| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Behavioral Ecology Vol. 13 No. 1: 109-124
© 2002 International Society for Behavioral Ecology
Territorial male bullfrogs (Rana catesbeiana) do not assess fighting ability based on size-related variation in acoustic signals
Division of Biological Sciences, University of Missouri, Columbia, MO 65211, USA
Address correspondence to M.A. Bee, who is now at AG Zoophysiologie und Verhalten, FB Biologie, Geo- und Umweltwissenschaften, Carl von Ossietzky Universitaet, Postfach 2503, 26111 Oldenburg, Germany. E-mail: mabf79{at}mizzou.edu .
Received 26 September 2000; revised 22 March 2001; accepted 4 May 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Some animals use communication signals to assess their opponent's size and fighting ability during aggressive conflicts. Male frogs assess their opponent's size based on the fundamental frequency (pitch) of advertisement calls, which is negatively correlated with body size, an important determinant of fighting ability in frogs. I conducted a field playback experiment to investigate whether territorial male bullfrogs assess the size of opponents based solely on size-related variation in fundamental frequency. I repeatedly broadcast synthetic bullfrog advertisement calls to three groups of males. Playback stimuli simulated a large male (n = 24), a small male (n = 24), or an acoustically size-matched male (n = 34). Neither the simulated size of the opponent, the subject's own size, nor the degree of size asymmetry between the subject and simulated intruder had significant effects on the magnitude of responses during the playback test or on the rate of habituation that occurred with repeated stimulation. Post-hoc analyses of effect sizes and statistical power indicated that the effects in this study were quite small compared to previous studies in other frogs. More important, power analyses indicated that this study had high power (1 - ß > 0.90) to detect the magnitude of effect sizes observed in previous studies. Thus, territorial male bullfrogs do not appear to assess an opponent's fighting ability based solely on the fundamental frequency of acoustic signals. These results contrast starkly with theoretical predictions and previous empirical work with frogs.
Key words: bullfrogs, communication, fighting ability, Rana catesbeiana, size assessment, territoriality.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Theoretical models of animal conflict predict that animals should attempt to assess asymmetries in fighting ability before engaging in escalated and potentially costly fights (Enquist and Leimar, 1983
; Maynard Smith,
1982; Maynard Smith and
Harper, 1995). Both the duration and intensity of agonistic
interactions are expected to vary depending on the degree of asymmetry between
opponents (Enquist and Leimar,
1983; Enquist et al.,
1990). Encounters should last longer and include more dangerous
behavioral elements when contestants possess similar fighting abilities and
asymmetries are more difficult to detect. When the asymmetries are larger and
more easily assessed, shorter, less intense encounters are expected on the
grounds that weaker individuals will assess their low probability of winning
and give up early in the contest. One important source of asymmetry in
fighting ability is a difference in body or weapon size (reviewed in
Andersson, 1994;
Archer, 1988). Larger size
often conveys advantages in physical contests.
Communication is common during animal conflicts, and animals often use
size-dependent or condition-dependent variation in displays to assess their
opponent's fighting ability during aggressive interactions (e.g.,
Clutton-Brock and Albon, 1979;
Davies and Halliday, 1978).
One property of acoustic communication signals that is constrained by body
size, and is therefore expected to be a reliable size-assessment signal, is
fundamental frequency, which is related to the percept of pitch
(Dawkins and Krebs, 1978;
Krebs and Dawkins, 1984;
Morton, 1977;
Wiley, 1983). Among anuran
amphibians (frogs and toads), fundamental frequency depends on the shape and
mass of the laryngeal apparatus, which is related to overall body size
(Martin, 1972). Fundamental
frequency is negatively correlated with body size in frogs, and this call
property is usually a better predictor of male body size than other call
properties (Bee and Gerhardt,
2001b; Bee et al.,
1999; Robertson,
1986; Wagner,
1989c). A number of studies demonstrate that large size confers
advantages in physical fights between male frogs
(Arak, 1983;
Davies and Halliday, 1978;
Given, 1988;
Howard, 1978;
Wagner, 1989a;
Wells, 1978). Therefore, in
frogs, accurate assessment of an opponent's relative size and fighting ability
based on the pitch of its calls would allow contestants to estimate their
likelihood of winning an escalated contest. Several studies demonstrate that
male frogs base decisions about continuing or escalating vocal and physical
interactions solely on size-related information conveyed by the spectral
properties of an opponent's acoustic signals
(Arak, 1983;
Given, 1987;
Ramer et al., 1983;
Robertson, 1986; Wagner,
1989a,b).
In these field playback experiments, male frogs were more likely to
persistently attack or direct aggressive vocalizations toward the high-pitched
calls of a perceived smaller frog.
Here I report results from a field playback study of acoustically mediated
size assessment by territorial males of the North American bullfrog (Rana
catesbeiana, Ranidae). During their breeding season, male bullfrogs
establish territories in permanent bodies of water
(Emlen, 1976;
Howard, 1978). From within
their territories, males emit advertisement calls, presumably to attract
gravid females and to repel rival males, and these signals are by far the most
common call type heard in a bullfrog breeding chorus. Territorial male
bullfrogs exclude other conspecific males from calling within their territory
using a combination of stereotyped aggressive movements, such as hops, jumps,
and lunges toward and onto an opponent, visual displays in the form of
presentations of the yellow gular sac, and vocalizations (see Emlen,
1968,
1976;
Howard, 1978;
Ryan, 1980;
Wiewandt, 1969). Advertisement
calls, encounter calls, and a low-frequency growl (see
Wells, 1978) are associated
with aggressive interactions. If neither contestant withdraws during a
contest, males engage in discrete wrestling bouts, in which they attempt to
clasp and submerge their opponent. Such fights are not uncommon and can last
more than an hour (Bee, personal observation;
Howard, 1978).
Agonistic encounters between bullfrogs may be costly in terms of time,
energy, and reduced vigilance for detecting predators, but the risk of
sustaining serious injury during a fight is probably low. Emlen
(1976) and Howard
(1978) found that larger male
bullfrogs win fights more frequently than smaller males. Additionally, Howard
(1978) reported that the
average size difference between male bullfrogs that engaged in physical fights
was significantly smaller than that between males engaged in aggressive
encounters that were settled by threats and displays before escalated
fighting. That is, aggressive encounters were more likely to escalate to
physical fighting when the size asymmetry between contestants was relatively
small, as predicted by some game theory models
(Enquist and Leimar, 1983).
Howard's data clearly suggest that some form of size assessment occurs before
escalated physical fights.
