42,43, thus supporting our hypothesis that genes showing shared genomic and expression divergence are likely targeted by natural selection.

About 10% of sequence differences in the Littorina array are expected to be copy number variants58. Thus, processes such as duplication and subsequent neofunctionalization might also play a role in the divergence among ecotypes4,22,88. Our results show that the Littorina microarray is able to detect more sequence differences among ecotype pairs than reported in a previous study using this same microarray58. Several reasons explain this gain in power. In the former study, a reference sample was not used and data was not filtered, thus increasing the inter-array variance due to technical noise effects89. Also, a probe-based analysis was not used to assess sequence differences. Instead, the relative hybridization signal for each gene represented on the array was calculated as the average intensity for each probe set. Thus, any mismatch signal resulting from a target DNA polymorphism affecting one single probe would be averaged with the remaining gene probes and therefore would be difficult to detect. In addition, sequence comparisons between ecotype pairs within a single locality were based on only four individuals from each ecotype. To obtain more power, in the present study the sample size was increased to 12 “crab” and 12 “wave” individuals per locality (72 individuals in total versus 8 in the former study for Galician snails). Finally, we used the limma package66 with empirical Bayes adjustement to the variance, that allows an improved estimation of variance respect to the conventional ANOVA tests previously used.

The chances of successfully capturing adaptive loci are greater when targeting functionally important regions. In this study, we simultaneously screened patterns of expression and sequence variation for the coding fraction of the genome. Thus, although some of the genes displaying parallel evolution could be false positives, the likelihood of identifying genes directly or indirectly (through linked selection) targeted by selection may be substantial. Remarkably, a large number of divergence events occurred in a single ecotype pair. An unknown proportion of this non-shared divergence could have resulted from stochastic processes, adaptive changes, or a combination of these factors. Overall, our results suggest that the genomic architecture underlying parallel phenotypic divergence probably followed a complex evolutionary path, affecting multiple loci in a mosaic pattern of both repeatable and idiosyncratic divergence, and where the repeated element involved many regions affected by natural selection.

Divergence in gene expression is decoupled from divergence in coding sequence

Our results showed that patterns of differentiation in gene expression and coding sequence were markedly dissimilar. The majority of divergent genes were divergent either for gene expression or genomic sequence, but not for both simultaneously. Remarkably, as few as 15 genes displayed simultaneous parallel changes in expression and genomic divergence, representing 4% of all genes with parallel changes. Such result is consistent with the hypothesis that expression and gene sequence differences underlying phenotypic divergence could, at least to a certain extent, be considered decoupled processes31.

One concern is that the comparison between expression and sequence variation could have been partly affected by misleading expression measurements resulting from sequence mismatches between the samples used for expression analysis and the reference upon which the array was designed. However, sequence mismatches cannot account for the dissimilarity in patterns of differentiation, since such mismatches should also be present in the samples used for sequence differentiation and would generate a correlated signal between gene expression and sequence divergence90. It might be also possible that our genome scan was not sensitive enough to pick up all the genes carrying a single nucleotide variant difference. However, this lack of sensitivity should equally affect the coding regions of genes displaying either expression or no expression differences, and thus cannot explain the dissimilarity. Also, for gene duplications where both genes are retained, similar patterns of differentiation are expected for gene expression and gene sequence if both diverge at clock-like rates91. It is also unlikely that power differences between expression and sequence divergence studies can account for the dissimilarity in patterns of differentiation, as they should lead to consistently larger differences between ecotype pairs for one such level (expression or sequence divergence) in the three localities examined and, therefore, genes with significant differences in the less powerful study should also display concordant significant differences for the most powerful one. These patterns are not observed in our data (Table 1). These considerations further support that, independently of the source of variation or error considered, gene expression and coding sequences appear to evolve differently as ecotypes repeatedly adapt to complex ecological gradients.

