Evolution of glandular structures on the scape of males in the genus Aphelinus Dalman (Hymenoptera, Aphelinidae)

The pores and associated glands on male antennae in species of Hymenoptera are involved in mate recognition and are diverse and widespread among taxa. However, nothing has been published about these structures in species of Aphelinus (Chalcidoidea: Aphelinidae), a genus of parasitoid wasps with a long history in biological control. Images from scanning electron microscopy (SEM) and transmission electron microscopy (TEM) of Aphelinus varipes revealed pores on the ventral side of the male scape that were connected to glands. A survey of the scapes of male antennae in 16 species in six species complexes of Aphelinus, as well as two outgroup species, Aphytis melinus and Centrodora sp., showed that pores were present in all except Centrodora sp. The pores varied in several characters: the shape of the structures that carried them, pore size, elevation of the cuticle surrounding the structures, the extent of a carina delimiting the area around the structures, and the number and position of pores. The shape of the pore-bearing structures, the elevation of cuticle around these structures, and the extent of the carina around them map well onto a molecular phylogeny of these Aphelinus species. Combinations of pore characters are diagnostic of species complexes, and in some cases, species of Aphelinus.


Introduction
Successful mating depends on discriminating between individuals of the same versus different species, as well as between potential mates that will yield progeny with high versus low fitness. Communication with pheromones that attract conspecifics and provide cues for their recognition is often a key component in the quest for mates by insects in general and Hymenoptera in particular (for review, see Ayasse et al. 2001). However, more or less elaborate courtship behavior may also be needed to secure mates (for reviews, see Brown 1999, Spieth 1974, Yuval 2006. In parasitic Hymenoptera, courting males often move their antennae in specific patterns and touch females directly with them (for review, see Gordh and DeBach 1978). Such behavior, usually called antennation, often occurs after a male has approached a female, she has stopped walking, and he has climbed on her back so that their bodies are parallel with his head above hers. Starting at this point, males perform various behaviors that may include head-bobbing, antennal waving and stroking, as well as other behaviors, like leg sweeps. Although male antennation during courtship was described at least as early as 1910 (Girault and Sanders 1910), Barrass (1960) in research on Nasonia vitripennis (Walker) (Hymenoptera: Pteromalidae) provided perhaps the first quantitative analysis of courtship behavior, which included quantitation of male antennation. In one of a series of papers on reproductive isolation among closely-related species of Aphytis Howard (Hymenoptera: Aphelinidae), Rao and Debach (1969) described antennation behavior in Aphelinidae. Gordh and DeBach (1978) later used a quantitative analysis of differences in courtship behavior, including antennation, to distinguish among cryptic species of Aphytis in the lingnanensis species complex.
The well-documented role of antennation in courtship triggered investigations into the morphology of antennae. Based on scanning electron microscopy (SEM) of antennae, Goodpasture (1975) suggested that the numerous pores on the scapes of males of some species of Monodontomerus Westwood (Hymenoptera: Torymidae) might secrete a pheromone. Dahms (1984a) found pores on the ventral surface of male scapes in Melittobia australica Girault (Hymenoptera: Eulophidae), for which he had described antennation a decade earlier (Dahms 1973). He proposed that these pores released a pheromone involved in mate recognition, rather than being sensory organs, as had been previously conjectured. This discovery prompted more studies on morphology of male antennae in parasitic Hymenoptera, using SEM and transmission electron microscopy (TEM). Bin et al. (1989) found three types of glands in antennae of Trissolcus basalis (Wollaston) (Hymenoptera: Platygastridae), and they suggested roles in courtship and mate recognition. On the ventral side of the fourth antennomere of Amitus spiniferus (Brèthes) (Hymenoptera: Platygastridae), Isidoro and Bin (1995) found numerous pores on an elevated plate, and these pores were attached internally to a complex of glandular cells. They suggested that this structure released and spread a mate-recognition pheromone during courtship, and Isidoro et al. (1996) suggested that such structures be called release-and-spread structures (RSS). Research has continued to delineate the taxonomic distribution and morphological diversity of these structures , Romani et al. 2010 (Guerrieri et al. 2001). They have also been found in parasitoids in other superfamilies: Trissolcus basalis (Wollaston) (Hymenoptera: Platygastridae) (Bin and Vinson 1986), a variety of species of Cynipidae and Figitidae (Cynipoidea) , Trichopria drosophilae (Perkins) (Hymenoptera: Diapriidae) (Sacchetti et al. 1999), and Pimpla turionellae (Linnaeus) (Hymenoptera: Ichneumonidae) . In the last two species, the RSS and glands were shown to be necessary for successful courtship and copulation , Romani et al. 2008.