The purpose of the present study was to investigate the importance of
size-related variation in the fundamental frequency of bullfrog advertisement
calls as a potential size assessment cue during aggressive territorial
interactions. The advertisement call is a broad-band signal that consists of a
series of harmonics that are integer multiples of a fundamental frequency
(range 90-135 Hz), which is usually absent from the frequency spectrum
(Figure 1A). Fundamental
frequency is strongly and negatively correlated with male body size
(Figure 1B; see below), and
fundamental frequency is the best acoustic predictor of male body size, with
the exception of call properties that are highly correlated with fundamental
frequency (e.g., harmonics; Bee and
Gerhardt, 2001b). In previous experiments based on the
habituation/discrimination paradigm, bullfrogs were shown to be capable of
perceptually discriminating between two advertisement calls that differ in
fundamental frequency by a magnitude typical of among-male differences in the
population (e.g., 9-12 Hz; Bee and Gerhardt,
2001a,c).
Hence, as in other frogs, the fundamental frequency of advertisement calls is
a reliable and discriminable signal of male body size in bullfrogs.
|
In this study, I repeatedly broadcast synthetic advertisement calls
simulating a new neighbor calling from a position adjacent to a subject's
territory. Subjects heard the sounds of a simulated large male, a small male,
or an acoustically size-matched male. I compared the initial and maximum
aggressive responses to broadcasts of these three stimuli. I also compared the
number of stimulus presentations required until males no longer responded
aggressively to the playback as a measure of whether the duration of
aggressive encounters depended on an assessment of an opponent's relative size
and fighting ability. In previous studies
(Bee, 2001;
Bee and Gerhardt, 2001a), we
demonstrated that the aggressive response of bullfrogs to repeated field
playbacks of conspecific advertisement calls exhibits characteristics of
response habituation. The primary goals of the present study were to determine
whether aggressive behavior directed toward an opponent depended on the
acoustically simulated size of the opponent, the size of the subject, and the
degree of size asymmetry between the subject and the opponent.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Playback experiments
Between 10 May and 1 August 1998-2000, I conducted a field playback experiment at the Little Dixie Lake Conservation Area (Callaway County, Missouri, USA) and the Thomas Baskett Wildlife Area (Boone County, Missouri; see Bee and Gerhardt, 2001a
,
for additional details about the study site). Playback tests were performed
under ambient light between 2130 and 0930 h and usually commenced between 2200
and 0000 h each night after males began actively calling. On a night before
testing, each subject was captured, and its snout-to-vent length (SVL) and
mass were measured. Measurements of mass were unavailable for two individuals.
Males were given individual toe clips on their hind feet, a cohort mark
indicating year of testing on their forefeet, and a waistband and tag with an
identifying number, which allowed visual identification of males from
distances of 2-3 m.
Equipment
The digital-to-analogue output of portable notebook computers (Samsung SENS
800, Dell Inspiron 3500 or 5000) and battery-powered amplifiers (Nagra DH or
Rockford Fosgate 2.6x) were used to broadcast synthetic bullfrog
advertisement calls through one of four 10-inch Optimus speakers mounted in
wooden boxes and floated on styrofoam platforms covered in black plastic. The
frequency response of each speaker was flat (± 4 dB) over the range of
frequencies in the stimuli. Playback levels were measured and calibrated in
the field with a GenRad 1982 sound-level meter or a Radio Shack sound-level
meter calibrated against the GenRad meter.
Stimuli
Previous playback experiments have established that territorial males
respond aggressively with encounter calls, aggressive movements, and
approaches toward a speaker broadcasting both natural, prerecorded
advertisement calls (Davis,
1987; Emlen, 1968;
Wiewandt, 1969) and synthetic
models of advertisement calls (Bee and Gerhardt,
2001a,c;
Davis, 1988). I broadcast
synthetic advertisement calls at a sound pressure level (SPL) of 87 dB
measured at a distance of 1 m (re 20 µPa, fast RMS, C-weighted), which
reflects the upper end of the range of variation in the SPL of natural calls
(Bee, unpublished data; Megela-Simmons,
1984). Sound pressure levels typically varied between 86-88 dB at
1 m and between 70-72 dB at the frog's original position. Stimuli were
generated at a sampling rate of 20 kHz with 16-bit resolution using
custom-designed software. All values of stimulus properties fall within the
range of natural variation for this species
(Bee and Gerhardt, 2001b;
Capranica, 1965). A stimulus
consisted of five consecutive advertisement calls separated by 30-s intercall
intervals (call duty cycle = 0.17; see
Figure 2A). Each call within a
stimulus consisted of five identical notes that were 700 ms in duration, had
symmetrical linear rise and fall times of 300 ms duration, and were separated
by 700-ms internote intervals (note duty cycle = 0.50). Each note within a
call consisted of a series of 10 harmonics
(f2-f4 and
f10-f16) that were integer multiples
of the fundamental frequency (f1) and had the same
starting phase relationships of 0°
(Figure 2B-D). The dominant
frequency was the second harmonic (f2). All other
harmonics were attenuated by 5-20 dB in relation to the dominant frequency.
The fifth and final call of the stimulus was followed by a silent
interstimulus interval (ISI) of 5 min. The combination of the stimulus and the
ISI was broadcast repeatedly using the sound-editing software GoldWave 4.02.
Together, a stimulus and the subsequent ISI constitute what is hereafter
referred to as a "stimulus period." All playback stimuli used in
this study shared the properties listed above, and differed only in
fundamental frequency (and correlated spectral differences).
|
I examined differences in aggressive responsiveness to simulated intruders of different sizes by broadcasting one of three types of acoustic stimuli to territorial males. For playback tests conducted in 1998 and 1999, subjects (n = 34) were presented with an acoustically size-matched opponent, for which the fundamental frequency of the stimulus was the same (± 1 Hz) as that of the subject's own advertisement calls. On the night each subject in this group was tested, prior to the beginning of the playback test, I recorded one or more of its advertisement calls with an HHB PDR 1000 DAT recorder (sampling rate = 32 kHz) or one of the notebook computers (sampling rate = 22.05 kHz), using Sennheiser MKH 70 or ME 66 directional microphones mounted on a tripod placed 1-2 m from the subject. Calls were digitized and stored on the notebook computer as 8-bit sound files at a sampling rate of 22.05 kHz. I used GoldWave 4.02 software to determine the fundamental frequency of one randomly chosen advertisement call from each subject by taking the reciprocal of the average period of 10 repetitions of the fine-temporal waveform from an oscillogram [fundamental frequency (Hz) = 1/waveform periodicity (s)]. I selected an appropriate stimulus from a series of synthetic stimuli with fundamental frequencies spanning the range of natural variation (90-135 Hz) that had been previously generated offline and stored on disk. The fundamental frequencies of the subjects' advertisement calls were strongly negatively related to male body size (Figure 1B). The average fundamental frequency of the size-matched stimuli was 113 Hz (SD = 9 Hz; range = 93-132 Hz).
