The problem with the Paleoptera problem: sense and sensitivity

While the monophyly of winged insects (Pterygota) is well supported, phylogenetic relationships among the most basal extant pterygote lineages are problematic. Ephemeroptera (mayflies) and Odonata (dragonflies) represent the two most basal extant lineages of winged insects, and determining their relationship with regard to Neoptera (remaining winged insects) is a critical step toward understanding insect diversification. A recent molecular analysis concluded that Paleoptera (Odonata+Ephemeroptera) is monophyletic. However, we demonstrate that this result is supported only under a narrow range of alignment parameters. We have further tested the monophyly of Paleoptera using additional sequence data from 18SrDNA, 28S rDNA, and Histone 3 for a broader selection of taxa and a wider range of analytical methodologies. Our results suggest that the current suite of molecular data ambiguously resolve the three basal winged insect lineages and do not provide independent confirmation of Odonata+Neoptera as supported via morphological data. 2003 The Willi Hennig Society. Published by Elsevier Inc. All rights reserved. Paleoptera (1⁄4Palaeoptera) refers to the grouping of extinct paleodictyopteroids, Ephemeroptera, and Odonata (Hennig, 1981; Kukalova-Peck, 1983, 1985, 1991, 1997; Riek and Kukalova-Peck, 1984). However, the monophyly of this group is still a controversial issue in insect evolution (Beutel and Gorb, 2001; Staniczek, 2000; Wheeler et al., 2001; Whiting et al., 1997). The extant paleopterous insects—dragonflies and damselflies (1⁄4Odonata), and mayflies (1⁄4Ephemeroptera)—lack the retractor muscle and wing sclerites necessary to fold the wings over the abdomen (Martynov, 1924). The absence of this feature has been suggested as evidence for the group s monophyly. However, this character may simply be symplesiomorphic because the muscles and sclerites allowing insects to fold wings over their abdomen were gained in the neopterous insects (Martynov, 1924). This innovation is presumably correlated with the huge explosion of neopterous species. Despite being one of the most important diversification events in all of evolution, the resolution of the relationships among Ephemeroptera, Odonata, and Neoptera remains ambiguous, and all resolutions of this three-taxon statement have been proposed. The first hypothesis will be referred to as the basal Ephemeroptera hypothesis and it suggests that Ephemeroptera is sister to Odonata+Neoptera (F€ urst von Lieven, 2000; Kristensen, 1991; Staniczek, 2000; Wheeler et al., 2001; Whiting et al., 1997). Six morphological characters proposed to support this hypothesis are (1) the anterior articulation of the mandible is a nonpermanent sliding groove and track system in Ephemeroptera, but in other pterygote lineages this articulation is more permanent; (2) subimago stage is present in Ephemeroptera but absent in other pterygotes; (3) tracheation is absent in arch of wing base and in posterior portion of the leg in Ephemeroptera but present in other insects; (4) direct spiracular musculature is absent in Ephemeroptera but present in odonates and neopterans; (5) never more than one tentorial-mandibular muscle is present in Odonata and Neoptera but multiple muscles are present in Ephemeroptera; (6) annulated caudal filament is presumably present in Archaeognatha, Monura, Zygentoma, and Ephemeroptera but absent in the remaining pterygotes; and (7) paired female genital openings are retained in Ephemeroptera and nowhere else among Pterygota. However, with some of these characters, it is unclear whether they are simply autapomorphies of Ephemeroptera or synapomorphies for Odonata+Neoptera. Corresponding author. Fax: +1-801-422-0090. E-mail address: [email protected] (T. Heath Ogden). Cladistics 19 (2003) 432–442 Cladistics www.elsevier.com/locate/yclad 0748-3007/$ see front matter 2003 The Willi Hennig Society. Published by Elsevier Inc. All rights reserved. doi:10.1016/S0748-3007(03)00083-5 The second hypothesis, termed the Paleoptera hypothesis, suggests that Ephemeroptera is sister group to Odonata, forming the group Paleoptera (Brodsky, 1994; Hennig, 1981; Kukalova-Peck, 1983, 1985, 1991, 1997; Martynov, 1924; Riek and Kukalova-Peck, 1984). This hypothesis is supported by the following characters: (1) short antennae; (2) fusion of galea and lacinia; (3) lack of the ability to fold back the wings over the abdomen; (4) veinal braces in the wings; (5) separated R and M wing veins; (6) wing fluting; and (7) aquatic larvae. Still, some of these characters (e.g., 1, 3, 7) have been regarded as plesiomorphic (Wheeler et al., 2001; Willmann, 1997). The third hypothesis places Odonata as sister to Ephemeroptera+Neoptera and will be referred to as the basal Odonata hypothesis (Boudreaux, 1979; Matsuda, 1970). This hypothesis is based primarily on the character that direct sperm transfer is synapomorphic for Ephemeroptera+Neoptera. Given that the ‘‘apterygotes’’ and Odonata have indirect sperm transfer, the gonopore-to-gonopore mode could be considered a shared derived character for mayflies and neopterous insects. However, the specific kind of indirect sperm transfer of the odonates appears to be quite different from those of the ‘‘apterygotes.’’ Odonate males deposit the sperm from segment 9 to an accessory gland on segment 2. Then, when in tandem (the position where the male grasps the female by the head with his terminalia), the female bends her abdomen down and forward to receive the sperm in her reproductive opening on segment 8. This complicated process does not resemble the indirect sperm transfer of ‘‘apterygotes’’ and is most likely autapomorphic, providing no phylogenetic information at an ordinal level (Beutel and Gorb, 2001). Due to the disagreement among, and questionable utility of, certain morphological characters, it is important to provide independent data that can corroborate one of these hypotheses to provide a more accurate estimate of phylogeny. We are particularly interested in the sensitivity of molecular topologies to perturbations of parameter values during phylogenetic analysis (Phillips et al., 2000; Wheeler, 2001, 1995). We specifically define robustness as a measure of stability of nodes to fluctuations in parameter values across an analytical landscape. A highly robust node is one that is supported under a wide range of parameter values, in contrast to a poorly supported node that is supported under only one or a few parameter values. We recognize that sensitivity analysis is only one measure of topological robustness and that other measures are currently in vogue (e.g., nonparametric bootstrap, Bremer support, posterior probabilities, etc.) (Archie, 1989; Bremer, 1988; Faith, 1991; Faith and Cranston, 1991; Felsenstein, 1985), each with their own pros and cons (Grant and Kluge, 2003). However, given that the current molecular data used to infer paleopteran phylogeny is primarily ribosomal DNA sequences and that the topologies generated via these sequences are strongly influenced by alignment methodologies, we are interested in addressing the question of whether any analytical method will robustly support one of the three hypotheses listed above under a wide range of parameter values. In approaching the question in this manner, we do not attach any particular significance to congruence among disparate analytical methodologies. We are interested only in determining whether a robust solution exists for the given data under any analytical methodology or whether the molecular data do not discriminate among the hypotheses. Independent tests (i.e., molecular data) have provided mixed support for the different hypotheses. For instance, Wheeler et al. (2001) published the most extensive formal analysis of ordinal relationships using molecular and morphological information. The 18S rDNA (18S) data and 18S+ 28S rDNA (28S) data supported a monophyletic Paleoptera, but the 28S data and the total-evidence analyses (including morphology) supported basal Ephemeroptera. This study, however, did not concentrate sampling on basal pterygotes, so the extent to which these results are influenced by the under sampling of taxa is not clear. In a recent molecular analysis, the relationship among basal pterygotes was specifically tested and the authors conclude that Paleoptera is monophyletic (Hovm€ oller et al., 2002). However, given the difficult nature of the Paleoptera problem, and some potential flaws in their analytical methodology, we were interested in determining the generality of their conclusion, given additional data and analyses. The overall objective, therefore, is to determine whether a robust solution to the Paleoptera problem exists given current data and analytical methods. This objective will be specifically examined by two subgoals: (1) test the generality of the claim that the current molecular data support the monophyly of Paleoptera as presented by Hovm€ oller et al. (2002); (2) provide additional data and analyses to test the sensitivity of the topology to data partitions, cost parameter values, and methods of data analysis. Materials and methods Reanalysis of Hovmöller et al. (2002) data In the Hovm€ oller et al. (2002) study, sequence data from 18S rDNA and partial 28S rDNA for 18 spp. of Odonata, 8 spp. of Ephemeroptera, 8 spp. of Neoptera, and 2 spp. of Archaeognatha were used to estimate phylogeny. This taxon sampling represents 22% (6 of 27) of the odonate family taxa and 14% (5 of 36) of the mayfly family taxa. No morphological data were incorporated in their analyses, though coded character matrices were available (Beutel and Gorb, 2001; T.H. Ogden, M.F. Whiting / Cladistics 19 (2003) 432–442 433