2006 (2007). The Journal of Arachnology 34:557–565

UNUSUALLY LONG (ARANEAE, ) SEQUENCE FOR SMALL SUBUNIT (18S) RIBOSOMAL RNA SUPPORTS SECONDARY STRUCTURE MODEL UTILITY IN

J. C. Spagna1 and R. G. Gillespie: Division of Insect Biology, Department of Environmental Science, Policy and Management, 137 Mulford Hall, UC Berkeley, Berkeley, California 94720, USA. E-mail: [email protected]

ABSTRACT. We report on the structure of the small-subunit ribosomal RNA (18S rRNA) sequence from Hyptiotes gertschi (Araneae, Uloboridae), which is the largest 18S gene sequenced in any to date. We compare this remarkable sequence to those from a range of other spiders and , and develop base-pairing models of its insert regions to determine its overall secondary structure. The H. gertschi sequence of 1902 bases is 86 nucleotides longer than any comparable sequence and contains 5 inserts between 5 and 28 bases in length, all at regions characterized as among the most variable in eukaryotic 18S genes. Inserts were also found in one of these variable regions in published sequences of 3 species of hard ticks (Acari, Ixodidae). Other arachnid taxa were remarkably uniform in 18S primary sequence length, ranging from 1802 to 1816 nucleotides. Thermodynamic modeling of the H. gertschi inserts suggests they are largely self-complementary, extending the stem portions of the variable regions. Keywords: , arachnids, Acari, gene inserts

The small subunit ribosomal RNA, or 18S at which different parts of the primary se- rRNA, is one of a small set of commonly used quence vary. sequences for molecular phylogenetic recon- The genes coding for ribosomal RNAs con- struction of relationships (reviewed tain regions that can accumulate and lose ba- in Caterino et al. 2000). The gene coding for ses through insertion and deletion events (in- the 18S rRNA contains sections that are high- dels) more easily than protein-coding genes, ly conserved, and these provide informative which are constrained by the requirement that characters for the assessment of relationships they maintain an open reading frame for prop- between distantly related taxa, such as meta- er translation into a functional primary amino zoan relationships (Giribet & Wheeler 2001; acid sequence. Indels can change the length of Mallatt et al. 2004). The 18S rRNA has also the gene and make homology assessments of been used in studies of divergence between individual base-pairs, and often long stretches arachnid orders, as well as studies of diver- of base-pairs, difficult. The proper methodo- gence between spider genera and species logical approach to this sequence-alignment (Wheeler & Hayashi 1998; Arnedo et al. problem is a source of controversy in phylo- 2004). In the context of arachnid molecular genetics, and opinions range from using static phylogenetics, to understand both the poten- alignments, with regions that are difficult to tial utility and drawbacks in the use of any align being either included or discarded (e.g., genetic marker, it is important to have knowl- Nardi et al. 2003); to using direct optimization edge of the amount of variation in both the (Wheeler 1996), which avoids the arbitrary re- primary sequence and secondary structures moval of data and possible evolutionary sig- across taxa at different levels. This is because nal and avoids the problem of multiple-se- the secondary structure can influence the rate quence alignment altogether. An additional 1 Current address: Dept. of Entomology, University feature of direct optimization, as implemented of Illinois Urbana-Champaign, 320 Morrill Hall, in the program POY (Wheeler 1999) is the 505 S. Godwin Ave., Urbana, Illinois 61801. ability to treat inserts as multistate characters E-mail: [email protected] with as many states as there are unique insert

557 558 THE JOURNAL OF ARACHNOLOGY sequences, with a matrix of character-state hard ticks (Acari, Ixodidae) that demonstrate transformation costs. See Giribet & Wheeler the evolutionary conservation of the overall (2001) for an example of implementation of structure of the arachnid 18s gene. the latter method. A very different approach, and one that has METHODS been promoted by many authors, is the use of Taxon sampling.—We sampled exemplar secondary structure models to aid in homol- sequences from all arachnid orders and major ogy assessment and relative weighting (Dixon spider lineages for which at least one full se- & Hillis 1993; Kjer 2004). Recently, it has quence of the entire 18S gene was available been proposed that elements of rRNA second- in GenBank (http://www.ncbi.nlm.nih.gov/ ary structure can provide a basis for choice Genbank/). To sample unusually long arachnid between multiple models to be used in a par- 18S sequences thoroughly, we extended our titioned Bayesian analysis (Telford et al. search to include all ‘‘partial sequences’’ that 2005). For this technique, bases from a variety came within 90 bases (ϳ5% of the gene of taxa would be compared to a secondary length) of the primer regions, since this region structure model for the gene for assignment to is highly conserved across and the ‘‘stem’’ and ‘‘loop’’ (and possibly transitional missing length could be estimated despite lack or ambiguous) partitions. These partitions of primary sequence. In taxa where 18S gene would be individually tested for the most ap- length was uniform and in the ϳ1800 base propriate likelihood models, then subsequent- pair (ϭ bp) length typical of arachnids, one ly analyzed using the partitioned, multiple- individual was sampled, while all complete or model approach. nearly-complete sequences with large (Ͼ 3 The RNAs produced by the 18S genes form bp) inserts relative to the Aphonopelma sp. se- secondary structure by base-pairing with their quence were included in the sample, since a own complementary stretches of sequence, complete secondary structure model is avail- forming helical structures commonly referred able for this taxon (Hendriks et al. 1988). to as stems, while the non-pairing regions Among spiders, lineages included represen- form loop structures that connect multiple tatives from (Aphonopelma stems, or form the terminal turn in the self- sp., Theraphosidae), (Liphistius complementary stems. Because the overall bicoloripes, Liphistiidae), Orbicularia (H. secondary structure is functionally important gertschi, Uloboridae), ‘‘derived orb-weavers’’ for the ribosome, and because two comple- (sensu Coddington 1990), (Nesticus cellulan- mentary changes in primary sequence are re- us, Nesticidae) and the RTA (retrolateral tibial quired to maintain a stem structure, it has been apophysis) (sensu Coddington et al. proposed that the stems should be treated dif- 2004) (Coelotes terrestris, Amaurobiidae) ferently in phylogenetic analysis than the less- (see Table 1). All spider sequences were from constrained loops (Dixon & Hillis 1993). To Genbank except for the H. gertschi data, for do so requires the ability to tell whether nu- which new sequence data were generated in cleotides are in a loop or a stem structure, the current study. Specimens of H. gertschi which can be modeled using computer algo- were collected by S. Lew at the Angelo Re- rithms that develop secondary structures serve, Mendocino County (39Њ43ЈN, based on comparative or thermodynamic in- 123Њ39ЈW), and from Del Norte County near formation (reviewed in Gardner & Giegerich the Oregon border (41Њ59ЈN, 123Њ43ЈW), Cal- 2004). ifornia, USA, in May 2003. Voucher speci- Here we present a comparison of 18S struc- mens are stored in the collection of the Essig tures sampled from arachnid orders and spider Museum of Entomology, UC Berkeley, under lineages to assess the utility of secondary the voucher codes EMEC50654 and structure information currently available for EMEC50993, respectively. making homology judgments across arachnid Molecular methods.—DNA was extracted taxa, and for determining data partitions used from two legs from each of the H. gertschi in model-based analyses. Included are new specimens using a Qiagen DNeasy Tissue data from the species Hyptiotes gertschi Kit’s standard protocol. The 18S gene was di- Chamberlin & Ivie 1935 (Araneae, Ulobori- rectly PCR-amplified in three parts using dae), and remarkably similar sequences from primer pairs 1F-5R, 5F-9R, and 3F-7R (after SPAGNA & GILLESPIE—HYPTIOTES 18S SECONDARY STRUCTURE 559

Table 1.—Taxa examined, with total gene length and p-distance between primary sequence of individ- uals and Aphonopelma sp. reference sequence. Length includes terminal primer regions (approximately 50 base pairs (ϭ bp); 23 per primer ϩ 4 bp downstream in the 3Ј direction from primer 9R). * indicates partial sequences (see text), with lengths estimated by assuming uniform sequence length relative to reference sequence at missing terminal regions.

Genbank Total Uncorrected Accession length p-distance to Order (lineage represented) Taxon sampled Number (bp) reference Araneae (Orbicularia) Hyptiotes gertschi Chamber- DQ015708 1902 10.8% lin & Ivie 1935 Araneae (Mygalomorphae) Aphonopelma sp. X13457 1814 — Araneae (Mesothelae) Liphistius bicoloripes Ono AF007104 1808 2.4% 1988 Araneae (derived orb-weavers) Nesticus cellulanus (Clerck AF005447 1816 7.8% 1757) Araneae (RTA clade) Coelotes terrestris (Wider AJ007986 1814 5.7% 1834) Opiliones Odiellus troguloides (Lucas X81441 1810 6.2% 1847) Scorpiones Androctonus australis (Lin- X77908 1812 6.7% naeus 1758) Pseudoscorpiones Roncus cf. pugnax (Nava´s AF05443 1808 11.0% 1918) Acari (Ixodidae) Amblyomma glauerti Keir- AF115372 1802 10.3% ans, King & Sharrad, 1994 Solifugae Eusimonia wunderlichi Pie- U29492 1802 7.7% per 1977 Amblypygi Paraphrynus sp. AF005445 1810 4.8% Uropygi Mastigoproctus giganteus AF005446 1810 4.9% (Lucas 1835) Schizomida Stenochrus portoricensis AF005444 1809 5.5% Chamberlin 1922 Ricinulei Pseudocellus pearsei (Cham- U91489 1813 5.1% berlin & Ivie 1938) Palpigradi Eukoenenia sp. AF207648 1810 7.4% Acari (Ixodidae) Sejus sp. AF287237 1880* 21.1% Acari (Ixodidae) Lohmannia sp. AF287234 1887* 12.3% Acari (Ixodidae) Alicorhagidia sp. AF022024 1872* 13.8%

Giribet et al. 1999), to amplify fragments of propriate buffers to stabilize the PCR reaction, length ϳ950 bp, ϳ850 bp, and ϳ1000 bp, which improved yield and quality of PCR covering the first half, second half, and an products. Molecular extraction vouchers were overlapping central portion of the gene re- stored at Ϫ80Њ C. spectively. The PCR protocol (modified from Gene size, purity and concentration were Hedin & Maddison 2001) used 35 cycles of assessed by running out a portion of the PCR 30 s at 95Њ C melting temperature, followed product on a 1.5% TBE/agarose gel. PCR by 30 s at an annealing temperature of 52Њ C, products were cleaned using Qiagen QiaQuick followed by an extension step of 45 s at 72Њ PCR Purification Kit, and cycle sequenced in C, with 3 s added to this extension for every both directions using dye terminators (after cycle after the first. In addition, these cycles Sanger et. al 1977). Cycle sequencing prod- were preceded by an initial melting at 95Њ C ucts were analyzed using an ABI 3730 capil- for 2 min, with a 7 min final extension at 72Њ lary autosequencing machine. Individual se- C. A MasterAmp PCR Optimization Kit (Ep- quences checked against their complementary icentre Technologies) was used to choose ap- sequences, using Sequencher 3.1.1 (Gene 560 THE JOURNAL OF ARACHNOLOGY

Codes Corporation). This program was also model using the same method, then calculated used to assemble contigs of all three overlap- the percentage of base-pairs and loop mem- ping primer regions for each H. gertschi spec- bers in the algorithm output that correctly imen, to create a single sequence for each. Ad- matched the model. ditional sequences (see Table 1 for accession RESULTS numbers) were acquired from GenBank for Comparisons of 18S sequence length and comparison. primary sequence divergence can be seen in Alignment and Comparisons.—Using Table 1. Most arachnid 18S sequences are in Clustal X version 1.83 (Chenna et al. 2003), the range of 1800–1810 base pairs (including a static alignment of multiple sequences was primer regions), with the H. gertschi and tick created using the default settings (gap inser- sequences being the exceptions (greater by 85 tion cost 15, gap extension cost 6.66, transi- and 56–64 bases, respectively). Figure 1 tion cost 0.5 times transversion cost). Treating shows a schematic of the Hendriks et. al the Aphonopelma sp. sequence and model as (1988) model of spider 18S, with regions with references, gaps in the aligned sequences were H. gertschi and tick inserts marked with black compared to the secondary structure model to arrows and a hatched bar, respectively. detect regions where insertions and deletions Hyptiotes gertschi and tick insertions.— have occurred. The two H. gertschi sequences are identical, Modeling of insert regions in H. gert- as are sequences from an additional set of am- schi.—In areas with insertions Ͼ 3 bp relative plifications of the Del Norte County specimen. to the Aphonopelma sp. reference sequence, However, in contrast to the marked similarity secondary structures were modeled using the in length and structure across the arachnid ex- web-interface version of the RNAfold pro- emplars from GenBank, the H. gertschi se- gram in the Vienna Package (http://www.tbi. quences are 86 nucleotides longer than the univie.ac.at/ϳivo/RNA/; Hofacker 2003). This next-longest spider 18S (Nesticus cellulanus), program uses a dynamic-programming algo- and we estimate it to be 21 nucleotides longer rithm with a variety of parameters to estimate than the next-longest arachnid 18S (Sejus sp., the secondary structure based on minimizing Acari, Ixodidae). This additional sequence is the free energy of the possible stem-loop found in two large inserts—a 28-base exten- structures from the primary sequence. The pa- sion of the helical arm E10-1 (Figure 2a), and rameters used include RNA base-pairing en- 27-base extension of E10-2 (Fig. 2b)—and the ergies plus a variety of experimentally deter- rest is found as smaller inserts of 5, 6, and 13 mined adjustments to this cost, including nucleotides in helices 6, 41, and 47 respec- energy estimates for loops of various sizes and tively (Figs. 3a, b, and c). Three near-com- locations, and for single- and double-stranded plete tick sequences were also found to have multi-base motifs known to affect local sta- large inserts in the terminal arms of branch 10 bility (Mathews et al. 1999). (see Table 1). The expanded regions of the H. gertschi Models of the H. gertschi inserts based on 18S gene were modeled using the primary se- free-energy minimization show that the in- quence of the insert region plus the four com- serted sequences extend the helices in all of plementary bases at each insertion site that the cases except structure 6, where the helical could be homologized with the reference se- stem is shortened and the terminal loop ex- quence and structure from Aphonopelma sp. panded. The RNAfold program also correctly Because the RNAfold program had difficulty replicated 90% (40 of 48) of the stem base- aligning the complementary 4-base-pair ends pairs in the Hendriks et al. (1988) model in properly in some cases, we added a comple- the arms in which the inserts were found, mentary 5-nucleotide extension at each ter- which is comparable with the 73% base-pair- minus to anchor the ends of the sequence and ing accuracy for this parameter set calculated stand in for the rest of the structure, so that by Mathews et al. (1999) using a variety of each primary sequence input took the form: other well-characterized genes. 5Ј-GGGGG- primary sequence -CCCCC-3Ј. We also tested the RNAfold program’s ability DISCUSSION to predict the structures in the variable arms The most remarkable results from the cur- of the Hendriks et al. (1988) Aphonopelma sp. rent study were the number of sizable inser- SPAGNA & GILLESPIE—HYPTIOTES 18S SECONDARY STRUCTURE 561

Figure 1.—General model for secondary structure of arachnid 18S ribosomal DNA. Small numbers refer to general stem-loop numbering system after (Nelles et al. 1984). Those features preceded by an ‘‘E’’ are specific to eukaryotic 18S sequences. Arrows and bold numbers refer to locations and size (in bases relative to tarantula model) of H. gertschi inserts, respectively. Hatched bar represents region with large inserts found in 3 tick taxa. tions in the H. gertschi sequences and, in par- able across a range of taxa (Van de Peer ticular, the shared expansion of stem E10 1996). The RNAfold models show that much found in both H. gertschi and ticks. In H. gert- of the inserted H. gertschi sequence is self- schi all insertions took place in known ‘‘var- complementary and these insertions function iable regions’’ of the eukaryote 18S structure; to extend the stem regions, rather than in- helices 6, E10-1 and -2, 41, and 47 belong to crease the loop size, except in Helix 6. This regions that have been characterized as vari- extension and maintenance of complementary 562 THE JOURNAL OF ARACHNOLOGY

Figures 2–4.—Region of arachnid 18S gene with unusually large inserts. 2. Basic structure for helices 10, E10-1 and E10-2 in arachnids. Primary se- quence and model after Hendriks et al. (1988); 3. Expanded E10-2 stem-loop structure in H. gertschi; 4. Expanded E10-1 stem loop in H. gertschi. Boxes represent areas of homology where expanded arms Figures 5–7.—Smaller inserts and secondary attach to conserved, adjacent parts of the structure. structure model comparisons. 5. Structure 6 with 5 base pair insertion in H. gertschi; 6. Structure 41 with 6 inserted bases in H. gertschi; 7. Structure 47 with 13 inserted bases in H. gertschi. stems has remarkable parallels with the find- ings of Hancock & Vogler (2000), who re- vealed a similar pattern of complementary not a pseudogene or an experimental artifact. stem expansion in the evolution of hypervari- One would expect to see decay of pseudogene able regions of 18S in tiger beetles. sequence relative to the selectively-con- A variety of factors suggest the H. gertschi strained functional gene (Giribet & Wheeler sequence represents the functional 18S rRNA 2001), and this decay (in the form of random gene rather than a pseudogene copy or exper- mutations) should be equally distributed imental artifact. The presence of insertions in throughout the sequence rather than restricted areas amplified by three different primer pairs, to known variable regions. Though some the replication of the sequence in individuals metazoans, such as helminths, are known to from two populations, the maintenance (and, carry two functional copies of the 18S gene in some cases, increase in length) of the base- with differing sequences (Carranza et al. pairing in stem structures, and the non-ran- 1996), we found only single bands in our aga- dom, self-complementary sequence found in rose gel visualizations of all three PCR prod- both large and small inserts, make it likely ucts, and no second allele or double-peak pat- that we have sequenced a functional gene and terns were seen in the raw sequence data, SPAGNA & GILLESPIE—HYPTIOTES 18S SECONDARY STRUCTURE 563 despite repeating the amplification and se- While the RNA folding algorithm used may quencing processes, suggesting the existence not predict intra-molecular folding dynamics of a single, uniform 18s sequence in H. gert- of rRNA accurately, it provides a repeatable schi. method for producing a plausible structure for How exceptional is the H. gertschi 18S sequence data which lacks obvious homologs rRNA? It seems to be an anomaly for spiders, from related taxa for comparison. The algo- which, based on those sampled in the current rithm also gives a reasonable basis for judging study, generally vary little even between the whether an insert extends a helix, a loop, or most distantly related lineages. Along with the both, particularly with short sequences such as tick data, these anomalous sequences show those modeled in this study. With changes af- that the presence of large inserts in arachnids, fecting sequence length concentrated in the though rare, is consistent with a general eu- terminal ends of known stem-loop structures, karyotic 18S evolutionary tendency—overall the H. gertschi and tick 18S genes appear to structure is maintained, while rare inserts ap- be the exceptions that prove the rule; bases pear in known variable areas. It is worth not- come and go, albeit very rarely, but the helical ing that only the H. gertschi gene contains backbone and location of stem and loop struc- expansions of several variable regions while tures of the small subunit ribosomal RNA the tick inserts are restricted to the E10 region. have remained conserved throughout arachnid With only two populations of H. gertschi evolution. sampled in the current study, we cannot es- Because the expanded sequence has only tablish the phylogenetic distribution of the been found in a single spider species, it is im- 18S rDNA extension, whether it occurs in oth- possible to argue that this is evidence of any er Hyptiotes species, other uloborid genera, or trend in 18S evolution in spiders. However, their sister group, the . Such ex- the increase in size of the 18S gene seen here tensions have not been described from other differs from the documented trend (relative to Orbicularia, the sister-group of the deinopoid other metazoans) toward reduction in size in (Uloboridae ϩ Deinopidae) clade, but may be the hypervariable region of the mitochondrial more widespread within the deinopoids. The 16S gene (Smith & Bond 2003) and in the large number of inserts found in H. gertschi arms of some tRNA genes (Masta & Boore in this study suggests that there might be var- 2004) of spiders. It should be noted that the iation in number and size of inserts found in cited trends are in mitochondrial genes, different taxa, since it seems unlikely that all whereas 18S is part of the nuclear genome, these inserts appeared simultaneously. It and the two genomes may be subject to dif- would be interesting to map any such varia- ferent selective pressures or constraints. tion onto the existing phylogeny of Deinopoid To understand the broader pattern of 18S genera (Coddington 1986) to determine the size change, it is useful to look at the evolu- pattern of evolution of the 18S gene in greater tion of the 18S gene in other metazoan taxa. detail. Alternatively, if variation seems to be Giribet & Wheeler (2001) showed that the limited to a subset of Hyptiotes or a small general pattern of 18S evolution across a number of uloborid genera, the inserts them- much broader sampling of taxa, including selves could be evaluated as phylogenetic hexapods, chelicerates, myriapods, and crus- characters within these groups. taceans, is one of conserved length in the Once the taxonomic range and amount of 1800 bp range, with occasional increases in variation of these inserts has been character- size. Deletion events appear to be exceedingly ized, hypotheses about causal biological, rare. The data presented here are in keeping physiological or ecological factors allowing with those findings, although the total length such inserts to occur can be tested rigorously. for the H. gertschi 18S gene is greater than Comparative studies could also be performed that of any of the 49 chelicerate taxa reported with the wide range of metazoan taxa which by Giribet & Wheeler (2001) or of any full show similar inserts, such as myriapods, pro- arachnid 18S sequence currently found in the turans, and helminth worms, to find general Genbank database. Complete or nearly-com- correlations between 18S size and phenotypes plete sequences are required for secondary (Carranza et al. 1996; Giribet & Wheeler structure modeling since there is long-range 2001). complementary base-pairing in 18S secondary 564 THE JOURNAL OF ARACHNOLOGY structure (Telford et al. 2005). Unfortunately gertschi specimens, to K. Rudolph for labo- for the pursuit of whole-gene comparisons and ratory assistance, and to S. Crews and S. structural modeling, many arachnid studies Longhorn for comments on the text. This have used only half of the 18S gene, (e.g., work was partially supported by grants from Wheeler & Hayashi 1998; Arnedo et al. the NSF, the Walker Fund for Systematics and 2004), and there is a sizable and important the Vincent Roth Fund for Spider Systematics. spider lineage, the haplogynae, for which no LITERATURE CITED complete 18S sequence is available. As for the overall utility of 18S data for Arnedo M.A., J. Coddington., I. Agnarsson & R.G. Gillespie. 2004. From a comb to a tree: phylo- arachnid phylogeny, this data set shows that genetic relationships of the comb-footed spiders the Hendriks et al. (1988) model is sufficient (Araneae, Theridiidae) inferred from nuclear and for locating structural changes in the 18S gene mitochondrial genes. for all known arachnid genotypes, though the and Evolution 31:225–245. value of modeling based solely on primary se- Carranza, S., G. Giribet, C. Ribera, N. Baguna & quence using thermodynamic predictions on a M. Riutort. 1996. Evidence that two types of 18S single taxon is limited. In the majority of cas- rDNA coexist in the genome of Dugesia es, alignment is trivial with a mean of less (Schmidtea) meditarranea (Platyhelminthes, Tur- than 1% sequence length divergence between bellaria, Tricladida). Molecular Biology and most taxa. This similarity in sequence-length Evolution 13:824–832. across Araneae is also corroborated by data Caterino, M. S., S. Cho & F.A.H. Sperling. 2000. from a family-level study of the RTA clade by The current state of insect molecular systematics: a thriving Tower of Babel. Annual Review of the authors (unpublished data). For data par- Entomology 45:1–54. titioning, relative-weighting, and model- Chenna, R., H. Sugawara, T. Koike, R. Lopez, T. J. choice purposes, the number and locations of Gibson, D.G. Higgins & J.D. Thompson. 2003. the stem-loop structures of the 18S gene re- Multiple sequence alignment with the Clustal se- main largely identical to that of most se- ries of programs. Nucleic Acids Research 31: quenced eukaryotes. Determination of stem 3497–3500. and loop regions should be achievable with Coddington, J.A. 1986. The monophyletic origin of some confidence for the large majority of the orb web. Pp. 319–363. In Spiders: Webs, Be- arachnid cases, using the Hendriks et al. havior, and Evolution. (W.A. Shear, ed.). Stan- (1988) model as a guide. ford University Press, Stanford, California. In exceptional situations where homology Coddington, J.A. 1990. Cladistics and spider clas- sification: araneomorph phylogeny and the assessments or stem-loop status of individual of orbweavers (Araneae: Araneo- bases are difficult, such as within the tick-spe- morphae; ). Acta Zoologica Fennica cific inserts seen here, removing the insert 190:75–87. data may be defensible, for two reasons: first, Coddington, J.A., G. Giribet, M.S. Harvey, L. Pren- the tendency of insertions to occur indepen- dini & D.E. Walter. 2004. Arachnida. Pp. 296– dently in the same areas across widely diver- 318. In Assembling the Tree of Life. (J. Cracraft gent taxa (such as spiders and ticks having & M. Donoghue, eds.). Oxford University Press, inserts in the E10 region) could plausibly add New York. homoplasy to a data matrix and place taxa in Dixon, M.T. & D.M. Hillis. 1993. Ribosomal RNA incorrect groupings. Second, because the in- secondary structure: compensatory mutations sertions tend to be small relative to the more and implications for phylogenetic analysis. Mo- lecular Biology and Evolution 10:256–267. easily homologized portions of the rRNA Gardner, P.P. & R. Giegerich. 2004. A comprehen- (much smaller, for instance, than the Ͼ 200 sive comparison of comparative RNA structure bp insertions found in many myriapods, see prediction approaches. BMC Bioinformatics 5: Giribet & Wheeler 2001), and only occur in a 140. small number of taxa, there is likely to be lit- Giribet G., S. Carranza, M. Riutort, J. Baguna & C. tle reduction in phylogenetic signal from the Ribera. 1999. Internal phylogeny of the Chilop- sequence if an insert region is excluded from oda (Myriapoda, Arthropoda) using complete analysis. 18S rDNA and partial 28S rDNA sequences. Philosophical Transactions of the Royal Society ACKNOWLEDGMENTS of London Series B: Biological Sciences 29:215– Thanks to S. Lew and the Essig Museum 222. of Entomology for the loan of the Hyptiotes Giribet, G. & W.C. Wheeler. 2001. Some unusual SPAGNA & GILLESPIE—HYPTIOTES 18S SECONDARY STRUCTURE 565

small-subunit ribosomal DNA sequences of Nardi, F., G. Spinsanti, J.L. Boore, A. Carapelli, R. metazoans. American Museum Novitates 3337: Dallai & F. Frati. 2003. Hexapod origins: mono- 1–14. phyletic or paraphyletic? Science 299:1887– Hancock, J. M. & A.P. Vogler. 2000. How slippage- 1889. derived sequences are incorporated into rRNA Nelles, L., B.L. Fang, G. Volckaert, A. Vanden- variable-region secondary structure: implications berghe & R. De Wachter. 1984. Nucleotide se- for phylogeny reconstruction. Molecular Phylo- quence of a crustacean 18S ribosomal RNA gene genetics and Evolution 14:366–374. and secondary structure of eukaryotic small sub- Hedin, M.C. & W.P. Maddison. 2001. A combined unit ribosomal RNAs. Nucleic Acids Research molecular approach to phylogeny of the jumping 12:8749–8768. spider subfamily Dendryphantinae (Araneae: Sanger, F., S. Nicklen & A.R. Coulson. 1977. DNA Salticidae). Molecular Phylogenetics and Evolu- sequencing with chain-terminating inhibitors. tion 18:386–403. Proceedings of the National Academy of Scienc- Hendriks, L., C. Van Broeckhoven, A. Vanden- es USA 74:5463–5467. berghe, Y. Van de Peer & R. De Wachter. 1988. Smith, S.L. & J. Bond. 2003. An analysis of the Primary and secondary structure of the 18S ri- secondary structure of the mitochondrial large bosomal RNA of the bird spider Eurypelma cal- subunit rRNA gene (16S) in spiders and its im- ifornica and evolutionary relationships among plications for phylogenetic reconstruction. Jour- eukaryotic phyla. European Journal of Biochem- nal of Arachnology 31:44–54. istry 177:15–20. Telford, M.J., M.J. Wise & V. Gowri-Shankar. Hofacker, I.L. 2003. Vienna RNA secondary struc- 2005. Consideration of RNA secondary structure ture server. Nucleic Acids Research 31:3429– significantly improves likelihood-based estimates 3431. of phylogeny: examples from the Bilateria. Mo- Kjer, K.M. 2004. Aligned 18S and insect phyloge- lecular Biology and Evolution 22:1129–1136. ny. Systematic Biology 53:506–514. Van de Peer, Y., G. Van der Auwera & R. De Wach- Mallatt, J.M., J.R. Garey & J.W. Shultz. 2004. Ec- ter. 1996. The evolution of Stramenopiles and dysozoan phylogeny and Bayesian inference: Alveolates as derived by ‘‘substitution rate cali- first use of nearly complete 28S and 18S rRNA bration’’ of small subunit ribosomal RNA. Jour- gene sequences to classify the arthropods and nal of Molecular Evolution 42:201–210. their kin. Molecular Phylogenetics and Evolution Wheeler, W.C. 1996. Optimization alignment: the 31:178–191. end of multiple sequence alignment in phyloge- Masta, S.E. & J.L. Boore. 2004. The complete mi- netics? Cladistics 12:1–9. tochondrial genome sequence of the spider Ha- Wheeler, W.C. 1999. Fixed character states and the bronattus oregonensis reveals rearranged and ex- optimization of molecular sequence data. Cladis- tremely truncated tRNAs. Molecular Biology and tics 15:379–385. Evolution 21:893–902. Wheeler, W.C. & C.Y. Hayashi. 1998. The phylog- Mathews, D.H., J. Sabina, M. Zuker & D.H. Turner. eny of the extant chelicerate orders. Cladistics 1999. Expanded sequence dependence of ther- 14:173–192. modynamic parameters improves prediction of RNA secondary structure. Journal of Molecular Manuscript received 13 May 2005, revised 21 Sep- Biology 288:911–940. tember 2005.