DIVERSIFICATION IN SEXUAL AND ASEXUAL ORGANISMS
2166
?2003The Society for the Study of Evolution.All rights reserved.
BRIEF COMMUNICATIONS
Evolution,57(9),2003,pp.2166–2172
DIVERSIFICATION IN SEXUAL AND ASEXUAL ORGANISMS
T IMOTHY G.B ARRACLOUGH ,1,2C.W ILLIAM B IRKY ,J R .,3,4AND A USTIN B URT 1,5
1Department
of Biological Sciences and NERC Centre for Population Biology,Imperial College London,Silwood Park Campus,Ascot,
Berkshire SL57PY,United Kingdom 2E-mail:t.barraclough@https://www.wendangku.net/doc/fc13364422.html,
3Department of Ecology and Evolutionary Biology and the Graduate Interdisciplinary Program in Genetics,University of Arizona,
Biological Sciences West,Tucson,Arizona 85721
4E-mail:birky@https://www.wendangku.net/doc/fc13364422.html, 5E-mail:a.burt@https://www.wendangku.net/doc/fc13364422.html,
Abstract.Sexual reproduction has long been proposed as a major factor explaining the existence of species and species diversity.Yet,the importance of sex for diversi?cation remains obscure because of a lack of critical theory,dif?culties of applying universal concepts of species and speciation,and above all the scarcity of empirical tests.Here,we use genealogical theory to compare the relative tendency of strictly sexual and asexual organisms to diversify into discrete genotypic and morphological clusters.We conclude that asexuals are expected to display discrete clusters similar to those found in sexual organisms.Whether sexuals or asexuals display stronger clustering depends on a number of factors,but in at least some scenarios asexuals should display a stronger pattern.Confounding factors aside,the only explanation we identify for stronger patterns of diversi?cation in sexuals than asexuals is if the faster rates of adaptive change conferred by sexual reproduction promote greater clustering.Quantitative comparisons of diversi?cation in related sexual and asexual taxa are needed to resolve this issue.The answer should shed light not only on the importance of the different stages leading to diversi?cation,but also on the adaptive consequences of sex,still largely unexplored from a macroevolutionary perspective.Key words.
Adaptation,ancient,bdelloid,diversity,genealogy,rotifers,speciation.
Received June 7,2002.
Accepted April 17,2003.
It is widely agreed that species are fundamental units of life.But why do species exist?Why does life fall into discrete clusters with respect to genotype,morphology,and behavior,rather than forming a continuous distribution?Coyne and Orr (1998)called this ‘‘one of the most important but neglected questions about speciation’’(p.287)and cite two main gen-eral explanations proposed by Maynard Smith and Szathmary (1995):
(1)Species arise by divergent selection leading to adap-tation to different ecological niches.There is increasing ex-perimental and theoretical support for ecological speciation in sexual organisms (Schluter 2001)and it has frequently been invoked to explain speciation in asexual organisms (Fisher 1930;Mayr 1957;Hutchinson 1968).
(2)Speciation is a consequence of sexual reproduction.Sexual reproduction (with at least some outcrossing)main-tains cohesion within populations,whereas reproductive iso-lation of different genotypes or populations leads to diver-gence and speciation (Mayr 1957).Maynard Smith and Szath-mary (1995)and Coyne and Orr (1998)regarded this hy-pothesis as most important,but they highlighted that empirical tests were scarce.
One way to distinguish the importance of ecological niches and sexual reproduction might be to test whether asexual organisms fall into discrete clusters to the same extent as sexual organisms.If they do,it would suggest that sex is not necessary for the formation of discrete clusters.Taxonomic evidence appears to support this in some groups.For ex-ample,in the famously asexual bdelloid rotifers there are often clear morphological differences among described spe-cies,qualitatively similar to those found in sexual groups (Hutchinson 1968),and taxonomists have identi?ed species more consistently than in sexual monogonont rotifers (Hol-man 1987).Moreover,bdelloids fall into distinct clades based on mitochondrial gene sequences (C.W.Birky,unpubl.data).These results show that clusters can form in asexual organ-isms,but not how effectively.More satisfying would be a quantitative comparison of the pattern of diversi?cation be-tween sexual and asexual clades.Such studies have not yet been made,but are likely to appear soon (Barraclough and Nee 2001).
