-
Behav Ecol Sociobiol (1995) 37: 249-254
-
250
manner. Therefore, the remaining two measures ofperformance
included the variability in the acousticstructure of the alpha
male's song contribution, andthe variability in the beta male's
contribution. We testedthe relationship between these four
properties of songperformance and female visitation rate using
correla-tion and step-wise discriminant function analysis.
Methods
Instead, they attempt to show a role of sexual selec-tion in
generating interpopulation and interspeciesdifferences in song
(Payne 1983; Eastzer et al. 1985;Baker et al. 1987).
In our study of long-tailed manakins (Chiroxiphialinearis) we
examined the relationship between songperformance and courtship
success. We avoided theabove limitations by measuring four acoustic
proper-ties of manakin songs independently of female
songpreference. Furthermore, we tested whether courtshipsuccess of
males within a population varied with thesefour measures of song
performance.
The lek mating system of the long-tailed manakinsis unusual in
that males display cooperatively to attractfemales to display
perch-zones (lek arenas) solely forthe purpose of mating.
Perch-zones are separated by75-300 m, and each is occupied by a
different alphamale who is dominant to other males in the lek.
Thealpha male displays most often with the beta male.Both of these
top-ranking males have definitiveplumage and are usually 8 years
old or older. As manyas 12 other males, including predefinitive
males, maybe affiliated with a given lek. The partnerships
betweenalpha and beta males develop and become stable overa period
of several years (McDonald 1989a). The part-ners perform a unison
song that attracts females anda joint dance that precedes
copulation (Trainer andMcDonald 1993). With very rare exceptions
all copu-lations are by alpha males. Alpha males from
differentperch-zones vary greatly in their success at
attractingfemales and at obtaining copulations (McDonald1989a,
1993b). We used the rate at which femalesvisited perch-zones as a
measure of courtship success.The relationships among visitation
rate, copulation rateand characteristics of the dance display were
examinedelsewhere (McDonald 1989b).
The female attraction song is given by two maleswho sing nearly
identical components almost in uni-son while perched 10-15 cm
apart. Thejoint song ono-matopoetically resembles the word
"toledo", with thefirst and last syllable sung at a flat F and the
middlesyllable rising to a flat A on the diatonic musical scale.The
songs of some teams sound to the human ear morecoordinated and
harmonious than those of otherteams. Harmonious songs sound like
one bird singing,but have a noticeably full sound, easily
distinguishedfrom the rare solo songs. Non-harmonious songs
sounddissonant, or out of tune, and sometimes sound
poorlysynchronized. Examination of sonagrams shows thatin songs
that sound harmonious to human ears, thefrequencies of the two
male's components are wellmatched. Two measures of song
performance, thedegree of frequency matching and the degree to
whichthe two males' song contributions were synchronized,were
designed to quantify the audible differences insong quality.
Quality song performance probablydepends not only on the ability to
sing in acoordi-nated fashion, but on the ability to sing in a
consistent
The study area, in Monteverde, Costa Rica (100 18'N, 840
48'W),is 80 ha of premontane tropical moist forest (Holdridge 1966)
atan elevation of 1300 m. Female visitation at seven of the most
activeperch-zones was observed over 2 years. Recordings of the
jointtoledo songs were made at five of the perch-zones in the 1st
year,and at two additional zones the 2nd year. At each perch-zone,
boththe alpha and the beta males were color banded, as well as
manyof the affiliate males. The sex and age of manakins were
determinedas described in McDonald (1989b, 1 993b).
Perch-zones were compared on the basis of visitation rates
perunit time because sample sizes varied. Observers conducted
sched-uled 2-h observations between 0600 and 1500 hours while
sitting inblinds made from black plastic placed 8-12 m from display
perches.From 9 to 19 scheduled observation periods were conducted
at eachof the seven perch-zones. During the scheduled observations,
anobserver recorded the number of females present on the perch
dur-ing each 5-min block. These numbers were added over all
obser-vation periods to determine the total number of female visits
to theperch-zone. The relative female visitation rate at a given
perch-zonein a given year was calculated as follows: number of
visits per obser-vation period at that zone/total number of visits
per observationperiod for all focal perch-zones during that year.
