Dendrocerus mexicali (Hymenoptera, Ceraphronoidea, Megaspilidae): Novel antennal morphology, first description of female, and expansion of known range into the U.S.

Abstract Dendrocerus mexicali has been described by Paul Dessart from a single male specimen collected in Mexico. Using 87 newly identified specimens we expand the known range to include the Southwestern United States and Florida, provide an expanded description of the species, and provide the first record of the female. We also use confocal laser scanning microscopy and in vitro hydrostatic pressure changes to investigate the functional morphology of apparently unique basally flexible antennal branches.


Introduction
Ceraphronoidea (Hymenoptera) is a widespread superfamily of parasitoid wasps comprised of two extant families: Ceraphronidae and Megaspilidae. Little is known about the biology of Ceraphronidae, but there are quite a few host records for Megaspilidae, especially for the genus Dendrocerus Ratzeburg, 1852 (Fergusson 1980;Dessart 1999).
Host records suggest that Dendrocerus parasitizes a broad range of orders, including Hemiptera, Neuroptera, Coleoptera, Diptera, Hymenoptera, (Fergusson 1980, Dessart 1995. Many of its hosts are predators or parasitoids of non-heteropteran Hemiptera, especially of aphids (Aphididae) (Fergusson 1980;Dessart 1995). Based on host records, some species are specialists, while others are generalists, and while a few may be primary parasitoids, many Dendrocerus are hyperparasitoids (Fergusson 1980). D. carpenteri, which has a very broad host range, has been recorded as being a secondary, tertiary, and even quaternary parasitoid (Fergusson 1980;Haviland 1920).
Dendrocerus mexicali was first described from a single male specimen collected on wild mustard in Mexicali, Mexico (Dessart 1999). Little is known about its natural history. The female has never been described and the host relationships of D. mexicali remain unknown.
Dendrocerus mexicali, like other Dendrocerus species of the halidayi species group, have antennae with long flagellar projections (flagellomeres are "branched" or "ramose"). The antennae of the male D. mexicali is perhaps its most distinguishing feature (Dessart 1999). While branched antennae are not uncommon across Hymenoptera, including Dendrocerus, the base of each flagellar process is wrinkled and is lighter than the flagellomere or the process ( Figure 1A). The function of this region is unknown, and even Dessart was not sure if it was an artifact of preservation (Dessart 1999). One of the aims of this study is to investigate the function of this region.

Methods
All specimens are point-mounted and air-dried. Specimens are deposited in the University of Central Florida Arthropod Collection (UCFC) (18 males and 5 females), the Canadian National Collection of Insects, Arachnids, and Nematodes (CNC) (9 males and 55 females) and Pennsylvania State University Collection Frost Entomological Museum (PSUC_FEM) identifier. All figures, OWL files, and supplementary files are available on Figshare (https://dx.doi.org/10.6084/m9.figshare.2063586.v1).

Confocal laser scanning microscopy (CLSM)
CLSM was used to image the male antenna and genitalia. Dissected male D. mexicali antennae and genitalia were placed in a droplet of glycerol between two no. 1.5 coverslips with a small amount of Blue-Tack as a spacer . Specimens were examined with an Olympus FV10i Confocal Laser Scanning Microscope. The antenna was imaged using three excitation wavelengths: 405 nm, 473 nm, and 559 nm. Autofluorescence was detected and assigned a pseudocolor using three channels with emission ranges of 420-520 nm (blue), 490-590 nm (green), and 570-670 nm (red), respectively. Volume rendered micrographs and media files were created in ImageJ (Schneider et al. 2012) using maximum intensity. The genitalia was imaged using two excitation lasers of 631 nm and 499 nm. Two channels were used to detect emissions of 647 nm (green) and 520 nm (red), respectively.

Bright field images
Bright field images were taken using an Olympus ZX41 compound microscope with an attached Olympus DP71 digital camera. Images were stacked and aligned using Zerene Stacker Version 1.04 Build T201404082055.

Antenna coiling experiment
Following the methods described by Steiner et al. (2010), we removed the antenna from one specimen stored in glycerol and one dried and pinned specimen. Both were macerated in 10% KOH for 10 minutes, and then stored in 80% ethanol overnight. We then placed the antenna in 100% ethanol for 10 minutes before transferring to distilled water.

mx autogenerated description
Specimen data, specimen images, OTU concepts and phenotypes expressed in natural language were compiled in mx (http://purl.org/NET/mx-database) and the description and material examined sections of this article were automatically generated from this software. Morphological terminology in the description and diagnosis are linked to classes in phenotype-relevant ontologies (Hymenoptera Anatomy Ontology (HAO), Phenotypic Quality Ontology (PATO), Biospatial Ontology (BSPO), OBO Relation Ontology (RO), Ontology for Biomedical Investigations (OBI), and Information Artifact Ontology (IAO); all of which are available at http://www.ontobee.org/).

Discussion
Branched antennae are common among various groups of insects. Many Diprionidae have pectinate and bipectinate antennae, though articulated branches have not been described (Benson 1939;Benson 1945). Some Chalcidoidea, such as Eucharitidae, have ramose antennae, though none have been reported to be capable of moving the branches (pers. comm. John Heraty 2015). Dendrocerus of the halidayi species group also have ramose antennae, though none besides D. mexicali have articulations (Dessart 1999).
This ramose flagellomere increases the surface area of the antenna, which could aid males in detecting female pheromones. Although nothing is known of D. mexicali mating behavior, male D. carpenteri have been shown to be attracted to sex pheromones released by the females (Schwörer et al. 1999). Heavy antennation during courtship has been observed, which implies the possible presence of chemosensilla on the antenna (Liebscher 1972).
Dessart postulated that the wrinkled regions at the bases of the male antennal branches were points of movement, which is extremely likely given the high resilin content of the cuticle that we found there (Figure 2) (Dessart 1999). This evolutionary novelty might allow the branches to fold, preventing the ramose antenna from getting caught on obstacles, allowing the wasp entry into a confined space, or as a mechanical defense against breakage. Hymenoptera do not have antennal pulsatory organs, but they can change the hemolymph pressure in their antenna indirectly through movements of their pharynx (Matus and Pass 1999). This movement may be controlled by the wasp via hemolymph pressure changes and hydraulics acting antagonistically against the the resilin at the base of the branch, though it may only be a passive movement of the branches when external force is applied. We replicated the Steiner et al. (2010) antennal coiling experiment to test whether the branches might be operated hydraulically and directly by the insect. Our results offer no evidence for hydraulic movement, but this could be due to damaged specimens or a more complicated mechanism.

Author contributions
Conceived the project: IM, KNB. Character concept generation, semantic statement generation, specimen visualization and creation of plates: KNB, IM. Wrote the manuscript: KNB, IM, ARD. Commented on the final stage of the manuscript: IM, ARD.