Homology and morphology in Poduromorpha ( Hexapoda , Collembola )

The detailed external morphology and general anatomy of the majority of families and subfamilies of Poduromorpha were investigated and compared. This examination was done in relation to the three remaining orders of Collembola: Entomobryomorpha, Symphypleona and Neelipleona. Within Poduromorpha, homologies among the different families and subfamilies are established for general chaetotaxy, and chaeotaxy of head, buccal cone, sensilla of fourth antennal segment, anal valves and tibiotarsus. A consistent and comprehensive nomenclature is proposed for these morphological features unifying those applied in scattered existing studies. Observations on first instars are reported. Comparison of first instars and adult chaetotaxy suggests a possible paedomorphic trend in the evolution of Poduromorpha.


INTRODUCTION
The external anatomy of springtails (class Collembola) has been investigated for a long time (e.g., Bourlet, 1839;De Geer, 1743;Lubbock, 1873), starting consistently with the work of Börner (e.g., Börner, 1901, 1906, 1913).Since then, our knowledge of springtail morphology has increased significantly.However, most studies have focused on single groups of different levels: Subfamily (e.g., Rusek, 1976), genus (e.g., Yosii, 1962) or even a single species (e.g., André, 1988).Almost no attempts have been made to generalize to the other groups, with the exception of a comprehensive work on the labial palps (Fjellberg, 1999).Consequently, several nomenclatures exist for the same structures according to different groups and usually the homologies have not been clarified.
The main goal of the present work is to establish homologies of the morphological features among the dif ferent groups in Poduromorpha, and to do this in relation to the other orders when possible.It is also my intention to provide a comprehensive nomenclature unifying the scattered and incomplete studies available in the literature.
Springtails are a diverse group with more than 6500 species described to date (Rusek, 1998), This probably represents a small part of their total specific diversity, a diversity that could reach more than 50 000 species (Hopkin, 1997;Hopkin, 1998).Poduromorpha comprises more than a third of the described species of springtails with about 2500 species in approximately 260 genera.
The class Collembola is traditionally divided in two groups (e.g.Börner, 1906;Cassagnau, 1990;Gisin, 1960;Hopkin, 1997;Massoud, 1971b;Uchida, 1971), Arthropleona and Symphypleona sensu lato.The Symphypleona s.l. are comprised of the orders Symphypleona sensu stricto and Neelipleona.Members of this group are globular in shape with fused abdominal segments.Arthro-pleona are characterized by possession of an elongate body with abdominal segments 1-4 clearly separated from each other by an intersegmental membrane.Arthropleona is comprised of the orders Poduromorpha and Entomobryomorpha.The Poduromorpha have a distinct prothorax that bears dorsal setae and third and fourth abdominal segments that are similar in size.Generally Entomobryomorpha have a greatly reduced prothorax that never bears setae, as in Symphypleona s.l., and the fourth abdominal segment is two to four times longer than the third segment.Exceptions to this are Isotomidae, Oncopoduridae and Tomoceridae, which have similar sized abdominal segments as in Poduromorpha.
Specimens were preserved in 95% ethyl alcohol; dissections of mouthparts and other structural features were made in that medium.They were cleared in successive 10% potassium hydroxide solution and Amann's Lactophenol (lactic acidglycerol-phenol: 25-50-25).The Lactophenol medium was slightly warmed up for the most pigmented specimens.Speci mens were mounted on microscope slides using Marc André II mounting medium (see Massoud, 1967).The slides were slightly warmed up with an alcohol burner in order to improve specimen position in the slide for a favorable observation (slight inflation).
Specimens were examined using a compound microscope with differential interference contrast optics at magnifications ranging form 250 to 1250 (oil immersion lens for the higher magnifications).Drawings were made with a drawing tube.

General aspects
Poduromorpha are elongate springtails with a clearly segmented body, cylindrical in shape or slightly flattened.Contrary to other springtails, the prothorax is always developed and bears dorsal setae.Members of this group are generally small, with some species as small as 0.4 mm in length (e.g. in the genera Mesaphorura Börner, 1901 andWillemia Börner, 1901).However, some species can grow to a centimeter in length (e.g. in the genera Tetrodontophora, Holacanthella Börner, and Caledonimeria Delamare Deboutteville & Massoud, 1962).
Poduromorpha are usually uniformly pigmented: many are white (Onychiurinae, Tullbergiinae, Willemia, some Anurida Laboulbène, 1865 etc.) while most of the others present a monochromatic pigmentation limited to blue or red.The degree of pigmentation is correlated to latitude and altitude; the number of dark species increases near the poles and in mountains (Rapoport, 1969).Rapoport (1971) considered that the darker species warm up more quickly in the sun and are more active than the lighter colored species.However, dark species are actually found on sun-exposed places as the littoral seashore (e.g. the dark-blue Anurida maritima (Guérin-Méneville, 1836), and the dark Actaletidae) and several studies showed that Collembola were sensitive to UV radiation and suggested that dark pigmentation plays a protective role in exposed species (Verhoef et al., 2000;Zinkler & Wilking, 1989).Species with patterns of different colors are exceptional: there are a few bi-colored species that are usually dark blue and yellow (Arlé & Rufino, 1976;Mendoça & Fer nandes, 1999), occasionally red and yellow or rarely tri colored red, yellow and white (Cassagnau & Deharveng, 1984).More generally, pigmentation in Poduromorpha is much more uniform than in Entomobryomorpha and Symphypleona.However, knowledge on pigmentation of Collembola remains very fragmentary insofar as it has not been studied extensively (e.g.Bouthier & Thibaud, 1974).
Cuticle.Examination of the surface of the cuticle using a scanning electron microscope indicates a characteristic granulation (Dallai, 1973;Lawrence & Massoud, 1974;Massoud, 1969;Massoud & Barra, 1980;Massoud & Betsch, 1973).The ornamentation of the cuticle presents mainly three degrees of complexity: 1.Primary granules: elementary granules, generally triangle-shaped, that are regularly distributed and linked to one another by ridges and thus forming a hex agonal pattern (Massoud, 1969).Sometimes, these ele mentary granules are square-shaped, resulting in a squared pattern (as in Podura aquatica); 2. Secondary granules: one or several of the primary granules are deformed and enlarged to form protuber ances (Dallai, 1973;Hale & Smith, 1966).Secondary granules are found in Poduromorpha and Symphy pleona but not in Entomobryomorpha (Massoud, 1969;Massoud & Thibaud, 1979); 3. Tertiary granules: the grouping of several secondary granules forms the tertiary granules characteristic of Neanurinae (Deharveng, 1983).The tubercles are a result of the tertiary granules structure deforming the cuticle.Tetrodontophora bielanensis also possess ter tiary granules that resemble "pseudo-scales" (Massoud & Betsch, 1973).However, this structure is not homologous to the tubercles found in Neanurinae.
Pseudocelli.Onychiuridae possess characteristic cuticular structures known as pseudocelli, which are unique to this group.Three main types of pseudocelli were defined (Rusek, 1984;Rusek & Weyda, 1981): 1. Simple pseudocelli are found in Tullbergiinae.The pseudocellar lid is weakly separated from the integu ment and weakly differentiated into one or a few flaps arranged into a rosette or tangentially.The granulation of the pseudocelli is a primary granulation or is the same as the surrounding integument (Rusek & Weyda, 1981).These pseudocelli are big in comparison to the other types of pseudocelli; In Tullbergiinae, pseudocelli were classified into four categories depending on the shape and the number of the flaps and their position (Weiner & Najt, 1991b); 2. The pseudocelli in Onychiurinae, and Tetrodontophorinae have a smooth pseudocellar lid, which is separated from the surrounding coarse secondary granulated integument by a distinct circular furrow of primary or low secondary granulation.The pseudo cellar lid bears 6+6 to 20+20 ribs in two parallel rows delineated by a central superficial groove (Dallai, 1973); 3. Rusek (1984) described a third type of pseudocelli found on the trochanters of members of the genus Onychiurus Gervais, 1841 and named them "trochanteral pseudocelli".They are smaller than type 2 pseudocelli and lack a circular furrow.These pseudocelli are also found on other part of the body: On the subcoxae in Allonychiurus Yoshii, 1995(= Paronychiurus Bagnall, 1948;see Weiner, 1989b); on anal valves in Onychiurinae (Christian, 1989); on the first antennal seg ment in Tetrodontophora bielanensis (see Christian, 1989); and on various body segments of Hymenaphorura Bagnall, 1948 both ventrally anddorsally (Pomorski, 1990;Weiner & Fjellberg, 1994).Weiner & Fjellberg (1994) proposed to name these structures parapseudocelli.
Pseudocelli are completely absent on the ventral side and legs of Tullbergiinae (Rusek, 1984) and Tetrodontophorinae.
Several hypotheses were developed to explain the role of pseudocelli.Koncek (1924) advocated a haemorragic function.Others proposed both an excretion and a defense function against predators by extruding repellent fluids (Hale & Smith, 1966;Usher & Balogun, 1966).Rusek and Weyda (1981) considered the excretion func tion only.
In addition to pseudocelli, tegument structures are also found in other groups; particularly the pseudopores observed in Oncopoduridae (Szeptycki, 1977), Entomobryidae, Tomoceridae, and Isotomidae (Deharveng, 1978).According to morphology and position of pseudo pores, one could wonder if these structures would be homologous to the pseudocelli.A detailed study using the electron microscopy would be necessary to establish or reject that hypothesis.
Chaetotaxy of s setae.Four main types of setae are possessed by springtails: Ordinary setae, sensilla (also named setae s), scales and trichobothria (Massoud & Ellis, 1977).Neither scales and trichobothria, nor multi laterally ciliate setae are found in Poduromorpha.Sensilla of the body and of the antennae are distinguished here: The term "sensilla" is used for those of the antennae; and "setae s" refers to sensilla of the body.In comparison to ordinary setae, setae s are smooth, blunt and have a dif ferent refractiveness under the compound microscope.They are usually subcylindrical, but can be candle flame shaped, "francisque"-shaped, etc. (Massoud & Ellis, 1977).
Within the Poduromorpha, setae s chaetotaxy is stable (with the exception of Onychiurinae and Tetrodontophorinae) compared to ordinary chaetotaxy, which is often plurichaetotic (Deharveng, 1979a).On the contrary, chaetotaxy of fourth antennal segment can present plurichaetoses or paurochaetoses of the sensilla independently in different groups.
The four families Hypogastruridae, Odontellidae, Neanuridae and Isotogastruridae possess a stable and scarce setae s chaetotaxy: 2+2 setae s (1+1 setae s in the m row, and 1+1 in the p row), and 0+0 or 1+1 microsensilla on meso-and metathorax, 1+1 setae s (in p row) on tergum of abdominal segments I to V. Podura aquatica has the same setae s chaetotaxy, even if setae s are indis tinct from ordinary setae.Brachystomellidae are slightly different: some species have 2+2 setae s on abdominal segments.Head, prothorax, and sixth abdominal segment never bear setae s in Hypogastruridae, Odontellidae, and Neanuridae.
Setae s chaetotaxy of Onychiuridae is not well differen tiated and difficult to interpret insofar that these setae are very difficult to distinguish from ordinary setae.More precisely, Onychiurinae have an abundant and not welldifferentiated chaetotaxy of setae s (Deharveng, 1979a).Some species, for example in the genera Psyllaphorura Bagnall, 1948, Oligaphorura Bagnall, 1949, and Tantulonychiurus Pomorski, 1996 possess setae s on the head.In the two latter, there is 2+2 setae s on the head (or 3+3 in some Psyllaphorura, see Weiner & Najt, 2000) and on abdominal segments I to III and 1+1 on the sixth (Pomorski, 1998;Weiner, 1996).Tullbergiinae have a setae s chaetotaxy which is difficult to determine, but it seems stable compared to Onychiurinae setae s chaetotaxy and closer to Hypogastruridae one.Particularly, 1+1 setae s are clearly distinct on the fifth abdominal segment.

Head
The position of the head has been a taxonomic criterion used by some authors to establish phylogenetic relation ships (e.g., Paclt, 1956;Salmon, 1964).Both authors con sidered Arthropleona prognathous and Symphypleona s.l. and Podura hypognathous.This was the major justifica tion for placing Podura aquatica within the Metaxypleona, an ensemble including Actaletidae, Metaxypleona being sister group to the Symphypleona s.l.(Moen & Ellis, 1984;Salmon, 1964).However, according to my own observations, Podura aquatica is prognathous when alive.The head bends downward with sudden death (when put in alcohol for example).It is important to note that the hypognathous position in Symphypleona is corre lated with the important development of the clypeal area.This is not the case with Podura aquatica.The number of clypeal setae, or pre-labral setae, is reduced to two or four in Poduromorpha whereas there are more in Podura The head also possess two different kinds of sense organs: eyes and PAO.
1. Springtails have eight simple eyes on each side of the head.These ocelli always have the same position with respect to each other.These have been annoted A to H (Guthrie, 1906).In many lineages, this number is reduced to complete absence, especially in euedaphic species (Thibaud, 1976;Thibaud & Massoud, 1973).2. The postantennal organs (PAO) are located on each side of the head in front of the ocular area, posterior to the base of the antennae.The precise function of this sense organ is unknown.It could have an olfactory function (Dallai, 1971), as well as sensitivity to tem perature and humidity (Altner & Thies, 1976).It is reduced to a setae-like structure, a "peg", in Symphypleona (Dallai, 1971) and absent in some Entomobryomorpha (Entomobryidae).In Poduromorpha, the PAO is composed of a variable number of vesicles and com prises a sensory cell in which the perikaryon is located in the central nervous system (Karuhize, 1971).The shape, number, structure of vesicles and the number of rows as well as the position of the vesicles in these rows can be highly variable in the different groups.These are fundamental criteria in Onychiuridae tax onomy (Pomorski & Skarzynski, 1992;Weiner, 1996), and to a lesser extent in other groups like Pseudachorutinae (Massoud, 1967).The postantennal organ is absent in Neanurinae.Based on the connections to the nervous system, some authors considered the PAO homologous to the Tomosváry organ in Myriapoda (Denis, 1926;Willem, 1900).François (1969) proposed that the pseudoculi in Protura are homologous to the Tomosvary organ, whereas Tuxen (1959) considered them vestiges of antennae.Recent studies tend to accept the homology of these organs among the three groups (Bitsch & Bitsch, 1998;Edgecombe et al., 2000;Haupt, 1979).
Head chaetotaxy.Homologies of cephalic setae have been established with several morphological landmarks, mainly on the basis of: 1. Relative positions of the setae to each other; 2. Relative position of the setae to the muscular insertion marks on the cuticle (Najt & Rubio, 1978); 3. Relative position of the setae to the transversal fissure separating the frontal area from the basis of the head and the fissures demarcating the ocular areas.
In Table 1, are proposed the correspondences between cephalic nomenclatures proposed by Cassagnau (1974) and Najt & Rubio (1978) for the Hypogastruridae and Tullbergiinae and the nomenclature proposed by Yosii (1956) and da Gama (1964) for the Neanurinae.
These correspondences were partially given by Cas sagnau (1980).However, he considered A = sd1, O = d1, B = d2, and E = sd4.The present proposal (Table 1) is probably more accurate insofar as in Cassagnau's system the unpaired setae replace the d1 setae.Then, it assumes sd row with 6 setae (one more posterior to sd1) in the genus Friesea Dalla Torre, 1895 (Fig. 2C), and B setae (supposed to be d2) would be distant from the axis of d row in Neanurinae (Fig. 2E).Primary homology criterion leads to rejection of that hypothesis.The replacement of two symmetric setae by one seta in the axis (paurochaetosis by axial substitution) presents an intermediary state in species with two asymmetrical setae (Deharveng, 1983), as in some species in the genus Anurida for exam ple.Jordana et al. (1997) proposed a nomenclature for cephalic setae in Onychiurinae.However, the interpreta tion of the d row is erroneous in this work as d5 do exist and d0 is the unpaired expression of setae d1, (already stated by Pomorski, 1996).Setae d5 is not homologous to the unpaired setae d in Friesea (in this genus d2 is unpaired).In the genus Triacanthella Schaffer, 1897 there is also an unpaired setae d.It is probably homolo gous with d0 setae in Onychiurinae.However, homolo gies are difficult to established in Triacanthella because of the plurichaetoses.
Part of the diversity of cephalic chaetotaxy in Poduromorpha is presented in Figs 1B-F and 2A-E.Chaetotaxy of a first instar of Folsomia candida (Entomobryomorpha, Isotomidae) is provided in Fig. 1A.Row sd' in Onychiuridae is probably homologous to row oc.According to observations in Onychiurus first instar, sd'1 = oc1 and sd'2 = oc2 (oc2 is always a big setae compared to the other oc setae) (personal observations and Pomorski, 1996).In Onychiurinae, there are several rows of setae (rows cm, cb, cp) between rows v and p (Fig. 1D).These rows are not found in other Poduromorpha, which possess row c only (Fig. 1C).Along the same vein, Onychiurinae can have the unpaired setae a0, a'0 and a0'    1E-F, and 2A-E).
Buccal cone.Collembola are entognathous, with a buccal cone on the anterior part of the head that encloses the mouthparts.This buccal cone is formed dorsally by the clypeus, which is extended by the labrum and ven trally by the labium, split into two parts by the linea ven tralis.There are two main kinds of buccal cone: short, truncated (Figs 3A-D, 4A-B), and hemispherical when viewed from the side.This condition is found in most Poduromorpha: Hypogastruridae, Onychiuridae, Brachys tomellidae, Frieseinae, and some Pseudachorutinae (i.e., Anurida); elongated, beak shaped (Figs 4C-D).The latter condition is found in Neanurinae (with simple mouthparts), some Pseudachorutinae (i.e., Pseudanurida Schött, 1901, Pseudachorutes Tullberg, 1871), and some Odontellidae.
Some species possess an intermediate state as that found in the genus Micranurida Börner.Elongation of the buccal cone could be correlated to the elongation of the maxillae, which could be linked to the feeding habits (Massoud, 1967).
Clypeal chaetotaxy is reduced to a few setae in most species.Podura and Paraxenylla have more setae and non-poduromorphan Collembola have many more.Prelabral chaetotaxy of Poduromorpha is formed by a row of four or two setae, or prelabral seate are absent (Najt & Weiner, 2002).Generally, it is composed of four setae in Entomobryomorpha (Yosii, 1976), thought three in some Isotomidae and six in Symphypleona (Betsch, 1980).
The nomenclature of Pomorski (1996) for the setae pos terior to labium is retained.Specifically, the number of setae in row i are of phylogenetic value showing a varia tion in number from one row to five rows: Four to five rows in Onychiurinae, Isotogastrura Thibaud &Najt, 1992 andFolsomia Willem, 1902; three in Tullbergiinae; three in Hypogastruridae; two in Neanurinae s.l.; one in the genus Microgastrura.
Mandibles.Springtails typically possess chewing-type mandibles.They are composed of a robust proximal com ponent, which is pierced by a foramen allowing the inser tion of mandibular muscles, and a distal part with a molar plate at the base and a toothed pars incisiva at the tip.Fol lowing the work of Stach (1949) and Gisin (1952), it was admitted that mandibles were lacking in Odontellidae (e.g., Massoud, 1967).However, they were observed in the genus Stachia Folsom, 1932. Deharveng (1981b) in his study on Odontellidae clearly established that mandi bles are always present in this family, even if they are reduced and do not have a molar plate (sometimes one asymmetrical mandibles only is present).All the prepared specimens of Odontellidae observed for this study had at least one mandible.In the same way, it was considered that Brachystomellidae did not have mandibles.But man dibles can be found in some genera (in Probrachystomellides Weiner &Najt, 1991 andWinterella Massoud, 1967), they move under the stipes when prepared and are then difficult to see (Judith Najt, pers.com., Weiner & Najt, 1991a).However, with the data to hand, it is diffi cult to assess if mandibles are absent in most Brachystomellidae or present and hidden by the stipes.The same goes for some Pseudachorutinae that seem to lack the mandibles as well (Najt & Weiner, 2002).
The molar plate can be present (Entomobryomorpha, Symphypleona, Onychiuridae, Hypogastruridae) or absent (Odontellidae, Neanuridae).It is worth noting that the Hypogastruridae genus Microgastrura has an interme diate state, with a reduced molar plate.The pars incisiva bears either a few sub-equal teeth or is formed of a big basal tooth and smaller teeth at the tip.It is reduced to a sclerified spike in Odontellidae.
Maxillae.In the genus Hypogastrura, the maxillary palps (maxillary outer lobes) are simple and found on each side of the mouth with a sublobal plate including none to three sublobal setae.The number of sublobal setae is highly variable and so useful for the lower-level taxonomy of species identification (Fjellberg, 1984a).For example, in the genus Xenylla, 0, 1, 2 or 3 sublobal setae can be found depending on the species (Fjellberg, 1984b).Maxillary palps are found in Onychiuridae, Hypogastruridae and Poduridae, reduced in Odontellidae, highly reduced in Brachystomellidae and absent in Neanuridae.
In almost every Collembola, cardo, stipes and fulcrum are articulated in the following order: Stipes cardo ful crum.In Odontellidae the stipes is directly articulated to the fulcrum (Deharveng, 1981b;Gisin, 1952).The cardo and the fulcrum can be either stocky (length/width ratio low) or thin (length/width ratio high).It is important to note that the length/width ratio of the cardo is small in Onychiuridae and Hypogastruridae, higher in Brachystomellidae and Frieseinae and still higher in Neanurinae.Actually, there is a cline from a stocky cardo (Hypogastruridae) to a thin cardo (Frieseinae) and even thinner one (Neanurinae).Cardo can be straight (Frieseinae, Brachystomellidae, and Neanurinae) or twisted (Hypogastruridae).The postero-internal extension of the fulcrum (p.p.i.) can be well developed as in Hypogastruridae and Onychiuridae, or reduced as in Neanuridae and Brachystomellidae (Deharveng, 1981b).

Antennae
The antennae are subdivided into four simple segments, each with its own intrinsic muscles.In some Symphypleona, the third and fourth antennal segments can be sec ondarily subsegmented (Betsch, 1980;Nayrolles, 1991), thus giving more flexibility to the antennae.Secondary segmentations are also observed in some Entomobryomorpha, namely Orchesellinae (Mari Mutt, 1988) and Tomoceridae.Antennae are inserted in the anterior part of the frontal area, and are in contact with the clypeal area.However, in Symphypleona they are inserted in the middle of the frontal area and make no contact with the clypeal area (Stebaeva, 1988).The region of the head near the antennae usually presents a distinct granulation (primary granules smaller) in Onychiuridae.Antennae are slightly shorter or sub-equal to the cephalic diagonal in Poduromorpha, significantly longer in Entomobryomorpha and Symphypleona, or significantly shorter than the cephalic diagonal in Neelipleona.It is worth noting that these ratios also reflect the proportions of the other appendices: legs and to a lesser extant, the furca.This is not true for Neelipleona, which have very short antennae despite the presence of well-developed legs and furcula.The proportions and shape of the four antennal segments are also potentially informative: the four segments can have the same width (resulting in subcylindrical-shaped antennae); segments can be decreasingly wide from base to tip, resulting in a more or less pronounced conical shape (a trend seen in Neanuridae and in Odontellidae).Antennae possess many sense organs, particularly at the tip: sense organ of the third antennal segment and above all the fourth antennal segment as a whole with many sensilla.
Generally, the first antennal segment contains seven setae: six arranged on one row and one more basally, near the external side.This conformation is observed almost uniformly in Poduromorpha (and Isotogastruridae) with few variations and with the exception of Onychiurinae, which possess one to three more setae.With ten setae, the first antennal segment chaetotaxy in the genus Triacanthella is very close to Onychiurinae.Primary chaetotaxy of Symphypleona is composed of seven setae as in Poduromorpha.There is only one seta on the first antennal seg ment in Neelipleona whereas Entomobryomorpha have a chaetotaxy with more than ten setae.The same kind of setae distribution is found on the second and third antennal segments.The second antennal segment bears from ten to 13 setae in most of Poduromorpha (including Triacanthella), with the exception of Onychiurinae, which have 15 to 17 setae.The third antennal segment bears from 21 to 22 setae in Onychiurinae and Triacanthella, and from 17 to 19 setae in the other Poduromorpha.
The sense organ of the third antennal segment is found in most Collembola.It is formed by five sensilla: two internal sensilla, two guard sensilla (one dorsal, one ven tral), and one small ventral sensillum.These sensilla are accompanied by guard setae.Guard sensilla bend in the same direction in Hypogastruridae, and one against another in Tullbergiinae.It has to be said that tribes of Tullbergiinae have been defined according to the number of guard sensilla (Luciáñez & Simón, 1993).However, according to their phylogenetic analyses, the three tribes were found to be paraphyletic.In Onychiurinae, Tetrodontophorinae and Isotogastruridae, the integumental papillae protect the sensilla of the third antennae sensory organ.In Tullbergiinae, the same structure is found but is not as developed as in Onychiurinae and Tetrodontophorinae.
A typical fourth antennal segment of Poduromorpha bears many sensilla, a subapical vesicle, a subapical organite and a dorso-external microsensillum.These sen sillary structures can be developed or sometimes absent.The microsensillum can be located in the distal third por tion of the fourth antennal segment, in the basal third por tion or it can be absent.Another way to describe the position of the microsensillum is to consider its proximity to the subapical organite.Different nomenclatural sys tems for the sensilla of the fourth antennal segment have been proposed for different groups: Neanurinae (Deharveng, 1981a), Tullbergiinae (Rusek, 1976), the genus Ceratophysella (Yosii, 1962) and the genus Anurida (Babenko, 1997;Fjellberg, 1985a -both confused sensilla f with c, g with d, c with e and e with f), the genus Willemia (Arbea & Jordana, 1986;Fjellberg, 1994;Pota pov, 1994) and for Xenylla jocquei André, 1988 (André, 1988).To date, no comprehensive system of nomencla ture, which would indicate homologies, has been pro posed.Table 5 summarizes the possible homologies for sensilla of the fourth antennal segment between the dif ferent groups and a proposition for a unified system.Seta x (Yosii, 1962) was named i by Deharveng (1981a), m by Fjellberg (1985a) and ad3 by André (1988).Microsensillum f (Rusek, 1976) was named Se by Deharveng (1981a), s by Fjellberg (1985a), and wa4 by André (1988).
Morulinae possess a chaetotaxy of sensilla which are not or only slightly distinct from ordinary setae (Deharveng, 1981a).
Important variations are observed in Odontellidae with plurichaetoses, mostly on the dorso-external side, or reductions.Xenyllodes Axelson, 1903 (Fig. 7A) and Superodontella Stach, 1949 possess sensilla S7, S8, and S9, the latter being relatively basal.Four other sensilla are found: S1, S2, S3, and S4 with slightly different positions in comparison to the other groups: S1 is slightly higher and more external, S3 has a more internal position and S4 is located a more interno-basally on the segment [a2-S3] and S2 positioned higher up under the subapical organit.
Compared to other families, it appears that sensilla S1, S2, and S4 rotated around the ordinary seta a2 clock-wise in a left-antennae, and S3 was pushed internally by S4 coming.Migration of these sensilla is perhaps correlated with the shortening of the basis of the fourth antennal segment (which causes the conical shape of the antennae).Another hypothesis would be that S3 (S4 in the previous hypothesis) does not really move and S4 (S3 in the pre vious hypothesis) migrates further, on the internal side.Both hypotheses require a rotation around ordinary seta a2.
Sensilla chaetotaxy of the fourth antennal segment of Tullbergiinae (Fig. 8B) is difficult to compare with sensilla chaetotaxy of Hypogastruridae, Odontellidae, Brachystomellidae, and Neanuridae.Five well-differentiated sensilla are found in Tullbergiinae: two dorso-externals and three dorso-internals.These could be homologous with sensilla S8 and S9 on one side and sensilla S1, S2, and S4 on the other side.It is worth noting that the micro- sensillum is located under the subapical organite, which has migrated to the external side of the antennae (the microsensillum is located between S7 and S8 in Hypogastruridae, Odontellidae, Brachystomellidae, and Neanuridae).It was not possible to identify seta i, as it is either absent or indistinct from the other ordinary setae.In Onychiurinae and Tetrodontophorinae the sensilla are indis tinguishable when viewed under a compound microscope.In some species, the sensilla are visible (as in Psyllapho-rura Weiner & Najt, 2000), however it was not possible to establish homologies.
Podura aquatica possesses a fourth antennal segment chaetotaxy very similar to Hypogastruridae, but the sen silla are not distinct from the ordinary setae.
Frequently, secondary losses or plurichaetoses of sen silla are observed independently in different groups.

Thorax
The prothorax is well developed in Poduromorpha and bears setae.It is well differentiated except in Caputanurina where it merges with the head (this is a diagnostic character of the genus); this merger is less complete in the other Caputanurinae.In Isotogastruridae, the prothorax is well developed but does not bear setae.The number of rows of setae and number of setae within these rows vary according to the different groups: 1 row of setae in Tullbergiinae, Hypogastruridae, Odontellidae, Brachystomel lidae, and Neanuridae, 2 rows in Onychiurinae, more than 2 rows in Tetrodontophora bielanensis.
An organ present on the pro thorax of Collembola's embryos may help hatching (Jura, 1967;Philiptschenko, 1912;Tamarelle, 1971Tamarelle, , 1981;;Thiegs, 1942;Wheeler, 1893).The organ has been found in the embryos of all Collembola studied thus far (Jura & Krzysztofowicz, 1978).This organ can persist in the first instar.According to Deharveng (1983), this dorsal organite of the prothorax would be characteristic of the first instar of Neanurinae, but was observed in Caputanurinae (Caputanurina tur bator Najt & Weiner, 1992) and Brachystomellidae (Najt & Weiner, 1992).According to my own observations, this structure is present in the first instar of Onychiurus ambulans, Mesaphorura macrocheta, Xenylla grisea, Neanura muscorum, and Folsomia candida (Entomobryomorpha, Isotomidae).Actually, this organite is widely present in Entomobryomorpha (Judith Najt, pers. com.).It is probably present in Collembola as a whole.The pro thorax is highly reduced and does not bear setae in Entomobryomorpha, Symphypleona, and Neelipleona.In the first instar of Folsomia candida, the second thoracic seg ment bears one more row of setae than the third thoracic segment.More generally, the mesothorax bears additional rows of setae on the anterior part compared to the meta thorax in the genera Tetracanthella, Isotomiella Bagnall, 1939, and probably in every Entomobryomorpha.The mesothorax and metathorax have an almost identical chaetotaxy in Poduromorpha.Additional rows of setae on the mesothorax may have originated ontogenetically from the prothorax.Some authors believe that the prothorax tergum is fused with the mesothorax (e.g., Denis, 1949).
The number of rows of setae on the meso-, and meta thorax is variable among the different families.In Poduromorpha, two to five rows are observed, and more than five in Entomobryomorpha.Microsensilla on the meta thorax are present in Onychiurinae (with few exceptions like Micraphorura pieninensis Weiner, 1988), Tetrodon tophorinae, Tullbergiinae, and Odontellidae.Microsensilla are absent in Hypogastruridae, Brachystomellidae, and Neanuridae.Thoracic sternites can bear setae.How ever, that character is probably not very informative for Poduromorpha as a whole since those setae can be present or absent in the same genus.
Onychiurinae and Tetrodontophorinae have a complete tibiotarsal chaetotaxy, with the presence of the rows A, B, C, and T as well as seta Y, which is absent in the other groups (Figs 10 & 11).This chaetotaxic scheme can be concealed by a secondary plurichaetosis (Tetrodontophora).Comparatively, other Poduromorpha have a less abundant chaetotaxy with row C and seta Y lacking.Hypogastruridae, Neanuridae, and Poduridae (Podura aquatica) are typical of this model, with generally 19 setae on tibiotarsi I and II, and 18 setae on tibiotarsi III split up into rows T, A, and B. Seta B7 is lacking on tibi otarsi III, except in Podura aquatica which possesses 19 setae on every tibiotarsi.Secondary losses are frequently observed in different groups, for example setae T2, and T3 are lacking in Willemia denisi, species of that genus have 17, 17, 16 setae at least (see Fig. 11).Rows A, and B can be relatively complete in Tullbergiinae (Fig. 10) and row T is always absent.Unguiculus (empodial appendage) is generally present in Onychiuridae, Odontellidae, and Hypogastruridae, but is often absent, particu larly in the genera Podura, Xenylla, and Paraxenylla (Figs 11B,11F & 11G).It is always absent in Neanuridae, and Brachystomellidae (Massoud, 1967), see Fig. 12.Generally, in Neanuridae and Brachystomellidae, rows T, A, and B are relatively complete and seta M is present, with the exception of Frieseinae in which M is always absent (Fig. 12) but in Friesea species of the south hemisphere (Deharveng & Bedos, 1991;Greenslade & Deharveng, 1997).
Leg chaetotaxy is fixed (no addition of setae during the development) from the first instar onward in Tullbergiinae, Odontellidae, Hypogastruridae, Brachystomellidae, and Neanuridae.During postembryonic development, setae are added on the legs in Onychiurinae (observations based on cultures of many species: Onychiurus, Kalaphorura Absolon, 1901, Protaphorura Absolon, 1901 etc.).Setae are also added during the development in nonpoduromorphan Collembola: in Symphypleona (Nayrolles, 1988), and more generally in Entomobryomorpha as a whole (cf.Fig. 13 for Isotomidae -personal observations).
Tibiotarsi have a decreasing antero-posterior gradient of the number of setae in Neanuridae, Hypogastruridae, and Onychiuridae (except for Protaphorura): seta B7 is absent on the third pair of tibiotarsi.It is the opposite in Isotomidae and Symphypleona.The tibiotarsi and femora also present a decreasing antero-posterior gradient of the number of setae in almost all Poduromorpha.The same applies for trochanters, with the exception of Onychiurinae with a postero-anterior gradient.In every Poduromorpha, the first and second subcoxae have a decreasing postero-anterior gradient of the number of setae.Coxae have a number of setae: leg I < leg II = leg III in every Poduromorpha except in Onychiuridae and Podura aquatica in which leg I < leg II < leg III.
In Poduromorpha and Isotomidae, the number of setae per row is seven (Deharveng, 1983).Eight setae per row are observed in Symphypleona (Nayrolles, 1998), and six in Neelipleona.The number of rows is fixed according to the different groups: four (rows T, A, B, and C) or less in Poduromorpha and Isotomidae (Deharveng, 1983), and five (rows I to V) in Symphypleona in first instar (Nayrolles, 1998).
The pretarsus terminates by a claw with three lamellae (these can also bear teeth) and an empodial tubercle, which bears the unguiculus.The unguiculus is formed by four lamellae in Symphypleona and Entomobryidae (Christiansen, 1965), three in Isotomidae and Tomoceridae and only one in Hypogastruridae.
The relative proportions of the different parts of the body are variable between the different groups of Collembola: Poduromorpha are relatively stocky, whereas Entomobryomorpha and Symphypleona possess elongate or very elongate appendages relatively to the body.This could be related to the habitat: generally edaphic in the former, more epigeic in the latter two.The legs are short compared to the length of the body in Poduromorpha, whereas they are long in Entomobryomorpha, Neelipleona, and Symphypleona.The ratio of tibiotarsus to claw reflects this: the ratio is near one in Poduromorpha, much higher in the others.

Abdomen
Entomobryomorpha and Poduromorpha possess an elongate and well-segmented body.On the contrary, the Symphypleona and Neelipleona are globular.The Symphypleona have a characteristic globular shape formed by the enlargement and fusion of the posterior thoracic and anterior abdominal segments forming the large abdomen, and the last abdominal segments are distinct, forming the small abdomen (e.g., Betsch, 1980).On the other hand, the globular body of Neelipleona is one whole structure and is mainly the result of the enlargement of the thoracic segments (Massoud, 1971a).
Poduromorpha have well-developed and sub-equal abdominal segments.Entomobryomorpha present fusions of some of the segments: fusion of fourth and fifth abdominal segments or fusion of the fourth, fifth, and sixth abdominal segments.The fourth abdominal segment is normally developed in Poduromorpha, Isotomidae and Tomoceridae, and longer in other Entomobryomorpha, Symphypleona, and Neelipleona (Massoud, 1971a).
Dorsal chaetotaxy is described here with the classic system with three rows of setae: anterior (a), middle (m), and posterior (p) proposed by Yosii (1956), and devel oped for Hypogastruridae by Cassagnau (1974) and Grow & Christiansen (1974), and applied to Poduromorpha as a whole.Fjellberg (1985b) established the correspondences between the classic rows a, m, and p system and the system of Di, De, Dl, and L that Cassagnau (1980) and Deharveng (1979b;1983) developed for Neanurinae.On the ventral side, the unpaired setae chaetotaxy is of interest.They are present in Onychiuridae but seem to be absent in Hypogastruridae, Odontellidae, Brachystomellidae, and Neanuridae.The number of ventral setae rows is more important in every segment in Onychiurinae, and Tetrodontophorinae compared to the other Poduromorpha.One or two unpaired setae are found on the sixth abdominal segment in Triacanthella, every Tullbergiinae, and most of Onychiurinae.Ventrally, between the anus and the anal spines, there are a maximum of three unpaired setae (the most distal seta can be transformed to an unpaired anal spine).There are no unpaired setae on sixth abdominal segment in Neanurinae.On the other hand, there are three unpaired setae between the anal spines and anus and two unpaired setae on the tergum in Tullbergiinae.Anal spines can be present or absent.How ever, this character is highly homoplastic in Poduromorpha since both character states can be found in the same genus (e.g., in the genus Willemia, D 'Haese, 2000).
Anal valves.Within Poduromorpha, two types of anal valves can be found: 1.The three valves are sub-equal, and non-overlapping; the supra-anal valve has a convex edge (Fig. 14A).These kinds of anal valves are found in Onychiurinae, Tetrodontophorinae, Neelipleona, Symphypleona, and Entomobryomorpha; 2. The paired valves partially overlap the supra-anal valve; the supra-anal valve has an almost straight edge (Fig. 14B).These kinds of anal valves are present in Hypogastruridae, Brachystomellidae, Odontellidae, Neanuridae, and Tullbergiinae.However, in the latter family, the supra-anal valve has a slightly convex edge but is recovered by the two other valves.
Three distinct nomenclatures were proposed to name the setae of paired anal valves: one established for the genus Willemia (see Huther, 1962), one for the sub family Tullbergiinae (see Rusek, 1976), and the last one for the genus Onychiurus (Yoshii, 1996).Anal valve chaetotaxy in Poduromorpha is quite homogeneous and therefore it is easy to propose homologies between the 3 nomenclatures, and generalized it to all Poduromorpha (see Table 6).Generally 1+1 setae z are present (there can be more in some Onychiurinae) or can be absent.In the same way, 3+3 setae hr are generally found, but some taxa have less.Lawrence (1978) proposed a nomenclature for the setae of the anal valves for Symphypleona (see also Yosii, 1969).It should be possible to establish homologies for the chaetotaxy of the anal valves of the four order of Collembola (Poduromorpha, Entomobryomorpha, Symphypleona, and Neelipleona).However, these comparisons are beyond the scope of the present work.Strong plurichaetoses can conceal the described chaetotaxic scheme (Tetrodontophora, Triacanthella...), making it difficult or impossible to establish homologies.This is also applicable to the chaetotaxy of the whole body.
Ventral tube.The ventral tube, or collophore, is formed by a pair of vesicles primitively derived from a pair of primitive appendages (the collophore is probably homologous to the coxae).The ventral tube is located on the sternum of the first abdominal segment, at the poste rior end of the linea ventralis.It plays an important role in   Hüther, 1962Rusek, 1976Yoshii, 1996 A 12' maintenance of water and ion balance.Urine is dis charged from the labial nephridia (ventral side of the head) into the linea ventralis and carried to the ventral tube by capillary action (Eisenbeis, 1974;Rusek, 1987;Verhoef et al., 1979;Verhoef et al., 1983).Fluid and salt are then exchanged at the apical part of the organ (water is reabsorbed).The coronal section of the collophore is oval in Poduromorpha and circular in Symphypleona and Entomobryomorpha (Massoud, 1967).In other words, the ventral tube is a short oval-shaped structure in Poduromorpha, whereas it is a rather elongate cylinder in Symphypleona, Neelipleona, and Entomobryomorpha.
Tullbergiinae have the same ventral tube chaetotaxy as Hypogastruridae.Frequently in Onychiurinae plurichaetoses are observed: row td comprising more than 4+4 setae, and row tp more than 2+2 setae.Distal setae td are homologous with "subapical setae" described by Pomorski (1996) in Onychiurinae, and proximal setae tp are homologous with "additional setae".
Tenaculum and furcula.The tenaculum is located on the third abdominal sternum.It is formed of a corpus and two rami bearing a variable number of teeth on the outer side.The tenaculum holds the furcula against the abdo men.In Poduromorpha, the corpus is rudimentary, and does not bear setae unlike Symphypleona and Entomobryomorpha, which possess a well-developed corpus with setae.
Reductions of furcula and retinaculum are largely cor related.
Furcula is the jumping organ characteristic of Collembola.It can either be well developed as in the more epigeic species, reduced or completely absent (trend in the more edaphic species).From an evolutionary point of view, the furcula is probably the result of the partial merging of a pair of primitive appendages located on the sternum of the fourth abdominal segment.The fusion of the proximal region formed the manubrium; the distal portions remain free, forming the dentes, with each one ending with a mucro.The degree of development of the furcula, the number of setae on the manubrium and on the dentes, as well as the shape of the mucro are variable characters among the species and are widely used in tax onomy.

REMARKS ON FIRST INSTAR
Postembryonic development in Arthropleona always involves the addition of setae, never their loss (Deharveng, 1983;Thibaud, 1970).Therefore, the first instars are fundamental to understand structures that can be diffi cult to interpret in adults (Collazo, 2000), like, for exam ple, in case of plurichaetosis (Szeptycki, 1972).
The first instar in the genus Onychiurus has relatively few setae, and its simple chaetotaxy appears to be very similar to the chaetotaxy of adult Hypogastruridae.A large number of setae appear when molting during postembryonic development.Pomorski (1996) described first instars of many species of Onychiurinae.Observa tions made on other species of the same sub-family for this study are in complete agreement with Pomorski's results.
Contrary to Onychiurinae, the first instar of Neanura muscorum has relatively complete chaetotaxy, which is almost identical with an adult one (labium, head, body, and all segments of the leg).However, some setae are not present in first instar, e.g.i' on the anal valves.The only present sensilla on the antennae are S5 and S6.Further more, in addition to present microsensillum, sensilla are always absent in first instar in the other Poduromorpha.Similar conclusions on chaetotaxy of the first instar were reached for Neanura tatricola Massoud, 1967(Gruia, 1974) and Thaumanura ruffoi Dallai, 1969(Dallai & Martinozzi, 1980).This may be applicable to the entire sub-family Neanurinae.The first instar of Caputanurina turbator (the Caputanurinae are phylogenetically close to Neanurinae, D'Haese, 2003) was described (Najt & Wei ner, 1992).This group shows relatively complete chaetotaxy, but not as complete as in Neanurinae.For example, sensilla are absent on the fourth antennal segment in the first instar.
If one confronts the observations on the first instars to the phylogeny of Poduromorpha (D'Haese, 2002(D'Haese, , 2003)), some interesting conclusions can be drawn.From Onychiurinae to Neanurinae, the chaetotaxy is increasingly complete (compared to the adult) in the first instar.Tullbergiinae, Hypogastruridae and non-neanurinae Neanuridae are intermediate: the first instar chaetotaxy is more complete than that in Onychiurinae but less complete than in Neanurinae.However, the adults have chaetotaxic schemes increasingly paurochaetotic beginning with the Onychiurinae through Tullbergiinae and then Hypogastruridae to Neanurinae.There appears to be a trend in Poduromorpha, which drives the adult descendants to maintain a chaetotaxy similar to the juvenile instars of its ancestor without change of the growing rate and size.In other words, phenomenon of paedomorphosis by neoteny could have been among the driving forces at the origin of diversification within Poduromorpha.
Fig. 5. Two-dimensional hypothetical scheme of a left fourth antennal segment of a Poduromorpha.

Table 1 .
Correspondences between existing systems of nomenclature for cephalic setae.

Table 2 .
Correspondences between existing systems of nomenclature for setae on papillae and guard setae on labial palps.

Table 4 .
Correspondences between existing systems of nomenclature for mentum and submentum setae.

Table 3 .
Correspondences between existing systems of nomenclature for proximal setae on labial palps.

Table 6 .
Correspondence between existing systems of nomenclature for setae of the anal valves.