Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

A desert songbird with no confamilials in the Western Hemisphere (Verdin, Auriparus flaviceps) investigates divergent conspecific songs

A desert songbird with no confamilials in the Western Hemisphere (Verdin, Auriparus flaviceps)... Abstract Most birds that show geographic variation in their songs discriminate between local and foreign songs, which may help them avoid unnecessary conflicts with vagrant individuals or similar-sounding congeners. However, some species respond equally to foreign and local songs, which may be useful if foreign individuals present territorial threats or if there are no sympatric congeners to avoid. Species without sympatric congeners are not commonly tested in playback studies, but they offer an opportunity to see how song variation and recognition unfolds when the pressure to avoid similar congeners is absent. Here, we use Verdins (Auriparus flaviceps), a monotypic genus of songbird with no confamilials in North America, to explore song variation and recognition in a species living without close relatives. We assessed geographic variation in song across the Verdin range and conducted a playback experiment using exemplars from 2 acoustically divergent and geographically distant regions as treatments. We found significant geographic variation in song that mapped well onto ecologically distinct desert regions. We found that Verdins had stronger vocal responses to local-sounding songs, but had equal movement responses to local-sounding and foreign songs. These results are similar to results found in other species without sympatric congeners and provide an example of a species that investigates acoustically divergent conspecific songs, despite recognizing salient differences in those songs. RESUMEN La mayoría de las aves que muestran variación geográfica en sus cantos discriminan entre los cantos locales y extranjeros, lo que puede ayudarlas a evitar conflictos innecesarios con individuos vagabundos o con congéneres que suenan similares. Sin embargo, algunas especies responden de mismo modo a los cantos extranjeros y locales, lo que puede ser útil si los individuos extranjeros representan una amenaza territorial o si no hay congéneres simpátricos a evitar. Las especies sin congéneres simpátricos no son comúnmente evaluadas en estudios de playback, pero ellas ofrecen una oportunidad para ver cómo se revela la variación y el reconocimiento del canto cuando la presión para evitar a congéneres similares está ausente. En este estudio usamos a Auriparus flaviceps, un género monotípico de ave canora sin familiares en Norteamérica, para explorar la variación y el reconocimiento del canto en una especie que vive sin parientes cercanos. Evaluamos la variación geográfica en el canto a través del rango de esta especie y realizamos un experimento de playback usando ejemplares prevenientes de dos regiones acústicamente divergentes y geográficamente distantes como tratamientos. Encontramos variación geográfica significativa en el canto, que se correspondió bien con las regiones desérticas ecológicamente distintas. Encontramos que A. flaviceps tuvo respuestas vocales más fuertes a los cantos que suenan como locales, pero tuvo las mismas respuestas de movimiento a los cantos que suenan como locales o extranjeros. Estos resultados son similares a los encontrados en otras especies sin congéneres simpátricos y brindan un ejemplo de una especie que indaga los cantos de individuos conespecíficos acústicamente divergentes, a pesar de reconocer diferencias marcadas en esos cantos. Lay Summary Many songbirds have geographic variation in their songs. Most species prefer songs that sound local, or familiar, but some will respond to foreign songs just as readily as they respond to local songs. This may be more likely if there are no similar-sounding relatives around, and therefore no risk of mistaken identity. Few species without close relatives have been studied, so we study one that is also a desert specialist, the Verdin (Auriparus flaviceps). We found that Verdins have geographic variation in song that mapped well onto large desert habitat regions, and they sing less to playbacks of songs from distant regions. However, despite recognizing salient differences in foreign songs, they are just as likely to approach foreign songs as local-sounding songs, as are other species that live without close relatives. When close relatives are absent, birds may be more permissive in what they consider a song worth investigating. INTRODUCTION Bird song is an important signal for mate and competitor recognition (Marler 1958), and understanding how song and recognition evolves has captured the attention of biologists for over a century (Allen 1919). Many species show a pattern of stronger response to acoustically similar song phenotypes and weaker response to dissimilar phenotypes (Podos and Warren 2007, Wilkins et al. 2013, Freeman and Montgomery 2017, Parker et al. 2018, Weir and Price 2019). This is reasonable because recognition of conspecifics is important for mate acquisition and resource defense, but responses to unimportant signals, such as signals from congeners, could be a waste of time and energy (Wiley 1994, Amézquita et al. 2011, Symes 2014, Hamao 2016). Even within species, more geographically proximate populations tend to sound more similar (e.g., may have the same dialect) and tend to receive stronger responses in territorial playbacks than more geographically distant populations (Podos and Warren 2007, Parker et al. 2018). Stronger territorial responses to local-sounding songs may be useful for many reasons, for example, if local individuals pose the greatest threat (Turcokova et al. 2011), if local individuals make the best mates (Rothstein and Fleischer 1987), or if local-sounding individuals are the highest quality (Searcy et al. 2002). Having a preference for the familiar need not even be adaptive and may simply be a byproduct of individual experience (Irwin and Price 1999). Even though many species show stronger responses to local-sounding songs, there are exceptions that highlight how response may depend upon circumstance. For example, Dickcissels (Spiza americana) respond equally to conspecific local and foreign songs, either because they tend to encounter foreign individuals in territorial interactions or because, as a monotypic genus, they face no selection to avoid similar-sounding congeners (Parra et al. 2017). Populations of Japanese Tits (Parus minor) and Varied Tits (Sittiparus varius) that live in isolation from other members of the family Paridae respond equally to local and foreign conspecific songs, presumably because they are released from selection to avoid competitive interactions with heterospecifics (Hamao 2016). Work in frogs, insects, and birds supports the idea that the likelihood of encountering phenotypically similar heterospecifics (e.g., congeners or confamilials) may influence responses to divergent conspecifics (Grether et al. 2009, Amézquita et al. 2011, Symes 2014, Parker et al. 2018). However, in birds, studies examining conspecific recognition in species without sympatric confamilials are relatively uncommon (Parker et al. 2018). This discrepancy is expected given that polytypic genera are more common than monotypic genera (Maruvka et al. 2013), but monotypic genera offer an opportunity to see how conspecific recognition unfolds when there is no pressure to avoid costly heterospecific interactions. We add to the literature on species without sympatric confamilials by assessing geographic variation in song and response to divergent conspecific signals in Verdins (Auriparus flaviceps), a species that is in a monotypic genus and is the only member of its family (Remizidae) in the Western Hemisphere. Verdins are an interesting study species not only because of their isolated evolutionary history, but also because of their unusual ecology as a non-migratory desert specialist (Webster 2020). Their range encompasses all 3 hot desert habitats in North America (Sonoran, Chihuahuan, and Mohave), and subspecies have been described based on subtle plumage variations, but no one has yet described geographic variation in their song (Webster 2020). We leveraged the recordings of Verdins from 2 large public audio collections to describe geographic variation in song, and we conducted a playback experiment with treatments from 2 acoustically divergent and geographically distant geographic regions to assess response to divergent conspecific songs in a central focal population. Finally, we assessed how Verdin responses compare to those of other species by qualitatively comparing the effect sizes from this study to effect sizes from a meta-analysis of 45 species with and without congeners (Parker et al. 2018). METHODS Study Species and System Verdins are small insectivorous passerines that live in desert habitat throughout the Southwestern United States and Mexico. They are non-migratory, though study populations have a relatively large turnover of individuals between breeding seasons, potentially because of high dispersal and/or low survival (Taylor 1971, Webster 2020, E. I. Greig personal observation). They breed and