EUROPEAN JOURNAL OF ENTOMOLOGY EUROPEAN ENTOMOLOGY Detailed morphological descriptions of the immature stages of the ant parasite Microdon mutabilis (Diptera: Syrphidae: Microdontinae) and a discussion of its functional morphology, behaviour and host speci ﬁ city

. The myrmecophilous immature stages of hover ﬂ ies of the genus Microdon Meigen, 1803 (Diptera, Syrphidae) are still poorly known and only about 15 species were previously incompletely described and/or illustrated using light microscopy based on occasional ﬁ ndings mainly of pupae and third instar larvae. The exceptional ﬁ nding of a large number of second and third instar larvae and pupae (159 specimens) of Microdon mutabilis (Linnaeus, 1758) inside the nest of a new host species, Formica cunicularia Latreille,1798, enabled us to rear them and obtain a great number of eggs and ﬁ rst instar larvae. We ﬁ lmed and described the feeding behaviour and locomotion of these highly derived slug-like larvae. Combining light, ﬂ uorescence and scanning electron (SEM) microscopy, we describe in detail and illustrate the external features of all the immature stages of M. mutabilis (eggs, larvae and pupae). Covering the entire chorion of the egg is a peculiar microsculpture composed of volcano-like processes. The three larval instars strongly differ from each other, especially at the level of the shape of the body, the posterior spiracular tubercle and the cephaloskeleton. SEM microscopy was used to describe in detail the microsculpture, sensorial structures, spiracles and cephalic appendages of larvae and pupae. Fluorescence microscopy was used to reveal the exceptional presence of resilin in the external layer of the posterior spiracular tubercle in ﬁ rst instar larvae. The possible functional signi ﬁ cance of these structures is discussed.


INTRODUCTION
A plethora of soil dwelling arthropods have evolved complex associations with ants, and this has happened several times independently in ants since they radiated in the Eocene (Parker & Grimaldi, 2014). These organisms, known as "myrmecophiles", belong to all the major extant lineages of arthropods like arachnids, mites, myriapods, crustaceans and, most importantly in terms of the number of species involved, hexapods (Thomas et al., 2005;Parker & Grimaldi, 2014;Lachaud et al., 2016). Interactions between ants and their myrmecophiles range from various degrees of mutualism, commensalism, to predation, parasitoidism or parasitism (Hölldobler & Wilson, 1990;Ivens et al., 2016).
The bulk of myrmecophilous diversity is however restricted to a handful of endopterygote taxa, chiefl y lepidopterans, hymenopterans, coleopterans and dipterans (Höll-et al., 1973). Since these larvae live hidden in ant nests, showing complex interactions with their hosts, the study of their biology and behaviour is challenging. This is why the life cycle of most Microdon species is still undescribed.
However, all previous studies are based only on light microscopy and mainly illustrate the general habitus, while scanning electron micrographs are seldom provided (Akre & Paulson, 1993;Witek et al., 2011).
In this work we describe in detail all the immature stages (egg, larva and pupa) of Microdon mutabilis using light, fl uorescence and scanning electron microscopy. Our aim is to increase the information on this fascinating group of hoverfl ies and provide a modern morphological standard for the immature characters that can be used as a reference for further descriptions (or re-descriptions) of other species of Microdon, and for a discussion of the immature taxonomy of this group.

Material examined
This study is based on the analysis of 44 specimens of Microdon mutabilis (10 eggs, 10 fi rst instar larvae, 4 second instar larvae, 10 third instar larvae and 10 puparia). Second and third instar larvae and pupae were collected at Pisoniano (Latium, Central Italy) between April and September 2015 (Table 1) from inside the underground nests of Formica cunicularia (Latreille, 1798), which are easily recognizable by their small earth mounds; eggs and fi rst instar larvae were obtained in the laboratory during the rearing described below. The material is preserved in the A. Di Giulio collection (Rome, Italy).

