Genetic architecture of a pollinator shift and its fate in secondary hybrid zones of two Petunia species.

BACKGROUND: Theory suggests that the genetic architecture of traits under divergent natural selection influences how easily reproductive barriers evolve and are maintained between species. Divergently selected traits with a simple genetic architecture (few loci with major phenotypic effects) should facilitate the establishment and maintenance of reproductive isolation between species that are still connected by some gene flow. While empirical support for this idea appears to be mixed, most studies test the influence of trait architectures on reproductive isolation only indirectly. Petunia plant species are, in part, reproductively isolated by their different pollinators. To investigate the genetic causes and consequences of this ecological isolation, we deciphered the genetic architecture of three floral pollination syndrome traits in naturally occurring hybrids between the widespread Petunia axillaris and the highly endemic and endangered P. exserta. RESULTS: Using population genetics, Bayesian linear mixed modelling and genome-wide association studies, we found that the three pollination syndrome traits vary in genetic architecture. Few genome regions explain a majority of the variation in flavonol content (defining UV floral colour) and strongly predict the trait value in hybrids irrespective of interspecific admixture in the rest of their genomes. In contrast, variation in pistil exsertion and anthocyanin content (defining visible floral colour) is controlled by many genome-wide loci. Opposite to flavonol content, the genome-wide proportion of admixture between the two species predicts trait values in their hybrids. Finally, the genome regions strongly associated with the traits do not show extreme divergence between individuals representing the two species, suggesting that divergent selection on these genome regions is relatively weak within their contact zones. CONCLUSIONS: Among the traits analysed, those with a more complex genetic architecture are best maintained in association with the species upon their secondary contact. We propose that this maintained genotype-phenotype association is a coincidental consequence of the complex genetic architectures of these traits: some of their many underlying small-effect loci are likely to be coincidentally linked with the actual barrier loci keeping these species partially isolated upon secondary contact. Hence, the genetic architecture of a trait seems to matter for the outcome of hybridization not only then when the trait itself is under selection.

Species divergence under competition and shared predation.

Species competing for resources also commonly share predators. While competition often drives divergence between species, the effects of shared predation are less understood. Theoretically, competing prey species could either diverge or evolve in the same direction under shared predation depending on the strength and symmetry of their interactions. We took an empirical approach to this question, comparing antipredator and trophic phenotypes between sympatric and allopatric populations of threespine stickleback and prickly sculpin fish that all live in the presence of a trout predator. We found divergence in antipredator traits between the species: in sympatry, antipredator adaptations were relatively increased in stickleback but decreased in sculpin. Shifts in feeding morphology, diet and habitat use were also divergent but driven primarily by stickleback evolution. Our results suggest that asymmetric ecological character displacement indirectly made stickleback more and sculpin less vulnerable to shared predation, driving divergence of antipredator traits between sympatric species.

Chromosomal Fusions Facilitate Adaptation to Divergent Environments in Threespine Stickleback.

Chromosomal fusions are hypothesized to facilitate adaptation to divergent environments, both by bringing together previously unlinked adaptive alleles and by creating regions of low recombination that facilitate the linkage of adaptive alleles; but, there is little empirical evidence to support this hypothesis. Here, we address this knowledge gap by studying threespine stickleback (Gasterosteus aculeatus), in which ancestral marine fish have repeatedly adapted to freshwater across the northern hemisphere. By comparing the threespine and ninespine stickleback (Pungitius pungitius) genomes to a de novo assembly of the fourspine stickleback (Apeltes quadracus) and an outgroup species, we find two chromosomal fusion events involving the same chromosomes have occurred independently in the threespine and ninespine stickleback lineages. On the fused chromosomes in threespine stickleback, we find an enrichment of quantitative trait loci underlying traits that contribute to marine versus freshwater adaptation. By comparing whole-genome sequences of freshwater and marine threespine stickleback populations, we also find an enrichment of regions under divergent selection on these two fused chromosomes. There is elevated genetic diversity within regions under selection in the freshwater population, consistent with a simulation study showing that gene flow can increase diversity in genomic regions associated with local adaptation and our demographic models showing gene flow between the marine and freshwater populations. Integrating our results with previous studies, we propose that these fusions created regions of low recombination that enabled the formation of adaptative clusters, thereby facilitating freshwater adaptation in the face of recurrent gene flow between marine and freshwater threespine sticklebacks.