roost in ball-shaped nests suspended from the tips of vegetation such as paloverde (Parkinsonia florida) branches (Taylor 1971, McCreedy and van Riper 2015). Verdins have a variety of vocalizations that are involved in intra-pair communication, mate attraction, and territory defense (Taylor 1971, Webster 2020). The “whistle” song, which is the focus of this study, is described as what is considered a typical song in oscines and is given often by males and sometimes by females (Taylor 1971). Although the function of the whistle song is uncertain, it is likely associated with territory maintenance or nesting behavior because it is used primarily during the breeding season and in the dawn chorus (Taylor 1971, E. I. Greig personal observation). We choose the whistle song as the focus for this study because it is the vocalization most analogous to what is usually considered a “song” in birds, based on context and acoustic structure (i.e., it is relatively loud and elaborate compared to most other Verdin vocalizations). Whistle songs are typically composed of 2–4 notes that are similar in structure (Figure 1). Although repertoire size has not been specifically quantified for Verdins, song variation within individuals seems minimal (E. I. Greig personal observation). An individual might sing a 2-note song and a 3-note song, but the structure of the notes within those songs appears consistent over time (e.g., male #119 in Figure 1). Figure 1. Open in new tabDownload slide Spectrograms of songs from the focal population (top row) and songs used for the playback experiment (lower 3 rows) showing representative geographic variation between regions. Dotted lines separate songs. Letters indicate playback type (S = Sonoran, B = Baja, C = Chihuahuan, OW = Old World confamilial), and numbers indicate playback exemplar ID (lower 3 rows) or individual ID (top row). Figure 1. Open in new tabDownload slide Spectrograms of songs from the focal population (top row) and songs used for the playback experiment (lower 3 rows) showing representative geographic variation between regions. Dotted lines separate songs. Letters indicate playback type (S = Sonoran, B = Baja, C = Chihuahuan, OW = Old World confamilial), and numbers indicate playback exemplar ID (lower 3 rows) or individual ID (top row). We studied a population of Verdins at Organ Pipe Cactus National Monument (ORPI) in Pima County, Arizona (31.942851°N, –112.813883°W, WGS 84). We located and color-banded nesting birds on 30 territories during the breeding seasons of March–April 2016 and 2017 (n = 18 territories in 2016; 14 with both members of the pair banded, 1 with the female only banded, and 3 with neither member of the pair banded and n = 12 territories in 2017; 7 with both members of the pair banded, 1 with the female only banded, and 4 with banded females, but in which the males were raising nestlings at other nests). Although we were not able to band every individual, we were sure that at least the female member of the pair was different on every territory because territories had active nests. We included the 4 territories in 2017 that had females raising nestlings alone because females engage in territory defense (E. I. Greig personal observation), so there was no a priori reason to exclude them. Acoustic Analysis To quantify the range of geographic variation of Verdin whistle songs and the acoustic divergence of playbacks relative to songs from the focal population, we conducted an acoustic analysis on a collection of songs from the focal population (n = 154 songs from 21 individuals recorded at ORPI in March 2015 and April 2019) and all available Verdin songs from the Macaulay Library (Cornell Lab of Ornithology, Ithaca, New York, n = 222 songs from 48 individuals), Xeno-canto (www.xeno-canto.org, n = 267 songs from 52 individuals) and colleagues (Lazarus Pomara, n = 12 songs from 3 individuals). Details of all recordings, including accession numbers, are in Supplementary Material Recording Metadata. We excluded songs that were a single note because they were not representative of typical whistle songs (n = 21 out of an original 676 identified songs). The final collection of 655 local and non-local whistle songs included the conspecific exemplars used for our playback experiment (n = 17 songs from 8 individuals; Figure 1). The collection of recordings spanned a variety of geographic regions, and we classified all songs as belonging to 1 of 4 geographic regions informed by the Bird Conservation Regions (BCR) (Bird Studies Canada and NABCI 2014); Sonoran Desert (BCR region 33, associated with the acaciarum subspecies, n = 232 songs plus 154 songs from the focal population), Sierra Madre Occidental (BCR region 34, a transition region between acaciarum and ornatus subspecies, n = 83 songs), Chihuahuan Desert (BCR regions 35 and 36, associated with the ornatus subspecies, n = 57 songs), and Baja California Sur (BCR regions 40 and 42, associated with the lamprocephalus subspecies, n = 129 songs). We grouped songs from the adjacent BCR regions 35/36 and 40/42 because recordings were sparse from each individual region. We measured the frequency contour of each note with an equal number of points (acoustic landmarks), which allowed us to assess note shape as well as calculate standard acoustic features such as length and high/low frequency (Taft 2011). This method is ideal for simple, tonal sounds such as Verdin whistle song notes because they can be represented well by a single frequency contour (Figure 2). Considering only the fundamental frequency contour of each note also minimizes any impact of recording quality on song variation, and when we tested for an effect of recording source (Macaulay Library, Xeno-Canto and personal collection) on song structure we did not find an effect. We used spectrograms digitized in Raven 1.5 (Bioacoustics Research Program 2004) with 16-bit sample format, discrete Fourier transform (DFT) = 512 samples, frequency resolution = 124 Hz, time resolution = 11.6 ms, frame overlap = 50%. We manually selected each note (n = 1,883 notes) to ensure we were measuring the target vocalizations, rather than background noise or other vocalizing birds. We described every note using landmarks that measured the approximate peak frequency of the note at 24 points in time along the note. To create these landmarks in Raven, we used the Split All Selection Borders function to create 24 “selections” for each note, and we used the Center Frequency measurement for each selection to extract frequency values at which the energy within the selection was equally divided above and below the value (this approximates the peak frequency contour of the note; Figure 2). We used the Center Time measurement for each selection to extract 24 time values along each note that we then scaled from the beginning of the note. Figure 2. Open in new tabDownload slide Example of a song note with acoustic landmarks made using the Split All Selection Borders function in Raven. Points represent the locations within each section where the Center Time measurement intersects with the Center Frequency measurement. This example shows 10 landmarks for clarity, but 24 were taken on each note for analysis. Figure 2. Open in new tabDownload slide Example of a song note with acoustic landmarks made using the Split All Selection Borders function in Raven. Points represent the locations within each section where the Center Time measurement intersects with the Center Frequency measurement. This example shows 10 landmarks for clarity, but 24 were taken on each note for analysis. We used R v. 3.3.3 (R Core Team 2017) for all calculations with acoustic landmarks. We used the landmarks to characterize song length, high frequency, and low frequency. We also used the landmarks to characterize note shape by running a principal component analysis on the centered and scaled landmarks for all 1,883 notes using the princomp function. The first 3 principal components from this analysis described 89.9% of the variation in notes. Because Verdin whistle songs are a series of repeated, similar notes, for every song we calculated the mean of the first 3 principal component scores across notes. The resulting mean scores (note shape pc1, pc2, and pc3) became our song-level metrics of note shape. To characterize song divergence relative to the focal population, we ran a second principal component analysis on the 6 song-level measurements (song length, song high frequency, song low frequency, note shape pc1, pc2, and pc3; Table 1) for all 655 songs. The first 3 principal components from this analysis described 80% of the variation in songs (Table 2). We used these 3 principal components to calculate a measure of acoustic divergence. We define acoustic divergence as the Euclidean distance in 3-dimensional acoustic space between every non-focal population song (n = 501 songs) and the point representing the mean song-level principal component scores of the focal population (based on n = 154 songs from ORPI). We calculated the geographic distance between every non-focal population song and the focal population using the rdist.earth function implemented in the package fields (Nychka et al. 2015). All acoustic data