Dorso-abdominal scent glands and metathoracic evaporatoria in adults of central European Rhopalidae (Hemiptera: Heteroptera), with a discussion of phylogeny and higher systematics

The reservoirs of dorso-abdominal scent glands and the occurrence of the metapleural scent gland evaporatoria in the adults of nine central European and one North American species in the family Rhopalidae (Hemiptera) were studied. All published data about the persistence of the dorso-abdominal scent glands in rhopalid adults are reviewed, and systematic and phylogenetic implications are derived from the patterns of variation. GENERAL CONSIDERATIONS Detailed studies of the dorso-abdominal scent glands of Western Hemisphere adults of the heteropteran family Rhopalidae have been reported in a number of papers. Formerly, authors commonly recorded completely reduced larval abdominal glands in heteropteran adults and their replacement by the metapleural scent glands (MTGs). However, several recent studies have reported the persistence of abdominal glands in adults. At least one gland, usually the anteriormost, has been found to be functional (Davidová-Vilímová, in prep., and Table 1). Dupuis (1953) first mentioned the persistence of the abdominal glands in rhopalid adults: “ ...two orifices of dorso-abdominal glands, medially on segment IV-V and V-VI.” He illustrated the abdominal dorsum with two conspicuous dorsal gland ostioles in two Rhopalini (Rhopalinae), Corizus hyoscyami (L., 1758) and Rhopalus (Aeschyntelus) maculatus (Fieber, 1837). Similarly, Schaefer (1965) illustrated the abdominal dorsum with two unpaired ostioles of three species, C. hyoscyami, Harmostes reflexulus (Say, 1832) (Rhopalinae: Harmostini), and Boisea trivittata (Say, 1825) (as Leptocoris trivittatus) (Serinethinae); other examples are in Table 1. Later, Aldrich et al. (1979) described the adult abdominal glands of Niesthrea louisianica Sailer, 1961 (Rhopalinae: Niesthreini) and Jadera haematoloma (Herrich-Schaeffer, 1847) (Serinethinae). Schaefer & Chopra (1982) used these data, the persistence of two adult functional abdominal glands, to characterize Seri­ nethinae. Aldrich et al. (1990a) mentioned scent glands in Jadera haematoloma and J. sanguinolenta (F., 1775), and Aldrich et al. (1990b) described in detail the adult abdominal glands of eight rhopalid species. The character “one gland, anterior, retains in the adults” was used to characterize the subfamily Rhopalinae, based on the study of one representative, N. louisianica. The authors described glands of two serinethine genera, Boisea Kirkaldy, 1910 (two spp.) and Jadera Stal, 1862 (five spp.). Aldrich (1988) studied the chemistry of adult scent glands of Leptocoris isolatus (Distant, 1914) (Serinethi­ nae), and of species in Niesthrea Spinola, 1837 and Jad­ era. He (1995) also mentioned briefly the Rhopalidae in his review of scent glands ofPentatomomorpha. Putshkov (1986), in his study of Palaearctic Rhopalidae, described briefly, and illustrated, the position of the adult dorsoabdominal scent gland ostioles in several species (see Table 1). Moulet (1995) mentioned generally the persis­ tence of two dorso-abdominal scent glands, with open ostioles, in adults ofRhopalidae. MTGs are generally developed, as autapomorphic structures, in adult heteropterans. However, the English name for Rhopalidae, scentless plant bugs, suggests an absence of MTGs. But this state is not general for all rhopalids. Chopra (1967) first summarized the states of MTGs in Rhopalidae, and used them in a new higher clas­ sification of the family. His opinion on anagenesis of MTG was later corroborated by Schaefer & Chopra (1982). The primitive state is represented by a separated metathoracic epimeron and episternum and a distinct ostiole, with a well-developed evaporatorium and/or peritreme (Rhopalinae: Niesthreini, Rhopalini). Reduction and loss of an externally visible ostiole and associated structures are more advanced stages. Reduction pro­ ceeded as follows: Fusion of epimeron and episternum accompanied by reduction of peritreme, and ostiole hidden between mesoand metacoxa (Rhopalinae: Chorosomatini); loss of peritreme (Rhopalinae: Harmostini, Maccevethini); loss of an externally visible ostiole, the gland opening into the coxal cavity (independently Rho­ palinae: Corizomorphini and Serinethinae). This strong reduction of the MTG is correlated with retention of func­ tional larval abdominal glands in serinethine adults (e.g., Aldrich et al., 1990b). Schaefer (1972) mentioned loss or

Detailed studies of the dorso-abdominal scent glands of Western Hemisphere adults of the heteropteran family Rhopalidae have been reported in a number of papers.Formerly, authors commonly recorded completely reduced larval abdominal glands in heteropteran adults and their replacement by the metapleural scent glands (MTGs).However, several recent studies have reported the persistence of abdominal glands in adults.At least one gland, usually the anteriormost, has been found to be functional in prep.,and Table 1).Dupuis (1953) first mentioned the persistence of the abdominal glands in rhopalid adults: " ...two orifices of dorso-abdominal glands, medially on segment IV-V and V-VI."He illustrated the abdominal dorsum with two conspicuous dorsal gland ostioles in two Rhopalini (Rhopalinae), Corizus hyoscyami (L., 1758) and Rhopalus (Aeschyntelus) maculatus (Fieber, 1837).Similarly, Schaefer (1965) illustrated the abdominal dorsum with two unpaired ostioles of three species, C. hyoscyami, Harmostes reflexulus (Say, 1832) (Rhopalinae: Harmostini), and Boisea trivittata (Say, 1825) (as Leptocoris trivittatus) (Serinethinae); other examples are in Table 1.
MTGs are generally developed, as autapomorphic structures, in adult heteropterans.However, the English name for Rhopalidae, scentless plant bugs, suggests an absence of MTGs.But this state is not general for all rhopalids.Chopra (1967) first summarized the states of MTGs in Rhopalidae, and used them in a new higher clas sification of the family.His opinion on anagenesis of MTG was later corroborated by Schaefer & Chopra (1982).The primitive state is represented by a separated metathoracic epimeron and episternum and a distinct ostiole, with a well-developed evaporatorium and/or peritreme (Rhopalinae: Niesthreini, Rhopalini).Reduction and loss of an externally visible ostiole and associated structures are more advanced stages.Reduction pro ceeded as follows: Fusion of epimeron and episternum accompanied by reduction of peritreme, and ostiole hidden between meso-and metacoxa (Rhopalinae: Chorosomatini); loss of peritreme (Rhopalinae: Harmostini, Maccevethini); loss of an externally visible ostiole, the gland opening into the coxal cavity (independently Rho palinae: Corizomorphini and Serinethinae).This strong reduction of the MTG is correlated with retention of func tional larval abdominal glands in serinethine adults (e.g., Aldrich et al., 1990b).Schaefer (1972) mentioned loss or Table 1.The persistence of the adult dorso-abdominal scent glands in the Rhopalidae.Median gland -gland between terga 4-5; posterior gland -gland between terga 5-6, glands and ostioles unpaired; + -presence of gland, and/or gland reservoir; il.-only illustration; sex.dim.-gland reservoir sexually dimorphic.All cited references report original data except for Schaefer & Chopra (1982) who quote already published data.Chopra (1967), has been basically accepted (e.g., Schaefer, 1994): Two subfamilies, Serinethinae and Rho palinae, the latter with six tribes.Schaefer (1993) summa rized data on the distribution of all the rhopaline tribes.He discussed centres of origin of individual tribes, in respect of Chopra's (1967) phylogeny and classification, their present distribution, and the Rhopalidae invasion of the New World.
The subfamily Serinethinae is generally considered the most advanced taxon of Rhopalidae (e.g., Chopra, 1967;Schaefer & Chopra, 1982;Putshkov, 1986).Ahmad & Afzal (1978) suggested that the data on MTG indicate that Serinethinae are more primitive than Rhopalinae.The former subfamily possesses a paired MTG reservoir, occurring also in the most primitive Heteroptera, com bined with an advanced character, the MTG ostiole opening in the coxal cavity.However, Schaefer & Chopra (1982) did not consider this character as primitive, and agreed with many other authors that the paired condition is derived from the single condition.Schaefer (1978), in a comprehensive morphological study of the Rhopalidae pygophore, did not find any pygophoral character clearly separating Serinethinae and Rhopalinae, and he suggested basing the classification of Rhopalidae on the tribal rather than subfamilial level.Putshkov (1986) mostly agreed with Chopra's (1967) conclusions, but he recommended also using characters of the eggs and larvae.Contrary to Chopra's (1967) conclu sions, Putshkov (1986) considered the Eastern Hemi sphere to have been the centre of the origin of Harmostini, which thus was the source area for dispersal to New World.Putshkov (1986) further supposed a closer relationship of Maccevethini to Rhopalini + Niesthreini than to Corizomorphini + Harmostini.He did not agree with close relationship of the genera classified in the tribe Maccevethini since their eggs and larvae are very dif ferent from each other.Putshkov classified these genera in the tribe Rhopalini, he thus synonymized Maccevethini with Rhopalini.Schaefer & Chopra (1982) analyzed the Rhopalidae cladistically and suggested one change in Chopra's (1967) cladogram: The clade Serinethinae is not a sister group of all other Rhopalidae (= Rhopalinae), because there is no apomorphy separating Rhopalinae from Seri nethinae.On the contrary, Maccevethini and Serinethinae share obvious synapomorphies (the number of conjunc tival phallic appendages).However, in this conception Rhopalinae is paraphyletic.Schaefer & Chopra (1982) concluded that the tribes Rhopalini and Niesthreini are closely related and distinctly isolated from the other clades.Therefore, it would be possible to lump them into one taxon.The Serinethinae should be reduced to a tribe, but the authors preferred to leave the higher classification suggested by Chopra (1967) because of the amount of anagenesis the clade Serinethinae has undergone.
The most recent study of rhopalid higher classification was published by Li & Zheng (1994).The developed peritreme represents a plesiomorphic character, and it is retained in the monophyletic tribe Rhopalini and in repre sentatives of the tribe Niesthreini.The absence of the MTG ostiolar peritreme and even the ostiolar opening are apomorphies.This advanced state (reduction) evolved in Maccevethini and Chorosomatini; the most advanced state (loss) in Harmostini and Serinethinae (the tribe Cori zomorphini was not studied).The authors elaborated the higher classification of Rhopalidae based on almost 40 somatic and genital characters.The genera Stictopleurus Stal, 1872 and Maccevethus Dallas, 1852 were classified by Chopra (1967) in a new tribe Maccevethini.Later, Putshkov (1986) did not accept this tribe and classified both genera in Rhopalini.These two genera were not placed in the same clade in the cladogram of Li & Zheng (1994), they were far distant from each other.Their clas sification is therefore still open.The serinethine genera are closely related to Chorosomatini; the former taxon is highly advanced, but if classified as a subfamily then the Rhopalinae becomes a paraphyletic taxon.Therefore the authors suggested that all clades in Rhopalidae, including Serinethini, be considered only tribes.Generally, the results of Li & Zheng (1994) correspond with the tribal classification of Chopra (1967), except for the distant relationship of the two Maccevethini genera.
The reservoirs of dorso-abdominal glands and their ostioles, and the presence or absence of evaporatoria of the metapleural scent glands, were studied by one of us (MN, particularly in her Master's thesis) in the adults of both sexes of nine central European Rhopalidae: Chorosoma schillingii (Schilling, 1829), Corizus hyoscyami, Liorhyssus hyalinus (F., 1794), Maccevethus caucasicus (Kolenati, 1845), Myrmus miriformis miriformis (Fallen, 1807), Rhopalus (Aeschyntelus) maculatus, Rhopalus (Rhopalus) subrufus (Gmelin, 1790), Rhopalus (Brachycarenus) tigrinus (Schilling, 1829), and Stictopleurus abutilon (Rossi, 1790); and the North American Boisea trivittata.The results we present here include both a survey and analysis of the literature on these glands throughout the Rhopalidae, as well as the first acount of the glands in the Rhopalini.The results of our study can support some published phylogenetic conclusions, how ever, we recognize that no final conclusion can be drawn from study of a single character.
In the present paper, we review persistence of the reser voirs of dorso-abdominal scent glands in adult Rho palidae and development of the evaporatoria of the metapleural scent glands.Our aim is to evaluate the pat tern of phylogenetic variation in these characters, and to employ the data obtained in considering the higher classi fication of the family.

Methods
The dry specimens were boiled for about 5 min in 10% KOH.They were removed to and dissected and drawn in dis tilled water.The abdomen was separated, the abdominal cavity was opened up laterally, and all tissues surrounding the gland reservoirs were removed with watchmaker forceps.The external structures associated with the metapleural glands (evaporatorium, ostiole, groove) were studied and illustrated on dry speci mens, with the second and third pairs oflegs removed.

Terminology
Three dorso-abdominal scent glands are generally developed in Heteroptera.However, only two glands occur in Rhopalidae; the topographical median and posterior ones.We modify the ter minology of Aldrich et al. (1990b) ("anterior" and "posterior" glands) to median and posterior glands, for those between terga 4-5 and 5-6, respectively.The anterior gland, of the three func tional in other heteropteran larvae, lies between terga 3-4, and is not functional in Rhopalidae.
The following terms are used for the structures associated externally with the metapleural glands: Ostiole (e.g., Hepburn & Yonke, 1971;Ahmad & Afzal, 1978), for the external opening of gland.Peritreme (all authors of studies on metapleural glands mentioned), for ostiolar groove, running from the ostiole, mostly laterally.Evaporatorium (as "evaporative area," e.g., Chopra, 1967;Schaefer, 1972), for a part of external surface of metapleura, connected with the ostiole, and covered with a spe cific, complex, mushroom-like structure.Groove-like structure, for a part of the border between the mesothorax and metathorax, where the sclerites are separated from one another and form lat eral walls of the groove-like structure, covered by the same structure as the evaporatorium.
The terms used for the dorso-abdominal glands are as follows: Gland reservoir, for the membranous structure, mostly sacshaped, on the ventral surface of the abdominal dorsum, beneath the ostiole, a chink-shaped opening, surrounded by elevated, roughly ring shaped cuticle.

Reservoirs of dorso-abdominal scent glands in adults (Figs 1-9)
The number and position of the reservoirs in adults of studied Rhopalidae are the same as in their larvae (unpubl.data).Two unpaired reservoirs are developed, each with an unpaired ostiole and surrounded by a sclerotized cuticular ring.The ostiole of the median gland is located between abdominal segments 4-5, that of the posterior gland between segments 5-6.Abdominal tergum 5 is narrowed between the ostioles (Figs 1-5, 7-9), except in Chorosoma (Fig. 6).
The reservoirs are small in relation to body size, but conspicuous, and roughly sac-shaped; either both reser voirs are of the same shape, or the posterior one has two small lateral appendices giving a trilobate outline to the reservoir.The reservoirs are colorless and almost trans parent, probably because the studied material was dried.

Rhopalini
Corizus hyoscyami (Fig. 1).Median reservoir shaped as a longitudinally elongated sac, slightly larger in male than in female.Posterior reservoir in both sexes globular, of the same size as female's median one.
Liorhyssus hyalinus (Fig. 2).Median and posterior res ervoirs very small but distinct in both sexes, developed only as small globular sacs beneath ostioles, only slightly larger than ostiolar diameter.
Rhopalus maculatus (Fig. 3).Median and posterior res ervoirs larger than their ostioles, conspicuous, slightly longitudinally elongated and sac-shaped, of same size in both sexes.
Stictopleurus abutilon (Fig. 9).Both reservoirs con spicuous, larger than their ostioles.Median reservoir sac- like, only slightly elongated, slightly smaller than poste rior in both sexes; in female of variable size, from one half to double of size in male.Posterior reservoir trilobate, larger than median in both sexes.

Serinethinae
Boisea trivittata (not illustrated).Median and posterior reservoirs very small, globular in both sexes, posterior a little larger than median.

Metapleural scent gland evaporatoria (Figs 10-14)
The species were examined with respect to the presence of an externally visible ostiole and peritreme, and the development of the evaporatorium, including groove-like structure.
In Rhopalini, one pair of ostioles is developed ventrally on the metathorax.Each ostiole lies between the mesocoxa and metacoxa, not visible ventrally.The peritreme is not developed.The evaporatorium covers only a small median part of the epimeron, and a larger part of the epis ternum (median area, and a narrow anterior area running along the metapleural anterior margin).The evaporatorium also extends as a very narrow strip on the posterior mesopleural margin.Groove-like structure developed.

CONCLUSIONS AND DISCUSSION
It is difficult to decide conclusively without histological study if the dorso-abdominal scent glands (= DAGs) are functional in the adults studied.We can only estimate functionality from the size of the gland reservoir and the presence of its ostioles.We assume that when both reser voirs persist as sacs of the same size, connected with ostioles, they are functional.
The abdominal glands of Boisea trivittata (Seri nethinae) were also studied by Aldrich et al. (1990b).They found two unpaired glands, the median reduced and empty, the posterior functional.We found two, sacshaped, unpaired gland reservoirs directly beneath the ostioles.The reservoirs are distinct, quite small, the median smaller than the posterior.Thus we agree with Aldrich et al. (1990b).The reservoir of functional poste rior gland is small, of the same size as the smallest gland reservoirs we found in Liorhyssus hyalinus, a much smaller bug than B. trivittata.We can assume, therefore, that at least one, but more probably two, dorso-abdominal scent glands are functional in adult Rhopalidae.
The existence of no more than two abdominal glands in larvae of all Rhopalidae represents an advanced condi tion; the possession of three glands, which occur in most heteropterans, is considered plesiomorphic (e.g., Aldrich et al., 1978).In Rhopalidae the anteriormost gland is completely reduced, including its ostiole.Two reservoirs of abdominal glands are developed in adults of all rhopalids studied.Sexual dimorphism occurs only in Corizus hyoscyami; the median reservoir is slightly larger in the male than in the female.The species is the only one studied with an aposematic colour pattern (see also below); it is questionable, however, if the sexual dimor phism is connected with this phenomenon.Sexual dimor phism of reservoirs of adult abdominal glands exists in several other heteropteran taxa (review Davidová- Vilímová, in prep.), but has not yet been described in Rhopalidae.
The tendency for reduction of MTG is considered advanced in Heteroptera.Also, the unusual chemical composition of gland secretion indicates an advanced phylogenetic position of Rhopalidae.The complete absence of secretional activity of MTG is known in Serinethinae.Gland secretion of only one rhopaline species, Niesthrea louisianica, was studied (Aldrich et al., 1979(Aldrich et al., , 1990b)).The pattern of MTG secretion fits that of the larval posterior DAG, which supports the presumption that the MTG functionally replaces the larval DAG.The possible groundplan of the adult Rhopalinae scent glands was proposed by Aldrich et al. (1990b) as follows: One repellent gland = MTG, one pheromonal gland = median DAG.The situation in Rhopalini (three genera and five species studied) agrees with the suggested Rhopalinae groundplan.MTGs are functional, two reservoirs of DAG were found, and the median gland could be functional at least in Liorhyssus hyalinus, Rhopalus maculatus, and Rhopalus subrufus; and almost surely it has a pheromonal function in C. hyoscyami, a species with sexual dimor phism (judging from the comparison with other Hetero ptera, review in Davidová-Vilímová, in prep.).We can suppose that the posterior gland is reduced functionally, and its reservoir persists only structurally as a small mem branous sac.Probably the posterior gland, or both glands, are functional in R. tigrinus, a species with a large, trilo bate posterior reservoir, but a small, globular median res ervoir.Published data about the occurrence of the MTG evaporatorium in the Rhopalini are listed in Table 2.
The situations in Chorosomatini and Maccevethini (two genera and two species studied of both tribes) do not fit the suggested Rhopalinae groundplan.MTGs in Chorosomatini are conspicuously reduced, and two elongated and sac-shaped reservoirs of DAG were found; the median gland is probably functional.Large differences were found in size; reservoirs are the largest in Myrmus miri formis, conspicuously larger than their ostioles, and very small in Chorosoma schillingii, of the same size as their ostioles.MTGs are also reduced in Maccevethini, which have two reservoirs of DAG too; functionality of the glands is uncertain.The posterior reservoir is larger than the median in both species.Aldrich et al. (1990b) supposed an evolution toward reduction of MTG in close connection with feeding spe cialization on toxic plants, and a consequent use of plant toxins in the form of blood toxins.The adult MTGs are then redundant.Serinethinae are, according to the authors, a typical example of such specialization (on Sapindaceae, see Schaefer & Chopra, 1982), accompa nied by aposematic colouration.But the situation with reduced MTG is surely not so easily explained.Corizus hyoscyami possesses a typical aposematic colour pattern, but has both MTGs and two reservoirs of DAG developed.Corizus hyoscyami is widely polyphagous, not fixed on toxic plants (for a list of known host plants see Nejedlá, 1997).The explanation of complete reduction of MTG in some Rhopalidae, and some other Heteroptera, is probably more complicated.
Other unanswered questions remain.Why in Seri nethinae is either the median (Jadera), or the posterior (Boisea) gland functional, when the adult DAG retains its repellent function?When the dorso-abdominal glands have a pheromonal function in adults, the anterior one is usually the functional one (review Davidová-Vilímová, in prep.; for information about Rhopalidae see Aldrich et al., 1979Aldrich et al., , 1990a)).Then why has the posterior gland the phe romonal function in Jadera?
It is commonly accepted that Niesthreini and Rhopalini are closely related and the most generalized.MTGs are functional and all the associated structures are developed only in these two tribes.Only a single DAG (median) is functional in Niesthreini (Aldrich et al., 1979;Aldrich, 1988).Chemical properties of secretion and thus the functionality of DAGs have not yet been studied in Rhopalini.Two unpaired reservoirs of DAG structurally per sist in the five species studied.The median gland reser voir is usually similar or larger than that of the posterior gland.The two unpaired glands in Niesthrea louisianica are structurally distinguishable; Aldrich et al. (1990a) recorded that only the anterior (= median) gland was functional, the posterior was inactive.We could assume that DAGs are morphologically homologous in the Niesthreini and Rhopalini.This result corroborates the hypothesis on close relationship of Rhopalini and Niesthreini.
The most advanced state of the MTG ostiole, occurring in Corizomorphini and Serinethinae, most probably evolved independently (Chopra, 1967).Corizomorphini, the most advanced tribe, has been studied only rarely, and the state of DAG is unknown.Most authors supposed this tribe to be closely related to Harmostini.