Observations on courtship behavior in several species of Aphelinus Dalman (Hymenoptera: Aphelinidae) showed that males antennate during courtship (Kazmer et al. 1996, Rhoades 2015, but nothing has been published on pores and glands on the antennae of Aphelinus. However, antennal pores and glands have been reported on the antennal clubs of males of Aphytis melinus DeBach , which is closely related to Aphelinus (Kim and Heraty 2012). Furthermore, observations of slide-mounted Aphelinus specimens using differential interference contrast microscopy (DICM) showed that RSS are widespread in the genus, and that the arrangement and shape of RSS differed among species complexes. Here we report results of an investigation of the glandular ultrastructure underlying RSS in scapes of male Aphelinus varipes (Förster) using TEM, and a survey of variation in the RSS in 16 species in six species complexes of Aphelinus using SEM, DICM, and macrophotography that documents their diversity and shows that they are taxonomically and phylogenetically informative.

Materials and methods
We follow the species complex classification of Hayat (1998), with one modification: we treat the daucicola species complex as distinct from the mali complex (Hopper et al. 2012).
With TEM, we studied the structure of the cells underlying the pores in the male scape of A. varipes. With SEM, we studied male scapes of two to three specimens each from nine species of Aphelinus in six species complexes, along with species in two other genera of Aphelininae: Aphytis melinus Debach and an undetermined species of Centrodora Förster (Hymenoptera: Aphelinidae) ( Table 1). With DICM, we studied male scapes of specimens from an additional seven species of Aphelinus in two species complexes (Table  1). Prior to preparation for imaging, the material was stored in molecular-grade ethanol.
For TEM, specimens were immersed in a solution of glutaraldehyde (2.5 ml/ 25 ml solution) and paraformaldehyde (1 g/25 ml solution) in 0.1M cacodylate buffer +5% sucrose, pH 7.2-7.3. The scape was removed from the rest of the antenna and cooled at 4 °C for 3 h. The specimens were placed in 0.1M cacodylate buffer +5% sucrose, pH 7.2-7.3, overnight at 4 °C, then the specimens were post-fixed in 1% OsO4 (osmium tetroxide) for 1 h at 4 °C and rinsed in the previous buffer. Dehydration in a graded ethanol series from 60% to 99% was followed by embedding in Epon-Araldite with propylene oxide as bridging solvent. Thin sections were made with a diamond knife DiATOME ultra 45° (DiATOME AG, Biel, Switzerland) on a LKB Bromma ultramicrotome (LKB®, Sweden), and mounted on formvar-coated, 50-mesh grids. The sections were stained with uranyl acetate (20 min, room temperature) and lead citrate (5 min, room temperature). Finally, the sections were investigated with a TEM Philips EM 208 (Thermo Fischer Scientific, Hillsboro, Oregon, USA). Digital images with 1376×1032 pixels, 8 bit, uncompressed greyscale in TIFF files were obtained using a high-resolution digital camera MegaViewIII (SIS) connected to the TEM.