are available in Supplementary Material Acoustic Data and Greig et al. (2021). Table 1. Mean values for acoustic characters of songs from the focal population and each geographical region. Acoustic divergence is the Euclidean 3-dimensional distance between each song in the first 3 song-level principal component scores relative to the mean scores of the focal population. . Focal . Sonoran . Sierra Madre . Baja . Chihuahuan . . (n = 154) . (n = 232) . (n = 83) . (n = 129) . (n = 57) . Song length (s) 0.66 0.70 0.61 0.53 0.75 High frequency (Hz) 5,579 5,403 5,165 6,478 4,558 Low frequency (Hz) 3,590 3,610 3,911 3,454 3,520 Note shape pc1 –4.49 –0.67 3.37 2.90 0.61 Note shape pc2 –1.77 0.28 0.59 –0.82 3.27 Note shape pc3 0.19 –0.12 –0.56 2.18 –3.49 Acoustic divergence 1.48 1.85 2.94 2.89 3.48 . Focal . Sonoran . Sierra Madre . Baja . Chihuahuan . . (n = 154) . (n = 232) . (n = 83) . (n = 129) . (n = 57) . Song length (s) 0.66 0.70 0.61 0.53 0.75 High frequency (Hz) 5,579 5,403 5,165 6,478 4,558 Low frequency (Hz) 3,590 3,610 3,911 3,454 3,520 Note shape pc1 –4.49 –0.67 3.37 2.90 0.61 Note shape pc2 –1.77 0.28 0.59 –0.82 3.27 Note shape pc3 0.19 –0.12 –0.56 2.18 –3.49 Acoustic divergence 1.48 1.85 2.94 2.89 3.48 Open in new tab Table 1. Mean values for acoustic characters of songs from the focal population and each geographical region. Acoustic divergence is the Euclidean 3-dimensional distance between each song in the first 3 song-level principal component scores relative to the mean scores of the focal population. . Focal . Sonoran . Sierra Madre . Baja . Chihuahuan . . (n = 154) . (n = 232) . (n = 83) . (n = 129) . (n = 57) . Song length (s) 0.66 0.70 0.61 0.53 0.75 High frequency (Hz) 5,579 5,403 5,165 6,478 4,558 Low frequency (Hz) 3,590 3,610 3,911 3,454 3,520 Note shape pc1 –4.49 –0.67 3.37 2.90 0.61 Note shape pc2 –1.77 0.28 0.59 –0.82 3.27 Note shape pc3 0.19 –0.12 –0.56 2.18 –3.49 Acoustic divergence 1.48 1.85 2.94 2.89 3.48 . Focal . Sonoran . Sierra Madre . Baja . Chihuahuan . . (n = 154) . (n = 232) . (n = 83) . (n = 129) . (n = 57) . Song length (s) 0.66 0.70 0.61 0.53 0.75 High frequency (Hz) 5,579 5,403 5,165 6,478 4,558 Low frequency (Hz) 3,590 3,610 3,911 3,454 3,520 Note shape pc1 –4.49 –0.67 3.37 2.90 0.61 Note shape pc2 –1.77 0.28 0.59 –0.82 3.27 Note shape pc3 0.19 –0.12 –0.56 2.18 –3.49 Acoustic divergence 1.48 1.85 2.94 2.89 3.48 Open in new tab Table 2. Correlations of song variables with the first 3 principal components of the song-level principal component analysis. . pc1 . pc2 . pc3 . Eigenvalue 1.46 1.16 1.16 Cumulative variation 0.36 0.58 0.80 Song length (s) 0.28 –0.34 0.35 High frequency (Hz) –0.64 0.00 0.08 Low frequency (Hz) 0.21 0.73 0.24 Note shape pc1 –0.08 0.48 –0.59 Note shape pc2 0.26 –0.32 –0.69 Note shape pc3 –0.63 –0.10 –0.05 . pc1 . pc2 . pc3 . Eigenvalue 1.46 1.16 1.16 Cumulative variation 0.36 0.58 0.80 Song length (s) 0.28 –0.34 0.35 High frequency (Hz) –0.64 0.00 0.08 Low frequency (Hz) 0.21 0.73 0.24 Note shape pc1 –0.08 0.48 –0.59 Note shape pc2 0.26 –0.32 –0.69 Note shape pc3 –0.63 –0.10 –0.05 Open in new tab Table 2. Correlations of song variables with the first 3 principal components of the song-level principal component analysis. . pc1 . pc2 . pc3 . Eigenvalue 1.46 1.16 1.16 Cumulative variation 0.36 0.58 0.80 Song length (s) 0.28 –0.34 0.35 High frequency (Hz) –0.64 0.00 0.08 Low frequency (Hz) 0.21 0.73 0.24 Note shape pc1 –0.08 0.48 –0.59 Note shape pc2 0.26 –0.32 –0.69 Note shape pc3 –0.63 –0.10 –0.05 . pc1 . pc2 . pc3 . Eigenvalue 1.46 1.16 1.16 Cumulative variation 0.36 0.58 0.80 Song length (s) 0.28 –0.34 0.35 High frequency (Hz) –0.64 0.00 0.08 Low frequency (Hz) 0.21 0.73 0.24 Note shape pc1 –0.08 0.48 –0.59 Note shape pc2 0.26 –0.32 –0.69 Note shape pc3 –0.63 –0.10 –0.05 Open in new tab Playback Experiment To assess the response of birds from the focal population to geographically and acoustically distant conspecific songs, we conducted a playback experiment from March 26 to April 3, 2016 and from March 27 to March 31, 2017. Focal pairs were nesting and generally responsive to playbacks at that time in the season. We used an experimental design in which each focal territory received each of 3 playback types: (1) a positive control song recorded from the Sonoran region within the range of 160–480 km from the study site and from the same subspecies as the focal population (acaciarum, n = 9 exemplars; Figure 1 and 3A); (2) a treatment song recorded from the Chihuahuan region or Baja region, 960–1,220 km from the playback site and from a different subspecies as the focal population (ornatus or