Behavioral Ecology and Sociobiology

, Volume 67, Issue 1, pp 145–152

Dealing with urban noise: vermilion flycatchers sing longer songs in noisier territories

Authors

    • Departamento de Ecología Evolutiva, Instituto de EcologíaUniversidad Nacional Autónoma de México
    • Department of Evolution and EcologyUniversity of California
  • Esmeralda Quirós-Guerrero
    • Departamento de Ecología Evolutiva, Instituto de EcologíaUniversidad Nacional Autónoma de México
  • Diego Gil
    • Departamento de Ecología EvolutivaMuseo Nacional de Ciencias Naturales (CSIC)
  • Constantino Macías Garcia
    • Departamento de Ecología Evolutiva, Instituto de EcologíaUniversidad Nacional Autónoma de México
Original Paper

DOI: 10.1007/s00265-012-1434-0

Cite this article as:
Ríos-Chelén, A.A., Quirós-Guerrero, E., Gil, D. et al. Behav Ecol Sociobiol (2013) 67: 145. doi:10.1007/s00265-012-1434-0

Abstract

In noisy conditions, several avian species modulate their songs in amplitude and in the temporal or frequency domains, presumably to improve communication. Most studies on how passerine birds perform such adjustments have been carried out in oscines, a group well known for the importance of learning in the development of their songs. On the other hand, suboscines, in which learning appears to have little influence on the development of their songs, have been largely neglected. We evaluated song adjustment to noise in the vermilion flycatcher (Pyrocephalus rubinus), a suboscine bird. We conducted song recordings and noise measurements at several territories within Mexico City during the length of the dawn chorus. Males living in noisier places sang long songs, while those males inhabiting quieter places sang both short and long songs. We also found evidence of individual song plasticity, as males sang less versatile songs (i.e., songs with more introductory elements) later in the morning when noise levels were higher. This individual shift in song seems to be more associated to time of the day rather than to the observed rise in noise. However, we cannot discard an effect of noise, which should be evaluated with an experiment. We discuss our results in the context of other studies with oscine passerines and other taxa and consider implications for signaling in intra- and intersexual contexts.

Keywords

Vermilion flycatcherPyrocephalus rubinusNoiseBird songSuboscineSong plasticity

Introduction

Increasing urbanization through the expansion of the city perimeters and increased development of highways have been related to a reduction in bird diversity and reproductive success (e.g., Reijnen and Foppen 1994; Reijnen et al. 1995; Kuitunen et al. 1998). Aside from direct effects of reduced habitat quality and poorer food resources, noise from vehicle traffic has been related to this reduction in fitness (Reijnen et al. 1995; Francis et al. 2009; Halfwerk et al. 2011a). In this context, many studies have investigated the influence of anthropogenic noise on bird song (reviews in Brumm and Slabbekoorn 2005; Patricelli and Blickley 2006; Ríos-Chelén 2009). This is because birds use their songs during social interactions, for instance, to attract females and repel intruders from their territories (Catchpole and Slater 2008), which results in differential reproductive success among individuals. Thus, factors affecting acoustic communication, such as noise, may impair breeding success. Urban noise may disrupt vocal communication, especially if animal acoustic signals occur near the same low frequencies of urban noise, in which case these signals may be masked and territory defense or mate attraction will be impaired. Even if a male is able to attract a female, noise may still erode female preferences (Swaddle and Page 2007), disrupting pair bonds and compromising reproduction.

