© 2004 Society of Systematic Biologists
Using Exon and Intron Sequences of the Gene Mp20 to Resolve Basal Relationships in Cicindela (Coleoptera:Cicindelidae)
Edited by Karl Kjer: Associate Editor
1 Department of Entomology, The Natural History Museum London SW7 5BD United Kingdom E-mail: joap{at}nhm.ac.uk(J.P.)
2 Department of Biological Sciences, Imperial College London Silwood Park Campus Ascot Berkshire SL5 7PY United Kingdom
3 Faculdade de Ciências da Universidade de Lisboa, Departamento de Zoologia e Antropologia e Centro de Biologia Ambiental Rua Ernesto Vasconcelos 1746–016 Campo Grande Lisboa Portugal
 |
Abstract |
|---|
 Top Abstract Material and MethodsResultsDiscussionAppendixAcknowledgmentsReferences |
|---|
The genus Cicindela (Coleoptera: Cicindelidae) is a species-rich cosmopolitan group of tiger beetles useful for comparing clade diversification worldwide. Knowledge about relationships of major groups is important for this analysis but basal nodes in Cicindela have been difficult to resolve with standard mtDNA markers. Here we developed the Mp20 gene, a single-copy nuclear marker coding for a muscle-associated protein in insects, for phylogenetic analysis of basal groups of Cicindela. Nearly full-length sequences were obtained for 51 cicindelids, including major taxonomic groups from all continents. Sequences of Mp20 were between 1.2 and 1.7 kb and spanning three introns. Phylogenetic signal of exon and intron sequences was compared with that from four gene regions of mtDNA (COI, COIII, Cytb, 16S rRNA; 2.4 kb total). Because introns differed in length, sequence alignment was conducted using various procedures of phenetic and parsimony-based character coding of indels to assess their phylogenetic information content, but major nodes were recovered consistently. Mp20 sequences contributed two thirds of the total support of the combined analysis, with most signal from the introns. We found major clades of Cicindela to be geographically largely coincident with continental regions, confined to Australasia, the Holarctic, the Indian subcontinent, Africa, and South and Central America. Clock estimates using various maximum-likelihood (ML) branch length calculations resulted in roughly similar divergence times whether Mp20 exon, introns, or mtDNA were used, and they were not greatly affected by different procedures for coding and optimizing indel characters. Based on existing clock calibrations in Cicindela, basal splits of continental lineages occurred in the mid-Miocene, placing the radiation of basal groups of Cicindela to a period when their open-vegetation habitats expanded globally.
Keywords: Congruence; DNA sequence alignment; ESTs; miocene; molecular clock; single copy genes
Received May 19, 2003; Revised September 19, 2003; Accepted February 24, 2004
Several recent studies have demonstrated the power of single-copy nuclear markers in molecular systematics, but for many groups the choice of such markers remains limited. In the Coleoptera (beetles), only a handful of nuclear markers have been used, including elongation factor-1alpha (Cognato and Vogler, 2001; Jordal, 2002), wingless (Ober, 2002), enolase (Farrell et al., 2001), and phosphoenolpyruvate carboxykinase (Sota and Vogler, 2003). As new gene sequences become available for polymerase chain reaction (PCR) primer design, additional markers can be tested for amplification and phylogenetic information content. Most nuclear markers contain intron sequences with a typically higher rate of sequence evolution and greater length variation, and these might provide different kinds of character variation and extend the range of hierarchical levels where a marker can potentially be useful (e.g., Hillis and Dixon, 1991).
Here we employed the Mp20 locus, encoding a muscle associated protein of about 20 kD (Ayme-Southgate et al., 1989). The Mp20 gene to date has only been described in D. melanogaster where it is located in a single position at polytene region 49F 9–13 (Ayme-Southgate et al., 1989). A close match of the Mp20 coding sequence was obtained repeatedly in cDNA libraries from taxonomically divergent species of Coleoptera (Theodorides et al., 2002) and hence sequence information was available for primer design. We used these primers to investigate relationships in tiger beetles (Cicindelidae, Coleoptera) of the genus Cicindela. This genus represents a spectacular worldwide radiation of nearly 1000 species. The beetles are very effective predators of small arthropods and have diversified in open habitats such as grasslands, salt flats, sand dunes, and river edges (Pearson and Vogler, 2001). We are specifically interested in questions about the factors promoting species diversification, in comparisons of subclades that are geographically confined to different continents (Barraclough and Vogler, 2002).
The genus Cicindela s.l. (subtribe Cicindelina of some authors) has been subdivided in some 55 subgenera by Rivalier (1950–1963) mainly based on male genitalic characters. However, Rivalier's work did not primarily attempt to establish relationships between these groups, and because he treated the major continental regions separately, the basal relationships in Cicindela were not addressed in much detail. Similarly, conventional mtDNA markers failed to resolve deeper relationships between subgenera, although they have been successful in resolving relationships within some of the North American species groups (Vogler and Kelley, 1998; Vogler et al., 2004). The poor understanding of basal relationships in Cicindela leaves open many questions about the early radiation and interchange between major biogeographic regions, and prohibits age estimations of deep branches based on molecular clocks.
The goal of this paper is to test the phylogenetic information content of exon and intron sequences of Mp20, in comparison to the better established mtDNA markers, for resolving relationships among the subgenera of Cicindela s.l. and estimating an evolutionary time frame for their diversification. The Mp20 gene, combined with four partial mtDNA genes provided a well-supported tree of basal lineages of Cicindela. Intron sequences in Mp20 played an important role in establishing tree topology and clock estimations.
|
Material and Methods |
|---|
|
TopAbstractMaterial and MethodsResultsDiscussionAppendixAcknowledgmentsReferences |
|---|
Taxon Sampling and DNA Procedures
This study included 47 species from 24 subgenera of Cicindela s.l. plus four outgroups (Appendix). We selected representatives of major continental groups, including the four independent North American radiations (Cicindela s. str., Cicindelidia, Ellipsoptera, Habroscelimorpha) (Barraclough and Vogler, 2002) plus several groups from South America (Brasiella, Cylindera), the Indian subcontinent (Calochroa, Jansenia, Eugrapha), Africa (Lophyra, Lophyridia), Madagascar (Hipparidium, Chaetotaxis), Southeast Asia (Monelica, Abroscelis, Callytron), Australia (Rivacindela), and New Zealand (Neocicindela). In some cases we included several closely related species from a subgenus to test the utility of Mp20 at shallow nodes. The outgroups included three genera (Odontocheila, Prothyma, Peridexia) from another subtribe (Prothymina) within the Cicindelini, and one genus from the tribe Megacephalini (Pseudoxycheila).
