The Mitochondrial Genome of the House Centipede Scutigera and the Monophyly Versus Paraphyly of Myriapods
http://www.100md.com
分子生物学进展 2004年第4期
Department of Biology and Centro Ricerche Interdepartimentale Biotecnologie Innovative (CRIBI), University of Padova, Padova, Italy
E-mail alessandro.minelli@unipd.it.
Abstract
Recent advances in molecular phylogenetics are continuously changing our perception of the phylogenetic relationships among the main arthropod lineages: crustaceans, hexapods, chelicerates, and myriapods. Besides the intrinsic interest in unraveling the evolution of the largest animal phylum, these studies are basic to an understanding of one of the major transitions in animal evolution—i.e., the conquest of land with all its associated structural and functional adaptations. Myriapods have been traditionally considered the closest relatives of hexapods, thus implying only one origin of terrestriality for the tracheate lineage, but this view is now challenged by molecular evidence. Sequence data available to date for centipedes and millipedes are very limited, and the taxon sampling is strongly biased. The most critical gap was the scutigeromorph centipedes, which are the sister group to all remaining Chilopoda from which they probably diverged in the Silurian if not earlier. We obtained the first complete mitochondrial sequence for a representative of this clade, the house centipede. In our phylogenetic analyses of the protein-coding genes in this mitochondrial genome, along with 16 further ones representing the other major arthropod clades plus two outgroups, the myriapods formed a clade with the chelicerates. This implies that water-to-land transition occurred at least three times (hexapods, myriapods, arachnids) during the evolution of the Arthropoda. In addition, in contrast to all previous studies, our best supported topologies favor paraphyly of the myriapods with respect to the chelicerates. This would increase to four the main events of land colonization in arthropods (once for centipedes, once for millipedes).
Key Words: Chilopoda ? Myriapoda ? Scutigera coleoptrata ? mitochondrial genome ? arthropod phylogeny
Introduction
The phylum Arthropoda is the most diversified and species-rich group in the animal kingdom. Although there is almost unanimous consensus on the monophyly of the phylum (Giribet and Ribera 2000), there is also hot debate about the phylogenetic relationships among the four main arthropodan lineages: chelicerates, myriapods, crustaceans, and hexapods. In addition, the monophyly of some of these groups has been repeatedly questioned. In particular, the paraphyly and even polyphyly of the Myriapoda has frequently been proposed (Edgecombe and Giribet 2002; Loesel, N?ssel, and Strausfeld 2002).
Traditional classifications placed the Chelicerata as sister group of the Mandibulata (Crustacea, Myriapoda, and Hexapoda, with the last two subphyla grouped together in a taxon named Atelocerata). Molecular analyses during the last decade (Boore et al. 1995; Friedrich and Tautz 1995; Boore, Lavrov, and Brown 1998; Burke et al. 1998; Cook et al. 2001; Hwang et al. 2001) have challenged this view and favored an alternative phylogeny in which the Crustacea and the Hexapoda form a monophyletic clade, Pancrustacea, whereas Myriapoda either feature as the sister group of the latter, or are grouped together with the Chelicerata in a not yet formally named clade. Within the Pancrustacea, the hexapods are sometimes found to be the sister group of the Malacostraca (Wilson et al. 2000; Hwang et al. 2001; Nardi et al. 2001): in this case, the Crustacea appears to be a paraphyletic taxon. Molecular data concerning mitochondrial genomes, both as primary sequences and as gene order (Boore et al. 1995; Boore, Lavrov, and Brown 1998; Hwang et al. 2001), has strongly contributed to these new phylogenetic views. However, taxonomic sampling of the main arthropod lineages has been uneven, with Chilopoda being very poorly represented and no known data on the mtDNA of the Scutigeromorpha, the sister group of all remaining centipedes, based on both morphological and molecular evidence (e.g., Edgecombe and Giribet 2002). Hence, we sequenced the complete mitochondrial genome of the common house centipede Scutigera coleoptrata, in order to know more about the higher phylogeny of arthropods, in particular to test whether the Myriapoda are monophyletic (Edgecombe and Giribet 2002) and what their phylogenetic relationships are with the other major arthropodan clades. These questions are obviously central to a reconstruction of the origin of terrestrial habits and corresponding morphological and physiological adaptations evolved in some of the most successful animal lineages on earth.
We show here the results of a phylogenetic analysis based on the complete mitochondrial gene-coding sequences and their products from representatives of these four taxa, including the house centipede Scutigera coleoptrata, whose sequence is discussed for the first time in this article.
Materials and Methods
Scutigera Coleoptrata mtDNA Sequencing
Total DNA was extracted from a specimen of Scutigera coleoptrata (Linnaeus, 1758) (Chilopoda, Myriapoda) collected in the regional park of the Euganean Hills (Italy).
A long polymerase chain rection (PCR) strategy was applied to amplify the whole mitochondrial genome. To start, a 449-bp fragment of the large ribosomal subunit rrnL was PCR amplified with the two primers 16Sfor (5'-CGG GTT GAA CTC AGA TCA-3') and 16Srev (5'-CGC GCC TGT TTA TCA AAA ACA T-3') which were designed on a multiple alignment of several rrnLs, including the centipede Lithobius forficatus. The DNA sequence obtained from this segment was used to select the two primers ScuHPK16Saa (5'-GAT TAT GCT ACC TTC GCA CGG TCA AAA TAC CG-3') and ScuHPK16Sbb (5'-CAT ATC GAC AAT AAG GGT TGC GAC CTC GAT GTT G-3'). The complete genome was then amplified using the Expand Long Template PCR (Roche Biochemicals) according to the protocol provided, with the ScuHPK16Saa and ScuHPK16Sbb primers. A genome walking strategy was successively applied to sequence directly the products obtained through consecutive long PCR amplifications. Sequencing was performed at the CRIBI Sequencing Service (University of Padova) on automated DNA sequencers.
The complete sequence of the Scutigera coleoptrata mitochondrial genome is available at the European Molecular Biology Laboratory (EMBL) database under accession number AJ507061.
Data Set Selection
Complete mitochondrial genome sequences for taxa other than Scutigera were obtained from GenBank. Taxa were selected according to two criteria: (1) a balanced and representative selection of the major arthropod clades and (2) a selection of genomes having distinct gene orders, thus representing the genomic order diversity within the phylum. The taxa analyzed are listed in table 1. Lumbricus terrestris (Annelida) and Katharina tunicata (Mollusca) were selected as outgroups. The resulting starting number of taxa was 21, including S. coleoptrata. The choice of an annelid and a mollusk as outgroups was made because no complete mitochondrial genome sequences are available for the species more closely related to the Arthopoda sensu stricto—i.e., the tardigrads and onychophorans. Complete mitochondrial genomes of nematodes (which, like the arthropods, belong to the ecdysozoan lineage) could have been considered as outgroups, but their extremely high rate of substitution and strong compositional bias made them unsuitable for this purpose (Foster and Hickey 1999; Hwang et al. 2001). The results of the analyses detailed below show that the sequences of the two outgroups are homogeneous for composition and evolved at a comparable rate with the ingroup sequences, thus excluding possible artifacts in the placement of the root within the ingroup.
Table 1 List of the Species and Their MtDNA Gene Order.
Sequence Alignment and Blocks Selection
Data sets relative to the 13 coded proteins were aligned using ClustalW (Thompson, Higgins, and Gibson 1994) and improved by visual inspection. In preliminary alignments it was noted that some proteins of Apis mellifera and Heterodoxus macropus, characterized by a high rate of divergence, affected negatively the alignment of informative and well-conserved portions of the sequence from other taxa; the two species were thus excluded from further analyses. Multiple alignments were analyzed with the Gblocks program (Castresana 2000), using default settings, to select conserved amino acid regions. These regions were concatenated in a single alignment spanning 2,226 amino acid residues (hereafter, PRO19T). PRO19T was used as a backbone to align the corresponding nucleotide sequences. Thus a single alignment of 6,678 bases was obtained (hereafter, NUC19T). Both PRO19T and NUC19T are available at EMBL under accession numbers ALIGN_000477, ALIGN_000478.
On these data sets we investigated the major factors that could potentially bias the phylogenetic analysis and therefore affect tree topology strongly. In particular, we conducted several preliminary investigations to verify possible effects resulting from (1) the rate of the nucleotide substitution process, (2) the amount of phylogenetic signal present in the data sets, (3) the compositional bias in NUC19T and PRO19T, (4) taxon sampling, and (5) usage of the protein data set versus the nucleotide data set.
Rate of the Nucleotide-Substitution Process
The level of saturation in the whole codons, and at the first, second, and third codon positions separately, was analyzed using scatter plot graphics, comparing the uncorrected p-distances (i.e., the distances obtained by dividing the number of observed differences between each couple of sequences by the length of the pairwise alignment) with the distances calculated by applying the best-fit evolutionary model (GTR + G + I) selected by the Modeltest program (Posada and Crandall 1998). In the absence of saturation in the nucleotide substitution process, GTR + G + I distances coincide with p-distances; conversely, GTR + G + I distances are larger than p-distances when the level of saturation increases. GTR + G + I distances are distinctly larger than p-distances in the case of saturation of the nucleotide-substitution process.
Phylogenetic Signal Detection
A priori estimation of the phylogenetic signal present in the analyzed data sets was performed according to the maximum likelihood mapping method (Strimmer and von Haeseler 1997) using the Tree-Puzzle 5.0 program (Schmidt et al. 2002). With this approach, the amount of phylogenetic signal present in a multiple alignment is mirrored by the percentage of the fully resolved quartets (Strimmer and von Haeseler 1997). Moreover, the length of the alignment affects the phylogenetic resolution. In the case of the nucleotide data sets, the amount of phylogenetic signal was estimated for first, second, third, and first + second positions separately.
Compositional Bias
The statistical significance of the nucleotide compositional biases was investigated for whole codons, as well as for all possible combinations of first, second, and third positions. The departure from homogeneity in base composition across the taxa was checked by applying the 2 test available in the PAUP 4.10 program (Swofford 2002) to nucleotide data sets. Conversely, the protein compositional bias was analyzed with the Tree-Puzzle 5.0 program (Schmidt et al. 2002).
Taxon Sampling
Many preliminary analyses were conducted on different intermediate data sets obtained from NUC19T or PRO19T by inclusion/exclusion of selected taxa and inclusion/exclusion of first/second and third positions or protein sequences (data not shown). Phylogenetic methods applied to these data sets are presented in detail below, under Tree Reconstruction. These analyses were performed to investigate the effects due to factors considered above in the sequence alignment and blocks selection paragraphs.
Among the numerous intermediate data sets investigated, we selected two final nucleotide data sets to (1) minimize departure from compositional homogeneity, (2) maximize taxon sampling, (3) maximize the amount of phylogenetic signal, (4) reduce the amount of homoplasy due to saturation, and (5) speed up the computational efforts, particularly in the ML bootstrap test. The two final data sets are described below. The first (hereafter named 14T4452) includes both first and second codon positions of 14 taxa (length: 4,452 bases) and was obtained by exclusion of the A. franciscana, D. pulex, R. sanguineus, I. hexagonus, and T. bielanensis sequences. The second data set (hereafter named 19T2226) includes all 19 taxa and is based on the second codon positions only (length: 2,226 bases). Parallel to the nucleotide data sets, we considered also two protein data sets. The first (named PRO14T) includes the same taxa present in 14T4452 data set and was obtained by removing from the PRO19T the above-mentioned taxa. The second data set was PRO19T itself. This choice allowed an easy comparison among trees obtained from nucleotide data sets and those obtained from protein data sets.
Tree Reconstruction
Phylogenetic analyses on the PRO19T, PRO14T, 14T4452, and 19T2226 data sets were performed according to the Bayesian inference (BI) (Huelsenbeck et al. 2001), maximum likelihood (ML), minimum evolution (ME) or Neighbor-Joining (NJ), maximum parsimony (MP) (Nei and Kumar 2000), and quartet-puzzling (QP) methods (Strimmer and von Haeseler 1996). Our results were derived mainly from analyses performed applying ML methods (including BI). In fact ML methods are very flexible due to their plasticity—i.e., the possibility to implement and apply complex evolutionary models that account for several biases faced by sequences during evolution. Moreover, ML methods are theoretically very sound and statistically consistent and have proved to be very efficient in recovering correct phylogenies, even when the sequences analyzed have evolved through very complicated evolutionary pathways (Whelan, Liò, and Goldman 2001).
