Ethology 108, 367—376 (2002)
2002 Blackwell Verlag, Berlin
ISSN 0179–1613
Dialect Discrimination by Male Orange-Tufted Sunbirds
(Nectarinia osea): Reactions to Own vs. Neighbor Dialects
Noam Leader*, Jonathan Wright & Yoram Yom-Tov*
*Zoology Department, Tel Aviv University, Israel; School of Biological Sciences,
University of Wales, Bangor, Wales
Abstract
Male orange-tufted sunbirds (Nectarinia osea) exhibit distinct song dialects
throughout Israel. Recently, two distinct local dialects with a sharp boundary
were discovered in a small (1.5 km2) urban neighborhood densely inhabited by 63
territorial sunbird pairs. We conducted playback experiments to determine song
dialect discrimination capability by sunbird males in this neighborhood. Males of
both dialects responded significantly more strongly to playback of their own
dialect than to that of the adjacent dialect. In spite of the extreme proximity
between the two dialect areas, we found no effect of distance to the neighboring
dialect on the intensity of any the behavioral responses. We suggest that due to
the complex acoustic properties of this urban neighborhood, sunbirds are
extremely limited in the number of neighboring males they can assess to establish
what the local song is. A stronger response to one’s own dialect is therefore
expected, and we discuss how local dialects could be maintained via this
mechanism regardless of the very small distances between territories and dialect
populations.
Corresponding author: N. Leader, Department of Zoology, Faculty of Life
Sciences, Tel Aviv University, Tel Aviv 69978, Israel. E-mail: [email protected]
Introduction
Many oscine species exhibit geographic variation in song. These variations in
bird song may also occur on a microgeographic scale, between neighboring
groups of birds, which might at least potentially interact or interbreed with each
other (reviewed in Mundinger 1982; Catchpole & Slater 1995).
The orange-tufted sunbird (Nectarinia osea) is a small (6–8 g), socially
monogamous and territorial passerine. It is a very common resident in gardens
near human dwellings throughout Israel, due to the abundance of ornamental
gardens, plentiful with cultivated nectar-secreting flowers all year round.
Populations of this species are characterized by discrete song dialects throughout
U. S. Copyright Clearance Center Code Statement: 0179-1613/2002/1084/0367/$15.00/0 www.blackwell.de/synergy
368
N. Leader, J. Wright & Y. Yom-Tov
Israel (Leader, unpubl. data). Recently, two distinct local dialects with a sharp
boundary between them were discovered in a study of a stable population of 63
territorial sunbird pairs in a small (1.5 km2) urban neighborhood in Ramat-Aviv,
Israel (Leader et al. 2000). The main distinction between the two dialects found in
the study population is the frequency of the trill, which comprises the last part of
the song (Fig. 1). A large difference of 2–3 kHz, with no overlap, was found in the
peak frequency of the trill, which in the ‘low’ dialect reaches a mean of 5 kHz; in
the ‘high’ dialect it reaches up to 8 kHz. These two dialects split the dense sunbird
population in the neighborhood in two, with about 60% of the males
(37 territorial males) singing one dialect and 33% (21 territorial males) singing
the other. Along the boundary that separates the two dialect populations,
neighboring birds sing different dialect songs, although they were only 20–30 m
apart. A few additional males occupying territories near the dialect boundary sing
both dialects or ‘hybrid’ songs.
The historical processes leading to the formation of this dialect system
probably result from the pattern of human settlement at the time of the
establishment of this neighborhood in the early 1950s (Leader et al. 2000).
However, several findings strongly point to the possibility that the current sunbird
dialect system in Ramat-Aviv is not merely the consequence of a non-adaptive
cultural founder effect 50 yr ago, and that an active mechanism (although still
undetermined) is responsible for the maintenance of these local dialects at the
present. Zilberman et al. (1999) have found high territorial fidelity and a high
degree of relatedness between neighboring individuals, as evidenced from high
proportions of band sharing discovered in genetic analysis of this population. In
addition, we have found very low dispersal rates of birds from their natal dialect
area to the adjacent one (Leader et al. 2000). This finding is extremely surprising,
considering the very small geographical scale.
