REVIEWS
Hybrid speciation
James Mallet1
Botanists have long believed that hybrid speciation is important, especially after chromosomal doubling (allopolyploidy).
Until recently, hybridization was not thought to play a very constructive part in animal evolution. Now, new genetic evidence
suggests that hybrid speciation, even without polyploidy, is more common in plants and also animals than we thought.
L
innaeus stated in Systema Naturae that species have remained
unchanged since the dawn of time, but he later experimented
with hybrids and convinced himself that hybridization provided
a means of species modification. One hundred and
eighty years later, Lotsy1
still argued that species were invariant
genetic types, and that novel lineages could evolve only by means
of hybridization. These peculiar ideas were overturned when the
concept emerged of species as reproductively isolated populations2–4
.
In zoology, this concept discouraged the view that hybridization and
gene flow (introgression) between species could be important evolutionary
forces2,5,6
, even while botanists continued to argue for their
significance4,7,8
. Today, armed with new and abundant molecular
marker data, biologists increasingly find new examples where hybridization
seems to facilitate speciation and adaptive radiation in animals,
as well as plants8–12
.
What is hybrid speciation?
‘Hybrid speciation’ implies that hybridization has had a principal
role in the origin of a new species. The definition applies cleanly to
hybrid species that have doubled their chromosome number (allopolyploidy):
derived species initially contain exactly one genome
from each parent, a 50% contribution from each, although, in older
polyploids, recombination and gene conversion may eventually lead
to unequal contributions. Furthermore, allopolyploids are largely
reproductively isolated by ploidy. Recombinational hybrid speciation,
in which the genome remains diploid (homoploid hybrid
speciation), is harder to define. The fraction from each parent will
rarely be 50% if backcrossing is involved. Homoploid hybrid species
may be only weakly reproductively isolated, and are hard to distinguish
from species that gain alleles by hybridization and introgression,
or from persistent ancestral polymorphisms. Although hybrid
speciation is sometimes inferred if any marker alleles originate from
different parents, I here restrict the term to cases where hybrid allelic
combinations contribute to the spread and maintenance of stabilized
hybrid lineages generally recognized as species.
This raises the question of what exactly we mean by ‘species’.
Hybrid speciation is only possible if reproductive isolation is weak;
if hybrids are intermediate, hybrid species will be even more weakly
isolated. In practice, we must recognize species as multi-locus ‘genotypic
clusters’ (Box 1)6,13
. A hybrid species will then be a third cluster
of genotypes, a hybrid form that has become stabilized and remains
distinct when in contact with either parent.
Hybridization can also influence speciation by means of ‘reinforcement’,
where mating barriers evolve owing to selection against
unfit hybrids6,14,15
. Although hybridization contributes to speciation,
I do not consider reinforcement to be hybrid speciation, because a
third species does not form. A related and highly relevant phenomenon
is ‘hybridogenesis’. The diploid or triploid edible frog Rana
esculenta is a well known example: it is heterozygous for complete
Rana lessonae and Rana ridibunda genomes16
. Here, I exclude hybridogenetic
species because they do not breed true.
Theory and background of hybrid speciation
Hybridization may be ‘‘the grossest blunder in sexual preference
which we can conceive of an animal making’’17
, but it is nonetheless
a regular event. The fraction of species that hybridize is variable, but
on average around 10% of animal and 25% of plant species are
known to hybridize with at least one other species18
. Hybridization
is especially prevalent in rapidly radiating groups: 75% of British
ducks (Anatidae)18
, for example. Recent, closely related species are
most likely to hybridize, although hybridization and introgression
1
Galton Laboratory, Department of Biology, University College London, 4 Stephenson Way, London NW1 2HE, UK.
Box 1 j Species as genotypic clusters versus reproductively isolated
populations
Species can be defined as distinguishable groups of genotypes that
remain distinct in the face of potential or actual hybridization and gene
flow6,13
. This is similar to Darwin’s usage of species to divide
biodiversity by means of gaps or troughs in the distributions of
phenotypes and genotypes. A very likely reason why a pair of genotypic
clusters in contact might remain distinct is, of course, ‘reproductive
isolation’, but this becomes a means of achieving speciation and
species maintenance rather than a definition of the species state itself.
It might seem that autopolyploids (species resulting from
chromosome doubling within a single parent species) cannot be
recognized as genotypic clusters because they have initial gene
frequencies like their diploid parents. Autopolyploids are, however,
genetically distinct in heritable traits such as chromosome number and
ploidy at individual loci. They can be regarded as distinct species
provided that euploids (for example, diploids and tetraploids) form
clusters that are more abundant than intermediates formed by
hybridization between them (for example, triploids and aneuploids).
Such species have no guarantee of permanence. Two genotypic
clusters might be stable for a long time, yet when ecological
circumstances change, gene flow may exceed some threshold,
eventually resulting in a single genotypic cluster that absorbs both
species13
. There are several examples of species fusion in the literature,
for example in Darwin’s finches and cichlid fish44,45
. ‘Despeciation’
itself could be classified as a form of hybrid speciation, as a new species
has resulted from the fusion of two old species. I exclude despeciation
here because, in my definition, hybrid species should remain distinct
when in contact with either parent. Many have argued that permanent
divergence is an important criterion of species2
. However, dropping
this a priori requirement seems reasonable to avoid the need to predict
an often unpredictable future for distinct, existing taxa6
, and allows for
extinction of species via genomic swamping, which seems as valid and
potentially important as other forms of extinction, especially in humanaltered
environments8
.
Vol 446j15 March 2007jdoi:10.1038/nature05706
279
Nature©2007 Publishing Group
may often persist for millions of years after initial divergence18
.
Hybridization is thus a normal feature of species biology, if at a rather
extreme end of the natural spectrum of sexuality5
; it is not merely an
unnatural ‘‘breakdown of isolating mechanisms’’2
. At the population
level, interspecific hybrids are, of course, unusual, forming ,0.1% of
individuals in a typical population2,18
; they are also ‘hopeful monsters’,
with hefty differences from each parent, no adaptive history to
any ecological niche, and little apparent scope for survival (Fig. 1).
Furthermore, hybrids are often sterile or inviable owing to divergent
evolution in each species2,6
. Even if a healthy hybrid is formed, it
normally suffers ‘minority cytotype disadvantage’ because it encounters
few mates of its own type, and backcrosses to the more abundant
parent species will often be unfit. For example, a rare tetraploid
hybrid will produce unfit triploid progeny with diploid parents19
.
Yet hybrid species exist. What advantages could outweigh the
catalogue of difficulties? This innocent-sounding question plunges
to the heart of controversies about adaptive evolution. Is saltational
evolution possible? Are maladaptive intermediates and genetic drift
involved? Common sense and prevailing opinion suggests that evolution
normally occurs by small adjustments rather than saltation, and
rarely involves maladaption6
. It is therefore extraordinary that hybrid
speciation can disobey both rules. Hybridization (or hybridogenesis)
can act as a multi-locus ‘macro-mutation’ that reaches out over
large phenotypic distances5
to colonize unoccupied ecological niches
or adaptive peaks (Fig. 1). Furthermore, random drift in small, localized
hybrid populations provides a parsimonious solution to
maladaptation, to enable local establishment, stabilization and ultimate
spread20
.
Two principal types of hybrid speciation are treated here: allopolyploidy
and homoploid hybrid speciation.
Hybrid speciation through allopolyploidy
Polyploidy is a well-established speciation mode in plants, although
many aspects of polyploid evolution are only today being
revealed3,19,21,22
. Speciation can be via autopolyploidy (duplication
of chromosomes within a species) or allopolyploidy (duplication of
chromosomes in hybrids between species), although the boundary is
blurred because of the ‘fuzzy’ nature of species. Polyploid species are
reproductively isolated from their parents because when polyploids
mate with diploids, progeny with odd-numbered ploidies, such as
triploids, are produced. These offspring may be viable but typically
produce sterile gametes with unbalanced chromosomal complements
(aneuploidy)3,4,22
. Polyploidy is thus a simple saltational
means of achieving speciation4
. The process may be repeated many
times, leading to lineages with .80-fold ploidy in some vascular
plants; 40–70% of all plant species are polyploids3,21
.
Allopolyploid speciation can result from somatic chromosome
doubling in a diploid hybrid, followed by selfing to produce a tetraploid.
This was the route taken by Primula kewensis, the allopolyploid
that arose spontaneously in 1909 among cultivated diploid
hybrids of Primula verticillata and Primula floribunda22
. However,
there are other possibilities, such as fusion of two unreduced gametes
Polyploid hybrid
Homoploid hybrid
Phenotype
Species 2
Species 1
Figure 1 | Hybridization and the adaptive landscape. The hyperspace of
possible phenotypes and genotypes can be represented as an adaptive
landscape20
. Fitness optima (‘adaptive peaks’) are coloured blue. Adaptive
landscapes are not rigid, but are readily distorted by environmental or biotic
changes, including evolutionary change. Mean phenotypes of species and
their hybrids are shown as crosses, and offspring distributions as dots.
Species 1 and 2 are adapted to different fitness optima. Natural selection acts
mainly within each species, so hybrids are ‘hopeful monsters’, far from
phenotypic optima (solid arrows). It is therefore hard to imagine how
hybrids often attain new optima unless unoccupied adaptive peaks are
abundant. Polyploid hybrids can have a variety of advantages over their
parents, including heterozygote advantage, extreme phenotypic traits and
reproductive isolation. Genetic variation in their offspring will initially be
similar to that of non-hybrid parents if recombination between parental
genomes is rare4,22
; such hybrids will not spread unless already near an
optimum. Homoploid hybrids have fewer initial advantages, but their
progeny can have extremely high genetic variances via recombination,
including phenotypes more extreme than either parent—transgressive
variation (not shown here). This burst of variation can help homoploids
attain new adaptive peaks (dotted arrow) far from parental optima30,32
.
REVIEWS NATUREjVol 446j15 March 2007
280
Nature©2007 Publishing Group
after failure of reduction divisions in meiosis. A third route is the
‘triploid bridge’, in which rare, unreduced (diploid) gametes fuse
with normal haploid gametes to form triploids. Triploids are normally
sterile, but can contribute to tetraploid formation by themselves
producing occasional, unreduced triploid gametes that can backcross
with a normal haploid gamete to form tetraploid progeny19,22
.
This was the route used to engineer the first, and maybe the only,
synthetic (that is, in the laboratory), self-sustaining bisexual animal
polyploid strain, a hybrid between silk moths (Bombyx mori and
Bombyx mandarina)23
.
After polyploid hybrids arise, they still face major hurdles. Diploid
and triploid hybrids are strongly disfavoured because their aneuploid
gametes are almost always sterile. Even when even-numbered allopolyploidy
is achieved, chromosome pairing is rarely perfect22
.
Furthermore, assuming new polyploids are rare, they will mate
mostly with incompatible parentals, leading to minority cytotype
disadvantage19
. These problems almost certainly explain why bisexual
polyploid speciation is more common in plants than animals: (1)
plants usually have indeterminate growth, and somatic chromosome
doubling can lead to germline polyploidy; (2) plants are also often
perennial or temporarily clonal, allowing multigenerational persistence
of hybrid cell lines within which polyploid mutations can occur;
(3) plants are more often hermaphrodites, allowing selfing as a
means of sexual reproduction of rare polyploids, once formed; and
(4) gene flow is weaker in plants than in animals, and local populations
with unusual ploidy (whether by local drift or selection) can
form more readily to overcome minority cytotype disadvantage. As
expected, polyploidy is strongly associated with asexual reproduction,
selfing and perenniality in plants, as well as in animals11,21,23
. In
clonal polyploid animals, reversion to out-crossing is rare, whereas in
plants, with frequent alternation between clonal and sexual phases,
bisexual polyploid species are common and themselves often give
rise to further species. Thus, animal allopolyploids such as stick
insects (Bacillus) and freshwater snails (Bulinus truncatus) are often,
although not always, parthenogenetic or selfers21
. Muller’s theory24
that sex chromosomes in animals prevent sexual polyploidy owing to
sex:autosome gene dosage is no longer given much credence6,21,23,25
.
How common is polyploid speciation? Otto and Whitton21
provided
new insights from the over-representation of even-numbered
chromosome counts. Recent polyploidy explains ,2–4% of speciation
events in flowering plants and ,7% of speciation events in
ferns21
, and these are probably underestimates6
(40–70% of plant
species overall are polyploid, but this includes the effects of much
non-polyploid speciation within already polyploid lineages3,21
). In
animals, there is no bias towards even-numbered chromosomal
counts, suggesting that animal polyploid speciation is very rare compared
with other speciation modes21
.
Traditional dogma has it that allopolyploids arise more readily
than autopolyploids because the latter are more prone to chromosome
pairing problems in meiosis4
; however, this view is no longer
generally accepted. Newly arisen autopolyploids have levels of infertility
and aneuploid gametes comparable to those of allopolyploids22
.
Furthermore, many autopolyploids probably lie unrecognized by taxonomy
within diploid progenitor species6
. Yet these discoveries give
little insight into a more important question: what fraction of polyploids
that spread successfully are allopolyploids? Autopolyploids
may often be doomed to extinction, perhaps through competition
with similar diploid relatives6
. Opinions differ, but it probably
remains true that allopolyploids are more successful than autopoly-
ploids6,26
; certainly allopolyploids are a sizeable fraction of wellstudied
crop cases, such as wheat, cotton and tobacco7,26
. There are
almost no surveys of entire floras, although a small-scale survey in the
United States revealed that 79–96% of 28 polyploid species were allo-
polyploids4
. Recently, the Arctic flora was surveyed, in which about
50% of the often clonal or selfing species are polyploids27
. In Svalbard
(Spitsbergen), 78% of the 161 species are polyploid, with the average
level of ploidy approximately hexaploid. Every one of the 47 polyploid
species studied genetically shows fixed marker heterozygosity, implying
100% allopolyploidy27
. The Arctic is, of course, an extreme environment,
but this remains the most comprehensive survey so far. If
Svalbard is typical, most successful polyploids are also hybrid species.
After formation, novel allopolyploids face the usual ‘hopeful monster’
difficulties (Fig. 1). It helps if they can exploit a new ecological
niche that is both vacant and also spatially separated to ameliorate
minority cytotype disadvantage. For example, recent allopolyploid
hybrids between introduced and native plants have successfully
spread from sites of origin (for example, Senecio cambrensis in
Wales28
and Spartina anglica in England29
). These invasive allopolyploids
were able to exploit vacant ecological roles with relatively little
evolutionary change (Fig. 1).
Stochastic drift may also be necessary to overcome minority cytotype
and other disadvantages. A few polyploids, usually from the same
hybridization event, must accumulate locally for the process to take
off, probably involving chance or an unusual local selective regime.
Stochastic effects are evident in nature. An independently derived
Scottish population of S. cambrensis became extinct in Edinburgh
some 20 yr after being discovered28
. Of two origins, only the Welsh
population now survives. Other allopolyploid hybrids can arise
repeatedly from the same parents26
, but many widespread polyploids
(for example, S. anglica) probably originated only once or a few
times21,28,29
, even though parent species are in broad contact, again
showing the importance of chance in the origins of hybrid species.
Recombinational and homoploid hybrid speciation
Homoploid hybrid speciation or recombinational speciation is wellknown
in flowering plants4,7,30
. Speciation takes place in sympatry (by
definition, as hybridization requires gene flow). Hybrids must then
overcome chromosome and gene incompatibilities, while lacking
reproductive isolation via polyploidy. For these reasons, the process
is often considered unlikely5,6,31
.
However, hybridization can boost genetic variance30,32
, allowing
colonization of unexploited niches (Fig. 1). Suppose 1 and 2 alleles
at genes affecting a quantitative trait differ between species, so that
each has fixed differences (11122 and 22211, say). Recombination
can then liberate ‘transgressive’ quantitative variation32
,
often more extreme than either parent (for example, 22222
and 11111). Most early recombinants will be unfit, but extreme
hybrids can colonize niches unavailable to parents. If ecological
opportunities are partially separated from the parental habitat, if like
hybrids tend to associate (for example, by means of seasonality or
drift in small populations), or if selfing or inbreeding is common,
gene flow between hybrids and parents will be reduced and hybrid
speciation becomes more likely7,32
. Successful hybrid species might
also displace one or more parent species ecologically30
, and obliterate
evidence of their own hybrid origin.
In plants, about 20 well-established homoploid hybrid species are
known31,33
, but they are hard to detect and may be more prevalent.
The best documented are the desert sunflowers Helianthus anomalus,
Helianthus deserticola and Helianthus paradoxus, which all derive from
hybrids between mesic-adapted Helianthus annuus and Helianthus
petiolaris30,33
. Selfing is rare and provides little assistance to establishment,
but the three hybrid species survive drought better than their
parents, suggesting recruitment of hybrid transgressive variation.
Synthetic hybrid populations are readily recreated with karyotypic
combinations like those in wild hybrid species, because selection
repeatedly favours similar combinations of compatible chromosomal
rearrangements. In addition, the wild contribution from each parent
of extreme adaptive traits for morphology, physiology and life history
of the hybrids (for example, small leaf size, seed dormancy, or tolerance
of drought and salt) matches experimental predictions32,33
. In
Helianthus, recombinant genotypes and spatial separation have
enabled the hybrids to flourish where their parents are absent.
Although bisexual polyploids are often barred in animals, there is
no reason why homoploid hybrid species would be rarer in animals
NATUREjVol 446j15 March 2007 REVIEWS
281
Nature©2007 Publishing Group
than in plants. The number of cases in animals is growing rapidly10,33
.
A recent example is the invasive sculpin, a hybrid fish derived from
the Cottus gobio group from the Scheldt River (compare Cottus perifretum)
and upper tributaries of the Rhine (compare Cottus rhenanum).
Sculpins are normally restricted to clear, well-oxygenated cold
waters in upper river tributaries across Europe. The Rhine and
Scheldt rivers became connected as a result of earlier canal building,
but invasive sculpins appeared in the warmer and muddier lower
Rhine only in the past fifteen years. Morphologically, the invasive
sculpin is intermediate, and its mitochondrial DNA, as well as nuclear
single nucleotide polymorphisms and microsatellites, are characteristic
of both Scheldt and Rhine forms34
. The hybrid form meets
upper Rhine sculpins in narrow hybrid zones, but remains distinct
despite gene flow, suggesting that it is adapted mainly to the lower
Rhine. Recent evolution and spread of the invasive sculpin, as well as
intermediacy, provides convincing evidence of adaptive hybrid origin.
A more ancient example is the cyprinid fish Gila seminuda, which
inhabits the Virgin River, a tributary of the Colorado River (USA).
This hybrid species contacts but does not overlap its parent species,
Gila robusta and Gila elegans, from the Colorado River. Gila seminuda
is morphologically and genetically intermediate, and similar to
synthetic hybrids. Intermediacy may allow it to out-compete both
parents in the Virgin River35
. A similar case, with well-documented
genetic intermediacy, is an unnamed form of the butterfly genus
Lycaeides. This homoploid hybrid species uses a different host plant
and inhabits high-elevation alpine habitats unoccupied by either
parent36
.
Rhagoletis fruitflies provide another historically documented
example. In 1997 flies were first found on introduced honeysuckle
(Lonicera spp.). Molecular markers in the fly are a blend from the two
parents: the blueberry maggot Rhagoletis mendax and the snowberry
maggot Rhagoletis zephyria37
. No F1 genotypes are detected between
the hybrid form and its parents where they overlap; the new fly is
reproductively isolated. Rhagoletis flies mate on host fruits, and host
choice ensures mating specificity. A different route to hybrid speciation
can be inferred from the Colombian butterfly Heliconius
heurippa38
. This form has a colour pattern like that of synthetic
hybrids between local Heliconius cydno and Heliconius melpomene.
Microsatellite alleles are shared across all three species, but H. heurippa
forms an allele frequency cluster distinguishable from either
parent. The hybrid wing coloration of H. heurippa is a cue in mating
discrimination, and directly causes reproductive isolation from both
parents. Similarly, in the fish Xiphophorus clemenciae, the ‘swordtail’,
a hybridization-derived trait, is involved in sexual selection and mate
choice39
and may be related to its speciation.
Hybrid speciation in animals is supported so far only by lowresolution
molecular data. Genomic mapping of ecological or speciation-related
hybrid traits, which so strongly supports hybrid speciation
in Helianthus, is not yet available for any animal case. Many
homoploid hybrid species fail to overlap with at least one parent
species, and reproductive isolation is weak, so species status could
be questioned (the Lonicera-feeding Rhagoletis is an exception).
Nonetheless, in the cases surveyed here, hybrid traits often contribute
strongly to maintenance or ecological expansion of the new form.
Is hybrid speciation important in evolution?
There are now many examples of hybrid species. We know that polyploidy
is common in plants, giving rise to $2–7% of vascular plant
species, but rarer in animals. Furthermore, ancient polyploidy has
been found at the root of many plant and animal groups. Genome
duplications probably facilitated the evolution of complex organisms
(although this is debated)21
, and we can infer that successful genome
duplications were mostly allopolyploid, provided that limited plant
community data are reliable4,27
. Hybridization would then be a catalyst
not only for speciation but also for major evolutionary innovations.
Polyploid speciation leaves a clear genomic signature, but we have
little idea how common homoploid hybrid species are. They could be
abundant: most speciation involves natural selection6
; natural selection
requires genetic variation; genetic variation is enhanced by
hybridization12
; and hybridization and introgression between species
is a regular occurrence, especially in rapidly radiating groups9,12,18
.
Enough suspected animal homoploid hybrid species exist to indicate
that it may be at least as common as in plants, in contrast to the
situation for polyploidy, where a variety of traits prevent its occurrence
(see above). It now seems intuitively unlikely that all biodiversity
arose as a result of recombination of existing diversity1
,
but homoploid hybrid species might still represent a large fraction.
Nonetheless, there are few convincing cases, probably, in part, because
of the difficulty of demonstrating that hybridization has led to speciation.
We clearly need more genomic analyses. As for hybrid species
as a whole, we have observed recent speciation in the laboratory or
nature in seven genera discussed here (Helianthus, Senecio, Primula,
Spartina, Rhagoletis, Bombyx and Cottus), and there are many other
cases. It would be hard to find another mode of speciation so readily
documented historically and so amenable to experimentation.
That hybrid species exist at all reveals something perhaps unexpected
about adaptive landscapes. If hybrid ‘hopeful monsters’,
with all their problems, are ever to survive in competition with their
parents, they must be able to hit (and for polyploid species, hit almost
exactly) new adaptive combinations of genes (Fig. 1). This implies
both that many adaptive peaks are scattered about in the adaptive
landscape, and also that many are unoccupied. Liberal adaptive landscapes
are further supported by the successes of many introduced
species, and by fossil evidence: for insects, angiosperms and many
other groups, diversity seems to have been increasing more or less
continuously over geological time40
.
The ability of hybrid species to invade hitherto unoccupied niches
also means that hybridization can contribute to adaptive radiations
such as African cichlid fish and Darwin’s finches7,9,12
. This principle
is well demonstrated by the ‘domestication niche’. Humans have
unwittingly created many allopolyploid and other hybrid crops and
domestic animals while selecting for transgressively high yields4,7
.
Even our own species may have a hybrid genomic ancestry41,42
,
although this is contested43
. Whichever way the debate about humans
is resolved, it would be hardly surprising if hybridization was one
trigger for the origin of Homo sapiens, the most invasive mammal on
the planet42
.