In 2000, I broadcast synthetic advertisement calls that simulated either a
large intruder with a low fundamental frequency (f1 = 95
Hz; n = 24 males) or a small intruder with a high fundamental
frequency (f1 = 125 Hz; n = 24 males). The large
and small stimuli were presented in random order during the 2000 breeding
season. Previous work has established that territorial males can perceive a
frequency difference of this magnitude (Bee
and Gerhardt, 2001a). Based on the regression of fundamental
frequency on SVL calculated for the 34 males that heard a size-matched
stimulus (Figure 1B), the large
stimulus acoustically represented a male with an SVL of 177 mm, and the small
stimulus represented a male with an SVL of 117 mm. Subjects in this study had
SVLs ranging from 115 to 171 mm (mean ± SD = 141 ± 13.5 mm;
population range = 109-181 mm; Bee, unpublished data).
Protocol
To begin a playback test, I positioned the playback speaker at a distance
of 6 m along the pond bank from an actively calling target male, in the
direction of the male's most distant neighbor. If there were no nearby
neighbors (e.g., < 15-20 m away), I determined speaker position (left or
right relative to the frog) by flipping a coin. Typical distances between
adjacent territorial male bullfrogs range between 3 and 16 m (Bee, personal
observations; Emlen, 1968,
1976). I preferentially tested
males that did not have a nearest neighbor within 10 m in at least one
direction along the pond bank to avoid interference from other territorial
males. I started playbacks 15-30 min after subjects resumed normal calling
behavior after speaker placement.
During each stimulus period of a playback test, I counted the number of
advertisement calls, encounter calls, and aggressive movements toward, around,
and away from the playback speaker; measured the maximum distance advanced
toward the speaker; and determined the latency to the first encounter call
with a stop watch. If no encounter call was produced during a stimulus period,
I assigned a latency value of 452 s, which was equivalent to the duration of
the stimulus period. For a male to be included in the data set, it had to give
at least one encounter call during the first stimulus period of the test.
Twelve males did not respond during the first stimulus period with an
encounter call in response to the size-matched (n = 3), small
(n = 1), large (n = 8) stimuli. Ten of these males were
successfully retested at a later date, and eight males were retested with the
same stimulus to which they did not respond on the first test (size-matched =
2, small = 1, and large = 5). Although there was a trend for males to be
nonresponsive when they heard the large stimulus
(
22 = 17.1, p <.05), most of these
males (five of eight) were successfully retested with the same stimulus at a
later date (one male was retested with the small stimulus, and two males were
not retested). Some factor other than stimulus size probably influenced the
behavior of these nonresponsive males.
I examined the magnitude of the aggressive response during a habituation phase that consisted of repeating the stimulus period as a continuous loop. In 1998 and 1999, the stimulus period was repeated until subjects met an arbitrary criterion of asymptotic response decrement, defined as no movement and no production of encounter calls during three consecutive stimulus periods (hereafter "response decrement criterion"). In 2000, I used a slightly different protocol that consisted of repeatedly broadcasting the small or large stimulus for a total of 30 stimulus periods (= 3.75 h).
Data analysis
I made between-group comparisons of three primary measures of response
strength: initial responses during the first stimulus period, maximum
responses during any stimulus period, and the rate of aggressive response
decrements. Initial responses were determined as the numbers of advertisement
calls, encounter calls, and movements, the distance advanced toward the
speaker, and the latency to the first encounter call during the first stimulus
period. Maximum aggressive responses were determined as the maximum numbers of
advertisement calls, encounter calls, and movements, the maximum distance
advanced toward the speaker and the minimum latency to an encounter call
during any stimulus period. The rate of habituation of the aggressive response
was measured as the number of stimulus periods required to reach the response
decrement criterion, including the three consecutive stimulus periods without
an encounter call or movement (Bee,
2001). In response to the size-matched stimulus, 30 of 34 males
(88.2%) reached the response decrement criterion after experiencing 30 or
fewer stimulus periods. In response to both the large stimulus and the small
stimulus in 2000, 20 of 24 males (83.3%) reached the response decrement
criterion in 30 or fewer stimulus periods. To make meaningful between-group
comparisons, I did not consider responses occurring after the 30th stimulus
period for the 12 of 82 males (14.6%) that had not yet met the response
decrement criterion. Eight of these 12 males were not tested after the 30th
stimulus period; the remaining 4 males tested with the size-matched stimulus
met the criterion after hearing 31, 36, 41, and 81 stimulus periods. This
procedure was necessary to make comparisons between groups that experienced
slightly different habituation training phases. For testing differences in the
rates of habituation, a maximum value of 30 stimulus periods to criterion was
assigned to the 12 males that had not yet met the criterion in 30 periods.
Following log-transformations [Y' = log10(Y +
k), where k = constant], most response variables met the
requisite assumptions of parametric statistical analyses. However, the
variance in the maximum number of advertisement calls was not homogenous
between groups, and the maximum approach distance departed from normality.
Because parametric tests are generally regarded as robust to moderate
violations of normality and homogeneity of variance at large sample sizes
(Rosenthal and Rosnow, 1991),
I chose to use parametric statistical methods. Statistical significance was
set at the 
= 0.05 level.
To examine the effects of a resident's size and the size of the simulated
opponent, I divided subjects into three size classes: small (SVL 
134 mm,
n = 29), medium (135 mm SVL 146 mm, n = 26), and
large (SVL 
147, n = 27). I performed 3 (stimulus) x 3
(size class) multivariate analyses of variance (MANOVA) to examine
between-group differences in the magnitude of the initial response and the
maximum response. Subsequent 3 x 3 ANOVAs were used to separately
analyze between-group differences in the five response variables for initial
and maximum aggressive responses. I performed a 3 x 3 ANOVA to compare
the number of stimulus periods required to reach the response decrement
criterion. These two-way ANOVAs were designed to answer three questions
(Bee et al., 2000;
Wagner, 1989b): First, do
males respond differently to different-sized opponents (stimulus effect)?
Second, do males of different sizes respond differently to opponents
(size-class effect)? Third, does a male's aggressive response depend on his
own size and the size of his opponent (stimulus x size class
interaction)?