Although local dialects on a microgeographic scale have been reported in a
number of bird species (reviewed in Mundinger 1982; Catchpole & Slater 1995), it
seems that the ultra-microgeographic pattern of the Ramat-Aviv sunbird dialect
system is extreme. Thus, inferring discrimination capability by sunbirds is a
necessary prerequisite for proposing an adaptive significance to dialect maintenance in this population.
Male sunbirds are highly aggressive in defense of their territory during the
breeding season. Males launch attacks from high song-posts at approaching male
rivals seeking extra-pair fertilizations (Goldstein & Yom-Tov 1988). Using song
playback, we simulated a male intrusion into the subject male’s territory. We
hypothesized that if males are capable of discrimination between songs of their
own dialect from those of the adjacent dialect, they will exhibit a differential
behavioral response to playback stimuli.
Methods
The experiments were conducted at Ramat-Aviv, a suburban neighborhood
of Tel Aviv, Israel (3205¢ N, 3447¢ E). Playback experiments were conducted
Dialect Discrimination by Male Orange-Tufted Sunbirds
369
Fig. 1: Spectrograms of orange-tufted sunbird songs used in playback experiments: (A–J) ‘low’ dialect
song; (K–T) ‘high’ dialect song
during the morning hours (08 : 00–12 : 00 h) between May and June 1999.
Experimental subjects were 23 color-ringed territorial males (13 of the low dialect
and 10 of the high dialect). Each male was tested once with a playback tape of
either the ‘low’ dialect or the ‘high’ dialect (Fig. 1), selected at random, and then
370
N. Leader, J. Wright & Y. Yom-Tov
tested again 1 h later, with a tape of the other dialect. Because the stage of the
nesting cycle may influence male response to song playbacks, all playbacks were
conducted on males whose female mate was in the incubation stage.
Ten typical songs were selected from recordings of each dialect for making the
playback tapes (Fig. 1). All songs used during our playback experiments were from
males other than the test males or their immediate neighbors, and were recorded at
close range (between 5 and 10 m) using a Sennheiser ME-67 ‘shotgun’ microphone
and a Sony WM-D6 cassette recorder. We then selected only high quality songs for
the playback recordings by inspecting spectrograms. We adjusted the song length
and the number of trill elements in each of the 10 songs to equal five trill elements
using Avisoft SASLab Pro for Windows computer software (Avisoft-SASLab Pro,
version 3.5 computer software (Raimund Specht, Berlin)).
Playback trials were conducted as follows: A Sony SRS-A31 self-amplified
speaker (70 Hz–20 kHz flat frequency response) connected to a Sony WM-D6
cassette recorder, was placed in the middle of a target male’s territory, at a height
of 1.5–2 m in a tree, within 5–10 m of a singing male. Each playback trial lasted
5 min and consisted of 30 songs of one male selected at random, delivered at a
natural singing rate of one song every 10 s. The playback level was adjusted by ear
to that of a naturally singing bird and thereafter kept constant throughout all the
tests.
Responses of the subject were dictated into a cassette recorder and later
transcribed onto data sheets. All behavioral responses were tallied in real-time.
We measured the following movement variables: number of flights (including
short flights and hops from branch to branch in a bush or tree), pause length (the
time in seconds where the subject male was stationary in one place) and the
distance from the speaker. In addition, we measured the following acoustic
variables: the number of full songs given by the male subject; the number of calls;
and the number of quick chatter bouts.
Behavioral measures of discrimination may vary in their reliability. Response
latency is a popular measure of response intensity, but poses significant problems
in interpretation (Stoddard 1996). Delayed latency to respond to playback may
indicate that a song stimulus has a low threat value, but it may also mean that the
song did not transmit well through the environment. This may be especially true
in our case, where the birds inhabit a noisy urban environment. We therefore did
not use this measure, and started each trial once we observed the target male
approaching the speaker.
Analysis
We first subjected the six response variables (following standardization) to a
principal components analysis (PCA). This was in order to extract fewer common
composite measures of response (McGregor 1992), and because response
measures correlated with each other. Each of the resulting principal components
measuring male response (having eigen values greater than 1) were then tested in a
between-groups and repeated measures analysis of variance (ANOVA) model. This
Dialect Discrimination by Male Orange-Tufted Sunbirds
371
involved two measures per male (own dialect playback vs. adjacent dialect
playback), tested as the within-subjects factor. The 23 males were distinguished by
a ‘dialect’ variable (low dialect vs. high dialect), tested as a between-subjects effect.
In this way, it is possible to test for both effects together in the same model, as well
as for any interaction, i.e. whether high or low dialect males responded differently
in their relative responses to own vs. adjacent dialects.
It is possible that the strength of response of experimental males of both
dialects to playback of their foreign dialect is influenced by differences in the
distance of their territories from the dialect boundary and other foreign males.
That is, males closer to foreign males are possibly more exposed to foreign dialect
song. We therefore used two measures of distance: aerial distance in meters; and
number of territories to the nearest foreign dialect male territory boundary, as
covariates in this analysis. Distance was measured by placing a ruler on a 1:2000
aerial photograph of the study site in which all territory boundaries were plotted.
A straight line was measured from the nearest foreign dialect male territory to
each of the experimental subjects. We also counted the number of territories
falling along this line, between the playback male and the nearest foreign dialect
male territory.
Results
Males typically reacted to playback by quickly approaching in a few short
flights. They first perched at a height of 5–8 m near the playback source, and then
constantly changing perches in an agitated manner, while often calling and
chattering vigorously.
Males of both dialects approached closer to the playback speaker when their
own dialect song was played (Table 1). In most cases, the male hovered or
perched very close to the playback speaker and tried pecking at it, while hearing
his own dialect song. Additionally, males of both dialects performed more flights
and hops and paused for shorter intervals at perches, while exposed to playback
of their own dialect (Table 1). Overall, these results suggest a more excited
response by males to the playback of their own song dialect.
In the PCA the two principal components extracted explained 43.1% and
21.1% of the total variation, respectively. The three movement variables
contributed the most to the first principal component, whereas the three acoustic
variables contributed about equally to the second (Table 2).
Analysis of variance on the first principal component revealed a significant
effect of song playback type (own dialect vs. adjacent dialect), no significant
difference in the responses of males from the two dialect groups, and no
significant interaction between song type and male dialect (Table 3, Fig. 2).
Therefore, both groups of males reacted similarly. That is, they approached the
playback speaker more closely, performed more flights and hops and paused for
shorter intervals at perches, while hearing their own dialect song than while
hearing the other dialect. No significant difference in male response was found for
the second principal component (Table 3).
372
N. Leader, J. Wright & Y. Yom-Tov
Table 1: Mean values ± SD of response variables to playback of ‘low’ and ‘high’ dialect
songs by male orange-tufted sunbirds of the ‘low’ (n ¼ 13 birds) and ‘high’ dialect (n ¼ 10
birds) populations in the Ramat-Aviv study site
Response
Dialect
Own dialect
Mean ± SD
Other dialect
Mean ± SD
Distance to speaker (m)
high
low
high
low
high
low
high
low
high
low
high
low
3.1
3.4
21.1
26.6
16.8
14.2
7.2
5.5
19.40
16.2
3.9
1.8
4.4
4.3
20.9
22.1
20.8
19.3
3.8
2.0
20.7
24.8
2.8
4.1
No. flights
Pause length (s)
No. full songs
No. calls
No. chatter bouts
±
±
±
±
±
±
±
±
±
±
±
±
1.1
0.9
8.4
14.4
5.5
10.0
7.6
7.0
26.7
17.8
4.6
2.5
±
±
±
±
±
±
±
±
±
±
±
±
1.3
1.9
10.2
10.7
14.2
11.2
6.9
4.4
23.6
24.5
4.4
5.1
Table 2: Loadings of the different measures of response on the first (PC1) and second
(PC2) principal components
Loadings
Response measures
PC1
PC2
Distance to speaker (m)
No. flights
Pause length (s)
No. full songs
No. calls
No. chatter bouts
0.670
) 0.860
0.900
0.275
0.499
) 0.374
) 0.450
0.108
0.062
) 0.650
0.670
0.548
We did not find any significant effect of aerial distance from experimental
males to their nearest neighbor of the other dialect on the magnitude of the
behavioral responses to playbacks of own vs. foreign dialect of experimental
males (Table 3).
The aerial distance of experimental males to the nearest territorial male of the
other dialect ranged from 80 to 700 m (450.4 ± 34.4 m). The number of territories
within this distance ranged from one to eight (4.4 ± 0.5). The two distance
variables, although highly correlated (Pearson correlation r ¼ 0.8, p < 0.001,
n ¼ 23), show discrepancies and therefore differ, in our opinion, in their usefulness
as informative measures of the distance of territorial birds. This is due to the
presence of a recreational park, lacking sunbird territories, which forms most of
the boundary between both dialect populations (Leader et al. 2000). Therefore, in
cases of experimental male territories located near this park, the aerial distance
from experimental males to their nearest neighbor of the other dialect can be quite
large, yet the number of territories will be small (e.g. male Z76441, whose nearest
373
Dialect Discrimination by Male Orange-Tufted Sunbirds
Table 3: The results of repeated measures ANCOVAs on the first and second principal
components of responses by male orange-tufted sunbirds, showing the effect of the ‘low’
and ‘high’ dialect males to treatment playbacks of their own dialect or the adjacent dialect,
and including aerial distance and number of territories from the dialect boundary as
covariates
Factor
Effect
df
F
p
PC1
Dialect
Treatment
Interaction
Covariate (r2 ¼ 0.20)
Dialect
Treatment
Interaction
Covariate (r2 ¼ 0.08)
19
21
21
19
19
21
21
19
0.677
6.515
0.161
2.436
0.076
0.995
0.965
0.793
0.421
0.019
0.692
0.114
0.786
0.330
0.337
0.467
a
Distance
PC2
a
Distance
a
Univariate test for both covariates (aerial distance and number of territories).
Standardized regression coefficients and their statistical significance for each covariate and
the dependent variable: PC1: aerial distance: beta ¼ 0.54, t(19) ¼ 1.67, p ¼ 0.112; no. territories: beta ¼ ) 0.13, t(19) ¼ ) 0.39, p ¼ 0.698. PC2: aerial distance: beta ¼ ) 0.42,
t(19) ¼ ) 1.19, p ¼ 0.248; no. territories: beta ¼ 0.42, t(19) ¼ 1.18, p ¼ 0.251.
Fig. 2: Response of male orange-tufted sunbirds of the ‘low’ and ‘high’ dialects to playbacks of song
recorded from males of their own dialect (OWN) or adjacent dialect (OTHER). Response is measured
as the mean (± SE) scores on the first principal component of responses to the playback speaker
during the playback period
foreign dialect male is located at an aerial distance of 420 m across the park, with
no sunbird territories in between). However, as with aerial distance, territorial
distance had no significant effect on the scale of male responses and had little effect
on the main experimental results when included as a covariate in Table 3.
374
N. Leader, J. Wright & Y. Yom-Tov
Discussion
The overall pattern arising from the results of these experiments is a
differential behavioral response by males to playbacks of their own and to
adjacent dialects. This behavioral discrimination implies capability for recognition of differences between the two dialects by the birds (and not only the
researcher). It is therefore a crucial first step before any attempt to examine how
and why these dialects are maintained in the present.
Furthermore, it appears that the bird’s home dialect songs elicit a stronger or
more excited response by territorial males of both dialects, as is evident from the
patterns shown in Table 1 and Fig. 2. This result conforms with results from
other experiments on different species, where a stronger response was elicited by a
male’s own dialect than by a dialect from some distance away (i.e. white-crowned
sparrows, Zonotrichia leucophrys: Milligan & Verner 1971; Baker et al. 1981;
Petrinovich & Patterson 1981; corn buntings, Miliaria calandra: McGregor 1983;
cardinals, Richmondena cardinalis: Lemon 1967; redwings, Turdus iliacus: Bjerke
1984). However, see Baker et al. (1981) for the opposite result.
Two primary hypotheses have been proposed for the general pattern of
stronger territorial responses to local compared to foreign song dialects in birds
(see Wright & Dorin 2001 for recent review). One explanation is that foreign males
are strongly discriminated against in mate and/or extra-pair choice by local
females, and so represent little threat to an established male’s breeding success
(Rothstein & Fleischer 1987). A second explanation is that males use the songs they
hear locally to learn what the species song ought to sound like, and respond less to
foreign songs to the extent that they differ from the standard (Dabelsteen &
Pedersen 1992; Nelson 1998). Lower response to foreign song under this hypothesis
is simply a non-adaptive consequence of song learning (Melman & Searcy 1999).
We do not aim at this stage to distinguish between these two hypotheses.
However, as both emphasize the term ‘local’, and given the very small geographic
scales in sunbirds (i.e. neighboring birds along the boundary having different
dialect songs, yet being only tens of meters apart), it would be surprising if local
females or males would not indeed be frequently exposed to both dialects. This
would therefore lower their discrimination value against any ‘foreign’ song. This
discrepancy can be resolved if we consider the acoustic properties of the local
habitat. The study site is located in the middle of an urban neighborhood. There is
thus an abundance of environmental noise of urban origin (e.g. automobile traffic,
planes from a nearby airport, construction and garden landscaping work, etc.).
More importantly, however, males and females occupy small territories in small
gardens between private homes and apartment buildings, most of them more than
two to three stories high. The whole study area is heavily vegetated, with many
mature human-planted trees (e.g. Ficus spp., Pinus spp., Eucalyptus spp.,
Erythrina spp., Melia azedarach, Araucaria excelsa, Tipuana tipu, Dalbergia
sissoo, Delonix regia). This complex acoustic environment poses problems for
song transmission over any distance, since vocalizations are prone to distortion
due to reverberations and frequency-dependent attenuation (reviewed in Catch-
Dialect Discrimination by Male Orange-Tufted Sunbirds
375
pole & Slater 1995). The high dialect, for example, is heavily distorted by
frequency-dependent attenuation, to such an extent that at a distance of 60–70 m
(one or two territories), it is virtually identical in its frequency range to that of the
low dialect (i.e. all frequencies above 5 kHz are totally attenuated; Leader,
unpubl. data). If such a song is heard and possibly interpreted as different by
females and/or males at a distance of just one or two territories away, then it may
seriously limit the number of neighboring males that a female/male sunbird can
assess in order to establish what their local song is. A stronger response to one’s
own dialect is therefore expected, and dialects could therefore be maintained in
such a complex acoustic environment regardless of the very small distance
between territories. This suggestion is further supported by the fact that we did
not find any effect of either measure of distance to the nearest foreign dialect male
neighbor, on the magnitude of the behavioral responses of experimental males in
response to playbacks of their own vs. foreign dialect songs.
The aerial distance to the nearest foreign dialect male territory reflects the
smallest direct route through which sound can travel. Although this measure does
not take into consideration any acoustic obstacles in its path (e.g. tall buildings,
large trees, a major road, etc.), it still provides a good estimate of the minimum
distance across which a male will have a chance of hearing any foreign dialect
song. The number of sunbird territories to the nearest foreign dialect male
territory is a less accurate measure of sound transmission distance. However, it
provides information on the number of neighboring males whose song dialect
resembles that of the experimental bird’s own song, that are in the way to the
nearest source of foreign dialect song. The greater the number of territories
between a male and the nearest foreign dialect song, the more birds a male will
hear singing his own dialect song from that direction. Therefore, the greater the
chance of any faraway song of a different dialect being masked. In addition, the
greater the number of ‘own’ dialect territories in between, the more difficult it is
for a male to approach the source of a different dialect song during ex-territorial
forays. This is because it requires crossing and encountering a greater number of
territorial males on the way. Thus, these two distance measures, and their lack of
any effect on male behavioral responses to our playback, supports our suggestions
that, even on such a small scale as our study site, it is the acoustic properties of the
environment that contribute to the maintenance of discrete dialects in sunbirds.
