Introduction

Many animals use distinctive odorants to ward off predators. Among insects, the repellents of Pentatomidae (Hemiptera), Carabidae (Coleoptera), and Coccinellidae (Coleoptera) are particularly well known. As Sasakia charonda (Hewitson, 1863) (Lepidoptera: Nymphalidae: Apaturinae) is the largest nymphalid species in terms of body size in Japan and its neighboring countries, some researchers have a special interest in this species. Sasakia charonda has been bred since 2012 in cages in the Kashihara City Museum of Insects, Kashihara, Nara, by Taro Hayashi.

While manually transferring S. charonda larvae to fresh leaves of Celtis sinensis (Rosales: Cannabaceae), the larvae released foul-smelling volatile compounds from their mouths without emitting a fluid orally. Moreover, in response to handling, they raised their heads, opened their mandibles, and made clicking sounds, resembling “Kachi, Kachi”, to threaten their predators. Furthermore, some larvae made movements as if they were trying to bite the predator. Many ants live in the same cage, and Hayashi (pers. comm.) also observed that the larvae showed similar behavior toward the ants. Specifically, S. charonda larvae were observed to successfully repel ant attacks by turning the opened mandible toward the ant's head (Fig. 1). Moreover, Kaori Holikawa (pers. comm.) verified the same behavior in Hestina assimilis (Linnaeus, 1758) and H. persimilis (Westwood, 1850) (Lepidoptera: Nymphalidae: Apaturinae), as well as in other Japanese Apaturinae species. Thus, we aimed to conduct a chemical analysis of the compounds released by these Apaturinae larvae. Additionally, in the field, the fourth author observed that H. assimilis fifth (final) instar larvae succeeded in repelling Polistes chinensis (Fabricius, 1793) (Hymenoptera: Vespidae) by opening their mandibles. Thus, these oral odorants may also be useful in evading attack by certain wasps. We, therefore, also aimed to confirm whether these compounds are effective deterrents to ants.

Fig. 1
figure 1

A second instar larva of Sasakia charonda that successfully repelled an ant. a Ant approached the larva to capture it. b Larva turned the opened mandible toward the ant’s head, and ant’s antenna touched the larva’s mandible. c Ant responded by retreating quickly

Full size image

In addition, during oral odorant sample collection, we noticed that the behavior to emit volatiles orally was transmitted to neighboring larvae. Based on this observation, the odorants may also act as an alarm pheromone, and we examined whether these substances elicited the same effect in S. charonda larvae.

Materials and methods

Collection of oral odorant samples from Japanese Apaturinae larvae

The first experiments were conducted at the Sasakia charonda Research Center, Kashihara, Nara. Sasakia charonda larvae were reared in a 6.5 m × 5 m × 3 m (L × W × H) cage covered with a 1-mm nylon mesh net and fed with fresh C. sinensis leaves. We also used S. charonda larvae reared in the Museum of Sasakia charonda in Hokuto, Yamanashi. The sources, dates, origins, and additional information of the butterflies used in these experiments, and the origin of larvae, sampling dates, larval stages, sampling temperature (°C), individual numbers (n), and sampling durations are shown in Table 1. Oral odorant samples of the 4th–6th instar larvae of S. charonda in butterfly cages were collected in sampling tubes (Tenax TA, GL Sciences, Japan) using an air suction pump with an initial suction capacity of approximately 20 L/min, as shown in Fig. 2. Oral odorant samples of the 4th–5th instar larvae of H. persimilis and A. metis were transported from their origin to the power supply point, and then collected in sampling tubes using the same air suction pump. For the collection of oral odorant samples from H. assimilis larvae, a manual pump was used, because the larvae of H. assimilis in Honshu are prohibited from being moved from their place of origin, and we could therefore not transport them to the power supply point. The collection time for each larval group is shown in Table 1. The air just above the fresh leaf surface of host plants—C. sinensis for S. charonda, H. assimilis, and H. persimilis, and Salix sp. (Salicaceae) for A. metis—was used as the control.

Table 1 Origin, collecting date, and additional information of caterpillars used
Full size table
Fig. 2
figure 2

Collection of oral odorant samples from the sixth instar (final) larvae of Sasakia charonda

Full size image

Analysis of volatile substances from host plant leaves

For this analysis, we collected 1.7 g of leaf material from both C. sinensis and S. babylonica var. babylonica. Each collected leaf was preserved in a 50 mL polyethylene tube and crushed using a medicine spoon. Then, the volatiles from leaves were trapped and extracted into a sampling tube for 30 s, as above. Trapped volatile substances were analyzed in the same way as the larval odorants.

Analysis of volatiles

Gas chromatography–mass spectrometry (GC–MS) analysis of the samples was performed in the Analyzing Center, Showa Denko Materials Techno Service, Hitachi, Japan. The oral odorant compounds were subjected to GC–MS analysis on an Agilent system consisting of a model 7890B gas chromatograph, a model 5977A mass-selective detector (EIMS, electron energy of 70 eV), and an Agilent ChemStation data system (Santa Clara, CA, USA). The GC column was a DB-VRX column (Agilent, USA) with a film thickness of 1.40 μm, length of 60 m, and internal diameter of 0.25 mm. The carrier gas was helium (He), supplied at a flow rate of 2.1 mL/min. The collected volatiles were analyzed using the thermal desorption method. The GC oven temperature program was as follows: 3 min at 40 °C, increasing 5 °C/min up to 260 °C, then 8 min at 260 °C. Sample components were identified by comparing their mass spectral-fragmentation patterns against those stored in the MS library (NIST14 database) and with reagents. The mass acquisition range is m/z 20 to 500. When we re-used these sampling tubes, they were re-initialized using the TC2-Tube Conditioner (GERSTEL K.K., Tokyo) under the following conditions: heating temperature 300 °C, heating time 2 h, purging gas N2, 50 mL/min. These were manualized by GERSTEL HP.

Reagents

We detected 22 substances in the oral odorants of S. charonda larvae. Among these, we were able to obtain and purchase 19 of those reagents. We also purchased reagents detected in samples from the three other species used in the same investigation. These are shown in Table 2 and were used to confirm the substances that we detected; they were also used in the behavioral testing of ants and larvae described below.

Table 2 Volatile substances detected and identified in the oral odorants of four Japanese Apaturinae species
Full size table

Calculation of the retention index of the detected substances

We could not find the retention index (RI) for DB-RVX, and therefore, we calculated the RI of the substances that we detected. First, we analyzed a mixture of pentane, heptane, and analytical standard, containing C8–C20, approximately 40 mg/L each, in hexane (MERCK), to determine the retention time of these standard substances. Next, mixed reagents were purchased and analyzed in the same manner as the oral odorant samples, and the RI of these reagents was calculated. The RI of other substances was estimated from their RI values before and after detection.

For this calculation, we used the equation shown below.

$$ I_{{\text{A}}} = {1}00 \times n + {1}00 \times \left( {{\text{log}}t_{{\text{A}}} {-}{\text{log}}t_{n} } \right)/\left( {{\text{log}}t_{n + 1} {-}\log t_{n} } \right), $$

where n is the carbon number of n-H-(CH2)n-H, tn is the retention time of n-alkane, and tA is the retention time of the substance whose RI was calculated; each retention time is located between tn+1 and tn.

Effect of oral odorants on predatory ants

This experiment was performed in the Sasakia charonda Research Center, Kashihara, in Nara, and the Museum of Sasakia charonda in Hokuto, Yamanashi. The effects of the 19 confirmed substances on Pristomyrmex punctatus (Smith, 1860) and Formica japonica Motschoulsky, 1866 (Hymenoptera: Formicidae) were investigated. Except in one case in Setagaya, Tokyo, Aug 1, 2020, where the experiment was carried out in the field, these ants’ nests were in butterfly cages where the ants constantly interacted with the butterfly larvae.

First, we attempted the following experiment. One side of a silicone tube was connected to the outlet of an air pump, while the other side of the tube was connected to polyethylene pipettes. Small pieces of filter paper (5 mm × 20 mm) soaked in 0.2 mL of each volatile were inserted into each pipette, to serve as a source for each volatile substance. Following that, vapor containing the substance (or without it, serving as the control) was blown onto the heads of examined individuals (hereafter, we refer to this as "Method A"). However, in this method, ants were frequently blown away by the vapor stream. Thus, we soaked a cotton swab in 0.2 mL of each substance and placed the swabs individually near the ants. We then observed whether the ants showed avoidance behavior of the swabs soaked with the sample (hereafter referred to as “Method B”). The method of assessment is described in the footnote of Table 3.

Table 3 Responses of ants to chemical samples
Full size table

Effect of oral odorants on predatory wasps

The instrument used in this experiment is shown in Fig. 3. The suction pump was continuously active throughout the experiment to prevent the volatile substances used from persisting in the glass tube containing the Polistes wasps. The method of assessment is described in the footnote of Table 4.

