The Phylogenetic Positions of Three Basal-Hexapod Groups (Protura, Diplura, and Collembola) Based on Ribosomal RNA Gene Sequences
http://www.100md.com
分子生物学进展 2005年第7期
* Institute of Plant Physiology & Ecology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, People's Republic of China; and School of Biological Sciences, Washington State University
Correspondence: E-mail: yxluan@sibs.ac.cn.
Abstract
This study combined complete 18S with partial 28S ribosomal RNA gene sequences (2,000 nt in total) to investigate the relations of basal hexapods. Ten species of Protura, 12 of Diplura, and 10 of Collembola (representing all subgroups of these three clades) were sequenced, along with 5 true insects and 8 other arthropods, which served as out-groups. Trees were constructed with maximum parsimony, maximum likelihood, Bayesian analysis, and minimum-evolution analysis of LogDet-transformed distances. All methods yielded strong support for a clade of Protura plus Diplura, here named Nonoculata, and for monophyly of the Diplura. Parametric-bootstrapping analysis showed our data to be inconsistent with previous hypotheses (P < 0.01) that joined Protura with Collembola (Ellipura), that said Diplura are sister to true insects or are diphyletic, and that said Collembola are not hexapods. That is, our data are consistent with hexapod monophyly and Collembola grouped weakly with "Protura + Diplura" under most analytical conditions. As a caveat to the above conclusions, the sequences showed nonstationarity of nucleotide frequencies across taxa, so the CG-rich sequences of the diplurans and proturans may have grouped together artifactually; however, the fact that the LogDet method supported this group lessens this possibility. Within the basal hexapod groups, where nucleotide frequencies were stationary, traditional taxonomic subgroups generally were recovered: i.e., within Protura, the Eosentomata and Acerentomata (but Sinentomata was not monophyletic); within Collembola, the Arthropleona, Poduromorpha, and Entomobryomorpha (but Symphypleona was polyphyletic); and in Diplura, the most complete data set (>2,100 nt) showed monophyly of Campodeoidea and of Japygoidea, and most methods united Projapygoidea with Japygoidea.
Key Words: Protura ? Diplura ? Collembola ? ribosomal RNA genes ? molecular phylogeny
Introduction
Hexapoda (Insecta sensu lato) includes four groups: Protura, Collembola, Diplura, and Insecta sensu stricto (with the latter containing Microcoryphia or Archeognatha, Zygentoma, and the pterygote insects). Protura, Collembola, and Diplura are ancient groups, known as fossils from the Devonian, Carboniferous, and Cretaceous periods up to 400 MYA (Delamare and Massoud 1968; Kukalová-Peck 1987, 1991; Whalley 1995). Based on their mouthparts, Hennig (1953) joined Protura, Collembola, and Diplura into Entognatha (with enclosed mouthparts), in contrast to Ectognatha (Insecta sensu stricto, with exposed mouthparts). All hexapods have a body divided into three basic tagmata—head, thorax, and abdomen—and three pairs of thoracic legs (R. C. Brusca and G. J. Brusca 2003). The three groups of entognath hexapods share (with many insect groups) the superficial similarities of small body size and a preference for living in soil, leaf litter, or under rocks. However, controversies about the interrelationships of these entognath groups have raged in recent years, with the monophyly of Diplura and the relative phylogenetic positions of Diplura, Protura, and Collembola being hot spots of dispute (Giribet et al. 2004).
Proturans are tiny animals with a conical head and a relatively long abdomen. They have no eyes, antennae, or caudal cerci but have a telson tail, which is common in crustaceans but absent in other hexapods. Protura includes three subgroups: Acerentomata, Eosentomata, and Sinentomata (Yin 1996), but within Sinentomata, the phylogenetic positions of the families Sinentomidae and Fujientomidae are controversial due to their special morphological characteristics. Unusual kinds of spermatozoa exist both within and between proturan species, which might reflect their long evolutionary course (Yin and Xue 1993).
Diplurans have a long, thin body that appears to have two paired tails. Actually, these "tails" are two long antennae projecting forward from the eyeless head and two cerci projecting back from the abdomen. Diplura is composed of Campodeoidea, Japygoidea, and Projapygoidea. Despite morphological features shared by these subgroups, the monophyly of Diplura has been questioned because Campodeoidea and Japygoidea differ in the type of cerci (a filiform vs. a grasping pincer), sperm morphology, and ovarian structure (tys and Bilinski 1990; Bilinski 1994). In addition, the phylogenetic position of the entire group is controversial: Does Diplura actually group with the true insects instead of with the other entognaths (Kukalová-Peck 1987)? What are the relationships between Diplura and the other "orders" of hexapods? Also, few studies have included Projapygoidea, whose phylogenetic position within Diplura is still a puzzle.
Collembola (springtails) are the most diverse of the entognath hexapods and are widely distributed in the world. They are defined by a characteristic forked "tail" (furcula), which enables them to spring, hence their English name. Collembola comprise two suborders: Arthropleona and Symphypleona. Agreement over this taxonomic scheme is not universal, however, and the phylogenetic position of Podura aquatica, an important species that has been considered as one of the most primitive collembolans, is also unclear (D'Haese 2002).
Several major hypotheses of basal hexapod relationships have been proposed based on morphological and molecular characters (fig. 1). Most of these hypotheses include a clade uniting Protura and Collembola, called Ellipura, based on a distinctive type of enclosed mouthparts and on the loss or absence of several characters (e.g., cerci) present in other hexapods (B?rner 1910; Hennig 1981; Kristensen 1981, 1991; Kukalová-Peck 1987, 1991; Koch 1997; Wheeler et al. 2001). The first hypothesis (fig. 1A), that of Hennig (1981), divides hexapods into Ectognatha and Entognatha, with the latter consisting of Ellipura and Diplura. The second hypothesis (Kukalová-Peck 1987) places Diplura with Ectognatha (fig. 1B) based on a Carboniferous fossil that was interpreted as japygid-like but had eyes and more exposed mouthparts, as in ectognaths. The next hypothesis (fig. 1C) proposed that Diplura is diphyletic (tys and Bilinski 1990). Based on the structure of the ovarioles, "Japygina" (=Japygidae) was placed basal to "Campodeina" (=Campodeidae) + Ellipura in Entognatha. The final hypothesis (fig. 1D) is from the molecular-phylogenetic study of Nardi et al. (2003), in which mitochondrial gene sequences suggested that Collembola are basal to a clade consisting of crustaceans and insects. According to this hypothesis, hexapods are not monophyletic, although some have disagreed with this suggestion (Delsuc, Phillips, and Penny 2003; Cameron et al. 2004).
FIG. 1.— Four hypotheses of basal hexapod relationships, which are evaluated in the present study.
Many studies have begun to explore the relations among arthropod groups using various molecular markers. These markers include 18S and 28S ribosomal RNA (rRNA) genes (Turbeville et al. 1991; Wheeler, Cartwright, and Hayashi 1993; Friedrich and Tautz 1995, 1997; Giribet and Ribera 1998, 2000; Mallatt, Garey, and Shultz 2004), which are thought to be especially appropriate for resolving higher level phylogenetic relationships (Hillis and Dixon 1991; Mallatt and Winchell 2002). Other studies used mitochondrial gene sequences and mitochondrial gene order (Ballard et al. 1992; Boore et al. 1995; Boore, Lavrov, and Brown, 1998), histone H3 and U2 small nuclear RNA (Colgan et al. 1998), Elongation Factor (EF) 1- and RNA Polymerase II (Regier and Shultz 1997; Shultz and Regier 2000), and EF-2 (Regier and Shultz 2001; Regier, Shultz, and Kambic 2004). Some of the studies are especially interesting in that they combine several sources of molecular data plus morphological characters (i.e., Zrzav et al. 1998; Giribet, Edgecombe, and Wheeler 2001; Wheeler et al. 2001). Others have started to focus on the relations among the basal hexapods (for example, Shao et al. 1999, 2000), and the most detailed of these studies have supported dipluran monophyly (Carapelli et al. 2000; Giribet and Ribera 2000; Luan et al. 2003, 2004; Regier, Shultz, and Kambic 2004; but not Shultz and Regier 2000), and in addition, some have suggested a new possibility: that Protura unites with Diplura (Giribet, Edgecombe, and Wheeler 2001; Giribet et al. 2004; Kjer 2004).
Here, we use complete 18S and partial 28S rRNA sequences to expand on these studies. We have obtained the most extensive sample of basal hexapods so far: namely, 10 proturans, 12 diplurans, and 10 collembolans, including all their major subgroups—plus a wide range of out-group taxa. Additionally, whereas most past studies used only parsimony-based methods of phylogenetic reconstruction, we add model-based and likelihood methods and use parametric bootstrapping to test the hypotheses in figure 1 in a statistical manner.
Materials and Methods
Most proturan, dipluran, and collembolan specimens were collected in 75% ethanol by the Tullgren funnel method. All were stored in ethanol in a freezer or refrigerator after identification. More details are given in table 1, which lists the species used. The choice of out-group taxa for the study was complicated by controversy over whether the closest relatives of hexapods are crustaceans or myriapods (Mallatt, Garey, and Shultz 2004). To overcome this problem, we used numerous representatives of each: four crustaceans, three myriapods, and even a chelicerate. As other out-group taxa, five major groups of true insects were also represented, bringing the total number of species used in this study to 45. Most of the out-group sequences were obtained from GenBank (see table 1).
Table 1 Information on Species Used in This Study
Genomic DNA was extracted from one individual of most species using the single-fly extraction method (Gloor et al. 1993) or from several individuals of some species (Campodeidae sp., Triacanthella, Sminthurus, Dilta) with the DNeasy Tissue Kit (Qiagen Inc., Valencia, Calif.). The 18S rRNA gene, ranging in length from 1,750 nt in collembolans to 2,200 nt in some diplurans, generally was amplified in four overlapping fragments of 600 nt each, using primer pairs 18S1L-18S1R, 18SL500-18SR1470, 18SL1210-18SR1790, and 18S3L-18S3R, respectively (table 2). The 28S rRNA fragment, 430 nt long and located in the 5' third of the gene from beyond the D3 to beyond the D5 divergent domains, was amplified using primer pair 28Sf and 28Sr. For Campodeidae sp., Triacanthella, Sminthurus, and Dilta, the complete 18S and 28S genes were each amplified in one or two pieces (Mallatt and Sullivan 1998). For the polymerase chain reaction–amplification and sequencing conditions see Luan et al. (2004) and Mallatt and Sullivan (1998). It should be noted that neither the 18S nor the 28S sequence was missing from any of the 45 taxa.
Table 2 Primers Used for Amplification and Sequencing of 18S rDNA and 28S rDNA
As in previous studies (Mallatt and Sullivan 1998; Winchell et al. 2002; Mallatt, Garey, and Shultz 2004), the genes of all taxa were aligned by eye in the GCG Seqlab program (Wisconsin Package Version 10.3; Accelrys Inc., San Diego, Calif.) against a reference set consisting of rRNA sequences from many other metazoans and arthropods, all rigidly aligned based on the rRNA secondary-structure models of Xenopus laevis and Strongylocentrotus purpuratus (Gutell 1994; Schnare et al. 1996) and confirmed by reference to models of yeast, Saccharomyces cerevisiae, rRNA (see http://www.psb.agent.be/rRNA/secmodel/Scer_SSU.html for yeast 18S and Gutell, Gray, and Schnare 1993 for yeast 28S). Thus, these reference models should be valid across eukaryotes. The secondary structure of rRNA genes is known for the insect Drosophila (Hancock, Tautz, and Dover 1988), and the Drosophila sequences are among those in our reference-alignment set; however, they were not used as our primary reference because Drosophila rRNA sequences are too divergent to produce suitable alignments.
About 10% of the sites in our data set were unalignable and were excluded from the analysis. More specifically, we excluded sites in which more than 30% of the taxa do not share the same nucleotide (Hillis and Dixon 1991), and this is made simpler by the fact that rRNA genes have long, conserved segments (in which almost all the sites are alignable) interrupted by about a dozen variable regions (in which many sites are obviously unalignable); therefore, most decisions about exclusion involve only short segments of just 5–10 nt in the transition zones between conserved and variable regions. The sites we excluded were almost entirely in the V2, V4, and V7–9 variable regions of 18S and in the D4 and D5 divergent domains of 28S. The remaining alignments contained 1,926 sites across the 45 taxa. Our alignments are available at http://www.wsu.edu/jmallatt/alignments.html (posted December 17, 2004) and from European Molecular Biology Laboratory (ftp://ftp.ebi.ac.uk/pub/databases/embl/align) under the accession numbers ALIGN_000854–000861.