To directly examine the importance of relative size differences in
determining male responses, I computed an index of size asymmetry (AI;
Enquist and Leimar, 1983;
Enquist et al., 1990):
 |
Because males tested with the large and small stimuli in 2000 experienced
the same number of stimulus periods (30), I was able to directly compare
changes in the responses of small, medium, and large males (between subjects)
that occurred with repeated presentations (within subjects) of the large and
small stimuli (between subjects). I performed 2 (stimulus) x 3 (size
class) x 30 (stimulus period) repeated-measures ANOVAs on each of the
five response variables. I used the Greenhouse and Geiser
(1959) method to adjust the
degrees of freedom for omnibus repeated-measures effects.
Two important caveats regarding the interpretation of the data presented here deserve consideration. First, although no males were tested multiple times in the same year, there is some small, but unknown, probability that males tested in 1998 or 1999 were also tested in a subsequent year. Individual toes clipped from the hind feet regenerated between breeding seasons and were indistinguishable from toes that had not been clipped. Therefore, identifying individuals between years was not possible. Toes clipped from the forefeet as cohort marks, however, did not regenerate between years. Hence, I could identify individuals that were marked in previous years, but I could not determine whether those individuals had also been tested in that year because only a subset of frogs marked each year were also tested that year. However, the number of males unknowingly tested twice could not exceed the total number of individuals that were tested in one year and also marked in a previous year, which was small (6 of 82 males; 7.3%). Therefore, the maximum potential for pseudoreplication in this study was rather small. For purposes of statistical analyses, I assume that any response to a stimulus in one year is independent of whether the male was also tested in a previous year.
Second, because size-matched stimuli were presented in 1998 and 1999 and
the large and small stimuli were presented in 2000, there is some potential
for differences in behavior to result from differences in any number of
factors that varied between years and produced differences between the males
tested with the size-matched or large and small stimuli. I believe the impact
of such factors was minimal in the present study for the following five
reasons. First, there were no between-group differences in SVL
(F2,79 = 0.14, p =.8685), mass
(F2,77 = 0.60, p =.5496), or an index of physical
condition (F2,77 = 1.78, p =.1748; after
Baker, 1992). Second, with the
single exception of minimum response latency (F2,79 =
7.64, p =.0009), there were no significant differences in response
variables based on year as a main effect (0.10 < F2,79
< 2.45,.09 < p <.91). Response latencies were significantly
lower in 1999 compared to both 1998 and 2000
(Scheffé's multiple comparisons test:
p <.02), and latencies in 1998 and 2000 were not different
(p =.2814). Third, playback tests were conducted over the duration of
the breeding season each year, which lasted between mid-May to late July in
all 3 years, and tests were started at similar times of night in all 3 years.
Fourth, based on maps of territory positions made nightly in all 3 years,
bullfrogs established territories in similar locations in the same ponds
across years. Fifth, based on these territory maps, there were no obvious
differences in male density and intermale spacing between years, although
precise measurements are unavailable. Hence, there is little evidence to
suggest that physical, temporal, spatial, and social variables covaried with
experimental treatments, and I assume that their impact on the data presented
here was negligible.
Power analysis and meta-analysis
For between-subjects comparisons based on ANOVA, I computed effect sizes
and the statistical power of the test following Cohen
(1988) and Rosenthal and
Rosnow (1991). The effect size
for the F statistic from ANOVA is 
2, which is the
proportion of variance explained by membership in two or more experimental
groups. The variable 2 is a generalization of the more
familiar coefficient of determination (r2) associated with
tests of differences between two groups (see
Cohen, 1988, for an extensive
discussion of this topic). Following Rosenthal and Rosnow
(1991), I calculated
2 as:
 |
I also calculated the effect sizes from five other studies of acoustically
mediated size assessment in frogs to determine how the effect sizes in the
present study compared to previous studies that have addressed similar
questions (Bee et al., 2000;
Davies and Halliday, 1978;
Given, 1987;
Robertson, 1986;
Wagner, 1989b). These studies,
which included five species from four families, were chosen because they met
the following criteria: they reported results from acoustic playback
experiments; they explicitly tested hypotheses about the function of spectral
call properties as assessment signals; and they either used ANOVA to make
statistical comparisons and reported F ratios and degrees of freedom,
or they provided raw data (Given,
1987), from which I generated F ratios using ANOVA. In
Table 1, I report effect sizes
for the variables in these studies that exhibited significant differences in
responses to playback stimuli differing in size-related spectral properties
(e.g., fundamental frequency or dominant frequency). The average effect size
for the significant effects from these studies was 2 =.29
(range: 2 =.12-.57). For comparison, I include results from
one-way ANOVAs and effect sizes for a between-subjects comparison of bullfrog
responses to the large and small stimuli presented in this study.
|
Statistical power (1 - ß, where ß = probability of a type II
statistical error) for a given effect size, sample size, and level
( = 0.05) was calculated using the power tables in Cohen
(1988). For most of the
statistical tests in the present study that yielded nonsignificant results, I
report the statistical power of the test to detect both the observed effect
size (1 - ßactual) and an expected effect size (1 -
ßexpected), determined as the average effect size from the
five previous studies listed in Table
1. Reports of the power to detect the expected effect size are
limited to univariate analyses of variance.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In response to broadcasts of the stimuli, males oriented toward the speaker, produced advertisement calls and encounter calls, and approached the speaker using stereotyped aggressive movements. Males often repeatedly charged toward the speaker during the broadcast of the stimulus and returned to their original calling site during a subsequent interstimulus interval. Similar aggressive behavior was originally described by Emlen (1968
; see also
Wiewandt, 1969). The
proportions of males that "attacked" the speaker by approaching
the entire 6 m distance were similar in responses to the small stimulus
(14/24, 58.3%), size-matched stimuli (20/34, 58.8%), and the large stimulus
(14/24, 58.3%; 22 <.01, p >.99). As
habituation training proceeded, aggressive responses exhibited marked
decrements, and the majority of males eventually returned to within 0.5 m of
their original calling position and resumed producing exclusively
advertisement calls. Previous work has ruled out sensory adaptation and
effector fatigue as explanations for aggressive response decrements (Bee and
Gerhardt,
2001a,c),
which result from stimulus-specific habituation.
