Morphology and biological notes of Macrosiagon octomaculata (Gerstaecker) triungulinid and a summary of the knowledge on immature stages of the genus (Ripiphoridae)

Species of Macrosiagon Hentz are known to be wasp and bee parasitoids. Here, we present information on Macrosiagon octomaculata (Gerstaecker, 1855), including eggs and triungulinids, found on Paepalanthus polyanthus (Bong.) Sano (Eriocaulaceae) blooming in restinga habitat on Santa Catarina Island, South Brazil. This is the first record of an Eriocaulaceae species playing an important role in the Macrosiagon bionomics and the first accurate record of Ripiphoridae for southern Brazil, Santa Catarina state. In the study area, P. polyanthus attracted an inordinate number of insect species and one Thynnidae wasp was captured with a triungulinid on its wing. Eggs and triungulinids were also found on one plant of P. polyanthus and associated with M. octomaculata. We provide descriptions of the female reproductive organs as well as the morphology of the triungulinid of M. octomaculata. A detailed examination of the triungulinid revealed novelties that complement information from previous publications regarding the Ripiphoridae. We also compare triungulinids of M. octomaculata with those of other species, and provide a table summarizing comparative characters. Finally, we provide a table with current information on the recorded host species and immatures of Macrosiagon.


Biology:
After hatching, individuals of Macrosiagon pass through at least two distinct larval stages (host-seeking or phoretic stage and feeding stage), which exhibit completely different biological and morphological traits (hypermetamorphosis). The host-seeking or phoretic stage is characterized by campodeiform first instar larva, known as a triungulinid, which is involved in the ectoparasite phase. The triungulinids attach themselves to adult hymenopteran hosts (Clausen, 1976) using their mandibles to grasp body hairs (Selander, 1991a;Tomlin & Miller, 1989), or to penetrate soft cuticles and wings . The host then transports them into its nest, where they change to the feeding stage. At this point, they enter the bodies of the host larvae and undergo enormous growth through the extension of their intersegmental membranes to assume their new form, named 'repleta' by Grandi (1936). Once the host larva approaches maturity, the pseudoeruciform Macrosiagon larvae emerge from their hosts and continue feeding externally (Grandi, 1936;Clausen, 1976;Selander, 1991a). It is not clear how many larval instars are involved in the whole cycle, nor if the triungulinid molts to the second instar within or externally to the host's body. In the most detailed biological description of a Macrosiagon species, Grandi (1936) did not mention their ecdysis, and did not observe the entire life cycle. Nevertheless, it is believed that Macrosiagon species most likely have six larval in-stars, like other Ripiphoridae, before molting to pupae. The larvae completely consume their hosts before pupating within the host's nest (Selander, 1991a). (Gerstaecker, 1855) Diagnosis: Adults of Macrosiagon octomaculata  can be distinguished from other species of Macrosiagon by the elevated process at the apex of the pronotal disc (Batelka, 2011b); concolorous pronotum and elytra that are yellow or orange with black marks or black with yellow or orange marks (Rivnay, 1929;Linsley & MacSwain, 1951;Vaurie, 1955); two foremost metatarsomere rounded (Vaurie, 1955); and second metatarsomere thicker and shorter than, or subequal to, the third (Rivnay, 1929;Linsley & MacSwain, 1951).

Study site and associated plant
Field work was conducted in Brazil, Santa Catarina state, Florianópolis city, Restinga da Joaquina (coastal sand dunes environment), between 27°36′S and 48°27′W, from 2007 to 2009. The Restinga da Joaquina comprises a large area of low-lying wetlands situated between two rows of fixed and semi-fixed sand dunes (Scherer & Castellani, 2004). The vegetation consists predominantly of herbs and shrubs, but a considerable area is devoid of plant life .
The genus Paepalanthus sect. Actinocephalus (Körn.) Ruhland. (Poales: Eriocaulaceae: Paepalanthoideae) consists of 28 species restricted to the open areas of Brazil. Paepalanthus polyanthus is the most widespread and morphologically variable species in the genus (Sano, 2001) with a population at the study site that is characterized by short-lived monocarpic individuals (Castellani et al., 2001) that have a unique fractal-like structure (Figs. 5-6).
Most have a central axis from which several scapes arise, each ending in an umbel. Sometimes the axis itself also bears an umbel. Each umbel has around 290 pedicels that terminate in capitula, which in turn consist of about 20 flowers (Castellani & d'Eça-Neves, 2000), each of which is made-up of numerous stamens and carpels. At the study site, plants tended to grow in non-continuous groups in areas subject to regular flooding (Castellani et al., 1996).

Sampling and field observations
Adults of Ripiphoridae were collected on October 2007 (12 th , 21 st , 23 rd and 28 th ) and November 2009. On November 10 th , 2007, eggs and triungulinids were observed and partially collected, and on 14 th one Myzinum sp. (Hymenoptera: Thynnidae: Myzininae) were collected bearing one triungulinid on its right hind wing. All specimens were collected on Paepalanthus polyanthus. Detailed information of the field expeditions and observations are presented in Table 2 in chronological order.
In addition to this, field observations were made on M. octomaculata behavior and on triungulinid behavior. The plant that possessed triungulinids was observed from a distance of two meters between 08:30 and 10:30 am on 19 November 2007. Notes of their behavior and any insects that visited the capitula were taken.

Procedures with material
We made Scanning Electron Micrographs (SEM) of triungulinids at the Universidade Federal de Santa Catarina/UFSC in the Laboratório Central de Microscopia Eletrônica (Central Laboratory of Electronic Microscopy) with a JEOL JSM-6390LV microscope as well as at the Museu de Zoologia da Universidade de São Paulo (MZSP) in the Laboratório de Microscopia Eletrônica with a Carl Zeiss microscope LEO 440. Triungulinids were dehydrated at 50°C for 12 hours and coated in gold (four specimens) or carbon (one specimen). We dehydrated a second batch of three specimens to the point of 100% ethanol. Two of these were coated in gold and one in carbon.
We examined and photographed triungulinids under a Leica DM 5000B optical microscope after treating them first with lactic acid and conducted the same procedure after treating then with KOH and acetic acid. Each specimen was placed on a slide with Hoyer's solution and a coverslip placed over them. We took photographs and measurements using Leica Application Suite 4.1.0 and made drawings from a Camera Lucida attached to a Zeiss Axioscope at 40× magnification. All measurements (observed measurement, mean/ and standard deviation/s.d.) are given in millimeters.

Identifications
Once collected, family and genera were defined by sight by authors. We identified all adult specimens of Ripiphoridae using the key provided by Linsley & MacSwain (1951). These were confirmed by Zachary H. Falin, from the Natural History Museum of University of Kansas. Wasp identifications were made using the key presented by Fernández & Sharkey (2006) and confirmed by Lynn S. Kimsey from the University of California. Classification of Paepalanthus polyanthus within Eriocaulaceae was based on Andrade et al. (2010).
Considering the abundance of Macrosiagon octomaculata observed and collected on P. polyanthus and the absence of any other species in any of the samples, we assumed that all the triungulinids collected belonged to M. octomaculata.

10.xi.2007
More than 300 eggs were found on the base of capitula scape of one umbel . A small amount of eggs (around 120) were collected with a plastic pot (2:30 hours searching).

14.xi.2007
Triungulinids observed very active among capitula of two umbels.  (Poole & Gentili, 1996). Linsley & MacSwain (1951) validated M. octomaculata and M. sayi as different species but argued that the last two and M. pectinata may be involved in a species complex. Batelka (2011b) recognized nine species groups within Macrosiagon; M. octomaculata species group is formed only by M. octomaculata and M. ukereweana Schilder, 1923. Due to this long and controversial discussion about the phenotypic differences between species and the possible existence of a species complex, we opted to present adult morphological characteristics and describe part of the female reproductive organs in order to characterize more accurately the specimens we studied to enable future comparisons.

