http://www.collembola.org/publicat/crustacn.htm - Last updated on 2006.02.20 by Frans Janssens

Checklist of the Collembola: Are Collembola terrestrial Crustacea?

Peter Nolan Lawrence, 9 Weald Close, Bromley Common, Kent, BR2 9PD, UK
Frans Janssens, Department of Biology, University of Antwerp (RUCA), Antwerp, B-2020, Belgium

Introduction

While there is consensus that Arthropoda is monophyletic, there is none on the relationships of its subordinates. Almost every possible alternative has been offered (Giribet & Ribera, 2000:205):
Arthropoda = Chelicerata + Mandibulata (Crustacea + Atelocerata (Myriapoda + Hexapoda)) according to Snodgrass, 1938; Weygoldt, 1979; Wågele, 1993; Wheeler et al., 1993; Wheeler, 1995, 1998 (cited from Giribet & Ribera, 2000:205).
Arthropoda = Chelicerata + Mandibulata (Myriapoda + Pancrustacea (Crustacea + Hexapoda)) according to Giribet et al., 1996; Giribet & Ribera, 1998; Zrzavy et al., 1998 (cited from Giribet & Ribera, 2000:205).
Arthropoda = (Chelicerata + Myriapoda) + Pancrustacea (Crustacea + Hexapoda) according to Turbeville et al., 1991; Friedrich & Tautz, 1995; Giribet et al., 1996 (cited from Giribet & Ribera, 2000:205); Dove & Stollewerk, 2003.
Arthropoda = Schizoramia (Chelicerata + Crustacea) + Atelocerata (Myriapoda + Hexapoda) according to Cisne, 1974; Briggs et al., 1992; Budd, 1993 (cited from Giribet & Ribera, 2000:205).
Many studies unite Crustacea and Atelocerata in the Mandibulata (Snodgrass, 1938; Weygoldt, 1979; Wågele, 1993; Wheeler et al., 1993; Wheeler, 1995,1998; Giribet et al., 1996; Giribet & Ribera, 1998; Zrzavy et al., 1998). Others align Crustacea with Chelicerata in the Schizoramia (Cisne, 1974; Briggs et al., 1992; Budd, 1993). Supporters of Mandibulata do not agree whether Atelocerata is monophyletic and sister of Crustacea (the traditional and majority position), or whether Crustacea is sister of Hexapoda (the Pancrustacea hypothesis). For a comprehensive overview of the rival hypotheses, see Giribet & Ribera (2000:222-225). Given the conflicting morphological (Atelocerata), molecular (Pancrustacea) and paleontological (Schizoramia) hypotheses, only a simultaneous analysis of all data from extinct and extant taxa might offer a solution (Giribet, Edgecombe & Wheeler, 1999:197).

Any direct relationship between Hexapoda and Myriapoda is becoming more and more doubtful and therefore the polyphyly of Atelocerata seems increasingly certain (Dove & Stollewerk, 2003). As more nuclear, mitochondrial gene order and protein-encoding gene data have been examined for an ever-wider set of taxa, little or no support has been found for any of the possible groupings alternative to Pancrustacea (Cranston & Gullan, 2003:883). In this paper, we will question whether or not Collembola are Insecta, Hexapoda or Crustacea within the context of the Pancrustacea theory. Hexapody originated at least twice independantly in Arthropoda, making the Hexapoda paraphyletic (Nardi et al., 2003). We will show that Collembola are not so called primitive apterous insects, but a unique and ancient group of well adapted terrestrial crustaceans.

Paleontological considerations

Fig.r. Rhyniella praecursor
After Scourfield (1940).

Hexapody did not evolve as an adaptation to terrestrial locomotion, but was already developed in the marine habitat, as demonstrated by the ectognathous marine hexapod Devonohexapodus bocksbergensis from the Lower Devonian Hunsrück Slates (Haas, Waloszek & Hartenberger, 2003:39,46). This finding demonstrates that stem lineage Insecta coexisted during the Devonian with the more derived Collembola (Haas, Waloszek & Hartenberger, 2003:52). Cambronatus brasseli and Wingertshellicus backesi from the Lower Devonian, are described by Briggs & Bartels (2001) as "crustaceanomorphs", while Haas, Waloszek & Hartenberger (2003:49) suggest a close relationship to Hexapoda. The marine Tesnusocaris goldichi Brooks, 1955, from the Carboniferous, redescribed by Emerson & Schram (1991) as a representative of the Remipedia, is condsidered to be more close to Hexapoda by Haas, Waloszek & Hartenberger (2003:49).

Reinterpretation of a fossil insect fragment Rhyniognatha hirsti Tillyard, 1928 from the early Devonian Rhynie cherts of Scotland indicates that insects originated in the Silurian period (Engel & Grimaldi, 2004:627-630). This is supported by the fossil records of Archaeognatha (= Microcoryphia) that have been listed from the early Devonian of Quebec (Labandeira & al., 1988 cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:387) and from the Devonian of New York (Shear & al., 1984 cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:387), suggesting a parallel evolution of Insecta and Collembola. Rhyniella praecursor Hirst & Maulik, 1926, the oldest known collembolan fossil from the Devonian red sandstone Rhynie chert beds, Scotland, resembles extant Collembola up to such an extent that this might be an indication that Collembola reached their evolutionary climax already 400 million years ago. Therefore, Handschin (1955:49) considers Collembola as living fossils.

Systematic overview

Historically, single morphological character systems formed the basis for organising the hexapods into groups (Carpenter & Wheeler, 1999:333,344). Since the characters taken into account in such studies are independent of each other, each study results in a different phylogeny. Combining sets of independent characters into one study is the approach of the 'total evidence' of Kluge (1989) and 'simultaneous analysis' of Nixon and Carpenter (1996) to maximise parsimony of the tree (Carpenter & Wheeler, 1999:333,344). Given the high number of characters involved, a numerical cladistic analysis will simplify the work. Computer assisted analyses allow for combining a hugh amount of characters (morphological, molecular, ecological, physiological, etc.) to be taken into account simultaneously, allowing in this way simultaneous treatment of all available evidence. In the following discussion, the single character based groupings that lead to taxonomic units used in the systematic classifications are questioned by integrating manually the result of many single and multiple character studies in an attempt to confront the different opinions preliminary.

Are Collembola Ellipura?

• Hennig (1953 cited from Wheeler, Whiting, Wheeler & Carpenter, 2001:115), (1965 and 1969 cited from Dallai & Callaini, 1979:64; Dallai, 1991:267) and Tuxen (1970 and 1972 cited from Dallai & Callaini, 1979:64; Dallai, 1991:267) regard Protura + Collembola as a monophyletic 'sister-group' of Diplura. Kukalová-Peck (1991 cited from Wheeler, Whiting, Wheeler & Carpenter, 2001:115), regard Parainsecta as a monophyletic sister-group of Diplura + Insecta. According to Hennig (cited from Dallai, 1991:267), the similarity was supported by a list of advanced characteristics present in Protura and Collembola. This view was revised by Kristensen (1975 cited from Dallai, 1991:267) who concluded that only the common features of entognathy, the modified tibiotarsi and the absence of abdominal spiracles and cerci could be regarded as true synapomorphies. In addition, François (1969 cited from Dallai, 1991:267) indicates the presence of the linea ventralis, and (cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:382) the reduction of the maxilliary and labial palpi, and the presence of large neurosecretory cells, the 'epipharyngeal ganglia'. In addition, Kristensen (1981, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:382) considers the pretarsus bearing only one unguis as the ultimate character.
• Ellipura is supported by Hennig (1953, 1981), François (1969), Kristensen (1975, 1981, 1991, 1995, 1997), Boudreaux (1979), Kukalová-Peck (1987, 1991), Stys & Bilinski (1990), Stys et al. (1993), Stys & Zrzavy (1994), Koch (1997), Kraus (1997) (all cited from Bitsch & Bitsch, 1998:356) and Giribet, Edgecombe & Wheeler (1999:203)). In addition by Bilinski (1993 cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:382) by Wheeler, Whiting, Wheeler & Carpenter (2001:113,130,131,133,135,142), by Kraus and Kraus (1994), Kraus (1998, 2001), Ax (1999), Carpenter & Wheeler (1999), Edgecombe & al. (2000) (all cited from Giribet, Edgecombe, Carpenter, D'Haese, & Wheeler, 2004:320). It is supported by the morphological phylogeny of Beutel & Gorb (2001:188,189,190,191,192) based on following five synapomorphies: 1. presence of a linea ventralis (Kristensen, 1991, 1997; Koch, 1997), 2. mouthparts comprising entognathous maxillae and mandibles in a single gnathal pouch (Tuxen, 1959; Koch, 1997, 2000), 3. absence of cerci, 4. absence of abdominal spiracles (Kristensen, 1975), 5. single claws.
• The real position of Protura has been obscured by their presumed relationship with the Collembola (Dallai, 1991:267).
• The assemblage Ellipura does not appear to form a natural group (Bitsch & Bitsch, 1998:353). Autapomorphy no longer supports the Ellipura (Bitsch & Bitsch, 1998:355). Ellipura is not parsimonious, based on internal anatomy (Bitsch & Bitsch, 1998:356).
• Ellipura is paraphyletic based on the analysis of 117 morphological characters combined with the sequences of 18S and 28S rRNA (Edgecombe, Giribet & Wheeler, 1999:299).
• The monophyly of the Ellipura is weakly supported by cladistic analysis based on external morphology (Bitsch & Bitsch, 2000:132). The statistical confidence level inferred from a parsimony analysis based on external morphology found for the grouping Ellipura was only 39% (Bitsch & Bitsch, 2000:146). The putative assemblage Ellipura is not firmly supported by a cladistic analysis based on external and internal anatomy characters (Bitsch & Bitsch, 2000:149). The Collembola form a monophyletic clade among the hexapods. They have a very ancient origin which split very early from an unknown ancestor (Bitsch & Bitsch, 2000:149-150).
• Data on the mitochondrial cytochrome c oxides subunit II gene does not lead to a close relationship between Protura and Collembola (Shao, Zhang, Ke, Yue & Yin, 2000:595).
• Ellipura is not supported by the molecular analysis of Wheeler, Whiting, Wheeler & Carpenter (2001:135,136), in which Collembola is sister-group of Insecta and Diplura, respectively.
• In Collembola the hypopharynx is often a short three lobed structure, composed of the lateral superlinguae on either side of the median lingua, while in Protura the superlingua is absent and the hypopharynx is represented by a narrow projection (Matsuda, 1965:106).
• François (1969) characterises the relationship between Collembola and Protura by the persistence of a post-antennal organ, by the the presence of a linea ventralis and by a stomatogastric nervous system (cited from Dallai, 1974:148).
• The neurosecretory system of Collembola is a primitive neurohemal organ, lacking proper secretory cells, while Protura have a system with rudimentary corpora cardiaca, as found also in some Diplura and Archaeognatha (Juberthie & Cassagnau, 1971:61-65).
• While Collembola have maintained a certain homogeneity in their spermatogenesis, Protura have accomplished their evolution through a long line of successive modifications (Dallai 1974:154). Collembola have sperm with a 9+2 flagellum, while the sperm of Protura has a strongly modified flagellum: 14+0 or 12+0 in Acerentomidae and Eosentomon has an aflagellate sperm (Dallai, 1974:155). Jamieson (1987, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381-382) concludes in a study of the evolution of the sperma in insects that the 9+2 axoneme is a plesiomorph character of Arthropoda.
• A true anterior tentorium is present in Collembola whilst in Protura it is reduced (Kristensen, 1975 cited from Dallai & Callaini, 1979:64). Bitsch & Bitsch (2000) regard the fulcrotentorium of Protura as non-homologous with the true tentorium of Insecta, and interpreted the endoskeletal formations of Collembola and Diplura to be a complex endosternite composed of connective fibres rather than a cuticular tentorium (cited from Edgecombe & Giribet, 2002:150). In contrast, Koch (2000) endorsed homology between the anterior tentorial apodemes of Collembola, Diplura, and Insecta, citing common points of origin, for example, identical sclerotic connections with the labrum (cited from Edgecombe & Giribet, 2002:150-151).
• Based on ultrastructures, it was not possible to find any close relationship between Protura and Collembola (Dallai 1976, 1980 cited from Bitsch & Bitsch, 1998:356).
• The pyloric region of Protura comprises a typical sphincter, a pyloric chamber with specialised cells having long microvilli and, above all, Malpighian papillae (Dallai, 1976 cited from Dallai & Callaini, 1979:64) while the pyloric region of Collembola comprises at the same level: posterior midgut cells, connecting cells, and pyloric ring cells (Dallai, 1979:175). Consequently, the pyloric region of Protura and Collembola are very differently organised (Dallai, 1979:175). The lack of Malpighian structures in Collembola might be explained by considering the excretory function of the midgut; the different organisation of the pyloric region in the two groups, however, still remains (Dallai & Callaini, 1979:64).
• Many other features make Protura and Collembola uncomparable: the type of the egg cleavage, which is complete only in Collembola; the smaller number of their abdominal segments; the presence of a true anterior tentorium in Collembola whilst in Protura it is incomplete (François, 1969 cited from Dallai, 1979:175); the sperm structure, peculiar in Protura (Dallai, 1974:155); the organisation of the ovary, which is meroistic in Collembola but panoistic in Protura (Jura, 1975 cited from Dallai, 1979:175,177); the presence of an unpair, rudimentary, supra-oesophageal corpus cardiacus, only in Protura (Juberthie & Cassagnau, 1971:65).
• By definition (A.W. Leftwich), the telson is the twelfth abdominal segment, which occurs only in Protura (Lawrence, 1978:69).
• The Protura must be regarded as an isolated clade among the hexapods (Dallai, 1991 cited from Bitsch & Bitsch, 2000:149).
• Bach de Roca, Gaju-Ricart & Compte-Sart (1999:393) conclude that Collembola and Protura each are monophyletic groups and that the grouping Ellipura is not supported by their review of available phylogenies and should not be accepted.
• The arthropod phylogeny based on ribosomal 18S rDNA does not support monophyly of Ellipura (Giberet & Ribera, 2000:222).
• Larval development in Protura is anamorphic (with segments added posteriorly during development), whilst in Collembola it is epimorphic (with segment number constant through development) (as in Diplura, Archaeognatha, Zygentoma and Pterygota) (Cranston & Gullan, 2003:884-885).
• Protura either is sister to Collembola, forming Ellipura in a weakly supported relationship based on similarity of the entognathous mouthparts and lack of cerci, or is sister to all remaining Hexapoda. In the latter case, Collembola form a more strongly supported relationship as sister to Diplura + Insecta (Cranston & Gullan, 2003:884).
• In a cladistic analysis based on 72 comparative morphological characters the Protura and the Collembola appear as independent clades (Bitsch & Bitsch, 2004).
• A cladistic analysis of 189 morphological characters is ambiguous with respect to the position of Collembola and Protura; in one topology Ellipura is monophyletic, but in an alternative resolution Protura is sister group of Diplura + Ectognatha (Giribet, Edgecombe, Carpenter, D'Haese, & Wheeler, 2004:324,325).
  The combined cladistic analysis of sequence data of five molecular markers and 189 morphological characters contradict the Ellipura hypothesis by confirming the monophyly of the Diplura and resolving Diplura and Protura as sister groups (Giribet, Edgecombe, Carpenter, D'Haese, & Wheeler, 2004:319,328,329).

Are Collembola Entognatha?

• Entognathy in apterygotans is a monophyletic characteristic according to Snodgrass (1938 cited from Wheeler, Whiting, Wheeler & Carpenter, 2001:114), Hennig (1953, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381) (1969, cited from Giribet, Edgecombe & Wheeler, 1999:203) (1981, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381 and Carpenter & Wheeler, 1999:335,345), Tuxen (1959, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381) (1968, cited from Dallai, 1974:148) (1970, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381), Lauterbach (1972, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381), Kristensen (1975, 1981, 1991, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381), Boudreaux (1979, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381), Stys & Bilinski (1990, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381 and Giribet, Edgecombe & Wheeler, 1999:203), and Machida & al. (1996, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381).
• The implied paraphyly of Entognatha should not be ruled out completely (Kristensen, 1997 cited from Beutel & Gorb, 2001:189). However, differences in entognathism (Koch, 1997, 2000 cited from Beutel & Gorb, 2001:189) do not necessarily mean that the advanced condition found in ellipurans is not derived from a hypothetical configuration of mouthparts in a common ancestor (Beutel & Gorb, 2001:189). Beutel & Gorb (2001:190) claim entognathous hexapods do not have adhesive foot structures [ignoring in this way the clavate tenent hairs of Collembola].
• Molecular analyses of the rRNA 18S supports the monophyly of Entognatha according to Friedrich & Tautz (1995, cited from Giribet, Edgecombe & Wheeler, 1999:203-204).
• Entognathy is a convergent characteristic acquired independently in Elliplura and Diplura according to Manton (1977, 1979 cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381), Koch (1997, 1998 cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381), and Kraus (1998 cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381).
• The monophyly of the Entognatha is rejected by Kukalová-Peck (1987 cited from Bitsch & Bitsch, 2000:151 and Ikeda & Machida, 2001:250), Kukalová-Peck (1991 cited from Wheeler, Whiting, Wheeler & Carpenter, 2001:115), Koch (1997 cited from Bitsch & Bitsch, 2000:151 and Ikeda & Machida, 2001:250), Kraus (1997 cited from Bitsch & Bitsch, 2000:151 and Ikeda & Machida, 2001:250), Ikeda & Machida (1998:114), Bach de Roca, Gaju-Ricart & Compte-Sart (1999:393), Giberet & Ribera (2000:222), and Wheeler, Whiting, Wheeler & Carpenter (2001:128,129,130,133,140,141,142,144).
• Molecular analyses of the 18S rRNA supports the paraphyly of Entognatha according to Giribet & al. (1996), Giribet & Ribera (1998) and Spears & Abele (1998) (all cited from Giribet, Edgecombe & Wheeler, 1999:204).
• Entognathy in apterygotans is a polyphyletic characteristic according to Stys & al. (1993, cited from Giribet, Edgecombe & Wheeler, 1999:203).
• Boudreaux (1979, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:381) considers Entognatha as a regressed Hexapoda, characterised by the following autapomorphies: modified mouthparts that are embedded into a buccal cavity in the head capsule, reduction of the palpi, reduction of the eyes in Collembola (cfr. Paulus 1972, 1979) even lacking in Protura and Diplura, reduction of the 'ciegos gástricos', and reduction of Malpighian tubes.
• The assemblage Entognatha (Ellipura and Diplura) does not appear to form a natural group (Bitsch & Bitsch, 1998:353). Based on internal anatomy, the monophyly is only weakly supported (Bitsch & Bitsch, 1998:357,358).
• Entognatha is neither supported by the cladistic analysis based on 256 morphological variables (Carpenter & Wheeler, 1999:335,336,345), nor on that of ribosomal DNA of two genes (Carpenter & Wheeler, 1999:339), while it is weakly supported in the combined analysis (Carpenter & Wheeler, 1999:340,341,346).
• The statistical confidence level inferred from a parsimony analysis based on external morphology found for the grouping Entognatha is only 21% (Bitsch & Bitsch, 2000:146).
• Janetschek (1969) is of the opinion that, in Apterygota, entognathy is realised after different schemes so it should be interpreted as the result of the parallel but independent evolution of the several groups of Apterygota; the polyphyletic appearence of entognathy in Pterygota seems to support this opinion (cited from Dallai, 1974:148).
• While in Collembola the epithelium of the midgut is renewed with every moult, in Campodea, the midgut epithelium is only partly regenerated per moult, resulting in a complete renewal only after 4 to 5 moults (Bareth, 1969 cited from Thibaud, 1970:154).
• Entognathy is a convergent feature which has appeared in widely different taxa and can hardly be regarded as a taxonomic character indicating affinity (Manton, 1977; cited from Dallai, 1991:263).
• Entognathy must have been developed independently in different groups (Remington, 1955; Manton, 1977; Lauterbach, 1980; Koch, 1997; Kraus, 1997; cited from Bitsch & Bitsch, 2000:135).
• Different opinions exist in the realisation of the entognathy condition in Collembola, Protura and Diplura (Manton, 1964, 1977; Janetschek, 1969; Tuxen, 1970; Lauterbach, 1972 cited from Dallai & Callaini, 1979:64) (Tuxen, 1970; Manton, 1977 cited from Dallai 1979:177).
• Dallai & Callaini (1979:64) suggest an independent evolution of the three Entognatha groups from a common ancestor.
• Kristensen (1995, cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:382) considers the 9+9+2 axoneme a synapomorphy of Diplura and Insecta, therefore invalidating Entognatha (Collembola have a 9+2 axoneme and Protura a 12+0 or 14+0 axoneme).
• In Collembola, the blastokinesis is completed before appendage formation. In Diplura, blastokinesis does not occur until the appendages are almost developed (Ikeda & Machida, 1998:109).
• Ikeda & Machida (1998:112) found a notable difference between the entognathy formation of Collembola and Diplura. In the collembolan Tomocerus ishibashii (Uemiya & Ando, 1987 cited from Ikeda & Machida, 1998:112), the labial terga participate in mouth fold formation and reach the mouth opening, in contrast to Diplura, in which the labial terga never do so. Moreover, a peculiar and unique labial morphogenesis is involved in entognathy formation in Diplura, but not in Collembola.
• Grimaldi (2001:1158) recognises two types of entognathy: one in Diplura, and another, more pronounced, in Collembola and Protura.
• In Diplura, the dorsal organ is formed by the concentration of whole serosal cells, after the stage where the embryo segregates the inner layer. In Collembola, the dorsal organ is formed much earlier, in the blastoderm stage when the germ band has yet to differentiate (Ikeda & Machida, 2001:247).
• In Diplura, as wel as in Ectognatha, the amnion is produced, while it is not produced in Collembola (Ikeda & Machida, 2001:250).
• The timing of blastodermic cuticle formation in Diplura is later in comparison with that of Collembola (Ikeda & Machida, 2001:250).
• During the stage of dorsal organ formation, three thin cuticular layers are secreted by the embryo of Diplura. In Collembola, the cuticular layers are formed by an area other than the dorsal organ (Tiegs, 1942; Tamarelle, 1971; Jura, 1972; Uemiya & Ando, 1987; cited from Ikeda & Machida, 2001:251).
• Traditionally, Collembola, Protura and Diplura were grouped as "Entognatha", based on the apparently similar morphology of the mouthparts, that are enclosed in folds of the head. However, two different types of entognathy now are recognised, one shared by Collembola and Protura and the second found only in Diplura. Other morphological evidence indicates that Diplura may be closer to Insecta than to other entognathans and thus Entognatha may be paraphyletic (Cranston & Gullan, 2003:884).
• A cladistic analysis of 189 morphological characters does not support Entognatha, with Diplura resolved as sister group of Ectognatha (Giribet, Edgecombe, Carpenter, D'Haese, & Wheeler, 2004:324).

Are Collembola Apterygota?

• Apterygotans comprise the Archaeognatha, Zygentoma, Diplura, Collembola and Protura, mainly characterised by the absence of wings (see Denis, 1949; Remington, 1955; Paclt, 1956; Sharov, 1966; cited from Bitsch & Bitsch 1998:340) (Rohdendorf, 1961; Kukalová:-Peck, 1991 cited from Grimaldi, 2001:1153).
• The system of Handlirsch (1903, 1925) is based on paleontological insect records from amber and on the formation of the wings and places the Collembola in the Apterygota at the basis of the Pterygota (Handschin, 1955:50).
• There is no relation between the nature of the palp of the maxillary lobus externus of Japyx and Campodea, and that of Collembola (Börner in Schultz, 1908:64).
• Egg cleavage is complete only in Collembola (Dallai & Callaini, 1979:64).
• The number of abdominal segments is reduced only in Collembola (Dallai & Callaini, 1979:64).
• Apterygota is considered as being an artificial assemblage of paraphyletic taxa (Moen & Ellis, 1984). Therefore, it is not accepted anymore as a valid formal taxon by the cladistic school of systematists (Hopkin, 1997:19) (Bitsch & Bitsch, 1998:340,358) (Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:393) (Grimaldi, 2001:1153).
• The Apterygota are not found to compose a natural group among the hexapods, based on a cladistic analysis of external and internal anatomy characters (Bitsch & Bitsch, 2000:153).
• Similar distances, calculated from the mitochondrial cytochrome c oxides subunit II gene, in Collembola, Protura, Diplura and Insecta suggest a parallel evolution among the wingless groups and the winged insects (Shao, Zhang, Ke, Yue & Yin, 2000:593).
• Apterygota is rejected by Wheeler, Whiting, Wheeler & Carpenter (2001:130,133,135,136,140,141,142,144),
• Basal hexapods undoubtedly include taxa whose ancestors were wingless and terrestrial. This grouping is not monophyletic, being based entirely on evident symplesiomorphies or otherwise doubtfully derived characters. Included groups are Protura, Collembola, Diplura, Archaeognatha, and Zygentoma (Cranston & Gullan, 2003:884). The once traditional group Apterygota comprising the primarily wingless taxa is paraphyletic and rejected (after Cranston & Gullan, 2003:885).

Are Collembola Insecta?

• Insecta Kukalová-Peck (1987, cited from Giribet, Edgecombe & Wheeler, 1999:203) is considered monophyletic.
• Monophyly of Insecta (Ectognatha) is confirmed by the morphological phylogeny of Beutel & Gorb (2001:189).
• It is very extraordinary for the spiracles of the tracheal system, as found in Smynthurus(sic), to open ventrally between head and prothorax. No insect has spiracles either in the head or between head and prothorax (Lubbock, 1873:77).
• The dorsal organ of the collembolan embryo is unrelated with that of the Pterygota (Tiegs, 1942 cited from Schaller, 1970:51).
• Collembola oldest fossils are from the Devonian, while Pterygota oldest fossils are from the upper Carboniferous (Handschin, 1955:45,47). But Pterygota appeared already in the Devonian (Grimaldi, 2001:1153), even in the Silurian (Rohdendorf, 1961 cited from Grimaldi, 2001:1153). This suggests an independent parallel evolution of Collembola and Pterygota from ametamorph precursors in the lower Silurian.
• The embryonal development of Collembola is, as Philipptchenko, Folsom and Imms have showed, completely opposed to that of Insecta (Handschin, 1955:44,46). A holoblastic cell cleavage continues up to the stage of 32 blastomers. Then most of the blastomers migrate to the egg surface and form the primitive blastoderm (Handschin, 1955:44). This for hexapodans, exceptional embryogenese is unique and characteristic for Collembola; it does not occur in the telolecithal eggs of Insecta; it can also be found in some terrestrial Acari (Pediculopsis) and in some primitive Crustacea, such as Branchiopoda (Handschin, 1955:44,46).
• The tarsi of Collembola are simple; the tarsi of Insecta are segmented (Handschin, 1955:46). To exclude influence from metamorphosis, it will be necessary to compare collembolan tarsi with larval insect tarsi.
• Tillyard coined the term 'holomery' for the unique post-embryonal development of the Collembola: they hatch from the egg with a number of abdominal segments (six) that remains stable during life (cited from Handschin, 1955:43). In Insecta, one finds 11 segments and a telson (Handschin, 1955:46). In Collembola, the number of abdominal segments does neither decrease nor increase during post-embryonal development (Handschin, 1955:46).
• Collembola males and females have gonopores on thesame (fifth) abdominal segment; Insecta males and females have gonopores on different (eighth and nineth) abdominal segment (Handschin, 1955:46).
• In Collembola, and Thysanura, some abdominal segments bear vestigial appendices; in Insecta: only gonopodes and gonapophyses, lacking in Collembola (Handschin, 1955:46).
• Absence of vasa malpighii in Collembola. (Handschin, 1955:43,46). In Collembola, the presence of structures resembling insect Malpighian tubules is definitely excluded (Dallai, 1966; Thibaud, 1968 cited from Dallai & Callaini, 1979:45; Dallai, 1979:175). The observations of Nicolet (1842), Heymons (1897) and Phyliptschenko (1912) describing Malpighian tubules in Collembola may be regarded as erroneous (Dallai & Callaini, 1979:45).
• Absence of respiratory system (tracheae). Exceptions in Actaletidae,Coenaletidae and Symphypleona; [abdominal] stigmata are absent (Handschin, 1955:43,46-47).
• Handschin (1955:43) considers the absence of compound eyes as one of the most important morphological features to differentiate Collembola from Insecta. Collembola have eye patches, each composed of maximum 8 ommatidia.
• Pterygota are characterised by a permanent adult morphological state. The final larval moult marks the final state of the individual life. Adult Pterygota do not continue to moult. Apterygota [implying Collembola] on the other hand, continue to moult even when they reached the adult state. The final larval moult does not mark the final state of the individual life. (Lindemann, Sweetman, von Orelli) (cited from Handschin, 1955:44-45).
• Collembola do not possess a true tentorium. It is generally considered to be of mesodermal origin. The tentorium of Insecta is a structure of cuticular formation that has migrated into the head capsule. (Wolter 1963 cited from Adams & Salmon, 1972:276). However, this opinion is challenged by Adams & Salmon (1972:276).
• Collembola are not on the same phylogenetic line as Thysanura and Pterygota, in particular, the mode of development of the endoskeletal tentorium is very peculiar (Matsuda, 1965:33).
• The structure of the head and mouthparts of Collembola has little in common with that of Thysanura and Pterygota (Matsuda, 1965:69).
• The ultrastructure of the sperm of Collembola shows a 9+2 pattern of the axial flagellar filament, which is not found in Insecta (Krzysztofowicz & Byczkowska-Smyk, 1966; Dallai, 1967; cited from Dallai, 1969:275).
• The maturation of the germ cells in Collembola, unlike in Insecta, occurs almost synchronously throughout the gonad (Dallai, 1969:275).
• The cephalic endoskeleton of Collembola, composed on one hand of ectodermic products, such as the suspensorium hypopharyngien and the sclerites of the oral folds, and on the other hand of conjuctive mesodermic formations, such as endosternites, is very different from the tentorium of Thysanura and Pterygota (François, 1971:47,48).
• The crystalline cone of the collembolan eyes is an extracellular secretion of proteinic nature while the crystalline cone of the ommatidia of insects are intracellular deposits of a mucopolysaccharid-protein complex (Barra, 1971:52-53).
• The cerebral neurosecretory system of Collembola lacks corpora cardiaca, while Insecta have a system with corpora cardiaca (Juberthie & Cassagnau, 1971:61,76).
• The oesophagus of Collembola is surrounded by an aortic sleeve, while in Insecta the aorta runs dorsally parallel to the oesophagus (Juberthie & Cassagnau, 1971:61,74).
• The corpora allata of Collembola are located suboesophageally, while in Insecta they are located supra-oesophageally (Juberthie & Cassagnau, 1971:71,72).
• Neelipleona have a permanent cuticular wax secretation system, based on detachable epicuticular tubercles, that is unique among Collembola and even among Insecta (Massoud, 1971:197);
• Altner & Ernst (1974:121) conclude that the mechanosensitive sensilla of the fourth antennomere of Collembola cannot be compared structurally with the mechanosensitive sensilla of the antenna of the Pterygota, and consider them phylogenetically primitive.
• Paulus (1974:124) notes that the high variability in the arrangement of the rhabdom of the entomobryomorph ommatidium is quite remarkable, since the structure of the rhabdom of the higher Insecta is mostly stable within each ordo.
• The spermatozoon of Collembola has a 9+2 axoneme, while in Insecta the spermatozoon has a 9+9+2 axoneme (Dallai, 1974:153).
• Kukalová-Peck (1991) (cited from Hopkin, 1997:20) coined the name Parainsecta for the grouping of the Collembola and Protura to stress the fact that it are not true Insecta.
• Unlike most insects, Collembola appear to lack specialised circulatory organs for pumping blood into the antenne (Pass, 1991 cited from Hopkin, 1997:59).
• Tracheae probably evolved independently after Collembola branched off from the stem leading to Insecta (Xué et al., 1994 cited from Hopkin, 1997:59).
• Insecta have a pair of antennal vessels. Collembola lack such vessels (Bitsch & Bitsch, 1998:347).
• Ikeda & Machida (1998:112) conclude that the abdominal ground plan with 11 segments should be assigned only to Ectognatha. The abdomen of Diplura is composed of 10 segments. The collembolan abdomen is composed of only six segments (Jura, 1972 cited from Ikeda & Machida, 1998:112). In Protura, the prelarva hatches with 9 abdominal segments, and the abdominal segments increase up to 12 during postembryonic development (Imadaté, 1991 cited from Ikeda & Machida, 1998:112).
• Many Ectognatha develop a pair of pleuropodia in the first abdominal segment as an embryonic organ. In Diplura and Collembola (cf. Jura, 1972), the pleuropodia do not develop (Ikeda & Machida, 1998:114).
• The evolutionary rate of the mitochondrial cytochrome c oxides subunit II gene in Collembola appeares to be faster than that in insect orders. Considering their long history of evolution (fossil record of 400 million years), it is unsurprising that Collembola has higher genetic divergence among its families than that among the orders of Insecta. This supports the view that Collembola should be elevated to a rank of class (Shao, Zhang, Ke, Yue & Yin, 2000:593).
• In the molecular phylogeny of hexapodans by Wheeler, Whiting, Wheeler & Carpenter (2001:135), Collembola appear as sister-group of Insecta. In combination with a morphological analysis, the total evidence strict consensus cladogram puts Collembola as sister-group of Protura + Diplura + Insecta (Wheeler, Whiting, Wheeler & Carpenter, 2001:141,144).
• A phylogenetical analysis of the mitochondrial sequence of 35 arthropodans strongly supports the monophyly of Insecta - excluding Collembola from Insecta (Nardi et al., 2003:1887).
• The phylogenetic analysis of nucleotide sequences of 35 arthropodans, supports the monophyly of Insecta - Collembola are excluded from Insecta (Delsuc et al., 2003:1482d).
• Because Insecta is treated as a class, the successively more distant sister groups Diplura and Collembola (with or without Protura) are of equal rank (Cranston & Gullan, 2003:884).
• Engel & Grimaldi (2004:627-630) reinterpretated a fragmentary insect fossil Rhyniognatha hirsti Tillyard, 1928 from the early Devonian Rhynie cherts of Scotland that shows that it is not only a true insect, but relatively derived, that is it had been around long enough to have accumulated some uniquely insect-like features. Although only the mandibles are preserved, it is possible that they once belonged to a winged insect. In any case, the fossil shows that the origin of insects is much earlier than previously thought.

Are Collembola Hexapoda?

• Hexapoda is a monophyletic group according to Hennig (1969), Boudreaux (1979), Kristensen (1981, 1991, 1998), Kukalová-Peck (1991), Stys & al. (1993), Wheeler & al. (1993), Kraus & Kraus (1994, 1996), Kraus (1998), Wheeler (1998), Willmann (1998) (all cited from Giribet, Edgecombe & Wheeler, 1999:203).
• Hexapoda is a monophyletic group according to Paulus (1979), Boudreaux (1979), Weygolt (1979, 1986), Brusca & Brusca (1990), Wägele (1993), Kristensen (1975, 1981, 1989, 1991, 1995, 1998), Bourgoin (1996), Deutsch (1997) (all cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:380), by the 18S manually aligned phylogeny of Kjer (2004), by the morphological phylogenies of Beutel & Gorb (2001:189) and Bitsch & Bitsch (2004).
• Among other deviations in Protura and Diplura from the basic ground plan, Collembola do not comply to the apomorphic characters of the Hexapoda, such as limbs 6-segmented (in Collembola: 5-segmented), and abdomen 11-segmented plus telson (in Collembola: 6-segmented); therefore, Wheeler, Whiting, Wheeler & Carpenter (2001:128-129) express some doubth on the monophyly of Hexapoda.
• Hexapoda is a polyphyletic group according to Anderson (1973, 1979), Manton (1977), Fryer (1992, 1996) (all cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:380).
• Hexapoda is a diphyletic group according to Tiegs & Manton (1958), Cisne (1974), Hessler & Newman (1975), Bergström (1980, 1992), Hessler (1992) (all cited from Bach de Roca, Gaju-Ricart & Compte-Sart, 1999:380).
• The molecular study of Wheeler (1998) does not support the monophyly of Hexapoda (Giribet, Edgecombe & Wheeler, 1999:204).
• The polyphyly of Hexapoda seems unacceptable based on morphological studies while many molecular studies indicate the reverse: Giribet & al (1996), Giribet & Ribera (1998), Spears & Abele (1998), Wheeler (1998) (all cited from Giribet, Edgecombe & Wheeler, 1999:204).
• The arthropod phylogeny based on ribosomal 18S rDNA does not support monophyly of Hexapoda (Giribet & Ribera, 2000:222).
• Support for the monophyly of Hexapoda is poor varying from 28% to 54%, depending on the model used for the phylogenetic molecular analysis (Buckley & Cunningham, 2002:397,399,400).
• Tillyard (1928) placed the Collembola at the basis of the Hexapoda (cited from Handschin, 1955:41). Based on the number of abdominal segments during embryological development (Collembola: 6; Protura: 8 to 11; Insecta: 11 or 12) he concludes that, contrary to Handlirsch, the low number of abdominal segments in Collembola, is not a secondary (derived due to reduction) but a primary character (protomorphic embryonical development). On the contrary, Bitsch & Bitsch, 2000:134 presumes that the low number of segments of the abdomen of Collembola is a secondarily acquired condition due to a more or less complete fusion of the segments.
• Except for the hexapody of the thorax, and the opisthogoneaty there is no other common character that supports the unification of Collembola and Hexapoda. Therefore, Handschin (1955:45,47) proposes to consider the Collembola as a separate, primitive arthropodan class.
• The six-legged condition so often found in insects is also characteristic of some other terrestrial arthropod taxa (Diplura, Protura, Collembola), but is probably convergently evolved in some or all of these as an "adaptive plateau minimum" reduction from a myriapodous condition. The name Hexapoda therefore covers a polyphyletic group and is best avoided for true insects (Brown in Parker, 1982:327).
• Hexapoda are polyphyletic according to Wheeler et al., 1993 (cited from Regier & Shultz, 1997:903).
• The monophyly of Hexapoda is provisionally accepted as a basic assumption by Bitsch & Bitsch, 1998:341,356 (based on Boudreaux, 1979; Kristensen, 1991, 1995, 1997; Bourgoin, 1996; Deutsch, 1997).
• Kluge (1999:352-359) supports the holophyly of Hexapoda based on six autapomorphies: hexapody, uniform structure of maxillae, fused labium, legs have a uniform set of segments, unique set of spiracles, and presence of corpora allata.
• Hexapoda is not supported as monophyletic group by the cladistic analysis based on sequence data of from two genes, ribosomal 18S and 28S rDNA (Carpenter & Wheeler, 1999:337-340,346).
• Molecular analysis rejects hexapodan monophyly (Giribet & Ribeira, 2000 cited from Nardi et al, 2001:1301).
• Phylogenetic analysis of the complete mitochondrial sequence of 15 arthropodans does not cluster Tetrodontophora bielanensis with Insecta, and therefore the monophyly of Hexapoda is rejected (Nardi et al., 2001:1301).
• The abscence of an antagonistic muscle to the depressor of the last podomere in Hexapoda, Chilopoda and Diplopoda has been considered as a common autapomorphy. However, Wolf & Harsch (2002:14) demonstrate that a similar arrangement is present in scorpions. Moreover, single muscles which lack an antagonistic counterpart but instead act against the passive bending of the joint by the body weight, haemolymph pressure, or elastic properties of the cuticle are common within all arthropod taxa. Rather than reflecting phylogenetic relationships, the emergences of these features may have been promoted by mechanical constraints, related to a marine versus a terrestrial life style. Indeed, reduction of distal muscles and proximal concentration of muscles that act on distal podomeres by a long tendon have a definite mechanical advantadge, namely, the avoidance of large inertial masses in distal positions which would be expensive to accelerate in each movement cycle (Wolf & Harsch, 2002:13-14).
• Phylogenetic analysis of the mitochondrial sequence of 35 arthropodans strongly supports the nonmonophyly of Hexapoda - that is the position of the Collembola outside the Crustacea + Insecta clade (Nardi et al., 2003:1888).
• Based on a phylogenetic analysis of nucleotide sequences, Collembola are placed at the base of Hexapoda, although with moderate to low support. With the data and methods currently available, the hypothesis of a common ancestry for extant hexapods cannot be rejected yet (Delsuc et al., 2003:1482d).
• The mono- or paraphyly of Hexapoda is disputable, since results differ depending on data sets and methods used in the phylogenetic reconstructions. (Nardi et al., 2003:1482e).
• A combined cladistic analysis of sequence data of five molecular markers and 189 morphological characters contradict the monophyly of Hexapoda by an alliance between Collembola, Crustacea and Ectognatha (excluding Diplura and Protura) (Giribet, Edgecombe, Carpenter, D'Haese, & Wheeler, 2004:319,325).

Are Collembola Crustacea?

• The among insects unique arrangement of the mouthparts of Collembola reminded Lubbock (1873:49) of the mouth of the Crustacea.
• Claypole (1898) conducted a detailed study on the embryonic development of Anurida maritma. Claypole (1898:278) concludes that Anurida shows characters allying it with crustaceans and myriapods rather than insects .
• Handlirsch (1908) considers Collembola as a more or less recent group of insects with an extreme specialisation. He considers them as forms with a retrograde development reaching maturity while in a larval state. (cited from Handschin, 1955:41,45). He points out that the antennae and mouthparts refer to a possible crustacean ancestry, and especially if one considers the number of abdominal segments (six) not as original but as a derived character (a reduction of the original number of segments) (cited from Handschin, 1955:41).
• The system of Tillyard (1930) adds the embryological and post-embryological development as a phylogenetic system from protomorphose, anamorphose to epimorphose. The protomorph Collembola are placed at the base of all forms. (Handschin, 1955:51).
• The dorsal organ of the collembolan embryo is more related with the embryonic organ of Crustacea, Limulus and Myriapoda than with the dorsal organ of the pterygotan embryo (Tiegs, 1942 cited from Schaller, 1970:51).
• Handschin (1955:51) considers Collembola as more primitive than the primitive Diplopoda and places them at the base of Diplopoda, who are separated from the Chilopoda that are more closely related to Insecta.
• Handschin (1955:52) concludes that Collembola and Crustacea especially have in common their embryological and post-embryological development.
• The segments of the headcapsule are characterised by the presence of an intercalary segment, bearing a specific sensorial organ, the post-antennal organ. The intercalary segment corresponds with the remnants of a premandibular segment, a segment typically well developed in Crustacea in which it bears the antennulae or secondary antennae (Handschin, 1955:42). Imms doubths its existance, even when Folsom observed the premandibula in Anurida. Denis demonstrated the presence of the vestigial premandibulae. The intercalary segment, between proto- and deuterocerebron, bears the post-antennal organ. Insecta lack such an organ (cited from Handschin, 1955:42).
  Young embryos of Anurida maritima have a small pair of appendages on the intercalary segment which is probably homologous to the second antennal segment of Crustacean (Tamarelle, 1984). These transient appendages may represent the second antennae of crustaceans which were 'lost' during evolution (cited from Hopkin 1997:22). A rudimentary pair of second antennae are briefly visible on the intercalary segment during development of the embryo of Tomocerus ishibashii (Uemiya & Ando, 1987 cited from Hopkin, 1997:142,143) but these quickly disappear, as in Anurida maritima (Tamarelle, 1984 cited from Hopkin, 1997:143).
• Tuxen showed that the tentorium supporting the mouthparts of the headcapsule does not correspond with that of Insecta, but that it corresponds with the suspension apparatus of the mouthparts of the Crustacea (Handschin, 1955:42-43,46). Handschin (1955:42) notes that the highly specialised and derived mouthparts have a doubthfull phylogenetic value.
• The post-embryonal development of Collembola is characterised by a periodical sexual maturity, synchronised by a specific moulting cycle. As in Crustacea, such as Astacus. "Le parallèle avec les Collemboles [and the Crustacea] est évident". (Handschin, 1955:47).
• Handschin (1955:48-49) discusses the relation between Collembola and Crustacea:
  - Collembola have holoblastic cleavage, as in Branchiopoda.
  - Collembola have a post-embryonal development cycle as in Astacus fluviatilis.
• Feustel (1958 cited from Schaller, 1970:29) points out that the rudimentary antennal nephridia of Collembola have much in common with those of the primitive crustacean Cladocera.
• in Hypogastruridae, Thibaud (1970:155-156) found that the vitellogenesis prolongs the intermoult cycle of females, such as in the crustacean Leander serratus (Drach 1955 cited from Thibaud, 1970:156) and in copepod crustacean Harpacticidae (Rouch, 1968 cited from Thibaud, 1970:156).
• The differentiation of the alimentary canal seems to relate the Collembola fairly closely to the conditions seen in Crustacea, where a gizzard in the form of a gastric mill is demarcated by cardiac and pyloric valves and a filter. The filter region in Collembola collectively appears to perform a similar function to that performed by the filter in Crustacea. (Adams & Salmon, 1972:281).
• Those Collembola species that lack a mandibular molar plate, although essentially terrestrial animals, persist with an aquatic type of feeding on the water film which surrounds soil particles. This type of feeding is called suspension feeding, refering to the enormous numbers of bacteria and protozoa suspended in the water film. (Adams & Salmon, 1972:283).
• Paulus (1972) compares the composed eye of Collembola with a reduced compound eye which is remarkably similar in structure to crustacean eyes (cited from Hopkin 1997:22).
• Paulus (1974:124) compares the additional 'functionless primary pigment cells' of the arthropleonic ommatidium with the corneagenous cells of the crustacean ommatidium.
• Paulus (1974:128) suggests a phylogeny of the corneagenous and primary pigment cells of the mandibulate ommatidium in a progressive pathway from protomandibulate crustacean ancestor to Insecta, followed by a regressive pathway from Insecta to Anurida maritima. Note that the ancestral protomandibulate crustacean ommatidium compares very well with the ommatidium of Anurida maritima (Paulus, 1974:128,fig.4A and 4H). This might suggest an alternative phylogeny in which the collembolan ommatidium evolved indepently from the insectal ommatidium. Both originated from the same ancestral crustacean ommatidium, in which the pancrustacean ommatidium remains monophyletic. Paulus (1974:130,131) does not consider the myriapodan eye as an ommatidium.
  The structure of the ommatidia is very similar to those of Crustacea lending support to the sister group status of Hexapoda/Crustacea (Hopkin, 1997:69).
• Paulus (1979) assumed that the basic number of frontal eyes (ocelli) was two pairs, innervated by the protocerebrum (as in Branchiopoda) (cited from Bitsch & Bitsch, 2000:136). Collembola have typically retained four frontal ocelli, showing a primitive structure (Paulus, 1979 cited from Bitsch & Bitsch, 2000:136).
• Physiological data show that springtails evolved directly from marine ancestors: haemolymph with high osmotic pressures and mainly composed of inorganic salts. The secondary transition from terrestrial habitats to fresh water is physiologically much more convincing (Little, 1983, 1990, cited from D'Haese, 2003:583).
• At the time springtails appeared, the soil and litter habitats were not yet present; the first terrestrial arthropods lived in algal mats and emergent vegetation; they then would have adapted to living in soaked terrestrial substrates and, only afterwards, colonised the ground (Shear & Kukalová-Peck, 1990, cited from D'Haese, 2003:565). 'Early Collembola were semi-aquatic' (Kukalová-Peck, 1991, cited from D'Haese, 2003:565).
• The question as to whether the myriapods or crustaceans are more closely related to hexapods is much debated (Osorio et al., 1995; Nielsen, 1995; Telford & Thomas, 1995; Dohle, 1997; Kraus, 1997; Wheeler, 1997; cited from Bitsch & Bitsch, 2000:132).
• Hexapoda are more closely related to branchiopod crustaceans than to myriapods, based on molecular phylogeny of Arthropoda (Regier & Shultz, 1997:902,907,908,909) (Field et al., 1988; Patterson, 1989; Friedrich & Tautz, 1995; Boore et al.,1995; cited from Regier & Shultz, 1997:903).
  Hexapoda may have originated from freshwater 'crustaceans' rather than from terrestrial lobopods or some unknown marine lineage (Manton, 1977; Little, 1990; Meglitsch & Schram, 1991; Averof & Cohen, 1997; cited from Regier & Shultz, 1997:911).
• The colonisation of terrestrial environments occured idependently for each apterygote hexapod lineage (Averof & Cohen, 1997, cited from D'Haese, 2003:566).
• Arthropod phylogeny inferred by 18S rDNA shows that Crustacea are not monophyletic. Collembola are a sistergroup of Branchiopoda (Spears & Abele, 1997 cited from Lange & Schram, 1999:236).
• Collembola have been clustered within Crustacea in other molecular and/or combined data sets (Spears & Abele, 1997; Giberet & Ribera, 2000:212-218; cited from Nardi et al., 2003:1887).
• Based on neurogenesis and molecular data, findings are in favour of a close relationship between Hexapoda and some crustacean groups (Whitington et al., 1993; Friedrich & Tautz, 1995; Osorio et al., 1995; cited from Bitsch & Bitsch, 1998:341).
• A total (holoblastic) and almost equal cleavage occurs in Collembola as also known to occur in many Crustacea (Bitsch & Bitsch, 1998:350-351; Bitsch & Bitsch, 2000:144).
• The absence of the amniotic cavity around the early germ band in Collembola, as in Crustacea (Bitsch & Bitsch, 1998:351; Bitsch & Bitsch, 2000:144).
• Mushroom bodies are prominent neuropils, found in annelids and all arthropod groups, except in Crustacea and Collembola (Strausfeld et al., 1998:11,23,25,26).
• The molecular study of Wheeler (1998) suggests relationships between Collembola (and Protura and Diplura) and certain crustacean groups (Giribet, Edgecombe & Wheeler, 1999:204).
• Lawrence (1999:2,13-18; 2004:622,627) discusses possible homologies between Crustacea and Collembola, such as the 'accessory flagellum' of the third segment of the first antenna of crustaceans with the sensory organ of the third segment of the antenna in Collembola, the apical sensory organ of the second antenna of crustaceans with the post-antennal organ of Collembola, the endopodites of Crustacea with the dental hooks of the furca in Collembola, the gills in crustaceans with the eversible vesicles of the collophore in Collembola.
  However, a teratological specimen of Brachystomella parvula, in which the most anterior ocellus of the eye patch is replaced by an additional post-antennal organ (Cardoso, 1966:83), makes the apical sensory organ of the crustacean second antenna homology of the post-antennal organ questionable.
• The arthropod phylogeny based on ribosomal 18S rDNA clusters Collembola with basal crustaceans (Giberet & Ribera, 2000:213,214,215,217,218).
• The intertidal Anurida maritima has an endogenous tidal rhythm of positive and negative phototaxis (Manica, McMeechan & Foster, 2000:371). This behaviour is often present in intertidal crustaceans, such as Cibanarius misanthropes - the hermit crab - (Drzewina, 1907 cited from Manica, McMeechan & Foster, 2000:371), Uca pugnas - the fiddler crab - (Palmer, 1964 cited from Manica, McMeechan & Foster, 2000:371), and many planktonic larval stages of crustaceans (Douglass & al., 1992 and Warman & al., 1993 cited from Manica, McMeechan & Foster, 2000:371).
• Molecular studies reject the monophyly of Crustacea and suggest that Malacostraca are more closely related to Insecta than Branchiopoda (Wilson et al, 2000 cited from Nardi et al., 2001:1301-1302).
• Phylogenetic trees based on the mitochondrial cytochrome c oxides subunit II gene show that Collembola branch out first from the Crustacea, then followed by Protura and Diplura. Thus, three terrestrial wingless groups are placed between archetypical aquatic Crustacea and terrestrial Pterygota. (Shao, Zhang, Ke, Yue & Yin, 2000:595).
• Phylogenetic analysis of the complete mitochondrial sequence of 15 Arthropoda rejects the monophyply of Crustacea. Malacostraca and Branchiopoda do not form a monophyletic group, with the former being more closely related to the Pterygota (Nardi et al., 2001:1301).
• A combined cladistic analysis of sequence data of five molecular markers and 189 morphological characters showed an alliance between Collembola, Crustacea and Ectognatha (excluding Diplura and Protura) (Giribet, Edgecombe, Carpenter, D'Haese, & Wheeler, 2004).

Are Collembola Pancrustacea?

• According to Beutel &: Gorb (2001:178), hypotheses suggesting closer relationships between Hexapoda and Crustacea are not consistent (Adoutte & Philippe, 1993 cited from Beutel & Gorb, 2001:178), (Averoy & Akam, 1995 cited from Beutel & Gorb, 2001:178), (Friedrich & Tautz, 1995 cited from Beutel & Gorb, 2001:178), (Regier & Shultz, 1997:902-913), (Wheeler, 1998 cited from Beutel & Gorb, 2001:178).
• The chiasma of the nervi corporis cardiaci 1 and the presence of the corpora allata relate Collembola more close with Insecta than with Myriapoda, in which the homolog nerves do not form a chiasma and in which corpora allata are absent (Juberthie & Cassagnau, 1971:76-77); this provides indirect support for the Pancrustacea hypothesis.
• The homology of tracheae between Myriapoda, Collembola, Protura, Diplura and Ectognatha has been rejected by several workers (Kraus & Kraus, 1994, 1996; Dohle, 1997; see Hilken, 1998 for thorough study; cited from Edgcombe & Giribet, 2002:151).
• The homology of the Tömösvary organs of Chilopoda, Pauropoda and Diplopoda with the postantennal organs of Collembola and pseudoculi of Protura, has been variably questioned (Bourgoin, 1996 cited from Edgecombe & Giribet, 2002:151) or defended (Haupt, 1979 cited from Edgecombe, Giribet & Wheeler, 1999:310,330) (Bitsch & Bitsch, 1998 cited from Edgecombe & Giribet, 2002:151).
• Analysis of the 18S rRNA gene sequences data of Arthropoda supports the monophyly of Pancrustacea (Crustacea and Hexapoda) and the polyphyly of Crustacea (Aleshin & Petrov, 1999:184,185). Bootstrap analysis shows that Collembola relate to Crustacea as sistergroup of Pentastomidia:Porocephalus and Crustacea:Argulus. Parsimonous trees show that Collembola relate to Crustacea as sistergroup of Crustacea:Limnadia and Crustacea:Artemis.
• Analysis of molecular marker expression reveals neuronal homology in Insecta, Malacostraca, Branchiopoda and Collembola (Duman-Scheel & Patel, 1999:2332) supporting a common pancrustacean ancestor.
• An arthropod phylogeny based on ribosomal 18S rDNA supports the monophyly of Pancrustacea (Giberet & Ribera, 2000:213,214,215,216,217,218,222-224) and places Collembola at the base of Pancrustacea (Giberet & Ribera, 2000:214).
• A phylogeny, applying the principle of total evidence, using molecular and morphological characters, strongly supports the monophyly of Pancrustacea (Giribet, Edgecombe & Wheeler, 2001:160).
• Support for the monophyly of Pancrustacea varies inbetween 26% to 77% depending on the model used for the phylogenetic molecular analysis (Buckley & Cunningham, 2002:397).
• A phylogenetic analysis of the mitochondrial sequence of 35 arthropodans strongly supports the monophyly of the Pancrustacea hypothesis (Nardi et al., 2003:1887). Based on a phylogenetic analysis of the complete mitochondrial genome of 35 arthropods, Collembola are placed at the base of Pancrustacea (Nardi et al., 2003:1888). This conclusion is challenged based on methodological grounds (Delsuc et al., 2003:1482d).

Exploring possible extra-pancrustacean ancestries of Collembola:

Do Collembola share characters with Arachnata?

• In most mites the tracheal orifices are situated at the base of the legs, yet in Trombidium holosericeum the spiracles are two in number, and, as in Smynthurus(sic), are situated at the lower side of the head, though not exactly in the same place, since in this species they open on the inner side of the basis of the mandibles [in stead of between head and prothorax] (Lubbock, 1873:78).
• The system of Versluys (1928) introduced the following morphological aspects: cerebralisation, development of the protocerebron, and development of the compound eye. He ommits Apterygota. He places Collembola with the Endotropha at the base of the third etage, parallel to Arachnida; Thysanura at the fourth etage, opposed to Crustacea (cited from Handschin, 1955:50-51).
• The dorsal organ of the collembolan embryo is more related with the embryonic organ of Crustacea, Limulus and Myriapoda than with the dorsal organ of the pterygotan embryo (Tiegs, 1942 cited from Schaller, 1970:51).
• The embryonal development of Collembola follows a holoblastic cell cleavage that continues up to the stage of 32 blastomers. Then most of the blastomers migrate to the egg surface and form the primitive blastoderm (Handschin, 1955:44). This for hexapodans exceptional embryogenese can also be found in some terrestrial Acari (Pediculopsis) (Handschin, 1955:44,46).
• The tentorium of Collembola should more correctly be termed an endosternum as it closely resembles this structure in the Arachnida (Tuxen cited from Adams & Salmon, 1972:276).
• The spermatozoon of Collembola exhibits certain peculiarities proper to the most primitive Arthropoda: the mature sperm is rolled up in a cyst like in Pseudoscorpions and Araneids, the bi-stratified acrosome lacks any extra-acrosomial material, the internal endo-nuclear cylinder is of the type found in Limulus, the mitochondria are almost unmodified, and above all the flagellum still has the basic 9+2 axoneme typical of all flagella (Dallai, 1974:153).
• In bootstrap analysis of the 18S rRNA gene sequences data of Arthropoda Collembola turns up, at least in one tree, as the sistergroup of Crustacea + Insecta (Aleshin & Petrov, 1999:184,185).
• The box-truss axial muscle system should be regarded as synapomorphic for Arthropoda and plesiomorphic for Chelicerata (Edgecombe & al., 2000 cited from Shultz, 2001:297). The loss of all anterior oblique axial muscles is synapomorphic for Arachnida (Shultz, 2001:297). Note that in Orchesella and Neanura all thoracic and abdominal anterior oblique axial muscles are lacking (Bretfeld in Schaller, 1970:30-33), except those of the furcal segment.
• Sperm winding in Collembola is shared with arachnidan pseudoscorpions and spiders; it differs by presence of an extracellular material around which the sperm components enroll (Dallai, Fanciulli, Frati, Paccagnini & Lupetti, 2002:497-498).

Discussion

Conventional hypothesis: Collembola are basal Hexapoda.

Tab.II. Classification of related and higher taxa of Collembola.

Pancrustacea
Crustacea	Hexapoda s.l.
	Collembola	Protura	Diplura	Insecta s.s. (= Ectognatha)
				Archaeognatha (= Monocondylia)	Dicondylia
					Zygentoma	Pterygota

Fig.1. Simplified scheme of the systematic position of
Collembola in the Arthropoda
(conventional phylogeny)

+---------- Insecta s.s.  +--+ | +---------- Diplura +--+ | | +---------- Protura | +--+ | +---------- Collembola +--+ | +---------------- Crustacea +--+ | | +------------- Arachnida | +-----+ ---+ +------------- Xiphosura | +---------------------- Myriapoda

Alternative hypothesis: Collembola are terrestrial Crustacea.

Tab.III. Classification of related and higher taxa of Collembola.
(modified after Shao, Zhang, Ke, Yue & Yin, 2000)

Pancrustacea
Crustacea s.l.		Hexapoda s.s.
Crustacea s.s.	Collembola	Protura	Diplura	Insecta s.s. (= Ectognatha)
				Archaeognatha (= Monocondylia)	Dicondylia
					Zygentoma	Pterygota

Fig.2. Simplified scheme of the systematic position of
Collembola in the Arthropoda
(modified after Shao, Zhang, Ke, Yue & Yin, 2000)

+------ Hexapoda s.s.  | +------------- Collembola | | +-----+------+------ Crustacea +--+ | | +------------- Arachnida | +-----+ ---+ +------------- Xiphosura | +---------------------- Myriapoda

Alternative hypothesis 2: Collembola are terrestrial basal Crustacea.

Tab.IV. Classification of related and higher taxa of Collembola.
(modified after Spears & Abele, 1997 cited from Lange & Schram, 1999; Aleshin & Petrov, 1999; Gibiret & Ribera, 2000:213,215,216,217,218)

Pancrustacea
Crustacea s.l.			Hexapoda s.s.
Crustacea in pars (Branchiopoda)	Collembola	Crustacea in pars (Malacostraca)	Protura	Diplura	Insecta s.s. (= Ectognatha)
					Archaeognatha (= Monocondylia)	Dicondylia
						Zygentoma	Pterygota

Fig.3. Simplified scheme of the systematic position of
Collembola in the Arthropoda
(modified after Spears & Abele, 1997
cited from Lange & Schram, 1999; Aleshin & Petrov, 1999;
Gibiret & Ribera, 2000:213,215,216,217,218)

+----- Hexapoda s.s.  | +---+----- Crustacea partim  | +------------- Collembola | | +-----+---+--------- Crustacea partim  +--+ | | +------------- Arachnida | +-----+ ---+ +------------- Xiphosura | +---------------------- Myriapoda

New hypothesis: Collembola are terrestrial basal Pancrustacea.

Tab.V. Classification of related and higher taxa of Collembola.
(modified after Aleshin & Petrov, 1999:184,185; Gibiret & Ribera, 2000:214; Nardi et al., 2003)

Pancrustacea
Collembola	Crustacea	Hexapoda s.s.
		Protura	Diplura	Insecta s.s. (= Ectognatha)
				Archaeognatha (= Monocondylia)	Dicondylia
					Zygentoma	Pterygota

Fig.4. Simplified scheme of the systematic position of
Collembola in the Arthropoda
(modified after Aleshin & Petrov, 1999:184,185;
Gibiret & Ribera, 2000:214; Nardi et al., 2003)

+------------- Hexapoda s.s.  +--+ | +------------- Crustacea +--+ | +---------------- Collembola +--+ | | +------------- Arachnida | +-----+ ---+ +------------- Xiphosura | +---------------------- Myriapoda

New hypothesis 2: Collembola are terrestrial basal Pancrustacea (Hexapoda rejected).

Tab.VI. Classification of related and higher taxa of Collembola.
(modified after Giribet & al., 2004:327)

Protura	Diplura	Pancrustacea
		Collembola	Crustacea	Insecta s.s. (= Ectognatha)
				Archaeognatha (= Monocondylia)	Dicondylia
					Zygentoma	Pterygota

Fig.5. Simplified schemes of the systematic position of
Collembola in the Arthropoda
(modified after Giribet & al., 2004:327)

+------------- Ectognatha   +--+ | +------------- Crustacea +--+ | +---------------- Collembola +--+ | | +------------- Arachnida | +- - -+ ---+ +------------- Xiphosura | | +----- Diplura +----------------+ | +----- Protura | +---------------------- Myriapoda

+---------------- Ectognatha   | +--+---------------- Collembola | | | +---------------- Crustacea +--+ | | +------------- Arachnida | +- - -+ ---+ +------------- Xiphosura | | +----- Diplura +----------------+ | +----- Protura | +---------------------- Myriapoda

Provisional conclusion

The recent discovery of the marine hexapod Devonohexapodus bocksbergensis Haas, Waloszek & Hartenberger, 2003:42-53 from the Lower Devonian challenges the traditional association between terrestralisation and hexapody, making room for alternative hypotheses concerning collembolan origins (Nardi et al., 2003:1482e). The recent reinterpretation of a fragmentary insect fossil from the early Devonian Rhynie cherts of Scotland shows that the origin of insects is much earlier than conventionaly accepted (Engel & Grimaldi, 2004:627-630). This indicates that insects evolved independently in parallel with collembolans. Given the edaphic origin of Collembola (D'Haese, 2002), the collembolan ancestor must have completed the transition from marine aquatic habitats to littoral soil habitats already before the Lower Devonian.

Acknowledgments

References

• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• Bitsch, C. & Bitsch, J. 2004. Phylogenetic relationships of basal hexapods among the mandibulate arthropods: a cladistic analysis based on comparative morphological characters. Abstract cited from the Web of Science., Zoologica Scripta 33 (6), (2004), p.511-550.
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• Dove, H. & Stollewerk, A. 2003. Comparative analysis of neurogenesis in the myriapod Glomeris marginata (Diplopoda) suggests more similarities to chelicerates than to insects., Development 130, (2003), p.2161-2171.
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• Kjer, K.M. 2004. Aligned 18S and insect phylogeny. Abstract cited from the Web of Science., Systematic Biology, 53 (3), (2004), p.506-514.
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• Shultz, J.W. 2001. Gross muscular anatomy of Limulus polyphemus (Xiphosura, Chelicerata) and its bearing on evolution in the Ararchnida., The Journal of Arachnology, 29, p.283-303.
• Wolf, H. & Harzsch, S. 2002. Evolution of the arthropod neuromuscular system. 1. Arrangement of muscles and innervation in the walking legs of a scorpion: Vaejovis spinigerus (Wood, 1863) Vaejovidae, Scorpiones, Arachnida., Arthropod Structure & Development, xx, 2002, p.yy-zz.