A comparative description of the mesosomal musculature in Sphecidae and Ampulicidae (Hymenoptera, Apoidea) using 3D techniques

Conflicting hypotheses about the relationships among the major lineages of aculeate Hymenoptera clearly show the necessity of detailed comparative morphological studies. Using micro-computed tomography and 3D reconstructions, the skeletal musculature of the mesoand metathorax and the first and second abdominal segment in Apoidea are described. Females of Sceliphron destillatorium, Sphex (Fernaldina) lucae (both Sphecidae), and Ampulex compressa (Ampulicidae) were examined. The morphological terminology provided by the Hymenoptera Anatomy Ontology is used. Up to 42 muscles were found. The three species differ in certain numerical and structural aspects. Ampulicidae differs significantly from Sphecidae in the metathorax and the anterior abdomen. The metapleural apodeme and paracoxal ridge are weakly developed in Ampulicidae, which affect some muscular structures. Furthermore, the muscles that insert on the coxae and trochanters are broader and longer in Ampulicidae. A conspicuous characteristic of Sphecidae is the absence of the metaphragma. Overall, we identified four hitherto unrecognized muscles. Our work suggests additional investigations on structures discussed in this paper.


Introduction
Hymenoptera form one of the largest insect orders and comprise more than 150,000 extant species (Aguiar et al. 2013). The group of interest examined in this paper constitutes a subclade of Hymenoptera, the Aculeata (stinging wasps, bees, and ants; Sharkey et al. 2012). Derived from the modified ovipositor, the stinger is a synapomorphy of aculeate Hymenoptera and a key innovation for their evolutionary success Schmidt 2016). The nature of phylogenetic relationships within the monophyletic Aculeata is still contested (e.g., Königsmann 1978;Lomholdt 1982;Rasnitsyn 1988;Alexander 1992;Brothers and Carpenter 1993;Ronquist et al. 1999;Peters et al. 2011Peters et al. , 2017Sharkey et al. 2012;Branstetter et al. 2017). Traditionally, Aculeata is divided into three lineages: Chrysidoidea, Vespoidea, and Apoidea (O'Neill 2001;Branstetter et al. 2017).
Morphological characters are still one of the major sources of phylogenetic inference (e.g., Friedrich and Beutel 2010;Ohl and Spahn 2010;Vilhelmsen et al. 2010;Zimmermann and Vilhelmsen 2016;Liu et al. 2019). Nevertheless, internal mesosomal structures are insufficiently studied across Hymenoptera, as predicated by Vilhelmsen et al. (2010), who provided detailed information for many apocritan wasps and other Hymenoptera; especially the mesosomal musculature of Pison chilense (Crabronidae) and external mesosomal characters for Pison chilense, Stangeella cyaniventris (Sphecidae), and Ampulex compressa (Ampulicidae) are described. They demonstrated, that the mesosomal region reveals considerable information for phylogenetic research. Previously, indispensable work about the mesosomal musculature in Hymenoptera was presented by Maki (1938), Snodgrass (1942; in particular, for Apis), Heraty (1989), and Matsuda (1970), followed by Prentice (1998). Recent substantial work was accomplished by Mikó et al. (2007). They dissected the musculature of the head and mesosoma in a review of the parasitic wasp family Scelionidae. Furthermore, a reinterpretation of the delimitation of the metapostnotum in Chrysidoidea was presented by Kawada et al. (2015). Moreover, Porto et al. (2016) defined internal mesosomal characters of bees and evaluated the potential of these structures, concluding that they are of great value to phylogenetic investigations. Garcia et al. (2017) described several body parts of three new species of the rare ant genus Zasphinctus, resulting in a comparative character matrix for species-level taxonomy. Subsequently, Liu et al. (2019) provided insights on the mesosoma of an ant worker of Myrmecia for comparisons with other Aculeata and to gain new information about evolution and body function.
A state-of-the-art method for morphological analyses is the three-dimensional imaging, using micro-computed tomography (microCT). It is a highly powerful technique (Faulwetter et al. 2013 and references therein;Garcia et al. 2017;Liu et al. 2019), as it makes internal structures visible without destroying the specimen. Moreover, the digital 3D models can be created repeatedly to work on different goals and the data can easily be shared worldwide.
By using 3D imaging, we aim to expand the basic morphological knowledge for phylogenetic investigations within Aculeata. In this paper we present data of muscular structures in the mesosoma of Sceliphron destillatorium (Illiger, 1807), Sphex (Fernaldina) lucae de Saussure, 1867 (both Sphecidae), and Ampulex compressa (Fabricius, 1781) (Ampulicidae) (Fig. 1). These wasps are solitary and nest-provisioning predators with different lifestyles (e.g., Williams 1942;Bohart and Menke 1976;Fouad et al. 1994;Haspel and Libersat 2003;Libersat 2003;Ohl and Spahn 2010). Both families were selected for their large number of plesiomorphic characters within digger wasps (Ohl and Spahn 2010), which might help to reconstruct the ancestral apoid anatomy. Primarily, we illustrate and describe mesosomal conformations of the skeletal musculature, with focus on the mesothorax, metathorax, and the first abdominal segment (propodeum). We also describe muscles that originate in the mesosoma and insert in the second abdominal segment (metasoma) because of strong interrelations of these muscles in this transition zone between both tagmata. The wasp waist allows for increased movability of the abdomen and, therefore, is an important anatomical cluster for various physical activities requiring precise movements of the abdomen below the body. This includes, for instance, stinging prey or enemies for defence, laying eggs (Williams 1942;Bohart and Menke 1976), carrying prey between mid or hind legs and abdomen while in flight, dragging prey forwards or backwards (Bohart and Menke 1976), and increasing balance in flight (at least when the second abdominal segment is petiolate; Bohart and Menke 1976).

