Synchronized interdigitated calling in the Kuvangu running frog, Kassina kuvangensis

Synchronized interdigitated calling in the Kuvangu running frog, Kassina kuvangensis

ANIMAL BEHAVIOUR, 2003, 66, 127–136 doi:10.1006/anbe.2003.2173 Synchronized interdigitated calling in the Kuvangu running frog, Kassina kuvangensis T...

243KB Sizes 0 Downloads 42 Views

ANIMAL BEHAVIOUR, 2003, 66, 127–136 doi:10.1006/anbe.2003.2173

Synchronized interdigitated calling in the Kuvangu running frog, Kassina kuvangensis T. ULMAR GRAFE

Department of Animal Ecology and Tropical Biology, University of Wu ¨ rzburg (Received 26 June 2002; initial acceptance 12 September 2002; final acceptance 5 November 2002; MS. number: 7386)

Male Kuvangu frogs show repetitive calling of pulsed advertisement calls in which up to seven calls are repeated in short succession. Recordings of pairwise interactions between males showed that calls were highly synchronized, with individual calls interdigitating with each other. Males frequently switched between the leader and follower role with neither male dominating the interaction. Interactive playback experiments using synthetic calls revealed that males slightly but significantly increased the number of calls per call group with increases in stimulus call number. Males also significantly increased call rate with the number of calls in the playback stimulus. Furthermore, when presented with shortened intercall intervals, males increased their own intercall intervals, thus ‘skipping’ a call and avoiding overlap with the playback. The low degree of call matching suggests that repetitive calling, apart from maintaining a male’s attractiveness to females relative to rival males, may mediate male–male competition. In addition, synchronized interdigitated calling may serve to reduce predation, while maintaining species-specific temporal features of advertisement calls important to females. Kuvangu running frogs may have reduced the costs associated with synchrony and alternation by using a signal timing scheme that allows them to do both simultaneously. 

2003 Published by Elsevier Ltd on behalf of The Association for the Study of Animal Behaviour.

One important form of competition between males is endurance rivalry whereby males outcompete and outsignal rivals by remaining reproductively active as long as possible throughout the mating season (Andersson 1994; Halliday & Tejedo 1995). Across taxa, signals that are given with high intensity and high repetition rate are highly attractive to females (Ryan & Keddy-Hector 1992). Furthermore, males that are active on the breeding site the longest most often have the highest reproductive success (Andersson 1994; Murphy 1994). Vocalizing is energetically expensive, and calling males risk detection by predators or parasites. Males should therefore adjust the amount of calling to the varying social milieu they encounter, that is the amount of competition with other signalling males. Furthermore, they should avoid overlap with the calls of neighbouring males to ensure that their calls are not masked but are easily detected and located by females. Signal timing is an important feature of animal communication systems and one target of female choice (e.g. Klump & Gerhardt 1992; Correspondence: T. U. Grafe, Department of Animal Ecology and Tropical Biology, Biozentrum, Am Hubland, University of Wu¨rzburg, 97074 Wu¨rzburg, Germany (email: grafe@biozentrum. uni-wuerzburg.de). 0003–3472/03/$30.00/0



Greenfield 1994; Grafe 1996, 1999; Greenfield et al. 1997; Backwell et al. 1998; Vencl & Carlson 1998). In anurans, alternating calls may reduce acoustic interference, such as masking, and thus serve to preserve information important to females and reduce the possibility of producing unattractive follower calls (reviewed by Schwartz 2001). I examined the calling behaviour of the Kuvangu running frog, the basal species within the genus Kassina (Drewes 1984). Other Kassina species, notably K. senegalensis and K. fusca, have highly synchronized calling behaviour, with calls of neighbouring males often overlapping (Wickler & Seibt 1974; Grafe 1999; U. Grafe & H. Lu ¨ ssow, unpublished data). I investigated whether male Kuvangu frogs synchronize their calls and whether leader and follower roles switch during an interaction. Since males show repetitive calling of pulsed advertisement calls (Tandy & Drewes 1985), I also investigated whether males match the number of calls within a call group with other males. Preliminary analysis revealed that males were able to vary intercall intervals. I therefore hypothesized that males would alter intercall intervals to avoid overlap with stimuli that had shortened intercall intervals. I also analysed in detail the advertisement calls and an aggressive call from 21 males.

127 2003 Published by Elsevier Ltd on behalf of The Association for the Study of Animal Behaviour.

128

ANIMAL BEHAVIOUR, 66, 1

METHODS

Study Area and Species I conducted the field work in the Nchila Wildlife Reserve and at Hillwood Farm near Ikelenge in northwestern Zambia (1115 S, 2420 E) in January 2000. The study took place at the height of the rainy season, which typically lasts from November to March. The area is characterized by a mosaic of open savannah, brachystegea woodlands and dense gallery forests. I conducted male interaction and playback experiments at Pauls’ fish ponds, Hillward Farm, and a savannah pond about 200 m away. Males were heard calling only from ponds and wetlands in the open savannah. In contrast to other species of Kassina, male Kuvangu frogs regularly started calling in mid-afternoon (1600 hours) when there was no acoustic competition from other anuran species. Males produce advertisement calls that are repeated in short succession (Tandy & Drewes 1985). The Kuvangu frog is the only member in the genus Kassina that consistently shows repetitive calling; K. fusca and K. senegalensis show such calling only at high density (unpublished data). Each sound impulse is termed a call because it is produced by an individual contraction of the trunk muscles. Sequences of calls are termed call groups and their length varies according to the number of calls grouped together. The Kuvangu frog is a terrestrial, cursorial, savannah hyperoliid and one of the larger members in the genus. During the study, I caught only three males because of their clandestine nature and the preponderance of dense vegetation at most of the breeding sites visited. Snout–vent and tibia–fibula length of males averaged 45.92.19 mm (XSD; range 44.3–47.4, N=3) and 10.40.14 mm (range 10.3–10.5, N=3), respectively.

