PERSPECTIVES OPINION Here,we discuss the relationship bety sms. diversification and speciation of orgar The evolutionary significance of WGDs and speciation and argue that most of the ancient WGDs that survived did so because they occurred at specific times: for ancient genome duplications instance, during major ecological upheavals and periods of extinction. At these times, competition with diploids was reduced and Yves Van de Peer, Steven Maere and Axel Meyer new ecological niches became available Furthermore, when WGDs survive they can Abstract Many organisms are currently polyploid, or have a polyploid ancestry greatly enhance the diversification potential and now have secondarily'diploidizedgenomes This finding is surprising of a lineage through the preferential because retained whole-genome duplications(WGDs)are exceedingly rare, retention of regulatory genes suggesting that polyploidy is usually an evolutionary dead end. We argue that ancient genome doublings could probably have survived only under very Competitive advantage of polyploid In the short term, polyploidy may lead specific conditions, but that, whenever established, they might have had a to transgressive segregation and increased pronounced impact on species diversification, and led to an increase vigour. In this section we argue that these in biological complexity and the origin of evolutionary novelties properties might give newly established genitors and a wider phenotypic range, M species of flowering plants and most polyploidy events have occurred thereby increasing their chances of survival. ates have descended from ancestors near the tips of the evolutionary tree of life who doubled their genomes, either through rather than at deeper branches. Although Reducing the risk of extinction. Crow and from cytogenetic analyses, morphological ancient polyploidy events have survived. tions could reduce the risk of extinction autopolyploidy or allopolyploidy. Evidence any species are currently polyploid, few agner have argued that genome dupli studies of fossil and extant species and, more During 500-600 million years of vertebrate through several by functional redun recently, whole-genome and EST analyses evolution, no more than two(or three for dancy, mutational robustness, and increased uggests that most(60-70%)flowering teleosts)WGDs have persisted. Since the rates of evolution and adaptation. Based on plants have a polyploid ancestry >-. In flow- rise of the flowering plants 150-200 million the work of Donoghue and Purnell, these ring plants, polyploids form at a frequency years ago(mya)3. the number of inferred authors observed that genome duplication of I per 100,000 individuals, and% of ancient WGDs in any angiosperm lineage events in vertebrate history seem to have speciation events involve polyploidization@. is at most four. 6. In the fungal lineage, for been preceded by multiple extinct lineages, As a result, many plants, and most of our genome sequences are sulting in pre-duplication gaps in the domesticated crop species, are polyploid known, there is only evidence for a single phylogeny of extant taxa. By analysing the Although polyploidy is much rarer in ancient WGD event". Paleopolyploidy numbers of families in extinct and extant animals than in plants, there are hundreds events therefore seem to be exceed- vertebrate lineages, they concluded that of known insects and vertebrate species ingly rare, and polyploids, or rather their extinction rates were considerably higher that are polyploid, mainly amphibians and descendants, have not been established for pre-duplication lineages than for fish. Whole-genome duplications(WGDs) tens or hundreds of times. However, all post-duplication lineages. ave also been documented for unicellular vertebrates seem to have shared two ancient The most compelling evidence that organisms: the first ancient WGD to be dis- WGD events, whereas all teleosts, and prob- genome duplications might aid in avoid covered in eukaryotes was that of the yeast ably also eudicots, are derived from ing extinction probably comes from Saccharomyces cerevisiae. More recently, lineage that experienced a WGD event 8-. flowering plants. Fawcett et al. showed it was shown that the unicellular ciliate This would suggest that, although descend- that various plants -including legumes, Paramecium tetraurelia has also undergone ants of WGD events do not survive often, cereals, Solanaceae(such as tomatoes and several wgDss when they do survive their evolutionary potatoes), lettuce and cotton -independ Because ancient WGDs in plants and neage can be very successful. ently underwent a WGD-60-70 mya animals gave rise to some particularly The observation that these WGDs pecies-rich groups, some have argued often gave rise to species-rich groups of me to the K-T boundary (BOX 1], suggest end but that it provides novelopportul.ad hat polyploidy is not an evolutionary de organisms,such as >25,000 species of fish ing that polyploid plants coped better with nd >350,000 species of flowering plants, the markedly changed environment than ties for evolutionary successb-.However, suggests that polyploidy can facilitate their diploid progenitors. Although many NATURE REVIEWS GENETICS VOLUME 10 lOCTOBER 20091725 22009 Macmillan Publishers Limited All rights reserved