The decoupling between gene expression and coding sequence differentiation is consistent with the existence of trans-regulation factors driving gene expression evolution, but also with the evolutionary decoupling of cis-regulatory regions and coding sequences. For example, in Drosophila melanogaster, gene expression differences among distinct strains are correlated with 5 prime sequences but not with coding sequences, thus supporting that differentiation at cis-regulatory regions is decoupled from differentiation at coding regions92. Our results are in line with what was observed among closely related ecotypes of lake whitefish20, rainwater killifish93, and woody sunflower29, where differentiation of gene expression and coding sequences were also decoupled. However, what distinguishes our study from these previous ones is that we focus on genes displaying parallel evolution across similar environmental gradients. As such, the genes we identify are more likely to underlie variability related with traits implied in a relevant adaptive response.

Functional interpretations of the decoupling between gene expression and sequence divergence should be taken cautiously, as array data do not allow to tell apart effects due to nonsynonymous mutations that alter the amino acid sequence from those due to synonymous mutations that do not affect the amino acid composition. Yet, even if most of changes occurred at synonymous sites, it would be needed to explain why in our data differentially expressed genes do not show such changes. This would point to the existence, even for synonymous sites, of selective constraints slowing down the evolution of coding sequences for genes displaying parallel changes in expression. Indeed, evidence exists indicating that synonymous sites appear to evolve slower than expected under neutrality in a way apparently consistent with weak selection in organisms as diverse as insects, yeast, worms, chicken or mammals94,95,96,97,98. For example, in D. melanogaster, 22% of four-fold synonymous sites are evolving under strong constraints, and genes with such constrained sites tend to be especially relevant, highly expressed, and often involved in developmental networks99. Therefore, our results may point to the possibility of some division of tasks underlying coding and regulatory regions, as previously hypothesized100.

Genome-wide data on expression variation versus sequence divergence are uncommon. Thus, this study provides a rare opportunity to determine the relative contribution of expression and coding changes underlying parallel phenotypic evolution. Our results suggest that both coding and expression changes contribute to parallel divergence among pairs of ecotypes. The relative contribution of expression and sequence changes varied among localities, but there was not an overall preeminent role of expression over coding sequence differences across all localities. Our results differ from other studies in three-spined sticklebacks providing a major role to gene expression variation (up to 83% of all differences) over coding sequence variation in the evolution of parallel phenotypic divergence16. This suggests that differences in life history features and the number, location and interactions among genes and regulatory regions, may generate very diverse outcomes in the molecular fingerprint underlying phenotypic adaptation23.

Data Availability

The expression and genomic divergence dataset is available in the NCBI gene expression Omnibus under the accessions GSE120697 and GSE120698 respectively.


  1. 1.

    Brandvain, Y. & Wright, S. I. The limits of natural selection in a nonequilibrium world. Trends Genet. 32, 201–210 (2016).

    CAS  PubMed  Article  Google Scholar 

  2. 2.

    Schluter, D. & Nagel, L. M. Parallel speciation by natural selection. Am. Nat. 146, 292–301 (1995).

    Article  Google Scholar 

  3. 3.

    Elmer, K. R. & Meyer, A. Adaptation in the age of ecological genomics: insights from parallelism and convergence. Trends Ecol. Evol. 26, 298–306 (2011).

    PubMed  Article  Google Scholar 

  4. 4.

    Stern, D. L. The genetic causes of convergent evolution. Nat. Rev. Genet. 14, 751–764 (2013).

    CAS  PubMed  Article  Google Scholar 

  5. 5.

    Faria, R. et al. Advances in ecological speciation: an integrative approach. Mol. Ecol. 23, 513–521 (2014).

    PubMed  Article  Google Scholar 

  6. 6.

    Westram, A. & Johannesson, K. Parallel speciation In: Encyclopedia of Evolutionary Biology (ed. Kliman, R.) 212–219 (Elsevier, 2016).

  7. 7.

    Losos, J. B. Convergence, adaptation, and constraint. Evolution 65, 1827–1840 (2011).

    PubMed  Article  Google Scholar 

  8. 8.

    Wagner, M. R. & Mitchell-Olds, T. Repeated phenotypic changes highlight molecular targets of convergent evolution. Genome Biol. 12, 124 (2011).

    PubMed  PubMed Central  Article  Google Scholar 

  9. 9.

    Conte, G. L., Arnegard, M. E., Peichel, C. L. & Schluter, D. The probability of genetic parallelism and convergence in natural populations. Proc. R. Soc. Lond. B Biol. Sci. 279, 5039–5047 (2012).