Recordings of Males and Call Analyses Single isolated males (N=21) and pairwise interactions of males (N=5) were recorded with Sennheiser MKE 300 directional microphones and a Marantz PMD 430 stereo recorder. For pairwise interactions, a microphone was placed close to each male and each male’s calls were recorded on a separate channel of the tape recorder. To avoid spurious data I recorded only pairs of males that were relatively isolated from other males, that is, whose calls were loudest to each other. Recordings were digitized at a sampling rate of 22 kHz and 8 bits. Call parameters were measured with Canary 1.2 (Bioacoustics Laboratory, Laboratory of Ornithology, Cornell University, Ithaca, New York, U.S.A.). I measured call length (ms), the lowand high-frequency peaks (Hz), frequency modulation (Hz), call rise time (ms), number of pulses/call, pulse length (ms) and pulse shape (pulse rise time/pulse length).

Playback Experiments Synthetic advertisement calls were used as playback stimuli. The synthetic call was based on measurements of the calls of 11 males from the same population and was

generated with SoundEdit 16. The acoustic properties measured for the synthetic call were call length, intercall interval, low-frequency peak, high-frequency peak, frequency modulation (dominant frequency of last 10% of call minus that of first 10% of call), number of calls/call group, modulation depth (degree of amplitude modulation of pulses), pulse shape and call envelope (call rise time/call length). Males were presented with 10 different stimuli in 10 different sequences of 30 repetitions of each stimulus. Each stimulus (a string of two to six calls) was given at a repetition rate of one stimulus every 5–7 s. I varied the timing of call presentation in a haphazard way to prevent males from predicting the exact timing of the playback stimuli and entraining to it. Five stimuli varied in the number of calls/call group from two to six, with a call duration of 32 ms and intercall intervals of 120 ms. To test whether males could adjust the timing of individual calls, I also presented males with three additional stimuli (made up of three calls) that had shortened intercall intervals of 100, 80 and 60 ms. To determine whether the synthetic stimuli were adequate in evoking a response by males, I included a three-call natural Kuvangu frog stimulus and a three-call white noise stimulus with the typical amplitude envelope in the playback sequence. The 10 blocks of stimuli were randomly presented to nine males in a repeated measures experimental design. Each male was presented with all 10 stimuli. Stimuli were presented to males at their call sites hidden underneath dense vegetation. Close neighbours that were likely to interact with focal males were disturbed from calling before and during playbacks to ensure selective attention of the focal male to the playback. Calls were played from a Sony tape recorder (WM D6C) and broadcast from a battery amplified loudspeaker with a flat frequency response over the range of call frequencies (Sony SRS-67). The intensity was set to correspond approximately to that of nearby neighbours. Because the exact position of males was unknown, playback levels could not be determined accurately. The response of males was recorded on one channel of a Marantz PMD 430 stereo tape recorder with a Sennheiser directional microphone. The playback stimulus was directly routed to the other channel of the tape recorder. Tapes were analysed with SoundEdit 16 and Canary 1.2. I conducted playbacks between 1700 and 2000 hours during peak calling activity. Temperature of the substrate near calling frogs varied between 18 and 26C. Variation in temperature did not influence temporal properties of calls (i.e. pulse rate or call rate; Fig. 1a).

Data Analysis and Statistics I analysed 5–15-min recordings of pairwise interactions for leader and follower interactions by calculating the relative phase of an individual’s calls with respect to his neighbour’s call period. I used circular statistics to determine the strength of coupling between males. I characterized the timing of one individual’s calls relative to that of his nearest neighbour by calculating the relative phase of the focal male’s call with respect to the neighbour’s call period. Phase angles of 0 and 360 indicate complete

