Queen Ants Make Distinctive Sounds That Are Mimicked by a Butterfly Social Parasite

See allHide authors and affiliations

Science  06 Feb 2009:
Vol. 323, Issue 5915, pp. 782-785
DOI: 10.1126/science.1163583


Ants dominate terrestrial ecosystems through living in complex societies whose organization is maintained via sophisticated communication systems. The role of acoustics in information exchange may be underestimated. We show that Myrmica schencki queens generate distinctive sounds that elicit increased benevolent responses from workers, reinforcing their supreme social status. Although fiercely defended by workers, ant societies are infiltrated by specialist insects that exploit their resources. Sounds produced by pupae and larvae of the parasitic butterfly Maculinea rebeli mimic those of queen ants more closely than those of workers, enabling them to achieve high status within ant societies. We conclude that acoustical mimicry provides another route for infiltration for ∼10,000 species of social parasites that cheat ant societies.

The main attribute that enables ants to dominate in most terrestrial ecosystems is their ability to live in complex societies whose cohesion is regulated by highly developed communication systems (1). Information is primarily transferred through exchanging distinctive semiochemicals, which, combined with physical contact, initiate and integrate different behaviors by colony members, providing a mechanism for caste determination and for recognizing and behaving appropriately to nestmates, non-kin ants, and potential intruders (13). In addition, the adults in 4 of the 11 subfamilies stridulate by scraping a plectrum located on an anterior segment of the abdomen (post-petiole) across a file (pars stridens) on the first segment of their gaster (1, 4).

About 10,000 other species of invertebrates from 11 orders have evolved adaptations to infiltrate ant societies and feed as social parasites on the rich resources concentrated inside nests (1, 5). Achieving this penetration generally includes the corruption through mimicry of their host's communication systems. The lycaenid butterfly Maculinea rebeli is among the better-understood examples. After briefly feeding on gentians, its final instar larvae are carried by Myrmica schencki workers (in Western Europe) into their nest, where the butterflies acquire ∼98% of their ultimate biomass before pupating 11 to 23 months later (5, 6). Inside the brood chambers, M. rebeli caterpillars beg like ant larvae and secrete semiochemicals that so closely mimic the surface hydrocarbons on Myrmica schencki larvae (7) that they are fed directly with regurgitations by the workers (6). However, neither begging nor chemical mimicry explains the high rank achieved by M. rebeli within its host's social hierarchy. For example, inert dummies painted with the surface pheromones of kin ant larvae or workers are retrieved in preference to dummies painted with M. rebeli's mimetic allomones (8), yet living M. rebeli larvae are rescued in preference to ant larvae when a colony is disturbed (9). Furthermore, nurse workers kill and feed their own brood to the social parasite if food is scarce (6, 10). We even observed Myrmica schencki queens treat Maculinea rebeli larvae or pupae like rivals (11) (fig. S1), whereas the workers regularly treat them like royalty (6). Lacking other cues, we speculated that the butterfly's elevated status might be achieved through mimicry of Myrmica schencki acoustics. Whereas Myrmica larvae are mute, distressed Maculinea larvae generate sounds that resemble the alarm stridulations of (adult) Myrmica workers (12). To be fully adaptive, the parasite would need to emulate the typical stridulation patterns of the most valued colony members, the queens. This, in turn, presupposes that queen stridulations differ from those of workers.

The substrate-borne (13) or airborne (14) calls of ants are generally considered to be a weakly developed means of communication (1), merely signaling alarm and the location of buried individuals, enhancing chemical recruitment, or (in Pogonomyrmex) indicating an end to mating (1, 3, 1517). Two studies, however, suggested that different castes might produce distinctive signals: The major workers of Atta cephalotes make sounds that are more intense and carry further than those of their smaller nestmates (16), and the space between the ridges of the pars stridens of queens exceeds that of workers in four Messor species (18).

