.

DOI 10.1002/bies

1) Department of OrgaBiology, Harvard UnUSA

2) Department of GeneEnvironment, UniverLondon, UK

3) Department of BioloInstitute for Global HDame, Notre Dame,

4) Department of BioloInformatics and ComBloomington, IN, US

*Corresponding authJames MalletE-mail: jmallet@oeb.ha

140 www.bioessa

Thinkagain

Insights & Perspectives

How reticulated are species?

James Mallet1)2)*, Nora Besansky3) and Matthew W. Hahn4)

Many groups of closely related species have reticulate phylogenies. Recent

genomic analyses are showing this in many insects and vertebrates, as well as

in microbes and plants. In microbes, lateral gene transfer is the dominant

process that spoils strictly tree-like phylogenies, but in multicellular eukaryotes

hybridization and introgression among related species is probably more

important. Because many species, including the ancestors of ancient major

lineages, seem to evolve rapidly in adaptive radiations, some sexual

compatibility may exist among them. Introgression and reticulation can thereby

affect all parts of the tree of life, not just the recent species at the tips. Our

understanding of adaptive evolution, speciation, phylogenetics, and compar-

ative biology must adapt to these mostly recent findings. Introgression has

important practical implications as well, not least for the management of

genetically modified organisms in pest and disease control.

admixture; homoplasy; introgression;

species concepts; tree of life

Keywords:

phylogenetic discordance; speciation;

Introduction

Not so long ago, analysis of microbial16S ribosomal RNA sequences led to arevolutionary new “Universal Tree ofLife,” consisting of three monophyleticdomains, here referred to as the Bacte-ria, the Archaea, and the Eukarya oreukaryotes [1, 2]. Yet almost as soon asthe new system was established, this

.201500149

nismic and Evolutionaryiversity, Cambridge, MA,

tics, Evolution andsity College London,

gical Sciences and Eckealth, University of NotreIN, USAgy and School ofputing, Indiana University,A

or:

rvard.edu

ys-journal.com Bioessaysopen accedistribution

tidy tree picture was threatened:sequencing of more microbial genesand then whole genomes quickly led toan understanding of the importance ofhorizontal or lateral gene transfer, theincorporation of foreign genes into thegenome. Some of the major transitionsin evolution were clearly due to lateraltransfer: the eukaryotes were formed byendosymbiosis of a-proteobacteria withArchaea to form the eukaryotes. Later,endosymbiosis of cyanobacteria witheukaryotes led to green algae andplants. Many other gene transferstogether with multiple other endosym-bioses have been inferred. Microbiolo-gists began to argue that the “tree” oflife wasmore like a web or network thana tree [3–5].

Today, whole genome sequencing isproviding unprecedented phylogeneticinformation about whole groups ofeukaryotes [6–14]. Here we reviewgenomic evidence suggesting that retic-ulate evolution may have considerableimpact in multicellular eukaryotes as

38: 140–149,� 2015 The Authors. BioEssays Puss article under the terms of the Creative Commoand reproduction in any medium, provided the o

well as microbes. Reproductively iso-lated species and bifurcating phyloge-nies have become an important basis forour understanding of evolution; nowthis bedrock seems threatened. As anideal, species are often taken to beevolutionarily independent populationsthat are reproductively isolated fromother such species, for example in the“biological species concept,” althoughit was always known that hybridizationdoes occur [15]. Reticulate evolution inplants has long been recognized [16],but recent genomic evidence fromanimals suggest that reticulation mightbe much more common than antici-pated [17, 18]. Given abundant newdata, it is time to enquire whether amajor shift in our understanding ofspecies, speciation, and phylogeneticsis taking place.

Prokaryotes: Is there auniversal tree of life?

Tree-like relationships among speciesarise because the genomeevolves within cells. When a celldivides, copies of the same genomeare found in each daughter cell. Ulti-mately, after populations of organismsdiverge or “speciate,” evolution alongeach branch will leave genomic signalsof that branching event in daughterlineages. Sex and recombination canobscure this picture, but in bothBacteria and Archaea sex (in theeukaryote-like sense of homologousgene exchange) is mostly a transactionbetween closely related individuals,mostly within the same populations or“species” [19–22]. Eukaryotes are simi-lar [23, 24]. Lateral transfer involvingnon-homologous exchange, on the

blished by WILEY Periodicals, Inc. This is anns Attribution License, which permits use,riginal work is properly cited.

..... Insights & Perspectives J. Mallet et al.Thinkagain

other hand, will lead to more wide-ranging phylogenetic discordance. Inprokaryotes, both sex and lateral trans-fer involve relatively few genes at a timeor even if more extensive, usually muchless than 50% of the genome. Neverthe-less, multiple exchanges may takeplace, and very large fractions ofthe genome might eventually beexchanged with other lineages or spe-cies over long periods. If so, it is possiblethat the signals of the organismicgenealogy (the original “tree of cells”)in the genome will be obliterated bymultiple phylogenetic signals from sexand lateral transfer.

Before assessing new genomic evi-dence for phylogenetic discordance inmulticellular eukaryotes, it is worthreviewing the controversy raging aboutthe microbial “Tree of Life” over the lastfew decades. Carl Woese [25] arguedthat in spite of considerable lateraltransfer, there is “a genealogy-definingcore of genes whose common historydates back to the root of the universaltree.” Woese suggested that the acqui-sition of sufficient co-adaptation amongthese key genes caused life to reach a“Darwinian threshold,” which permit-ted divergence into separate species andallowed us to trace the organismalhistory, even while lateral transferobscures the universal tree for manyother genes. According to Woese, beforethe Darwinian threshold was reached,divergence and speciation could nottake place, and no tree of genes wouldallow us to trace the organismal history.

It quickly became apparent thatlateral transfer does indeed swamp thesignal of the Universal Tree in micro-bial genomes: in fact no other genessupport Woese’s original 16S RNAtree [26]. Many microbiologists nowdeny a tree-like phylogeny of micro-bial evolution; instead the phylogenyof life looks more like a web or a ring[3, 27–29]. By excluding all genes thatdisagree with the Universal Tree, onecan select 20–30 largely informationalgenes that more or less rescue theribosomal RNA Tree [29–31]. But thisalmost seems like cheating, and isitself obtained only by pruning out anumber of clear cases of lateraltransfer in even these genes. As thisanyway only applies to a tiny fractionof the genome, these recent incarna-tions of the Universal Tree have been

Bioessays 38: 140–149,� 2015 The Authors. B

derided as “the tree of one percent” [32]. Around 80% of eukaryoticproteins are actually more closelyrelated to homologs in the Bacteriathan in the Archaea; the UniversalTree’s closer archaeal-eukaryote affin-ity is reflected in only about 15% ofeukaryote proteins [28, 32], includingthose used by Ciccarelli et al. [30].Because of concerns such as these, theexistence of species and of the Univer-sal Tree in microbes has been dis-missed as a “myth” in the prokaryoteliterature [33]. Whether species or theUniversal Tree exist in prokaryotes hasbecome almost a philosophical ratherthan a biological issue [29], but it doesseem clear that most of the originalUniversal Tree, whether identifiable ornot, is located on the far side of whatWoese originally intended by theDarwinian Threshold.

What causes phylogeneticincongruence ineukaryotes?