For SEM, specimens were critical-point-dried (CPD) using a Tousimis Samdri-790 (Tousimis Research Corporation, Rockville, Maryland, USA), following the manufacturer's protocol. After CPD, two to three males from each accession (Table 1) were mounted onto standard 12.7mm Ted Pella pin stubs using black carbon tape. Specimens were then gold sputter-coated using a Technics Hummer I (Anatech Ltd, Battle Creek, Michigan, USA), following the manufacturer's protocol. Gold was sputtered on specimens for a total of five minutes, in one-minute intervals separated by one minute. All antennomeres on both the left and the right antennae were examined on each specimen. SEM images were acquired using a Tescan Vega 3 microscope (Tescan USA, Warrendale, Pennsylvania, USA) with secondary electron-emission in high vacuum and beam-acceleration voltages ranging from 15 kV to 30 kV. The beam intensity was either 3 or 5 (range is 1-20), and the scan speed was either 6 or 7 (corresponding to 32 or 100 msec/pixel). For DICM, specimens were slide-mounted in Canada balsam using a protocol modified from Noyes (1982) and observed with an Olympus BH2 compound microscope equipped with differential interference contrast and planapochromat objectives. DICM images were captured with a Jenoptik ProGres CT5 digital camera using Image Pro Plus 7.0. Macrophotographic (MP) images were taken of card-mounted specimens using a Macropod Pro setup (https://macroscopicsolutions.com/) consisting of Mitutoyo long-working distance 50× and 100× planapochromat objectives, a Canon 200 mm prime lens, and a Canon EOS 5D Mark III camera. DICM and MP images were focus-stacked using Zerene Stacker 1.04 and final preparation of images from SEM, DICM, and MP was done in Adobe Lightroom. Sample size for specimens examined by DICM and MP varied from a few specimens to ten specimens, or more in cases in which abundant material was available.
Voucher specimens of material examined have been deposited in the Texas A&M University Insect Collection with Voucher #733. Voucher numbers in Table 1 refer to specimens deposited in the Beneficial Insect Introduction Research Laboratory (USDA/ ARS), as voucher specimens of the original collections from which colonies were founded.
Five morphological traits of the RSS were recorded for 14 species of Aphelinus, as well as for Aphytis melinus ( Table 2). Eight of these 14 species of Aphelinus were studied using SEM, the remainder were studied using DICM. The data were entered into mx, an open-source, web-based database management system (code and documentation available at http://mx.phenomix.org), and the results were also exported as a Nexus file. Diagnostics (Consistency Index and Retention Index) were calculated for each trait using PAUP* with the traits mapped onto a molecular phylogeny from Heraty et al. (2007), modified with additional unpublished data. PAUP* (4.a build 159: Swofford 2002) and the Trace Character History algorithm using the Parsimony Ancestral States in Mesquite (Maddison and Maddison 2018) provided the most parsimonious reconstructions of character state changes on the phylogeny.

Results
The male scape in A. varipes is characterized by the presence of cuticular modifications located on the ventral side, defining a specialised area (Fig. 1A). The ventral margin presents three subconical elevated areas with truncated and slightly concave tops that carry single pores and which are surrounded by a depressed region delimited by a carina ( Fig 1B). In some specimens, a secretion oozing from the pore can be observed accumulating on top of the flattened area (Fig. 1C). Internal observations revealed the presence of a large cell that occupies about half of the whole internal scape volume behind each pore (Fig. 1D). This cell has a large nucleus located in the basal region, a cluster of electron-lucid vesicles located in the middle region, and a straight evacuating duct connected with the external pore (Figs 1D, E). The antennal muscles located within the scape are not connected with the glands, and they are located in the remaining half of the internal scape volume together with the antennal nerve (Fig. 1D, F). Sections taken in a different plane show the presence of a well-developed end apparatus located in the middle region of the cell and characterized by an aporous structure surrounded by microvilli (Fig. 1F, G). Large clumps of electron-lucid vesicles surround the end apparatus (Fig. 1G). The cytological features of the cell are those typical of intensively active cells, i.e. the presence of a large nucleus and cytoplasm showing the presence of abundant mitochondria and ribosomes (Fig. 1H). The cuticular evacuating duct is connected to the end apparatus and is produced by an associated duct cell with very reduced cytoplasm and a small nucleus (Fig. 1H). The secretion gathered at the level of the end apparatus flows through the evacuating duct and reaches the external pore where it is released. The morphology of these release-and-spread structures varies among Aphelinus species and Aphytis melinus (Figs 2, 3; Table 2). The pores in the asychis complex are sessile, i.e. not raised above surface of surrounding cuticle, and are on a convex region not surrounded by a carina (Fig. 2A, B). In the other species complexes of Aphelinus, the pores are in a depressed region of the cuticle, but individual pores are on elevated structures (Figs 2C, D, 3A-E), which vary from smoothly rounded (Fig. 2C), to crenulated ridges (Fig. 2D), or truncated and volcano-like (Fig. 3A-F). In the two outgroup species, pores are present and sessile in Aphytis melinus (Fig. 2E), and absent in Centrodora sp. (Fig. 2F).