Before performing such studies,it is vital to consider the expected effects of presence or absence of sexual reproduc-tion on patterns of diversi?cation.Previous accounts have made general statements but there have been few detailed treatments.For example,recombination tends to oppose ge-netic divergence within populations,by eroding associations across multiple loci (Felsenstein 1981).Without compensat-ing bene?ts,therefore,genetic divergence might be harder in sexuals than in asexuals.Moreover,the comparison be-tween sexuals and asexuals is complicated by the dif?culty of applying a consistent species concept.The biological spe-cies concept cannot apply to asexual organisms because all individuals are reproductively isolated entities.Attempts have been made to devise a universal species concept that will be widely accepted,but the literature bears witness to the dif?culty of this endeavor (Mishler and Donoghue 1982;Hey 2001).
Here,we present a critical account of how sex versus asex is likely to affect the tendency of lineages to diversify.To
2167
BRIEF
COMMUNICATIONS
F I
G .1.Hypothetical gene trees.(A)A gene tree from a single population.(B)A gene tree showing the existence of three discrete clusters on the end of long stem branches.
avoid problems with de?ning species,we focus instead on comparing patterns of genetic and morphological diversi?-cation between sexuals and asexuals.Our framework is to consider expected gene trees under alternative scenarios.This approach provides a link between pattern-based surveys of genetic and morphological diversity and the population ge-netic processes underlying the observed patterns.After con-sidering the patterns of diversity expected in single popu-lations in simple environments,we compare the relative re-sponse of sexual and asexual populations to conditions fa-voring diversi?cation.Finally,we discuss likely confounding factors that differ between sexual and asexual clades beyond the presence or absence of recombination.
G ENE T REES :B ACKGROUND
AND
D EFINITIONS
A gene tree is a representation of the ancestry of alleles from the same locus sampled across a set of individuals.Under the simplest null model that the individuals can be treated as a random sample from a single population,the expected gene tree conforms to the genealogy for a single population (Fig.1A).For example,if a locus has evolved under a neutral model,the expected tree would be the neutral coalescent (Kingman 1982;Hudson 1991).The expected gene trees for a fairly wide range of conditions are known,including changing population size and various selection re-gimes (Hudson 1991;Neuhauser and Krone 1997;Fu and Li 1999).From this,we can predict expected patterns of genetic variation in a population,and thereby predict expected var-iation in morphology and other traits.
We ?rst consider gene trees for a single population with a constant size of N individuals,which is either entirely sex-ual (with random mating)or entirely asexual.For present purposes we do not consider partial or rare recombination (e.g.,bacterial speciation,Cohan 1998;Cohan 2001).We begin with a haploid model for simplicity,discussing the effects of diploidy later.Hence,in both the sexual and asexual populations,for a single locus there are N genes in the pop-ulation at any time.Our account is based primarily on work assuming the Fisher-Wright model,namely that generations are discrete and that the genes present in an offspring gen-eration were obtained by sampling N genes with replacement from the parent generation.However,we believe the general points are qualitatively robust to other formulations.One
property that we will focus on is the time to most recent common ancestor (Tmrca ).This is the time back from the present at which the gene tree coalesces to a single common ancestor for all genes in the sample.
Our main interest is in the formation of discrete clusters beyond levels expected under simple null models that indi-viduals come from a single population.By clusters,we mean groups of individuals with similar genetic and/or morpho-logical characteristics,separated from other such groups by gaps in genotypic/morphological space.Classic accounts have visualized clusters as peaks in the n -dimensional fre-quency distribution of morphological traits (Hutchinson 1968).In the context of gene trees,discrete clusters are ap-parent as clumps of terminals on the end of long internal branches (Fig 1B).Importantly,we are interested in clusters found consistently across multiple loci,for example,in sur-veys of several DNA markers.A few loci may not cluster in the same way,for example,due to ancestral polymorphisms or lack of variation,but overall we require the expected pop-ulation genealogy to exhibit clusters,not just a given gene tree (Avise 2000).For now,we consider idealized gene trees,that is,knowledge of the true genealogy of a locus.