Chromosomal inversions can limit adaptation to new environments.

Chromosomal inversions are often thought to facilitate local adaptation and population divergence because they can link multiple adaptive alleles into non-recombining genomic blocks. Selection should thus be more efficient in driving inversion-linked adaptive alleles to high frequency in a population, particularly in the face of maladaptive gene flow. But what if ecological conditions and hence selection on inversion-linked alleles change? Reduced recombination within inversions could then constrain the formation of optimal combinations of pre-existing alleles under these new ecological conditions. Here, we outline this idea of inversions limiting adaptation and divergence when ecological conditions change across time or space. We reason and use simulations to illustrate that the benefit of inversions for local adaptation and divergence under one set of ecological conditions can come with a concomitant constraint for adaptation to novel sets of ecological conditions. This limitation of inversions to adaptation may contribute to the maintenance of polymorphism within species.

Widespread intersex differentiation across the stickleback genome - The signature of sexually antagonistic selection?

Females and males within a species commonly have distinct reproductive roles, and the associated traits may be under perpetual divergent natural selection between the sexes if their sex-specific control has not yet evolved. Here, we explore whether such sexually antagonistic selection can be detected based on the magnitude of differentiation between the sexes across genome-wide genetic polymorphisms by whole-genome sequencing of large pools of female and male threespine stickleback fish. We find numerous autosomal genome regions exhibiting intersex allele frequency differences beyond the range plausible under pure sampling stochasticity. Alternative sequence alignment strategies rule out that these high-differentiation regions represent sex chromosome segments misassembled into the autosomes. Instead, comparing allele frequencies and sequence read depth between the sexes reveals that regions of high intersex differentiation arise because autosomal chromosome segments got copied into the male-specific sex chromosome (Y), where they acquired new mutations. Because the Y chromosome is missing in the stickleback reference genome, sequence reads derived from DNA copies on the Y chromosome still align to the original homologous regions on the autosomes. We argue that this phenomenon hampers the identification of sexually antagonistic selection within a genome, and can lead to spurious conclusions from population genomic analyses when the underlying samples differ in sex ratios. Because the hemizygous sex chromosome sequence (Y or W) is not represented in most reference genomes, these problems may apply broadly.

Phylogenomics of an extra-Antarctic notothenioid radiation reveals a previously unrecognized lineage and diffuse species boundaries.

BACKGROUND: The impressive adaptive radiation of notothenioid fishes in Antarctic waters is generally thought to have been facilitated by an evolutionary key innovation, antifreeze glycoproteins, permitting the rapid evolution of more than 120 species subsequent to the Antarctic glaciation. By way of contrast, the second-most species-rich notothenioid genus, Patagonotothen, which is nested within the Antarctic clade of Notothenioidei, is almost exclusively found in the non-Antarctic waters of Patagonia. While the drivers of the diversification of Patagonotothen are currently unknown, they are unlikely to be related to antifreeze glycoproteins, given that water temperatures in Patagonia are well above freezing point. Here we performed a phylogenetic analysis based on genome-wide single nucleotide polymorphisms (SNPs) derived from restriction site-associated DNA sequencing (RADseq) in a total of twelve Patagonotothen species. RESULTS: We present a well-supported, time-calibrated phylogenetic hypothesis including closely and distantly related outgroups, confirming the monophyly of the genus Patagonotothen with an origin approximately 3 million years ago and the paraphyly of both the sister genus Lepidonotothen and the family Notothenidae. Our phylogenomic and population genetic analyses highlight a previously unrecognized linage and provide evidence for shared genetic variation between some closely related species. We also provide a mitochondrial phylogeny showing mitonuclear discordance. CONCLUSIONS: Based on a combination of phylogenomic and population genomic approaches, we provide evidence for the existence of a new, potentially cryptic, Patagonotothen species, and demonstrate that genetic boundaries between some closely related species are diffuse, likely due to recent introgression and/or incomplete linage sorting. The detected mitonuclear discordance highlights the limitations of relying on a single locus for species barcoding. In addition, our time-calibrated phylogenetic hypothesis shows that the early burst of diversification roughly coincides with the onset of the intensification of Quaternary glacial cycles and that the rate of species accumulation may have been stepwise rather than constant. Our phylogenetic framework not only advances our understanding of the origin of a high-latitude marine radiation, but also provides the basis for the study of the ecology and life history of the genus Patagonotothen, as well as for their conservation and commercial management.