    Article  Google Scholar 

  10. 10.

    Kautt, A. F., Elmer, K. R. & Meyer, A. Genomic signatures of divergent selection in a “natural experiment”, the young parallel radiations of Nicaraguan crater lake cichlid fishes. Mol. Ecol. 21, 4770–4786 (2012).

    PubMed  Article  Google Scholar 

  11. 11.

    Roda, F. et al. Convergence and divergence during the adaptation to similar environments by an Australian groundsel. Evolution 67, 2515–29 (2013).

    PubMed  Article  Google Scholar 

  12. 12.

    Perreault-Payette, A. et al. Investigating the extent of parallelism in morphological and genomic divergence among lake trout ecotypes in Lake Superior. Mol. Ecol. 26, 1477–1497 (2017).

    CAS  PubMed  Article  Google Scholar 

  13. 13.

    Soria-Carrasco, V. et al. Stick insect genomes reveal natural selection’s role in parallel speciation. Science 344, 738–742 (2014).

    ADS  CAS  PubMed  Article  Google Scholar 

  14. 14.

    Bradic, M., Teotónio, H. & Borowsky, R. L. The population genomics of repeated evolution in the blind cavefish Astyanax mexicanus. Mol. Biol. Evol. 30, 2383–2400 (2013).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  15. 15.

    Manceau, M., Domingues, V. S., Linnen, C. R., Rosenblum, E. B. & Hoekstra, H. E. Convergence in pigmentation at multiple levels: mutations, genes and function. Philos. Trans. R. Soc Lond B. Biol. Sci. 365, 2439–2450 (2010).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  16. 16.

    Jones, F. C. et al. The genomic basis of adaptive evolution in threespine sticklebacks. Nature 484, 55–61 (2012).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  17. 17.

    Renaut, S., Grassa, C., Moyers, B., Kane, N. & Rieseberg, L. The population genomics of sunflowers and genomic determinants of protein evolution revealed by RNA-seq. Biology 1, 575–596 (2012).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  18. 18.

    Holliday, J. A., Zhou, L., Bawa, R., Zhang, M. & Oubida, R. W. Evidence for extensive parallelism but divergent genomic architecture of adaptation along altitudinal and latitudinal gradients in Populus trichocarpa. New Phytol. 209, 1240–1251 (2016).

    CAS  PubMed  Article  Google Scholar 

  19. 19.

    Khaitovich, P. et al. Parallel patterns of evolution in the genomes and transcriptomes of humans and chimpanzees. Science 309, 1850–1854 (2005).

    ADS  CAS  Article  Google Scholar 

  20. 20.

    Jeukens, J., Renaut, S., St-Cyr, J., Nolte, A. W. & Bernatchez, L. The transcriptomics of sympatric dwarf and normal lake whitefish (Coregonus clupeaformis spp., Salmonidae) divergence as revealed by next-generation sequencing. Mol. Ecol. 19, 5389–5403 (2010).

    PubMed  Article  Google Scholar 

  21. 21.

    Zhao, L., Wit, J., Svetec, N. & Begun, D. J. Parallel gene expression differences between low and high latitude populations of Drosophila melanogaster and D. simulans. Plos Genet. 11, e1005184, https://doi.org/10.1371/journal.pgen.1005184 (2015).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  22. 22.

    Yeaman, S. et al. Convergent local adaptation to climate change in distantly related conifers. Science 353, 1431–1433 (2016).

    ADS  CAS  PubMed  Article  Google Scholar 

  23. 23.

    Hoekstra, H. E. & Coyne, J. A. The locus of evolution: evo-devo and the genetics of adaptation. Evolution 61, 995–1016 (2007).

    PubMed  Article  Google Scholar 

  24. 24.

    Nuzhdin, S. V., Wayne, M. L., Harmon, K. L. & McIntyre, L. M. Common pattern of evolution of gene expression level and protein sequence in Drosophila. Mol. Biol. Evol. 21, 1308–1317 (2004).

    CAS  PubMed  Article  Google Scholar 

  25. 25.

    Jordan, I. K., Mariño-Ramírez, L. & Koonin, E. V. Evolutionary significance of gene expression divergence. Gene 345, 119–126 (2005).