DNA extractions and procedures for amplification of three mitochondrial fragments (cytochrome b [cytb], cytochrome oxidase III [COIII], and 16S rRNA) were performed as described previously (Vogler and Welsh, 1997). A fragment of 771 bp of cytochrome oxidase I (COI) was amplified as described by Ribera et al. (2001). Amplification was not successful in a few taxa for part of the data set, as follows: C. pimeriana (16S rRNA), C. guerrensis, M. captotriola (COI), P. chaudori, R. salicursoria, Macfarlandia (COIII), L. chloris (Cytb), N. parryi (16S rRNA and COIII), and N. ginevi (COI and COIII).
Primers for polymerase chain reaction (PCR) amplification of Mp20 were based on expressed sequences tags (ESTs) from four species of taxonomically divergent beetles, including the adephagan Carabus granulatus (Carabidae), and the polyphagan Agriotes lineatus (Elateridae), Mycetophagus quadripustulatus (Mycetophagidae), and Curculio glandium (Curculionidae), in combination with the D. melanogaster genomic sequence. The following oligonucleotide pair was designed for amplification of Mp20: Mp20-5' (5'-ATG TCT CTK GAA CGT CAA GTC C-3') and Mp20-3' (5'-TGN CCG GCY TGK GTG GCR CCC TTG-3'). The primer binding sites correspond to sequences at the 1st and 4th exons to amplify almost the complete coding sequence except for 19 amino acids at the 3' end (Fig. 1). PCR followed a touchdown protocol: after an initial denaturation step of 94°C for 3.5 min, 8 cycles were performed at 94°C for 30 s, 61°C for 35 s (decreasing by 0.5°C every cycle), and 72°C for 2 min and 30 s. This was followed by 32 cycles at 94°C for 30 s, 57°C for 35 s, and 72°C for 2.5 min and a final extension step at 72°C for 10 min.
|
PCR amplified fragments were cloned blunt ended in pMOSBlue using a cloning kit (Amersham Pharmacia Biotech). Inserts were detected with PCR using T7 and U-19 primers and sequenced from both strands on an ABI3700 DNA Analyzer (Applied Biosystems). Specific cicindelid internal Mp20 primers in the 2nd exon of Mp20 gene were designed to obtain complete sequences of both strands: MP20-F2 (5'-GTT CTC TGC CAG GTT ATG AA-3') and MP20-R1 (5'-ACC AGG AGG GAA CTT YTT GC-3'). The extent of intron sequences in Mp20 was predicted by comparing cicindelid sequences with the EST from Carabus granulatus. Genbank accession numbers of mitochondrial and Mp20 sequences for each species are given in Appendix.
Phylogenetic Analysis
Tree alignments were conducted using POY vers. 3.0 (Gladstein and Wheeler, 2002). For all searches reported, we present the best (lowest cost) trees from 100 random addition replicates and collecting no more than 3 shortest trees in each replicate (commands: –random 100 –maxtrees 3). Aligned sequence matrices can be produced from the POY tree alignment with the –impliedalignment command. This alignment ("implied alignment") is reconstructed from the list of synapomorphies at internal nodes, which is based on the initial cladogram (Wheeler, 2003). The aligned matrix obtained from this procedure is a representation of the homologies established in the direct optimization, and in contrast to most standard alignment procedures does not represent homoplastic insertions in the same column. The resulting matrix of aligned sequences is a synapomorphy scheme that approximates the character transformation in the dynamic homology search of POY (Wheeler, 2003). The implied alignment does not represent an alignment in the traditional sense, and it should not be used as a matrix of fixed character correspondences for searches of shorter trees (although parsimony searches on the implied alignment would frequently find shorter trees than reported by POY [unpublished observation] and these could be used as a starting tree for another round of searches in POY [Wheeler, 2003]). However, the implied alignment is a sufficiently close representation of the character optimization in the initial cladogram search that it can be used as input for subsequent analysis of character variation, and we use it here for estimates of molecular rates.
Support of trees were based on an approximate method to establish Bremer Support implemented in POY, using tree-bisection-reconnection (TBR) swapping on a constraint file obtained with the program Jack2Hen available with the POY software.
Tree searches were also conducted on fixed matrices from standard multiple alignment procedures performed using the advanced ClustalW form (Higgins et al., 1996) available online at the Institute Pasteur (http://bioweb.pasteur.fr/seqanal/interfaces/clustalw.html). Parsimony searches on these matrices were carried out in PAUP version 4.0b10 (Swofford, 2002), with gaps coded as 5th character state. In addition, gaps were coded according to Simmons and Ochoterena (2000) using the GapCoder software (Young and Healy, 2003), whereby each indel of a particular length and position in the aligned matrix is coded as a separate character and the presence/absence of these indels represents binary character states in the parsimony analysis.
Bayesian phylogenetic analyses were performed using MrBayes 3.0 software (Huelsenbeck and Ronquist, 2001). Searches were performed based on 2 million generations with four Markov chain Monte Carlo (MCMC) chains starting from random trees which were sampled every 100th generations. The log-likelihood scores of sample points were plotted against generation time, and stationarity of Markov chains was assumed when the log-likelihood values reached a stable equilibrium (Huelsenbeck and Ronquist, 2001). The stationarity of the chains was confirmed by plotting the remaining log-likelihood values (sump command). All sample points prior to stationarity were discarded as burn-in values, and remaining points were used to generate a strict consensus tree, with each clade posterior probability value represented by the proportion of nodes recovered in the sample of trees. Each search was conducted three times independently, starting from random trees, to explore the tree space which retrieved identical topologies and very similar parameters and credibility values in all analyses reported here.
Bayesian analysis was implemented in several ways. First, a given ClustalW alignment was subjected to analysis using MODELTEST (Posada and Crandall, 1998), which performs a hierarchical test of likelihood fits under 56 different models of character variation. A GTR+ 
+ I model was selected in this analysis, and bayesian searches were performed under this preferred model. Further, bayesian searches were performed under a GTR model calculating site specific rates for eight partitions showing greatly different dynamics of character change in parsimony optimization: mitochondrial structural RNA, protein 1st, 2nd, and 3rd positions, Mp20 exon 1st, 2nd, and 3rd positions, and introns. These default Mr Bayes analyses treat indel characters as missing, but information from indels can be incorporated in the search as a separate set of binary characters (analogous to the treatment of morphological characters). To incorporate phylogenetic information from indels, we used the binary recoded character matrix from the Simmons and Ochoterena (2000) gap coding procedure in the bayesian analysis, using a GTR + +I model for the DNA data, and for the binary characters using a model estimating the among site rate variation according to a gamma distribution.
Estimating the Relative Ages of Nodes from Sequence Data
Branch lengths were calculated based on the tree and model parameters calculated by bayesian inference under the three treatments describe above (ClustalW single base coding, binary gap coding, and POY implied alignment). In all cases, comparisons of models constrained for a molecular clock were significantly worse when assessed with the likelihood ratio test (P < 0.001, data not shown). Therefore branch lengths were fitted to a molecular clock using Sanderson's (1997) nonparametric rate smoothing (NPRS) algorithm. This method does not assume a strict molecular clock but that neighboring branches on the tree tend to have similar rates. NPRS was applied as implemented in TreeEdit v. 1.0a9 (Rambaut and Charleston, 2002). We also estimated branch lengths from the implied aligment and tree obtained with POY, estimating maximum-likelihood (ML) branch lengths in PAUP based on a GTR+ +I model selected by MODELTEST, and correcting these branch lengths by the NPRS method. As the number of characters is finite, stochastic variation may affect branch length estimates. To take this type of error into account we applied a resampling scheme by generating 1000 bootstrap replicates of the data and calculating branch length on each of these new data sets given the original tree topology and GTR+ +I model, fitted to a clock using NPRS (Baldwin and Sanderson, 1998). Absolute node ages were calculated with reference to a mtDNA calibration from North American taxa based on the Pleistocene biogeographic divergence along the Florida Peninsula and the closure of the isthmus of Panama (Barraclough and Vogler, 2002).
|
Results |
|---|
|
TopAbstractMaterial and MethodsResultsDiscussionAppendixAcknowledgmentsReferences |
|---|
Mp20 Sequences and Gene Organization
PCR amplification of Mp20 was successful for taxa from a wide taxonomic range of cicindelids, but in most cases sequencing was only possible after cloning of PCR products. On average, five clones were sequenced per species, but for nine species only a single clone was obtained (Appendix). Multiple clones usually differed in length, probably representing allelic variation within individuals responsible for the problems with direct sequencing. Intraindividual length variation was entirely confined to the introns, and inferred indels were very short (1 to 3 bp), except for a single clone of L. catena exhibiting an insertion of 83 bp. Mp20 sequences differed between clones obtained from a single specimen by between 0% and 2%. Sequences obtained from a single individual were monophyletic and their divergences were lower than sequence divergence between species, generally by a factor of 10 or more. Intraindividual variation in some species such as C. mathani, E. marginata, and N. ginevi is likely due to the presence of two recognizable different alleles. Within these major types, minor nucleotide discrepancies in singleton clones were encountered which we attributed to PCR errors (calculated to about 1.7 discrepancies per 1000 nucleotides). We cannot exclude that the variation of clones represents more than two alleles in a few species such as L. chloris, P. fulvia, and B. hemichrysea, perhaps indicating the presence of multiple copies of the Mp20 gene (paralogs), but this is unlikely. Only a single copy of Mp20 is present in the full genome sequence of D. melanogaster, and no evidence for paralogs was obtained in the cDNA libraries from which these sequences were obtained originally. Further, the monophyly of gene copies obtained from each individual demonstrates that if gene duplication had affected the locus, it would be recent relative to the divergence of the species under investigation here. To simplify the phylogenetic analysis, species were represented by a 50% majority rule consensus sequence of the more frequently isolated allele (i.e., removing the putative PCR errors). Where the same number of clones was isolated for each allele we randomly picked one of them.
The length of Mp20 PCR products was in the range of 1.2 to 1.7 kb. Sequences from all species exhibited three introns (Fig. 1), recognizable by the presence of the canonical splicing motif ‘GT...AG’ in all cases. The third intron separated a codon, with two nucleotides located in the second and the remaining nucleotide in the third exon. Intron length varied substantially among taxa (Fig. 1) whereas all exons were of equal length. The first intron of Prothyma (798 bp in length) and the second intron of Chaetotaxis rugicollis (516 bp) were not included in the analysis and treated as missing data, because of their great divergence from all other species. No length variation was observed in mtDNA, except for single nucleotide indels in the 16S gene region of M. arachnoides, C. maroccana, I. labeoaneae, L. chloris, and P. fulvia, five single-nucleotide indels in the P. chaudoiri, and a single-plus a two-nucleotide indel in O. confusa.
Sequence Alignment and Tree Topology
Searches for tree alignments were initially conducted in POY on all intron and exon data and the four mitochondrial regions combined, and topologies were assessed for the recovery of certain nodes (Table 1). Alignment parameters were varied extensively with respect to the relative cost of indel versus nucleotide change, and the weight of the introns relative to the remaining (exons plus mtDNA) partitions. Three main clades were recovered under equal weighting. One group included the subgenera Cicindela s. str., Cicindelidia, Lophyridia, Lophyra, Calochroa, and Hipparidium (henceforth clade I). A second group was centered around the subgenus Cylindera (s.l.), including Old World and New World groups (Ellipsoptera, Brasiella, Eugrapha, and Ifasina; clade II). The third group included a number of mostly Australian and Southeast Asian taxa (Rivacindela, Neocicindela, Macfarlandia, and Abroscelis) plus in many cases Hyphaetha (clade III). In addition, two smaller clades were frequently recovered, including a group of two species of the Indian endemic radiation of Jansenia (J. chloropleura and J. rostrula) and the Malagasy Chaetotaxis rugicollis, which frequently also included Taenidia circumdata (henceforth Jansenia clade); and a group of two species of Myriochile (M. undulata and M. mastersi) and Monelica fastidiosa.
|
Increasing the gap costs relative to nucleotide changes greatly affected tree topologies. Gap cost = 3 and especially gap cost = 4 generally resulted in the loss of clades I, II, and III, the outgroup was polyphyletic, and relationships within the three clades were increasingly inconsistent with the traditional taxonomy (Table 1). When the intron regions were downweighted relative to the other partitions by a factor of between 2 and 4, this partly reconstituted the recovery of the three main clades at higher gap costs. At lower gap costs, downweighting the intron regions generally resulted in better recovery of major taxonomic groups, although their relative positions varied greatly, and one outgroup species (Prothyma) frequently appeared in a derived position as sister to clade I (Table 1). The tree in Figure 2 (equal gap cost, introns downweighted by a factor of three) best illustrates features commonly encountered in various analyses.
|
Alternative alignment strategies were conducted with a two-step protocol, clustering sequences with ClustalW followed by tree searches on prealigned data matrices. The alignment space was explored with a hierarchical design, initially testing a wide range of gap opening penalties (from 10 to 1) and extension penalties (0.05 and 0.001). We selected those which produced the lowest incongruence between intron and exon partitions (as measured by the incongruence length difference (ILD) per character change in the combined data; ILD/change ratio), followed by variation of the parameters in a narrower range and detailed analysis of the resulting trees. The set of trees retrieved from these alignments recovered clades I, II, and III, but differed mostly within clade I and with regard to the relationship of clade II and the Jansenia clade, and the presence of Prothyma in the ingroup. The common nodes shared from all trees based on these alignments were labeled with asterisks in Figure 3. The alignment based on gap opening penalty 6 and gap extension penalty 0.001, in both the pairwise and multiple alignment menu of ClustalW produced the lowest incongruence (ILD/step = 0.0056). This alignment was the only one that accurately reflected exon and intron boundaries, and in the tree derived from it none of the outgroups species were included in the ingroup. Minor manual modifications of this alignment further reduced the incongruence (0.0049), and also produced the shortest tree obtained in any of the analyses (5133 steps). This Mp20 alignment was combined with mitochondrial sequences for tree searches and resulted in a single tree of 13440 steps (Fig. 3).
|
The preferred ClustalW alignment was also used as input for bayesian tree searches using a GTR+
+I model of nucleotide substitution. The resulting tree (Fig. 4) was very similar to the preferred parsimony tree but also resolved the reciprocal monophyly of Cicindela and Cicindelidia that was otherwise only recovered in a few POY alignments. Tree searches were also performed using a site specific model, and with the GTR+ +I model and gaps coded as binary state (see Material and Methods). These analyses retrieved similar tree topologies and credibility values (Fig. 4), and those nodes not recovered in all three searches showed low credibility values in all three trees.
|
, and nodes not supported when gaps are coded as binary characters are marked by #. See Material and Methods and Phylogenetic Information Content and Dynamics of Sequence Evolution in mtDNA and Mp20
The data were separated according to functional criteria to test for the phylogenetic signal of various data partitions and potential conflict in a parsimony framework. In the ClustalW alignment obtained under the preferred parameters, the Mp20 provided a far smaller number of positions than mtDNA (1767 versus 2655; Table 2) and contributed fewer steps to the total cost of the simultaneous analysis (5133 versus 8146). However, the phylogenetic information content of Mp20 was greater due to lower internal consistency (RI = 0.66 versus 0.30) and the higher proportion of informative sites (55.2% versus 33.7%). Consequently, the Partitioned Bremer Support (PBS) attributed to Mp20 in the simultaneous analysis tree was greater than that of mtDNA (719 versus 436). Most of the phylogenetically informative variation in Mp20 was contributed by the introns, with Intron 1 providing a total PBS of 454, slightly higher than that of all mtDNA combined (Table 2). The Mp20 partition was in closer agreement with the combined data than the mtDNA; the single tree from Mp20 shared more nodes (37 of 49 total) with, and required fewer extra steps (0.85% versus 1.42% for all mtDNA) to fit the simultaneous analysis tree. Based on PBS, the Mp20 partitions (mainly exons) provided most of the support for basal relationships whereas support from mitochondrial partitions for these nodes was weak or negative (Fig. 3). Alternative coding of indels as ‘missing’ characters had only a very slight impact on the tree topology. Similarly, the gap coding procedure according to Simmons and Ochotorena (2000) also produced trees remarkably similar to those obtained with indels coded as 5th character, with 45 of 49 nodes shared (Table 2).
|
Separate analysis of introns retrieved similar topologies as in the combined analysis, resolving deep and shallow nodes. Trees based only on the intron partitions shared more nodes with the combined analysis (32 nodes) than trees from the exons (20 nodes). The signal in introns was derived both from the indel free and indel containing sites, with trees from the latter sharing a greater number of nodes with the combined analysis tree than the tree from indel free positions only (31 versus 22 nodes, Table 2), although the latter (based on 223 sites of which 124 were potentially informative) still recovered clades I, II, III and most of the relevant subclades within these groups. Although intron positions in total contributed greater signal, the exons partitions showed nearly three times higher PBS per site than intron partitions (Table 2).
Character incongruence as determined by ILD was significant for most of the partitions, but generally internal conflict was greater in mtDNA partitions than in Mp20. The ILD (corrected for the number of character changes) in mtDNA was 0.0390, compared to values of 0.0049 or 0.0169 for Mp20 when analyzed as two (exons versus introns) or four (exons/intron1/intron2/intron3) partitions, respectively. However, the level of incongruence was decreased if mtDNA plus Mp20 exons or mtDNA plus Mp20 exons and non-indel positions were analyzed in combination (ILD/change 0.0067, P < 0.96; and ILD/change = 0.0109, P < 0.66, respectively). This demonstrated that combining all data in simultaneous analysis produced a more consistent character distribution than any of the partitions separately. This was confirmed by the interesting observation that incongruence between the four mitochondrial partitions was reduced when analyzed together with the Mp20 data (ILD/change = 0.0390 versus 0.0120).
Rate Variation Within and Between Data Partitions, and Evidence for Saturation
Mp20 and mtDNA partitions differed in base composition and rates of substitution. The Mp20 introns and mtDNA were A+T rich (average 67.07% and 75.84%, respectively), whereas the exons were G+C rich (64.69% on average) based on informative characters only (Table 3). These values differed across taxa, when assessed with the heterogeneity chi square test in PAUP, but only due to the apparent bias in 3rd codon positions of the mtDNA (Table 3). This bias was due mostly to a small number of species with relatively low A+T content. These species were widely scattered throughout the tree, indicating that A+T bias is unlikely to have resulted in artificial groupings.
|
MtDNA and Mp20 partitions also differed in their distribution of character changes and site-to-site rate heterogeneity. For mtDNA, likelihood models to describe character variation produced a better fit to the data with increasing complexity of the models, and for most partitions a GTR+
+I was found to be better than simpler models. However, for the Mp20 exons single parameter or two parameter models produced a fit that was equally good than complex models, in particular if the three codon positions were considered separately (Table 3). Whereas the estimated proportion of invariant sites was high for all partitions, except the highly variable mtDNA 3rd positions and the Mp20 introns, the shape of the gamma distribution was greatly different between mtDNA and Mp20. The alpha parameter was much greater in Mp20 (Table 3), indicating a more homogeneous among-site rate variation.
