An acoustic analysis of the 1988 song of the humpback whale, Megaptera novaeangliae, off eastern Australia

10 Pages·1991·2.1 MB·English

by  A Mednis

AN ACOUSTIC ANALYSIS OFTHE 988 SONG OFTHE HUMPBACK WHALE,
1
MEGAPTERA NOVAEANGLIAE, OFF EASTERN AUSTRALIA
ASTRIDAMEDNIS
Mednis, A. 1991 07 01: An acoustic analysis ofthe 1988 song ofthe Humpback Whale,
Megaptera novaeangliae, off east coast Australia. Memoirs ofthe Queensland Museum
30(2): 323-332. Brisbane. ISSN 0079-8835.
Thispaperpresentsanacousticalanalysisofthe 1988HumpbackWhalesongoffeastcoast
Australia,particularlyofthesound typeswhichcomprisethesong.Theanalysisshowsthat
the song represented generally a well structured form, broadly similar to that of 1982/83,
even though there was a change to a less structured song in 1984 (Cato, this memoir). A
relatively detailedanalysisofthestructuralcomponentsofthe song,specifically thesound
types, has been undertaken. Although the song is relatively stereotyped, variation in the
acousticalcharacteristicsofthe sound typesisobservedwhich may be significant in terms
ofstructuralcomponentsofthesong.ThefullrangestructuralcomplexityintheHumpback
Q
Whalesongmay notbefully known. Humpback Whale,song, marineacoustics, animal
behaviour, marinemammals.
Astrida Mednis, Centre for Marine Science, University ofNew South Wales. Present
address: Australian Heritage Commission, G.P.O. Box 1567, Canberra 2601; 18
February, 1991.
Payne and McVay (1971) recognised that limited. Information transmission depends on
HumpbackWhalesoundsrecordedoffBermuda variations in the song.
occurred in fixed patterns to form songs. They
METHODS
found that individual sound units were grouped
to form phrases which were repeated to form
themes. This ordered sequence comprised the Song recording
song. Song data for 1988 were collected by Dr D.
Based on studies of Humpback Whale songs Cato, during the southward migration, 2—4 Oc-
in eastern Australian waters since 1979, Cato tober, 1988, offStradbroke Island latitude 27°S.
(1984, this memoir) established the existence of Thiswasclose to thepeak timeofthe southward
songstructuresthatwerebroadly similartothose migrationasidentifiedbyChittlcborough(1965)
of the Northern Hemisphere. The song of and confirmedby Paterson and Paterson (1989).
1982/83waswellstructuredwithsix themes,but Recordings were made from a 3m motor boat,
satrraupcitdurcehdansogneg.inC1a9t8o4ilnetardotdoucaerdeltahteivceolnycpeopotrloyf uamspilnigfiaerh,yadrSoopnhyoWnaelaktmaabnouPtro2f0esmsidoenpatlh,WaMpDre6-
"sound type" as a means of categorising the tape recorder. The frequency responses are
sound units according to acoustical charac- uniform within ±3dB over the frequency range
teristics. He found that sound units of the 50Hz-15kHz.
1982/83 song could be classified into 12 types.
I present here an analysis ofthe 1988 songoff Song analysis
easternAustraliawithparticularemphasisonthe Song analysis involved aural and spectro-
characteristics of sound types. These are given graph^ analysis. Most of the energy of Hump-
descriptive names which arc definitive to this back Whale sounds lies in the range 100-4000
paper. Sound types are classified according to Hz and the sounds are distinctive to our ears
acoustical characteristics and some of the sub- (human hearing is within the range ofc. 20Hz-
tleties in variation of characteristics are ad- 20kHz). A considerable amount of analysis of
dressed.This may allow interpretation in greater the songstructurewas therefore done aurally by
detail andcontributetofunctional understanding noting down sound units sequentially using
of structural components of the song. Cato graphic symbols, where each symbol represents
(1984) noted that, because the song is ste- a sound type, thus identifying the order and
reotyped, its potential to carry information is timingofsound units, phrases, themes and song
324 MEMOIRS OFTHE QUEENSLAND MUSEUM
TABLE 1. Phrase structuresofthe 1988 song. cycles of an individual singer and between the
songs ofpresumably different singers. Acousti-
THEME SOUND UNITS IN ORDER OCCUR- cal characteristics for each sound type and
OF OCCURRENCE RENCES phrase structure were determined from
n-chug 19-28 sonagrams produced on a Kay Elemetrics
sonagraph.