A functional trade-off between trophic adaptation and parental care predicts sexual dimorphism in cichlid fish.

Although sexual dimorphism is widespread in nature, its evolutionary causes often remain elusive. Here we report a case where a sex-specific conflicting functional demand related to parental care, but not to sexual selection, explains sexual dimorphism in a primarily trophic structure, the gill rakers of cichlid fishes. More specifically, we examined gill raker length in a representative set of cichlid fish species from Lake Tanganyika featuring three different parental care strategies: (i) uni-parental mouthbrooding, whereby only one parental sex incubates the eggs in the buccal cavity; (ii) bi-parental mouthbrooding, whereby both parents participate in mouthbrooding; and (iii) nest guarding without any mouthbrooding involved. As predicted from these different parental care strategies, we find sexual dimorphism in gill raker length to be present only in uni-parental mouthbrooders, but not in bi-parental mouthbrooders nor in nest guarders. Moreover, variation in the extent of sexual dimorphism among uni-parental mouthbrooders appears to be related to trophic ecology. Overall, we present a previously unrecognized scenario for the evolution of sexual dimorphism that is not related to sexual selection or initial niche divergence between sexes. Instead, sexual dimorphism in gill raker length in uni-parental mouthbrooding cichlid fish appears to be the consequence of a sex-specific functional trade-off between a trophic function present in both sexes and a reproductive function present only in the brooding sex.

Meta-analysis of chromosome-scale crossover rate variation in eukaryotes and its significance to evolutionary genomics.

Understanding the distribution of crossovers along chromosomes is crucial to evolutionary genomics because the crossover rate determines how strongly a genome region is influenced by natural selection on linked sites. Nevertheless, generalities in the chromosome-scale distribution of crossovers have not been investigated formally. We fill this gap by synthesizing joint information on genetic and physical maps across 62 animal, plant and fungal species. Our quantitative analysis reveals a strong and taxonomically widespread reduction of the crossover rate in the centre of chromosomes relative to their peripheries. We demonstrate that this pattern is poorly explained by the position of the centromere, but find that the magnitude of the relative reduction in the crossover rate in chromosome centres increases with chromosome length. That is, long chromosomes often display a dramatically low crossover rate in their centre, whereas short chromosomes exhibit a relatively homogeneous crossover rate. This observation is compatible with a model in which crossover is initiated from the chromosome tips, an idea with preliminary support from mechanistic investigations of meiotic recombination. Consequently, we show that organisms achieve a higher genome-wide crossover rate by evolving smaller chromosomes. Summarizing theory and providing empirical examples, we finally highlight that taxonomically widespread and systematic heterogeneity in crossover rate along chromosomes generates predictable broad-scale trends in genetic diversity and population differentiation by modifying the impact of natural selection among regions within a genome. We conclude by emphasizing that chromosome-scale heterogeneity in crossover rate should urgently be incorporated into analytical tools in evolutionary genomics, and in the interpretation of resulting patterns.

Genomics of adaptive divergence with chromosome-scale heterogeneity in crossover rate.

Genetic differentiation between divergent populations is often greater in chromosome centres than peripheries. Commonly overlooked, this broadscale differentiation pattern is sometimes ascribed to heterogeneity in crossover rate and hence linked selection within chromosomes, but the underlying mechanisms remain incompletely understood. A literature survey across 46 organisms reveals that most eukaryotes indeed exhibit a reduced crossover rate in chromosome centres relative to the peripheries. Using simulations of populations diverging into ecologically different habitats through sorting of standing genetic variation, we demonstrate that such chromosome-scale heterogeneity in crossover rate, combined with polygenic divergent selection, causes stronger hitchhiking and especially barriers to gene flow across chromosome centres. Without requiring selection on new mutations, this rapidly leads to elevated population differentiation in the low-crossover centres relative to the high-crossover peripheries of chromosomes ("Chromosome Centre-Biased Differentiation", CCBD). Using simulated and empirical data, we then show that strong CCBD between populations can provide evidence of polygenic adaptive divergence with a phase of gene flow. We further demonstrate that chromosome-scale heterogeneity in crossover rate impacts analyses beyond that of population differentiation, including the inference of phylogenies and parallel adaptive evolution among populations, the detection of genetic loci under selection, and the interpretation of the strength of selection on genomic regions. Overall, our results call for a greater appreciation of chromosome-scale heterogeneity in crossover rate in evolutionary genomics.