For the phylogenetic analyses, the 18S and partial 28S sequences were combined because, as parts of the same gene family, the 18S and 28S genes of animals routinely yield similar trees (Mallatt and Winchell 2002; Winchell et al. 2002; Mallatt, Garey, and Shultz 2004). As in previous studies, phylogenetic trees relating the 45 taxa were estimated with several different algorithms, all executed in PAUP* 4.0 beta 10 (Swofford 2002): equally weighted maximum parsimony (MP), minimum evolution using LogDet-Paralinear distances (LogDet), and maximum likelihood (ML) in which the general time reversible (GRT) +I + model was found to fit our combined rRNA data best (by the Akaike information criterion (AIC) approach to model evaluation in Modeltest; Posada and Crandall 1998). For more information on these methods, see Mallatt and Winchell (2002), Mallatt, Garey, and Shultz (2004), and Posada and Buckley (2004).
For the LogDet and MP methods, support for clades was evaluated with ordinary, nonparametric bootstrapping (1,000 replicates), but for ML, due to high computational demands, only the single best tree could be calculated; however, ML nonparametric bootstrapping with 100 replicates was performed on the trimmed, 34-taxon set introduced below.
Likelihood-based Bayesian inference (Markov Chain Monte Carlo analysis; Huelsenbeck and Ronquist 2001) was also used on the 45-taxon set, as described in Winchell, Martin, and Mallatt (2004), and Mallatt, Garey, and Shultz (2004), using MrBayes 3.01. Briefly, we used a single, invgamma model, for which the number of substitution types (nst) was set at six, four Markov chains were run for 106 generations, and posterior probabilities were calculated from these trees, after the first 200,000 were discarded as burn-in. In addition to this single-model test, we ran a partitioned Bayesian analysis, after using AIC in Modeltest to find the best models for the 18S and 28S partitions; these models were "nst = 6 invgamma" for both the partitions.
The four hypotheses in figure 1, plus a general hypothesis that hexapods are monophyletic, were tested with our data using ML-based parametric bootstrapping (Efron 1985; Huelsenbeck, Hillis, and Jones 1996). To accommodate the computational demands of this method, it was necessary to trim the 45 taxa down to 34, by removing some of the out-groups as well as those in-group taxa that were most closely related to others (in the same genus, for example). Care was taken to assure that the retained 34 sequences yielded the same basic topology as the 45-taxon tree (compare figs. 2 and 3). After ML was used to find optimal constrained and unconstrained trees, the parametric-bootstrap tests were performed, with 100 simulations per hypothesis tested, as described in Mallatt, Garey, and Shultz (2004). The necessary simulations were done with the Seq-Gen version 1.2.5 program (Rambaut and Grassly 1997).
FIG. 2.— Phylogenetic tree based on the entire data set: 45 taxa; 18S + partial 28S genes; 1,926 characters. The basic tree, with its branch lengths, was calculated by ML (–Ln L = 14,707.69). The four values at each node are, respectively, (1) Bayesian posterior probability based on a partitioned analysis with separate models for 18S and 28S partitions (1.5 x 106 generations), (2) Bayesian posterior probability based on a single model for the combined 18S + 28S data (106 generations), (3) LogDet-bootstrap value (1,000 bootstrap replicates), and (4) MP-bootstrap value (1,000 bootstrap replicates).
FIG. 3.— ML tree based on an abbreviated data set; same set as in figure 2 but with just 34 taxa (–Ln L = 13,050.0651). ML-bootstrap values (100 replicates) are indicated at the nodes.
Finally, we further examined the relationships of taxa within each basal hexapod group: in Protura, then in Diplura, then in Collembola. Because these analyses considered only within-group relations, there was no need to include an out-group, so a larger number of alignable characters could be recognized and included in the analysis—expanding each of the character sets from 1,926 to >2,100 sites and, presumably, increasing the amount of phylogenetic signal. The relations within each group were analyzed with the ML, LogDet, and MP tree-building algorithms.
As in previous studies (Mallatt, Garey, and Shultz 2004; Winchell, Martin, and Mallatt 2004), we accepted clades in the Bayesian tree at 95% posterior probability, while accepting values around 60%–70% as significant in the nonparametric-bootstrap trees. For parametric bootstrapping, rejection was at P 0.05.
Results
Nucleotide Composition
When the 2 test of stationarity of nucleotide frequencies in PAUP* was applied to our aligned data set, the frequencies were found to be highly nonstationary across the 45 taxa (2 = 648.1; P = 0.00000000) (table 3), and the same was found for the abbreviated data set of 34 taxa (not shown). This extreme nonstationarity must be due to the high proportion of C and G nucleotides in the diplurans because when all diplurans were removed, the remaining sequences were stationary (P = 0.43). Then, however, the proturan sequences were more CG rich than any of the other remaining taxa. When nucleotide frequencies were tested within each of the three basal hexapod groups (i.e., within Protura, within Diplura, within Collembola), they were always stationary (P = 0.98 for each group).
Table 3 Chi-square Test of Stationarity of Base Frequencies in the 18S + 28S Sequences Used in This Study
Trees
The calculated phylogenetic trees are shown in figures 2–4. Similar results were obtained by all methods: Bayesian inference, LogDet, MP, and ML. Protura, Diplura, and Collembola were each monophyletic with 100% statistical support (figs. 2 and 3). Protura grouped strongly with Diplura, again with universal 100% support. The position of collembolans was not clearly resolved, although the Bayesian, ML, and LogDet methods joined them with "Protura + Diplura" with weak support. Within Protura, the subgroups Eosentomata and Acerentomata were each monophyletic (and were joined together by the likelihood-based methods), but Sinentomata was not monophyletic as its two genera, Sinentomon and Fujientomon, did not group together (figs. 2 and 4A). In Diplura, a monophyletic Campodeoidea was evident with the expanded sequences (fig. 4B) but not with the shorter sequences of figure 2. Japygoidea was universally monophyletic, and the projapygid Octostigma was grouped with Japygoidea by all methods except LogDet (figs. 2 and 3). Within Collembola, Arthropleona was monophyletic but Symphypleona was not, being represented by a grade of taxa at the base of the collembolans. Entomobrya and Isotoma, which are sometimes placed together in the order Entomobryomorpha (D'Haese 2002), formed a monophyletic clade of entomobryomorphs separate from the rest of the Arthropleona (the Poduromorpha) (figs. 2 and 4C).
FIG. 4.— ML trees, presented as star diagrams, for the three basal-hexapod groups. Based on expanded alignments of >2,100 characters, which should reveal the relationships within these groups better than in figure 2, which was based on fewer characters. The "veils" of radiating lines are artistic devices to separate major subgroups of taxa. Bootstrap values are indicated at the nodes: ML (100 replicates), MP (1,000 replicates), and LogDet (LD) (1,000 replicates). The thick arrows show where these trees can be rooted, as determined from figure 2. (A) Ten proturans, best ML tree with bootstrap values: ML/MP/LD, 2,237 nt, –Ln L= 7,055.3501, 2 = 0.98; (B) 12 diplurans, best ML tree with bootstrap values: ML/MP/LD, 2,211 nt, –Ln L= 4,839.8029, 2=0.98; (C) 10 collembolans, best ML tree with bootstrap values: ML/MP/LD, –Ln L= 5,736.3947, 2,169 nt, 2 = 0.98.
Among the out-groups, the following groupings are evident and received some support, at least from the likelihood-based techniques (figs. 2 and 3): crustaceans plus hexapods (Pancrustacea), millipedes, millipedes + centipede Lithobius, and pterygote insects. However, every method placed the Artemia sequence (a crustacean) with some hexapod, apparently violating hexapod monophyly.
Hypothesis Testing
Parametric-bootstrap analysis of the rRNA sequences from the 34-taxon subset (taxa shown in fig. 3) was used to test the four alternate hypotheses of hexapod relationships shown in figure 1 as well as the hypothesis that hexapods are monophyletic. Results are presented in table 4. All these hypotheses were rejected by our data, except for hexapod monophyly (Hypothesis 5).
Table 4 Results of Hypothesis Testing by Parametric Bootstrapping
Discussion
General
The main findings of this study, which used the largest taxon sample to date to explore the relationships of basal hexapods, are that Protura and Diplura are sister taxa and that diplurans are monophyletic—thereby upholding the conclusions of some other recent molecular and morphological studies (Giribet, Edgecombe, and Wheeler 2001; Giribet et al. 2004; Kjer 2004). Bootstrap-support values for these clades were high and were consistent across the four tree-building methods (figs. 2–4), suggesting that the rRNA genes contain a large amount of signal. All alternate hypotheses that unite proturans with collembolans as Ellipura were rejected by parametric bootstrapping (table 4), as was the hypothesis of Nardi et al. (2003) that collembolans are not hexapods (for more discussion of this latter point, see C. Bitsch and J. Bitsch 2000, 2004; Delsuc, Phillips, and Penny 2003; and Cameron et al. 2004).
A potential problem with these conclusions must be pointed out. Nonstationarity of nucleotide frequencies across taxa can mislead both likelihood-based and parsimony-based methods (Jermin et al. 2004), and our nucleotide frequencies were nonstationary (table 3). Thus, because both the dipluran and proturan sequences are CG rich, these two sequences might have been united artifactually by homoplasy; that is, they may actually be unrelated but have independently evolved C's and G's at the same sites (Swofford et al. 1996). The LogDet method is designed to avoid this particular "long-branch attraction" artifact (Lake 1994; Lockhart et al. 1994), so our finding that LogDet grouped diplurans with proturans adds credibility to the "Protura + Diplura" clade. Also, a recent simulation study (Jermin et al. 2004) indicated that this attraction artifact is most likely to occur where stem branches are short—but the stem branch leading to "Protura + Diplura" in our study appears rather long (fig. 2), increasing the chance that this clade is valid after all. Nonetheless, caution must be exercised because the true length of that stem branch is unknown given our CG–biased data, and LogDet is not always able to solve the nonstationarity problem (Foster and Hickey 1999). Furthermore, the LogDet minimum-evolution method employed in PAUP* is weak at modeling among-site rate variation within genes—a weakness that itself can cause long-branch attraction artifacts in rRNA-based analyses of arthropod phylogeny (Mallatt, Garey, and Shultz 2004). Therefore, our support for a clade of proturans plus diplurans is not absolutely solid. On the other hand, the fact that nucleotide composition is stationary within Diplura (P = 0.98) means that the other major conclusion of this study—that Japygoidea, Projapygoidea, and Campodeoidea form a monophyletic Diplura—is solid.
In the 18S sequences of our basal hexapods, the differences between individuals of the same species (from different geographical locations) are no more than 10 nt and are far smaller than those of any comparison between the different species of a genus. This supports the morphological identification of species and suggests conservation of the 18S rRNA gene at the species level. Therefore, rRNA genes may be helpful for identifying some odd species.
Specific Groups
Out-group Taxa and Hexapod Monophyly
Most of the relations we obtained among the out-groups are reasonable, based on past morphological and molecular studies (for example, monophyly of Pancrustacea, of Insecta, and of Myriapoda; Giribet, Edgecombe, and Wheeler 2001; Ruppert, Fox, and Barnes 2004; Regier, Wilson, and Shultz 2005). However, the fact that our trees show the branchiopod crustacean, Artemia, in Hexapoda (figs. 2 and 3) should not be interpreted as evidence for a paraphyletic Hexapoda for several reasons. First, it is notoriously difficult to obtain reasonable relationships between the hexapods and the groups of crustaceans using just 18S (and partial 28S) genes (Giribet and Ribera 2000; Wilson et al. 2000; Giribet et al. 2005). Second, the support for "Artemia plus insects" in our trees was below the level of significance for most of the tree-building methods, and parametric bootstrapping showed our data to be fully consistent with hexapod monophyly (table 4). Finally, an expanded data set, which includes nearly complete 28S and 18S sequences from more taxa, supports branchiopods as distinct from a monophyletic Hexapoda (J. Mallatt, unpublished data; also see Mallatt, Garey, and Shultz 2004). Thus, the placement of Artemia in Hexapoda in this study seems to be an artifact of insufficient signal in the 18S gene.