GRAFE: ACOUSTIC COMMUNICATION IN RUNNING FROGS

overlap of calls and thus high synchronization, whereas a phase angle of 180 indicates precise call alternation. Intermediate phase angle values indicate varying amounts of lead or lag between individuals. The mean angular deviation was calculated using Zar (1999, page 599). There was reason to expect a priori that there was a strong coupling between pairs of males and that the very short response latencies would result in a relative phase angle approaching 0. I therefore used the V test for circular uniformity and a specified mean direction as described by Zar (1999, page 618) to examine the strength of the coupling and its direction. This test is more powerful than the Rayleigh test. The grand mean (mean value of mean values) and a second-order test for significance was calculated according to Zar (1999, pp. 608 and 638, respectively). To determine whether males consistently adjusted their call cycle to that of their neighbours, I correlated the delay of an individual’s calls (1) with the corresponding call period of the other caller (2) and vice versa (cf. Wickler & Seibt 1974; Klump & Gerhardt 1992). If male 2 uses the calls of male 1 as a trigger for his own call, then the response latency of male 2 (the interval between the calls of males 1 and 2) will remain unchanged regardless of the call cycle of male 1 (the intercall interval of male 1). If male 2 triggers the calls of male 1, then the response latency of male 2 will be only slightly shorter than the call cycle of male 1 and the data points will lie on a line with a 45 angle. The sequence of leader–follower relationships was analysed from recordings for interactions lasting 5–15 min. To determine whether there was an order in these sequences, I calculated transition probabilities (A/A, B/A, A/B and B/B) and determined whether these differed significantly from chance (equal probability of occurrence for each of the four transition types; 0.25) using a chi-square test. To determine whether a war of attrition could explain role shifts, that is, both males wait for the other to call, I compared the intercall group intervals preceding calls in which males switched roles with intercall group intervals preceding calls without a role shift. I measured responses to the stimuli that varied in call number in four ways. First, the proportion of triggered responses was noted. A call was regarded as having been triggered when the response latency was less than 200 ms. I determined this cutoff from poststimulus histograms of response latencies. Pooling over all males, 51.2% of calls produced during the playback experiments were given within 200 ms. Similarly, 59.1% of calls in pairwise interactions of males recorded in the field were within 200 ms. Calls that had a longer response latency were not regarded as having been given in direct response to the stimulus call. Second, I calculated the median number of calls/call group. Third, I determined response latencies, measured from the beginning of the stimulus, for each response. Finally, the call rate (call groups/min) was determined. I used the same four response measures to determine whether the synthetic call stimuli are perceived by males as conspecific calls. I analysed the effects of shortened intercall intervals of stimuli on calling behaviour by comparing the intercall

intervals (interval between first and second call) of calls given in response. Significance was tested with a Friedman ANOVA and the Wilcoxon–Wilcox test for pairwise multiple comparisons (Ko ¨ hler et al. 1996, pp. 186–188). Experiment-wide error rates were adjusted with a sequential Bonferroni (Rice 1989). If not stated otherwise meansSD are given, and all tests are two tailed. RESULTS

Calls and Calling Behaviour Advertisement calls were short and partly amplitude modulated, and showed substantial frequency modulation (Fig. 1a, Table 1). In contrast, the only aggressive call heard and recorded was much longer, contained many more pulses, and had less energy at high frequencies (Fig. 1b, Table 1). The aggressive call was elicited at the beginning of a playback session with high sound pressure level. Advertisement calls of two interacting neighbouring males were always interdigitated with complete integrity of calls and short response latencies (Fig. 2). Males produced advertisement calls that were grouped together in strings of up to seven calls. The modal number of calls/call group was four, and the range of the maximum number for individual males was two to seven. Males consistently gave calls that were highly synchronized with those of neighbours (defined as any call given within 200 ms of call onset; Fig. 3). In five pairs of males, 59.1% of calls were synchronized with calls always interdigitating. The mean relative phase angle of an individual’s call with respect to the neighbour’s call period varied between 0.05 and 8.6 (N=10; Table 2). For each of 10 males, the relative phase angles were not uniformly distributed; they matched significantly the predicted 0 (V test: u=4.8–7.8, N=13–53, P<0.0005). The grand mean relative phase angle of all males was 3.02.4 and the distribution of angles was highly nonrandom (F2,8 =793, P<0.0005). The average strength of the coupling (vector r) was 0.8450.162 (range 0.514– 0.997). Scattergrams showing the response latencies of a male to a neighbour’s calls in relation to the corresponding call period of that neighbour indicated that males called highly synchronously without one male consistently leading or following (Fig. 4). Instead, males frequently switched between the leader and follower roles (Fig. 5). The transition probabilities between leader and follower roles of all five pairs investigated did not differ significantly from chance (23 =0.4–5.5, NS). In addition, there was no significant difference in the number of leading calls between pairwise interacting males (23 =0–3.1, NS), suggesting that a male was equally likely to be a leader or follower and that there was no asymmetry in the interaction within the relatively short periods investigated.

Adequacy of Synthetic Stimuli Males showed significant differences in response to the three playback stimuli (synthetic, natural and white

129

130

ANIMAL BEHAVIOUR, 66, 1

Figure 1. Energy spectrum, sonagram and waveform of (a) a typical advertisement call and (b) an aggressive call of Kuvangu frogs. LF and HF denote the low- and high-frequency peaks in the energy spectrum of the advertisement call, respectively. Table 1. Call measurement summary (X±SD) for male Kuvangu frogs

Call parameter

Advertisement call (N=21)

Aggressive call (N=1)

34.9±4.6 789±59 1583±104 — 672±137 5.2±4.3 2.4±0.5 5.6±0.6 0.46±0.05

76 — — 1100 453 2.4 12 6.3 0.38

simple noise burst of the same amplitude envelope as a natural call, however, was much less adequate. Three of nine males were never triggered to call by the noise stimulus.