Demography and genome divergence of lake and stream populations of an East African cichlid fish.

Disentangling the processes and mechanisms underlying adaptive diversification is facilitated by the comparative study of replicate population pairs that have diverged along a similar environmental gradient. Such a setting is realized in a cichlid fish from southern Lake Tanganyika, Astatotilapia burtoni, which occurs within the lake proper as well as in various affluent rivers. Previously, we demonstrated that independent lake and stream populations show similar adaptations to the two habitat regimes. However, little is known about the evolutionary and demographic history of the A. burtoni populations in question and the patterns of genome divergence among them. Here, we apply restriction site-associated DNA sequencing (RADseq) to examine the evolutionary history, the population structure and genomic differentiation of lake and stream populations in A. burtoni. A phylogenetic reconstruction based on genome-wide molecular data largely resolved the evolutionary relationships among populations, allowing us to re-evaluate the independence of replicate lake-stream population clusters. Further, we detected a strong pattern of isolation by distance, with baseline genomic divergence increasing with geographic distance and decreasing with the level of gene flow between lake and stream populations. Genome divergence patterns were heterogeneous and inconsistent among lake-stream population clusters, which is explained by differences in divergence times, levels of gene flow and local selection regimes. In line with the latter, we only detected consistent outlier loci when the most divergent lake-stream population pair was excluded. Several of the thus identified candidate genes have inferred functions in immune and neuronal systems and show differences in gene expression between lake and stream populations.

A high-density genetic map reveals variation in recombination rate across the genome of Daphnia magna.

BACKGROUND: Recombination rate is an essential parameter for many genetic analyses. Recombination rates are highly variable across species, populations, individuals and different genomic regions. Due to the profound influence that recombination can have on intraspecific diversity and interspecific divergence, characterization of recombination rate variation emerges as a key resource for population genomic studies and emphasises the importance of high-density genetic maps as tools for studying genome biology. Here we present such a high-density genetic map for Daphnia magna, and analyse patterns of recombination rate across the genome. RESULTS: A F2 intercross panel was genotyped by Restriction-site Associated DNA sequencing to construct the third-generation linkage map of D. magna. The resulting high-density map included 4037 markers covering 813 scaffolds and contigs that sum up to 77 % of the currently available genome draft sequence (v2.4) and 55 % of the estimated genome size (238 Mb). Total genetic length of the map presented here is 1614.5 cM and the genome-wide recombination rate is estimated to 6.78 cM/Mb. Merging genetic and physical information we consistently found that recombination rate estimates are high towards the peripheral parts of the chromosomes, while chromosome centres, harbouring centromeres in D. magna, show very low recombination rate estimates. CONCLUSIONS: Due to its high-density, the third-generation linkage map for D. magna can be coupled with the draft genome assembly, providing an essential tool for genome investigation in this model organism. Thus, our linkage map can be used for the on-going improvements of the genome assembly, but more importantly, it has enabled us to characterize variation in recombination rate across the genome of D. magna for the first time. These new insights can provide a valuable assistance in future studies of the genome evolution, mapping of quantitative traits and population genetic studies.

Testing the stages model in the adaptive radiation of cichlid fishes in East African Lake Tanganyika.

Adaptive radiation (AR) is a key process in the origin of organismal diversity. However, the evolution of trait disparity in connection with ecological specialization is still poorly understood. Available models for vertebrate ARs predict that diversification occurs in the form of temporal stages driven by different selective forces. Here, we investigate the AR of cichlid fishes in East African Lake Tanganyika and use macroevolutionary model fitting to evaluate whether diversification happened in temporal stages. Six trait complexes, for which we also provide evidence of their adaptiveness, are analysed with comparative methods: body shape, pharyngeal jaw shape, gill raker traits, gut length, brain weight and body coloration. Overall, we do not find strong evidence for the 'stages model' of AR. However, our results suggest that trophic traits diversify earlier than traits implicated in macrohabitat adaptation and that sexual communication traits (i.e. coloration) diversify late in the radiation.

The genomic signature of parallel adaptation from shared genetic variation.