Field sampling and captive breeding
The fi eld sampling was carried out in a small area of wet grassland (Pisoniano, Latium, Central Italy), used sporadically for grazing. In April 2015, respectively 72 pupae and 44 pupae were found in two different nests (Table 1). All pupae were attached to grass stems emerging from the nests.
This material was transferred to the laboratory and kept in cages (40 × 30 cm), at room temperature (24-27°C). Each cage was provided with a container full of earth, taken from the original ant nest, on the surface of which the pupae were placed. Periodically this arena was humidifi ed with distilled water. An artifi -Members of the family Syrphidae, also known as hover fl ies or fl ower fl ies, are nearly ubiquitous and belong to one of the largest groups of Diptera [about 6,200 known species, 828 of which are present in Europe (Pape et al., 2015)], especially known for their unsurpassed textbook examples of Batesian mimicry of Hymenoptera (Speight, 2008). Within this family, Microdontinae is the group with highest diversity of myrmecophiles, with about 110 documented records of associations with ants (Reemer, 2013). Most of these species are known to be social parasites or predators of ant brood, with only one being a parasitoid (Pérez-Lachaud et al., 2014). The most representative genus is Microdon Meigen, 1803, the larvae of which are social parasites associated with fi ve ant subfamilies: Ponerinae, Dolichoderinae, Pseudomyrmecinae, Myrmicinae and Formicinae (Reemer, 2013). Although Microdon is a speciose genus of about 300 species mainly occurring in South America, it is still poorly known, with the majority of studies on this genus on a few species mainly from Europe and North America (Wheeler, 1924;Akre et al., 1973;Garnett et al., 1990;Rotheray, 1991;Barr, 1995;Doczkal & Schmid, 1999;Schönrogge et al., 2002;Schmid, 2004;Gammelmo & Aarvik, 2007;Speight & Sarthou, 2011;Witek et al., 2011;Wolton, 2011;Speight, 2013). In Europe only six species are known: M. analis (Macquart, 1842), M. major (Andries, 1912), M. devius (Linnaeus, 1761), M. miki Doczkal & Schmid, 1999, M. mutabilis (Linnaeus, 1758 and M. myrmicae Schönrogge et al., 2002(Doczkal & Schmid, 1999Schmid, 2004;Speight, 2004Speight, , 2013Gammelmo & Aarvik, 2007). Another species from Bulgaria, M. sophianus Drensky, 1934 is listed in Fauna Europaea (Speight, 2004). However, it was never recorded again after description and the validity of this species is still under discussion. Of these, M. mutabilis and M. myrmicae can be identifi ed based only on the characters of the pre-imaginal instars, like the length of the puparium and spiracular tubercles (Schönrogge et al., 2002). The validity of these species is also based on their different ecologies and especially their host ants: Microdon mutabilis has been known as a parasite of Formica lemani Bondroit, 1917, whereas M. myrmicae was mainly found in the nests of Myrmica scabrinodis Nylander, 1846 and sometimes those of other species of Myrmica (Schönrogge et al., 2002;Speight , 2013). Adult morphology of European Microdon has been thoroughly studied, whereas their myrmecophilous maggots and their behavioural relationships with ants is much less well known.
Adult hover fl ies feed mostly on fl owers, sugary liquids or decaying plant matter and their apodous larvae show a huge spectrum of feeding habits ranging from phytophagous, mycophagous, saprophagous to predators and parasitoids. Microdon larvae are highly modifi ed, slug-like predators of ant larvae (Garnett et al., 1990), one of the most striking examples of feeding specialization in this group. Like other obligate myrmecophiles, these larvae are able to successfully infi ltrate into ant colony, feed on the ant brood and also gain other benefi ts like shelter, favourable climatic conditions and protection from predators (Akre cial ant nest with ants from the original F. cunicularia host colony was installed next to the cage with the Microdon pupae. Each nest consisted of a box lined with plaster, connected to an external feeding arena via a plastic tube. The feeding arena (plastic container without lid and with the inner sides treated with Fluon) was inserted inside the cage with the fl y pupae. Flies started to emerge almost synchronously from puparia 14 days after they were collected, and immediately mated. Eggs were laid about 4 days after mating, and the preferred oviposition site was the ant feeding arena. Furthermore, several M. mutabilis adults tried repeatedly to enter the ant nest through the plastic tube to the feeding arena, but were immediately recognized and killed by F. cunicularia ants.
Egg masses were collected using a fi ne brush, isolated in sterile vials, and incubated at room temperature in a humid chamber. Larvae started to hatch about 8 days after laying, but none of them reached the second instar.
Only 4 second instar larvae were found in the fi eld at the end of July. Three were immediately fi xed and one was reared in the laboratory to observe and record its feeding behaviour (Supplementary fi le 2), as described below (see Results). In September, 39 third instar larvae were found; 6 were kept alive for 2 months and put in a cylindrical artifi cial nest with about 50 workers of F. cunicularia plus ant larvae and pupae from the same nest. Interestingly, the ant workers aggregated and stayed on the dorsal surfaces of the syrphid larvae (up to 6-8 workers on one larva) as well as on the piles of their own brood.

