Phylogenetic Systematics of Odonata

The systematic analysis and classification strictly followed the principles of consequent Phylogenetic Systematics (sensu HENNIG, 1966, 1981; analysis by hand and brain!), rather than so-called "numerical cladistics" (for reasons see WÄGELE, 1994, BORUCKI, 1996, and Glossary of Phylogenetic Systematics), since a careful character selection, a thorough character analysis, and a priori character weighting on the basis of complexity and compatibility has to be regarded by as essential for a sound phylogenetic analysis. The application of parsimony as mere minimization of the number of homoplasies is over reductionistic, pseudo objective and rather formalistic. Therefore, parsimony must include the different weights of characters which do not express a different "value" of putative synapomorphies, but only take account of the simple circumstance that we should have different trust in the truth of our homology hypotheses, according to the amount and quality of the available evidence. The dogma of "equal weighting" of characters is regarded by me as a myth since equality of the truth probability of all characters (= homology hypotheses) seems to be one of the least probable cases. The unavoidable subjective impact on character selection and character delimitation makes preference of cladograms which are only few steps shorter than alternative cladograms untenable since the result can be simply overturned by a different delimitation of the used characters.

Redundant taxa and the assignment of formal categorial ranks have been omitted whenever possible without violation of the International Rules of Zoological Nomenclature because they are absolutely arbitrary and superfluous (WILLMANN, 1989). To indicate the hierarchical relations the taxa have been indented, so that sistergroups are always placed on the same level. All recognized monophyla have been named because they represent natural entities that deserve a proper name. The alternative sequencing of the fossil stemgroup representatives (e.g. recommended by AX, 1984, 1988; and recently endorsed by LOHMANN, 1996) has to be rejected because of the logical and practical reasons described by WILLMANN (1989). The counter argument that such a strictly cladistic classification leads to an inflation of names is absurd since one could use the same argument against the naming of hundred-thousands of species. Why should a very subordinate entity like a new species be named, while other natural entities should remain unnamed although they are monophyletic, higher in relative rank, containing many species and eventually are of much more general interest. Names serve the purpose of communication, and it is much more convenient to refer to a (new) taxon called «Italoansida» than to a «monophyletic group of dragonflies that contains the corduliid subfamily "Corduliinae" (sensu Fraser) and the sistergroups Urothemistidae and Libellulidae, and that is characterised by the synapomorphic boot-shaped anal loop ("italian loop")». For each proposed taxon a recommended usage (list of included taxa) and a list of the referring autapomorphies (the synapomorphies of its members) is provided, according to Phylogenetic Taxonomy (sensu Bechly, nec sensu DE QUEIROZ & GAUTHIER, 1990, 1992).

The taxonomical nomenclature is generally based on the most recent catalogue of Odonata by BRIDGES (1994).

The used wing venation nomenclature is based on RIEK (1976) and RIEK & KUKALOVÁ-PECK (1984), amended by KUKALOVÁ-PECK (1991), NEL et al. (1993) and BECHLY (1994, 1995, 1996a).

In the list of potential autapomorphies I have distinguished between wing venational and other characters since only the former category of characters is of paleontological relevance. Nevertheless all proposed monophyla with extant representatives are supported by many more (and often stronger) autapomorphies that are ...

• partly already published (CALVERT, 1893; NEEDHAM, 1903; SCHMIDT, 1916; TILLYARD, 1917; KENNEDY, 1919, 1929; HENNIG, 1953, 1969, 1981, 1986; ASAHINA, 1954; FRASER, 1957; COMPTE SART, 1971; KRISTENSEN, 1975, 1981, 1989, 1991; BOUDREAUX, 1979; CARLE, 1982a-b, 1995, 1996; DAVIES & TOBIN, 1985; PFAU, 1991; CARLE & WIGHTON, 1990; KUKALOVÁ-PECK, 1991; CARPENTER, 1992; NEL & MARTÍNEZ-DELCLÒS, 1993; CARLE & LOUTON, 1994; MAY, 1995a-b; BECHLY, 1993a-d, 1995, 1996a-d, 1997a-i, 1998a-b; BECHLY & NEL & BECHLY & MARTÍNEZ-DELCLÒS, 1996; NEL & MARTÍNEZ-DELCLÒS, 1996; BECHLY & NEL & MARTÍNEZ-DELCLÒS & FLECK, 1998; TRUEMAN, 1996; LOHMANN, 1996);
  
• partly in press (BECHLY, in press a-c; JARZEMBOWSKI & MARTÍNEZ-DELCLÒS & BECHLY & NEL & CORAM & ESCUILLIÉ, in press; NEL & BECHLY & JARZEMBOWSKI & MARTÍNEZ-DELCLÒS, in press);
  
• and partly still unpublished (BECHLY, 1994, and in prep.; FLECK & BECHLY & NEL, in prep.; BECHLY & LINDEBOOM, in prep.; BECHLY & NEL & MARTÍNEZ-DELCLÒS, in prep.; BECHLY & NEL & MARTÍNEZ-DELCLÒS & JARZEMBOWSKI & CORAM & MARTILL & FLECK & WISSHAK & MAISCH, in prep.; LINDEBOOM, 1996; LOHMANN, 1995; MAY, in prep.; NEL & BECHLY & MARTÍNEZ-DELCLÒS, in prep.; NEL & BECHLY & MARTÍNEZ-DELCLÒS & PAPIER, in prep.; SMITH, in prep.; STANICZEK & BECHLY, in prep.).

PLEASE NOTE:

The proposed phylogenetic system and the referring characters are mainly derived from preliminary results of an ongoing PhD thesis by the author (BECHLY, 1996a, and in prep.). In BECHLY (1996a) there were 65 new higher taxa proposed above the family-group level, as well as 20 new families, 14 new subfamilies, 6 new tribes, 5 new subtribes, and two new genera. Furthermore there were 60 taxa status novum (nomina translata), 81 taxa sensu novum, 4 taxa status rest., and 3 new synonymies. The proposed classification should be seen as a first attempt towards a phylogenetic system of Odonatoptera. The list of characters does not include detailed descriptions or figures and is only meant as brief justification of the proposed system. A more detailed analysis shall be supplied with the final version of the mentioned thesis. The list of of characters includes "constituting" autapomorphies that are strong evidence for the monophyly of the referring taxa, as well as derived groundplan characters (autapomorphies) that are only a posteriorly interpreted as such, but are rather weak evidence for the monophyly of this taxon. The citations shall indicate the source for certain characters, but it was not possible for practical reasons to include them all in the literature section yet. Characters that are derived from standard publications, e.g. NEEDHAM (1903) or FRASER (1957), are often not yet supplied with the referring citation.

The † Cretapetaluridae, † Eumorbaeschnidae and † Proterogomphidae that have been proposed as familiae novae and † Voltzialestinae that has been proposed as subfamilia nova in BECHLY (1996a) have to be regarded as nomina nuda till the descriptions of the referring type genera that are in press or in preparation have been properly published according to art. 8 of the ICZN. Besides some older taxa, the relatively recently created taxa Anisozygopteria (taxon novum in BECHLY, 1995; nec "Anisozygoptera" Handlirsch, 1906), † Tarsophlebioidea (sensu NEL et al., 1993) and Epiophlebioidea (sensu NEL et al., 1993) that were still endorsed by BECHLY (1995) as monophyla, have been dismissed as paraphyla by BECHLY (1996a). Likewise, the taxon Euryoptera BECHLY, 1996a is here dismissed as paraphylum, in favour of the new taxon Euepiproctophora.

The following taxa have been transferred to a new position in BECHLY (1996a):

• † Dissurus from Gomphidae to † Sonidae sens. nov.
• † Dysagrioninae from Pseudolestidae to Thaumatoneuridae sens. nov.
• † Euphaeopsis from Epallagidae to † Steleopteridae
• † Euthemistidae from † Tarsophlebioidea to † Isophlebioptera taxon nov.
• † Gomphaeschnaoides is retransfered from Gomphidae to Gomphaeschnidae
• Olpogastra lugubris from Zygonychinae to Onychothemistinae - Onychothemistina
• † Petrolestes from Amphipterygidae to Thaumatoneuridae sens. nov.
• † Prostenophlebia from † Stenophlebiidae to † Erichschmidtiidae fam. nov.
• † Pseudophaea from Epallagidae to † Steleopteridae
• † Reisia from Palaeoptera incertae sedis to † Triadotypidae
• † Steleopteridae from Zygoptera to Epiproctophora taxon nov.
• † Triassolestodes from † Triassolestidae to † Triassoneuridae sens. nov.
• † Triassologus from † Paralogidae to † Triadophlebioidea incertae sedis
• † "Triassoneura" okafujii from † Triassoneuridae to † Progonophlebiidae
• † "Triassothemis" nipponensis † Triassothemidae to † Progonophlebiidae
• † Yixiangomphus from Gomphidae to † Sonidae sens. nov.

Main differences of the proposed system to other recently proposed phylogenetic systems of Odonata:

• CARLE (1995, 1996): The only difference to Carle's phylogeny of Anisoptera is the position of gomphids which are regarded by this author as most basal branch of Anisoptera. The characters that have been used by Carle to support this alternative phylogenetic position of gomphids have been discussed and rejected by LOHMANN (1995, 1996) and NEL & BECHLY & JARZEMBOWSKI & MARTÍNEZ-DELCLÒS (in press).
• TRUEMAN (1996): There are numerous differences to Trueman's cladogram of Odonata which are mostly due to an insufficient character basis which is restricted to wing venational characters only, and partly also due to serious flaws of the analysis (e.g. errors in the data matrix and use of derived species as terminal taxa, instead of using the groundplans of the respective families), and especially due to the applied methodology of so-called "Computer Cladistics" (for a critical discussion of the methodological issues see: Glossary of Phylogenetic Systematics). Trueman's analysis does neither include an appropriate character weighting, nor any a priori homologisation and polarisation of the characters, so that it has to be regarded as a basically phenetic analysis rather than a truly phylogenetic analysis. Further critical points in Trueman's analysis are: incorrect homologisations of the wing venation; erroneous generalisations of characters due to under-sampling (e.g. the alleged synapomorphies of Aeshnidae and Cordulegastridae); and neglectance of conflicting evidence (e.g. the strong larval characters that clearly support a relationship of cordulegastrids with libelluloid dragonflies rather than with aeshnid dragonflies). The most important point is that Trueman regards Zygoptera as a paraphyletic group, with Hemiphlebiidae being even more basal than the Permian Protozygoptera. Further differences concern the position of nearly all zygopteroid taxa, as well as the Tarsophlebiidae (Tarsophlebiopsis), Liassophlebiidae (Liassophlebia), Aeschnidiidae (Wightonia), Chlorogomphoidea (Chlorogomphus), Synthemistidae (Synthemis) and the analysed taxa of "corduliids" and Libellulidae (Hemicordulia, Cordulephya, Nannophlebia, and Diplacodes). Trueman's hypothesis implies that 15 (!) branches of crown-group Zygoptera were already separated in the Triassic or even Permian since the oldest representatives of the anisopteran stemgroup are known from the Upper Triassic (BECHLY, unpubl.). Since no crown-group Zygoptera are known from any pre-Cretaceous deposits at all, although there are numerous localities with an excellent fossil record of Permian and early Mesozoic odonates, the fossil record clearly contradicts the phylogeny proposed by Trueman. This contradiction is much too significant to be simply explained as artefact, due to an incomplete fossil record.
• LOHMANN (1996): The only differences regarding Lohmann's system of extant Anisoptera are the respective positions of Austropetaliida and Petaluridae which are mostly based on characters with uncertain polarity, due to an absence of the most important character complexes (e.g. segmented vesicula spermalis) in the available outgroups. Further differences concern the positions of several fossil Anisoptera like e.g. Cymatophlebiidae, Sonidae, Hemeroscopidae, Araripegomphidae, Eocorduliidae, and Condaliidae which are mostly based on incorrect character states, e.g. "compound eyes widely separated" for Cymatophlebiidae, although they are broadly confluent.