    CAS  Article  Google Scholar 

  26. 26.

    Hunt, B. G., Ometto, L., Keller, L. & Goodisman, A. D. Evolution at two levels in fire ants: the relationships between patterns of gene expression and protein sequence evolution. Mol. Biol. Evol. 30, 263–271 (2012).

    PubMed  Article  Google Scholar 

  27. 27.

    Hodgins, K. A., Yeaman, S., Nurkowski, K. A., Riesenberg, L. H. & Aitken, S. N. Expression divergence is correlated with sequence evolution but not positive selection in conifers. Mol. Biol. Evol. 33, 1502–1516 (2016).

    CAS  PubMed  Article  Google Scholar 

  28. 28.

    Tirosh, I. & Barkai, N. Evolution of gene sequence and gene expression are not correlated in yeast. Trends Genet. 24, 109–113 (2008).

    CAS  PubMed  Article  Google Scholar 

  29. 29.

    Moyers, B. T. & Riesenberg, L. H. Divergence in gene expression is uncoupled from divergence in coding sequence in a secondarily woody sunflower. Int. J. Plant. Sci. 174, 1079–1089 (2013).

    CAS  Article  Google Scholar 

  30. 30.

    Warnefors, M. & Kaessmann, H. Evolution of the correlation between expression divergence and protein divergence in mammals. Genome Biol. Evol. 5, 1324–1335 (2013).

    PubMed  PubMed Central  Article  Google Scholar 

  31. 31.

    Tirosh, I., Reikhav, S., Levy, A. & Barkai, N. A yeast hybrid provides insight into the evolution of gene expression regulation. Science 324, 659–62 (2009).

    ADS  CAS  PubMed  Article  Google Scholar 

  32. 32.

    Haygood, R., Babbitt, C. C., Fedrigo, O. & Wray, G. A. Contrasts between adaptive coding and noncoding changes during human evolution. Proc. Natl. Acad. Sci USA 107, 7853–7857 (2010).

    ADS  CAS  PubMed  Article  Google Scholar 

  33. 33.

    Rolán-Alvarez, E. et al. Nonallopatric and parallel origin of local reproductive barriers between two snail ecotypes. Mol. Ecol. 13, 3415–24 (2004).

    PubMed  Article  Google Scholar 

  34. 34.

    Panova, M., Hollander, J. & Johannesson, K. Site-specific genetic divergence in parallel hybrid zones suggests non-allopatric evolution of reproductive barriers. Mol. Ecol. 15, 4021–4031 (2006).

    CAS  PubMed  Article  Google Scholar 

  35. 35.

    Butlin, R. K. et al. Parallel evolution of local adaptation and reproductive isolation in the face of gene flow. Evolution 68, 935–49 (2014).

    PubMed  Article  Google Scholar 

  36. 36.

    Rolán-Alvarez, E., Austin, C. & Boulding, E. G. The contribution of Littorina to the field of Evolutionary Ecology. Oceanography and Marine Biology, an Annual Review 53, 157–214 (2015).

    Google Scholar 

  37. 37.

    Johannesson, K., Rolán-Alvarez, E. & Ekendahl, A. Incipient reproductive isolation between two sympatric morphs of the intertidal snail Littorina saxatilis. Evolution 49, 1180–1190 (1995).

    PubMed  Article  Google Scholar 

  38. 38.

    Johannesson, K. et al. Repeated evolution of reproductive isolation in a marine snail: unveiling mechanisms of speciation. Philos. Trans. R. Soc. Lond. B Biol. Sci. 365, 1735–1747 (2010).

    PubMed  PubMed Central  Article  Google Scholar 

  39. 39.

    Rolán-Alvarez, E., Johannesson, K. & Erlandsson, J. The maintenance of a cline in the marine snail Littorina saxatilis: the role of home site advantage and hybrid fitness. Evolution 51, 1838–1847 (1997).

    PubMed  Article  Google Scholar 

  40. 40.

    Quesada, H., Posada, D., Caballero, A., Morán, P. & Rolán-Alvarez, E. Phylogenetic evidence for multiple sympatric ecological diversification in a marine snail. Evolution 61, 1600–1612 (2007).