1 zp 1 Seventeen hours of recorded data were
oink 1
n-chugevolve intowhistles 11-12 analysed, which included 64 song cycles able to
be analysed. Intensity levels for different sound
cry 2
units were determined with a Bruel and Kjaer
n-whistles 17-20
levelrecordertype 2305,playedthroughaRock-
cry 5-8
n-whistles 3-4 land low pass filter, to remove unwanted low
frequency noise.
cry 3-12
upmoan 3 RESULTS
n-vioiin 1
3
coarseroar 2
down moan SongStructure
upmoan The 1988songstructure (Table I)comprises7
downmoan themes in a fixed order, where each theme com-
upmoan prisesaparticularphrasewith avariable number
downmoan of renditions. This is a well structured song,
4 upmoan similar to that of 1982/83 (Cato,1984,this
flatroar memoir). It has 7 themes and 15 sound types
moanstaccato 3-4 compared with 6 themes and 12 sound types in
1982/83. This shows that the song off east
screal
Australia has returned to a well structured form
upmoan
after the change to a relatively poorly structured
down moan song in 1985. On average, the song consists of
upmoan
about 160sound units,dependingon the number
downmoan
5 of phrase and sound unit renditions in different
upmoan
themes.
flhatmrmoar 5-7 The Humpback Whale song is complex and
relatively stereotyped; considerable analysis
screal maybedone inthedescriptionand interpretation
upmoan ofthesongcharacteristics. Even thoughthesong
downmoan
is relatively stereotyped, there are certain sub-
upmoan tleties in the complex song which are not
downmoan
6 stereotyped.
upmoan
flat roar
whoomp 5-8 Descriptionofthe SoundTypes.
The frequencies of all sound types is in the
screal 1 range 50-8000Hz (more than 7 octaves), the
whoomp
7 lowest in frequency being the "whoomps" and
chain-sawgrowl
1 "downmoan"andthehighestin frequencybeing
the "n-whistles" and "screal" (Table 2). In the
1988 songthe range in the fundamental frequen-
cycles. The aural analysiswas repeated until the cy was 50-4000Hz while Payne and Payne
song structure was fully established. This in- (1985) observed a range of30^-OOOHz in songs
volvedadetailedanalysisofthedata,lookingfor in Bermudan waters over 19 years.
Thesound typesvaryfrom acousticallysimple
patterns in the occurrence of the sound units.
tocomplexandareclassifiedinto fourcategories
Sound units were grouped to form phrases and (Table 2). Thesecategoriesarc determined from
similar phrases were then grouped into themes. sonagrams and are audibly quite distinctive.
Aspatternswereestablished, they were checked Each category can be subdivided according to
for consistency between the different song thefrequencyrange, fundamental frequencyand
1
ACOUSTICANALYSIS, 1988 HUMPBACK WHALE SONG, E. AUST 325
TABLE2. General sound type spectral and lempural characteristics.
SOUNDTYPE FREQUENCY FREQUENCYOF NO. OF TIME
RANGE FUNDAMENTAL HARMONICS
(s)
(Hz) OR LOWESTS.L.* /S.L.*
(H/j
HARMONIC
—risingfrequency
n-v\rustles 600-8000 600-3000 2-5 0.1-0.3
crv 500-7500 500-700 10-11 0.3-1,35
upmoan :oo-4onu 2(10-.150 2-<l6 0.8-1.7
n-violin LOO-6000 1ini -i 0,6
oink 150-2500 150 7 0.4
—fallingfrequency
down moan 50-5^00 50-200 2-19 0.8-2.2
—steadyfrequency
hmm 12i1-4(100 120-250 5-16 0.45-0.7
-collective 0.9-1.7
BROADBANDWITH
SPECTRALLINES
coarseroar 40(^-3750 400-600 2-R 1.2-4.45
Halroar 250-4500 250-350 4-20 0.8-2.4
chainsawprowl 100-41)00 100-1000 6 3.0-5.7
serea 900-8000 900-1450 3-S 2,0-3.5
BROAD BAND
IMPULSIVE
7-P 20I.HJ SO 7 0.5
n-chugs 80-6000 impulsive - 0.2-0.35
COMPLEX
—moanstaccato i 1-3 ii
moan 100-5000 100-340 3-15 0.N-1.8
staccato 300-6800 300—150 3-12 0.3
-whoomp
harmonic 50-1300 50-100 5-11 0,3-0.5
impulsive 600-4000 50-4000 impulsive 0.1-0.5
-pulse 600-4000 600-1100 pulsation 0.1-0.2
*S.L.=spectral lines
duration. The harmonic sound types "cry'" and Harmonicsoundtypes with a risingfrequency
"up moan", and the complex sound type "moan The "cry" sound type is representative ofthis
staccato"(Table2)(whichisactuallyacombina- subcategory (Fig. 1). It lasts 0.3-1.35 sees
tion of two harmonic sounds) have similarities; depending on its position in the phrase in theme
however, differences in fundamental frequen- 2, the only theme in which it occurs. Silences
between cries range from 1.1-2.8 sees.
