© 2006 Society of Systematic Biologists
Differentiating between Hypotheses of Lineage Sorting and Introgression in New Zealand Alpine Cicadas (Maoricicada Dugdale)
Edited by Tim Collins: Associate Editor
1 Landcare Research Private Bag 92170, Auckland New Zealand buckleyt{at}landcareresearch.co.nz
2 Department of Ecology and Evolutionary Biology, University of Connecticut Storrs Connecticut 06269–3043 USA michael.cordeiro{at}huskymail.uconn.edu (M.C.); david.marshall{at}uconn.edu (D.C.M.); csimon{at}uconn.edu (C.S.)
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
Lineage sorting and introgression can lead to incongruence among gene phylogenies, complicating the inference of species trees for large groups of taxa that have recently and rapidly radiated. In addition, it can be difficult to determine which of these processes is responsible for this incongruence. We explore these issues with the radiation of New Zealand alpine cicadas of the genus Maoricicada Dugdale. Gene trees were estimated from four putative independent loci: mitochondrial DNA (2274 nucleotides), elongation factor 1-
(1275 nucleotides), period (1709 nucleotides), and calmodulin (678 nucleotides). We reconstructed phylogenies using maximum likelihood and Bayesian methods from 44 individuals representing the 19 species and subspecies of Maoricicada and two outgroups. Species-level relationships were reconstructed using a novel extension of gene tree parsimony, whereby gene trees were weighted by their Bayesian posterior probabilities. The inferred gene trees show marked incongruence in the placement of some taxa, especially the enigmatic forest and scrub dwelling species, M. iolanthe. Using the species tree estimated by gene tree parsimony, we simulated coalescent gene trees in order to test the null hypothesis that the nonrandom placement of M. iolanthe among gene trees has arisen by chance. Under the assumptions of constant population size, known generation time, and panmixia, we were able to reject this null hypothesis. Furthermore, because the two alternative placements of M. iolanthe are in each case with species that share a similar song structure, we conclude that it is more likely that an ancient introgression event rather than lineage sorting has caused this incongruence.
Keywords: Cicada; combined data; hybridization; introgression, lineage sorting; reconciled trees
Received May 11, 2005; Revised August 8, 2005; Accepted November 26, 2005
Recent studies have shown that incongruence among different gene trees is a widespread problem in phylogenetic studies of recent and rapidly radiating species, including insects (e.g., Kliman et al., 2000; Sota and Vogler, 2001; Ballard et al., 2002; Beltràn et al., 2002; Davies and Bermingham, 2002; Shaw, 2002; Besansky et al., 2003; Broughton and Harrison, 2003; Gómez-Zurita and Vogler, 2003; Machado and Hey, 2003; Wilder and Hollocher, 2003; Hughes and Vogler, 2004). Much of this incongruence is the result of the stochastic nature of the coalescence process, causing the random retention and extinction of alleles that is known as lineage sorting (e.g., Tateno et al., 1982; Pamilo and Nei, 1988; Maddison, 1997; Nichols, 2001). In addition, the presence of semipermeable species boundaries following speciation allows at least limited gene flow or introgression (e.g., Maddison, 1997; Nichols, 2001; Seehausen, 2004), further complicating the inference of species phylogenies.
Given the ubiquity of incongruent gene trees, the need to develop methods for estimating the species tree from different gene trees and for quantifying the relative roles of introgression and lineage sorting are pressing (e.g., Funk, 1985; Sang and Zhong, 2000; Holder et al., 2001; Degnan and Salter, 2005). Distinguishing between hypotheses of introgression and lineage sorting is difficult due to the large variance associated with the coalescent process and the fact that both processes can produce similar phylogenetic patterns (Holder et al., 2001). Here we have integrated an a posteriori coalescent-based simulation test (see also Knowles and Maddison, 2002; Knowles, 2004; Carstens et al., 2004, 2005) with information from species-specific courtship songs to tease apart the relative roles of introgression and lineage sorting for the New Zealand alpine cicada genus Maoricicada Dugdale. We have also reconstructed the species tree using reconciled trees/gene tree parsimony, which explicitly accounts for lineage sorting events in a nonprobabilistic manner (Page, 1994; Maddison, 1997; Ronquist, 1995; Page and Charleston, 1997). To account for uncertainty in the individual gene trees, we have estimated the species trees using sets of gene trees weighted by their posterior probabilities (see also Cotton and Page, 2002).
The fourteen species of the genus Maoricicada (Fleming, 1971; Dugdale, 1972; Dugdale and Fleming, 1978; Buckley et al., 2001a, 2001b, 2001c) are predominantly montane; however, five species are found almost exclusively in low-altitude habitats such as dry river-beds, clay banks, and, in one case, forest and scrub. Fleming (1971) suggested the forest and scrub species M. iolanthe was sister species to the remaining Maoricicada species. In a previous study, Buckley et al. (2001c) examined the phylogenetic position of M. iolanthe using mitochondrial DNA (mtDNA) data and found it occupied a derived position in the Maoricicada radiation, being nested within a largely alpine clade. We have further examined the phylogenetic position of M. iolanthe by increasing the number of Maoricicada populations sampled, increasing the length of the mtDNA sequences, adding data from three putatively unlinked nuclear loci, and integrating information from the species-specific courtship songs. We use these analyses to investigate the relative roles of lineage sorting and introgression in the radiation of this genus.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Laboratory Methods and Data Collection
Adult cicadas were collected from the localities listed in Table 1. For almost all species we sequenced individuals from at least two localities, and in some cases up to six localities, to adequately sample over the geographic range of each species (Funk and Omland, 2003). To root the resulting phylogenetic trees, we used the species Rhodopsalta leptomera and R. cruentata, which have been shown to be suitable outgroups based on more extensive studies of cicadettine cicadas (Buckley et al., 2002; Arensbuger et al., 2004; Vanderpool et al., unpublished data). The mitochondrial cytochrome oxidase subunit I (COI), ATP synthetase subunits 6 (A6) and 8 (A8), and tRNAAsp sequences were obtained using the primers described in Buckley et al. (2001b) and Table 2. The cytochrome oxidase subunit II (COII), tRNALeu, and tRNALys sequences were obtained using the primers TL2-J-3033 and TK-N-3786 from Simon et al. (1994). We designed PCR primers to amplify three nuclear markers: elongation factor 1-
(EF1-), period, and calmodulin (Table 2). The period region was amplified and sequenced in two fragments using the primers Per-205-for + Per-925-rev, and Per-491-for + Per-1201-rev. The EF1- region was also amplified and sequenced in two fragments using the primers EF1a-97-for + EF1a-189-rev and EF1a-236-for + EF1a-352-rev. We used ABI BigDye versions 3.0 and 3.1 following the manufacturers instructions. Electrophoresis was performed on an ABI 310 or an ABI 3700 DNA sequencer. The DNA sequences have been submitted to GenBank under the accession numbers DQ178417
[GenBank]
to DQ178592
[GenBank]
and DQ194391
[GenBank]
to DQ194552.
|
|
Phylogenetic Inference
Alignment of the mtDNA sequences was trivial because only one indel was observed in the tRNAAsp gene. The nuclear intron sequences did contain some indels, mainly short repeated motifs. These were aligned using ClustalX (Thompson et al., 1994) and adjusted manually. The nuclear exons were all length invariant. A small number of individuals displayed heterozygous genotypes for the nuclear loci, and these were coded using standard degeneracy codes.
The assumption of neutrality was tested for the mtDNA locus using a range of codon-based models (e.g., Yang et al., 2000) as implemented in Paml 3.14 (Yang, 1997). The first (M0) was a model where all sites are assumed to have the same nonsynonymous/synonymous ratio (
). The second, a neutral model (M1), assumes a proportion of sites where = 0 and a proportion of sites where = 1. The third, a selection model (M2), has an additional parameter where a proportion of sites have a value of estimated from the data. The fourth model (M3) allows for a proportion of sites to be under positive selection by having the value of at a site a random variable drawn from a three-parameter discrete distribution. In the fifth model, or branches model (Yang, 1998), we allowed the value of to vary independently for every branch in the tree and assumed a distribution for as in M0. The significance of these comparisons was assessed using a likelihood-ratio test.
Each of the sampled loci encodes gene products that perform very different metabolic functions and are therefore likely to be under different selective constraints. Furthermore, we assumed all these loci are physically unlinked and therefore each represents a "linkage partition" (Slowinski and Page, 1999), so a separate phylogenetic analysis of each is valid. We also performed a "combined analysis" where all of the loci were concatenated into a single character matrix.
All maximum likelihood (ML) calculations and tree searches were performed using PAUP*4.0b10 and b11 (Swofford, 1998). The best-fit model for each data set was identified using the Akaike information criterion (AIC; Akaike, 1973; Posada and Buckley, 2004). For determining the best-fit model, we estimated the ML tree under the JC69 model (Jukes and Cantor, 1969) with a single random addition replicate followed by TBR branch swapping. We then calculated the likelihood of this tree under the 56 substitution models implemented in ModelTest 3.0 (Posada and Crandall, 1998). Our final maximum likelihood trees under the resulting best-fit models were inferred using TBR branch swapping from 10 random addition replicates in order to limit the computational burden. For each data set we performed a bootstrap analysis (Felsentein, 1985) with 200 pseudoreplicates, with 4 to 10 random addition replicates followed by TBR branch swapping.
Bayesian phylogenetic analyses were performed using MrBayes 3.0b4 (Ronquist and Huelsenbeck, 2003). Because some Bayesian methods of assessing statistical support may be more susceptible to model misspecification than conservative nonparametric methods, such as the bootstrap (e.g., Shimodaira, 2001; Waddell et al., 2001; Buckley, 2002; Aris-Brosou, 2003; Huelsenbeck and Rannala, 2004), we implemented more complex substitution models for the Bayesian analysis of the mtDNA than for the maximum likelihood analyses. We implemented two classes of models: first, for each of the nuclear genes, we assumed all sites were evolving under the same model because each nuclear gene contained very low levels of homoplasy; and second, for the mtDNA and the combined gene analyses, we partitioned 3rd codon positions from non-3rd positions (1st, 2nd, and tRNA sites) and assigned different substitution models to these different site classes (e.g., Yang, 1996; Nylander et al., 2004; Brandley et al., 2005). For the mtDNA data, the 1st codon positions, 2nd codon positions, and tRNA sites were partitioned from the 3rd codon positions because it was known from previous studies that the substitutional dynamics are radically different between these two classes of sites (Buckley et al., 2001a). For the combined analyses, all the nuclear sites were assigned to a single partition, separate from the mitochondrial 3rd and non-3rd sites.
The prior distributions for the Bayesian analyses are given in Table 3. The chain mixing was poor for some of the MCMC runs on the nuclear loci using vague branch length priors (
= 10). These analyses were therefore repeated with a much more informative prior distribution on the branch lengths ( = 100), which improved chain mixing considerably, and gave parameter estimates much closer to those from maximum likelihood. We ran four MCMC chains simultaneously, one cold and three incrementally heated (heating parameter = 0.2), with a thinning interval of 100 starting from a random tree. All the MCMC chains were run for 10 million cycles, with the exception of the calmodulin data set, which was run for 20 million cycles. Every MCMC analysis was performed twice. To assess convergence of the topology we compared the posterior probabilities of different splits between pairs of identical runs (e.g., Huelsenbeck et al., 2001; Miller et al., 2002) and examined the chain exchange rates, low values of which can be indicative of poor chain mixing. We also used an expanding window approach where splits frequencies were calculated in increments of one million cycles and monitored over time (Warren et al., 2003)
|
Because the mtDNA partition exhibited the most extreme levels of homoplasy, and the outgroup taxa had by far the longest branches in the data set, we repeated the ML and Bayesian analyses on the mtDNA data by removing the two outgroup taxa to examine the effects that the presence or absence of these taxa had on the analyses (Holland et al., 2003; Jordan et al., 2003).
We used the Shimodaira and Hasegawa test (1999; SH test) to test hypotheses concerning the fit of individual gene trees to other partitions. The SH tests were implemented using the RELL approximation with 10,000 bootstrap replicates. We tested the hypothesis that the combined analysis tree explains the data from a given partition as well as the maximum likelihood tree for that particular partition (see Buckley et al., 2002). This was achieved by building a set of candidate topologies that included the four gene trees from each locus and the combined analysis tree. This candidate set of five trees was then examined using the data from each of the five data sets.
We used gene tree parsimony (Page, 1994; Maddison, 1997; Ronquist, 1995; Page and Charleston, 1997) as implemented in the program GeneTree 1.3.0 (Page, 1998) to infer species-level relationships within Maoricicada. In gene tree parsimony we start by assuming that an association exists between alleles and the species that they are sampled from. The alleles have shared relationships defined by a gene tree and the species are related by a species tree. The gene trees will track the species tree to some extent, depending on the frequency of a range of biological events. In the implementation here, the only biological events allowed are "deep coalescent" or lineage sorting events. The species tree has a score, which is calculated by counting the number of events implied by fitting the gene tree to the species tree. Where multiple gene trees from different loci are available, the score of the species tree is the sum of scores across individual gene trees. The resulting optimal species tree is the topology that minimizes the number of lineage sorting events required to fit all of the gene trees to a given species tree. Note that this method does not require the gene trees to be monophyletic with respect to a given species. All sampled alleles were associated with their respective species, with the exception of M. nigra, which we did not assume to be a single species because none of the sampled loci supported its monophyly, and there are marked differences between the courtship songs of the two subspecies (see below).
To accommodate phylogenetic uncertainty, we searched for the shortest species tree using 10,000 draws from the MrBayes-generated posterior distribution of trees for each locus using the "gene tree bootstrapping" option of GeneTree (see Cotton and Page [2000] for a frequentist approach based on the bootstrap). The frequency that each node is recovered can be regarded as the posterior probability that each node occurs in the true species tree for these four loci. The approach we have taken here ignores the error that would be associated with estimating each species tree using a fully probabilistic approach (e.g., Arvestad et al., 2003), and therefore these posterior probabilities are likely to be somewhat liberal (M. Holder, personal communication). Each species tree search was repeated 5 to 10 times from random trees, followed by alternating NNI and SPR rearrangements with steepest ascent and all trees of equal cost retained.
Because our analyses suggest that uncertainty surrounding the phylogenetic position of M. iolanthe may be due to introgression, which is not accounted for by the assumptions of gene tree parsimony as implemented here, we repeated the analyses by pruning this species from all of the gene trees prior to estimating the species tree. We also tested the effect the mtDNA data set had on the results by performing a search with this partition excluded. This analysis was performed because it has been argued that the mitochondrial genome is more susceptible to introgression than nuclear loci (Machado and Hey, 2003; Ballard and Whitlock, 2004).
Coalescence Simulations
Our phylogenetic analyses demonstrated that the species M. iolanthe occupied two different phylogenetic positions in the estimated gene trees. For this reason we used an a posteriori coalescent-based simulation test to further investigate the cause of this incongruence. If we assume that each ancestral species possesses alleles that assort into descendant species independently among loci, then we expect the nature of incongruence among loci to be random. On the other hand, if two species have hybridized and alleles from one species have successfully introgressed into the other species, then the pattern of incongruence will be nonrandom and similar across loci. We performed coalescent simulations to test the null hypothesis that there has been no introgression with respect to M. iolanthe. We refer to this null hypothesis as the "lineage-sorting hypothesis." The alternative hypothesis states that that the nonrandom variation in phylogenetic placement of M. iolanthe among the different gene trees is unlikely to be the result of chance alone and therefore more likely to be caused by introgression rather than lineage sorting. We refer to this alternative hypothesis as the "introgression hypothesis." Introgression events involving more than two species could also create a pattern of a random incongruence across loci, and thus failure to reject the null hypothesis will not always be indicative of a lack of introgression.
We used the species tree topology with molecular-clock branch lengths optimized using one randomly selected sequence from each species, although the test itself was performed using all individuals. For the taxa M. m. mangu and M. clamitans, we did not choose the individuals that had aberrant haplotypes. We assumed a generation time of 3 years, based on the known life cycle length of the related New Zealand cicada species Kikihia muta (Cumber, 1952) and the North American cicadettine species Cicadetta calliope (Beamer, 1920). The depth of the species tree was set to four million generations based on this generation time estimate and the molecular dating study of Buckley and Simon (unpublished data). We performed three simulations with panmictic and constant effective population sizes of 1 x 105, 1 x 106, and 1 x 107 individuals, covering the range of biologically realistic population sizes.
Using Mesquite (Maddison and Maddison, 2004) we simulated 100 sets of four coalescent trees, mimicking the four sampled loci, for the 44 individuals sampled in this study. This process operates by starting with each gene copy and simulating the coalescence of these gene copies under a simple coalescent model of panmixia, fixed effective population size, and generation time. For each of these 100 draws of four gene trees, we counted the number of times M. iolanthe fell in one phylogenetic position for two of the gene trees and another position for the other two gene trees, which we refer to as a nonrandom pattern. The expectation under the null hypothesis is that the frequency of nonrandom patterns will not exceed 5%. If the frequency at which nonrandom patterns occurs exceeded 5%, then the lineage sorting hypothesis was rejected (Fig. 1).
|
Analysis of Species-Specific Songs
For each species male songs were recorded in the field from multiple populations using a SONY TCD-D8 digital audio recorder and later analyzed using Canary v1.2.4 (Cornell Bioacoustics Workstation, Ithaca, NY, USA). Advertisement songs of cicadas (Claridge, 1985) and other acoustically signaling insects (e.g., Walker, 1964) tend to be species specific, with little variation among calls of the same individual or across different males of the same species. In all New Zealand species, these songs facilitate pair formation by eliciting timed wing-flick signals from females (e.g., Lane, 1995). Male songs are organized into an "introductory phrase," which may communicate male species identity, followed by a "cueing phrase," which in Maoricicada consists solely of one or more short notes that trigger the female responses (Fig. 2) (Marshall et al., unpublished data).
|
To characterize similarities between the songs of different species, nine song characters were identified and coded (Table 4). Only temperature-independent traits involving the number and arrangement of song units were used. Character descriptions were completed by adapting the structural terminology of Ragge and Reynolds (1998), with "syllable" referring to the sound from a single in-out tymbal click and "echeme" referring to a first-order assemblage of syllables. The term "phrase" used here corresponds to the "echeme sequence" of Ragge and Reynolds (1998). Some character states for M. hamiltoni and the M. nigra subspecies are tentative because the songs of these species are highly divergent. Character 8 was derived from examination of oscillograms in light of the discussion in Fleming (1975).
|
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Properties of Sequence Evolution and Model Selection
The likelihood scores for the codon models calculated from the mtDNA data were –10,358.574426, –10,789.888807, –10,206.976806, –10,197.297848, and –10,306.608661 for models M0, M1, M2, M3, and branches models, respectively. When we tested M0, M1, and M2 against M3, all likelihood ratios are significant at P < 0.0001. These results suggest the value of
varies among sites; however, under the M3 model, no sites were inferred to be under positive selection. The high number of sites with values close to 0 indicates most sites are under strong purifying selection. We tested M0 against the branches model, which gave a P-value of 0.07977, indicating that allowing to vary across lineages does not yield a significant improvement in fit. The nuclear markers showed substantial differences in substitutional dynamics from the mitochondrial genes. The nuclear loci had much less extreme among-site rate variation, transition bias, and slightly less nucleotide bias (Table 5), consistent with other studies (e.g., Lin and Danforth, 2004). The homoplasy indices calculated on the maximum likelihood trees are much lower for the mitochondrial partition than for the nuclear loci; however, the nuclear loci contain far fewer parsimony informative sites (Table 5). Analysis of chain exchange frequencies indicated the MCMC chains were exchanging states at appropriate frequencies. Plots of clade posterior probabilities through time and comparison of posterior probabilities between runs indicated the chains had reached stationarity (Table 3).
|
Phylogenetic Relationships among Maoricicada Species
Each of the different partitions gave quite different estimates of the phylogenetic position of some taxa (Fig. 3). The strong nature of this incongruence is supported by the pairwise SH tests, which were all highly significant with the exception of when we compared the combined tree (Fig. 4) versus the mtDNA tree using the mtDNA and combined data sets (Table 6). Because we were able to reject the null hypothesis that the combined analysis tree and the optimal trees from the individual loci are equally good explanations of the data, we conclude that for each locus, with the exception of the mtDNA locus, the combined analysis tree is a poor fit to the observed data. We also note that the posterior probabilities tended to be more extreme than the bootstrap proportions (e.g., Huelsenbeck and Rannala, 2004). Furthermore, the presence of hard polytomies in our ML trees suggests that this effect could be at least in part due to the bias described by Lewis et al. (2004), whereby the prior distribution on branch lengths excludes zero length branches.
|
|
|
All loci supported the monophyly of the genus, a pattern confirmed in other studies with a much greater range of outgroups but with fewer loci (Buckley et al., 2002; Arensburger et al., 2004a). Despite the marked topological incongruence, some species groupings are supported by two or more loci. For example, the two low-altitude species M. myersi and M. lindsayi always form a clade (with calmodulin, M. iolanthe is included in this clade). Other well-supported clades are M. clamitans + M. oromelaena (mtDNA, period), M. m. mangu + M. m. multicostata (mtDNA, EF1-
), and M. otagoensis + M. n. nigra + M. n. frigida + M. phaeoptera (mtDNA, EF1-). Most of the above relationships are not strongly contradicted by other loci. The results from the combined analysis of all loci (Fig. 4) were very similar to those of the mtDNA data (Fig. 3). There are very few phylogenetic relationships that appear in the combined analysis that are not supported by analyses of the individual loci. One exception to this is the clade containing M. tenuis and most populations of M. mangu, a relationship hypothesized by Dugdale and Fleming (1978). This clade is not recovered in the separate analysis of any locus, yet it receives a very high posterior probability (100%), albeit with weak bootstrap support (56%) in the combined analysis.
Removal of the outgroups had a dramatic effect on bootstrap support values for some nodes in the mtDNA phylogeny (Fig. 4). For example, support for the monophyly of M. mangu, excluding the Hakataramea (M. mangu 13) and Awakino Ski Field (M. mangu 6), populations increased from 76% to 98%. The most extreme increase was for the node that supported the monophyly of most of the alpine species, including M. oromelaena, M. clamitans, M. alticola, M. mangu (6 and 13), M. nigra, M. otagoensis, M. phaeoptera, M. iolanthe, and M. campbelli, which increased from < 50% bootstrap support to 92% when the outgroups were excluded.
We also recovered some phylogenetic relationships that strongly conflict with our a priori expectations based on the present taxonomy. For example, populations of M. m. mangu from Hakataramea and Awakino Ski Field fail to group with other M. m. mangu and the other subspecies of M. mangu for all loci with the single exception of the EF1- locus. Furthermore, the individual of M. clamitans collected from Awakino Ski Field also fails to group with other sampled individuals of this species at the period locus. Future research in our laboratories will attempt to understand the uniqueness of these M. mangu and M. clamitans populations at the southern extremity of their respective ranges.
Perhaps the most striking example of incongruence involves the enigmatic species M. iolanthe. This low-altitude species, the only Maoricicada species to inhabit forest and scrub, groups with either M. campbelli or with the M. lindsayi and M. myersi clade. Interestingly, all of these four species tend to inhabit low-altitude habitats, have similar genitalic morphology, are of a similar size, and, additionally, all have similar songs (Fleming, 1971; Dugdale and Fleming, 1978; Fig. 5). Using the coalescent-based simulation test, we tested the null hypothesis that the non-random variation in phylogenetic placement of M. iolanthe is due to lineage sorting. Under population sizes of 1 x 105, 1 x 106, and 1 x 107, we were able to reject the null hypothesis with P-values of < 0.01, 0.01, and < 0.01, respectively.
|
The species tree, assuming that deep coalescent/lineage sorting events are the only cause of gene tree conflicts, grouped M. iolanthe, M. myersi, and M. lindsayi in a clade with a posterior probability of 79% (Fig. 5). Other relatively well-supported clades are M. clamitans + M. oromelaena + M. alticola with 89% support, and M. n. nigra + M. n. frigida + M. phaeoptera with 71% support (Fig. 5). Removal of M. iolanthe from the gene tree parsimony analysis had a profound effect on several nodes, and the resulting species tree was more similar to the combined analysis tree (Fig. 4). Removal of M. iolanthe caused the position of M. campbelli to shift from being sister group to the M. lindsayi + M. myersi + M. hamiltoni clade to being sister group to the M. alticola + M. oromelaena + M. clamitans clade (84% posterior probability). In general, resolution improved with removal of M. iolanthe. Those taxa that had different placements in the two trees tended to differ by a small number of branch rearrangements.
Exclusion of the set of gene trees estimated from the mtDNA had several noticeable effects on the species tree. Support for the placement of M. iolanthe as sister group to the M. myersi + M. lindsayi clade increased from 79% to 100% posterior probability, almost certainly because the signal from the mtDNA gene trees for placing M. iolanthe with M. campbelli was removed. Support for the clade containing all the alpine species was still low (< 50%); however, support for some phylogenetic relationships within this clade actually increased. Overall, exclusion of the mtDNA gene trees had little effect on many nodes, suggesting that the loci are beginning to converge on a common species tree.
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Differentiating between Hypotheses of Lineage Sorting and Introgression
By comparing phylogenetic patterns at multiple gene loci, we have been able to reveal a complex pattern of both lineage sorting and ancient introgression among Maoricicada species. Although the inferred relationships among some sampled individuals are incongruent across loci and methods of analysis, some general patterns are evident. The low-altitude species M. lindsayi and M. myersi tend to group together and there is also a tendency for the alpine species M. clamitans and M. oromelaena to form a clade. A clade containing the species M. mangu and M. tenuis, a relationship hypothesized by Dugdale and Fleming (1978), was only supported by the combined analysis. Given that the other analyses tended to place these two species separately, this observation may be an example of "hidden support" (e.g., Gatesy et al., 1999), revealed by combining the data sets. Other species show highly unstable positions within the Maoricicada radiation, namely M. alticola, M. cassiope, M. campbelli, and M. iolanthe.
Because many of the conflicting nodes are well supported, and levels of sequence divergence are relatively low, it seems unlikely that errors in phylogenetic reconstruction are to blame for much of the observed incongruence with respect to the phylogenetic position of M. iolanthe. We were unable to detect any mtDNA sites undergoing positive selection, and, furthermore, a model where the pattern of selection was allowed to vary across the tree did not fit the data significantly better than a model where the selective regime across branches was homogeneous.
Excluding the possibilities of an unusual selective event and gross errors in phylogenetic reconstruction, we conclude this incongruence is the product of either lineage sorting or introgression. Differentiating between hypotheses of introgression and lineage sorting is extremely difficult because both processes can generate very similar phylogenetic patterns (Holder et al., 2001). Two lines of evidence suggest introgression is more likely to be responsible for the ambiguity of the phylogenetic position of M. iolanthe. First, separate phylogenetic analyses of all four loci always place M. iolanthe with either M. campbelli or with the M. lindsayi and M. myersi clade. Furthermore, unpublished sequence data from the actin locus also places M. iolanthe with M. lindsayi and M. myersi rather than M. campbelli, reinforcing this nonrandom pattern. If lineage sorting was responsible for the incongruence, then we might expect the phylogenetic position of M. iolanthe to show more variation across loci, as is observed for some other Maoricicada species such as M. cassiope. In fact, the coalescent-based simulation test we implemented does indeed suggest this nonrandom pattern is highly unlikely to have arisen by chance as might be expected under many models of lineage sorting. Furthermore, we were able to reject the null hypothesis under all assumed effective populations sizes. Below we discuss the interpretation of this test and its limitations.
The distribution of gene trees generated from the coalescence simulations was strongly dependent on the assumed population size. For example, when we assumed a population size of 1 x 104, all species for which we sampled at least two individuals were monophyletic in 100% of the simulated gene trees. When the population size was increased to 1 x 105, this value was 100% for seven species and 99% for two species. When the population size was further increased to 1 x 106, the percentage of simulated gene trees with a given species monophyletic ranged from 17% to 67%. Therefore, as expected, increasing the effective population size caused a concomitant increase in the frequency of lineage sorting events; however, the results of the hypothesis test itself did not vary with changes in the assumptions regarding population size.
Despite this result, the true population sizes are unlikely to be constant over time and across species. Although we could have allowed the ancestral population sizes to vary in a step-wise manner, it is not possible to estimate ancestral population sizes without intensive intraspecific sampling. Additionally, the effective population size is not equal among loci, because we have a mixture of nuclear and mitochondrial genes. A further complication is that we set the depth of the species tree to be equal to that of the gene tree. The gene trees are likely to coalesce deeper than the speciation events and therefore the depth of the species tree is likely to be somewhat overestimated (Edwards and Beerli, 2000), and this may have an effect on the test. As with the ancestral population sizes, we are not able to correct for this source of error. The depth of the species tree was fixed following the molecular dating study of Buckley and Simon (unpublished data). However, we note that a depth of 12 Myr is rather deep and, furthermore, the branch leading to the outgroups is one of the longest in all gene trees (Fig. 3). In fact 100% of the simulated gene trees coalesce such that Maoricicada is monophyletic, an observation consistent with the empirical data (Fig. 3). By compressing the depth of the species tree, we would increase the frequency that gene trees incongruent to the species tree are generated. This is not expected to affect the power of the test because changing the depth of the species tree is not likely to increase or decrease the frequency of non-random patterns, only of patterns where the species tree and gene tree are incongruent. The effect of compressing the species tree with respect to time is expected to have a similar effect to increasing the population size.
We have also conditioned the test on the single species tree with the highest posterior probability. However, because our test statistic evaluates the presence of M. iolanthe grouping in any two phylogenetic positions at four loci, the test should not be sensitive to the shape of the assumed species tree. A further possible extension of this test would be to simulate coalescent trees, simulate sequences on these trees, and then perform phylogenetic reconstruction before proceeding to the calculation of the test statistic. This added layer to the simulations would incorporate uncertainty generated by errors in phylogenetic reconstruction. We would not expect this addition to the test to alter the power of rejecting the null hypothesis because, as we have argued above, we do not believe that phylogenetic error is responsible for the nonrandom phylogenetic placement of M. iolanthe. To conclude, although the test is conditioned on a number of simplistic assumptions, these assumptions are unlikely to render this particular application of the test misleading.
The second line of evidence that points to introgression as the cause of the incongruent phylogenetic position of M. iolanthe is song structure. Species with similar songs are presumably more likely to interbreed with each other than species with highly differentiated songs (e.g., Wells and Henry 1998). The observed phylogenetic position of M. iolanthe is with M. campbelli or M. lindsayi and M. myersi, the three species within Maoricicada that have the most similar song types (Fig. 5; Table 7). The species-specific courtship songs of Maoricicada demonstrate a high degree of structural variability among species, with some making extensive use of frequency modulation (Dugdale and Fleming, 1978). Despite this, all Maoricicada songs conform to a general "urr-chip" structural model consisting of an introductory section followed by a cueing section containing one cueing pulse or, in some species, multiple repeated pulses (Figs. 2 and 5). A large number of Maoricicada species share a particular version of this ground plan characterized by a simple pulsed introductory section; this song type, which we refer to as the campbelli type, includes M. campbelli, M. iolanthe, M. lindsayi, M. myersi, M. tenuis, M. otagoensis, and M. mangu.
|
Among species with the campbelli type song, M. campbelli, M. iolanthe, M. myersi, and M. lindsayi are all highly similar to one another (Table 7; Fig. 5). The two species M. campbelli and M. iolanthe are the most similar of the four, except that M. iolanthe tends to be higher in dominant pitch (up to 44 kHz) than M. campbelli (36 kHz; Fleming, 1971) and appears to produce preceding pulses to the introductory section somewhat more loudly and with greater regularity. The species M. lindsayi is also similar to M. campbelli but with slower pulses in the introductory section. On the other hand, M. myersi has a pulse rate in the introductory section that is in between that of M. campbelli and M. lindsayi, but also produces multiple cueing pulses. The species M. lindsayi appears to be the most divergent of the three because it has a slower pulse rate and longer pulses. The similarity in song structure between M. iolanthe, M. campbelli, M. myersi, and M. lindsayi is consistent with an introgression event involving these four species. For these reasons, it seems introgression rather than lineage sorting is more likely to explain the pattern of incongruence with respect to the phylogenetic position of M. iolanthe. However, lineage sorting is undoubtedly responsible for at least some of the incongruence in other regions of the phylogeny due to the rapid nature of the alpine radiation (Buckley and Simon, unpublished data).
The hypothesis of introgression involving M. iolanthe appears inconsistent with two observations that indicate hybridization is an uncommon event in Maoricicada. First, F1 hybrid individuals can be relatively easy to detect in cicadas because they may display songs with some combination of the characteristics of both parental species (Lane, 1995; Marshall et al., unpublished data b). Nevertheless, not one Maoricicada individual with a hybrid-like song has ever been observed, in contrast to the closely related genus Kikihia, which has undergone a similar species radiation in song divergence, number of species, and timing of speciation events (Arensburger et al., 2004b; Marshall et al., unpublished data a). In Kikihia, up to 12 hybrid zones have been located, most of which were identified from such anomalous songs; however, few of these Kikihia hybrid zones are between pairs of species that are routinely sympatric (Lane, 1995; Marshall et al., unpublished data b). Second, we observed no alleles shared between pairs of sympatric Maoricicada species. Many of the localities at which we sampled had more than one species of Maoricicada present. For example, at Rainbow Ski Field, M. m. celer, M. tenuis, M. alticola, and M. oromelaena are all sympatric, yet we did not observe any hybrid songs or any shared alleles at any of the sampled loci, albeit with limited intraspecific sampling. In summary, it appears alleles from some of the loci we sampled have introgressed from M. campbelli or the common ancestor of M. myersi and M. lindsayi into M. iolanthe, or vice versa. A similar pattern of introgression has been hypothesized to explain the lack of concordance between mtDNA gene trees and morphologically defined species boundaries in New Zealand alpine cockroaches of the genus Celatoblatta (Chinn and Gemmell, 2004).
Our conclusion that introgression is responsible for the uncertain phylogenetic position of M. iolanthe raises the question of whether its correct position is as sibling species to M. campbelli, or to the M. myersi and M. lindsayi clade. One possible solution to this problem is to select the phylogenetic position supported by the greatest number of loci—in essence, a majority-rule consensus approach. However, this approach can be unreliable with only a small number of loci (Wu, 1991). For example, the species M. mangu is highly distinctive in both morphology and song structure relative to other Maoricicada species (Dugdale and Fleming, 1978; Fig. 5, Table 7), suggesting that it is indeed monophyletic. Despite this, the monophyly of this species is supported solely by analyses of the EF1- locus, so a consensus tree approach to estimating the species tree would lead us to believe that M. mangu is polyphyletic.
Estimating the Pattern of Diversification of Recent Species Radiations
With studies of species radiations and intraspecific variation increasingly focusing on multiple loci (e.g., Hare, 2001; Machado and Hey, 2003; Zhang and Hewitt, 2003), the development of methods for inferring the species trees from discordant gene trees is becoming more critical. Concatenating such data sets into a single alignment is clearly inadequate because in searching for the species, tree loci are effectively weighted in proportion to the number of informative sites sampled (Doyle, 1992, but see Seo et al., 2005). Consensus methods are very conservative and ignore much information in the data, such as signal for marginally suboptimal nodes (Barrett et al., 1991; Gatesy et al., 1999). Gene tree parsimony is an attractive approach; however, there are only ad hoc methods for selecting the optimal weights for deep coalescence events relative to horizontal transfer (Arvestad et al., 2003). Ultimately, a coalescent-based approach that integrates over uncertainty in individual gene trees, effective population sizes, migration rates and other pertinent parameters will become feasible for large numbers of species and loci (e.g., Wu, 1991; Maddison, 1997; Nielsen and Wakeley, 2001; Holder et al., 2001; Rosenberg, 2002; Felsenstein, 2004; Degnan and Salter, 2005). Such a method would facilitate the quantification of crucial evolutionary events such as introgression and lineage sorting that occur during the speciation process.
|
|
Acknowledgements |
|---|
This work was supported by funding from the New Zealand Foundation for Research, Science and Technology (LCRX0004), and the National Science Foundation (DEB-0089946 and DEB-0422386). Collecting permits were provided by the Department of Conservation. Dianne Gleeson provided much support at all stages of this project and Robyn Howitt assisted with data collection. Karen Slon provided unpublished sequences and primers. Michael Hickerson, Kathy Hill, Richard Leschen, and Neil Gemmell provided comments on the manuscript and Dan Vanderpool and John Cooley helped with fieldwork. Advice on methods of analysis and access to computational resources and unpublished software were provided by David Swofford, Jim Wilgenbusch, and Mark Holder. Rod Page, Tim Collins, and an anonymous reviewer gave many helpful comments on the manuscript.
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
-
Akaike H. Information theory as an extension of the maximum likelihood principle. In: Second International Symposium on Information Theory—Petrov B. N., Csaki F., eds. (1973) Budapest: Akademaia Kiado. Pages 267–281.
Arensburger P., Buckley T. R., Simon C., Moulds M., Holsinger K. E. Biogeography and phylogeny of the New Zealand cicada genera (Hemiptera: Cicadidae) based on nuclear and mitochondrial DNA data. J. Biogeog. (2004a) 31:557–569.
Arensburger P. C., Simon, Holsinger K. Evolution and phylogeny of the New Zealand cicada genus Kikihia (Homoptera: Auchenorrhyncha: Cicadidae) with special reference to the origin of the Kermadec and Norfolk Islands' species. J. Biogeog. (2004b) 31:1769–1783.[CrossRef]
Aris-Brosou S. How Bayes tests of molecular phylogenies compare with frequentist approaches. Bioinformatics (2003) 19:618–624.
Arvestad L., Berglund A.-C., Lagergren J., Sennblad B. Bayesian gene/species tree reconciliation and orthology analysis using MCMC. Bioinformatics (2003) 19:7–15.[CrossRef]
Ballard J. W. O., Chernoff B., James A. C. Divergence of mitochondrial DNA is not corroborated by nuclear DNA, morphology, or behavior in Drosophila simulans. Evol. (2002) 56:527–545.
Ballard J. W. O., Whitlock M. The incomplete natural history of mitochondria. Mol. Ecol. (2004) 13:729–744.[CrossRef][Medline]
Barrett M., Donoghue M., Sober E. Against consensus. Syst. Zool. (1991) 40:486–493.
Beamer R. H. Studies on the biology of Kansas Cicadidae. Univ. Kansas Sci. Bull. (1920) 18:155–263.
Betrán M., Jiggins C. D., Bull V., Linares M., Mallet J., McMillan W. O., Bermingham E. Phylogenetic discordance at the species boundary: Comparative gene genealogies among rapidly radiating Heliconius butterflies. Mol. Biol. Evol. (2002) 19:2176–2190.
Besansky N. J., Krzywinski J., Lehmann T., Simard F., Kern M., Mukabayire O., Fontenille D., Touré Y., Sagnon N. F. Semipermeable species boundaries between Anopheles gambiae and Anopheles arabiensis: Evidence from multilocus DNA sequence variation. Proc. Nat. Acad. Sci. USA (2003) 100:10818–10823.
Brandley M. C., Schmitz A., Reeder T. W. Partitioned Bayesian analyses, partition choice, and the phylogenetic relationships of scincid lizards. Syst. Biol. (2005) 54:373–390.
Broughton R. E., Harrison R. G. Nuclear gene genealogies reveal historical, demographic and selective factors associated with speciation in field crickets. Genetics (2003) 163:1389–1401.
Buckley T. R. Model misspecification and probabilistic tests of topology: Evidence from empirical data sets. Syst. Biol. (2002) 51:492–508.
Buckley T. R., Arensburger P., Simon C., Chambers G. K. Combined data, Bayesian phylogenetics, and the origin of the New Zealand cicada genera. Syst. Biol. (2002) 51:4–18.
Buckley T. R., Simon C. Evolutionary radiation of the cicada genus Maoricicada and the age of the New Zealand alpine biota. Unpublished data.
Buckley T. R., Simon C., Chambers G. K. Exploring among-site rate variation models in a maximum likelihood framework using empirical data: Effects of model assumptions on estimates of topology, branch lengths and bootstrap support. Syst. Biol. (2001a) 50:67–86.
Buckley T. R., Simon C., Chambers G. K. Phylogeography of the New Zealand cicada Maoricicada campbelli based on mitochondrial DNA sequences: Ancient clades associated with Cenozoic environmental change. Evol. (2001b) 55:1395–1407.
Buckley T. R., Simon C., Shimodaira H., Chambers G. K. Evaluating hypotheses on the origin and evolution of the New Zealand alpine cicadas (Maoricicada) using multiple comparison tests of tree topology. Mol. Biol. Evol. (2001c) 18:223–234.
Carstens B. C., Degenhardt J. D., Stevenson A. S., Sullivan J. Accounting for coalescent stochasticity in testing phylogeographic hypotheses: Testing models of Pleistocene population structure in the Idaho giant salamander Dicamptodon aterrimus. Mol. Ecol. (2005) 14:255–265.[Medline]
Carstens B. C., Sullivan J., Davalos L. M., Larsen P. A., Pedersen S. C. Exploring population genetic structure in three species of Lesser Antillean bats. Mol. Ecol. (2004) 13:2557–2566.[Medline]
Chinn W. G., Gemmell N. J. Adaptive radiation within New Zealand endemic species of the cockroach genus Celatoblatta Johns (Blattidae): A response to Plio-Pleistocene mountain building, and climate change. Mol. Ecol. (2004) 13:1507–1518.[Medline]
Claridge M. F. Acoustic signals in the Homoptera: Behavior, taxonomy, and evolution. Ann. Rev. Entomol. (1985) 30:297–317.[CrossRef][Web of Science]
Cotton J. A., Page R. D. M. Going nuclear: Gene family evolution and vertebrate phylogeny reconciled. Proc. R. Soc. Lond. B (2002) 269:1555–1561.[Medline]
Crosby T. K., Dugdale J. S., Watt J. C. Area codes for recording specimen localities in the New Zealand subregion. N. Z. J. Zool. (1976) 25:175–183.
Cumber R. A. Notes on the biology of Melampsalta cruentata Fabricius (Hemiptera-Homoptera: Cicadidae), with special references to the nymphal stages. Trans. R. Entomol. Soc. Lond. (1952) 103:219–237.
Davies N., Bermingham E. The historical biogeography of two Caribbean butterflies (Lepidoptera: Heliconidae) as inferred from genetic variation at multiple loci. Evol. (2002) 56:573–589.
Degnan J. H., Salter L. A. Gene tree distributions under the coalescent process. Evolution. (2005) 59:24–37.[Web of Science][Medline]
Doyle J. J. Gene trees and species trees: Molecular systematics as one-character taxonomy. Syst. Biol. (1992) 17:144–163.
Dugdale J. S. Genera of New Zealand Cicadidae (Homoptera). N. Z. J. Sci. (1972) 14:856–882.
Dugdale J. S., Fleming C. A. New Zealand cicadas of the genus Maoricicada (Homoptera: Tibicinidae). N. Z. J. Zool. (1978) 5:295–340.
Edwards S. V., Beerli P. Perspective: Gene divergence, population divergence, and the variation in coalescence time in phylogeographic studies. Evolution. (2000) 54:1839–1854.[CrossRef][Web of Science][Medline]
Felsenstein J. Confidence limits on phylogenies: An approach using the bootstrap. Evol. (1985) 39:783–791.[CrossRef]
Felsenstein J. Inferring phylogenies (2004) Sunderland, Massachusetts: Sinauer Associates.
Fleming C. A. A new species of cicada from rock fans in southern Wellington, with a review of three species with similar songs and habitat. N. Z. J. Sci. (1971) 14:443–479.
Fleming C. A. Acoustic behaviour as a generic character in New Zealand cicadas (Hemiptera: Homoptera). J. R. Soc. N. Z. (1975) 5:47–64.
Funk D. J., Omland K. E. Species-level paraphyly and polyphyly: Frequency, causes, and consequences, with insights from animal mitochondrial DNA. Ann. Rev. Ecol. Syst. (2003) 34:397–423.[CrossRef][Web of Science]
Funk V. A. Phylogenetic patterns and hybridization. Ann. Missouri Bot. Gard. (1985) 72:681–715.[CrossRef][Web of Science]
Gatesy J., O'Grady P., Baker R. H. Corroboration among data sets in simultaneous analysis: Hidden support for phylogenetic relationships among higher level Artiodactyl taxa. Cladistics (1999) 15:271–313.[CrossRef][Web of Science]
Gómez-Zurita J., Vogler A. P. Incongruent nuclear and mitochondrial phylogeographic patterns in the Timarcha goettingensis species complex (Coleoptera, Chrysomelidae). J. Evol. Biol. (2003) 16:833–843.[CrossRef][Web of Science][Medline]
Hare M. P. Prospects for nuclear gene phylogeography. Trends Ecol. Evol. (2001) 16:700–706.[CrossRef]
Holder M. T., Anderson J. A., Holloway A. K. Difficulties in detecting hybridization. Syst. Biol. (2001) 50:978–982.
Holland B. R., Penny D., Hendy M. D. Outgroup misplacement and phylogenetic inaccuracy under a molecular clock—A simulation study. Syst. Biol. (2003) 52:229–238.
Huelsenbeck J. P., Rannala B. Frequentist properties of Bayesian posterior probabilities of phylogenetic trees under simple and complex substitution models. Syst. Biol. (2004) 53:904–913.
Huelsenbeck J. P., Ronquist F., Nielsen R., Bollback J. P. Bayesian inference of phylogeny and its impact on evolutionary biology. Science (2001) 294:2310–2314.
Hughes J., Vogler A. P. The phylogeny of acorn weevils (genus Curculio) from mitochondrial and nuclear DNA sequences: The problem of incomplete data. Mol. Phylogenet. Evol. (2004) 32:601–615.[CrossRef][Web of Science][Medline]
Jordan S., Simon C., Polhemus D. Molecular systematics and adaptive radiation of Hawaii's endemic damselfly genus Megalagrion. Syst. Biol. (2003) 52:89–109.
Jukes T. H., Cantor C. R. Evolution of protein molecules. In: Mammalian protein metabolism—Munro N. H., ed. (1969) New York: Academic Press. Pages 21–123.
Kimura M. Estimation of evolutionary distances between homologous nucleotide sequences. Proc. Natl. Acad. Sci. USA (1981) 78:454–458.
Kliman R. M., Andolfatto P., Coyne J. A., Depaulis F., Kreitman M., Berry A. J., McCarter J., Wakeley J., Hey J. The population genetics of the origin and divergence of the Drosophila simulans complex species. Genetics (2000) 156:1913–1931.
Knowles L. L. The burgeoning field of statistical phylogeography. J. Evol. Biol. (2004) 17:1–10.[CrossRef][Web of Science][Medline]
Knowles L. L., Maddison W. P. Statistical phylogeography. Mol. Ecol. (2002) 11:2623–2635.[CrossRef][Medline]
Lane D. H. The recognition concept of species applied in an analysis of putative hybridization in New Zealand cicadas of the genus Kikihia (Insecta: Hemiptera: Tibicinidae). In: Speciation and the recognition concept: Theory and applications—Lambert D. M., Spencer H. G., eds. (1995) Baltimore: The John Hopkins University Press. Pages 367–421.
Lewis P. O., Holder M. T., Holsinger K. E. Polytomies and Bayesian phylogenetic inference. Syst. Biol. (2005) 54:241–253.
Lin C.-P., Danforth B. N. How do insect nuclear and mitochondrial gene substitution patterns differ? Insights from Bayesian analyses of combined datasets. Mol. Phylogenet. Evol. (2004) 30:686–702.[CrossRef][Web of Science][Medline]
Machado C. A., Hey J. The causes of phylogenetic conflict in a classic Drosophila species group. Proc. R. Soc. Lond. B (2003) 270:1193–1202.[Medline]
Maddison W. P. Gene trees in species trees. Syst. Biol. (1997) 46:523–536.
Maddison W. P., Maddison D. R. Mesquite: A modular system for evolutionary analysis (2004) Version 1.04. http://mesquiteproject.org.
Marshall D. C., Cooley J. R., Hill K., Simon C. Phylogenetic utility of acoustic sexual signals in New Zealand cicadas. Unpublished data a.
Marshall D. C., Cooley J. R., Hill K., Simon C. Phylogeography of the New Zealand grass cicadas of the Kikihia muta complex: Song and mtDNA concordance in a parapatric radiation. Unpublished data b.
Miller R. E., Buckley T. R., Manos P. S. An examination of the monophyly of morning glory taxa using Bayesian phylogenetic inference. Syst. Biol. (2002) 51:740–753.
Neilsen R., Wakeley J. Distinguishing migration from isolation: A Markov chain Monte Carlo approach. Genetics (2001) 158:885–896.
Nichols R. Gene trees and species trees are not the same. Trends Ecol. Evol. (2001) 16:358–364.[CrossRef][Medline]
Nylander J. A. A., Ronquist F., Huelsenbeck J. P., Nievea-Aldrey J. L. Bayesian phylogenetic analysis of combined data. Syst. Biol. (2004) 53:47–67.
Page R. D. M. Maps between trees and cladistic analysis of historical associations among genes, organisms, and areas. Syst. Biol. (1994) 43:58–77.
Page R. D. M. GeneTree: Comparing gene and species trees using reconciled trees. Bioinformatics (1998) 14:819–820.
Page R. D. M., Charleston M. A. From gene to organismal phylogeny: Reconciled trees and the gene tree/species tree problem. Mol. Phylogenet. Evol. (1997) 7:231–240.[CrossRef][Web of Science][Medline]
Pamilo P., Nei M. Relationships between gene trees and species trees. Mol. Biol. Evol. (1988) 5:568–583.[Abstract]
Posada D., Buckley T. R. Model selection and model averaging in phylogenetics: Advantages of Akaike information criterion and Bayesian approaches over likelihood ratio tests. Syst. Biol. (2004) 53:793–808.