Findings of promiscuous gene exchangeamong prokaryotes have usuallybeen contrasted with supposedly well-behaved trees in eukaryotes [33, 34].Eukaryote genomes originated when anarchaeal cell acquired many bacterialgenes, in part but certainly not onlyassociated with the bacterial endosym-biotic origins of mitochondria andchloroplasts [35]. Eukaryotes alsoinvented meiosis, which allows recom-bination of whole genomes. In multicel-lular eukaryotes, reproduction itselfoften involves meiosis. This innovationeffectively destroys the tree-like signalin an organismal (“tree of cells”)phylogeny. In every meiosis recombi-nant haploid genomes from two suc-cessful, independent cells are throwntogether to form diploid zygotes, beforethe sum of the genetic material ishaphazardly and approximately equallyrecombined into haploid daughter cells.A “tree of cells” justification for theeukaryote Tree of Life is no longerpossible.

While tree-like patterns are readilydiscernible in eukaryote phylogenies,we here highlight recent evidencesuggesting that a number of regions ofthe eukaryotic tree show similar

ioEssays Published by WILEY Periodicals, Inc.

pathologies to those found in prokar-yotes. This raises doubt about theeukaryotic Tree of Life as a whole. Apartfrom phylogenetic estimation error andhomoplasy, there are three main causesof phylogenetic incongruence: lateralgene transfer, incomplete lineage sort-ing, and introgression.

Lateral transfer

In Eukaryotes, lateral or horizontalgene transfer is widespread, but isusually thought to be rare comparedto that in prokaryotes [8, 36, 37]. Itseems to be associated mainly withsingle-celled eukaryotes (the “pro-tists”), especially those that engulftheir food, or in multicellular organ-isms with parasites in close cellularcontact with their hosts. Eukaryotesclearly seem to have acquired impor-tant genes via lateral transfer from bothmitochondrial and chloroplast endo-symbionts, but transfers also originatefrom other endosymbionts, parasites,and close associates [35]. Lateral pro-cesses in eukaryotes, in contrast toother possible causes of reticulation,may transfer genes between distantlyrelated species, but typically involverelatively few genes at a time, as inprokaryotes. Lateral transfer is com-mon in some multicellular groups [36],such as bdelloid rotifers, which, inter-estingly, lack meiotic sex [38, 39].Horizontal gene transfer in the mito-chondrial genomes of plants and yeastsis also widespread [40]. However,horizontal transfer is probably not anoverriding factor in the evolution of thenuclear protein coding genes of mostmulticellular eukaryotes, unlike thoseof prokaryotes.

In contrast to the genes, eukaryoticgenomes often consist largely of non-coding DNA, and 30–60% of thisconsists of recognizable mobile ele-ments [41, 42]. Intergenic and intronicDNA is thought to originate largely viaactive or inactivated mobile geneticelements [43–45], most of which arethought to enter lineages via lateraltransfer [46]. Mobile elements are par-ticularly likely to be important in theevolution and spread of regulatoryelements. Nonetheless, the introductionof new mobile elements via lateraltransfer is rare, and the lifespan of

141

J. Mallet et al. Insights & Perspectives.....Thinkagain

active proliferation via transposition iscut short by relatively rapid loss,inactivation and sequence degradationin the host genome [46].

Incomplete lineage sorting

The two main causes of gene tree –species tree discordance, at least forprotein-coding genes in closely relatedgroups of eukaryotes, are incompletelineage sorting and introgression. In-complete lineage sorting occurs whenpolymorphisms persist between specia-tion events, so that the actual (true)genealogical relationship of a gene orgenome region differs from the truespecies branching pattern. As an exam-ple of incomplete lineage sorting,around 15% of human genes are moreclosely related to homologs in gorillasthan to those in our true sister lineage,the chimpanzees, while another 15% ofgenes group gorilla and chimpanzee.This is expected from what we knowabout the ancestral effective populationsizes of these species and the short timebetween human-gorilla and human-chimpanzee speciation events [47, 48].

In some cases, incomplete lineagesorting occurs as a result of balancingselection maintaining polymorphisms:when speciation occurs, both daughterspecies may maintain the same “trans-specific” polymorphisms, even thoughwith recombination, the signal of an-cestral origin may erode over time [49].Good examples of shared polymor-phisms between humans and apes areMHC [50] and ABO blood group loci [51],among other genes. In the speciescomplex including the major mosquitovector, Anopheles gambiae, a very largechromosomal inversion, 2La (22Mb inlength, 8.5% of the total genome size) ismaintained as a balanced polymor-phism that has persisted across severalspeciation events [18].

Unlike lateral transfer and introgres-sion, however, discordance created byincomplete lineage sorting does notimply phylogenetic reticulation at thelevel of species. It merely muddles thegenomic signal of what might be a trulybifurcating phylogeny. In some treeswith four or more taxa and rapidsuccessive speciation events (the“anomaly zone” of phylogenetics),the species tree estimated from the gene

142 B

trees has been shown to converge on anincorrect but highly significant solu-tion [52, 53]. In spite of this “tyranny ofthemajority” in phylogenetic analysis, acoalescent-based analysis should none-theless be able to retrieve the truebifurcation signal in spite of the con-fused gene tree signal [54, 55].

Introgression and reticulatedevolution

The third source of phylogenetic incon-gruence, introgression, occurs whenhybrids backcross and transfer geneticmaterial between species. Hybridizationmay occur without strongly affectingthe genomes of recipient populations ifstrongly resisted by selection, butgenomic admixture results if the intro-gressed alleles are established.

Hybridization between related eu-karyote species does occur reasonablyfrequently in nature; it is known to affectaround 25% of the species of floweringplants andabout 10%of animals [56–58].The fraction of hybrids in natural pop-ulations, nevertheless, is usually low:natural interspecific hybridization ratesin animals are typically 0.1% or less pergeneration in any species [57, 59]. Pergeneration hybridization rates can bemuch higher in some populations ofplants and animals, where it reachesseveral per cent, for example in someoaks (Quercus), Darwin’s finches, andsome cases in Heliconius butter-flies [60–63]; but these are probablyexceptional. While some hybrids aresterile, a substantial fraction of suchhybrids are at least partly fertile, leadingto observed cases of backcrossing andintrogression. It is important to realizethat hybridization and introgressionmayoccur amongnon-sister species aswell asbetween sister species, especially duringrapid adaptive radiations.

Closely related species hybridizemore readily than more distant spe-cies [64]. The decline of natural hybrid-ization rates with genetic distance, whilenoisy,maybeveryroughlyapproximatedas exponential [59], mirroring the noisydecline of compatibility inmeta-analysesof transformation experiments in prokar-yotes and laboratory crosses in animalsand plants [19–23, 64–67]. Thus, intro-gression tends to generate phylogeneticdiscordance mainly among closely

ioessays 38: 140–149,� 2015 The Authors. Bio

related groups of species, unlike lateralgene transfer. This is a major differencebetween reticulate evolution in prokar-yotes and eukaryotes: while lateral genetransfer weaves lineages together acrossdisparate parts of a tree, introgressionmerely results in tangled knots on a localscale. Nonetheless, introgression haspotentially important effects throughoutthe tree of life by obscuring relationshipsamonglineages thatdiversifiedrapidlyatany time, not just in those that did sorecently.

Introgression was well known be-fore the advent of genetic markers orgenomics, and was long believed animportant catalyst for adaptive evolu-tionary change in plants [16]. Introgres-sion was thus familiar by the 1960s, butideas of “coadapted gene complexes,”and “the unity of the genotype” associ-ated with the biological species conceptled to a belief that hybridization hadlittle importance in animals, at least.When hybridization did occur, it wasoften assumed to be unnatural andwas attributed to environmentalchanges wrought by humans [68].Because hybrids are mostly unfit, itwas assumed that introgression amonganimal species very rarely had any long-term evolutionary impact [15].

With the potential for introgression,not only will individual gene trees telldifferent stories, but the actual organis-mal branching pattern between specieswill be reticulate, rather than strictlybifurcating. The true phylogeny may beapproximately tree-like if introgressionis rare and affects only a very smallfraction of the genome, but will not betree-like if introgression is common.However, the importance of introgres-sion is only now becoming apparentwith rapid genome sequencing.

Gene transfer is importantin eukaryote genomes

The extent of introgressionacross the eukaryote tree

As we have seen, meiotic fertility has anincreasing tendency to fail with geneticdistance, but failure is often not com-plete in the closest hybrids. For thisreason, introgression, which requiressome fertile hybrid offspring, is mostlikely to occur among closely related

Essays Published by WILEY Periodicals, Inc.

Figure 1. A: “Whole genome” versus “species” tree topologies of the Anopheles gambiaecomplex in Africa. B: The tree based on the X chromosome only, showing introgressionevents and estimated node divergence times. The average phylogeny of the whole genomeis distorted by autosomal introgression between A. gambiaeþ coluzzii and A. arabiensis, butthis was prevented on the X chromosome by X-linked hybrid incompatibilities and multipleoverlapping inversions that prevent recombination. Modified and reprinted from [18] withpermission from AAAS.

..... Insights & Perspectives J. Mallet et al.Thinkagain

species. Hybridization between sisterspecies will not usually affect thespecies tree topology, but will makethe apparent divergence time betweenthe species appear more recent [69].However, if two widely distributedspecies interact in populations wherethey overlap, it may be possible thatindividual populations become on aver-age more closely related locally to asister species than to more distantconspecific populations. In contrast,hybridization and introgression amongnon-sister lineages can readily distortthe species tree topology. If introgres-sion between non-sister lineages iswidespread across the genome, it maybe very hard to retrieve the truebifurcation history of the species. Thisis because a unitary history of thegenome may not exist; if inferred frommultiple loci or whole genomes, thisspecies tree may be meaningless ormisleading. Here we discuss severalrecent examples from multicellulareukaryotes where this may have oc-curred. Interestingly, most of theseexamples come from rapid speciesradiations; these are exactly the casesin which closely related but non-sisterspecies may be hybridizing.

The group of eight African mosquitospecies known as the A. gambiaecomplex radiated within the last 2million years [18]. Species distributionsoverlap extensively, and in areas ofsympatry hybrids have been recorded atrates of �0.02–0.75% [70, 71]. DespiteF1 hybrid male sterility in most cases,introgression is plausible through thebackcrossing of vigorous and fertile F1hybrid females. When genomes ofmultiple members of the A. gambiaecomplex were sequenced and com-pared, the inferred species tree wasevident in only 2% of the genome,mainly on the X chromosome, whereasthe majority tree in the rest of thegenome yields a completely contradic-tory tree [18]. While some of thesedifferences are due to incomplete line-age sorting, much of this discordance isdue to introgression between two non-sister species (Fig. 1). This is particularlyclear for the 2La inversion mentionedabove, which is inferred to have beenpolymorphic in the ancestor of thecomplex, but is affected by three lossesof 2Lþ and one of 2La, as well as onefairly recent (1 Mya) introgression of 2La

Bioessays 38: 140–149,� 2015 The Authors. B

from A. gambiae to A. arabiensis [18].Introgression is on-going, and is anexcellent explanation for the phyloge-netic discordance, because wild hybridsand backcrosses between the latter twospecies are �0.22% of the individualscaptured in sympatry [71, 72]. In decid-ing between conflicting topologies, thespecies tree was inferred from regions ofthe genome with the deepest coales-cence times between species [18]. If thisinformation had not been available, or ifintrogression had been even morecomplex, it would have been hard toinfer the species tree at all.

In Heliconius butterflies, the “mel-pomene-silvaniform” clade consists ofaround 15 species. Most of these are“good” species that co-occur over largesympatric regions, and are somewhatinterfertile with other members of theclade. However, rare hybrids and back-crosses are known from the wild and incaptivity across this whole group,suggesting the possibility that a slowtrickle of introgression is constantlyoccurring among the largely sympatricspecies in the group [59]. This sugges-tion has now been confirmed: becauseof introgression, a local population ofH. melpomene can be more closelyrelated to the locally overlapping popu-lation of its sister H. cydno than it isto conspecifics at over 40% of thegenome [17, 73].

Rapidradiationssuchas these tendtoproduce many closely related speciesthat may be partially interfertile. For

ioEssays Published by WILEY Periodicals, Inc.

example, per generation hybridizationrates among closely related species ofDarwin’s finches can be as much as 6%,with high fertility of hybrids. The Dar-win’s finches began to diversify on theCocos and Galapagos Islands less than 1million years ago, and there is stronggenomic evidence for past and continu-ing introgression across almost the entiregroup [74]. Other vertebrate groups suchas African lake cichlids, Xiphophorusfishes, horses, and even hominins showsimilar phylogenetic discordance in-ferred to be due to introgression [75–78].

Much deeper evidence of reticulateevolutionary patterns also exists. Forexample, there is considerable phyloge-netic discordance at the base of theNeoaves, ormodern birds [79–81]. In factnone of the thousands of individual genetrees support the various conflictingestimates of the species tree [79, 81].Trees built from indels and stable mobileelement insertions (which are less proneto homoplasy than nucleotide or aminoacid substitutions) show similar conflict,suggesting that thegene treediscordanceis real, rather than due to phylogeneticerror [79]. The authors of these papersargued that the tangle at the base of thisancient radiation was due to incompletelineage sorting, but did not address thepossibilityof introgression.Yet introgres-sion seems a likely additional cause:around 9% of today’s bird species areknown to hybridize in the wild [56], andbirds retain some hybrid compatibilitywith congeners for �10 My after

143

J. Mallet et al. Insights & Perspectives.....Thinkagain

speciation [65]. After the demise of thedinosaurs, the early Neoaves had fewcompetitors, and it isnotunlikely that thefirst species in today’s lineageswere ableto hybridize with one another duringtheir global ecological diversification,much as the Darwin’s finches do todayon the Galapagos Islands [74]. Anexplanation for the strong signals ofdiscordance at the base of the Neo-aves [79, 81] may therefore lie partly ingene flow among the lineages after theydiversified. Given that other majorgroups, such as the placental mam-mals [13], or the animals as a whole [82],appear also to have evolved in rapidradiations, it seems likely that ourpersistent problems with estimation oftrees for the deepest branches of theseradiations is due to historical introgres-sionaswell as incomplete lineage sortingduring their initial diversification.

Is introgression adaptive?