Focused Ion Beam / Scanning Electron Microscopy (FIB/SEM)
Eggs, larvae (all instars) and puparia of M. mutabilis were examined using a Dual-Beam (FIB/SEM) Helios Nanolab (FEI Company, Eindhoven, The Netherlands) at the L.I.M.E. (University of Roma Tre, Rome, Italy). This instrument incorporates both a focused ion beam (FIB) and a scanning electron beam (SEM) in the same microscope. The standard dual-beam column confi guration consists of a vertical electron column with a 52° tilted ion column (typically employing a gallium, Ga source) both focused on the same point in the sample. This combination allows one to selectively ablate parts of a sample using the ion beam, and to observe and take high resolution images of a cross-section of the surface using the electron beam. This technique was used to investigate the internal structure of the processes of the marginal band of larvae, while the FIB/SEM was operated only with the SEM column to acquire high resolution images of the other structures.
Samples were prepared as follows: Larvae were immersed in 70% ethanol and then gradually dehydrated by placing in higher concentrations of ethanol up to 100%, with intervals of 10 min between each step. Then they were critical-point dried using a Bal-Tec CDP 030, mounted on double-sided carbon discs on standard stubs and gold sputtered using an Emitech K550 unit.

Light microscopy
An optical microscopy analysis was carried out on ten slidemounted fi rst instar larvae as follows: Specimens preserved in 70% ethanol were fi rst rehydrated in 3 descending consecutive baths (ethanol 50%, 20%, 10%), for about 10 min each, and washed in distilled water. Afterwards, they were placed in a 10% solution of KOH for about 30 min at 30°C. To facilitate the penetration of KOH into the body of the larvae, their skins were perforated using a minute probe. Later the specimens were transferred to hot lactic acid for 30 min. The larvae were pressed to remove their internal contents. When suffi ciently clean and clear they were transferred to 10% ethanol and dehydrated through a series of ethanol washes (20%, 50%, 70%, 85%, 95%, 100%) of 10 min each. Later, the specimens were immersed in Clove Oil for at least 1 h and fi nally mounted in Canada Balsam on a slide and put in an oven at 40°C for 3 days. These preparations were observed using an Olympus BX51 light microscope.
Measurements reported in the descriptions of eggs, fi rst and third instar larvae and pupae are means of 10 specimens, except for second instar larvae for which only 3 were measured.

Histology
For the histological analysis, fi rst instar larvae were killed in Bouin's solution, dehydrated in a graded ethanol series (as describe above) and embedded in paraffi n. The small paraffi n blocks were serially cut using a rotary microtome into sections 7 μm thick. Sections were stuck on slides using albumin and stained with haematoxylin and eosin. Slides were studied using an Olympus BX51 light microscope and a fl uorescence microscope Zeiss Axio Zoom V16.

Locomotion
A third instar larva of M. mutabilis was placed in an empty small Petri dish (5 cm of diameter). The locomotion movies were shot using an Olympus OM-D camera. In some videos the third instar larva was fi lmed through the bottom of the Petri dish in order to observe the movement of the muscular foot (Supplementary fi le 1).

Feeding behaviour
A second instar larva of M. mutabilis was placed in a small Petri dish (5 cm of diameter) with a wet disk of fi lter paper on the bottom together with 5 larvae and pupae of Formica cunicularia. The Petri dish was observed using an Olympus SZX2-ILLT stereo microscope connected to an Olympus camera U-CMAD3. The fi lm of the feeding behaviour was obtained using software CellD (Supplementary fi le 2).

Acronyms
In the description we used the following terminology and nomenclature of anatomical parts proposed by Courtney et al. (2000) for the larvae of Diptera and those of Garnett et al. (1990) for larvae of Microdon.
For structures not described previously we have used new terminology that refers to their peculiar shapes.
Body features. Body oval in dorsal view, distinctly convex dorsally, fl attened ventrally. Anterior part slightly narrower than posterior and bearing 2 raised lobes (Figs 2A1-A2). Marginal band, surrounding the whole perimeter of the body except for the anteromedial furrow, distinctly separates dorsal from the ventral side of the body (Figs 2A1-A2-A3). Dorsal surface rough, bumpy, transversely corrugated, deeply marked by subequal, conical, rugulose structures ( Fig. 2A1-B1, 3A). Thoracic and abdominal tergites fused and not recognizable (Fig. 2A1). Sculpticels on anterior part of the body distinctly pointed and posteriorly directed (Fig. 2B1). Four longitudinal grooves present dorsally (Fig. 2A1) dividing dorsal body surface into 5 main longitudinal fi elds: 1 medial, 2 lateral and 2 external marginal fi elds. Medial fi eld partially divided into 2 halves by a longitudinal, medial line (ecdysial suture?). Dorsal surface with regularly spaced "fl ower-like" sensilla ( Fig. 3A): medial fi eld with 2 longitudinal rows of 9 sensilla; each lateral fi eld with 13 sensilla arranged in 2 rows (7 along lateral groove and 6 along medial groove). Each marginal fi eld with 1 row of 10 sensilla. Each sensillum (Fig. 3C) composed of a cylindrical base, with many imbricate, thick sculpticels, apically with a medial fl ower-like structure with a variable number (5-10) of lobes, pointed at tip, encircling a medial dome with a lateral pore. Ventral surface is wide, soft (Fig. 2A3), transversally multi-folded, markedly furrowed by a deep, longitudinal, medial groove (Figs 2A3, 3B), running along abdominal sterna and surrounded by a hairy marginal stripe. Ventral surface covered by pointed microsculpture medially, fi nely pilose on sides. The thoracic sterna possibly represented by the fi rst 3 narrow anterior segments, separated by deep transverse furrows (Fig. 2B3). First 7 abdominal sterna, possibly recognized by the presence of 7 transverse rows of 6 fl owerlike sensilla, 3 on each side of medial groove. Ventral fl ower-like sensilla (Fig. 3D) similar to dorsal ones except for fl at, soft, unsculptured base and fl at, thin, distinctly pointed lobes. Suboval marginal stripe covered by elongated hairs (Fig. 2A3), posteriorly directed and bearing some small fl ower-like sensilla. Anal opening wide, transverse, subtriangular (Fig. 2A3). Four fl ower-like sensilla posterior to the anus.