    PubMed  Article  Google Scholar 

  41. 41.

    Pérez-Pereira, N., Quesada, H. & Caballero, A. Can parallel ecological speciation be detected with phylogenetic analyses? Mol. Phylogenet. Evol. 166, 149–156 (2017).

    Article  Google Scholar 

  42. 42.

    Westram, A. M. et al. Do the same genes underlie parallel phenotypic divergence in different Littorina saxatilis populations? Mol. Ecol. 23, 4603–4616 (2014).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  43. 43.

    Westram, A. M., Panova, M., Galindo, J. & Butlin, R. K. Targeted re-sequencing reveals geographic patterns of differentiation for loci implicated in parallel evolution. Mol. Ecol. 25, 3169–3186 (2016).

    PubMed  Article  Google Scholar 

  44. 44.

    Ravinet, M. et al. Shared and nonshared genomic divergence in parallel ecotypes of Littorina saxatilis at a local scale. Mol. Ecol. 25, 287–305 (2016).

    PubMed  Article  Google Scholar 

  45. 45.

    Goodwin, S., McPherson, J. D. & McCombie, R. C. Coming of age: ten years of next-generation sequencing technologies. Nat. Rev. Genet. 17, 333–351 (2016).

    CAS  PubMed  Article  Google Scholar 

  46. 46.

    Gresham, D., Dunham, M. J. & Botstein, D. Comparing whole genomes using DNA microarrays. Nat. Rev. Genet. 9, 291–302 (2008).

    CAS  PubMed  Article  Google Scholar 

  47. 47.

    Renn, S. C. P. et al. Using comparative genomic hybridization to survey genomic sequence divergence across species: a proof-of-concept from Drosophila. BMC Genomics 11, 271 (2010).

    PubMed  PubMed Central  Article  Google Scholar 

  48. 48.

    Bottomly, D. et al. Evaluating gene expression in C57BL/6J and DBA/2J mouse striatum using RNA-seq and microarrays. Plos One 6, e17820 (2011).

    ADS  CAS  PubMed  PubMed Central  Article  Google Scholar 

  49. 49.

    Sirbu, A., Kerr, G., Crane, M. & Ruskin, H. J. RNA-seq vs dual- and single-channel microarray data: sensitivity analysis for differential expression and clustering. Plos One 7, e50986 (2012).

    ADS  CAS  PubMed  PubMed Central  Article  Google Scholar 

  50. 50.

    Robinson, D. G., Wang, J. Y. & Storey, J. D. A nested parallel experiment demonstrates differences in intensity-dependence between RNA-seq and microarrays. Nucleic. Acids. Res. 43, e131 (2015).

    PubMed  PubMed Central  Article  Google Scholar 

  51. 51.

    Hoen, P. A. et al. Deep sequencing-based expression analysis shows major advances in robustness, resolution and inter-lab portability over five microarray platforms. Nucleic. Acids. Res. 36, e141 (2008).

    PubMed  PubMed Central  Article  Google Scholar 

  52. 52.

    Zhao, S., Fung-Leung, W.-P., Bittner, A., Ngo, K. & Liu, X. Comparison of RNA-Seq and microarray in transcriptomic profiling of activated T cells. Plos One 9, e78644 (2014).

    ADS  PubMed  PubMed Central  Article  Google Scholar 

  53. 53.

    Willenbrock, H. et al. Quantitative miRNA expression analysis: comparing microarrays with next-generation sequencing. RNA 15, 2028–2034 (2009).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  54. 54.

    McIntyre, L. M. et al. RNA-seq: technical variability and sampling. BMC Genomics 12, 293 (2011).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  55. 55.

    Tirado, T., Saura, M., Rolán-Alvarez, E. & Quesada, H. Historical biogeography of the marine snail Littorina saxatilis inferred from haplotype and Shell morphology evolution in NW Spain. Plos One, 11, e0161287, https://doi.org/10.1371/journal.pone.0161287 (2016).

    Article  Google Scholar 

  56. 56.

    Diz, A. P., Páez de la Cadena, M. & Rolán-Alvarez, E. Proteomic evidence of a paedomorphic evolutionary process within a marine snail species: a strategy for adapting to extreme ecological conditions? J. Evol. Biol. 25, 2569–2581 (2012).