The ML, ME, and MP phylogenetic analyses on the nucleotide data sets were performed with PAUP 4.10 (Swofford 2002). The branch swapping was performed applying the tree bisection-reconnection TBR algorithm with the steepest descent option activated. For the ML and ME trees, the (GTR + G + I) models fitting best to the data sets were selected with the Modeltest program and according to the akaike criterion (Posada and Crandall 1998). Parsimony analyses were limited to parsimony-informative characters. Bayesian phylogenetic analyses were performed with MrBayes 2.1 (Huelsenbeck and Ronquist 2001). A GTR + G + I evolutionary model was applied in all analyses. The Metropolis-coupled Markov chain Monte Carlo sampling approach was used to calculate posterior probabilities. Prior probabilities for all trees were equal, starting trees were random, tree sampling was done every 20 generations, and burn-in values were determined empirically from the likelihood values. To check for consistency of results, four Markov chains were run simultaneously for 100,000 generations.
The NJ, MP, ML, and QP analyses on protein data sets were performed with the PHYLIP 3.6a3 and Tree-Puzzle programs (Felsenstein 2002; Schmidt et al. 2002).
Neighbor-Joining analysis was conducted in two steps: the ML distances were calculated with Tree-Puzzle, applying the mtREV24 substitution matrix (Adachi and Hasegawa 1996), then the ML distances matrices were provided to the NEIGHBOR program in the PHYLIP package to built up the final NJ trees. The same strategy was used for the nonparametric bootstrap, starting from 100 data matrices obtained respectively from PRO14T and PRO19T, using the program SEQBOOT in the PYLIP package. Maximum parsimony analysis was conducted using PROTPARS with the Jumble option active. Maximum likelihood trees were built up using PROML, with the Jumble and Global rearrangement options active, applying both the Dayhoff-PAM and JTT substitution matrices (Dayhoff, Schwartz, and Orcutt 1978; Jones, Taylor, and Thornton 1992). Alternatively, QP trees were built up with Tree-Puzzle, applying the mtREV24 substitution matrix and a four rate approximated gamma distribution of among-site rate heterogeneity with or without a portion of the sites considered invariable. In this case the best tree was selected by the Kishino-Hasegawa test (Kishino and Hasegawa 1989).
Bayesian inference analysis on PRO14T and PRO19T data sets were done using the MrBayes 3 program (Ronquist and Huelsenbeck 2003). An mtREV substitution matrix (Adachi and Hasegawa 1996) was used while the invgamma option corresponding to G + I was used to treat the among-site rate variation. All other options were identical to those applied to the nucleotide data sets.
Nonparametric Bootstrap Test
The nonparametric bootstrap test (BT) (Felsenstein 1985) was performed to test the robustness of tree topologies. In the case of the 14T4452 and 19T2226 nucleotide data sets we performed 100 replicates for the ML trees and 1,000 replicates for the ME and MP trees. For the protein PRO14T and PRO19T data sets, 100 replicates were always performed.
Testing for Alternative Tree Topologies
To evaluate alternative phylogenetic hypotheses, the almost unbiased test (AU) and the weighted Shimodaira-Hasegawa test (WSH) (Shimodaira 2002) were calculated for different tree topologies. Calculations were done on 14T4452 and 19T4452 data sets using the CONSEL program (Shimodaira and Hasegawa 2001). The 19T4452 data set was obtained from NUC19 by deleting third codon positions. This set presents a distinctly higher level of phylogenetic signal than the 19T2226 data set (see Results). Maximum likelihood and BI tree topologies obtained from 19T4452 match perfectly with those obtained from the 19T2226 data set. We could not use the 19T4452 data set in our ML bootstrap test because of the extremely long computational time required with the available computer facilities. Alternative tree topologies obtained from protein data sets were evaluated with the Bayesian posterior probability (BPP) as implemented in MrBayes 3.
Results
The mitochondrial genome of Scutigera coleoptrata is 14,922 bp long (a detailed analysis is presented in Negrisolo, Minelli, and Valle 2004). It codes for 13 proteins, 22 tRNAs, and two rRNAs—i.e. the standard genetic content of the metazoan mitochondrial genome (Boore 1999). The gene order, however, is unique with respect to all of the arthropodan genomes so far investigated (see table 1). Comparison of the S. coleoptrata mtDNA arrangement with the corresponding arrangement in Limulus, which is regarded as primitive in the context of the Arthropoda (Staton, Daehler, and Brown 1997; Boore 1999; Lavrov, Boore, and Brown 2002), reveals that it was reached through at least 10 translocations involving four protein-encoding genes (nad3, nad4L, nad6, and nad1) and six tRNAs genes (trnN, trnS2, trnI, trnM, trnC, and trnY) (Negrisolo, Minelli, and Valle 2004).
Rate of the Nucleotide Substitution Process
The level of nucleotide substitution in the NUC19T data set, as evaluated by scatter plot graphics, is presented in figure 1. Second positions show the lowest level of saturation. First positions present a higher level of saturation, especially in those species (T. bielanesis, A. franciscana, D. pulex, I. hexagonus, and R. sanguineus) that have GTR + G + I distances >1 from many other taxa of the ingroup. Third positions are fully saturated.
FIG. 1. Scatter plot graphics performed to test the saturation of the nucleotide substitution process. x axis: GTR + G + I distances calculated by applying the best-fit evolutionary model (GTR + G + I) selected by Modeltest (Posada and Crandall 1998). Y axis: uncorrected p-distances. All distances were calculated with PAUP* (Swofford 2002). A, scatter plot graphic for whole codons; B, C, D, same for first, second, and third positions in the NUC19T alignment. When there is no saturation in the nucleotide-substitution process—i.e. when GTR + G + I distances = p-distances—points are located along the line of equation y = x; conversely, when the saturation increases—i.e., GTR + G + I distance > p-distance—the departure of the points from the y = x line increases too
Phylogenetic Signal Detection
An a priori estimation of the phylogenetic signal present in all analyzed data sets is shown in figure 2. In the NUC19 data set, first and second positions shows a good and comparable amount of phylogenetic signal, having, respectively, 87.4% and 91.6% of the quartets analyzed fully resolved (fig. 2A and 2B). Note that the second positions' alignment corresponds to the 19T2226 data set. Conversely, third positions present a very poor amount of phylogenetic signal, with only 38.2% of the quartets fully resolved and 49.2% of the quartets totally unresolved (fig. 2C). Thus the saturation of the nucleotide substitution process at third positions (see above) corresponds to a strong drop of the phylogenetic signal. This circumstance invites us to exclude third positions from further analyses. Finally, first + second position alignment, i.e. the 19T4452 data set (see Materials and Methods), presents 94.1% of all quartets analyzed fully resolved (fig. 2D); moreover, its length is twice that of each single position (4,452 bases vs. 2,226 bases). Because the length of the alignment affects the phylogenetic resolution and the percentage of resolved quartets reflects the final phylogenetic signal (Strimmer and von Haeseler 1997), the 19T4452 data set presents a distinctly higher level of phylogenetic signal than single position alignment. The percentage of fully resolved quartets in the 14T4452 (fig. 2E) and 19T2226 (fig. 2B) data sets is comparable (94.7% vs. 91.6%). Fully or partly unresolved quartets are 4.3% in 14T4452 and 8.2% in 19T2226. However, the length of the first alignment is twice that of the second. Thus, for the same reasons given above for the 19T4452 alignment, the phylogenetic signal in the 4T4452 data set is distinctly higher than that in the 19T2226 data set. Conversely, both PRO14T and PRO19T present good phylogenetic signal.
FIG. 2. Likelihood mapping for the NUC19T, 14T4452, PRO14T, and PRO19T data sets. A, first positions in NUC19T; B, second positions in NUC19T (=19T2226 data set); C, third positions in NUC19T; D, first + second positions in NUC19T (=19T4452 data set); E, in 14T4452; F, PRO14T; G, PRO19T. All possible quartets were evaluated. Values in percent. According to the likelihood mapping method (Strimmer and von Haeseler 1997), the amount of phylogenetic signal present in the multiple alignment considered is reflected by the percentage of fully resolved quartets—i.e., to the sum of the values in the corner regions of the triangle and to the length of the alignment. Conversely, the partly unresolved quartets (lateral rectangles) and the fully unresolved quartets (central triangle) indicate a loss of the signal, particularly the latter
Compositional Bias
A graphical representation of base composition in NUC19T is presented in figure 3. Second positions show the most regular compositional profile within the whole data set, whereas third positions appear to be the most uneven. The statistical significance of the nucleotide compositional bias was investigated for whole codons, as well as for all possible combinations of first, second, and third positions. The departure from homogeneity in base composition across the taxa was checked by applying the 2 test. This test was not significant (P = 0.418) for second position data among all taxa considered, the tick R. sanguineus excepted. The data set becomes even more homogeneous if the sequence of the other tick I. hexagonus is also excluded (P = 0.718). Conversely, the inclusion of both ticks into the second position (19T2226) data set reduces the homogeneity (P = 0.021). For all remaining nucleotide data sets, base compositional homogeneity was detected for some closely related taxa at most. The 2 test on PRO14T was not significant, and all 14 sequences showed a homogeneous result (P > 0.05). Conversely, in the PRO19T data set the sequences of L. terrestris (P = 0.01), D. pulex (P = 0.01), I. hexagonus (P = 0.00), and R. sanguineus (P = 0.00) are not homogeneous with respect to the other species.
FIG. 3. Base composition values (in percent for each base) in NUC19T. A, whole codons; B, first positions; C, second positions; D, third positions. Taxon identifiers: 1 = Lumbricus terrestris; 2 = Katharina tunicata; 3 = Limulus polyphemus; 4 = Ixodes hexagonus; 5 = Rhipicephalus sanguineus; 6 = Lithobius forficatus; 7 = Scutigera coleoptrata; 8 = Narceus annularis; 9 = Thyropygus sp.; 10 = Pagurus longicarpus; 11 = Penaeus monodon; 12 = Artemia franciscana; 13 = Daphnia pulex; 14 = Tetrodontophora bielanensis; 15 = Locusta migratoria; 16 = Triatoma dimidiata; 17 = Ceratitis capitata; 18 = Drosophila yakuba; 19 = Anopheles gambiae
Phylogenetic Analysis
The results of the phylogenetic reconstructions performed on data sets 14T4452 and PRO14T are summarized in figure 4. Most analyses were congruent in recovering two major clades within the phylum Arthropoda: one including Chelicerata plus Myriapoda, the other comprising Crustacea plus Hexapoda. The BI and ML trees obtained from 14T4452 data set have the same topology and favor myriapod paraphyly, the Chilopoda showing up as sister group of the Chelicerata, and the Diplopoda as a sister group of Chilopoda + Chelicerata. This topology is strongly supported by bootstrap and BI values. The MP and ME trees derived from the same data set are in almost complete agreement with the BI and ML trees. However, the MP tree favors a monophyletic myriapod clade with sister-group relationships to the Chelicerata, whereas the ME tree shows a myriapod paraphyletic clade, with the Diplopoda as sister group of Chelicerata and Chilopoda basal to Diplopoda + Chelicerata. These alternative topologies, however, are not supported by bootstrap analysis. Moreover, the single most parsimonious tree obtained from the 14T4452 data set had quite low parsimony scores (Consistency Index = 0.4708; Retention Index = 0.3625; Rescaled Consistency Index = 0.1706). The BI tree obtained from 14PROT differs from the nucML tree in recovering a monophyletic Myriapoda (BI support 79%) sister group to L. polyphemus and the clade L. migratoria + T. dimidiata (BI support 75%). The protML tree was the same, irrespective of the substitution matrix used and the treatment of site rate variation. Both the protML and the protMP trees favored a monophyletic Myriapoda sister group to L. polyphemus. The former received very marginal QP support (51%), but neither tree was BT supported. Finally, the protNJ tree exhibits a rather different topology, with the Diplopoda as basal clade to all other Arthropoda and the Chilopoda + L. polyphemus sister group of Crustacea + Hexapoda. However this topology does not receive any BT support al all.
[in this window]
[in a new window]
FIG. 4. Phylogenetic analysis performed on the 14T4452 and PRO14T data sets. The phylogram shown is the best maximum likelihood tree (-lnL = 33988.49) obtained from the nucleotide 14T4452 data set. Bar represents 0.1 mutation per site. Nodes receiving support by one or more of the applied phylogenetic methods, i.e., Bayesian inference (BI), maximum likelihood (ML) minimum evolution (ME), maximum parsimony (MP), Neighbor-Joining (NJ), and quartet-puzzling (QP) are labeled with lowercase letters. nt = values obtained from 14T4452 nucleotide data set; pr = values obtained from PRO14T data set. Bootstrap and Bayesian inference values, in percent, are listed in the table
The results of phylogenetic analyses based on the 19T2226 and PRO19T data sets are summarized in figure 5. The subdivision of Arthropoda in the Crustacea + Hexapoda and Chelicerata + Myriapoda clades was supported by these analyses too.