The dialect discrimination capability of male sunbirds reported in this paper
adds to strengthen our hypothesis that the current sunbird dialect system in
Ramat-Aviv is not merely the consequence of a non-adaptive cultural founder
effect 50 yr ago, when this neighborhood was first inhabited by sunbirds. The
present differential response to own vs. foreign conspecific song of a neighboring
dialect strongly suggests the existence of a mechanism which currently maintains
these dialects at the current boundaries. All of this is consistent with the current
spatial distribution of the two dialect populations, the low dispersal rates of birds
from their natal dialect area (Leader et al. 2000), and the acoustic properties of
the habitat (Leader, unpubl. data). Further research is now needed to investigate
the specific nature of this mechanism.
376
N. Leader, J. Wright & Y. Yom-Tov
Acknowledgements
We thank A. Feldman for assistance in the field. This study was supported by a grant from the
Tel Aviv University Research Fund to Y. Y.
Literature Cited
Baker, M. C., Thompson, D. B., Sherman, G. L. & Cunningham, M. A. 1981: The role of male vs.
male interactions in maintaining population dialect structure. Behav. Ecol. Sociobiol. 8, 65—69.
Bjerke, T. K. 1984: The response of male redwings Turdus iliacus to playback of different dialects.
Fauna Norv. Series C. Cinculus 7, 24—27.
Catchpole, C. K. & Slater, P. J. B. 1995: Bird Song: Biological Themes and Variations. Cambridge
Univ. Press, Cambridge.
Dabelsteen, T. & Pedersen, S. B. 1992: Song features essential for species discrimination and behaviour
assessment by male blackbirds (Turdus merula). Behaviour 121, 259—287.
Goldstein, H. & Yom-Tov, Y. 1988: Breeding biology of the orange-tufted Sunbird in Israel. Ardea 76,
169—174.
Leader, N., Wright, J. & Yom-Tov, Y. 2000: Microgeographic song dialects in the orange-tufted
sunbird (Nectarinia osea). Behaviour 137, 1613—1627.
Lemon, R. E. 1967: The response of cardinals to songs of different dialects. Anim. Behav. 15,
538—547.
McGregor, P. K. 1983: The response of corn buntings to playback of dialects. Z. Tierpsychol. 62,
256—260.
McGregor, P. K. 1992: Quantifying responses to playback: one, many, or composite multivariate
measures. In: Playback and Studies of Animal Communication (McGregor, P. K., ed.). Plenum
Press, New York, pp. 79—96.
Melman, D. S. & Searcy, W. A. 1999: Microgeographic song discrimination in a nonterritorial
passerine, the Boat-tailed Grackle. Condor 101, 845—848.
Milligan, M. M. & Verner, J. 1971: Inter-populational song dialect discrimination in the Whitecrowned sparrow. Condor 73, 208—213.
Mundinger, P. C. 1982: Microgeographic and macrogeographic variation in the acquired vocalizations
of birds. In: Acoustic Communication in Birds (Kroodsma, D. E. & Miller, E. H., eds). Academic
Press, New York, pp. 147—208.
Nelson, D. A. 1998: Geographic variation in song of Gambel’s white-crowned sparrow. Behaviour 135,
321—342.
Petrinovich, L. & Patterson, T. L. 1981: The responses of White-crowned Sparrows to songs of
different dialects and subspecies. Z. Tierpsychol. 57, 1—14.
Rothstein, S. I. & Fleischer, R. C. 1987: Vocal dialects and their possible relation to honest status
signalling in the brown-headed cowbird. Condor 89, 1—23.
Stoddard, P. K. 1996: Vocal recognition of neighbors by territorial passerines. In: Ecology and
Evolution of Acoustic Communication in Birds (Kroodsma, D. E. & Miller, E. H., eds). Cornell
Univ. Press, Ithaca, pp. 356—376.
Wright, T. F. & Dorin, M. 2001: Pair duets in the yellow-naped Amazon (Psittaciformes: Amazona
auropalliata): responses to playbacks of different dialects. Ethology 107, 111—124.
Zilberman, R., Moav, B. & Yom-Tov, Y. 1999: Extra-pair paternity in the socially monogamous
orange-tufted sunbird (Nectarinia osea osea). Israel J. Zool. 45, 407—421.
Received: May 24, 2001
Initial acceptance: July 18, 2001
Final acceptance: October 4, 2001 (A. Kacelnik)