Thyridanthrax fenestratus (Fallén, 1814) is a bombyliid with poorly understood biology. It was recently shown to locally but frequently parasitize Pemphredon fabricii (M. Müller, 1911) (Hymenoptera: Crabronidae), a crabronid wasp that abundantly nests in old Lipara-induced galls on the common reed Phragmites australis (Cav.) Trin. ex Steud., 1840. The parasitism modes in Bombyliidae and Thyridanthrax spp. are not uniform. Here, we report that Th. fenestratus switches facultatively between killing the host almost immediately (idiobiont strategy) and killing the host at a later developmental stage (koinobiont strategy). We document the koinobiont parasitoid strategy for a series of Th. fenestratus larvae parasitizing P. fabricii. We found that a significant portion of Th. fenestratus larvae spend winter as young larvae and start feeding on fully developed and defecated prepupae of P. fabricii only after the end of cold-induced winter diapause. The time needed for the development of Th. fenestratus larvae exceeds several times the time needed for pupation of P. fabricii prepupae; the parasitized prepupae, therefore, remain paralyzed until the parasitic larva completes feeding. Fungicides, which alter the pupation of the host larva, seem to have negligible effects on Th. fenestratus larvae. The ability to switch between the two parasitism strategies has already been reported for several Anthrax spp., though the ability to block the host in the defecated prepupa stage and prevent its pupation following cold-induced diapause is herein reported for the first time.

Diptera, eclosion, ectoparasite, idiobiont, koinobiont, parasitoid, pupation

Thyridanthrax fenestratus (Fallen, 1814) is a Palearctic species of bee fly (Evenhuis and Greathead 2015). Larval stages are thought to live in the soil (Oldroyd 1969) and be parasitoids of cocoons and predators of eggs of locusts (Acrididae: Dociostaurus maroccanus in Spain and Algeria, see Künckel d`Herculais 1893–1905, 1894; Delassus et al. 1929; Del Cañizo 1943; Arcyptera microptera in various parts of Russia and at Crimea, see Mokržeckij 1895; Bezrukov 1922; Porčinskij 1894, 1895; Podisma pedestris in Russia, see Porčinskij 1894, 1895, and Schistocerca gregaria in Algeria, see Sergent 1916) and other grasshoppers (Pamphagidae: Ocneridia volxemii in Algeria, see Künckel d`Herculais 1893–1905, 1894). Some butterfly species are probably parasited too (Noctuidae: Apamea anceps, see Engel 1932–37; Grigorieva 1958; Zaitsev 1966; Tomaj 1977; Greathead and Evenhuis 1998; Motyčková 2012). Reports from the Netherlands and Britain describe Th. fenestratus as parasitizing Ammophila spp. larvae (Pontin 1961; Miles and Muggleton 2010; Noordijk et al. 2016; Slikboer et al. 2019). More recently, abundant Th. fenestratus larvae in nests of the crabronid wasp Pemphredon fabricii were reported (Bogusch et al. 2015, 2018). In contrast, Th. fenestratus larvae were rarely reported in nests of the crabronid wasp Trypoxylon deceptorium Antropov, 1991 (Bogusch et al. 2015). The larvae are thought to be generalists attacking the nests of various Crabronidae and Sphecidae (Bogusch et al. 2018). For an unclear reason, Th. fenestratus host specialization has some local consequences. Although P. fabricii, Tr. deceptorium, and Th. fenestratus distribution ranges overlap in a large part of Europe, the P. fabricii and Tr. deceptorium nests attacked by Th. fenestratus have been reported only in the Pannonian lowland, despite intensive research having been conducted in other regions of occurrence of both these species (e.g., Bohemia, i.e., western part of Czechia) (Bogusch et al. 2018). In Czechia, Th. fenestratus adults are active from mid-June until mid-August (Motyčková et al. 2012; Čelechovský et al. 2020). Adults are typically found in sun-exposed forest ecotones, dirt roads (Čelechovský et al. 2020), heathlands (Drake 1991; Lake 2002; Čelechovský et al. 2020), and other dry sandy habitats (Noordijk et al. 2016; Slikboer et al. 2019), where they visit flowers and feed on nectar and pollen.

The bee fly Th. fenestratus was abundant in old common reed galls induced by Lipara lucens and has been previously reported from the tailing pond of ash and slug from the lignite powerplant near Hodonín (Czechia; 21% of 29 examined P. fabricii nests were parasitized) and at a fishpond with sandy bedrock near Sekule (Slovakia; 24% of 89 examined P. fabricii nests were parasitized) (Bogusch et al. 2015; Heneberg et al. 2022). They were abundantly present near Hodonín, even though this sampling site was contaminated with arsenic (40 µg g^-1), copper (26 µg g^-1), and sulfur (4616 µg g^-1) (Heneberg et al. 2022). When focusing on Lipara-induced common reed galls, Th. fenestratus was most abundantly associated with galls 10–14 mm in diameter (also preferred by P. fabricii) and present more abundantly in galls on the stems of small or intermediate thickness (also preferred by P. fabricii). These stems typically occur in reed beds less stressed by drought or other factors (Astapenková et al. 2017). The larvae of Th. fenestratus in galls induced by L. lucens frequently suffer from fungal infections (Heneberg et al. 2016).

The mature larva of Th. fenestratus was described by Séguy (1932); a drawing of the mature larva was also published by Oldroyd (1969). However, more recently, Bogusch et al. (2015) reported deviations from the original descriptions. Séguy (1932) and Oldroyd (1969) described the larval mandibles of Th. fenestratus as bidentate, with lateral hooks, which is in contrast to the findings by Bogusch et al. (2015) and to other larval Bombyliidae (Oldroyd 1969). Prominent sharp mandibles with three teeth oriented backward were present (Bogusch et al. 2015). Mature larvae reach 4.9 ± 0.8 mm in length. The development and immature larvae of Th. fenestratus have been documented by Künckel d`Herculais (1893–1905). There is a report of Th. fenestratus rearing from naturally infected Ammophila pubescens Curtis, 1836 cocoons (Miles and Muggleton 2010). However, as A. pubescens larva spins cocoons, direct observation of their development was impossible. Nevertheless, the above report was the first proof that Th. fenestratus allows parasitized larvae to complete their growth, feed on the food supplies provided by the parent wasp, and to spin a cocoon. The cocoon was not intact, and therefore, the authors were uncertain whether the larva was attached to the host larva during the whole time or whether it had eaten some of its host`s provisions, bored into the cocoon from below after the cocoon was formed and then only consumed the wasp prepupa or pupa (Miles and Muggleton 2010).

In the present study, we document the development of Th. fenestratus on naturally infected P. fabricii prepupae under laboratory conditions. The larvae, including host prepupae, were exposed to various field-realistic concentrations of azole fungicides that are commonly used in agriculture (ECDC 2013), namely, difenoconazole, penconazole, and tebuconazole. These azoles have been shown to affect the timing of metamorphosis into adults of the host species (Heneberg and Bogusch 2022), but their effects on Th. fenestratus and dipterans, in general, are unknown. Selected other azole fungicides were previously documented to kill dipterans (Kenneke et al. 2009; Saraiva et al. 2018; do Prado et al. 2023). Development was documented, and differences in its timing were tracked.

To analyze the geographic distribution of the Th. fenestratus population parasitizing P. fabricii, we considered the data on reed galls induced by L. lucens. These consisted of 176 reed beds in Czechia, Poland, Hungary, Slovakia, Austria, Slovenia, and Italy. We included only those reed beds in which the reed galls induced by L. lucens were positive for P. fabricii. The list of coordinates of examined reed beds that were positive for P. fabricii is provided in Suppl. material 1. Numbers of P. fabricii obtained by cutting the galls, numbers of Th. fenestratus obtained using the same procedure, and the total numbers of galls cut are indicated for sampling sites that were Th. fenestratus-positive. We did not show the numbers of P. fabricii and Th. fenestratus obtained from galls from which the inquilines were allowed to rear but were not cut individually, as Th. fenestratus often remain in the galls after rearing in rearing bags, and the detected parasitation intensities would, therefore, be underestimated.