Most species of flowering plants and vertebrates have descended from ancestors who doubled their genomes, either through autopolyploidy or allopolyploidy. Evidence from cytogenetic analyses, morphological studies of fossil and extant species and, more recently, whole-genome and EST analyses suggests that most (60–70%) flowering plants have a polyploid ancestry1,2–4. In flowering plants, polyploids form at a frequency of 1 per 100,000 individuals5 , and ∼2–4% of speciation events involve polyploidization6 . As a result, many plants, and most of our domesticated crop species, are polyploid7 . Although polyploidy is much rarer in animals than in plants, there are hundreds of known insects and vertebrate species that are polyploid, mainly amphibians and fish6 . Whole-genome duplications (WGDs) have also been documented for unicellular organisms: the first ancient WGD to be discovered in eukaryotes was that of the yeast Saccharomyces cerevisiae8 . More recently, it was shown that the unicellular ciliate Paramecium tetraurelia has also undergone several WGDs9 . Because ancient WGDs in plants and animals gave rise to some particularly species-rich groups, some have argued that polyploidy is not an evolutionary dead end but that it provides novel opportunities for evolutionary success10–13. However, most polyploidy events have occurred near the tips of the evolutionary tree of life rather than at deeper branches. Although many species are currently polyploid, few ancient polyploidy events have survived. During 500–600 million years of vertebrate evolution, no more than two (or three for teleosts) WGDs have persisted. Since the rise of the flowering plants 150–200 million years ago (mya)13,14, the number of inferred ancient WGDs in any angiosperm lineage is at most four15,16. In the fungal lineage, for which many more genome sequences are known, there is only evidence for a single ancient WGD event17. Paleopolyploidy events therefore seem to be exceedingly rare, and polyploids, or rather their descendants, have not been established tens or hundreds of times. However, all vertebrates seem to have shared two ancient WGD events, whereas all teleosts, and probably also eudicots, are derived from a lineage that experienced a WGD event15,18–21. This would suggest that, although descendants of WGD events do not survive often, when they do survive their evolutionary lineage can be very successful. The observation that these WGDs often gave rise to species-rich groups of organisms, such as >25,000 species of fish and >350,000 species of flowering plants, suggests that polyploidy can facilitate diversification and speciation of organisms. Here, we discuss the relationship between WGDs and speciation and argue that most of the ancient WGDs that survived did so because they occurred at specific times: for instance, during major ecological upheavals and periods of extinction. At these times, competition with diploids was reduced and new ecological niches became available. Furthermore, when WGDs survive they can greatly enhance the diversification potential of a lineage through the preferential retention of regulatory genes. Competitive advantage of polyploids In the short term, polyploidy may lead to transgressive segregation and increased vigour. In this section we argue that these properties might give newly established polyploids an edge over their diploid progenitors and a wider phenotypic range, thereby increasing their chances of survival. Reducing the risk of extinction. Crow and Wagner22 have argued that genome duplications could reduce the risk of extinction through several means: by functional redundancy, mutational robustness, and increased rates of evolution and adaptation. Based on the work of Donoghue and Purnell23, these authors observed that genome duplication events in vertebrate history seem to have been preceded by multiple extinct lineages, resulting in pre-duplication gaps in the phylogeny of extant taxa. By analysing the numbers of families in extinct and extant vertebrate lineages, they concluded that extinction rates were considerably higher for pre-duplication lineages than for post-duplication lineages. The most compelling evidence that genome duplications might aid in avoiding extinction probably comes from flowering plants. Fawcett et al.24 showed that various plants — including legumes, cereals, Solanaceae (such as tomatoes and potatoes), lettuce and cotton — independently underwent a WGD ~60–70 mya. This wave of WGDs occurred close in time to the K–T boundary (BOX 1), suggesting that polyploid plants coped better with the markedly changed environment than their diploid progenitors. Although many OpiniOn The evolutionary significance of ancient genome duplications Yves Van de Peer, Steven Maere and Axel Meyer Abstract | Many organisms are currently polyploid, or have a polyploid ancestry and now have secondarily ‘diploidized’ genomes. This finding is surprising because retained whole-genome duplications (WGDs) are exceedingly rare, suggesting that polyploidy is usually an evolutionary dead end. We argue that ancient genome doublings could probably have survived only under very specific conditions, but that, whenever established, they might have had a pronounced impact on species diversification, and led to an increase in biological complexity and the origin of evolutionary novelties. PersPecTives nATurE rEvIEWS | Genetics voluME 10 | oCToBEr 2009 | 725 © 2009 Macmillan Publishers Limited. All rights reserved
ERSPECtIVes Box 1 Whole-genome duplications across the phylogeny of eukaryotes Angiosperms Animals Fungi 铺 Cenozoic Cretaceous Monocots ore)Eudicots 145 mya Ascomycetes Devonian 409 mya vertebrates Angiosperms-moss split ebrates split Cambrian Precambrian Whole-genome duplications (WGDs)seem to have been followed by a Similarly, early polyploidization events in one or more angiosperm plant lbstantial increase in morphological complexity (see the figure, lineages might explain the rapid rise and diversification of angiosperms paleopolyploidy events are indicated as red bars and are based on studies in the Early Cretaceous period 240. Fundamental innovations that published previously for plants 6, 24, fish2,5556, vertebrates, fungi and occurred early in angiosperm evolution are the invention of the closed ciliates"). The two rounds of genome duplication(1R and 2R)in the carpel, and the emergence of flowers and of double fertilization.These vertebrate stem were followed by a period of rapid morphological early innovations were elaborated specialized pollination n, which led to: enhanced nervous, endocrine and circulatory strategies and fruits. The evolution of xylem vessels is also believed stems; enhanced sensory organs; more complex brains; and the skull, have been important in early erm diversification, but their origin vertebrae, the endoskeleton and teeth. These were followed in the is less clear. Some basal angiosperms, such as Amborella spp, lack vessels, jawed vertebrate lineage by innovations such as paired appendages, hinged whereas vessel structures have been discovered in members of the 0o2-l0s. Many of these innovations are Gnetales order and in ferns. 0. However, the diversity of vasculature related to the emergence in vertebrates of the neural cresto. Since Ohno in at rms is unparallelled. The timing of the early angiosperm first suggested that these innovations are facilitated by genome polyploidizations is unclear. It is possible that they contributed to the duplications", a causal link between the 2R duplication and the emergence elaboration and diversification of the afe of vertebrates has been suggested (see, for example, REFS 78, 107) than to their establishment 22009 Macmillan Publishers Limited All rights reserved