In addition, toshow how visitation changed over time, we calculated
the numberof visits per observation period at perch-zone Z over a
period of 5years.
Toledo songs of seven alpha-beta partnerships were recordedwith
a Sony WM D6C cassette recorder and a Sennheiser ME 80directional
microphone. Ten songs from each team were analyzedusing a Kay
Elemetrics Model 5500 Sonagraph with a grey scaleprinter. Singers
were identified on the basis of their color bands.Occasionally
during a bout of singing, one of the two males wouldgive a small
number of solitary toledos. By analyzing the
frequencycharacteristics of these solitary toledos, we were able to
identify theindividual singer of each contribution to the toledo
songs.
The two males of a team sing almost identical song
contribu-tions, each approximately 0.6 s in duration. The song
contributionsare sung nearly in unison, with one male commencing
approxi-mately 0.1 s after the first male begins. In sonagrams,
each male'scontribution begins at about 1700 Hz and sweeps down to
form atrough, then sweeps rapidly up to form an arch, and finally
endswith an element at about 1450 Hz. The trough in a sonagram
hastwo nodes of highest amplitude corresponding to two dips in
fre-quency (Fig. la).
Our analysis of song structure was based on samples of ten
uni-son songs from each team of males. For each male, we
character-ized his contribution to a toledo song by measuring four
parameters:the minimum frequency reached in the first dip of the
trough, theminimum frequency reached in the second dip of the
trough, themaximum frequency reached in the arch, and the time
intervalbetween the onset of the song and the rapid frequency
ascent(Fig. Ib). To determine the minimum or maximum sound
fre-quency, the Sonagraph was set to display the narrowband (15
Hz)audiospectrogram and the power spectrum of a selected
O.IO-ssample. The power spectrum where the minimum or
maximumfrequency occurred was displayed, and the frequency with the
high-est amplitude was noted. Frequency was measured with a
minimum
-
251
Table Variables of song performance in long-tailed manakins
Variable DescriptionN:I:~>(.)CQ);J0-Q)It
Index of frequency matching
Onset interval(singing synchrony)
0.5 1.0
Time (seconds) Song variance of alpha male(inversely related to
singingconsistency)Song variance of beta male
Sum of the disparities betweenthe alpha and beta males in
3frequency parameters subtractedfrom 200
Time interval between theonsets of alpha and beta males'song
contributions
Generalized variance in 4parameters of alpha male's
songcontribution
Generalized variance in 4parameters of beta male's
songcontribution
b2.0, C GDN
:t:~>t)CGO='c-O)
It
tested using Pearson's coefficient, r. To determine which of
theabove variables best explained the variance among teams in
thefemale visitation rate, we used a step-wise discriminant
function.We divided the seven teams into fwO groups, one containing
theteams with visitation rates above the mean (x = 0.82), and one
con-taining the teams with visitation below the mean. Because the
songvariables were not normally distributed, we used the rank
values inthe discriminant function analysis. This analysis revealed
whichvariables or combinations of variables helped discriminate
betweenteams with high or low visitation rates.
A E J ~1.5
1.00.5 1.0 1.5
Time (seconds)
Fig. 1 a Audiospectrogram of a long-tailed manakin dual
"toledo"song. b Drawing of the above spectrogram with the
contributionof one male shown solid, and the other male shown
outlined.Parameters A, B, C, and D were measured for each manakin
male.D indicates the time interval between the onset and the
ascendingfrequency sweep of a male's song contribution. A, B, and C
indi-cate time points at which the sound frequencies were
measured.These frequencies were compared to similar measurements at
E, Fand G of the partner's contribution to generate the index of
fre-quency matching. The index of frequency matching in Hz = 200
-(iA -EI + IB -FI + IC -GD
Results
Either male in a team could commence the duet song.In most of
the teams, the alpha was the male that ini-tiated nearly all of the
duet songs. However, in twoteams (A and Y), the alpha and beta
males initiatedsongs about equally often. No significant
differenceexisted between the song variance of alpha males(x =
19.39; SD = 2.59) and that of the beta males(x = 19.17; SD =
2.51).
There were no significant correlations among thefour variables
of singing performance (Table 2). Whenall four variables were
included in a discriminant func-tion analysis, average frequency
matching alone wasuseful in discriminating between teams with high
visi-tation rates and those with low visitation rates (step-wise
discriminant function analysis; F(I.5) = 15.0;P = 0.012) (Table 3).
In addition, only average fre-quency matching was significantly
correlated withvisitation rate (Pearsons r = 0.92; P = 0.003)
(Table 2).
After the 1st year of the study, the beta male atperch-zone Z
disappeared and was replaced by thegamma male in subsequent years.
This gave us theopportunity to observe the impact of a partner
changeon frequency matching and courtship success. Prior tothe
disappearance of the beta, team Z had the songswith the highest
degree of frequency matching, high-est visitation rate (Table 2),
highest song rate, bestdance, and highest copulation rate (McDonald
1989b).The replacement beta had an abnormal, incomplete
resolution of 10 Hz and time was measured with a minimum
res-olution of 6.25 m. These variables could be repeated reliably
formany teams of males.
To quantify the audible differences in song quality, we
devisedan index of frequency matching. The index was calculated by
tak-ing the sum of the disparities between the two males' song
contri-bution in the three minimum and maximum frequency
variables.This sum was then subtracted from 200 so that the
magnitude ofthe frequency matching index would increase in
magnitude as thedisparities in frequencies decreased. In addition,
we measured threeother variables of singing performance (Table 1).
One of these, thetime interval between the onsets of each
contribution to the song,indicated the extent to which the
partners' songs were synchronized.The other two variables measured
the amount of variability in thesinging of each partner, and
indicated how consistently each malesang his contribution to the
duet song. To estimate the amount ofvariability in the singing of
each male, a multivariate measure ofvariance in the four song
parameters was calculated for each male'ssample of songs. The
measure was calculated as the log of the deter-minant of the
variance/covariance matrix of the four parameters(Sneath and Sokal
1973). To ensure that each parameter wasweighted equally, the time
interval measurements (ms) were multi-plied by 10 so that they
would have a magnitude similar to the fre-quency measurements
(Hz).
The correlations among female visitation rate, average
frequencymatching index, average onset interval between song
contributions(synchrony), song variance of alpha, and song variance
of beta were
-
252
Table 2 Correlation (Pearsonscoefficient, r) matrix amongfemale
visitation rate and fourvariables of song performancein long-tailed
manakins
Frequency matchingOnset intervalSong variance IXSong variance
p
0.92-().45-0.59-0.03
0.0030.310.170.95
-0.53
-0.53
-0.30
0.220.220.95
0.550.40
0.280.38 0.57-0.26
Table 3 Female visitation and song performance variables of
long-tailed manakins. (+) designates visitation rates above the
mean
(x = 0.82); ( -) designates visitation rates below the mean.The
num-bers in parentheses are ranks
Songvariance ;x
Team Visitationrate
Average frequencymatching
Average onsetinterval
Songvariance P
16.8 (6)21.1 (3)22.0 (2)18.1 (4)16.4 (7)22.2 (1)17.6 (5)
0.10.10.10.]0.]0.]0.]
18.3 (5)19.3 (4)15.0 (7)18.0 (6)20.6 (3)22.1 (2)22.4 (I)
17411710799829146
abnormal song contribution of the new male did notcontain the
analogous elements necessary for compar-ison with his partner's
contribution, an index of fre-quency matching could not be
computed. Figure 2shows the complete lack of frequency matching in
thedual song. The number of visits per observation perioddeclined
dramatically during the new male's first sea-son as beta, and
continued to decline over the next fewyears (Fig. 3). Copulations
per observation period fellfrom 0.41 in year 1 to 0 in succeeding
years, after thenew beta male took over.
2.0
J::>u 1.5c:~:JC"..It
0.5 1.0 1.5
Time (seconds)
Fig. 2 Audiospectrogram of the song of team Z in 1986. The
abnor-mal song contribution of the beta male occurs above
approximately1.75 kHz
Discussion
"0.g 3.00 I
0)Co 2.50
CII.00 2.00
'Q;Co 1.50CII...
:§ 1.00>0)
"iij 0.50E0)
LL. 0.00
YearFig. 3 Decline in female visitation at perch-zone Z after
the replace-
ment of the beta male in year 2
song, about 300 Hz higher in frequency than that ofany other
male sampled. Not only was the degree offrequency matching in the
songs of the new teamextremely low, songs were not as loud and were
deliv-ered from perches closer to the ground. Because the
The measure of song performance that most directlyaccounted for
the harmonious sound of well-coordi-nated manakin song \,!,as
frequency matching. Songswith extremely poorly matched frequency
not onlysound dissonant to human ears, but sound unsyn-chronized as
well. The apparent lack of synchrony wasnot due to unusually long
intervals between the onsetsof the two males' song contributions;
onset intervalswere similar in harmonious and unharmonious
songs.Rather, the frequency mismatch renders the temporaloffset
between the two males' contributions more per-ceptible, making the
song sound less synchronous.Likewise, the amount of variability in
the singing ofalpha and beta males was not related to the degree
offrequency matching. All males in this sample weredefinitive (at
least 4 years old) and sang quite con-sistently. Frequency matching
appears to be the bestmeasure of how harmonious and coordinated
manakinsongs sound to the human ear.
Teams of long-tailed manakins with songs that werewell matched
in terms of frequency attracted more
1011,21
119102127
134
(6)(5)(3)(4)(7)(2)(1)
(1)(2)(3)(4)(6)(5)(7)
-
253
propagate would be less than this for two reasons.
First,temporal overlap of matched frequencies is not perfect,so the
intensity of a dual-male song is probably notdoubled at the source.
Second, sound intensity atten-uates considerably in a dense forest
(Wiley andRichards 1978), and this excess attenuation is linearwith
distance. Thus, the increased distance of audibil-ity of a dual
song would be somewhat less than thetheoretical maximum of 1.41
times as far. Therefore, iffemale visitation preferences were based
solely on audi-bility, a male with a partner might potentially
attractmore females than a solitary male, but the benefit islikely
to be considerably less than a twofold increasein visitation.
Solitary songs are rare, however, and ourinterest is in whether
duet songs with close frequencymatching may propagate further than
poorly matchedsongs. It seems unlikely that a small difference in
fre-quency matching would increase the area of audibilityenough to
impact the rate of female visitation, espe-cially since temporal
overlap of the two males' songcontributions is imperfect in both
well and poorlymatched songs. It is more likely that active choice
byfemales explains the higher visitation at perch-zoneswhere well
matched songs are heard.
Frequency matching may provide females with anunambiguous ideal
against which dual-male perfor-mance can be judged. Like the
morphological charac-ters assessed under fluctuating asymmetry
arguments(Watson and Thornhill 1994), frequency matching hasa clear
maximum of perfect matching. Thus, dual-maleperformances provide
females with the opportunity forenhanced discrimination among
potential mates.Furthermore, such performances may provide
femaleswith a criterion useful in assessing the quality of
poten-tial mates.
The expression of frequency matching may be re-lated to the
cooperative abilities of the participatingmales. The partnership
between an alpha and beta maledevelops over several years. Young,
predefinitive malesinitially participate in several partnerships,
fluidly mov-ing among perch-zones (McDonald 1989a). In this
way,they become established in a dominance system inwhich status
increases with age. As males grow olderand their status
relationships become better defined,they spend more time displaying
at fewer perch-zones.The dominance hierarchy at a perch-zone
appears tofunction as a queue of variously aged males
awaitingopportunities to move into beta and alpha
positions(McDonald 1993a). By the age of about 8 years, malestend
to be established in stable, long-term partnershipsas beta males.
As long as these queues remain orderly,little overt aggression is
observed. The orderliness ofqueues may be reinforced by female
choice, sincefemales seldom remain in a perch-zone while
chasingamong males occurs. Opportunities for males to movebetween
queues, however, appears to be limited byaggression, particularly
among the younger males(McDonald 1993a). We have some evidence to
suggest
females. Neither the degree of synchrony nor the con-sistency
with which team members sang was related tothe ability to attract
females. Although a significantrelationship between visitation rate
and frequencymatching is suggestive, it is insufficient to allow us
toconclude that well matched singing per se causes highvisitation
by females. Females may have been respond-ing to other correlated
behavioral cues. For example,teams of males that sang more
persistently attractedmore females (McDonald 1989b), and this may
havebeen the primary basis for females' choice of perch-zone.
Characteristics of the duet song, includingamount of output, appear
to be important in deter-mining which perch-zones a female will
visit. Once shearrives in the display area, characteristics of the
dance,including a "butterfly" display, appear to determinewhether
or not she will choose to mate with the alphamale (McDonald 1989b).
Nevertheless, our results sug-gest that females may have a
preference for songs witha high degree of frequency matching.
Song quality in long-tailed manakins is a functionof the
behavior of two males. Mate choice based on adual-male phenotype is
an intriguing possibility, espe-cially since one of the males does
not mate. The suc-cess of the alpha male may depend on the ability
ofhis partner to sing in a coordinated fashion. The effectof the
non-mating partner on courtship success wasillustrated the year
that the beta partner disappearedfrom team Z, the most successful
partnership in thestudy area. Visitation and copulation rates
declinedwhen the alpha male established a partnership with amale
whose song was abnormal. The individual phe-notype of the formerly
extremely successful alpha maledid not change, yet females ceased
responding to hiscourtship. The change in the dual-male
phenotype,therefore, is most likely to explain this decline
incourtship success.
The above observations raise questions about theevolution of
mate choice based on a dual-male pheno-typic character. How does a
dual-male character suchas frequency matching help a female find or
select amate? We consider two possibilities, which are notmutually
exclusive: frequency matched songs are moreaudible than unmatched
songs, or frequency matchingprovides females with a criterion
useful in assessing thequality of potential mates.
By matching sound frequencies, two males may pro-duce a song
with up to 2 times the intensity of thatsung by a single male (a
fourfold increase in intensitytheoretically is possible if coherent
sound waves areproduced by two closely spaced sources; however,
thedistance between manakin singers, 10-15 cm, is toolarge for such
near-field interference to occur). A dualsong with twice the source
intensity would be audiblefrom a maximum of 1.41 times as far
(sound intensitydecreases in proportion with the square of the
dis-tance), and would be audible over twice the area. Inreality,
the additional distance a manakin song could
-
254
that frequency matching may be related to the dura-tion of a
partnership. First, frequency matching in-creases with age of the
younger partner. Second, songsof established teams of males have
better frequencymatching than songs resulting from randomly
com-bining the measurements of individual males' song por-tions to
form artificial teams (J.M. Trainer andD.B. McDonald, unpublished
work). Because partnersare not genetically related (McDonald and
Potts 1994),frequency matching is very unlikely to result
fromgenetic similarity. These observations suggest that jointsongs
of established teams become better matched asthe partnership
develops.
Thus, females may be able to acquire informationabout the
duration and stability of an alpha male's part-nership from the
degree of frequency matching in hisjoint songs. A male whose songs
have a higher degreeof frequency matching has demonstrated that he
hassurvived eight or more years, acquired sufficient statusto be a
member of an established team, has the abilityto perform highly
coordinated courtship display, andexhibits the tolerance of other
adult males necessaryto form a stable partnership. To the extent
that thesequalities are heritable, females may benefit from
choos-ing males with a high degree of frequency matching.Our
studies of the long-tailed manakin help us under-stand how mate
choice based on dual-male phenotypemight have evolved.
