The Wilson Journal of Ornithology 120(3):519–524, 2008
MALE SONG VARIATION OF GREEN VIOLETEAR
(COLIBRI THALASSINUS) IN THE TALAMANCA MOUNTAIN
RANGE, COSTA RICA
GILBERT BARRANTES,1,4 CÉSAR SÁNCHEZ,1 BRANKO HILJE,2 AND
RODOLFO JAFFÉ3
ABSTRACT.—We studied variation in acoustic and temporal characteristics of the static male song of the
Green Violetear (Colibri thalassinus) in a single population in Costa Rica. The static song of 19 males was
extremely variable. The song has two elements: the first was delivered exclusively at the beginning of each song
while the second was present once, twice, or three times in the song of different males. Low frequency (LF),
song duration (T), and high frequency (HF) varied significantly among most individuals. The male population
of Green Violetear has four song types that differ in acoustic and temporal characteristics. The great inter-male
song variation suggests this type of vocalization may be under sexual selection. Received 2 February 2007.
Accepted 17 January 2008.
Song variation among individual birds is aggressive physical interactions with other
well known for oscine and psittacid species males during the breeding season (e.g., darting
(Farabaugh and Dooling 1996, Kroodsma chases) and visual displays are apparently ab-
1996). Learning in these birds has an impor- sent. Territorial males sing nearly continuous-
tant role in syntaxes and structure of the song. ly during courtship from before dawn until
This variation may have evolved to facilitate sunset (Slud 1964, Feinsinger 1977). Males
social interactions and/or by intra- or inter- begin to sing in September and some continue
sexual selection (Kroodsma 2004). Male re- until the end of March, investing up to 84%
productive success in birds is often associated of daily time to this activity (Skutch 1967,
with striking displays, such as complex songs, Wolf 1976). The objective of our study was
resulting from sexual selection (Catchpole to describe the variation in male song features
1982, Searcy and Yasukawa 1983, Johnsgard within a population of Green Violetear.
1994, Kroodsma 2004).
Individual song variation is relatively un- METHODS
explored in hummingbirds, another song-
We conducted fieldwork at the Estación
learning bird group (Baptista and Schuchmann
Biológica Cuericı́, Talamanca Mountains,
1990, Gaunt et al. 1994, Jarvis et al. 2000).
Costa Rica (09 33 N, 83 40 W; elevation
Gaunt et al. (1994) showed that male Green
2,600 m) during the dry season in January
Violetear (Colibri thalassinus) in neighboring
2004. The area is dominated by oak (Quercus
populations share song types and similarity
spp.) forest intermixed with several succes-
decreases with geographical distance. The ex-
sional growth stages with abundant flowering
tent of intra-population variation in male song
plants: Fuchsia paniculata, Bomarea costari-
has not previously been reported for this hum-
censis, Lamourouxia lanceolata, and Centro-
mingbird.
pogon spp. We recorded the static songs (song
Male Green Violetear show little (if any)
uttered by perched birds) from 19 males sing-
ing from exposed perches, ranging from 8 to
1 Escuela de Biologı́a, Universidad de Costa Rica, 25 m in height, on a 1.5-km transect along the
Ciudad Universitaria, Costa Rica. primary road (4 m wide); 12 males were re-
2 Instituto Tecnológico de Costa Rica, P. O. Box corded one morning and seven the next morn-
159-7050, Cartago, Costa Rica.
3 ing. Singing males were separated by 20–100Apartado 68941, Altamira, Caracas 1062, Vene-
zuela. m and perched at most 10 m into the forest
4 Corresponding author; (n  17), facing the main road, or in a forest
e-mail: gilbert.barrantes@gmail.com gap (n  2). Each bird was recorded once for
519
520 THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 120, No. 3, September 2008
at least 2 min or until the bird became silent. were conducted to examine which particular
We recorded the hummingbirds sequentially variables differ among individuals. We clas-
as we walked along the road to avoid record- sified the song of the hummingbirds into four
ing the same individual more than once. Be- types based on sonogram images, and used
cause of their high density, we could, in most stepwise Discriminant Function Analysis
cases, listen to the hummingbird previously (DFA) to test whether hummingbirds could be
recorded when we began to record the next separated into four groups by their song fea-
focal individual. We considered singing males tures. We also recorded the following infor-
as territorial individuals following Skutch mation during 6 days: non-systematic obser-
(1967); however, identifying territory limits vations of aggressive interactions and pres-
was difficult, due to few aggressive interac- ence of flower patches near perches of singing
tions by males of this species. We define a males. Means  SD are presented.
male territory as a circular area of 20 m in
diameter around the perch the singing male RESULTS
was most frequently using. The 19 male Green Violetears apparently
We recorded songs using a Telinga Pro II held the same territories for at least 6 days as
Parabola and a Marantz PMD-222 tape re- the same perches were occupied by singing
corder, and analyzed the songs using the pro- males day after day and we assumed the same
gram CANARY Version 1.2.1 (Charif et al. males were present. Territories included 2–3
1993). Tapes were deposited at the Laborato- perches from which males sang; they spent
rio de Bioacústica, Escuela de Biologı́a, Univ- most of their time on one of them. The num-
ersidad de Costa Rica. Spectrograms were ob- ber of males singing appeared to decrease
tained with a filter bandwidth of 349.70 kHz, from the forest border to the interior, except
frame length of 256 points, and a sampling when a gap was present, in which case the
rate of 44,100 Hz. We identified the elements number of males singing increased near the
of the song, defining a song as a string of gap edge.
notes or elements, temporally separated from Territories of the 19 recorded males varied
other such strings; an element was the shortest by presence of flower patches used as nectar
consistently recognizable trace on a sound sources for hummingbirds. Twelve territories
spectrogram (Marler and Peters 1982). We (main song perches) were within 10 m of
randomly selected five songs separated by at large patches of Fuchsia paniculata, which
least 10 sec to analyze the among-individual were frequently visited by Green Violetears
song variation from the total record of each and Volcano Hummingbirds (Selasphorus
hummingbird. We measured eight variables flammula). Attacks between individual Green
for each complete song: high frequency (HF), Violetears were rarely seen although the pres-
low frequency (LF), frequency range or band- ence of an individual in a flower patch occa-
width (F) (HF-LF), peak frequency (PF, fre- sionally triggered an attack by a singing male.
quency with highest amplitude), peak time The absence of obvious sexual dimorphism
(PT, time with maximum amplitude), duration prevented us from knowing whether these few
in seconds (T), time elapsed between two attacks were directed preferentially at males.
consecutive songs (T1), and time between the Males sang from before dawn until dusk,
first and the second element (T2). We also only abandoning their perches to briefly visit
quantified the number of songs and elements nearby flower patches for feeding or to cap-
delivered in a random selected minute for ture insects on the wing with short sallies. The
each individual. song of all males included two different ele-
We used a multiple analysis of variance ments (Fig. 1). The first element (a) was pro-
(MANOVA) on six song variables (T, LF, duced only once in the song of all individuals.
HF, PF, T1, and T2) to examine the difference Element (b) occurred once in the song of nine
of these variables among individuals (based males (song type 1), twice in the song of sev-
on a Discriminant Function Analysis (DFA), en individuals (type 2), three times in two oth-
F and PT were excluded because they did er males (type 3), and one or two times in
not differ among individuals). One-way anal- songs produced by another male (type 4). On
yses of variance and posteriori tests (Duncan) average, males produced 64.2 ( 8.8, range:
Barrantes et al. • GREEN VIOLETEAR MALE SONG VARIATION 521
FIG. 1. Spectrograms of the variation in number of elements, and structural characteristics of elements and
songs of three male C. thalassinus. Element (a) was present only once, at the beginning of each male’s song,
while presence of element (b) varied from one to three in songs of different males.
42–76.5) songs and 163.6 ( 33.7, range: among individuals using a MANOVA. The
126–231) elements per minute. comparison of all song features showed sig-
Structurally, the frequency range and high- nificant differences among individuals
est frequency were features of the static song (F108,414  21.66, P  0.001). Significant dif-
that had the largest range (max-min values) ferences were also found among individuals
for the 19 males: 11.06 and 9.08 kHz respec- when variables were analyzed separately (one-
tively (Table 1). The time between the first way ANOVA and Duncan tests). Variables
and the second element (T2), followed by fre- that differed among most individuals were LF
quency range (F), had the largest variation (16 males: F18,76  57.64, P  0.001), T (14
(CV) among individuals. Number of songs de- males: F18,76  47.27, P  0.001), and HF (13
livered by individual per minute decreased males: F18,76  77.56, P  0.001). Variation
with song duration (r  0.41, n  19, P  among individuals was lower for PF (8 males:
0.08). However, the number of elements (r  F18,76  6.32, P  0.001), T1 (4 males: F18,76
0.17, n  19, P  0.49) produced per minute  2.59, P  0.002), and T2 (4 males: F18,76 
was not related to T. The large variation in 26.62, P  0.001).
F was primarily caused by individual varia- Male Green Violetears were divided in sub-
tion in both HF and LF. groups based on structural and temporal char-
We compared six characteristics of the acteristics of their song. DFA separated the
complete song (T, LF, HF, PF, T1, and T2) males of this population into four groups
522 THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 120, No. 3, September 2008
TABLE 1. Characteristics of high frequency (HF), low frequency (LF), bandwidth (F), peak frequency
(PF), time duration (T), time between two consecutive songs (T1), and time between the first and the second
element (T2) for the complete song of 19 male Green Violetears in Costa Rica.
HF (kHz) LF (kHz) F (kHz) PF (kHz) T (sec) T1 (sec) T2 (sec)
Mean 10.98 3.16 7.82 6.11 0.58 0.37 0.33
SD 2.19 0.57 2.36 0.81 0.10 0.03 0.15
CV 19.90 17.98 30.17 13.31 17.73 8.86 44.41
Range 9.04–18.12 2.08–4.82 4.28–15.34 4.05–8.53 0.42–0.95 0.30–0.51 0.069–0.84
(F15,240  11.98, P  0.0001) based on six pothesis’’ proposed by Kroodsma (2004) to
variables (T, LF, HF, PF, T1, and T2), and explain temporal song variation in dialects of
classified 82% of the songs correctly. the Three-wattled Bellbird (Procnias tricarun-
culatus). This hypothesis proposes that adult
DISCUSSION birds ‘‘relearn’’ their songs throughout their
The difference in characteristics of the stat- life, and modify their songs through compe-
ic song among male Green Violetears was tition (by imitation) with familiar rivals. For
high. This difference individualizes most sing- this hypothesis to be possible, groups of males
ing males in the population, suggesting a pos- have to hear and recognize neighboring males
sible role of inter- and intra-sexual selection (Aoki 1989); this is the case with Green Vi-
(Morton 1986, Searcy and Andersson 1986); oletear. Consequently, at a particular time lag,
however, other factors such as age may also different songs could be present in a single
affect song characteristics. Frequency range in population. However, spatial distribution of
the song of C. thalassinus is possibly the fea- song types is needed to begin testing this hy-
ture from which the receiver (e.g., a female) pothesis.
obtains greater information, particularly when The information available allows us to
the song is composed by short, broad band- compare variation of song traits of C. thal-
width elements or notes (Fig. 1, Table 1). This assinus with other Trochilinae hummingbirds.
characteristic permits more precise location of The number of elements in the song of C.
singing males (Smith et al. 1978, Richards and thalassinus is small compared to songs of
Wiley 1980). Additionally, males producing Anna’s Hummingbird (Calypte anna) (Stiles
songs with a wide frequency range may con- 1982), Blue-throated Hummingbird (Lampor-
ceal the songs of other males, if these males’ nis clemenciae) (Ficken et al. 2000), Ame-
songs have a narrower bandwidth that fit in thyst-throated Hummingbird (L. amethystin-
part of the range of frequencies of other us), and Green-throated Mountaingem (L. vir-
males. idipallens). The number of elements is similar
Despite the high variation in this population to Magnificent Hummingbird (Eugenes ful-
of male Green Violetears, groups of males gens) (Ornelas et al. 2002) and more complex
produced similar song types. This suggests the (e.g., more elements and frequency modula-
presence of ‘‘different dialects’’ in a single tion) than the static song of its congener Spar-
population. However, causes of inter-group kling Violetear (Colibri coruscans) (Gaunt et
song differentiation are difficult to explain. al. 1994). A characteristic present in all songs
Dialects usually evolve through geographic or of this small sample of hummingbirds is the
microgeographic isolation (Kroodsma 1996) wide frequency range (bandwidth), although
but, in this case, geographic isolation seems both species of the genus Colibri present the
not to be the cause, since this species breeds narrowest bandwidth. Song complexity is not
in the area and then migrates to lower eleva- higher in those species lacking visual displays,
tion during the non-breeding season (Skutch such as C. thalassinus, when compared to spe-
1967). In addition, young hear songs of adult cies with complex visual displays, such as C.
males as males sing well beyond when fledg- anna and L. amethystinus. These results
lings abandon their nests (GB, pers. obs.). A should be viewed with caution because num-
possible explanation is the ‘‘song relearn hy- ber of elements and acoustic structure of hum-
Barrantes et al. • GREEN VIOLETEAR MALE SONG VARIATION 523
mingbird songs may be strongly influenced by FEINSINGER, P. 1977. Notes on the hummingbirds of
environmental and phylogenetic features (Ir- Monterverde, Cordillera de Tilarán, Costa Rica.
win 1988, McCracken and Sheldon 1997). Wilson Bulletin 89:159–164.
FICKEN, M. S., K. M. RUSH, S. J. TAYLOR, AND D. R.
These aspects can be more closely analyzed POWERS. 2000. Blue-throated Hummingbird song:
within the monophyletic genus Colibri. When pinnacle on nonoscine vocalizations. Auk 117:
features of the song are compared between C. 120–128.
thalassinus and C. coruscans, the two high- GAUNT, S. L. L., L. F. BAPTISTA, J. E. SÁNCHEZ, AND
land species of the genus (Gaunt et al. 1994), D. HERNÁNDEZ. 1994. Song learning as evidenced
the static song of C. thalassinus has a higher from song sharing in two hummingbird species
(Colibri coruscans and C. thalassinus). Auk 111:
number of elements. However, C. coruscans 87–103.
produces a ‘‘dynamic song’’ that males sing IRWIN, R. E. 1988. The evolutionary importance of be-
during a diving flight as part of the courtship havioral development: the ontogeny and phylog-
display, which is absent in C. thalassinus. eny of bird song. Animal Behaviour 36:814–824.
This supports Wagner’s (1954) suggestion that JARVIS, E. D., S. RIBEIRO, M. L. DA SILVA, D. VENTURA,
species lacking elaborate dynamic songs have J. VIELLIARD, AND C. V. MELLO. 2000. Behaviour-
complex static songs, and suggests the two ally driven gene expression reveals song nuclei in
hummingbird brain. Nature 406:628–632.
highland species of the genus have evolved JOHNSGARD, P. A. 1994. Arena birds: sexual selection
different courtship strategies. The other two and behavior. Smithsonian Institution Press,
species in the genus, C. delphinae and C. ser- Washington, D.C., USA.
rirostris are mid-elevation and apparently ter- KROODSMA, D. 1996. Ecology of passerine song de-
ritorial (Stiles and Skutch 1989, Schuchmann velopment. Pages 3–19 in Ecology and evolution
1999). Little is known about the vocalizations of acoustic communication in birds (D. E. Kroods-
ma and E. H. Miller, Editors). Cornell University
of these two species. Press, Ithaca, New York, USA.
KROODSMA, D. 2004. The diversity and plasticity of
ACKNOWLEDGMENTS birdsong. Pages 108–131 in Nature’s music. The
We thank J. R. Eberhard, D. E. Kroodsma, C. E. science of birdsong (P. Marler and H. Slabbe-
Braun, and two anonymous reviewers for helpful and koorn, Editors). Academic Press, New York, USA.
critical comments on the manuscript, and Luis San- MARLER, P. AND S. PETERS. 1982. Subsong and plastic
doval for helping with the statistical analysis. We also song: their role in the vocal learning process. Pag-
thank the Universidad de Costa Rica for financial sup- es 25–50 in Acoustic communication in birds.
port and The Organization for Tropical Studies and Production, perception, and design features of
Estación Biológica Cuericı́ for logistical support. sounds (D. E. Kroodsma and E. H. Miller, Edi-
tors). Academic Press, New York, USA.
MCCRACKEN, K. G. AND F. H. SHELDON. 1997. Avian
LITERATURE CITED vocalizations and phylogenetic signal. Proceed-
AOKI, K. 1989. A sexual-selection model for the evo- ings of the National Academy of Science USA 94:
lution of imitative learning of song in polygynous 3833–3836.
birds. American Naturalist 134:599–612. MORTON, E. S. 1986. Prediction from the ranging hy-
BAPTISTA, L. F. AND K. L. SCHUCHMANN. 1990. Song pothesis for the evolution of long distance signals
learning in the Anna Hummingbird (Calypte in birds. Behaviour 99:65–86.
anna). Ethology 84:15–26. ORNELAS, J. F., C. GONZÁLES, AND J. URIBE. 2002.
CATCHPOLE, C. K. 1982. The evolution of bird sounds Complex vocalizations and aerial displays of the
in relation to mating and spacing behaviour. Pages Amethyst-throated Hummingbird (Lampornis
297–319 in Acoustic communication in birds. Pro- amethystinus). Auk 119:1141–1149.
duction, perception, and design features of sounds RICHARDS, D. G. AND R. H. WILEY. 1980. Reverbera-
(D. E. Kroodsma and E. H. Miller, Editors). Ac- tion and amplitude fluctuations in the propagation
ademic Press, New York, USA. of sound in a forest: implications for animal com-
CHARIF, R. A., S. MITCHELL, AND C. W. CLARK. 1993. munication. American Naturalist 115:381–399.
Canary 1.1 user’s manual. Cornell Laboratory of SCHUCHMANN, K. L. 1999. Family Trochilidae (hum-
Ornithology, Ithaca, New York, USA. mingbirds). Pages 468–680 in Handbook of birds
FARABAUGH, S. M. AND R. J. DOOLING. 1996. Acoustic of the world. Volume 5. Barn Owls to humming-
communication in parrots: laboratory and field ex- birds (J. Del Hoyo, A. Elliott, and J. Sargatal, Ed-
periments of Budgerigars, Melopsittacus unulatus. itors). Lynx Edicions, Barcelona, Spain.
Pages 97–117 in Ecology and evolution of acous- SEARCY, W. A. AND M. ANDERSSON. 1986. Sexual se-
tic communication in birds (D. E. Kroodsma and lection and the evolution of song. Annual Review
E. H. Miller, Editors). Cornell University Press, of Ecology and Systematics 17:507–533.
New York, USA. SEARCY, W. A. AND K. YASUKAWA. 1983. Sexual se-
524 THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 120, No. 3, September 2008
lection and Red-winged Blackbirds. American STILES, F. G. 1982. Aggressive and courtship displays
Scientist 71:166–174. of the male Anna’s Hummingbird. Condor 84:
SKUTCH, A. F. 1967. Life histories of Central American 208–225.
highland birds. Nuttall Ornithological Club Num- STILES, F. G. AND A. F. SKUTCH. 1989. A guide to the
ber 7. Cambridge, Massachusetts, USA. birds of Costa Rica. Cornell University Press, Ith-
SLUD, P. 1964. The birds of Costa Rica. Distribution aca, New York, USA.
and ecology. Bulletin of the American Museum of WAGNER, H. O. 1954. Versuche einer analyse der ko-
Natural History 128:1–430. libribalz. Zeitschrift für Tierpsychologie 11:182–
SMITH, W. J., J. PAWLUKIEWICZ, AND S. T. SMITH. 1978. 212.
Kinds of activities correlated with singing patterns WOLF, L. L. 1976. Avifauna of the Cerro de la Muerte
of the Yellow-throated Vireo. Animal Behaviour region Costa Rica. American Museum Novitates
26:862–884. 2606:1–37.