lamprocephalus, n = 8 exemplars; Figure 1 and 3A); and (3) a negative control heterospecific song recorded from an Old World confamilial Remizid (Remiz pendulinus, R. consobrinus, Anthoscopus caroli, A. minutus or A. musculus, n = 9 exemplars; Figure 1). We chose exemplar recordings from the Macaulay Library and Xeno-canto based on the location of the recording and the clarity of the spectrograms (details of exemplar recordings in Supplementary Material Recording Metadata). We standardized the recordings in Raven to be the same sample rate, bit-depth, and amplitude for playback (16-bit and 44 kHz). We used an amplified speaker (Pignose Legendary 7-100, Pignose-Gorilla, Las Vegas, Nevada, USA; frequency response: 0.1–12.0 kHz) for playback. We tested the amplitude of playbacks in the field using a sound-pressure level meter (model number 33-2050, Radio Shack Corporation, Fort Worth, Texas, USA), set at C-weighting, fast response (~89.0 dB at 1 m for all playbacks). This amplitude was similar to natural Verdin song levels. Figure 3. Open in new tabDownload slide (A) Map of geographic distribution of song recordings (small dots) and playback exemplars (large labeled dots) overlayed on the range of Verdin (gray) and the Bird Conservation Regions (black lines) and colored by our geographic region classification (yellow = Sonoran, orange = Sierra Madre, blue = Baja, pink = Chihuahuan). Black dot shows the focal population site. (B) Song-level principal component scores 1 and 2 for every song recording and playback exemplar colored by geographic region. Black dots are recordings from the focal population. Sierra Madre songs are not labeled with larger dots because they were not used in the playback experiment. Figure 3. Open in new tabDownload slide (A) Map of geographic distribution of song recordings (small dots) and playback exemplars (large labeled dots) overlayed on the range of Verdin (gray) and the Bird Conservation Regions (black lines) and colored by our geographic region classification (yellow = Sonoran, orange = Sierra Madre, blue = Baja, pink = Chihuahuan). Black dot shows the focal population site. (B) Song-level principal component scores 1 and 2 for every song recording and playback exemplar colored by geographic region. Black dots are recordings from the focal population. Sierra Madre songs are not labeled with larger dots because they were not used in the playback experiment. We presented all 3 playback types to every focal territory (n = 30 territories) so that we could account for between-individual variation in response. We presented playbacks to focal territories on separate days and we balanced the order of playback presentations so that each playback type (i.e., positive control, treatment, and negative control) was equally represented in every order position (i.e., first, second, or third). A trial consisted of 1 or 2 observers arriving at a focal territory, placing the playback speaker at the base of a bush ~15–30 m from the focal birds’ nest, concealing themselves behind nearby vegetation ~20 m from the speaker, and turning on the playback. The playback consisted of a single exemplar song repeated every 10 s for 5 min, during which the territory was monitored. We chose a 5-min playback period because we began each playback irrespective of the location of the focal birds, and it often took several minutes for focal birds to notice the playback (mean latency to approach positive control songs was 113 s). This means there was some probability of false-negative responses, in which birds did not respond because they were not within the range of the speaker during the playback, but this was equally possible across playback types and thus is accounted for in our experimental design. Additionally, there was some probability of false-positive responses because some birds may have come to the area of their nests (which was near the speaker) irrespective of the playback being presented, but again this was equally possible across playback types and thus is accounted for in our experimental design. The observers recorded all focal bird vocal behavior and dictated all movement behavior possible using a digital recorder (Marantz PMD 661 solid-state digital recorder at 96 kHz sampling rate, 24-bit depth, D&M Professional, Itasca, Illinois, USA combined with ME66 shotgun microphone capsules and K6 power modules, Sennheiser Electronic Corporation, Old Lyme, Connecticut, USA; frequency response: 0.04–20.0 kHz). From these dictated recordings of the trials, we later transcribed the following response variables: (1) whether or not at least 1 member