FIG. 5. Phylogenetic analysis performed on the 19T2226 and PRO19T data set. The phylogram shown is the best maximum likelihood tree (-lnL = 17654.15) obtained from the nucleotide 19T2226 data set. Bar represents 0.1 mutation per site. Nodes receiving support by one or more of the applied phylogenetic methods, i.e., Bayesian inference (BI), maximum likelihood (ML), minimum evolution (ME), maximum parsimony (MP), Neighbor-Joining (NJ), and quartet-puzzling (QP) are labeled by lowercase letters. nt = values obtained from 19T2226 nucleotide data set; pr = values obtained from PRO19T data set. Bootstrap and Bayesian inference values, in percent, are listed in the table. Stars on branches indicate a water-to-land transition within the Arthropoda clade
The nucBI and nucML tree topologies agree perfectly. Bayesian inference support in favor of the topology presented in figure 5 is higher than the BT support. The latter is weak or absent (nodes f, i, l), especially at the deepest nodes of the tree, a result expected as a consequence of the lower phylogenetic signal present in the 19T2226 data set (see above). Maximum parsimony analysis conducted on the 19T2226 data set gave two equally most parsimonious trees with low parsimony scores (Consistency Index = 0.4272; Retention Index = 0.3729; Rescaled Consistency Index = 0.1593). The strict consensus tree of the two shows some very unusual outputs: (1) A. franciscana + D. pulex is recovered as basal clade of all other Arthropoda; (2) L. forficatus does not group with S. coleoptrata; (3) T. bielanensis is basal to all Arthropoda except A. franciscana + D. pulex. But this topology is not supported by BT values. The nucME tree favors a monophyletic Myriapoda sister group of Chelicerata, but with no BT support. Moreover, this tree presents some unconventional placements—i.e., A. franciscana + D. pulex as basal clade of other Arthropoda with 61% BT support; and T. bielanensis as sister taxon of Myriapoda + Chelicerata with no BT support.
The BI tree obtained from PRO19T is mostly congruent with the nucML tree, from which it however differs in favoring the L. migratoria + T. dimidiata clade (BI support 95%). The protML tree was the same, irrespective of the substitution matrix used and the treatment of site rate variation. This tree differs from the nucML tree in recovering a monophyletic Myriapoda (without QP or BT support) sister group of Chelicerata and the L. migratoria + T. dimidiata clade (BT support 70% only). Maximum parsimony analysis on PROT19T produced a single tree, which presented some unusual groupings: Chelicerata are disrupted with L. polyphemus sister taxon of Diplopoda, whereas the two ticks form a clade with T. bielanensis (marginally BT supported, 57%). Moreover, this latter group is placed as sister taxon to all remaining Arthropoda. Finally, the protNJ tree mostly agrees with the nucML tree, except for the myriapod clade that is recovered as monophyletic, but not receiving BT support, and sister group of Chelicerata.
Testing for Alternative Tree Topologies
To avoid overinterpretation of our results, we performed on the nucleotide data sets the AU and WSH statistical tests that allow comparison of alternative phylogenetic hypotheses. Possible controversial points in our tree topologies can thus be evaluated as follows. AU as well as WSH tests were not significant in rejecting (1) the monophyly of the Myriapoda, (2) the paraphyly of Crustacea, or (3) the placement of T. bielanensis within the Hexapoda. The monophyly of Myriapoda received also marginal support in the ML analysis, and a more robust support in BI reconstructions on protein data sets. Conversely AU tests strongly reject (P < 0.001) the hypothesis of the Myriapoda as sister group of the Hexapoda, or as basal clade of the Mandibulata (P < 0.001). These results were also supported by the very conservative WSH test (P < 0.05). Bayesian posterior probability tests performed on the protein data sets give results very similar to those obtained on the 14T4452 and 19T4452 nucleotide data sets. In particular, these tests rejected strongly the placement of the Myriapoda as either sister group of Hexapoda (P < 0.001) or as a basal clade of the Mandibulata (P < 0.05), whereas they were not significant in solving the above-mentioned points—i.e., the monophyly vs. paraphyly of Myriapoda—as well as the placement of T. bielanensis and the paraphyly of Crustacea.
Discussion
Factors Affecting the Reconstruction of Arthropod Phylogeny
Arthropod mtDNA sequences underwent a complex evolution that in several taxa involved gene order rearrangements or was accompanied by strong compositional bias (Boore 1999; Lavrov, Brown, and Boore 2000; Hickerson and Cunningham 2000; Nardi et al. 2001; Shao, Campbell, and Barker 2001; Lavrov, Boore, and Brown 2002; Roehrdanz, Degrugillier, and Black 2002). However, this should be expected because of the very ancient divergence of the major arthropodan clades. Chelicerates and crustaceans are well documented since the Early Cambrian period (Briggs, Erwin, and Collier 1994), whereas scutigeromorph centipedes are known to have existed since the Upper Silurian period (Shear, Jeram, and Selden 1998). The average A + T content in the mtDNA is around 70% in most taxa. To properly explain these changes, an approach is necessary that accounts for multiple substitutions at single sites and that accommodates compositional biases. This type of approach proved to be matched by the ML and BI methods in this study. In fact, the analyses conducted on the various data sets were consistent when performed with ML and BI methods, whereas results obtained applying the MP, NJ, and ME approaches were uneven. Irrespective of the phylogenetic method applied, the inclusion of third positions was particularly severe in affecting tree topology. Even the application of a specific codon-based model (BI analyses not shown) did not allow us to circumvent topology artifacts resulting from inclusion of third positions. Taxon sampling had no strong effect when third positions were excluded.
Analyses performed on the protein data sets give results that agree, in large part, with those obtained from nucleotide data sets when analyzed with BI and ML methods. Differences observed (see details in the Results) cannot be solved with the data sets at hand. The ML topologies obtained from the PRO14T and PRO19T multiple alignments receive poor bootstrap support in some of the more basal and phylogenetically critical nodes. These results were expected as a consequence of the selection process of conserved amino acid blocks (Castresana 2000). In fact, the choice made by the Gblocks program of conserved amino acid portions has two positive effects: (1) the level of homogeneity among the sequences is markedly increased and (2) chances are very low that positional homology is lost. These are, of course, very desirable properties of the final alignment. Conversely, this method may remove the more diverging or some misleading coincidentally biased positions, thus reducing particularly the support of basal nodes in the ML tree (see Castresana 2000 for a more detailed discussion). This effect, however, does not appear to influence the nucleotide multiple alignments, which produced topologies well supported even at the basal nodes whenever a high phylogenetic signal was present (see fig. 2).
Main Phylogenetic Results
In our analysis based on the 19T2226 and PRO19T data sets, the Crustacea were recovered as a monophyletic group (nucBI, nucML, protML, protNJ, and protMP analyses) whereas some previous molecular analyses had indicated a paraphyletic Crustacea with respect to Hexapoda (Wilson et al. 2000; Cook et al. 2001; Nardi et al. 2001). Moreover, the spring-tail Tetrodontophora bielanensis does not group with the other hexapods. This result is in agreement with previous molecular studies (Nardi et al. 2001, 2003), suggesting an isolated position for the Collembola within the Pancrustacea. Heterometabolous Hexapoda were recovered, alternatively monophyletic or paraphyletic, but this point was not settled by statistical tests. Finally, our analysis adds strong support to previous molecular studies (Friedrich and Tautz 1995; Burke et al. 1998; Cook et al. 2001; Hwang et al. 2001) that grouped myriapods with chelicerates, as well as crustaceans with hexapods. However, we recovered a paraphyletic myriapod clade with the result that the Chilopoda were more closely related to the Chelicerata than to the Diplopoda. This is the first molecular evidence suggesting a paraphyletic Myriapoda with respect to Chelicerata, even if this finding is not conclusive.
Myriapod Paraphyly and the Arthropod Water-to-Land Transition
Paraphyly of the Myriapoda in the context of the Mandibulata has been repeatedly suggested, mainly using morphological data (Borucki 1996; Kraus 1998; Edgecombe et al. 2000; Giribet and Ribera 2000). A monophyletic Myriapoda as basal clade of the Mandibulata has been recently recovered (Giribet, Edgecombe, and Wheeler 2001) following a combined analysis of morphological and molecular data, and a monophyletic Myriapoda as basal clade of all Arthropoda was obtained (Regier and Shultz 2001) in a tree based on the nuclear gene elongation factor-2. Finally, immunocytochemical and neuroanatomical studies (Loesel, N?ssel, and Strausfeld 2002) provided evidence in favor of a paraphyletic Myriapoda as basal clade within the Arthropoda. Our phylogenetic analyses, in agreement with a variety of molecular and developmental studies (Boore et al. 1995; Friedrich and Tautz 1995; Boore, Lavrov, and Brown 1998; Burke et al. 1998; Cook et al. 2001; Hwang et al. 2001; Dove and Stollewerk 2003), support a close relationship between Myriapoda and Chelicerata and go even further favoring a paraphyletic myriapodan clade, although this finding requires more decisive evidence. Conversely, our results are conclusive in rejecting the monophyly of the mandibulate clade. This has strong implications for the origin of terrestriality within the Arthropoda. According to our results, a water-to-land transition occurred at least three times during the evolution of this phylum (not to mention the woodlice and other more or less strictly terrestrial crustaceans). The more striking circumstance is the independent water-to-land transition of myriapods and hexapods that led to the parallel acquisition of many morphological features whose similarities require new interpretations.
Our analysis could even suggest (fig. 5) two additional independent transitions from water to land, one for the collembolans, represented here by T. bielanensis, if separate from the true insects, and the other for the diplopods, if these are the sister group of centipedes + chelicerates. The phylogenetic support for these two latter events is still open to question. Nevertheless, comparative morphology suggests a few possible synapomorphies of Chilopoda + Chelicerata, including the peculiar feeding mechanism, the segmental position of the excretory organs and a morphological instability frequently observed around mid-trunk.
The vast majority of chelicerates and centipedes, but not the millipedes, feed on fluid food, mostly of animal origin and digested preorally or extraorally (Brusca and Brusca 2002).
As for the excretory system, arachnids have "coxal glands" that mostly open behind legs I and III (Kaestner 1968), whereas centipedes have a "maxillary rein" opening on the second maxillary segment (Rilling 1968). This segmental distribution might represent an instance of positional homology. In fact, despite the different kind of appendages borne on these segments, the third leg pair of arachnids and the second maxillary segment of the centipedes are both regarded as segmentally homologous to the insect labium (respectively, Telford and Thomas 1998; Kraus and Kraus 1994).
Finally, in chelicerates and centipedes, but not in millipedes, one or a few trunk segments around mid-body are either reduced or distinguished by unique morphological markers. In several chelicerates, the sternum of one or more anterior segments of the opisthosoma becomes "subducted" under neighboring sclerites (Shultz 1993). This feature has a segmental equivalent in the "mid-body anomaly" of centipedes (Minelli et al. 2000). In the centipede Lithobius, at least, this segmental anomaly does probably correspond to an early, transient embryonic expression of Abdominal-B (Hughes and Kaufmann 2002). Further comparative studies, morphological, embryological, and molecular, will eventually check the robustness of the phylogenetic relationships suggested by our analysis.
Acknowledgements
We are grateful to Charles E. Cook, Diethard Tautz, Richard Thomas, and two anonymous referees for their insightful comments on a previous version of this paper. E.N. was supported by a postdoctoral fellowship from the University of Padova. We thank the BMR sequencing team at the CRIBI (University of Padova) for major help in the sequencing of the mitochondrial genome of Scutigera coleoptrata. E.N. is also grateful to Dr. Paolo Laveder and Dr. Riccardo Schiavon (University of Padova) for helpful suggestions during the bench work. Research supported by a MIUR grant to G.V. (FIRB 2001—RBNE01F5W7_007).
Literature Cited
Adachi, J., and M. Hasegawa. 1996. Model of amino acid substitution in proteins encoded by mitochondrial DNA. J. Mol. Evol. 42:459-468.
Beard, C. B., D. M. Hamm, and F. H. Collins. 1993. The mitochondrial genome of the mosquito Anopheles gambiae, DNA sequence, genome organization and comparisons with mitochondrial sequences of other insects. Insect Mol. Biol. 2:103-124.
