Pii: s0169-5347(01)02161-9

TRENDS in Ecology & Evolution Vol.16 No.7 July 2001 Phylogenetics and speciation
Timothy G. Barraclough and Sean Nee
Species-level phylogenies derived from molecular data provide an indirect record
Our review discusses recent progress in this area.
of the speciation events that have led to extant species. This offers enormous
We discuss methodological issues in reconstructing potential for investigating the general causes and rates of speciation within
species histories, and move on to the two broad clades. To make the most of this potential, we should ideally sample all the species
questions concerning the rates and causes of in a higher group, such as a genus, ensure that those species reflect evolutionary
speciation. Finally, we consider the future entities within the group, and rule out the effects of other processes, such as
possibilities of incorporating fine-scale GENEALOGICAL extinction, as explanations for observed patterns. We discuss recent practical and
data into large-scale phylogenetic studies.
theoretical advances in this area and outline how future work should benefit from
incorporating data from genealogical and phylogeographical scales.
Species-level phylogenies
Phylogenetic studies of speciation have focused on
SPECIATION (see Glossary) is the creative process reconstructing species-level phylogenies, namely the leading to the build up of species diversity; relationships among species within higher groups understanding the general patterns and processes of such as genera13–15. The INTERNAL NODES of the tree speciation is fundamental to explaining the diversity reflect speciation events: it is known ‘who split from of life1. Central questions include: what are the whom’ and, ideally, the timing of those events.
general causes of speciation, and how do rates of Because reconstruction relies on living species, there speciation vary over time and among TAXONOMIC is no record of speciation events involving species that groups and geographical regions? Although subsequently went extinct, although extinction can fundamental, such questions are difficult to answer leave a signature on the shape of the tree.
because direct observation is usually impossible and Phylogenetic reconstruction from molecular data is many extant groups do not have adequate fossil an advanced science16, but two issues have been records with which to investigate these topics2 (but particularly important when applying phylogenetics see Benton and Pearson3, this issue).
The expansion of molecular PHYLOGENETICS over the The first is the issue of sampling. To obtain an past decade has opened up a powerful new approach accurate view of speciation in a higher group, nearly to this problem. PHYLOGENETIC TREES, particularly all the species from that group should be sampled.
those including all the living species in a higher Missing species reduce the sample size of taxonomic group, provide an indirect record of the reconstructed speciation events available, and can speciation events that have led to present-day introduce bias, for example by tending to remove the species4. Together with information on the most recent speciation events17 or those involving geographical and ecological attributes of species, they rare species7. In particular, the ability to consider the can provide information on the causes of speciation effects of other processes, such as extinction, on the within the group5–7. In addition, trees derived from observed patterns relies crucially on a very complete DNA sequences contain information about the sample of species. However, most phylogenetic relative timing of reconstructed speciation events, studies do not sample all the described species within and can be used to estimate speciation rates8–10.
a group, and, for reasons outlined in Box 1, special This has led to considerable interest in molecular phylogenetics as a tool for solving speciation The second issue relates to the status of species problems, but success rests on two fundamental included in phylogenetic trees. Most phylogenetic issues. First, the approach relies on reconstructing studies of higher clades sample one individual from evolutionary species relationships within a CLADE.
each species named in the checklist for the group.
Few biologists deny that the evolutionary entities Using the resulting tree to study speciation assumes Timothy G. Barraclough*
referred to as species do exist, but the taxonomically that taxonomic species generally reflect the recognized species in a clade might not correspond to evolutionary entities whose origin we wish to explain, those entities11 (see Hey12, this issue). Second, but this need not be so. The number and identity of processes in addition to speciation influence the taxonomic species in a group depends upon the PHYLOGENY and attributes of present-day species1,6.
judgement of the taxonomist who described them, For example, species are lost by extinction, and subsequent phenotypic evolution might obscure the splitting–lumping continuum). More fundamentally, ecological pattern of speciation events. Hence, a it might not always be possible to resolve the true phylogenetic tree cannot simply be read as the history history of speciation into a tree of MONOPHYLETIC of speciation: analyses must factor out alternative species, because some modes of speciation lead to processes as explanations for observed patterns. PARAPHYLETIC species18 or even phylogenetic http://tree.trends.com 0169–5347/01/$ – see front matter 2001 Elsevier Science Ltd. All rights reserved. PII: S0169-5347(01)02161-9 TRENDS in Ecology & Evolution Vol.16 No.7 July 2001 described species appear to conform well to Box 1. Reconstructing species-level phylogenies
evolutionary units14,21, but less so in others22. In the Several methodological problems have faced researchers wanting to absence of detailed information, taxonomically reconstruct species-level phylogenies for studying speciation. The first is defined species can at best be regarded as hypotheses obtaining DNA from all the species of a sufficiently large monophyletic of evolutionary groups (see Hey12, this issue).
group to detect any patterns. Collection of DNA from rare species might be To date, time and expense have forced studies to difficult or impossible and it might be hard to guarantee PCR amplification reach a compromise between these two issues.
from every species using universal primers. Statistical methods exist to Phylogenetic studies have looked at speciation on a investigate incomplete samples of speciesa, but complete samples will still broad scale, allowing investigation into general be needed to detect many patterns. Second, even if samples can be causes and rates, but have usually relied on obtained, it might be difficult to obtain a fully resolved tree. Many existing DNA regions were developed for higher level systematics and might evolve genealogical studies have delimited evolutionary too slowly to resolve the species level. Faster evolving regions are groups more precisely, but, in most cases, have been available, but can have associated problems, such as the existence of restricted to much fewer speciation events by the alternative multiple copies or the difficulty of using universal primers in the need for intensive within-species sampling (but see chosen target groupb. Alternative methods are being explored, such as Ref. 11). As sequencing technology becomes faster genetic fingerprinting techniques or the use of transposable elementsc,d.
and cheaper, the need to compromise is lessening.
Alternatively, we might have to accept that 100% resolution is not We discuss primarily the phylogenetic approach, but always possible at this level, and develop analytical approaches that take also highlight those areas that should benefit most uncertainty into account. For example, Bayesian approaches are being from integrating phylogenetic and genealogical developed that allow estimation of evolutionary parameters, such as speciation and extinction rates, as an integral part of the phylogeny A final area of phylogenetics important for studies reconstruction process, taking account of uncertainty in the treee. Finally, of speciation is the use of molecular data to date nodes existing studies are usually limited to a single exemplar for each described on a phylogeny. Most phylogenetic studies of species in a group, and to one or a few molecular markers. Several speciation rely on information of the relative ages of processes can complicate reconstruction of species history from the speciation events, whether for estimating speciation resulting gene trees, including modes of speciation producing paraphyletic rates or identifying recent nodes most likely to reveal species, lineage sorting of ancestral polymorphisms and hybridizationf,g.
possible causes of speciation. Although methods for Genealogical methods based on reconstructing multi-gene trees and estimating relative node ages from sequence data are sampling larger numbers of individuals per named species are available to advancing rapidly and no longer rely on the cope with these issuesh–j. These approaches allow accurate reconstruction assumption of a strict molecular clock, it can still be of species histories, but to date have been too time-consuming or costly to difficult to calibrate the tree in real time (Box 2).
apply to studies of large groups. Aside from difficulties of collection, the above problems are becoming less severe as molecular methods become Estimating speciation rates
faster and cheaper, making it easier to obtain comprehensive samples for A central question concerning speciation within clades relates to how often speciation occurs. A largeliterature has developed describing methods for References
a Pybus, O.G. and Harvey, P.H. (2000) Testing macroevolutionary models using estimating speciation rates from trees containing all incomplete molecular phylogenies. Proc. R. Soc. London B Biol. Sci. 267, 2267–2272 the species within a clade10. These methods use b Hillis, D.M. et al., eds (1996) Molecular Systematics, Sinauer Associates information on the time elapsed between successive c Verneau, O. et al. (1998) Determining and dating recent rodent speciation events by branching events, and can be illustrated graphically using L1 (LINE-1) retrotransposons. Proc. Natl. Acad. Sci. U. S. A. 95, 11284–11289 by plotting the number of LINEAGES through time d Albertson, R.C. et al. (1999) Phylogeny of a rapidly evolving clade: the cichlid fishes of Lake Malawi, East Africa. Proc. Natl. Acad. Sci. U. S. A. 96, 5107–5110 (Fig. 1). If speciation rates have been constant over e Yang, Z.H. and Rannala, B. (1997) Bayesian phylogenetic inference using DNA time and among lineages, and there has been no sequences: a Markov Chain Monte Carlo method. Mol. Biol. Evol. 14, 717–724 extinction, a straight line with slope equal to the f Avise, J.C. (2000) Phylogeography: the History and Formation of Species, Harvard average per lineage speciation rate, b is expected g Brower, A.V.Z. et al. (1996) Gene trees, species trees, and systematics: a cladistic perspective. Annu. Rev. Ecol. Syst. 27, 423–450 This approach has been used23 to estimate the h Maddison, W.P. (1997) Gene trees in species trees. Syst. Biol. 46, 523–536 average speciation rate during the radiation of i Koufopanou, V. et al. (1997) Concordance of gene genealogies reveals reproductive Hawaiian silverswords (family Asteraceae). This isolation in the pathogenic fungus Coccidioides immitis. Proc. Natl. Acad. Sci. U. S. A. plant group has diversified into a wide range of taxa on the archipelago, from cushion plants and vines to j Kliman, R.M. et al. (2000) The population genetics of the origin and divergence of the Drosophila simulans complex species. Genetics 156, 1913–1931 trees. Using an inferred maximum age for thedivergence between silverswords and theirCalifornian sister clade, a minimum per lineage NETWORKS19. The precise description of species speciation rate of 0.56 ± 0.17 species my−1 was histories has been the focus of genealogical and estimated. This suggests that the silverswords have PHYLOGEOGRAPHICAL studies of speciation11 (see speciated at rates comparable to peak origination Nichols20, this issue), which rely on more intensive rates observed from fossil evidence during sampling within species (Box 1). In some groups, TRENDS in Ecology & Evolution Vol.16 No.7 July 2001 Box 2. Dating phylogenies
DNA sequence data can be used to estimate the relative ages of nodes on a phylogeny. Assuming that nucleotide substitutions between taxa accumulate randomly over time, molecular distances reconstructed onto the phylogeny are expected to be roughly proportional to the time elapseda. However, variation in substitution rates among lineages means that we cannot assume a strict molecular clock in most cases. Recently, a variety of methods have been proposed for estimating the relative ages of nodes, with confidence intervals, from sequence data even in the absence of molecular clock. Most are likelihood methods that fit node ages under explicit or heuristic models of how rates change among lineagesb–d. The exception is Sanderson’s Non-Parametric Rate Smoothing algorithme, which converts an unconstrained tree (in which branch lengths reflect rate as well as time) into an ultrametric tree (in which branch lengths only reflect time) by minimizing rate changes across the tree. These methods have not yet been fully evaluated on real data, but represent an important step forward. Additional complexities can arise when considering very recent speciation events, in which case genetic diversity within populations can have a large effect on estimates of divergence times (see Nicholsf, this Calibration of the tree in real time remains difficult, relying on the availability of fossil dates or biogeographical evidence, which can be lacking in some groups. The traditional approach of using blanket calibrations for rates of molecular evolution, such as the widely used insect mitochondrial DNA clock of 2% pairwise sequence divergence per millionyearsg, is confounded by rate variation among taxa. However, more Fig. 1. Lineages-through-time plot for studying speciation rates. The
sophisticated calibrations should be possible in future, using tests for rate log of the number of lineages is plotted against the relative time of each node since the root node (*) (other graphical representations are variation between study groups and reference clades with calibrated dates.