Initial and maximum responses
MANOVA did not reveal any significant differences in the magnitude of
initial responses that were due to the main effect of stimulus (Wilks's
= 0.94, R10,138 =.40, p =.9443), the
main effect of size class (Wilks's = 0.89,
R10,138 =.86, p =.5690), or the stimulus x
size class interaction (Wilks's = 0.72, R20,229 =
1.22, p = 0.2402). Subsequent univariate ANOVAs also failed to reveal
any significant differences for these effects
(Figure 3A,C; Table 2). Similar results were
obtained for the maximum aggressive response, for which there was no
multivariate effect of stimulus (Wilks's = 0.79,
R10,138 = 1.72, p =.0813), size class (Wilks's
= 0.81, R10,138 = 1.58, p =.1187), or a
stimulus x size class interaction (Wilks's = 0.81,
R20,229 =.74, p =.7830). There were also no
significant differences in univariate comparisons of maximum responses
(Figure 3B,D;
Table 3).
|
|
|
MANOVA also did not reveal any significant differences, nor were there
univariate differences (Table
4), depending on whether the AI was positive, negative, or equal
to zero for initial responses (Wilks's = 0.92,
R10,150 =.62, p =.7914;
Figure 3E) and maximum
responses (Wilks's = 0.86, R10,150 = 1.22,
p =.2831; Figure 3F).
In general, there were no linear or nonlinear trends relating the degree of
size asymmetry to the initial and maximum values of the response variables
(Figure 4). However, with a
large sample size of 82 individuals, the initial number of movements was
significantly correlated with the size asymmetry index (r = -.23,
p =.036), but the degree of asymmetry explained less than 6% of the
variation in the initial number of movements. This correlation was not
significant following a Bonferroni correction procedure to account for
multiple comparisons of the AI. All other correlations had absolute values
less than|r| =.19 (p >.10; 1 - ßactual
< 0.43).
|
|
Rates of habituation
ANOVA did not detect any significant differences in the rate of habituation
related to the main effects of stimulus (F2,73 = 1.12,
p =.3315, 2 =.03, 1 - ßactual =.26, 1
- ßexpected >.98; Figure
5A) or size class (F2,73 = 0.47, p
=.6255, 2 =.01, 1 - ßactual =.10, 1 -
ßexpected >.98; Figure
5B), and there was no significant interaction
(F4,73 = 1.54, p =.1982, 2 =.08,
1 - ßactual =.53, 1 - ßexpected >.95). The
number of stimulus periods to reach criterion also did not depend on whether
the AI was positive, negative, or equal to zero (F2,79 =
1.09, p =.3412, 2 =.03, 1 - ßactual
=.27, 1 - ßexpected >.98;
Figure 5C), and there was no
correlation between the AI and the number of periods to criterion (r
= -.13, p =.23, 1 - ßactual =.23;
Figure 4F).
|
The results from the 2 (stimulus) x 3 (size class) x 30
(stimulus periods) ANOVA, which compared responses to the large and small
stimuli broadcast to males in 2000, are presented in
Table 5. There were no
significant effects of stimulus, size class, or stimulus x size class
interactions for any response variables. The repeated measure of stimulus
period was highly significant for all five response variables. For
advertisement calls, there was also a significant stimulus period x
stimulus interaction. As Figure
6A illustrates, the number of advertisement calls started
moderately high, then exhibited a sharp decrease over the first four periods,
suggesting the rapid habituation of evoked advertisement calling during
initial presentations of the stimulus. For males that heard the large
stimulus, advertisement calling remained at this reduced level, while the rate
of advertisement calling by males that heard the small stimulus exhibited a
general increase over periods 5-24, and then a final decrease during the last
few periods (Figure 6A). The
trend for males presented with the small stimulus is explained by an increase
in advertisement calling concomitant with an overall increase in nightly
chorus activity, which begins between 2100 and 2300 h, peaks around 0000-0200
h, and then declines steadily until dawn
(Bee and Gerhardt, 2001a). Why
males presented with the large stimulus did not also exhibit this nightly
trend in calling activity is unclear.
|
|
In contrast to advertisement calling, repeated stimulus broadcasts resulted in decreases in encounter calling, movements, and approach toward the speaker and an increase in response latency, reflecting the overall trend for aggressive responses to habituate with repeated playbacks (Figure 6). The changes in these four variables that occurred with repeated stimulation did not depend on the size of the simulated opponent (nonsignificant stimulus period x stimulus interactions) or the size of the subject (nonsignificant stimulus period x size class interactions), and there were no significant three-way interactions between stimulus, size class, and the repeated measure (Table 5).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The results of the present study strongly suggest that the aggressive response of territorial male bullfrogs does not depend solely on the opponent's acoustically simulated size, the size of the resident male, or the degree of size asymmetry. Because previous work has shown that male bullfrogs can perceive differences in fundamental frequency as small as 9-12 Hz (Bee and Gerhardt, 2001a
,c),
we can be confident that the fundamental frequencies used to simulate a large
male and a small male in the present study were perceived as different. Hence,
the male bullfrogs in this study appeared to ignore size-related variation in
fundamental frequency. Based on results from this field study, I conclude
that, unlike many other frogs, male bullfrogs do not assess their opponent's
size based solely on information conveyed by the fundamental frequency of
advertisement calls.
A comment on statistical power
The conclusion that male bullfrogs do not use fundamental frequency as an
assessment signal is, of course, equivalent to accepting the null hypothesis
of no difference between treatment groups. Therefore, a comment on the power
of the statistical tests reported above is in order. Statistical power refers
to the probability that a statistical test will yield a significant result
(Cohen, 1988). Obtaining
statistical significance depends on the specified type I error rate (),
the size of the study (i.e., the sample size), and the size of the effect,
which is a measure of the degree to which the null hypothesis is false
(Cohen, 1988;
Rosenthal and Rosnow, 1991).
Based on an analysis of effect sizes from this study, the different treatments
in the playback tests had small or negligible effects on the territorial
aggressive response of male bullfrogs compared to the effects demonstrated in
other frogs (.00 2 .11; see
Table 1). The effects of the
subject's body size and the degree of size asymmetry had similar small
effects.
In Table 1, I compare results from previous studies to an analysis of the effect of stimulus size in the present study. For the present study, I computed a one-way (between-subjects) ANOVA on each response variable for males that heard the large and small stimuli in 2000. This analysis provides for the most direct comparison of my results with those of earlier studies. All of these studies share the fact that stimuli simulated large and small males from near the ends of the range of natural variation in body size. Notice in Table 1 that the effect sizes associated with a difference in opponent size in the present study are smaller than those found in other studies, while the sample size of the present study (indicated by the degrees of freedom) is larger than previous studies. Clearly, manipulating a simulated opponent's size by varying fundamental frequency had smaller effects on behavior in bullfrogs than it did in other frogs.
The nonsignificant results reported here are unlikely to represent type II
statistical errors, in which I failed to reject a false null hypothesis.
Although the present study lacked sufficient statistical power to detect the
small effects reported above (.00 1 - ß .68), this study had
extremely high power (1 - ß .97; Tables
2,3,4,5)
to detect the magnitude of effects reported in previous studies that have
demonstrated size assessment in frogs (2 =.29; see above).
