Hiding among the palms: the remarkable discovery of a new palm bug genus and species (Insecta: Heteroptera: Thaumastocoridae: Xylastodorinae) from remote Norfolk Island; systematics, natural history, palm specialism and biogeography

A total of 27 males, 21 females, 249 nymphs and 11 batches of eggs of the Norfolk Island palm bug were collected during fieldwork in 2019–2020. All except one adult, which was taken in a malaise trap, were collected on infructescences of Rhopalostylis baueri. The first individual palm sampled in February 2021 was the same plant from which the initial chance collection was made in December 2019 (Fig. 2c). This bore four post-flowering low infructescences of different ages (Fig. 2d), three of which bore fruit of various ages and could be examined directly at head height. No adults or nymphs were detected after close visual examination for 15 min by G. B. Monteith and A. Postle. Sheets were subsequently spread on the ground and the infructescences sprayed with aerosol pyrethrin. This yielded 10 adults (e.g. Fig. 3a–c) and 83 nymphs of all instars, demonstrating the crypsis of this xylastodorine species and how difficult the bugs were to detect by eye (e.g. Fig. 3d).

Fruiting rachillae from the same palm were taken for microscopic examination on the same day, yielding several batches of eggs. These were all inserted around the base of the fruit in the narrow fissure between the outer surface of the petiolar scales and the raised edge of the cupule in the rachilla surface where the fruit develops (Fig. 3a–d). Eggs were most often deposited in tight-spaced single rows of up to seven eggs in a linear configuration (e.g. Fig. 4a–f), albeit the occasional presence of single eggs. Eggs were always lodged in a droplet of adhesive secretion that coated the side of the egg against the petiolar scale and embedded approximately the lower third of the egg, with the operculum of the egg free (Fig. 4b, e, f). A clear boundary is visible between the adhesive associated with each egg, implying that the female deposits a separate droplet of adhesive for each egg.

Fig. 4.

Eggs and oviposition sites of Latebracoris norfolcensis sp. nov. (a) Dorsal view of a row of four eggs deposited on surface of a basal perianth lobe of Rhopalostylis baueri fruit and embedded in adhesive secretion. (b) Lateral view of an egg. (c) Scanning electron microscope (SEM) image of closed (left) and ecloded (right) eggs. (d) SEM image of aerolate fine structure of operculum. (e) Pair of eggs laid on R. baueri, with remnant of adhesive secretion. (f). Higher magnification of egg in lateral view, with fine structure of micropylar process. Abbreviations: as, adhesive secretion; mp, micropylar process; o, operculum; O(c), operculum closed; O(o), operculum open, egg apparently ecloded; O(ae), aerolate fine structure of operculum.

Subsequent to these initial observations, rachillae from unsprayed palms were examined under the microscope and living adults were located, almost always resting on the petiolar scales (Fig. 3a–c). Adults were sometimes seen moving between the scales on the green surface of the rachillae but only once on the surface of a fruit.

A developing palm spathe is exposed when a leaf has recently been shed (Fig. 3e). This develops as a leaf axillary structure under a sheathing leaf base. When the leaf is finally shed from the tree, the spathe expands, splits open and the branched inflorescence is exposed, and subsequently flowers and sets fruit. We found a cluster of a dozen adult palm bugs under the new spathe when we pulled this away from the trunk. We postulate that the palm bugs that bred on the old inflorescence have migrated to beneath the new inflorescence, awaiting opening.

The eggs of Latebracoris norfolcensis are shiny, dark brown, urn-shaped, taper a little at top and bottom (Fig. 4a–f) and are slightly flattened at the widest part. The egg has a height of 0.5 mm, the widest diameter 0.34 mm and narrowest diameter 0.25 mm. The operculum (Fig. 4a–d) is flat, circular, 0.19 mm in diameter and bordered by ~30 close-set, subquadrate micropylar processes, and has an areolate texture, with the raised margins of each areole hexagonal in outline (Fig. 4a–f).

All nymphal instars of Latebracoris norfolcensis are greatly flattened and semi-transparent with bright red eyes, with all preserved specimens having a yellow colour, with alternating transverse brown and reddish stripes on the abdomen (Fig. 5b, c). The antennae are yellowish-brown, with darker enbrownment on AII and apex of AIV in later instars.

Fig. 5.

Dorsal view of nymphal stages and dorsal abdominal gland of Latebracoris norfolcensis sp. nov. (a) Detail of dorsal abdominal scent gland of fifth instar; (b) First and second instar; (c) Third to fifth instars. Abbreviations: AT(b), antenniferous tubercle, apically bifurcate; AT(u), antenniferous tubercle, apically undivided; ALM, anterolateral margin of pronotum; CY, clypeus; DAG(Re), dorsal abdominal gland reservoir; MP, mandibular plate; III-V, abdominal terga III-V.

Dorsal views of each instar are shown in Fig. 5. A single, red, dorsal abdominal scent gland is conspicuously present in all instars (Fig. 5a). This consists of a subspherical reservoir within abdominal segment 4 that opens to a single subcircular aperture slightly posterior to the anterior margin of abdominal tergum 5. In all instars the labium reaches to slightly beyond the hind margin of the metacoxae and is four-segmented in later instars.

Following the procedures of the Brooks–Dyar Rule (Floater 1996), both head width across the eyes and overall head–body length provided segregation into five discrete instars (Fig. 6). Size range (L, body length from apex of clypeus to tip of abdomen in millimetres) and the following morphological attributes that also differentiate the five instars are as follows.

Fig. 6.

Size class distribution of Latebracoris norfolcensis sp. nov. segregated into five instars, based on head/body length (top) and head width (bottom) measurements of 131 nymphs taken in a single pyrethrin collection.

Latebracoris norfolcensis Cassis, Monteith & Postle sp. nov., by original designation.

Prefix from the Latin ‘latebra’ meaning hidden, in reference to the cryptic habit and the suffix ‘coris’ meaning bug; noun in apposition.

Latebracoris is recognised by the following combination of characters: body ovoid (Fig. 7a, b), strongly dorsoventrally flattened; ocelli postocular (7A); eyes small, suboval, substylate (Fig. 7a, b, 8a); labium reaching abdominal SIII; acute anterolateral angles of pronotum projecting well beyond posterior margin of head, reaching near anterior margin of eyes; bifurcate antenniferous tubercles present (Fig. 7a, b, 8a); abdominal sterna III and IV long (Fig. 7b); without tibial appendix (Fig. 9e); tarsi with apical spinose comb (Fig. 9e); pretarsus with elongated lamellate pseudopulvilli (Fig. 9f); hemelytra with short median flexion line (Fig. 10a); costal area broad, with corium reaching tip of hemelytron (Fig. 10a); corium with four cells (Fig. 10a); male abdominal SVIII and pygophore asymmetrical, latter right oriented (Fig. 7b, d, 11a); external female genitalia absent (Fig. 12b).

Fig. 7.

Adult morphology of male of Latebracoris norfolcensis sp. nov. (a) Dorsal habitus. (b) Ventral view of body. (c) Ventral view of pretarsus of hind leg. (d) Ventral view of male pygophore, with aedeagus partly inflated. Abbreviations: ALA, anterolateral angle of pronotum; ALM, anterolateral margin of pronotum; AT, antenniferous tubercle; BU, bucculae; CA, costal area of hemelytron; CL, clavus of hemelytron; CO, corium of hemelytra; CY, clypeus; EN, endosoma of aedeagus; LA, labium; ME, hemelytral membrane; MES, mesosternum; MP, mandibular plate; PC, pretarsal claw; PH, phallotheca; PLM, posterolateral margin of pronotum; PP, pseudopulvillus of pretarsus of hind leg; PR, proctiger; PR(S), proctiger spine; PY, pygophore; R + M, radius and medial vein; R + M + CuA, radius + medial + cubitus vein; SII, abdominal sternum II; SIII, abdominal sternum III; SVIII, abdominal SVIII. Scale bars as indicated.

Fig. 8.

Scanning electron micrographs of key head, thoracic and hemelytron characters of male of Latebracoris norfolcensis sp. nov. (a) Head in dorsal view. (b) Head, and pro- and mesothorax in lateral view. (c) Hemelytron in dorsal view. (d) Proximal third of hemelytron in dorsal view. (e) Scutellum and frena in dorsal view. (f) Prosternum in ventral view. (g) Propleuron in lateral view. (h) External efferent system of metathoracic gland in lateral view (pterothoracic legs removed). Abbreviations: ALA, anterolateral angle of pronotum; ALM, anterolateral margin of pronotum; AT, antenniferous tubercle; CA, costal area; CL, clavus; CO, corium; CS, claval suture; CY, clypeus; FR, frena; LA, labium; LAR, labrum; ME, hemelytral membrane; MECX, mesocoxa; MESL, meso-supracoxal lobe; MFL, median flexion line; MGO, metathoracic gland ostiole; MGP, metathoracic gland peritreme; MP, mandibular plate; MTCX, metathoracic coxa; PEP, proepimeron; PES, proepisternum; PLM, posterolateral margin of pronotum; POM, postocular margin of head in dorsal view; PRCX, procoxa; PST, prosternum; SC, scutellum; ST(p), setiferous puncture; V, vertex.

Fig. 9.

Scanning electron micrographs of key pterothoracic, abdominal and leg characters of male of Latebracoris norfolcensis sp. nov. (a) Pterothoracic pleura in lateral view. (b) Abdominal venter in lateral view. (c) Abdominal apex in ventral view. (d) Ventrolateral tergite 8 and spiracle in ventral view. (e) Proleg pretarsus, tarsi and tibial apex in lateral view. (f) Proleg pretarsus in apical view. Abbreviations: MEP, mesopleuron; MTP, metapleuron; PA, parempodium; PC, pretarsal claw; PP, pseudopulvillus; SV–SVI, abdominal sterna V–VI; SII–IV(sp), abdominal sterna II–IV spiracles; SVII(sp), abdominal spiracle SVII; TA1–2, tarsomeres 1 and 2; TC, tibial comb; VLT8, ventrolateral tergite 8; VLT8(sp), ventrolateral tergite 8 spiracle.

Fig. 10.

Hemelytron and hind wing of Latebracoris norfolcensis sp. nov. (a) Hemelytron; (b) Hind wing. Abbreviations: BC1, first basal cell of corium; BC2, second basal cell of corium; CA, costal area; CL, clavus; CM, costal margin; CO, corium; Cu, cubitus vein; M, medial vein; M + Cu, medial + cubital vein; MC, medial cell of corium; ME, hemelytral membrane; MFL, median flexion line; PCu, post cubital vein; RC, radial cell of corium; R + M, radial + medial vein; R + M + CuA, radial + medial + cubital anterior vein; R + Sc, radial + subcostal vein; SCV, subcostal vein; 1A, first anal vein.

Fig. 11.

Male genitalia of Latebracoris norfolcensis sp. nov. (a) Abdominal terminalia, including pygophore, ventral view. (b) Pygophore in lateral view. (c) Proctiger. (d) Aedeagus in lateral view. (e) Apex of endosoma and ductus seminis. Abbreviations: BP, basal process; DS(D), ductus seminis distalis; DS(P), ductus seminis proximalis; DS(SP), ductus seminis spines; EN, endosoma; PB, phallobase; PH, phallotheca; PH(CL), phallothecal cleft; PR, proctiger; PR(S), proctiger spine; PY, pygophore; SG, secondary gonopore; SVII, SVIII, abdominal sternites VII and VII.

Fig. 12.

Female genitalia of Latebracoris norfolcensis sp. nov. (a) Abdomen in dorsal view. (b) Abdominal venter. (c) Bursa copulatrix, oviducts and ovarioles. Abbreviations: BC, bursa copulatrix; CO, common oviduct; LO, lateral oviduct; LO(R), lateral oviduct reservoir; OV, ovariole; SII–SVII, abdominal sterna 2–7; T2–T7, abdominal terga 2–7; VLT8, ventrolateral tergite 8.