Specimens
Posterior spiracular tubercle. Elongate, cylindrical, strongly sclerotized (Fig. 5D) structure, emerging perpendicularly from posterodorsal part of abdomen (Fig. 2A2), with apical part wider than basal and medially incised, and sides carinated and sharply edged (Fig. 7A2). Main part of tubercle furrowed longitudinally, both anteriorly and posteriorly, by a deep longitudinal groove separating 2 subparallel structures, each circular in section and containing one tracheal tube (Figs 5D, 7A1-A2). Wall of spiracular tubercle very thick and multi layered (Figs 11A1-A2). Fluorescence microscopy showing the presence of resilin in the external layer (Figs 5E-F). Surface of spiracular tubercle with peculiar microsculpture, completely covered by imbricate, sclerotized scales with an indented superior edge (Figs 7A1-A2); dimensions of scales decreasing basally. Apex of tubercle with 2 circular smooth plates, slightly convex, each with 1-2 respiratory narrow fi ssures that communicate with the distal part of tracheal trunks (Fig.  7A1). The remaining 2 tracheae separated for their entire length, each serving one side of the body (Fig. 5D).
Marginal band. Appearing as an undulated fringe of elongate, parallel, radially projecting processes, continuously surrounding the body laterally and posteriorly, only absent on the small V-shaped anterior part of the tergum (Figs 2A1, 8A1). Length of processes regularly varying, showing 8 waves on each side (Figs 2A1-A3, 8B1). At the apices of the 8 waves, the longest processes appear thicker as a result of partial lateral fusion of 2 adjacent simple processes (Figs 8B2-B3), and bear dorsally one apical and one subapical spiniform seta (Fig. 8A2). Each simple process composed of an elongate stem and an apical fringed brush (Fig. 8A3); the stem showing two very different surfaces: dorsal surface apparently articulated with 4-5 imbricated joints, the last one fringed apically (Fig. 8A3); ventral surface completely smooth (Fig. 8B3). Cross sections of these structures (Fig. 8B3) showing complex cuticular projections.
Second instar differs from fi rst instar as follows: Dorsal surface rough, deeply marked by sub equal, irregularly wrinkled microsculpture (Figs 9B-F), with some scattered, round and multiperforate depressed plates (Figs 9B-C-D), which are slightly rugulose. Dorsal reticulation processes abundant, forming an irregular pattern of intersecting rows (Fig. 9A). Each simple reticulation process crown-shaped (Figs 9C-E), bearing 4-5 extended, pointed projections. Longitudinal dorsal grooves reduced to a couple of medial furrows (Fig. 9A). Fewer fl ower-like sensilla than in fi rst instar larvae, but similar in shape (Fig. 3E). External perimeter of abdomen resembling a thick, raised, cuticular frame, completely covered by pointed, rugulose and scallop-like sculpticels (Figs 10A1-A2). Pseudocephalon small compared to the rest of body. Posterior spiracular tubercle short, dome-shaped, almost hemispherical, reddishbrown (Figs 7B1-B2; 13A1-A2), deeply sulcated anteriorly and posteriorly by an incomplete medial depression (Figs 7B1-B2). Spiracular tubercle surface covered by many independent polygonal plates, with serrate margins, fi tting together like pieces of a puzzle; subapical plates irregular in shape (Figs 7B1-B2); polygonal plates representing fl attened apices of more complex cuticular arborescent structures, each characterized by a multi-branched stem emerging from the internal surface of a spiracular tubercle; length of stems becoming shorter from base to apex (Figs 11B1-B2). Apical surface of spiracular tubercle smooth and butterfl y-shaped with irregular margins, furrowed by 4 groups of narrow respiratory fi ssures radially disposed and medially with two round holes (Figs 7B1-B2). Mandibles completely separated with no sign of anterior fusion. Each mandible antero-ventrally bearing 15-18 conical sharp, inwardly curved denticles, the 6 anterior ones more fl attened and not distinctly bidentate at apex.
Processes on the marginal band short and subequal in length except for pairs of longer processes [on anterior part of body bifurcate at apex (Fig. 10A1)], regularly spaced with 7-10 shorter ones in between (Figs 10A2-A3); each process composed by a smooth, cylindrical stem and a fl attened, apical brush; longer processes combined with 3-4 spiniform setae (Fig. 10A3).
Third instar differs from second instar as follows: Body strongly convex dorsally, nearly semi circular in transverse section (Figs 13B1-B2); surface entirely covered by papillary sculpticels without multiperforate and depressed plates. Dorsal reticulation reduced to a narrow, lateral strip along the perimeter of abdomen (Figs 10B1-B2, 13B1-B2). Dorsal reticulation processes forming semi circular or polygonal shapes (Figs 13B1-B2). Each reticulation process showing stringy, extended projections (Figs 10B1-B2). Dorsal convex surface with scattered sub circular groups of 5-9 umbrella-like structures (Fig. 12F) with a fl attened, wrinkled, circular apex bearing a cylindrical stem. Dorsal grooves absent. Flower-like sensilla similar in shape to those of previous stages, with lobes slightly narrower (Fig. 3F). External perimeter without raised, cu- ticular frame. Surface of the posterior spiracular tubercle covered laterally with many polygonal plates; a subapical stripe of irregular plates dividing the apex into 2 halves (Figs 7C1-C2); polygonal plates representing the fl attened apices of cuticular cylindrical structures (Figs 11C1-C2). The base of posterior spiracular tubercle encircled by a smooth cuticular crown (Fig. 7C2). Apical surface of posterior spiracular tubercle smooth with irregular margins, furrowed by numerous groups of narrow respiratory fi ssures, radially arranged (Figs 7C1-C2). Mandibles dorsally fused together by a thin membranous belt (Fig. 6B3). Each mandible antero-ventrally bearing 26 conical, sharp,       inwardly curved denticles, the posterior bigger and more spiniform than the anterior ones (Fig. 6B2). Processes on the marginal band short, smooth, with cylindrical stem and distal portion with a double, alternate conformation: single or biramous (Figs 10B1-B2-B3). No spiniform setae along marginal band.