Posada D., Crandall K. A. ModelTest: Testing the model of DNA substitution. Bioinformatics (1998) 14:817–818.
Ragge D. R., Reynolds W. J. The songs of the grasshoppers and crickets of Western Europe (1998) Colchester, Essex: Harley Books.
Ronquist F. Reconstructing the history of host-parasite associations using generalised parsimony. Cladistics (1995) 11:73–89.[CrossRef][Web of Science]
Ronquist F., Huelsenbeck J. P. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinfomatics (2003) 19:1572–1574.[CrossRef]
Rosenberg N. A. The probability of topological concordance of gene trees and species trees. Theor. Pop. Biol. (2002) 61:225–247.[CrossRef][Web of Science][Medline]
Sang T., Zhong Y. Testing hybridization hypotheses based on incongruent gene trees. Syst. Biol. (2000) 49:422–434.
Seehausen O. Hybridization and adaptive radiation. Trends Ecol. Evol. (2004) 19:198–207.[CrossRef][Medline]
Seo T.-K., Kishino H., Thorne J. L. Incorporating gene-specific variation when inferring and evaluating optimal evolutionary tree topologies from multilocus sequence data. Proc. Nat. Acad. Sci. USA (2005) 102:4436–4441.
Shaw K. L. Conflict between nuclear and mitochondrial DNA phylogenies of a recent species radiation: What mtDNA reveals and conceals about modes of speciation in Hawaiian crickets. Proc. Nat. Acad. Sci. USA (2002) 99:16122–16127.
Shimodaira H. Multiple comparisons of log-likelihoods and combining nonnested models with applications to phylogenetic tree selection. Comm. Stat. Theory Meth. (2001) 30:1751–1772.[CrossRef]
Shimodaira H., Hasegawa M. Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Mol. Biol. Evol. (1999) 16:1114–1116.[Web of Science]
Simon C., Frati F., Crespi B., Liu H., Flook P. K. Evolution, weighting, and phylogenetic utility of mitochondrial gene sequences and a compilation of conserved PCR primers. Ann. Entomol. Soc. Am. (1994) 87:651–701.[Web of Science]
Slowinski J. B., Page R. D. M. How should species phylogenies be inferred from sequence data? Syst. Biol. (1999) 48:814–825.
Sota T., Vogler A. P. Incongruence of mitochondrial and nuclear gene trees in the carabid beetles Ohomopterus. Syst. Biol (2001) 50:39–59.
Swofford D. L. PAUP*. Phylogenetic analysis using parsimony (*and other methods). Version 4 (1998) Sunderland, Massachusetts: Sinauer Associates.
Tateno T., Nei M., Tajima F. Accuracy of estimated phylogenetic trees from molecular data. J. Mol. Evol. (1982) 18:387–404.[CrossRef][Web of Science][Medline]
Thompson J. D., Higgins D. G., Gibson T. J. CLUSTAL W: Improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position specific gap penalties and weight matrix choice. Nucleic Acids Res. (1994) 22:4673–4680.
Waddell P. J., Kishino H., Ota R. A phylogenetic foundation for comparative mammalian genomics. Genome Inform. Series (2001) 12:141–154.
Walker T. J. Cryptic species among sound-producing Ensiferan Orthoptera (Gryllidae and Tettigoniidae). Quart. Rev. Biol. (1964) 39:345–355.[CrossRef]
Warren D. L., Wilgenbusch J., Swofford D. L. Converge: A program for implementing MCMC convergence diagnostics. (2003) Available from the authors at danwarren{at}ucdavis.edu.
Wells M. M., Henry C. S. Songs, reproductive isolation and speciation in cryptic species of insects: A case study using green lacewings. In: Endless forms: Species and speciation—Howard D., Berlocher S., eds. (1998) New York: Oxford University Press. Pages 217–233.
Wilder J. A., Hollocher H. Recent radiation of endemic Caribbean Drosophila of the dunni subgroup inferred from multilocus DNA sequence variation. Evolution (2003) 57:2566–2579.[Web of Science][Medline]
Wu C.-I. Inference of species phylogeny in relation to segregation of ancient polymorphisms. Genetics (1991) 127:429–435.[Abstract]
Yang Z. Maximum likelihood models for combined analyses of multiple sequence data. J. Mol. Evol. (1996) 42:587–596.[CrossRef][Web of Science][Medline]
Yang Z. PAML: A program package for phylogenetic analysis by maximum likelihood. Comp. Appl. Biosci. (1997) 13:555–556.[Medline]
Yang Z. Likelihood ratio tests for detecting positive selection and application to primate lyzosyme evolution. Mol. Biol. Evol. (1998) 15:568–573.[Abstract]
Yang Z., Nielsen R., Goldman N., Pederson A.-M. Krabbe. Codon-substitution models for heterogeneous selection pressure at amino acid sites. Genetics (2000) 155:431–449.
Zhang D.-X., Hewitt G. M. Nuclear DNA analyses in genetic studies of populations: Practice, problems and prospects. Mol. Ecol. (2003) 12:563–584.[CrossRef][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  E. W. Bloomquist and M. A. Suchard Unifying Vertical and Nonvertical Evolution: A Stochastic ARG-based Framework Syst Biol, November 9, 2009; (2009) syp076v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. McCormack, H. Huang, and L. L. Knowles Maximum Likelihood Estimates of Species Trees: How Accuracy of Phylogenetic Inference Depends upon the Divergence History and Sampling Design Syst Biol, October 1, 2009; 58(5): 501 - 508. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. S. Kubatko Identifying Hybridization Events in the Presence of Coalescence via Model Selection Syst Biol, October 1, 2009; 58(5): 478 - 488. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Q. Spinks and H. B. Shaffer Conflicting Mitochondrial and Nuclear Phylogenies for the Widely Disjunct Emys (Testudines: Emydidae) Species Complex, and What They Tell Us about Biogeography and Hybridization Syst Biol, May 28, 2009; (2009) syp005v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Joly and A. Bruneau Measuring Branch Support in Species Trees Obtained by Gene Tree Parsimony Syst Biol, May 25, 2009; (2009) syp013v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. M. Bossu and T. J. Near Gene Trees Reveal Repeated Instances of Mitochondrial DNA Introgression in Orangethroat Darters (Percidae: Etheostoma) Syst Biol, May 22, 2009; (2009) syp014v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. T. Brumfield, L. Liu, D. E. Lum, and S. V. Edwards Comparison of Species Tree Methods for Reconstructing the Phylogeny of Bearded Manakins (Aves: Pipridae, Manacus) from Multilocus Sequence Data Syst Biol, October 1, 2008; 57(5): 719 - 731. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. J. Maureira-Butler, B. E. Pfeil, A. Muangprom, T. C. Osborn, and J. J. Doyle The Reticulate History of Medicago (Fabaceae) Syst Biol, June 1, 2008; 57(3): 466 - 482. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. C. Carstens and L. L. Knowles Estimating Species Phylogeny from Gene-Tree Probabilities Despite Incomplete Lineage Sorting: An Example from Melanoplus Grasshoppers Syst Biol, June 1, 2007; 56(3): 400 - 411. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Liu and D. K. Pearl Species Trees from Gene Trees: Reconstructing Bayesian Posterior Distributions of a Species Phylogeny Using Estimated Gene Tree Distributions Syst Biol, June 1, 2007; 56(3): 504 - 514. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||