For rearing analysis, we collected prepupae of P. fabricii parasitized by Th. fenestratus at sampling sites Hajnáčka (48.2126403°N, 19.9546442°E; 20 Jan 2022) and Šíd (48.2660036°N, 19.8795681°E; 21 Jan 2022). Both localities are in the Cerová vrchovina mountains in southern Slovakia. The collected nests were present in ≥1-year-old common reed galls induced by L. lucens.

When larvae of P. fabricii are 7–20 days old, they stop feeding and line up in the nest one after the other in a row. All of them defecate, shed to become a prepupa, and build a cocoon from their silk on the inner side of the gall, without septa between the larval chambers. Therefore, when collected in January, the reed galls contained defecated fully grown prepupae in cold-induced diapause (Bogusch et al. 2018). The prepupae require only a few days (usually four to six) to pupate and develop synchronously to adults when exposed to the laboratory temperature.

We cut all the collected galls on 22–23 Jan 2022 and kept the excised prepupae at 4 °C until placed in 96-well plates (Brand, Wertheim, Germany) on 25 Feb 2022. We placed the prepupae individually in the wells and allowed them to acclimate to their new environment for 24 h at 23 °C and > 90% humidity. We performed the experiment together with exposure of the nonparasitized prepupae of P. fabricii to the same compounds (see the analysis published in Heneberg and Bogusch (2022) for more details). On 26 Feb 2022 (Day 0), we exposed the parasitized prepupae to three triazole formulations, each at a concentration equal to or lower than those recommended by their manufacturers for use in spraying crops to eliminate fungi. The tested compounds were penconazole (Merck, Darmstadt, Germany; batch BCBZ4909, 99.0% purity), difeconazole (Merck; batch BCCD4900, 95.5% purity), and tebuconazole (Merck; batch BCCF9398, 99.1% purity). Before the final dilution, all compounds were first diluted at 50 mg mL^-1 in dimethyl sulfoxide (DMSO), and DMSO alone was therefore used as a control. We sprayed the compounds using Potter Precision Laboratory Spray Tower (Burkard Scientific, Uxbridge, UK). The application rate of the applied solutions was constant at 2 μL cm^-2.

On Days 4 through 48, following the application of triazole fungicides, we checked the treated digger wasps at least every other day for molting, wing development, and mortality (with a break between Days 32 and 42). During the experiment, we kept the digger wasps at 23 °C with >90% humidity. We recorded the time until the full development of wings and mortality.

We found Th. fenestratus to parasitize P. fabricii at 15 of the 176 reed beds that were positive for P. fabricii. These sites were located mainly in Slovakia (6) and Hungary (4), but several sampling sites were also positive for Th. fenestratus in southeastern Czechia (3), eastern Italy (1), and eastern Austria (1). Except for the single sampling site at the border of Italy and Slovenia, all other Th. fenestratus-positive P. fabricii sampling sites were located in the Pannonian lowland. The restriction of Th. fenestratus-positive P. fabricii sampling sites to the Pannonian lowland is surprising given that the distribution ranges of both examined species are much broader and many other sampling sites were located in regions where both species are present (Fig. 1). The Th. fenestratus prevalence at Th. fenestratus-positive P. fabricii sampling sites was 11.6%±4.1% (mean±SE), max 52.9% (reed beds of Neusiedler See near Oggau, Austria), min 0.8% (sampling site Tát, Hungary).

Figure 1. 

Map of examined sampling sites that were positive for P. fabricii (circles). Empty circles indicate sampling sites where only P. fabricii was found. Green circles indicate sites where P. fabricii was parasitized by Th. fenestratus. Source data are available in Suppl. material 1.

At sampling sites where P. fabricii was parasitized by Th. fenestratus, two asynchronous populations of this dipteran were present. First, some larvae were overwintering as fully grown; we included one of them in the examined dataset (Fig. 2, P-400-D8). These overwintering larvae were previously found in P. fabricii nests at other sampling sites as well (e.g., Hodonín and Sekule mentioned in the Introduction (Bogusch et al. 2015; Heneberg et al. 2022)) and represented the previously described populations parasitizing this host. To date, it is unclear whether Th. fenestratus is a true parasitoid or whether it eats part of its host`s provisions (Miles and Muggleton 2010). Moreover, as host use in Bombyliidae varies, it was unclear whether these parasitoids act as koinobionts, allowing some further development of infected hosts, or idiobionts, which kill and consume their host in the state in which it is attacked (Yeates and Greathead 2008).

Figure 2. 

Time-lapse evidence of the development of Th. fenestratus under laboratory conditions following cold-induced diapause. The X-axis indicates days elapsed since the termination of cold-induced diapause. Individual lines represent photographs of the same individual. The individual in the second line spent the cold-induced diapause as a fully grown larva; all others spent the winter attached to the defecated prepupa but did not grow before the cold-induced diapause. Letters indicate the treatment (DMSO = control, P = penconazole, D = difeconazole, T = tebuconazole; numbers (200, 400, 500, 800, 1200, 2000) indicate the dilution of the study compounds (for more details, refer to Heneberg and Bogusch 2022); letters with numbers indicate the positions in 96-well plates.

A part of the Th. fenestratus population, however, adopts a different overwintering strategy. A planidium finds the larva or prepupa of its host and overwinters on the defecated host prepupa. We did not observe the planidium directly, but when we sorted out individual overwintering P. fabricii prepupae, tiny attached Th. fenestratus larvae were already present on them. The attached larvae induce a block of the development of parasitized prepupae. Under laboratory conditions, the P. fabricii prepupae mostly (>80%) developed into pupae four days after removal from cold conditions. Nearly all (>98%) developed into pupae at six days following removal from cold conditions (Heneberg and Bogusch 2022). All the parasitized prepupae remained blocked at this developmental stage – see the illustration figure of one of the analyzed 96-well plates at Day 10 (Fig. 3). The nonparasitized prepupae pupated at the indicated time and metamorphosed to adults at Days 12–14.

Figure 3. 

Representative figure of the 96-well plate at D10 with typically developing P. fabricii (already pupated, two to four days ahead of the metamorphosis to the adults) treated with penconazole (50 μg mL⁻¹) and with prepupae in wells D7, D8, and G5, which are parasitized by Th. fenestratus. The parasitized prepupae did not pupate and remained paralyzed and were killed within the ensuing four days.

We observed the development of 11 individuals spending diapause as first- or second-instar larvae. Seven completed their development successfully, whereas four failed to develop. Two of those that failed to develop had issues with the integrity of their paralyzed prey – the prey changed color and later was overgrown with mold, which also killed the Th. fenestratus larvae. In the other two cases, the fully grown larvae failed to pupate, became infected, and died (Fig. 2).

A single host prepupa is sufficient to support the development of a single Th. fenestratus larva. The time to pupation was 12 days in the case of the larva that was fully grown at the end of diapause. The seven individuals that grew out only after diapause pupated at Days 18, 20 (3×), 22, 24, and 26 (21.4±1.0 days, median = 20 days) (Fig. 2).

All seven individuals that spent diapause as a first- or second-instar larvae and completed subsequent pupation also succeeded in metamorphosis to adults. Additionally, the single individual observed as overwintering as fully grown succeeded in metamorphosis to adult. It took 20 days for the latter individual to metamorphose from pupa to adult. It took the individuals spending diapause as the first- or second-instar larvae and completed subsequent pupation 20 (2×), 21, 22 (2×), and 24 (2×) days (21.9±0.6 days, median = 22 days) to metamorphose from pupa to adult (Fig. 2). Representative images illustrating critical events in the development of Th. fenestratus on overwintering prepupae of P. fabricii are shown in Fig. 4.