Locomotion (Supplementary fi le1)
Movement is peristaltic (similar to the foot of snails), mainly performed by using the ventral muscular plate, which is covered with hairy microsculpture. The peristalsis can involve the whole body in fi rst instar larvae, which can quite rapidly completely stretch and contract their bodies. Due to the hardening and thickening of the dorsal part of their bodies, movement in second and third instar larvae is slower and only consists of repeated contraction and relaxation of muscles that propagate an anterograde wave, starting in the anal region of the ventral plate. However, fi rst instar larvae can also move slowly. During this movement the larva slides forward the hairy medial part of the ventral plate on a wet, mucous, adhesive layer. This larva can also move backwards and sideways, or just rotate its body by contracting and partially folding its sides, which determines the direction. For the backward and sideward movements the waves of contraction start on the anterior part of the plate and go backwards. The mandibles are not involved in locomotion and remain retracted, while the head, completely protracted during locomotion, is swung from side to side and explores the environment.

Feeding behaviour (Supplementary fi le 2)
We observed and fi lmed a second instar larva of M. mutabilis feeding on pupae and larvae of Formica cunicularia. The feeding behaviour consists of fi rst approaching prey at a 90 degree angle while swinging its pseudocephalon from side to side. When the prey is reached, M. mutabilis partially raises the anterior part of its body and starts a sequence of vigorous and deep lunges into the prey's integument via repeated and complete protractions and retractions of its mouthparts. After lacerating the prey's integument, the Microdon larva starts sucking the body fl uids of its prey by means of a pumping action of the pharynx,. The extreme mobility of the head skeleton enables the larva to reach different areas inside the prey.

DISCUSSION
The study of myrmecophiles is challenging because they are rare, live in concealed environments (ant nests) and the interactions with their hosts are complex (Di Giulio et al., 2011). While it is diffi cult to study their behaviour and life cycle in nature, it is also diffi cult to rear them in laboratory. Because of these diffi culties that hamper direct observations, the nature of most interactions between myrmecophiles and ants, and the function of many structural adaptations still remain a mystery or a matter of speculation (e.g., Di Giulio & Moore, 2004). Most myrmecophiles have evolved ways of being accepted by ants and of surviving and developing in their nests. Such adaptations include: chemical and morphological mimicry; specialized feeding behaviour and ways of inducing ants to feed them; and structural and chemical modifi cations that enable them to avoid being attacked by ants (Thomas et al., 2005;Lachaud et al., 2013). This is also the case for species of Microdon, whose biology has been inferred mostly from occasional observations in the fi eld and laboratory (Garnett et al., 1985;Barr, 1995;Elmes et al., 1999;Wolton, 2011).
The exceptional fi nding of a large number of second and third instar larvae and pupae (159 specimens, see Table 1) of M. mutabilis inside Formica cunicularia nests in Central Italy, enabled us to rear them and obtain eggs and fi rst instar larvae. By using light, fl uorescence and SEM microscopy, we have described in detail and illustrated the external features of all the immature stages of M. mutabilis (eggs, larvae and pupae).