    CAS  PubMed  Article  Google Scholar 

  57. 57.

    Wilding, C. S., Butlin, R. K. & Grahame, J. Differential gene exchange between parapatric morphs of Littorina saxatilis detected using AFLP markers. J. Evol. Biol. 14, 611–619 (2001).

    CAS  Article  Google Scholar 

  58. 58.

    Panova, M. et al. Species and gene divergence in Littorina snails detected by array comparative genomic hybridization. BMC Genomics 15, 687 (2014).

    PubMed  PubMed Central  Article  Google Scholar 

  59. 59.

    Canbäck, B. et al. The Littorina sequence database (LSD) – an online resource for genomic data. Mol. Ecol. Res. 12, 142–148 (2012).

    Article  Google Scholar 

  60. 60.

    Carvalho, B. S. & Irizarry, R. A. A framework for oligonucleotide microarray preprocessing. Bioinformatics 26, 2363–2367 (2010).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  61. 61.

    Irizarry, R. A. et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4, 249–264 (2003).

    MATH  Article  Google Scholar 

  62. 62.

    Stolc, V., Samanta, M. P., Tongprasit, W. & Marshall, W. F. Genome-wide transcriptional analysis of flagellar regeneration in Chlamydomonas reinhardtii identifies orthologs of ciliary disease genes. Proc. Natl. Acad. Sci. USA 102, 3703–3707 (2005).

    ADS  CAS  PubMed  Article  Google Scholar 

  63. 63.

    Toedling, J. et al. Ringo - an R/Bioconductor package for analyzing ChIP-chip readouts. BMC Bioinformatics 8, p. 221 (2007).

    PubMed  PubMed Central  Article  Google Scholar 

  64. 64.

    Ritchie, M. E. et al. A comparison of background correction methods for two-color microarrays. Bioinformatics 23, 2700–2707 (2007).

    CAS  Article  Google Scholar 

  65. 65.

    Yang, Y. H. & Thome, N. P. Normalization for two-color cDNA microarray data. Lect. Notes Monogr. Ser. 40, 403–418 (2003).

    MathSciNet  Article  Google Scholar 

  66. 66.

    Smyth, G.K. Limma: Linear Models for Microarray Data in Bioinformatics and Computational Biology Solutions using R and Bioconductor (eds Gentleman, R., Carey, V. J., Dudoit, S., Irizarry, R. & Huber, W.) 397–420 (Springer, New York, 2005).

  67. 67.

    Carvajal-Rodríguez, A., De Uña-Alvarez, J. & Rolán-Alvarez, E. A new multitest correction (SGoF) that increases its statistical power when increasing the number of tests. BMC Bioinformatics 10, 209 (2009).

    PubMed  PubMed Central  Article  Google Scholar 

  68. 68.

    Manly, B. F. J. Randomization, bootstrap and Monte Carlo methods in biology, 3rd edn. (Chapman & Hall, London, 2006).

  69. 69.

    Derome, N., Duchesne, P. & Bernatchez, L. Parallelism in gene transcription among sympatric lake whitefish (Coregonus clupeaformis Mitchill) ecotypes. Mol. Ecol. 15, 1239–1249 (2006).

    CAS  PubMed  Article  Google Scholar 

  70. 70.

    Draghici, S. Data Analysis Tools for DNA Microarrays (Chapman & Hall, London, 2003).

  71. 71.

    Conesa, A. et al. Blast2GO: A universal tool for annotation, visualization and analysis in functional genomics research. Bioinformatics 21, 3674–3676 (2005).

    CAS  Article  Google Scholar 

  72. 72.

    Sarashina, I. et al. Molecular evolution and functionally important structures of molluscan Dermatopontin: Implications for the origins of molluscan shell matrix proteins. J. Mol. Evol. 62, 307–318l (2006).

    ADS  CAS  PubMed  Article  Google Scholar 

  73. 73.

    Grégoire, C. Structure of the molluscan shell in Chemical Zoology IV (eds Florkin, M. & Sheer, B. T.) 45–102 (Academic Press, NY, 1972).

  74. 74.