Figure 4. 

Representative figures illustrating critical events in the development of Th. fenestratus on overwintering prepupae of P. fabricii. In bombyliid flies, the first stage, triungulin, is active and immediately molts after attaching to its host to the second stage (likely D6), which has already attached itself and is not distinctly curved. This stage can overwinter (as does the pupa). Early third instar (likely D8) and late third instar (likely D14) can be diagnosed by their characteristic curvature. D24 and D34 represent pupae.

A long-standing question is whether the larvae of bombyliid flies paralyze and kill the host immediately or at the very late stages of feeding. Adult females lack features that would allow them to paralyze the host. Therefore, it is believed that most bombyliid fly larvae contact the host in the state in which it is consumed (as typical idiobionts) and that the host is not immediately paralyzed or killed (which would be a koinobiont strategy) (Yeates and Greathead 2008). Evidence for other bombyliid species suggests that the host is killed only in the final stages of parasitoid development. However, here, we show clear evidence that parasitic larvae induce a development block of their host prepupae at very early stages, at the latest immediately after diapause. Therefore, while nonparasitized prepupae need only four days to pupate after diapause, the parasitized prepupae never pupate and remain alive and paralyzed for several weeks, which only allows for the complete growth of the parasitic larva until pupation. Correspondingly, reports of some Anthrax spp. claim that their larvae may adopt different strategies when parasitizing tiger beetle (Coleoptera: Cicindelidae) larvae. While they parasitized any tiger beetle larval instar, they remained in a second instar for up to nine months, depending on the host stage and time of year. Once the tiger beetle larva constructs a pupal chamber and becomes a prepupa, the Anthrax larva begins rapid growth and reaches full size in one or two weeks (Palmer 1982; Shelford 1913). Thus, the larva appears to behave as a koinobiont or idiobiont based on the time of the year and host developmental stage. Given that two stages of Th. fenestratus larvae are found in Lipara-induced reed galls, this parasitic species appears to employ a similar strategy. Whether the first-instar larva feeds on the aphid provisions remains unclear, as some other bombyliid species are reported to feed on provisions and develop only on the prepupa or pupa (Bohart 1960; Du Merle 1979). Regarding our experiments with Th. fenestratus, we removed all P. fabricii prepupae from their cocoons before beginning experiments. Thus, the parasitic larvae must have been attached to them during the cold-induced diapause.

We exposed P. fabricii prepupae with Th. fenestratus larvae to three topically applied fungicides before their growth. However, we did not observe any adverse effects of exposure to the indicated fungicides on the bee flies. The larva treated with DMSO (control) died during development because of fungal infection. There were no concentration-dependent differences in the development and survival of the larvae treated with difeconazole and penconazole, including a lack of effects at the highest concentrations used (Fig. 2). The only unclear result is with regard to the single tebuconazole-treated larva, which died after it completed its growth, just before pupation. However, more larvae treated with the same compound are needed to provide a definitive answer on tebuconazole’s effects on this dipteran.

In conclusion, we found that Th. fenestratus facultatively switches between killing the host almost immediately (idiobiont strategy) and killing the host at a later developmental stage (koinobiont strategy). We documented the koinobiont parasitoid strategy in detail using a series of Th. fenestratus larvae parasitizing their hymenopteran host P. fabricii, which nests in Lipara-induced reed galls. We found that a significant portion of Th. fenestratus larvae overwinter as young larvae and start feeding on fully developed and defecated prepupae of P. fabricii only after the end of cold-induced winter diapause. The time needed for the development of Th. fenestratus larvae exceeds several times the time needed for pupation of P. fabricii prepupae; the parasitized prepupae, therefore, remain paralyzed until the parasitic larva completes feeding.

The study was supported by the Czech Science Foundation project 23-07303S and the University of Hradec Králové, Faculty of Science project Specific Research 2101/2022. We thank Wojciech Pulawski and Neal Evenhuis for their helpful and constructive peer-review comments.