Phylogenetic or genealogical studies ofthe extent of introgression across thegenome do not, however, revealwhether the process is largely neutralor whether it is aided by a selectiveadvantage on the new genomic back-ground. The relative importance ofselection in introgression across thegenome is still not known, and is anarea of active research [83], but manyintrogression events are now known tohave involved adaptation. A number oftransfers of mimicry-determining locihave been documented in Heliconiusbutterflies (Fig. 2A and B), and inAnopheles the many cases of insecticideresistance alleles crossing speciesboundaries (see below) and the exis-tence of balancing selection at the 2Lainversion make it rather hard to believethat selection is only rarely involved inintrogression.

Adaptive introgression may alsointroduce adaptive combinations thatlead to new species, or hybrid specia-tion [84, 85]. Plant examples have longbeen known [16, 85], but animal exam-ples are no longer rare. For example, theHeliconius pardalinus-like ancestor ofH. elevatus seems to have recentlyacquired the majority of its defensivecolor pattern mimicry from H. melpom-ene (Fig. 2C), subsequently proving ableto coexist in sympatry with both

144 B

parents [17]. That case remains to befully worked out, but similar cases havebeen put forward for cichlid fish,monkeyflowers, and other hybridizingadaptive radiations [86, 87]. In one case,the beginnings of the process have beenobserved in “real time”: a new hybridfinch species that breeds strictly endo-gamously has now been followed on aGalapagos island for seven generationssince its formation via initial hybridiza-tion events in the early 1980s [88].

Introgression challengesnotions of species andphylogeny

The meaning of species andspeciation

We are thus confronted by extraordi-nary levels of introgression found in thegenomes of rapidly radiating species(such as Anopheles, Heliconius, andDarwin’s finches). Yet these taxa arecurrently readily identifiable to speciesusing morphology or genetics: none ofus doubt that the species is a usefulrank, at least in multicellular eukar-yotes. We recognize these taxa asspecies not because of reproductiveisolation per se, nor because theyrepresent phylogenetic branchingevents, but because of the simplerobservation that hybrids and intermedi-ates between the clusters we callspecies [89] are rare. While most ofthe introgression that has resulted inreticulate relationships occurred in thepast – and may or may not be ongoing –these results suggest that species arelike the Ship of Theseus in philosophy,which can progressively but almostcompletely be rebuilt with new wood,and yet remain the same ship. We do notyet know how common these effects areamong genomes of other eukaryotes,but the recent discoveries in mammals,birds, fish, insects, plants, and fungisuggest that they may be widespreadthroughout the eukaryotic Tree of Life.

The “true phylogeny” versusthe species tree

In introgressing species, different genetrees vary in the story they tell about

ioessays 38: 140–149,� 2015 The Authors. Bio

their genealogical history. The truephylogeny will trace the disparatehistories of every gene, and cannotreadily be represented on a page,certainly not as a single tree. Yetwe propose that there may still be atrue bifurcating tree of species outthere (Fig. 3), in spite of the back-ground chaos of gene trees. Only ifspecies fuse either wholly or in somegeographic region to become a singlecluster (e.g. in sticklebacks [90] or inhybrid speciation), does the speciesphylogeny itself become reticulateunder this view.

Possible alternatives to the speciestree is some consensus of gene trees, orperhaps the tree based on the “demo-cratic majority” of the genome [91].Obtaining the maximum likelihood ormost probable species tree from a seriesof genes is in fact the aim of manyphylogenetic and phylogenomic stud-ies, at least among eukaryote system-atists [92, 93]. This program assumesthat the true species tree is more likelyto emerge via analysis of larger fractionsof the genome. Under the viewpointproposed here, this is not necessarilytrue if there is abundant introgression(Fig. 3). For example, as shown above,the single most common tree inferredfrom whole genomes of the Anophelesgambiae complex in Africa gives anincorrect rendering of the group’shistory [18] (Fig. 1).

Historical introgression events intaxa such as Anopheles have beeninferred to affect the majority of thegenome, even though natural hybridsare relatively rare among the contem-porary species (see above for rates ofhybridization). Nonetheless, hybrid-ization can introduce variation at ratesmuch higher than mutation, so thatsignificant levels of genomic replace-ment may accrue over long periods,even at the low hybridization ratesknown in Anopheles today. Similarresults also apply in some Heliconiusspecies. If we wish the species tree tobe determined by the democraticopinion of the genes, we are thereforeforced to accept a peculiar speciesdefinition that perhaps applies only toterminal taxa, rather than the originalbifurcating ancestors, because thebranches of the tree change theirspecies identity whenever accumula-tion of introgressed regions flips the

Essays Published by WILEY Periodicals, Inc.

Figure 2. Phylogenetic discordance B/D mimicry region of Heliconius genomes. A: FST plotshows divergent optix regulatory region determining mimicry differences between geographicraces within H. melpomene. Mimicry has been shown to have very strong adaptive value inHeliconius. B: The same region shows a strong excess of ABBA phylogenetic sites overBABA sites, implicating introgression between H. melpomene and H. timareta. C: Further-more, the non-sister species H. elevatus shows a phylogenetic topology indicatingintrogression of the rayed mimicry pattern from the melpomene-timareta clade in the samegenomic region. Modified and reprinted with permission from [17].

..... Insights & Perspectives J. Mallet et al.Thinkagain

democratic majority of the genes toanother topology. It is perhaps defen-sible to argue that the “democraticopinion” tree is more predictive of theorigins of the genes, though it ismarred by potential inferences ofancestral species (pale green) thatnever existed (Fig. 3). We instead favor

Bioessays 38: 140–149,� 2015 The Authors. B

the idea that the species tree is thebifurcation history (Fig. 3). This wewould argue is closer to what we meanby the speciation history, in spite ofthe difficulty of its discovery, andacknowledging a lowered expectationof its predictiveness for the histories ofits component genes.

ioEssays Published by WILEY Periodicals, Inc.

Are species incompatible?

Another conclusion that arises fromthese findings is that large fractions ofdifferent species’ genomes may in factbe compatible. The genomic distribu-tion of “intrinsic” incompatibilities(such as “Dobzhansky-Muller incom-patibilities” [94, 95]) is poorly knownexcept in a few species [96]. In Saccha-romyces yeasts, it is possible to replacewhole chromosomes with little effect onviability, while in Drosophila manyhybrid sterility loci seem scattered verywidely across the genome [97–101]. It ispossible that the situation in Drosophilais unusual, perhaps a result of “fastermale” sexual selection that leads to

145

Figure 3. A simple case where introgression can distort the history of species andspeciation. By “the true phylogeny,” in this paper, we mean the totality of true histories ofevery part of the genome. This is not readily depicted: our simplified cartoon of the truephylogeny network above indicates abundant introgression between species 1 and 2 aftertheir bifurcation, but little between sister species 2 and 3. It does not, however, show whichgene travels in which direction and when, all of which is surely important information aboutthe “true phylogeny” as well. If introgression is extensive, the whole genome tree (bottom left)may indicate an incorrect bifurcation history, as well as ancestral species that never existed(such as the apparent ancestor of 1 and 2 in the diagram). The true bifurcation history ofspecies is shown bottom right.