Parallel adaptation is common and may often occur from shared genetic variation, but the genomic consequences of this process remain poorly understood. We first use individual-based simulations to demonstrate that comparisons between populations adapted in parallel to similar environments from shared variation reveal a characteristic genomic signature around a selected locus: a low-divergence valley centred at the locus and flanked by twin peaks of high divergence. This signature is initiated by the hitchhiking of haplotype tracts differing between derived populations in the broader neighbourhood of the selected locus (driving the high-divergence twin peaks) and shared haplotype tracts in the tight neighbourhood of the locus (driving the low-divergence valley). This initial hitchhiking signature is reinforced over time because the selected locus acts as a barrier to gene flow from the source to the derived populations, thus promoting divergence by drift in its close neighbourhood. We next empirically confirm the peak-valley-peak signature by combining targeted and RAD sequence data at three candidate adaptation genes in multiple marine (source) and freshwater (derived) populations of threespine stickleback. Finally, we use a genome-wide screen for the peak-valley-peak signature to discover additional genome regions involved in parallel marine-freshwater divergence. Our findings offer a new explanation for heterogeneous genomic divergence and thus challenge the standard view that peaks in population divergence harbour divergently selected loci and that low-divergence regions result from balancing selection or localized introgression. We anticipate that genome scans for peak-valley-peak divergence signatures will promote the discovery of adaptation genes in other organisms.

Predictability, an Orrery, and a Speciation Machine: Quest for a Standard Model of Speciation.

Accurate predictions are commonly taken as a hallmark of strong scientific understanding. Yet, we do not seem capable today of making many accurate predictions about biological speciation. Why? What limits predictability in general, what exactly is the function and value of predictions, and how might we go about predicting new species? Inspired by an orrery used to explain solar eclipses, we address these questions with a thought experiment in which we conceive an evolutionary speciation machine generating new species. This experiment highlights complexity, chance, and speciation pluralism as the three fundamental challenges for predicting speciation. It also illustrates the methodological value of predictions in testing and improving conceptual models. We then outline how we might move from the hypothetical speciation machine to a predictive standard model of speciation. Operationalizing, testing, and refining this model will require a concerted shift to large-scale, integrative, and interdisciplinary efforts across the tree of life. This endeavor, paired with technological advances, may reveal apparently stochastic processes to be deterministic, and promises to expand the breadth and depth of our understanding of speciation and more generally, of evolution.

Recombination in the threespine stickleback genome--patterns and consequences.

Heterogeneity in recombination rate may strongly influence genome evolution and entail methodological challenges to genomic investigation. Nevertheless, a solid understanding of these issues awaits detailed information across a broad range of taxa. Based on 282 F(2) individuals and 1872 single nucleotide polymorphisms, we characterize recombination in the threespine stickleback fish genome. We find an average genome-wide recombination rate of 3.11 cm/Mb. Crossover frequencies are dramatically elevated in the chromosome peripheries as compared to the centres, and are consistent with one obligate crossover per chromosome (but not chromosome arm). Along the sex chromosome, we show that recombination is restricted to a small pseudoautosomal domain of c. 2 Mb, spanning only 10% of that chromosome. Comparing female to male RAD sequence coverage allows us to identify two discrete levels of degeneration on the Y chromosome, one of these 'evolutionary strata' coinciding with a previously inferred inverted region. Using polymorphism data from two young (<10 000 years old) ecologically diverged lake-stream population pairs, we demonstrate that recombination rate correlates with both the magnitude of allele frequency shifts between populations and levels of genetic diversity within populations. These associations reflect genome-wide heterogeneity in the influence of selection on linked sites. We further find a strong relationship between recombination rate and GC content, possibly driven by GC-biased gene conversion. Overall, we highlight that heterogeneity in recombination rate has profound consequences on genome evolution and deserves wider recognition in marker-based genomic analyses.

Repeated genetic divergence plays a minor role in repeated phenotypic divergence of lake-stream stickleback.

Recent studies have shown that the repeated evolution of similar phenotypes in response to similar ecological conditions (here "parallel evolution") often occurs through mutations in the same genes. However, many previous studies have focused on known candidate genes in a limited number of systems. Thus, the question of how often parallel phenotypic evolution is due to parallel genetic changes remains open. Here, we used quantitative trait locus (QTL) mapping in F2 intercrosses between lake and stream threespine stickleback (Gasterosteus aculeatus) from four independent watersheds on Vancouver Island, Canada to determine whether the same QTL underlie divergence in the same phenotypes across, between, and within watersheds. We find few parallel QTL, even in independent crosses from the same watershed or for phenotypes that have diverged in parallel. These findings suggest that different mutations can lead to similar phenotypes. The low genetic repeatability observed in these lake-stream systems contrasts with the higher genetic repeatability observed in other stickleback systems. We speculate that differences in evolutionary history, gene flow, and/or the strength and direction of selection might explain these differences in genetic parallelism and emphasize that more work is needed to move beyond documenting genetic parallelism to identifying the underlying causes.

