Academia.eduAcademia.edu

Figure 2 – uploaded by Rebecca Dunlop

See full PDFdownloadDownload figure
FIG. 2. Spectrogram (x=time (s), y=frequency (Hz)) of sounds in the nonsong, low-frequency, narrowband category; “wop” (a), “thwop” (b), “snort” (c) “grumble” (d) and “sigh” (e). Spectrograms were generated using a FFT of 4096 and frequency resolution of 5.4 Hz. Social vocalizations which were also part of the song structure are identified by’.

Figure 2 Spectrogram (x=time (s), y=frequency (Hz)) of sounds in the nonsong, low-frequency, narrowband category; “wop” (a), “thwop” (b), “snort” (c) “grumble” (d) and “sigh” (e). Spectrograms were generated using a FFT of 4096 and frequency resolution of 5.4 Hz. Social vocalizations which were also part of the song structure are identified by’.

Related Figures (11)

FIG. 1. Map showing the position of the study site at Peregian Beach, southeast Queensland, Australia (Fig. 1(a)) and (Fig. 1(b)) the arrangement of the base station, visual station (Emu Mountain) and hydrophone array within the study site (see inset, Fig. 1(a)).
FIG. 1. Map showing the position of the study site at Peregian Beach, southeast Queensland, Australia (Fig. 1(a)) and (Fig. 1(b)) the arrangement of the base station, visual station (Emu Mountain) and hydrophone array within the study site (see inset, Fig. 1(a)).
TABLE I. Measurements and a description of the measurements made on all vocalizations. Some of the measurements are illustrated in Fig. 2.
TABLE I. Measurements and a description of the measurements made on all vocalizations. Some of the measurements are illustrated in Fig. 2.
TABLE II. Mean (SD) spectrogram parameters (see Table I) of measured low-frequency, amplitude-modulated, noisy, complex and repetitive sounds, the number measured of each vocalization and the number groups in which they were heard (see Figs. 2-4 for associated spectrograms). All frequency measurements are shown in the linear scale (Hz). Social vocalizations which were also part of the song structure a highlighted along with the song year of which they were part.
TABLE II. Mean (SD) spectrogram parameters (see Table I) of measured low-frequency, amplitude-modulated, noisy, complex and repetitive sounds, the number measured of each vocalization and the number groups in which they were heard (see Figs. 2-4 for associated spectrograms). All frequency measurements are shown in the linear scale (Hz). Social vocalizations which were also part of the song structure a highlighted along with the song year of which they were part.
TABLE III. Mean (SD) spectrogram parameters (see Table I) of measured repetitive and harmonic sounds, the number measured of each vocalization type and the number of pods they were heard in (see Figs. 5—7 for associated spectrograms). All frequency measurements are shown are in the linear scale (Hz). Social vocalizations which were also part of the song structure are highlighted along with the song year in which they were heard.
TABLE III. Mean (SD) spectrogram parameters (see Table I) of measured repetitive and harmonic sounds, the number measured of each vocalization type and the number of pods they were heard in (see Figs. 5—7 for associated spectrograms). All frequency measurements are shown are in the linear scale (Hz). Social vocalizations which were also part of the song structure are highlighted along with the song year in which they were heard.
FIG. 3. Spectrogram (x=time (s), y=frequency (KHz)) of mid-frequency harmonic sounds; “groan” (a), “siren” (b), “short moan” (c), “ascending moan” (d), “modulated moan” (e), “horn” (f), “cry” (g) “modulated cry” (h), “ascending cry” (i), “violin” (j) and “trumpet” (k). Spectrograms were generated using a FFT of 1024 samples and frequency resolution of 21.5 Hz. Social vocalizations which were also part of the song are identified by “.