Fig. 3
figure 3

Instrument used in the experiment investigating wasp behavior. a Tube connected to a funnel for air intake and odorant substances. b Glass or PET tubes to enclose wasps. c: Tube connected to the suction pump

Full size image
Table 4 Responses of wasps to chemical samples
Full size table

Effect of oral odorants on the larvae of the four Apaturinae species

The oral odorants identified were presented to the four Apaturinae larvae at the Sasakia charonda Research Center, Kashihara, Nara, the Museum of Sasakia charonda in Hokuto, Yamanashi, and the Park Parada Saku, Nagano. In some cases, both Methods A and B were applied to the same individuals. We observed the behavior of the larvae in response to these substances. Almost all behavioral responses of ants, wasps, and larvae were recorded using a video camera (Standard resolution NTSC video, 720 × 480 pixels, HDR-HC7, SONY, Japan).

Results

Identified oral odorants and their retention indices

The individual substances detected in the oral odorants of the larvae of the four species are summarized in Table 2. We were able to identify the following 13 volatiles from S. charonda: 2,3-butanedione, 2-butanol, 2-methyl-3-buten-2-ol, 1-penten-3-ol, 3-pentanol, 3-pentanone, 3-hydroxy-2-butanone, 3-methyl-3-buten-1-ol, (±)-2-methyl-1-butanol, 2,3-butandiol, 3-methyl-2-butenal, 3-hexen-1-ol, and cyclohexanol. None of these volatiles were detected in the control air samples. The compounds, 1-penten-3-ol, 1-penten-3-on? and 3-hexen-1-ol?, were detected as volatile substances from the pulverized leaves of Celtis sinensis and Salix babylonica var. babylonica, and were not considered to be the important components in larval oral odorants. In contrast, 3-hexen-1-ol and 2-hexenal, which are thought to be the main components from crushed leaves, could not be detected in larval oral odorants. There seemed to be no relationship between oral odorants and leaf volatiles.

We also analyzed the oral odorants of the other three Apaturinae butterfly species in Japan (Table 2). Among these, 3-methyl-2-buten-1-ol, hexanal, benzaldehyde, 6-methyl-5-hepten-2-one, nonanal, and (E)-geranylacetone were confirmed using reagents. The calculated RI of each substance is shown in Table 2.

Effect of oral odorants produced by S. charonda on predator ants, wasps, and larvae of four Apaturinae species

Results of ant responses were obtained from ten nests of P. punctatus and five nests of F. japonica that were tested, which showed contrasting responses to different oral odorants produced by S. charonda, as well as temporal and/or spatial variation in their responses to a particular compound (Table 3). These ants were strongly repelled by 2,3-butanedione, 2-butanol, 2-methyl-3-buten-2-ol, 1-penten-3-ol, 3-pentanol, 3-pentanone, and 3-hydroxy-2-butanonea.

We obtained response results from 24 Polistes jokahamae Radoszkowski, 1887, three P. nipponensis Saussure, 1858, and 12 P. chinensis during Apr and Sep 2022 (Table 4).

The responses of the larvae of four species of Apaturinae are summarized in Table 5 (S. charonda) and Table 6 (other three Apaturinae species). We could obtain results from 63 S. charonda, 11 H. assimilis, 19 H. persimilis, and 14 A. metis larvae between Jun 2020 and Nov 2022.

Table 5 Responses of Sasakia charonda larvae to chemical samples
Full size table
Table 6 Responses of other three Apaturinae larvae to chemical samples
Full size table

Positive responses of the larvae to the substances were based on a specific behavior; that is, the larvae raised their heads from the host leaf and opened their mandibles toward the pipette or swab. This behavior was only displayed by individuals that were resting on a scaffold formed by themselves during the larval stage (Figs. 4, 5). All individual ants, wasps, and larvae responded almost instantly if the substances had an effect, and did not respond to only air or a dry swab (Fig. 6).

Fig. 4
figure 4

Responsive behavior of ants (Pristomyrmex punctatus) to the volatiles. Setagaya, Tokyo, Aug 01, 2020

Full size image
Fig. 5
figure 5

Non-responsive behavior to the substances. A For Sasakia charonda larvae. B For larvae of Hestina assimilis and the other two species

Full size image
Fig. 6
figure 6

Hostile response of Sasakia charonda larvae to the examined volatiles

Full size image

Discussion

Although S. charonda is popular among butterfly breeders in Japan, it is likely that oral odorants emitted by S. charonda and the three other Apaturinae larvae had not been previously identified because breeders rear the larvae in the absence of predators. Among butterflies, Papilionidae larvae have a defensive tongue-like organ known as the osmeterium, which secretes a mixture of strong odorant substances that repel ants (Honda 1983a) and wasps (Hirose 1981). However, the emission of such defensive substances has not been reported in other butterfly families before (Fig. 7).

Fig. 7
figure 7

.

Full size image

We could not collect oral odorants from some individuals, because they did not consider us as predators and did not open their mandibles in response to our handling of them. It is well known (at least in Japan) that Papilionidae larvae that are handled often by their breeders become accustomed to them, and, eventually, they do not extend their osmeterium in response to handling. It appears that the larvae no longer consider the breeders a threat. Thus, we assumed that Apaturinae larvae also have this learning ability. We also noticed that non-responsive individuals were in the process of walking or eating, activities during which their aggression may be reduced.

The chemical composition of the oral odorants produced by H. assimilis and H. persimilis was similar to that of S. charonda, but that of A. metis was distinct from the other three species. Despite some similarities between S. charonda and the two Hestina spp., the latter mainly contained edulan-group substances instead of low-molecular-weight alcohols and ketones. In contrast, the chemical composition of A. metis only contained 2-butanol and benzaldehyde, although the former compound was also found in the oral odorant samples of the other three Apaturinae larvae. These differences may occur because the larvae utilize different host plants.

Another reason might be the phylogenetic relationships of these species. However, as a phylogenetic study (Ohshima et al. 2010) on Apaturinae species did not include S. charonda, we cannot draw a conclusion regarding the relationship between oral odorant composition and phylogeny. We could conclude that as the control samples taken from above the host plant leaves did not contain these chemicals, they must be synthesized in the larval body de novo. The same has already been confirmed for the components of the osmeterium in Papilionidae butterfly larvae (Honda 1983b).

Considering the reactions to each substance, most of the individuals used in the experiment (ants, wasps, and Apaturinae larvae) responded to 2,3-butanedione, 2-butanol, and 3-methyl-2-buten-1-ol. Of these, 2-butanol was detected in all four Japanese Apaturinae species, and 2,3-butanedione was detected in S. charonda, H. assimilis, and H. persimilis. It can be proposed that these two substances play a major role in repelling predatory ants and wasps in the larvae of these four Japanese Apaturinae. However, 3-methyl-3-buten-1-ol and 2-methyl-1-butanol could only be detected in S. charonda. These substances were not listed in Knudsen et al. (2006). In contrast, benzaldehyde and 6-methyl-5-hepten-2-one were detected from Salix flowers by Füssel et al. (2007) and Tollsten and Knudsen (1992), respectively (6-methyl-5-hepten-2-one was in trace amounts). Benzaldehyde, 6-methyl-5-hepten-2-one were also not listed in previous studies by either Füssel et al. (2007) or Leme et al. (2020). However, a considerable number of individuals used in the experiment responded to 2,3-butanedione, 2-butanol, and 3-methyl-2-buten-1-ol., which may have been included in unidentified RI peaks. The results of H. persimilis were different from those of other species, which is likely due to the small number of experimental individuals by stage, and the small number of total examinations, necessitating a follow-up test. However, Hayashi observed that larvae of S. charonda were successful in repelling an attack by a small hunting spider (Araneae). Thus, these oral odorants may also be effective against smaller generalist predators.

In this study, we did not use a large number of specimens. In the case of ants, all individuals used from each institute were the same nest members, and the genetic diversity may be poor. In the case of S. charonda, although individuals are morphologically similar, they have been bred for many generations in the same institute, and the genetic diversity of each institute is also poor. Thus, in ants and S. charonda, we did not try to increase the number of individuals held at any one institute but instead increased the number of sites where we performed our experiments. In the case of the other three Apaturinae species, they are not established as experimental insects. However, we could collect several larvae of these species in the field and perform our experiments. Moreover, we collected a number of wasps in the field and successfully completed our experiments.

A comparison between Method A and Method B for the same individual larva showed a more sensitive response to Method A than to Method B. This indicates that the wind generated by the wasp wing, not just the odor response, may be involved in the defensive behavior of the larvae.

A comparison of the results of S. charonda and A. metis revealed that the response of the larva becomes stronger as it ages. It has been pointed out that the final instar is the most vulnerable period for larvae from wasp attack, and it is understandable that the response becomes stronger at this time. Although the larvae of Papilionidae project the osmeterium even during the first instar stage, Apaturinae larvae appear to differ from Papilionidae larvae by having a reduced capacity for oral odorant production in their early instars. Considering the individual responses of the larvae, individuals that responded strongly to one substance tended to respond to many other substances too. As the genetic differences between the individuals of each institution used in the experiment do not seem to be large, these individual differences were considered to be linked to differences in their physiological condition. The results from Ōiso, Kanagawa, Aug 15, 2022, suggest that the sensitivity was considerably lower at the prepupal stage. The weak response of individuals after October may be the result of the drop in temperature, and it is also possible that various individuals began to prepare for overwintering and would have altered metabolic tempos as a result.

