Acoustic Characteristics of Lekking Male Greater Prairie-Chicken (Tympanuchus cupido pinnatus) Vocalizations
The vocalizations of relatively few bird species have been described in detail, even though vocalizations are an integral aspect of avian life history. We studied the vocalizations of lekking male greater prairie-chickens (Tympanuchus cupido pinnatus) in 2013 and 2014 to provide baseline information for this species of conservation concern. Our objective was to characterize the acoustic components of the four primary vocalizations in terms of their duration (s), sound pressure level (dB SPL), peak frequency (Hz), dominant frequency (Hz), fundamental frequency (Hz), bandwidth (Hz), and nonlinearities (frequency jumps, biphonations, subharmonics, and deterministic chaos). The boom, cackle, whine, and whoop vocalizations were complex and dominated by low-frequency energy. Booms had the longest duration (x̄ = 1.89 s, SD = 0.18), whines and whoops had shorter durations (x̄whine = 0.32, SD = 0.17; x̄whoop = 0.36, SD = 0.07), and cackles had the shortest duration (x̄ = 0.07, SD = 0.02). Booms had the highest sound pressure level (x̄ = 95 dB SPL, SD = 5), followed by whoops (x̄ = 88, SD = 7). Cackles and whines had the lowest sound pressure levels (x̄cackle = 71, SD = 5; x̄whine = 73, SD = 6). Booms had the lowest fundamental frequency (x̄ = 299 Hz, SD = 13), followed by the cackles and whines (x̄cackle = 355, SD = 32; x̄whine = 430, SD = 81). Whoops had the highest fundamental frequency (x̄ = 622, SD = 69). We suggest characteristics that may be useful in future studies of acoustics in the context of behavior or conservation.
bioacoustics, bird, call, grouse, lek, song, Tympanuchus cupido pinnatus
Vocalizations are an integral aspect of the behavior and life history of avian species. The two overarching functions of bird vocalizations are mate attraction and territory defense (Catchpole and Slater 2008). Specifically, vocalizations are essential for assessing potential mates, finding extra-pair copulations, defending territories, alerting conspecifics to potential dangers, maintaining group cohesion, finding prey, and detecting predators (Barber et al. 2009). Bird songs and calls are complex, and although a rich literature of bird vocalization studies exists, the vocalizations of only a limited number of species have been quantified and described in detail. There is great value in studying bird vocalizations, and more research is needed, especially for species other than songbirds (Benedict and Krakauer 2013).
Study of the acoustic characteristics of avian vocalizations is critical to understanding the behavioral functions of bird songs and calls. Numerous studies have measured bird song or call acoustic characteristics, and the information has led to important findings about the species. For example, Christie et al. (2004) found that pitch shifts and structure of male black-capped chickadee (Poecile atricapillus) songs are indicators of male quality. Lein (1978) suggested that the song variation of male chestnut-sided warblers (Andropadus virens) allows males to transmit more specific signals than otherwise would be possible with only one song type. Ritschard and Brumm (2012) demonstrated that male zebra finch (Taeniopygia guttata) song rate, amplitude [End Page 93] and fundamental frequency, and latency to sing are affected by food availability, suggesting that male zebra finch songs may serve as an indicator of habitat and/or male quality. Each of these studies uses a detailed analysis of a species' vocalizations to provide important information about the species.
Many components of bird vocalizations are shaped by sexual selection, either male-female intersexual selection or male-male intrasexual functions (Catchpole and Slater 2008). Male vocalizations may be an indicator of male quality and territory condition (Catchpole and Slater 2008). Females often use male vocalizations to actively choose a mate, and females often prefer songs that are more complex (Catchpole et al. 1986) or require more energy to produce (Ryan 1988). Although vocalizations are energetically costly to produce, bird song often deters rival males from invading a territory, and the energy expenditure of vocalizing is often less than that of physically fighting intruders (Catchpole and Slater 2008).