FIG. 3. Spectrogram (x=time (s), y=frequency (KHz)) of mid-frequency harmonic sounds; “groan” (a), “siren” (b), “short moan” (c), “ascending moan” (d), “modulated moan” (e), “horn” (f), “cry” (g) “modulated cry” (h), “ascending cry” (i), “violin” (j) and “trumpet” (k). Spectrograms were generated using a FFT of 1024 samples and frequency resolution of 21.5 Hz. Social vocalizations which were also part of the song are identified by “.
curred in the 2003 and 2004 song. “Croaks” (Fig. 7(b)) and “yaps” (Fig. 7(c)) were part of the 2002 song (Table III) though were only heard in 2003 and 2004 as part of the social vocalization repertoire. “Yelps” (Fig. 7(d)), “pulses” (Fig. 7(e)) and “low yaps” (Fig. 7(f)) were not heard in the song during the three years (Table III) and were compara- tively uncommon vocalizations (Table III).
curred in the 2003 and 2004 song. “Croaks” (Fig. 7(b)) and “yaps” (Fig. 7(c)) were part of the 2002 song (Table III) though were only heard in 2003 and 2004 as part of the social vocalization repertoire. “Yelps” (Fig. 7(d)), “pulses” (Fig. 7(e)) and “low yaps” (Fig. 7(f)) were not heard in the song during the three years (Table III) and were compara- tively uncommon vocalizations (Table III).
FIG. 5. Spectrogram (x=time (s), y=frequency (Hz)) of amplitude-modulated sounds; “growl” (a), “purr” (b) and “trill” (c). Spectrograms were generated using a FFT of 4096 and frequency resolution of 5.4 Hz (Figs. 4(a)—4(c)) and FFT of 1096 and frequency resolution of 21.5 Hz (Fig. 4(d)). Social vocalizations which were also part of the song are identified by *”. The DFA process correctly classified 78.6% of calls (n :-660) correctly when using the factor scores generated by A one-factor analysis of variance (ANOVA) was con- ducted on all of the measured parameters to compare be- tween subjectively categorized signals. T he means of all measured parameters were significantly different between the 34 proposed signal types (P<0.001). Of t he measured fre- quency parameters (start and end frequencies, minimum and maximum frequencies, frequency trend range ratio, frequency of the spectral pea value was for the maximum frequency o ratio, k) the quency component of the sound, suggesting this parameter varied most between signal types (Fig. 8). The maximum frequency of the lowest frequency com the sounds ranged from 40 to 2500 Hz. High (>100) were also obtained for the end frequency o est frequency component (range of 40-2500 Hz), which also suggests a high degree of variability in this parameter among signal types. Of all the sounds parameters, F values for the frequency highest F f the lowest fre- frequency ponent of F values f the low-
FIG. 5. Spectrogram (x=time (s), y=frequency (Hz)) of amplitude-modulated sounds; “growl” (a), “purr” (b) and “trill” (c). Spectrograms were generated using a FFT of 4096 and frequency resolution of 5.4 Hz (Figs. 4(a)—4(c)) and FFT of 1096 and frequency resolution of 21.5 Hz (Fig. 4(d)). Social vocalizations which were also part of the song are identified by *”. The DFA process correctly classified 78.6% of calls (n :-660) correctly when using the factor scores generated by A one-factor analysis of variance (ANOVA) was con- ducted on all of the measured parameters to compare be- tween subjectively categorized signals. T he means of all measured parameters were significantly different between the 34 proposed signal types (P<0.001). Of t he measured fre- quency parameters (start and end frequencies, minimum and maximum frequencies, frequency trend range ratio, frequency of the spectral pea value was for the maximum frequency o ratio, k) the quency component of the sound, suggesting this parameter varied most between signal types (Fig. 8). The maximum frequency of the lowest frequency com the sounds ranged from 40 to 2500 Hz. High (>100) were also obtained for the end frequency o est frequency component (range of 40-2500 Hz), which also suggests a high degree of variability in this parameter among signal types. Of all the sounds parameters, F values for the frequency highest F f the lowest fre- frequency ponent of F values f the low-