1. Lotsy, J. P. Evolution by Means of Hybridization (Martinus Nijhoff, The Hague, 1916).
2. Mayr, E. Animal Species and Evolution (Harvard Univ. Press, Cambridge,
Massachusetts, 1963).
3. Stebbins, G. L. Processes of Organic Evolution (Prentice-Hall, Englewood Cliffs, New
Jersey, 1971).
4. Grant, V. Plant Speciation (Columbia Univ. Press, New York, 1981).
5. Barton, N. H. The role of hybridization in evolution. Mol. Ecol. 10, 551–568 (2001).
Medline
6. Coyne, J. A. & Orr, H. A. Speciation (Sinauer Associates, Sunderland,
Massachusetts, 2004).
7. Anderson, E. & Stebbins, G. L. Hybridization as an evolutionary stimulus. Evolution
8, 378–388 (1954).
8. Arnold, M. L. Natural Hybridization and Evolution (Oxford Univ. Press, Oxford,
1997).
9. Seehausen, O. Hybridization and adaptive radiation. Trends Ecol. Evol. 19, 198–207
(2004).
10. Dowling, T. E. & Secor, C. L. The role of hybridization and introgression in the
diversification of animals. Annu. Rev. Ecol. Syst. 28, 593–620 (1997).
11. Bullini, L. Origin and evolution of animal hybrid species. Trends Ecol. Evol. 9,
422–426 (1994).
12. Grant, P. R., Grant, B. R. & Petren, K. Hybridization in the recent past. Am. Nat. 166,
56–57 (2005).
13. Mallet, J. A species definition for the modern synthesis. Trends Ecol. Evol. 10,
294–299 (1995).
14. Butlin, R. Speciation by reinforcement. Trends Ecol. Evol. 2, 8–12 (1987).
15. Ortı´z-Barrientos, D., Counterman, B. A. & Noor, M. A. F. The genetics of speciation
by reinforcement. PLoS Biol. 2, e416 (2004).
16. Tunner, H. G. & Nopp, H. Heterosis in the common European water frog.
Naturwissenschaften 66, 268–269 (1979).
17. Fisher, R. A. The Genetical Theory of Natural Selection (Clarendon Press, Oxford,
1930).
REVIEWS NATUREjVol 446j15 March 2007
282
Nature©2007 Publishing Group
18. Mallet, J. Hybridization as an invasion of the genome. Trends Ecol. Evol. 20,
229–237 (2005).
19. Husband, B. C. Constraints on polyploid evolution: a test of the minority cytotype
exclusion principle. Proc. R. Soc. Lond. B 267, 217–223 (2000).
20. Wright, S. The roles of mutation, inbreeding, crossbreeding and selection in
evolution. Proc. XI Int. Congr. Genet. Hague 1, 356–366 (1932).
21. Otto, S. P. & Whitton, J. Polyploid incidence and evolution. Annu. Rev. Genet. 34,
401–437 (2000).
22. Ramsey, J. & Schemske, D. W. Neopolyploidy in flowering plants. Annu. Rev. Ecol.
Syst. 33, 589–639 (2002).
23. Astaurov, B. L. Experimental polyploidy in animals. Annu. Rev. Genet. 3, 99–126
(1969).
24. Muller, H. J. Why polyploidy is rarer in animals than in plants. Am. Nat. 59,
346–353 (1925).
25. Mable, B. K. ‘Why polyploidy is rarer in animals than in plants’: myths and
mechanisms. Biol. J. Linn. Soc. 82, 453–466 (2004).
26. Soltis, D. E., Soltis, P. S. & Tate, J. A. Advances in the study of polyploidy since plant
speciation. New Phytol. 161, 173–191 (2004).
27. Brochmann, C. et al. Polyploidy in arctic plants. Biol. J. Linn. Soc. 82, 521–536
(2004).
28. Abbott, R. J. & Lowe, A. J. Origins, establishment and evolution of new polyploid
species: Senecio cambrensis and S. eboracensis in the British Isles. Biol. J. Linn. Soc.
82, 467–474 (2004).