We calculated rates of change per branch in various partitions for the tree based on GTR+ +I model of Figure 4. Rates were expressed as the average rate per branch in a given clade, and hence the values permit a direct comparison of likelihood rates of change between different partitions. Rates in mtDNA were on average nearly four times higher than Mp20, but with great variation of rates within each of the major partitions. For example, the average rate of change in intron positions was approximately four times higher than in exon sequences, and nearly twice as high as in Mp20 3rd codon positions. In mtDNA, rates of change were very high in 3rd codon positions, whereas rates in 1st codon positions (not shown) and structural RNA regions were at least four times lower (Table 4). Rates were also calculated separately for clades I, II, and III, the ingroup and the outgroup. In these comparisons rates were found to differ by similar ratios (Table 4), i.e., the partitions exhibited an intrinsic rate that differed in concert throughout the tree, even for the unlinked Mp20 and mtDNA genes.
|
Whereas higher rates in mtDNA were widely found within the ingroup, in the outgroup this ratio was reversed, as rates were faster in Mp20 than in mtDNA (0.144 versus 0.121, respectively; Table 4). Rate of change in mtDNA in the outgroup was very similar to the ingroup (0.121 versus 0.098), but in the outgroup Mp20 rate was about six times faster than in the ingroup (0.144 versus 0.022). Interestingly, estimated rates in Mp20 based on parsimony branch length were very similar to those estimated by ML, but in mtDNA parsimony rates are lower by a factor of 2 to 4 (not shown), suggesting a much greater level of saturation in mtDNA. Saturation in mtDNA relative to Mp20 was also evident in plots of branch length optimized under ML models of different complexity (Fig. 5). A plot of branch length calculated under the simple F81 model against a GTR+
+I model produced a largely linear correlation for Mp20. In contrast, for mtDNA this correlation is weak, as the estimates of branch length are much higher under the complex model, indicating the difficulty of reconstructing the full extent of character variation without taking into account the great rate heterogeneity in this data set. This effect was almost entirely explained by the behavior of 3rd position in this analysis (not shown).
|
Estimation of Node Ages
Branch length estimates were used for calculation of node ages. Only the Mp20 exon data were marginally consistent with a molecular clock (0.05 < P > 0.025; likelihood ratio test), whereas all other partitions and the combined data set under any of the treatments were not (P < 0.001, see Table 5). Hence for estimation of node ages we used branch lengths based on the NPRS algorithm, although results were closely correlated with clock estimates assuming rate constancy (r2 = 0.98). Ages were estimated for major nodes in the tree, with focus on sister lineages confined to different biogeographic regions and including clades I, II and III. Calibration of absolute dates were based on a estimate of 5 Mya for the split of Ellipsoptera marginata from E. sperata plus E. puritana (Barraclough and Vogler, 2002).
|
Based on these extrapolations, the Mp20 and mtDNA combined data with the GTR+
+I model indicated an age of the ingroup node (Cicindela s.l.) in the range of 12.7 to 15.9 Mya (based on bootstrap replicates), with the estimates for the ages of clades I, II and III just below this range (Table 5). Separation of several sister groups from India and Madagascar were estimated to range between 5.1 and 6.7 and 11.2 and 13.8 Mya, and the split of the Australian Rivacindela from Neocicindela of New Zealand in the range of 6.6 to 8.6 Mya. Ages were also estimated under different likelihood models and gap coding procedures: a site specific model and a model including gaps as binary characters. Moreover, the implied alignment and tree obtained with POY were also used to estimate the ages with the model selected by MODELTEST (GTR+ +I). In all cases the values obtained were within, or very slightly outside of, the ranges of bootstrapped data and analysis under the GTR+ +I model (Table 5). Ages calculated for the Mp20 and mtDNA partitions separately also closely matched the calculations from the combined analysis (Table 5). Because variation in exons was clock-like, ages from this partition were also estimated without the NPRS correction, and produced dates that were slightly younger at lower nodes, and slightly older at deep nodes (not shown). Our study included 14 nodes that had also been estimated in the diversification of the North American tiger beetles based on mtDNA only (Barraclough and Vogler, 2002) and these data showed good correlation with the previously estimated ages (r2 = 0.85, Table 5). |
Discussion |
|---|
|
TopAbstractMaterial and MethodsResultsDiscussionAppendixAcknowledgmentsReferences |
|---|
Utility of Mp20 in Molecular Systematics of Cicindela
The Mp20 locus was shown here to produce a phylogenetic signal largely congruent with mtDNA. It added in particular to knowledge about basal relationships in Cicindela that had been difficult to resolve with mtDNA alone. The total number of base pairs sequenced for Mp20 was only about half that for mtDNA, and Mp20 showed a roughly four times lower rate of nucleotide variation. Yet, in the combined analysis Mp20 contributed a larger amount of phylogenetic signal, as it showed greater overall character consistency, greater information content per base pair, and contributed a greater amount of total support. Variation in Mp20 showed several features that probably contributed to the greater utility in the phylogenetic analysis, including lower site-to-site rate heterogeneity and fewer invariable sites, a lower level of saturation of character variation, and a less pronounced AT bias than was found in mtDNA.
Major elements of the tree were supported by both the Mp20 and mtDNA, including the major clades I, II, and III, and several groups within clade I. However, topological congruence decreased towards the root, where nodal support values were generally lower and the mtDNA partitions produced very low or negative PBS values. In contrast, at the tips mtDNA generally was in agreement with Mp20, showing a much greater proportion of nodes recovered by all partitions and higher total PBS. Recent work has questioned that Bremer Support values are comparable, even within a single tree (DeBry 2001), and indeed these values should not be considered to be comparable in a statistical sense. However, BS and PBS remain well-established parameters for estimating relative degree of corroboration among competing hypotheses (Grant and Kluge, 2003), and the trends in these nodal support values (rather than any particular value) are useful for establishing weaker parts of our phylogenetic knowledge. Therefore we accept that incongruence is greater at deep nodes, presumably due to the poor performance of mtDNA and the various biases of character variation and nucleotide composition in this marker. The underlying phylogenetic signal in mtDNA, however, may not be very different from that in Mp20. This is corroborated by the fact that interactions with Mp20 in simultaneous analysis reduced the level of incongruence between the various mtDNA partitions, suggesting that the various mtDNA partitions support a common signal that was more easily apparent in the context of the Mp20 data.
Other causes typically affecting incongruence between mitochondrial and nuclear markers, such as differential introgression and incomplete lineage sorting, are unlikely to play a role here. These effects should be strongest between closely related taxa, rather than deeper in the tree. In addition, the presence of multiple genomic copies and their diversification could result in the sequencing of paralogs, and in this case it would be the Mp20 partition that contributes the false signal. However, this again is unlikely, as one of the main criteria for selecting Mp20 was that it represented only a small number of candidates from the EST libraries which apparently were lacking any close paralogs within the D. melanogaster genome or in the Coleoptera cDNA libraries (unpublished). Further, as the Mp20 tree is congruent with mtDNA at the tip level, the Mp20 sequences can be confidently assumed to be orthologs at least at this hierarchical level, and it is difficult to conceive a scenario where different paralogs would be picked up by PCR deeper in the tree only. Hence the phylogenetic history of both markers was probably largely the same, and incongruence could be attributed mostly to differences in parameters of sequence variation.