Female characterization and description of reproductive organs (Figs. 1-3, 7-17)
Six specimens analyzed: Body size varied considerably, from 6.4 to 9.1 mm (from head to apex of elytra).
Color: Head, sternites of thorax and legs black; antennae with the first two antennomeres and the proximal region of third yellowish, remaining antennomeres black; abdomen black (four individuals) or reddish (two individuals); scutellum black. General body color was not identical in any specimens, Figs. 7-10 illustrate four patterns of dorsal color of pronotum and elytra. Specimens varied from having most of the dorsal surface light brown in color, to most of it darkish to black (Figs. 7-10).
Head: frontal surface and lateral to eyes with no setae, but punctated, area below eyes (gena) and vertex covered with bright setae; one longitudinal smooth stripe from middle of head to vertex. Clypeus slightly invaginated and with a triangular depression medially. Labrum densely covered with bright setae. Mandibles covered with bright setae on proximal area of external sides.  (7) to (10) shows lighter to darker specimens.
Thorax: Pronotum punctated and covered with bright setae, denser and longer setae anteriorly, short or absent posteriorly; medially with longitudinal and elevated process abruptly ending near posterior margin of pronotal disc; area after process always black; if pronotum and elytra are in their natural and adjacent position, the scutellum is barely visible; thoracic sternites punctated and together with the legs covered by bright setae. Elytra punctated, if bearing setae, then strongly reduced; umbo smooth; one transverse depression on each elytron, directed from umbo to elytral suture. imally (spiculum gastrale region) and medially invaginated distally 17); sternite IX bearing many short setae (Fig. 14), integument soft, mostly membranous, each side with a more sclerotized area from which originates a sclerotized, sharp filet, extending proximally 17). Tergite VIII relatively sclerotized laterally. Ovipositor bearing many short setae associated with glands (Fig. 15); distally the most sclerotized part (gonopore), bearing setae of different sizes and a fringe of setae on the apical border (Fig. 13).

Egg description (Figs. 35, 73-74)
Eggs were collected with the embryo completely formed, as pharate triungulinids. Although the color observed, dark brown to black, was mostly the embryo colorsome of them were more yellowish; elliptical, about three times as long as wide; ultrastructure (SEM images) of egg surface smooth. Head chaetotaxy: dorsal surface with three pairs of setae (st 1-3) -one pair anteriorly near labrum, one laterally and one longer latero-posteriorly behind stemmata; and numerous microsetae -two/three anteriorly, in each side of median longitudinal line, four medially forming an arcline, two central-posteriorly, and two pairs near the stemmata. Labrum bearing one pair of short setae and, within a slight concavity, one pair of longer setae, each one between two sensilla (red semi-circle on Figs. 52, 54). Stipes anteriorly bearing a long seta contiguous to one microseta. Palpifer with one microseta at middle. Mala with one microseta, one medium seta contiguous to one microseta and a pair of peg-like setae distally. First maxillary palpomere with one pore distally, 3 rd palpomere with one pore near base. Some specimens lack one or two dorsal microsetae (such as in Fig. 34, black arrow).

