The Auk 110(4):702-715, 1993
STRUCTURE, SEQUENCE AND EVOLUTION
ELEMENTS
IN WILD
AUSTRALIAN
RICHARD
OF SONG
ZEBRA FINCHES
ZANN
Departmentof Zoology,La TrobeUniversity,Bundoora
Victoria3083,Australia
ABSTRACT.--Songs
from 402 Zebra Finches(Taeniopygia
guttatacastanotis)
were sampledin
order to describethe structureof the songphraseand the relationshipof its elementsto the
call repertoire.The songof wild birds was alsocomparedto that of 47 domesticatedZebra
Finches from two Europeanlaboratoriesin order to examine the effectsof domesticationon
songstructure.The stereotypedphrase,which is the repetitive unit of the song,had a mean
number of 6.75 elementsand a mean duration of 0.86 s in wild birds. Elementswere sung
in sequencethat defined three parts to the phrase--a start, a middle and an end. Fourteen
types of elementswere identified of which four were sung by the vast majority of males;
three of these "primary" elements were "borrowed" unmodified from the call repertoire,
and formed the startand end sectionsof the phrase."Secondary"elements,which were less
frequently representedacrossmales,constitutedthe middle of the phraseand appearedto
be modified versions of the Distance-call Element, the loudest element in the phrase. I
tentatively concludethat Zebra Finch songmay have evolved from the callsassociatedwith
flight intention and take-off. Domesticationhas led to changesin element morphology,
frequencyof occurrence,and rate of singing (elements/s),but not in number of elements
per phrase.Received
9 March 1992,accepted
30 November1992.
ALTHOUGH
THE SONG of domesticated
Zebra
repeated one or more times to form the song
Finches(Taeniopygia
guttatacastanotis)
has be- ("bout," Price 1979; "strophe," B•hner 1983,
come a classicalsubject for laboratory-based 1990).The first phraseof the songis preceded
studiesof the development (Immelmann 1969,
by severalidentical notes,the Introductory Elements, some of which are incorporated into
tebohmet al. 1990),and function(Clayton1990a, the phrase pattern itself. Macrostructuralfeab, Clayton and PrOve 1989) of behavior, few tures of song (number of elementsper phrase,
studieshave investigatedthe structuralorga- number of phrasesper song,elementssung per
nization of the song itself, and little informa- second,and number of Introductory Elements
tion about song of the speciesin the wild has per song)have beenstudiedand found to differ:
been published. In order to make senseof the (1) between those songs sung in premating
laboratory-baseddiscoveries that detail the courtshipand those undirected songssung in
specificationsand constraints on the mecha- noncourtshipcontexts(Sossinkaand B•hner
nisms for song learning and its perceptionin 1980, Bischof et al. 1981); and (2) between unthe Zebra Finch, it is necessary,as a first step, directed songsof domesticatedand wild birds
to make a detailed descriptive study of the (SlaterandClayton1991).Bycontrast,therehave
acousticstructure of the song itself and its re- been no studiesof microstructuralaspectsof the
lationship with other acousticsignalsemitted song (namely, the song elementsthemselves,
by the species.
their acousticstructureand sequentialorganiCynx (1990)found that the elementsor syl- zation within the phrase)in either domesticatlables-the smallesttracingson the sonagram ed or wild birdsof the Australian(T. g. castanothat are temporally distinct from neighboring tis) or Timor (T. g. guttata)subspecies.
Thus, my
elements (Eales1985)--are the functional units aim is to describethe structuralfeaturesof song
of song production in domesticated Zebra in wild Australian Zebra Finches and to draw
Finches.Consequently,they are the mostbasic contrasts between two focal colonies some 1,700
unit of analysisin this study.Eachmale has,on km apart in order to examinestereotypyin miaverage,sevendifferentelementsthat he sings crostructure.Songsof wild birds will be comin a set order and together they constitutethe pared with those of domesticatedbirds from
phrase("song-unit,"Price 1979;"motif," B•h- Europe in order to determine what aspectsof
net 1990; "song," Cynx et al. 1990), which is songchangeunder domestication.Variation in
Slater et al. 1988, Williams 1990), control (Not-
702
October
1993]
Song
Structure
ofWildZebra
Finches
703
I analyzed songson a Kay Sonagraph7030A with
more macrostructuralfeatures of song within
and among 34 wild populations is the subject the 150-Hz filter, which provided the most appropriate resolutionfor this study. Phraseswere defined
of a companion paper (Zann 1993).
Investigationsinto the evolutionary precur- as the repetitive unit of the song, and at least three
phrasesof each male were analyzed from the onset
sorsof songand its elementshave not received of singing to establishwhich elementswere repeated
much attention from researchers in this or other
consistently and constituted the phrase. Successive
speciesof songbirds,although Price (1979), phrasessometimeswere separatedby pausesor calls
while studying determinants of song devel- (e.g. Fig. lc), but frequently were not (e.g. Fig. la).
opmentin domesticated
Zebra Finches,noticed Song (bouts) varied considerably in duration and
similaritiesbetween someelementsof the song structurewithin individuals, but the single phrase,
and certain calls from the vocal repertoire.
once learned, was fixed in structure and duration,
However, I have shown that at least one call of
making it an appropriate unit for analysis of song
organization in this species(Zann 1990).
To determinethe durationof the phrasethe sonagram was measuredwith a ruler to the nearestmillimeter (equivalent to 8 ms) from the start of the first
wild-caught Zebra Finches, namely the male
DistanceCall, the loudest and most penetrating
sound emitted, is actually learned from the fa-
ther early in life, and aviary confinementand
domestication affects this processso that the
Distance
Calls of almost all domesticated
birds
element
to the end of the last element.
In order
to
characterizethe types of elements sung by Zebra
Finches, all of the elements from 20 males, each cho-
published to date are aberrant in structure by
comparisonwith thoseof wild birds(Zann 1985).
Consequently,a secondaim is to comparethe
structureof the elementsof the songphraseof
sen at random from the focal colonies (Da and As),
wild (nondomesticated) birds with that of their
calls in order to understand the possiblederivation of the song.
to the nearestmillimeter (equivalent to 80 Hz). Furthermore, the tone and noise componentsthat constitute the Distance-callElement were measuredseparately (see Zann 1984).
I used visual inspection of sonagramsto classify
elementsin the phrasesfrom songsof all 402 males
in the sample into 14 different categoriesor types.
METHODS
Songsfrom 402 males were recorded at breeding
coloniesin two geographicallydistantareasof Australia: the centralarid zone at Alice Springs;and the
semiarid southeasterncorner bordering the Murray
River some 1,700 km away. For comparison, large
sampleswere taken from two focal colonies:40 males
from colonyAs (24ø45'S,133ø52'E)
in centralAustralia;
and 99 males from colony Da (36ø09'S,145ø26'E)in
southeasternAustralia.SeeZann (1993:fig.1) for further details on study sites.