FIG. 6. Spectrograms (x=time (s), y=frequency (Hz)) of “broadband,” “noisy” and complex sounds; “presumed underwater blow” (a), “bark” (b), “bellow” (c), “creek” (d) and “screech” (e) and “scream” (f). Spectrograms were generated using a FFT of 4096 and frequency resolution of 5.4 Hz. Social vocalizations which were also part of the song are identified by *”.
FIG. 6. Spectrograms (x=time (s), y=frequency (Hz)) of “broadband,” “noisy” and complex sounds; “presumed underwater blow” (a), “bark” (b), “bellow” (c), “creek” (d) and “screech” (e) and “scream” (f). Spectrograms were generated using a FFT of 4096 and frequency resolution of 5.4 Hz. Social vocalizations which were also part of the song are identified by *”.
FIG. 8. Mean (+ scanning electron microscropy) plot of log maximum fre- quency of the 34 proposed different vocalization types. The maximum fre- quency was significantly different between vocalization type (ANOVA: P <0.0001). calization types, of which 27 were heard from more than one group. This catalogue is much larger than previously de- scribed social vocalization repertoires (e.g. Thompson et al., 1986). Contrary to previous suggestions (e.g. Tyack 1983; Silber 1986), they were commonly heard in mother and calf pairs and single animals as well as in other social groups such as mother-calf and escorts, pairs of adults, groups of three and groups of four adults. The use of social vocaliza- tions within these different group compositions will be the subject of a further study. However, there are problems with subjectively analyzing sounds. There is a degree of both in-
FIG. 8. Mean (+ scanning electron microscropy) plot of log maximum fre- quency of the 34 proposed different vocalization types. The maximum fre- quency was significantly different between vocalization type (ANOVA: P <0.0001). calization types, of which 27 were heard from more than one group. This catalogue is much larger than previously de- scribed social vocalization repertoires (e.g. Thompson et al., 1986). Contrary to previous suggestions (e.g. Tyack 1983; Silber 1986), they were commonly heard in mother and calf pairs and single animals as well as in other social groups such as mother-calf and escorts, pairs of adults, groups of three and groups of four adults. The use of social vocaliza- tions within these different group compositions will be the subject of a further study. However, there are problems with subjectively analyzing sounds. There is a degree of both in-
FIG. 7. Spectrograms (x=time (s), y=frequency (Hz)) of repeated sounds; “grunts” (a), “croaks” (b) and “yelps” (c). Spectrograms were generated using a FFT of 4096 and frequency resolution of 5.4 Hz. “Pulses” (d), “low yaps” (e) and “yaps” (f) spectrograms were generated using a FFT of 1024 samples and frequency resolution of 21.5 Hz. Social vocalizations which were also part of the song are identified by **.
FIG. 7. Spectrograms (x=time (s), y=frequency (Hz)) of repeated sounds; “grunts” (a), “croaks” (b) and “yelps” (c). Spectrograms were generated using a FFT of 4096 and frequency resolution of 5.4 Hz. “Pulses” (d), “low yaps” (e) and “yaps” (f) spectrograms were generated using a FFT of 1024 samples and frequency resolution of 21.5 Hz. Social vocalizations which were also part of the song are identified by **.
TABLE IV. Synopsis of humpback whale nonsong vocalizations recorded from previous studies (including the fundamental frequency and duration measur ments if stated). Visual comparisons of spectrograms for vocalizations recorded in the following studies were made to spectrograms generated for vocaliza tions recorded in this study.
TABLE IV. Synopsis of humpback whale nonsong vocalizations recorded from previous studies (including the fundamental frequency and duration measur ments if stated). Visual comparisons of spectrograms for vocalizations recorded in the following studies were made to spectrograms generated for vocaliza tions recorded in this study.
Related topics:
GeographyUnderwater AcousticsMedicineAnimal migrationTime Factors
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved