Mouthpart adaptations of antlion larvae facilitate prey handling and fluid feeding in sandy habitats

Several insect groups feed exclusively on fluids during their larval or adult stages (Kingsolver and Daniel, 1995; Krenn, 2019). Recent multidisciplinary studies of the feeding mechanisms of megadiverse fluid-feeding groups, such as butterflies and moths (Lepidoptera) (Monaenkova et al., 2012; Zhang et al., 2018; Lehnert and Wei, 2019), bees (Hymenoptera) (Liao et al., 2020; Wei et al., 2020; Sun et al., 2021) and flies (Diptera) (Lehnert et al., 2017, 2022; Krenn et al., 2021) have revealed themes of complex structural and wetting dynamics of mouthparts that optimize fluid uptake. The larvae of antlions (Neuroptera: Myrmeleontidae) also feed solely on fluids; however, they are faced with challenges that differ from those of the aforementioned groups because of complications associated with residing in sandy habitats coupled with predatory behaviors (Griffiths, 1980; Fertin and Casas, 2007; Büsse et al., 2020 preprint).

Antlions represent approximately 30% of the nearly 6000 described neuropteran species (Stange, 2004; Aldrich and Zhang, 2016; Badano et al., 2017). The larvae of some species build pits in sandy substrates and remain at the bottom with their bodies concealed and mouthparts exposed, waiting for arthropods to fall into the trap for them to ambush (Griffiths, 1980). Although several aspects of antlion biology have been investigated, including pit construction (Lucas, 1982; Burgess, 2009; Franks et al., 2019), competition (Prado et al., 1993; Gotelli, 1997; Barkae et al., 2010) and morphology (Devetak et al., 2010; 2013; Zimmerman et al., 2019), the adaptations that enable fluid uptake in sandy habitats remain unexplored.

Residing in habitats composed of a sandy substrate presents unique difficulties for a fluid-feeding insect: for example, they are exposed to small substrate particles that could obstruct the flow of ingested fluids or hinder mouthpart movements. In addition, unlike other fluid-feeding insects (e.g. Hymenoptera; Linghu et al., 2015), antlion larvae are unable to use their legs to clean their mouthparts, which suggests other strategies for cleaning. Here, we explored the feeding mechanism of larvae of the antlion Myrmeleon crudelis, with an emphasis on revealing the adaptations that facilitate fluid feeding and prey handling in sandy habitats. We hypothesized that the mouthparts have a structural organization and wettability that play critical roles in prey handling and in channeling incoming fluids while keeping the mouthparts clean from sandy substrate particles.

Third instar larvae of Myrmeleon crudelis Walker 1853 were obtained from AntlionFarms.com (Pensacola, FL, USA). All individuals were kept in ventilated plastic containers and maintained at 21–23°C and 30–40% relative humidity (RH). Individuals were fed fruit flies, Drosophila melanogaster, but were not fed for at least 24 h prior to the feeding experiments. Individuals not used for feeding experiments were stored in a −80°C freezer. Unless stated otherwise, the antlions previously stored at −80°C were placed at room temperature (approximately 21–23°C) for at least 30 min prior to experimentation, which was conducted at 30–40% RH.

Larvae had their abdomens secured to a glass coverslip with hot glue. Forceps were used to provide larvae with fruit flies; the flies were pre-fed a 25% sucrose solution mixed with blue food coloring to increase the visibility of ingested fluids. The mouthpart movements and fluid uptake processes were video recorded with a brightfield setting at 10 frames s⁻¹ on an inverted confocal microscope (Olympus IX81, Center Valley, PA, USA) with cellSens^TM software. The videos were analyzed in Adobe^® Premiere^® Pro v.14.0 to quantify mouthpart movements and fluid uptake rates.

Antlion mouthparts were removed with forceps and dissecting scissors and placed through an ethanol dehydration series (70%, 80%, 90%, 95%, 100% ethanol, 24 h each) followed by critical point drying (Leica CPD300). The mouthparts were then secured to an aluminium stub with carbon graphite tape and sputter-coated (EMS 150TS) with 10 nm of platinum. The mouthparts were imaged by scanning electron microscopy (SEM; JEOL 6010LV) at 20 kV to observe the mouthpart structures. 3D images were acquired with the scanning electron microscope and InTouch Scope^TM software. In order to generate the 3D image, we first acquired a SEM image, then manually tilted the specimen holder to a 5% tilt and acquired a second image. The 5% tilt value was entered into the software value box that then created a 3D image using both images, i.e. stereo pair imaging with the before and after tilt images, which was used to assess the surface roughness of the dorsal side of the mandibles.

Mouthpart and head structures were studied with laser ablation tomography (LATscan) at L4iS (State College, PA, USA) and the University of Southern Maine (Portland, ME, USA) (Hall and Lanba, 2019). A linear drive stage was used to feed the sample through an ultraviolet laser (355 nm wavelength, pulse duration <30 ns, pulse energy of approximately 260 µJ, pulse repetition rate 15–30 kHz) at increments of 4 µm. The size of the RGB images was 6720×4480 pixels and the magnification resulted in a resolution of 1.1 µm per pixel. Image processing was performed with FIJI software (Schindelin et al., 2012). The 3D reconstructions and analyses were performed in FEI Avizo software (v.2019.4, ThermoFisher Scientific, Inc., Waltham, MA, USA).

Autofluorescence with confocal microscopy was used to correlate the extent of sclerotization of the mouthparts and their mechanical properties. The mouthparts were removed with dissecting scissors and forceps and placed into a droplet of dH₂O on a glass slide and a high-precision coverslip (12 mm diameter, thickness no. 1.5H, Neuvitro Corporation, Vancouver, WA, USA) was placed on top. The autofluorescence of mouthpart structures was studied with an Olympus IX81 inverted confocal microscope. Mouthparts were imaged with autoexposure settings and a 15-slice Z-stack in Olympus cellSens^TM software using the blue (DAPI, 377–360 nm excitation, 447–460 nm emission), red (CY3, 531–540 nm excitation, 593−640 nm emission) and green (GFP, 469–535 nm excitation, 525–639 nm emission) filters. Structures that appeared red are assumed to correlate with heavy sclerotization that is indicative of hard structures, green autofluorescence possibly relates to moderate levels of sclerotization, and those structures that show blue autofluorescence suggest the presence of the elastomeric protein resilin (Michels and Gorb, 2012; Eshghi et al., 2018).

The capillary-rise technique was used to assess a general circumferential wetting profile of the mouthparts, similar to the protocol outlined by Lehnert et al. (2013). Individual mandibles, maxillae and mandible–maxilla couplets were removed from antlions with dissecting scissors and forceps, and the base of the mouthparts was adhered to carbon graphite tape on a micromanipulator. The mouthparts were positioned above a dish and lowered so that the distal tip of the mouthparts was nearly level with the top of the dish. A syringe was then used to fill the dish with dH₂O and the interactions between the rising dH₂O and the mouthparts were recorded at 10 frames s⁻¹ (Dino-Lite AM4815ZTL digital microscope). The videos were analyzed with Adobe^® Premiere^® Pro v.14.0 to determine mouthpart wettability based on meniscus shape.

Mandibles were removed from the head with forceps and dissecting scissors and secured to carbon graphite tape on a glass slide in a horizontal position in front of a digital microscope for video recording at 10 frames s⁻¹ (Dino-Lite AD3713TB digital microscope). We administered droplets of dH₂O onto the mandibles using a spray bottle and used contact angles of the droplets (measured from the lateral side) to assess wettability. The mandibles were divided into three zones to investigate wettability along their length: zone 1, base of the mandible to tooth 1; zone 2, tooth 1 to tooth 3; and zone 3, tooth 3 to distal tip. The contact angles of three droplets were measured per zone on the dorsal and ventral sides to determine average droplet contact angles per zone for each individual, which were further used for statistical analyses. A second droplet experiment was performed on the same mandibles after they were washed in hexane (two washes, 15 min each) to remove cuticular waxes; the same methodology was used to assess wettability. The videos were analyzed in Adobe^® Premiere^® Pro v.14.0 to isolate frames where the droplet profile was in focus, and the contact angles and droplet diameters were then measured (Contact Angle plugin, ImageJ; Schneider et al., 2012). Droplets were measured only if they were not touching the setae. We compared the average contact angles among zones on the dorsal and ventral sides within treatments (the control with waxes present, or the hexane treatment without waxes) and compared average contact angles between treatments within each zone using analysis of variance (ANOVA) and paired t-tests (P<0.05), respectively, in JMP software v.14.0.

Minuten insect pins were placed through the heads of antlions with their mouthparts still attached and were sonicated for 1 min in dry sand (particle size of 184.93±22.82 µm, mean±s.e.m., acquired with SEM). After sonication, only the pins were handled to transfer specimens, to help prevent sand removal from the mouthparts prior to imaging. The pins were used to attach the heads to a micromanipulator and positioned so the distal tips of the mouthparts were facing downward, approximately 5 cm above a nebulizer filled with dH₂O. The mouthparts were exposed to water vapor from the nebulizer for 10 min. The dorsal surface of the mandibles was imaged with a Dino-Lite AM4815ZTL digital microscope before and after exposure to water vapor. The total area of the dorsal surface of the mandibles was measured in pixels using Adobe^® Photoshop^® CS6 v.13.0.1. Sand grains on the mandibles were colored in Photoshop, then the area of color, representing the sand coverage, was measured in pixels. As a control, we also performed the same experiment with the dorsal surface of the antlion heads. We calculated the percentage of the mandibles and heads covered in sand pre- and post-exposure to water vapor and compared them with a paired t-test (P<0.05) in JMP software v.14.0.

