Stonefly duets: vibrational sexual mimicry can explain complex patterns

Abstract

Stoneflies of the suborder Arctoperlaria exchange vibration signals for intersexual communication. The exchange takes the form of a duet with a male Call and female Answer. Some species and populations have three-way duets, with the male responding to the Answer with a third signal, termed Reply. We review duet descriptions of 106 species in terms of their temporal organisation and mate guarding tactics as identified by Bailey (2003), and propose a functional explanation in terms of concealment of the female for three recurrent phenomena: (1) In the case of large temporal overlap between Call and Answer, the overlapping part of the Call masks the Answer; (2) In three-way duets, the Reply mimics the female; (3) In a Call with two distinct phases, one phase mimics the female. It turns out that the Reply can be interpreted as mimicking the Answer in virtually all three-way duets. In two-way duets with a diphasic Call, the first phase typically mimics the Answer. Three-way duets with a diphasic Call are heterogeneous. Evidence for eavesdropping behaviour comes from the observation that males of the genus Zwicknia produce a Call in response to duetting sequences more often than in response to single male Call signals. We conclude with proposals for further research.

Appendices

Box 1: Overlapping signals

Examples from Supplement 1, Table S1.4: Several species in the genus Isoperla have an Answer signal consisting of a single beat interspersed during the male’s prolonged Call (Berthélemy 1979b; Sandberg and Stewart 2003, 2006). The same may apply to Clioperla clio, though only two Answers were recorded (Graham 1982: 15, 20, 22–23). Examples of two-way duets with longer Answers embedded in a prolonged Call include Capnioneura mitis (Tierno de Figueroa et al. 2011b), Pteronarcys biloba from Pennsylvania (Zeigler 1989), P. scotti (Sandberg and Stewart 2006), Chernokrilus misnomus (Maketon and Stewart 1984a) and Paraperla wilsoni (Stewart et al. 1995a), illustrated in Fig. 3d. Further cases are listed in Supplement 1, Table 3. A Reply that overlaps entirely with the Answer is reported for Pteronarcys princeps, based on one observation (Zeigler and Stewart 1985) and for Pteronarcys biloba from Tennessee, based on two observations (Stewart et al. 1982a). Examples of overlapping signals in more complex duetting types are given in Supplement 1, Tables S1.5–7.

Box 2: Three-way duets with a monophasic Call

Examples from Supplement 1, Table S1.5: Categories I (Call different from Answer and Reply) and II (Call, Answer and Reply similar): mimicry interpretation straightforward. The duets of two populations of Taeniopteryx burksi further illustrate the mimicking nature of the Reply. The Call has similar beat intervals around 120 ms in both Nebraska (Stewart et al. 1995b) and Oklahoma (Zeigler and Stewart 1985). In Nebraska, the Answer and Reply also have this frequency, but in Oklahoma, both the Answer and Reply have a lower frequency with intervals averaging 140 ms. (More accurately, all signals have gradually lengthening beat intervals). We consider the duet with unisex signals to be the ancestral condition; in Oklahoma, the Answer acquired a lower frequency and the mimic followed the model.

Category III: low beat numbers. Most species in this category have few beats in both Answer and Reply, but a remarkable case is presented by Perlinella drymo, with an Answer of 1 beat and Reply of 10 beats.

Category V: potential counter-examples. Zeigler (1989) describes two signal exchange patterns for the same specimens of Tallaperla maria from Pennsylvania: either three-way with the Call and Reply both consisting of one rub and the Answer consisting of one beat, or two-way with a diphasic Call consisting of a single rub followed by one beat. Neither pattern follows a common or theoretically predicted pattern, cf. Sect. “Two-way duets with a diphasic Call” on two-way duets with diphasic Call. There are several examples of species alternating between two-way and three-way duetting, or between monophasic and diphasic Calls—we interpret such cases as inconsistent performance of the mimicking signal. The alternation described for T. maria, however, seems very unusual. We suspect that the rub ‘Reply’ may in fact be a subsequent Call signal. The male single beat, whether preceding or following the Answer, can be interpreted as sexual mimicry. Confusingly, Maketon and Stewart (1988) describe the Call of T. maria from Virginia as monophasic, consisting of seven rubs.

Box 3: Two-way duets with a diphasic Call

Examples from Supplement 1, Table S1.6: Arguable mimicry: The Answer of Isoperla cf. grammatica from Norway consists of only 1 beat (Fig. 3). In the case of I. morenica, the first phase has beat intervals that gradually shorten from 120 to 20 ms, the second phase has intervals of 16 ± 9 ms and the Answer has 28 ± 5 ms intervals (Tierno de Figueroa et al. 2011a). Therefore, only the central part of the first phase resembles the Answer.

Counter-example: Paragnetina media from Arkansas: The second phase resembles the Answer, with which it is either overlapping or contiguous. It represents the single example of the pattern sketched in Fig. 5e. A Reply was recorded from a Wisconsin population, but not in Arkansas. See further notes in Box 4 under category II and in Supplement 1.

Apparent counter-examples: (1) Acroneuria lycorias. The descriptions by Stewart et al. (1982b) and Maketon and Stewart (1984b) do fit the predicted pattern, but Graham (1982) reports that both phases of the Call have the same frequency. This is an unusual pattern. Moreover, the frequency in her data is much higher for both sexes than reported in both other descriptions. (2) Tallaperla maria, discussed in Box 2.