    Nosil, P. Ecological speciation. Oxford University Press, Oxford. (2012).

  75. 75.

    Nielsen, R. Molecular signatures of natural selection. Ann. Rev. Genet. 39, 197–218 (2005).

    CAS  PubMed  Article  Google Scholar 

  76. 76.

    Falconer, D. S. Introduction to quantitative genetics. Longman, Harlow, 438 pp. (1989).

  77. 77.

    Ekblom, R. & Galindo, J. Applications of next generation sequencing in molecular ecology of non-model organisms. Heredity 107, 1–15 (2011).

    CAS  PubMed  Article  Google Scholar 

  78. 78.

    García, C., Avila, V., Quesada, H. & Caballero, A. Candidate transcriptome sources of inbreeding depression in Drosophila melanogaster. Plos One 8, e70067 (2013).

    ADS  PubMed  PubMed Central  Article  Google Scholar 

  79. 79.

    Edelman, G. M. & Gally, J. A. Degeneracy and complexity in biological systems. Proc. Natl. Acad. Sci. USA 98, 13763–13768 (2001).

    ADS  CAS  PubMed  Article  Google Scholar 

  80. 80.

    Renaut, S., Owens, G. L. & Riesenberg, L. H. Shared selective pressure and local genomic landscape lead to repeatable patterns of genomic divergence in sunflowers. Mol. Ecol. 23, 311–324 (2014).

    CAS  PubMed  Article  Google Scholar 

  81. 81.

    Moczek, A. P. & Nagy, L. M. Diverse developmental mechanisms contribute to different levels of diversity in horned beetles. Evol. Dev. 7, 175–185 (2005).

    PubMed  Article  Google Scholar 

  82. 82.

    Steiner, C. C., Römpler, H., Boettger, L. M., Schöneberg, T. & Hoekstra, H. E. The genetic basis of phenotypic convergence in beach mice: similar pigment patterns but different genes. Mol. Biol. Evol. 26, 35–45 (2009).

    CAS  PubMed  Article  Google Scholar 

  83. 83.

    Ralph, P. L. & Coop, G. Convergent evolution during local adaptation to patchy landscapes. Plos Genet. 11, e1005630 (2015).

    PubMed  PubMed Central  Article  Google Scholar 

  84. 84.

    Hendry, A. P. Eco-evolutionary dynamics. Princeton University Press (2016).

  85. 85.

    Renaut, S., Nolte, A. W., Rogers, S. M., Derome, N. & Bernatchez, L. SNP signatures of selection on standing genetic variation and their association with adaptive phenotypes along gradients of ecological speciation in lake whitefish species pairs (Coregonus spp.). Mol. Ecol. 20, 545–559 (2010).

    PubMed  Article  Google Scholar 

  86. 86.

    Deagle, B. E. et al. Population genomics of parallel phenotypic evolution in stickleback across stream-lake ecological transitions. Proc. R. Soc Lond. B Biol. Sci. 279, 1277–1286 (2012).

    CAS  Article  Google Scholar 

  87. 87.

    St-Cyr, J., Derome, N. & Bernatchez, L. The transcriptomics of life-history trade-offs in whitefish pairs (Coregonus sp.). Mol. Ecol. 17, 1850–1870 (2008).

    CAS  PubMed  Article  Google Scholar 

  88. 88.

    Zhen, Y., Aardema, M. L., Medina, E. M., Schumer, M. & Andolfatto, P. Parallel molecular evolution in an herbivore community. Science 28, 1634–1637 (2012).

    ADS  Article  Google Scholar 

  89. 89.

    Eisen, M. B. & Brown, P. O. DNA arrays for analysis of gene expression. Methods Enzymol. 303, 179–205 (1999).

    CAS  PubMed  Article  Google Scholar 

  90. 90.

    Lemos, B., Meiklejohn, C. D., Caceres, M. & Hartl, D. L. Rates of divergence in gene expression profiles of primates, mice, and flies: Stabilizing selection and variability among functional categories. Evolution 59, 126–137 (2005).

    CAS  PubMed  Article  Google Scholar 

  91. 91.