Black, W. C., IV, and R. L. Roehrdanz. 1998. Mitochondrial gene order is not conserved in arthropods: prostriate and metastriate tick mitochondrial genomes. Mol. Biol. Evol. 15:1772-1785.
Boore, J. L. 1999. Animal mitochondrial genomes. Nucleic Acids Res. 27:1767-1780.
Boore, J. L. 2001. Mitochondrial gene arrangement source guide, version 6.0. Department of the Environment Joint Genome Institute, Walnut Creek, Calif. (Available from http://www.jgi.doe.gov/index.html).
Boore, J. L., and W. M. Brown. 1994. Complete DNA sequence of the mitochondrial genome of the black chiton, Katharina tunicata. Genetics 138:423-443.
Boore, J. L., and W. M. Brown. 1995. Complete sequence of the mitochondrial DNA of the annelid worm Lumbricus terrestris. Genetics 141:305-319.
Boore, J. L., T. M. Collins, D. Stanton, L. L. Daehler, and W. M. Brown. 1995. Deducing the pattern of arthropod phylogeny from mitochondrial DNA rearrangements. Nature 376:163-165.
Boore, J. L., D. V. Lavrov, and W. M. Brown. 1998. Gene translocation links insects and crustaceans. Nature 392:667-668.
Borucki, H. 1996. Evolution und phylogenetisches System der Chilopoda (Mandibulata, Tracheata). Verh. Naturwiss. Ver. Hamburg NF 35:95-226.
Briggs, D. E. G., D. H. Erwin, and F. J. Collier. 1994. The fossils of the Burges Shale. Smithsonian Institution Press, Washington.
Brusca, R. C., and G. J. Brusca. 2002. Invertebrates. Second edition. Sinauer Associates, Sunderland, Mass.
Burke, W. D., H. S. Malik, W. C. Lathe, III, and T. H. Eickbush. 1998. Are retrotransposons long-term hitchhikers? Nature 392:141-142.
Castresana, J. 2000. Selection of conserved blocks from multiple alignments for their use in phylogenetic analysis. Mol. Biol. Evol. 17:540-552.
Clary, D. O., and D. R. Wolstenholme. 1985. The mitochondrial DNA molecule of Drosophila yakuba: nucleotide sequence, gene organization, and genetic code. J. Mol. Evol. 22:252-271.
Cook, C. E., M. L. Smith, M. J. Telford, A. Bastianello, and M. Akam. 2001. Hox genes and the phylogeny of the arthropods. Curr. Biol. 11:759-763.
Crease, T. J. 1999. The complete sequence of the mitochondrial genome of Daphnia pulex (Cladocera: Crustacea). Gene 233:89-99.
Crozier, R. H., and Y. C. Crozier. 1993. The mitochondrial genome of the honeybee Apis mellifera: complete sequence and genome organization. Genetics 133:97-117.
Dayhoff, M. O., R. M. Schwartz, and B. C. Orcutt. 1978. A model of evolutionary change in proteins. Pp. 345–352 in M. O. Dayhoff, ed., Atlas of protein sequence structure, vol. 5, National Biomedical Research Foundation, Washington, D.C.
Dotson, E. M., and C. B. Beard. 2001. Sequence and organization of the mitochondrial genome of the Chagas disease vector, Triatoma dimidiata. Insect Mol. Biol. 10:205-215.
Dove, H., and A. Stollewerk. 2003. Comparative analysis of neurogenesis in the myriapod Glomeris marginata (Diplopoda) suggests more similarities to chelicerates than to insects. Development 130:2161-2171.
Edgecombe, G. D., and G. Giribet. 2002. Myriapod phylogeny and the relationships of Chilopoda. Pp. 143–168 in J. J. Llorente Bousquets and J. J. Morrone, eds., Biodiversidad, taxonomía y biogeografia de artrópodos de México: hacia una síntesis de su conocimiento, vol. 3, Prensas de Ciencias, Universidad Nacional Autónoma de México.
Edgecombe, G. D., G. D. F. Wilson, D. J. Colgan, M. R. Gray, and G. Cassis. 2000. Arthropod cladistics: combined analysis of histone H3 and U2 sn RNA sequences and morphology. Cladistics 16:204-231.
Felsenstein, J. 1985. Confidence limits on phylogenies: an approach using bootstrap. Evolution 39:783-791.
Felsenstein, J. 2002. PHYLIP (Phylogeny Inference Package) version 3.6a3 (Univ. of Washington, Seattle).
Flook, P. K., C. H. F. Rowell, and G. Gellissen. 1995. The sequence, organization, and evolution of the Locusta migratoria mitochondrial genome. J. Mol. Evol. 41:928-941.
Foster, P. G., and D. A. Hickey. 1999. Compositional bias may affect both DNA-based and protein-based phylogenetic reconstructions. J. Mol. Evol. 48:284-290.
Friedrich, M., and D. Tautz. 1995. Ribosomal DNA phylogeny of the major extant arthropod classes and the evolution of myriapods. Nature 376:165-167.
Giribet, G., G. D. Edgecombe, and W. C. Wheeler. 2001. Arthropod phylogeny based on eight molecular loci and morphology. Nature 413:157-161.
Giribet, G., and C. Ribera. 2000. A review of arthropod phylogeny: new data based on ribosomal DNA sequences and direct character optimization. Cladistics 16:204-231.
Hickerson, M. J., and C. W. Cunningham. 2000. Dramatic mitochondrial gene rearrangements in the hermit crab Pagurus longicarpus (Crustacea, Anomura). Mol. Biol. Evol. 17:639-644.
Huelsenbeck, J. P., and F. Ronquist. 2001. MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics 17:754-755.
Huelsenbeck, J. P., F. Ronquist, R. Nielsen, and J. P. Bollback. 2001. Bayesian inference of phylogeny and its impact on evolutionary biology. Science 294:2310-2314.
Hughes, C. L., and T. C. Kaufman. 2002. Exploring the myriapod body plan: expression patterns of the ten hox genes in a centipede. Development 129:1225-1238.
Hwang, U. W., M. Friedrich, D. Tautz, C. J. Park, and W. Kim. 2001. Mitochondrial protein phylogeny joins myriapods with chelicerates. Nature 413:154-157.
Jones, D. T., W. R. Taylor, and J. M. Thornton. 1992. The rapid generation of mutation data matrices from protein sequences. Comput. Appl. Biosci. 8:275-282.
Kaestner, A. 1968. Invertebrate zoology, 2. Wiley-Interscience, New York.
Kishino, H., and M. Hasegawa. 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J. Mol. Evol. 29:170-179.
Kraus, O. 1998. Phylogenetic relationships between higher taxa of tracheate arthropods. Pp. 295–303 in R. A. Fortey, and R. H. Thomas, eds., Arthropod relationships, Chapman & Hall, London.
Kraus, O., and M. Kraus. 1994. Phylogenetic system of the Tracheata (Mandibulata): on "Myriapoda"—Insecta interrelationships, phylogenetic age and primary ecological niches. Verh. naturwiss. Ver. Hamburg N.F. 34:5-31.
Lavrov, D. V., J. L. Boore, and W. M. Brown. 2000. The complete mitochondrial DNA sequence of the horseshoe crab Limulus polyphemus. Mol. Biol. Evol. 17:813-824.
Lavrov, D. V., J. L. Boore, and W. M. Brown. 2002. Complete mtDNA sequences of two millipedes suggest a new model for mitochondrial gene rearrangements: duplication and nonrandom loss. Mol. Biol. Evol. 19:163-169.
Lavrov, D. V., W. M. Brown, and J. L. Boore. 2000. A novel type of RNA editing occurs in the mitochondrial tRNAs of the centipede Lithobius forficatus. Proc. Natl. Acad. Sci. USA 97:13738-13742.
Loesel, R., D. R. N?ssel, and N. J. Strausfeld. 2002. Common design in a unique midline neuropil in the brains of arthropods. Arthropod Struct. Dev. 31:77-91.
Minelli, A., D. Foddai, L. A. Pereira, and J. G. E. Lewis. 2000. The evolution of segmentation of centipede trunk and appendages. J. Zool. Syst. Evol. Res. 38:103-117.
Nardi, F., A. Carapelli, P. P. Fanciulli, R. Dallai, and F. Frati. 2001. The complete mitochondrial DNA sequence of the basal hexapod Tetrodontophora bielanensis: evidence for heteroplasmy and tRNA translocations. Mol. Biol. Evol. 18:1293-1304.
Nardi, F., G. Spinsanti, J. L. Boore, A. Carapelli, R. Dallai, and F. Frati. 2003. Hexapod origins: monophyletic or paraphyletic? Science 299:1887-1889.
Negrisolo, E., A. Minelli, and G. Valle. 2004. Extensive gene order rearrangement in the mitochondrial genome of the centipede Scutigera coleoptrata. J. Mol. Evol. 58 (in press).
Nei, S., and K. Kumar. 2000. Molecular evolution and phylogenetics. Oxford University Press, New York.
Posada, D., and K. A. Crandall. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14:817-818.
Regier, J. C., and J. W. Shultz. 2001. Elongation factor-2: useful gene for arthropod phylogenetics. Mol. Phylogenet. Evol. 20:136-148.
Rilling, G. 1968. Lithobius forficatus L., in Grosses zoologisches Praktikum. Fischer, Stuttgart, 13B:1-136.
Roehrdanz, R. L., M. E. Degrugillier, and W. C. Black, IV. 2002. Novel rearrangements of arthropod mitochondrial DNA detected with long-PCR: applications to arthropod phylogeny and evolution. Mol. Biol. Evol. 19:841-849.
Ronquist, F., and J. P. Huelsenbeck. 2003. MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19:1572-1574.
Schmidt H. A., K. Strimmer, M. Vingron, and A. von Haeseler. 2002. Tree-Puzzle: maximum likelihood phylogenetic analysis using quartets and parallel computing. Bioinformatics 18:502-504.
Shao, R., N. J. H. Campbell, and S. C. Barker. 2001. Numerous gene rearrangements in the mitochondrial genome of the wallaby louse, Heterodoxus macropus (Phthiraptera). Mol. Biol. Evol. 18:858-865.
Shear, W. A., A. J. Jeram, and P. A. Selden. 1998. Centipede legs (Arthropoda, Chilopoda, Scutigeromorpha) from the Silurian and Devonian of Britain and the Devonian of North America. Am. Mus. Novitates. 3231:1-16.
Shimodaira, H. 2002. An approximately unbiased test of phylogenetic tree selection. Syst. Biol. 51:492-508.
Shimodaira, H., and M. Hasegawa. 2001. CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinformatics 17:1246-1247.
Shultz, J. M. 1993. Muscular anatomy of the giant whipscorpion Mastigoproctus giganteus (Lucas) (Arachnida: Uropygi) and its evolutionary significance. Zool. J. Linn. Soc. 108:335-365.
Spanos, L., G. Koutroumbas, M. Kostyfakis, and C. Louis. 2000. The mitochondrial genome of the Mediterranean fruit fly, Ceratitis capitata. Insect Mol. Biol. 9:139-144.
Staton, J. L., L. L. Daehler, and W. M. Brown. 1997. Mitochondrial gene arrangement of the horseshoe crab Limulus polyphemus L.: conservation of major features among arthropodan classes. Mol. Biol. Evol. 14:867-874.
Strimmer, K., and A. von Haeseler. 1996. Quartet puzzling: a quartet maximum likelihood method for reconstructing tree topologies. Mol. Biol. Evol. 13:964-969.
Strimmer, K., and A. von Haeseler. 1997. Likelihood-mapping: a simple method to visualize phylogenetic content of a sequence alignment. Proc. Natl. Acad. Sci. USA 94:6815-6819.
Swofford, D. L. 2002. PAUP*. Phylogenetic analysis using parsimony (*and other methods). Version 4.10. Sinauer Associates, Sunderland, Mass.
Telford, M. J., and R. H. Thomas. 1998. Expression of homeobox genes shows chelicerate arthropods retain their deutocerebral segment. Proc. Natl. Acad. Sci. USA 95:10671-10675.
Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. ClustalW: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-4680.
Valverde, J. R., B. Batuecas, C. Moratilla, R. Marco, and R. Garesse. 1994. The complete mitochondrial DNA sequence of the crustacean Artemia franciscana. J. Mol. Evol. 39:400-408.
Whelan, S., P. Liò, and N. Goldman. 2001. Molecular phylogenetics: state-of-the art methods for looking into the past. Trends Genet. 17:262-272.