Body strongly dorsoventrally flattened, ovoid in shape (Fig. 7a, b). Head: preocular region strongly dissected, apically with tripartite outline (Fig. 7a, b, 8a, b); mandibular plates arcuate, widely separated, subequal in length to clypeus (Fig. 7a, b, 8a); clypeus conical, tapered apically, extending posteriorly to near anterior margins of eyes, lateral margins weakly sinuate, margins posteriorly contiguous with weak grooves, leading to ocelli (Fig. 7a, b, 8a); bifurcate antenniferous processes, extending to near apices of first antennal segment, anterior branch longest (Fig. 7a, b, 8a); postocular region of head with margins strongly convergent posteriorly, sublinear (Fig. 7a, b, 8a); maxillary plates subtriangular, ventral in position, posterior margin obsolete; genae depressed; bucculae elongated, subparallel, almost reaching posterior margin of head, carinate (Fig. 8b). Eyes and ocelli: eyes suboval, substylate; ocelli postocular, widely separated (Fig. 7a, b, 8a). Antennae: 4-segmented; antennal length formula: AIII > AII > AIV = AI; AI extending anteriad of antenniferous tubercle, subcylindrical, thickest segment; AII subcylindrical, little thinner than AI, weakly bowed; AIII and AIV subequal in thickness, thinner than AII; AIII cylindrical; AIV elongated, weakly fusiform (Fig. 7a, b). Labium: 4-segmented; LI shorter than bucculae, broadest segment, not dilated; LII–LIV subequal in length; LII bicompressed, reaching near posterior margin of prosternum; LIII–LIV flattened; LIII reaching near posterior margin of metasternum; LIV reaching near posterior margin of abdominal sternum III (Fig. 7b, 8b). Pronotum: exaggerated, with lateral margins explanate, anterolaterally projected near front of eyes, apically acute, posterolaterally short; posterior margin convex, evenly rounded (Fig. 7a, b, 8b). Scutellum: broadly triangular, transverse, short, mostly flattened (Fig. 7a); frena elongated, subtended apically beneath apex of scutellum (Fig. 8e); mesonotum weakly exposed (Fig. 8e). Hemelytra: clavi prominent, short, weakly expanded distally (Fig. 7a, 8c, 10a); claval commissure reaching abdominal T5 (Fig. 7a); greatly enlarged costal region, basally with short subcostal vein; broadest proximally, apically tapering strongly, reaching tip of forewing, with 3–5 incomplete rows of setiferous tubercles (Fig. 7a, 8c, 10a); coria greatly enlarged, with short median flexion line (Fig. 7a, 8c, 10a); corium with four cells (two basal, medial and radial); hind wing with M + Cu not reaching R + SC; PCu present (Fig. 10b). Thoracic pleura: propleura bipartite, with horizontal ventrally facing region and shorter vertical region, with ventral margin rounded (Fig. 8b); pterothoracic pleura short, weakly convex, with pleural sutures entire (Fig. 9a). Thoracic sterna: prosternum hour-glass shaped, with posterior region widened (Fig. 7b, 8f); shallow labial groove extending to abdominal SIII (Fig. 7b). Metathoracic gland: small ostiole proximal and anteriad to metacoxal cavity, projecting posterolaterally into depressed canal-like peritreme, with minute papillate dermal processes, without evaporative areas (Fig. 8h). Legs: coxae globose, short, widely separated, with procoxae closer to midline than pterothoracic coxae; procoxae open distinctly separated from pterothoracic coxae, paired pterothoracic coxae subcontiguous on each side (Fig. 7b, 8g, h); trochanters prominent, not fused with femora (Fig. 7b); fore and middle legs homomorphic, femora weakly expanded, tibiae cylindrical (Fig. 7b); metafemora weakly elbowed (Fig. 7b); legs with apical tibial combs (Fig. 9e); all tibiae without tibial appendix; tarsi 2-segmented, first tarsomere short, second tarsomere expanded distally, ~5× longer than first tarsomere (Fig. 9e, f); pretarsus with setiform parempodia, subcontiguous; large fleshy pseudopulvilli, subequal in length to claws, attached to unguitractor plate and inner base of claws; claws evenly arcuate (Fig. 9e, f). Abdominal venter: broad (Fig. 7a, b); SII elongate, ~½ length of SIII medially, anvil-shaped, anterior margin straight medially, sublaterally with depressions housing metacoxae; posterior margin excavate (Fig. 7b, 12b); abdominal SIII longest sternite in males, posterior margin straight (Fig. 7a); male SIV–SVI homomorphic, margins straight (Fig. 7b); male SVI posterior margin weakly excavate (Fig. 7b); male SVII short, weakly asymmetrical, posterior margin weakly excavate (Fig. 7b, d); abdominal SVIII strongly asymmetrical, accommodating pygophore on left side (Fig. 7b, d); female abdominal terga 2–7 bilaterally symmetrical, T2 second longest tergum, T7 longest tergum, medially pinched (Fig. 12a); female venter bilaterally symmetrical, abdominal SVII longest sternum, large, apically tapered, with narrowly rounded apex (Fig. 12b); female ventrolateral tergite VIII narrowly visible in ventral view (Fig. 9c, d). Abdominal spiracles: ventral in position, sublateral on abdominal SII–SVII (Fig. 9a–c); female ventrolateral tergite 8 with spiracle (Fig. 9d).

Genitalia as in species description.

Latebracoris is differentiated from Proxylastodoris by the bifurcate v. undivided antenniferous tubercles (cf. Fig. 8a and van Doesburg et al. 2010, fig. 3), apically free v. contiguous mandibular plates (cf. Fig. 8a and van Doesburg et al. 2010, fig. 3), anterolateral angles reaching midpoint v. not reaching eyes (cf. Fig. 8a and van Doesburg et al. 2010, fig. 3), greatly explanate and serrated anterolateral margins v. not expanded and smooth anterolateral margins of pronotum (cf. Fig. 7a and van Doesburg et al. 2010, fig. 5), costal area of hemelytra broad v. narrow (cf. Fig. 7a and 10a with Doesburg et al. 2010, fig. 1 and 2), pygophore short v. squat (cf. Fig. 7b, d, 11a with Fig. 8a and van Doesburg et al. 2010, fig. 33–34), proctiger with v. without spines (cf. Fig. 11b, c with van Doesburg et al. 2010, fig. 33), common oviduct elongated v. short (cf. Fig. 8a and van Doesburg et al. 2010, fig. 3).

The monotypic genera Latebracoris gen. nov. and Thaicoris Kormilev are differentiated by the following characters: ovoid v. elongated body (cf. Fig. 7a and Heiss and Popov 2002, fig. 1), clypeus elongated and greatly surpassing mandibular plates v. subequal in length (cf. Fig. 7a, 8a and Heiss and Popov 2002, fig. 1), antenniferous tubercle reaching ~½ length of mandibular plates v. subequal in length (cf. Fig. 7a, 8a and Heiss and Popov 2002, fig. 1), second antennomere ~2× length of first v. subequal in length (cf. Fig. 7a and Heiss and Popov 2002, fig. 1), labium reaching abdominal SIII v. mesosternum (cf. Fig. 7b and Heiss and Popov 2002, fig. 2), corium unpartitioned v. divided into radial, medial and distal cells (cf. Fig. 10a and Heiss and Popov 2002, fig. 1), pygophore at rest reaching abdominal SVII v. SVI (cf. Fig. 7a and Heiss and Popov 2002, fig. 1) and phallotheca without elongated apical processes v. with processes (cf. Fig. 7a and Heiss and Popov 2002, fig. 2).

The species epithet is based on the restricted distribution of this species on Norfolk Island.

Holotype. NORFOLK ISLAND: ♂: −29.020°S, 167.940°E, 210 m, lower Mt Pitt Rd, 5–6 Feb 2021, Monteith & Postle, ex fruiting palm, 40341, from fruiting inflorescence of Rhopalostylis baueri (In QM, REG. NO. T258332). PARATYPES: 3♂3♀, same data as holotype (in QM); 1♂, same data but 8 Feb 2021, 40350 (in QM); 7♂5♀, −29.017°S, 167.946°E, 165 m, lower Palm Glen, 4 Feb 2021, Monteith & Postle, ex sprayed palm fruit, 40328 (7♂4♀in AM and UNSW; 1♀ in QM); 6♂2♀, −29.0152°S, 167.9387°E, Mt Pitt Reserve, 14 Dec 2019, A. Postle, from fruiting inflorescence of Rhopalostylis baueri, 39721 (5♂2♀ in QM; 1♂ in AM); 1♂1♀, −29.016°S, 167.939°E, 260 m, lower Mt Bates track, 3 Feb 2021, Monteith & Postle, ex sprayed palm fruit, 40322, from fruiting inflorescence of Rhopalostylis baueri (in QM); 5♂3♀, −29.014°S, 167.945°E, 205 m, upper Palm Glen, 6 Feb 2021, Monteith & Postle, spraying palm fruit 40338, from fruiting inflorescence of Rhopalostylis baueri (1♂1♀ in ANIC;1♂1♀ in AMNH; 1♂1♀ in NHM; 2♂ in QM); 3♂4♀, −29.016°S, 167.938°E, 265 m, lower Mt Bates track, 3 Feb 2021, Monteith & Postle, ex sprayed palm fruit. 40323, from fruiting inflorescence of Rhopalostylis baueri (in QM, MNHN); 1♀, −29.0117°S, 167.9536°E, McLachlans Lane, NINP, 115 m, 17.ix.2022, J.M.H. Tweed, beaten from green palm berries from fruiting inflorescence of Rhopalostylis baueri (in QM).

NORFOLK ISLAND: 1♀ 3 nymphs, data as for sample 40323 above (DNA quality specimens in UNSW); 1♀, data as for sample 40341 above (DNA quality specimen in AM); 140 nymphs, data as for sample 40328 above (in QM and AM); 6 nymphs and eggs on palm fruit bracts, data as for sample 40322 above (in QM, AM and UNSW); 25 nymphs, data as for sample 40338 above (in QM); 76 nymphs, data as for sample 40323 above (in QM); eggs on palm fruit bracts, data as for sample 40341 above (in QM); Eggs on palm fruit bracts, −29.016°S, 167.939°E, 260 m, lower Mt Bates track, 5 Feb 2021, Monteith & Postle, 40333 (in QM).

Latebracoris norfolcensis is recognised by the following combination of characters: body dark greyish-brown in life (Fig. 3a–d), pale yellowish-brown in ethanol preserved specimens, with translucent lateral margins of pronotum and costal region of hemelytra (Fig. 7a); body with dense distribution of setiferous punctures (Fig. 7a, b, 8d), each with a single hair-like seta; first and second antennomeres and lateral margins of head with setiferous tubercles (Fig. 7a); pygophore geniculate, tubiform in outline (Fig. 11a, b); proctiger cap-like, with spine (Fig. 11c); parameres absent (Fig. 11c); ductus seminis proximalis narrowly annulated (Fig. 11d); ductus seminis distalis with tile-like texture (Fig. 11d); endosoma with scythe-like sclerotised basal process (Fig. 11d); common oviduct elongated and annulated (Fig. 12c); lateral oviducts short and annulated, with enlarged round reservoirs (Fig. 12c).

Latebracoris norfolcensis is differentiated from Thaicoris sedlaceki and Proxylastodoris kuscheli by non-genitalic differences as provided in the Latebracoris generic remarks section, including the shape of the head and pronotum (Fig. 7a, 8a). Latebracoris norfolcensis is unequivocally separated from P. kuscheli by the male genitalia, including the following character states: proctiger with a spine v. not spinose (cf. Fig. 11b, c and van Doesburg et al. 2010, fig. 33), pygophore elongated v. squat (cf. Fig. 8d, 11b, c and van Doesburg et al. 2010, fig. 33), ductus seminis distalis elongated and tile-like v. short and tubular (cf. Fig. 11c and van Doesburg et al. 2010, fig. 33), and common oviduct elongated v. short (cf. Fig. 12c and van Doesburg et al. 2010, fig. 32).

Heiss and Popov (2002) provided few details of the male genitalia of Thaicoris sedlaceki and there is no information on the female genitalia. Based on the illustration, the pygophore is more elongated overall and has an elongated apical spur, differentiating this from that of L. norfolcensis that is shorter and lacking an elongated apical spur (cf. Fig. 11a–c and Heiss and Popov 2002, fig. 3–5). Also, the proctiger spine in L. norfolcensis is not present in T. sedlaceki.