Several oscine species have been shown to undergo presumably adaptive song adjustments to noise, which may enable acoustic communication in noisy conditions. For instance, European robins (Erithacus rubecula) sing at times when anthropogenic noise levels are lower (Fuller et al. 2007), nightingales (Luscinia megarhynchos) sing at a higher amplitude when facing high levels of urban noise (Brumm 2004), and red-winged blackbirds (Agelaius phoeniceus) are able to increase signal tonality when noise suddenly increases (Hanna et al. 2011). However, most studies so far have dealt with another song modification, namely, an increase in song pitch with noise (Slabbekoorn and Peet 2003; Fernández-Juricic et al. 2005; Slabbekoorn and den Boer-Visser 2006; Wood and Yezerinac 2006; Bermúdez-Cuamatzin et al. 2009, 2011; Hu and Cardoso 2009; Nemeth and Brumm 2009; Gross et al. 2010; Potvin et al. 2011). These studies have shown that individuals living in more noisy environments sing at a higher pitch than those living in quieter ones. Because urban noise is mainly composed of low frequencies, singing at higher frequencies, above those of urban noise, may lead birds to avoid noise masking. Nevertheless, most studies addressing how ambient noise levels influence songs in passerines have been conducted in oscines. In contrast, studies on the sister group of the oscines, the suboscines, representing more than 30 % of the tropical avifauna (Chesser 2004) have been largely neglected (exceptions being Francis et al. 2011a; Ríos-Chelén et al. 2012). The study by Francis et al. (2011a) showed that, of two suboscines investigated, the ash-throated flycatcher Myiarchus cinerascens and the gray flycatcher Empidonax wrightii, only the first species presented a positive significant association between song frequency and noise levels, which may be explained in part by differences in noise masking between the two species (see “Discussion”). Recent data on several species indicate that suboscines display a weaker relationship between song frequency and noise than oscine species (Ríos-Chelén et al. 2012). The latter study focused on frequency attributes of the songs and would have missed variation in temporal song attributes which may also help birds to contend with environmental noise. For instance, the Japanese quail (Coturnix japonica) increases its call rate and call bout length as noise increases (Potash 1972), whereas the song of the chaffinch (Fringilla coelebs) becomes more redundant in places with high natural noise levels (Brumm and Slater 2006). Other vertebrates increase their call duration in noisy conditions, such as marmosets (Callithrix jacchus, Brumm et al. 2004) and Cope’s gray tree frogs (Hyla chrysoscelis), which also increase its call rate (Love and Bee 2010).

In this study, we investigated whether males of a suboscine species, the vermilion flycatcher (Pyrocephalus rubinus), display the same type of frequency and temporal song modifications under noisy conditions as found in oscine birds and in other taxa. The vermilion flycatcher (Tyrannidae, Fluviconilae) is a socially monogamous, fiercely territorial, and sexually dimorphic species, which sings only one type of song while perched (Smith 1967; Ríos Chelén et al. 2005), mostly before sunrise (Ríos Chelén et al. 2005). The song of this species is composed of two parts: the first formed by a variable number of introductory elements and the second composed of three types of elements: two middle elements, one high frequency, and one terminal element (Fig. 1). Vermilion flycatchers can modify their songs in several ways, more commonly (1) they produce a flight-song display, which is a short song whose initial elements may or may not be substituted by a short call (Ríos-Chelén and Macias-García 2004) and (2) they modify the length of their songs through the addition of introductory elements, seemingly to adjust the intensity of the territorial display (Fig. 1, Ríos Chelén et al. 2005).
https://static-content.springer.com/image/art%3A10.1007%2Fs00265-012-1434-0/MediaObjects/265_2012_1434_Fig1_HTML.gif
Fig. 1

Songs of vermilion flycatcher males. Songs are composed of a variable number of introductory elements, two middle elements, one high-frequency element, and one terminal element. Thus, each song is composed of four different types of elements (i.e., introductory elements, middle elements, high-frequency element, and terminal element). Males can change the length of their songs by modifying the number of introductory elements. In a and b, we show a song with seven and four introductory elements, respectively. This means that the song in a has a lower value of versatility (number of different element types / total number of elements = 4/11 = 0.36) than the song in b (4/8 = 0.50). Modified from Ríos Chelén et al. (2005) with permission from Brill

If vermilion flycatchers are able to perform similar temporal and frequency song modifications under noisy conditions as found in oscines and other animals, we expect that in noisier territories, they will sing (1) songs with higher minimum frequencies and (2) longer songs (i.e., with more introductory elements), since longer signals are expected to be more easily detected in places with higher levels of noise. Similarly, if vermilion flycatchers can individually modulate their songs, we expect that in moments of higher noise levels, they will (1) increase the pitch of their songs and (2) increase the length of their songs by adding introductory elements.