I recordedsongsusingseveralrecorders(Uher 4000
report L, Nagra 4.2 L, Marantz Superscope,Sony
Walkman Professionaland Sony TCD5 Pro) coupled
with either a parabolicreflector(SonyPBR330 with
a BeyerM69 microphone),or a shotgunmicrophone
(SennheiserElectretor MKH 816 T). Most songswere
the undirectedtype, but a few were directed at femalesduring courtship.I recordedfree-flying birds
at their nestingor roostingbushes,but somemales
were temporarilyconfinedto a portableaviary in order to obtain quality recordings.All birds were sexually mature when recorded.Eachsong included in
the study was assumedto have originated from a
different male; this certainly was the casewhere the
birdswere colorbandedor where a colonywasvisited
only once.The possibilityexistsin one colony that
some males recorded one year were rerecordedthe
next, but the high turnover among breeding birds
(Zann and Runcimanin press)makesthis unlikely.
were measured
for duration
If an element
had harmonic
between
to the nearest millimeter.
structure
the first and second harmonic
the distance
was measured
Classification
of acousticsignalsby visualinspection
of sonagramsis a reliable and proven method for
defining natural categoriesand results can be comparable with objectivenumerical methods(Nowicki
and Nelson 1990). Elements were first classified according to the presenceor absenceof tonal structure
and those with tonal structure were further
classified
accordingto the morphologyof the first or second
harmonic.Durationof the elementand the frequency
of the fundamental
were then taken into considera-
tion. The majority of maleshad unique variantsof
each type of element (Zann 1990), but these were
ignored in this study. Repeatabilityof element classificationwas testedin a replicateddouble-blind trial
of 83 elementsrepresenting the 10 most common element types sung by Zebra Finches;elements in the
trial were chosenhaphazardly from a referencecollection and presentedrandomly to the assessor.The
Kappa coefficientof concordance(K; Lehner 1979)
between the original classificationand the classification in the blind trial over all element types was
0.93, with a range of 0.66 to 1.00, depending on the
type of element.
To investigateamplitude of the song elements,I
scoredsongsfrom a sampleof five maleseachfrom
the two focal coloniesusing the energy calculation
optionon a Kay CSLComputerized
SpeechLaboratory 4300 application(vet. 4.0). Recordingsat both
704
R•cH•a•t>
Z.•iN
[Auk, Vol. 110
(b)
(a)
Phrase
1
Phrase
2
0.5s
oc
(C)
Phrase1
Phrase
I
•
NN H NST •
DC
$2
I
I
S3 h
h h
S
C
(d)
2
NN H NST S•
I
DC
S= S3
OC
S
C
Fig. 1. Wide-band-filter
(300-Hz)sonaRrams
of songphrasesand callsof two wild ZebraFinchesfrom
differentpartsof ranse:(a and b) southeastern
Australia;and (c and d) centralAustralia.Callsrecorded
betweenboutsof song:DC,DistanceCall;S,StackCall;C, Communication
Call.Songelements:
I, Introductory;
LN, Ladder-noise;NN, Noise-noise;NS, Noise-structure;NST, Noise-structureTone; H, High; $, Stack;DC,
Distance-call.Different versions of same element type have numerical subscripts.
songsof thesedomesticated
birdswerecompared
with
a subsample
of sonagrams
of comparable
qualityfrom
ary andsang25 to 30 cmfrom the microphone.I chose wild birds (22 songsfrom colonyDa and 14 from
malesat random and analyzed the elementsof the colony As) that yielded 110 elementsin total. Each
five best-qualityphrasesof each.The cursorwasmoved element was cut from its sonagramsheet and any
throughthe amplitudeenvelopeof eachelementan- identifying marks removed. Single elements were
colonieswere made with the sameequipment under
similar conditions; birds were held in an outdoor avi-
(SPL) recorded. A mean was calculated for the five
alyzed on the screenand the loudest reading in dB
drawn from the combined pool (n = 262) and classifiedaccordingto type and origin (domesticatedvs.
replicatesof eachtype of elementsungby eachmale.
wild) in a randomized blind trial.
Data were analyzed with JMP (software for statistical
visualizationon the Apple Macintosh,vet. 2.04;SAS
RESULTS
Institute 1991). Transformations of the data failed to
produceequal variancesso a nonparametrictest was
used (Kruskal-Wallis test).
Photocopiesof sonagramsof undirected songsof
24 domesticated Zebra Finches from St. Andrews,
Phrase macrostructure.--The
number
of ele-
mentsper phraseranged from 3 to 14, with a
mean of 6.76; correspondingly, the duration
ranged from 0.32 to 1.66 s, with a mean of 0.86
s (Table 1). The Pearsonproduct-momentcorrelationbetweennumberof elementsand phrase
United Kingdom,were providedby P. J.B.Slaterand
22 from Bielefeld, Germany, were provided by N. S.
Clayton(e.g.Fig. 2). Songswere recordedunder simduration was 0.816 (P < 0.0001). The regression
ilar conditionsat the two laboratoriesand analyzed
with similar settings. These 46 song phraseswere equation was
subjectedto the samemeasurements
and elementD = 0.167 + 0.1084E,
(1)
classification
proceduresusedon wild birds.The morphology of 152 elements (repeatsomitted) from the where D is duration in seconds and E is element
October
1993]
Song
Structure
ofWild
Zebra
Finches
705
(a)
kHz
'
h
S
?
12 TO H 13 N
S
8
Io
DC
(b)
I•
I1
12
H N N
I
Sl
0.5s
S2
DC
DC
I
Fig.2. Wide-band-filter
(300-Hz)sonagrams
of songphrases
sungby twodomesticated
ZebraFinches
fromtwo European
laboratories:
(a) St.Andrews;(b) Bielefeld.Songelements:
I, Introductory;
H, High;NN,
Noise-noise;TO, Tone; S, Stack;DC, Distance-call;?, unclassifiableelement.
number. The tempo ranged from 5.22 to 12.38
with a mean of 7.90 elements/s.
The durationof the phrasedid not differ significantly between colony As and colony Da.
However, birds from the former had significantly more elementsper phraseand sangthe
phrasesfaster(Table 1).
Typesof elements.--Based
on morphology of
spectralstructure,97.1%of all songelements(n
= 2,711) from 402 males analyzed were successfullyallocatedto 14 categoriesor types;the
remainder were unclassifiable (Table 2). Most
elementsfell into discretecategories,but a few
gradedfrom one to anothervia intermediates,
706
RICHARD
Z•,NN
[Auk,Vol. 110
TABLE
1. Comparison
of macrostructural
parameters
(mediansandinterquartileranges)of songphrases
from
wild (Danaherand Alice Springs)and domesticated(Bielefeldand St. Andrews)colonies.