cies, frequency contour on the sonagram and The "up moan" occurs in 4 different themes
duration separate them (Figs 1-3). The "coarse with some variation in the sound character be-
roar" and "flat roar" (Table 2) are similar; how- tween different themes and within the same
ever, the former has a higher fundamental fre- phrase ofa theme depending on the position of
quency range. The "down moan" is distinctive the sound unit in the phrase, i.e. inter theme and
relative to all othersound types (Fig. 4). intra phrase variation (Figs 2,4). This variation
326 MEMOIRS OFTHE QUEENSLAND MUSEUM
TIME (s)
FIG. I. Sonagram ofan harmonic with a rising frequency sound type "cry".
is in the fundamental frequency and duration. quencies ofthese lines are not harmonically re-
The duration of the sound type ranges from lated. The uscreal" (Table 2) has 3-10 spectral
0.6—2.1 sees. Silences of following units in lines (Fig. 5) and occurs once in 3 different
themes 4 and 5 are 1.5-1.9 sees. themes, consistently as the last sound unit in a
phrase.Soundcharactervariationinthedifferent
Harmonicsoundtypes with afallingfrequency themes is shown in Table 3.
The fctdown moan?1 (Table 2) occurs in four The "chain saw growl" (Table 2) generally
different themes with some variation in fun- occurs at a number of frequencies between 100
damental frequency and duration between and 1600 Hz (Fig. 6). It generally increases in
themes (Fig. 4). duration with each phrase rendition in the
theme. The sound type occurs singularly in the
Broadbandwithspectrallines notharmonic- phrase oftheme 7.
allyrelated
These sound types are not harmonic but have Harmonic sound types with a steadyfrequency
energy spread over some bandwidth, and The <vhmm" (Table 2) occurs in a group of
spectral peaks evident as spectral lines. The fre- units, like the "n-chugs", where 2-^4sound units
TABLE3. Variationof"screal"between themes.
THEME. FREQUENCY FUNDAMENTAL NO. OF TIME
RANGE FREQUENCY SPECTRALLINES (s)
(Hz) (Hz)
4 uixi-Snuo 900-1400 3-10 2.5-3.5
5 900-5500 900-1400 3-6 3.0-3.3
6 1000-6800 1000-1450 5-8 2.0-3.0
ACOUSTIC ANALYSIS, 1988 HUMPBACK WHALE SONG, E. AUST 327
N
3
zn
j**'
£
2
q:
1
2
1
TIME (s)
FIG. 2. Sonagram ofan harmonic with a rising frequency sound type "up moan".
FIG. 3. Sonagram ofacomplex sound type "moan staccato" including a, moan and b, staccato endings.
—
328 MEMOIRS OFTHE QUEENSLAND MUSEUM
(a) (b)
N
HI
>-. 1
>-^
UJ
cr
—
-*»«*
"»,' ••qst
i
o 3
TIME (s)
FIG. 4. Sonagram ofan harmonic with a falling frequency sound type a, "upmoan" and b, "down moan".
may occur in one group. For example, (hmm Broadbandimpulsivesoundtypes
hmm hmm)(hmm hmm hmm hmm)(hmm hmm The impulsive sound types are the "n-chugs",
hmm),whereeachhmmrepresentsonerendition occurring in themes 1 and 2 and "zp" occurring
of the sound unit. Each sound unit is 0.45-0.7 in theme 1. The "zp" occurs only once in the
sees in duration while a collective group may phrase oftheme 1.
extend 0.9-1.7 sees. The period of silence be- The"n-chug"soundtypeoccursasacollective
tween unitsofthe samegroup isc.0.05-0.1 sees unit. Forexample acollection of"n-chugs" may
and between groups there are longer silences of comprise28soundunitrepetitionsinaparticular
0.4-0.9 sees. It occurs in theme 5 onlv. pattern i.e. (Ill) (1111) (1111) (11111)
4
^. <<"_»tt . '
>-
TIME (s)
FIG. 5. Sonagram ofabroad band with spectral lines sound type "screal".
ACOUSTICANALYSIS. 1988 HUMPBACK WHALE SONG. E. AUST 329
N
X
_*:
>
LJ
<3
TIME (s
FIG. 6. Sonagram ofabroad band wilh speciral lines sound type "chain-saw growl .
(Ill 111) where 1=1 chug. So that the whole components which may be harmonic or impul-
collectiveconsistsofsubgroupseachwithavari- sive. The "whoomp'" sound type starts as a har-
able number ofchug repetitions (Fig. 7). These monic and evolves to an impulse with a
unitsare impulsive instructurewith mostenergy pulsating,"possum" ending(Fig. 8).This sound
between 80-2000Hz and extending to 6000Hz. type occurs in themes 6 and 7 with sound char-
Silences of 0.20-0.55 sees between subgroups acter variation between these themes beingofa
and0.15 seesbetween renditionsofunitswithin lower frequency, less harmonic in structure and
one group were observed without (he terminal impulse in theme 7. While
in theme 6 the impulse ending may or may not
Complexsoundtypes occur.
Complex sound types are combinations of Tt is evident from the above discussion that
TIME (s)
FIG. 7. Sonagram ofabroad band impulsive sound type "n-chugs".
330 MEMOIRS OFTHE QUEENSLAND MUSEUM
M (C) (b) <a)(b) (c) (c;
52
TIME (s
KIG.8. Sonagramo1"acomplexsound type - "whoomp' including(a) harmonic, (b)impulse,and(e)pulsating
ending.
there are variations in the characteristics of the VariationinSouxdTypfCharactfr-Tnter-
sound I)pes in the song. tllemf and intraphrasp variation
As sound types are repeated throughout a
DURATIONOFTHESOUNDTYPFS phrase or as they occur in different themes, the
Sound types vary in duration from 0.1-5.7 sound character (e.g. a slight change in fun-
sees, the shortest being the "n-whistles" (theme damental frequency, frequency range or dura-
2) and the longest being the "chain saw growl** lion) may vary depending on the sound type's
(theme 7). Cato observed sound type durations position in thesong. Forexample, in theme4the
of 0.1-4.3 sees in the 1^82 song, where the ''up moan" is emitted 3 times, each rendition
"whistles" werethe shortestand the "chain saw" becoming shorter. However, the same sound
the longest. However, the 'Yhain saws" of the type in theme 5 is generally shorter in duration.
1982 song are not the same sound type as the Additionally, the "moan staccato" may vary in
"chain saw growls" in the 1988 song. In Ber- sound character with repetition in a phrase of
mudan watersthe longest sound type was K sees theme 4 and the number of staccato units may
?
and the shortest <2% of the longest (Payne and vary with each repetition. The staccato endings
Payne 19*5) are mostly present but sometimes absent.
Similarly, the "whoomp" in theme 6, is repre-
Variation inSound Type Abundance and Dis- sented with "possum-like" endings in some
trihitionthroughoutvhp. sonc phrases but not in others,
1'he sound types are variable in terms of dis- The character of a sound type may therefore
tribution through the different phrases of the change with respect to its position in the phrase
song (Table 1). Some sound types are particular or in the song. Although there arc subtle varia-
tocertain themes i.e. "zp" and'oink" to theme 1 tions in the character ofa particularsound type,
and "n-whistle" to theme 2, whilst other sound the sound types are generally similar from one
types are relatively ubiquitous, i.e. 'up moan". rendition of the song to the next. Even though
"downmoan","scrca!"and"flat roar",occurring there issomevariation in theacousticalcharacter
in three different themes and the "whoomp" ofasound type, thisvariation issmall compared
occurring in two different themes. to the differences in acoustical characteristics
Nine ofthe sound types occur more than once between different sound types.
in a phrase. The remaining occur only once in a Perhaps these observed variations can be re-
phrase, for example "n-violin" in theme 3 and, lated to accentuation of meaning i.e. possibly
"screaP and "flat roar" in themes 4, 5 and 6. these variations havespecific functions in terms
1
ACOUSTIC ANALYSIS. I98S HUMPBACK WHALE SONG, E. AUST 331
ofcommunication and general behaviour of the This paper presents an analysis of the song at
singer. A slight modification in the sound type a certain level of magnification, however, it is
may change the signification ofthe significr i.e. not fully understood whether this level ofdetail
the meaning ofthe sound type. isat a scale appropriate to the Humpback Whale
physiological sensory apparatus. It is yet to be
SoundIntensity determined how representative these identified
From a limited sample of6 songcyclesover2 song components are for communication.
days, intensity varied between different sound Analysisofthese structural detailsmay lead to a
types. The "moan staccato" has the highest level greaterunderstandingofHumpbackWhalesong
of all sound types while the "flat roar", "hmm" evolution, learning capacity and behaviour, and
and"screal"arethenexthighest.Thesearcmean the role ofsong in the reproductive success and
values and not always consistent with each possible social structure ofthe species.
phrase rendition. It is important to consider this
feature relative to overall rhythm (Guinee and ACKNOWLEDGEMENTS
Payne, 1988)andpossible"accentuation"ofcer-
tain sound units. Additionally, sound level I would like to thank Dr D. Cato (Defence
diminishes a-t theme 7, which may be due to ScienceandTechnology Organization )forshar-
attenuationofsoundwhen thewhale approaches ing his data and for his enthusiastic encourage-
the surface to breath (Tyack.1981). mentandguidanceand DrM. Augee(University
ofNew South Wales ) for his general guidance.
SilencesbetweenSoundUnits Additionally, I would like to thank Mr G. Plant,
In the 1988song,silencesbetween sound units National Acoustics Laboratory,fortheuseofthe
ofupto3.2seeswereobservedwherethe longest Kay ElemetricsSonagraph.ThanksalsogotoDr
silencesgenerally occurred between sound units R. Paterson and Mr L. Nash for their support in
ofthemes 1 and 7. Shortest silenceswere c. 0. 5 collecting the song data at Stradbroke Island.
sees between "n-whistles" in theme 2 and 0.2 Thanks are also due to the DSTO for providing
sees between "n-chugs" in theme 1. Payne and financial support for this study.
Payne (1985) observed silences as very short or
lasting up to 6 sees. The range in duration of LITERATURECITED
silences between sound units may be important
structural features ofthe song. CATO, D.H. 1984. Recording Humpback Whale
sounds off Stradbroke Island. 285-290. In R.J.
CONCLUSIONS Coleman, J. Covacevich and P. Davie, (eds),
'Focuson Stradbroke',(Boolarong,;Brisbane).
The 1988 songofeast Australiashowsa well- CATO. D.H. this memoir Songs of Humpback
developed structure similar to that in the 1982- Whales; the Australian perspective. Mem. Qd
early 1984,compared tothe"unstructured"song Mus. 30(2): 277-290.
of 1985. CHITTLEBOROUGH, R.G. 1965. Dynamicsoftwo
The 1988 song consisted of 15 sound types. populations of the Humpback Whale, Megap-
Soundtypesrepresentative ofparticularacousti- tera novaeangliae (Borowski). Aust. J. Mar.
cal groups are described with regard to spectral Freshw. Res. 16:33-128.
and temporal characteristics, showing the dif- GUINEE. L.N. AND PAYNE, K.B. 1988. Rhyme-
ferences between sound types and between ren- like repetitions in songs of Humpback Whales.
ditions of the same sound type. The analysis Ethology 79: 295-306.
indicates that while the sound types do not vary PATERSON, R. AND PATERSON, P. 1989. The
to a great extent with different song renditions, Status of the recovering stock of Humpback
there are subtle variations with renditions ofthe Whales, Megaptera novaeangliae, in east
same sound type which may contribute to the Australian waters. Biological Conservation 47:
overall function of the song, for example the 33-48.
informationcontent. Itispossible thatthesesub- PAYNE, R.S. AND McVAY, 1971. Songs of
tletiesfunction ascomponentsofthesongstruc- Humpback Whales. Science 173: 585-597.
ture. A more detailed analysis of the song PAYNE, K. AND PAYNE, R. 1985. Large scale
structure and the sound units may reveal the changes over 19 years in songs of Humpback
function of these structures and thereby con- Whales in Bermuda. Z. Tierpsychol. J. Com-
tribute to the understanding ofthe song. parativeEthology 68:89-114.
332 MEMOIRS OFTHE QUEENSLAND MUSEUM
TYACK, P. 1981. Interactions between singing population comparisons. Behav. Ecol.
Humpback Whales and conspecifics nearby. Sociobiol. 8: 41-46.
Behav. Ecol. Sociobiol. 8: 105-1 16. WINN,H.E.ANDWINN,L.K. 1978.Thesongofthe
WINN, H.E., THOMPSON, T.J., CUMMINGS, HumpbackWhale,Megapteranovaeangliae, in
W.C., HAIN, J., HAYS, H. AND STEINER, the West Indies. Marine Biology 47: 97-114.
W.W. 1981. Song of the Humpback Whale -