G ENE T REES IN A S INGLE P OPULATION ,N O
D IVERSIFYING S ELECTION
Neutral Case
The expected gene tree for a single locus under a neutral model is given by neutral coalescent theory (Kingman 1982;Hudson 1991).For a haploid population,Tmrca is exponen-tially distributed with expectation 2N(1?1/k)generations,in which k is the number of genes sampled in the gene tree.Hence,for large samples,the expected time until all genes in the sample coalesce is approximately 2N generations.This is the same for haploid sexuals and asexuals.The expected Tmrca would be doubled in a diploid sexual population be-cause a diploid population harbors twice as many genes.The difference between haploid sexual and asexual pop-ulations is apparent when we consider gene trees across mul-tiple loci.In asexuals,there is no recombination,therefore all loci have the same gene tree.Hence,there exists a tree structure with recognizable clades that differ by varying amounts from other such clades (Fig 2A,Higgs and Derrida 1992).This leads to inherent discreteness within asexual pop-ulations simply as the result of stochastic individual repli-cation.In the sexual case,although each locus has a gene tree similar to a single locus in the asexual case,gene trees differ among incompletely linked loci (loci show different realizations of the random process).Hence,there is no tree structure within the sexual population because the overall genotype comprises the average of the set of different random gene trees (Fig 2B).Sexual populations should display a unimodal distribution of neutral genetic or morphological variation based on multiple loci,without the hierarchical var-iation and gaps expected in an asexual population (Higgs and Derrida 1992).
Selection against Deleterious Mutations
For a locus experiencing selection against deleterious mu-tations arising at a constant rate,the expected gene tree is
2168
BRIEF
COMMUNICATIONS
F I
G .2.Hypothetical gene trees for single populations under a range of scenarios.(A)Neutral coalescent for a single locus in a haploid population.The same tree is expected for a consensus tree of multiple loci in asexuals,assuming neutral coalescence.(B)Consensus tree for multiple loci in a sexual haploid population.The polytomy re?ects incongruence among unlinked loci.(C)Gene tree following completion of a selective sweep that occurred at the locus or linked loci.(D)Gene tree sampled midway through a selective sweep.(E)Frequency-dependent selection at a single locus.The two clusters represent the two alleles favored by selection.
very similar to the neutral coalescent.High and low selection coef?cients have little effect,but intermediate selection co-ef?cients reduce the Tmrca slightly below 2N generations (Neuhauser and Krone 1997;Przeworski et al.1999).Back-ground selection at loci linked to the region of interest could have the same effect,and this is likely to be stronger in asexuals because the entire genome is linked.However,the qualitative comparison of sexual and asexual populations is unchanged from the neutral case.
Selective Sweeps
The spread of a bene?cial variant through a population tends to homogenize variation at linked loci,referred to as a selective sweep (Fig.2C,Barton 2000).Sweeps occurring within,on average,the last 2N generations will reduce the Tmrca of a gene tree to the time since the sweep began (Gal-tier et al.2000).In asexuals,a selective sweep homogenizes the entire genome.In sexuals,a single sweep only homog-enizes the part of the genome that is tightly linked to the bene?cial variant.Multiple sweeps can occur at the same time but will rarely homogenize the entire genome.Once a sweep is ?nished,the expected gene trees return to those for the neutral model.
The effect of sweeps on surveys of genetic and morpho-logical diversity will depend on their frequency.If sweeps occur rarely,then most populations we sample will be un-affected and the qualitative pattern of variation will be similar to the neutral case.If sweeps are frequent,some populations will be sampled during a sweep or recovery from a sweep.In this case,an asexual population will display a single strong cluster of individuals,those with the new variant within a wider tree (Fig.2D,Barton 2000).No such pattern would be expected for multilocus measures in sexuals because sweeps would only affect regions linked to the selected locus.Fi-nally,if a change in environment leads to sweeps at multiple loci,sexuals are likely to evolve faster towards the new op-
timum (assuming the rate of origin of bene?cial alleles is not limiting,Barton and Charlesworth 1998;Burt 2000),perhaps increasing the chance of catching asexuals during a sweep.
Conclusions
Within a single population in a simple environment,as-exuals display a hierarchical pattern of variation,in contrast to the unimodal distribution expected for multilocus traits or measures in sexual populations.The broad pattern is unaf-fected by selection against deleterious mutations or selective sweeps.This tree-like ancestry could give the appearance of discrete taxa in asexual populations,but there is a limit to the degree of diversi?cation that can evolve in a single pop-ulation,determined by the expected Tmrca .Therefore,a more interesting comparison is the extent to which asexuals and sexuals diversify beyond levels expected in these simple null models.Two conditions cause diversi?cation at this level:diversifying selection and geographic isolation.
D IVERSIFYING S ELECTION
IN A
S INGLE P OPULATION
Most models of sympatric speciation investigate how di-versifying selection might cause a single sexual population to split into two reproductively isolated species.But how does the likelihood of sympatric divergence depend on the presence or absence of sex?Genealogical theory developed for frequency-dependent selection applies to this scenario.If two or more variants at a single locus experience frequency-dependent selection,then providing mutational shifts be-tween them are relatively rare,the expected gene tree is shown in Figure 2E (Hudson 1991).We expect to observe two clades in the tree,corresponding to the two variants maintained by selection.The Tmrca is greater than in the neutral case.In asexual populations,frequency dependent selection at a single locus with rare mutation would lead to the formation of two clusters recognizable across the whole
2169
BRIEF
COMMUNICATIONS
F I
G .3.Consensus gene trees from multiple loci sampled across two populations geographically isolated at time T s ,each with population size N/2.(A)Asexual haploid population.Assuming neutral coalescent,the Tmrca within each population is N generations.(B)Sexual haploid population.Same as (A)except for lack of resolution in consensus trees.(C)Sexual diploid population.Assuming neutral coalescent,the Tmrca within each population is 2N generations.Note we assume that T s k 2N ;that is,that coalescence within populations is complete at time of sample.
genome.In sexuals,frequency dependent selection at a single locus will only produce clustering at linked loci.
If diversifying selection acted on multiple gene traits,for example,caused by the existence of two habitat types,then divergence in the asexual population could proceed by step-wise accumulation of successive bene?cial changes in each cluster.In contrast,adaptive divergence could only occur in the sexual population if assortative mating evolved concur-rently (Felsenstein 1981;Dieckmann and Doebeli 1999).If assortative mating did arise,for example,mediated by local mating or pleiotropic effects of adaptive change (Rice 1987),then divergence might proceed faster in sexuals than asexuals because sweeps can occur simultaneously at multiple loci.However,sympatric adaptive divergence in sexuals,although possible,is widely believed to be less frequent than allopatric divergence because of the need for assortative mating (Turelli et al.2001).To the extent that this belief is true,sympatric divergence will be easier in asexuals,which have no such requirement.
G ENE T REES F OLLOWING G EOGRAPHIC I SOLATION
Barrier to Gene Flow in the Neutral Case
We consider an ancestral population of size N that at time Ts splits into two populations of size N/2with no gene ?ow.For simplicity,we consider an equal split,but our conclusions do not depend on this assumption.In asexuals,the barrier to migration means that after roughly N generations from the onset of isolation,each population coalesces to a single an-cestor (because the population sizes are N/2).Therefore,we expect two discrete genotypic clusters,with the age of the split between them equal to Ts ?N,in which N is the expected Tmrca for two alleles chosen at random from the ancestral population at the time of split (Fig.3A;Edwards and Beerli 2000).
In haploid sexuals,we expect a similar pattern because the
neutral coalescent is the same.The only difference is that each asexual population would display hierarchical variation of multilocus genotypes or traits,whereas the sexual popu-lations would not (Fig.3B).Therefore,the rate and extent to which sexuals and asexuals form discrete clusters in geo-graphic isolation is the same,all other things being equal.If the sexuals were diploid,the Tmrca within each cluster would be twice that in the asexual case.This would increase the size of each cluster in genotypic space relative to the gap between them (Fig.3C).
The important feature of this scenario is that a single lim-ited population is split into two populations of smaller size,whose population sizes are limited independently of each other.After the split,the probability that an individual con-tributes to future generations depends only on the number of individuals in the same population,and hence the populations evolve independently.Constant population size is not nec-essary for these predictions,just some form of limitation on population growth.Similarly,assuming that sexuals and as-exuals are equally likely to form splits of a particular size,allowing unequal splits of the ancestral population would not affect the qualitative predictions.
Barrier to Gene Flow with Selection
Divergent selection between the two populations,for ex-ample,due to environmental differences between areas,can speed up the formation of clusters (Cohan 1998).In asexuals,the entire genome will be homogenized by the most recent selective sweep,and therefore we might observe stronger clustering across multiple loci than in sexuals,depending on what proportion of the sexual genome experiences hitchhik-ing.However,if sexuals adapt faster to their environment than asexuals,the rate and magnitude of adaptive divergence between the two populations might be greater in the sexual case.Although discrete clusters in adaptive traits,such as
2170
BRIEF
COMMUNICATIONS
F I
G .4.Expected variation among clusters for two adaptive characters coded by multiple loci in (A)asexual and (B)sexual cases.Each ellipse represents the range of values observed within a single recognizable cluster.Note that the asexuals display further discreteness within each major cluster,due to the inherent tree structure of quantitative variation in asexual populations.The traits could be mor-phological measures or the ?rst two principal components from a principal components analysis of a set of morphological characters.The number of clusters is the same in the sexual and asexual case,but the magnitude of differences among clusters (disparity)is greater in sexuals.
functional morphology,are expected in asexuals,they might be based on a smaller number of characters or smaller quan-titative differences than in sexuals (Fig.4).The same logic applies to divergence by differential responses to uniform selection.
Other types of selection will have less effect on the relative strength of cluster formation in sexuals and asexuals.For example,background selection against deleterious alleles might reduce Tmrca in asexuals slightly compared to sexuals (because of complete linkage),principally at intermediate selection coef?cients (Neuhauser and Krone 1997).Similarly,diversifying selection in the ancestral population that re-mained in each daughter population could create a compli-cated pattern of coalescence and clustering,but the relative performance of sexuals and asexuals is a direct combination of scenarios discussed already.
Re-establishment of Sympatry
Suppose the barrier to migration is removed.There is now a population of N individuals,N/2of which come from each population.In the asexual case,assuming that all individuals have an equal chance of contributing to the next generation,the expected gene tree will return gradually to a single co-alescent with common ancestor at 2N generations (assuming neutral coalescence).There will be an equal chance that the common ancestor came from either population 1or 2.If the two populations remain independently limited,that is adapted to different niches,the clusters will be stable in the face of population mixing (Cohan 2001).
In the sexual case,with or without interbreeding,the com-bined population will also return to a single coalescent with common ancestor 2N generations ago unless the two popu-lations are independently limited.With interbreeding,the ge-nome-wide pattern of clustering will be lost within a few generations,but each locus will retain the signature of two clusters for up to roughly 2N generations (Avise 2000).Un-linked loci could have ancestral genes from different popu-lations.With no interbreeding,the ancestral gene at all loci will come from the same population (as in the asexual case).This scenario is similar to competitive extinction of one pop-ulation by the other,but it occurs with no difference in com-petitive ability between them,simply by the chance contri-butions of individuals to future generations.
Therefore,for sexuals and asexuals,independent limitation of population size is a key requirement for the stable coex-istence of discrete clusters.This requires ecological differ-ences between the two populations,that is that they are lim-ited by different factors in their shared environment.Without such differences,any clusters formed in isolation will even-tually collapse back to a single cluster.The only alternative would be that rates of cluster formation were fast enough for clusters to exist at a steady state between formation and ex-tinction without the need for stable coexistence (Hubbell 2001).
The signi?cance of sexual reproduction for coexistence is twofold.First,if individuals from the two populations in-terbreed when isolation is removed,differences based on multiple incompletely linked loci will be eroded quickly (Fel-senstein 1981).Hence,if the rate of evolution of reproductive isolation is limiting,sex will counteract the tendency for clusters to persist in sympatry.Second,by speeding up adap-tive divergence between isolated populations,sexual repro-duction might increase the chance that enough adaptive dif-ferences exist between the two populations for them to co-exist.The net outcome will depend on the balance among rates at each step of cluster formation.For example,if en-vironmental events causing isolation and secondary contact are rare,there might be time for both sexuals and asexuals to evolve into discrete clusters with suf?cient divergence to persist even if contact is re-established.Instead,if environ-mental events are frequent,stable clusters might only have time to form in sexual lineages,and then only if reproductive isolation keeps step with rates of adaptive divergence.
2171 BRIEF COMMUNICATIONS
D ISCUSSION
We conclude that asexual organisms are expected to di-versify into discrete genetic and morphological clusters sim-ilar to those found in sexual organisms,as a result of diver-gent selection and/or geographic isolation.In both sexuals and asexuals,stability of clusters in sympatry requires in-dependent limitation of population size,implying different niches,otherwise one cluster will be lost within a few N generations.This is a simple consequence of standard pop-ulation models,and does not rely on special assumptions about ecology or competition.In sexuals,reproductive iso-lation is also necessary otherwise genetic differences between clusters will be lost in just a few generations.Whether as-exuals or sexuals are likely to display more clusters following a given series of events will depend on the balance among different stages of cluster formation.