Wild a
All colonies
Domesticated b
ColonyDa
ColonyAs
ColonyB1
ColonySa
pc
0.86 (0.71-0.97) 0.78 (0.69-0.91) 0.83 (0.70-0.99) 0.65 (0.56-0.77) 0.67 (0.53-0.80) 0.001
7 (6-8)
6 (5-7)
7 (6-8)
6 (6-7)
5 (5.5-8)
0.003
7.4 (6.7-8.2)
8.5 (7.6-9.2)
10.1 (9.1-10.7)
9.9 (8.8-11.6) 0.001
7.9 (7.1-8.7)
Duration(s)
No. elements
Elements/s
No. distance-
call elements
Repeats(per-
1 (1-1)
cent of birds)
I (1-1)
66
99
73
n
! (1-1)
402
! (0-1)
75
40
I (0-1)
95
22
0.001
83
24
• Davs. As (Wilcoxontwo-sample
testwith continuitycorrection):
duration,P = 0.31;numberof elements,P = 0.002;elements/s,p < 0.001;
distance calls, P < 0.001.
• BI vs. Sa (Wilcoxon):duration,P = 0.99;number of elements,P • 0.72;elements/s,P • 0.74;distancecalls,P = 0.10.
ßKruskal-Wallistest (chi-squareapproximation)acrossall colonies(SAS Institute 1990).
ementsthat introducethe phrase.They arebrief,
soft, low-frequencysounds(Table 2) with mulis given by the Kappa(K) coefficientof concor- tiple harmonicsof a chevron-shapewith a proand their classificationwas lessprecise;the degree of ambiguity in classifyingeach element
dance
listed
below.
I found
that
73% of the
nounced
down-slur.
Their classification
was un-
phrases had "repeats" (i.e. two or more ele- ambiguous(K = 1.00). In 60% of males there
mentsbelonging to one type); however, these was only one variant of the Introductory Elewere usually different variantswithin the type ment, but 40% had two (e.g. Figs. la and c) or
and not exact replicates.
even three; these repeats often were sung in
Measuresof duration, frequencyand rate of sequence.Introductory Elementsdid not differ
occurrenceare given in Table 2 for a sampleof significantlyin duration,but the fundamental
all 14 element types,and in Table 3 quantitative frequency of those from central Australia was
comparisonsof five of the most frequent ele- significantly lower than those from southeastments are made for the two focal colonies.
are described
These
below:
(1) Introductory Elements(I) are commonel-
ern Australia (Table 3).
(2) Introductory Diads (II) are Introductory
Elementsin contextand configurationthat oc-
TABLE2. Frequencyof occurrenceof Zebra Finch song elementswith measureof their acousticstructure.
Sample of 2,711 elements from 402 males.
Duration (ms)
Occurrence
(%)
Element type
Elements Males
Frequency (kHz) •
Interquartile
n
Median
Interquartile
range
Median
range
Introductory (I)
23.9
90.0
40
54
46-52
0.40
0.32-0.48
Distance-call (DC)
Stack (S)
16.4
17.0
87.1
80.1
40
40
155
70
139-163
62-85
1.04
0.56
High (H)
8-29
10.!
66•6
40
8
6.40
0.90-1.36
0.48-0.56
5.30-6.72
Noise-noise (NN)
Ladder-noise (LN)
6.9
3.9
39.6
24.9
40
20
124
155
110-147
139-142
-0.96
-0.88-0.96
Noise Structure (NS)
Tone-noise (TON)
3.6
3.2
23.9
21.6
20
20
155
116
141-168
93-116
-0.92
-0.88-1.80
3.2
2.8
2.2
21.6
17.9
13.4
20
20
20
224
163
101
194-253
149-170
85-132
0.96
0.72
0.56
0.88-0.96
0.72-0.80
0.48-0.64
1.8
11.7
20
89
77-101
0.48
0.48-0.54
1.2
0.8
7.7
5.7
20
8
31
244
23-39
207-267
1.52
0.92
0.98-1.76
0.68-0.96
2•9
5.7
Noise-structure
Distance-call (NSDC)
Noise-structure Tone (NST)
Down-slur (DS)
Introductory Dyad (II)
Tone (TO)
Noise-noise Distance-call (NNDC)
Unclassified
•'Fundamentalfrequencymeasuredon 150 Hz filter sonagramsbetweenfirst and secondharmonics,exceptfor H element,which had only one
harmonic.
October
1993]
Song
Structure
ofWild
Zebra
Finches
707
TABLE
3. Duration
andfrequency
offivesong-element
typesfromtwocolonies
(Da,southeastern
Australia;
and As, centralAustralia)1,700km apart(n = 20 for all measures).
Colony •
Variable
Duration (ms)
Frequency(kHz)
Da
As
Introductory Element
54 (46-70)
54 (46-52)
0.48 (0.48-0.48)
0.32 (0.24-0.32)
Distance-call
Duration (ms)
Frequency(kHz)
pb
0.647
0.0001
Element
151 (139-168)
0.96 (0.88-1.16)
155 (139-163)
1.12 (1.00-1.36)
0.957
0.001
Tonal componentof Distant-callElement
Duration (ms)
58 (42-70)
35 (23-46)
0.001
Stack Element
Duration (ms)
Frequency (kHz)
64 (64-78)
0.48 (0.48-0.56)
88 (72-96)
0.56 (0.56-0.64)
0.0005
8 (8-8)
6.68 (6.48-7.08)
0.0001
0.0001
135 116-151)
0.031
0.0004
High Element
Duration (ms)
Frequency(kHz)
27 (16-31)
5.32 (3.12-5.80)
Noise-noise
Duration (ms)
Element
112 (85-143)
Median,with interquartilerangein parentheses.
Wilcoxon
two-sample
testwith continuity
correction
of 0.5(SASInstitute1990).
cur in doublets;usually,both are differentvariants of IntroductoryElements.
(3) Distance-callElements (DC) are the loud-
est and one of the longestelementssung by
Zebra Finches(Table2). There is normally one
Distant-callElementper phrase(Figs.la andc),
but 13% have none and 20% have more than
bands have little frequency modulation. The
versionsgiven by birds in centraland southeasternAustraliaweresignificantlydifferentin
both durationand fundamentalfrequency(Table 3), but the visual classification was unam-
biguous(K = 1.00).
(5) High Elements(H) are the shortestand
highestpitchedof all ZebraFinchsounds(Table
one;colonyAs had fewer than did colonyDa
(Table1). The elementis diagnostic(K = 1.00) 2), and one of the most conspicuouson sonabecauseof two distinct parts to its structure: a
tonal componentof multiple harmonicswith
little frequencymodulationthat precedesa second part, the noise component,which has a
suddendownwardsweepin frequencyof the
harmonics(Figs.la and c). The proportionsof
the componentsvary greatlybetweenindividuals.The Distance-call
Elementis normallythe
ultimate(Fig.la) or penultimateelementin the
phrase.Therewereno significantdifferences
in
grams,but too brief for humansto detectby ear
(K = 1.00). The High Element in central Australia was significantlyshorterin duration and
the fundamental frequency was significantly
higherthan that in southeastern
Australia(Table 3). The harmonic configuration also varied
accordingto location.In centralAustralia,all
35 maleshad a straight,narrow vertical line on
the sonagramand rangedfrom i to 2 kHz (Fig.
duration and fundamental frequency of Dis-
lc), whereas males from southeasternAustralia
had fundamentalswith shapesranging from a
tance-call Elements between
narrow vertical line to chevrons with several
colonies Da and
As, but duration of the tonal component was
significantlylongerin the latter (Table3) and
the noise componentcorrespondinglyshorter.