Wilson, K., V. Cahill, E. Ballment, and J. Benzie. 2000. The complete sequence of the mitochondrial genome of the crustacean Penaeus monodon: are malacostracan crustaceans more closely related to insects than to branchiopods? Mol. Biol. Evol. 17:863-874.(Enrico Negrisolo, Alessan)
E-mail alessandro.minelli@unipd.it.
Abstract
Recent advances in molecular phylogenetics are continuously changing our perception of the phylogenetic relationships among the main arthropod lineages: crustaceans, hexapods, chelicerates, and myriapods. Besides the intrinsic interest in unraveling the evolution of the largest animal phylum, these studies are basic to an understanding of one of the major transitions in animal evolution—i.e., the conquest of land with all its associated structural and functional adaptations. Myriapods have been traditionally considered the closest relatives of hexapods, thus implying only one origin of terrestriality for the tracheate lineage, but this view is now challenged by molecular evidence. Sequence data available to date for centipedes and millipedes are very limited, and the taxon sampling is strongly biased. The most critical gap was the scutigeromorph centipedes, which are the sister group to all remaining Chilopoda from which they probably diverged in the Silurian if not earlier. We obtained the first complete mitochondrial sequence for a representative of this clade, the house centipede. In our phylogenetic analyses of the protein-coding genes in this mitochondrial genome, along with 16 further ones representing the other major arthropod clades plus two outgroups, the myriapods formed a clade with the chelicerates. This implies that water-to-land transition occurred at least three times (hexapods, myriapods, arachnids) during the evolution of the Arthropoda. In addition, in contrast to all previous studies, our best supported topologies favor paraphyly of the myriapods with respect to the chelicerates. This would increase to four the main events of land colonization in arthropods (once for centipedes, once for millipedes).
Key Words: Chilopoda ? Myriapoda ? Scutigera coleoptrata ? mitochondrial genome ? arthropod phylogeny
Introduction
The phylum Arthropoda is the most diversified and species-rich group in the animal kingdom. Although there is almost unanimous consensus on the monophyly of the phylum (Giribet and Ribera 2000), there is also hot debate about the phylogenetic relationships among the four main arthropodan lineages: chelicerates, myriapods, crustaceans, and hexapods. In addition, the monophyly of some of these groups has been repeatedly questioned. In particular, the paraphyly and even polyphyly of the Myriapoda has frequently been proposed (Edgecombe and Giribet 2002; Loesel, N?ssel, and Strausfeld 2002).
Traditional classifications placed the Chelicerata as sister group of the Mandibulata (Crustacea, Myriapoda, and Hexapoda, with the last two subphyla grouped together in a taxon named Atelocerata). Molecular analyses during the last decade (Boore et al. 1995; Friedrich and Tautz 1995; Boore, Lavrov, and Brown 1998; Burke et al. 1998; Cook et al. 2001; Hwang et al. 2001) have challenged this view and favored an alternative phylogeny in which the Crustacea and the Hexapoda form a monophyletic clade, Pancrustacea, whereas Myriapoda either feature as the sister group of the latter, or are grouped together with the Chelicerata in a not yet formally named clade. Within the Pancrustacea, the hexapods are sometimes found to be the sister group of the Malacostraca (Wilson et al. 2000; Hwang et al. 2001; Nardi et al. 2001): in this case, the Crustacea appears to be a paraphyletic taxon. Molecular data concerning mitochondrial genomes, both as primary sequences and as gene order (Boore et al. 1995; Boore, Lavrov, and Brown 1998; Hwang et al. 2001), has strongly contributed to these new phylogenetic views. However, taxonomic sampling of the main arthropod lineages has been uneven, with Chilopoda being very poorly represented and no known data on the mtDNA of the Scutigeromorpha, the sister group of all remaining centipedes, based on both morphological and molecular evidence (e.g., Edgecombe and Giribet 2002). Hence, we sequenced the complete mitochondrial genome of the common house centipede Scutigera coleoptrata, in order to know more about the higher phylogeny of arthropods, in particular to test whether the Myriapoda are monophyletic (Edgecombe and Giribet 2002) and what their phylogenetic relationships are with the other major arthropodan clades. These questions are obviously central to a reconstruction of the origin of terrestrial habits and corresponding morphological and physiological adaptations evolved in some of the most successful animal lineages on earth.
We show here the results of a phylogenetic analysis based on the complete mitochondrial gene-coding sequences and their products from representatives of these four taxa, including the house centipede Scutigera coleoptrata, whose sequence is discussed for the first time in this article.
Materials and Methods
Scutigera Coleoptrata mtDNA Sequencing
Total DNA was extracted from a specimen of Scutigera coleoptrata (Linnaeus, 1758) (Chilopoda, Myriapoda) collected in the regional park of the Euganean Hills (Italy).
A long polymerase chain rection (PCR) strategy was applied to amplify the whole mitochondrial genome. To start, a 449-bp fragment of the large ribosomal subunit rrnL was PCR amplified with the two primers 16Sfor (5'-CGG GTT GAA CTC AGA TCA-3') and 16Srev (5'-CGC GCC TGT TTA TCA AAA ACA T-3') which were designed on a multiple alignment of several rrnLs, including the centipede Lithobius forficatus. The DNA sequence obtained from this segment was used to select the two primers ScuHPK16Saa (5'-GAT TAT GCT ACC TTC GCA CGG TCA AAA TAC CG-3') and ScuHPK16Sbb (5'-CAT ATC GAC AAT AAG GGT TGC GAC CTC GAT GTT G-3'). The complete genome was then amplified using the Expand Long Template PCR (Roche Biochemicals) according to the protocol provided, with the ScuHPK16Saa and ScuHPK16Sbb primers. A genome walking strategy was successively applied to sequence directly the products obtained through consecutive long PCR amplifications. Sequencing was performed at the CRIBI Sequencing Service (University of Padova) on automated DNA sequencers.
The complete sequence of the Scutigera coleoptrata mitochondrial genome is available at the European Molecular Biology Laboratory (EMBL) database under accession number AJ507061.
Data Set Selection
Complete mitochondrial genome sequences for taxa other than Scutigera were obtained from GenBank. Taxa were selected according to two criteria: (1) a balanced and representative selection of the major arthropod clades and (2) a selection of genomes having distinct gene orders, thus representing the genomic order diversity within the phylum. The taxa analyzed are listed in table 1. Lumbricus terrestris (Annelida) and Katharina tunicata (Mollusca) were selected as outgroups. The resulting starting number of taxa was 21, including S. coleoptrata. The choice of an annelid and a mollusk as outgroups was made because no complete mitochondrial genome sequences are available for the species more closely related to the Arthopoda sensu stricto—i.e., the tardigrads and onychophorans. Complete mitochondrial genomes of nematodes (which, like the arthropods, belong to the ecdysozoan lineage) could have been considered as outgroups, but their extremely high rate of substitution and strong compositional bias made them unsuitable for this purpose (Foster and Hickey 1999; Hwang et al. 2001). The results of the analyses detailed below show that the sequences of the two outgroups are homogeneous for composition and evolved at a comparable rate with the ingroup sequences, thus excluding possible artifacts in the placement of the root within the ingroup.
Table 1 List of the Species and Their MtDNA Gene Order.
Sequence Alignment and Blocks Selection
Data sets relative to the 13 coded proteins were aligned using ClustalW (Thompson, Higgins, and Gibson 1994) and improved by visual inspection. In preliminary alignments it was noted that some proteins of Apis mellifera and Heterodoxus macropus, characterized by a high rate of divergence, affected negatively the alignment of informative and well-conserved portions of the sequence from other taxa; the two species were thus excluded from further analyses. Multiple alignments were analyzed with the Gblocks program (Castresana 2000), using default settings, to select conserved amino acid regions. These regions were concatenated in a single alignment spanning 2,226 amino acid residues (hereafter, PRO19T). PRO19T was used as a backbone to align the corresponding nucleotide sequences. Thus a single alignment of 6,678 bases was obtained (hereafter, NUC19T). Both PRO19T and NUC19T are available at EMBL under accession numbers ALIGN_000477, ALIGN_000478.
On these data sets we investigated the major factors that could potentially bias the phylogenetic analysis and therefore affect tree topology strongly. In particular, we conducted several preliminary investigations to verify possible effects resulting from (1) the rate of the nucleotide substitution process, (2) the amount of phylogenetic signal present in the data sets, (3) the compositional bias in NUC19T and PRO19T, (4) taxon sampling, and (5) usage of the protein data set versus the nucleotide data set.
Rate of the Nucleotide-Substitution Process
The level of saturation in the whole codons, and at the first, second, and third codon positions separately, was analyzed using scatter plot graphics, comparing the uncorrected p-distances (i.e., the distances obtained by dividing the number of observed differences between each couple of sequences by the length of the pairwise alignment) with the distances calculated by applying the best-fit evolutionary model (GTR + G + I) selected by the Modeltest program (Posada and Crandall 1998). In the absence of saturation in the nucleotide substitution process, GTR + G + I distances coincide with p-distances; conversely, GTR + G + I distances are larger than p-distances when the level of saturation increases. GTR + G + I distances are distinctly larger than p-distances in the case of saturation of the nucleotide-substitution process.
Phylogenetic Signal Detection
A priori estimation of the phylogenetic signal present in the analyzed data sets was performed according to the maximum likelihood mapping method (Strimmer and von Haeseler 1997) using the Tree-Puzzle 5.0 program (Schmidt et al. 2002). With this approach, the amount of phylogenetic signal present in a multiple alignment is mirrored by the percentage of the fully resolved quartets (Strimmer and von Haeseler 1997). Moreover, the length of the alignment affects the phylogenetic resolution. In the case of the nucleotide data sets, the amount of phylogenetic signal was estimated for first, second, third, and first + second positions separately.
Compositional Bias
The statistical significance of the nucleotide compositional biases was investigated for whole codons, as well as for all possible combinations of first, second, and third positions. The departure from homogeneity in base composition across the taxa was checked by applying the 2 test available in the PAUP 4.10 program (Swofford 2002) to nucleotide data sets. Conversely, the protein compositional bias was analyzed with the Tree-Puzzle 5.0 program (Schmidt et al. 2002).
Taxon Sampling
Many preliminary analyses were conducted on different intermediate data sets obtained from NUC19T or PRO19T by inclusion/exclusion of selected taxa and inclusion/exclusion of first/second and third positions or protein sequences (data not shown). Phylogenetic methods applied to these data sets are presented in detail below, under Tree Reconstruction. These analyses were performed to investigate the effects due to factors considered above in the sequence alignment and blocks selection paragraphs.
Among the numerous intermediate data sets investigated, we selected two final nucleotide data sets to (1) minimize departure from compositional homogeneity, (2) maximize taxon sampling, (3) maximize the amount of phylogenetic signal, (4) reduce the amount of homoplasy due to saturation, and (5) speed up the computational efforts, particularly in the ML bootstrap test. The two final data sets are described below. The first (hereafter named 14T4452) includes both first and second codon positions of 14 taxa (length: 4,452 bases) and was obtained by exclusion of the A. franciscana, D. pulex, R. sanguineus, I. hexagonus, and T. bielanensis sequences. The second data set (hereafter named 19T2226) includes all 19 taxa and is based on the second codon positions only (length: 2,226 bases). Parallel to the nucleotide data sets, we considered also two protein data sets. The first (named PRO14T) includes the same taxa present in 14T4452 data set and was obtained by removing from the PRO19T the above-mentioned taxa. The second data set was PRO19T itself. This choice allowed an easy comparison among trees obtained from nucleotide data sets and those obtained from protein data sets.
Tree Reconstruction
Phylogenetic analyses on the PRO19T, PRO14T, 14T4452, and 19T2226 data sets were performed according to the Bayesian inference (BI) (Huelsenbeck et al. 2001), maximum likelihood (ML), minimum evolution (ME) or Neighbor-Joining (NJ), maximum parsimony (MP) (Nei and Kumar 2000), and quartet-puzzling (QP) methods (Strimmer and von Haeseler 1996). Our results were derived mainly from analyses performed applying ML methods (including BI). In fact ML methods are very flexible due to their plasticity—i.e., the possibility to implement and apply complex evolutionary models that account for several biases faced by sequences during evolution. Moreover, ML methods are theoretically very sound and statistically consistent and have proved to be very efficient in recovering correct phylogenies, even when the sequences analyzed have evolved through very complicated evolutionary pathways (Whelan, Liò, and Goldman 2001).