We have intentionally avoided using the term‘‘species’’because of the dif?culty of applying a universal species con-cept.Instead,we focused on the formation of genotypic and phenotypic clusters.Although such clusters are a part or con-sequence of most species de?nitions,we emphasize that de-?ning species is not necessary for our argument.We believe our conclusions offer general insights into diversi?cation, irrespective of exact species concept subscribed to.
Our account makes several simplifying assumptions.We assumed extreme cases,but of course there can be inter-mediates,such as partial recombination,partial gene?ow, etc.An explicit model of diversi?cation could be devised to look at these complexities and the interactions among dif-ferent stages.In particular,the frequency of environmental events causing changes in population structure or selection regime is an important parameter.Details of the external world are often left out of population genetic models or treat-ed as given,but these details can have a major effect on outcomes.
Also,we assumed that recombination is the only difference between sexuals and asexuals,but in real comparisons this is unlikely to be true.For example,asexuals may be more prone to gene duplication or rearrangements,because there is no constraint for genomes to align at meiosis(Blackman et al.2000).If these changes play a role in adaptive evolution (Lynch and Conery2001),this could affect relative rates of adaptive change in sexuals and asexuals.
Another difference is that sexuals and asexuals might oc-cupy different habitats.For example,bdelloid rotifers tend to be found in unpredictable,desiccating habitats and have dispersive,drought-resistant stages(Ricci1987).Asexuality is one of a suite of characters that allow bdelloids to thrive in these habitats,permitting fast reproduction without the need for a mate.In contrast,sexual monogononts occupy habitats with more predictable water-availability and survive drought in the form of resting eggs.Differences in dispersal rates could affect the tendency to form clusters by geographic isolation,and also how organisms experience spatial envi-ronmental variation.Interestingly,most ancient asexuals are in the1-mm size range often linked to global dispersal ca-pabilities(Finlay and Clarke1999;but see Walter and Lind-quist1995;Schon et al.1998).
So how do our predictions compare to patterns in the real world?The most striking feature of asexuality is that strictly asexual lineages are relatively scarce and tend to be of recent origin(Bell1982;Judson and Normark1996).For example, most asexual plant species result from recent,highly poly-phyletic origins of asexuality(Mogie1992).The resulting diversity can probably be explained mainly by diversi?cation in sexual progenitors,rather than by strictly asexual diver-si?cation.However,some origins of asexuality might have undergone subsequent diversi?cation,for example,Taraxa-cum section hamata comprises a well-established monophy-letic group of25named apomictic species(Kirschner and Stepanek1996).Whether these taxa represent diversi?cation beyond levels that are expected in a single asexual population is unclear.Other examples appear to conform well to ex-pected patterns in a single asexual population,for example, the taxonomic continuum of clones found in the genus Al-chemilla(Maynard Smith and Szathmary1995;Sepp and Paal 1998).
The lack of diversity in asexual clades could be interpreted as evidence for a reduced tendency to form clusters in as-exuals,that is lower origination rates.However,the alter-native is that asexuals have higher extinction rates than sex-uals because of their failure to keep pace with changing biotic and abiotic environments and mutational loads(Burt2000). In this case,many asexual lineages might simply not last long enough for clusters to form,even though their rates of cluster formation could be the same as or greater than that of sexuals.
Fortunately,a few long-lived asexual clades are known that can be compared to sexual relatives.The largest is the bdelloid rotifers,comprising around360named species.Re-cent genetic tests support their asexual status,?nding no evidence for cryptic sex(Mark Welch and Meselson2000). Bdelloid species have been consistently recognized by tax-onomists(Holman1987),and preliminary mtDNA data sup-port the existence of genetic clusters in the group(C.W. Birky,unpubl.data).However,morphological differences be-tween bdelloid species can be slight,and gross morphology is more uniform in bdelloids than in their closest sexual rel-atives,the monogonont rotifers(Ricci1987).Monogononts also comprise more species and occupy a broader range of trophic niches(Donner1966;Wallace and Snell1991).These differences might re?ect lower rates of adaptive evolution in bdelloids,or a more recent origin for bdelloids.Phylogenetic tests are needed to test these alternatives.