Description of the free-living first instar larva
Antennae (Figs. 18-20, 22, 32-34, 36, 38, 40-47, 49-50 falciform, covered by head, strongly overlapping each other when closed, inner margin without teeth or toothlike ridges, one pore-like sensillum near condyle on external margin and one on ventral surface (Fig. 52, red arrows), ventral surface irregular; condyle well developed; dorsal surface not examined. Maxillae weakly sclerotized and widened posteriorly; cardo and stipes not fused; car-do narrow and rectangular; stipes longer than wider and dilated basally. Maxillary palpi 3-articulated; first wider than long; second longer than wide, almost twice the length of the first palpomere; third palpomere slightly longer than second, with one lateral digitiform sensillum concealed in a groove on each side (Fig. 54, 57,, prosternal plate the narrowest and longest, metasternal plate the widest. Thoracic chaetotaxy (Figs. 18-19, 22-23, 27-30, 32-33): dorsal surface with numerous microsetae and two pairs of setae near lateral margins of each segment, one pair at middle and one posteriorly. Prosternum bearing one pair of short and stout setae near each lateral margin (Fig. 57, red arrow); meso-and metaventrite laterally each bearing one long seta on each side. Coxal surface with six setae and two pores. Trochanter with four pores, three on a row and one apical, one apical spur and one long seta, slightly longer than femora. Femur bearing several sparse setae. Tibia with three setae on anterior margin, one seta on posterior margin, and one pore on outer face.
Abdomen (Figs. 18-19, 22, 24-25, 32-33, 66-69): length 0.235-0.261,  = 0.246, s.d. = 0.008. Elongate, gradually attenuated from base to apex, segments I to IX wider than long, X subquadrate, slightly wider than long, bearing a bilobate appendage. Lateral spiracles on segments I-VII, the largest located dorsally on segment VIII. tral-posteriorly and one pair of lateral setae posteriorly; sternites II-III same configuration, but with two (against one) pairs of lateral setae posteriorly; sternite VII the same as I, but with one (against two) pair of medial setae central-posteriorly; VIII with one pair of microsetae anteriorly positioned, one pair of medial setae and one pair of long lateral setae; IX with one pair of medial and one pair of lateral setae; X with one pair of lateral setae. In some spec-imens, one or two ventral setae absent, especially on segments I, III and VI (setal socket or alveolus missing).