(4) StackElements(S; Figs. la and c) are similar to thosecalls given by birds when taking
changesin frequencyover a rangeof lessthan
i kHz in most males(e.g. Fig. la).
(6) Noise-noise Elements (NN) have the en-
ergymoreor lessuniformlydistributed
overall
frequencieswithout any spectralstructureand
off (Figs.lb and d). It is a brief soundwith a
lower fundamental frequency than the Dis-
resemblea band of white noise (Figs. lc and 3).
tance-callElement and the multiple harmonic
Element where amplitude modulation pro-
Birds from central Australia
had a Noise-noise
708
RICHARD
ZANN
NSDC
[Auk,Vol. 110
NNDC
--L
H
.t•.'••'•
Distan
Element
ß
DCDC
•':
•li I • •
'•;if•
'.'
•t
LN
TON
TO
H
Fig. 3. Schemashowing possibleorigins of eight song element types from DistanceCall: DC, Distancecall;DCDC,DoubleDistance-call;TON, Tone-noise;DS,Down-slur;H, High; NSDC,Noise-structure
Distancecall; NNDC, Noise-noise Distance-call; NS, Noise-structure; NN, Noise-noise; LN, Ladder-noise; TO, Tone.
duced three to four pulses of noise and were
(10) Noise-structure Tone Elements (NST) are
significantlylonger than thosefrom southeast-
formed
ern Australia (Table 3).
(7) Ladder-noise Elements (LN) are so called
tured component of the Noise-structure Element with the Tone Element (Fig. lc). One of
because
18 was misclassified as a Tone Element (K =
0.94).
(11) Tone-noise Elements (TON) are formed
horizontal
harmonic
bands sit between
two vertical columns of unstructured noise, thus
resemblinga ladder (Figs. la and 3). The duration of the harmonicsvaried considerablybetween singerswith gradationsto Noise-noise
Elements
and Noise-structure
Distance-call
El-
from
the fusion
of the chevron-struc-
from the fusion of a Tonal Element and a shortduration
Noise Element
that lacks the down-
ward sweep in frequency characteristicof the
ements(Fig. 3), producingsomemisclassifica- Distance-callElement (Fig. 3; K = 1.00).
tions (K = 0.88).
(8) Noise-structure Elements (NS) resemble
Noise-noiseElements,but are significantly lon-
(12) Noise-structure Distance-call Elements
(NSDC) are formed from the union of two oth-
ers: the structural component of the Noise-
ger (Wilcoxon two-sample test, z = 3.33, P =
structure
0.001, n• = 20, n2= 40; Table 2) and have several
This is one of the two longestelementssung
chevron-shaped
fundamentalsin the secondhalf
of the sound (Fig. la; K = 0.93).
by Zebra Finches (Table 2). The amount of unstructured noise varies considerablybetween
individuals and gradationsoccurwith the Lad-
(9) Tonal Elements (TO) are brief elements
with a simple harmonic structure whose fundamental frequency is secondonly to that of
the High Element (Table 2 and Fig. 3). Two of
18 were misclassifiedas brief Introductory Elements in the blind trials (K = 0.75).
Element
and a Distance-call
Element.
der-noiseElementleading to misclassifications
(K = 0.66); however, the latter is significantly
shorter in duration (Wilcoxon two-sample test,
z = 4.37, P = 0.0001, n• = 20, n2 = 20; Table 2).
(13) Down-slur Elements (DS) are rare ele-
October
1993]
Song
Structure
ofWildZebra
Finches
709
ments that resemble the noise component of
Distance-callElements,where there is a steep
down-sweepin frequency (Fig. 3). Superficially, they resemble brief Noise-structure Elements,but are significantlyshorterin duration
(Wilcoxontwo-sampletest,z = 4.71,P = 0.0001,
median (interquartile range) amplitude of the
n• = 20, n2 = 20; Table 2).
(14) Noise-noise Distance Calls (NNDC) are
56.1 (52.6-59.1).
Distance-call Element was 67.7 dB SPL (58.1-
83.3),which wasmuchhigher than that of other
elements:Noise-noise, 57.0 (55.0-59.9); High,
56.7 (50.1-62.6);
Noise-structure
Tone, 55.3
(52.8-58.6); Stack,56.0 (52.6-61.9); Introductory
Phrasestructureand elementsequence.--Elements that form the Zebra Finch song phrase
basicallyhave a more or lessdefined location
and order that divide the phrase into three
parts--a start, a middle and an end (Table 4).
The structureof the phrase can be best understood by first consideringthe pattern in the
middle where the type of elementsand their
sification trials.
sequenceis more determined than elsewhere.
Simple measurementscould not encompass The middle consistsof two types of elements:
somefine details of element morphologythat the conspicuousHigh Element; and various
were hard to categorizeand even harderto mea- typesof elementswith unstructurednoise(e.g.
surebut, nevertheless,still might be diagnostic Figs. la and c). It is the presenceand location
features of the phrases from colonies Da and of thesetwo typesof elementsthat dividesthe
As. In order to assess
this possibility,I set up a phraseinto three partsin 95%of males.In 54%
trial in which sonagramsof songphrasesfrom of males, the middle of the phrase had three
both colonieswere assessed
subjectivelyin a elements,the High Element precededand foldouble-blind matching experiment. Phrases lowed by elementswith a significantnoisecomfrom 20 males, which were recorded with the
ponent. In 29% of males,there were only two
sameequipmentunder identicalcircumstances, middle elements,mostwith a noisecomponent;
were chosenat random from each colony and only 30%of thesemaleshad High Elementsand
the sonagrams
analyzedwith the samesettings. they were placedin the secondposition(Table
The task of the assessorwas to sort sonagrams 4).
In colony As the three middle elements
subjectivelyinto two groupsbasedon similarity
of the detailed morphologyof the elements.I (Noise-noise,High, Noise-structureTone) were
ran two trials with different and independent identical in type and order for 35 of 40 birds
assessors. Each trial was run such that there was
sampledsothat the threeelementscouldalmost
random sampling with replacement between be considereda one-elementcomplex(Fig. lc).
the 30 matchingdecisionssothat eachdecision Birdsfrom colonyDa were more variable in the
was independentof thosethat precededit. The type of noise elements (Ladder-noise, Toneproceduresallowed chancefrequenciesof the noise, Noise-structure), but the High Element
result to be calculatedusing the binomial dis- was still in its central position (e.g. Fig. la) or
tribution where the samplesizewas 30 and the the secondpositionif only two middleelements
probability0.5. In eachtrial assessors
allocated were present.