Other clades of putative ancient asexuals with apparent diversity exist(Judson and Normark1996;Schon et al.1998; Burt2000),particularly in protists,fungi,and mites,some with genetic evidence for prolonged asexuality(Kuhn et al. 2001).These examples are suggestive of diversi?cation in long-lived asexual lineages,and more speculatively of low rates of adaptive divergence,but at present the evidence is largely anecdotal.Quantitative comparisons of adaptive di-vergence and degree of clustering in related sexual and asex-ual clades are desperately needed,together with genetic tests to con?rm the sexual or asexual history of the clades.Current phylogenetic and genealogical methods,combined with func-tional morphometrics,provide the tools for these studies(Av-ise2000;Monteiro et al.2000;Barraclough and Nee2001). Studies are needed on both recent and ancient asexual groups,
2172BRIEF COMMUNICATIONS
to obtain a comprehensive view of the effects of sexuality on diversi?cation.
Returning to the questions raised in the introduction,we conclude that sexual reproduction is not a necessary condition for diversi?cation into discrete genetic and morphological entities.Asexuals are not expected to display a continuum of forms,not only because of their tree-like ancestry,but also because the processes causing diversi?cation in sexuals are expected to have similar effects in asexuals.In at least some cases,these processes will be more effective in asexuals than in sexuals,but whether sexuals or asexuals display stron-ger overall patterns of diversi?cation is an empirical question. Irrespective of sexuality,diversi?cation is an inevitable con-sequence of the complexity of the physical world,either through isolation in different geographic areas or adaptive divergence in a heterogeneous environment.But comparisons between sexuals and asexuals could shed light on the im-portance of adaptive evolution for the origin and persistence of diversity.Confounding factors aside,the only explanation we can see for stronger patterns in sexuals rather than as-exuals is if faster rates of adaptive change promote greater clustering.
A CKNOWLEDGMENTS
We thank J.Coyne for comments on an early draft.The work is supported by Natural Environment Research Council (NERC)grant NER/A/S/2001/01133and funds from the Uni-versity of Arizona.TGB is a Royal Society University Re-search Fellow.
L ITERATURE C ITED
Avise,J.C.2000.Phylogeography:the history and formation of species.Harvard Univ.Press,Cambridge,U.K. Barraclough,T.G.,and S.Nee.2001.Phylogenetics and speciation.
Trends Ecol.Evol.16:391–399.
Barton,N.H.2000.Genetic hitchhiking.Philos.Trans.R.Soc.
Lond.B355:1553–1562.
Barton,N.H.,and B.Charlesworth.1998.Why sex and recom-bination?Science281:1986–1990.
Bell,G.1982.The Masterpiece of Nature.Univ.of California Press, Berkeley,U.K.
Blackman,R.L.,J.M.Spence,and B.B.Normark.2000.High diversity of structurally heterozygous karyotypes and rDNA ar-rays in parthenogenetic aphids of the genus Trama(Aphididae: Lachninae).Heredity84:254–260.
Burt,A.2000.Perspective:sex,recombination,and the ef?cacy of selection-was Weismann right?Evolution54:337–351. Cohan,F.M.1998.Genetic structure of prokaryotic populations.
Pp.475–489in R.S.Singh and C.B.Krimbas,eds.Evolutionary genetics from molecules to morphology.Cambridge Univ.Press, Cambridge,U.K.
Cohan,F.M.2001.Bacterial species and speciation.Syst.Biol.
50:513–524.
Coyne,J.A.,and H.A.Orr.1998.The evolutionary genetics of speciation.Philos.Trans.R.Soc.Lond.B353:287–305. Dieckmann,U.,and M.Doebeli.1999.On the origin of species by sympatric speciation.Nature400:354–357.
Donner,J.1966.Rotifers.Frederic Warne and Co.,London. Edwards,S.V.,and P.Beerli.2000.Perspective:Gene divergence, population divergence,and the variance in coalescence time in phylogeographic studies.Evolution54:1839–1854. Felsenstein,J.1981.Skepticism towards Santa Rosalia,or why are there so few kinds of animals?Evolution35:124–138. Finlay,B.J.,and K.J.Clarke.1999.Ubiquitous dispersal of mi-crobial species.Nature400:828.Fisher,R.A.1930.The genetical theory of natural selection.Oxford Univ.Press,Oxford,U.K.