The ML, ME, and MP phylogenetic analyses on the nucleotide data sets were performed with PAUP 4.10 (Swofford 2002). The branch swapping was performed applying the tree bisection-reconnection TBR algorithm with the steepest descent option activated. For the ML and ME trees, the (GTR + G + I) models fitting best to the data sets were selected with the Modeltest program and according to the akaike criterion (Posada and Crandall 1998). Parsimony analyses were limited to parsimony-informative characters. Bayesian phylogenetic analyses were performed with MrBayes 2.1 (Huelsenbeck and Ronquist 2001). A GTR + G + I evolutionary model was applied in all analyses. The Metropolis-coupled Markov chain Monte Carlo sampling approach was used to calculate posterior probabilities. Prior probabilities for all trees were equal, starting trees were random, tree sampling was done every 20 generations, and burn-in values were determined empirically from the likelihood values. To check for consistency of results, four Markov chains were run simultaneously for 100,000 generations.
The NJ, MP, ML, and QP analyses on protein data sets were performed with the PHYLIP 3.6a3 and Tree-Puzzle programs (Felsenstein 2002; Schmidt et al. 2002).
Neighbor-Joining analysis was conducted in two steps: the ML distances were calculated with Tree-Puzzle, applying the mtREV24 substitution matrix (Adachi and Hasegawa 1996), then the ML distances matrices were provided to the NEIGHBOR program in the PHYLIP package to built up the final NJ trees. The same strategy was used for the nonparametric bootstrap, starting from 100 data matrices obtained respectively from PRO14T and PRO19T, using the program SEQBOOT in the PYLIP package. Maximum parsimony analysis was conducted using PROTPARS with the Jumble option active. Maximum likelihood trees were built up using PROML, with the Jumble and Global rearrangement options active, applying both the Dayhoff-PAM and JTT substitution matrices (Dayhoff, Schwartz, and Orcutt 1978; Jones, Taylor, and Thornton 1992). Alternatively, QP trees were built up with Tree-Puzzle, applying the mtREV24 substitution matrix and a four rate approximated gamma distribution of among-site rate heterogeneity with or without a portion of the sites considered invariable. In this case the best tree was selected by the Kishino-Hasegawa test (Kishino and Hasegawa 1989).
Bayesian inference analysis on PRO14T and PRO19T data sets were done using the MrBayes 3 program (Ronquist and Huelsenbeck 2003). An mtREV substitution matrix (Adachi and Hasegawa 1996) was used while the invgamma option corresponding to G + I was used to treat the among-site rate variation. All other options were identical to those applied to the nucleotide data sets.
Nonparametric Bootstrap Test
The nonparametric bootstrap test (BT) (Felsenstein 1985) was performed to test the robustness of tree topologies. In the case of the 14T4452 and 19T2226 nucleotide data sets we performed 100 replicates for the ML trees and 1,000 replicates for the ME and MP trees. For the protein PRO14T and PRO19T data sets, 100 replicates were always performed.
Testing for Alternative Tree Topologies
To evaluate alternative phylogenetic hypotheses, the almost unbiased test (AU) and the weighted Shimodaira-Hasegawa test (WSH) (Shimodaira 2002) were calculated for different tree topologies. Calculations were done on 14T4452 and 19T4452 data sets using the CONSEL program (Shimodaira and Hasegawa 2001). The 19T4452 data set was obtained from NUC19 by deleting third codon positions. This set presents a distinctly higher level of phylogenetic signal than the 19T2226 data set (see Results). Maximum likelihood and BI tree topologies obtained from 19T4452 match perfectly with those obtained from the 19T2226 data set. We could not use the 19T4452 data set in our ML bootstrap test because of the extremely long computational time required with the available computer facilities. Alternative tree topologies obtained from protein data sets were evaluated with the Bayesian posterior probability (BPP) as implemented in MrBayes 3.
Results
The mitochondrial genome of Scutigera coleoptrata is 14,922 bp long (a detailed analysis is presented in Negrisolo, Minelli, and Valle 2004). It codes for 13 proteins, 22 tRNAs, and two rRNAs—i.e. the standard genetic content of the metazoan mitochondrial genome (Boore 1999). The gene order, however, is unique with respect to all of the arthropodan genomes so far investigated (see table 1). Comparison of the S. coleoptrata mtDNA arrangement with the corresponding arrangement in Limulus, which is regarded as primitive in the context of the Arthropoda (Staton, Daehler, and Brown 1997; Boore 1999; Lavrov, Boore, and Brown 2002), reveals that it was reached through at least 10 translocations involving four protein-encoding genes (nad3, nad4L, nad6, and nad1) and six tRNAs genes (trnN, trnS2, trnI, trnM, trnC, and trnY) (Negrisolo, Minelli, and Valle 2004).
Rate of the Nucleotide Substitution Process
The level of nucleotide substitution in the NUC19T data set, as evaluated by scatter plot graphics, is presented in figure 1. Second positions show the lowest level of saturation. First positions present a higher level of saturation, especially in those species (T. bielanesis, A. franciscana, D. pulex, I. hexagonus, and R. sanguineus) that have GTR + G + I distances >1 from many other taxa of the ingroup. Third positions are fully saturated.
FIG. 1. Scatter plot graphics performed to test the saturation of the nucleotide substitution process. x axis: GTR + G + I distances calculated by applying the best-fit evolutionary model (GTR + G + I) selected by Modeltest (Posada and Crandall 1998). Y axis: uncorrected p-distances. All distances were calculated with PAUP* (Swofford 2002). A, scatter plot graphic for whole codons; B, C, D, same for first, second, and third positions in the NUC19T alignment. When there is no saturation in the nucleotide-substitution process—i.e. when GTR + G + I distances = p-distances—points are located along the line of equation y = x; conversely, when the saturation increases—i.e., GTR + G + I distance > p-distance—the departure of the points from the y = x line increases too
Phylogenetic Signal Detection
An a priori estimation of the phylogenetic signal present in all analyzed data sets is shown in figure 2. In the NUC19 data set, first and second positions shows a good and comparable amount of phylogenetic signal, having, respectively, 87.4% and 91.6% of the quartets analyzed fully resolved (fig. 2A and 2B). Note that the second positions' alignment corresponds to the 19T2226 data set. Conversely, third positions present a very poor amount of phylogenetic signal, with only 38.2% of the quartets fully resolved and 49.2% of the quartets totally unresolved (fig. 2C). Thus the saturation of the nucleotide substitution process at third positions (see above) corresponds to a strong drop of the phylogenetic signal. This circumstance invites us to exclude third positions from further analyses. Finally, first + second position alignment, i.e. the 19T4452 data set (see Materials and Methods), presents 94.1% of all quartets analyzed fully resolved (fig. 2D); moreover, its length is twice that of each single position (4,452 bases vs. 2,226 bases). Because the length of the alignment affects the phylogenetic resolution and the percentage of resolved quartets reflects the final phylogenetic signal (Strimmer and von Haeseler 1997), the 19T4452 data set presents a distinctly higher level of phylogenetic signal than single position alignment. The percentage of fully resolved quartets in the 14T4452 (fig. 2E) and 19T2226 (fig. 2B) data sets is comparable (94.7% vs. 91.6%). Fully or partly unresolved quartets are 4.3% in 14T4452 and 8.2% in 19T2226. However, the length of the first alignment is twice that of the second. Thus, for the same reasons given above for the 19T4452 alignment, the phylogenetic signal in the 4T4452 data set is distinctly higher than that in the 19T2226 data set. Conversely, both PRO14T and PRO19T present good phylogenetic signal.
FIG. 2. Likelihood mapping for the NUC19T, 14T4452, PRO14T, and PRO19T data sets. A, first positions in NUC19T; B, second positions in NUC19T (=19T2226 data set); C, third positions in NUC19T; D, first + second positions in NUC19T (=19T4452 data set); E, in 14T4452; F, PRO14T; G, PRO19T. All possible quartets were evaluated. Values in percent. According to the likelihood mapping method (Strimmer and von Haeseler 1997), the amount of phylogenetic signal present in the multiple alignment considered is reflected by the percentage of fully resolved quartets—i.e., to the sum of the values in the corner regions of the triangle and to the length of the alignment. Conversely, the partly unresolved quartets (lateral rectangles) and the fully unresolved quartets (central triangle) indicate a loss of the signal, particularly the latter
Compositional Bias
A graphical representation of base composition in NUC19T is presented in figure 3. Second positions show the most regular compositional profile within the whole data set, whereas third positions appear to be the most uneven. The statistical significance of the nucleotide compositional bias was investigated for whole codons, as well as for all possible combinations of first, second, and third positions. The departure from homogeneity in base composition across the taxa was checked by applying the 2 test. This test was not significant (P = 0.418) for second position data among all taxa considered, the tick R. sanguineus excepted. The data set becomes even more homogeneous if the sequence of the other tick I. hexagonus is also excluded (P = 0.718). Conversely, the inclusion of both ticks into the second position (19T2226) data set reduces the homogeneity (P = 0.021). For all remaining nucleotide data sets, base compositional homogeneity was detected for some closely related taxa at most. The 2 test on PRO14T was not significant, and all 14 sequences showed a homogeneous result (P > 0.05). Conversely, in the PRO19T data set the sequences of L. terrestris (P = 0.01), D. pulex (P = 0.01), I. hexagonus (P = 0.00), and R. sanguineus (P = 0.00) are not homogeneous with respect to the other species.
FIG. 3. Base composition values (in percent for each base) in NUC19T. A, whole codons; B, first positions; C, second positions; D, third positions. Taxon identifiers: 1 = Lumbricus terrestris; 2 = Katharina tunicata; 3 = Limulus polyphemus; 4 = Ixodes hexagonus; 5 = Rhipicephalus sanguineus; 6 = Lithobius forficatus; 7 = Scutigera coleoptrata; 8 = Narceus annularis; 9 = Thyropygus sp.; 10 = Pagurus longicarpus; 11 = Penaeus monodon; 12 = Artemia franciscana; 13 = Daphnia pulex; 14 = Tetrodontophora bielanensis; 15 = Locusta migratoria; 16 = Triatoma dimidiata; 17 = Ceratitis capitata; 18 = Drosophila yakuba; 19 = Anopheles gambiae
Phylogenetic Analysis
The results of the phylogenetic reconstructions performed on data sets 14T4452 and PRO14T are summarized in figure 4. Most analyses were congruent in recovering two major clades within the phylum Arthropoda: one including Chelicerata plus Myriapoda, the other comprising Crustacea plus Hexapoda. The BI and ML trees obtained from 14T4452 data set have the same topology and favor myriapod paraphyly, the Chilopoda showing up as sister group of the Chelicerata, and the Diplopoda as a sister group of Chilopoda + Chelicerata. This topology is strongly supported by bootstrap and BI values. The MP and ME trees derived from the same data set are in almost complete agreement with the BI and ML trees. However, the MP tree favors a monophyletic myriapod clade with sister-group relationships to the Chelicerata, whereas the ME tree shows a myriapod paraphyletic clade, with the Diplopoda as sister group of Chelicerata and Chilopoda basal to Diplopoda + Chelicerata. These alternative topologies, however, are not supported by bootstrap analysis. Moreover, the single most parsimonious tree obtained from the 14T4452 data set had quite low parsimony scores (Consistency Index = 0.4708; Retention Index = 0.3625; Rescaled Consistency Index = 0.1706). The BI tree obtained from 14PROT differs from the nucML tree in recovering a monophyletic Myriapoda (BI support 79%) sister group to L. polyphemus and the clade L. migratoria + T. dimidiata (BI support 75%). The protML tree was the same, irrespective of the substitution matrix used and the treatment of site rate variation. Both the protML and the protMP trees favored a monophyletic Myriapoda sister group to L. polyphemus. The former received very marginal QP support (51%), but neither tree was BT supported. Finally, the protNJ tree exhibits a rather different topology, with the Diplopoda as basal clade to all other Arthropoda and the Chilopoda + L. polyphemus sister group of Crustacea + Hexapoda. However this topology does not receive any BT support al all.
[in this window]
[in a new window]
FIG. 4. Phylogenetic analysis performed on the 14T4452 and PRO14T data sets. The phylogram shown is the best maximum likelihood tree (-lnL = 33988.49) obtained from the nucleotide 14T4452 data set. Bar represents 0.1 mutation per site. Nodes receiving support by one or more of the applied phylogenetic methods, i.e., Bayesian inference (BI), maximum likelihood (ML) minimum evolution (ME), maximum parsimony (MP), Neighbor-Joining (NJ), and quartet-puzzling (QP) are labeled with lowercase letters. nt = values obtained from 14T4452 nucleotide data set; pr = values obtained from PRO14T data set. Bootstrap and Bayesian inference values, in percent, are listed in the table
The results of phylogenetic analyses based on the 19T2226 and PRO19T data sets are summarized in figure 5. The subdivision of Arthropoda in the Crustacea + Hexapoda and Chelicerata + Myriapoda clades was supported by these analyses too.
FIG. 5. Phylogenetic analysis performed on the 19T2226 and PRO19T data set. The phylogram shown is the best maximum likelihood tree (-lnL = 17654.15) obtained from the nucleotide 19T2226 data set. Bar represents 0.1 mutation per site. Nodes receiving support by one or more of the applied phylogenetic methods, i.e., Bayesian inference (BI), maximum likelihood (ML), minimum evolution (ME), maximum parsimony (MP), Neighbor-Joining (NJ), and quartet-puzzling (QP) are labeled by lowercase letters. nt = values obtained from 19T2226 nucleotide data set; pr = values obtained from PRO19T data set. Bootstrap and Bayesian inference values, in percent, are listed in the table. Stars on branches indicate a water-to-land transition within the Arthropoda clade
The nucBI and nucML tree topologies agree perfectly. Bayesian inference support in favor of the topology presented in figure 5 is higher than the BT support. The latter is weak or absent (nodes f, i, l), especially at the deepest nodes of the tree, a result expected as a consequence of the lower phylogenetic signal present in the 19T2226 data set (see above). Maximum parsimony analysis conducted on the 19T2226 data set gave two equally most parsimonious trees with low parsimony scores (Consistency Index = 0.4272; Retention Index = 0.3729; Rescaled Consistency Index = 0.1593). The strict consensus tree of the two shows some very unusual outputs: (1) A. franciscana + D. pulex is recovered as basal clade of all other Arthropoda; (2) L. forficatus does not group with S. coleoptrata; (3) T. bielanensis is basal to all Arthropoda except A. franciscana + D. pulex. But this topology is not supported by BT values. The nucME tree favors a monophyletic Myriapoda sister group of Chelicerata, but with no BT support. Moreover, this tree presents some unconventional placements—i.e., A. franciscana + D. pulex as basal clade of other Arthropoda with 61% BT support; and T. bielanensis as sister taxon of Myriapoda + Chelicerata with no BT support.
The BI tree obtained from PRO19T is mostly congruent with the nucML tree, from which it however differs in favoring the L. migratoria + T. dimidiata clade (BI support 95%). The protML tree was the same, irrespective of the substitution matrix used and the treatment of site rate variation. This tree differs from the nucML tree in recovering a monophyletic Myriapoda (without QP or BT support) sister group of Chelicerata and the L. migratoria + T. dimidiata clade (BT support 70% only). Maximum parsimony analysis on PROT19T produced a single tree, which presented some unusual groupings: Chelicerata are disrupted with L. polyphemus sister taxon of Diplopoda, whereas the two ticks form a clade with T. bielanensis (marginally BT supported, 57%). Moreover, this latter group is placed as sister taxon to all remaining Arthropoda. Finally, the protNJ tree mostly agrees with the nucML tree, except for the myriapod clade that is recovered as monophyletic, but not receiving BT support, and sister group of Chelicerata.
Testing for Alternative Tree Topologies
To avoid overinterpretation of our results, we performed on the nucleotide data sets the AU and WSH statistical tests that allow comparison of alternative phylogenetic hypotheses. Possible controversial points in our tree topologies can thus be evaluated as follows. AU as well as WSH tests were not significant in rejecting (1) the monophyly of the Myriapoda, (2) the paraphyly of Crustacea, or (3) the placement of T. bielanensis within the Hexapoda. The monophyly of Myriapoda received also marginal support in the ML analysis, and a more robust support in BI reconstructions on protein data sets. Conversely AU tests strongly reject (P < 0.001) the hypothesis of the Myriapoda as sister group of the Hexapoda, or as basal clade of the Mandibulata (P < 0.001). These results were also supported by the very conservative WSH test (P < 0.05). Bayesian posterior probability tests performed on the protein data sets give results very similar to those obtained on the 14T4452 and 19T4452 nucleotide data sets. In particular, these tests rejected strongly the placement of the Myriapoda as either sister group of Hexapoda (P < 0.001) or as a basal clade of the Mandibulata (P < 0.05), whereas they were not significant in solving the above-mentioned points—i.e., the monophyly vs. paraphyly of Myriapoda—as well as the placement of T. bielanensis and the paraphyly of Crustacea.
Discussion
Factors Affecting the Reconstruction of Arthropod Phylogeny
Arthropod mtDNA sequences underwent a complex evolution that in several taxa involved gene order rearrangements or was accompanied by strong compositional bias (Boore 1999; Lavrov, Brown, and Boore 2000; Hickerson and Cunningham 2000; Nardi et al. 2001; Shao, Campbell, and Barker 2001; Lavrov, Boore, and Brown 2002; Roehrdanz, Degrugillier, and Black 2002). However, this should be expected because of the very ancient divergence of the major arthropodan clades. Chelicerates and crustaceans are well documented since the Early Cambrian period (Briggs, Erwin, and Collier 1994), whereas scutigeromorph centipedes are known to have existed since the Upper Silurian period (Shear, Jeram, and Selden 1998). The average A + T content in the mtDNA is around 70% in most taxa. To properly explain these changes, an approach is necessary that accounts for multiple substitutions at single sites and that accommodates compositional biases. This type of approach proved to be matched by the ML and BI methods in this study. In fact, the analyses conducted on the various data sets were consistent when performed with ML and BI methods, whereas results obtained applying the MP, NJ, and ME approaches were uneven. Irrespective of the phylogenetic method applied, the inclusion of third positions was particularly severe in affecting tree topology. Even the application of a specific codon-based model (BI analyses not shown) did not allow us to circumvent topology artifacts resulting from inclusion of third positions. Taxon sampling had no strong effect when third positions were excluded.
Analyses performed on the protein data sets give results that agree, in large part, with those obtained from nucleotide data sets when analyzed with BI and ML methods. Differences observed (see details in the Results) cannot be solved with the data sets at hand. The ML topologies obtained from the PRO14T and PRO19T multiple alignments receive poor bootstrap support in some of the more basal and phylogenetically critical nodes. These results were expected as a consequence of the selection process of conserved amino acid blocks (Castresana 2000). In fact, the choice made by the Gblocks program of conserved amino acid portions has two positive effects: (1) the level of homogeneity among the sequences is markedly increased and (2) chances are very low that positional homology is lost. These are, of course, very desirable properties of the final alignment. Conversely, this method may remove the more diverging or some misleading coincidentally biased positions, thus reducing particularly the support of basal nodes in the ML tree (see Castresana 2000 for a more detailed discussion). This effect, however, does not appear to influence the nucleotide multiple alignments, which produced topologies well supported even at the basal nodes whenever a high phylogenetic signal was present (see fig. 2).
Main Phylogenetic Results
In our analysis based on the 19T2226 and PRO19T data sets, the Crustacea were recovered as a monophyletic group (nucBI, nucML, protML, protNJ, and protMP analyses) whereas some previous molecular analyses had indicated a paraphyletic Crustacea with respect to Hexapoda (Wilson et al. 2000; Cook et al. 2001; Nardi et al. 2001). Moreover, the spring-tail Tetrodontophora bielanensis does not group with the other hexapods. This result is in agreement with previous molecular studies (Nardi et al. 2001, 2003), suggesting an isolated position for the Collembola within the Pancrustacea. Heterometabolous Hexapoda were recovered, alternatively monophyletic or paraphyletic, but this point was not settled by statistical tests. Finally, our analysis adds strong support to previous molecular studies (Friedrich and Tautz 1995; Burke et al. 1998; Cook et al. 2001; Hwang et al. 2001) that grouped myriapods with chelicerates, as well as crustaceans with hexapods. However, we recovered a paraphyletic myriapod clade with the result that the Chilopoda were more closely related to the Chelicerata than to the Diplopoda. This is the first molecular evidence suggesting a paraphyletic Myriapoda with respect to Chelicerata, even if this finding is not conclusive.
Myriapod Paraphyly and the Arthropod Water-to-Land Transition
Paraphyly of the Myriapoda in the context of the Mandibulata has been repeatedly suggested, mainly using morphological data (Borucki 1996; Kraus 1998; Edgecombe et al. 2000; Giribet and Ribera 2000). A monophyletic Myriapoda as basal clade of the Mandibulata has been recently recovered (Giribet, Edgecombe, and Wheeler 2001) following a combined analysis of morphological and molecular data, and a monophyletic Myriapoda as basal clade of all Arthropoda was obtained (Regier and Shultz 2001) in a tree based on the nuclear gene elongation factor-2. Finally, immunocytochemical and neuroanatomical studies (Loesel, N?ssel, and Strausfeld 2002) provided evidence in favor of a paraphyletic Myriapoda as basal clade within the Arthropoda. Our phylogenetic analyses, in agreement with a variety of molecular and developmental studies (Boore et al. 1995; Friedrich and Tautz 1995; Boore, Lavrov, and Brown 1998; Burke et al. 1998; Cook et al. 2001; Hwang et al. 2001; Dove and Stollewerk 2003), support a close relationship between Myriapoda and Chelicerata and go even further favoring a paraphyletic myriapodan clade, although this finding requires more decisive evidence. Conversely, our results are conclusive in rejecting the monophyly of the mandibulate clade. This has strong implications for the origin of terrestriality within the Arthropoda. According to our results, a water-to-land transition occurred at least three times during the evolution of this phylum (not to mention the woodlice and other more or less strictly terrestrial crustaceans). The more striking circumstance is the independent water-to-land transition of myriapods and hexapods that led to the parallel acquisition of many morphological features whose similarities require new interpretations.
Our analysis could even suggest (fig. 5) two additional independent transitions from water to land, one for the collembolans, represented here by T. bielanensis, if separate from the true insects, and the other for the diplopods, if these are the sister group of centipedes + chelicerates. The phylogenetic support for these two latter events is still open to question. Nevertheless, comparative morphology suggests a few possible synapomorphies of Chilopoda + Chelicerata, including the peculiar feeding mechanism, the segmental position of the excretory organs and a morphological instability frequently observed around mid-trunk.
The vast majority of chelicerates and centipedes, but not the millipedes, feed on fluid food, mostly of animal origin and digested preorally or extraorally (Brusca and Brusca 2002).
As for the excretory system, arachnids have "coxal glands" that mostly open behind legs I and III (Kaestner 1968), whereas centipedes have a "maxillary rein" opening on the second maxillary segment (Rilling 1968). This segmental distribution might represent an instance of positional homology. In fact, despite the different kind of appendages borne on these segments, the third leg pair of arachnids and the second maxillary segment of the centipedes are both regarded as segmentally homologous to the insect labium (respectively, Telford and Thomas 1998; Kraus and Kraus 1994).
Finally, in chelicerates and centipedes, but not in millipedes, one or a few trunk segments around mid-body are either reduced or distinguished by unique morphological markers. In several chelicerates, the sternum of one or more anterior segments of the opisthosoma becomes "subducted" under neighboring sclerites (Shultz 1993). This feature has a segmental equivalent in the "mid-body anomaly" of centipedes (Minelli et al. 2000). In the centipede Lithobius, at least, this segmental anomaly does probably correspond to an early, transient embryonic expression of Abdominal-B (Hughes and Kaufmann 2002). Further comparative studies, morphological, embryological, and molecular, will eventually check the robustness of the phylogenetic relationships suggested by our analysis.
Acknowledgements
We are grateful to Charles E. Cook, Diethard Tautz, Richard Thomas, and two anonymous referees for their insightful comments on a previous version of this paper. E.N. was supported by a postdoctoral fellowship from the University of Padova. We thank the BMR sequencing team at the CRIBI (University of Padova) for major help in the sequencing of the mitochondrial genome of Scutigera coleoptrata. E.N. is also grateful to Dr. Paolo Laveder and Dr. Riccardo Schiavon (University of Padova) for helpful suggestions during the bench work. Research supported by a MIUR grant to G.V. (FIRB 2001—RBNE01F5W7_007).
Literature Cited
Adachi, J., and M. Hasegawa. 1996. Model of amino acid substitution in proteins encoded by mitochondrial DNA. J. Mol. Evol. 42:459-468.
Beard, C. B., D. M. Hamm, and F. H. Collins. 1993. The mitochondrial genome of the mosquito Anopheles gambiae, DNA sequence, genome organization and comparisons with mitochondrial sequences of other insects. Insect Mol. Biol. 2:103-124.
Black, W. C., IV, and R. L. Roehrdanz. 1998. Mitochondrial gene order is not conserved in arthropods: prostriate and metastriate tick mitochondrial genomes. Mol. Biol. Evol. 15:1772-1785.
Boore, J. L. 1999. Animal mitochondrial genomes. Nucleic Acids Res. 27:1767-1780.
Boore, J. L. 2001. Mitochondrial gene arrangement source guide, version 6.0. Department of the Environment Joint Genome Institute, Walnut Creek, Calif. (Available from http://www.jgi.doe.gov/index.html).
Boore, J. L., and W. M. Brown. 1994. Complete DNA sequence of the mitochondrial genome of the black chiton, Katharina tunicata. Genetics 138:423-443.
Boore, J. L., and W. M. Brown. 1995. Complete sequence of the mitochondrial DNA of the annelid worm Lumbricus terrestris. Genetics 141:305-319.
Boore, J. L., T. M. Collins, D. Stanton, L. L. Daehler, and W. M. Brown. 1995. Deducing the pattern of arthropod phylogeny from mitochondrial DNA rearrangements. Nature 376:163-165.
Boore, J. L., D. V. Lavrov, and W. M. Brown. 1998. Gene translocation links insects and crustaceans. Nature 392:667-668.
Borucki, H. 1996. Evolution und phylogenetisches System der Chilopoda (Mandibulata, Tracheata). Verh. Naturwiss. Ver. Hamburg NF 35:95-226.
Briggs, D. E. G., D. H. Erwin, and F. J. Collier. 1994. The fossils of the Burges Shale. Smithsonian Institution Press, Washington.
Brusca, R. C., and G. J. Brusca. 2002. Invertebrates. Second edition. Sinauer Associates, Sunderland, Mass.
Burke, W. D., H. S. Malik, W. C. Lathe, III, and T. H. Eickbush. 1998. Are retrotransposons long-term hitchhikers? Nature 392:141-142.
Castresana, J. 2000. Selection of conserved blocks from multiple alignments for their use in phylogenetic analysis. Mol. Biol. Evol. 17:540-552.
Clary, D. O., and D. R. Wolstenholme. 1985. The mitochondrial DNA molecule of Drosophila yakuba: nucleotide sequence, gene organization, and genetic code. J. Mol. Evol. 22:252-271.
Cook, C. E., M. L. Smith, M. J. Telford, A. Bastianello, and M. Akam. 2001. Hox genes and the phylogeny of the arthropods. Curr. Biol. 11:759-763.
Crease, T. J. 1999. The complete sequence of the mitochondrial genome of Daphnia pulex (Cladocera: Crustacea). Gene 233:89-99.
Crozier, R. H., and Y. C. Crozier. 1993. The mitochondrial genome of the honeybee Apis mellifera: complete sequence and genome organization. Genetics 133:97-117.
Dayhoff, M. O., R. M. Schwartz, and B. C. Orcutt. 1978. A model of evolutionary change in proteins. Pp. 345–352 in M. O. Dayhoff, ed., Atlas of protein sequence structure, vol. 5, National Biomedical Research Foundation, Washington, D.C.
Dotson, E. M., and C. B. Beard. 2001. Sequence and organization of the mitochondrial genome of the Chagas disease vector, Triatoma dimidiata. Insect Mol. Biol. 10:205-215.
Dove, H., and A. Stollewerk. 2003. Comparative analysis of neurogenesis in the myriapod Glomeris marginata (Diplopoda) suggests more similarities to chelicerates than to insects. Development 130:2161-2171.
Edgecombe, G. D., and G. Giribet. 2002. Myriapod phylogeny and the relationships of Chilopoda. Pp. 143–168 in J. J. Llorente Bousquets and J. J. Morrone, eds., Biodiversidad, taxonomía y biogeografia de artrópodos de México: hacia una síntesis de su conocimiento, vol. 3, Prensas de Ciencias, Universidad Nacional Autónoma de México.
Edgecombe, G. D., G. D. F. Wilson, D. J. Colgan, M. R. Gray, and G. Cassis. 2000. Arthropod cladistics: combined analysis of histone H3 and U2 sn RNA sequences and morphology. Cladistics 16:204-231.
Felsenstein, J. 1985. Confidence limits on phylogenies: an approach using bootstrap. Evolution 39:783-791.
Felsenstein, J. 2002. PHYLIP (Phylogeny Inference Package) version 3.6a3 (Univ. of Washington, Seattle).
Flook, P. K., C. H. F. Rowell, and G. Gellissen. 1995. The sequence, organization, and evolution of the Locusta migratoria mitochondrial genome. J. Mol. Evol. 41:928-941.
Foster, P. G., and D. A. Hickey. 1999. Compositional bias may affect both DNA-based and protein-based phylogenetic reconstructions. J. Mol. Evol. 48:284-290.
Friedrich, M., and D. Tautz. 1995. Ribosomal DNA phylogeny of the major extant arthropod classes and the evolution of myriapods. Nature 376:165-167.
Giribet, G., G. D. Edgecombe, and W. C. Wheeler. 2001. Arthropod phylogeny based on eight molecular loci and morphology. Nature 413:157-161.
Giribet, G., and C. Ribera. 2000. A review of arthropod phylogeny: new data based on ribosomal DNA sequences and direct character optimization. Cladistics 16:204-231.
Hickerson, M. J., and C. W. Cunningham. 2000. Dramatic mitochondrial gene rearrangements in the hermit crab Pagurus longicarpus (Crustacea, Anomura). Mol. Biol. Evol. 17:639-644.
Huelsenbeck, J. P., and F. Ronquist. 2001. MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics 17:754-755.
Huelsenbeck, J. P., F. Ronquist, R. Nielsen, and J. P. Bollback. 2001. Bayesian inference of phylogeny and its impact on evolutionary biology. Science 294:2310-2314.
Hughes, C. L., and T. C. Kaufman. 2002. Exploring the myriapod body plan: expression patterns of the ten hox genes in a centipede. Development 129:1225-1238.
Hwang, U. W., M. Friedrich, D. Tautz, C. J. Park, and W. Kim. 2001. Mitochondrial protein phylogeny joins myriapods with chelicerates. Nature 413:154-157.
Jones, D. T., W. R. Taylor, and J. M. Thornton. 1992. The rapid generation of mutation data matrices from protein sequences. Comput. Appl. Biosci. 8:275-282.
Kaestner, A. 1968. Invertebrate zoology, 2. Wiley-Interscience, New York.
Kishino, H., and M. Hasegawa. 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J. Mol. Evol. 29:170-179.
Kraus, O. 1998. Phylogenetic relationships between higher taxa of tracheate arthropods. Pp. 295–303 in R. A. Fortey, and R. H. Thomas, eds., Arthropod relationships, Chapman & Hall, London.
Kraus, O., and M. Kraus. 1994. Phylogenetic system of the Tracheata (Mandibulata): on "Myriapoda"—Insecta interrelationships, phylogenetic age and primary ecological niches. Verh. naturwiss. Ver. Hamburg N.F. 34:5-31.
Lavrov, D. V., J. L. Boore, and W. M. Brown. 2000. The complete mitochondrial DNA sequence of the horseshoe crab Limulus polyphemus. Mol. Biol. Evol. 17:813-824.
Lavrov, D. V., J. L. Boore, and W. M. Brown. 2002. Complete mtDNA sequences of two millipedes suggest a new model for mitochondrial gene rearrangements: duplication and nonrandom loss. Mol. Biol. Evol. 19:163-169.
Lavrov, D. V., W. M. Brown, and J. L. Boore. 2000. A novel type of RNA editing occurs in the mitochondrial tRNAs of the centipede Lithobius forficatus. Proc. Natl. Acad. Sci. USA 97:13738-13742.
Loesel, R., D. R. N?ssel, and N. J. Strausfeld. 2002. Common design in a unique midline neuropil in the brains of arthropods. Arthropod Struct. Dev. 31:77-91.
Minelli, A., D. Foddai, L. A. Pereira, and J. G. E. Lewis. 2000. The evolution of segmentation of centipede trunk and appendages. J. Zool. Syst. Evol. Res. 38:103-117.
Nardi, F., A. Carapelli, P. P. Fanciulli, R. Dallai, and F. Frati. 2001. The complete mitochondrial DNA sequence of the basal hexapod Tetrodontophora bielanensis: evidence for heteroplasmy and tRNA translocations. Mol. Biol. Evol. 18:1293-1304.
Nardi, F., G. Spinsanti, J. L. Boore, A. Carapelli, R. Dallai, and F. Frati. 2003. Hexapod origins: monophyletic or paraphyletic? Science 299:1887-1889.
Negrisolo, E., A. Minelli, and G. Valle. 2004. Extensive gene order rearrangement in the mitochondrial genome of the centipede Scutigera coleoptrata. J. Mol. Evol. 58 (in press).
Nei, S., and K. Kumar. 2000. Molecular evolution and phylogenetics. Oxford University Press, New York.
Posada, D., and K. A. Crandall. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14:817-818.
Regier, J. C., and J. W. Shultz. 2001. Elongation factor-2: useful gene for arthropod phylogenetics. Mol. Phylogenet. Evol. 20:136-148.
Rilling, G. 1968. Lithobius forficatus L., in Grosses zoologisches Praktikum. Fischer, Stuttgart, 13B:1-136.
Roehrdanz, R. L., M. E. Degrugillier, and W. C. Black, IV. 2002. Novel rearrangements of arthropod mitochondrial DNA detected with long-PCR: applications to arthropod phylogeny and evolution. Mol. Biol. Evol. 19:841-849.
Ronquist, F., and J. P. Huelsenbeck. 2003. MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19:1572-1574.
Schmidt H. A., K. Strimmer, M. Vingron, and A. von Haeseler. 2002. Tree-Puzzle: maximum likelihood phylogenetic analysis using quartets and parallel computing. Bioinformatics 18:502-504.
Shao, R., N. J. H. Campbell, and S. C. Barker. 2001. Numerous gene rearrangements in the mitochondrial genome of the wallaby louse, Heterodoxus macropus (Phthiraptera). Mol. Biol. Evol. 18:858-865.
Shear, W. A., A. J. Jeram, and P. A. Selden. 1998. Centipede legs (Arthropoda, Chilopoda, Scutigeromorpha) from the Silurian and Devonian of Britain and the Devonian of North America. Am. Mus. Novitates. 3231:1-16.
Shimodaira, H. 2002. An approximately unbiased test of phylogenetic tree selection. Syst. Biol. 51:492-508.
Shimodaira, H., and M. Hasegawa. 2001. CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinformatics 17:1246-1247.
Shultz, J. M. 1993. Muscular anatomy of the giant whipscorpion Mastigoproctus giganteus (Lucas) (Arachnida: Uropygi) and its evolutionary significance. Zool. J. Linn. Soc. 108:335-365.
Spanos, L., G. Koutroumbas, M. Kostyfakis, and C. Louis. 2000. The mitochondrial genome of the Mediterranean fruit fly, Ceratitis capitata. Insect Mol. Biol. 9:139-144.
Staton, J. L., L. L. Daehler, and W. M. Brown. 1997. Mitochondrial gene arrangement of the horseshoe crab Limulus polyphemus L.: conservation of major features among arthropodan classes. Mol. Biol. Evol. 14:867-874.
Strimmer, K., and A. von Haeseler. 1996. Quartet puzzling: a quartet maximum likelihood method for reconstructing tree topologies. Mol. Biol. Evol. 13:964-969.
Strimmer, K., and A. von Haeseler. 1997. Likelihood-mapping: a simple method to visualize phylogenetic content of a sequence alignment. Proc. Natl. Acad. Sci. USA 94:6815-6819.
Swofford, D. L. 2002. PAUP*. Phylogenetic analysis using parsimony (*and other methods). Version 4.10. Sinauer Associates, Sunderland, Mass.
Telford, M. J., and R. H. Thomas. 1998. Expression of homeobox genes shows chelicerate arthropods retain their deutocerebral segment. Proc. Natl. Acad. Sci. USA 95:10671-10675.
Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. ClustalW: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-4680.
Valverde, J. R., B. Batuecas, C. Moratilla, R. Marco, and R. Garesse. 1994. The complete mitochondrial DNA sequence of the crustacean Artemia franciscana. J. Mol. Evol. 39:400-408.
Whelan, S., P. Liò, and N. Goldman. 2001. Molecular phylogenetics: state-of-the art methods for looking into the past. Trends Genet. 17:262-272.
Wilson, K., V. Cahill, E. Ballment, and J. Benzie. 2000. The complete sequence of the mitochondrial genome of the crustacean Penaeus monodon: are malacostracan crustaceans more closely related to insects than to branchiopods? Mol. Biol. Evol. 17:863-874.(Enrico Negrisolo, Alessan)