Varying Number of Calls/Call Group Call length (ms) Low-frequency peak (Hz) High-frequency peak (Hz) Dominant frequency (Hz) Frequency modulation (Hz) Call rise time (ms) Number of pulses/call Pulse length (ms) Pulse shape (rise time/length)

noise), in the proportion of triggered responses and call rate (Friedman ANOVA: 22 =10.1 and 11.6, respectively, P<0.01) as well as in response latency (Friedman ANOVA: 22 =9.0, P<0.05; Table 3). In pairwise multiple comparisons there were no significant differences in any response measure between the synthetic and the natural call (Wilcoxon–Wilcox: NS). However, males were triggered significantly less often by the noise stimulus than by either the synthetic or natural call (P<0.01 and P<0.05, respectively), showed a significantly lower call rate to the noise stimulus than to either the synthetic or natural call (P<0.05 and P<0.01, respectively) and called with a significantly higher latency to the noise stimulus (both P<0.05). These results suggest that the synthetic calls evoked the same kind of response as natural calls. A

During playback, males responded on average to 35.53.8% of the stimuli. The average response latency was 812 ms. When presented with stimuli varying in the number of calls/call group, males showed a significant increase in the number of calls/call group with increases in the number of calls in the stimulus (Friedman ANOVA: 24 =12.3, P<0.05; Fig. 6a). The number of calls/ call group in response to the stimuli with four or more calls was significantly higher than to the two-call stimulus (Wilcoxon–Wilcox: P<0.05). Males also showed a significant increase in call rate with the number of calls in the stimulus (Friedman ANOVA: 24 =10.0, P<0.05; Fig. 6b). Pairwise comparisons revealed that call rate increased significantly only between the two- and six-call stimuli (Wilcoxon–Wilcox: P<0.05). Males showed no significant differences in both the proportion of times they responded (Friedman ANOVA: 24 =7.6, NS; Fig. 6c) and the response latencies between treatments (Friedman ANOVA: 24 =1.7, NS; Fig. 6d).

Varying Intercall Intervals When responding to stimuli with shortened intercall intervals, males significantly increased their own intercall

GRAFE: ACOUSTIC COMMUNICATION IN RUNNING FROGS

(a)

A

B

A

0

A

B

A

A

B

A

250 Time (ms)

500

(b) Number of responses

30 25 20 15 10 5 0 0

200

400

600

800

1000

1200

1400

Response latency (ms) Figure 2. Call-triggered timing pattern in the repetitive calls of Kuvangu frogs. (a) Two sound tracks showing interdigitated calling by two males (A and B). Both males matched their number of calls. (b) Representative example of the latencies in response by one male to repetitive conspecific advertisement calls (indicated by the sonagram). Bars represent timing of call onset.

interval between the first and second call within a call group (Friedman ANOVA: 24 =14.8, P<0.05; Fig. 7). Pairwise comparisons showed that intercall intervals increased significantly between the nonstimulus period and the stimuli with 60- and 80-ms intervals (Wilcoxon– Wilcox: P<0.05). This lengthening of intercall intervals resulted in males ‘skipping’ a call (Fig. 8). DISCUSSION Male Kuvangu running frogs showed repetitive and highly synchronized advertisement calling with calls interdigitating with those of neighbouring males. Advertisement calls showed a strong upward frequency modulation, a distinctive feature in the genus Kassina, as well as a pulsatile beginning (see also Tandy & Drewes 1985). In contrast, the aggressive call was much longer and completely pulsatile. It is similar in structure to that described for K. senegalensis (Fleischack & Small 1978; U. Grafe & H. Lu ¨ ssow, unpublished data). Aggressive calls have not been reported for other Kassina species, despite extensive playback experiments (Bishop 1994; Grafe 1999). It remains to be explained why aggressive calls are rare or absent in the genus Kassina, despite their utility in mediating aggressive interactions in other anuran species. Reciprocal pairs of males showed a strong coupling in the timing of their calls. The average phase angle of calls was 3.02.4, and 59% of calls (N=five pairs of males) were synchronized (within 200 ms of call onset) with calls always interdigitating. This high degree of synchronization was supported by playback experiments.

Although males responded on only about a third of the playback stimuli, the response latencies were very short for all the playback treatments. Short response latencies have also been documented in other anurans that show high degrees of synchronization (Gerhardt & Huber 2002). When male Kuvangu frogs responded to white noise, the response latency was significantly longer, suggesting that males can distinguish between conspecific calls and noise. This may be useful to avoid acoustic interference from calls of heterospecific males. Furthermore, males increased both their call rate and the number of calls per call group with increases in the number of calls per call group in the stimulus. Finally, as in natural interactions, males interdigitated their calls with the playback stimuli.

Functional Significance of Synchronous Calling Synchronous calling can be adaptive in several ways. Explanations for its function can be divided into cooperative and competitive hypotheses. Prominent cooperative explanations are maximizing the peak amplitude of group signalling, making detection by predators more difficult and preserving call features important to females. These hypotheses are not mutually exclusive, however, as there are likely to be multiple selective pressures on signal timing. Most evidence so far supports the competitive hypothesis that males attempt to jam the calls of neighbours (Greenfield et al. 1997). Although the definition varies, synchronous calling is found in many vocalizing species. Among these are

131

ANIMAL BEHAVIOUR, 66, 1

(a)

20

270° 300°

240°

(a) 18 330°

16

180°

0°/360°

30°

150° 120°

90°

Interval 7_7 (s)

210°

60°

14 12 10 8 6

(b)