discussed in Ref. 10). Under the simplest model (the constant This will rely on the availability of a library of well-dated phylogenies speciation rate model), where the probability of a speciation event occurring in a given time is constant both over time and among species, a straight line with slope equal to the per lineage speciation References
rate, b is expected. This corresponds to the pure birth stochastic a Hillis, D.M. et al., eds (1996) Molecular Systematics, Sinauer Associates process, an early statistical model that first arose in precisely this b Thorne, J.L. et al. (1998) Estimating the rate of evolution of the rate of molecular context10. The maximum likelihood estimate of b equals the number of evolution. Mol. Biol. Evol. 15, 1647–1657 reconstructed speciation events that have occurred since the root c Huelsenbeck, J.P. et al. (2000) A compound Poisson process for relaxing the molecular node, divided by the total lineage time available for such events to occur. Confidence intervals for the estimate based on the fact that only d Yoder, A.D. and Yang, Z.H. (2000) Estimation of primate speciation dates using local a finite sample of nodes is available can also be calculated10,23. The molecular clocks. Mol. Biol. Evol. 17, 1081–1090 probability theory underlying statistical inference from phylogenies is e Sanderson, M.J. (1997) A nonparametric approach to estimating divergence times in the closely related, and often identical to population genetics theory used absence of rate constancy. Mol. Biol. Evol. 14, 1218–1231 to make inferences from gene genealogies9,17, but there are some f Nichols, R. (2001) Gene trees and species trees are not the same. Trends Ecol. Evol. 16, differences. For example, population geneticists largely rely on a coalescence approach, which reverses time and imagines the tree shrinking as its branches coalesce at the nodes. However, this g Brower, A.V.Z. (1994) Rapid morphological radiation and convergence among races of the approach does not allow one to theorize about trees that grow butterfly Heliconius erato inferred from patterns of mitochondrial-DNA evolution. Proc. according to a birth–death process (i.e. with extinction), for which a Natl. Acad. Sci. U. S. A. 91, 6491–6495 forward perspective on time is needed8.
The situation is more complicated if the data are indistinguishable, the range of diversification models not consistent with a constant speciation rate model.
consistent with the data can be narrowed down, For example, extinction occurring randomly and at a thereby gaining better estimates of speciation rates.
roughly constant rate over time is expected to cause At present, few studies have applied these an apparent acceleration in speciation rate towards techniques to estimating speciation rates in real clades, the present (Box 3). In this case, it is possible to possibly mainly because of the rarity of sufficiently estimate speciation and extinction rates separately, complete trees. As more data become available, rather than just to estimate the net DIVERSIFICATION particularly those demonstrating the evolutionary RATE17,24. However, as described in Box 3, other status of included species, variations of these methods processes, and sampling and taxonomic artefacts in will be used for broad surveys of speciation rates. In particular, can affect the shape of the plots. Estimates addition, links between genealogical approaches and of speciation rates and their interpretation rely the methods we describe should allow estimation of a heavily on these issues. Nonetheless, although the broader range of parameters, such as the relative rates outcomes of some processes might be statistically of paraphyletic and monophyletic modes of speciation, TRENDS in Ecology & Evolution Vol.16 No.7 July 2001 Box 3. Departures from the constant speciation rate model
predictions will differ for clades with b−d <0.4 species per million years. In the tiger speciation rate model outlined in Fig. 1.
Ellipsoptera group of tiger beetlesa (Fig. I), can be caused by: (1) an actual increase in (Fig. II), can be caused by a decrease in background extinction rate, d, that is high speciation rate or an increase in extinction relative to the (constant) speciation rate, b rate. If background extinction has occurred (Ref. b). In the latter case, the slope near recent upturn above the levelling off stage, the speciation rate, whereas the slope in but beyond that it is probably not possible to distinguish the two explanationsf.
speciation minus extinction rate (i.e. the References
a Barraclough, T.G. et al. (1999) Testing whether ecological factors promote cladogenesis in a result of extinction is expected to start at Cicindelidae). Proc. R. Soc. London B Biol. Sci. b Nee, S. et al. (1994) Extinction rates can be estimated from molecular phylogenies. Philos. Trans. R. Soc. London Ser. B 344, 77–82 c Harvey, P.H. et al. (1994) Phylogenies without could also lead to artefactual upturns or d Klicka, J. and Zink, R.M. (1999) Pleistocene effects on North American songbird evolution.
Proc. R. Soc. London B Biol. Sci. 266, 695–700 e Lovette, I.J. and Bermingham, E. (1999) Dendroica warblers. Proc. R. Soc. London B multiple substitutions at the loci used to f Kubo, T. and Iwasa, Y. (1995) Inferring the saturation of genetic distances, producing molecular phylogenies. Evolution 49, 694–704 shorter branch lengths deeper in the tree.
g Pybus, O.G. and Harvey, P.H. (2000) Testing macroevolutionary models using incomplete molecular phylogenies. Proc. R. Soc. London B problemh, but there has been little general h Hillis, D.M. et al., eds (1996) Molecular and the frequency of hybrid speciation. Even without divergences would be clustered towards the present.
these extensions, the approach offers a means to It is therefore vital to test for a significant increase in investigate otherwise unanswerable questions in speciation rate during the Pleistocene compared to the null model of constant speciation rate. Such a testwas performed using mitochondrial DNA phylogenies of 11 lineages of passerine birds27. It was found that Speciation rates change over time for a variety of net diversification rates decreased towards the reasons and the methods described can be extended to present rather than increased, contrary to the test for such changes. For example, one widespread predictions of the Late Pleistocene origins hypothesis.
hypothesis is that glacial cycles during the Many studies have used similar patterns to argue Pleistocene (2.5 to 0.01 million years ago) increased that diversification rates typically decline during the speciation rates in Northern Hemisphere groups25,26.
radiation of clades28,29. This might occur because the The main evidence for this is that many species and opportunity for speciation decreases as ecological or populations display genetic divergences from their geographical space becomes filled30, or as a nearest relatives consistent with separation during consequence of decreased range sizes following the late Pleistocene. However, even if speciation rates successive subdivisions of ancestral ranges31.
remained constant, it is expected that many However, several artefacts can also create the illusion TRENDS in Ecology & Evolution Vol.16 No.7 July 2001 of early bursts of ‘explosive’ speciation. These include traditional evidence for answering this has been to incomplete sampling, taxonomic underestimates of look at recently split species to infer what processes the number of evolutionary species, and possible might have been important. For example, evidence biases in estimates of node ages (Box 3). Even if none that sexual selection can play a key role in speciation of these problems apply, it might be impossible to initially came from observations of large differences distinguish falling speciation rates from rising in SECONDARY SEXUAL TRAITS among closely related species39. Molecular phylogenetics allows moreprecise tests: species relationships can be more Variation in speciation rate among regions accurately known and information on the timing of Species richness varies widely among regions, but speciation allows us to focus on the most recent what causes hotspots of diversity? Phylogenetic data events. However, a key limitation with this approach can provide insights into this problem. For example, is that changes can occur since speciation: any studies in South American and African tropical birds patterns observed for even closely related species have used phylogenetic data to infer precisely where could be the incidental outcome of the independent in the tropics the most recent speciation occurred33,34.
evolutionary histories of those species, rather than indicative of the forces promoting speciation1,6. Also, phylogenetically youngest species do not tend to be extinction can affect the characteristics displayed by found in the central areas of lowland rain forest, but surviving species. Hence, some way is needed to rather in the surrounding topographically complex, factor out the effects of incidental change and montane areas. This appears to rule out lowland extinction on the patterns observed. We discuss two areas as hotbeds of recent speciation.
With complete species-level phylogenies for study groups, it should be possible to estimate how speciation and extinction rates vary among regions.
Several authors have used species-level phylogenies One example is the recent investigation of the to assess the relative frequencies of different evolutionary basis of a species–area relationship, by geographical modes of speciation, based on the comparing speciation rates of Anolis lizards among geographical distributions of recently formed sister Caribbean islands of different sizes35. Future studies species5,7,40. For example, in a recent study of several could apply similar approaches in continental regions bird, insect and fish groups41, all of the most recent and at a global scale. For example, does high splits had no range overlap, suggesting that speciation and/or low extinction explain the ALLOPATRIC SPECIATION is the norm (with some possible occurrence of floristic hotspots in Mediterranean exceptions). Also, range size differences between climate regions36, and, what are the relative roles of recently split species suggest that speciation often speciation and extinction in causing latitudinal involves the isolation of small populations, so-called peripatric speciation. Other measures can be used toexplore other modes, such as the frequency of PARAPATRIC SPECIATION, or the role of external Speciation rates of taxa might depend on their boundaries, such as rivers and mountain ranges42.
biological characteristics, such as body size or the The Achilles heel of these studies is that ranges of degree of sexual selection. To date, most tests of such species can move, even over very short timescales. If range movements are common, present-day ranges phylogenetic information, simply comparing the might not preserve a record of the mode of speciation.
numbers of species in sister taxa that differ in the In the above study, several clades showed the same trait of interest6. As a result, it cannot be ruled out qualitative pattern of range overlaps as would be that extinction, rather than speciation, might have expected if species ranges had moved to random caused the observed patterns. As more species-level locations within the area occupied by the whole phylogenies become available, it will be possible to clade41. Recent work has tried to deal with this estimate speciation and extinction rates separately problem in several ways: (1) by looking at situations for multiple sister taxa pairs. Similarly, it might be where range movements are unlikely to cloud the possible to test for the effects of rapidly changing results, such as speciation on small oceanic islands43; traits on speciation rates, even in the absence of sister (2) by checking the success of biogeographical taxa differing unambiguously in their expression of methods in cases known to have a particular mode of the trait38. For hypotheses where the effect of the trait speciation, for example presumed SYMPATRIC is uncertain, such as body size, these tests will be SPECIATION of polyploid plants7; or (3) by proposing crucial in narrowing down possible mechanisms.
tests for historical signals, for example randomizingranges of species among the tips of the phylogeny and Causes of speciation
comparing observed overlaps of sister species to those Perhaps the most fundamental question in the study obtained between random pairs of species41. Although of speciation is what causes a single ancestral species choosing a realistic null model can be difficult44, the to split into two (or more) daughter species? The third approach could offer a general way forward.
TRENDS in Ecology & Evolution Vol.16 No.7 July 2001 Box 4. Character shifts and speciation
The locations of shifts in ecological characters can be between recently split species than expected, given average rates reconstructed onto a tree using maximum parsimony or across the tree and the short divergence time of those splitsd; and likelihood methodsa. Figure I shows a hypothetical example of a (2) that the amount of change within lineages is more directly clade with three reconstructed changes (marked by vertical bars) proportional to the number of nodes than to branch lengthsc. If in the state of a discrete character. If ecological shifts occur ecological differences are associated instead with the long-term incidentally over time with no effects on speciation or extinction, persistence and subsequent radiation of lineages, we might we expect shifts to be distributed across the phylogeny roughly in expect that most shifts would be observed between more proportion to the time available for them to occur, that is, the distantly related lineages within the claded.
branch lengths of the treeb,c. If shifts are associated with Any test of these predictions must take into account that the speciation, we expect to see: (1) more ecological differences location of shifts are reconstructed and are therefore prone to uncertainties or biases of the reconstruction methoda. Other discrete or continuous species traits thought to be involved in speciation, such as reproductive morphology or genetic characters, could be treated in the same way.
References
a Cunningham, C.W. and Omland, K.E. (1998) Reconstructing ancestral character states: a critical reappraisal. Trends Ecol. Evol. 13, 361–366 b Pagel, M. (1997) Inferring evolutionary processes from phylogenies. Zool. c Mooers, A.O. et al. (1999) Using phylogenies to test macroevolutionary hypotheses of trait evolution in Cranes (Gruinae). Am. Nat. 154, 249–259 d Barraclough, T.G. et al. (1999) Testing whether ecological factors promote cladogenesis in a group of tiger beetles (Coleoptera: Cicindelidae). Proc. R.
Soc. London B Biol. Sci. 266, 1061–1067 Also, tests for particular modes can look at additional findings at scales where detailed experiments are not predictions, for example that peripatric speciation possible (see Schluter46, this issue). The basic should lead to imbalanced trees45 or that sympatric approach is to map ecological characters onto a tree speciation should involve host shifts6.
(Box 4), with the prediction that recently split species Extinction can also affect geographical patterns, will tend to occupy different niches or habitats. For but its effects have been little explored. Random example, a large body of work on herbivorous insect extinction occurring at a constant rate over time groups has shown that closely related species are tends to prune older lineages17, leaving many recent often found on different host-plant species13,47.
speciation events ‘untouched’, and therefore could Similarly, phylogenetic evidence supports a possible have a fairly minor effect. However, selective role of habitat shifts in several Hawaiian groups, extinction, for example for species with smaller including silversword plants48, Drosophila, and ranges or those that interact with large numbers of spiders from the genus Tetragnatha49. However, close relatives, could have a much larger effect, except in the most extreme cases, there is again the perhaps even biasing evidence for geographical problem that ecological differences between even modes. Another source of bias might come from closely related species could have evolved since taxonomic artefacts, for example if taxonomists were speciation, as a by-product of independent more likely to recognize taxa as species when they are One solution is to compare observed ecological Despite the problems, phylogenetics provides a differences among species to those expected under a framework for considering the effects of other null model in which ecological traits evolve randomly processes and has led to refinements of existing tests.
through time with no effect on speciation (Box 4). This approach is stricter in the criteria needed to phylogeographical data are available for all the demonstrate that ecological shifts are linked to species in a higher clade. This should make it possible speciation, but there remain limitations. Characters to determine, for example, what proportion of that change predominantly at speciation events need populations go on to form reproductively isolated and not play a causal role in the speciation process: the morphologically recognizable species, and what evolution of some characters might be speeded up by distinguishes the winners from the losers.
conditions during speciation, such as smallpopulation size50. Also, even using null models it might be difficult to distinguish rapidly evolving Recent work has renewed interest in the role of characters from those involved in speciation. Hence, ecological shifts in speciation, and phylogenetics theoretical and experimental evidence is vital for provides a means to test the generality of these supporting the likelihood that a given trait is involved TRENDS in Ecology & Evolution Vol.16 No.7 July 2001 promoting REPRODUCTIVE ISOLATION and diversity insexually selected characters? What role do geneticprocesses, such as changes in chromosome number,gene rearrangements and duplications, play in speciation54? Considering the evolution of geneticcharacters with respect to phylogeny will allowestimates of the relative frequencies of different typesof change and tests for their intimate association withspeciation. Work on these questions has alreadystarted, but major advances should be seen over thenext few years, as more complete reconstructions of evolutionary relationships in clades becomeavailable.
Linking genealogical and phylogenetic scales
The previous discussion has shown how broad-scale
phylogenetic studies can help to identify general
trends in speciation, but a potential weakness to date
has been that most studies at this level rely ontaxonomic species as surrogates for evolutionary species. Current technology for DNA sequencing means it is becoming feasible to sample genealogies of individuals within all the species of a higher Fig. 2. Speciation of rock-dwelling cichlid fish (mbuna) in Lake Malawi.
taxonomic group, such as a genus. These data Ecological diversification has been thought to be important in the will allow tests of the validity of using taxonomic speciation of cichlids in the African Great Lakes, but among a group of species in speciation studies, but more importantly mbuna from Lake Malawi, ecological differences in jaw morphology and associated feeding style are observed primarily between distantly will allow a broader range of questions to be related species (i.e. between genera28). For example, Pseudotropheus answered. By identifying boundaries between spp. use their downward-pointing mouths to pinch off mouthfuls of evolutionary species, it will be possible to ask what algae, whereas Metriaclima spp. use their terminal mouths to comb features are associated with species boundaries, food from attached algae and Labeotropheus spp. use their robust sub- terminal jaws to scrape algae from rocks. Field studies confirm a whether the multifarious aspects of species tend to frequent lack of resource partitioning within genera and that closely evolve in concert, and what role the external related species use broadly overlapping resources51. Hence, although environment plays in producing genetic and ecological divergence might have played a role in the early radiation or phenotypic clusters55. Two areas would seem long-term persistence of lineages, it appears to have been less important for recent speciation events that led to the majority of particularly hard to address without integrating diversity in the lake. Instead, closely related species often differ in male genealogical and phylogenetic approaches: the role of nuptial colouration, suggesting that female choice for male colouration hybridization in speciation, and speciation in might be a more important factor in recent speciation. A similar pattern has been found in Nicaraguan crater-lake cichlids15. Adapted, with permission, from Ref. 28. The tree is based on amplified fragment length polymorphisms (AFLP), a DNA fingerprinting technique. Scale bar = 1% difference in the AFLP profile.
Hybrid speciation occurs when hybridization betweentwo species leads to the formation of a new, third in speciation. More work is needed to establish a species. It has been long considered important in strong statistical framework for these kinds of tests, plants, with 11% of plant species richness attributed but, in some cases, interpretation can be more to this mode by recent authors19, but it might also straightforward. For example, ecological shifts play a role in animals56. The main tool of current appear to have played little role in recent speciation research is detailed genetic analysis, but events of Lake Malawi rock-dwelling cichlids, because phylogenetics could play a key role in the future ecological differences are primarily observed between resolution of the general prevalence of hybrid speciation. Current work suggests two ways in which These examples show the potential and some such tests might proceed. First, extensions of pitfalls of using phylogenies to investigate causes of traditional methods of phylogeny reconstruction speciation. Studies that deal critically with the could be used to reconstruct detailed histories of problems have the potential to answer major hybridization and CLADOGENESIS in terms of questions concerning the generality of mechanisms networks57,58. As yet, the practicalities of this outlined by theory or experiments. For example, what approach for large data sets are uncertain: allowing are the relative roles of geographical isolation and for lateral connections among taxa increases the DIVERGENT SELECTION between environments in number of possible solutions among which to search promoting speciation? What are the relative roles of for the optimum. The second approach is to estimate species interactions52 and ecological shifts53 in the frequency of hybridization without reconstructing TRENDS in Ecology & Evolution Vol.16 No.7 July 2001 an explicit history of those events. Current methods questions about speciation. Current studies have have been developed to quantify levels of been caught between the competing demands of recombination among groups of bacterial and fungal sampling enough species to detect any patterns and sequences59,60, but similar tests could be applied to establishing the evolutionary status of the included hybridization between species. A problem common to species. It can still be difficult to reconstruct an both approaches is that it might be hard to accurate species phylogeny for higher groups, but distinguish hybrid formation of a new species from ongoing practical and theoretical advances are gene flow between two existing species that does not making it feasible to meet both demands in a single study. Even so, there is still the need for studies thatsacrifice detail for scale, and vice versa.
Assuming a suitable tree is available, theoretical Sexual reproduction might explain the existence of the questions remain about how best to extract discrete units recognized as species: interbreeding information on speciation and in particular how to maintains coherence within populations, whereas rule out the effects of other processes such as reproductive isolation leads to genetic and phenotypic extinction. The basic theory on rates of speciation is discontinuities among isolated populations (see well established, but extensions are needed to answer Turelli et al.61, this issue). However, if similar genetic questions at finer and broader scales, for example, and morphological clusters were found in asexual what proportion of populations go on to form fully taxa, external ecological factors might be more fledged species, and how does the probability of important than is sexual reproduction in explaining speciation vary geographically? The theory for why species exist. One way to test this idea would be to investigating the causes of speciation is much more compare genealogical histories between closely related diffuse, and a stronger framework is needed before sexual and asexual taxa. The data could be used to: the generality of mechanisms outlined in this special (1) identify whether discrete genetic and issue can be firmly assessed. Crucially, ways are morphological clusters exist in asexual taxa62; (2) test needed to identify when the signal of speciation has whether the degree of clustering is more extreme than been lost entirely from present species.
expected simply from a stochastic birth–death model Finally, we have outlined how major advances Acknowledgements
of asexual division63; (3) compare the rate of origin of over the next few years will probably result from observed clusters between sexual and asexual taxa; bridging the gap between genealogical and and (4) test what role divergent selection from phylogenetic scales. Genealogies allow accurate environmental variables might play in the process.
reconstruction of species histories and insights into population processes, but a phylogenetic perspective Conclusions
is needed to identify general trends and to consider Our aim has been to show the enormous potential of the full array of processes leading to species diversity molecular phylogenetics for answering long-standing References
10 Nee, S. On inferring speciation rates from 20 Nichols, R.A (2001) Gene trees and species 1 Mayr, E. (1963) Animal Species and Evolution, trees are not the same. Trends Ecol. Evol. 16, 11 Avise, J.C. (2000) Phylogeography: the History and 2 Panchen, A.L. (1992) Classification, Evolution Formation of Species, Harvard University Press 21 Avise, J.C. and Walker, D.E. (1999) Species and the Nature of Biology, Cambridge University 12 Hey, J. (2001) The mind of the species problem.
realities and numbers in sexual vertebrates: Perspectives from an asexually transmitted 3 Benton, M.J. and Pearson, P.N. (2001) Speciation 13 Sato, A. et al. (1999) Phylogeny of Darwin’s finches genome. Proc. Natl. Acad. Sci. U. S. A. 96, in the fossil record. Trends Ecol. Evol. 16, 405–411 as revealed by mtDNA sequences. Proc. Natl. 4 Hennig, W. (1966) Phylogenetic Systematics, Acad. Sci. U. S. A. 96, 5101–5106 22 Trewick, S.A. (2000) Mitochondrial DNA 14 Scheffer, S.J. and Wiegmann, B.M. (2000) sequences support allozyme evidence for cryptic Molecular phylogenetics of the holly leaf miners radiation of New Zealand Peripatoides Phylogeny, Ecology, and Behaviour: a Research (Diptera: Agromyzidae: Phytomyza): species (Onychophora). Mol. Ecol. 9, 269–281 Program in Comparative Biology, Chicago limits, speciation, and dietary specialization. Mol. 23 Baldwin, B.G. and Sanderson, M.J. (1998) Age and rate of diversification of the Hawaiian 6 Barraclough, T.G. et al. (1998) Revealing the 15 Wilson, A.B. et al. (2000) Incipient speciation in silversword alliance (Compositae). Proc. Natl. factors that promote speciation. Philos. Trans. R. sympatric Nicaraguan crater lake cichlid fishes: Acad. Sci. U. S. A. 95, 9402–9406 sexual selection versus ecological diversification.
24 Purvis, A. et al. (1995) Macroevolutionary 7 Berlocher, S.H. (1998) Can sympatric speciation Proc. R. Soc. London B Biol. Sci. 267, 2133–2141 inferences from primate phylogeny. Proc. R. Soc. be proven from phylogenetic and biogeographic 16 Hillis, D.M. et al., eds (1996) Molecular London B Biol. Sci. 260, 329–333 evidence? In Endless Forms: Species and 25 Avise, J.C. and Walker, D. (1998) Pleistocene Speciation (Howard, D.J. and Berlocher, S.H., 17 Nee, S. et al. (1994) Extinction rates can be phylogeographic effects on avian populations and eds), pp. 99–113, Oxford University Press estimated from molecular phylogenies. Philos. the speciation process. Proc. R. Soc. London B 8 Hey, J. (1992) Using phylogenetic trees to study Trans. R. Soc. London Ser. B 344, 77–82 speciation and extinction. Evolution 46, 627–640 18 Omland, K.E. et al. (2000) Cryptic genetic 26 Klicka, J. and Zink, R.M. (1999) Pleistocene 9 Purvis, A. (1996) Using interspecies phylogenies variation and paraphyly in ravens. Proc. R. Soc. effects on North American songbird evolution.
to test macroevolutionary hypotheses. In New London B Biol. Sci. 267, 2475–2482 Proc. R. Soc. London B Biol. Sci. 266, 695–700 Uses for New Phylogenies (Harvey, P.H. et al., eds), 19 Rieseberg, L.H. (1997) Hybrid origins of plant 27 Zink, R.M. and Slowinski, J.B. (1995) Evidence species. Annu. Rev. Ecol. Syst. 28, 359–389 from molecular systematics for decreased avian TRENDS in Ecology & Evolution Vol.16 No.7 July 2001 diversification in the Pleistocene epoch. Proc. speciation in birds – a test of Lynch’s method.
through niche partitioning? Oecologia 121, Natl. Acad. Sci. U. S. A. 92, 5832–5835 28 Albertson, R.C. et al. (1999) Phylogeny of a 41 Barraclough, T.G. and Vogler, A.P. (2000) 52 Coyne, J.A. and Orr, H.A. (1997) ‘Patterns of rapidly evolving clade: the cichlid fishes of Lake Detecting the geographical pattern of speciation speciation in Drosophila’ revisited. Evolution 51, Malawi, East Africa. Proc. Natl. Acad. Sci. U. S. A. from species-level phylogenies. Am. Nat. 155, 53 Podos, J. (2001) Correlated evolution of 29 Lovette, I.J. and Bermingham, E. (1999) Explosive speciation in the New World Dendroica Biogeographic and ecological forces responsible for Darwin’s finches. Nature 409, 185–188 warblers. Proc. R. Soc. London B Biol. Sci. 266, speciation in Ateles. Int. J. Primatol. 21, 421–444 54 Mascheretti, S. et al. (2000) Phylogenetics of the 43 Coyne, J.A. and Price, T.D. (2000) Little evidence speciose and chromosomally variable rodent 30 Schluter, D. (2000) The Ecology of Adaptive for sympatric speciation in island birds. Evolution genus Ctenomys (Ctenomyidae, Octodontoidea), Radiation, Oxford University Press based on mitochondrial cytochrome b sequences.
31 Rosenzweig, M.L. (1996) Species Diversity in 44 Gotelli, N.J. and Graves, G.R. (1996) Null Models Space and Time, Cambridge University Press in Ecology, Smithsonian Institution 55 Giessler, S. et al. (1999) Morphological evolution 32 Kubo, T. and Iwasa, Y. (1995) Inferring the rates 45 Chan, K.M.A. and Moore, B.R. (1999) Accounting and genetic differentiation in Daphnia species of branching and extinction from molecular for mode of speciation increases power and complexes. J. Evol. Biol. 12, 710–723 phylogenies. Evolution 49, 694–704 realism of tests of phylogenetic asymmetry. Am. 56 Dowling, T.E. and Secor, C.L. (1997) The role of 33 Fjeldsa, J. (1994) Geographical patterns for relict hybridization and introgression in the diversification and young species of birds in Africa and South 46 Schluter, D. (2001) Ecology and the origin of of animals. Annu. Rev. Ecol. Syst. 28, 593–619 America and implications for conservation species. Trends Ecol. Evol. 16, 372–380 57 Strimmer, K. and Moulton, V. (2000) Likelihood priorities. Biodiv. Conserv. 3, 207–226 47 Bush, G.L. and Smith, J.J. (1998) The genetics analysis of phylogenetic networks using directed 34 Roy, M.S. (1997) Recent diversification in African and ecology of sympatric speciation: a case study.
graphical models. Mol. Biol. Evol. 17, 875–881 greenbuls (Pycnonotidae: Andropadus) supports 58 Xu, S.Z. (2000) Phylogenetic analysis under a montane speciation model Proc. R. Soc. London 48 Baldwin, B.G. (1997) Adaptive radiation of the reticulate evolution. Mol. Biol. Evol. 17, 897–907 Hawaiian silversword alliance: congruence and 35 Losos, J.B. and Schluter, D. (2000) Analysis of an conflict of phylogenetic evidence from molecular measurement of recombination from sequence evolutionary species-area relationship. Nature and non-molecular investigations. In Molecular Evolution and Adaptive Radiation (Givnish, T.J.
60 Goddard, M.R. and Burt, A. (1999) Recurrent 36 Cowling, R.M. et al. (1996) Plant diversity in and Sytsma, K.J., eds), pp. 103–128, Cambridge invasion and extinction of a selfish gene. Proc. Mediterranean-climate regions. Trends Ecol. Natl. Acad. Sci. U. S. A. 96, 13880–13885 49 Roderick, G.K. and Gillespie, R.G. (1998) 61 Turelli, M. (2001) Theory and speciation. Trends 37 Chown, S.L. and Gaston, K.J. (2000) Areas, Speciation and phylogeography of Hawaiian cradles and museums: the latitudinal gradient in terrestrial arthropods. Mol. Ecol. 7, 519–531 62 Taylor, D.J. and Foighil, D.O. (2000) Transglobal species richness. Trends Ecol. Evol. 15, 311–315 50 Cheverud, J.M. et al. (1999) Epistasis and the comparisons of nuclear and mitochondrial genetic 38 Pagel, M. (1997) Inferring evolutionary processes evolution of additive genetic variance in structure in a marine polyploid clam (Lasaea, from phylogenies. Zool. Sci. 26, 331–348 populations that pass through a bottleneck.
Lasaeidae). Heredity 84, 321–330 39 Darwin, C. (1871) The Descent of Man and 63 Higgs, P.G. and Derrida, B. (1992) Genetic Selection in Relation to Sex, John Murray 51 Genner, M.J. et al. (1999) Foraging of rocky distance and species formation in evolving 40 Chesser, R.T. and Zink, R.M. (1994) Modes of habitat cichlid fishes in Lake Malawi: coexistence populations. J. Mol. Evol. 35, 454–465 TREE online – making the most of your personal subscription
Links to other articles, other journals and cited software and databases To claim your FREE online access to Trends in Ecology and Evolution:
Go to http://www.bmn.com/general/subkey and select Trends in Ecology Enter your own BioMedNet login details when prompted (if you are not yet a member, joining takes minutes and is FREE) Follow the instructions on the Trends in Ecology and Evolution page You only need to register once.
For subsequent visits, bookmark: http://journals.bmn.com
Tip: If you do not use a shared terminal, you can tick the ‘save password‘ box
when you first log on to BioMedNet so that you only need to register once If you have any questions e-mail: [email protected]

Source: http://evolution.powernet.ru/library/Scienc2.pdf