29. Ainouche, M. L., Baumel, A. & Salmon, A. Spartina anglica C. E. Hubbard: a natural
model system for analysing early evolutionary changes that affect allopolyploid
genomes. Biol. J. Linn. Soc. 82, 475–484 (2004).
30. Buerkle, C. A., Morris, R. J., Asmussen, M. A. & Rieseberg, L. H. The likelihood of
homoploid hybrid speciation. Heredity 84, 441–451 (2000).
31. Rieseberg, L. H. Hybrid origins of plant species. Annu. Rev. Ecol. Syst. 28, 359–389
(1997).
32. Rieseberg, L. H., Raymond, O., Rosenthal, D. M., Lai, Z. & Livingstone, K. Major
ecological transitions in wild sunflowers facilitated by hybridization. Science 301,
1211–1216 (2003).
33. Gross, B. L. & Rieseberg, L. H. The ecological genetics of homoploid hybrid
speciation. J. Hered. 96, 241–252 (2005).
34. Nolte, A. W., Freyhof, J., Stemshorn, K. C. & Tautz, D. An invasive lineage of
sculpins, Cottus sp. (Pisces, Teleostei) in the Rhine with new habitat adaptations
has originated from hybridization between old phylogeographic groups. Proc. R.
Soc. Lond. B 272, 2379–2387 (2005).
35. DeMarais, B. D., Dowling, T. E., Douglas, M. E., Minckley, W. L. & Marsh, P. C.
Origin of Gila seminuda (Teleostei: Cyprinidae) through introgressive
hybridization: implications for evolution and conservation. Proc. Natl Acad. Sci.
USA 89, 2747–2751 (1992).
36. Gompert, Z., Fordyce, J. A., Forister, M., Shapiro, A. M. & Nice, C. C. Homoploid
hybrid speciation in an extreme habitat. Science 314, 1923–1925 (2006).
37. Schwarz, D., Matta, B. M., Shakir-Botteri, N. L. & McPheron, B. A. Host shift to an
invasive plant triggers rapid animal hybrid speciation. Nature 436, 546–549
(2005).
38. Mava´rez, J. et al. Speciation by hybridization in Heliconius butterflies. Nature 441,
868–871 (2006).
39. Meyer, A., Salzburger, W. & Schartl, M. Hybrid origin of a swordtail species
(Teleostei: Xiphophorus clemenciae) driven by sexual selection. Mol. Ecol. 15,
721–730 (2006).
40. Labandeira, C. C. & Sepkoski, J. J. Insect diversity in the fossil record. Science 261,
310–315 (1993).
41. Patterson, N., Richter, D. J., Gnerre, S., Lander, E. S. & Reich, D. Genetic evidence
for complex speciation of humans and chimpanzees. Nature 441, 1103–1108
(2006).
42. Evans, P. D., Mekel-Bobrov, N., Vallender, E. J., Hudson, R. R. & Lahn, B. T. Evidence
that the adaptive allele of the brain size gene microcephalin introgressed into
Homo sapiens from an archaic Homo lineage. Proc. Natl Acad. Sci. USA 103,
18178–18183 (2006).
43. Barton, N. H. How did the human species form? Curr. Biol. 16, R647–R650
(2006).
44. Seehausen, O., van Alphen, J. J. M. & Witte, F. Cichlid fish diversity
threatened by eutrophication that curbs sexual selection. Science 277, 1808–1811
(1997).
45. Grant, B. R. & Grant, P. R. High survival of Darwin’s finch hybrids — effects of beak
morphology and diets. Ecology 77, 500–509 (1996).
Acknowledgements I thank C. Brochmann, K. Dasmahapatra, B. Husband,
S. Knapp, M. Linares, J. Mava´rez, A. Meyer, P. Nosil, S. Otto, C. Salazar and S. Turelli
for discussions and comments. The work was supported, in part, by grants from
NERC and the DEFRA Darwin Initiative programme.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The author declares no competing financial interests.
Correspondence should be addressed to J.M. (j.mallet@ucl.ac.uk).
NATUREjVol 446j15 March 2007 REVIEWS
283
Nature©2007 Publishing Group