Utility of Intron Sequences and Alignment
Intron sequences add to the technical difficulties of PCR amplification from nuclear genes, and they also tend to complicate phylogenetic analysis as they generally exhibit higher rates of variation and suffer from length variation. Most authors have therefore disregarded intron sequences in phylogenetic analysis, or have amplified from exons only, by designing primers that avoid introns or by using cDNA templates. However, in our analysis the introns made an important contribution and clearly added to the strengths of Mp20, as seen in other nuclear genes in recent studies (e.g., Sanchis et al., 2001; Kawakita et al., 2003). The introns contributed about five times greater total PBS than exons, and their fairly high rate of change, exceeding the rate in exons by threefold, and the lower site-by-site rate heterogeneity, were useful to resolve relationships over a range of hierarchical levels. In addition, the fact that introns and exons differ in types and rates of sequence variation may have beneficial effects, as interaction of data of different kinds frequently leads to improved recovery of phylogenetic signal when analyzed simultaneously (e.g., Olmstead and Sweere, 1994). Finally, the length variation in the intron sequences provided a major source of phylogenetic signal, which was largely congruent with other characters, as the indel containing sites produced similar topologies as the exons and the combined analysis.
However, sequence alignment remains a difficult issue when using intron sequences. It is an inevitable step in such analysis to choose a value for the cost of indels prior to the tree search, but there are no simple criteria for this choice. The selection of parameters after empirical exploration of parameter values based on congruence analysis (Wheeler, 1995) has been criticized because the differential weighting required in these procedures ultimately reduces (not increases) total character congruence, and because of the poorly justified use of only a small portion of a potentially infinite parameter space (Grant and Kluge, 2003). Here we addressed the problem of sequence alignment in two very different procedures. First, the dynamic homology assessment of POY was used to explore a range of pertinent gap cost parameters. The selection of a tree from this (by no means exhaustive) set of tree searches, however, was arbitrary in the current case, and the tree shown in Figure 2 is merely representing an example of the trees obtained. Following Grant and Kluge (2003), the analysis probably cannot be taken further, and as such the various resulting hypotheses are presented in tabulated form (Table 1).
We found, however, that low gap costs produced trees which are generally more in agreement with existing ideas of relationships. If gap costs were set higher, it was preferable to downweight the intron (alignment variable) regions, reducing the overall costs of indels. There was also a tendency for larger gaps to be preferred (lower extension cost in the Clustal alignment). These findings indicate that indels should not be weighted with a high cost as afforded to them by the POY method (Wheeler, 1996; Giribet and Wheeler, 1999), which treats each single-nucleotide indel separately, and where the penalty for the introduction of a gap in the alignment conflates the weight of the indels in the phylogenetic analysis.
The Clustal analysis was performed with a slightly different perspective. Gap costs here represent a penalty when aligning two or more sequences to each other, and the resulting character matrices are independent of the tree search. The resulting matrices can be compared directly, without the problems from differential weighting as in the POY analysis. To the degree that the ILD represents the congruence of different partitions (see for potential problems Dolphin et al., 2000; Yoder et al., 2001; Darlu and Lecointre, 2002; Dowton and Austin, 2002), we could then show that alignment conditions could be obtained under which the phylogenetic signal in the introns was very compatible with the other data. This applies both to the topologies, which were very similar to those obtained from non-intron partitions, and also to the rates of change throughout the tree which differed in concert among intron and non-intron partitions. For example, the approximately fourfold higher rate of change in introns compared to exons was observed consistently for all subclades tested (Table 4). Because intron and exons as part of a linkage group are expected to have a similar phylogenetic history, this suggests that the optimal alignments produced here are not simply a fortuitous result of phenetic procedures but represent a good reflection of base pair homologies.
Phylogenetic and Biogeographic History of Cicindela
Molecular clock estimates in insects have relied almost exclusively on mtDNA but this marker has been problematic, in particular for inferences about deeper nodes. The Mp20 data increased the confidence in these calculations, by providing a better topology for the deep nodes and also providing data apparently less affected by saturation. Therefore, with the Mp20 data in hand it was possible for the first time to calculate ages in the deeper nodes in Cicindela, for a test of competing biogeographic scenarios explaining the cosmopolitan distribution of the group.
We found that the main groupings in Cicindela s.l. recovered in this analysis conform to major geographic regions. For example, clade I included the Holarctic subgenus Cicindela s. str. and its Nearctic sister Cicindelidia, which combined were sister to a clade of Cosmodela, Lophyridia, and the C. japonica group (Cicindela s. str. subgroup 6 of Rivalier), with distribution mostly in the eastern Palearctic. Sister to this Holarctic clade were the Indian/Magadascan/African Lophyra plus Calochroa (India) and Hipparidium (Madagascar). Clade II is a grouping of species considered by Rivalier to be part of the cosmopolitan (sub)genus Cylindera. Within this globally distributed group the major subclades were again confined to particular biogeographic regions. The two Indian subgenera Eugrapha and Ifasina (the former also including the Malagasy E. zaza) were monophyletic and sister to a New World clade including the South American Brasiella and Hemichrysea, which were sister to the Cylindera debilis group plus Ellipsoptera from North and Central America. Clade III included taxa from Australia and New Zealand plus a few subgenera with wide distribution in Southeast Asia. These associations had not been recognized before and they remain to be tested further. The analysis also established the deep separation of the Australian clade from all others, although several groups (Jansenia, Habroscelimorpha, Taenidia, and Chaetotaxis) remained difficult to place.
These relationships of taxa in major biogeographic regions now permit testing alternative scenarios of dispersal or continental break-up to explain continental distributions of Cicindela s.l. Based on the molecular clock calibration, the age of Cicindela s. l. was placed between 12.7 and 15.9 Mya in the analysis of all data, and in a very similar range in the analyses based on Mp20 and mtDNA alone (Table 5). For nodes of known geological ages, such as the separation of India and Madagascar, or Australia and New Zealand, the calculated node ages were generally 10 Mya or less, and hence about an order of magnitude lower than the time of break-up of major land masses estimated from geological evidence. This indicates that the cosmopolitan distribution of Cicindela involved occasional faunal interchange between continents. However, this exchange is comparatively rare, as radiations can be identified which are limited to a particular region.
If the molecular clock calibration is correct, the origin of the Cicindela s.l. clade is placed in the mid-Miocene, a period of expanding dry climate and spread of grasslands and savanna at the expense of forest habitat world-wide (Behrensmeyer, 1992). These changes of temperate and tropical ecosystems may have promoted the radiation of Cicindela that have conspicuously diversified in open habitats, in contrast to the mostly forest dwelling basal lineages of Cicindelidae. The group therefore might be one of several that have undergone diversification and colonization of new biogeographic areas during the early and middle Miocene with the changes in habitat availability. For instance, hypsodont mammals were able to expand their habitat and food resources in Europe (Jernvall and Fortelius, 2002), shrubland and fynbos flora radiated in South Africa (Goldblatt et al., 2002), and the diversification of hominids in savanna habitat coincided with this period (Pickford, 2002). These changes also led to the dispersal of several groups between Africa and Eurasia such as tiger frogs (Kosuch et al., 2001), rodents and lagomorph (Winkler, 2002), and hominids (Chaimanee et al., 2003), and could also have prompted the dispersal of early Cicindela between continents.
In conclusion, the use of Mp20, a molecular marker not previously used in phylogenetics, greatly advanced our knowledge of basal relationships in Cicindela. Yet, the deep nodes remain poorly supported, perhaps due to fast radiation during the early evolution of the group that may have coincided with a period where their habitat greatly expanded. Better resolution will likely come from additional nuclear gene markers, and it will be interesting to establish if intron sequences are more generally useful for establishing Miocene relationships in insects.
|
Appendix |
|---|
|
TopAbstractMaterial and MethodsResultsDiscussionAppendixAcknowledgmentsReferences |
|---|
|
) |
Acknowledgments |
|---|
|
TopAbstractMaterial and MethodsResultsDiscussionAppendixAcknowledgmentsReferences |
|---|
We thank D. Broszka, D. Pearson, D. Sumlin, and F. Cassola for specimens, and A. de Riva for contributing to the development of primers for Mp20. We are indebted to J. Gomez-Zurita and M. Arnedo for their help with phylogenetic methods and comments on the manuscript. We are also grateful for extremely constructive criticism during the review process by K. Kjer, K. Ober, C. Simon, and an anonymous referee that greatly improved this manuscript. This study was supported by NERC grant NER/A/S/2000/00489. Funding for AC was through a stipend of Fundaçõ para a Ci
cia e a Tecnologia (PRAXIS XXI/BD/18409/98), and KT was funded through BBSRC grant G15548 (to APV and P. Foster). Associate Editor: Karl Kjer
|
References |
|---|
|
TopAbstractMaterial and MethodsResultsDiscussionAppendixAcknowledgmentsReferences |
|---|
-
Ayme-Southgate A., Lasko P., French C., Pardue M. L. Characterization of the gene for Mp20—a Drosophila muscle protein that is not found in asynchronous oscillatory flight-muscle. J. Cell Biol. (1989) 108:521–531.
Baldwin B. G., Sanderson M. J. Age and rate of diversification of the Hawaiian silversword alliance (Compositae). Proc. Natl. Acad. Sci. U.S.A. (1998) 95:9402–9406.
Barraclough T. G., Vogler A. P. Recent diversification rates in North American tiger beetles (genus Cicindela). Mol. Biol. Evol. (2002) 19:1706–1716.
Behrensmeyer A. K. Terrestrial ecosystems through time (1992) Chicago, Illinois: Chicago University Press.
Chaimanee Y., Jolly D., Benammi M., Tafforeau P., Duzer D., Moussa I., Jaeger J. J. A Middle Miocene hominoid from Thailand and orangutan origins. Nature (2003) 422:61–65.[CrossRef][Web of Science]
Cognato A. I., Vogler A. P. Exploring data interaction and nucleotide alignment in a multiple gene analysis of Ips (Coleoptera: Scolytinae). Syst. Biol. (2001) 50:758–780.
Darlu P., Lecointre G. When does the incongruence length difference test fail? Mol. Biol. Evol. (2002) 19:432–437.
De Bry R. W. Improving interpretation of the decay index for DNA sequence data. Syst. Biol. (2001) 50:742–752.
Dolphin K., Belshaw R., Orme C. D. L., Quicke D. L. J. Noise and incongruence: Interpreting results of the incongruence length difference test. Mol. Phylogenet. Evol. (2000) 17:401–406.[CrossRef][Web of Science][Medline]
Dowton M., Austin A. D. Increased congruence does not necessarily indicate increased phylogenetic accuracy—the behavior of the incongruence length difference test in mixed-model analyses. Syst. Biol. (2002) 51:19–31.
Farrell B. D., Sequeira A. S., O'Meara B. C., Normark B. B., Chung J. H., Jordal B. H. The evolution of agriculture in beetles (Curculionidae : Scolytinae and Platypodinae). Evolution (2001) 55:2011–2027.[Web of Science][Medline]
Giribet G., Wheeler W. C. On gaps. Mol. Phylogenet. Evol. (1999) 13:132–143.[CrossRef][Web of Science][Medline]
Gladstein D., Wheeler W. C. POY. Program and documentation, version 3.0 (2002) New York: American Museum of Natural History.
Goldblatt P., Savolainen V., Porteous O., Sostaric I., Powell M., Reeves G., Manning J. C., Barraclough T. G., Chase M. W. Radiation in the Cape flora and the phylogeny of peacock irises Moraea (Iridaceae) based on four plastid DNA regions. Mol. Phylogenet. Evol. (2002) 25:341–360.[CrossRef][Web of Science][Medline]
Grant T., Kluge A. G. Data exploration in phylogenetic inference: Scientific, heuristic, or neither. Cladistics (2003) 19:379–418.[CrossRef][Web of Science]
Higgins D. G., Thompson J. D., Gibson T. J. Using CLUSTAL for multiple sequence alignment. Methods Enzymol. (1996) 266:383–401.[Web of Science][Medline]
Hillis D. M., Dixon M. T. Ribosomal DNA: Molecular evolution and phylogenetic inference. Quart. Rev. Biol. (1991) 66:411–453.[CrossRef][Medline]
Huelsenbeck J. P., Ronquist F. MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics (2001) 17:754–755.
Jernvall J., Fortelius M. Common mammals drive the evolutionary increase of hypsodonty in the Neogene. Nature (2002) 417:538–540.[CrossRef]
Jordal B. H. Elongation factor 1 alpha resolves the monophyly of the haplodiploid ambrosia beetles Xyleborini (Coleoptera : Curculionidae). Insect Mol. Biol. (2002) 11:453–465.[CrossRef][Web of Science][Medline]
Kawakita A., Sota T., Ascher J. S., M M. I., Tanaka H., Kato M. Evolution and phylogenetic utility of alignment gaps within intron sequences of three nuclear genes in bumble bees (Bombus). Mol. Biol. Evol. (2003) 20:87–92.
Kosuch J., Vences M., Dubois A., Ohler A., Bohme W. Out of Asia: Mitochondrial DNA evidence for an Oriental origin of tiger frogs, genus Hoplobatrachus. Mol. Phylogenet. Evol. (2001) 21:398–407.[CrossRef][Web of Science][Medline]
Ober K. A. Phylogenetic relationships of the carabid subfamily Harpalinae (Coleoptera) based on molecular sequence data. Mol. Phylogenet. Evol. (2002) 24:228–248.[CrossRef][Web of Science][Medline]
Olmstead R. G., Sweere J. A. Combining data in phylogenetic systematics: An empirical approach using three molecular data sets in the Solanaceae. Syst. Biol. (1994) 43:467–481.
Pearson D. L., Vogler A. P. Tiger beetles: The ecology and evolution of the Cicindelidae (2001) Ithaca, New York: Cornell University Press.
Pickford M. Palaeoenvironments and hominoid evolution. Z. Morphol Anthropol. (2002) 83:337–348.[Medline]
Posada D., Crandall K. A. MODELTEST: Testing the model of DNA substitution. Bioinformatics (1998) 14:817–818.
Rambaut A., Charleston M. (2002) TreeEdit. Phylogenetic Tree Editor v1.0 alpha 10. Oxford University.
Ribera I., Barraclough T. G., Vogler A. P. The effect of habitat type on speciation rates and range movements in aquatic beetles: Inferences from species-level phylogenies. Mol. Ecol. (2001) 10:737–750.[CrossRef][Medline]
Rivalier E. Démembrement du genre Cicindela Linné. Rev. Franc. Entomol. (1950) 17:217–244.
Rivalier E. Démembrement du genre Cicindela Linné. V. Faune australienne. Rev. Franc. Entomol. (1963) 30:30–48.
Sanchis A., Michelena J. M., Latorre A., Quicke D. L. J., Gardenfors U., Belshaw R. The phylogenetic analysis of variable-length sequence data: Elongation factor-1alpha introns in European populations of the parasitoid wasp genus Pauesia (Hymenoptera: Braconidae: Aphidiinae). Mol. Biol. Evol. (2001) 18:1117–1131.
Sanderson M. J. A nonparametric approach to estimating divergence times in the absence of rate constancy. Mol. Biol. Evol. (1997) 14:1218–1231.[Web of Science]
Simmons M. P., Ochoterena H. Gaps as characters in sequence-based phylogenetic analyses. Syst. Biol. (2000) 49:369–381.
Sota T., Vogler A. P. Reconstructing species phylogeny of the carabid beetles Ohomopterus using multiple nuclear DNA sequences: Heterogeneous information content and the performance of simultaneous analyses. Mol. Phylogenet. Evol. (2003) 26:139–154.[CrossRef][Web of Science][Medline]
Swofford D. L. When are phylogeny estimates from molecular and morphological data incogruent? In: Phylogenetic analysis of DNA sequences—Miyamoto M. M., Cracraft J., eds. (1991) Oxford: Oxford University Press. 295–333. Pages.
Swofford D. L. PAUP*: Phylogenetic Analysis using Parsimony. Version 4.0b. (2002) Sunderland, Massachusetts: Sinauer Associates.
Theodorides K., Riva A. d., Gómez-Zurita J., Foster P. G., Vogler A. P. Comparison of EST libraries from seven beetle species: Towards a framework for phylogenomics of the Coleoptera. Insect Mol. Biol. (2002) 11:467–475.[CrossRef][Web of Science][Medline]
Vogler A. P., Cardoso A., Barraclough T. G. Distribution of homoplasy and the effect of weighting when the tree is fully sampled: Species level phylogenetics of North American tiger beetles (genus Cicindela) Syst. Biol. submitted. (2004).
Vogler A. P., Kelley C. K. Covariation of defensive traits in Cicindela tiger beetles: A phylogenetic approach using mtDNA. Evolution (1998) 52:529–538.[CrossRef][Web of Science]
Vogler A. P., Welsh A. Phylogeny of North American Cicindela tiger beetles inferred from multiple mitochondrial DNA sequences. Mol. Phylogenet. Evol. (1997) 8:225–235.[CrossRef][Web of Science][Medline]
Wheeler W. C. Sequence alignment, parameter sensitivity, and the phylogenetic analysis of molecular data. Syst. Biol. (1995) 44:321–331.
Wheeler W. C. Optimization alignment: The end of multiple sequence alignment in phylogenetics? Cladistics (1996) 12:1–9.[CrossRef][Web of Science]
Wheeler W. C. Implied alignment: A synapomorphy-based multiple-sequence alignment method and its use in cladogram search. Cladistics (2003) 19:261–268.[CrossRef][Web of Science][Medline]
Winkler A. J. Neogene paleobiogeography and East African paleoenvironments: Contributions from the Tugen Hills rodents and lagomorphs. J. Human Evol. (2002) 42:237–256.[CrossRef][Web of Science][Medline]
Yoder A. D., Irwin J. A., Payseur B. A. Failure of the ILD to determine data combinability for slow loris phylogeny. Syst. Biol. (2001) 50:408–424.
Young N. D., Healy J. GapCoder automates the use of indel characters in phylogenetic analysis. BMC Bioinformatics (2003) 4:6.[CrossRef][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. Pons, T. G. Barraclough, J. Gomez-Zurita, A. Cardoso, D. P. Duran, S. Hazell, S. Kamoun, W. D. Sumlin, and A. P. Vogler Sequence-Based Species Delimitation for the DNA Taxonomy of Undescribed Insects Syst Biol, August 1, 2006; 55(4): 595 - 609. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Creer, C. E. Pook, A. Malhotra, and R. S. Thorpe Optimal Intron Analyses in the Trimeresurus Radiation of Asian Pitvipers Syst Biol, February 1, 2006; 55(1): 57 - 72. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Hughes, S. J. Longhorn, A. Papadopoulou, K. Theodorides, A. de Riva, M. Mejia-Chang, P. G. Foster, and A. P. Vogler Dense Taxonomic EST Sampling and Its Applications for Molecular Systematics of the Coleoptera (Beetles) Mol. Biol. Evol., February 1, 2006; 23(2): 268 - 278. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Pons and A. P. Vogler Complex Pattern of Coalescence and Fast Evolution of a Mitochondrial rRNA Pseudogene in a Recent Radiation of Tiger Beetles Mol. Biol. Evol., April 1, 2005; 22(4): 991 - 1000. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||