Mapping six characters of the release-and-spread structures (Table 2) onto a molecular phylogeny of these Aphelinus species and Aphytis melinus revealed two broad phylogenetic patterns for different sets of characters; one set had shared states for the mali and varipes complexes and different states in the other species complexes, and the other set had different states in all the species complexes (Fig. 5). For the shape of the structure bearing each pore (Fig. 5A) and the elevation of the region around the pores (Fig. 5C), consistency and retention indexes are 1.0, indicating no homoplasy and potentially useful phylogenetic signal. States for these characters are the same in the mali and varipes complexes, as well as Aphelinus daucicola Kurdjumov, the shape being raised, conical, truncated, and flat or slightly concave on the top with the region around these structures being depressed. The shape varies among the other species, with Aphelinus abdominalis (Dalman) having raised structures that are rounded on top, and Aphelinus perpallidus Gahan having pores on a crenulate ridge, in both cases surrounded by a depressed region. Species in the asychis complex, as well as Aphytis melinus have sessile pores, suggesting that sessile pores might be ancestral. However, the surrounding region is elevated in the asychis complex, but neither depressed nor elevated in Aphytis melinus. Pore size (Fig. 5B) is also the same in the mali and varipes complexes, as well as A. daucicola, and the same in the asychis complex, A. abdominalis and Aphytis melinus, but the consistency index is 0.5 and the retention index is 0.67 because of parallel increase in pore size in A. perpallidus and in the lineage consisting of A. daucicola through Aphelinus certus Yasnosh. For the carina around the pores (Fig. 5D), the consistency and retention indexes are 1.0, indicating no homoplasy and strong phylogenetic signal. However, the state of the carina differs between the varipes complex and mali complex. Species in the varipes complex have a carina around the proximal end of the pores, whereas species in the mali complex, as well as A. daucicola, A. perpallidus, and Aphytis melinus have a carina completely surrounding the pores, suggesting that a complete carina is ancestral in Aphelinus, although it is absent in the asychis complex. Both the number (Fig. 5E) and location of pores (Fig. 5F) have consistency and retention indexes less than 1 (0.6 and 0.5 for pore number; 0.4 and 0.25 for pore location), indicating a fair amount of Figure 1. Structure of the male scape in Aphelinus varipes. A-C Scanning electron microscope images showing in A the modified male scape (Sc) with the presence of a ventral carina on which three specialized structures (arrowheads) are observed B detail of the carina revealing three elevated areas (Ea) C close up of the previous in which a single apical pore (Po) per Ea is clearly visible, as well as secretion (Scr) oozing from the pore itself. D-H Transmission electron microscope images showing internal ultrastructural features of the male scape D cross section of the scape taken through one of the elevated areas, showing the secretory cell (Scl) that occupies about half of the scape internal volume; Scl has a large nucleus (Nu) located basally, a central cluster of electron-lucid secretory vesicles (Sv) and a straight evacuating duct (Ed) connected with the external pore (Po); the rest of the scape volume is occupied by muscles (Mu) and the antennal nerve (An) E close-up view of the previous image, showing the cuticular evacuating duct (Ed) running straight towards the external pore F cross section of the scape taken in a different view, showing the large secretory cell (Scl) with a centrally positioned end apparatus (Ea) surrounded by numerous secretory vesicles (Sv); the evacuating duct (Ed), muscles (Mu), and antennal nerve (An) can be seen G detail of the end apparatus (Ea), which appears perforated and surrounded by microvilli (Mv); secretory vesicles (Sv) are above H detail of the duct cell (Dc) characterised by a very reduced cytoplasm and a small nucleus (Nu); the duct cell is surrounded by the cytoplasm of the secretory cell, that reveals the presence of ribosomes (Ri) and mitochondria (Mt); the evacuating duct (Ed) is visible in cross section. Scale bars: 50 µm (A); 10 µm (B, D, F); 5 µm (C); 2.5 µm (E); 1 µm (G); 2 µm (H).    homoplasy for these characters. The varipes complex is consistent in pore number with all species having three pores. Nine species in the varipes and mali complexes have the most proximal pore at the scape midpoint. However, Aphelinus atriplicis Kurdjumov, Aphelinus mali (Haldeman), and A. daucicola, have the most proximal pore proximal to the scape midpoint. For A. mali and A. daucicola this may have arisen because they both have five pores, which might force the most proximal pore further towards the head. Species in the asychis complex have four or five pores, but like most species in the varipes and mali complexes, the most proximal pore is at the scape midpoint. However, species in the asychis complex have small, sessile pores that are close together, which may allow them to fit more distally than found in A. mali and A. daucicola. On the other hand, both A. perpallidus and Aphytis melinus have only two pores but have the most proximal pore proximal to the scape midpoint, suggesting that fit to the available length may be not be important for pore placement.