This point is best illustrated by comparing the power of the test from the
present study reported in Table
1, in which I compared responses to the large and small stimuli in
2000. Given a sample size of n = 24 males per treatment (total
n = 48), the power of the statistical test in this study (at
= 0.05) ranged from 1 - ß =.67 for the smallest effect from other studies
reported in Table 1
(2 =.12; Wagner,
1989b), to 1 - ß >.99 for the largest effect
(2 =.57; Robertson,
1986). The power to detect the average effect from the previous
studies in Table 1
(2 =.29) was greater than 1 - ß =.98. Perhaps a more
accurate summary of the evidence from this study is that the effects of
size-related variation in acoustic signals are quite small in bullfrogs
relative to the effects found in other frogs.
Comparison with other frogs
Behavioral discriminations between two conspecific signals based solely on
differences in size-related spectral call properties has been demonstrated in
field playback experiments in five species of frogs from four families. Male
natterjack toads (Bufo calamita, Bufonidae) and male cricket frogs
(Acris crepitans, Hylidae) were more likely to abandon calling or
retreat in response to broadcasts of low-pitched calls, while a higher
proportion of males attacked a speaker broadcasting high-pitched calls
(Arak, 1983; Wagner,
1989a,b).
Robertson (1986) demonstrated
differences in the threshold playback amplitude required to evoke encounter
calling and fighting or retreat that depended on the size of the simulated
opponent in Uperoleia rugosa (Myobatrachidae). In the present study,
the same proportion of bullfrogs "attacked" the playback speaker
in response to all three stimuli.
Previous field playback studies conducted with territorial male carpenter
frogs (Rana virgatipes, Ranidae) and green frogs (Rana
clamitans, Ranidae) have also demonstrated that males respond differently
to calls simulating males of different size. Like bulllfrogs, both carpenter
frogs and green frogs have several distinct call types in their repertoire
(Bee and Perrill, 1996;
Given, 1987;
Wells, 1978). Given
(1987) found that male
carpenter frogs produced more single-note aggressive calls and more total call
notes in responses to the calls of a small male (dominant frequency = 622 Hz)
compared to the call of a large male (dominant frequency = 524 Hz). In green
frogs, the type II high-intensity advertisement call is considered an
agonistic signal (Ramer et al.,
1983; Wells,
1978). In response to the calls of a small male (dominant
frequency = 505 Hz), other small males produced more type II calls, but large
males did not. Large males produced more type II calls in response to the
calls of another large male (dominant frequency = 288 Hz), whereas small males
failed to increase their rate of type II calling in response to the large
opponent (Ramer et al., 1983).
Bee and Perrill (1996)
demonstrated that male green frogs lower the dominant frequency of their calls
during simulated territorial intrusions. In follow-up studies, Bee et al.
(1999,
2000) found that the magnitude
of frequency alteration depended on both the size of the simulated intruder
and on the size of the subject. Males lowered their frequency more in response
to simulated large intruders (dominant frequency = 350 Hz versus 400 Hz or 450
Hz), and this trend was most pronounced for smaller males. Wagner
(1989a,b)
also found that the properties of a male cricket frog's calls in response to
playbacks depended on the simulated size of the opponent.
Both carpenter frogs and green frogs are closely related to bullfrogs
(Hillis and Davis, 1986), and
the vocal repertoire of all three species includes distinct aggressive
vocalizations. However, I failed to demonstrate that bullfrogs respond with
encounter calls differently depending on either their own size, their
opponent's size, or the degree of size asymmetry. Thus, it is interesting that
both of these close relatives exhibit behavioral differences in elicited
vocalizations depending on the size of a simulated opponent, while the
bullfrogs in this study did not.
Comparison with game theory predictions
The results of this study are particularly interesting in light of
expectations based on theoretical models of animal communication that predict
the evolution of assessment signaling to settle asymmetric conflicts (e.g.,
Maynard Smith 1982). In
bullfrogs, large body size confers an advantage in physical fights
(Howard, 1978), and the
fundamental frequency of advertisement calls is the best acoustic predictor of
male body size (Bee and Gerhardt,
2001a). At a general level, game theory would predict that the
size-related information conveyed in bullfrog advertisement calls should
function as an assessment signal during aggressive encounters, as it does in
other frogs. However, there was no evidence that the fundamental frequency of
advertisement calls functioned as an assessment signal in bullfrogs.
Enquist's sequential assessment game
(Enquist and Leimar, 1983;
Enquist et al., 1990) makes
predictions about the influence of asymmetries in fighting ability on the
duration and intensity of aggressive encounters. For example, in conflict
situations where the asymmetry in relative fighting ability is large and,
presumably, clear to both contestants, the weaker individual is expected to
quickly realize its low likelihood of winning the contest and decide to give
up early in the interaction. The relatively stronger opponent should be more
likely to persist and escalate to more aggressive behaviors. In contrast, when
the asymmetry in relative fighting ability is small and more difficult to
discern, interactions are expected to last longer and include a higher number
of repetitions of aggressive behaviors.
Based on these model predictions, two predictions could be made for the aggressive response of male bullfrogs to repeated playbacks simulating intruders of different sizes. First, in cases where the asymmetry index is negative (e.g., when small males responded to the large stimulus), we might expect the duration of interactions to be relatively shorter because smaller males would give up early in the conflict. Second, in cases where the asymmetry in fighting ability was small (e.g., when males responded to an acoustically size-matched stimulus), we might expect interactions to be relatively more intense and to last relatively longer. However, there were no indications that the magnitude of the aggressive response and the number of stimulus periods until males stopped responding aggressively to the stimulus were related to the degree of size asymmetry.
Possible explanations for the apparent lack of size assessment in
bullfrogs
Although there are no a priori reasons that bullfrogs should differ from
other frogs in terms of acoustically mediated size assessment, there are at
least three possible explanations for why this study failed to demonstrate
acoustically mediated size assessment. First, this study can be criticized for
using synthetic stimuli that may not have included acoustic elements necessary
for size assessment and, therefore, did not elicit aggressive responses
equivalent to those evoked by vocalizations of real intruders. Although this
valid criticism is difficult to dismiss, I point out that bullfrogs respond
aggressively to these stimuli, and some males continue to respond for several
consecutive nights and for several consecutive hours each night
(Bee, 2001;
Bee and Gerhardt, 2001a).
Differences in the harmonic structure of natural calls and the synthetic
stimulus (cf. Figure 1A and
Figure 2) probably had
negligible effects on size assessment. The bullfrog auditory periphery does
not encode individual spectral components using a rate-place code, but instead
appears to extract spectral information using a temporal code of phase-locked
responses to the fundamental frequency and spectral components primarily below
the fourth harmonic (Schwartz and Simmons,
1990; Simmons et al.,
1992,
1993).