Protura with Diplura
If Protura and Diplura form a clade, as our molecular results suggest (also see Luan et al. 2003, 2004), then an attempt should be made to identify some shared morphological characters of this clade and the clade should be named. Few shared characters of proturans and diplurans have been proposed in the literature, but we have tentatively identified the following: (1) Protura, Campodeoidea, and some Projapygoidea have malpighian tubules only as short papillae (in other projapygids and in japygids, the tubules are absent) (Dallai 1976; Boudreaux 1979; Hennig 1981; Dallai and Burroni 1982; Ferguson 1990) and (2) the lack of eyes, meaning neither simple nor compound eyes are present (Tuxen 1964; Boudreaux 1979; Hennig 1981). We name the clade "Nonoculata" ("no eyes") after this second character.
Subgroups recovered within Protura, Diplura, and Collembola will be considered next. Because the nucleotide frequencies within each of these groups were stationary, there is no chance of bias due to convergent evolution of nucleotide composition.
Protura
There are three taxonomic groups in Protura: Acerentomata, Eosentomata, and Sinentomata (Yin 1996). These possess three types of pseudoculus, respectively, the ophiso-neuroforamen, mid-neuroforamen, and multi-neuroforamen types (Yin et al. 2002). Additionally, Eosentomata possess meso- and metathoracic spiracles with a primitive tracheal system, while members of Acerentomata lack these structures. Within Sinentomata, species of Fujientomidae lack a tracheal system, but species of Sinentomidae possess one (though this system differs obviously from that of Eosentomata). Our rRNA-based phylogenetic analyses of proturan species supported the monophyly of Acerentomata and of Eosentomata, but Sinentomata (represented by one species of Sinentomidae and one of Fujientomidae) was not monophyletic (figs 2, 3, and 4A). Traditionally, Fujientomidae was considered the primitive group among Acerentomata (Yin 1984). However, its spermatozoon and pseudoculus are patterned as in Sinentomidae, so Yin and Xue (1993) proposed the clade Sinentomata. Although not supported by high bootstrap values, most of our trees show Fujientomon as basal, the sister group to "Sinentomon plus the other proturans."
Diplura
As mentioned, controversy surrounds the question of whether Diplura is monophyletic or diphyletic. Most morphological studies have suggested that Diplura is monophyletic. The anatomical characters shared by Japygoidea and Campodeoidea include the special structure of the labium and of the oral folds, the interlocking between the superlingua and the maxillary galea (Koch 1997), unique muscles and pivots in the legs (Manton 1972, 1977), the absence of eyes (C. Bitsch and J. Bitsch 2000, 2004), and embryonic features of the head (Ikeda and Machida 1998). However, evidence from the ultrastructure of germ cells suggested that the sperm of Campodeidae resembled that of Insecta sensu stricto, yet their ovarian structure was similar to that of Collembola and differed from that of Insecta sensu stricto. In contrast, Japygidae was similar to Collembola in sperm ultrastructure but resembled Insecta sensu stricto in ovarian structure. Based on these discrepancies and emphasizing ovarian structure, tys and Bilinski (1990) suggested that "Diplura" is a paraphyletic group. Our rRNA-based findings do not support this. In fact, tys and Bilinski's hypothesis that separated japygids from campodeids was the most strongly rejected of all hypotheses tested in this study (table 4).
Projapygids are an interesting and controversial group of Diplura. Specimens of Projapygoidea are so difficult to find that only a few studies have included them (Rusek 1982; C. Bitsch and J. Bitsch 2000, 2004; Luan et al. 2004), and their phylogenetic position is controversial because they harbor many morphologically intermediate characters between Campodeoidea and Japygoidea (Rusek 1982). Different morphological studies have concluded that they are basal diplurans (Rusek 1982) or that they group with Campodeoidea (Pagés 1997) or with Japygoidea (tys and Bilinski 1990; tys, Zrzav, and Weyda 1993). In the present study, all methods except LogDet united our projapygid Octostigma with Japygoidea. Another, non-sequence–based characteristic shared by Projapygoidea and Japygoidea is that their 18S genes were longer than in any of the species of Campodeoidea by more than 300 bp.
Collembola
Collembola is more diverse than either proturans or diplurans, so our sampling within its major subgroups is less complete. The two traditional subgroups are Arthropleona and Symphypleona: in Arthropleona, the thoracic and abdominal segments are easily separable, whereas Symphypleona generally have a globular body with the thoracic and first four abdominal segments fused. Our trees support monophyly of the Arthropleona (figs. 2, 3, and 4C). Within Arthropleona, Poduromorpha ([Poduridae + Neanuridae] + Onychiuridae + Hypogastruridae) was monophyletic and P. aquatica was the sister group of Neanuridae. Entomobrya and Isotoma were united as the sister group to the poduromorphs, supporting other studies that unite them as Entomobryomorpha (C. Bitsch and J. Bitsch 2000; D'Haese 2002). Symphypleona, by contrast, may be paraphyletic because Bayesian analysis placed Sminthurus as the sister group to Arthropleona (fig. 2; also see fig. 4C); however, this violates the strong morphological evidence for uniformity of the Symphypleona (D'Haese 2003), so it is a tentative conclusion. Finally, our trees suggested that the Hypogastruridae is polyphyletic, with Triacanthella separate from Hypogastrura and alone at the base of the poduromorphs.
The above findings are remarkably similar to those of D'Haese (2002), who also used 28S rRNA but an entirely different part of that gene (about 700 nt at the 5' end). Every one of the relations mentioned in the previous paragraph is evident in D'Haese's tree. In fact, the only difference is that his tree grouped Triacanthella with Onychiurus, whereas ours did not. Like him, we found P. aquatica to be a derived collembolan, supporting his contention that ancestral collembolans were not semiaquatic but lived in soils (D'Haese 2002, 2003).
Conclusions
Our investigation inferred the relations of 10 proturans, 12 diplurans, 10 collembolans, and 13 other arthropods from complete 18S and partial 28S rRNA genes (2,000 nt) by using several model-based tree-building methods (ML, Bayesian inference, LogDet) as well as MP plus parametric bootstrapping to test alternate phylogenetic hypotheses. The different tree-building methods produced similar results. The main findings are that Diplura is monophyletic and that Protura unites with Diplura in a clade named here as"Nonoculata," in agreement with the recent studies of Giribet, Edgecombe, and Wheeler (2001), Kjer (2004), and Giribet et al. (2004). The position of Collembola within Hexapoda is uncertain, but some of the methods place collembolans with "Protura and Diplura" in a new version of Entognatha, and an ongoing study based on complete 28S and 18S sequences supports this placement (J. Mallatt, unpublished data). The present rRNA data are inconsistent with traditional, morphology-based hypotheses that unite Protura with Collembola as Ellipura and are extremely inconsistent with hypotheses that split japygids from campodeids into a paraphyletic Diplura (table 4).
However, the Protura + Diplura clade must be accepted with caution because the data set showed strong nonstationarity of nucleotide composition, meaning that the high CG content in the genes of the proturans and diplurans may have artifactually attracted these groups to one another—although the fact that the LogDet method also recovered "Protura + Diplura" lowers this possibility.
The 18S + 28S sequences also contained phylogenetic signal at lower taxonomic levels. Within Protura, the traditional Acerentomata and Eosentomata subgroups were recovered, but the basal Sinentomata was not monophyletic. Within Diplura, Japygoidea was monophyletic with all methods, Campodeoidea was monophyletic when the most complete data set (>2,100) was used, and most methods joined Projapygoidea with Japygoidea. Within Collembola, the traditional Arthropleona was monophyletic as were its subgroups Poduromorpha and Entomobryomorpha; however, Symphypleona was not monophyletic.
Acknowledgements
This work was supported by the National Natural Science Foundation of China (Grant No. 30130040). Gonzalo Giribet from the Museum of Comparative Zoology at Harvard provided the Triacanthella, Sminthurus, and Dilta specimens and kindly offered comments on the manuscript.
References
Abele, L. G., W. Kim, and B. E. Felgenhauer. 1989. Molecular evidence for inclusion of the phylum Pentastomida in the Crustacea. Mol. Biol. Evol. 66:685–691.
Ballard, J. W., G. J. Olsen, D. P. Faith, W. A. Odgers, D. M. Rowell, and P. W. Atkinson. 1992. Evidence from 12S ribosomal RNA sequences that onychophorans are modified arthropods. Science 258:1345–1348.
Bilinski, S. 1994. The ovary of Entognatha. Pp. 7–30 in J. Büning, ed. The insect ovary. Chapman & Hall, London.
Bitsch, C., and J. Bitsch. 2000. The phylogenetic interrelationships of the higher taxa of apterygote hexapods. Zool. Scr. 29:131–156.
———. 2004. Phylogenetic relationships of basal hexapods among the mandibulate arthropods: a cladistic analysis based on comparative morphological characters. Zool. Scr. 33:511–550.
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.
B?rner, C. 1910. Die phylogenetische Bedeutung der Protura. Biol. Zbl. 30:633–641.
Boudreaux, H. B. 1979. Arthropod phylogeny, with special reference to insects. John Wiley & Sons, New York.
Brusca, R. C., and G. J. Brusca. 2003. Invertebrates. 2nd edition. Sinauer Associations, Sunderland, Mass.
Cameron, S. L., K. B. Millar, C. A. D'Haese, M. F. Whiting, and S. C. Barker. 2004. Mitochondrial genome data alone are not enough to unambiguously resolve the relationships of Entognatha, Insecta and Crustacea sensu lato (Arthropoda). Cladistics 20:534–557.
Carapelli, A., F. Frati, F. Nardi, R. Dallai, and C. Simon. 2000. Molecular phylogeny of apterygotan insects based on nuclear and mitochondrial genes. Pedobiologia 44:361–373.
Colgan, D. J., A. McLauchlan, G. D. F. Wilson, S. P. Livingston, G. D. Edgecombe, J. Macaranas, G. Cassis, and M. R. Gray. 1998. Histone H3 and U2 snRNA DNA sequences and arthropod molecular evolution. Aust. J. Zool. 46:419–437.
Dallai, R. 1976. Fine structure of the pyloric region and malpighian papillae of Protura (Insecta Apterygota). J. Morphol. 150:727–762.
Dallai, R., and D. Burroni. 1982. Fine structure of the pyloric region and malpighian papillae of Diplura. Mem. Soc. Entomol. Ital. 60:125–135.
Delamare, D. C., and Z. Massoud. 1968. Révision de Protentomobrya walkeri Folsom, Collembole du Crétacé, et remarques sur sa position systématique. Rev. Ecol. Biol. Sol. 4:619–630.
Delsuc, F., M. J. Phillips, and D. Penny. 2003. Comment on "Hexapod origins: monophyletic or paraphyletic." Science 301:1482d.
D'Haese, C. A. 2002. Were the first springtails semi-aquatic? A phylogenetic approach by means of 28S rDNA and optimization alignment. Proc. R. Soc. B 269:1143–1151.
———. 2003. Morphological appraisal of Collembola phylogeny with special emphasis on Poduromorpha and a test of the aquatic origin hypothesis. Zool. Scr. 32:563–586.
Efron, B. 1985. Bootstrap confidence intervals for a class of parametric problems. Biometrika 72:45–58.
Ferguson, L. M. 1990. Insecta: Diplura. Pp. 951–963 in D. L. Dindal, ed. Soil biology guide. John Wiley & Sons, New York.
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.
Frati, F., and E. Dell'Ampio. 2000. Molecular phylogeny of three subfamilies of the Neanuridae (Insecta, Collembola) and the position of the Antarctic species Friesea grisea Sch?ffer. Pedobiologia 44:342–360.
Friedrich, M., and D. Tautz. 1995. Ribosomal DNA phylogeny of the major extant arthropod classes and the evolution of myriapods. Nature 376:165–167.
———. 1997. An episodic change of rDNA nucleotide substitution rate has occurred during the emergence of the insect order Diptera. Mol. Biol. Evol. 14:644–653.
Giribet, G., and C. Ribera. 1998. The position of arthropods in the animal kingdom: a search of reliable outgroup for internal arthropod phylogeny. Mol. Phylogenet. Evol. 9:481–488.
———. 2000. A review of Arthropoda phylogeny: new data based on ribosomal DNA sequences and direct character optimization. Cladistics 16:204–231.
Giribet, G., G. D. Edgecombe, J. M. Carpenter, C. D'Haese, and W. C. Wheeler. 2004. Is Ellipura monophyletic? A combined analysis of basal hexapod relationships with emphasis on the origin of insects. Org. Divers. Evol. 4:319–340.
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., S. Richter, G. D. Edgecombe, W. C. Wheeler. 2005. The position of crustaceans within the Arthropoda—evidence from nine molecular loci and morphology. Crustac. Issues 16:307–352.
Gloor, G. B., C. R. Preston, D. M. Johnson-Schlitz, N. A. Nassif, R. W. Phillis, W. K. Benz, H. M. Robertson, and W. R. Engels. 1993. Type I repressors of P element mobility. Genetics 135:81–95.
Gutell, R. R. 1994. Collection of small subunit (16S- and 16S-like) ribosomal RNA structures. Nucleic Acids Res. 22:3502–3507.
Gutell, R. R., M. W. Gray, and M. N. Schnare. 1993. A compilation of large subunit (23S and 23S-like) ribosomal RNA structures: 1993. Nucleic Acids Res. 21:3055–3074.
Hancock, J. M., D. Tautz, and G. A. Dover. 1988. Evolution of the secondary structures and compensatory mutations of the ribosomal RNAs of Drosophila melanogaster. Mol. Biol. Evol. 5:393–414.
Hendricks, L., C. Van Broeckhoven, A. Vandenberghe, Y. Van de Peer, and R. De Wachter. 1988. Primary and secondary structure of the 18S ribosomal RNA of the bird spider Eurypelma californica and evolutionary relationships among eukaryotic phyla. Eur. J. Biochem. 177:15–20.
Hennig, W. 1953. Kritische bemerkungen zum phylogenetischen system der insekten. Beitr. Entomol. 3 (Sonderheft):1–85.
———. 1981. Insect phylogeny. John Wiley & Sons, New York.
Hillis, D. M., and M. T. Dixon. 1991. Ribosomal DNA: molecular evolution and phylogenetic inference. Q. Rev. Biol. 66:411–453.
Huelsenbeck, J. P., D. M. Hillis, and R. Jones. 1996. Parametric bootstrapping in molecular phylogenetics: applications and performance. Pp. 19–45 in J. E. Ferraris, and S. R. Palumbi, eds. Molecular zoology: advances, strategies, and protocols. Wiley-Liss, New York.
Huelsenbeck, J. P., and F. Ronquist. 2001. MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics 17:754–755.
Ikeda, Y., and R. Machida. 1998. Embryogenesis of the dipluran Lepidocampa weberi Oudemans (Hexapoda, Diplura, Campodeidae): external morphology. J. Morphol. 237:101–115.
Jermin, L. S., S. Y. W. Ho, F. Ababneh, J. Robinson, and A. W. D. Larkum. 2004. The biasing effect of compositional heterogeneity on phylogenetic estimates may be underestimated. Syst. Biol. 53:638–643.
Kjer, K. M. 2004. Aligned 18S and insect phylogeny. Syst. Biol. 53:506–514.
Koch, M. 1997. Monophyly and phylogenetic position of the Diplura (Hexapoda). Pedobiologia 41:9–12.
Kristensen, N. P. 1981. Phylogeny of insect orders. Annu. Rev. Entomol. 26:135–157.
———. 1991. Phylogeny of extant hexapods. Pp.125–140 in I. D. Naumann, ed. The insects of Australia, Vol. 1, 2nd edition. Melbourne University Press, Carlton, Australia.
Kukalová-Peck, J. 1987. New Carboniferous Diplura, Monura, and Thysanura, the hexapod ground plan, and the role of thoracic side lobes in the origin of wings (Insecta). Can. J. Zool. 65:2327–2345.
———. 1991. Fossil history and the evolution of hexapod structures. Pp. 141–179 in Naumann, I. D., ed. The insects of Australia, Vol. 1, 2nd edition. Melbourne University Press, Carlton, Australia.
Lake, J. A. 1994. Reconstructing evolutionary trees from DNA and protein sequences: paralinear distances. Proc. Natl. Acad. Sci. USA 91:1455–1459.
Lockhart, P. J., M. A. Steel, M. D. Hendy, and D. Penny. 1994. Recovering evolutionary trees under a more realistic model of sequence evolution. Mol. Biol. Evol. 11:605–612.
Luan, Y. X., Y. G. Yao, R. D. Xie, Y. M. Yang, Y. P. Zhang, and W. Y. Yin. 2004. Analysis of 18S rRNA gene of Octostigma sinensis (Projapygoidea: Octostigmatidae) supports the monophyly of Diplura. Pedobiologia 48:453–459.
Luan, Y. X., Y. P. Zhang, Q. Y. Yue, J. F. Pang, R. D. Xie, and W. Y. Yin. 2003. Ribosomal DNA gene and phylogenetic relationship of Diplura and lower hexapods. Sci. China Ser. C 46:67–76.
Mallatt, J., J. R. Garey, and J. W. Shultz, 2004. Ecdysozoan phylogeny and Bayesian inference: first use of nearly complete 28S and 18S rRNA gene sequences to classify the arthropods and their kin. Mol. Phylogenet. Evol. 31:178–191.
Mallatt, J., and J. Sullivan. 1998. 28S and 18S rDNA sequences support the monophyly of lampreys and hagfishes. Mol. Biol. Evol. 15:1706–1718.
Mallatt, J., and C. J. Winchell. 2002. Testing the new animal phylogeny: first use of large-subunit and small-subunit rRNA gene sequences to classify the protostomes. Mol. Biol. Evol. 19:289–301.
Manton, S. M. 1972. The evolution of arthropodan locomotory mechanisms. Part 10. Locomotory habits, morphology and evolution of the hexapod classes. Zool. J. Linn. Soc. 51:203–400.
———. 1977. The Arthropoda: habits, functional morphology, and evolution. Oxford University Press, Oxford.
Nardi, F., G. Spinsanti, J. L. Boore, A. Carapelli, R. Dallai, and F. Frati. 2003. Hexapoda origins: monophyletic or paraphyletic? Science 299:1887–1889.
Nelles, L., B. L. Fang, G. Volckaert, A. Vandenberghe, and R. De Wachter. 1984. Nucleotide sequence of a crustacean 18S ribosomal RNA gene and secondary structure of eukaryotic small subunit ribosomal RNAs. Nucleic Acids Res. 12:8749–8768.
Pagés, J. 1997. Notes sur les Diploures Rhabdoures Insectes, Apterygotes no.1—Diplura Genavensia XXII. Rev. Suisse Zool. 104:869–896.
Posada, D., and T. R. Buckley. 2004. Model selection and model averaging in phylogenetics: advantages of the AIC and Bayesian approaches over likelihood ratio tests. Syst. Biol. 53:793–808.
Posada, D., and K. A. Crandall. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14:817–818.
Rambaut, A., and N. C. Grassly. 1997. Seq-Gen sequence: An application for the Monte Carlo simulation of DNA sequence evolution along phylogenetic trees. Comput. Appl. Biosci. 13:235–238.
Regier, J. C., and J. W. Shultz. 1997. Molecular phylogeny of the major arthropod groups indicates polyphyly of crustaceans and a new hypothesis for the origin of hexapods. Mol. Biol. Evol. 14:902–913.
———. 2001. Elongation factor-2: a useful gene for arthropod phylogenetics. Mol. Phylogenet. Evol. 20:136–148.
Regier, J. C., J. W. Shultz, and R. E. Kambic. 2004. Phylogeny of basal hexapod lineages and estimates of divergence times. Ann. Entomol. Soc. Am. 97:411–419.
Regier, J. C., H. M. Wilson, and J. W. Shultz. 2005. Phylogenetic analysis of Myriapoda using three nuclear protein-coding genes. Mol. Phylogenet. Evol. 34:147–158.
Ruppert, E. E., R. S. Fox, and R. D. Barnes. 2004. Invertebrate zoology, a functional evolutionary approach. Thomson, Brooks/Cole, Belmont, Calif.
Rusek, J. 1982. Octostigma herbivora n. gen. & sp. (Diplura: Projapygoidea: Octostigmatidae n. fam.) injuring plant roots in the Tonga Islands. N. Z. J. Zool. 9:25–32 .
Schnare, M. N., S. H. Damberger, M. W. Gray, and R. Gutell. 1996. Comprehensive comparison of structural characteristics in eukaryotic cytoplasmic large subunit (23S-like) ribosomal RNA. J. Mol. Biol. 256:701–719.
Shao, H. G., Y. P. Zhang, X. Ke, Q. Y. Yue, and W. Y. Yin. 2000. Sequences of mitochondrial DNA cytochrome oxidase II in Cryptopygus nanjiensis and phylogeny of Apterygota. Sci. China Ser. C 43:589–596.
Shao, H. G., Y. P. Zhang, R. D. Xie, and W. Y. Yin. 1999. Mitochondria cytochrome b sequences variation of Protura and molecular systematics of Apterygota. Chin. Sci. Bull. 44:2031–2036.
Shultz, J. W., and J. C. Regier. 2000. Phylogenetic analysis of arthropods using two nuclear protein-encoding genes supports a crustacean + hexapod clade. Proc. R. Soc. B 267:1011–1019.
Spears, T., and L. G. Abele. 1997. Crustacean phylogeny inferred from 18S rDNA. Pp. 169–187 in R. A. Fortey, and R. H. Thomas, eds. Arthropod relationships. Chapman & Hall, London.
tys, P., and S. Bilinski. 1990. Ovariole types and the phylogeny of hexapods. Biol. Rev. 65:401–429.
tys, P., J. Zrzav, and F. Weyda. 1993. Phylogeny of the Hexapoda and ovarian metamerism. Biol. Rev. 68:365–379.
Swofford, D. L. 2002. PAUP*: phylogenetic analysis using parsimony (*and other methods). Version 4.0b10. Sinauer Associates, Sunderland, Mass.
Swofford, D. L., G. J. Olsen, P. J. Waddell, and D. M. Hillis. 1996. Phylogenetic inference. Pp. 407–514 in D. M. Hillis, C. Moritz, and B. K. Mable, eds. Molecular systematics, 2nd edition. Sinauer Associates, Sunderland, Mass.
Turbeville, J. M., D. M. Pfeifer, K. G. Field, and R. A. Raff. 1991. The phylogenetic status of arthropods, as inferred from 18S rRNA sequences. Mol. Biol. Evol. 8:669–686.
Tuxen, S. L. 1964. The Protura. A revision of the species of the world with keys for determination. Hermann, Paris.
Whalley, P. 1995. Unfair to ancient fossil springtails! Antenna 19:2–3.
Wheeler, W. C., P. Cartwright, and C. Y. Hayashi. 1993. Arthropod phylogeny: a combined approach. Cladistics 9:1–39.
Wheeler, W. C., M. Whiting, Q. D. Wheeler, and J. M. Carpenter. 2001. The phylogeny of the extant Hexapod orders. Cladistics 17:113–169.
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.
Winchell, C. J., A. P. Martin, and J. Mallatt. 2004. Phylogeny of elasmobranchs based on LSU and SSU ribosomal RNA genes. Mol. Phylogenet. Evol. 31:214–224.
Winchell, C. J., J. Sullivan, C. B. Cameron, B. J. Swalla, and J. Mallatt. 2002. Evaluating hypotheses of deuterostome phylogeny and chordate evolution with new LSU and SSU ribosomal DNA data. Mol. Biol. Evol. 19:762–776.
Yin, W. Y. 1984. A new idea on phylogeny of Protura with approach to its origin and systematic position. Sci. Sin. Ser. B XXVII:149–160.
———. 1996. New considerations on systematics of Protura. Pp. 60 in Proceedings of XX International Congress of Entomology, Firenze, Italy.
Yin, W. Y., R. D. Xie, Y. M. Yang, Q. Y. Yue, and Y. X. Luan. 2002. Analysis of the main characters for regrouping of the class Protura (Hexapoda). Acta Zootaxonomica Sin. 27:649–658 .
Yin, W. Y., and L. Z. Xue. 1993. Comparative spermatology of Protura and its significance on proturan systematics. Sci. China Ser. B 36:575–587.
Zrzav, J., S. Mihulka, P. Kepka, A. Bezdk, and D. Tietz. 1998. Phylogeny of the Metazoa based on morphological and 18S ribosomal DNA evidence. Cladistics 14:249–285.(Yun-xia Luan*, Jon M. Mal)
Correspondence: E-mail: yxluan@sibs.ac.cn.
Abstract
This study combined complete 18S with partial 28S ribosomal RNA gene sequences (2,000 nt in total) to investigate the relations of basal hexapods. Ten species of Protura, 12 of Diplura, and 10 of Collembola (representing all subgroups of these three clades) were sequenced, along with 5 true insects and 8 other arthropods, which served as out-groups. Trees were constructed with maximum parsimony, maximum likelihood, Bayesian analysis, and minimum-evolution analysis of LogDet-transformed distances. All methods yielded strong support for a clade of Protura plus Diplura, here named Nonoculata, and for monophyly of the Diplura. Parametric-bootstrapping analysis showed our data to be inconsistent with previous hypotheses (P < 0.01) that joined Protura with Collembola (Ellipura), that said Diplura are sister to true insects or are diphyletic, and that said Collembola are not hexapods. That is, our data are consistent with hexapod monophyly and Collembola grouped weakly with "Protura + Diplura" under most analytical conditions. As a caveat to the above conclusions, the sequences showed nonstationarity of nucleotide frequencies across taxa, so the CG-rich sequences of the diplurans and proturans may have grouped together artifactually; however, the fact that the LogDet method supported this group lessens this possibility. Within the basal hexapod groups, where nucleotide frequencies were stationary, traditional taxonomic subgroups generally were recovered: i.e., within Protura, the Eosentomata and Acerentomata (but Sinentomata was not monophyletic); within Collembola, the Arthropleona, Poduromorpha, and Entomobryomorpha (but Symphypleona was polyphyletic); and in Diplura, the most complete data set (>2,100 nt) showed monophyly of Campodeoidea and of Japygoidea, and most methods united Projapygoidea with Japygoidea.
Key Words: Protura ? Diplura ? Collembola ? ribosomal RNA genes ? molecular phylogeny
Introduction
Hexapoda (Insecta sensu lato) includes four groups: Protura, Collembola, Diplura, and Insecta sensu stricto (with the latter containing Microcoryphia or Archeognatha, Zygentoma, and the pterygote insects). Protura, Collembola, and Diplura are ancient groups, known as fossils from the Devonian, Carboniferous, and Cretaceous periods up to 400 MYA (Delamare and Massoud 1968; Kukalová-Peck 1987, 1991; Whalley 1995). Based on their mouthparts, Hennig (1953) joined Protura, Collembola, and Diplura into Entognatha (with enclosed mouthparts), in contrast to Ectognatha (Insecta sensu stricto, with exposed mouthparts). All hexapods have a body divided into three basic tagmata—head, thorax, and abdomen—and three pairs of thoracic legs (R. C. Brusca and G. J. Brusca 2003). The three groups of entognath hexapods share (with many insect groups) the superficial similarities of small body size and a preference for living in soil, leaf litter, or under rocks. However, controversies about the interrelationships of these entognath groups have raged in recent years, with the monophyly of Diplura and the relative phylogenetic positions of Diplura, Protura, and Collembola being hot spots of dispute (Giribet et al. 2004).
Proturans are tiny animals with a conical head and a relatively long abdomen. They have no eyes, antennae, or caudal cerci but have a telson tail, which is common in crustaceans but absent in other hexapods. Protura includes three subgroups: Acerentomata, Eosentomata, and Sinentomata (Yin 1996), but within Sinentomata, the phylogenetic positions of the families Sinentomidae and Fujientomidae are controversial due to their special morphological characteristics. Unusual kinds of spermatozoa exist both within and between proturan species, which might reflect their long evolutionary course (Yin and Xue 1993).
Diplurans have a long, thin body that appears to have two paired tails. Actually, these "tails" are two long antennae projecting forward from the eyeless head and two cerci projecting back from the abdomen. Diplura is composed of Campodeoidea, Japygoidea, and Projapygoidea. Despite morphological features shared by these subgroups, the monophyly of Diplura has been questioned because Campodeoidea and Japygoidea differ in the type of cerci (a filiform vs. a grasping pincer), sperm morphology, and ovarian structure (tys and Bilinski 1990; Bilinski 1994). In addition, the phylogenetic position of the entire group is controversial: Does Diplura actually group with the true insects instead of with the other entognaths (Kukalová-Peck 1987)? What are the relationships between Diplura and the other "orders" of hexapods? Also, few studies have included Projapygoidea, whose phylogenetic position within Diplura is still a puzzle.
Collembola (springtails) are the most diverse of the entognath hexapods and are widely distributed in the world. They are defined by a characteristic forked "tail" (furcula), which enables them to spring, hence their English name. Collembola comprise two suborders: Arthropleona and Symphypleona. Agreement over this taxonomic scheme is not universal, however, and the phylogenetic position of Podura aquatica, an important species that has been considered as one of the most primitive collembolans, is also unclear (D'Haese 2002).
Several major hypotheses of basal hexapod relationships have been proposed based on morphological and molecular characters (fig. 1). Most of these hypotheses include a clade uniting Protura and Collembola, called Ellipura, based on a distinctive type of enclosed mouthparts and on the loss or absence of several characters (e.g., cerci) present in other hexapods (B?rner 1910; Hennig 1981; Kristensen 1981, 1991; Kukalová-Peck 1987, 1991; Koch 1997; Wheeler et al. 2001). The first hypothesis (fig. 1A), that of Hennig (1981), divides hexapods into Ectognatha and Entognatha, with the latter consisting of Ellipura and Diplura. The second hypothesis (Kukalová-Peck 1987) places Diplura with Ectognatha (fig. 1B) based on a Carboniferous fossil that was interpreted as japygid-like but had eyes and more exposed mouthparts, as in ectognaths. The next hypothesis (fig. 1C) proposed that Diplura is diphyletic (tys and Bilinski 1990). Based on the structure of the ovarioles, "Japygina" (=Japygidae) was placed basal to "Campodeina" (=Campodeidae) + Ellipura in Entognatha. The final hypothesis (fig. 1D) is from the molecular-phylogenetic study of Nardi et al. (2003), in which mitochondrial gene sequences suggested that Collembola are basal to a clade consisting of crustaceans and insects. According to this hypothesis, hexapods are not monophyletic, although some have disagreed with this suggestion (Delsuc, Phillips, and Penny 2003; Cameron et al. 2004).
FIG. 1.— Four hypotheses of basal hexapod relationships, which are evaluated in the present study.
Many studies have begun to explore the relations among arthropod groups using various molecular markers. These markers include 18S and 28S ribosomal RNA (rRNA) genes (Turbeville et al. 1991; Wheeler, Cartwright, and Hayashi 1993; Friedrich and Tautz 1995, 1997; Giribet and Ribera 1998, 2000; Mallatt, Garey, and Shultz 2004), which are thought to be especially appropriate for resolving higher level phylogenetic relationships (Hillis and Dixon 1991; Mallatt and Winchell 2002). Other studies used mitochondrial gene sequences and mitochondrial gene order (Ballard et al. 1992; Boore et al. 1995; Boore, Lavrov, and Brown, 1998), histone H3 and U2 small nuclear RNA (Colgan et al. 1998), Elongation Factor (EF) 1- and RNA Polymerase II (Regier and Shultz 1997; Shultz and Regier 2000), and EF-2 (Regier and Shultz 2001; Regier, Shultz, and Kambic 2004). Some of the studies are especially interesting in that they combine several sources of molecular data plus morphological characters (i.e., Zrzav et al. 1998; Giribet, Edgecombe, and Wheeler 2001; Wheeler et al. 2001). Others have started to focus on the relations among the basal hexapods (for example, Shao et al. 1999, 2000), and the most detailed of these studies have supported dipluran monophyly (Carapelli et al. 2000; Giribet and Ribera 2000; Luan et al. 2003, 2004; Regier, Shultz, and Kambic 2004; but not Shultz and Regier 2000), and in addition, some have suggested a new possibility: that Protura unites with Diplura (Giribet, Edgecombe, and Wheeler 2001; Giribet et al. 2004; Kjer 2004).
Here, we use complete 18S and partial 28S rRNA sequences to expand on these studies. We have obtained the most extensive sample of basal hexapods so far: namely, 10 proturans, 12 diplurans, and 10 collembolans, including all their major subgroups—plus a wide range of out-group taxa. Additionally, whereas most past studies used only parsimony-based methods of phylogenetic reconstruction, we add model-based and likelihood methods and use parametric bootstrapping to test the hypotheses in figure 1 in a statistical manner.
Materials and Methods
Most proturan, dipluran, and collembolan specimens were collected in 75% ethanol by the Tullgren funnel method. All were stored in ethanol in a freezer or refrigerator after identification. More details are given in table 1, which lists the species used. The choice of out-group taxa for the study was complicated by controversy over whether the closest relatives of hexapods are crustaceans or myriapods (Mallatt, Garey, and Shultz 2004). To overcome this problem, we used numerous representatives of each: four crustaceans, three myriapods, and even a chelicerate. As other out-group taxa, five major groups of true insects were also represented, bringing the total number of species used in this study to 45. Most of the out-group sequences were obtained from GenBank (see table 1).
Table 1 Information on Species Used in This Study
Genomic DNA was extracted from one individual of most species using the single-fly extraction method (Gloor et al. 1993) or from several individuals of some species (Campodeidae sp., Triacanthella, Sminthurus, Dilta) with the DNeasy Tissue Kit (Qiagen Inc., Valencia, Calif.). The 18S rRNA gene, ranging in length from 1,750 nt in collembolans to 2,200 nt in some diplurans, generally was amplified in four overlapping fragments of 600 nt each, using primer pairs 18S1L-18S1R, 18SL500-18SR1470, 18SL1210-18SR1790, and 18S3L-18S3R, respectively (table 2). The 28S rRNA fragment, 430 nt long and located in the 5' third of the gene from beyond the D3 to beyond the D5 divergent domains, was amplified using primer pair 28Sf and 28Sr. For Campodeidae sp., Triacanthella, Sminthurus, and Dilta, the complete 18S and 28S genes were each amplified in one or two pieces (Mallatt and Sullivan 1998). For the polymerase chain reaction–amplification and sequencing conditions see Luan et al. (2004) and Mallatt and Sullivan (1998). It should be noted that neither the 18S nor the 28S sequence was missing from any of the 45 taxa.
Table 2 Primers Used for Amplification and Sequencing of 18S rDNA and 28S rDNA
As in previous studies (Mallatt and Sullivan 1998; Winchell et al. 2002; Mallatt, Garey, and Shultz 2004), the genes of all taxa were aligned by eye in the GCG Seqlab program (Wisconsin Package Version 10.3; Accelrys Inc., San Diego, Calif.) against a reference set consisting of rRNA sequences from many other metazoans and arthropods, all rigidly aligned based on the rRNA secondary-structure models of Xenopus laevis and Strongylocentrotus purpuratus (Gutell 1994; Schnare et al. 1996) and confirmed by reference to models of yeast, Saccharomyces cerevisiae, rRNA (see http://www.psb.agent.be/rRNA/secmodel/Scer_SSU.html for yeast 18S and Gutell, Gray, and Schnare 1993 for yeast 28S). Thus, these reference models should be valid across eukaryotes. The secondary structure of rRNA genes is known for the insect Drosophila (Hancock, Tautz, and Dover 1988), and the Drosophila sequences are among those in our reference-alignment set; however, they were not used as our primary reference because Drosophila rRNA sequences are too divergent to produce suitable alignments.
About 10% of the sites in our data set were unalignable and were excluded from the analysis. More specifically, we excluded sites in which more than 30% of the taxa do not share the same nucleotide (Hillis and Dixon 1991), and this is made simpler by the fact that rRNA genes have long, conserved segments (in which almost all the sites are alignable) interrupted by about a dozen variable regions (in which many sites are obviously unalignable); therefore, most decisions about exclusion involve only short segments of just 5–10 nt in the transition zones between conserved and variable regions. The sites we excluded were almost entirely in the V2, V4, and V7–9 variable regions of 18S and in the D4 and D5 divergent domains of 28S. The remaining alignments contained 1,926 sites across the 45 taxa. Our alignments are available at http://www.wsu.edu/jmallatt/alignments.html (posted December 17, 2004) and from European Molecular Biology Laboratory (ftp://ftp.ebi.ac.uk/pub/databases/embl/align) under the accession numbers ALIGN_000854–000861.
For the phylogenetic analyses, the 18S and partial 28S sequences were combined because, as parts of the same gene family, the 18S and 28S genes of animals routinely yield similar trees (Mallatt and Winchell 2002; Winchell et al. 2002; Mallatt, Garey, and Shultz 2004). As in previous studies, phylogenetic trees relating the 45 taxa were estimated with several different algorithms, all executed in PAUP* 4.0 beta 10 (Swofford 2002): equally weighted maximum parsimony (MP), minimum evolution using LogDet-Paralinear distances (LogDet), and maximum likelihood (ML) in which the general time reversible (GRT) +I + model was found to fit our combined rRNA data best (by the Akaike information criterion (AIC) approach to model evaluation in Modeltest; Posada and Crandall 1998). For more information on these methods, see Mallatt and Winchell (2002), Mallatt, Garey, and Shultz (2004), and Posada and Buckley (2004).
For the LogDet and MP methods, support for clades was evaluated with ordinary, nonparametric bootstrapping (1,000 replicates), but for ML, due to high computational demands, only the single best tree could be calculated; however, ML nonparametric bootstrapping with 100 replicates was performed on the trimmed, 34-taxon set introduced below.
Likelihood-based Bayesian inference (Markov Chain Monte Carlo analysis; Huelsenbeck and Ronquist 2001) was also used on the 45-taxon set, as described in Winchell, Martin, and Mallatt (2004), and Mallatt, Garey, and Shultz (2004), using MrBayes 3.01. Briefly, we used a single, invgamma model, for which the number of substitution types (nst) was set at six, four Markov chains were run for 106 generations, and posterior probabilities were calculated from these trees, after the first 200,000 were discarded as burn-in. In addition to this single-model test, we ran a partitioned Bayesian analysis, after using AIC in Modeltest to find the best models for the 18S and 28S partitions; these models were "nst = 6 invgamma" for both the partitions.
The four hypotheses in figure 1, plus a general hypothesis that hexapods are monophyletic, were tested with our data using ML-based parametric bootstrapping (Efron 1985; Huelsenbeck, Hillis, and Jones 1996). To accommodate the computational demands of this method, it was necessary to trim the 45 taxa down to 34, by removing some of the out-groups as well as those in-group taxa that were most closely related to others (in the same genus, for example). Care was taken to assure that the retained 34 sequences yielded the same basic topology as the 45-taxon tree (compare figs. 2 and 3). After ML was used to find optimal constrained and unconstrained trees, the parametric-bootstrap tests were performed, with 100 simulations per hypothesis tested, as described in Mallatt, Garey, and Shultz (2004). The necessary simulations were done with the Seq-Gen version 1.2.5 program (Rambaut and Grassly 1997).
FIG. 2.— Phylogenetic tree based on the entire data set: 45 taxa; 18S + partial 28S genes; 1,926 characters. The basic tree, with its branch lengths, was calculated by ML (–Ln L = 14,707.69). The four values at each node are, respectively, (1) Bayesian posterior probability based on a partitioned analysis with separate models for 18S and 28S partitions (1.5 x 106 generations), (2) Bayesian posterior probability based on a single model for the combined 18S + 28S data (106 generations), (3) LogDet-bootstrap value (1,000 bootstrap replicates), and (4) MP-bootstrap value (1,000 bootstrap replicates).
FIG. 3.— ML tree based on an abbreviated data set; same set as in figure 2 but with just 34 taxa (–Ln L = 13,050.0651). ML-bootstrap values (100 replicates) are indicated at the nodes.
Finally, we further examined the relationships of taxa within each basal hexapod group: in Protura, then in Diplura, then in Collembola. Because these analyses considered only within-group relations, there was no need to include an out-group, so a larger number of alignable characters could be recognized and included in the analysis—expanding each of the character sets from 1,926 to >2,100 sites and, presumably, increasing the amount of phylogenetic signal. The relations within each group were analyzed with the ML, LogDet, and MP tree-building algorithms.
As in previous studies (Mallatt, Garey, and Shultz 2004; Winchell, Martin, and Mallatt 2004), we accepted clades in the Bayesian tree at 95% posterior probability, while accepting values around 60%–70% as significant in the nonparametric-bootstrap trees. For parametric bootstrapping, rejection was at P 0.05.
Results
Nucleotide Composition
When the 2 test of stationarity of nucleotide frequencies in PAUP* was applied to our aligned data set, the frequencies were found to be highly nonstationary across the 45 taxa (2 = 648.1; P = 0.00000000) (table 3), and the same was found for the abbreviated data set of 34 taxa (not shown). This extreme nonstationarity must be due to the high proportion of C and G nucleotides in the diplurans because when all diplurans were removed, the remaining sequences were stationary (P = 0.43). Then, however, the proturan sequences were more CG rich than any of the other remaining taxa. When nucleotide frequencies were tested within each of the three basal hexapod groups (i.e., within Protura, within Diplura, within Collembola), they were always stationary (P = 0.98 for each group).
Table 3 Chi-square Test of Stationarity of Base Frequencies in the 18S + 28S Sequences Used in This Study
Trees
The calculated phylogenetic trees are shown in figures 2–4. Similar results were obtained by all methods: Bayesian inference, LogDet, MP, and ML. Protura, Diplura, and Collembola were each monophyletic with 100% statistical support (figs. 2 and 3). Protura grouped strongly with Diplura, again with universal 100% support. The position of collembolans was not clearly resolved, although the Bayesian, ML, and LogDet methods joined them with "Protura + Diplura" with weak support. Within Protura, the subgroups Eosentomata and Acerentomata were each monophyletic (and were joined together by the likelihood-based methods), but Sinentomata was not monophyletic as its two genera, Sinentomon and Fujientomon, did not group together (figs. 2 and 4A). In Diplura, a monophyletic Campodeoidea was evident with the expanded sequences (fig. 4B) but not with the shorter sequences of figure 2. Japygoidea was universally monophyletic, and the projapygid Octostigma was grouped with Japygoidea by all methods except LogDet (figs. 2 and 3). Within Collembola, Arthropleona was monophyletic but Symphypleona was not, being represented by a grade of taxa at the base of the collembolans. Entomobrya and Isotoma, which are sometimes placed together in the order Entomobryomorpha (D'Haese 2002), formed a monophyletic clade of entomobryomorphs separate from the rest of the Arthropleona (the Poduromorpha) (figs. 2 and 4C).
FIG. 4.— ML trees, presented as star diagrams, for the three basal-hexapod groups. Based on expanded alignments of >2,100 characters, which should reveal the relationships within these groups better than in figure 2, which was based on fewer characters. The "veils" of radiating lines are artistic devices to separate major subgroups of taxa. Bootstrap values are indicated at the nodes: ML (100 replicates), MP (1,000 replicates), and LogDet (LD) (1,000 replicates). The thick arrows show where these trees can be rooted, as determined from figure 2. (A) Ten proturans, best ML tree with bootstrap values: ML/MP/LD, 2,237 nt, –Ln L= 7,055.3501, 2 = 0.98; (B) 12 diplurans, best ML tree with bootstrap values: ML/MP/LD, 2,211 nt, –Ln L= 4,839.8029, 2=0.98; (C) 10 collembolans, best ML tree with bootstrap values: ML/MP/LD, –Ln L= 5,736.3947, 2,169 nt, 2 = 0.98.
Among the out-groups, the following groupings are evident and received some support, at least from the likelihood-based techniques (figs. 2 and 3): crustaceans plus hexapods (Pancrustacea), millipedes, millipedes + centipede Lithobius, and pterygote insects. However, every method placed the Artemia sequence (a crustacean) with some hexapod, apparently violating hexapod monophyly.
Hypothesis Testing
Parametric-bootstrap analysis of the rRNA sequences from the 34-taxon subset (taxa shown in fig. 3) was used to test the four alternate hypotheses of hexapod relationships shown in figure 1 as well as the hypothesis that hexapods are monophyletic. Results are presented in table 4. All these hypotheses were rejected by our data, except for hexapod monophyly (Hypothesis 5).
Table 4 Results of Hypothesis Testing by Parametric Bootstrapping
Discussion
General
The main findings of this study, which used the largest taxon sample to date to explore the relationships of basal hexapods, are that Protura and Diplura are sister taxa and that diplurans are monophyletic—thereby upholding the conclusions of some other recent molecular and morphological studies (Giribet, Edgecombe, and Wheeler 2001; Giribet et al. 2004; Kjer 2004). Bootstrap-support values for these clades were high and were consistent across the four tree-building methods (figs. 2–4), suggesting that the rRNA genes contain a large amount of signal. All alternate hypotheses that unite proturans with collembolans as Ellipura were rejected by parametric bootstrapping (table 4), as was the hypothesis of Nardi et al. (2003) that collembolans are not hexapods (for more discussion of this latter point, see C. Bitsch and J. Bitsch 2000, 2004; Delsuc, Phillips, and Penny 2003; and Cameron et al. 2004).
A potential problem with these conclusions must be pointed out. Nonstationarity of nucleotide frequencies across taxa can mislead both likelihood-based and parsimony-based methods (Jermin et al. 2004), and our nucleotide frequencies were nonstationary (table 3). Thus, because both the dipluran and proturan sequences are CG rich, these two sequences might have been united artifactually by homoplasy; that is, they may actually be unrelated but have independently evolved C's and G's at the same sites (Swofford et al. 1996). The LogDet method is designed to avoid this particular "long-branch attraction" artifact (Lake 1994; Lockhart et al. 1994), so our finding that LogDet grouped diplurans with proturans adds credibility to the "Protura + Diplura" clade. Also, a recent simulation study (Jermin et al. 2004) indicated that this attraction artifact is most likely to occur where stem branches are short—but the stem branch leading to "Protura + Diplura" in our study appears rather long (fig. 2), increasing the chance that this clade is valid after all. Nonetheless, caution must be exercised because the true length of that stem branch is unknown given our CG–biased data, and LogDet is not always able to solve the nonstationarity problem (Foster and Hickey 1999). Furthermore, the LogDet minimum-evolution method employed in PAUP* is weak at modeling among-site rate variation within genes—a weakness that itself can cause long-branch attraction artifacts in rRNA-based analyses of arthropod phylogeny (Mallatt, Garey, and Shultz 2004). Therefore, our support for a clade of proturans plus diplurans is not absolutely solid. On the other hand, the fact that nucleotide composition is stationary within Diplura (P = 0.98) means that the other major conclusion of this study—that Japygoidea, Projapygoidea, and Campodeoidea form a monophyletic Diplura—is solid.
In the 18S sequences of our basal hexapods, the differences between individuals of the same species (from different geographical locations) are no more than 10 nt and are far smaller than those of any comparison between the different species of a genus. This supports the morphological identification of species and suggests conservation of the 18S rRNA gene at the species level. Therefore, rRNA genes may be helpful for identifying some odd species.
Specific Groups
Out-group Taxa and Hexapod Monophyly
Most of the relations we obtained among the out-groups are reasonable, based on past morphological and molecular studies (for example, monophyly of Pancrustacea, of Insecta, and of Myriapoda; Giribet, Edgecombe, and Wheeler 2001; Ruppert, Fox, and Barnes 2004; Regier, Wilson, and Shultz 2005). However, the fact that our trees show the branchiopod crustacean, Artemia, in Hexapoda (figs. 2 and 3) should not be interpreted as evidence for a paraphyletic Hexapoda for several reasons. First, it is notoriously difficult to obtain reasonable relationships between the hexapods and the groups of crustaceans using just 18S (and partial 28S) genes (Giribet and Ribera 2000; Wilson et al. 2000; Giribet et al. 2005). Second, the support for "Artemia plus insects" in our trees was below the level of significance for most of the tree-building methods, and parametric bootstrapping showed our data to be fully consistent with hexapod monophyly (table 4). Finally, an expanded data set, which includes nearly complete 28S and 18S sequences from more taxa, supports branchiopods as distinct from a monophyletic Hexapoda (J. Mallatt, unpublished data; also see Mallatt, Garey, and Shultz 2004). Thus, the placement of Artemia in Hexapoda in this study seems to be an artifact of insufficient signal in the 18S gene.
Protura with Diplura
If Protura and Diplura form a clade, as our molecular results suggest (also see Luan et al. 2003, 2004), then an attempt should be made to identify some shared morphological characters of this clade and the clade should be named. Few shared characters of proturans and diplurans have been proposed in the literature, but we have tentatively identified the following: (1) Protura, Campodeoidea, and some Projapygoidea have malpighian tubules only as short papillae (in other projapygids and in japygids, the tubules are absent) (Dallai 1976; Boudreaux 1979; Hennig 1981; Dallai and Burroni 1982; Ferguson 1990) and (2) the lack of eyes, meaning neither simple nor compound eyes are present (Tuxen 1964; Boudreaux 1979; Hennig 1981). We name the clade "Nonoculata" ("no eyes") after this second character.
Subgroups recovered within Protura, Diplura, and Collembola will be considered next. Because the nucleotide frequencies within each of these groups were stationary, there is no chance of bias due to convergent evolution of nucleotide composition.
Protura
There are three taxonomic groups in Protura: Acerentomata, Eosentomata, and Sinentomata (Yin 1996). These possess three types of pseudoculus, respectively, the ophiso-neuroforamen, mid-neuroforamen, and multi-neuroforamen types (Yin et al. 2002). Additionally, Eosentomata possess meso- and metathoracic spiracles with a primitive tracheal system, while members of Acerentomata lack these structures. Within Sinentomata, species of Fujientomidae lack a tracheal system, but species of Sinentomidae possess one (though this system differs obviously from that of Eosentomata). Our rRNA-based phylogenetic analyses of proturan species supported the monophyly of Acerentomata and of Eosentomata, but Sinentomata (represented by one species of Sinentomidae and one of Fujientomidae) was not monophyletic (figs 2, 3, and 4A). Traditionally, Fujientomidae was considered the primitive group among Acerentomata (Yin 1984). However, its spermatozoon and pseudoculus are patterned as in Sinentomidae, so Yin and Xue (1993) proposed the clade Sinentomata. Although not supported by high bootstrap values, most of our trees show Fujientomon as basal, the sister group to "Sinentomon plus the other proturans."
Diplura
As mentioned, controversy surrounds the question of whether Diplura is monophyletic or diphyletic. Most morphological studies have suggested that Diplura is monophyletic. The anatomical characters shared by Japygoidea and Campodeoidea include the special structure of the labium and of the oral folds, the interlocking between the superlingua and the maxillary galea (Koch 1997), unique muscles and pivots in the legs (Manton 1972, 1977), the absence of eyes (C. Bitsch and J. Bitsch 2000, 2004), and embryonic features of the head (Ikeda and Machida 1998). However, evidence from the ultrastructure of germ cells suggested that the sperm of Campodeidae resembled that of Insecta sensu stricto, yet their ovarian structure was similar to that of Collembola and differed from that of Insecta sensu stricto. In contrast, Japygidae was similar to Collembola in sperm ultrastructure but resembled Insecta sensu stricto in ovarian structure. Based on these discrepancies and emphasizing ovarian structure, tys and Bilinski (1990) suggested that "Diplura" is a paraphyletic group. Our rRNA-based findings do not support this. In fact, tys and Bilinski's hypothesis that separated japygids from campodeids was the most strongly rejected of all hypotheses tested in this study (table 4).
Projapygids are an interesting and controversial group of Diplura. Specimens of Projapygoidea are so difficult to find that only a few studies have included them (Rusek 1982; C. Bitsch and J. Bitsch 2000, 2004; Luan et al. 2004), and their phylogenetic position is controversial because they harbor many morphologically intermediate characters between Campodeoidea and Japygoidea (Rusek 1982). Different morphological studies have concluded that they are basal diplurans (Rusek 1982) or that they group with Campodeoidea (Pagés 1997) or with Japygoidea (tys and Bilinski 1990; tys, Zrzav, and Weyda 1993). In the present study, all methods except LogDet united our projapygid Octostigma with Japygoidea. Another, non-sequence–based characteristic shared by Projapygoidea and Japygoidea is that their 18S genes were longer than in any of the species of Campodeoidea by more than 300 bp.
Collembola
Collembola is more diverse than either proturans or diplurans, so our sampling within its major subgroups is less complete. The two traditional subgroups are Arthropleona and Symphypleona: in Arthropleona, the thoracic and abdominal segments are easily separable, whereas Symphypleona generally have a globular body with the thoracic and first four abdominal segments fused. Our trees support monophyly of the Arthropleona (figs. 2, 3, and 4C). Within Arthropleona, Poduromorpha ([Poduridae + Neanuridae] + Onychiuridae + Hypogastruridae) was monophyletic and P. aquatica was the sister group of Neanuridae. Entomobrya and Isotoma were united as the sister group to the poduromorphs, supporting other studies that unite them as Entomobryomorpha (C. Bitsch and J. Bitsch 2000; D'Haese 2002). Symphypleona, by contrast, may be paraphyletic because Bayesian analysis placed Sminthurus as the sister group to Arthropleona (fig. 2; also see fig. 4C); however, this violates the strong morphological evidence for uniformity of the Symphypleona (D'Haese 2003), so it is a tentative conclusion. Finally, our trees suggested that the Hypogastruridae is polyphyletic, with Triacanthella separate from Hypogastrura and alone at the base of the poduromorphs.
The above findings are remarkably similar to those of D'Haese (2002), who also used 28S rRNA but an entirely different part of that gene (about 700 nt at the 5' end). Every one of the relations mentioned in the previous paragraph is evident in D'Haese's tree. In fact, the only difference is that his tree grouped Triacanthella with Onychiurus, whereas ours did not. Like him, we found P. aquatica to be a derived collembolan, supporting his contention that ancestral collembolans were not semiaquatic but lived in soils (D'Haese 2002, 2003).
Conclusions
Our investigation inferred the relations of 10 proturans, 12 diplurans, 10 collembolans, and 13 other arthropods from complete 18S and partial 28S rRNA genes (2,000 nt) by using several model-based tree-building methods (ML, Bayesian inference, LogDet) as well as MP plus parametric bootstrapping to test alternate phylogenetic hypotheses. The different tree-building methods produced similar results. The main findings are that Diplura is monophyletic and that Protura unites with Diplura in a clade named here as"Nonoculata," in agreement with the recent studies of Giribet, Edgecombe, and Wheeler (2001), Kjer (2004), and Giribet et al. (2004). The position of Collembola within Hexapoda is uncertain, but some of the methods place collembolans with "Protura and Diplura" in a new version of Entognatha, and an ongoing study based on complete 28S and 18S sequences supports this placement (J. Mallatt, unpublished data). The present rRNA data are inconsistent with traditional, morphology-based hypotheses that unite Protura with Collembola as Ellipura and are extremely inconsistent with hypotheses that split japygids from campodeids into a paraphyletic Diplura (table 4).
However, the Protura + Diplura clade must be accepted with caution because the data set showed strong nonstationarity of nucleotide composition, meaning that the high CG content in the genes of the proturans and diplurans may have artifactually attracted these groups to one another—although the fact that the LogDet method also recovered "Protura + Diplura" lowers this possibility.
The 18S + 28S sequences also contained phylogenetic signal at lower taxonomic levels. Within Protura, the traditional Acerentomata and Eosentomata subgroups were recovered, but the basal Sinentomata was not monophyletic. Within Diplura, Japygoidea was monophyletic with all methods, Campodeoidea was monophyletic when the most complete data set (>2,100) was used, and most methods joined Projapygoidea with Japygoidea. Within Collembola, the traditional Arthropleona was monophyletic as were its subgroups Poduromorpha and Entomobryomorpha; however, Symphypleona was not monophyletic.
Acknowledgements
This work was supported by the National Natural Science Foundation of China (Grant No. 30130040). Gonzalo Giribet from the Museum of Comparative Zoology at Harvard provided the Triacanthella, Sminthurus, and Dilta specimens and kindly offered comments on the manuscript.
References
Abele, L. G., W. Kim, and B. E. Felgenhauer. 1989. Molecular evidence for inclusion of the phylum Pentastomida in the Crustacea. Mol. Biol. Evol. 66:685–691.
Ballard, J. W., G. J. Olsen, D. P. Faith, W. A. Odgers, D. M. Rowell, and P. W. Atkinson. 1992. Evidence from 12S ribosomal RNA sequences that onychophorans are modified arthropods. Science 258:1345–1348.
Bilinski, S. 1994. The ovary of Entognatha. Pp. 7–30 in J. Büning, ed. The insect ovary. Chapman & Hall, London.
Bitsch, C., and J. Bitsch. 2000. The phylogenetic interrelationships of the higher taxa of apterygote hexapods. Zool. Scr. 29:131–156.
———. 2004. Phylogenetic relationships of basal hexapods among the mandibulate arthropods: a cladistic analysis based on comparative morphological characters. Zool. Scr. 33:511–550.
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.
B?rner, C. 1910. Die phylogenetische Bedeutung der Protura. Biol. Zbl. 30:633–641.
Boudreaux, H. B. 1979. Arthropod phylogeny, with special reference to insects. John Wiley & Sons, New York.
Brusca, R. C., and G. J. Brusca. 2003. Invertebrates. 2nd edition. Sinauer Associations, Sunderland, Mass.
Cameron, S. L., K. B. Millar, C. A. D'Haese, M. F. Whiting, and S. C. Barker. 2004. Mitochondrial genome data alone are not enough to unambiguously resolve the relationships of Entognatha, Insecta and Crustacea sensu lato (Arthropoda). Cladistics 20:534–557.
Carapelli, A., F. Frati, F. Nardi, R. Dallai, and C. Simon. 2000. Molecular phylogeny of apterygotan insects based on nuclear and mitochondrial genes. Pedobiologia 44:361–373.
Colgan, D. J., A. McLauchlan, G. D. F. Wilson, S. P. Livingston, G. D. Edgecombe, J. Macaranas, G. Cassis, and M. R. Gray. 1998. Histone H3 and U2 snRNA DNA sequences and arthropod molecular evolution. Aust. J. Zool. 46:419–437.
Dallai, R. 1976. Fine structure of the pyloric region and malpighian papillae of Protura (Insecta Apterygota). J. Morphol. 150:727–762.
Dallai, R., and D. Burroni. 1982. Fine structure of the pyloric region and malpighian papillae of Diplura. Mem. Soc. Entomol. Ital. 60:125–135.
Delamare, D. C., and Z. Massoud. 1968. Révision de Protentomobrya walkeri Folsom, Collembole du Crétacé, et remarques sur sa position systématique. Rev. Ecol. Biol. Sol. 4:619–630.
Delsuc, F., M. J. Phillips, and D. Penny. 2003. Comment on "Hexapod origins: monophyletic or paraphyletic." Science 301:1482d.
D'Haese, C. A. 2002. Were the first springtails semi-aquatic? A phylogenetic approach by means of 28S rDNA and optimization alignment. Proc. R. Soc. B 269:1143–1151.
———. 2003. Morphological appraisal of Collembola phylogeny with special emphasis on Poduromorpha and a test of the aquatic origin hypothesis. Zool. Scr. 32:563–586.
Efron, B. 1985. Bootstrap confidence intervals for a class of parametric problems. Biometrika 72:45–58.
Ferguson, L. M. 1990. Insecta: Diplura. Pp. 951–963 in D. L. Dindal, ed. Soil biology guide. John Wiley & Sons, New York.
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.
Frati, F., and E. Dell'Ampio. 2000. Molecular phylogeny of three subfamilies of the Neanuridae (Insecta, Collembola) and the position of the Antarctic species Friesea grisea Sch?ffer. Pedobiologia 44:342–360.
Friedrich, M., and D. Tautz. 1995. Ribosomal DNA phylogeny of the major extant arthropod classes and the evolution of myriapods. Nature 376:165–167.
———. 1997. An episodic change of rDNA nucleotide substitution rate has occurred during the emergence of the insect order Diptera. Mol. Biol. Evol. 14:644–653.
Giribet, G., and C. Ribera. 1998. The position of arthropods in the animal kingdom: a search of reliable outgroup for internal arthropod phylogeny. Mol. Phylogenet. Evol. 9:481–488.
———. 2000. A review of Arthropoda phylogeny: new data based on ribosomal DNA sequences and direct character optimization. Cladistics 16:204–231.