Here, we report that the stridulatory organ of Myrmica schencki shows a morphological distinction between workers and queens greater than those seen in Messor, with the queen possessing a 44% longer pars stridens (P = 0.04) and a 33% wider gap between the ridges (P < 0.001) (Fig. 1). We recorded the signals produced by unstressed workers and queens (11) [supporting online material (SOM) audio S1 and S2] and found that the sounds of M. schencki queens differ significantly from those of workers in their dominant frequency (Fig. 1, A and B) (P = 0.014) and overall acoustics (Fig. 2) (P = 0.035), although the pulse lengths and the pulse repetition frequencies (PRFs) of both castes are similar (Fig. 1). The first two attributes reflect the morphology of the stridulatory organs, whereas the PRF reflects the rhythm and speed with which the plectrum is scraped across the file.

Fig. 1.

Location, morphology, and sounds of the acoustical organs of (A) queens and (B) workers of the ant Myrmica schencki and (C) pupae and (D) larvae of its social parasite Maculinea rebeli. Measurements of the stridulatory file are the average distance between the parallel ribs and of length along the central line. Characteristic sound spectra are shown with measurements ± SE of the average pulse length (PL), PRF, and dominant frequency (DF). Linear mixed effect models show that pupal and larval sounds differ from sounds from queens and workers [likelihood ratio (LR) = 12.63, P = 0.0004 for PL and LR = 20.67, P = 0.0005 for PRF]; the DF of worker sounds differs from those of queens (LR = 10.16, P = 0.0014), pupae (LR = 16.65, P = 0.0001), and larvae (LR = 34.37, P = 0.0002); whereas the DF of queens is more similar to pupae (LR = 4.12, P = 0.042) than to larvae (LR = 13.43, P = 0.0002). Pupal and larval sounds do not differ significantly (LR = 0.81).

Fig. 2.

Comparisons of the overall acoustics of Myrmica schencki queens and workers and Maculinea rebeli pupae and larvae. (A) Three-dimensional plot of the three sound parameters for each category of insect; large orbs indicate mean values. (B) The combined effect of the three parameters shown as the first and second component plot of a principal components analysis over all individual measurements; ellipses indicate 95% confidence intervals; large circles, the centroid for each source. Pairwise tests in an analysis of similarities of normalized euclidean distances, with measurements nested within individuals, showed that workers differ from queens (R = 0.13, P = 0.035), larvae (R = 0.58, P < 0.001), and pupae (R = 0.64, P = 0.001). Queens also differ from larvae (R = 0.62, P < 0.001) and pupae (R = 0.90, P = 0.001). There is no significant difference between larval and pupal sounds (R = 0.02). Mean larval and pupal calls are significantly closer (P < 0.0001 and P < 0.001, respectively) to queen sounds than to workers.

We played the sounds of both castes to undisturbed laboratory nests of M. schencki workers, together with controls consisting of a third miniature speaker that played computer-generated white noise and a fourth that was silent (11). No antagonistic or alarmed behavior was observed, except for a small repulsion of workers by white noise (F5,66 = 4.33, P = 0.002) (Fig. 3 and table S1). In contrast, the ant acoustics resulted in workers aggregating nonaggressively around the speaker, then tapping it with their antennae (antennation) or standing motionless on its surface in a posture similar to that adopted when they attend queens and objects of high value to their society (on-guard attendance) (Fig. 3 and table S1). The sounds of workers and queens elicited similar amounts of antennation, the main functions of which are to induce worker-worker recruitment, to smell nestmates, and to facilitate oral exchanges of food or pheromones (1). However, queen sounds induced significantly higher occurrences of on-guard attendance than did worker calls (posthoc test, least significant distance for workers–queens P = 0.014) (table S1), consistent with the exalted status and protection afforded to queens in the hierarchy of a colony (1). Combining these results, we suggest that acoustical communication within the vast subfamily Myrmicinae, to which Messor and Myrmica schencki belong, is more variable and conveys more social information within ant colonies than has previously been recognized. They also fulfill the strict adaptationist definition of biological communication, in which both the signal and response are adaptive (19).

Fig. 3.