The palm associations for eight Xylastodorinae species are given in Fig. 15 and Table 2, confirming palm specialism at a transoceanic scale (van Doesburg et al. 2010). All but one of the palm bugs with documented palms are host plant specific, with Discocoris drakei reported from two genera and five palm species, in the subfamilies Arecoideae and Ceroxyloideae. Discocoris is the most species-rich xylastodorinae genus (5 spp.), not restricted to any one neotropical lineage of palms and occurs on five different tribal-level palms lineages, belonging to subfamilies Arecoideae, Calamoideae and Ceroxyloideae. All six palm genera on which Discocoris occur are restricted to the Neotropical region but belong to distantly related palm lineages. Discocoris fernandezi is found on a palm species from the subfamily Calamoideae and tribe Lepidocaryeae that is distantly related from the Arecoideae that harbour the remaining xylastodorines. Most Xylastodorinae species, including the Neotropical and Southwest Pacific species are associated with the subfamily Arecoideae that is the most diverse for the palm family Arecaceae, with 14 tribes and 61 genera. The Neotropical species Xylastodoris luteolus occurs on only one palm species, Roystonea regia that is native to Cuba and widely planted as an ornamental in southern Florida.

Both the Southwest Pacific xylastodorine species (Proxylastodoris gerdae and Latebracoris norfolcensis) occur on palms of the highly speciose tribe Areceae that is widespread in the Eastern Hemisphere from Africa–Madagascar through Asia to the Indo-Pacific and Australia but does not occur in the Neotropics. Proxylastodoris gerdae occurs on Burretiokentia, a genus restricted to New Caledonia that belongs to the subtribe Basselininae that is restricted to Lord Howe, New Caledonia, Vanuatu, Fiji and Solomons. Latebracoris occurs on Rhopalostylis that belongs to the subtribe Rhopalostylidinae comprising two genera ranging from Lord Howe and Norfolk islands to New Zealand, and the Kermadec and Chatham Island archipelagos.

In Fig. 15, the host palms and xylastodorine-affiliated species (aside from D. fernandezi) are mapped to an abridged version of the modern suprageneric phylogeny of the Arecaceae by Baker and Dransfield (2016). This demonstrates that the Xylastodorinae on current knowledge, aside from D. fernandezi, exhibit host plant conservatism at the subfamilial level, for both Neotropical and southwest Pacific xylastodorines. At the tribal level, the associations are restricted to 4 of the 14 Arecoideae tribes and this does not support an hypothesis of preferential host switching, with a caveat of the likelihood of sampling inadequacy.

Fig. 15.

Mapping of palm association data for all known Xylastodorinae species to an abridged palm tribal phylogeny (not including Calamoideae, Nypoideae and Coryphoideae) redrawn from Baker and Dransfield (2016). Oenocarpus species given in full in Table 2.

For a subset of the palm–palm bug associations (Table 2), the sister-group palm tribes Euterpeae and Areceae each harbour a pair of Neotropical and Southwest Pacific xylastodorines respectively (see Fig. 15). This establishes that the palm hosts of the Southwest Pacific xylastodorines are not shared with Neotropical relatives at the tribal level.

Latebracoris norfolcensis is similar to Discocoris spp. and Proxylastodoris kuscheli within the Xylastodorinae in that adults and nymphs are found on reproductive structures of palms rather than on leaves, as in Xylastodoris luteolus (Weissling et al. 2012). Latebracoris norfolcensis and Discocoris drakei (Couturier et al. 2002) have the most detailed biological information within the subfamily, both of which exhibit ecomorphological and lifestyle similarities, including: (1) cryptozoic and dorsoventrally flattened body; (2) cling tightly to plant surface; (3) found in palm cupule depressions (cf. Fig. 3d and 4 with Couturier et al. 2002); and (4) eggs oviposited in the fissure between the cupule rim and the base of the flower or fruit that rests in the cupule.

These two species exhibit differences in life cycle strategy that correspond with palm reproductive phenology. A palm inflorescence is a large structure and development through flowering, fruiting, fruit fall and final shedding can occur over many months. Discocoris drakei are early flower feeders. In this species, adults fly to young inflorescences when flowers first open, where eggs are laid and hatchlings feed only on flowers. Hatchlings vacate the palm host by flight a month later as adults before fruit has developed (Couturier et al. 2002). We could not study the activity of Latebracoris norfolcensis at early flowering stage because no palms were in flower during fieldwork. However, insecticide spraying of infructescences with young (white), medium (green) or old (red) fruit consistently yielded catches of adults of both sexes and all nymphal instars. This indicates that L. norfolcensis has continuous, overlapping life stages and possibly generations that persist for the lifespan of an inflorescence or infructescence. This hypothesis is supported by the fact that the dark greyish-brown adults of L. norfolcensis are camouflaged against the brown perianth scales of the palm host (Fig. 3a–c) that are a feature of mature fruit. In contrast, the much paler Discocoris drakei adults resemble the white flowers and rachillae of early inflorescences of the palm host. This is also supported by the location of eggs of L. norfolcensis that are inserted into the fissure around the base of mature fruit, inferring that nymphs do not feed on flowers. Discocoris drakei inserts eggs into the minute fissure in soft tissues around the base of small unopened flowers, whereas the fissure used by L. norfolcensis is wider and bounded by the tough, mature tissues of the rachilla and petiole scale. This may explain why L. norfolcensis glues eggs in position with a copious deposit of adhesive secretion that may be an adaptation for life on mature infructescences. Egg adhesive was not noted for Discocoris species and in the wider Thaumastocoridae this has only been recorded as a superficial device to anchor eggs to leaf surfaces in a few thaumastocorine species (Kumar 1964; Hill 1988).

In good seasons, mature palms may flower each time a leaf is shed resulting in an individual palm having several inflorescences of different ages, with the youngest at the top of the series. This was conspicuous in the Rhopalostylis baueri palms on Norfolk Island at the time of our fieldwork with some having four infructescences at different developmental stages (e.g. Fig. 2d and see caption). This indicates that the population of L. norfolcensis on a palm may not disperse often to infest new palms but can simply move upwards on the same individual palm and infest progressively younger fruiting structures. Our observation of adults clustering on the outside of a soon-to-open spathe further up the palm from an older inflorescence may be an example of this behaviour (Fig. 2c). Discocoris drakei are frequent fliers since life on an individual palm is brief. We did not observe L. norfolcensis in flight, but wing buzzing and capture in a malaise trap, and no evidence of wing shortening indicate that these are capable of dispersion. Continuous sessile populations on one palm over long periods may not often need flight.

Discocoris drakei feed on flowers whereas the precise feeding site of L. norfolcensis is not clear. When living on fruit-bearing rachillae there are only two options: the rachilla surface or the fruit surface. Adults usually rest on the hard dried petiolar scales and one was seen to apparently be feeding on the rachilla surface from the pose in Fig. 3b. Nymphs were almost invisible but occasionally seen on the rachilla surface. First instars hatching from the eggs would emerge directly onto the rachilla surface, therefore the rachilla surface is the most likely feeding site. Presumably the palm vascular system would be transmitting a rich flow of nutrients to the fruit, at least during the fruit growth period and this supports the rachilla feeding site hypothesis.

In this work we verify the hypothesis of palm specialism for Recent Xylastodorinae (Couturier et al. 1998, 2002; Cassis et al. 1999) with the inclusion of palm host records for the Southwest Pacific species Proxylastodoris kuscheli (van Doesburg et al. 2010) and Latebracoris norfolcensis. In Fig. 15 we show in Discocoris, the only multi-species xylastodorinae genus, that these are not restricted to any one Neotropical lineage of palms. The best-known species of Discocoris, D. drakei, is known from four species of the palm genus Oenocarpus but also an unrelated palm species in the genus Phytelephas from Peru (Schuh 1975) (Table 2). Couturier et al. (2002) postulated that D. drakei is an Oenocarpus specialist but this may be influenced by the association with multiple cultivated foodcrop species of this palm genus. This may have unnaturally expanded the host range of D. drakei and encounters of the species by insect collectors. All the other xylastodorines are host plant-specific (Table 2) but more survey is needed to verify that these are not oligophagous or polyphagous, and to test whether restricted host breadth is an artefact of sampling inadequacy.

The strongly supported sister-species relationship of Latebracoris norfolcensis and Thaicoris sedlaceki (Fig. 13 and 14) predicts that the host of the latter is a palm species. This is a South-East Asian xylastodorine from two widely separated localities: (1) ‘50 km N of Trak’ Thailand (Kormilev 1969) and (2) ‘Ciantar 1000 m, 20 km N of Bandung’ Java (Heiss and Popov 2002). If this is a palm feeder then the elongated body may indicate that the species is a leaf-feeder like the elongated Xylastodoris luteolus, rather than an inflorescence feeder, as in the oval species of Discocoris, Latebracoris and Proxylastodoris. In our phylogenetic analysis, Xylastodoris luteolus is sister to Latebracoris norfolcensis + Thaicoris sedlaceki but this is not supportive of a leaf feeding prediction, given that the relationship is without symmetric resampling support. By contrast, the fossil species Paleodoris lattini is elongated, yet is nested within the oval body shape of Discocoris with resampling support (Fig. 13 and 14) but its position is compromised by a lack of genital characters that could be codified.

Numerous xylastodorine species are known from few specimens, indicating that collection has been serendipitous, as was the initial collection of Latebracoris norfolcensis. Considering the crypsis of xylastodorines on inflorescences and infructescences, high out of reach of orthodox collecting techniques, the probability of discovery of new xylastodorine species on more palm species is high, particularly with the application of the novel xylastodorine collecting methods designed for palm bugs in this work (Fig. 2). This is particularly the case in the Southwest Pacific where xylastodorine discovery has occurred in only the past 10+   and the presence of xylastodorines in South-East Asia is restricted to only two specimens of Thaicoris sedlaceki, both without host plant data. We predict that other palms of the Areceae may serve as hosts for the latter species, given the tribe's diversification in the Indomalayan bioregion and the tribal group upon which the related L. norfolcensis and P. kuscheli feed.

The discovery of two Xylastodorinae endemic species in the Southwest Pacific brings into question the origins and diversification of the subfamily. Van Doesburg et al. (2010) discussed this for Proxylastodoris kuscheli based on temporally disconnected and alternative hypotheses about the origins of the New Caledonia biota in general; viz. the dispersalist Pliocene derivation and post-Palaeocene submersion v. the Gondwanan vicariance hypotheses (see Edgecombe and Giribet 2009 for discussion). Van Doesburg et al. (2010) came to no conclusion about the two models but suggested that present-day xylastodorine distribution may be a result of extinction and the Southwest Pacific occurrence may be a case of relictualism. These authors, by recognising P. kuscheli as a living fossil (because the congener P. gerdae is a Baltic amber fossil), also suggested that Xylastodorinae may have been widespread in the Eastern Hemisphere during the Cenozoic. The discovery of Latebracoris norfolcensis herein not only fulfills that prediction but also specifically adds Norfolk Island to the xylastodorine puzzle, an oceanic island that is both remote and comparatively recent in age, ~3 million years old (Jones and McDougall 1973). Furthermore, our recognition that Latebracoris norfolcensis and Thaicoris sedlaceki are sister-species supports the notion that Recent xylastodorines are currently more widespread (van Doesburg et al. 2010; Schuh and Weirauch 2020) but likely undiscovered in the Eastern Hemisphere, with the latter species being present west of Wallace’s line.

The biogeographic connections of the xylastodorines between the Indo-Australian and Neotropical bioregions is an open question without the benefit of phylogenomics and divergence dating. The absence of xylastodorine-palm associations offers another avenue of enquiry. Van Doesburg et al. (2010) highlighted the stasis of xylastodorine-palm affiliation at a suprageneric level on the opposite sides of the Pacific and speculated that long distance dispersal for both lineages was unlikely. The mapping of all xylastodorine-palm records (Fig. 15) to the palm tribal phylogeny of Baker and Dransfield (2016) reveals no overlap of palm associations at the tribal level. This does however show that the Discocorisdrakei—Oenocarpus palm species and Southwest Pacific xylastodorine associations are shared between the sister palm tribes Euterpeae and Areceae respectively (Fig. 15). Dowe (2010) reported on the first fossil palm pollen records in Australia and New Zealand from the Palaeocene (c. 60 Ma), and that palms were diverse in the Eocene—Palaeocene but with decline in palm diversity and distribution with the onset of aridity in the late Miocene. Palm genera, including Rhopalostylis were also suggested to possibly have been present as an autochthonous element on the Australian continent prior to the Miocene. Pole (1994) argued that the Norfolk Island flora is a result of dispersal, with connections to New Zealand and that the flora shares the palm species Rhopalostylis baueri with the Kermadec Islands, the host of Latebracoris norfolcensis.

The narrow range endemism of Latebracoris norfolcensis within the 460-ha area of Norfolk Island National Park, coupled with the possible impacts of cleared vegetation and invasive species (Director of National Parks 2010, 2018), requires consideration under the IUCN Red List methodology (International Union for Conservation of Nature 2023). The vulnerability of the species may satisfy the area and extent of occupancy criteria. This would likely be assessed as data deficient, requiring future sampling in forest remnants outside the Norfolk Island National Park and more extensive sampling within the reserve.

The Western Hemisphere taxa (Xylastodoris luteolus and Discocoris spp.) within Xylastodorinae notably have nymphs with two abdominal scent glands and eggs without a circle of erect micropylar processes around the perimeter of the operculum. In comparison, the two Southwest Pacific species, Proxylastodoris kuscheli and Latebracoris norfolcensis have nymphs with only one scent gland and eggs with a prominent circle of erect micropylar processes. Although characters of both egg and nymphs have differences between east and west taxa, observations of more taxa are required prior to their inclusion in phylogenetic analyses.

In both our analyses, without (Fig. 13) and with (Fig. 14) fossil taxa, the Tingidae + Thaumastocoridae relationship is supported, with corroborated resampling and synapomorphy support. None of the eight synapomorphies are of the genitalia or venation of the forewing. Given the importance of these character systems in heteropteran systematics in general, we discuss below difficulties in the homologisation, with the prospect that future resolution will verify the sister-group relationships in the Miroidea presented herein.

The asymmetrical abdominal SVIII and pygophore in Thaumastocoridae are unlike any in the Heteroptera, let alone the Miroidea. Likewise for females, with the absence of external female genitalia, enlarged abdominal SVII and absence of a subgenital plate, indicating specialised mating and oviposition behaviour, and undoubtedly in relation to the laminaphile habits. Female thaumastocorids lack an ovipositor for the species investigated and the bursa copulatrix is sac-like without sclerotisation. In comparison, the external genitalia of the Miridae are alike, with a flap-like subgenital plate and a laciniate ovipositor overlapped by the eighth and ninth ventrolateral tergites (Davis 1955; Drake and Slater 1957). Even more perplexing, the sperm storage organs in the three miroid families are highly divergent. In the Miridae the sperm storage organ is a large, sac-like structure attached broadly to the bursa copulatrix and is referred to as the seminal depository (Davis 1955). In the Tingidae there are competing hypotheses, with the so-called paired pseudospermathecae leading into the bursa copulatrix (Drake and Davis 1960; Lis 1999) that are alternatively interpreted as accessory glands (Livingstone and Yacoob 1990; Schuh and Weirauch 2020). In the Thaumastocoridae there are no such organs leading into the bursa copulatrix and there are ball-like structures near the base of each of the lateral oviducts instead that we refer to as reservoirs but are of unknown function.

Forewing homology across the three target families is also vexed and this is true more broadly for the Heteroptera (Schuh and Weirauch 2020) and Hemiptera (Nel et al. 2012). Foremost in the Miroidea is the challenge of homologising the Tingidae forewing components, with specialised areolation and carination, with those of the Miridae and Thaumastocoridae. This is confounded in the Tingidae due to competing hypotheses of forewing venation by Drake and Davis (1960) and Lis (1999). Zhang et al. (2005) proposed using the fossil taxon Ignotingis mirifica Zhang, Golub, Popov & Shcherbakov, 2005 (Ignotingidae) forewing venation, essentially as a ground plan but we found the application to Recent taxa of little use.

Equating the most anterior vein that is posteriad to the wing margin in Miridae and Thaumastocoridae with that of Tingidae is currently at an impasse. These have been named as R + M in Miridae (cf. Schuh and Slater 1995) and subcosta in Tingidae (Lis 1999), under the hypothesis that the anterior forewing margin is the costal vein. This is confounded by the recognition of an area in the Cantacaderinae that is anterior to the costal area, referred to as the stenocostal area by tingid specialists (Drake and Davis 1960; Froeschner 1996), that is hypothesised to be of pre-costal origin (Lis 1999). In comparison, the R + M in Thaumastocoridae is equivalent to that named in the Miridae, with the first recognisable anterior vein posteriad to the costal margin, spatially demarcated by the adjacent median flexion line in both families (cf. Fig. 10a for L. norfolcensis and Schuh and Slater 1995, fig. 10.1 for Miridae).

Although we did not codify corial cells in our phylogenetic analysis, there is promise in homologising the cells in the corium of Thaumastocoridae and Tingidae. The corium has four closed cells (radial, medial and two basal) in Latebracoris norfolcensis (Fig. 10a). This cellular configuration is very similar to that found in all Xylastodorinae, including Discocoris drakei (Cassis, personal observation), Proxylastodoris gerdae (Bechly and Wittmann 2000, fig. 1) and Xylastodoris luteolus (Schuh and Slater 1995, fig. 52.2D). The Dominican fossil species Paleodoris lattini is similar in cell number but differs in cell area and configuration (Poinar and Santiago-Blay 1997, fig. 2). We could not homologise the corial cells of the Xylastodorinae either with the extra subdivisions that are common in the Cantacaderinae forewing (Lis 1999) or the subcostal area in Tingidae because of the apparent absence in the Miridae and Thaumastocoridae. What also appears to be confounding is the elongation of the corium in the Xylastodorinae and Tingidae, resulting in a prolonged subcostal area in lace bugs but no equivalent in the palm bugs. This also has ramifications in homologising the corium in Miridae and Thaumastocorinae with the discoidal area in Tingidae.