Functional morphology
The three larval instars strongly differ from each other, in particular in the shape of their body, the posterior spiracular tubercle (Fig. 7) and the cephaloskeleton (Fig. 6). The modifi cations of these parts during development are probably linked to the different behaviours of the larval instars inside the nest of the host and to specifi c morpho-functional constraints.
First, a body, which changes from soft and fl attened (fi rst instar) to hard (though not sclerotized) and strongly domeshaped (third instar), constrains their locomotion, which in the third instar involves slug-like slipping on a hairy, mucous ventral plate. Since eggs are laid outside the nests of their host, the newly eclosed fi rst instar larvae possibly need to be highly mobile in order to enter a nest and reach the brood of the ant. The slow and bulged third instar larva is, instead, already in the brood chamber, where it develops up to pupal stage, and its strongly convex shape and hard and thick dorsal cuticle are likely to protect them from occasional attacks by ants. Furthermore, it is hypothesized that the presence of parasites inside the nest can stimulate the ants to leave the parental nest and move to another place. The empty nest, successively, can be re-occupied by other species of ants, which are not hosts of Microdon and consequently able to recognize this parasite (Schönrogge et al., 2000). In these rare cases the protective structure of the larvae could help them to resist the attacks of ants, which physically cannot bite and hold the larval body because it is too big, thick and without areas that can be gripped and held by ants.
Second, the function of resilin in the external layer of the posterior spiracular tubercle in fi rst instar larvae is still unclear, but it certainly increases the elasticity of this structure and may be related to its exceptional length (Fig. 7A2) and thus to frequent mechanical stress. It is noteworthy that the short and strongly sclerotized, dome shaped posterior spiracles of second and third instar larvae do not have resilin. We can speculate that the increased fl exibility may be related either to possible manipulation by the host's mandi-bles (it is unknown whether the ants actively transport the fi rst instar larvae, but if they do, the long tubercle might be used as a "handle"), or associated with active infi ltration of these small larvae into the nest through the soil where such a long "snorkel" could be damaged.
Third, as already stressed, the structure of the cephaloskeleton undergoes great modifi cation during larval devel- opment, mostly in terms of the mandibles. In fi rst instar larvae the mandibles are fused and form a unique serrated blade while in second and third instar larvae the mandibles are separated into two blades that are mesodorsally connected.
The SEM analysis revealed the fi ne morphology of the peculiar fl ower-like structures present on both the dorsal and ventral surfaces of the bodies of larvae of M. mutabilis. Garnett et al. (1990) describe homologous structures as sensilla in other North American species, and for this reason in this work we refer to those of M. mutabilis as fl ower-like sensilla. The presence of a pore in the medial dome of these fl ower-like sensilla could indicate they have a chemo receptive function, though a possible glandular function cannot be excluded. A glandular function is very likely for the dorsal processes that make up the reticulations typical of most third instar larvae of Microdon, including M. mutabilis in which pores and a sticky substance associated with these processes were recorded in this study. The functional role of these secretions in these predators is still obscure, since we did not see an ant licking the bodies of larvae, as is reported in other myrmecophiles that offer ants appeasing substances.