Behavioral responses of Myrmica schencki colonies (10 workers) to sound recordings of M. schencki workers, M. schencki queens, Maculinea rebeli pupae, and M. rebeli larvae and to two controls (white noise and silence). Benevolent behaviors are in order of increasing respect, culminating in on-guard attendance. A significant overall difference in responses occurred within all four behaviors (repel F5,66 = 4.33, P = 0.002; aggregate F5,66 = 4.82, P = 0.001; antennate F5,66 = 20.20, P < 0.001; on-guard F5,66 = 8.64, P < 0.001). The letters above each column indicate pairwise posthoc tests of least significant differences (P < 0.05): The same letter indicates no significant difference within each type of behavior; different letters indicate significantly different responses (table S1).

We next recorded the sounds made by Myrmica schencki's host-specific social parasite, Maculinea rebeli, and, unlike an earlier study of Maculinea acoustics (12), we recorded pupae as well as larvae and did not stress the ants or butterflies (11). In contrast to the ants, pupal calls were generated by tooth-and-comb stridulatory organs (Fig. 1C and fig. S2) situated in the grooves between abdominal segments 5 and 8. We are unsure of the source of larval sounds, which may emanate from muscular contractions in the abdomen (20) or from sclerotized structures between segments 4 and 7 (Fig. 1D and fig. S2) that resemble the stridulatory organ of the mutualistic lycaenid Arhopala madytus (21). Despite the dissimilar sources, pupal and larval sounds resembled those made by the queens and workers of their host (Fig. 2A), but the similarity was 23% and 27% closer, respectively, to queen sounds than to those of workers (Fig. 2B) [mean normalized euclidean distances between individual butterflies' and ants' sounds are as follows: pupa-queen 2.47 ± (SE) 0.10, pupa-worker 3.03 ± 0.15, t = –3.14, df 87, distancepupa-queens < distancepupa-workers, P < 0.001; larva-queen 2.52 ± 0.11, larva-worker 3.21 ± 0.12, t = –4.32, df 237, distancelarva-queens < distancelarva-workers, P < 0.001]. The distributions in Fig. 2 also satisfy the concept that the perfect mimic should have maximal overlap with queen acoustics and minimal overlap with those of workers.

Playing recordings of Maculinea pupal calls to the same naïve cultures of Myrmica schencki workers resulted in enhanced benevolent responses similar to those elicited by queen ant sounds. We found no significant differences toward Maculinea pupal and Myrmica queen calls in any of the four behaviors scored, and pupal calls elicited six times more instances of royal on-guard attendance than occurred when worker sounds were played (Fig. 3 and table S1) (P < 0.001). Recordings of M. rebeli larvae induced lower worker responses and, despite eliciting 2.3 times more on-guard attendances than worker calls, did not differ significantly from responses toward worker sounds (Fig. 3 and table S1). We did not play Maculinea calls to queen ants but predict that they would provoke rivalry similar to that observed when live Maculinea pupae were artificially enclosed with Myrmica schencki queens (11) (fig. S1).

We suggest that regional host specificity in Maculinea populations is mediated first through chemical mimicry (6, 22); but once the intruder is admitted and accepted as a member of a host society, it mimics adult ant acoustics (particularly queens) to advance its seniority toward the highest attainable position in the colony's hierarchy. Selection for accurate acoustical mimicry may have been stronger in pupae, which lack the main secretory organs of M. rebeli larvae and offer only weak rewards to tending workers.

The young stages of other Maculinea species make similar pulsed sounds to M. rebeli (12): All differ substantially from those of other studied Lycaenidae, most of which are commensals or mutualists or have no known relationship with ants (12, 2327). None of the latter mimics the acoustics of associated ants in obvious ways, although the sound of one strongly mutualistic species attracts workers (2326). Thus, the use of acoustics to signal superior status to ants is unlikely to be a basal trait in the Lycaenidae, although we might expect it in Phengaris, the sister genus to Maculinea.

Beyond the Lycaenidae, ∼10,000 species of ant social parasites may exist (5), particularly among other Lepidoptera, Coleoptera, Diptera, and inquiline ants (1, 6). If acoustics plays the role that we suggest in reinforcing an ant's hierarchical status, it seems likely that this cue has evolved in other social parasites to infiltrate and exploit their societies.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 and S2

Table S1


Audio S1 to S4

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article