Introduction

Eusocial insects, including termites, construct colonies, where physiologically and morphologically specialized individuals execute specific tasks (Prestwich 1979; Wilson 1971). In general, they utilize various types of semiochemicals as the communicative media to maintain their society, while they also rely on auditory and visual media. The subterranean termites are mostly blind and have softer bodies than the other eusocial hymenopteran insects, so it is most likely that they rely heavily on chemical and tactile communications (Costa-Leopnard and Hifig 2010; Stuart 1969).

The colonies usually consist of kin members, so that they have a well-developed alarm-defense communication system to defend the colony members from various threats. A typical chemical signal involved in the alarm-defense communication system is an alarm pheromone (Blum 1969, 1981; Verheggen et al. 2010). This pheromone effectively transmits danger warnings to the other colony members and triggers various behavioral responses in the receivers depending on the amounts secreted. It recruits nestmate workers and soldiers with relatively small amounts, and often disperses them away from the signal with large amounts (Bradshaw 1981; Bradshaw et al. 1975, 1979; Brand et al. 1974; Verheggen et al. 2010). The pheromone usually enhances the aggressiveness of the recruited colony members, which may serve to produce an accurate response to dangers (Hölldobler and Wilson 1990; Vander Meer and Alonso 1998; Wilson and Regnier 1971).

Concerning termites, the soldier castes have specific morphs and exocrine glands for colony defense (Prestwich 1979, 1983, 1984; Prestwich et al. 1977). The cephalic capsules of the soldier caste are often heavily sclerotized and their mandibles are larger than those of the workers. It is most likely that this morphological feature is applicable to the physical colony defense in termites (Prestwich 1983, 1984). Soldiers also have an additional colony defense technique, i.e. defensive chemical secretion (Costa-Leonardo 1998a, b, 2001, 2004; Costa-Leonardo and Kitayama 1991; Costa-Leonardo et al. 2009; Prestwich 1979, 1983, 1984; Prestwich et al. 1977; Roisin et al. 1990; Vrkoc et al. 1978). Such defensive chemicals often come from the frontal gland, which is an unpaired cephalic organ and specific to termites.

The Japanese subterranean termite, Reticulitermes speratus Kolbe, which is the most common termite species in Japan, might have an alarm-defense communication system in which the defensive chemicals serve not only a purely defensive purpose but also an informative role as an alarm signal (Hölldobler and Wilson 1990). The soldier castes have sclerotized cephalic capsules that are larger than those of the worker castes, and possess the caste-specific chemical compound β-selinene in the frontal gland (Nguyen et al. 2011). This compound has been confirmed to serve as an effective alarm pheromone for the colony members, recruiting the soldier castes but simultaneously dispersing the worker castes (Nguyen et al. 2011), although it is not yet clarified if the chemicals were actually defensive.

Ironically, however, this alarm communication system is eavesdropped on and used by the Japanese garden ant, Lasius japonicas Santschi, in order to locate and hunt the termites through mass recruitment (Nguyen and Akino 2012). Therefore, although the termite soldiers secrete the alarm pheromone in an attempt to defend the colony, it actually has the opposite effect, and results in attracting the predatory ants. However, it does not seem evolutionally adaptive for the termites to acquire chemical signals to attract their predators. Since Lasius ants are originally distributed over the temperate zone (Antmaps.org. 2015; Bolton 1995; Bolton et al. 2006; Wilson 1955), it is most likely that the ants were encountered by the termites when they were expanding their distribution area northwards, which might be the reason why the alarm communication system of R. speratus did not work to defend the colony members against the Lasius ants. We therefore hypothesized that the soldier-specific alarm substances of R. speratus might have an additional ethological role, such as a defensive function, against the termite-hunting ants that are sympatric species distributed over the native regions of the termites.

In general, the Reticulitermes species has a Holarctic distribution, except in dry regions (Eggleton 2000). Distribution of the Japanese subterranean termite R. speratus is, however, restricted to northern areas of the Tokara Strait in the Japanese archipelago (Austin et al. 2002). According to molecular phylogeographic study on the Asian Reticulitermes termites, it is most likely that ancestral R. speratus had migrated from east China to the southern Japanese main islands, i.e. Kyushu, and then expanded their distributions northwards (Park et al. 2006). It is therefore necessary to focus on termitophagous ants whose native regions overlap that of ancestral R. speratus in order to confirm our hypothesis on the defensive function of the soldier-specific alarm signals.

In this study, we focused on Brachyponera chinensis Emery that is native to Southeastern Asian countries (Antmaps.org. 2015; Bolton et al. 2006; Yashiro et al. 2010) to confirm our hypothesis. The ponerine ants, including the genus Brachyponera, are known to be termitophagous, while some are obligate and others are facultative (Hölldobler and Wilson 1990; Leal and Olivera 1995). Concerning B. chinensis, Teranishi (1929) suggested it was a facultative termitophagous ant. Concerning the colony defense by Reticulitermes termites against the termitophagous B. chinensis ants, Matsuura (2002) reported the contribution made by the cephalic capsule size of the termite soldiers. Acquisition of such physical colony defense suggests that the mutual competition between R. speratus and B. chinensis had been extended over a long time. It also suggests that the termites might have acquired some chemical colony defense system. We discuss this through the behavioral responses of B. chinensis ants to the termite soldiers and the caste-specific chemical signals.

Materials and methods

Termites and ants

Colonies of R. speratus (Isoptera: Rhinotermitidae) nesting in rotten tree trunks, each of which consisted of thousands of colony members, were collected in Takaraga-ike Park, Kyoto City, Japan from April to June 2011. The rotten tree trunks were broken into small pieces, and then the pieces containing termites were kept in plastic containers with dimensions of 350 × 255 × 45 mm. Each colony was divided into 4–6 containers, and all the containers were kept isothermally at 28 ± 0.5 °C in constant darkness. Water was occasionally supplied in order to retain moisture.

A total of 7 colonies of the Asian needle ant B. chinensis (Hymenoptera: Formicidae) were collected from the rotten wood trunks in Takaragaike-Park, Kyoto City, Japan from May to June 2011. Each colony consisted of a queen, ≥50 workers, and dozens of brood. In the laboratory, the colonies were nested in a plastic container with dimensions of 60 × 50 × 20 mm, the bottom of which was covered with a layer of ca. 5 mm plaster to retain moisture. The top of the container, an artificial nest, was covered with red film to keep the inside dark, and a small hole (ca. 7 mm in diam.) was made in one of the lateral sides. As shown in Fig. 1, the smaller container was placed in a larger-sized plastic container, with dimensions of 175 × 80 × 30 mm, as an arena, the inner wall of which was treated with aqueous Fluon solution (Asahi Glass co. Ltd., Tokyo, Japan) to prevent the ponerine ants escaping. All the containers were kept isothermally at 28 ± 0.5 °C with 14L10D. Either house crickets, Acheta domesticus L., or the mealworm beetle, Zophobas atratus Fabricius, were supplied every 2–4 days. Water was occasionally supplied to the plaster layer in order to retain moisture.

Fig. 1
figure 1

Equipment to test defensive effects of the subterranean termite soldiers on termitophagous ant foragers

Full size image

Behavioral bioassay

To compare the foraging activity of B. chinensis workers with the termite workers and soldiers, three small clear glass tubes (7 mm in diam., 20 mm long) were simultaneously presented to B. chinensis. In one of the tubes, ten fresh worker whole bodies of R. speratus were placed, and ten fresh soldier whole bodies were put in another glass tube. The third tube was kept empty as a control (experiment I). In a separate experiment, instead of fresh whole bodies, the respective hexane extracts from termite workers and soldiers were applied to the inside of the glass tubes at one individual equivalent of 10 μl. Those extracts were prepared beforehand by immersing 100 heads of workers and soldiers in 1 ml of n-hexane for 30 min, respectively. As a solvent control, the last glass tube was treated with 10 μl of n-hexane. After evaporating the solvents, those glass tubes were placed in the arena (experiment II).

To evaluate the soldiers’ effectiveness in avoiding attacks by B. chinensis, two samples were simultaneously presented: one was ten fresh bodies of termite workers, and the other was ten fresh bodies of termite workers placed with three crushed bodies of termite soldiers (experiment III). Next, instead of crushed soldier bodies, the equivalent of three soldier head extracts (30 μl) was placed inside the open end of the glass tube. As a control, 30 μl of n-hexane was placed in the other glass tube in the same manner. Ten fresh bodies of termite workers were placed in both tubes (experiment IV).

Those tubes were lined up ca. 50 mm away from the entrance of the artificial nest of B. chinensis. The number of B. chinensis workers in each tube was counted 5 times every 1 min.