The xylastodorine corial cells are however very similar to those found in the fossil taxon Golmonia pater Popov, 1989 (cf. L. norfolcensis, Fig. 10a and Golub and Popov 2008, fig. 1), with the former having a disputed taxonomic alliance (Lis 1999, Thaumastocoridae; Nel et al. 2004, incertae familiae; Golub and Popov 2008, Tingidae). In the illustration of the forewing of this fossil species, there are discernible elongated costal and subcostal areas, suggesting annectant morphology between Xylastodorinae that lack a subcostal area and Tingidae. We note that Golub and Popov (2016) hypothesised a relationship between Thaumastocoridae and G. pater based on the corial cells but did not name either the costal or subcostal areas in their illustration.

In comparison, the forewing of Thaumastocorinae lacks corial subdivision (e.g. Noack et al. 2011). There are also differences between the two Thaumastocoridae subfamilies, with the Xylastodorinae having the R + M fusing with CuA distally, extending the corium to near the apex of the forewing that differentiates this from the much shorter corium in the Thaumastocorinae.

The phylogenetic analysis herein verifies, for both taxon partitions, a monophyletic Thaumastocoridae in alignment with the hypotheses of Schuh and Štys (1991) and Weirauch et al. (2019). In both analyses the family has 99% resampling support and 13 supporting synapomorphies, including non-genitalic and genitalic characters, such as the highly autapomorphic characters of the asymmetrical pygophore and male abdominal SVIII, and the absence of external female genitalia and subgenital plate. These results support retention of the Thaumastocorinae and Xylastodorinae as sister taxa at subfamilial rank, and the raising of the latter by Viana and Carpintero (1981) to family rank is not supported, in alignment with the ranking argumentation of Schuh et al. (2006).

The Xylastodorinae is monophyletic in both phylogenetic analyses, with >90% resampling support and four synapomorphies, including the loss of both parameres. This verifies the concept of xylastodorines as a transoceanic lineage. This also establishes sister-group relationships between the South-East Asian species Thaicoris sedlaceki and the Norfolk Island species Latebracoris norfolcensis, based on three head synapomorphies. This corroborates the synonymy of the Thaicorinae with the Xylastodorinae, in accordance with Schuh and Weirauch (2020). The phylogenetic analysis of the partition with fossil taxa corroborated the monophyly of the Xylastodorinae based on the Recent taxa partition, although this was compromised by the lack of genitalic characters for fossils. Uncertainty exists in the systematic position of the genus Proxylastodoris that is not monophyletic in the fossil taxa partition analysis (although without genital data for the fossil P. gerdae) and in the Recent taxa partition analysis, with P. kuscheli sister to the remaining Xylastodorinae species exemplars. Neither result is supported by resampling support or uncontradicted synapomorphies. Likewise, in both analyses, the clade Xylastodoris luteolus (Thaicoris sedlaceki + Latebracoris norfolcensis) has no resampling support nor synapomorphies.

The Thaumastocorinae was also found to be monophyletic based on 98% resampling support and five synapomorphies, with the reduced tarsi and presence of a tibial appendix correlated, suggesting a diminution of attachment and locomotory function of the tarsi. In this work, we provide the first analysis of ingroup relationships based on two species of the three Australian genera (Baclozygum, Onymocoris and Thaumastocoris) but with monophyly only for Onymocoris. The genus Baclozygum is lacking a modern taxonomic treatment and study of the internal genitalia of any species, hampering diagnosis and determination of its systematic position. The Indian genus Wechina Drake & Slater, 1957 was not analysed given its rarity and lack of pertinent morphological information, although this is undoubtedly a member of the Thaumastocorinae (Drake and Slater 1957; Schuh and Slater 1995; Schuh and Weirauch 2020).

The fossil subfamily Thaumastotinginae is not a member of Thaumastocoridae, despite the reported presence of pretarsal pseudopulvilli and asymmetrical male genitalic characters (Heiss and Golub 2015). Based on our examination of the images of Heiss and Golub (2015, fig. 3) and those of T. areolatus provided by Heiss (pers. comm.), we hypothesise that the pygophore is positioned centrally and not laterally as in Thaumastocoridae, and the left orientation of the pygophore may be a result of preservation. Also, the paired symmetrical parameres, that may be directed caudally as in Tingidae if the pygophoral genital opening is artefactual, are not present in Thaumastocoridae, with the left paramere present at most as in the Thaumastocorinae. The female genitalia of the second specimen of Thaumastotingis areolatus are unlike that of Recent Thaumastocoridae and may be of the Tingidae-type, with an ovipositor and overlapping laterotergites (see Heiss and Golub 2015, fig. 6). However, given the state of preservation and our observation of only images of the fossil, we refrained from codifying this as such.

Heiss and Golub (2015) recognised the annectant morphology of T. areolatus, including Tingidae characters (e.g. areolation and cells in the corium) that are not present in Recent or fossil Thaumastocoridae. In addition, there are characters that are not found in both Thaumastocoridae and Tingidae, such as the three-segmented tarsi. However we found character support for the transfer of Thaumastotinginae to Tingidae in our phylogenetic analysis (Fig. 14) but there was only modest resampling support for inclusion in the latter family. Also, we did not have the opportunity to observe the specimens directly. For this reason, we regard Thaumastotinginae as incertae familiae and not a member of Thaumastocoridae.

We note that the discovery of critical fossil Tingidae and allies over the past 35   (see Schuh and Weirauch 2020 for documentation) has highlighted a rich ancient morphology (Golub and Popov 2016) that is not apparent in Recent Tingidae, including that found in fossil taxa such as the Ignotingidae (Zhang et al. 2005). As a result, tingidologists of both Recent (e.g. Lis 1999) and fossil (Golub and Popov 2016) taxa have recognised these within a superfamily Tingoidea that is not recognised by phylogeneticists (e.g. Weirauch et al. 2019; Schuh and Weirauch 2020). We recognise the benefits of integrating the classification of fossil and Recent Hemiptera, as in Szwedo (2016; including the recognition of the superfamily Tingoidea). However, a nexus between both is confounded by character inapplicability, either as preservation inadequacy or characters that are unique or cannot be codified with confidence.

In this work, we have sought to assess not only the position of Latebracoris norfolcensis within the Thaumastocoridae but also test the phylogenetic relationships within the superfamily Miroidea. The classification of Miroidea: Miridae, Tingidae and Thaumastocoridae, with the latter comprising the Thaumastocorinae and Xylastodorinae, is supported.