Methods

Study site and subjects

We recorded songs of territorial vermilion flycatcher males and registered noise levels at different moments of the predawn and dawn chorus. Males may modify their songs depending on the social context, as songs with more elements are likely to be more threatening (Ríos-Chelén and Macías Garcia 2007). Therefore, it is possible that males with more neighbors produce songs with more introductory elements. Consequently, we also registered the number of neighbors that surrounded each male (Ríos-Chelén and Macias-García 2004).

We recorded a total of 29 males in different parks and urban areas of Mexico City. Twenty-three were recorded in the Reserva del Pedregal de San Ángel, Universidad Nacional Autónoma de México (UNAM) (19°20′01″ N, 99°11′54″ W). This urban area is composed of mixed vegetation patches, trees, and grasslands, but it also contains roads and busy avenues. Four males were recorded in the “Bosque de Chapultepec” park (19°24′32.4″ N, 99°12′38.3″ W). This consists of 243.9 ha, of which 53 ha represents green areas harboring different tree species, and is crisscrossed and encircled by highways and avenues. Two additional individuals were recorded at the Centro Nacional de las Artes (19°26′8.3″ N, 99°8′28.6″ W), a set of buildings surrounded by parks containing abundant trees and shrubs. It lies at the intersection of two very busy avenues. Because all of our recorded males were in the same urban area (Mexico City), the degree of urbanization (i.e., urban vs rural vs natural areas) was not a potential confounding factor and so was not included in our analyses.

Birds were not ringed but were identified by their territories, roosting and singing perches (as they sing from their roosts before dawn). Long-term studies in other Mexican populations show that color-ringed individuals are recaptured near the same singing perches between different days and even years (AAR-C, personal observation), and at UNAM, we have recorded a ringed individual occupying the same territory for over 6 years.

Song recordings and noise measurements

We recorded birds and registered noise during the predawn and dawn chorus (overall hour range = 0540 to 0710 hours), from March 30, 2009 to June 26, 2009. Total time from the first to the last recording was (mean ± standard error (SE)) 44 ± 2.8 min (n = 29 males; range between males, 20–78 min). Each male was recorded several times (for recording times, see “Statistical analyses”), in only 1 day, with a Sennheiser™ ME66 shotgun microphone connected to a digital recorder (Marantz™ PMD660, sampling rate 44 kHz, 16-bit accuracy), from a distance of 5–10 m. In each sample, we recorded around five to ten songs and then measured noise levels with a SEW™2310 SL digital sound level meter (range, 30–130 dB; frequency weighting: A; slow response; ANSI S1.4, Class II). With the sound meter pointing upwards (above the head of the observer), we registered noise levels every 10 s for 1 min. The sound level meter was at a distance of 5–10 m from the bird that was perching on a tree at around 45° with respect to the sound meter. For the analysis, we used the averages of these six measurements. We repeated this sequence (song recordings and six noise measures) several times (see below) for each individual since noise levels change during the day, increasing through the early morning as the activity in the city rises. Thus, our recording protocol is expected to provide us with song recordings for each individual in several noise conditions, as required to evaluate the hypothesis that individuals can modulate their songs in real time in response to ambient noise. From now on, we will refer to each of these samples (i.e., song recording + noise measures) as a period. We obtained an average (±SE) of 9.8 ± 0.8 periods per individual (range = 4–20).