Biological notes
Macrosiagon octomaculata collected in Restinga da Joaquina, Santa Catarina state, represents the first accu-  ids. Most the triungulinids later hatched in the laboratory. Eggs were laid between the bases of the pedicels of a single umbel (Figs. 73-74, eggs with pharate larva) and, when they were found, the capitula did not contain any active triungulinids. However, on hatching, the triungulinids were very active and occupied the capitula of the two neighboring umbels four days after being discovered (Figs. 75-76). This is the first record of Ripiphoridae oviposition on a species of Eriocaulaceae.
Ripiphoridae beetles are known to be short-lived as adults (Batelka, 2013). This was corroborated in this study as adults were only found during a short period within the duration of field work. Nine specimens were found resting on P. polyanthus (Fig. 77) and one was found dead on a bract, whilst the reared individuals lived for only two days.
During the behavioral observation of the triungulinids, most stayed hidden among the tiny flowers of each capitulum but when the umbel was touched would hastily emerge and head towards the most apical parts of the flowers, namely the apex of the filets and styles. Interestingly, they were somehow able to distinguish     Avulsos Zool., 2020;v.60.special-issue: e202060(s.i.).25 16/26 movements of the umbel caused by the wind from those caused by being touched by hands or two flies landing on nearby umbels.
Although the feeding habits of adult Macrosiagon are mostly unknown, most are presumed to consume nectar, given their elongate proboscis-like galeae (Falin, 2002a(Falin, , 2002bLawrence et al., 2010). However, we observed a female of M. octomaculata, in situ, feeding on the anthers and entire flowers of P. polyanthus, which would result in the ingestion of pollen and nectar (Fig. 78). This individual was first seen touching the capitula with its extroverted distal abdominal segments (Fig. 79), despite this behavior, no eggs were found. Eventually the female was testing by the ovipositor, the suitability of substrate to lay eggs, a common behavior in female beetles (Batelka, pers. comm.). The beetle was collected and kept alive for behavioral observation in the laboratory but, once in captivity, it barely moved and died the following day.
We found a few wasp species on P. polyanthus. Among the few specimens of Myzinum sp. observed, three were collected besides one Scotaena sp. and two males Scotaenini indet. One Myzinum sp. collected feeding on P. polyanthus flowers bores a triungulinid on the ventral side of its right hind wing (80)(81)(82)(83). The triungulinid had its head pointing towards the host's body and its longitudinal axis parallel to the primary wing veins (Fig. 81). This is identical to what was observed by Tomlin & Miller (1989) for the triungulinids of Ripiphorus sp. attached to the wings of Halictidae bees. We postulate that Myzinum sp. is a potential triungulinid host of M. octomaculata because of its compatible size, its preference for P. polyanthus, and for being previously reported as host to M. sayi triungulinids (Rivnay, 1929). But it is noteworthy that this record can be accidental since usually multiple triungulinids of Ripiphorinae were reported loading on their hymenopteran host species at the same time (Batelka, 2011a).
Despite the sharing of several characters between triungulinids of Macrosiagon and Metoecus, subtle differences are nevertheless discernible in the SEM images between M. octomaculata and Me. paradoxus (see Švácha, 1994 for Me. paradoxus). Compared to Macrosiagon octomaculata (in parentheses) the head of Metoecus paradoxus is dorsally irregularly striated transversally, with fine and feebly elevated lines (vs. head striated posteriorly, after ending of longitudinal mid-dorsal line); longest setae on sternum VIII, and tergum IX smaller than in Macrosiagon; abdominal segment X rectangular, longer than wide (vs. subquadrate, slightly wider than long); keels at the ventral margin of median and hind legs absent (vs. present). Nevertheless, as molecular evidence recovers Metoecus internal to Macrosiagon (Batelka et al., 2016), these differences may only concern to species boundaries of the same genus.
The abdominal appendage is a membranous region which can be partially invaginated (Grandi, 1936;Tomlin & Miller, 1989) and has not been described for Macrosiagon (Grandi, 1936), Ripiphorus Batelka & Straka, 2011), Metoecus triungulinids (Chapman, 1870;Švácha, 1994), and Pelecotoma fennica (Švácha, 1994). The term pygopod was used to designate the membranous and retractable terminal region of the abdomen (Grandi, 1936), which Chobaut (1891a, b) referred to as a modification related to host adhesion and for which Tomlin & Miller (1989) confirmed the capacity for adhesion to smooth substrates. Tomlin & Miller (1989) defined the pretarsus as being characterized by leafshaped pulvilli or an inflatable lobe and Pohl & Beutel (2004) termed it the pretarsal bladder, which they defined as a highly developed membranous lobe or pad. However, the most recent study on extant Ripidiinae larvae (Batelka et al., 2019a) applied the term abdominal appendage, which we follow. Batelka (2011a) highlighted the need for