27 out of 30 and 28 out of 30 matchescorrectly;
Despitethe pivotal nature of the High Elethe chanceprobabilitiesof suchan outcomeare ment in the organizationof the phrase,it was
less than 0.0001 in both instances. "Setrations"
presentin only 66.6%of 402 birdsanalyzed (Taproducedby rapid frequencymodulationof the ble 2). This value is biasedby the large sample
tonal partsof elements,in particular,Distance- from colony Da where the High Element was
call and Stackelements,were diagnosticof birds present in only 27.2% of 99 males. The High
from central Australia.
Elementwaspresentin 79.3%of birds recorded
Amplitudeof elements.--Nineelement types from seven other colonies within 40 km of colwere analyzed,but only six were sung by more ony Da (n = 174). These proportions are sigthan six of the malesinvestigatedand included nificantly different (G --- 73.04, df = 1, P <
in the results.There was a significantdifference 0.0001).
in amplitude among the elements (KruskalThe firstand lastsectionsof the phraseconsist
Wallis test, X 2= 689.23, df = 5, P < 0.0001). The of three typesof elements,the Introductory Elelements where the Distance Call component
is precededby a noise componentto form one
of the longestelementssungby Zebra Finches.
They resembleLadder-noiseElements,but are
muchlongersuchthat the durationsof the two
elementsdo not overlap. Its rarity prevented
sufficientsamplesbeing availablefor the clas-
710
R•Cm•RVZANN
[Auk, Vol. 110
TABLE
4. Mainsequence
of elements
• in threesections
thatconstitute
songphraseof wild ZebraFinches
(n
= 402).Of males,5%hadphrases
with no middlesection,
5%hadno startsection,
and 15%hadno end
section.Percentage
frequencyof mainsequences
within sections
shown.
Start b
1
Middle c
3
Percent
-I/S/DC
53
15
NN
NN
2
End a
3
Percent
H
H
NST/NN
NS/NSDC
21
7
S
S
1
2
I
I
--/I
S
I
DC
--
4
TON
H
NS/NSDC
14
I
S
DC
DC
I/DC
I/S
I/S/DC
I/S/DC
I/DC
7
4
5
LN
LN
NN
H
NS/NN/H
NS/NN/H
NS/NSDC
---
5
9
6
1
2
3
Percent
--/DC
DC
-I/S/DC
28
8
DC
--
--
14
DC
DC
I
S
S
S/DC
-I/S/DC
--/DC/S
13
5
11
"Element
abbreviations:
I, Introductory;
S,Stack;
DC,Distax•ce-call;
H, High;NN, Noise-noise;
NS,Noise-structure;
NST,Noise-str•cture
Tone;
TON, Tone-noise;NSDC, Noise-stvacture
distance-call;
LN, Ladder-noise;
--, no element;/, alternatives.
•'27%of startshadoneelementonly,36%two elements,
23%threeelements,
and8%fourelements.
• I0%of middleshadoneelementonly,29%two elements,
54%threeelements,
and4%fourelements.
• 28%of endshadoneelementonly,41%two elements,19%threeelements,and 13%fourelements.
ement, the Stack Element and the Distance-call
tory (G = 9.95, df = 4, P = 0.041) and High
Element; however, 80% of all Introductory El-
elements (G = 20.95, df = 2, P < 0.001), and
more elements that could not be classified (G =
ements occur in the first section, and 81% of
Stack Elements
and 80% of Distance-call
Ele-
mentsoccurin the last part of the phrase(Table
4). The Distance-call Element, which is the
loudest in the phrase and most resistantto at-
tenuation,appearsto serve as a "coda" or
"punctuation" separatingone phrasefrom the
next.
I found that 15%of maleshad songsthat end-
26.48,df = 1, P < 0.001;Fig. 2a). The latter were
often new typesformed by the fusion of High
Elementswith others,frequently, Introductory
Elements(Figs.2aand b). In contrast,wild birds
sangmoreDistance-callElements(G = 34.61,df
= 1, P < 0.001), Stack Elements (G = 7.90, df =
1, P = 0.005) and Ladder Elements (G = 17.56,
df = 1, P < 0.001).
In the subjectiveblind-matchingtrial, 85%of
elementswere classifiedcorrectlyaccordingto
no end section; however, in these cases there
whether they originatedfrom domesticstocks
was a significantly greater number of elements or wild colonies;there was no significant difin the first section of the phrase than those ference in the proportion of correctand incorfor the two categoriesof singphraseswhere three sectionswere distinguish- rectclassifications
able (Wilcoxon two-sampletest, z = 6.76, P = ers (G = 1.12, df = 1, P = 0.29). Thus, elements
birds
0.0001,n• = 52, n2= 321). This suggeststhat the of thesametypein wild anddomesticated
end section has not been omitted, but simply are distinct in detailed morphology.Incorrect
involved all element types but
placed in the wrong order, thereby adding to classifications
similarelementtypesfound in the first part of mainly Introductory, High and Noise-structure
elements.
the phrase.
Songsof domesticated
Zebra Finches.--There
were no significant differencesin macrostructural featuresof songbetweenthe samplesfrom
DISCUSSION
Bielefeld and St. Andrews (Table 2), so they
Fairly clearrulesof songorganizationemerge
were pooledand comparedto the whole sample
(n = 402) of wild birds. There were no signifi- from thisanalysisof the stereotyped
songphrase
cant differencesin the number of elementsper of wild Zebra Finches: (a) song elements most
ed with
elements
characteristic
of the middle
sectionof the phrase,thus appearingto have
phrase(Wilcoxontwo-sampletest,P = 0.758), similar to callsbracketthe phrase;(b) two of
but wild birds sang more slowly (P = 0.0001)
and, hence, had a significantly longer (P =
0.0001)phrasethan the domesticatedbirds (Table 1). Differences also existed in the rate of
occurrenceof different element types.Domesticated birds sang significantly more Introduc-
these,the StackElement and the very loud Distance-callElement often end the phrase;(c) another set of elements that are similar to soft calls
begin the phrase;(d) the compoundelements
that makeup the middle of the phraseand confer most of the variation to the phrase do not
October1993]
Song
Structure
of WildZebraFinches
closelyresembleany callsand could be consid-
711
last second before take-off; the sustained har-
ered the most "derived."
monic structureshown in Figureslb and d is
The IntroductoryElement, the Distance-call normally distorted by rapid frequency modu-
Element,
and the Stack Element
are the most
lation (ca. 60 Hz) at the moment of lift-off. This
frequentlysungelementsand arefoundin more
than 80% of males.They appear to have been
borrowed almostunchangedin structurefrom
the mostfrequentlyutteredvocalizations
in the
callrepertoire,namelythe CommunicationCall,
call, which is the same in both sexes, is invari-
of males sampled. It has no equivalent in the
call repertoireand its derivation is problematical; it may have been modified from the tonal
cessivephrasesof an Alice Springsindividual,
two distinct Distance-callElementswere sung,
ably followed by one or more Distance Calls
(see Zann 1984).
