The soapberry bug, Jadera haematoloma (Insecta, Hemiptera, Rhopalidae): First Asian record, with a review of bionomics

Abstract The soapberry bug, Jadera haematoloma (Herrich-Schäffer, 1847) (Insecta: Hemiptera: Heteroptera: Rhopalidae: Serinethinae), a species native in tropical and subtropical regions of the New World and accidentally introduced to Hawaii, is reported for the first time from Asia (Taiwan). This record represents the first occurrence of the species in Asia. Stable populations composed of hundreds of specimens were found in seven localities of Kaohsiung City and one locality in Tainan City, and a single specimen was observed in Chiayi County. Aggregating adults and larvae fed in large numbers on the sapindacean plants Cardiospermum halicacabum L. and Koelreuteria elegans (Seem.) A. C. Smith ssp. formosana (Hayata) F. G. Meyer. Diagnostic characters of adults and larvae of Jadera haematoloma are discussed. A review of its bionomics and a bibliography are provided. Initial observations on the populations in southern Taiwan are presented. The species is potentially invasive, and further extension of its range is anticipated in Southeast Asia.


Diagnostic characters of adult
The genus Jadera can be recognized within Serinethinae by the long bucculae which approach base of head posteriorly; in the two other genera of the subfamily, Leptocoris and Boisea, they are short, at most extending to middle of ventral surface of head (Schaefer 1965, Göllner-Scheiding 1979. No native Asian member of Serinethinae shares this character. Jadera haematoloma is a medium-sized species within the genus (9.5-14.5 mm) readily recognized by its colour (Figs 1-5): dorsal ground colour black, head with a narrow red stripe along each eye, and pronotum broadly margined with bright red laterally; abdominal venter black, lateral margins, posterior margin of sternite VI and posterior third of sternite VII broadly red (occasionally more extensively red). Only two other species of the genus have a uniformly black dorsum with contrasting red lateral margins of the pronotum: J. pyrrholoma Stål, 1870 andJ. diaphona Göllner-Scheiding, 1982. The South American J. pyrrholoma differs from J. haematoloma, among others, by its considerably greater size (14.0-18.5 mm) and its uniformly red abdomen. The Central American J. diaphona is similar to J. haematoloma, but it has a uniformly orange abdominal venter. Detailed morphological redescriptions of J. haematoloma and other congeners were provided by Göllner-Scheiding (1979); for distinguishing it from J. diaphona, the subsequent paper by Göllner-Scheiding (1982) also should be consulted. The male genitalia of the species are diagnostic; they were illustrated by Deay (1928), Schaefer (1965Schaefer ( , 1977Schaefer ( , 1978, Chopra (1967), Göllner-Scheiding (1979), and Hoebeke and Wheeler (1982).

Larval instars
A brief description and photo of the fifth instar were provided by Mead (1985).
Intraspecific variability Body size. Both males and females in regions of southcentral USA (Oklahoma) are significantly smaller than those in tropical areas (southern Florida) Loye 1987, Carroll 1988). Macropterous morphs are also significantly larger than brachypters (Carroll et al. 2003b). In Florida, members of populations on a native host plant (Cardiospermum corindum) are slightly greater than those on an introduced host plant (Koelreuteria elegans); the difference is not significant (Carroll et al. 1997(Carroll et al. , 1998(Carroll et al. , 2001. Colour pattern varies only slightly within a population. Caribbean specimens (Bahamas, Cuba) usually have broader vitta along the lateral margin of pronotum, red pattern is present on thoracic pleuron, and the apex of the clypeus also is red (Carroll and  Boyd 1992). Colour variants were observed in Mexico (McLennan et al. 1994); some of them have the abdominal venter extensively red (Carroll and Boyd 1992) but this latter record needs confirmation because of possible confusion with J. diaphona. Two colour variants, 'orange' and 'lemon', were described from laboratory cultures, but they are rare in nature; inheritance of these colour morphs apparently follows a two-locus/two-allele mode, with the two loci interacting epistatically (McLennan et al. 1994).
Wing polymorphism. Usually macropterous (Figs 1-2); approximately 20% of the population in the southern USA is brachypterous (Carroll et al. 2003b). The brachypterous morph was illustrated by Glover (1876), Mead (1985) and Carroll et al. (2003b); such specimens have the corium and membrane shortened  and their flight muscles are lacking (Carroll et al. 2003b). Macropterous morph encompasses three states of flight muscle development: flight muscles developed and retained; flight muscles histolysed; and flight muscles congenitally lacking (Dingle and Winchell 1997, Winchell et al. 2000, Carroll 2003b). As a result, a considerable proportion (about one half in average) of the macropterous individuals is cryptic flightless in some populations.
Frequency of wing morphs is under complex genetic and physiological regulation. Crossing experiments indicate a polygenic inheritance of wing morphs (Dingle and Winchell 1997). Frequency of the flightless (brachypterous and histolyzing macrop- terous) morphs does not differ in populations colonizing native and introduced host plant in Florida; however, congenitally flightless macropterous specimens were more common on the introduced host plant (Carroll et al. 2003b). In populations feeding on native host plants flightless morphs have significantly lower activities of selected enzymes involved in glycolysis, oxidative metabolism and fatty acid oxidation than flyers, but there is no difference in populations feeding on introduced host plants ( Winchell et al. 2000). In laboratory cultures of developing larvae from populations on native the host plant, a significant negative correlation between food level and macroptery ratio was documented: increase in available food results in a decrease in the percentage of macropterous individuals. Treatment with a juvenile hormone analog (methoprene) tends to increase the proportion of brachypterous morphs, but response of the different populations is different (Dingle and Winchell 1997). Rearing at different temperatures does not affect wing-morph frequencies (Dingle and Winchell 1997), but wing development is influenced by photoperiod (Carroll et al. 2003b).
Length of labium. The labium is significantly longer in macropterous specimens (Carroll et al. 2003b). Significant differences in the length of the labium correlating with fruit morphology of the host plant were documented in local populations; see 'Coevolution with host plants'.

Habitat
Jadera haematoloma colonizes various habitats where host plants are available and can be found in city parks and other human-dominated environments (Carroll and Boyd 1992). Adults and larvae usually are found in the canopy and on the trunk of host plants Loye 1987, Carroll 1988), but they also are found on the ground in leaf litter where they feed on fallen seeds (Gagné 1971b, Carroll 1988.

Diet
As all other members of the subfamily Serinethinae, J. haematoloma is an oligophagous seed-predator that develops exclusively on plants of the soapberry family (Sapindaceae s. lato, including the former Hippocastanaceae and Aceraceae). All of its hosts belong to the subfamily Sapindoideae. In contrast to several other congeners, which are restricted to members of the tribe Paullinieae, J. haematoloma also feeds on plants of the subfamilies Sapindeae (Sapindus) and Koelreuteriae (Koelreuteria) ( Table 1) (Schaefer and Mitchell 1983, Carroll and Loye 1987. In the southwestern USA its primary native host plant is the western soapberry (Sapindus saponaria var. drummondii), but it also can be found in large numbers on the littlefruit slipplejack (Serjania brachycarpa). Within its native area it also successfully colonizes several sapindaceous trees introduced to that region, e.g. large aggregations are commonly found on the goldenrain tree (Koelreuteria paniculata) and the Chinese  Carroll and Loye (1987) Oklahoma + Carroll and Loye (1987), Carroll and Boyd (1992) Koelreuteria elegans (Seem.) A.C. Smith Florida + Carroll and Loye (1987), Carroll and Boyd (1992), Carroll et al. (1997Carroll et al. ( , 1998Carroll et al. ( , 2003b Loye (1987, 2012), Aldrich et al. (1990b), Carroll and Boyd (1992), Carroll et al. (1997Carroll et al. ( , 1998Carroll et al. ( , 2003b Carroll and Loye (1987), Carroll and Boyd (1992) rain tree (K. elegans), which are native to eastern Asia and introduced in the southern part of the United States (Carroll and Loye 1987). The heartseed vine (Cardiospermum halicacabum), a widely distributed subtropical climbing plant of uncertain provenance is also present to the southern part of the USA, and is frequently colonized by J. haematoloma in Louisiana and Mississippi where the tree is common, but this plant is apparently not used as a host in southern Oklahoma and northern Texas where it is less common (Carroll and Boyd 1992).
In Florida the bug is common on the native balloon vine (Cardiospermum corindum) and also feeds on the introduced Sapindus mukorossii but avoids a native congener, S. saponaria (Carroll and Loye 2012). After it was introduced to Hawaii, J. haematoloma quickly colonized sapindaceans not occurring in its native area, some of which are native to Hawaii (Sapinus oahuensis); other hosts are introduced (Koelreuteria elegans subsp. formosana and Cardiospermum grandiflorum) Loye 1987, 2012).
Under laboratory conditions, J. haematoloma cultures can be maintained for several generations on seeds of Koelreuteria paniculata and water; seeds of Cardiospermum corindum and C. grandiflorum were also successfully used for such purposes (Aldrich et al. 1990a, b). Sunflower seeds are suitable for rearing at least one generation (Aldrich et al. 1979).
Occasionally the bugs feed on various disabled or freshly dead arthropods (Carroll and Loye 1987). Cannibalism in the field or in captivity also is not rare. Entomophagy mostly involves larvae or reproductive females feeding on freshly moulted larvae or teneral adults, or much smaller larvae Loye 1987, Ribeiro 1989). First instar larvae often cannibalize eggs soon after hatching under laboratory conditions (Ribeiro 1989).

Feeding
It feeds exclusively on the mature and nearly mature seeds of host plants (Carroll and Boyd 1992). Adult females feed more frequently than males (Carroll 1991). On host plants whose fruit is a bladdery capsule with seeds attached to the septum and situated close to the middle (Koelreuteria, Cardiospermum), adults, most frequently females, access the seeds through the pericarp using their rostrum; smaller males and larvae usually feed on fruits that are damaged or dehiscent and, therefore, open (Carroll and Boyd 1992). In Florida, oviposition sometimes occurs into the capsule of C. corindum through small openings of the pericarp, and larvae develop there until at least the fourth instar Loye 1987, Carroll 1988).
The bugs cannot access seeds of Sapindus oahuensis through the fleshy, hardened drupe; therefore, it feeds only on the pericarp (Gagné 1971b). For similar reasons it does not colonize S. saponaria in Florida; however, in captivity it readily feeds on manu-ally opened seeds (Carroll and Loye 1987). In populations feeding on S. saponaria var. drummondii in the southcentral part of the USA, young larvae cannot access to the seed; therefore, they are restricted to feeding on fallen and damaged fruits, but 4th-5th instar larvae and adults can penetrate the drupe with their labium (Carroll and Loye 1987).

Coevolution with host plants
In several cases length of the labium differs significantly between populations on native host plants and nearby populations on introduced host plants. In some populations the change in the average length of the labium can be nearly 25%. The increase or decrease in the length of the labium is consistent with the difference in fruit size and morphology of the native and introduced hosts (Carroll and Boyd 1992). Cross-rearing experiments indicate that the differences are evolved rather than induced by developing on a particular host species (Carroll et al. 1997). Examination of museum specimens also provides evidence for such morphological changes and indicates directional selection and a rapid adaptive evolution in the relatively close past (20-50 years, about 100 generations) following the bug's colonization of host plants introduced into its range. In some populations the allometric change is restricted to the labium; in other populations the shape of the head and the thorax is also slightly different. Laboratory crossing and backcrossing experiments indicate considerable additive genetic variation in length of the labium in populations on both native and introduced host plants; epistatic and dominance variance for the length of the labium was proven (Carroll and Boyd 1992, Carroll and Dingle 1996, Carroll et al. 2001, 2003a, Dingle et al. 2009). The length of the labium and the wing morph frequency also show genetic correlation (Dingle et al. 2009). The rate of microevolution in length of the labium was estimated by Hendry and Kinnison (1999) and Carroll et al. (2001).

Aggregation
Feeding and reproductive adults and larvae form prominent, mixed-instar aggregations on host plants, most commonly on the trunks and on fallen seeds Loye 1987, Ribeiro 1989). The size of the aggregations can reach hundreds or thousands individuals on hosts with large seed crops (Carroll and Salamon 1995), but in warm temperate regions of southcentral USA (Oklahoma) they tend to be larger than those in tropical areas (southern Florida) (Carroll 1988). The tendency of larvae to disperse increases with age, but larvae of every instar rejoin aggregations to moult (Ribeiro 1989).
The structure of aggregations formed by diapausing adults in the canopies of goldenrain trees (Koelreuteria sp.) in Florida was studied by Zych (2010). Aggregations were composed of as many as 300 indivituals, usually on more terminal branches more or less parallel to the ground and always on the undersides of leaves.

Phenology and lifespan
Data are available only from the USA Loye 1987, Carrol 1988). Populations in warm temperate regions of the southcentral part (Oklahoma) and tropical areas (southern Florida) differ greatly in phenology, partly due to thermal seasonality and partly to differences in host-plant phenology.
In Oklahoma (where the population feeds on K. paniculata and S. saponaria var. drummondii with seeds ripening in late July-August and mid-August-September, respectively) reproduction is highly seasonal. Adults and larvae overwinter in dense clusters, mostly on the ground among leaf litter. They leave their refugia around February or March, and overwintered females generally oviposit in March; then the overwintering adults decline in May and June. Adults of the new generation start to emerge in late July; mating and oviposition continue until early October. In October, while food is still available, they enter diapause Loye 1987, Carroll 1988), diapauses which cannot be interrupted (Carroll 1988).
In Florida (where the population feeds on C. corindum with most seeds ripening in May and in November-December) it breeds year round. Adults start to feed and reproduce in late April and May, with bugs (mainly adults of the new generation) entering a starvation diapause in early summer when the seed base is exhausted. A second reproductive period follows from November until January. From January, as food again becomes unavailable, they enter starvation diapause, spending the period in clusters, mostly on herbaceous plants (Carroll and Loye 1987, Carrol 1988, Zych 2010. Diapausing individuals occasionally take nectar from flowering Bidens sp. (Asteraceae) or fluid from petioles of K. elegans (Carroll and Loye 1987).
Individuals are inactive but the moulting of larvae is continuous during diapause in both populations (Carroll 1988). Reproductive adults may live for as long as 2 months (Carroll 1991).

Development
Mean adult development time does not different between sexes (Carroll 1988). In Florida, development time on the native host plant (C. corindum) is longer, age to first reproduction is longer than on the introduced host plant (K. elegans) (Carroll 1988, Carroll et al. 1997. Survivorship of both populations is higher on the 'home' host plant, suggesting the existence of populations adapted to the introduced host (Carroll et al. 1998). Group effects were observed under laboratory conditions: young larvae reared in groups moulted significantly earlier and more synchronously than isolated larvae and mortality was lower (Ribeiro 1989).

Population structure
Data are available only from the USA. Adult sex ratio in populations in humid parts of the southcentral region (Oklahoma) is generally strongly male-biased (ranging from 1:1 to 5:1, average 2.73±0.95 males per female), while in populations in tropical areas (southern Florida), it is close to 1:1 (Carroll 1988, Carroll and Corneli 1995, Carroll and Salamon 1995. The male-bias of the sex ratio in Oklahoma is mainly due to greater female mortality (Carroll 1988).

Mating behaviour
Mating behaviour was studied in detail by Carroll (1988Carroll ( , 1991Carroll ( , 1993. Reproduction takes place within the aggregation. Males search for mates on the ground and in the can-opy, but sit-and-wait searching also occurs, with males remaining stationary until they detect an approaching individual (Carroll 1991). After approaching, the male mounts the female's back, attempts intromission, and, if successful, turns around and attains an endto-end mating position. The pairs generally remain connected for several hours, but duration of copulation is highly variable (from 20 minutes up to 11 days with an average of 20.5±24.5 hours under laboratory conditions). The prolonged copulation is much longer than needed for sperm transfer alone and serves as postinsemination mate guarding (Carroll 1988(Carroll , 1991. Female resistance appears not to have a major influence on the duration copulations (Carroll 1991, Carroll and Corneli 1995, Zych 2012. Average duration of copulations of virgin females is significantly shorter than those of the same females during subsequent copulations. Under laboratory conditions, duration of the copulation tends to be greater in groups where sex ratios are more male biased because of intense male-male competition (Carroll 1991). In male-biased populations there is strong positive sexual selection for male body size. In such populations the ratio of large to small males mating mating is greater. Similarly, mating males are significantly larger than single males. These differences are not observed in unbiased populations. The mating advantage of large males results from their increased locomotion activity (Carroll and Salamon 1995) Oviposition Eggs generally are laid in a hole about 1 cm deep, which the female digs with its fore legs in dry soil close to the host tree. After completing oviposition, the female covers the eggs with soil using its fore legs (Sanderson 1906, Carroll 1988, 1991. In Florida, oviposition also commonly occurs into the capsule of C. corindum through emergence holes made by lepidopteran larvae (Carroll 1988); similar behaviour was observed in Texas (Sanderson 1906). The male interrupts copulation but climbs the back of the female and guards it during oviposition, holding its phallus close to the female's vulva. Ovipositing females are commonly targets of searching males, but the guarding male usually effectively prevents takeover by quickly recopulating (Carroll 1991).
Egg clutches typically are laid at 1-to 2-day intervals for 2-3 weeks; a clutch contains 1-20 eggs (14±4.1 in average) in Oklahoma (Carroll 1988(Carroll , 1991(Carroll , 1993. Maximum lifetime fecundity is estimated to be 400-800 eggs. After oviposition, pairs usually recouple, but males generally guard mates for only one or two ovipositions. In male-biased populations males guard their mates significantly more frequently than those from unbiased populations (Carroll 1993, Carroll andCorneli 1995).
If the male departs it remains sexually active, and often mounts the next available female encountered. Most females also copulate with several males (Carroll 1991(Carroll , 1993. Maximum lifespan of twice-mated females after the last egg is laid is about 6 days (Carroll 1991).
In Oklahoma, females produce significantly more and smaller eggs than those from southern Florida (Carroll et al. 1998). Florida females on the native host plant (C. corindum) produce significantly larger eggs than those on the introduced host plant (K. elegans). However, egg production of females from populations on the native host is the same on native or introduced hosts, whereas females from populations on the introduced host lay significantly fewer eggs per day on the native host but exhibit enhanced fecundity on the introduced host. This suggests the existence of populations adapted to the introduced host (Carroll et al. 1998).

Aposematism, natural enemies and interspecific competitors
The conspicuous aggregations of the red larvae are aposematic. Laboratory experiments with toads (various Bufo spp., Bufonidae) and blue jays (Cyanocitta cristata (Linnaeus, 1758), Corvidae), as well as field observations on Mantidae, showed that after having tasted larvae these predators avoided other larvae. Although adults also are distasteful, their effectiveness alone in causing avoidance is uncertain (Ribeiro 1989, Aldrich et al. 1990a).
In U.S. populations there is little or no predation on the bugs (Aldrich et al. 1990a) and no parasitoids have been observed at any phase of the life cycle (Carroll 1988). Caterpillars of two lycaenid butterflies, Chlorostrymon maesites (Herrich-Schäffer, 1865) and Cyclargus thomasi (Clench, 1941), consume immature seeds of Cardiospermum corindum in southern Florida, and because they cause considerable (occasionally more than 50%) loss in production, they are likely to be significant interspecific competitors of J. haematoloma (Carroll 1988).

Allomones, sequestration, attractants
Pinching the bugs causes them to discharge haemolymph from the rostrum and intersegmentally, and also to emit secretions from the scent glands (Aldrich et al. 1990a). Dorsal abdominal scent glands persist and they are functional in the adult. The volatile compounds of scent gland secretions were analyzed by Aldrich et al. (1979Aldrich et al. ( , 1990b. In addition to (E)-2-hexenal and (E)-2-octenal, compounds often found in Heteroptera, several monoterpene hydrocarbons were identified. The secretion is not sexually different, but compounds from glands of segment IV and V differ: unsaturated carbonyl compounds are produced only by the anterior and monoterpene hydrocarbons only by the posterior gland. Secretions from the ventral abdominal gland of the male again differ from those of the dorsal abdominal glands (Aldrich et al. 1990b). No clear alarmreleasing activity of the compounds on larvae could be proven (Aldrich et al. 1990b).
The haemolymph of J. haematoloma sequesters glycosides. These are not truly cyanogenic; HCN is released from crushed individuals only if they were reared on Cardiospermum grandiflorum and if β-glucosidase is added (Aldrich et al. 1990a). Feces of individuals that develop on Koelreuteria paniculata contains 4-methyl-2(5H)-furanone, which attracts conspecific individuals (Aldrich 1990a).

Stridulation
Stridulation was recorded and documented by Zych et al. (2012). A raised surface at the lateral margin of abdominal tergite I functions as a plectrum; fused abdominal tergites I+II are moved rapidly (15-25 Hz) anteriorly and posteriorly, opposing partly the posterior edge of metanotum, partly the ventral side of the anterior margin of the wing functioning as stridulitrum. Thus a low-frequency and a high-frequency signal, respectively, are produced. Sound producing structures are present and sound is produced in both sexes. The sound is produced as a response to a rapidly approaching conspecific individual, especially if it climbs on the top of the signaller. Apparently sound indicates that the female or male is unreceptive for mating. Interspecific encounters or other threat stimuli do not elicit signals (Zych et al. 2012).

Pest status, control
Large populations around habitations may alarm people (Mead 1985); it was documented in Texas and Oklahoma as a nuisance insect, occasionally entering houses especially in the summer and early autumn (Wheeler 1982 cited by Mead 1985, Reinert et al. 1999. No control measures are necessary. Removing the fallen seeds from under trees and manual collecting and destroying the bugs are usually enough in case they are a nuisance. If chemical control is needed, diazinon EC could be effective (Mead 1985). Biocontrol products containing formulations of Beauveria bassana Vuill. strains show promise for a low-impact and environmentally sound control (Reinert et al. 1999).

Distribution
The distribution range of Jadera haematoloma is determined by the geography of its native and introduced host plant species (Carroll 1988). It is the only species of the genus that enters temperate regions of North America (Fig. 6). It occurs throughout the Gulf Plain, and northward it broadly extends into the area of mixed open forests and temperate grasslands in the western part of the Interior Plains. The northern extent of its range is somewhat indistinct because only a few scattered records are available from the Great Lakes region; these records most likely represent isolated adventitious individuals rather than established populations. The bug does not enter the regions characterized by semi-desert and shrubland vegetation in the Western Mountains and Mexican Plateau, but its range is more or less continuous throughout the subtropical and tropical forests of southern Mexico and Central America.
Although J. haematoloma is common in the peninsular part of Florida, it does not enter the 'panhandle'; therefore, this population is apparently disjunct from that of the southcentral USA (cf. Carroll 1988). No published records are known from the Atlantic Plain prior to 1974, although museum specimens indicate its presence in Virginia as early as 1932 (S.P. Carroll, pers. comm.). In the 1970s it apparently started to gradually expand along the Atlantic Coast towards the northeast (Hoffman and Steiner 2005); the northernmost published localities are in Maryland. It does not occur in higher parts of the Appalachian Highlands.
Records from northern South America are scarce, but most likely its area is bordered in the south by the Northern Andes.
Records from the sub-Amazonian South America, e.g. southern Brazil (Banho et al. 2011), Uruguay, Paraguay (Berg 1892), Argentina (Pennington 1922, Carroll and Dingle 1996, Bressa et al. 2001, Pall and Coscarón 2012 are apparently erroneous and probably at least partly pertain to the superficially similar J. pyrrholoma Stål, 1870(Göllner-Scheiding 1979. The recent record from Buenos Aires is accompanied with a photo (Pall andCoscarón 2012: 1447, fig. 4F). The much broader pronotum and fore wing, the rather distinct dark dots on the pronotum and the different shape of the marginal vitta, and the reddish head of the specimen are sufficient to exclude the possibility that the photographed specimen is J. haematoloma. It apparently represents J. pectoralis Stål, 1862 or J. parapectoralis Göllner-Scheiding, 1979 (opinion confirmed by U. Göllner-Scheiding in litt.). Jadera haematoloma colonizes several islands of the Caribbean. The single record from Antigua is based on an incompletely coloured specimen, the record therefore is uncertain (Barber 1923), but its occurrence on Antigua is likely. It was inadvertently introduced to Hawaii in the 1960s; it was detected on O'ahu Island in August 1968 (Davis 1969) and subsequently recorded on Kauai and Kona Islands (Gagné 1971a, b). It is recorded for the first time from Southeast Asia (Taiwan) in the present paper.

Material and methods
Populations of J. haematoloma were observed at 7 sites in Kaohsiung City, southern Taiwan (J.F. Tsai, Y.X. Hsieh, November 2012-January 2013) and at one site in Tainan City (January 2013, U. Ong). Single individuals were recorded from two additional localities.
Specimens were examined using a SteREO Discovery.V20 microscope with a Pla-nApo S 1.0x FWD 60mm objective. Measurements of larvae were taken using a calibrated Leica stage micrometer (10310345); they were preserved by freezing in order to maintain their shape. Photographs were taken with Nikon D300 and Canon EOS 5D digital cameras equipped with AF-S Nikkor 60mm micro-lens and MPE-65 mm lens, respectively.
Measurements of populations of specimens were compared using non-parametric Wilcoxon-Mann-Whitney two-sample rank-sum test; all presented U and p values were obtained using this test.
Plant names are used following the online database of the International Plant Names Index (www.ipni.org, accessed December 2012).
Voucher specimens of J. haematoloma collected during the present study have been deposited in the following public collections: National Museum of Natural

Distribution and habitat in Taiwan
Single individuals of Jadera haematoloma were observed at the following localities: Kaohsiung Populations of Jadera haematoloma were collected or observed at the following sites (Table 2): Site 1. Kaohsiung City: Cishan, Ci-nan Third Road (N22°49'23", E120°27'44.17"), 30.xi.2012, Y.X. Hsieh, J.F. Tsai (Fig. 21). Around a lychee (Litchi chinensis Sonn., Sapindaceae) orchard. The orchard was bordered by a chain-link fence climbed by several plants, the dominant among them was the heartseed vine (Cardiospermum halicacabum), mixed with some Passiflora foetida L. (Passifloraceae), Mikania micrantha Kunth and Bidens pilosa L. var. radiata Sch.Bip. (both Asteraceae). The fallen leaves of the litchee trees were removed from under the trees and moved to the margin of the orchard under the fence. Several macropterous and brachypterous adults and larvae of all instars were observed to actively walk on and in the leaf litter and feed on C. halicacabum.
Site 3. Kaohsiung City: Ciaotou, corner of Gong-yuan Road and Ciao-chung Street (N22°45'23", E120°18'30"), 30.xi.2012, Y.X. Hsieh, J.F. Tsai. A vegetable garden bordered by a plastic mesh fence fixed to cemented pillars, climbed by C. halicacabum only, with a layer of dead cucurbitacean leaves under the fence (the garden was apparently used for growing melon earlier). Adults and larvae (first to third instars) were observed mainly on the heartseed vine, only a few specimens in the leaf litter under the plant. Several mating pairs and brachypterous individuals were found.
Site 4. Kaoshiung City: Ciaotou, at the junction of Shu-he Road and Tong-shu Road (N22°45'17", E120°18'16"), 30.xi.2012, Y.X. Hsieh, J.F. Tsai. A fallow ground owned by Taiwan Sugar Corporation, with some herbs on the ground, among them C. halicacabum. Several adults, including mating pairs were observed on 16.xi.2012 by Y.X. Hsieh, but the abundance of adults was very low two weeks later: only 7 adults were collected; however, about a hundred larvae were found.
Site 5. Kaohsiung City: Ciaotou (N22°44'27", E120°19'24"), 30.xi.2012, Y.X. Hsieh, J.F. Tsai. A flower farm of Taiwan Sugar Corporation; a public recreation farm with several cultivated vegetables, flowers and trees. Several adults and 2nd-4th instar larvae were collected in an old-growth patch of Koelreuteria elegans subsp. formosana with a thick layer of fallen leaves and seed pods under the trees. Several dozens of adults were collected by Y.H. Hsieh at the same locality on 3.xii.2012. One month later (14.i.2013, Y.X. Hsieh) hundreds of adults (clearly more males than females), including several mating pairs, and larvae of all instars forming aggregations near the base of the trunks were observed. Careful searching on all dates yielded no brachypterous individuals. Site 6. Kaoshiung City: Nanzih (N22°43'51", E120°20'08"), 3.xii.2012, Y.X. Hsieh. A patch of K. elegans subsp. formosana trees (with mature fruits in this season) planted along the street, opposite the building of Kaohsiung High Administrative Court. Adults were actively walking and feeding on the seeds on and among the fallen leaves and fruits under the tree.
Site 8. Tainan City: East District, near Sheng-chen Road (N22°57'55.28", E120°13'23.33"), 13.i.2013, U. Ong. A large fallow ground owned by Taiwan Sugar Corporation, with a large number of C. halicacabum mixed with Bidens pilosa var. radiata. A large number of adults, including several mating pairs, and larvae were observed feeding on C. halicacabum and nectar of B. pilosa. Brachypters were much more abundant than macropters.

Intraspecific variation of adult
Colour. Only slight variation in the colour was observed. In males middle portion of abdominal sternites II-VI was usually black, but several specimens, especially females, had sternites III-VI more or less broadly margined with red posteriorly (Fig. 2) as reported by Göllner-Scheiding (1979).
Body measurements. Adults (n = 187) from various localities in Kaohsiung were measured (Table 3). Body length of males was significantly smaller than females in both the macropterous (U = 16.74, p < 0.001) and brachypterous (U = 6.45, p < 0.001) specimens. Width of pronotum of males was also significantly smaller than that of females in both the macropterous (U = 16.93, p < 0.001) and brachypterous (U = 6.52, p < 0.001) individuals. Humeral width of pronotum of macropterous specimens was significantly larger than that of brachypterous individuals in both males (U = 5.26, p < 0.001) and females (U = 4.30, p < 0.001).
Adults collected on two different host plants (C. halicacabum, K. elegans subsp. formosana) at various sites in Kaohsiung were compared (Table 4). Males collected on C.
halicacabum were slightly smaller on average than those collected on K. elegans subsp. formosana, but neither the difference in total length (U = 0.69, p = 0.488), nor length measured from apex of the clypeus to the apex of abdomen (U = 0.93, p = 0.353) was statistically significant. Females collected on C. halicacabum were slightly larger on average than those collected on K. elegans subsp. formosana (U = 0.93, p = 0.353) in respect to total length, but the relationship was opposite (U = 1.89, p = 0.059) when measuring from apex of the clypeus to the apex of abdomen; these differences are also not statistically significant. The fact that on one of the host plants the mean total lengths of specimens of one sex were greater than those of the opposite sex also suggests that there is no substantial difference in the body size of individuals from the two host plants. Measurements to apex of abdomen might reveal comparative differences in food level, hydration or reproductive condition of females. However, because it is very plastic, it is not as useful a measure for assessing developmental or genetic size differences among adults within or between populations.
Variation in the relative length of labium. The same specimens as in the previous paragraph were examined (Table 3). The apex of the labium in resting position attains at least the posterior margin of sternite II (♂) or the middle of abdominal sternite III (♀), and in extreme cases it approaches the posterior margin of abdominal sternite IV (♂, ♀). Both macropterous and brachypterous females had a relatively longer labium on average than the males. In both sexes macropterous individuals had a relatively longer labium on average than brachypterous individuals of the same sex. The relative length of the labium seems to be slightly longer in both males and females of the populations on C. halicacabum than those on K. elegans subsp. formosana (Table 4), but no conclusion can be drawn for our data and careful testing is needed based on absolute lengths.
Wing polymorphism. At most sites macropterous (Figs 1-2) and brachypterous (3-5) specimens were observed and collected too. Forty-four adults were counted at site 1 on 30.xi.2012; 5 (11.4%) were brachypterous. On one occasion (site 8, 13.i.2013) (Table 2). Slight variability was observed in the development of the fore wing of the brachypterous individuals. The apex of the wing can reach the anterior (Fig. 5) or posterior portion (Figs 3-4) of abdominal sternite VI; in some individuals the membrane is rather broad and subtriangular (Fig. 3), shorter and broadly rounded in others (Fig.  4), and in others is reduced to a narrow band (Fig. 5).

Diagnosis
Larvae of Rhopalidae can readily be recognized using the family keys of Jordan (1951), Leston and Scudder (1956), Herring andAshlock (1971), or Yonke (1991); their unique diagnostic character is the posterior margin of abdominal tergite V deeply emarginate cranially; therefore, the abdominal tergite is longitudinally shortened along midline. Larvae of Jadera haematoloma are more or less similar in size, colour, and shape to those of two Leptocoris species, L. augur (Fabricius, 1781) andL. vicinus (Dallas, 1852), both native and common in Taiwan. The diagnostic characters of the three species are provided in Table 5.
Body colour similar to J. haematoloma but frequently darker red, body without pruniosity.
Mandibular plates strongly narrowed distally, portion of head anteriad of antenniferous tubercles broadly rounded anteriorly. 3 With a single, broadly interrupted series of setae along dorsal margin of eye.
With a single, uninterrupted series of setae along dorsal margin of eye.
Ecdysial suture of head rather U-shaped, with its contralateral branches less diverging.

5
Postocular portion of head of somewhat angulate lateral outline in dorsal view, provided with a single series of setae at each side, without protuberance.
Postocular portion of head of rounded lateral outline in dorsal view, provided with at least two irregular series of setae or irregular pilosity at each side, with a pair of blunt, angular protuberance dorsolaterally.

6
Apex of labial segment I reaching posterior margin of eye.
Apex of labial segment I reaching base of head.
Apex of labial segment I extending to postocular part of head, approaching base of head. 7 All legs uniformly grey to black.
Coxae red to brownish, remaining segments of legs chestnut-coloured to black.
Coxae brownish red, remaining segments of legs black.
Intersegmental suture IV/V slightly curved posteriad at middle.
Intersegmental suture IV/V strongly curved posteriad at middle.

9
Openings of dorsal abdominal scent glands of segments IV and V close to each other.
Openings of dorsal abdominal scent glands of segments IV and V far from each other.
Openings of dorsal abdominal scent glands of segments IV and V rather close to each other.
of eye, visible in dorsal view, antenniferous tubercle distinct, with a tuft of setae; buccula undeveloped (1st-5th instars); eye rounded, prominent, distinctly separated from pronotum by a relatively long postocular margin provided with a single series of setae.
Abdomen composed of 11 visible segments (tergites I and II distinct, sternite I absent); venter distinctly more convex than dorsum. Dorsal abdominal scent glands with two single minute openings situated between tergites IV/V and V/VI, intersegmental suture between tergites IV and V nearly straight, that between tergites V and VI deeply curved anteriad along midline; therefore, tergite V short along midline and gland openings situated close to each other; spiracles II-VIII situated posterolaterally on the respective sternites; trichobothrial formula 0-0-0-3-3-2 (sternites II-VII) in all stages; trichobothria on sternites III and IV situated submedially (rarely 3+4 trichobothria present on sternite IV), trichobothria of sternites V-VII situated on anterior portion of respective sternites, arranged transversely; genital segment distinguishable in 5th instars of both sexes: posterior margin of sternite VIII with slight (4th instar) to deep (5th instar) incision along midline, sternite IX depressed in female, abdominal sternite IX undivided (4th-5th instars), much swollen (5th instar) in male; ring-like segment XI usually exposed. Table 6.

Morphometric changes during larval development
The body is short and oval in newly hatched larvae (Figs 7-8). Abdomen of older first instar larvae is considerably extended because of feeding, the body therefore more elongate; shape of older larvae gradually becoming more similar to that of adult . Second to fifth instar larvae undergo in rather conspicuous changes during each developmental stage (cf. Tables 6-7): larvae of each instar soon after moulting are brighter red, the body appearing smaller because of the shorter abdomen; therefore, the labium is apparently longer in relation to the abdominal sternites. After moulting, the body appears less bright (a dust-like substance on the thorax makes it appear pru-inose) and the abdomen extends so it becomes longer and the labium appears relatively shorter when compared to the body length. These changes are demonstrated in two specimens of second instar larvae in Figs 9-10 (freshly moulted) and 11-12 (older). Because of the extension of the abdomen, a small difference can be observed in the relative length of the mesothoracic wing pads of the fourth and fifth instars.

Host plants and feeding
Several adults and larvae were observed feeding on the ripe fruits of Cardiospermum halicacabum (sites 1-4) and Koelreuteria elegans subsp. formosana (sites 5-7). Table 6. Measurements of larval instars (in mm) and relative lengths of their mesothoracic wing pads collected at site 4. Abbreviations: L1-L5 = 1st-5th larval instars, l = length, w = width. At sites 1, 4, 5, 6 and 7, the thick layer of dead leaves accumulated below the host plant offered an ideal microhabitat for adults and larvae. First and second larval instars were never observed on the plants; they hid among the leaf litter (Fig. 27) and fed mostly on fallen, mature fruits (with brownish pericarp), which were open (Fig. 30). At site 4 most of the first to third instar larvae aggregated within the ripe and open fruits. Third instar and older larvae were more vagile than the first two instars; they walked around and frequently climbed and formed aggregations on the stem of the heartseed vine, on the trunk of K. elegans subsp. formosana (commonly hiding in the crevices), and occasionally on the shadow side of cement pillars around the plants. During feeding, stylets of adults and fourth and fifth instar larvae penetrated deeply into fruit through the pericarp (Figs 22: arrow,28), and reached the seeds. Brown spots appeared on fruits where it was damaged by the feeding of adults or older larvae (Fig. 24). All larval instars accessed seeds by climbing into the fruit through an opening or injury to the pericarp (Fig. 32), or consumed fallen seeds; several adults also fed similarly. Adults and all (including first) larval instars were frequently sucked the fruit stalks (Fig. 26). Adults and at least older larvae commonly drank nectar from flowers of C. halicacabum (Fig. 29) too. Frequent nectar consumption from flowers of Bidens pilosa var. radiata, an asteracean weed, was observed at site 8 (U. Ong, pers. comm.).
As in several other Pentatomomorpha, especially those consuming seeds or feeding from other hard surfaces, the feeding is of the stylet-sheath type (Miles 1972, Cobben 1978. The feeding cones formed by solidified saliva on a seed of C. halicacabum are shown in Fig. 25 (arrow). Feeding cones also were observed in the case of specimens preying on other arthropods ( Fig. 36: arrow), but they never appeared when the specimens fed on the fruit stalks.
Dozens of specimens were kept for several days in captivity on seeds of C. halicacabum, but no cannibalism was observed. Nevertheless, several instances of zoophagy on other rhopalid species were observed; these are discussed below under 'Competitors'.

Condition of populations in Taiwan
At some places the abundance of J. haematoloma was rather high on and around its host plant. At site 1 an hour of searching along the fence (in an area of about 10 × 1 m, cf. Fig. 21) yielded 43 adults and dozens of larvae. A few minutes of searching resulted in hundreds of larvae at site 4; aggregations composed by about 10-30 specimens were observed at this locality.
In the time of the observations (between 6.xi.2012 and 14.i.2013) specimens were still walking, running and feeding actively, and did not show any sign of diapause. Several mating pairs were found (6 at site 1, 2 at site 3, 5 at site 4, 6 at site 7, several at site 8) (Fig. 28). Many of the copulating females were gravid, with greatly enlarged abdomens (Fig. 33); a large number of gravid females was found on K. elegans subsp. formosana during December 2012 and January 2013.
Figures 21-27. Jadera haematoloma on and around its host plant (C. halicacabum) (site 1). 21 Chain-link fence with heartseed vine 22 Adult male feeding on a fruit (arrow: stylet with its basal portion ensheathed in the concavity of labrum) 23 Adult male feeding on a seed 24 fruit damaged by J. haematoloma (arrows: feeding scars) 25 seed damaged by J. haematoloma (arrow: feeding cone) 26 fourth instar larva feeding on the stalk of a fruit 27 larvae walking and feeding on a leaf stem of heartseed vine among leaf litter.
Figures 28-34. Jadera haematoloma on and around its host plant (C. halicacabum). 28 A mating couple, the female (in the right) feeding on fruit of the host plant 29 Adult feeding on a flower of C. halicacabum 30 second instar larva feeding on seed of the host plant 31 Adults feeding on a fourth instar larva of Leptocoris augur (arrow: another fourth instar larva of L. augur) 32 fifth instar larvae in the fruit of C. halicacabum 33 a male guarding a gravid female 34 aggregation of a brachypterous male, a fourth instar larva (arrow) of L. augur and two fourth instar larvae of J. haematoloma.
Several dozen larvae were observed and approximately one hundred were captured. At site 1 (30.ix) only first to third instars were found; careful searching did not yield any older instars. All larval instars were observed at site 4 on the same day, forming aggregations of dozens of larvae at ground level.

Competitors
In all localities J. haematoloma co-occurred with Leptocoris augur (Fabricius, 1781), a taxonomically closely related serinethine species native and rather abundant in Taiwan. In all of the above localities L. augur was estimated to be clearly more abundant than J. haematoloma. Individuals of both species frequently occur within the same aggregation (Fig. 34).
According to our subjective observations, adults and especially larvae of J. haematoloma are more vagile than L. augur. Although first and second instars usually do not walk much, they ran quickly when approached, and they were distinctly quicker than larvae of L. augur of the same instar. The difference in older instars also was evident.
Leptocoris augur was observed to feed on both C. halicacabum and K. elegans subsp. formosana in a manner similar to that described above for J. haematoloma; nectar feeding in L. augur also was observed. In some cases direct interference between individuals of the two species was observed. At site 2, adults and larvae were observed to feed on a freshly moulted adult of L. augur (Fig. 31). A similar phenomenon was observed in a plastic container where the two species were reared together: a freshly moulted adult of L. augur was attacked and consumed by six larvae (representing all larval instars) of J. haematoloma. At site 1 a female of J. haematoloma was observed to approach a female of L. augur, climb its back, and penetrate its labium into its neck (Figs 35-36). They remained in this position for about 15 minutes; after that the individual of L. augur was still alive but it stopped moving and died a few hours later.
At site 5 another related native rhopalid having similar host plants and habits, Leptocoris vicinus (Dallas, 1852) was observed. Adults and all larval instars of L. vicinus were found on the ground, but in smaller numbers than J. haematoloma and L. augur.

Discussion
Our field observations indicate that J. haematoloma has probably already established in southern Taiwan. Because of the large number of adults, high frequency of mating pairs, presence of several gravid females, and most importantly the large numbers of all larval instars, it is apparent that strong, reproducing populations are present in southern Taiwan. The number and the condition of the observed populations suggest that J. haematoloma was not introduced in 2012, but at least one or two years earlier. From the current geographic distribution within Taiwan it seems probable that the species entered through the seaport of Kaohsiung, the largest harbour of the country where most of Taiwan's marine import and export passes.
Apparently the populations in Taiwan have a host range similar to those in North America. Cardiospermum halicacabum and Koelreuteria elegans subsp. formosana were identified as host plants of J. haematoloma in Taiwan; both plants previously were reported as hosts in the continental USA, the Caribbean and Hawaii (Table 1). Frequent nectar consumption from host flowers and Bidens pilosa var. radiata was observed.
Little is known about the bug's phenology in Taiwan. Active, reproducing populations fed on both C. halicacabum and K. elegans subsp. formosana from late November to mid-January. Because of the subtropical and tropical climate of Taiwan, no winter diapause is expected. Because fruits of balloon vine are available year round in Taiwan, and seeds of K. elegans subsp. formosana also are available until late March (Chou and Chen 2010), the population presumably will not enter starvation diapause but remain active, at least on C. halicacabum. Further field observations are needed to confirm or reject this hypothesis.
At least several populations in Taiwan seem more or less male-biased and show variation similar to those in the southern USA. Females are significantly larger than males in both wing morphs and macropterous morphs are significantly larger than brachypters, which is similar to the North American populations (Carroll et al. 2003b). Carroll et al. (1997Carroll et al. ( , 1998Carroll et al. ( , 2001 reported slight differences in the body size of populations feeding on different host plants (Cardiospermum corindum, Koelreuteria elegans) in Florida; no difference in body size was found in populations feeding on C. halicacabum and K. elegans subsp. formosana in Taiwan. 11.4% of the individuals in the population at site 1 observed on 30.xi.2012 were brachypterous; this ratio is about 20% in the southern USA (Carroll et al. 2003b). No inference can be drawn from this apparent difference because our observations are based on a much smaller sample. In some populations on C. halicacabum (site 8, 13.i.2013) the majority of the specimens occasionally is brachypterous. In spite of considerable effort and observations at several localities no brachypterous individuals were observed on K. elegans subsp. formosana. We suggest that wing polymorphism can be explained as a response to host-plant phenology: the percentage of the brachypterous specimens is higher on C. halicacabum, which produces seeds year round, but brachypters are rare or absent on K. elegans subsp. formosana, which is highly seasonal with respect to seed production. A similar negative correlation between food availabil-ity and macroptery ratio was demonstrated in laboratory experiments by Dingle and Winchell (1997).
Jadera haematoloma occurs in the same habitats and uses the same food in the same manner as do Leptocoris augur and L. vicinus, two taxonomically closely related native rhopalid species in Taiwan. Mixed-species aggregations of J. haematoloma and one or both of the native species were commonly observed at several localities. Although no interspecific competition between J. haematoloma and other hemipterans was reported in North America (Carroll 1988), at least scramble competition with the two Leptocoris species is expected if resources are limited. We observed direct interference between individuals of J. haematoloma and L. augur; based on our preliminary observations J. haematoloma is usually more successful in such interferences. Although J. haematoloma also readily feeds on various disabled or freshly dead arthropods in its native area Loye 1987, Ribeiro 1989), feeding on L. augur seems particularly common in Taiwan. Further investigation is needed on the biological interaction between J. haematoloma and the two native rhopalid species and its effect on their populations.
Koelreuteria elegans subsp. formosana originally was found mainly at lower altitudes (Chen 1993), but during the past few decades it became a popular ornamental tree planted extensively in Taiwan along roads in major cities and in the country (Chang 2005, Chen 2006). Cardiospermum halicacabum is common throughout the main island of Taiwan and occurs in various habitats, most frequently along the coast and in wastelands and fallows. An additional 7 genera of Sapindaceae, each with a single species, have been recorded from Taiwan (Chen 1993). Among them, Sapindus mukorossii was listed as a host plant of J. haematoloma in the USA (Carroll and Loye 2012). Allophylus timorensis Blume, an evergreen shrub common in thickets along the coast of southern Taiwan (Chen 1993), is also a likely host plant because adults and larvae of Leptocoris vicinus frequently feed on its seeds in large numbers in the Pratas Islands and the main island of Taiwan (J.F. Tsai, pers. observ.). Colonization of further sapindacean species also seems likely. Because the tropical and subtropical climate of Taiwan is suitable for the species and several of its host plants are readily available, further rapid spread is expected. Jadera haematoloma probably will colonize all of the main island of Taiwan. A specimen from central Taiwan (Chiayi County) apparently indicates that such spread is in progress.
Several of the sapindacean plants that have already been reported as host plants of J. haematoloma in the USA also occur in southeast China (Lo and Chen 1985). Cardiospermum halicacabum is of circumtropical distribution and is common in the eastern, southern and western parts of China; it also occurs in the northern and northeastern parts of the country but is more rare. The genus Koelreuteria is represented by K. paniculata and K. bipinnata in China, both of which have been reported as hosts of J. haematoloma in the USA (Table 1). Sapindus mukorossii and three additional congeners are widely distributed in eastern, southern and western China. Several other members of the rich sapidancean flora of China, comprising 25 genera and 53 species (Lo and Chen 1985), could potentially be consumed by J. haematoloma in case of an eventual invasion. Because the climate of a great part of Southeast Asia and even the neighbouring Palaearctic areas are presumably suitable for J. haematoloma, and various host plants occur in the region, an eventual introduction might also result in establishment of the species in other regions of Southeast Asia.