Each experiment conducted was simultaneously repeated 10 times by using nestmate workers from 2 to 3 colonies kept in the laboratory. As there was no difference in the responses of the ponerine ants among colonies, all the data was pooled for statistical analyses.

All the statistical analyses were performed on Stat Macros working on Excel 2003. The Friedman test, along with a post hoc Nemenyi test was used to compare the foraging ant numbers among three test tubes (experiments I and II), while a Mann–Whitney U test was used to compare the foraging ant numbers between two test tubes (experiments III and IV).

Results

The foraging activity of B. chinensis was compared by providing fresh bodies of termite workers and soldiers (Fig. 2). The forager ants immediately started hunting the worker bodies, as well as leading their nestmate ants to the tube, so that the number of foraging ants in the glass tube providing the termite workers was significantly larger than that in the tube providing termite soldiers (the respective Friedman’s χ 2 were 15.5, 17.2, 13.9, 13.9, and 13.2 to the elapsed time for 1, 2, 3, 4, and 5 min. N = 10, df = 2, p < 0.05, Friedman with post hoc Nemenyi test). Few foraging ants were recruited not only to the tube providing the termite soldiers, but also to the empty tube.

Fig. 2
figure 2

Number of foraging Brachyponera chinensis ants in glass tubes providing the subterranean termite Reticulitermes speratus soldiers (gray), workers (black), and nothing (white). The different letters signify statistically significant differences (Friedman with post hoc Nemenyi test, p < 0.05) between the treatments. Double asterisks mean the p value was less than 0.01

Full size image

This was also confirmed when the inner surfaces at the entrance of those glass tubes were treated with the head extract of the termite workers and soldiers (Fig. 3). Few B. chinensis ants foraged in the tubes treated with the soldier head extract or with only the solvent, n-hexane. On the other hand, significantly larger numbers of B. chinensis ants foraged in the tube treated with the worker head extract, especially at the elapsed times of 1 and 4 min, even though there were no rewards in any of the glass tubes (the respective Friedman’s χ 2 were 11.5, 4.9, 4.9, 7.3, and 2.5 to the elapsed time for 1, 2, 3, 4, and 5 min. N = 10, df = 2, p < 0.05, Friedman with post hoc Nemenyi test).

Fig. 3
figure 3

Number of foraging Brachyponera chinensis ants in glass tubes treated with head extracts of the subterranean termite Reticulitermes speratus soldiers (gray) and workers (black). Another glass tube was treated with the solvent, n-hexane, as a control (white). The different letters signify statistically significant differences (Friedman with post hoc Nemenyi test, p < 0.05) between the treatments

Full size image

When the termite worker bodies were presented along with crushed termite soldier bodies, B. chinensis ants clearly avoided feeding on the termite bodies (Fig. 4). They preferred to forage in the tube where only worker bodies were presented (the respective U values were 15, 6.5, 10, 0.5, and 1 to the elapsed time for 1, 2, 3, 4, and 5 min. N = 10, p < 0.01, Mann–Whitney U test). This was also true when the extract of termite soldier bodies was placed in the glass tubes. B. chinensis ants clearly preferred to feed on the worker bodies in the intact glass tube (the respective U values were 7.5, 21, 5, 3, and 1.5 to the elapsed time for 1, 2, 3, 4, and 5 min. N = 10, p < 0.01 or p < 0.05, Mann–Whitney U test, Fig. 5).

Fig. 4
figure 4

Number of foraging Brachyponera chinensis ants in glass tubes providing intact Reticulitermes speratus worker bodies (black, left) and the bodies of crushed soldiers (gray, right). The asterisks signify statistically significant differences (Mann–Whitney U test, **p < 0.01)

Full size image
Fig. 5
figure 5

Number of foraging Brachyponera chinensis ants in glass tubes providing intact Reticulitermes speratus worker bodies (black, left) and the bodies with the soldier head extract (gray, right). The asterisks represent statistically significant differences (Mann–Whitney U test, *p < 0.05, **p < 0.01)

Full size image

Discussion

The foraging responses of B. chinensis ants clearly showed that they avoided the soldiers of R. speratus (Figs. 2, 4). Furthermore, they also avoided the solvent extract of the soldier termites, but not that of the worker termites (Figs. 3, 5). It seems most likely, therefore, that the hexane-soluble chemicals peculiar to the soldier caste create a repellent effect on the ponerine ants. As B. chinensis is a facultative but powerful termitophagous ant in Japan (Teranishi 1929), the soldiers of R. speratus would have acquired a chemical defence system against such a termite-hunting predator. Since β-selinene is the only abundant compound found in the hexane-soluble constituents peculiar to the soldier caste (Nguyen et al. 2011), we suppose that it might function as a defensive chemical against the termitophagous B. chinensis ants.

Chemical colony defence based on the frontal gland secretion is also reported in some termite species (Costa-Leonardo 1998a, b, 2001). Chemical compounds contained in the termite frontal gland often correspond to the ant alarm pheromones (Billen and Morgan 1998; Blum 1981; Roisin et al. 1990). Such chemicals may, therefore, enhance the alerting responses of the ants, which results in reducing their attack efficiency (Šobotonik et al. 2008). In our case of R. speratus, the quantitative major component of the frontal gland secretion, β-selinene, has multiple functions which means that it can act as an alarm pheromone as well as a defensive signal (“allomones”) against the termitophagous ants. This means that just one semiochemical might enable the termite soldiers to evoke appropriate alerting responses depending on the castes of the colony members, and, at the same time, to repel the termitophagous ants. This appears to be a very effective and sophisticated unified alarm and colony defense system, which might work more efficiently to defend the colony when combined with different types of colony defense system, including the physical one described by Matsuura (2002).

Although the alarm and colony defence system of R. speratus acts quite effectively against the ponerine ant, it also brings unfortunate results to the termite colony when trying to defend it against Lasius ants (Nguyen and Akino 2012). In this case, the termite chemical defence system actually results in increasing the damage to the colony, because β-selinene is ineffective in repelling Lasius ants. In fact, the Lasius ants can utilize the terpenoid hydrocarbon β-selinene in order to locate the termite workers and soldiers. Such an unfortunate interaction is presumably because of the inappropriate combination of defensive tactics used by R. speratus. We therefore suppose that ancestral R. speratus would have acquired their chemical colony defense system against B. chinensis, which was native to Southeastern Asian countries (Antmaps.org. 2015; Bolton et al. 2006; Yashiro et al. 2010). However, R. speratus later encountered Lasius ants, which were originally distributed over the temperate zone, as they expanded their distribution area northwards from the south of Japan’s main island Honshu, where they had moved from east China, as Park et al. (2006) suggested. It would be logically possible, however, to interpret our experimental results in a different way, and to conclude that predation pressure by Lasius ants was not severe enough for R. speratus to have acquired any chemical defense system against Lasius ants, although this seems less likely, because the soldier-secreted alarm signals of the termites result in actually recruiting Lasius ants, which might be followed by mass recruitment by Lasius ants themselves to hunt the termites (Nguyen and Akino 2012). Some Lasius ants are known to utilize terpenoid hydrocarbons in their pheromone communication, but ponerine ants do not (Bergström and Löfqvist 1970; Billen and Morgan 1998). These temperate ants can have a different sense of chemicals in order to adapt to their habitat. As a result, a termite defense system using the terpenoid hydrocarbons against the ponerine ants would lose its effect on predatory ants, including Lasius, in the temperate zone.

The termite alarm communication system is still a complex matter, because the alarm pheromone has multiple roles, ethologically speaking. In general, the alarm and alert signals are used to inform the colony members as quickly as possible of coming danger, and this may be the reason why volatile chemicals are so often applied as an alarm pheromone in various terrestrial insects (Verheggen et al. 2010). However, as a result, the signal sender that secreted the alarm pheromone faces more danger, because the signal sender has made it easier for predators to locate it by secreting chemicals; as is also well known in the case of the so-called alarm call of birds (Dewsbury 1978). Thus, defensive chemical secretion, combined with the alarm signals, may be one of the counter-measures against these eavesdropping predators. Soldiers of various termite species often secrete such defensive chemicals, including insecticides (Prestwich 1983, 1984). These defensive chemicals, as well as the alarm chemicals, can work effectively only when detected by the appropriate receivers, because they might either be ineffective or be cribbed by unexpected receivers. In other words, it is necessary to select an evolutionary appropriate combination of signal sender and receiver in order to understand the natal ethological meanings of such chemical interactions. Further chemical and behavioral studies on termites and predatory ants could enable us to illustrate the termites’ alarm and chemical defense systems against predators.