Nature Reviews | Genetics Land vertebrates Angiosperms–moss split Fish–land vertebrates split Teleosts Ascomycetes 3R 2R 1R Fish Angiosperms Moss Animals Fungi Ciliates Cenozoic Cretaceous Jurassic Triassic Permian Carboniferous Devonian Silurian Ordovician Cambrian Precambrian Physcomitrella patens Kluyveromyces lactis Paramecium spp. Tetrahymena spp. Lampreys Hagfish Neurospora crassa Aspergillus fumigatus Acorus americanus Sor Zea mays ghum bicolor Oryza sativa Hordeum vulgare Triticum aestivum Musa spp. Eschscholzia californica Arabidopsis thaliana Carica papaya Gossypium hirsutum Populus trichocarpa Medic Glycine max ago truncatula Vitis vinif Lotus japonicus era Lactuca sativa Centaurea solstitialis Solanum lycopersicum Solanum tuberosum Bichir (Polypteriformes) T Medaka akifugu rubripes Zebrafish Bony tongues (Osteoglossiformes) Gar (Semionotiformes) Sturgeon (Acipenseriformes) Mammals Amphibi Lobe-finned fish a Birds Saccharomyces cerevisiae Saccharomyces spp. Candida glabrata Monocots (Core) Eudicots 65 mya 145 mya 208 mya 245 mya 290 mya 363 mya 409 mya 439 mya 510 mya 542 mya >542 mya Box 1 | Whole-genome duplications across the phylogeny of eukaryotes Whole-genome duplications (WGDs) seem to have been followed by a substantial increase in morphological complexity (see the figure, paleopolyploidy events are indicated as red bars and are based on studies published previously for plants3,16,24, fish20,55,56, vertebrates101, fungi17 and ciliates9 ). The two rounds of genome duplication (1R and 2R) in the vertebrate stem were followed by a period of rapid morphological innovation, which led to: enhanced nervous, endocrine and circulatory systems; enhanced sensory organs; more complex brains; and the skull, vertebrae, the endoskeleton and teeth. These were followed in the jawed vertebrate lineage by innovations such as paired appendages, hinged jaws and an adaptive immune system40,102–105. Many of these innovations are related to the emergence in vertebrates of the neural crest40,102. Since Ohno first suggested that these innovations are facilitated by genome duplications106, a causal link between the 2R duplication and the emergence of vertebrates has been suggested (see, for example, REFS 78,107). Similarly, early polyploidization events in one or more angiosperm plant lineages might explain the rapid rise and diversification of angiosperms in the Early Cretaceous period6,11,13,14,52,108. Fundamental innovations that occurred early in angiosperm evolution are the invention of the closed carpel, and the emergence of flowers and of double fertilization109. These early innovations were elaborated on to create specialized pollination strategies and fruits. The evolution of xylem vessels is also believed to have been important in early angiosperm diversification85, but their origin is less clear. Some basal angiosperms, such as Amborella spp., lack vessels, whereas vessel structures have been discovered in members of the Gnetales order and in ferns85,110,111. However, the diversity of vasculature in angiosperms is unparallelled. The timing of the early angiosperm polyploidizations is unclear. It is possible that they contributed to the elaboration and diversification of the aforementioned inventions rather than to their establishment. Pers P ectives 726 | oCToBEr 2009 | voluME 10 www.nature.com/reviews/genetics © 2009 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES anges associated with polyploidization be a selective advantage when sexual mate ld are probably disadvantageous or deleteri. are scarce. Following this logic, environ in reproductive isolatic oncluded ous., it seems that many K-T polyploids mental upheaval may have been a driving that RGL at duplicated contribute outcompeted their diploid progenitors, force in shaping survivorship probabilities to speciation events that occurred after the probably owing to a higher tolerance of a associated with genome duplication the teleost WGD wider range of environmental conditions, 26. clustered genome duplications in flower No similar studies have been performed Alternatively, in a more 'neutral scenario, ing plants at the K-T boundary provide r plants, but recent experimental work has one could assume that environmental stress tantalizing example. provided evidence that reciprocal silenc leads to an increased incidence of polypro However, owing to uncertainties in the ing or loss of duplicated genes provides an formation: for instance, through the produc- dating of most ancient WGDs, a more gen- important source of epistatic interactions that tion of unreduced, 2n gametes". In this case, eral link between WGDs and major extinc- follow the Bateson-Dobzhansky-Muller model the cataclysmic events that were responsible tions cannot be ascertained. The 2R WGD Bikard et al. show that, in crosses between for the K-T extinction could have increased event in vertebrates may date from 520-550 different accessions of Arabidopsis thaliana, the establishment of polyploid lineages by mya,, close to the mass extinction at loci interact in an epistatic manner to control chance. However, it is unclear whether such the dawn of the Cambrian explosion a recessive embryo lethality. This effect is an increase alone could explain the extent (542 mya-), and the genome duplication explained by divergent evolution occurring to which polyploid plants replaced or in teleosts, which according to the most mong paralogues of an essential duplicated overshadowed their diploid relatives recent estimate happened 226-316 mya", gene when the functional copy is not located may have occurred close to the Permian at the same locus in the different accessions Increased vigour In the adaptive scenario, Triassic(P-T)mass extinction event this results in lowered fitness in the first or eterotic effects and rapid genomic and(250 mya). For other paleopolyploidies- second filial (f, or F, )generations of certain epigenetic changes underlie the ability of for example, in S cerevisiae and the core rosses, which contributes to reproductive polyploids to quickly adapt to more extreme eudicots-there is no indication that they isolation By demonstrating the link between environments. In allopolyploids and autop are linked to mass extinction events gene duplication and genetic incompatibil ploids, increased heterozygosity can lead to ity, the authors provide direct evidence for increased variation in gene expression and Increased species diversity duplicate gene loss as a neutral mechanism in regulatory wiring", which may result in Genome duplications often seem to be that generates post-zygotic isolating barriers increased vigour and faster adaptation to novel accompanied by marked and sudden within existing species or populations onditions=3. Rapid genomic and epigenetic increases in species richness. Although a changes after WGD may similarly lead to link between any specific genome duplica- Subfunctionalization. Other neutral sce increased variation and transgressive traits. tion event and increased species diversit narios might also promote speciation. One Transgressive segregation in polyploids remains correlational rather than causal example would be a case in which both might serve as a pre-adaptation for survival several mechanisms might explain how opies of a gene that has multiple functions in habitats that were not accessible to their gene duplication facilitates the formation(for instance, it is expressed at different diploid parent species. Several studies of novel species. stages in development or in different tissues) have suggested that polyploid plants are are retained in different populations after a more tolerant to a wide of environ- Reciprocal gene loss. Both Werth and duplication event. Should the populations mental conditions compared with thei Windham and Lynch and Force proposed become geographically isolated, the two diploid relatives. Furthermore, many that the loss of different copies of a dupli- duplicate genes in each population could polyploids are invasive 233 and can exploit cated gene in separated populations might subfunctionalize and the orthologues in the habitats that their diploid progenitors can- genetically isolate these populations (FIG. 2). different populations might evolve different not26 34. Polyploid insects also have a wider Divergent resolution of the thousands to tens functions. The resulting F, hybrids from the geographical distribution than their diploid of thousands of genes and regulatory rnAs two populations would develop correctly progenitors, often colonizing northern and that are produced by a genome duplication because each subfunction is performed mountain regions.One of the rare exa event could therefore potentially facilitate by one of the genes from each population. ples of relatively recent polyploidy establish- speciation. However, one-eighth of the F, zygotes will ment in vertebrates is given by the tetraploid Scannell et al. "studied gene loss in lack one of the subfunctions and will die if frog Xenopus laevis, which is a highly three yeast species that have undergone a this function is essentials(FIG. 2b). As a invasive species that colonizes disturbed WGD and showed that, at many loci, dif- result, lineage-specific subfunction parti- and man-made habitats. It is also extremely ferent species lost different members of a tioning could accelerate rates of speciation tolerant to salt, drought, cold and starvation, duplicated pair, so that 4-7% of single-copy and is more disease resistant than its diploid genes compared between any two species Speciation. There seems to be a correlation relative Silurana tropicalis are not orthologues but paralogues. Such a between WGDs in plants and increased In summary, increased phenotypic pattern provides strong evidence for specia- rates of speciation or divergence. First, there variability and heterotic effects have the tion through the reciprocal gene loss(RGl) seems to be a correlation between the old potential to enable polyploids to survive modelb(FIG 2a]. Similar findings have WGDs and the early and fast diversifica environmental conditions that do not favour been reported for duplicated fish genomes, tion of flowering plants$2.53. Second, Soltis their diploid ancestors"(FIG. 1). Polyploidy in which it is estimated that-1, 700(8%) et al. found a strong correlation between is also known to facilitate self-fertilization estral loci of Tetraodon nigroviridi diversification rates and polyploidy follow and the formation of asexually reproduc nd zebrafish underwent RGL". Because ing recent WGDs in many plant lineages. For ing(apomictic)species.,which might RGL at only a few pairs of loci that encode instance, the WGD in the Poaceae lineage URE REVIEWS GENETICS 22009 Macmillan Publishers Limited All rights reserved
changes associated with polyploidization are probably disadvantageous or deleterious6,11,12, it seems that many K–T polyploids outcompeted their diploid progenitors, probably owing to a higher tolerance of a wider range of environmental conditions25,26. Alternatively, in a more ‘neutral’ scenario, one could assume that environmental stress leads to an increased incidence of polyploid formation: for instance, through the production of unreduced, 2n gametes27. In this case, the cataclysmic events that were responsible for the K–T extinction could have increased the establishment of polyploid lineages by chance. However, it is unclear whether such an increase alone could explain the extent to which polyploid plants replaced or overshadowed their diploid relatives. Increased vigour. In the adaptive scenario, heterotic effects and rapid genomic and epigenetic changes underlie the ability of polyploids to quickly adapt to more extreme environments. In allopolyploids and autopolyploids, increased heterozygosity can lead to increased variation in gene expression and in regulatory wiring28, which may result in increased vigour and faster adaptation to novel conditions29,30. rapid genomic and epigenetic changes after WGD may similarly lead to increased variation and transgressive traits28. Transgressive segregation in polyploids might serve as a pre-adaptation for survival in habitats that were not accessible to their diploid parent species22,31. Several studies have suggested that polyploid plants are more tolerant to a wider range of environmental conditions compared with their diploid relatives25,26. Furthermore, many polyploids are invasive32,33 and can exploit habitats that their diploid progenitors cannot26,34. Polyploid insects also have a wider geographical distribution than their diploid progenitors, often colonizing northern and mountain regions35. one of the rare examples of relatively recent polyploidy establishment in vertebrates is given by the tetraploid frog Xenopus laevis, which is a highly invasive species that colonizes disturbed and man-made habitats. It is also extremely tolerant to salt, drought, cold and starvation, and is more disease resistant than its diploid relative Silurana tropicalis36,37. In summary, increased phenotypic variability and heterotic effects have the potential to enable polyploids to survive environmental conditions that do not favour their diploid ancestors38 (FIG. 1). Polyploidy is also known to facilitate self-fertilization and the formation of asexually reproducing (apomictic) species35,39, which might be a selective advantage when sexual mates are scarce. Following this logic, environmental upheaval may have been a driving force in shaping survivorship probabilities associated with genome duplication22; the clustered genome duplications in flowering plants at the K–T boundary provide a tantalizing example. However, owing to uncertainties in the dating of most ancient WGDs, a more general link between WGDs and major extinctions cannot be ascertained. The 2r WGD event in vertebrates may date from 520–550 mya19,40, close to the mass extinction at the dawn of the Cambrian explosion (542 mya41–43), and the genome duplication in teleosts, which according to the most recent estimate happened 226–316 mya44, may have occurred close to the Permian– Triassic (P–T) mass extinction event (250 mya). For other paleopolyploidies — for example, in S. cerevisiae and the core eudicots — there is no indication that they are linked to mass extinction events. increased species diversity Genome duplications often seem to be accompanied by marked and sudden increases in species richness. Although a link between any specific genome duplication event and increased species diversity remains correlational rather than causal, several mechanisms might explain how gene duplication facilitates the formation of novel species. Reciprocal gene loss. Both Werth and Windham45 and lynch and Force46 proposed that the loss of different copies of a duplicated gene in separated populations might genetically isolate these populations (FIG. 2). Divergent resolution of the thousands to tens of thousands of genes and regulatory rnAs that are produced by a genome duplication event could therefore potentially facilitate speciation. Scannell et al.47 studied gene loss in three yeast species that have undergone a WGD and showed that, at many loci, different species lost different members of a duplicated pair, so that 4–7% of single-copy genes compared between any two species are not orthologues but paralogues. Such a pattern provides strong evidence for speciation through the reciprocal gene loss (rGl) model45,46 (FIG. 2a). Similar findings have been reported for duplicated fish genomes, in which it is estimated that ~1,700 (8%) ancestral loci of Tetraodon nigroviridis and zebrafish underwent rGl48. Because rGl at only a few pairs of loci that encode essential genes would be sufficient to result in reproductive isolation, it was concluded that rGl at duplicated loci might contribute to speciation events that occurred after the teleost WGD48. no similar studies have been performed for plants, but recent experimental work has provided evidence that reciprocal silencing or loss of duplicated genes provides an important source of epistatic interactions that follow the Bateson–Dobzhansky–Muller model. Bikard et al.49 show that, in crosses between different accessions of Arabidopsis thaliana, loci interact in an epistatic manner to control a recessive embryo lethality. This effect is explained by divergent evolution occurring among paralogues of an essential duplicated gene when the functional copy is not located at the same locus in the different accessions; this results in lowered fitness in the first or second filial (F1 or F2 ) generations of certain crosses, which contributes to reproductive isolation. By demonstrating the link between gene duplication and genetic incompatibility, the authors provide direct evidence for duplicate gene loss as a neutral mechanism that generates post-zygotic isolating barriers within existing species or populations. Subfunctionalization. other neutral scenarios might also promote speciation. one example would be a case in which both copies of a gene that has multiple functions (for instance, it is expressed at different stages in development or in different tissues) are retained in different populations after a duplication event. Should the populations become geographically isolated, the two duplicate genes in each population could subfunctionalize46 and the orthologues in the different populations might evolve different functions. The resulting F1 hybrids from the two populations would develop correctly because each subfunction is performed by one of the genes from each population. However, one-eighth of the F2 zygotes will lack one of the subfunctions and will die if this function is essential50,51 (FIG. 2b). As a result, lineage-specific subfunction partitioning could accelerate rates of speciation. Speciation. There seems to be a correlation between WGDs in plants and increased rates of speciation or divergence. First, there seems to be a correlation between the older WGDs and the early and fast diversification of flowering plants52,53. Second, Soltis et al.13 found a strong correlation between diversification rates and polyploidy following recent WGDs in many plant lineages. For instance, the WGD in the Poaceae lineage Pers P ectives nATurE rEvIEWS | Genetics voluME 10 | oCToBEr 2009 | 727 © 2009 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES possibly coincides with the origin and In fish, the correlation between WGD Late Cretaceous and Tertiary periods, mo divergence of the core Poaceae, a large clade and species diversification rates is less clear. than 150 million years later. This observa containing-10,000 species. Early-branching Fish constitute half of all vertebrate species tion could be taken to indicate that genome subclades of the Poaceae, as well as closely and are a highly successful and diverse evo- duplication was not an important factor in related non-Poaceae families, contain only lutionary lineage. The fish-specific genome the rapid radiation of teleosts. However, both a small number of species. Whole-genome duplication(3R)in the teleost lineage is esti- RGL and subfunction partitioning can occur duplications have also been reported for mated to have occurred 226-350 myas rer tens of millions of years after a WGD the Brassicaceae(3, 700 species), Asteraceae The inferred phylogenetic timing of 3R and can continue to promote speciation over (23,000 species), the Fabaceae(19, 400 spe- seems to separate the species-poor, early- long periods of times. It is conceivable that cies)and the Solanaceae(>3,000 species), to branching lineages of ray-finned fish from 3R continued to increase the propensity for name but a few, and these WGDs also seem the species-rich teleost lineage, and therefore speciation until a suitable ecological occa- to correlate with species-rich plant families, seems to provide evidence that 3R might sion presented itself, such as the K-Tmass Ithough the precise phylogenetic placement be causally related to an increase in species extinction. As an example of such stored of these WGDs is unclear. Furthermore, and biological diversity. However, there is diversifying potential, X laevis still main the rate of diversification is also high in these a large period of time between 3R and the tains-32-47%of its genes in duplicate, families compared with other families in the main teleost radiations, which, according ome 40 million years after its most recent same orders to fossil evidence, did not occur until the polyploidization event, and its genome 姒“ Figure 1 Survival of the fittest. The figure illustrates one of many.2-s a new niche(the new peak is indicated by an arrow in c). None of the exi simplified fitness landscape models. The upper and lower panels show the ing species has the evolutionary potential to fill this niche, but a polyploid fitness landscape with tw ary phenotype axes, 1 and 2. These axes species(white dot in b and d) may be able to develop the necessary phe- do not represent single quantitative traits but rather a flattened version of notypic innovationse, f In another scenario, the fitness landscape changes phenotype space. The black dots It well-adapted organisms that drastically. for example, through a catastrophic event. Most organisms can- ccupy the peaks in phenotype space(red indicates the most well adapted, not adapt to the changed environment and perish (red crosses). Some blue the least well adapted), which correspond to niches in which that par- organisms (near the centre of the landscape) live in relatively unaltered ticular combination of phenotypic characters is advantageous. The full niches and can adapt enough to survive. Others may manage to survive ircles represent the phenotypes accessible to the organisms, whereas the initially through polyploidization (white dots), outcompeting their diploid dashed circles are a simplified representation of the phenot of parents because of, for example, heterotic effects. These polyploi their polyploid relatives. Blue regions of the phenotype space are not via- harbour the potential to develop innovations that in time may enable them ble, so there is little room for successful genome duplication events. to colonize empty niches in phenotype space that cannot be reached by a-d In one scenario, there is an unoccupied peak in the fitness landscape other organisms. Differential realization of this potential among the polyploid offspring may lead to phenotype diversifi 22009 Macmillan Publishers Limited All rights reserved
Nature Reviews | Genetics 0 0.5 0.1 1.5 Phenotype 1 Phenotype 1 Phenotype 2 Phenotype 1 Phenotype 2 Phenotype 1 Phenotype 2 Phenotype 2 0 0.5 0.1 1.5 c Phenotype 1 Phenotype 2 d 0 0.5 0.1 1.5 e Phenotype 1 Phenotype 2 f Fitness Fitness Fitness a b possibly coincides with the origin and divergence of the core Poaceae, a large clade containing ~10,000 species. Early-branching subclades of the Poaceae, as well as closely related non-Poaceae families, contain only a small number of species. Whole-genome duplications have also been reported for the Brassicaceae (3,700 species), Asteraceae (23,000 species), the Fabaceae (19,400 species) and the Solanaceae (>3,000 species), to name but a few, and these WGDs also seem to correlate with species-rich plant families, although the precise phylogenetic placement of these WGDs is unclear13. Furthermore, the rate of diversification is also high in these families compared with other families in the same orders54. In fish, the correlation between WGD and species diversification rates is less clear. Fish constitute half of all vertebrate species and are a highly successful and diverse evolutionary lineage21. The fish-specific genome duplication (3r) in the teleost lineage is estimated to have occurred 226–350 mya44,55–57. The inferred phylogenetic timing of 3r seems to separate the species-poor, earlybranching lineages of ray-finned fish from the species-rich teleost lineage, and therefore seems to provide evidence that 3r might be causally related to an increase in species and biological diversity. However, there is a large period of time between 3r and the main teleost radiations, which, according to fossil evidence, did not occur until the late Cretaceous and Tertiary periods, more than 150 million years later. This observation could be taken to indicate that genome duplication was not an important factor in the rapid radiation of teleosts. However, both rGl and subfunction partitioning can occur over tens of millions of years after a WGD and can continue to promote speciation over long periods of time47,48. It is conceivable that 3r continued to increase the propensity for speciation until a suitable ecological occasion presented itself, such as the K–T mass extinction. As an example of such stored diversifying potential, X. laevis still maintains ~32–47%58 of its genes in duplicate, some 40 million years after its most recent polyploidization event, and its genome Figure 1 | survival of the fittest. The figure illustrates one of many92,95,112–115 simplified fitness landscape models. The upper and lower panels show the fitness landscape with two imaginary phenotype axes, 1 and 2. These axes do not represent single quantitative traits but rather a flattened version of phenotype space. The black dots represent well-adapted organisms that occupy the peaks in phenotype space (red indicates the most well adapted, blue the least well adapted), which correspond to niches in which that particular combination of phenotypic characters is advantageous. The full circles represent the phenotypes accessible to the organisms, whereas the dashed circles are a simplified representation of the phenotype space of their polyploid relatives. Blue regions of the phenotype space are not viable, so there is little room for successful genome duplication events. a–d | in one scenario, there is an unoccupied peak in the fitness landscape (a,b) or a new fitness peak emerges (c,d), for instance, through evolution of a new niche (the new peak is indicated by an arrow in c). None of the existing species has the evolutionary potential to fill this niche, but a polyploid species (white dot in b and d) may be able to develop the necessary phenotypic innovations. e,f | in another scenario, the fitness landscape changes drastically, for example, through a catastrophic event. Most organisms cannot adapt to the changed environment and perish (red crosses). some organisms (near the centre of the landscape) live in relatively unaltered niches and can adapt enough to survive. Others may manage to survive initially through polyploidization (white dots), outcompeting their diploid parents because of, for example, heterotic effects. These polyploids also harbour the potential to develop innovations that in time may enable them to colonize empty niches in phenotype space that cannot be reached by other organisms. Differential realization of this potential among the polyploid offspring may lead to phenotype diversification and speciation. Pers P ectives 728 | oCToBEr 2009 | voluME 10 www.nature.com/reviews/genetics © 2009 Macmillan Publishers Limited. All rights reserved
PERSPECTIVES shows little evidence of du tionalization or neofunctionalization s-60 Any theory that attempts to link wGD to species diversity should take into account the fact that radiations are not always pre- Genome duplication t ceded by genome duplications. Invertebrates and vertebrates have diversified at similar ates, despite the fact that the vertebrates underwent two rounds of genome duplication and the invertebrates none Evolutionary innovations In the longer run, polyploidy may pave the way for evolutionary innovations or elabora- tions of existing morphological structures that allow exploration of fundamentally Geographic different regions of phenotype space ene loss solaria Subfunctionalization Genome duplication favours gene retention One of the prerequisites for developing more complex systems is an increase in the CDCD DO number of gene regulators. Intriguingly, duplications are the preferred way to ccomplish such an increase. Transcriptional Population1\ POpulation 2 lation\ y and developmental regulators and signal transducers have been preferentially retained in duplicate after all genome duplications in bidopsis thaliana63-65, after the IR and 21 WGDs in vertebrates 9,66 after 3R in fish, 67, and after the WGD in yeast.69. Mo these regulatory gene classes cannot be cations, which accentuates the importane of genome duplications in expanding the regulatory gene repertoire. Maere et al. as estimated that more than 90% of the increase in regulatory genes in the Arabidopsis lineage in the last-150 million years caused by genome duplications. Both under-retention of regulators after gene duplications and their over-retention after genome duplications can be explained by dosage balance effects.. Freeling and Thomas"and Freeling"argue that, after 個DD( modules are inherently retained in duplicate Figure 2I Reciprocal gene loss or subfunctionalization facilitates speciation. Red bands on after which they can adaptively evolve novel tion event. a After diploidization, the duplicated gene is present on two different chromosomes. After geographic isolation, both populations have lost one of the duplicates on different chromo- functions and might ultimately cause an somes. If individuals from isolated populations mate. their hybrid progeny would be heterozygous. increase in morphological complexity possessing a functional allele at each locus of the duplicated gene. However, one-sixteenth The study of individual gene families also (approximately 6%)of crosses between the first filial( F, individuals produce second filial(F)indi- points to the importance of genome duplica- viduals that have null alleles at both loci in question(dark grey square)and therefore lack viability tions in expanding the regulatory gene rep- and/or fertility. Others might receive one allele(light grey squares), which might reduce functional- ertoire of an organism. In plants, important ity when a gene is haploinsufficient, or might receive three or four functional alleles (mid-grey developmental regulators, such as the AUX/ juares), which might have a negative dosage effect. All these outcomes might lead to post-mating certain MADS-box transcription factor sub. cated genes in the different populations have subfunctionalized (orange and yellow bands on families?,76, seem to have expanded mainly but one-sixteenth of the F, generation will be homozygous for alleles lacking one essential sub through genome duplications In vertebrates, function, and another one-sixteenth will be homozygous for alleles lacking the other essential IR and 2R are thought to be responsible for subfunction (dark grey squares), thus reducing the fitness of hybrids. Other F, individuals might, the expansion of the number of homeobox as in a, show reduced fitness caused by dosage or haploinsufficiency effects. URE REVIEWS GENETICS VOLUME 10 lOCTOBER 20091729 22009 Macmillan Publishers Limited All rights reserved
Nature Reviews | Genetics F2 F1 F1 a Geographic isolation Geographic isolation Gene loss Subfunctionalization Population 1 Population 2 Population 1 Population 2 Diploid Polyploid Paleopolyploid b Genome duplication Genome duplication Diploidization Diploidization Figure 2 | Reciprocal gene loss or subfunctionalization facilitates speciation. red bands on chromosomes represent a locus that is duplicated (along with all other loci) during a tetraploidization event. a | After diploidization, the duplicated gene is present on two different chromosomes. After geographic isolation, both populations have lost one of the duplicates on different chromosomes. if individuals from isolated populations mate, their ‘hybrid’ progeny would be heterozygous, possessing a functional allele at each locus of the duplicated gene. However, one-sixteenth (approximately 6%) of crosses between the first filial (F1 ) individuals produce second filial (F2 ) individuals that have null alleles at both loci in question (dark grey square) and therefore lack viability and/or fertility. Others might receive one allele (light grey squares), which might reduce functionality when a gene is haploinsufficient, or might receive three or four functional alleles (mid-grey squares), which might have a negative dosage effect. All these outcomes might lead to post-mating reproductive isolation46. b | in this scenario, after diploidization and geographic isolation, the duplicated genes in the different populations have subfunctionalized (orange and yellow bands on chromosomes). Hybrids between the two populations should in general develop normally, but one-sixteenth of the F2 generation will be homozygous for alleles lacking one essential subfunction, and another one-sixteenth will be homozygous for alleles lacking the other essential subfunction (dark grey squares), thus reducing the fitness of hybrids. Other F2 individuals might, as in a, show reduced fitness caused by dosage or haploinsufficiency effects. shows little evidence of duplicate subfunctionalization or neofunctionalization58–60. Any theory that attempts to link WGD to species diversity should take into account the fact that radiations are not always preceded by genome duplications. Invertebrates and vertebrates have diversified at similar rates61, despite the fact that the vertebrates underwent two rounds of genome duplication and the invertebrates none. Evolutionary innovations In the longer run, polyploidy may pave the way for evolutionary innovations or elaborations of existing morphological structures that allow exploration of fundamentally different regions of phenotype space. Genome duplication favours gene retention. one of the prerequisites for developing more complex systems is an increase in the number of gene regulators62. Intriguingly, genome duplications are the preferred way to accomplish such an increase. Transcriptional and developmental regulators and signal transducers have been preferentially retained in duplicate after all genome duplications in Arabidopsis thaliana63–65, after the 1r and 2r WGDs in vertebrates19,66, after 3r in fish66,67 , and after the WGD in yeast68,69. Moreover, these regulatory gene classes cannot be expanded easily through single-gene duplications, which accentuates the importance of genome duplications in expanding the regulatory gene repertoire. Maere et al.63 estimated that more than 90% of the increase in regulatory genes in the Arabidopsis lineage in the last ∼150 million years is caused by genome duplications. Both the under-retention of regulators after singlegene duplications and their over-retention after genome duplications can be explained by dosage balance effects70,71. Freeling and Thomas72 and Freeling73 argue that, after genome duplication, entire functional modules are inherently retained in duplicate through non-adaptive dosage balance effects, after which they can adaptively evolve novel functions and might ultimately cause an increase in morphological complexity. The study of individual gene families also points to the importance of genome duplications in expanding the regulatory gene repertoire of an organism. In plants, important developmental regulators, such as the AuX/ IAA family of auxin response regulators74 and certain MADS-box transcription factor subfamilies75,76, seem to have expanded mainly through genome duplications. In vertebrates, 1r and 2r are thought to be responsible for the expansion of the number of homeobox Pers P ectives nATurE rEvIEWS | Genetics voluME 10 | oCToBEr 2009 | 729 © 2009 Macmillan Publishers Limited. All rights reserved