    Wagner, A. Decoupled evolution of coding region and mRNA expression patterns after gene duplication: implications for the neutralist-selectionist debate. Proc. Natl. Acad. Sci. USA 97, 6579–6584 (2000).

    ADS  CAS  PubMed  Article  Google Scholar 

  92. 92.

    Kohn, M. H., Shapiro, J. & Wu, C. I. Decoupled differentiation of gene expression and coding sequence among Drosophila populations. Genes Genet. Syst. 83, 265–273 (2008).

    CAS  PubMed  Article  Google Scholar 

  93. 93.

    Kozak, G. M., Brennan, R. S., Berdan, E. L., Fuller, R. C. & Whitehead, A. Functional and population genomic divergence within and between two species of killifish adapted to different osmotic niches. Evolution 68, 63–80 (2013).

    PubMed  Article  Google Scholar 

  94. 94.

    Akashi, H. Inferring weak selection from patterns of polymorphism and divergence at “silent” sites in drosophila DNA. Genetics 139, 1067–1076 (1995).

    CAS  PubMed  PubMed Central  Google Scholar 

  95. 95.

    Chamary, J., Parmley, J. L. & Hurst, L. D. Hearing silence: Non-neutral evolution at synonymous sites in mammals. Nat. Rev. Genet. 7, 98–108 (2006).

    CAS  Article  Google Scholar 

  96. 96.

    Eöry, L., Halligan, D. L. & Keightley, P. D. Distributions of selectively constrained sites and deleterious mutation rates in the hominid and murid genomes. Mol. Biol. Evol. 27, 177–192 (2010).

    PubMed  Article  Google Scholar 

  97. 97.

    Zhou, T., Gu, W. & Wilke, C. O. Detecting positive and purifying selection at synonymous sites in yeast and worm. Mol. Biol. Evol. 27, 1912–1922 (2010).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  98. 98.

    Künstner, A., Nabholz, B. & Ellegren, H. Significant selective constraint at 4-fold degenerate sites in the avian genome and its consequence for detection of positive selection. Genome Biol. Evol. 3, 1381 (2011).

    PubMed  Article  Google Scholar 

  99. 99.

    Lawrie, D. S., Messer, P. W., Hershberg, R. & Petrov, D. A. Strong purifying selection at synonymous sites in D. melanogaster. Plos Genet. 9, e1003527 (2013).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  100. 100.

    Stern, D. L. & Orgogozo, V. The loci of evolution: how predictable is genetic evolution. Evolution 62, 2155–2177 (2008).

    PubMed  PubMed Central  Article  Google Scholar 

Download references


We would like to thank the ECIMAT Marine Reseach Center (University of Vigo) for providing marine laboratory facilities. We are greateful to Pierre Duchesne for extending from two to three localities the algorithm for calculating the probability that the observed parallelism could be due to chance alone and help in calculating the corresponding p-values. This work was supported by Ministerio de Economía y Competitividad (codes BFU2013-44635-P, CGL2016-75482-P and CGL2016-75904-C2-1), Axudas do programa de consolidación e estruturación de unidades de investigacións competitivas do SUG, Xunta de Galicia (ED431C 2016-037), Fondos Feder: “Unha maneira de facer Europa”, Xunta de Galicia (INCITE09 310 006 PR) and the Swedish Research Councils VR and Formas (Linnaeus grant Formas 217-2008-1719).

Author information




H.Q., K.J. and E.R.-A. conceived the study. H.Q. wrote the manuscript, with input from all authors. M.J.R. and M.S. conducted the experiments. M.J.R. and A.P.-F. performed the bioinformatic analyses with advises and support from H.Q., E.R.-A. and A.C. M.P., C.A. and T.J. designed the microarray and assisted in data analyses.

Corresponding author

Correspondence to Humberto Quesada.

Ethics declarations

Competing Interests

The authors declare no competing interests.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Rivas, M.J., Saura, M., Pérez-Figueroa, A. et al. Population genomics of parallel evolution in gene expression and gene sequence during ecological adaptation. Sci Rep 8, 16147 (2018). https://doi.org/10.1038/s41598-018-33897-8

Download citation


  • Ecotypes
  • Parallel Changes
  • Nondirectional Changes
  • Empirical Bayes Adjustment
  • NimbleScan

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing