STRUCTURE AND DEVELOPMENT OF THE INSECT ...

/111. 1./nsecr Morpho!. & Embryo!., VoL 17, No.3, pp. Printed in Great Britain

243-294, 1988

0020-7322/88 $3.00 + .00 Pergamon Press pic

STRUCTURE AND DEVELOPMENT OF THE INSECT ANTENNODEUTOCEREBRAL SYSTEM

JEAN PIERRE RosPARS Laboratoire de B iometrie, Institut National de la Recherche Agronomique, Centre de Recherches de Versailles, route de Saint Cyr, F78000 Versailles, France

(Accepted2 December 1987)

Abstract- The structure and postembryonic development of antennae and deutocere­

brum in various insect orders are reviewed. First, the number and/or size of system components, i.e. antenna! sensilla, neuroreceptors, deutocerebral neurons and synaptic complexes (glomeruli), are compared in adult insects. Second, the neuronal organization of the system is examined. Evidence of projection of the neuroreceptors from the 2 basal antenna! segments (scape and pedicel) in the antenna! mechanosensory and motor center and from the distal segments (flagellum) in the antenna! lobe is discussed. Third, the types of cerebral neurons found in the antennal lobe are described. Fourth, all synaptic contacts between neurons in the antenna! lobe take place in discrete glomeruli whose ultrastructure and neurotransmitters are examined. Evidence of individual identifiability of glomeruli is given. Fifth, various sexual dimorphisms present in the antenna and antennal lobe, related or not to sex pheromone perception, are described at sensillar, neuronal, and glomerular levels. Sixth, the postembryonic development of neurons and glomeruli is analyzed in halo- and hemimetabolous insects. In conclusion, the columnar organization of the system is emphasized, alternative models of antennal-neuron projection into glomeruli are considered, and the functional significance of identified glomeruli is discussed. Index descriptors (in addition to those in title): Antenna, deutocerebrum, antenna! lobe, glomeruli, olfaction, sexual dimorphism, postembryonic development.

INTRODUCTION

SENSORY information gathered by antennae is processed in the deutocerebrum, one of

the 3 main subdivisions of the insect brain (Fig. 1). Antennae represent the main tactile, gustatory, and olfactory organ of insects. Sensory neurons of both antennae project to 2 distinct deutocerebral areas on each side of the brain : an anterior one, called "antenna! lobe" (AL), and a posterior or dorsal one called here "antenna! mechanosensory and motor center" (AMMC, this name is justified later). Antenna! neurons that enter the ALs terminate within them, except for rare exceptions. They synapse with AL neurons within characteristic spheroidal structures, the "glomeruli". A part of these cerebral neurons give rise to tracts that connect the glomeruli to the protocerebrum. This set of Abbreviations used in text: ACT = antenna-cerebral tract; AL = antenna! lobe; AMMC = antenna! mechanosensory and motor center; CNS = central nervous system; EAG = electoantennogram; GABA = gamma aminobutyric acid; MGC= macroglomerular complex; LP =lateral protocerebrum; LPO= labial-pit organ; SEG = subesophageal ganglion; TOG= tractus olfactorio-globularis. 243

-

244

JEAN PIERRE ROSPARS antenna

antenna! nerve brain

antenna! lobe "----- compound eye '------ esophageal . canal '----- subesophageal ganglion

1-----

maxillary palp

""'--- 1

A sensill

antenna! nerv e

antenna! se•gm9 Strausfeld, 1976 Strausfeld, 1976 Mciver, 1971

2F 64 c 338 c

NR/ALd Authorse

4.7 5.6

14 Esslen and Kaissling, 1976 60 Witthoft, 1967

a. Male individuals unless otherwise stated. b. Number of neuroreceptors per antenna (in 1,000s), for flagellum (F), or olfactory + gustatory neuroreceptors (C). c. Number of neurons per hemibrain (in 1,000s), in hemideutocerebrum (hDTC AL + AMMC), or in antenna! lobe (AL). d. Ratio of number of neurorecePtors to number of AL neurons. e. Authors for each species: NR on first line, hDTC orAL on second line. =

Neuron number. The number of AL neurons has been determined only in a very few species (Table 1). Comparing these results which vary in a 10-fold range, is not easy because they apply either to the whole deutocerebrum (AMMC included: Apis Acromyrmex, Musca), or to the AL (Periplaneta) or to the neuron groups of the lobe (Blaberus, Manduca). These data must be cautiously interpreted as the rarely described counting methods are likely to involve numerous errors (see discussion in Chambille and Rospars, 1985). The number of neurons, which leave the lobe via the TOG in Periplaneta, is estimated at 250 (adults, Ernst et al., 1977) or 320-350 (first instar nymphs, Prillinger, 1981) i.e. 25-35% of the neuron population (Table 1). It is about 700-800 in Locusta migratoria (Ernst et al., 1977) and 200 in Drosophila melanogaster (Borst and Fischbach, unpublished observations). In Manduca (Homberg et al., 1986), 760 fibers leave the lobe in 3 main tracts and more than half of them (400) are in the TOG, i.e. 35% of the AL neurons. 1.2.2. Population of glomeruli. Size of antenna/ lobe and distribution of glomeruli. ALs greatly vary in size (Table 2) but their often succinct descriptions in the literature give evidence of their uniform appearance in species of various orders, so much so that Ehnbom (1948) regards lepidopteran lobes as being of very poor interest from a comparative point of view. Most authors have noted that glomeruli are closely packed at the lobe periphery and are less dense or missing in the central part of the lobe. In some species glomeruli are regularly distributed and no glomerulus subsets can be distinguished (e.g. Blaberus, Rospars and Chambille, 1981; Mamestra, Rospars, 1983).

248

JEAN PIERRE RoSPARS TABLE 2. DIAMETER AND VOLUME OF ANTENNAL LOBE

Order

Species

Dictyoptera

Dia.a

Vol.b

Authors

Peripla neta americana {male) Peripla neta americana {female) Blatta orie ntalis Blattella germanica Blaberu s craniifer

471 417 388 222 461

55 38 31 5.7 52

Neder, 1959 Neder, 1959 Neder, 1959 Neder, 1959 Rospars and Chambille, 1986

Lepidoptera

Manduca sexta Mamestra brassicae Pieris brassicae

40QC 253 166

34 8.6 2.4

Sanes et al., 1977 Rospars, 1983 Rospars, 1983

Diptera

Calliphora erythrocephala Musca dome stica Dro soph ila melanogaster Aedes aegypti

330 134 58 140

19 1.3 0.1 2.9

Boeckh et al., 1970 Strausfeld, 1976 Power, 1946 Childress and Mciver, 1984

Hymenoptera

Apis mellifera {worker) Apis mellifera (worker) Apis mellifera (male) Acromyrmex octospinosus

350 265 285 200

22 9.9 9.9 4.1

Pareto , 1972 Budharugsa, 1984, Arnold et al., 1985 Delabie, 1984

a. Diameter in f.1m. b. Volume in 103 1-1m3, vol. (-rr/6) x Dia3. c. Approximate value measured on figures published. =

In other species, such subsets have been described. In Periplaneta, there are 2 subsets of equal sizes (Prigent, 1966). In Hymenoptera, glomeruli subsets are common: 2 in Bombus hypnorum (Fonta, 1984), 3 in Acromyrmex (Delabie, 1984), at most 3 in Camponotus, at least 4 in Mesoponera caffraria (Masson, 1972), and 4 in Apis mellifera, 2 of which are larger and innervated by their own tracts of antenna! fibers (Suzuki, 1975; Mobbs, 1982). Number of glomeruli. There is no glomerulus in the ALs of adult Ephemeroptera, Odonata and Plecoptera (Pauov, 1961). In other orders, this lack of glomeruli is exceptional. Glomerular counts have been carried out in some species (Table 3). Four categories may be distinguished: species with no glomerulus, those with 10-20 glomeruli, those with 50-200 glomeruli (the most numerous), and those with roughly 1,000 glomeruli. The 2 latter types do not match any taxonomic subdivision, e.g. the 4th one includes an orthopterous species and a hymenopterous one, which are phylogenetically quite distinct. Moreover, one may wonder whether the term "glomerulus" describes the same reality in these 2 categories. Glomerular size. Glomeruli vary greatly in size according to species (Table 3): the diameter of the biggest known glomeruli (Manduca) is 7 times larger than that of the smallest ones (Camponotus), i.e. a ratio of 300 by volume. The frequency histogram of glomerulus diameter is shown for 3 species in Fig. 2. It displays a nearly symmetrical, approximately Gaussian distribution. The diameter range, which is within 3 standard deviations from the mean in the 3 species (Chambille et ale ;1980; Rospars, 1983), 1s remarkably similar.