Fu,Y.X.,and W.H.Li.1999.Coalescing into the21st century: An overview and prospects of coalescent theory.Theor.Popul.
Biol.56:1–10.
Galtier,N.,F.Depaulis,and N.H.Barton.2000.Detecting bottle-necks and selective sweeps from DNA sequence polymorphism.
Genetics155:981–987.
Hey,J.2001.The mind of the species problem.Trends Ecol.Evol.
16:326–329.
Higgs,P.G.,and B.Derrida.1992.Genetic distance and species formation in evolving populations.J.Mol.Evol.35:454–465. Holman,E.W.1987.Recognizability of sexual and asexual species of rotifers.Syst.Zool.36:381–386.
Hubbell,S.P.2001.The uni?ed neutral theory of biodiversity and Biogeography.Princeton Univ.Press,Princeton,NJ. Hudson,R.R.1991.Gene genealogies and the coalescent process.
Oxf.Surv.Evol.Biol.7:1–44.
Hutchinson,G.E.1968.When are species necessary?Pp.177–186 in R.C.Lewontin,ed.Population biology and evolution.Syr-acuse Univ.Press,Syracuse,NY.
Judson,O.P.,and B.B.Normark.1996.Ancient asexual scandals.
Trends Ecol.Evol.11:41–46.
Kingman,J.F.C.1982.The coalescent.Stoch.Proc.Appl.13: 235–248.
Kirschner,J.,and J.Stepanek.1996.Modes of speciation and evo-lution of the sections in Taraxacum.Folia Geobot.Phytotaxon.
31:415–426.
Kuhn,G.,M.Hijri,and I.R.Sanders.2001.Evidence for the evo-lution of multiple genomes in arbuscular mycorrhizal fungi.Na-ture414:745–748.
Lynch M.,and J.S.Conery.2000.The evolutionary fate and con-sequences of duplicate genes.Science290:1151–1155.
Mark Welch,D.M.,and M.Meselson.2000.Evidence for the evolution of bdelloid rotifers without sexual reproduction or genetic exchange.Science288:1211–1215.
Maynard Smith,J.,and E.Szathmary.1995.The major transitions in evolution.W.H.Freeman,Oxford,U.K.
Mayr,E.1957.The species problem.Ed.Am.Assoc.Adv.Sci., Washington,D.C.
Mishler,B.D.,and M.J.Donoghue,1982.Species concepts:a case for pluralism.Syst.Zool.31:491–503.
Mogie,M.1992.The evolution of asexual reproduction in plants.
Chapman and Hall,London.
Monteiro,L.R.,B.Bordin,and S.F.dos Reis.2000.Shape dis-tances,shape spaces and the comparison of morphometric meth-ods.Trends Ecol.Evol.15:217–220.
Neuhauser,C.,and S.M.Krone.1997.The genealogy of samples in models with selection.Genetics145:519–534. Przeworski,M.,B.Charlesworth,and J.D.Wall.1999.Genealogies and weak purifying selection.Mol.Biol.Evol.16:246–252. Ricci,C.1987.Ecology of bdelloids—How to be successful.Hy-drobiologia147:117–127.
Rice,W.R.1987.Speciation via habitat specialization:the evo-lution of reproductive isolation as a correlated character.Evol.
Ecol.1:301–314.
Schluter,D.2001.Ecology and the origin of species.Trends Ecol.
Evol.16:372–380.
Schon,I.,R.K.Butlin,H.I.Grif?ths,and K.Martens.1998.Slow molecular evolution in an ancient asexual ostracod.Proc.R.Soc.
Lond.B265:235–242.
Sepp,S.,and J.Paal.1998.Taxonomic continuum of Alchemilla (Rosaceae)in Estonia.Nordic J.Bot.18:519–535.
Turelli,M.,N.H.Barton,and J.A.Coyne.2001.Theory and spe-ciation.Trends Ecol.Evol.16:330–343.
Wallace,R.L.,and T.W.Snell.1991.Rotifera.Pp.187–248in J.
H.Thorpe and A.P.Couich,eds.Ecology and classi?cation of
North American Freshwater Invertebrates.Academic Press,New York.
Walter,D.E.,and E.E.Lindquist.1995.The distributions of par-thenogenetic ascid mites(Acari,Parasitiformes)do not support the biotic uncertainty hypothesis.Exp.Appl.Acarol.19: 423–442.
Corresponding Editor:R.Harrison