Giribet, G., G. D. Edgecombe, J. M. Carpenter, C. D'Haese, and W. C. Wheeler. 2004. Is Ellipura monophyletic? A combined analysis of basal hexapod relationships with emphasis on the origin of insects. Org. Divers. Evol. 4:319–340.
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., S. Richter, G. D. Edgecombe, W. C. Wheeler. 2005. The position of crustaceans within the Arthropoda—evidence from nine molecular loci and morphology. Crustac. Issues 16:307–352.
Gloor, G. B., C. R. Preston, D. M. Johnson-Schlitz, N. A. Nassif, R. W. Phillis, W. K. Benz, H. M. Robertson, and W. R. Engels. 1993. Type I repressors of P element mobility. Genetics 135:81–95.
Gutell, R. R. 1994. Collection of small subunit (16S- and 16S-like) ribosomal RNA structures. Nucleic Acids Res. 22:3502–3507.
Gutell, R. R., M. W. Gray, and M. N. Schnare. 1993. A compilation of large subunit (23S and 23S-like) ribosomal RNA structures: 1993. Nucleic Acids Res. 21:3055–3074.
Hancock, J. M., D. Tautz, and G. A. Dover. 1988. Evolution of the secondary structures and compensatory mutations of the ribosomal RNAs of Drosophila melanogaster. Mol. Biol. Evol. 5:393–414.
Hendricks, L., C. Van Broeckhoven, A. Vandenberghe, Y. Van de Peer, and R. De Wachter. 1988. Primary and secondary structure of the 18S ribosomal RNA of the bird spider Eurypelma californica and evolutionary relationships among eukaryotic phyla. Eur. J. Biochem. 177:15–20.
Hennig, W. 1953. Kritische bemerkungen zum phylogenetischen system der insekten. Beitr. Entomol. 3 (Sonderheft):1–85.
———. 1981. Insect phylogeny. John Wiley & Sons, New York.
Hillis, D. M., and M. T. Dixon. 1991. Ribosomal DNA: molecular evolution and phylogenetic inference. Q. Rev. Biol. 66:411–453.
Huelsenbeck, J. P., D. M. Hillis, and R. Jones. 1996. Parametric bootstrapping in molecular phylogenetics: applications and performance. Pp. 19–45 in J. E. Ferraris, and S. R. Palumbi, eds. Molecular zoology: advances, strategies, and protocols. Wiley-Liss, New York.
Huelsenbeck, J. P., and F. Ronquist. 2001. MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics 17:754–755.
Ikeda, Y., and R. Machida. 1998. Embryogenesis of the dipluran Lepidocampa weberi Oudemans (Hexapoda, Diplura, Campodeidae): external morphology. J. Morphol. 237:101–115.
Jermin, L. S., S. Y. W. Ho, F. Ababneh, J. Robinson, and A. W. D. Larkum. 2004. The biasing effect of compositional heterogeneity on phylogenetic estimates may be underestimated. Syst. Biol. 53:638–643.
Kjer, K. M. 2004. Aligned 18S and insect phylogeny. Syst. Biol. 53:506–514.
Koch, M. 1997. Monophyly and phylogenetic position of the Diplura (Hexapoda). Pedobiologia 41:9–12.
Kristensen, N. P. 1981. Phylogeny of insect orders. Annu. Rev. Entomol. 26:135–157.
———. 1991. Phylogeny of extant hexapods. Pp.125–140 in I. D. Naumann, ed. The insects of Australia, Vol. 1, 2nd edition. Melbourne University Press, Carlton, Australia.
Kukalová-Peck, J. 1987. New Carboniferous Diplura, Monura, and Thysanura, the hexapod ground plan, and the role of thoracic side lobes in the origin of wings (Insecta). Can. J. Zool. 65:2327–2345.
———. 1991. Fossil history and the evolution of hexapod structures. Pp. 141–179 in Naumann, I. D., ed. The insects of Australia, Vol. 1, 2nd edition. Melbourne University Press, Carlton, Australia.
Lake, J. A. 1994. Reconstructing evolutionary trees from DNA and protein sequences: paralinear distances. Proc. Natl. Acad. Sci. USA 91:1455–1459.
Lockhart, P. J., M. A. Steel, M. D. Hendy, and D. Penny. 1994. Recovering evolutionary trees under a more realistic model of sequence evolution. Mol. Biol. Evol. 11:605–612.
Luan, Y. X., Y. G. Yao, R. D. Xie, Y. M. Yang, Y. P. Zhang, and W. Y. Yin. 2004. Analysis of 18S rRNA gene of Octostigma sinensis (Projapygoidea: Octostigmatidae) supports the monophyly of Diplura. Pedobiologia 48:453–459.
Luan, Y. X., Y. P. Zhang, Q. Y. Yue, J. F. Pang, R. D. Xie, and W. Y. Yin. 2003. Ribosomal DNA gene and phylogenetic relationship of Diplura and lower hexapods. Sci. China Ser. C 46:67–76.
Mallatt, J., J. R. Garey, and J. W. Shultz, 2004. Ecdysozoan phylogeny and Bayesian inference: first use of nearly complete 28S and 18S rRNA gene sequences to classify the arthropods and their kin. Mol. Phylogenet. Evol. 31:178–191.
Mallatt, J., and J. Sullivan. 1998. 28S and 18S rDNA sequences support the monophyly of lampreys and hagfishes. Mol. Biol. Evol. 15:1706–1718.
Mallatt, J., and C. J. Winchell. 2002. Testing the new animal phylogeny: first use of large-subunit and small-subunit rRNA gene sequences to classify the protostomes. Mol. Biol. Evol. 19:289–301.
Manton, S. M. 1972. The evolution of arthropodan locomotory mechanisms. Part 10. Locomotory habits, morphology and evolution of the hexapod classes. Zool. J. Linn. Soc. 51:203–400.
———. 1977. The Arthropoda: habits, functional morphology, and evolution. Oxford University Press, Oxford.
Nardi, F., G. Spinsanti, J. L. Boore, A. Carapelli, R. Dallai, and F. Frati. 2003. Hexapoda origins: monophyletic or paraphyletic? Science 299:1887–1889.
Nelles, L., B. L. Fang, G. Volckaert, A. Vandenberghe, and R. De Wachter. 1984. Nucleotide sequence of a crustacean 18S ribosomal RNA gene and secondary structure of eukaryotic small subunit ribosomal RNAs. Nucleic Acids Res. 12:8749–8768.
Pagés, J. 1997. Notes sur les Diploures Rhabdoures Insectes, Apterygotes no.1—Diplura Genavensia XXII. Rev. Suisse Zool. 104:869–896.
Posada, D., and T. R. Buckley. 2004. Model selection and model averaging in phylogenetics: advantages of the AIC and Bayesian approaches over likelihood ratio tests. Syst. Biol. 53:793–808.
Posada, D., and K. A. Crandall. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14:817–818.
Rambaut, A., and N. C. Grassly. 1997. Seq-Gen sequence: An application for the Monte Carlo simulation of DNA sequence evolution along phylogenetic trees. Comput. Appl. Biosci. 13:235–238.
Regier, J. C., and J. W. Shultz. 1997. Molecular phylogeny of the major arthropod groups indicates polyphyly of crustaceans and a new hypothesis for the origin of hexapods. Mol. Biol. Evol. 14:902–913.
———. 2001. Elongation factor-2: a useful gene for arthropod phylogenetics. Mol. Phylogenet. Evol. 20:136–148.
Regier, J. C., J. W. Shultz, and R. E. Kambic. 2004. Phylogeny of basal hexapod lineages and estimates of divergence times. Ann. Entomol. Soc. Am. 97:411–419.
Regier, J. C., H. M. Wilson, and J. W. Shultz. 2005. Phylogenetic analysis of Myriapoda using three nuclear protein-coding genes. Mol. Phylogenet. Evol. 34:147–158.
Ruppert, E. E., R. S. Fox, and R. D. Barnes. 2004. Invertebrate zoology, a functional evolutionary approach. Thomson, Brooks/Cole, Belmont, Calif.
Rusek, J. 1982. Octostigma herbivora n. gen. & sp. (Diplura: Projapygoidea: Octostigmatidae n. fam.) injuring plant roots in the Tonga Islands. N. Z. J. Zool. 9:25–32 .
Schnare, M. N., S. H. Damberger, M. W. Gray, and R. Gutell. 1996. Comprehensive comparison of structural characteristics in eukaryotic cytoplasmic large subunit (23S-like) ribosomal RNA. J. Mol. Biol. 256:701–719.
Shao, H. G., Y. P. Zhang, X. Ke, Q. Y. Yue, and W. Y. Yin. 2000. Sequences of mitochondrial DNA cytochrome oxidase II in Cryptopygus nanjiensis and phylogeny of Apterygota. Sci. China Ser. C 43:589–596.
Shao, H. G., Y. P. Zhang, R. D. Xie, and W. Y. Yin. 1999. Mitochondria cytochrome b sequences variation of Protura and molecular systematics of Apterygota. Chin. Sci. Bull. 44:2031–2036.
Shultz, J. W., and J. C. Regier. 2000. Phylogenetic analysis of arthropods using two nuclear protein-encoding genes supports a crustacean + hexapod clade. Proc. R. Soc. B 267:1011–1019.
Spears, T., and L. G. Abele. 1997. Crustacean phylogeny inferred from 18S rDNA. Pp. 169–187 in R. A. Fortey, and R. H. Thomas, eds. Arthropod relationships. Chapman & Hall, London.
tys, P., and S. Bilinski. 1990. Ovariole types and the phylogeny of hexapods. Biol. Rev. 65:401–429.
tys, P., J. Zrzav, and F. Weyda. 1993. Phylogeny of the Hexapoda and ovarian metamerism. Biol. Rev. 68:365–379.
Swofford, D. L. 2002. PAUP*: phylogenetic analysis using parsimony (*and other methods). Version 4.0b10. Sinauer Associates, Sunderland, Mass.
Swofford, D. L., G. J. Olsen, P. J. Waddell, and D. M. Hillis. 1996. Phylogenetic inference. Pp. 407–514 in D. M. Hillis, C. Moritz, and B. K. Mable, eds. Molecular systematics, 2nd edition. Sinauer Associates, Sunderland, Mass.
Turbeville, J. M., D. M. Pfeifer, K. G. Field, and R. A. Raff. 1991. The phylogenetic status of arthropods, as inferred from 18S rRNA sequences. Mol. Biol. Evol. 8:669–686.
Tuxen, S. L. 1964. The Protura. A revision of the species of the world with keys for determination. Hermann, Paris.
Whalley, P. 1995. Unfair to ancient fossil springtails! Antenna 19:2–3.
Wheeler, W. C., P. Cartwright, and C. Y. Hayashi. 1993. Arthropod phylogeny: a combined approach. Cladistics 9:1–39.
Wheeler, W. C., M. Whiting, Q. D. Wheeler, and J. M. Carpenter. 2001. The phylogeny of the extant Hexapod orders. Cladistics 17:113–169.
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.
Winchell, C. J., A. P. Martin, and J. Mallatt. 2004. Phylogeny of elasmobranchs based on LSU and SSU ribosomal RNA genes. Mol. Phylogenet. Evol. 31:214–224.
Winchell, C. J., J. Sullivan, C. B. Cameron, B. J. Swalla, and J. Mallatt. 2002. Evaluating hypotheses of deuterostome phylogeny and chordate evolution with new LSU and SSU ribosomal DNA data. Mol. Biol. Evol. 19:762–776.
Yin, W. Y. 1984. A new idea on phylogeny of Protura with approach to its origin and systematic position. Sci. Sin. Ser. B XXVII:149–160.
———. 1996. New considerations on systematics of Protura. Pp. 60 in Proceedings of XX International Congress of Entomology, Firenze, Italy.
Yin, W. Y., R. D. Xie, Y. M. Yang, Q. Y. Yue, and Y. X. Luan. 2002. Analysis of the main characters for regrouping of the class Protura (Hexapoda). Acta Zootaxonomica Sin. 27:649–658 .
Yin, W. Y., and L. Z. Xue. 1993. Comparative spermatology of Protura and its significance on proturan systematics. Sci. China Ser. B 36:575–587.
Zrzav, J., S. Mihulka, P. Kepka, A. Bezdk, and D. Tietz. 1998. Phylogeny of the Metazoa based on morphological and 18S ribosomal DNA evidence. Cladistics 14:249–285.(Yun-xia Luan*, Jon M. Mal)