The Distance-call
Element
is identical
to the
sexuallydimorphic DistanceCall in about 50%
the StackCall and the DistanceCall, respec- of males(e.g.Figs.la and b), with the remainder
tively. Exceptfor colonyDa, the High Element having a slightly different variant (e.g. Figs. lc
was alsofound in the songsof more than 80% and d; see Zann 1990). However, in two suc-
component of the Distance-call Element. Wil-
one of which
was identical
to his Distance
Call.
Price(1979),while investigatingsongdevel-
liamset al. (1989)believedthat the High Ele-
opment in domesticatedZebra Finches,was the
ment represents an extreme form of harmonic
suppression in which the fundamental fre-
first to notice the strongresemblancebetween
somesongelementsandthreetypesof callnotes,
which he termedshort,medium and long. These
correspondto the Communication,Stackand
quency and many lower harmonicsare suppressedby the syrinx, but a higher one is em-
phasized. Variable degrees of harmonic
suppressioncharacterizethe structureof many
elementsof Zebra Finch songand thesecanbe
perceivedby both sexes(Cynx et al. 1990).The
structureof the High Element is exceptional
becauseit has a fundamentalfrequencyfive
timesgreaterthan the next-highest-pitchedel-
Distance calls in the wild birds studied here;
however,the harmonicmodulationof the long
call of the domesticatedbirds was completely
different from that of the Distance Calls in: (a)
omitting the noisy downward sweep component and, thus, resembling the female version
of the call; or (b) placing it at the start of the
ement in the song;nevertheless,it is still in the
range of frequenciesto which Zebra Finches
are mostsensitive(Okanoyaand Dooling 1987).
I have termed these four frequent and widely
sung elements the "primary elements" of the
phrasebecausethey do not appearto havebeen
recently derived from other elements in the
call rather
than
at the end.
These and other
configuration;
thebriefestandmostfrequently
versions of the Communication
aberrations(e.g. Fig. 2b) are widespreadin domesticated
Zebra Finches and are associated with
errors in the learning of this call in captivity
(Zann 1984, 1990).
In contrast to the primary elements of the
song phrase, the remaining 10 element types
song.
are sungfar lessfrequently within and among
The structureof the IntroductoryElements males (Table 2), and it is difficult to find their
strongly resemblesthat of the "Tet" Commu- exact counterpartsin the call repertoire, but it
nicationCalls first identifiedby Immelmann is possiblethat these"secondaryelement" types
(1962).The callis not individuallystereotyped may have been derived from them. The Introin duration, but all harmonics have the same ductory Diad is simply an associationof two
Call, and the
uttered version(C in Fig. lb) matchedmost Noise-structure Tone Element appears to be a
IntroductoryElements(e.g.I2in Fig.la), where- fusion of a noise componentand the Tone Elasotherelementsare longer(e.g.I• in Figs.la ement.
andb), but thesealsoresemblelongerversions It is possiblethat the remainingeight of these
of thecall.Thus,theIntroductory
Elements
are secondaryelements originated from the Dislargely unmodified versions of those found
among the Communication Calls. These soft
tanceCall through systematicrepetition, modificationand omissionof the spectraland noise
calls, which are identical in structure between
componentsof the element (Fig. 3). Through
the sexes,accompany
mostnonflyingmove- omissionof the tonal componentit is possible
mentsand serveas a closecontactsignal between the partners.The StackElementis identical to the Stack Call, which is uttered in the
to derive the Down-slur Element and, conceiv-
ably, through omissionof the lower harmonics
in the noisy down-slur component a version of
712
RIcn•l•I•Z•,II•I
[Auk,Vol. 110
the High Element could form. Other elements
may have arisen from a rare "mistake" in the
sion of a Noise-noise/Noise-structure
Distance Call (Zann 1984, 1985) and/or
The absenceof matchingof derived elements
to the duration and frequency measuresof the
Double Distance-callElement may be due to
drift or the extent of the evolutionary changes.
Dis-
tance-call Element, where two distinct but un-
usually brief DistanceCalls have been fused to
form a doublet with four components, tonenoise-tone-noise,
the Double Distance-call
El-
ement (Fig. 3; DCDC); although the samplesize
is small, the doublet is only about 20% longer
than the mean duration for a single Distancecall Element. Through progressive loss of the
first tonal componentand modification of the
first noisy down-slur component,it is possible
with
a Distance-call
Element
Element.
Noise-noise Distance Calls, Noise-structure
Distance Calls and the Down-slurs all have small,
but significant (P < 0.05) negative correlations
(Pearson product-moment correlations -0.15,
to derive the Ladder-noise Element, the Noisenoise Distance Call, and the Noise-structure
-0.12, and -0.14, respectively)with the frequency of occurrenceof the Distance-callElement. Conceivably, these three elements are
more recently derived than other secondaryelementsthat have no significantcorrelationswith
Distance-call
the Distance-call Element (Noise-structure, 0.12;
Element.
The
Ladder-noise
Ele-
ment may reduce then omit entirely the second
tonal componentso that it forms two contiguous noise components, the Noise-noise Element.
The Tone-noise
Element
Ladder-noise,0.06;High, 0.04;Tone, 0.05;Tonenoise, 0.05; Noise-structure
Tone, 0.02; Noise-
noise, 0.00). Further clarification of the deri-
can be formed
vation of song elements may be possiblewith
studiesof song development.
The sequenceof primary elementsin the song,
second Distance Call, and the loss of the second
namely Introductory, Stack and Distance-call
noise component forms the Tone Element, elements,followsthe sequenceof callnotesgivwhich may conceivablyhave produceda High en by birds just before and during take-off,
Element with the loss of its lower harmonics.
whereby with increasingarousalthe soft series
by the lossof the noisy down-slursof the Ladder-noiseElement, leaving the tonal part of the
In the Noise-structure
Distance-call
Element
the
of Communication
Calls switch over to the Stack
first tonal componentis omitted and both noisy Calls that are followed by the very loud Disdown-slurs have developed clear spectrawith tanceCalls just after take-off.All three callsare
the first having a chevron-shapedfundamental. not given at every take-off,but the sequenceis
The origin of the Noise-structure Element is the same.The sequencewithin the songphrase
ambiguous;it may have arisenfrom spectraap- is not as stereotyped: 88% of songs have the
pearing in the Noise-noise Element or, con- Introductory Element preceding the Stackand
versely,it may have arisenfrom the lossof the Distance-callelements,but only 60% of songs
spectralcomponentsor by the lossof all tonal have the StackElementprecedingthe Distancefeatures
in the Ladder-noise
Elements.
call Element.