Specimens and body parts examined
Sphex and Ampulex were taken from the collection of the Museum für Naturkunde Berlin (MfN) and Sceliphron was collected in the field (Table 1). To examine and compare the muscle sets, specimens of the same sex (females) were selected. We analysed the musculature of the mesothorax, metathorax, and the first and second abdominal segments.

Preparation, microCT, and 3D reconstruction
The extremities of the specimens were removed to minimize the scan field for optimizing the resolution of the data sets. Furthermore, the tip of the gaster was removed to facilitate the infiltration of the iodine, which intensifies the visibility of the musculature in the scan. Following Metscher (2009) and Gignac et al. (2016), our specimens were contrasted in a 25% iodine solution in pure ethanol  (100%) for three days and washed out with pure ethanol for 30 seconds. The wasps were dried using a critical point dryer (Leica EM CPD300; Table 1). Afterwards, the three specimens were scanned at the Visualisation Laboratory of the MfN using a Phoenix nanotom X-ray|s tube (General Electric) at 48-50 kV and 250-275 µA. At 1 second per image 1000-1440 projections were generated per scan. The different kV-and projection-settings depended on the respective specimen size, which was also responsible for the range of the effective voxel size between 3.4-5 μm ( Table 1). The cone beam reconstruction was performed using the CT reconstruction software PHOENIX|X-RAY DATOS|X version 2.0 (GE Sensing & Inspection Technologies GmbH).

3D segmentation and post-processing
The raw microCT image data were visualised and analysed by using a Wacom Cintiq 22HD interactive pen display and the software AMIRA ZIB EDITION 2020.02 and former versions (provided by the Zuse Institute Berlin). All muscles were segmented and labelled manual-ly by using appropriate segmentation tools in AMIRA. Segmented materials were transformed into high-resolution surfaces using the Isosurface-Tool in AMIRA. The reconstruction was accomplished for one body side of the specimens, as no structural asymmetries were observed in this region. Therefore, the number of muscles given in the results refers to one-half of the body. For post-editing (e.g., picture artefacts, file size reduction, file converting, figure compilation) we exported TIF-files from AMIRA into ADOBE PHOTOSHOP CS6.

Terminology
Skeletal musculature was categorised based on insertion sites. The muscle terminology of the Hymenoptera Anatomy Ontology (HAO; http://portal.hymao.org/projects/32/ public/ontology/) (Mikó et al. 2007;Vilhelmsen et al. 2010;Yoder et al. 2010;Seltmann et al. 2012) has been adopted here. In this connection, we provide a list of Universal Resource Identifiers (URI) for each muscular and cuticular term (Suppl. material 1: Table S1). It was created by using the "analyze" tool on the HAO website. Newly detected muscles, not listed in the HAO so far or found in other literature, were also named in the HAO-scheme by the areas of origin and insertion with additional topographical orientation, if required ( Order; mostly stated for functional groups of muscles: a or 1 first b or 2 second c or 3 third Descriptions, that involve the meso-and metafurca, are based on the terminology of Porto et al. (2016). The descriptions in the results were ordered by the point of insertion from mesosoma towards metasoma and by relevant functional groups, if possible (Table 2). In this comparative work, Sceliphron destillatorium serves as reference species (Fig. 2). In addition, a homologisation with the generalised nomenclature for the thoracic musculature of Neoptera following Friedrich and Beutel (2008) is presented in Table 2.

Data availability
The large image data sets accomplished for this study are available online as a data publication in conjunction with this paper. Thus, our images and raw data are freely accessible via the MfN data repository (Willsch 2019; https:// doi.org/10.7479/dft0-yy6m). Moreover, images will be available on the HAO portal (http://portal.hymao.org).

Results
We found 42 muscle pairs within the analysed tagmata of the three species (Table 2). There are 37 muscles in Sceliphron (mesothorax 18, metathorax 14, first and second abdominal segments 5), 39 in Ampulex (mesothorax 19, metathorax 16, first and second abdominal segments 4), and 40 muscles in Sphex (mesothorax 20, metathorax 15, first and second abdominal segments 5). The following description of the skeletal musculature in Sceliphron serves as structural basis. Subsequently, comparative descriptions of differing muscles in Sphex and Ampulex are given. Each muscle absent in one or two of the compared species examined is mentioned below (see also Table 2):
Second abdominal segment. Median mesophragmo-metaphragmal muscle (ph2m-ph3) is absent. The mesophragma in Ampulex is rectangular like the outer cuticle and lacks a posterior notch for the insertion of a muscle. Metaphragmo-second abdominal tergal mus-cle (ph3-T2; Fig. 5A-C, F) arises from the metaphragma and propodeum, inserts dorsally on the second abdominal tergite; broad, large muscle extended to the posterior region. Metafurco-second abdominal sternal muscle (fu3-S2; Fig. 5A) arises posteriorly from the metafurcal arm, positioned posteromedial to fu3-tr3, inserts anteroventrally on the second abdominal sternite. In length and width distinctly more gracile than in Sceliphron. Metasterno-second abdominal sternal muscle (s3-S2; Fig. 5A) arises from the metadiscrimenal lamella and inserts on the anterolateral margin of the second abdominal sternite. It is noticeably smaller and neither fan-like nor bent, as in Sceliphron.
Mesothorax. The mesopleural pit in Sceliphron presumably developed by muscle and spiracle reduction. According to Vilhelmsen et al. (2010), the occurrence of the mesopleural pit shows high variances within and amongst superfamilies. Spiracle reduction likely occurred independently in different groups. Snodgrass (1942), for instance, found the posterior thoracic spiracle in honeybee workers without a closing apparatus. Each of the other spiracles is equipped with an occlusor muscle (Snodgrass 1942). Vilhelmsen et al. (2010) documented the absence of the posterior thoracic spiracle in Stephanidae and Pteromalidae, while they evidenced its presence (without sp3occ) in the apoid family Crabronidae, as well as in Rhopalosomatidae (Vespoidea), and the non-aculeate families Cynipidae, Evaniidae, and Trigonalidae. Hence, not only Apoidea but also Spheciformes sensu lato bear a high variance of the development of this spiracle-muscle-complex. Duncan (1939) presented an illustration of the closing mechanism of the posterior thoracic spiracle in Vespula. The occlusor muscles we found in Sphex and Ampulex (Figs 4A-D, 5F-H) show wider attachment points than the fan-shaped muscle described in Duncan's work. In the neopteran representatives, like Zorotypus, examined by Friedrich and Beutel (2008; Table 2), sp3occ was not revealed. Concluding, other related specimens should be examined to exclude all doubts about the homologisation of the posterior thoracic spiracle and sp3occ and to gain further insights into the different formations.
In all species examined, pl2-cx2 is located as described by the HAO, with origin on the mesopleuron and anterolateral insertion on the mesocoxa (Figs 3D, 4A, D, 5A, C, D, G, H). However, it is larger and extending farther anteriorly in Ampulex (Fig. 5A, C, D, G, H). Ampulex distinct-ly shows the additional and slender mesocoxal muscle pl2-cx2b, which we describe here for the first time ( Fig.  5C-E, G, H). In Sphex it is broader and closely adjacent to pl2-cx2 (Fig. 4A, C, D). It is absent in Sceliphron. Consequently, the development of pl2-cx2b should be examined in other species to clarify the phylogenetic relevance.