Box 4: Three-way duets with a diphasic Call

Examples from Supplement 1, Table S1.7: Reply apparently non-mimicking: (1) In the case of Hydroperla  fugitans neither phase closely resembles the Answer, nor does the Reply. However, this description of the Reply is apparently based on only two observations, while the illustration provided (Graham 1982: 25, fig. 11) suggests that the last beat intervals of the Answer and the first intervals of the Reply may in fact be similar in length, around 80 ms. (2) Sandberg (2009) collected nymphs of Hesperoperla pacifica from two sites in California, Butte Creek (BC) and Domingo Springs (DS), and adult specimens at DS. The Reply mimics the Answer in the reared specimens from both localities: Answer and Reply have inter-beat intervals of around 45 ms at DS, and intervals ranging from 70 to 100 ms at BC. However, in the case of adults collected at DS, the Answer is reportedly of the BC type with intervals averaging 70 ms (Supplement 1, Table 3). The difference between collected and reared adults from DS can hardly result from the treatment. The incongruence might be due to sympatric occurrence of both types at DS.

Call, Category I: First phase is a mimic. Examples are Nemurella pictetii, Acroneuria abnormis and A. evoluta (authors’ type A). There are indications that females respond only to the second phase: Males of N. pictetii sometimes drop the first phase of the Call (Gnatzy and Rupprecht 1972), while the second phase alone can elicit the female response (Rupprecht 1976). Likewise, a quarter of the Call signals of A. evoluta contain only the second phase, and females respond to both types of Call (Maketon and Stewart 1984b).

Call, Category II: neither phase is a mimic. The duet of Acroneuria carolinensis may have evolved from a duet like those of A. abnormis and A. evoluta (authors’ type A) in the following way: the Answer acquired a lower frequency; the Reply followed, but the first phase did not. Conversely, in Hesperoperla pacifica (reared specimens, Sandberg 2009), the first phase may have acquired a lower frequency than the Answer (sketched in Fig. 1h). Acroneuria evoluta (authors’ type B) might have evolved from A. evoluta (type A) by replacing the first phase with a rubbing signal. Rubbing is considered to be a derived signal (Stewart and Sandberg 2005) and therefore likely addresses the female. The Paragnetina media population from Wisconsin has a long and variable latency (936 ± 521 ms) separating Call and Answer; it has two non-mimicking Call phases and a mimicking Reply (Graham 1982: 17, 37). It could have evolved from the duetting pattern described for the same species in Arkansas. The latter population has no Reply but a mimicking second phase that is contiguous or overlapping with the Answer (Maketon and Stewart 1988). A longer female latency may have rendered the mimicking phase ineffective and a Reply necessary.

Call, Category III: the second phase is a potential mimic. Temporally overlapping signals are exemplified by two species: (1) The second phase of Viehoperla ada is divided into 1–3 beat groups, each of which resembles the Answer in frequency, except that subsequent beat intervals shorten in the second phase but lengthen in the Answer. All recorded Answers were either “interspersed after the 1st phase call (n = 8) or between one or more of the 2nd phase groups (n = 29)” (Maketon and Stewart 1988: 92). (2) In Paraperla frontalis, the Call and Reply were both described as diphasic with a female-mimicking second phase. A single female beat is hidden among beats of the second phase of the Call (Sandberg and Stewart 2003). Notwithstanding the temporal overlap, the second phase appears to be a mimic rather than a mask. Note that, in these two species, the female clearly responds to the first phase.

Within category III, two species have a more derived signal or higher frequency in the non-mimicking first phase as expected: Viehoperla ada has higher frequency in the first phase, and Leuctra pseudosignifera has grouped beats in the first phase and simple beats in the other signals. In Zealeuctra hitei, however, the first phase has 20 ms intervals, and the second phase, the Answer and the Reply have 12.5 ms intervals. In spite of the lower frequency, the first phase must be the stimulus for the female because the second phase contains only “the last few beats” (Snellen and Stewart 1979: 75). Remarkably, the interval between the second phase and the Answer is likewise 12.5 ms, and may be part of the mimicking behaviour. Also in Agnetina capitata, the female responds to the first phase even though that has lower frequency. Importantly, a male specimen that produced only the first phase also elicited female responses (Zeigler 1989).

In two species, the second phase resembles the Answer, but the first phase is not a likely stimulus for the female: (1) The first phase of Soliperla fenderi consists of low-frequency beats that seem too few and too irregularly spaced to be the real Call. The most unexpected pattern is found in Nemoura spiniloba: the first phase is a simple but long series of beats, whereas the second phase, the Answer and the Reply are similar signals with grouped beats. Yet it seems unlikely that the second phase is only a mimic aimed at beta males, for two reasons. Firstly, if the second phase were a mimic, the first phase must be the real Call. It is atypical for the Call to have the least derived beat pattern in the duet. Secondly and more importantly, females do respond to a signal like the second phase. Stewart et al. (1991) mention that females in some instances respond to the Reply with a second Answer, giving rise to a so-called four-way duet. The interpretation of these second Answers, not observed by Sandberg (2011), needs further study (see “Discussion” section). However, it shows that females respond to a signal with grouped beats. The function of the first phase in N. spiniloba is not clear.