The Distance-callElement is not significantly
different
in duration
from the Ladder-noise
El-
Structuring of the song phrase into subsections in which call-like elements precede and
ement (Wilcoxon two-sample test, z = 0.65, P = follow a sequenceof secondaryor noncall-like
0.51),but the frequencyof the tonal component elements has a functional basis.Young domesis significantly higher (z = 2.52, P = 0.01). If ticated males learn their songsin "chunks" or
the Ladder-noise
Element is derived from the
blocksof elements(Williams and Staples1992),
two very brief calls that constitutethe Double and experimentaldisruption to the production
Distance Call, one would expect it to be about of songin adultsresultsin lossand recoveryof
20%longer than a normal DistanceCall. In con- chunks of elements (Williams and McKibben
trast, the durations of the Noise-noise Distance
1992). It is significant that these chunks correCalls and the Noise-structure Distance Calls,
spond to sectionsof call-like and noncall-like
the longestelementssung,are about60%longer elementsthat have been identified in this study
than the Distance-call Elements (z = 4.31, P =
of wild Zebra Finch song.
0.0001, and z = 5.75, P = 0.0001, respectively),
It is possiblethat the ancestralsong of Zebra
but not 20%longer asone might expectif they Finchessimply consistedof three types of call
were derived from the Double Distance-call Elnotes given in a stereotypedsequence,since
ement, or 100%longer if derived from the fu- most elements in the phrase clearly have their
October
1993]
Song
Structure
ofWildZebra
Finches
713
originsin the call notes.The primitive ancestral presumably,this is due to learning from the
song may have been derived unmodified in father.
According to Simpson and Vicario (1990),
structureand sequencefrom the callsassociated
with preflight and take-off. Subsequentevolu- there are important sex differencesin Zebra
tion involved the addition of soundsunique to
Finches in the neural control of Distance Calls.
the songby systematicmodificationsto the Distance-callElement and their incorporationinto
the sequence positioned between the Com-
DistanceCalls in femalesare controlled by an
evolutionarilyprimitive pathway,found wide-
munication
Calls and the Stack Call. These char-
acteristicsof song structureare not evident in
the Timor subspecies
T. guttataguttatanor have
they occurredin relatedAustralianestrildines.
The Timor Zebra Finch does not incorporate
DistanceCalls into its song phrase(Clayton et
al. 1991), nor is there any pronounced sexual
dimorphismin thesecalls(Zann 1984).Poephila
acuticauda,
P. cinta and P. personata,
closerelatives of the Zebra Finch, do not have sexually
dimorphic Distance Calls (Zann 1975), nor do
they incorporatethem into their songs(Zann
1976).I concludethat the evolutionarychanges
that have led to songorganizationin the Australian Zebra Finch are relatively recent.
The Distance Call appears to have had a
unique role in the evolution of Zebra Finch
song. It not only appears to serve as a source
for new song elements, but has a similar developmentalprogramto thatof song(Zann 1985)
and sharesthe sameneural pathways(Simpson
and Vicario 1990). Cross-fosteringexperiments
with offspring of wild-caught Zebra Finches
have shown that the information required for
the development of the sexually specific features of the male Distance Call are acquired
from the father before 40 daysof age.This information provides for the masculinization of
its much simpler,juvenile, and more feminine
precursorbetween 40 to 60 days of age. The
masculinization
leads to sex differences
in call
duration, fundamental frequency and modulation (Zann 1985). Only about 30% of laboratory malescopythe father'sDistanceCall exactly,but the remainderdevelopcallsthatshow
a strong resemblance.Femalesdo not learn the
Distance Call from the mother, but retain the
juvenile version, more or less.The structure of
the DistanceCall can convey,potentially,information about the subspecies,sex, colony,
family and individual (Zann 1984).In about80%
of free-living malesthe song'sDistance-callEl-
ly in bothpasserines
andnonpasserines,
whereby a structurein the brain stemgovernsthe
respiratorypatterning necessaryfor the production of the call. In contrast, male Zebra
Finches have an additional pathway superim-
posedon the primitiveone,wherebycomplex
syringealcontrolis providedfor productionof
the Distance Call. This second pathway includes the same neural structures involved
in
control of Zebra Finch song, namely the telencephalicnuclei (HVC and RA) and the tracheosyringealnerves innervating the syrinx,
which allow complexvocalizationsto be producedand imitated. When Simpsonand Vicario
(1990) disruptedthis pathway by nerve section
and lesion,they removedall the learned featuresof the male call, leaving a basiccall characteristic
of the female.
The primitive, ancestralsong phraseprobably consisted
of the unlearnedcall notesafter
which the undifferentiated,
bisexual Distance
Call wassubjectedto sexualselectionto produce
the dimorphicmale call under the additional
controlof the higher vocalcenters.Subsequent
selectionmay have been responsiblefor the
modification of the Distance Call during de-
velopmentwith referenceto an externalmodel
providedby the father'sDistanceCall.At a later
stageall callsin the primitive songphrasemay
havebeenbroughtunder the neuralcontrolof
the higher centersso that they too could be
copiedas a set from the father'ssongor some
external
model. The Distance-call
Element
re-
mains the most evolutionarily labile of elementsin the phrase,possiblyhaving given rise
toat leasteightsecondary
elementsin wild birds.
Its open developmentalprogramhasproduced
versions of the call in domesticated
birds that
are almostunrecognizablebut for their greater
amplitudeand durationrelative to thoseof other elements (Simpsonand Vicario 1990, Zann
1985, 1990).
Songsof wild ZebraFinchesinvestigatedhere
are distinct from domesticated
stocks in dura-
ement matches that of the father's Distance Call
tion, tempo and frequency of some element
or his Distance-call Element (Zann 1990) and,
types,but not in number of elementsin the
714
RICHARD
ZANN
phrase. These findings contrastwith those of
Slater and Clayton (1991), who found no differencesbetweensongsof wild birds (n -- 18)
and domesticatedbirds from Bielefeld (n = 20);
birds from both locations differed from domes-
ticatedbirdsfrom St. Andrews(n = 20). Sample
sizes for the domesticated
stocks are similar to
[Auk,Vol. 110
(Taeniopygia
guttata):Selectivityin choiceof a tutor and accuracyof song copies.Anita. Behav.
31:231-237.
BOHNER,
J. 1990. Early acquisitionof song in the
Zebra Finch,Taeniopygia
guttata.Anita. Behav.39:
369-374.
CLAYTON,
N.S. 1990a. Assortativemating in Zebra
Finch subspecies,
Taeniopygia
guttataguttataand
T. g. castanotis.
Philos. Trans. R. Soc.Lond. Set. B
thosein the presentstudy.Consequently,
I conclude that: (a) samplesof this size are not adBiol. Sci. 330:351-370.
equate;and/or (b) important methodological CLAYTON,
N. S. 1990b. Subspecies
recognitionand
song learning in Zebra Finch subspecies.Anita.
differencesexistin the scoringof sonagrams.
Behav. 40:1009-1017.
Problemsmay arisein defining where one element starts and another stops; Cynx (1990)
foundthat about6%of elementsin his sample
had silent intervals of less than 10 ms, which
is equivalent to about 1 ram, the limit of effective measurement,on a standardKay sonagram
CLAYTON,
N. S., ANDE. PR6VE. 1989. Song discrim-
ination in femaleZebra Finchesand Bengalese
Finches.