Discussion
There is considerable variation among Aphelinus species in the characters of the releaseand-spread structures on male scapes. Although three characters show some homoplasy with CI values of 0.50 to 0.75 (pore number, size, and location), three characters have CI = 1.0 (supporting structure, elevation of cuticle, and extent of carina). Combinations of characters are diagnostic for species complexes of Aphelinus (Table 2), and some combinations are diagnostic for species within complexes. Furthermore, the morphologically distinctive aspects of the male scape in these Aphelinus species mapped reasonably well onto the molecular phylogeny. It is intriguing that pores were found on the male scapes of one outgroup taxon, Aphytis melinus, but were absent from the male scapes of Centrodora sp. Along with Aphelinus, both outgroups are members of Aphelininae, but Aphytis and Centrodora are in the subtribe Aphytini (Hayat 1998) and are closely related members of a clade separate from the lineage containing Aphelinus in the morphological phylogeny of Kim and Heraty (2012). A different RSS structure on the antennal club (terminal antennomere) of Aphytis males may also play a role in male antennation during courtship . A broader survey of RSS structures on male antennae in Aphelinidae would be worthwhile.
Aphelinus varipes males have specialized secretory structures on the scape, which are connected with the external pores. The glandular units have secretory cells releasing secretions in electron-lucid vesicles. These vesicles surround an end apparatus connected with the cuticular evacuating duct (produced by a duct cell) that allows the external release of the secretion. These features are typical of secretory cells belonging to class III, according to the classification of insect epidermal glands proposed by Quennedey (1974, 1991) and Quennedey (1998). Male antennal glands in this class have been reported for other groups of Hymenoptera, including Symphyta and Aculeata (Bin and Vinson 1986, Romani et al. 2005, Romani et al. 2008, and functionally, these structures have been related in most cases with courtship behavior by males. However, despite the numerous cases reported, occurrence of male antennal glands on the scape is relatively uncommon. Dahms (1984a) described the presence of antennal glands on the profoundly modified scape of males of Melittobia australica, which are related to stereotyped antennal movements during courtship involving the scape itself. Dahms (1984b) used differences among structures on the ventral area of male scapes in a revision of the genus Melittobia Westwood (Hymenoptera: Eulophidae). Dahms compared his work on Melittobia with that of van den Assem et al. (1982) on Melittobia and noted that both antennation and morphology of the structures on male scapes varied among Melittobia species groups. The species group that did not have extensively modified structures did not antennate as strongly in its courtship as the species groups that had more modified male scapes.
In Aphelinidae, male antennal glands have been reported in Encarsia asterobemisiae Viggiani and Mazzone (Hymenoptera: Aphelinidae), Encarsia aurantii (Howard) and Encarsia opulenta (Silvestri) (Pedata and Isidoro 1993), however, in these species RSS are located on the flagellar segments, not the scape. The functional significance of RSS on the scapes of Aphelinus and Aphytis (and perhaps other Aphelinidae) presents an intriguing problem in functional morphology associated with mate recognition. By demonstrating the diversity in forms of RSS and by placing this diversity in a phylogenetic context, our results are a first step in unraveling this puzzle. Knowledge of these structures may help in understanding mate selection in this genus, as well as adding to the knowledge about sex glands in parasitic Hymenoptera in general.