a more detailed description of this phoretic character. Based on the illustrations presented in the above-mentioned literature and this study for all Ripiphorinae genera we noticed that the abdominal appendages of Macrosiagon are bilobate and one-segmented, Metoecus seem to be bilobate and one-segmented, and Ripiphorus appear to be undivided and one-segmented. In Ripidiinae the abdominal appendage shape is more variable between genera and in some cases absent. In Blattivorus inquirendus is tubular and one-segmented (see fig. 10 in Silvestri, 1906), in the two Ripidius species are three-segmented and tubular, and absent in the longipedes larva of Ripidiinae gen. et sp. indet. For Pelecotoma the images do not allow comparison. This indicates the potential use of the abdominal appendage (if present) as a taxonomic character to distinguish Ripidiinae from Ripiphorinae.
In the literature, the number of abdominal segments of ripiphorid triungulinids was assumed to be nine (Silvestri, 1906;Böving & Craighead, 1931;Besuchet, 1956;Bologna et al., 2010), nine or 10 (Chapman, 1870) and 10 (Grandi, 1936;Costa et al., 1988;Linsley et al., 1952;Selander, 1991a, b;Švácha, 1994;Lawrence et al., 2010;Batelka et al., 2019a, b;this study). This variation is likely related to the retractable nature of the abdominal appendage and the methods used to study the larvae. Based on our observations, we could correlate methods of specimen preparation with the pleura distension or contraction. Depending on the preparation performed (ethanol, KOH + acetic acid or lactic acid) the form of the abdominal segments varies, especially the distal segments, explaining this discrepancy between the numbers of visible segments. Linsley et al. (1952) was the first to point out the variations between the triungulinids of Macrosiagon and Ripiphorus. Compared to Ripiphorus (in parentheses), the triungulinids of Macrosiagon are more elliptical (vs. subcylindrical), bear a longitudinal mid-dorsal line on the head (vs. absent); possess eight pairs of spiracles on abdominal pleura and three pairs on the thorax (vs. spiracles on mesothorax and abdominal segment VIII); one pair of long setae on sternum VIII and tergum IX, setae are longer than segments VI to X (vs. short setae, setae shorter or about as long as segment X) (apicolateral setae in Batelka & Straka, 2011) and abdominal segment X is subquadrate with a bilobate appendage (conical, not accessed). Also, following our interpretation of the antennomeres (see below), the antennomere II of Macrosiagon triungulinids are one third longer than in Ripiphorus; and the cardo and stipes are not fused (vs. fused).
Distinctions between Macrosiagon and the Ripidiinae genera Blattivorus and Ripidius are remarkable. Among the most noticeable are (Macrosiagon in parenthesis): head length almost equal or shorter than head width (almost equal or greater); sternal thoracic plates absent, or apparently present but small in Blattivorus (present and well developed going beyond coxa); tibia strongly thinned compared to femur, longer than femur and bearing several setae, posterior tibia longer than others in Blattivorus (tibia of subequal thickness compared to femur, almost as long as femur and bearing a couple of setae, equal in length); the chaetotaxy of Ripidiinae is composed of several stout setae dorsally on head, dor-sally and ventrally on thoracic segments and on few abdominal sternites, besides filiform setae (only one pair of stout setae on prosternal plate); one long seta present distally on apical maxillary palpomere, as long as or longer than all palpomeres together (seta absent). However, still more different from Macrosiagon is Pelecotoma. The latter is supposed not to bear phoretic habit and thus, does not bear the common triungulinid forms of first instar Ripiphoridae (Švácha, 1994). The body form of Pelecotoma first instar larva is slender and elongate, some parts are barely sclerotized, antennae reduced comparing to head length, only two stemmata are present and pretarsus bears a sickle-like claw. Further comparisons are present in Table 3. Batelka et al. (2019a) discussed on the discrepancy between the interpretations of the number of antennomeres in Ripiphoridae larvae by different authors - Silvestri (1906), Grandi (1936), Besuchet (1956) and Švácha (1994). Batelka concluded that it is impossible to decide whether the arrangement of larval antennomeres of Ripidius sp. is the same from other described Ripidiinae -Blattivorus inquirendus (Silvestri, 1906) and Ripidius quadriceps (Besuchet, 1956) -or if it has been overlooked in these Ripidiinae, or yet, if it is an artefact of fossilization. Moreover, Batelka pointed out that his  interpretation (3 antennomeres plus 1 terminal bristle-like seta) is corroborated by Švácha (1994) (studying Pelecotoma) and Grandi (1936) (studying Macrosiagon). Batelka et al. (2019a) interpretation also agrees with those of Chobaut (1906), Linsley et al. (1952) and Batelka & Straka (2011). However, based on our SEM images and on observations under optic microscopy after lactic and acetic acid treatments of the triungulinids, our interpretation agrees with the interpretation of Silvestri (1906) and Besuchet (1956). This last interpretation also understands an antennae with 3 antennomeres antennae, but, in another way, considering the terminal bristle like setae as the third whip-like antennomere, and what was previously interpreted as antennomeres II and III is here, considered the antennomere II. These previous interpretations seem to be the result of the constriction on the basal third of the antennomere II that appears as a division between segments under optic microscopy, but is evident in SEM images. Unfortunately, the antennal descriptions of Ripiphorus do not provide SEM images and relevant information on these characters to determine whether they too share the same antennal structure and thus, further investigation is required to confirm this.