J. Mallet et al. Insights & Perspectives.....Thinkagain

genome-wide effects on male hybridsterility [102, 103]. Even though incom-patibility loci have been mapped incrosses between A. gambiae and A.arabiensis [104], genomic evidence forvery widespread homologous replace-ment between species in the autosomesof Anopheles and Heliconius [18, 73]suggests either that incompatibilitieswere not very common in thosegenomes, or that some introgressedalleles are advantageous enough toovercome initial incompatibility. Al-though autosomal genes introgressreadily in both groups, the preponder-ance of “species tree” genealogies in thesex chromosome in the Anophelesgambiae complex [18] is likely due tomultiple overlapping inversions thatdiffer between A. gambiaeþ coluzziiand A. arabiensis. These inversionssuppress recombination and so inhibitintrogression of small chromosomalfragments on this chromosome. Ifadaptive alleles are widely available tointrogress, determining the number andeffect of incompatibilities will not be

146 B

adequate to assess the potential forintrogression between species: we willalso need to know the number andselective effects of these variants.

As far as is known, classical lateraltransfer from distantly related species isnot a major recent source of phyloge-netic incongruence in multicellulareukaryotes, and most of the phyloge-netic reticulation we observe is due tohomologous exchange via hybridiza-tion. The selective advantages of sexwithin species remain contentious, butsex surely optimizes some balancebetween benefits and costs of recombi-nation [105, 106]. Typically, hybridizingwith another species is viewed as “thegrossest blunder in sexual preference,”and mate choice (reinforcement) isexpected to evolve to limit hybridizationamong sympatric species [107]. How-ever, given that hybridization does stilloccur, and sometimes leads to benefi-cial effects, we should now perhapsbroaden our view of sex across thespecies boundary, where the samecost/benefit function is confronted by

ioessays 38: 140–149,� 2015 The Authors. Bio

individuals seeking sexual partners,albeit with different parameter values.If outcrossing within and betweenspecies is regulated by the same cost/benefit equation, a sexual selectionprocess similar to reinforcement shouldapply to interactions within as well asbetween species.

Practical implications ofintrogression

The prevalence of laterally transferredantibiotic resistance genes among bac-terial species is a well-known problemfor human health [108, 109]. Similarproblemsmight therefore be expected toresult from introgression or lateraltransfer among related eukaryotic pestand disease species. The African ma-laria-carrying mosquitoes provide someworrying examples. For example, ratesof hybridization between Anophelesgambiae and A. arabiensis are only�0.22% per generation [71]. However,because this introduces foreign allelesat a rate far higher than mutation, thereare persistent concerns that insecticideresistance evolution in one species maylead to the rapid spread of that resis-tance to others via introgression [72].Multiple cases of introgression of allelesencoding both organophosphate andpyrethroid insecticide resistance arecertainly known between the sisterspecies A. gambiae and A. coluz-zii [110–113]; these two are known tohybridize and backcross much morefrequently [63] than do A. gambiae withA. arabiensis. Similarly, large siblingspecies complexes of the black fly genusSimulium transmit river blindness inAfrica and tropical America, and mayalso exchange genes. Among sympatricspecies of the African S. damnosumcomplex, hybridization rates may reach0.1% per generation. Introgression isthought likely to explain the rapidspread of insecticide resistance amongmultiple Simulium species inAfrica [114].The same problem occurs even invertebrate pests: a genomic regioncontaining a rodenticide resistanceallele spread via introgression betweentwo partially interfertile mouse speciesin Western Europe [115].

Recent advances in genetics andgenetic engineering are revolutionizingpest control, allowing for “designer

Essays Published by WILEY Periodicals, Inc.

..... Insights & Perspectives J. Mallet et al.Thinkagain

organisms” in agriculture and humanhealth. Several major transgenic crops,especially those expressing herbicide orinsect resistance, have been released inmany countries. At the same time, newmolecular marker and genomic analy-ses let us gather evidence on gene flowbetween crops and wild relatives for thefirst time. The results are clear: intro-gression does occur, and weedy rela-tives are acquiring novel geneticvariation from crops, including trans-genes that are liable to make theseweeds more noxious [116].

The use of transgenic organisms ismore advanced in agriculture than inhuman health. However, a variety ofgenetic control measures of vectorshave been suggested and in some casesare being used to engineer diseasevector populations [117]. For example,infection of Aedes mosquitoes by Wol-bachia causes refractoriness to denguevirus proliferation [118], while Wolba-chia-infected Anopheles mosquitoesshow reduced Plasmodium infec-tion [119]. In addition, there is thepossibility of manipulating the geneticsof mosquito innate immunity in order toreduce their efficiency as a vector [120].Of these, probably the most successfullyused cases so far are a number ofreleases of Wolbachia-infected Aedesaegypti to control dengue ([118] www.eliminatedengue.com). As with trans-genic plants, because the transmissionof genetic traits requires mating, thesegenetic traits may “leak” to relatedspecies via introgression. This may nothave negative impacts, especially incomparison to the potentially positivebenefits of the engineered trait on thetarget species. However, given genomicevidence for introgression of many othertraits, its importance should not beunderemphasized when seeking regula-tory approval for release of geneticallymodified organisms (recognizing thatWolbachia infection does not techni-cally qualify as a genetic modification tomost regulatory bodies).

Conclusions and outlook

Our main conclusion is that many morespecies are likely exchanging genesthan has been appreciated. It is notonly sister species that hybridize andundergo genomic introgression: whole

Bioessays 38: 140–149,� 2015 The Authors. B

groups of rapidly radiating species mayexchange adaptive as well as non-adaptive genomic regions, as in Heli-conius, Anopheles, cichlids, Xiphopho-rus, Darwin’s finches, horses, andhominins. In fact, because hybridiza-tion between sister species does notalways affect the species tree – andbecause introgression between sisterspecies is more likely – it may be thatestimates of introgression rates fromspecies tree topologies alone vastlyunderestimate the amount of gene flowoccurring in nature. For many systemswe may think we are able to infer aspecies tree signal, but we must recog-nize that this signal may only berepresented by a small fraction of genes.

As well as causing problems forphylogenetics, abundant introgressionand incomplete lineage sorting mightgreatly weaken inferences in compara-tive analysis. When we map charactertraits onto the tree of a rapidly radiatinggroup, we should be cautious. Forinstance, the raptorial habit is thoughtto be ancestral to the entire core land-birds, but today it is present in severalmonophyletic groups, each moreclosely related to birds that haveapparently lost the habit [79–81]. Alter-natively, core landbirds may have beenancestrally non-raptorial, and a numberof raptorial traits could have could havebeen shared at the base of these lineagesby introgression among the early spe-cies of each lineage. This is notdissimilar to what we observe in mim-icry patterns in Heliconius or in beakmorphology of Darwin’s finches, amongspecies of radiations that we see hy-bridizing today [17, 74]. Similarly,inferences from phylogeography – suchas geographic origins of rapidly radiat-ing groups inferred from phylogeneticmethods – should be affected as well.The origins of traits, and the genes thatdetermine them can have very differenthistories from that of the species tree.

AcknowledgmentsWe thank Kanchon Dasmahapatra, BillHanage, Robin Hopkins, James McIner-ney, Jesse Shapiro, Leonie Moyle, LuayNakhleh, Tim Sackton, and Ziheng Yangfor discussions. This paper benefitedfrom the comments of three anonymousreviewers. The authors were supportedby BBSRC, Harvard University, SPARC

ioEssays Published by WILEY Periodicals, Inc.

funding from the Broad Insititute ofHarvard and MIT, and NIH grant R01AI76584.The authors have declared no conflictsof interest.