The defensive substances in the osmeterial emissions of Papilionidae species have been studied (Ômura et al. 2006), and their effects on two species of ants in Japan have been analyzed in detail (Honda 1983a). According to this previous study, these substances mainly consist of monoterpenoids, sesquiterpenoids, aliphatic acids, and their esters. Thus, the oral odorant components of the four Japanese Apaturinae larvae are very different from those of Papilionidae. As shown in Table 2, the defensive odors of the larvae of S. charonda and the three related species are mainly derived from C4–C5 alcohols. The synthesis pathways are unknown, and the mechanisms involved merit further investigation. Further, Kandori et al. (2022) reported that the long horn-like projections on the heads of Hestina japonica (Felder, 1862) butterfly larvae protect them from their natural enemies, but they did not make any mention of oral odorants produced by these larvae. As above, Hayashi and Holikawa were able to perceive the smell, we wonder as to why Kandori et al. (2022) could not notice these odorants.

The odor-detecting systems of some ants have been researched morphologically and electrophysiologically (Hashimoto 1992; Ozaki et al. 2005; Ruchty et al. 2009; Sharma et al. 2015), but this research was limited to the substances that are involved in cuticular wax. The olfactory system of ants should still be researched to assess their responses to additional substances. Our own results for ants responses were inconsistent. This might be related to the fact that ants will assess individuals as probable nest mates using very minute features of their cuticular hydrocarbon pheromones (Ruchty et al. 2009). However, among the chemicals in larval oral odorants, 2-butanol induced a repellent behavior in almost all of the ants examined. As 2-butanol was the only substance detected in all four Japanese Apaturinae larval oral odorants, this chemical must be the key repellent. Notably, these substances may also affect Apaturinae larvae by serving as alarm cues. Research on the olfactory responses of Lepidoptera larvae has only been performed with substances emitted from their host plants (e.g., Dethier 1973; Van Loon 1990; Malmgren 2011; Rharrabe et al. 2014; Asaoka and Shibuya 1995). Our results indicate that the larvae are also sensitive to substances that they themselves produce and emit. Thus, our results may enrich the study of Lepidoptera olfaction. Many wasps responded to 3-methyl-2-buten-1-ol, benzualdehyde, 6-methyl-5-hepten-2-one, Nonanal, and (E)-geranylacetone, but considering that these substances are found in the volatiles produced by flowers, these responses seem to have a distinct significance toward 2,3-butanedione, etc. In fact, all five Polistes jokahamae and one P. nipponensis used on Sep 26, 2022, and Oct 2, 2022, respectively, which were collected on Cayratia japonica (Vitaceae: Vitales) flowers, showed notable responses to the substances found from flower scents (Knudsen et al. 2006). Sasagawa et al. (1989) showed that JH is implicated in the physiological regulation responsible for the age-linked division of labor or polyethism in Apis mellifera Linneaus, 1758 (Apidae: Hymenoptera); such a phenomenon may also exist in wasps. Although some of the tested individuals were observed to respond to hexane, we also considered that 2,3-butanedione and 2-butanol influenced them, as those trials were performed just prior to hexane exposure. In this experiment, we examined the reaction to each substance in the order of elution of the substance by GC. Actually, in the experiment performed on Nov 15, 2022 (Column j of Table 5 and Column j of Table 6), we first examined the reaction to hexane. Further experiments with electro-antennograms, as well as with ants, would be necessary to validate this.

The larvae of Apaturinae, Charaxinae, Boblinae, and Satyrinae species have a pair of horns on their heads, which they use defensively when attacked by a predator (Teshirogi 2016). Nymphalinae, Heliconinae, and Limenitinae larvae have many spines on their body, and these can be used for physical defense against their predators. Moreover, Heliconius larvae have a chemical defense system that is also found in Danainae and Ithomiinae larvae. In contrast, an effective chemical defense system is not known in Charaxinae, Boblinae, and Satyrinae larvae. In the present study, we showed that the four Japanese Apaturinae larvae have a chemical defense system, raising the question of whether Apaturinae, Charaxinae, Boblinae, and Satyrinae larvae found outside Japan, which are morphologically similar to Japanese Apaturinae larvae, also bear such a chemical defense system. This could be a direction for future research.

In our study, we could not obtain sufficient numbers of H. assimilis, H. persimilis, or A. metis larvae, because a method of mass-breeding of these three species, such as that used for S. charonda, has not yet been established successfully. We hope that other researchers will tackle this vital issue. In conclusion, we showed that S. charonda and three other Apaturinae larvae use oral odorants to repel their predatory ants and wasps.