Here, we present a quantitative analysis of the vocalizations of male greater prairie-chickens (Tympanuchus cupido pinnatus; hereafter "prairie-chicken"). The prairie-chicken, a species of conservation concern (Schneider et al. 2011), is a medium-sized grouse that resides in the open prairies and oak savannah of central North America (Aldrich 1963; Johnson et al. 2011). Prairie-chickens are known for their polygynous mating system in which males congregate in groups each spring to perform courtship displays and attract females (Breckenridge 1929; Schwartz 1945). These areas, or leks, are distributed throughout the landscape. Males maintain territories within the leks, which they actively defend from other males (Breckenridge 1929; Schwartz 1945; Hamerstrom and Hamerstrom 1960). Females visit leks to observe the lekking males until a suitable mate is found and copulation is accomplished (Schwartz 1945; Hamerstrom and Hamerstrom 1960). After mating, females leave the leks to nest and raise the young independently (Schwartz 1945). Most leks typically have one or two dominant males, positioned near the center, that achieve 71% to 89% of the copulations (Robel 1966).
Vocalizations are an essential component of prairie-chicken leks. Male prairie-chickens primarily use four distinctly different types of vocalizations while lekking: "boom," "cackle," "whine," and "whoop." The boom is a low-frequency, three-syllable sound produced by the syrinx and amplified by the air sacs on the sides of the males' necks (Johnson et al. 2011). Males produce the boom during courtship displays and mildly aggressive encounters with other males (Sparling 1981, 1983). Because of its low frequency, the boom travels a long distance across the landscape and may advertise the presence of a lek to other prairie-chickens (Hamerstrom and Hamerstrom 1960; Sparling 1983). The whoop is a clear and musical call typically produced when females are present on the leks and is often associated with flutter jumps and foot-stomping (Hamerstrom and Hamerstrom 1960; Sparling 1983). A likely function of the whoop is to attract females at close range (Sparling 1981, 1983). The cackle and whine are short vocalizations that are often used together in long, repetitive strings. Cackles and whines are frequently associated with aggressive behavior, and cackles are considered to be slightly more aggressive than whines (Sparling 1981, 1983).
Greater prairie-chickens are the focus of many empirical studies, and interest in the acoustic ecology of the species has grown in recent years. However, very few studies of male prairie-chicken vocalizations have been completed. These studies provide qualitative descriptions (Breckenridge 1929; Hamerstrom and Hamerstrom 1960), information content and functions (Sparling 1981), quantitative information on the number of notes per sequence, duration, internote interval, strongest frequency, and frequency range (Sparling 1983), and variation among individuals (Hale et al. 2014), but there are gaps in the characterization of sound pressure level, fundamental frequency, dominant frequency, bandwidth, and nonlinearities. Our goal was to provide a detailed characterization of the acoustic components of the four primary vocalizations used by male prairie-chickens at leks as baseline data for future studies. Our objectives were (1) to characterize the four vocalizations in terms of their duration (s), sound pressure level (dB SPL), peak frequency (Hz), dominant frequency (Hz), fundamental frequency (Hz), bandwidth (Hz), and nonlinearities (frequency jumps, biphonations, subharmonics, and deterministic chaos), and (2) to use our results to provide suggestions for acoustic measures that would be useful to other studies. Our study was conducted in the absence of obvious influences to establish a baseline data set for future studies of vocalizations. [End Page 94]
We studied male prairie-chicken vocalizations near Ainsworth, Brown County, NE, USA, from March through June 2013 and 2014, as part of a larger project investigating the effects of wind turbine noise on male prairie-chicken vocalizations (Whalen 2015). The study site was associated with a 36-turbine wind energy facility (42.45°N, –99.89°W). We recorded vocalizations at 11 prairie-chicken leks in the area surrounding the wind energy facility. These leks were located at least 1 km from the wind turbines, and we selected these sites to minimize potential acoustic influences of the wind turbines (Whalen 2015).
We used SM2+ Song Meter audio recorders and SMX-II electret, omnidirectional microphones housed in weatherproof enclosures with foam windscreens (Wildlife Acoustics, Concord, MA, USA) to record male prairie-chicken vocalizations at the 11 study leks. The audio settings were the same for all recordings: +36 dB preamplifier gain, 44,100 Hz sampling rate for digitization, and a 2.5 V bias was used to power the microphones. The audio signals were high-pass filtered using a 3 Hz cutoff frequency and were stored in standard, uncompressed 16-bit WAV files. We recorded vocalizations at one lek per morning during 2013 and two leks per morning during 2014. We visited the leks on a rotating schedule and recorded vocalizations at each lek between 7 and 11 times over the two field seasons (Whalen 2015).