Second, the stimuli in this study were presented from speakers positioned 6
m from the subject, whereas other studies have presented stimuli from much
smaller distances (e.g., 25 cm to 3 m;
Ramer et al., 1983;
Robertson, 1986). Aggressive
responses in frogs are known to vary depending on the perceived proximity of
another calling male (Given,
1987; Robertson,
1986; Schwartz,
1989; Wagner,
1989b), and thus it is conceivable that stimuli presented at a
distance of 6 m were perceived as relatively nonthreatening. However,
individual male bullfrogs exhibit repeated, aggressive approaches toward
playback speakers positioned 6 m away, both within a night and across multiple
nights of testing (Bee and Gerhardt,
2001a). Moreover, males discriminate between familiar territorial
neighbors and strangers over similar distances
(Davis, 1987), and
interactions between adjacent territorial males sometimes occur over these
distances (Bee, personal observation). These observations suggest that male
bullfrogs perceive a new male calling from a distance of 6 m to be a threat to
territory ownership and that males are willing to engage an opponent over this
distance. Whether size assessment in bullfrogs and other species varies as a
function of perceived proximity is a question open to further experimental
study (see Robertson,
1986).
Another explanation stems from the recent finding that male bullfrogs
actively lower the fundamental frequency of their advertisement calls in the
context of male—male aggression (Bee
and Bowling, in press). In sequences of consecutively recorded
calls, males produced advertisement calls with significantly lower fundamental
frequencies following the production of an encounter call, a signal associated
with territory defense. A number of other frog studies have also demonstrated
socially mediated reductions in spectral properties of advertisement calls
during male—male vocal interactions
(Bee and Perrill, 1996; Bee et
al., 1999,
2000;
Given, 1999;
Howard and Young, 1998;
Wagner, 1989a,
1992). Because spectral
properties are negatively related to body size, which often determines
fighting ability, male frogs might lower the frequency of spectral properties
of their calls during aggressive encounters as an attempt to acoustically
inflate their apparent size (Wagner,
1989a,
1992;
Bee et al., 2000).
If frequency alteration represents a dishonest signal of size in frogs,
then natural selection would favor receivers that devalue fundamental
frequency as reliable size assessment cues, especially when other forms of
assessment are not too costly. It is interesting to speculate that in
bullfrogs, the costs of probing an opponent in a close-range interaction to
more directly assess its fighting ability using visual or tactile cues may be
sufficiently low to have permitted some devaluation of fundamental frequency
as a reliable size assessment signal. This is not to say, of course, that
size-related variation in advertisement calls does not play some role in size
assessment. But there is little evidence to suggest that males use this
information before more escalated encounters in which assessment could also be
based on additional visual or tactile information. It is interesting that
Davies and Halliday (1978)
found that male common toads (Bufo bufo, Bufonidae) were more likely
to persistently attack a small male in amplexus when that male was paired with
broadcasts of a small male's high-pitched calls than when paired with the
low-pitched calls of a large male. However, no such discrimination occurred
when these playback stimuli were paired with a large opponent. Presumably,
visual and tactile information also contributed to the assessment of an
opponent's size and fighting ability.
The pitch of frog calls is often cited as a classic example of an
unbluffable assessment signal in animal communication (e.g.,
Alcock, 1998;
Bradbury and Vehrencamp, 1998;
Krebs and Dawkins, 1984;
Wiley, 1983). However, the
results from recent studies demonstrating socially mediated plasticity in
fundamental frequency during agonistic encounters call into question the
reliability of call pitch as an assessment signal during aggressive
interactions between male frogs. The importance of investigating whether call
pitch functions as a size assessment cue is highlighted by the present study,
which demonstrates the absence of size assessment based solely on size-related
variation in acoustic signals. Clearly, frequency alteration and the use of
fundamental frequency as an assessment signal by male frogs deserve additional
experimental and theoretical consideration.
|
ACKNOWLEDGEMENTS |
|---|
I thank Matt Dyer, Susan Harris, and especially Chris Bowling for their indefatigable efforts in the field. Carl Gerhardt, Mike Keller, Vince Marshall, and two anonymous referees provided helpful comments on a previous version of the paper; Josh Schwartz provided generous help in synthesizing stimuli, and Jeff Koppleman and Don Martin provided access to the Little Dixie Lake Conservation Area. I was supported by a National Science Foundation (NSF) Graduate Research Fellowship and an NSF Doctoral Dissertation Improvement Grant. This study was approved by the University of Missouri Animal Care and Use Committee (IACUC 2944 and 3479).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Alcock J, 1998. Animal behavior. Sunderland, Massachusetts: Sinauer Associates.
Andersson M, 1994. Sexual selection. Princeton, New Jersey: Princeton University Press.
Arak A, 1983. Sexual selection by male-male competition in natterjack toad choruses. Nature 306: 261-262.
Archer J, 1988. The behavioural biology of aggression. Cambridge: Cambridge University Press.
Baker JMR, 1992. Body condition and tail height in great crested newts Triturus cristatus. Anim Behav 43: 157-159.
Bee MA, 2001. Habituation and sensitization of aggression in territorial bullfrogs (Rana catesbeiana): testing the dual-process theory of habituation. J Comp Psychol 115: 307-316.[Medline]
Bee MA, Bowling AC, in press. Socially mediated pitch alteration by territorial male bullfrogs, Rana catesbeiana. J Herpetol.
Bee MA, Gerhardt HC, 2001a. Habituation as a mechanism of reduced aggression between neighboring territorial male bullfrogs (Rana catesbeiana). J Comp Psychol 115: 68-82.[Medline]
Bee MA, Gerhardt HC, 2001b. Neighbour-stranger discrimination by territorial male bullfrogs (Rana catesbeiana): I. Acoustic basis. Anim Behav (in press).
Bee MA, Gerhardt HC, 2001c. Neighbour-stranger discrimination by territorial male bullfrogs (Rana catesbeiana): II. Perceptual basis. Anim Behav (in press).
Bee MA, Perrill SA, 1996. Responses to conspecific advertisement calls in the green frog (Rana clamitans) and their role in male-male communication. Behaviour 133: 283-301.
Bee MA, Perrill SA, Owen PC, 1999. Size assessment in simulated territorial encounters between male green frogs (Rana clamitans). Behav Ecol Sociobiol 45: 177-184.
Bee MA, Perrill SA, Owen PC, 2000. Male green frogs
lower the pitch of acoustic signals in defense of territories: a possible
dishonest signal of size? Behav Ecol 11:
169-177.
Bradbury JW, Vehrencamp SL, 1998. Principles of animal communication. Sunderland, Massachusetts: Sinauer Associates.
Capranica RA, 1965. The evoked vocal response of the bullfrog. Cambridge: MIT Press.
Clutton-Brock TH, Albon SD, 1979. The roaring of red deer and the evolution of honest advertisement. Behaviour 69: 145-169.
Cohen J, 1988. Statistical power analysis for the behavioral sciences, 2nd ed. Hillsdale, New Jersey: Lawrence Erlbaum.
Davies NB, Halliday TR, 1978. Deep croaks and fighting assessment in toads Bufo bufo. Nature 274: 683-685.
Davis MS, 1987. Acoustically mediated neighbor recognition in the North American bullfrog, Rana catesbeiana. Behav Ecol Sociobiol 21: 185-190.
Davis MS, 1988. Neighbor recognition in bullfrogs (PhD dissertation). Columbia: University of Missouri.
Dawkins R, Krebs JR, 1978. Animal signals: Information or manipulation? In: Behavioural ecology: an evolutionary approach (Krebs JR, Davies NB, eds). Oxford: Blackwell; 282-309.
Emlen ST, 1968. Territoriality in the bullfrog, Rana catesbeiana. Copeia 1968: 240-243.
Emlen ST, 1976. Lek organization and mating strategies in the bullfrog. Behav Ecol Sociobiol 1: 283-313.
Enquist M, Leimar O, 1983. Evolution of fighting behaviour: decision rules and assessment of relative strength. J Theor Biol 102: 387-410.
Enquist M, Leimar O, Ljungberg T, Mallner Y, Segerdahl N, 1990. A test of the sequential assessment game: fighting in the cichlid fish Nannacara anomala. Anim Behav 40: 1-14.
Given MF, 1987. Vocalizations and acoustic interactions of the carpenter frog, Rana virgatipes. Herpetologica 43: 467-481.
Given MF, 1988. Territoriality and aggressive interactions of male carpenter frogs, Rana virgatipes. Copeia 1988: 411-421.
Given MF, 1999. Frequency alteration of the advertisement call in the carpenter frog, Rana virgatipes. Herpetologica 55: 304-317.
Greenhouse SW, Geiser S, 1959. On methods in the analysis of profile data. Psychometrica 24: 95-112.[Web of Science]
Hillis DM, Davis SK, 1986. Evolution of ribosomal DNA: Fifty million years of recorded history in the frog genus Rana. Evolution 40: 1275-1288.
Howard RD, 1978. The evolution of mating strategies in bullfrogs, Rana catesbeiana. Evolution 32: 850-871.
Howard RD, Young JR, 1998. Individual variation in male vocal traits and female mating preferences in Bufo americanus. Anim Behav 55: 1165-1179.[Web of Science][Medline]
Krebs JR, Dawkins R, 1984. Animal signals: Mind reading and manipulation. In: Behavioural ecology: an evolutionary approach, 2nd ed (Krebs JR, Davies NB, eds). Oxford: Blackwell; 380-402.
Martin WF, 1972. Evolution of vocalization in the genus Bufo. In: Evolution in the genus Bufo (Blair WF, ed). Austin: University of Texas Press; 279-309.
Maynard Smith J, 1982. Evolution and the theory of games. Cambridge: Cambridge University Press.
Maynard Smith J, Harper D, 1995. Animal signals: models and terminology. J Theor Biol 177: 305-311.
Megela-Simmons A, 1984. Behavioral vocal response thresholds to mating calls in the bullfrog, Rana catesbeiana. J Acoust Soc Am 76: 676-681.[Medline]
Morton ES, 1977. On the occurrence and significance of motivational-structural rules in some bird and mammal sounds. Am Nat 111: 855-869.
Ramer JD, Jenssen TA, Hurst CJ, 1983. Size-related variation in the advertisement call of Rana clamitans (Anura: Ranidae), and its effect on conspecific males. Copeia 1983: 141-155.
Robertson JGM, 1986. Male territoriality, fighting and assessment of fighting ability in the Australian frog Uperoleia rugosa. Anim Behav 34: 763-772.
Rosenthal R, Rosnow RL, 1991. Essentials of behavioral research: methods and data analysis, 2nd ed. New York: McGraw-Hill.
Ryan MJ, 1980. The reproductive behavior of the bullfrog (Rana catesbeiana). Copeia 1980: 108-114.
Schwartz JJ, 1989. Graded aggressive calls of the spring peeper, Pseudacris crucifer. Herpetologica 45: 172-181.
Schwartz JJ, Simmons AM, 1990. Encoding of a spectrally-complex communication sound in the bullfrog's auditory cortex. J Comp Physiol A 166: 489-499.[Medline]
Simmons AM, Reese G, Ferragamo M, 1993. Periodicity extraction in the anuran auditory nerve. II: Phase and temporal fine structure. J Acoust Soc Am 93: 3374-3389.[Medline]
Simmons AM, Schwartz JJ, Ferragamo M, 1992. Auditory nerve representation of a complex communication sound in background noise. J Acoust Soc Am 91: 2831-2844.[Web of Science][Medline]
Wagner WE Jr, 1989a. Fighting, assessment, and frequency alteration in Blanchard's cricket frog. Behav Ecol Sociobiol 25: 429-436.
Wagner WE Jr, 1989b. Graded aggressive signals in Blanchard's cricket frog: vocal responses to opponent proximity and size. Anim Behav 38: 1025-1038.
Wagner WE Jr, 1989c. Social correlates of variation in male calling behavior in Blanchard's cricket frog, Acris crepitans blanchardi. Ethology 82: 27-45.
Wagner WE Jr, 1992. Deceptive or honest sigalling of fighting ability? A test of alternative hypotheses for the function of changes in call dominant frequency by male cricket frogs. Anim Behav 44: 449-462.
Wells KD, 1978. Territoriality in the green frog (Rana clamitans): Vocalizations and agonistic behaviour. Anim Behav 26: 1051-1063.
Wiewandt TA, 1969. Vocalization, aggressive behavior, and territoriality in the bullfrog, Rana catesbeiana. Copeia 1969: 276-285.
Wiley RH, 1983. The evolution of communication: information and manipulation. In: Animal behaviour, vol. 2. Communication (Haliday TR, Slater PJB, eds). New York: W. H. Freeman; 156-189.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. L. Basolo Congruence between the sexes in preexisting receiver responses Behav. Ecol., November 1, 2002; 13(6): 832 - 837. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||