Uninformative polymorphisms bias genome scans for signatures of selection.

BACKGROUND: With the establishment of high-throughput sequencing technologies and new methods for rapid and extensive single nucleotide (SNP) discovery, marker-based genome scans in search of signatures of divergent selection between populations occupying ecologically distinct environments are becoming increasingly popular. METHODS AND RESULTS: On the basis of genome-wide SNP marker data generated by RAD sequencing of lake and stream stickleback populations, we show that the outcome of such studies can be systematically biased if markers with a low minor allele frequency are included in the analysis. The reason is that these 'uninformative' polymorphisms lack the adequate potential to capture signatures of drift and hitchhiking, the focal processes in ecological genome scans. Bias associated with uninformative polymorphisms is not eliminated by just avoiding technical artifacts in the data (PCR and sequencing errors), as a high proportion of SNPs with a low minor allele frequency is a general biological feature of natural populations. CONCLUSIONS: We suggest that uninformative markers should be excluded from genome scans based on empirical criteria derived from careful inspection of the data, and that these criteria should be reported explicitly. Together, this should increase the quality and comparability of genome scans, and hence promote our understanding of the processes driving genomic differentiation.

Genome divergence during evolutionary diversification as revealed in replicate lake-stream stickleback population pairs.

Evolutionary diversification is often initiated by adaptive divergence between populations occupying ecologically distinct environments while still exchanging genes. The genetic foundations of this divergence process are largely unknown and are here explored through genome scans in multiple independent lake-stream population pairs of threespine stickleback. We find that across the pairs, overall genomic divergence is associated with the magnitude of divergence in phenotypes known to be under divergent selection. Along this same axis of increasing diversification, genomic divergence becomes increasingly biased towards the centre of chromosomes as opposed to the peripheries. We explain this pattern by within-chromosome variation in the physical extent of hitchhiking, as recombination is greatly reduced in chromosome centres. Correcting for this effect suggests that a great number of genes distributed widely across the genome are involved in the divergence into lake vs. stream habitats. Analyzing additional allopatric population pairs, however, reveals that strong divergence in some genomic regions has been driven by selection unrelated to lake-stream ecology. Our study highlights a major contribution of large-scale variation in recombination rate to generating heterogeneous genomic divergence and indicates that elucidating the genetic basis of adaptive divergence might be more challenging than currently recognized.

Evolution: Predictability and the promise of ancient DNA.

Is evolution predictable? Genomes from thousands-of-years-old stickleback suggest that, despite substantial stochasticity in the course of evolution, our understanding of the recent evolutionary past of this species was generally valid.

Constraints on speciation suggested by comparing lake-stream stickleback divergence across two continents.

Adaptation to ecologically distinct environments can coincide with the emergence of reproductive barriers. The outcome of this process is highly variable and can range along a continuum from weak population differentiation all the way to complete, genome-wide divergence. The factors determining how far diverging taxa will move along this continuum remain poorly understood but are most profitably investigated in taxa under replicate divergence. Here, we explore determinants of progress towards speciation by comparing phenotypic and molecular divergence within young (<150 years) lake-stream stickleback pairs from Central Europe to divergence in older (thousands of years) archetypal lake-stream pairs from Vancouver Island, Canada. We generally find relatively weak divergence in most aspects of foraging morphology (gill raker number, body shape) in the European pairs, although substantial adaptive divergence is seen in gill raker length. Combined with striking overall phenotypic differences between the continents, this argues for genetic and time constraints on adaptive divergence in the European pairs. The European lake-stream pairs also do not display the strong habitat-related differentiation in neutral (microsatellite) markers seen in the Canadian watersheds. This indicates either the lack of strong reproductive barriers owing to weak adaptive divergence, or alternatively that neutral markers are poorly suited for detecting reproductive barriers if these emerge rapidly. Overall, our comparative approach suggests constraints on speciation due to genetic architecture and limited time for divergence. The relative importance of these factors remains to be quantified by future investigation.