270°

4

300°

240°

2 –2

330°

210°

180°

0

2

4 6 Interval 7_8 (s)

8

10

12

0

2

4 6 Interval 8_7 (s)

8

10

12

20

0°/360°

(b) 18 16

30°

150° 120°

90°

60°

Figure 3. Distribution of relative phase angles (a°) and strength of coupling (r) of calls shown by a pair of interacting male Kuvangu frogs. Direction and length of arrows indicate mean phase angles and strength of coupling, respectively. (a) Male 1 with r=0.861, a°=1.8, angular deviation s=30.21 and N=39. (b) Male 2 with r=0.796, a°=2.0, angular deviation s=36.6 and N=41. Table 2. Coupling of the calls of 10 male Kuvangu frogs to the calls of the nearest neighbouring males Frog

N

r



s

U

P

1 2 3 4 5 6 7 8 9 10

39 41 21 19 24 25 53 47 15 13

0.861 0.796 0.911 0.997 0.997 0.919 0.514 0.615 0.871 0.968

1.8 2.0 1.6 1.3 4.4 2.5 3.6 8.6 0.1 4.3

30.21 36.60 24.17 4.44 4.44 23.06 56.55 50.28 29.22 14.50

7.6 7.2 5.9 6.1 6.9 6.5 5.3 5.9 4.8 4.9

<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001

r=strength of the coupling ; a°=mean phase angle; s=angular deviation; U=test statistic (V test).

gibbons (e.g. Mitani 1988), birds (e.g. Sonnenschein & Reyer 1983; Levin 1996a, b; Langmore 2000), insects and anurans (examples in Greenfield 1994; Gerhardt & Huber 2002). In many insects and anurans the degree of synchronization is so large that signals overlap (e.g. Oecanthus fultoni: Walker 1969; Neoconocephalus spiza: Greenfield & Roizen 1993; Hyla ebraccata: Wells & Schwartz 1984; Bufo punctatus: Sullivan 1985; S. sila: Ryan 1986; Centrolenella granulosa: Iba´n ˜ ez 1993; Kassina fusca: Grafe 1999).

Interval 8_8 (s)

132

14 12 10 8 6 4 2 –2

Figure 4. Scattergrams of the call period of one male in relation to the delay of calls of the other interacting individual. (a) Duration of the call period of male 7 (the intercall interval) in relation to the time interval between his call and the call of his neighbour, male 8. (b) Duration of the call period of male 8 (the intercall interval) in relation to the time interval between his call and the call of his neighbour, male 7. C: Alternating calls; : leading calls; x: following calls.

In insects and anurans, competition between males for leading calls often explains call timing because females strongly prefer signals that lead (Grafe 1996; Greenfield et al. 1997; Bosch & Ma´rquez 2002). This preference for leading calls has been termed the precedence effect: it is taxonomically widespread and can occur whether or not signals overlap (Gerhardt & Huber 2002). Thus, at least in anurans, where males typically respond to the concurrent signal of a rival (homoepisodic mechanism), males should generally avoid calling highly synchronously. It is puzzling, therefore, why males in some species risk being discriminated against by calling with such short response latencies. This paradox is resolved when one looks at female preferences in these species. In K. fusca, females

GRAFE: ACOUSTIC COMMUNICATION IN RUNNING FROGS

Figure 5. Sequence of leading calls given by five pairs of males. Numbers on the right indicate number of leading calls given during the interaction. Table 3. Response measures to three playback stimuli Synthetic

Natural

Noise

32.0 21.7–66.7

26.7 13.8–60.0

6.7 0.0–23.3

Call rate (calls/min)

8.7 6.8–9.7

9.7 6.3–11.0

5.0 2.2–6.7

Response latency (ms)

80 75–86

75 67–84

93 82–99

Percentage of responses

Values are median and interquartile range.

discriminate against follower calls at high degrees of overlap but prefer follower calls at low degrees of overlap, that is, of about 21% (Grafe 1999). Female Hyla ebraccata also prefer overlapping follower calls (Wells & Schwartz 1984) and in Alytes obstetricans follower calls are preferred even after a considerable delay (600 ms; Bosch & Ma´rquez 2001). Such preference exerts a selective pressure on males to overlap or follow the calls of competing males. Unfortunately, female choice trials have not been conducted for the Kuvangu frog. It remains to be seen, therefore, whether close synchronization benefits males by making them more attractive to females. Close synchronization may also be a response to high levels of acoustic competition. Males in several anuran families respond to sudden drops in background noise (i.e. conspecific calls) by initiating their own calls with varying delays (Zelick & Narins 1985; Schwartz 1993; Grafe 1996). This enables males to make use of short gaps of reduced noise. This strategy is driven not by an attempt to jam a neighbour’s call but to find the right moment to propagate one’s own signal. In this sense, call timing is a cooperative endeavour that is enforced in large choruses because males need to attend selectively to several nearby males and it is difficult to predict the precise timing of their calls. Because of the precedence effect, most males

need to avoid overlapping the calls of neighbours. The best strategy in this situation is to make sure one’s own call is placed in gaps of reduced noise. Indeed, species that form loose aggregations of relatively low density such as midwife toads show much higher relative phase angles (i.e. they alternate calls; 78 in Alytes cisternasii and 87 in A. obstetricans; Bosch & Ma´rquez 2001, 2002) than the aforementioned species that respond to gaps in background noise. Within the genus Kassina, synchronized calling is found in some but not all species and the pattern of call timing is likely to depend on the sensory ecology of each species, in particular chorus density (Grafe, in press). Apart from variation in female preferences for leading or lagging signals, variation in exposure to predators could also explain patterns of call timing. Synchronous calling can influence the ability of vertebrate predators to localize sound sources. In the Neotropical frog Smilisca sila, synchronized overlapping calls attract fewer predatory bats than alternating calls (Tuttle & Ryan 1982). The hypothesis that synchronous calling has a cooperative function in deterring predators clearly needs more investigation. It may explain some of the variation in signal timing so far attributed to other causes.

Interdigitized Calling Behaviour Although they synchronized call groups, male Kuvangu frogs alternated calls within call groups. Such interdigitized calling is also known from several other species of anurans. Schwartz & Wells (1985) found that Hyla microcephala may interdigitate 15 or more secondary notes of their advertisement calls, and males of Afrixalus vittiger also interdigitate notes of advertisement calls (unpublished data). In contrast, other species of frogs that have multinote advertisement calls do not interdigitate notes (e.g. Philautus leucorhinus: Arak 1983; Rana nicobariensis: Jehle & Arak 1998; Crinia georgiana: Gerhardt et al. 2000a). Interdigitation or interleaving may be possible only if the internote intervals are significantly longer than the notes. Species that do not interdigitate have internote intervals that approach note duration. In these species interdigitation would probably cause a high degree of overlap resulting in acoustic interference. Male Kuvangu frogs were able to adjust their intercall intervals, thus avoiding overlap with the playback stimuli. This shows that males are not committed to calling with fixed intercall intervals but can adjust them, to a certain degree, to avoid overlap. Thus, males adjust the timing not only of a call group but potentially of each call within a call group. This parallels Schwartz’s (1991) findings that male H. microcephala were able to adjust calls and notes to avoid overlap with playback stimuli.

Switching of Leader and Follower Roles Male Kuvangu frogs frequently switched between the leader and follower roles with neither male dominating the interaction. Role switching between pairs of signalling males was described by Walker (1969) in his pioneering work on tree crickets and has since been investigated

133

ANIMAL BEHAVIOUR, 66, 1

18 b

5 ab

4

b

a

3

2

80 70

3

2

4

(b)

16

b

ab

14

ab

ab

12

a

10 8 6 4 2

3

4

5

6

65 6 2 3 Number of calls/call group in stimulus

4

5

6

5

6

100

(c)

2 (d)

95 Response latency (ms)

90 Percentage of responses

b

Call rate (call groups/min)

Calls/call group in response

(a)

60 50 40 30 20 10

90 85 80 75 70

0 3

2

4

5

Figure 6. Median response measures of nine male Kuvangu frogs to playbacks of synthetic stimuli varying in the number of calls/call group. (a) Number of calls/call group, (b) call rate, (c) percentage of responses, (d) response latency. Box plots show the median response with interquartile range, 10th and 90th percentiles and minimum and maximum values. Different letters above box plots indicate significant differences.

200 Response intercall interval (ms)

134

a 180 160

a ab

140

ab b

120 100

60

80 100 120 Nonstimulus Stimulus intercall interval (ms)

Figure 7. Median intercall intervals of nine male Kuvangu frogs to playbacks of three-call stimuli with average (120 ms) and reduced (60–100 ms) intercall intervals. Box plots show the median response with interquartile range, 10th and 90th percentiles and minimum and maximum values. Different letters above box plots indicate significant differences.

in many species of insects and anurans (e.g. Schneider et al. 1988; Greenfield & Roizen 1993; Bosch & Marquez 2000, 2001). Role switching has been discussed as resulting from males adjusting their call timing to

Figure 8. Representative response of a male to a three-call stimulus with 80-ms intercall intervals. (a) Playback stimulus (A) and (b) two calls given in response (B). Dashed line indicates when the second call would have been initiated (on average 111 ms after call onset) without interference by the playback.

become leaders in vocal interactions with other males. Alternatively, call leadership per se may not be important to males. Instead, they may try to synchronize for the reasons mentioned above, with leadership being irrelevant. Noise of the internal oscillator, thought to drive rhythmic calling behaviour, can cause the calls between males to drift apart, reduce synchrony and

GRAFE: ACOUSTIC COMMUNICATION IN RUNNING FROGS

necessitate a resetting of the oscillator (Schwartz 2001). A resetting to the common rhythm will often lead to shifts in call leadership. According to this scenario, switching of call leadership is a by-product of the attempt to remain synchronized.

Call Matching and Call Rate Adjustment Although males increased the number of calls per call group with increasing number of calls from the stimulus, the correlation was not close. A significant increase in response was found between two and four to six calls per call group, however, with a total average increase of only 0.35 calls per call group. Only female choice experiments would show if such an increase were biologically relevant. Small differences in call rate can have profound effects on the attractiveness of males (reviewed in Gerhardt & Huber 2002). Increasing call rate, even by small amounts, can be a good indicator of viability to females, since calling is energetically very expensive in anurans (reviewed in Wells 2001). When presented with two-call stimuli, male Kuvangu frogs did not reduce their own calling effort compared to the nonstimulus period, suggesting that such calls would be unattractive to females. Call matching, to varying degrees, has been demonstrated in several frog species (reviewed in Gerhardt & Huber 2002). It may allow males to produce calls that are as attractive to females as those of rivals without wasting energy on more attractive calls (e.g. Gerhardt et al. 2000b). Alternatively, call matching might serve to signal the propensity to escalate a dispute with a rival (Arak 1983). It is difficult to argue for or against these hypotheses for Kuvangu frogs without further experiments. However, by not matching the calls of the playback stimulus, males might have been signalling an intention to de-escalate, a situation that may parallel that in some songbirds in which switching to a different song type signals de-escalation (Burt et al. 2001). In summary, by synchronizing call groups while at the same time alternating calls within call groups, male Kuvangu frogs may be able to maintain signal integrity while synchronizing calls either to increase their attractiveness to females or to avoid acoustically hunting predators. Of particular interest is the proficiency with which males adjusted the timing of individual calls to avoid overlap. The lack of call matching suggests that additional functions for repetitive calling, apart from attracting females, such as to mediate male–male competition, should be considered. Acknowledgments I thank Stefan Kaminsky for assistance in the field, Pete and Lynn Fisher and Steve and Debbie Wolford for their generous hospitality during my stay at Hillwood Farm and Hedje Lu ¨ ssow for help in analysing the data. Bob Drewes assisted in locating the study site. This work was supported by the Deutsche Forschungsgemeinschaft (Gr 1584).

References Andersson, M. 1994. Sexual Selection. Princeton, New Jersey: Princeton University Press. Arak, A. 1983. Vocal interactions, call matching and territoriality in a Sri Lankan treefrog, Philautus leucorhinus (Rhacophoridae). Animal Behaviour, 31, 292–302. Backwell, P., Jennions, M., Passmore, N. & Christy, J. 1998. Synchronized courtship in fiddler crabs. Nature, 391, 31–32. Bishop, P. J. 1994. Aspects of social organization in anuran choruses. Ph.D. thesis, University of the Witwatersrand. Bosch, J. & Ma´rquez, R. 2000. Acoustical interference in the advertisement calls of the midwife toads (Alytes obstreticans and Alytes cisternasii). Behaviour, 137, 249–263. Bosch, J. & Ma´rquez, R. 2001. Call timing in male-male acoustical interactions and female choice in the midwife toad Alytes obstetricans. Copeia, 2001, 169–177. Bosch, J. & Ma´rquez, R. 2002. Female preference function related to precedence effect in an amphibian anuran (Alytes cisternasii): tests with non-overlapping calls. Behavioral Ecology, 13, 149–153. Burt, J. M., Campbell, S. E. & Beecher, M. D. 2001. Song type matching as threat: a test using interactive playback. Animal Behaviour, 62, 1163–1170. Drewes, R. C. 1984. A phylogenetic analysis of the Hyperoliidae (Anura): treefrogs of Africa, Madagascar, and the Seychelles Islands. Occasional Papers of the California Academy of Science, 139, 1–70. Fleischack, P. C. & Small, C. P. 1978. The vocalizations and breeding behaviour of Kassina senegalensis (Anura, Rhacophoridae) in summer breeding aggregations. Koedoe, 21, 91–99. Gerhardt, H. C. & Huber, F. 2002. Acoustic Communication in Insects and Anurans: Common Problems and Diverse Solutions. Chicago: Chicago University Press. Gerhardt, H. C., Roberts, J. D., Bee, M. A. & Schwartz, J. J. 2000a. Call matching in the quacking frog (Crinia georgiana). Behavioral Ecology and Sociobiology, 48, 243–251. Gerhardt, H. C., Tanner, S. D., Corrigan, C. M. & Walton, H. C. 2000b. Female preference functions based on call duration in the gray tree frog (Hyla versicolor). Behavioral Ecology, 11, 663–669. Grafe, T. U. 1996. The function of call alternation in the African reed frog Hyperolius marmoratus: precise call timing prevents auditory masking. Behavioral Ecology and Sociobiology, 38, 149–158. Grafe, T. U. 1999. A function of synchronous calling and a novel female preference shift in an anuran. Proceedings of the Royal Society of London, Series B, 266, 2331–2336. Grafe, T. U. In press. Acoustic interactions among chorusing male anurans. In: Animal Communication Networks (Ed. by P. K. McGregor). Cambridge: Cambridge University Press. Greenfield, M. D. 1994. Synchronous and alternating choruses in insects and anurans: common mechanisms and diverse functions. American Zoologist, 34, 605–615. Greenfield, M. D. & Roizen, I. 1993. Katydid synchronous chorusing is an evolutionary stable outcome of female choice. Nature, 364, 618–620. Greenfield, M. D., Tourtellot, M. K. & Snedden, W. A. 1997. Precedence effects and the evolution of chorusing. Proceedings of the Royal Society of London, Series B, 264, 1355–1361. Halliday, T. R. & Tejedo, M. 1995. Intrasexual selection and alternative mating behavior. In: Amphibian Biology: Social Behavior (Ed. by H. Heatwole & B. K. Sullivan), pp. 419–468. Chipping Norton: Surrey Beatty. Iba´n ˜ ez, D. R. 1993. Female phonotaxis and call overlap in the Neotropical glassfrog Centrolenella granulosa. Copeia, 1993, 846– 850. Jehle, R. & Arak, A. 1998. Graded call variation in the Asian cricket frog Rana nicobariensis. Bioacoustics, 9, 35–48.

135

136

ANIMAL BEHAVIOUR, 66, 1

Klump, G. M. & Gerhardt, H. C. 1992. Mechanisms and function of call-timing in male–male interactions in frogs. In: Playback and Studies of Animal Communication (Ed. by P. K. McGregor), pp. 153–174. New York: Plenum. Ko ¨ hler, W., Schachtel, G. & Voleske, P. 1996. Biostatistik. Berlin: Springer-Verlag. Langmore, N. E. 2000. Why female birds sing. In: Animal Signals: Signalling and Signal Design in Animal Communication (Ed. by Y. Espmark, T. Amundsen & G. Rosenqvist), pp. 317–327. Trondheim: Tapir Academic Press. Levin, R. N. 1996a. Song behavior and reproductive strategies in a duetting wren, Thryothorus nigricapillus: I. Removal experiments. Animal Behaviour, 52, 1093–1106. Levin, R. N. 1996b. Song behavior and reproductive strategies in a duetting wren, Thryothorus nigricapillus: II. Playback experiments. Animal Behaviour, 52, 1107–1117. Mitani, J. C. 1988. Male gibbon (Hylobates agilis) singing behavior: natural history, song variation and function. Ethology, 79, 177– 194. Murphy, C. G. 1994. Determinants of chorus tenure in barking treefrogs (Hyla gratiosa). Behavioral Ecology and Sociobiology, 34, 285–294. Narins, P. M. 1982. Behavioral refractory period in Neotropical treefrogs. Journal of Comparative Physiology A, 148, 337–344. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution, 43, 223–225. Ryan, M. J. 1986. Synchronized calling in a treefrog (Smilisca sila). Brain, Behavior and Evolution, 29, 196–206. Ryan, M. J. & Keddy-Hector, A. 1992. Directional patterns of female mate choice and the role of sensory biases. American Naturalist, 139, S4–S35. Schneider, H., Hoermann, G. & Ho ¨ dl, W. 1988. Calling and antiphonal calling in four Neotropical anuran species of the family Leptodactylidae. Zoologische Jahrbu¨cher Physiologie, 92, 77–103. Schwartz, J. J. 1991. Why stop calling? A study of unison bout singing in a Neotropical treefrog. Animal Behaviour, 42, 565–577. Schwartz, J. J. 1993. Male calling behavior, female discrimination and acoustic interference in the Neotropical treefrog Hyla microcephala under realistic acoustic conditions. Behavioral Ecology and Sociobiology, 32, 401–414.

Schwartz, J. J. 2001. Call monitoring and interactive playback systems in the study of acoustic interactions among male anurans. In: Anuran Communication (Ed. by M. J. Ryan), pp. 183–204. Washington, D.C.: Smithsonian Institution Press. Schwartz, J. J. & Wells, K. D. 1985. Intra- and interspecific vocal behavior of the Neotropical treefrog Hyla microcephala. Copeia, 1985, 27–38. Sonnenschein, E. & Reyer, H.-U. 1983. Mate-guarding and other functions of antiphonal duets in the slate-coloured boubou (Laniarius funebris). Zeitschrift fu¨r Tierpsychologie, 63, 112–140. Sullivan, B. K. 1985. Male calling behavior in response to playback of conspecific advertisement calls in two bufonids. Journal of Herpetology, 19, 78–83. Tandy, M. & Drewes, R. C. 1985. Mating calls of the ‘kassinoid’ genera Kassina, Kassinula, Phlyctimantis and Tornierella (Anura: Hyperoliidae). South African Journal of Science, 81, 191–195. Tuttle, M. D. & Ryan, M. J. 1982. The role of synchronized calling, ambient light, and ambient noise, in anti-bat-predator behavior of a treefrog. Behavioral Ecology and Sociobiology, 11, 125–131. Vencl, F. V. & Carlson, A. D. 1998. Proximate mechanisms of sexual selection in the firefly Photinus pyralis (Coleoptera: Lampyridae). Journal of Insect Behavior, 11, 191–207. Walker, T. J. 1969. Acoustic synchrony: two mechanisms in the snowy tree cricket. Science, 166, 891–894. Wells, K. D. 2001. The energetics of calling in frogs. In: Anuran Communication (Ed. by M. J. Ryan), pp. 45–60. Washington, D.C.: Smithsonian Institution Press. Wells, K. D. & Schwartz, J. J. 1984. Vocal communication in a Neotropical treefrog, Hyla ebraccata: advertisement calls. Animal Behaviour, 32, 405–420. Wickler, W. & Seibt, U. 1974. Rufen und Antworten bei Kassina senegalensis, Bufo regularis und anderen Anuren. Zeitschrift fu¨r Tierpsychologie, 34, 524–537. Zar, J. H. 1999. Biostatistical Analysis. Upper Saddle River, New Jersey: Prentice-Hall. Zelick, R. & Narins, P. M. 1985. Characterization of the advertisement call oscillator in the frog Eleutherodactylus coqui. Journal of Comparative Physiology A, 156, 223–229.