We arrived at a lek before dawn, prior to the arrival of the prairie-chickens, and remained until the prairie-chickens finished lekking for the day, or 1000 hours CDT. We placed the audio recorders on the lek at locations where we expected the males would be vocalizing. We used two, two-channel audio recorders and connected 50 m cables to one microphone port on each audio recorder, which allowed us to place microphones at four widely spaced locations on the lek. We placed the audio recorders and microphones on wooden stakes at the approximate head height of a prairie-chicken (25 cm above the ground) (Toepfer and Eng 1988). We placed 12-cm-tall white plastic stakes at 1 m intervals for 5 m extending in four directions around each microphone, to help us estimate the distance between a vocalizing male and the microphone (Whalen 2015).
The audio recorders recorded vocalizations continuously during each lek session. We observed the prairie-chickens from a blind placed on the edge of the lek and listened for "good-quality" vocalizations, which we defined as occurring when a vocalizing male was within 5 m of the microphone and there was minimal background noise. When we heard a good-quality vocalization, we noted the time and microphone number, and estimated the distance between the vocalizing male and the microphone to the nearest 0.5 m. We also video-recorded the lekking prairie-chickens so we could later confirm our distance estimations as well as collect data on additional vocalizations that were not noted in the field. We attempted to record vocalizations from multiple males during each visit to a lek, and to collect different types of vocalizations (boom, cackle, whine, and whoop) from each of those males.
Acoustic Characteristic Measurements
We selected 20 booms, 20 cackles, 20 whines, and 10 whoops from the sound recordings at each of 11 leks to include in the analysis. We selected vocalizations recorded at different times of day, different days of the lekking season, different years, produced by different males, and with minimal background noise and wind contamination (Whalen 2015).
We used Raven Pro 1.4 sound analysis software (Cornell Lab of Ornithology, Ithaca, NY, USA) to measure the duration, maximum power, peak frequency, fundamental frequency, dominant frequency, bandwidth, and nonlinearities of the boom, cackle, whine, and whoop (Fig. 1). See the appendix or Whalen (2015) for additional details about the measurement protocols; the measurement process was similar for all four vocalization types. Duration is the time span of the vocalization in seconds, measured by manually identifying the time interval at which the pressure varied from baseline in the waveform; time points were confirmed by viewing a broadband spectrogram of the call (Fig. 2). Maximum power is the highest power occurring in the vocalization spectrogram, measured in decibels (Charif et al. 2010). Peak frequency (in Hz) is the frequency component of the vocalization corresponding to the peak power and was measured manually using a narrowband spectrum of the call (Fig. 3). Bandwidth (Hz) is the difference between upper and lower frequencies identified at specific levels below the maximum power of the vocalization spectrum (Fig. 4). Fundamental frequency (Hz) is the lowest frequency in the vocalization and was measured [End Page 95] using a narrowband spectrogram, and verified by estimating the frequency difference between harmonics confirming that the two estimates matched; fundamental frequency was measured at discrete times throughout the narrowband spectrogram, and averaged values are reported (Fig. 5). Dominant frequency (Hz) is the frequency with the highest power, measured at specific times throughout the vocalization spectrogram, and average values are reported (Fig. 6). Finally, we evaluated vocalization spectrograms for the presence of four types of nonlinearities: frequency jumps, biphonations, sub-harmonics, and deterministic chaos (Fig. 7) as defined by Riede et al. (2004).
The power measurements acquired from the Raven software are referenced to dimensionless sample units, and therefore do not provide specific information regarding the sound pressure levels of vocalizations. We used different microphones and different audio recorders to record acoustic signals, and prairie-chickens were located different distances from recording microphones. To address these disparities, we calibrated each audio recorder/microphone (see the appendix or Whalen  for details about the calibration process) and presented all levels in dB SPL re 20 μPa normalized to 1 m distance from the source.