Song measurements

We measured song attributes from spectrograms (window: window Hann, DFT size 512 samples, overlap 50, filter bandwidth 124 Hz) using Raven Pro™ v. 1.3 (Cornell Laboratory of Ornithology). For each song, we measured song length (in second) and counted the total number of elements. We also calculated song versatility, obtained by dividing 4 (the number of different element types in the vermilion flycatcher song, which is 4; see “Introduction”) by the total number of elements in the song. Since songs mainly vary in the number of introductory elements, a larger number of these elements in a song would lead to a reduction in song versatility (Fig. 1). We also measured peak or dominant frequency (the frequency with maximum amplitude, in hertz) and frequency range (song maximum frequency–song minimum frequency, in hertz). To obtain a measure of song minimum and maximum frequency (in hertz), we used the power spectrum function in Raven. This power spectrum allowed us to calculate the frequency at which the song amplitude dropped by a value of 20 dB from the maximum amplitude towards the lower or the higher part of the song in the frequency axis, which corresponded to our measure of song minimum or maximum frequency, respectively (Podos 1997). Some song attributes were highly intercorrelated (Table 1). To reduce the number of statistical tests (and thus type I errors), we reduced the number of song parameters and analyzed those that were not highly intercorrelated: maximum and minimum song frequencies, and song versatility. However, because background noise was more intense at low frequencies, several of our songs had too low signal-to-noise ratios to allow us to measure song minimum frequency using power spectra, and this decreased our available sample size (our sample size of minimum frequency dropped from 29 to 13 males, and from 29 to 4 males for the inter- and intraterritory analyses respectively). Thus, we were prevented from analyzing minimum song frequency in the intraterritory comparisons. However, we also analyzed song peak frequency, since this measure is correlated with song minimum frequency (Table 1). Because of similar problems with background noise, the peak frequency has previously been used instead of the song minimum frequency (Nemeth and Brumm 2009). Additionally, since peak frequency has the largest amount of energy in the song, it makes sense to consider this song attribute. We did not correct for multiple testing (Moran 2003; Nakagawa 2004). Table 2 shows descriptive statistics of measured song parameters.
Table 1

Pearson’s correlation coefficient of measured song attributes

 

Min freq

Max ,freq

Freq range

Song length

# elem

Song ver

Peak freq

Min freq

 

0.54

−0.48

0.06

0.21

−0.22

0.64*

Max freq

  

0.42

0.04

0.11

−0.10

0.46*

Freq range

   

−0.18

−0.31

0.31

−0.13

Song length

    

0.78**

0.74**

0.11

# elem

     

0.99**

0.11

Song ver

      

−0.03

Some song parameters were highly intercorrelated. Highly correlated song characteristics are shown in italics. n = 29 in song length, song ver, and # elem. n = 28 in peak freq and max freq. n = 13 in min freq and freq range

Min freq minimum frequency, Max freq maximum frequency, Freq range frequency range, # elem number of elements, Song ver song versatility (number of different element types/total elements), Peak freq peak frequency

*P < 0.05; **P < 0.001

Table 2

Descriptive statistics of measured song parameters

Song parameter

Number

Mean

SE

Range

Song length (s)

29

0.846

0.026

0.568–1.19

# elem

29

9.184

0.153

7.56–10.63

Song ver

29

0.443

0.007

0.38–0.53

Min freq (Hz)

13

2923

59

2532–3,274

Max freq (Hz)

28

6,062

30

5,680–6,570

Peak freq (Hz)

28

4,677

32

4,191–5,093

Freq range (Hz)

13

3,082

59

2,762–3,532

Abbreviation names as in Table 1

SE standard error

Statistical analyses

Since data were normally distributed, we used parametric tests. For the between male comparisons, we used a stepwise multiple regression analysis to take into account first the predictor variables that were more significantly related to the dependent variable, while excluding those that were not significantly related. We entered to the model the song attributes as dependent variables, and the number of neighbors, elapsed time registering each bird, date, and noise level as predictor variables. For each individual, we averaged the noise levels and the song characteristics from the first, the last, and the intermediate randomly selected period. These averages were used in the regression analysis. This sampling of periods was designed to obtain a representative, unbiased selection of noise experienced, and songs produced by each bird. The recordings of the first, last, and intermediate periods lasted in average (±SE) 0.87 ± 0.1, 1.15 ± 0.1, and 1.00 ± 0.1 min, respectively. We chose those songs of the best quality available. The mean (±SE) numbers of measured songs were 5.1 ± 0.64, 6.3 ± 0.86, and 7.5 ± 0.96 for the first, last, and intermediate periods, respectively.

To investigate whether each individual modulates its songs in relation to noise levels (individual plasticity), we compared the averaged attributes of songs produced during the period with the lowest vs the period with the highest noise levels. For this purpose, we used a repeated measures general linear model with the song attributes as dependent variables and noise as a factor with two levels (lowest and highest). Our extreme recordings, during the lowest and highest noise periods, lasted on average (±SE) 0.78 ± 0.11 and 1.41 ± 0.24 min, respectively. We measured a mean (±SE) of 4.9 ± 0.63 and 6.1 ± 0.75 songs per male in the periods of less and more noise, respectively. Because these were intraterritory comparisons, other factors such as elapsed time registering each bird, number of neighbors, habitat type, and body size were controlled. However, for most birds, the lowest noise period was recorded early in the morning, whereas the highest noise period took place later in the morning. Therefore, a possible shift in song attribute(s) could be due to a time effect rather than to an increase in noise. To investigate this possibility, we also analyzed our data using a different set of songs. For this purpose, we split our males into two groups: one group (n = 14 males) in which the high-noise period selected for analysis was early in the morning (and the low-noise period later in the morning) and the other group (n = 15 males) where the low-noise period was early in the morning (and the high-noise period later in the morning). If noise has an effect on any song attribute, we expect a shift in song with noise irrespective of whether the direction of noise change is from high to low, or from low to high levels. Because we had many recorded periods for each male (see “Song recordings and noise measurements”), we could select other periods to perform this analysis. Statistical analyses were conducted with SPSS v. 17.

Results

Noise levels

Noise varied within territories (all territories considered) from a mean (±SE) of 49.6 ± 1.0 dB(A) (range, 42.7–61.7) during the quietest periods to 57.3 ± 1.3 dB(A) (range, 48.7–73.9) during the loudest; as expected, these within-territory differences in noise between the quietest and the loudest periods were significantly different (paired t test: df = 28, t = −8.4, P < 0.001). The mean (±SE) difference in noise level between these two extreme periods was 7.67 ± 1.0 dB(A) (range, 1.5–22.9). Although intraterritory variation in noise was significant, variation in noise levels across territories was far larger (ANOVA: F28,58 = 4.96, P < 0.001); mean noise levels ranged between territories from 47.7 to 73.1 dB(A) (mean ± SE, 53 ± 0.7).

Variation between territories

The stepwise multiple regression revealed a positive and significant association between noise and number of song elements (β = 0.41, P = 0.025, n = 29, Fig. 2). All other predictor variables were not significantly related to this song attribute (P > 0.3 in all). For minimum song frequency, no significant predictors were detected and none were entered into the model. On the other hand, there was a negative association between song maximum frequency and both date (β = −0.56, P = 0.002, n = 28) and elapsed time registering each bird (β = −0.32, P = 0.047, n = 28). For song peak frequency, no predictor variables were kept in the model.
https://static-content.springer.com/image/art%3A10.1007%2Fs00265-012-1434-0/MediaObjects/265_2012_1434_Fig2_HTML.gif
Fig. 2

Mean number of song elements and noise level. Most males in noisier territories sang with many (more than nine) song elements. Some birds in quieter places sang with many elements while others sang with few

Within-territory variation

Birds sang less versatile songs in situations of high noise levels (repeated measures general linear model: F1,26 = 5.80, P = 0.023). They sang songs with a mean (±SE) of 9.0 1 ± 0.21 and 9.39 ± 0.17 song elements during the quietest and noisiest periods, respectively, showing a nonsignificant tendency to emit songs with more elements (repeated measures general linear model: F1,26 = 3.60, P = 0.069), during the highest noise period. Because these were intraterritory comparisons, other factors (e.g., elapsed time registering each bird, habitat type, male body weight) were controlled for. However, since in this extreme period comparison the loudest periods were generally recorded later in the morning than the quietest ones, the shift in song versatility could conceivably be due to a temporal drift (i.e., a time effect).

To explore a possible effect of time, we performed new within-male analyses for song versatility using other periods instead of the extreme ones reported above. For half of the males, we chose songs that were recorded during a high-noise period early in the morning, and for the other half of the males, we used songs recorded at high noise levels late in the morning (see “Methods” for more details). With these new data, we found no association between song versatility and period (repeated measures general linear model: F1,28 = 1.90, P = 0.178), suggesting that the decrease in song versatility during the highest noise period was a by-product of time of the day (i.e., singing later in the morning), instead of an increase in noise. Furthermore, using also this song sample, we found that only those males for which the low-noise period was early in the morning decreased song versatility later in the morning during the high-noise period (paired t test: t = −2.5, df = 14, P = 0.022; Fig. 3a); those males from which we used their recorded high noise level period early in the morning did not show a change in song versatility later in the morning during the low-noise period (paired t test: t = 0.83, df = 13, P = 0.419; Fig. 3b). No other song attributes showed a significant shift during the two extreme periods (P > 0.1 in all).
https://static-content.springer.com/image/art%3A10.1007%2Fs00265-012-1434-0/MediaObjects/265_2012_1434_Fig3_HTML.gif
Fig. 3

Song versatility as a function of times of day. a When the periods of high noise level (open circles) occurred later in the morning, individual birds sang less versatile songs than during the quieter periods (black circles) earlier in the morning. b Males still sang less versatile songs under low-noise conditions when the quieter period took place later in the morning, following the same time direction as in a (albeit the effect was nonsignificant). *P < 0.05

Discussion

In this study, we show that males occupying territories with relatively higher noise levels consistently produced long songs, whereas males in quieter places sang both long and short songs. We found that males sang less versatile songs (i.e., with more elements) later in the morning when noise level was higher, but time of the day seemed to play a more important role for driving this shift in song than noise levels (see below).

The lack of correlation between noise level and song pitch across territories is at odds with previous studies in oscines which have shown that males in noisy territories sing at a higher pitch than those in less noisy locations (Slabbekoorn and Peet 2003; Fernández-Juricic et al. 2005; Slabbekoorn and den Boer-Visser 2006; Wood and Yezerinac 2006; Bermúdez-Cuamatzin et al. 2009; Hu and Cardoso 2009; Nemeth and Brumm 2009; Potvin et al. 2011; Francis et al. 2011a, b). However, some of these studies measured minimum frequencies by hand in spectrograms, which can lead to measuring errors (Zollinger et al. 2012). Hence, it is becoming increasingly necessary to verify if this method yields different results than the method of measuring frequencies at a set threshold below the peak that was used in this study and others (Podos 1997; Nemeth and Brumm 2009). At any rate, our results suggest that, considering long-term adaptations through natural or sexual selection (i.e., long-term process, Brumm and Slabbekoorn 2005), vermilion flycatchers do not adjust their song pitch (nor song minimum, maximum, or peak frequencies) to ambient noise. These results are in line with a recent comparative analysis, in which frequency shifts were found to be greater in oscines than in suboscines, suggesting that song learning may play a role in song adaptation to noisy environments (Ríos-Chelén et al. 2012). On the other hand, Francis et al. (2011a) found contrasting patterns between two suboscine species: whereas the ash-throated flycatcher (M. cinerascens) singed higher pitched songs in noisier places, that was not the case for the gray flycatcher (E. wrightii). This difference may be partly explained by considering that the relatively higher pitched songs of the gray flycatcher are probably less masked by noise than those of the ash-throated flycatcher (Francis et al. 2011a, see also Parris and Schneider 2008 for a similar example). In the case of the vermilion flycatcher, this is also a possibility. Most of the energy of vermilion flycatcher songs is found between 2,923 and 6,061 Hz (song minimum and maximum frequencies, respectively) and concentrated around 4,677 Hz (song peak frequency), which may leave most of the songs unmasked by urban noise (Table 2, see also Fig. 1). Instead of singing higher pitched songs in noisy locations, vermilion flycatchers sang songs with more elements. One interpretation is that this would make songs easier to hear because they last longer (i.e., the introductory elements are repeated more times in a song, perhaps resulting in an increased probability of being heard, Fig. 1). Similarly, Francis et al. (2011b) found that the gray vireo (Vireo vicinior) sings longer songs in noisy locations; however, the plumbeous vireo (Vireo plumbeus) showed the opposite pattern (i.e., shorter songs in noisy places, Francis et al. 2011b). Thus, our study adds to the body of evidence showing that both birds (e.g., Potash 1972; Jouventin et al. 1999; Lengagne et al. 1999; Brumm and Slater 2006) and other vertebrates (e.g., Brumm et al. 2004; Love and Bee 2010) show noise-related changes in temporal attributes of vocalizations.

We found that males individually shifted the versatility of their songs later in the morning (individual plasticity), when noise levels were higher. However, it seems that this shift was a time effect: no association between song versatility and noise was found when we compared songs recorded during a high-noise period early in the morning with songs recorded later in the morning during a low-noise period (Fig. 3b). While these results show that time of the day has an effect on individual song versatility, we cannot discard an influence of noise on versatility. We found no evidence of males changing frequency related song attributes in response to noise, corroborating the idea that vermilion flycatchers are more plastic in temporal song attributes than in frequency parameters (Ríos Chelén et al. 2005). So far, individual song plasticity in response to noise has been demonstrated in several species, none of them are suboscines. For instance, with increased noise levels, Bengalese finches (Lonchura striata var. domestica) change syllable pitch (Tumer and Brainard 2007); great tits (Parus major) switch to song types that are less masked by noise (Halfwerk and Slabbekoorn 2009); reed buntings (Emberiza schoeniclus, Gross et al. 2010), chiffchaffs (Phylloscopus collybita, Verzijden et al. 2010), and house finches (Carpodacus mexicanus, Bermúdez-Cuamatzin et al. 2011) sing at a higher minimum song frequency; and red-winged blackbirds (A. phoeniceus, Hanna et al. 2011) sing with increased tonality.

While individual plasticity in vocal attributes may improve acoustic communication in noisy conditions (Halfwerk et al. 2011b), the interaction between noise and other ecological factors may not be straightforward; in noisy areas, some species experience a higher reproductive success (Francis et al. 2009) while others face a decline (Halfwerk et al. 2011a). Halfwerk et al. (2011b) showed that female great tits may prefer low-pitched songs, but under noisy conditions, it pays males to sing higher pitched songs because they evoke stronger responses by females. Thus, a possible trade-off between producing songs preferred by females and songs that are more easily heard in noise may arise (but see a critique for this interpretation in Eens et al. 2012). In the case of the vermilion flycatcher, it is not known whether an increase in the number of elements may impact positively or negatively mate attraction and territorial defense. Nevertheless, it is known that male songs are longer, with more elements, after the onset of nest construction, when females are potentially fertile (Ríos Chelén et al. 2005), and songs with more introductory elements are probably perceived by males as more threatening signals (Ríos-Chelén and Macías Garcia 2007). As shown in Fig. 2, there is a wider spread of song lengths in quiet territories than in noisier ones, which probably reflects the natural variation of song length (Ríos Chelén et al. 2005) and the possibility that different messages are conveyed by these song variants (Ríos-Chelén and Macias García 2007). Thus, singing longer songs even in quiet places may be a way to convey information about individual quality. These individuals may be those capable of colonizing noisy territories, in which case, the observed association across territories between noise level and number of elements may be a by-product of birds singing longer songs being able to colonize noisy areas. While loud noise may be an undesirable attribute of a territory, other variables such as abundance of food, shelter, etc. may override this effect. In fact, if predators avoid noisy places, birds settling territories in those areas would benefit from a low predation level (Francis et al. 2009). The possibility that vermilion flycatchers signal individual quality through their songs is in line with the several proposed associations between song attributes and individual quality, both in oscines (e.g., Hasselquist et al. 1996; Galeotti et al. 1997; Forstmeier et al. 2002; Christie et al. 2004) and suboscines (e.g., Murphy et al. 2008), and awaits further studies.

Our results show that a suboscine bird that probably does not learn to sing displays temporal (i.e., number of elements) intermale variation across territories varying in noise. This study supports the idea that suboscine adaptation to noise is different in degree and mode to that taking place among oscines (see also Ríos-Chelén et al. 2012), suggesting heterogeneity in the capacity of bird species to colonize and survive in the urban environment.

Acknowledgments

Fundación Banco Bilbao Vizcaya Argentaria provided financial support during this work. The authors thank Aaron Goodman for checking the English grammar of the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical standards

This study complies with the current laws of the country where it was performed.

Copyright information

© Springer-Verlag Berlin Heidelberg 2012