Anita.
Behav.
38:352-362.
CLAYTON, N. S., D. HODSON, AND R. A. ZANN.
1991.
Geographicvariationin Zebra Finch subspecies.
Emu 91:2-11.
using the 300-Hz filter. Slater and Clayton's CYNX,J. 1990. Experimental determination of a unit
(1991)samplesof wild birdswere probablynot
of songproductionin the Zebra Finch (Taenioplarge enough(12/99 from colonyDa and 6/40
ygiaguttata).J. Comp. Psychol.104:3-10.
fromcolonyAs),sincetheyfoundnosignificant CYNX, J., H. WILLIAMS, AND F. NOTTEBOHM. 1990.
Timbre discriminationin Zebra Finch (Taeniopdifferencesbetweenthem, whereasmy larger
samplesdiffered in all but the number of ele-
ments(Table2). Clearly,largesamplesareneeded for comparisons
of songsbetweenpopulations, and methodological procedures for
measuringsonagramsneed to be standardized
acrossstudies.Differencesin songbetweenwild
and domesticatedbirds are not necessarilydue
to domesticationprocesses
per se but may be a
consequenceof the number of wild birds that
foundedthe colonyand its subsequent
isolation
from other populations.
ACKNOWLEDGMENTS
I thank B. and M. Danaherand the Alice Springs
ygiaguttata)songsyllables.J.Comp.Psychol.104:
303-308.
EALES,
L.A.
1985. Song learning in Zebra Finches:
Someeffectsof songmodelavailabilityon what
is learnt and when.
Anita.
Behav. 33:1293-1300.
IMMELMANN, K. 1962. Australian
finches in bush and
aviary. Angus and Robertson,Sydney.
IMMELMANN,
K. 1969. Songdevelopmentin the Zebra Finch and other estrildid finches.Pages6474 in Bird vocalizations (R. A. Hinde, Ed.). Cam-
bridge Univ. Press,Cambridge.
LEHNER,
P.N. 1979. Handbook of ethologicalmethods. Garland STPM Press, London.
NOTTEBOHM,
F., A. ALVAREZ-BUYLLA,
J. CYNX, J. KIRN,
C.-Y. LING, M. NOTTEBOHM,R. SUTER,A. TOLLES,
ANDH. WILLIAMS.1990. Song learning in birds:
C.S.I.R.O.Division of Wildlife and Ecologyfor allowing me to record Zebra Finches on their properties.
Invaluable technicalassistance
was provided by E.
The relationbetweenperceptionand production.
Tanger; D. Runciman, E. B. Male and A.M. Dunn
assistedwith recordings.P. J. B. Slater and N. S. Clay-
NOWICKi,S.,ANDD. A. NELSON.1990. Defining nat-
ton kindly provided sonagramsof domesticatedZebra Finchesfrom Europe.La Trobe University and
the Australian ResearchCouncil funded the project.
N. S. Clayton, M.P. Schwarz,P. J. B. Slater, I. W. B.
Thornton, and three anonymousreviewerssuggested
of three methodsapplied to 'chick-a-dee'call
notes.Ethology86:89-101.
OKANOYA,
K., ANDR. J. DOOLIN½.1987. Hearing in
important improvements to the manuscript.
Psychol. 101:7-15.
PRICE,P. H. 1979. Developmental determinants of
LITERATURE CITED
BISCHOF,H.-J., J. BOHNER, AND R. SOSSINKA. 1981.
Influence of external stimuli on the quality of
song of the Zebra Finch (Taeniopygia
guttatacastanotisGould). Z. Tierpsychol.57:261-267.
BOHNER,
J. 1983. Song learning in the Zebra Finch
Philos.
Trans. R. Soc. Lond.
Set. B Biol. Sci. 329:
115-124.
ural categoriesin acoustic signals: Comparison
passerine and psittacine birds: A comparative
studyof absoluteand maskedthresholds.
J.Comp.
structurein Zebra Finch song.J. Comp. Physiol.
Psychol. 93:260-277.
SASINsTITUTE. 1990. SASstatistics, version 6 ed. SAS
Institute,Inc., Cary, North Carolina.
SAS INSTITUTE. 1991. JMP, version 2.04. ed. Software
for statistical visualization. SAS Institute, Inc.,
Cary, North Carolina.
October1993]
Song
Structure
of WildZebra
Finches
SIMPSON,
H. B., ANDD. S. VICARIO. 1990. Brain pathways for learned and unlearned vocalizationsdiffer in Zebra Finches. J. Neurosci. 10:1541-1556.
SLATER,P. J. B., L. A. EALES,AND N. S. CLAYTON. 1988.
Songlearning in Zebra Finches(Taeniopygia
guttata):Progressand prospects.Adv. Study Behav.
18:1-34.
WILLIAMS,H., ANDK. STAPLES.1992. Syllable chunking in Zebra Finch (Taeniopygia
guttara)song.J.
Comp. Psychol.106:278-286.
ZANN,R. 1975. Inter- and intraspecificvariation in
the calls of three speciesof grassfinches
of the
subgenusPoephila
(Gould) (Estrildidae).Z. Tierpsychol. 39:85-125.
SLATER,P. J. B., AND N. S. CLAYTON. 1991. Domes-
ticationand songlearning in Zebra FinchesTaeniopygiaguttata.Emu 91:126-128.
SOSSINK^,
R., AND J. BtSHNER.1980. Song types in
the Zebra Finch (Poephilaguttata castanotis).
Z.
Tierpsychol. 53:123-132.
WILLIAMS,
H. 1990. Models for songlearning in the
Zebra Finch: Fathers or others. Anim.
715
Behav. 39:
747-757.
WILLIAMS, H., J. CYNX, AND F. NOTTEBOHM. 1989.
Timbre control in Zebra Finch (Taeniopygia
guttata) song syllables.J. Comp. Psychol. 103:366380.
WILLIAMS,H., AND J. R. McKIBBEN.1992. Changes
in stereotypedcentralmotor patternscontrolling
vocalizationare inducedby peripheral nerve injury. Behav. Neural Biol. 57:67-78.
ZANN,R. 1976. Variation in the songsof three speciesof estrildine grassfinches.Emu 76:97-108.
ZANN, R. 1984. Structural variation in the Zebra Finch
distancecall. Z. Tierpsychol.66:328-345.
ZANN, R. 1985. Ontogeny of the Zebra Finch distancecall:I. Effectsof crossfostering
to Bengalese
finches.Z. Tierpsychol.68:1-23.
ZANN,R. 1990. Songand call learningin wild Zebra
Finches.
Anim.
Behav.
40:811-828.
ZANN,R. 1993. Variation in songstructurewithin
and among populations of Australian Zebra
Finches.
Auk
110:716-726.
ZANN,g., ANDD. RUNCIMAN.In press. Survorship,
dispersaland sex ratios of Zebra Finches Taeniopygiaguttata in southeastAustralia. Ibis 136.