References

1. Woese CR, Kandler O, Wheelis ML. 1990.Towards a natural system of organisms:proposal for the domains Archaea, Bacteria,and Eucarya. Proc Natl Acad Sci USA 87:4576–9.

2. Woese CR, Fox GE. 1977. Phylogeneticstructure of the prokaryotic domain: theprimary kingdoms. Proc Natl Acad Sci USA74: 5088–90.

3. Doolittle WF. 1999. Phylogenetic classifi-cation and the universal tree. Science 284:2124–9.

4. Hilario E, Gogarten JP. 1993. Horizon-tal transfer of ATPase genes – the tree oflife becomes a net of life. BioSyst 31:111–9.

5. Martin W. 1999. Mosaic bacterial chromo-somes: a challenge en route to a tree ofgenomes. BioEssays 21: 99–104.

6. Clark AG, Eisen MB, Smith DR, BergmanCM, et al. 2007. Evolution of genes andgenomes on the Drosophila phylogeny.Nature 450: 203–18.

7. Lindblad-Toh K,Garber M, Zuk O, Lin MF,et al. 2011. A high-resolution map of humanevolutionary constraint using 29 mammals.Nature 478: 476–82.

8. Dujon B, Sherman D, Fischer G, DurrensP, et al. 2004. Genome evolution in yeasts.Nature 430: 35–44.

9. Maia VH, Gitzendanner MA, Soltis PS,Wong GK-S, et al. 2014. Angiospermphylogeny based on 18S/26S rDNA se-quence data: constructing a large data setusing next-generation sequence data. Int JPlant Sci 175: 613–50.

10. Lee EK,Cibrian-Jaramillo A,KolokotronisS-O, Katari MS, et al. 2011. A functionalpylogenomic view of the seed plants. PLoSGenet 7: e1002411.

11. Marcussen T, Sandve SR, Heier L, Span-nagl M, et al. 2014. Ancient hybridizationsamong the ancestral genomes of breadwheat. Science 345: 1250092.

12. Brandvain Y, Kenney AM, Flagel L, CoopG, et al. 2014. Speciation and introgressionbetween Mimulus nasutus and Mimulusguttatus. PLoS Genet 10: e1004410.

13. McCormack JE, Faircloth BC, CrawfordNG,Gowaty PA, et al. 2012. Ultraconservedelements are novel phylogenomic markersthat resolve placental mammal phylogenywhen combined with species-tree analysis.Genome Res 22: 746–54.

14. Yang Y,MooreMJ,Brockington SF, SoltisDE, et al. 2015. Dissecting molecularevolution in the highly diverse plant cladeCaryophyllales using transcriptome se-quencing. Mol Biol Evol 32: 2001–14.

15. Mayr E. 1963. Animal Species and Evolu-tion. Cambridge, MA: Harvard UniversityPress.

16. Anderson E, Stebbins GL. 1954. Hybrid-ization as an evolutionary stimulus. Evolution8: 378–88.

147

J. Mallet et al. Insights & Perspectives.....Thinkagain

17. Heliconius Genome Consortium. 2012. But-terfly genome reveals promiscuous ex-change of mimicry adaptations amongspecies. Nature 487: 94–8.

18. Fontaine MC, Pease JB, Steele A, Water-house RM, et al. 2015. Extensive introgres-sion in a malaria vector species complexrevealed by phylogenomics. Science 347:1258524.

19. Zawadzki P, Roberts MS, Cohan FM.1995. The log-linear relationship betweensexual isolation and sequence divergence inBacillus is robust. Genetics 140: 917–32.

20. Majewski J, Zawadzki P, Pickerill P,Cohan FM, et al. 2000. Barriers to geneticexchange between bacterial species: Strep-tococcus pneumoniae transformation.J Bacteriol 182: 1016–23.

21. Eppley JM, Tyson GW,Getz WM, BanfieldJF. 2007. Genetic exchange across aspecies boundary in the archaeal genusFerroplasma. Genetics 177: 407–16.

22. Naor A, Lapierre P, Mevarech M, PapkeRT, et al. 2012. Low species barriers inhalophilic Archaea and the formation ofrecombinant hybrids. Curr Biol 22: 1444–8.

23. Gourbi�ere S, Mallet J. 2010. Are speciesreal? The shape of the species boundarywith exponential failure, reinforcement, andthe “missing snowball”. Evolution 64:1–24.

24. Morales L,Dujon B. 2012. Evolutionary roleof interspecies hybridization and geneticexchanges in yeasts.Microbiol Mol Biol Rev76: 721–39.

25. Woese CR. 2002. On the evolution of cells.Proc Natl Acad Sci USA 99: 8742–7.

26. Creevey CJ, Doerks T, Fitzpatrick DA,Raes J, et al. 2011. Universally distributedsingle-copy genes indicate a constant rateof horizontal transfer. PLoS ONE 6: e22099.

27. Rivera MC, Lake JA. 2004. The ring of lifeprovides evidence for a genome fusionorigin of eukaryotes. Nature 431: 152–5.

28. McInerney JO, O’Connell MJ, Pisani D.2014. The hybrid nature of the Eukaryotaand a consilient view of life on Earth.Nat RevMicrobiol 12: 449–55.

29. Puigb�o P, Wolf YI, Koonin EV. 2009.Search for a ‘Tree of Life’ in the thicket ofthe phylogenetic forest. J Biol 8: 1–17.

30. Ciccarelli FD, Doerks T, von Mering C,Creevey CJ, et al. 2006. Toward automaticreconstruction of a highly resolved tree oflife. Science 311: 1283–7.

31. Lang JM, Darling AE, Eisen JA. 2013.Phylogeny of bacterial and archaealgenomes using conserved genes: super-trees and supermatrices. PLoS ONE 8:e62510.

32. Dagan T, Martin W. 2006. The tree of onepercent. Genome Biol 7: 118.

33. Lawrence JG, Retchless AC. 2010. Themyth of bacterial species and speciation.Biol Philos 25: 569–88.

34. Mindell DP. 2013. The Tree of Life: meta-phor, model, and heuristic device. Syst Biol62: 479–89.

35. Huang J. 2013. Horizontal gene transfer ineukaryotes: the weak-link model. BioEssays35: 868–75.

36. Keeling PJ, Palmer JD. 2008. Horizontalgene transfer in eukaryotic evolution. NatRev Genet 9: 605–18.

37. Crisp A, Boschetti C, Perry M, TunnacliffeA, et al. 2015. Expression of multiplehorizontally acquired genes is a hallmark

148 B

of both vertebrate and invertebrategenomes. Genome Biol 16: 50.

38. Gladyshev EA,MeselsonM,Arkhipova IR.2008. Massive horizontal gene transfer inbdelloid rotifers. Science 320: 1210–3.

39. Boschetti C, Carr A, Crisp A, Eyres I, et al.2012. Biochemical diversification throughforeign gene expression in bdelloid rotifers.PLoS Genet 8: e1003035.