The muscles fu2l-tr2 and fu2m-tr2 in Ampulex, which insert on the mesotrochanter, seem to have been coalesced completely, making a separation impossible (compare Fig. 7A, B). Because of the insertion and the rather medial position, we reasonably homologized the structure with fu2m-tr2 by excluding fu2l-tr2 for Ampulex. The unambiguous identification of both muscles in Sphecidae appears to indicate an autapomorphic feature of Ampulicidae. However, Vilhelmsen et al. (2010; see also references therein) stated that both muscles were found in Evaniidae, Platygastroidea, most Proctotrupoidea, Plumarius, and Apoidea, which might include all genera they examined (i.e., Ampulex, Apis, Bombus, Pison, Stangeella). However, the authors noted the absence of fu2l-tr2 in Orthogonalys (Trigonaloidea) and of fu2m-tr2 in Ceraphronoidea, Chalcidoidea, and Stephanoidea. Nevertheless, they explained that a secondary subdivision of fu2m-tr2 may have led to the development of fu2l-tr2. In summary, the contrariness referring to fu2l-tr2 needs to be clarified by additional studies on Ampulex, in particular.
In addition, fu2l-tr2 fills the mesopleural area in Sceliphron (Fig. 3C), whereas this muscle is smaller in Sphex (Fig. 4A, D). In contrast, pl2-cx2b extends over the mesopleural region in Sphex and Ampulex (Figs 4A, C, D, 5C-E, G, H). In Ampulex, the origin of this muscle is the same spiracle apodeme as that from which sp3occ arises (Fig. 5E-H); in Sphex it partly originates from the posterior thoracic spiracle and partly from the mesopleuron (Fig. 4A, C, D). However, we recommend a closer look at these different formations in other species before drawing phylogenetic conclusions.
Metathorax. The different constructions of the metathoracic muscles mainly depend on variations of the skeletal structures. The slight difference in the metapleural origin of pl3a-ba3 in Ampulex (Fig. 5C, E) is a consequence of the less distinct development of the paracoxal ridge (Fig.  6). As shown by Vilhelmsen et al. (2010), the paracoxal ridge is weakly developed in Ampulicidae and non-apocritan Hymenoptera, whereas it is highly variable within apocritan groups. Orthogonalys (Trigonalidae), which serves as reference species in the paper of Vilhelmsen et al. (2010), has a weakly developed paracoxal ridge, except for the ventralmost part. As no other information about the structure in Pison (Crabronidae) is available, it should be identical. We confirm the differences noted by Vilhelmsen et al. (2010), as the paracoxal ridge is weakly developed in Ampulicidae and well-marked in Sphecidae (Fig. 6). Additionally, Vilhelmsen et al. (2010) described a distinct paracoxal ridge in Chrysidoidea, Evanioidea, and Stephanoidea.
The muscle t2p-t3 inserts laterally on a spine, which is located dorsally on the mesophragma in Sphecidae (Fig.  Figure 7. Comparison of fu2m-tr2 -median mesofurco-mesotrochanteral muscle and fu2l-tr2 -lateral mesofurco-mesotrochanteral muscle, anterolateral view. A. Sceliphron destillatorium; B. Ampulex compressa. Scale bars: 0.4 mm (A), 0.5 mm (B). 8A). Vilhelmsen et al. (2010) revealed in Apoidea and Vespoidea a typical lateral insertion on the metanotum, which is not yet observed in other groups; this might indicate that this feature is synapomorphic in both superfamilies. Although we found the mesoscutellum to be of similar shape in all analysed species, t2p-t3 in Ampulex is instead located entirely between the upper and lower mesoscutellar sclerite (Fig. 8B). So far, this modification seems to be unique. To verify this, further representatives of Ampulicidae should be examined.
The metanotal muscle pl3la-t3 in Ampulex differs from that in Sphecidae because of the weakly developed metapleural apodeme, which leads to a rather more lateral than submedial position on the thorax (Fig. 5B, F). We found a fusion of the lateral metafurcal arms with the metapleural apodeme in Ampulex (Fig. 6D), as already observed by Vilhelmsen et al. (2010) in the same species, other apoid taxa (Stangeella, Apis, Bombus, Pison), and in Vespoidea. Vilhelmsen et al. (2010) stated that most apocritan Hymenoptera have a metapleural apodeme that is often fused with the lateral metafurcal arms. In non-ap-ocritan Hymenoptera, the metapleural apodeme shows high morphological diversity. In many cases, this may not be easy to recognize (Vilhelmsen et al. 2010). Studies on more species from both families are necessary to determine if the structures found in the present study are family-specific. Sphecidae has a well-developed metapleural apodeme, similar to Cynipoidea (Vilhelmsen et al. 2010), which is an important characteristic. Our results corroborate the conclusion by Vilhelmsen et al. (2010), that the development of the metapleural apodeme is highly variable within Apocrita and, moreover, even within Apoidea.
Additionally, the weakly developed metapleural apodeme in Ampulex influenced the origin of pl3-sa3, which only originates from the metapleuron and inserts on the metasubalare (Figs 5C, D, 9B). The origin of the metatrochanteral muscle pl3-tr3 is also affected in Ampulex (Figs 5D, E, 6C, D). This muscle originates from a delicate sclerite, which provides a narrow surface of origin. This sclerite arose from the fusion of the metafurcal arm and metapleural apodeme and is equal to the medial margin of the metapleural apodeme and metafurcal arm.  The homology of the metanotal muscle, which we tentatively assign to pl3lp-t3, according to the HAO terminology, cannot be assured. In the HAO, it is described as fan-shaped and posterolaterally originating from the metapleuron. However, size, structure, and position of pl3lp-t3 are different among the species examined (Figs 3D, 5C, D). In Ampulex, pl3lp-t3 shows great similarity to the description of it by the HAO (wide, fan-shaped, and arises laterally from the metapleuron), whereas in Sphecidae, pl3lp-t3 is very small and compact but still fan-shaped and located sublaterally. It appears to originate from the metanotum and to insert on the metapleu-ron. Additional examination of pl3lp-t3 in other specimens is required to resolve the homology of this muscle.
The muscle s3-cx3 is clearly identifiable in Sphex (Fig.  4B). It is located ventrally to fu3m-cx3 and might serve to strengthen the metacoxal function from the lower centre of the body. From fu3m-cx3, s3-cx3 might be subdivided. This possibly forms a genus-specific character of Sphex, but not of the family Sphecidae.
First and second abdominal segment. The metaphragma is conspicuously absent in Sphecidae among all studied taxa. Nevertheless, ph2m-ph3 (Fig. 3A) and ph3-T2 (Fig. 3A-D) in Sphecidae are homologue muscles. The metaphragma is usually located between the metanotum and the first abdominal segment (Snodgrass 1942). The HAO describes the metaphragma as the site of origin of the mesophragmo-metaphragmal and metaphragmo-second abdominal tergal muscles. Although the third phragma was found to be absent in honeybees by Snodgrass (1942). However, Vilhelmsen et al. (2010) stated that most Hymenoptera have at least a weak laterally developed metaphragma. This has been observed in Mymarommatoidea (Terebrantia) and Chrysidoidea (Aculeata). Vilhelmsen et al. (2010) described a metaphragma medially continuous adjacent to the lateral metapleural apodeme for other apocritan taxa (i.e., Vespoidea, Trigonaloidea, Megalyroidea, Stephanoidea, Evanioidea, most Ichneumonoidea, and Apoidea: Stangeella (Sphecidae), Pison (Crabronidae), and Ampulex (Ampulicidae)). Stangeella and Ampulex were analysed by dissection but not figured. We cannot confirm this specific pattern for Ampulex (Fig. 10A-D). The absence of the metaphragma we observed in Sceliphron and Sphex may be a potential autapomorphy or an independent reduction. Consequently, further investigation of this phragma is highly recommended.

Conclusions
We recommend additional investigations of the structures and features presented in this paper. It would be of great value to analyse the tagmata and other characteristics in the family Heterogynaidae and additional species of Crabronidae, Ampulicidae, and Sphecidae. Due to the unresolved phylogenetic position of Heterogynaidae and the paraphyly of Crabronidae, the study of more species from these taxa might be desirable. Structural investigations of more species of Vespoidea and Chrysidoidea would be helpful for clarifying controversial assumptions about phylogenetic relationships within Aculeata. Structures of phylogenetic significance were mainly found in the metathorax, i.e., the metapleural apodeme, paracoxal ridge, metaphragma, and the origin and insertion of associated muscles. Future studies should also focus on: the muscles that insert into the legs, the posterior thoracic spiracle as well as the occlusor muscle in closely related species, and the four muscles described here for the first time in Sphecidae and Ampulicidae.