Structure of Antennodeutocerebral System

249

TABLE 3. NUMBER AND SIZE OF GLOMERULI Order

Species

Number

Odonata

My stacideilongicornis

Orthoptera

Locusfa migratoria

Dictyoptera

Periplaneta americana Periplaneta ameri cana Blaberu s craniifer

Lepidoptera

Manduca sexta Manduca sexta Mame stra brassicae Pieris brassicae Bombyx mori Bombyx mori

Hymenoptera

Calliphora erythrocephala Dro sophila melanogaster Aedes aegypti Formica praten sis Camponotu s vagu s Mesoponera cafraria Acromyrmex octo sponosus Vespa crabro A pi s mellifera (worker) A pis mellifera (worker) A pis mellifera (male)

Ehnbom, 1948

0 ca 1000 ca 125 107 57-61 67± I 62± 3 50-60 55-60

25

19 9

ca 200 ca 200

Ernst et al., 1977

30-100 Boeckh eta!., 1970 50-90 Prillinger, 1981 45-100 Chambille et al., 1980 50-100 Hildebrand et al., 1980 Schneiderman et al., 1983 24--90 Rospars, 1983 12--48 Rospars, 1983 Kanzaki and Shibuya, 1983 ca 35 30-50 Koontz and Schneider, 1987 30-60

>

Authors Panov, 1961

0

Trichoptera

Diptera

Dia.a

20 10-12 4--18

ca 120 900-1000 10-80 165± 2 27-80 103± I 23-82

Boeckh et al., 1970 Stocker et al., 1983 Childress and Mciver, 1984 Goll, 1967 Masson , 1973 Masson, 1973 Delabie, 1984 Hanstrtim , 1928 Pareto, 1972 Arnold eta!., 1984 Budharugsa , 1984

a. Diameter in 1-1m (average or minimum-maximum).

NB. GLOMERULI (%)

-d'

---- �

RADIUS

FIG. 2. Histograms of glomerular radii. Mean radii (R) are 68 1-1m (B.c.), 47 1-1m (M.b.) and 30 r-tm (P .b.) respectively. cr, standard deviations (from Chambille eta!., 1981 and Rospars, 1983) .

250

JEAN PIERRE ROSPARS

1.2.3. Significance of the antenna! lobe quantitative features. It seems premature to draw general conclusions from the few simple properties of neuron and glomerulus populations summarized above. For example, there is no clear relation between the numbers of antenna! and AL neurons: their ratios vary in the range 200-370 in Dictyoptera and Lepidoptera, but in the range 10-60 in Diptera and Hymenoptera (Table 1). If one excepts male Apis, there is a rough correlation, however, between AL volume and number of incoming antenna! neurons (Tables 1 and 2). This is in line with qualitative observations. The reduced ALs and even the lack of glomerulus in Ephemeroptera parallel a reduction of antennae, which bear few sensilla (Grasse, 1975). Leptoceridae, an atypical family of Trichoptera, is an apparent exception to this rule with their long antennae and undifferentiated ALs (Ehnbom, 1948). In fact, the antenna! nerve branch that enters the AL is thinner than in other Trichoptera and behavioral observations suggest that antennae are used as a balancing pole in the slow or stationary flight, which is typical of this family. The development of this mechanical or mechanosensory role of the antenna at the expense of its chemosensory functions is also reflected in the large size of the AMMC (see Sections 1.3 and 2.4). Another approach to the same problem consists of comparing the AL diameters in different species after correction of insect size variations. Head width or protocerebral central body diameter may be considered as convenient measures (there is a very good correlation between these 2 measures, Howse, 1974; Rospars, 1983). It is then observed that the lobe of a diurnal species, such as the butterfly, Pieris, is only 2.5 times larger than its central body, whereas it is about 8 times larger in nocturnal or crepuscular species, such as the moth, Mamestra and the cockroach, Blaberus. This is accompanied by a reverse variation of the optic lobes (Rospars, 1983). Finally, few quantitative data on ALs and their neuron and glomerulus populations are available. More thorough investigations on a wider sample are needed for a better understanding of the parameters governing number and size of components.

1.3. Antenna/ mechanosensory and motor center (AMMC) This is a cerebral compartment with ill-defined borders, initially recognized and named "dorsal lobes" by Viallanes (1887). Because this center is not always in a dorsal position with respect to antenna! lobe, it is also called the "posterior lobe" (Sanchez, 1937; Power, 1946) or "posterior antenna! center" (e.g. Strausfeld and Singh, 1980). Other names are based on its admitted function: "antenna-motor" (Sanchez, 1937; Pflugfelder, 1937), "mechano-sensory" (Strausfeld, 1976) (see Section 2). Still fewer quantitative data are available than for ALs. The neuropilar volumes in percent of total brain volume of the AMMC and AL in females Musca domestica and Aedes aegypti are reversed, that is 2.5 (AMMC) and 1.1 (AL) for the fly (Strausfeld, 1976) and 1.1 (AMMC) and 2.6 (AL) for the mosquito (Childress and Mciver, 1984). The fiber tracts that enter these centers do not match these ratios with 7,000 (AMMC) and 14,500 (AL) axons in the fly (Strausfeld, 1976), 4,000-5,000 (AMMC, Mciver, 1982) and more than 2,050 (AL, Mciver, 1978) in the mosquito. These data suggest that, in this case, there is no simple quantitative relationship between number of fibers and volume of neuropil. Other factors, such as number of cerebral neurons and volume of neurites, must be taken into account. The number of AMMC neurons per hemibrain is 2,650 in

Structure of Antennodeutocerebral System

251

both the worker bee and the drone (Witth6ft, 1967). Other data relative to motoneuron number in this area are discussed in Section 2.4. The AMMC neuropil does not show glomeruli but "microglomeruli," 5-10 11-m in diameter (Pareto, 1972; Masson and Strambi, 1977), which stain more deeply than the surrounding neuropil. They look like those of the calyces of the corpora pedunculata (cf. Section 3. 2.2) and are likely to correspond to synaptic terminals. The insufficient neuroanatomical characterization of the AMMC, in contrast with the clear limits and conspicuous internal structure of ALs, probably explains the relative lack of attention paid to this brain area so far.

2. NEURORECEPTORS AND MOTOR NEURONS The antennodeutocerebral system appears divided into 2 subsystems: the flagellar subsystem and the scapopedicellar subsystem. The flagellar subsystem is composed of the antenna! flagellum and of the "antenna!" lobe (flagellar lobe would be more accurate), and is exclusively sensory, because the flagellum is devoid of muscles (Section 2.1). The scapopedicellar subsystem is composed of the scape, pedicel, and AMMC. It is mixed, sensory (Section 2.2) and motor (Section 2.4). Evidence for this subdivision is examined by studying the origin and destination of the neurons that connect antenna and deutocerebrum. Confirmatory evidence is derived from neuroreceptors, which come from other head parts (Section 2.3) (see also review by Mobbs, 1985).

2.1. Neuroreceptors from the antenna/flagellum 2.1.1. The general model and exceptions to it. The antenna! nerve is composed of sensory axons that project centripetally into the AL. It also contains some motor axons that project centrifugally to innervate the muscles at the base of the antenna. The antenna! nerve enters the brain at the leva! of the AL, where it divides into 2 tracts (for a comparative study, see Youssef, 1971). The shorter of these tracts innervates the AL, whereas the longer runs laterally to the AMMC. All flagellar fibers probably follow the short tract and most of them terminate in the glomeruli of the ipsilateral AL. This has been demonstrated by silver impregnation, degeneration experiments, cobalt migration, and electron microscopy. For example, Boeckh et al. (1970) have shown in Periplaneta that after amputation of the flagellum (the scape and pedicel remain undisturbed), the degenerating axons project solely to the ipsilateral AL. This unilateral projection has also been obserbed in Hymenoptera (Pareto, 1972; Mobbs, 1982) and Lepidoptera (Matsumoto and Hildebrand, 1981). These experiments are summarized in Section 2.1.3. The model described above does, however, show exceptions (Fig. 3): (i) Passive migration of cobalt chloride in the sectioned antenna! nerve of the moth, Manduca sexta, has stained some isolated axons apparently coming from flagellar neuroreceptors (possibly mechanoreceptors) that project into the ipsilateral AMMC (Hildebrand et al., 1980). Also in Bombyx, Lymantria and Antheraea (Koontz and Schneider, 1987) flagellar fibers were found that project mostly to the ipsilateral AMMC and also to the contralateral AMMC. Some of these fibers probably originate from sensilla coeloconica sensitive to C02 and warm humid air (Schneider, 1969). (ii) Pearson (1971) has noted in Golgi preparations of Sphinx ligustri and other

-

252

JEAN PIERRE ROSPARS

flagellar neuroreceptors

FLAGELLUM

antenna! nerve

PEDICEL

glomerulus

SCAPE

dorsal-lobe

mechanoreceptor

tractus

ANTENNAL LOBE

DORSAL LOBE

FIG. 3. Diagrammatic sagittal view of antennodeutocerebral system to show projections of flagellar

neuroreceptors. Most projections are in antennal-lobe glomeruli. Projections in AMMC and protocerebrum {long fibers) are rare.

Lepidoptera that some antenna! fibers entering the AL pass directly into the TOG. According to Strausfeld (1976), these "long olfactory fibers" terminate in the calyces of the corpora pedunculata. However, these projections have not been confirmed by other authors (e. g. Mobbs, 1984 in honeybee). (iii) In Diptera, a part of the fibers terminating in AL show bilateral projections. This has been shown in Calliphora vicina (Boeckh eta!., 1970) and Drosophila (Stocker et a!., 1976, 1983; Borst and Fischbach, 1987 and unpublished observations) by degeneration studies after unilateral ablation of the flagellum, and by Golgi impregnations and cobalt backfilling from the cut end of the antenna at the pedicel/flagellum level. Neuroreceptors with bilateral projection make up about '14 of the antenna! fibers (Borst and Fischbach, 1987). They branch within, or close to, the glomerulus where the ipsilateral branch terminates, while the contralateral branch runs along the supraoesophageal AL commissure (Stocker et al., 1983; Borst and Fischbach, 1987). As projections within glomeruli are maintained, this is a mere variation of the general model. 2.1.2. Tracts associated with antenna! lobes. In all the species studied, the neuroreceptor fibers coming from different parts of the antenna! nerve gather into a series of tracts when entering the AL. This results in typical figures of fiber tract crossings. In Dictyoptera (Ernst et a!., 1977; Chambille and Rospars, 1981), Diptera (Power, 1946) and Lepidoptera (Boeckh and Boeckh, 1979), tracts are numerous. In Periplaneta, they can be assigned to groups of different glomeruli (Ernst eta!., 1977). In Hymenoptera they are initially less numerous and therefore more conspicuous: 3 main tracts (Jawlowski, 1948; Pareto, 1972; Masson, 1973) and more recently a 4th one (Suzuki, 1975; Mobbs, 1982), noted Tl-T4, have been described that supply different subsets of glomeruli. 2.1.3. Neuroreceptor terminals in glomeruli. In how many glomeruli does a neuroreceptor terminate? The functional specificity of glomeruli depends on this property of neuroreceptors. Goll (1967) in Formica and Mobbs (1982) in Apis, using the Golgi method, have maintained that fibers terminate in only one glomerulus, although biglomerular terminals may occasionally be seen, according to Mobbs. This specificity has also been evidenced in Periplaneta (Ernst and Boeckh, 1983; see camera Iucida

Structure of Antennodeutocerebral System

253

drawings in SchaUer-Selzer, 1984) and Drosophila (Stocker et al., 1983) by means of cobalt axonal migration. In the fly, each fiber with bilateral projection terminates in 2 symmetrical glomeruli (see also Borst and Fischbach, 1987). However, multiglomerular projections have been reported in Locusta. Antenna! fibers ramify shortly before terminating and up to 6 endings per axon have been observed that belonged to up to 3 distinct glomeruli (Ernst et al., 1977). The similar branching reported in Golgi preparations of Manduca (Ferguson et al., cited by Tolbert and Hildebrand, 1981) was not confirmed by dye impregnation of individual fibers (J. G. Hildebrand, personal communication). Are neuroreceptor terminals identical in all glomeruli? In the glomeruli of Musca, there are at least 4 types of terminals and only one type per glomerulus. The most common type consists of small varicose terminals (Strausfeld, 1976). In the bee (Mobbs, 1982), each glomerulus generally receives only one kind of terminal. Five terminal types have been distinguished by their branches (present or not, diffuse or not) and varicosities (large or small, sparse or dense). Each subset of glomeruli defined by tracts T1-T3 preferentially contains terminals of one kind. However, in Drosophila (Borst and Fischbach, unpublished observations) the terminals, which have only a few branches, were found to be very similar in different glomeruli, except in one glomerulus (probably innervated by aristal axons, see below) where they have many branches. 2.1.4. Relationships between flagellum and glomeruli. As each segment of the flagellum bears sensilla of several structural types and modalities, the problem is to determine which rules control the projection of neurons to the AL(s). Is there a preferential projection of certain neuroreceptors to the ALs? In particular, wh�re do the flagellar mechanoreceptors with a proprioceptive function project? As scape and pedicel receptors with this function project to the AMMC (cf. Section 2.1.2), a similar projection area may be hypothesized for those of the flagellum. In fact, the available data show that whatever their modalities, flagellar neuroreceptors project in the ipsilateral AL, mechano- and thermoreceptors included (except for some fibers, see Section 2.1.1(i) above). In Periplaneta, electrophysiological responses (reviewed by Boeckh et al., 1983; Boeckh and Ernst, 1983; Boeckh, 1984) are recorded in ALs after olfactory, thermal or mechanical stimulation of the flagellum (e.g. Boeckh, 1974; Waldow, 1975; Ernst and Boeckh, 1983). Evidence for the projection of the corresponding flagellar neuroreceptors to the AL is thus given (provided that the recorded neurons have no dendritic arborizations outside the lobes, which actually happens, as will be seen). In Manduca, some rare AL neurons respond to a puff of air (Matsumoto and Hildebrand, 1981); their small number is probably due to the predominantly olfactory function of the flagellum in this species. Neuroanatomical evidence is provided in Drosophila by cobalt fills of a sensillum with a probably proprioceptive function located on the arista (terminal part of the flagellum in Diptera). Its neurons project to 2 specific glomeruli in the ipsilateral AL (Stocker et al., 1983). Is there a spatial representation of the flagellum in the AL? If this were so, then different areas of the lobe (different glomeruli) would receive terminals from different parts of the flagellum. The available evidence does not support this view. Firstly, each area of a lobe is connected to all flagellar segments: a given neuron responds to stimuli applied at different points along the flagellum (Boeckh, 1974). But this is no proof for direct connection as the dendritic arborizations of the recorded neuron can be widely

254

JEAN PIERRE RoSPARS

distributed throughout the lobe. Secondly, each flagellar segment is connected to all glomeruli. Pareto (1972) demonstrated this by sectioning the last segment of the bee antenna and subsequently observing degenerations in all glomeruli, which is clearly inconsistent with the hypothesis of a spatial representation. Is there a functional representation? Stocker et al. (1983) gave evidence for this hypothesis. The first article of the flagellum of Drosophila (or funiculus) is very large and exhibits a characteristic pattern such that only one sensillum type is present in each of 3 areas, s. trichodea, coeloconica, and basiconica respectively. When these areas are slightly lesioned selective migrations of cobalt chloride in the neuroreceptors of a single sensillum type may be obtained. It is thus observed that the neurons from each sensillum type project predominantly to 3 specific glomeruli. A total of 7 glomeruli are involved, 2 of which receive terminals from 2 types of sensilla. Moreover, the area bearing s. basiconica is large enough to fill separately from it proximal, medial, and distal parts: the heavily stained glomeruli are the same in the 3 cases. It may be concluded that the representation of this flagellar region in the lobe is functional rather than spatial, thus indicating a functional specialization of glomeruli. In summary, whatever their sensory modalities, the flagellar neuroreceptors project to the glomeruli of the ipsilateral AL (and contralateral AL in Diptera). As an exception, some fibers might project to the AMMC and others directly to the corpora pedunculata. Each glomerulus receives fibers from sensilla located on a large part and even on the whole flagellar surface, which would explain why fibers from different points of the antenna! nerve section gather when entering the AL. Each neuroreceptor generally terminates in only one glomerulus and each glomerulus might receive only one kind of terminal. Neuroreceptors in sensilla of a given type terminate in specific glomeruli. This specificity will be analyzed from a different point of view in Section 3. 2.2. Neuroreceptors from scape and pedicel The neuroreceptors of the 2 basal segments of the antenna are in close relationship with the AMMC. 2.2.1. Tracts associated with the AMMC (Figs 4; 5). In the bee, the ventral part of the main antenna! nerve goes to the AMMC where it divides into 2 tracts, T5 and T6, which have been described with increasing accuracy (Jawlowski, 1948; Pareto, 1972; Suzuki, 1975; Mobbs, 1982; Budharugsa, 1984; Arnold et al., 1985). T5 seems to terminate in the AMMC, whereas T6 continues and splits into 3 branches. T6--1 projects to the protocerebrum and T6--2 to the subesophageal ganglion (SEG) where it connects to contralateral T6--2. T6--3 innervates the medial and posterior protocerebral area. Similar tracts have been described (except for T6--3) in the bumblebee (Fonta, 1984) and an ant (Delabie, 1984). These tracts are not restricted to the Hymenoptera. Power (1946) distinguished 2 types of fibers in the antenna] nerve of Drosophila. The thin fibers leave the nerve when it enters the brain and go to the AL. The large ones, more numerous, go into the AMMC. Some of them terminate there (like T5), others turn dorsally to the central commissure of the protocerebrum (one recognizes T6--1), ventrally to the thoracic nerve mass (like T6--2), or medially towards the symmetrical contralateral tract (like T6--3). Jawlowski (1948) described T5 (he noted MD) and T6--2 (noted MB) in the cockroach Periplaneta and in the bee, but identified T6--1 (noted MO) only in the bee.

255

Structure of Antennodeutocerebral System

anterior tegumentary nerve

---- PROTOCEREBRUM ANTENNAL LOBE

nerve of Janet chordotonal organ

l

OEUTOCEREBRUM SCAPE

I �

DORSAL LOBE

antenna! nerve

-

TRITOCEREBRUM PERIOSOPHAGEAL CONNECTIVE dorsal-lobe tractus

FrG. 4. Diagrammatic sagittal view of brain within head capsule to show nerves associated with · deutocerebrum. Secondary nerves (anterior tegumentary, Janet) are not present in all insect orders.

PROTOCEREBRUM

ANTENNAL LOBE

DORSAL LOBE

SUBESOPHAGEAL GANGLION

FIG. 5. Fiber tracts associated with antennomotor and mechanosensory center {AMMC) (modified from Mobbs, 1982).

2.2.2. Scapo-pedicellar projections (Fig. 6). In the fly, Calliphora and the cockroach, Periplaneta, convincing evidence for the innervation of AMMC from scape and pedicel was first given by Boeckh eta!. (1970). Cell degeneration is observed in the AMMC only after total removal of antenna; if the 2 basal segments are maintained, no degeneration occurs there. Stocker et a[. (1976) have confirmed these results on Drosophila. The ipsilateral AMMC shows few or no degeneration spots if the flagellum alone is removed, but a large number of spots if the antenna is sectioned between scape and pedicel and a still larger number if sectioned between pedicel and flagellum. The contralateral AMMC remains normal in appearance, but degenerating figures appear in the subesophageal

256

JEAN PIERRE RoSPARS

commissure and ganglionic mass (probably via T6-2). Strausfeld (1976) has reported from Go!gi preparations of Mus ca that most axons of the lateral part of the antenna! nerve terminate in the ipsilateral AMMC, whereas a few branch to the contralateral center. However, in Drosophila lateral antenna! fibers have been found that project directly to the pro- and mesothoracic ganglia (Strausfeld and Singh, 1980). The cobalt-filled axons of scape and pedicel mechanoreceptors (Johnston's and B6hm's organs) in Mandu ca project to characteristic regio�s of the AMMC (Hildebrand et a!., 1980; their further projection to the SEG and the ventral nerve cord is now considered uncertain, Arbas and Hildebrand, personal communication). None projects to ALs. In Locusta migratoria and S chisto c er ca gr egaria (Braiinig et a!., 1983; Gewecke, 1979) the selective infusion of cobalt chloride in B6hm's organ of scape and pedicel has shown that their neuroreceptors go to the brain via the antenna! nerve and give rise to collaterals, which all branch in the same areas (within the AMMC as it seems). Some terminate there, while others project further to the SEG. The fibers from sensilla campaniformia that surround the distal edge of pedicel have a similar course and branch in the same areas. Pareto (1972) mentioned that in the bee, the destruction of the scape hair plates (Mark!, 1962) brings about degenerations in the tract he called 4a (probably corresponding to T5) but fewer in 4b (T6). In S chisto c er ca gr egaria, a proprioceptor organ, which is elongated when the scape is depressed, is innervated by 4-5 neuroreceptors. Their cell bodies do not lie in the periphery as usual but in the protocerebrum, close to the pars intercerebralis. However, their dendrites are located in the AMMC, in accordance with the observations above (Briiunig, 1985). The projection of another organ with a proprioceptive function, Janet's chordotonal organ (1905, 1911) has been studied. It is located on the hymenopteran head close to the scape and records its movements. Its axons (about 12 in Camponotus, Masson, 1973) form a small nerve, which joins the main antenna! nerve shortly before entering the brain (Gall, 1967; Pareto, 1972) and is not sectioned when antenna is removed. Pareto showed that T6 carries at least some of these fibers to protocerebrum and SEG. A consistent picture emerges from these investigations. Neurons from mechano­ sensory sensilla with a proprioceptive function on scape and pedicel project to the AMMC via a tract called T5 in Hymenoptera and to the SEG via tract T6-2. Their protocerebral projections via T6-1, as suggested by Pareto, remain to be ascertained. 2:3. N euror e c eptorsfrom non a- nt enna! origin The main sensory input to deutocerebrurn comes from antennae. However, recent investigations have shown that neuroreceptor fibers do come from other head parts. They provide further dramatic examples of projections of afferent fibers according to their function rather than their topographical origin. Neuroreceptors associated with wall pore sensilla basiconica of the maxillary palps of Drosophila project to both ipsi- and contralateral ALs via the labial nerve. These sensilla are probably olfactory. The other sensilla of the maxillary palps are probably mechanosensory and project to the SEG (Singh and Nayak, 1985). The labial palps of Lepidoptera possess a pit, which houses sensilla with wall pore and characteristic lamellated dendrites (Lee et a!., 1985; Kent et a!., 1986; Bogner et a!., 1986). Their reaction to complex odors from host-plants and conspecific individuals (Lee et a!., 1986) are interpreted as reactions to C02 by Bogner et a!. (1986). In Pi eris (Lee

257

Structure of Antennodeutocerebral System FLAGELLUM

PEDICEL

Johnston organ

SCAPE

ANTENNAL LOBE

DORSAL LOBE dorsal-lobe tract motoneurone

Tract T6-iii

Btshm organ (hair plates)

axon from a hair-plate mechanoreceptor

FIG. 6. Diagrammatic view of antennodeutocerebral system to show projections of neuroreceptors from basal antennal segments. Scape and pedicel have external (hair plates) and internal (chordotonal organs like Johnston's organ) proprioceptors and few mechanoreceptors (not drawn). Their neurons only project to AMMC.

and Altner, 1986), Rhodogastria (Bogner et a{,, 1986), Manduca, Bombyx andAntheraea (Harrow et at., 1983; Kent et al., 1986), these neuroreceptors project, via the labial nerve and the SEG to a single glomerulus in each ALs. This glomerulus receives apparently no fiber from antennae and develops independently of other glomeruli (Kent and Hildebrand, 1983). It provides additional evidence of uniglomerular termination of neuroceptors. Similarly, scapo-pedicellar organs are not the only input to AMMC. Neuroceptors associated with wind-sensitive hairs on the head capsule are known to project via tegumentary nerves to the AMMC and further to the SEG and the thoracic ganglia in flies (Strausfeld, 1976), locusts (e.g. Griss and Rowell, 1986) and Manduca (Arbas and Hildebrand, 1986). 2.4. Motor neurons of AMMC In numerous insects, a nerve of small diameter is found that terminates on the antenna! extrinsic muscles moving the scape (Fig. 6; Viallanes' accessory antenna! nerve, Kenyon's internal antenna-motor nerve; see also Brousse-Gaury, 1971 for cockroaches, Gewecke, 1967 for Diptera, Janet, 1905 and Jonescu, 1909 for Hymenoptera). It appears to be motor only. The motor neurons of the intrinsic muscles moving the pedicel are carried by the main antenna! nerve or by a distinct but parallel nerve as in Formica (Goll, 1967). In Calliphora (Boeckh et al., 1970), after complete section of the antenna and degeneration of neuroreceptors, undegenerated axons remain, which have, therefore, their cell bodies in the brain. In the bee (Pareto, 1972), under the same experimental conditions, only about 20 thick fibers remain. Some (6) innervate the 2 intrinsic muscles via the antenna] nerve. The others (9-11) reach the 4 extrinsic muscles via the accessory nerve. Suzuki (1975) repeated the experiment, but injected Procion Yellow in the sectioned nerve after degeneration of the neuroreceptors. He thus revealed the dendritic arborizations and cell bodies (20) of the motor neurons innervating intrinsic muscles (the neurons of the unsectioned accessory nerve are not revealed).

258

JEAN

PIERRE ROSPARS

In Musca, Golgi-stained antenna! motor neurons, apparently identified by their characteristic dendritic arborizations, have cell bodies and dendrites located in the rear part of the AMMC (Strausfeld, 1976). In the moth, Manduca, the cell bodies (about 12) of these neurons revealed by means of cobalt chloride, are gathered in a deutocerebral area adjacent to the AL. Their dendritic arborizations lie in the AMMC (Hildebrand et al., 1980). In the deutocerebrum of the butterfly, Danaus plexippus, Nordlander and Edwards (1968), counted 20 cell bodies, which can be ascribed to motoneurons following morphological criteria defined by Wigglesworth (1959) and Cohen and Jackie! (1965). In conclusion, the cell bodies and dendritic arborizations of the motor neurons that control the antenna! muscles attached to scape and pedicel, are located in the AMMC and are 20-30 in number. An insect muscle is usually innervated by 3 neurons (slow, fast, inhibitory) but there can be as many as 8 (Tyrer, 1971). Therefore, one can predict 18--48 motoneurons for (typically) 6 antenna! muscles, which is in fairly good agreement with the above data,

3. NEURONAL ORGANIZATION OF ANTENNAL LOBES Three main types of interneurons have been demonstrated in ALs. The cell bodies of the first 2 types are in the AL. Intrinsic neurons (Section 3.1) have all their processes in the lobe(s). Projection neurons (Section 3.2) also have their dendrites in the AL, but their axons leave it. Neurons of the third type (Section 3.3) have cell bodies outside AL and their axons either terminate or send collaterals in the AL. Sexually dimorphic neurons are treated in Section 5.

3.1. Intrinsic neurons These neurons were initially recognized in Golgi preparations of Hymenoptera ("segmental interneurons", Gall, 1967; Mobbs, 1982) and Dictyoptera ("local interneurons", Ernst et al., 1977). Their cell bodies are in the AL. Their branches, which can differ in size and shape, are entirely restricted to the AL neuropil and always invade several glomeruli. In Diptera at least 4 neuron types can be distinguished according to their dendritic morphologies (Strausfeld, 1976). In Musca (Strausfeld, 1976) and Drosophila (Borst and Fischbach, 1987) some of these neurons are unilateral, while others (at least in Musca) are bilateral and pass apparently into the AL commissure (which carries the antenna! neuroreceptors with bilateral terminals, see Section 2.1.1). In honeybees, about 30 AL fibers were also found that cross the brain to the opposite lobe (Jawlowski, 1957; Mobbs, 1984; Schafer and Bicker, 1986). Selective staining confirmed these descriptions. In Locusta, some local interneurons were stained with antibodies to serotonin. Their cell bodies lie in the lobe and their branches invade several glomeruli (Tyrer et al., 1984). In Periplaneta, Selzer (1979) described one intrinsic AL neuron both anatomically and physiologically: its process innervates numerous glomeruli in the dorsal subset and it responds only to food odors. Its soma is in the front of the lobe (Ernst and Boeckh, 1983). Two other sets of such neurons were also demonstrated in subgroups A and B of the main dorsal cell group. Cells in A ramify in the macroglomerular complex (see Section 5) and small dorsal glomeruli and respond only to female odor. Processes of cells in B invade many glomeruli, which are all located in the dorsal subset and respond to mechanical and olfactory stimulations of the antennae (Ernst and Boeckh, 1983).

259

Structure of Antennodeutocerebral System

In Manduca, Matsumoto and Hildebrand (1981) described many types of local interneurons, intracellularly stained with cobalt and horseradish peroxidase. Some, present in males and females, are the most frequent neuronal type with 60% of sampled neurons. They are anaxonic; their primary neurite divides into several branches, which invade many glomeruli and most, if not all, glomeruli are innervated by them. They respond to odor stimuli applied to the ipsilateral antenna, and sometimes very weakly to mechanical stimuli. The same neuron may have multiple sites of spike initiation in different parts of its arborization (Christensen and Hildebrand, 1986). In conclusion, intrinsic neurons have been found in all insect orders studied. They have their cell body in the AL and they invade several glomeruli. The distinction of different classes of these neurons requires further anatomical, morphometrical and physiological investigations, which may profit by the available results on this type of neuron in the vertebrate olfactory bulb. 3.2. Projection neurons "Projection neurons" (Hoskins et al., 1986), also called "segmental interneurons" (Goll, 1967), "TOG neurons" (Ernst et al., 1977), "output neurons" (Matsumoto and Hildebrand, 1981), "principal neurons" or "relay neurons" (Burrows et al., 1982), or "primary antenna! neurons" (Mobbs, 1982) have their cell bodies in the AL and send their axons out of the AL through one of the output tracts (Section 3.2.1). Their AL origin (Section 3.2.2) and protocerebral destination (Section 3.2.3) are described (see also the classification scheme proposed by Christensen and Hildebrand, 1987). 3.2.1. Tracts connecting antenna! lobe to other parts of the CNS. Three main tracts connect the AL to the ipsilateral protocerebrum (Fig. 7). They are not specific to the AL and contain fibers from other origins (e.g. Tyrer et al., 1984). central complex calyces of mushroom bodies

-------._

"""'""-

PROTOCEREBRUM

____

lateral lobe of protocerebrum

OPTIC LOBE

antennal nerve

glomerulus

FIG. 7. Tracts of AL neurons that project to the protocerebrum. A projection neuron with soma in cell group, uniglomerular dendritic arborization and axon projecting via TOG to calyces of corpora pedunculata (mushroom bodies) and lateral protocerebrum is shown (modified from Ernst et al., 1977 and Mobbs, 1982).

260

JEAN PIERRE ROSPARS

The main tract is called inner antenna-cerebral tract (inner ACT, Kenyon, 1896), tractus olfactorio-globularis (TOG, Hanstrom, 1928), antenna-glomerular tract (AGT, Pearson, 1971) median AGT (Mobbs, 1982). It has been described in all investigated species (Hanstr6m, 1940; Power, 1946; Masson, 1977). It runs near the sagittal plane of the brain and projects to the calyces of the corpora pedunculata and further to a circumscribed region of the lateral protocerebrum (LP) called the lateral glomerular zone (Williams, 1975), lateral horn (Strausfeld, 1976; Strausfeld et al., 1980), lateral protocerebral lobe (Ernst et al., 1977) or lateral deutocerebral neuropile area (Mobbs, 1982; because this neuropile is thought to be derived from the deutocerebrum). The second tract is called middle ACT (Kenyon, 1986) or mediolateral AGT (Mobbs, 1982). In Apis, it branches out of the inner ACT, then divides into 2 branches. One branch projects to the LP (Mobbs, 1984), while the other runs to the corpora pedunculata behind the calyx (Mobbs, 1982, 1984). In Manduca, the middle ACT projects directly to the LP (Hoskins et al., 1986) and apparently corresponds to the first branch. In P eriplan eta, the posterior branch of the TOG going behind the calyx (noted VP in Jawlowski, 1948, 1954) might correspond to the second branch but is in fact of tritocerebral origin in this species (Ernst et al., 1977). The third tract in Hymenoptera (outer ACT of Kenyon, 1896; exterior olfactoroglobular tract TE of Jawlowski, 1948; lateral AGT of Mobbs, 1982) emerges from the AL independently of the inner ACT. It sends a collateral to the LP (Mobbs, 1984; possibly TR ramification of Jawlowski, 1948), then projects to the calyx. Some fibers pass further to the inner ACT and loopback to !he AL (Mobbs, 1984). The outer ACT has also been described in Orthopteroids (lateral antenna! tract TA of Jawlowski, 1954), in Lepidoptera (Hoskins et al., 1986) and in Diptera (at least partly homologous to the "broad root" of Power, 1946). Other output tracts are known. Arnold et al. (1985) described in Apis a posterior mediolateral ACT, which projects to the LP. In Drosophila, the broad root connects the AL to various brain parts and to the SEG (Power, 1946) also mentioned individual AL fibers, which run to the AMMC, tritocerebrum and SEG, and others, which run toward the ocellar tracts. In Manduca (Hoskins et al., 1986), a dorsal ACT has been described that divid�s into 3 branches, one of them probably projecting to the contralateral side of the brain. Individual fibers follow a tract, which connects the AL to the SEG, the tritocerebrum and probably the AMMC. However, lesions of cell groups and fiber tracts in the AL of P eriplan eta and Locusta reveal degenerations in the ipsilateral calyx and the LP only and none in the contralateral proto- or deutocerebrum or in the central body (Ernst et al., 1977).

3.2.2. D endritic tr e es in th e ant enna/ lob e. Hym enopt era. Goll (1967) found in the AL of Formica pratensis neurons which contribute to the inner ACT and has drawn them as neurons with only one dendritic tree in one glomerulus and with cell bodies gathered into only one group. In the bee lobe, the most frequent neurons in Golgi preparations project into the TOG and have very dense globular dendritic fields. Some have branches in several glomeruli located in different regions of the lobe, but most of them have uniglomerular terminals (Mobbs, 1982). They arborize either in the outer part of glomeruli where neuroreceptors terminate, or in the

Structure of Antennodeutocerebral System

261

core (Mobbs, 1984). However, dye-injected neurons with multiglomerular terminals that project into the protocerebrum have also been found (Iwama and Shibuya, 1986). Intracellular recording and staining with Lucifer Yellow show that the neurqns of the bee inner ACT with uniglomerular dendritic trees are generally bi- or multimodal (Homberg, 1984). They respond to mechanical, olfactory and gustatory stimulations of the antenna but also, in half of the cases, to an air puff directed at the head or the abdomen. The latter responses indicate that sensory information from non-antenna! origin is processed in the lobe.

L epidopt era. In Golgi preparations of Sphinx ligustri, neurons are found with large processes invading several glomeruli that send a thin fiber into the TOG (Pearson, 1971). . In the sphingid, Manduca (Matsumoto and Hildebrand, 1981), these neurons have dense uniglomerular arborization and cell bodies in AL cell groups. Their arborization is limited to the peripheral half of the glomerulus (these neurons respond to plant odor when excitable) or extends to the whole glomerulus (their cell bodies are then found in the small anterior cell group). However, recent observations have indicated that there are also multiglomerular output neurons in this species that send their axons through the inner ACT (J. G. Hildebrand, personal communication). The cell bodies of the output neurons that send their axons in this tract lie in all 3 AL cell groups, whereas those of the middle and outer ACTs lie in the ventrolateral group (Montague, et al., 1983; cf. Section 5.2.2 for Bombyx). -

Dipt era. In Golgi preparations of Drosophila, Borst and Fischbach (1987) found that each output neuron innervates only one glomerulus with extremely dense ramifications.

Orthopt eroid ea. In Locusta (Ernst et al., 1977), the primary neurites of Golgi-stained TOG neurons divide into several branches, which run along the margin between central and glomerular neuropils. These ring-like fibers emit processes arborizing within glomeruli (several in at least some cases). In P eriplan eta, the dendritic arborizations of cobalt-stained output neurons are uniglomerular (Selzer, 1979). Their cell bodies form a cluster in the dorsal group and the connected glomeruli are also close together in the ventral subset. They only respond to food odors. Ernst and Boeckh (1983) confirm that their dendritic trees are uniglomerular: they found an output neuron terminating in 3 (non-adjacent) glomeruli in only 1 case out of 70. They distinguished among output neurons 2 functional types common to both sexes: the unimodal (olfactory) and the multimodal (olfactory and mechanical), which both have their cell bodies in dorsal subgroup A. The innervated glomeruli also lie in the dorsal subset where they are gathered in 2 distinct groups. In the cricket, Ach eta dom esticus, output neurons stained with cobalt or Lucifer Yellow display dendritic trees in only one glomerulus in most cases (Schildberger, 1983). They respond to mechanical and olfactory stimulations of antenna but also to cereal and acoustic stimulations (Schildberger, 1984).

262

JEAN PIERRE RoSPARS

3.2.3. Terminal arborizations in the protocerebrum. Lepidoptera. In Sphinx (Pearson, 1971), the TOG runs to the base of the calyces of the ipsilateral corpus pedunculaturn. Secondary fibers arise at right angle to output axons, enter the calyces and ramify throughout. Most axons of the TOG proceed further and end in a 100 fJ-rn3 area where they produce a network of thin branches. Similarly in Manduca, output neurons with axons in the inner ACT send out collaterals ramifying mainly in the "lips" of the calyx, and terminate in the LP. Axons in the outer ACT ramify diffusely in a region rostral to the calyx (Montague et al., 1983). (See also Section 5.2.2). Diptera. In Musca, Strausfeld (1976) distinguished 4 types of fibers corning from ALs and terminating in different calyx regions. Varicose-spiny and varicose-smooth terminals are thick and look like those described by Pearson (1971). Type 3 neurons have small­ diameter endings and some of them send collaterals to the contralateral calyx. Type 4 are the thin "long olfactory fibers." In Drosophila (Borst and Fischbach, 1987) output neurons send one or several "club-shaped" side branches into the calyx and terminate iu the LP with dense ramifications. Orthopteroidea. Degeneration experiments in Periplaneta (and Locusta, Ernst et al., 1977), complemented by dye injection (Ernst and Boeckh, 1983), demonstrated that TOG neurons coming from ALs terminate in the ipsilateral calyces and in the LP. Each neuron sends a collateral to both calyces and continues to the LP. Terminals in calyces are small swellings, hardly visible by light microscopy that appear presynaptic to a large number of neighboring profiles in electron micrographs. These synapses are characterized by a large divergence: each axon has 60-80 terminals and each terminal is in synaptic contact with 10-20 postsynaptic elements (Boeckh et al., 1984). The LP terminals resemble those of calyces. In Acheta (Schiirmann, 1973; Schildberger, 1983), TOG output neurons project to the ventral part of calyces, usually to the anterior calyx, and are diffuse in the LP. Calyx terminals appear as characteristic thick knobs. 3.3. Neurons with cell bodies outside AL Strausfeld (1976) mentions in Musca descending centrifugal fibers corning from the TOG and terminating in the AL glomeruli. These elements are also present in bees (Mobbs, 1982); they are less frequent than output neurons and their thinner axons always branch over several glomeruli with rather spaced varicosities. The origin of these input neurons has not yet been determined (Strausfeld suggests that their cell bodies lie in the calyces, but with incomplete evidence). Ernst and Boeckh (1983) describe a neuron stained by diffusion of cobalt from the AL with a large cell body located in the pars intercerebralis. Its primary neurite divides into 2 branches close to the deutocerebrurn. One branch ramifies into the ipsilateral AL, whereas the other sends out a collateral to the contralateral lobe and proceeds through the tritocerebrurn to the SEG. A neuron with a seemingly identical cerebral morphology has been described in Gryllus bimaculatus (Richard et al., 1985). Its axon projects to the 3 thoracic ganglia where it shows large bilateral arborizations. It does not seem to proceed toward the abdominal ganglia. It responds to visual, auditory, and antenna!

Structure of Antennodeutocerebral System

263

(tactile) stimuli and would play a part in the control of leg motoneurons. According to these authors, it corresponds to neuron PI(2) : 5 or PI(2) : 6 in Schistocerca gregaria, 2 neurons identified by Williams (1975), which come from a small cluster of 7 pairs of cell bodies in the middle pars intercerebralis. It appears that neuron PI(2) : 5 is activated when the animal is moving and that it responds to various sensory stimulations (Kien and Altman, 1984). The physiology of this neuron type is also comparable to that studied by Olberg (1983a) in Bombyx (see Section 5). Rehder et al. (1987) identified in the honeybee a paired serotinin-immunoreactive neuron (called deutocerebral giant DCG), which apparently belongs to the same class. Its cell body is located in the cortex of the AMMC. Its anterior projection ramifies extensively in the ipsilateral AL, while its posterior projection sends off arborizations into the AMMC and runs into the SEG. One pair of neurons were stained in the SEG using antibodies against a neuropeptide hormone. These neurons project into the AL and further into the protocerebrum (Homberg et al., 1985). Further investigations are needed to classify these neurons according to their dendritic (output neurons) or terminal (input neurons) arborizations in the AL. Hamberg's (1984) and Schildberger's (1984) data (see Section 3.2.2) show that input neurons from non­ antenna! origin are present and that ALs do not only receive information from the antenna. The complexity of the system is thereby dramatically enhanced. 4. ANTENNA L - LOBE GLOMERULI 4.1. Structure of glomeruli What is a glomerulus? A general definition is given by Shepherd (1974); this is a "a synaptic complex enclosed in glial membranes or otherwise set apart," and a synaptic complex is "a set of specific synaptic connections between axonal and dendritic terminals, which terminals may themselves be interconnected." This definition originally applied to complexes observed in the vertebrate brain. It also applies to AL glomeruli. 4.1.1. Internal organization of glomeruli. Glomeruli present an internal organization, which is apparent from the spatial localization of synapses and the dendritic trees of intrinsic and projection neurons. In Diptera and Hymenoptera, this organization is well established. A glomerulus typically presents 2 regions: a central region, essentially made of neuroreceptor fibers and a peripheral region covering the former like a glove covering a finger, where neuroreceptors terminate and synapses are located. This has been first demonstrated in the fly, Calliphora by Boeckh et a!. (1970) who compared the glomerulus to a mushroom. Its large cap is made of thin fibers, which come from the flagellum as shown by their massive degeneration after the flagellum has been cut. The stem is made of thick cerebral fibers, which do not degenerate. The shape of neuron terminals in Drosophila (Borst and Fischbach, unpublished observations) are in good agreement with this description. Pareto (1972), also using the degeneration technique, observed this organization in almost all the glomeruli of workerbees. The thickness of the peripheral region decreases progressively in the direction of its opening, which is always located on the central side. This shows that glomeruli are radially oriented. The number of degenerating figures increases with the number of antenna! segments cut. In Golgi preparations isolated

264

JEAN PIERRE RoSPARS

fibers, probably neuroreceptors, are strictly limited to the external region, whereas branched fibers, probably AL neurons, are only found in the internal region. Subsequent observations by means of Procion Yellow and cobalt on Apis (Suzuki, 1975; Mobbs, 1982, 1984; Budharugsa, 1984; Arnold et al., 1985} and other Hymenoptera (Delabie, 1984; Fanta, 1984) have confirmed these results. However, Mobbs (1984} has shown that some bee projection neurons branch in the peripheral part of the glomerulus, whereas others branch in its core. Moreover, all Hymenopteran glomeruli do not exhibit this characteristic subdivision. In the worker and male bees, there is a cluster of 7 glomeruli whose external region is little differentiated or non-existent (Suzuki, 1975; Budharugsa, 1984; Arnold et at., 1985). Can the above scheme be generalized to all insects? This is unclear in Dictyoptera and Lepidoptera. The photomicrographs of deafferented ALs in Sphinx ligustri (Lund and Collett, 1968) and Periplaneta (Boeckh et al., 1970) do not show any intraglomerular pattern of degenerating figures. However, monoaminergic synapses, as seen by fluorescence microscopy, lie at the periphery of the glomerulus in Periplaneta (Frontali and Norberg, 1966; Rutschke and Thomas, 1975}. Similarly, in Blaberus, silver impregnation (Bodian) shows strongly stained dots and lines located at the periphery of glomeruli (Chambille and Rospars, 1981). A different organization has been evidenced in Manduca (Hoskins et al., 1986}. The dendritic arborizations of projection neurons are either distributed throughout the glomerulus or restricted to the external hemiglomerulus, i.e. the glomerulus half which is nearest to the external surface of the lobe. Moreover, the arborizations of local interneurons are more densely concentrated in the core and the internal half of each glomen.ilus than in the external half, a pattern which is similar to the arborization of processes containing GABA. 4.1.2. Intraglomerular synaptic connections. In all the cases where electron microscope observations of ALs have been made, it has been found that synaptic connections are restricted to glomeruli (Boeckh et al., 1970; Masson, 1972; Rutschke and Thomas, 1975; Schiirmann and Wechsler, 1969, 1970; Tolbert and Hildebrand, 1981}. Glomeruli are composed of neurites closely packed together where synapses may be recognized from their membrane differentiations and their associated vesicles. Small clear vesicles and large dense-cored vesicles are present in presynaptic elements. Tolbert and Hildebrand (1981} have distinguished 2 main types of presynaptic terminals. Type A, the most frequent, mainly contains small clear vesicles. Type B contains many large dense-cored vesicles. A third type (C), which is relatively rare, contains pleiomorphic clear vesicles. These data cannot be currently correlated with the transmitter type. Each presynaptic element contains an electron-dense structure close to or in contact with the presynaptic membrane (Boeckh et al., 1970}; Schiirmann and Wechsler, 1970) described in Manduca as a bar with a triangular cross-section (Tolbert and Hildebrand, 1981}. The most common synaptic contact in this sphinx has one presynaptic element of type A, B or C to which multiple postsynaptic elements (up to 7 were observed) are apposed by pairs. The postsynaptic element is sometimes presynaptic for other profiles, thus forming a serial synapse. A similar organization has been observed in Musca, Periplaneta, Locus/a, and Camponotus (Boeckh et al., 1970; Schiirmann and Wechsler, 1970; Masson, 1972; Ernst et al., 1977) with a comparable number of postsynaptic

Structure of Antennodeutocerebral System

265

elements associated with each presynaptic element on individual sections. Specific fillings of neuroreceptors with horseradish peroxidase (Tolbert and Hildebrand, 1981) reveal that their presynaptic terminals contain small, round, electron­ lucent vesicles and only a small number of dense-cored vesicles and are probably of type A. Terminals of types B and C, as well as some of type A, are tentatively attributed to CNS neurons associated with the AL. Presynaptic specializations are found on AL neurons, i.e. they are not mere postsynaptic elements with respect to antenna! axons. This has been shown indirectly by degeneration experiments (Boeckh et a!., 1970; Schiirmann and Wechsler, 1970; Ernst et a!., 1977) and directly by peroxidase fills of AL intrinsic neurons (Tolbert and Hildebrand, 1981). After removal of the flagellum and subsequent degeneration of the sensory neurons, Boeckh et a!. (1970) concluded that most of the synapses do not involve antenna! neurons, but only AL neurons. 4.1.3. Transmitters acting in antenna/ lobes. Several different neurotransmitters were found in the lobes (see reviews by Evans, 1980; Hildebrand and Maxwell, 1980; Klemm, 1976 and Sattelle, 1980). The terminals of chemo- and mechanosensory neurons very likely use acetylcholine (Sanes and Hildebrand, 1976c; Sanes et a!., 1977; Hildebrand et a!., 1979; Buchner and Rodrigues, 1983; Buchner et a!., 1986). Acetylcholinesterase has been found in some AL neurons projecting directly to the LP in Manduca (Hoskins and Hildebrand, 1983) and in the TOG of Drosophila (with choline acetyltransferase-like immunoreactivity, Buchner et a!., 1986). The transmitter 'Y-aminobutyric acid (GABA) mainly occurs in inhibitory synapses in Crustacea and mammals (Kuffler and Nicholls, 1976). Neuroreceptors lack GABA (Sanes and Hildebrand, 1976b; Schafer and Bicker, 1986). High levels of it are found in the ALs of Manduca (Maxwell et a!., 1978; Kingan and Hildebrand, 1985) where many of the synapses are inhibitory as shown by intracellular recordings (Matsumoto and Hildebrand, 1981; Christensen and Hildebrand, 1986; Waldrop et a!., 1987). More than half of local interneurons contain GABA acting as the principal or exclusive inhibitory transmitter in the lobes. Few fibers in the inner and outer ACTs, but many in the middle ACT are immunoreactive (Hoskins et a!., 1986). In bee lobes (Schafer and Bicker, 1986), 750 neurons are GABA-immunoreactive. The median ACT contains a few reactive fibers, one of which goes to the calyx via the LP. The inner and outer ACT are not stained, although the inner ACT is flanked by reactive fibers. Monoaminergic fibers have been demonstrated in the glomeruli of various orders (Frontali, 1968; Klemm, 1974). Such fibers have not been found in antenna! nerves. Dopamine has been identified at the periphery of glomeruli (Rutschke and Thomas, 1975; Rutschke et a!., 1976 in Periplaneta). Serotoninergic neurons of several types were also described: intrinsic neurons with arborizations in the glomerular periphery (Klemm et a!., 1984 in Periplaneta, Schiirmann and Klemm, 1984 in Apis), a unique neuron (one in each AL) with arborizations in the contralateral AL and in the LP (Hamberg et a!., 1986 in Manduca), a AMMC output neuron in the TOG (Tyrer et a!., 1984 in Locusta), and a protocerebral neuron entering the AL (Klemm and Sundler, 1983 in Schistocerca). Neuropeptides have also been detected with immunocytochemical methods in AL local interneurons and neurons coming from the SEG (Hamberg et a!., 1985, 1986 in Manduca).

266

JEAN PIERRE ROSPARS

From these data, a simplified model of intraglomerular organization may be proposed. Terminals of neuroreceptors have presynaptic areas with many small vesicles filled with acetylcholine (type A), which are apposed to multiple postsynaptic elements arranged in pairs. Each neuroreceptor would excite several AL neurons, probably both intrinsic and projection neurons. Terminals of AL neurons are quite similar, except that at least some of them contain large vesicles (type B) with various kinds of transmitters. Their postsynaptic elements are also arranged in pairs and most of them likely belong to AL neurons. A large number of feedback loops is therefore to be expected. These local circuits process sensory information within and between glomeruli. The information carried on a large number of input channels into glomeruli can be thus summarized on a much smaller number of output channels. The convergence of neuroreceptors on projection neurons appears as the anatomical consequence of the efficiency of this processing. The intraglomerular circuitry is a promising theme for further investigations. 4.2. Three-dimensional organization of the glomerular population 4.2.1. Identification of glomeruli. A number of anatomical arguments show that the glomerulus is an identifiable unit (Chambille et a/., 1978; Rospars et a/., 1979; see also Pareto, 1972). This notion was originally derived from the concept of identified neurons (e.g. Hoyle, 1975; Kater and Nicholson, 1973; Rowell, 1976). Unless otherwise stated, the glomerular properties considered in this section are common to both sexes. Differences between sexes are presented in Section 5. Evidence. The identifiability of glomeruli in the cockroach, Blaberus craniifer (Chambille et a/., 1980; Rospars and Chambille, 1981; Chambille and Rospars, 1981) and the moth, Mamestra brassicae (Rospars, 1983) is based on the following arguments (see also morphometric arguments in Section 4.2.2): (i) The number of glomeruli per lobe for animals of the same sex is basically constant: 107 in Blaberus and 67 in male Mamestra (cf. 1.2.2(b)), although slight variations have been observed (Rospars, 1983; Chambille and Rospars, 1985). (ii) Some glomeruli may be recognized from their own morphological characteristics, such as shape, size, orientation, proximity of fiber tracts of cell groups and TOG. In Blaberus (Chambille and Rospars, 1981), these anatomical criteria have made it possible to identify 21 glomeruli in silver impregnated (Bodian) serial sections. Taking these glomeruli as landmarks some of their neighbors could also be repeatedly recognized. Most of the 36 glomeruli (i.e. one third of the population) directly identified on sections lie in the dorsal part of the lobes. In this area, glomeruli form a unique layer and fiber tracts enter the glomerular neuropil, which makes identification easier. (iii) Other glomeruli are generally not identifiable on serial sections because of their number and similarity and of the slightly different orientation of sections between preparations. However, on computer reconstructions of ALs similar spatial patterns of neighboring glomeruli give clues as to homologous glomeruli (Fig. 8). This allows the "visual" identification of all glomeruli. (iv) These "visual" pairs determined independently from a series of lobes are internally consistent, i.e. when 3 lobes are compared two by two, I with 2 and 2 with 3,

Structure of Antennodeutocerebral System

267

FIG. 8. Bidimensional model of glomerular organization in adult antenna! lobes of Mamestra brassicae (top) and Blaberus cra niifer (bottom). 2 different individuals are shown for each species.

Internal sagittal views (observer is on axis perpendicular to sagittal plane and looks at right lobe from a point located in left lobe). Shaded glomeruli in Blaberus are identified from their morphological characteristics (from Rospars, 1983 and Chambille and Rospars, 1985).

pairs between lobes 1 and 3 deduced logically via lobe 2 are identical to pairs 1 : 3 established by direct visual comparison (Rospars and Chambille, 1981; Rospars, 1983). Comparison b etw e en sp eci es. The constancy of the number of glomeruli has been examined in P eriplan eta (Prillinger, 1981), and Manduca (Schneiderman et a!., 1983) and in each of the 4 subsets of glomeruli of Apis (Budharugsa, 1984; Arnold et a!., 1985). A variation of only 1 or 2 glomeruli has been observed in the most precise counts (Table 3). A more or less complete identification of glomeruli in ALs of various species has been done. In Manduca, the constancy of the glomerular organization is stated (Tolbert and Hildebrand, 1981; Matsumoto and Hildebrand, 1981) and a specific example is given with the 2 symmetrical glomeruli in each AL that are innervated by neuroreceptors from the labial pit organ (Kent et a!., 1986). The same pair of glomeruli has also been observed in 3 other species of Manduca, in Anth era ea, in Bombyx (Kent et a!., 1986), in the moth, Rhodogastria (Bogner et a!., 1986) and in the butterfly, Pi eris (Lee and Altner, 1986). In Bombyx (Koontz and Schneider, 1987), 4 glomeruli were morphologically identified (cluster of medial small glomeruli MSG 1-4; see also Section 5). In P eriplan eta (Prillinger, 1981), glomerular identification from position, shape and size would also be possible as illustrated by 3 glomeruli. In Drosophila, 19 glomeruli have been identified by means of cobalt backfilling from flagellar sensilla (Stocker et al., 1983) and 5 from maxillary palps (Singh and Nayak, 1985). In Apis worker, Pareto (1972) mentioned that some glomeruli can be recognized individually from their sizes and shapes, and

268

JEAN PIERRE RoSPARS E � ro 0 "" � • "

""

" • 40 � 0 " • E 0 � � "' N

0 z • � • "

20

Male No 1 glomerular radius (pm)

FIG. 9. Example of diagram to compare glomerular dimensions, here in Mamestra brassicae (from Rospars, 1983).

Budharugsa (1984) and Arnold et a!. (1985) have identified in this way 57 glomeruli of about 166. They have obtained similar results in male bees. Glomeruli thus appear as individually identifiable units, i.e. each glomerulus is present in lobes of different individuals. Since evidence for this stable organization of ALs has been collected in orders as different as Dictyoptera, Lepidoptera, Diptera, and Hymenoptera, it can be suggested that glomerulus identifiability is a widespread property in insects. This raises in turn new questions on the specificity of flagellum-glomeruli connections and on the functional specificity of glomeruli.

4.2.2. Invariance and variability of glomerular organization. Geometrical invariance. The 3-D organization of glomeruli in ALs is highly invariant. This is shown by morphometrical analyses based on the positions (coordinates X, Y, Z in a standard system) and dimensions (radii) R) of glomeruli (Rospars and Chambille, 1981; Rospars, 1983). The following results apply to comparisons between lobes of different individuals of the same sex. The radii of homologous glomeruli are statistically equal in Blaberus (Rospars et a!., 1979; Chambille et al., 1980) and in Mamestra (Rospars, 1983; Fig. 9). A similar correlation was also shown in 2 samples of about 50 homologous glomeruli from a worker and a male Apis (Arnold et al., 1985). However, this property does not seem to be confirmed in Pieris (Rospars, 1983). The coordinates of homologous glomeruli are also statistically equal (Rospars et al., 1979; Chambille et a!., 1980; Rospars, 1983). More precisely, the visual homolog is in most cases the nearest glomerulus (rank 1) to the point where it should lie if the lobes were identical. Consequently, a large number of glomeruli (80% in Blaberus, 75% in

Structure of Antennodeutocerebral System

269

Mam estra) can be identified by a computer algorithm from their spatial location alone (Rospars and Chambille, 1981). These automated results are consistent with the visual ones: less then 6% of glomeruli are paired differently in cockroaches and less than 2% in moths. Variability. Most glomeruli are invariant in location and size, at least in Blab erus and Mam estra. How then can the low number of variant glomeruli be interpreted? After size differences between individuals and experimental errors have been taken into account, a source of variability remains that explains at least half of the total variability and leads a given glomerulus of Blab erus to fluctuate by 6 fLm about its mean radius and by 27 fLm about its mean location (95% confidence limits; Chambille et a/., 1980). Because of this true biological variability, the expression of the invariant building scheme of ALs presents inter- and intraspecific variations. Invariance is apparently not maintained to the same degree in all species. Whatever the criterion used, it is slightly better in Mam estra than in Blab erus and much less good in the butterfly Pi eris. Does it mean that the lobes of Pi eris do not contain the same glomeruli? Probably not. But the positional variability is higher and does not lead to quantitative invariance. The glomerular organization of a given species is subject to a variability, which is minimum between lobes of a given individual, slightly higher between lobes of 2 individuals of the same sex, and maximum between individuals of different sexes. This is true for both location and size in Mam estra and in Blab erus (e.g. variant glomeruli in position amount to 20% in intraindividual, 24% in intrasexual and 34% in intersexual comparisons; from data in Rospars and Chambille, 1981). The excess of intersexual variability over intrasexual variably implies a true difference in the building schemes between sexes (see Section 5). In conclusion, there is morphometric evidence in some species that location and size of identified glomeruli are highly constant. The geometry of the glomerular arrays differ between species and perhaps also between sexes. It is affected by random factors that do not always act identically, so that the precision with which the building schemes are expressed also depends on the species. These observations need to be interpreted at the neuronal level (cf. Sections 6 and 7). For the species with a highly stable organization stereotactical applications may be considered.

5 . SEXUAL D I MORPHISM 5 . 1 . S exual dimorphism of ant enna e The sexual dimorphism of antennae displayed in some insect species may be related to sexual (Section 5 . 1 . 1) and other (Section 5 . 1 .2) behaviors. The results reviewed below mainly aim at providing the necessary background for interpreting AL structure. 5 . 1 . 1 . D imorphisms r esulting from s exual b ehavior. All propagated signals (visual, auditory, olfactory) are used in the long-range attraction of a mate, but each species preferentially or exclusively uses one of them (e.g. Otte, 1977; Fig. 10). Visual and auditory signals are not detected by the antenna! flagellum and therefore are not processed in the ALs. For example, in the butterfly family Pieridae, males find females

270

JEAN PIERRE RosPARs OLFACTION

Enslfera A. Acndoldea

0

Q

HEARING

I

Bloltldae

VISION

FIG. 10. Distribution of 3 major orthopteran groups on a triangle representing 3 major signaling modes. Each adaptative zone is dominated by one group (from Otte, 1977).

visually (Tinbergen eta!., 1972; Obara, 1964; Obara and Hidaka, 1964; 1968; Rutowski, 1977a; Silberglied, 1977; Silberglied and Taylor, 1978) and their antennae are not sexually dimorphic (Grula and Taylor, 1980; also Myers, 1968; see however, Section 5.1.2). The attracting signal may also be a sound wave detected by mechanoreceptors, e.g. in some male mosquitoes, the large Johnston's organ of the pedicel (Belton, 1974) is used to detect the flight sound of females (Ewing, 1977). Sexual attraction based on olfactory signals explains the large size and complex shape of the male flagellum in numerous species of Lepidoptera and Coleoptera (Schneider, 1964). This larger surface area accommodates a large number of olfactory sensilla specialized for trapping (Vogel, 1983) and detecting the attractant pheromone emitted by females. However, sexual dimorphism due to specialized sensilla may occur without overall modification of the antenna, as a result of an increased density of sensilla or because different types of sensilla occur in both sexes. Pheromone receptor cells have been extensively studied in moths (see reviews by Schneider, 1969, 1974; Kaissling, 1971; Kramer, 1978; Steinbrecht and Schneider, 1980; Seabrook, 1978; Priesner, 1979a, b; O'Connel, 1981; Mayer and Mankin, 1985). Extreme antenna! sexual dimorphisms are found in Bombyx mori (Steinbrecht, 1970) and Anth era ea polyph emus (Boeckh et a!., 1960). In cockroaches, sexual attraction by pheromones (reviewed by Barth, 1968; Persoons and Ritter, 1979; Schal et a!., 1984) and sexual dimorphism in favor of males (Chapman, 1982) are also known. In P eriplan eta, the only dimorphism evidenced involves olfactory sensilla (Schafer and Sanchez, 1976; Altner et a!., 1977). There is evidence that a special type of sensillum (long sw B; Schaller, 1978), houses the neuroreceptors that are sensitive to sex-pheromone (Sass, 1980). In Blab erus craniif er, such a pheromone has been found in behavioral experiments (Barth, 1961, 1970; see also Nutting, 1953; Barth, 1963; Grillou, 1973; Sreng, 1983). The antenna! sexual dimorphism of olfactory sensilla was recently demonstrated in this species (Chambille, 1986). In the same family, L eucopha ea mad era e displays no dimorphism (Schafer, 1971) and does not seem to make use of a volatile sex­ pheromone (Roth and Barth, 1967).

Structure of Antennodeutocerebral System

271

5.1.2. Dimorphisms from other origins. Not all antenna! sexual dimorphisms are related to attractant sex-pheromone or are shown by males. Species of various orders, in which females have less than ca 1,500 olfactory sensilla, are generally characterized by more sensilla in females then in males and no attractant sex-pheromones are known to be used (the reverse is found for species with more than 1,500 olfactory sensilla in females, Chapman 1982). The antenna] dimorphisms of flagellar sensilla unrelated to attractant sex-pheromone detection may involve olfactory sensilla, e.g. in mosquitoes (Ismail, 1962; Mciver, 1971) and moths; in Bombyx mori the female has twice as many short trichoid sensilla as the male (Steinbrecht, 1970). Antenna! dimorphism may also involve contact chemo­ receptors, e.g. in Blaberus craniifer whose male has more of these sensilla than the female (Chambille, 1986) and in Mamestra configura/a whose female antenna bears 2 more terminal-pore sensilla per segment than the male (Liu and Liu, 1984). These differences may be related to female behaviors, which involve olfactory signals, such as the blood meal of mosquitoes (Davis and Sokolove, 1976), the search for a host-plant after fecundation (Visser, 1986) and the perception of the aphrodisiac pheromone emitted by male Lepidoptera during courtship (Birch, 1974). In the pierid butterflies, for example, females need unimpaired antennae for successful mating (Taylor, 1973) because an aphrodisiac pheromone (Rutowski, 1977b) is emitted during courtship by specialized male structures (Bergstrom and Lundgren, 1977). This olfactory signal is used for the specific recognition of mates already in close vicinity. In conclusion, various types of antenna! dimorphism have been observed. The most conspicuous are related to the remote detection of mates by auditory or olfactory (attractant sex-pheromone) signals. Other flagellar dimorphisms have also been reported for olfactory and gustatory sensilla, which do not respond to attractant sex-pheromones but probably play a role in characteristic female behaviors. 5.2. Macroglomerular dimorphism of antenna/ lobes The sexual dimorphisms observed on the flagellum give rise to neuronal and glomerular variations in the AL. The most conspicuous difference involves a male macroglomerular mass (Section 5.2.1) but other differences are present (Section 5.2.2). 5.2.1. Macroglomerular dimorphism. The presence of a glomerular mass apparently specific to adult male ALs was initially noticed by Bretschneider (1924) in Lepidoptera (Bombyx mori, Lasiocampa quercus). It was also mentioned by Jawlowski (1948) in Hymenoptera (Vespa germanica, V. crabro, Apis mellifera) and Dictyoptera (Periplaneta americana, P. orienta/is). More recently, this mass has been described in other species (see below) and named "macroglomerulus" (Boeckh et a!., 1977) or "macroglomerular complex" (MGC, Hildebrand eta!., 1980). The available data are still too fragmentary and the sample of species studied too small to allow definitive conclusions. However, they permit some remarks: (i) The dimorphic glomerular mass varies in morphology. It can be made either of 1 or several distinct glomeruli: 1 in Blaberus craniifer (Chambille and Rospars, 1981, 1985); 2 in Mamestra brassicae (Rospars eta!., 1983; Rospars, 1983); 3 in Bombus terrestris and B. hypnorum (Fanta, 1984); or of several closely apposed or fused glomeruli -at least 3 in Lymantria dispar, 4 in Bombyx and Antheraea polyphemus (Koontz and Schneider, 1987; see also Boeckh, 1979; Kanzaki and Shibuya, 1983); or of a unique, very bulky

272

JEAN PIERRE RosPARS TABLE 4. CORRELATION BETWEEN ANTENNAL DIMORPHISM AND MACROGLOMERULAR DIMORPHISM

Species

NR

NR%

MGC

Blaberu s craniifer