We described the following acoustic characteristics: duration (s), sound pressure level (dB SPL), peak frequency (Hz), fundamental frequency (Hz), dominant frequency (Hz), bandwidth at 10 dB below peak (Hz), bandwidth at 20 dB below peak (Hz), bandwidth at 30 dB below peak (Hz), bandwidth at 40 dB below peak (Hz), frequency jumps (presence/absence), biphonations (presence/absence), subharmonics (presence/absence), and deterministic chaos (presence/absence). For numerical data, we calculated the average and standard deviation of each acoustic characteristic for each vocalization (PROC MEANS; SAS Institute, Cary, NC, USA). For categorical data, we calculated proportions and associated standard deviations. We used standard deviations and confidence intervals to make general comparisons of characteristics among vocalization types. We used t-tests to evaluate potential year effects. The differences between years were minor and not critical for characterization of the measures of each characteristic. Thus, we pooled data from 2013 and 2014 for consistency across all analyses.
This research was approved by the University of Nebraska–Lincoln's Institutional Animal Care and Use Committee (permit #901).
Boom vocalizations are spectrally simple, narrowband tonal calls with relatively long durations (Fig. 1A, Supplemental Material Audio File 1; audio files for all vocalizations are available online at http://digitalcommons.unl.edu/natrespapers/624/). Boom vocalizations occupy a narrower range of frequencies than cackles and whines. The fundamental frequency is low and represents the dominant frequency of the call, and successive harmonics are attenuated rapidly. The amplitude and frequency of booms change over the course of the vocalization, and the calls are best described as heavily modulated in the amplitude domain with minor modulation in the frequency domain. Boom vocalizations commonly display frequency jumps and biphonations.
Cackle vocalizations are very short in duration and are often repeated in long strings (Fig. 1B, Supplemental Material Audio File 2). Cackles are tonal in nature and display distinctive broadband, harmonic structure with a frequency range greater than booms and whoops. Dominant frequency is represented by either the fundamental frequency or a higher harmonic. Fundamental frequency and successive harmonics are frequency modulated, and the frequency changes follow a simple arc pattern in which frequency begins low, reaches a peak in the middle of the call, and decreases until the finish. Cackles also commonly exhibit biphonations.
Whine vocalizations (Fig. 1C, Supplemental Material Audio File 3) are also often repeated in long strings, and are sometimes used in conjunction with the cackle. Similar to the cackle, whine vocalizations are tonal, and exhibit multiple harmonics, extending across a broad frequency range, which is greater than the range of frequencies of booms and whoops. Whines also exhibit a complex frequency modulated pattern, and nonlinearities including biphonations and deterministic chaos are typically present at the start and finish of the vocalization.
Whoop vocalizations (Fig. 1D, Supplemental Material Audio File 4) are similar in duration to whines and are also tonal in nature, with a clear harmonic structure, and have a bandwidth between that observed for booms and cackles/whines. They are both amplitude-and [End Page 96]
[End Page 97]
[End Page 98]
[End Page 99]
[End Page 100]
[End Page 101] frequency-modulated, and the frequency modulation follows a similar pattern to that of the cackle. Whoop vocalizations also commonly contain frequency jumps and biphonations.
We measured the acoustic characteristics of 220 boom, 220 cackle, 220 whine, and 110 whoop vocalizations from prairie-chickens located on 11 different leks. Only seven of 35 characteristics (boom sound pressure level, boom fundamental frequency, cackle sound pressure level, whine duration, whine fundamental frequency, whine dominant frequency, whoop bandwidth at 20dB; P < 0.05) showed statistically significant differences between years.
Booms had the longest duration, whines and whoops had shorter durations, and cackles had the shortest duration (Table 1). Booms had the highest sound pressure level, followed by whoops, and cackles and whines had the lowest sound pressure levels (Table 1). The peak frequencies of the booms and whoops had little variation, whereas the peak frequencies of the cackles and whines had large amounts of variation (Table 1). The booms had the lowest fundamental frequency, followed by the cackles and whines, and the whoops had the highest fundamental frequency (Table 1). Booms had the lowest dominant frequency and little variation, cackles and whines had higher dominant frequencies with high levels of variation, and whoops had a dominant frequency similar to the whines but with less variation (Table 1). Booms are narrowband vocalizations, whereas cackles and whines are broadband in nature, and whoops exhibit intermediate spectral width (Table 2).
Each vocalization had one or two nonlinearities that were commonly present; the remaining nonlinearities occurred less often or were absent (Table 3). Frequency jumps and biphonations were common features of booms, subharmonics were observed less frequently, and deterministic chaos was observed rarely. Biphonation was a common feature of cackles, whereas frequency jumps were uncommon, and subharmonics and deterministic chaos events were absent. Biphonation and deterministic chaos were observed frequently in whines, and frequency jumps and subharmonics were rarely observed. Frequency jumps and biphonation were common events in whoops, and subharmonics and deterministic chaos were observed infrequently.
The four vocalizations of lekking male prairie-chickens are distinctive across many of the measured acoustic characteristics. The boom vocalization is markedly longer in duration than the cackle, whine, and whoop (Table 1). Although the cackle and whine are shorter in duration, individual cackles and whines are frequently repeated in long strings, so the duration differences may not be quite as large if the call rate and overall "airtime" of each vocalization are considered. The results of this study are similar to the prairie-chicken vocalization durations previously reported by Sparling (1983) (boom: this study 1.89 ± 0.18 s, previous 2.73 ± 0.89 s; cackle: this study 0.07 ± 0.02 s, previous 0.05 ± 0.01 s; whine: this study 0.32 ± 0.17 s, previous 0.19 ± 0.06 s; whoop: this study 0.36 ± 0.07 s, previous 0.27 ± 0.10 s). The slight differences between the vocalization characteristics of this study and those reported by Sparling (1983) might be due to geographic variation, or differences in recording equipment, or the number of individuals sampled. Sparling recorded booms from 11 individuals, cackles from 10 individuals, whines from 9 individuals, and whoops from 9 individuals, and we recorded vocalizations from more than 50 individuals.
The sound pressure levels of prairie-chicken vocalizations had not been previously measured until this study, so these results fill an important knowledge gap. The higher sound pressure levels of the boom and whoop compared to the cackle and whine (Table 1) may be related to the functions of the vocalizations. The two vocalizations with higher sound pressure levels, the boom and whoop, are used to attract females (Sparling 1983). In contrast, the vocalizations with lower sound pressure levels, the cackle and whine, are used for aggression and territory defense (Sparling 1981, 1983). Higher sound level vocalizations are more energetically expensive to produce (Brumm 2004), so the results suggest that male prairie-chickens expend more energy on vocalizations that are used to attract females than on vocalizations that may be directed at other males. Females often prefer male vocalizations with more energy (Ryan 1988), so it is possible that boom and whoop vocalizations with high sound pressure levels may indicate the quality of the vocalizing male and hence may be an important piece of information in the female's mate choice decision. Amplitude is often overlooked in studies of bird vocalizations (Zollinger and Brumm 2015, but see [End Page 102]
[End Page 103] Shannon et al. 2015 and Whalen 2015), although it is an important acoustic characteristic. Sound pressure levels should be considered in future research in the context of anthropogenic disturbances.
The three frequency measurements (peak, fundamental, and dominant) demonstrate that all four vocalizations are dominated by low-frequency energy (Table 1). Low-frequency vocalizations are fairly unusual among birds. Avian communities often have frequency segregation to avoid heterospecific competition among bird species with temporal overlap in their songs, a phenomenon known as the "acoustic niche hypothesis" (Farina et al. 2011). Although there are other bird vocalizations in the same habitat with frequencies similar to prairie-chicken vocalizations (e.g., sharp-tailed grouse, Tympanuchus phasianellus, "coo" peak frequency: 297 ± 54 Hz; Sparling 1983), prairie-chickens occupy a fairly unique frequency range relative to other avian species. The cackle, whine, and whoop peak frequencies in this study (Table 1) are similar to the "strongest frequencies" previously reported (cackle 760 ± 480 Hz; whine 989 ± 542 Hz; whoop 623 ± 134 Hz; Sparling 1983). The boom peak frequency in the present study (302 ± 13 Hz) is slightly different than the boom strongest frequency previously reported (268 ± 17 Hz; Sparling 1983). Whalen (2015) reported an effect of proximity to a wind energy facility on the fundamental frequency of whines of males, so the assessment of frequency could be useful in future studies.