40. Wu B, Buljic A, Hao W. 2015. Extensivehorizontal transfer and homologous recom-bination generate highly chimeric mitochon-drial genomes in yeast. Mol Biol Evol 32:2559–70.

41. Sela N, Kim E, Ast G. 2010. The role oftransposable elements in the evolution ofnon-mammalian vertebrates and inverte-brates. Genome Biol 11: R59.

42. Kidwell MG. 2002. Transposable elementsand the evolution of genome size ineukaryotes. Genetica 115: 49–63.

43. Lynch M. 2007. The Origins of GenomeArchitecture. Sunderland, MA: SinauerAssociates.

44. Lynch M. 2006. The origins of eukaryoticgene structure. Mol Biol Evol 23: 450–68.

45. Rogozin IB, Carmel L, Csuros M, KooninEV. 2012. Origin and evolution of spliceo-somal introns. Biol Direct 7: 11.

46. Schaack S, Gilbert C, Feschotte C. 2010.Promiscuous DNA: horizontal transfer oftransposable elements and why it mattersfor eukaryotic evolution. Trends Ecol Evol25: 537–46.

47. Burgess R, Yang Z. 2008. Estimation ofhominoid ancestral population sizes underBayesian coalescent models incorporatingmutation rate variation and sequencingerrors. Mol Biol Evol 25: 1979–94.

48. Dutheil JY, Munch K, Nam K, Mailund T,et al. 2015. Strong selective sweeps on the Xchromosome in the human-chimpanzeeancestor explain its low divergence. PLoSGenet 11: e1005451.

49. Wiuf C, Zhao K, Innan H, Nordborg M.2004. The probability and chromosomalextent of trans-specific polymorphism. Ge-netics 168: 2363–72.

50. Teixeira JC, de Filippo C, Weihmann A,Meneu JR, et al. 2015. Long-term balancingselection in LAD1 maintains a missensetrans-species polymorphism in humans,chimpanzees, and bonobos. Mol Biol Evol32: 1186–96.

51. Saitou N, Yamamoto F. 1997. Evolution ofprimate ABO blood group genes and theirhomologous genes. Mol Biol Evol 14:399–411.

52. Degnan JH, Rosenberg NA. 2006. Discor-dance of species trees with their most likelygene trees. PLoS Genet 2: e68.

53. Yang Z. 2014. Molecular Evolution. AStatistical Approach. Oxford: Oxford Uni-versity Press.

54. Yang Z, Rannala B. 2014. Unguidedspecies delimitation using DNA sequencedata from multiple loci. Mol Biol Evol 31:3125–35.

55. Degnan JH, Rosenberg NA. 2009. Genetree discordance, phylogenetic inferenceand the multispecies coalescent. TrendsEcol Evol 24: 332–40.

56. Grant PR, Grant BR. 1992. Hybridization ofbird species. Science 256: 193–7.

57. Mallet J. 2005. Hybridization as an inva-sion of the genome. Trends Ecol Evol 20:229–37.

ioessays 38: 140–149,� 2015 The Authors. Bio

58. Arnold ML, Sapir Y, Martin NH. 2008.Genetic exchange and the origin of adapta-tions: prokaryotes to primates. Philos TransR Soc Lond B Biol Sci 363: 2813–20.

59. Mallet J, Beltr�an M, Neukirchen W,Linares M. 2007. Natural hybridization inheliconiine butterflies: the species boundaryas a continuum. BMC Evol Biol 7: 28.

60. Bacilieri R, Ducousso A, Petit RJ, KremerA. 1996. Mating system and asymmetrichybridization in a mixed stand of Europeanoaks. Evolution 50: 900–8.

61. Grant PR. 1993. Hybridization of Darwin’sfinches on Isla Daphne Major, Gal�apagos.Philos Trans R Soc Lond B Biol Sci 340:127–39.

62. Mallet J, McMillan WO, Jiggins CD. 1998.Estimating the mating behavior of a pair ofhybridizing Heliconius species in the wild.Evolution 52: 503–10.

63. Lee Y,Marsden CD, Norris LC, Collier TC,et al. 2013. Spatiotemporal dynamics ofgene flow and hybrid fitness between the Mand S forms of the malaria mosquito,Anopheles gambiae. Proc Natl Acad SciUSA 110: 19854–9.

64. Coyne JA, Orr HA. 1997. “Patterns ofspeciation inDrosophila” revisited. Evolution51: 295–303.

65. Price TD, Bouvier MM. 2002. The evolutionof F1 postzygotic incompatibilities in birds.Evolution 56: 2083–9.

66. Moyle LC, Nakazato T. 2010. Hybridincompatibility “snowballs” between Sola-num species. Science 329: 1521–3.

67. Moyle LC, Olson MS, Tiffin P, Baum D.2004. Patterns of reproductive isolation inthree angiosperm genera. Evolution 58:1195–208.

68. Anderson E. 1948. Hybridization of thehabitat. Evolution 2: 1–9.

69. Leach�e AD,Harris RB,Rannala B, Yang Z.2013. The influence of gene flow on speciestree estimation: a simulation study. Syst Biol63: 17–30.

70. Coluzzi M, Sabatini A, Petrarca V,Di DecoMA. 1979. Chromosomal differentiation andadaptation to human environments in theAnopheles gambiae complex. Trans R SocTrop Med Hyg 73: 483–97.

71. Weetman D, Steen K, Rippon EJ,MawejjeHD, et al. 2014. Contemporary gene flowbetween wild An. gambiae s.s. and An.arabiensis. Parasites Vectors 7: 345.

72. Mawejje HD, Wilding CS, Rippon EJ,Hughes A, et al. 2013. Insecticide resis-tance monitoring of field-collected Anophe-les gambiae s.l. populations from Jinja,eastern Uganda, identifies high levels ofpyrethroid resistance. Med Vet Entomol 27:276–83.

73. Martin SH, Dasmahapatra KK, NadeauNJ, Salazar C, et al. 2013. Genome-wideevidence for speciation with gene flow inHeliconius butterflies. Genome Res 23:1817–28.

74. Lamichhaney S, Berglund J, Almen MS,Maqbool K, et al. 2015. Evolution ofDarwin’s finches and their beaks revealedby genome sequencing. Nature 518: 371–5.

75. BrawandD,Wagner CE, Li YI,MalinskyM,et al. 2014. The genomic substrate foradaptive radiation in African cichlid fish.Nature 513: 375–81.

76. J�onsson H, Schubert M, Seguin-OrlandoA, Ginolhac A, et al. 2014. Speciation withgene flow in equids despite extensive

Essays Published by WILEY Periodicals, Inc.

..... Insights & Perspectives J. Mallet et al.Thinkagain

chromosomal plasticity. Proc Natl Acad SciUSA 111: 18655–60.

77. Cui R, Schumer M, Kruesi K, Walter R,et al. 2013. Phylogenomics reveals exten-sive reticulate evolution in Xiphophorusfishes. Evolution 67: 2166–79.

78. Green RE, Krause J, Briggs AW, Mari-cic T, et al. 2010. A draft sequence of theNeandertal genome. Science 328:710–22.

79. Jarvis ED,Mirarab S,Aberer AJ, Li B, et al.2014. Whole-genome analyses resolve earlybranches in the tree of life of modern birds.Science 346: 1320–31.

80. Suh A, Smeds L, Ellegren H. 2015. Thedynamics of incomplete lineage sortingacross the ancient adaptive radiation ofneoavian birds. PLoS Biol 13: e1002224.

81. Prum RO, Berv JS, Dornburg A, Field DJ,et al. 2015. A comprehensive phylogeny ofbirds (Aves) using targeted next-generationDNA sequencing. Nature 526: 569–73.

82. Whelan NV, Kocot KM, Moroz LL, Hala-nych KM. 2015. Error, signal, and theplacement of Ctenophora sister to all otheranimals. Proc Natl Acad Sci USA 112:5773–8.

83. Hedrick PW. 2013. Adaptive introgressionin animals: examples and comparison tonew mutation and standing variation assources of adaptive variation. Mol Ecol 22:4606–18.

84. Mallet J. 2007. Hybrid speciation. Nature446: 279–83.

85. Rieseberg LH. 1997. Hybrid origins of plantspecies. Annu Rev Ecol Syst 28: 359–89.

86. Seehausen O. 2004. Hybridization andadaptive radiation. Trends Ecol Evol 19:198–207.

87. Stankowski S, Streisfeld MA. 2015. Intro-gressive hybridization facilitates adaptivedivergence in a recent radiation of monkey-flowers. Proc Biol Sci 282: 20151666.

88. Grant PR, Grant BR. 2014. Synergism ofnatural selection and introgression in theorigin of a new species. Am Naturalist 183:671–81.

89. Mallet J. 2013. Species, concepts of. InLevin SA, ed; Encyclopedia of Biodiversity,Second Edition, Volume 6. Waltham, MA:Academic Press. p. 679–91.

90. Behm JE, Ives AR, Boughman JW. 2010.Breakdown in postmating isolation and thecollapse of a species pair through hybrid-ization. Am Naturalist 175: 11–26.

91. Maddison WP. 1997. Gene trees in speciestrees. Syst Biol 46: 523–36.

Bioessays 38: 140–149,� 2015 The Authors. B

92. Liu L, Yu L, Kubatko L, Pearl DK, et al.2009. Coalescent methods for estimatingphylogenetic trees. Mol Phylogen Evol 53:320–8.

93. Edwards SV. 2009. Is a new and generaltheory of molecular systematics emerging?Evolution 63: 1–19.

94. Orr HA, Turelli M. 2001. The evolution ofpostzygotic isolation: accumulating Dobz-hansky-Muller incompatibilities. Evolution55: 1085–94.

95. Coyne JA, Orr HA. 2004. Speciation.Sunderland, MA: Sinauer Associates.

96. Orr HA, Madden LD, Coyne JA, GoodwinR, et al. 1997. The developmental geneticsof hybrid inviability: a mitotic defectin Drosophila hybrids. Genetics 145:1031–40.

97. Greig D. 2007. A screen for recessivespeciation genes expressed in the gametesof F1 hybrid yeast. PLoS Genet 3: 281–6.

98. Greig D. 2008. Reproductive isolation inSaccharomyces. Heredity 102: 39–44.

99. True JR, Weir BS, Laurie CC. 1996.A genome-wide survey of hybrid incompati-bility factors by the introgression of markedsegments of Drosophila mauritiana chromo-somes into Drosophila simulans. Genetics142: 819–37.

100. Hollocher H, Wu CI. 1996. The genetics ofreproductive isolation in the Drosophilasimulans clade: X vs. autosomal effectsand male vs. female effects. Genetics 143:1243–55.

101. Masly JP, Presgraves DC. 2007. High-resolution genome-wide dissection of thetwo rules of speciation in Drosophila. PLoSBiol 5: e243.

102. Wu CI, Palopoli MF, Johnson NA. 1996.Haldane’s rule and its legacy: why are thereso many sterile males? Trends Ecol Evol 11:281–4.

103. Chapman T, Partridge L. 1996. Sexualconflict as fuel for evolution. Nature 381:189–90.

104. Slotman M, della Torre A, Powell JR.2004. The genetics of inviability and malesterility in hybrids between Anopheles gam-biae and An. arabiensis. Genetics 167:275–87.

105. Maynard Smith J. 1978. The Evolution ofSex. Cambridge: Cambridge UniversityPress.

106. Otto SP. 2009. The evolutionary enigma ofsex. Am Naturalist 174: S1–S14.

107. Fisher RA. 1930. The Genetical Theory ofNatural Selection. Oxford: Clarendon Press.

ioEssays Published by WILEY Periodicals, Inc.

108. Davies J, Davies D. 2010. Origins andevolution of antibiotic resistance. MicrobiolMol Biol Rev 74: 417–33.

109. Finley RL, Collignon P, Larsson DGJ,McEwen SA, et al. 2013. The scourge ofantibiotic resistance: the important role ofthe environment. Clin Infect Dis 57: 704–10.

110. Djogb�enou L, Chandre F, Berthomieu A,Dabir�e R, et al. 2008. Evidence of introgres-sion of the ace-1R mutation and of the ace-1duplication in West African Anophelesgambiae s.s. PLoS ONE 3: e2172.

111. Weill M,Chandre F, Brengues C,ManguinS, et al. 2000. The kdr mutation occurs in theMopti formofAnophelesgambiaes.s. throughintrogression. Insect Mol Biol 9: 451–5.

112. Norris LC,Main BJ, Lee Y,Collier TC, et al.2015. Adaptive introgression in an Africanmalaria mosquito coincident with the in-creased usage of insecticide-treated bednets. Proc Natl Acad Sci USA 112: 815–20.

113. Clarkson CS, Weetman D, Essandoh J,Yawson AE, et al. 2014. Adaptive introgres-sion between Anopheles sibling specieseliminates a major genomic island but notreproductive isolation. Nat Commun 5: 4248.

114. Adler PH, Cheke RA, Post RJ. 2010.Evolution, epidemiology, and populationgenetics of black flies (Diptera: Simuliidae).Infect Genet Evol 10: 846–65.

115. Song Y, Endepols S, Klemann N, RichterD, et al. 2011. Adaptive introgression ofanticoagulant rodent poison resistance byhybridization between Old World mice. CurrBiol 21: 1296–301.

116. Ellstrand NC, Meirmans P, Rong J,Bartsch D, et al. 2013. Introgression ofcrop alleles into wild or weedy populations.Annu Rev Ecol Evol Syst 44: 325–45.

117. Alphey L. 2014. Genetic control of mosqui-toes. Annu Rev Entomol 59: 205–24.

118. Hoffmann AA, Montgomery BL, PopoviciJ, Iturbe-Ormaetxe I, et al. 2011. Success-ful establishment of Wolbachia in Aedespopulations to suppress dengue transmis-sion. Nature 476: 454–7.

119. Bian G, Joshi D, Dong Y, Lu P, et al. 2013.Wolbachia invades Anopheles stephensipopulations and induces refractoriness toPlasmodium infection. Science 340: 748–51.

120. Pike AD, Cirimotich CM, Dimopoulos G.2013. Impact of transgenic immune deploy-ment on mosquito fitness. In Koenraadt S,Takken W, eds; Ecology and Control ofVector-Borne Diseases, Volume 3, Ecologyof Parasite-Vector Interactions. Wagenin-gen: Academic Publishers. p. 19–33.

149