Larvae of M. crudelis seized the fruit flies with their mandibular teeth and the distal tips of the mandibles (n=8 individuals) (Fig. 1). The mid and distal regions of the mandibular teeth are likely heavily sclerotized, as suggested by their red autofluorescence with confocal microscopy (n=3 individuals) (Fig. 2), and the teeth did not pierce the cuticle, but instead secured the prey. The base of each tooth had blue autofluorescence, suggesting that the elastomeric protein resilin might be present. If there is resilin, it could provide elasticity and bendability to help prevent the teeth from breaking off during prey handling, but this requires further study. The distal tip of the mandible also had red autofluorescence and was observed to pierce the prey tissue (Fig. 2). Previous confocal microscopy studies of the mouthparts of other insect groups have revealed similar possible correlative themes of heterogeneous mechanical properties that relate to structure–function relationships and feeding habits (Michels et al., 2016; Lehnert et al., 2021, 2022).

While the mandibular teeth secured the prey and the tip pierced the prey tissue, the sclerotized maxillae (suggested by the red autofluorescence shown in Fig. 2) performed sliding antiparallel movements at an amplitude of 136±6 µm and exposed the food canal at a rate of 3±0 exposures s⁻¹ (all values presented as means±s.e.m., n=8 individuals) (Fig. 1; Movie 1). The serrated teeth near the tip region of the maxillae (Fig. S1; see also van Zyl and van der Linde, 2000) and mandible point posteriorly and likely assist in macerating and moving the prey contents into the food canal. SEM (n=5) and LATscan (n=3) show that the sliding mechanism is enabled by grooves and ridges that connect the maxilla and mandible (Fig. 1; Fig. S1, Movie 2; van Zyl and van der Linde, 2000). The mandibles and maxillae have a medial conduit with a C-shaped cross-section that, when brought together, create a circular food canal (described by van Zyl and van der Linde, 2000; Aspöck and Aspöck, 2003; Krenn, 2019), textured with cuticular projections approximately 1 µm in length that might assist in the flow of fluids and particulates (Fig. S1). We are not aware of previous reports of cuticular projections in the food canal. In addition to providing the piercing mechanism, the antiparallel movements also might assist in affecting the pressure differential in the food canal for fluid uptake, as observed with mouthparts of Lepidoptera (Tsai et al., 2014; Zhang et al., 2018).

The transparency of the fruit flies allowed observations of the inside of the flies during the feeding process. The liquified contents inside the fly became foam-like near the tip of the maxillae and then entered the food canal during the antiparallel movements. Antlions are known to be venomous (Yoshida et al., 1999; 2001; Nishiwaki et al., 2007), and the observation of the ingested fluids becoming foam-like prior to food canal entry suggests the presence of surfactants in the venom that would reduce viscosity and increase flow rates. In addition, the opening and closing of the food canal by the antiparallel movements of the maxillae might further break up the liquid column to reduce effective viscosity. The fluids in the food canal were observed to travel unidirectionally towards the head with a maximum speed of approximately 2 mm s⁻¹ (n=8 individuals) (Movie 1). Our observations of fluids flowing through the food canals indicate that the feeding mechanism is dictated by pressure differentials initiated by a sucking pump, and thus similar to the drinking straw model (Kingsolver and Daniel, 1995). Incoming fluids, also visible in the semi-transparent head, entered the head from both mandible–maxilla couplets and traveled through conduits and valves before pooling together in the cibarium (source of the pressure differential) and then continuing to flow posteriorly through the alimentary canal (Fig. 1).

The capillary-rise technique revealed that the outer surface of the mandible–maxilla couplets is hydrophobic (n=3 individuals, one mandible–maxilla couplet per individual) (Fig. 3). We further verified the hydrophobicity by measuring the contact angles of administered dH₂O droplets (96±3 µm diameter), which had a mean contact angle of 141±2 deg (n=5 individuals, includes mean measurements from all studied zones on dorsal and ventral surfaces); the contact angles did not significantly differ among the tested zones on the dorsal (F=0.5476, d.f.=2,12, P=0.5921) and ventral sides (F=0.9948, d.f.=2,12, P=0.3983) (Fig. 3). In addition, there were no significant differences in contact angles between the ventral and dorsal sides of the mandibles (F=0.5158, d.f.=1,28, P=0.4786). The contact angles of droplets in the hexane-wash treatment, however, were significantly lower than those in the control, with a mean of 94±2 deg (n=5 individuals, includes mean measurements from all studied zones on dorsal and ventral surfaces; measurements are from the same individuals studied for contact angle measurements prior to being washed in hexane) (Table S1). The contact angles did not differ among zones on the ventral side (F=1.6925, d.f.=2,12, P=0.2252), but did significantly differ among zones on the dorsal side (F=5.0455, d.f.=2,12, P=0.0257), with the tip region (zone 3) having a higher average contact angle than zone 2 (Fig. 3). The contact angles in the hexane-wash treatment did not differ significantly between the dorsal and ventral sides (F=0.0798, d.f.=1,28, P=0.7797).

The hydrophobicity of the mouthparts is likely due to a combination of wax layers and surface roughness of the cuticle, similar to other hydrophobic structures (Darmanin and Guittard, 2015). We confirmed the presence of cuticular waxes with SEM and surface roughness with SEM-3D imaging of antlion mouthparts (Fig. S2), which would assist hydrophobicity via Cassie–Baxter effects (Murakami et al., 2014), as found with many self-cleaning materials (Geyer et al., 2020). Self-cleaning properties would allow small droplets to form on the mouthparts, possibly during periods of high humidity, that could roll on the surface and remove sand particles. Without their removal, these particles could impede the antiparallel movements necessary for feeding and block fluid flow.

We tested the self-cleaning abilities of the dorsal side of the mandibles and heads by first exposing them to dry sand in a sonicator; however, the sand particles did not adhere well to the mandibles, which remained relatively clean even after sonication with only 5±0% of the surface covered in sand (n=3 individuals, one mandible per individual). The head, in contrast, was heavily covered by sand particles (45±8% of the head's surface was covered in sand) (Fig. 3). Droplets were observed to form on the mandibles during exposure to water vapor from the nebulizer. After exposure to the water vapor, there was significantly less sand coverage on the mandibles (3±0%; n=3 individuals, one mandible per individual, paired t-test, t₂=−4.0999, P=0.0273), while there was no significant difference in sand coverage on the surface of the head (n=3 individuals, paired t-test, t₂=−2.4943, P=0.0649), which remained covered in sand (42±6% coverage) (Fig. 3). We suggest that the setae on the head help trap sand particles to aid in camouflage, but the mouthparts remain clean for optimal feeding efficiency, but this requires further study. Overall, the mandibles appeared to have anti-dirtying properties that prevented adherence of sand particles before exposure to water vapor, which was not observed on the heads.

The capillary-rise technique also revealed a hydrophilic contact angle with the food canal when the mandibles and maxillae were separated from each other (n=3 individuals, one mandible and one maxilla per individual) (Fig. 3). In addition, when the distal tip of the food canal encountered the dH₂O, the dH₂O rapidly traveled along the entire length of the food canal that has the C-shaped cross-section (Fig. 3; Movie 3). The ability to perform capillarity in an open conduit with a C-shaped cross-section is an interesting phenomenon (Kolliopoulos et al., 2021), and although we did not find evidence that capillarity plays a role in feeding, additional studies of antlion mouthparts could inspire the production of novel microfluidic devices.

These experiments reveal that the mouthparts possess a wetting dichotomy: a hydrophobic surface with a hydrophilic food canal. The antiparallel movements of the maxillae serve as the mechanism for exposing the hydrophilic food canal, which is otherwise protected from debris when the mouthparts are sealed. The only time the food canals are exposed is when the maxillae shift anteriorly or posteriorly, relative to the mandible, to ingest the nutritive fluids from prey, but this occurs inside the prey, not where sandy particles could be problematic. Therefore, larvae of M. crudelis employ a strategy for keeping venom and incoming nutritive fluids isolated from the external sandy habitat, where sand particles could serve as a source of contamination and obstruct fluid flow. A wetting dichotomy of mouthparts modified for fluid feeding also has been reported for Lepidoptera (Lehnert et al., 2013), likely as a result of the common selective pressure of having to feed on fluids while keeping the mouthparts clean of debris.

The ability of antlions to feed on fluids in sandy habitats relies, at least in part, on the combined effects of the wetting dichotomy, morphology and mechanical properties of the mouthparts, thus supporting our hypothesis. The ability of the mandible – maxilla couplets to seal and perform antiparallel movements, thus determining when to expose the hydrophilic food canal, is particularly important because this takes full advantage of the wetting dichotomy: simultaneous fluid uptake and self-cleaning.

We thank Connie Kramer and Michelle Kane-Sutton (Kent State University at Stark) for assistance in acquiring the antlion specimens. We also thank Daytona Johnson and Ellen Camarato for laboratory assistance, and Valerie Kramer-Lehnert and Elaine Bast for reading an early version of the manuscript.