The bandwidths of prairie-chicken vocalizations had not been previously measured until this study, although Sparling (1983) measured "frequency range" defined as "the absolute difference between the lowest and highest trace of the strongest or carrier frequency." We believe the bandwidth measured in this study is different than Sparling's frequency range metric, although the two measurements may be related. The boom is a narrowband vocalization (Table 2), which means the acoustic energy is concentrated in a narrow frequency range, whereas the whoop has an intermediate frequency bandwidth. In contrast, the cackle and whine are broadband vocalizations, indicating that the acoustic energy is spread over a larger frequency range. In swamp sparrows (Melospiza georgiana), frequency bandwidth is thought to be an important aspect of vocal performance and is subject to sexual selection (Ballentine et al. 2004). The relevance of bandwidth to prairie-chicken sexual selection is unknown, but it is likely that aspects of these complex vocalizations have been shaped by sexual selection. As such, bandwidth may serve as a useful consideration for studies of mate selection and fitness in prairie-chickens or other birds. The vocalization bandwidth is an important determinant of whether or not the effects of anthropogenic noise will mask individual vocalizations, as vocalizations with broader bandwidths are more difficult to detect in the presence of noise (Lohr et al. 2003).
The presence of nonlinearities in prairie-chicken vocalizations had not been previously investigated until this study. Each vocalization type had one or two nonlinearities that were commonly present, while the remaining nonlinearities occurred less often or were absent (Table 3). Thus, we encourage future research to focus on nonlinearities that are not fixed (omnipresent or absent) among vocalizing males.
The boom often contained frequency jumps and biphonations, the cackle typically contained biphonations, the whine had a high presence of biphonations and deterministic chaos, and the whoop often contained frequency jumps and biphonations. Animals living in groups may become habituated to each other's vocalizations, and nonlinearities provide a complexity and unpredictability to the vocalizations that may help them stand out and be more noticeable (Fitch et al. 2002). Blumstein and Récapet (2009) found that yellow-bellied marmots (Marmota flaviventris) had increased responsiveness to calls containing nonlinearities, suggesting that one purpose of nonlinearities in vocalizations is to prevent habituation. Fitch et al. (2002) suggest that nonlinearities in vocalizations promote individual recognition. Because prairie-chickens are a lekking species, the ability of an individual to stand out and be recognized may be important and may serve as a potential explanation for the high occurrence of nonlinearities in their vocalizations. Sexual selection in lek mating systems is very strong and is a complex dynamic between male-male competition and female choice (Höglund and Alatalo 1995). It is likely that selection pressures have resulted in vocalization complexity, which may account for the presence of nonlinear elements in prairie-chicken vocalizations. Thus, the study of nonlinearities in the context of mate selection may prove to be useful for future studies.
Male prairie-chicken vocalizations are complex and dominated by low-frequency energy. The results presented [End Page 104] here are similar to those previously reported by Sparling (1983) for the acoustic characteristics measured in both studies. However, we fill important knowledge gaps with the addition of sound pressure level, bandwidth, and nonlinearity descriptions for male greater prairie-chicken vocalizations, which were not previously known. It is likely that these complex vocalizations have been shaped by sexual selection and are influenced by the functions of the vocalizations. We encourage future studies, in the context of behavior or conservation, to consider measures of acoustics, and our characterizations suggest that inclusion of measures of sound pressure level, frequency, frequency bandwidth, and nonlinearities may be especially useful.
Cara E. Whalen, School of Natural Resources, University of Nebraska, Lincoln, NE 68583
Mary Bomberger Brown, School of Natural Resources, University of Nebraska, Lincoln, NE 68583
JoAnn McGee, Boys Town National Research Hospital, Omaha, NE 68131
Larkin A. Powell, School of Natural Resources, University of Nebraska, Lincoln, NE 68583
Edward J. Walsh, Boys Town National Research Hospital, Omaha, NE 68131
Descriptions of methods used to measure the acoustic characteristics of male greater prairie-chicken vocalizations recorded at leks near Ainsworth, Brown County, NE, USA, in 2013 and 2014.
Duration is the length of the vocalization, measured in seconds. We measured duration by visually estimating the earliest time point indicating an increase in pressure, and the offset as the time point at which the pressure wave returned to baseline values. The duration is the difference between the onset values and offset values (Fig. 2). Time points were verified by examining spectrograms using a Hann window type, with a window size of 10 ms, 3 dB filter bandwidth of 144 Hz, DFT size of 512 samples, and grid spacing of 86 Hz.
Maximum power is the highest power occurring in the vocalization, measured in decibels (Charif et al. 2010). Maximum power measurements were taken from spectrograms viewed in Raven and automatically recovered from Raven software. When measuring boom maximum power, we used a Hann type window, window size of 100 ms, 3 dB filter bandwidth of 14.4 Hz, DFT size of 8192 samples, and grid spacing of 5.38 Hz. When measuring cackle, whine, and whoop maximum power, we used a Hann type window, window size of 50 ms, 3 dB filter bandwidth of 28.8 Hz, DFT size of 4096 samples, and grid spacing of 10.8 Hz.
Sound Pressure Level Calibration Methods
The power measurements acquired from the Raven software are referenced to dimensionless sample units, and therefore do not provide specific information regarding the sound pressure levels of vocalizations. In addition, we used different microphones and different audio recorders to record acoustic signals, and prairie-chickens were located different distances from recording microphones. To address these disparities, we calibrated each recording system in an acoustically and electrically shielded booth located in a quiet laboratory setting. Digitally synthesized tones of known frequency and sound pressure level (dB SPL re 20 μPa), which were confirmed using a precision sound pressure level meter, were used as calibration signals. The calibration signals were recorded using the audio settings used in field recordings for both microphones attached to each audio recorder, in sound files that were 1 minute in length.
Raven software was used to measure the maximum power of each tone recorded on each channel of the audio recorders at the frequency used for calibration (Hann window type, 100 ms window size, 14.4 Hz 3 dB filter bandwidth). The difference between the known sound pressure level (dB SPL re 20 μPa) and the maximum power (dB) was used to compute calibration correction factors for each microphone/channel of each audio recorder. We performed calibrations both before and after the field season, and the resulting values were averaged to produce one correction factor for each microphone for the entire field season. The calibration correction factors were used to calculate sound pressure levels in dB SPL re 20 μPa. In addition, based upon the approximated distance between the microphone and vocalizing male, the levels were normalized to correspond to the level one meter from the vocalizing male.
Peak frequency, measured in hertz, is the frequency component of the vocalization corresponding to the peak power identified in the spectrum of the call. Peak frequency was determined by measuring the frequency [End Page 105] corresponding to the highest power of a narrowband selection spectrum (Fig. 3). A selection spectrum is a graph with power on the y-axis and frequency on the x-axis, and the selection spectrum displays the averaged spectrum of the entire selected vocalization.
When measuring boom peak frequency, we used a Hann type window, window size of 100 ms, 3 dB filter bandwidth of 14.4 Hz, DFT size of 8192 samples, and grid spacing of 5.38 Hz. When measuring cackle, whine, and whoop peak frequency, we used a Hann type window, window size of 50 ms, 3 dB filter bandwidth of 28.8 Hz, DFT size of 4096 samples, and grid spacing of 10.8 Hz.
We used the power spectrum of each vocalization to measure bandwidth (Fig. 4). Bandwidth is the difference between upper and lower frequencies at specific levels below the peak value of the power spectrum, measured in hertz. We measured bandwidth at 10, 20, 30, and 40 dB below the power spectrum peak, when possible. When measuring boom bandwidth, we used a Hann type window, window size of 100 ms, 3 dB filter bandwidth of 14.4 Hz, DFT size of 8,192 samples, and grid spacing of 5.38 Hz. When measuring cackle, whine, and whoop bandwidth, we used a Hann type window, window size of 50 ms, 3 dB filter bandwidth of 28.8 Hz, DFT size of 4,096 samples, and grid spacing of 10.8 Hz.