Results of fi eld sampling and captive breeding
Although Microdon mutabilis is a relatively rare species, considered endangered in many countries (Schönrogge et al., 2002;Van de Meutter et al., 2009), we found a large number of specimens (72 inside a single ant nest!), if compared with Microdon myrmicae, for which the maximum number of specimens recorded per nest is 27 (Witek et al., 2012). The detection of parasitized nests was, however, diffi cult, and so far only 4 nests with M. mutabilis have been found. The fi eld site in Central Italy is very similar to the one described for M. myrmicae (Gammelmo & Aarvik, 2007;Van de Meutter et al., 2009;Wolton, 2011).
During our laboratory rearing several Microdon adults were observed entering a nest of Formica cunicularia where they were immediately recognized and attacked by ants. In the literature it is documented that the methyl 6-methylsalicylate, a constituent of the mandibular gland secretion of many Formicidae, could have a role in the recognition of a suitable Formica lemani nest (Schönrogge et al., 2008). It seems plausible that the M. mutabilis we reared could have perceived this or a similar compound and used it to locate the artifi cial nest of F. cunicularia.
In the laboratory we placed some third instar M. mutabilis larvae in an artifi cial nest with F. cunicularia workers and brood from the nest in which these parasitic larvae were found, and noted that all the ant workers aggregated on the dorsal surface of the syrphid larvae in the same way as they aggregate on their brood. Similar behaviour is described by Barr (1995) after having placed 6 M. mutabilis larvae found in a F. lemani nest within an artifi cial nest of Myrmica ruginodis (Nylander, 1846); in this case the author interpreted this as an attempt by ants to prevent the social parasites from feeding on the ant brood. However, alternative hypotheses are also possible and a protective behaviour of the ant workers toward brood seems more likely based on the chemical mimicry reported in the literature for other Microdon species (Howard et al., 1990a, b) and their host ants.

Feeding behaviour
The hemicryptocephalic condition, with no sclerotization of external parts of the head coupled with a further development of an internal cephaloskeleton that characterizes larvae of the Muscomorpha (= cyclorrhaphous Brachycera) is a derived condition within the order Diptera (Stehr, 1991). A protractile and highly mobile pseudocephalon is advantageous for fl y larva, since it can maximize the exploitation of a food resource and, as suggested by Rotheray & Lyszkowski (2015), is energy-effi cient because the rest of the body remains immobile. The mandibles of species of Microdon differ from those of some other Cyclorrapha, because they are enlarged and fi nely and tightly serrated. This is likely a derived condition, which strongly characterizes Microdon. Most predatory cyclorraphans use hooklike mandibles to perforate the integument of their prey. Phaonia goberti (Mik) and Phaonia subventa (Harris), two predatory species of the family Muscidae, are able to pierce the prey's integument using their mandibles, supported by the parastomal bar, almost as scissors (Rotheray & Wilkinson, 2015). In M. mutabilis, as in other species of Microdon (Rotheray & Lyszkowski, 2015), the bladelike mandibles saw through their prey's tissues like a knife. The presence of a medial lobe on the labial sclerite bearing apically small teeth could help to stabilize the mandibles gripping the prey and thus facilitate feeding. After piercing the integument, M. mutabilis starts to suck the body fl uids of its prey.

Locomotion
The absence of thoracic legs strongly affects the mobility of dipteran larvae. Many fl ies, especially those in brachyceran groups with hook-like mandibles, move their mouthparts vertically and obliquely to hook on to the substrate and establish an anchor point on which they slide the rest of the body (Berrigan & Pepin, 1995, Schneeberg & Beutel, 2015, whereas, locomotion in M. mutabilis is performed without using the head skeleton. The highly developed ventral musculature in M. mutabilis, coupled with the enlarged body with a fl at "foot", is suffi cient to allow an autonomous movement without the need of mandibles, in way analogous to slugs with which they were initially confused (Reemer, 2012). Furthermore, the blade-like mandibles of Microdon are not suitable for attaching to the substrate for supporting the larval body during a peristaltic wave. Concerning the marginal band, which encircles the larval body, from the behavioural observations in the laboratory of the different larval instars, it does not seem to be directly involved in locomotion, but its primary role seems to be to perceive mechanical information with the radial setae in order to orientate and freely move inside the galleries.