Acknowledgments This project was funded by NSF grant BNS-8814038
to J.M.T., a Jessie Smith Noyes Foundation grant fromthe
Organization for Tropical Studies, the Frank M. Chapman fundof the
American Museum of Natural History and the AmericanPhilosophical
Society. D.B.M. is grateful to the Harry FrankGuggenheim
Foundation, the Organization for Tropical Studiesand the National
Geographic Society for support. Joe and JeanStuckey graciously
permitted the use of their farm as a study site.Members of the
Monteverde community have helped in many ways.Personnel in the
offices of Vida Silvestre were remarkably courte-ous, prompt and
helpful in securing necessary permits. We thankPeter Hoekje for
discussions of acoustics.
Baker MC, Bjerke TK, Lampe H; Espmark Y (1987) Sexualresponse of
female yellowhammers to differences in regionalsong dialects and
repertoire sizes. Anim Behav 35:395--401
Catchpole CK (1986) Song repertoires and reproductive success
inthe great reed warbler Acrocephalus arundinaceus. Behav
EcolSociobiol19:439-445
Eastzer DH, King AP, West MJ (1985) Patterns of courtshipbetween
cowbird subspecies: Evidence for positive assortment.Anim Behav
33:30-39
Gibson RM, Bradbury JW (1985) Sexual selection in lekking
sagegrouse: phenotypic correlates of male mating success. BehavEcol
Sociobiol 18: 117-123
Holdridge L (1966) The life zone system. Adansonia 6:
199-203.King AP, West MJ (1983) Dissecting cowbird song potency:
assess-
ing a song's geographic identity and relative appeal.
ZTierpsycho163:37-50
McDonald DB (1989a) Cooperation under sexual selection:
Age-graded changes in a lekking bird. Am Nat 134:709-730
McDonald DB (1989b) Correlates of male mating success in
alekking bird with male-male cooperation. Anim
Behav37:1007-1022
McDonald DB (1993a) Delayed plumage maturation and orderlyqueues
for status: a manakin mannequin experiment. Ethology94:31-45
McDonald DB (1993b) Demographic consequences of sexual
selec-tion in the long-tailed manakin. Behav Eco14:297-309
McDonald DB, Potts WK (1994) Cooperative display and
related-ness among males in a lek-mating bird. Science 266:
1030-1032
Payne RB (1983) Bird songs, sexual selection, and female
matingstrategies. In: Wasser SK (ed) Social behavior of female
verte-brates. Academic Press, New York, pp 55-91
Payne RB, Payne K (1977) Social organization and mating
successin local song populations of village indigobirds, Vidua
chaly-beata. Z Tierpsycho145:113-173
Searcy WA (1992) Song repertoire and mate choice in birds.
AmZoo132:71-80
Searcy W A, Andersson M (1986) Sexuai selection and the
evolu-tion of song. Annu Rev Ecol Syst 17: 507-533
Sneath PHA, Sokal RR (1973) Numerical taxonomy. Freeman,
SanFrancisco
Trainer JM, McDonald DB (1993) Vocal repertoire of the
long-tailed manakin and its relation to male-male
cooperation.Condor 95:769-781
Watson PI, Thornhill R (1994) Fluctuating asymmetry and
sexualselection. Trends Ecol Evol 9:21-25
Wiley RH, Richards DG (1978) Physical constraints on
acousticcommunication in the atmosphere: implications for the
evolu-tion of animal vocalizations. Behav Ecol Sociobiol 3:
69-94
Communicated by R. GibsonReferences
Andersson S (1989) Sexual selection and cues for female choice
inleks of Jackson's widowbird Euplectes jacksoni. Behav
EcolSociobioI25:403-410