Fundamental frequency is the lowest frequency in the vocalization, measured in hertz. We identified the fundamental frequency visually in the spectrum (Fig. 5) and verified the assessment by estimating the difference in hertz between harmonics, confirming that the two estimates matched. We measured the fundamental frequency at different time points throughout a vocalization. For booms, we measured the fundamental frequency at 100 ms intervals, and the average value was used in subsequent analyses. For cackles, whines, and whoops, we measured the fundamental frequency at the beginning and end of the call, as well as the maximum fundamental frequency, and the average value for each call was used in subsequent analyses. The maximum fundamental frequency is the highest value of the fundamental frequency, which typically occurred in the middle of the cackle, whine, and whoop vocalizations. We measured the fundamental frequency with the spectrogram slice view, which is a graph of power versus frequency at a specific time point in the vocalization. After creating the spectrogram slice view, we measured the frequency of the peak that corresponded with the fundamental frequency we had located in the spectrogram.
When measuring boom fundamental frequency, we used a Hann type window, window size of 100 ms, 3 dB filter bandwidth of 14.4 Hz, DFT size of 8,192 samples, and grid spacing of 5.38 Hz. When measuring cackle, whine, and whoop fundamental frequency, we used a Hann type window, window size of 50 ms, 3 dB filter bandwidth of 28.8 Hz, DFT size of 4,096 samples, and grid spacing of 10.8 Hz.
Dominant frequency is the frequency with the highest power, measured at specific times throughout the vocalization spectrogram, measured in hertz. In contrast to the peak frequency, which was averaged over the entire call, we measured dominant frequency at specific times in the call. We used the same time points within the spectrogram to measure dominant frequency as we did to measure fundamental frequency (Fig. 6). The averaged value for each call was used in subsequent analyses.
When measuring boom dominant frequency, we used a Hann type window, window size of 100 ms, 3 dB filter bandwidth of 14.4 Hz, DFT size of 8,192 samples, and grid spacing of 5.38 Hz. When measuring cackle, whine, and whoop dominant frequency, we used a Hann type window, window size of 50 ms, 3 dB filter bandwidth of 28.8 Hz, DFT size of 4,096 samples, and grid spacing of 10.8 Hz.
We evaluated vocalization spectrograms for the presence of four types of nonlinearities: frequency jumps, biphonations, subharmonics, and deterministic chaos as defined by Riede et al. (2004) (Fig. 7). We documented whether each type of nonlinearity was present or absent in each call. We only evaluated vocalizations with no background noise, or background noise that occurred at a different frequency than the vocalization, [End Page 106] because we did not want to confuse background noise with nonlinear elements.
We thank Maggie DeLauter, Jeff Lusk, Ian Hoppe, Dan Leger, Jocelyn Olney Harrison, Bill Vodehnal, and Chris Walnofer for their assistance with the project. We thank Jennifer Smith for her valuable input throughout the planning and data collection phases of this project. We thank Matt Gonnerman, Jackie Mather, Taylor Montgomery, Lindsey Sanders, Elsie Shogren, and Nate Turner for their assistance in finding leks. We also thank the numerous private landowners who graciously provided land access for the study as well as the Nebraska Public Power District for providing access to the wind energy facility. Funding was provided by Federal Aid in Wildlife Restoration Project W-99-R administered by the Nebraska Game and Parks Commission. This research was approved by the University of Nebraska–Lincoln's Institutional Animal Care and Use Committee (permit #901).
(1) Conceived the idea, design, or experiment: M.B.B., J.M., L.A.P., E.J.W. (2) Collected the data: C.E.W. (3) Wrote the paper: C.E.W., M.B.B., J.M., L.A.P., E.J.W. (4) Developed or designed methods: C.E.W., M.B.B., J.M., L.A.P., E.J.W. (5) Analyzed the data: C.E.W.