Likewise, light microscopy only enables the observation of abdominal spiracles on segment eight, which are located dorsally and are the easiest to observe. However, SEM images provided the first observations of spiracles on all thoracic segments as well as on abdominal segments I to VII. Therefore, the usage of SEM images is an essential tool for the study and characterization of triungulinid morphology.
The phoretic habits of M. octomaculata triungulinids involve their attachment to flying hosts for transport to the host species' nest. Tomlin & Miller (1989) examined the strength of forces involved in their attachment to the host's wing and found at least three forces involved: centripetal acceleration ("the tendency to 'throw' the attached ripiphorid towards wing apex because of rotation around wing hinge"), tangential acceleration ("approximately perpendicular to the wing"), and aerodynamic drag. As a result, structures are present that ensure that these forces are overcome and the triungulinid reaches the host's nest. Most relevant are the sucker-like pretarsi, which they share with Strepsiptera triungulinids; and the soft, sucker-like and moveable abdominal appendage (Selander, 1991a;Pohl & Beutel, 2004, 2008Tomlin & Miller, 1989). However, these structures alone may not be enough to prevent detachment from the Hymenoptera wing during flight.
Other characters that are likely associated with the phoresy include: 1) The antenna. The prolongation of the antennomere after the sensorium is uncommon in Coleoptera larvae (Costa et al., 1988), however, there is relatively scarce information regarding first instar morphology. We suggest that the prolongation of the antennomere II to beyond sensorium in association with the spine adjacent the sensorium and the whip-like third antennomere are morphological features relat-ed to the phoretic habits of first instar Coleoptera. It is presumably involved in anchoring the larvae to the host during its phoretic phase . In families in which the first instar larvae show similar phoretic behavior, namely the Ripiphoridae and Meloidae, the antennae share these similar morphological characters. On the other hand, triungulinids of Strepsiptera lack developed antennae. However, Strepsiptera first instar larvae display a wide range of mechanisms to attach themselves to their hosts, besides having more specialized morphological structures covering the entire body that enhance their anchorage to the host. In the families Xenidae and Stylopidae, which display phoresy, like Ripiphoridae, morphological structure associated to phoretic adhesion are remarkable, as modified labium structures and the well-developed rows of robust spine-like prolongations on body segments (Batelka, pers. comm.;Linsley et al., 1952;Selander, 1991a, b;Švácha, 1994;Pohl, 2002;Pohl & Beutel, 2004;Batelka & Straka, 2011). 2) Ventral setae. The ventral setae are all posteriorly orientated. This may assist in attachment to the host and provide an answer to the question of "Why do triungulins lie parallel to the main vein and always face their heads towards the body of the halictid host" (Tomlin & Miller, 1989). In this position the longitudinal axis of a triungulinid is parallel to the direction of centripetal acceleration, and thus, the ventral setae oppose this force, keeping it attached. 3) Attachment to a vein. As for all the triungulinids found on host wings by Tomlin & Miller, (1989), those that we found were attached to a wing vein. Veins are sclerotized and therefore less susceptible to surface instability than the membranous area. Furthermore, the stability of the triungulinid body is improved by the four long setae on sternum VIII and tergum IX, which may act similarly to the tail of a kite, to enhance balance and reduce drag. 4) The design of the integument surface is important to reduce friction (Luo et al., 2015). For triungulinids, the integument surface resembles other surfaces that are known to increase slip velocity, decreasing drag at the high air speeds experienced by the larvae (Chear & Dol, 2015;Razzak, 2018). Nevertheless, the aerodynamic qualities of the triungulinid integument surface requires testing using experimental models.

ACKNOWLEDGMENTS
We are very grateful to Sônia Casari and Gabriel Biffi for the invitation to submit a manuscript in this special issue of PAZ celebrating the 80 th birthday of Prof. Cleide Costa; and to Jan Batelka and one anonymous reviewer for providing comments and corrections on the manuscript. We also thank Elise L. Galitizk