of the pair approached within 20 m of the speaker, (2) the latency of the first member of the pair approaching within 20 m of the speaker, (3) the total time that at least 1 member of the pair spent within 20 m of the speaker, (4) whether or not at least 1 member of the pair sang a whistle song, (5) the latency of the first member of the pair to sing a whistle song, and (6) the total number of whistle songs given by both members of the pair. We continued to observe territories for 5 min after the playback had stopped, but for our analysis, we only use responses that began within the 5-min playback period. We combined male and female responses in our analysis for 2 reasons. First, male and female Verdins have similar enough plumage that seeing color bands on birds sexed in the hand is the only way to be certain of the sex of individuals in the field. Observers were not always able to identify color bands because focal birds would sometimes sing from locations out of sight (n = 31 sing instances) or approach for a period of time too fleeting to see their bands (n = 15 approach instances). By considering responses blind to the sex of the focal birds we could treat all trials equally, despite variation in observer success identifying color bands and variation in whether or not a focal pair had been banded. We have high confidence that responses were from the focal pair rather than from intruding birds because we never observed more than 2 birds respond simultaneously and all birds that we were able to identify by their bands were focal pair members. Second, because every focal territory received all 3 trial types, any between-territory variation in male/female responsiveness (e.g., focal territories where only a female was present or focal territories at different stages of the nesting cycle) was accounted for in mixed models. All playback data are available in Supplementary Material Playback Data and Greig et al. (2021). Statistical Analysis We used R v. 3.3.3. for all statistical analyses. To assess how well the 6 song-level measurements categorized all songs (n = 655) as originating from 1 of the 4 different geographical regions, we used a linear discriminant function analysis implemented in the package MASS (Venables and Ripley 2002). We standardized all song measurements using the scale function before analysis. To assess the importance of geographic region and geographic distance from the focal population on acoustic divergence (Euclidean distance) for the 501 non-focal population songs, we used a model-selection approach with linear mixed models implemented with the lmer function in the package lme4 (Bates et al. 2015). We used the Akaike information criterion of ΔAIC < 2 indicating equivalent models. We excluded the 154 songs from the focal population from this analysis because their geographic distance was, by definition, zero for every song. We included bird ID as a random effect. We initially included recording sources (Macaulay Library, Xeno-canto or personal collection) as a fixed effect, but it was not a significant predictor. Therefore, we removed it from final models for simplicity. To determine if the playback exemplars from different geographic regions were significantly different from one another in song-level principal component scores and acoustic divergence from the focal population, we used non-parametric pairwise Wilcoxon rank-sum tests implemented with the pairwise.wilcox.test function. To assess differences in responses to the playbacks, we used mixed models implemented in the package lme4. We included focal pair ID, playback ID (because we presented playbacks more than once), and recording ID (because we used different songs from the same recording for some playbacks) as random effects in all models. We initially included playback order and year as fixed effects in models, but they were not significant. Therefore, we removed them from the final models for simplicity. We assessed the probability of approaching and singing as binomial response variables for all playback trials (n = 90 trials). For these binary responses, we used generalized linear mixed logistic regression models, using binomial distributions and logit-link functions. For the subset of trials in which a bird approached (n = 48 trials), we assessed the strength of approach behavior based on latency to approach within 20 m and time spent within 20 m of the speaker. For the subset of trials in which a bird sang (n = 52 trials), we assessed the strength of singing behavior based on latency to first song and number of songs. For models assessing strength of response, we used linear mixed models. We evaluated effect sizes using Cohen’s d and 95% confidence intervals for the strength of response variables following Nakagawa and Cuthill (2007). To put our playback results for Verdin in the context of previous studies on other species, we calculated Fisher’s transformation of the correlation coefficients (Zr) for all response variables following Parker et al. (2018). We binned Verdin response variables into movement or vocal response types and calculated a mean Zr value for each type so we could compare Verdin Zr values to the equivalent values calculated for 45 species in Parker et al. (2018), using the data deposited in Open Science Framework doi:10.17605/OSF.IO/W2MVP. We did not conduct statistical analyses on Zr values because a formal analysis of this dataset is the focus of Parker et al. (2018) and is beyond the scope of this paper. In their analysis, Parker et al. (2018) found that movement responses tended to have larger effect sizes than vocal responses, and that species living without confamilials tended to have smaller effect sizes than species living with congeners or confamilials. This provided us with a framework for qualitative comparison of Verdin Zr values in relation to the Zr values of the 45 species included in Parker et al. (2018). RESULTS Acoustic Analysis Songs were distinguishable as originating from 1 of the 4 geographical regions with 86% accuracy based on the linear discriminant function analysis with jackknifed validation (Table 3). The likelihood of being correctly assigned was high (~90%) for Sonoran, Chihuahuan, and Baja songs, but lower for Sierra Madre songs (54% classified correctly). Sierra Madre songs were most often misclassified as Sonoran songs (76% of misclassifications). Of the 17 songs used for playbacks, 70% were classified correctly: 5/5 Chihuahuan playbacks, 6/9 Sonoran playbacks, and 1/3 Baja playbacks. The 2 misclassified Baja playbacks were classified as Sonoran and the 3 misclassified Sonoran playbacks were classified as Sierra Madre (1) or Baja (2). These misclassifications are consistent with the overlap in song characteristics evident in Figure 3B. Table 3. Coefficients of linear discriminants from the discriminant function analysis. . ld1 . ld2 . ld3 . Proportion of trace 0.68 0.18 0.14 Song length (s) –0.22 0.15 –0.06 High frequency (Hz) 0.49 –0.41 –1.16 Low frequency (Hz) –0.11 0.26 0.96 Note shape pc1 0.51 –1.18 0.23 Note shape pc2 –0.39 –0.20 –0.14 Note shape pc3 1.10 0.77 1.18 . ld1 . ld2 . ld3 . Proportion of trace 0.68 0.18 0.14 Song length (s) –0.22 0.15 –0.06 High frequency (Hz) 0.49 –0.41 –1.16 Low frequency (Hz) –0.11 0.26 0.96 Note shape pc1 0.51 –1.18 0.23 Note shape pc2 –0.39 –0.20 –0.14 Note shape pc3 1.10 0.77 1.18 Open in new tab Table 3. Coefficients of linear discriminants from the discriminant function analysis. . ld1 . ld2 . ld3 . Proportion of trace 0.68 0.18 0.14 Song length (s) –0.22 0.15 –0.06 High frequency (Hz) 0.49 –0.41 –1.16 Low frequency (Hz) –0.11 0.26 0.96 Note shape pc1 0.51 –1.18 0.23 Note shape pc2 –0.39 –0.20 –0.14 Note shape pc3 1.10 0.77 1.18 . ld1 . ld2 . ld3 . Proportion of trace 0.68 0.18 0.14 Song length (s) –0.22 0.15 –0.06 High frequency (Hz) 0.49 –0.41 –1.16 Low frequency (Hz) –0.11 0.26 0.96 Note shape pc1 0.51 –1.18 0.23 Note shape pc2 –0.39 –0.20 –0.14 Note shape pc3 1.10 0.77 1.18 Open in new tab Including geographic region as a fixed effect significantly improved the fit of linear mixed models explaining acoustic divergence (Euclidean distance) from the focal population (full model df = 7, AIC = 770.9 vs. model without geographic region df = 4, AIC = 778.2, χ 2 = 14.10, P = 0.003) but including geographic distance from focal population did not (full model df = 7, AIC = 770.9 vs. model without geographic distance df = 6, AIC = 769.8, χ 2 = 1.72, P = 0.189). After correcting for multiple comparisons using the Benjamini-Hochberg adjustment, Baja (n = 3) and Chihuahuan (n = 5) playbacks were significantly different from Sonoran (n = 9) playbacks in at least 1 principal component score (song pc1: Sonoran mean = 0.082, Chihuahuan mean = 2.85, Wilcoxon P = 0.003; song pc3: Sonoran mean = 0.483, Baja mean = –0.854, Wilcoxon P = 0.027). The acoustic divergence (Euclidean distance) of Baja playbacks was not significantly different from the acoustic divergence of Sonoran playbacks (Sonoran mean = 1.86, Baja mean = 2.37, Wilcoxon P = 0.209), but the acoustic divergence of Chihuahuan playbacks was significantly greater than the acoustic divergence of Sonoran playbacks (Chihuahuan mean = 3.50, Wilcoxon P = 0.036). Playback Analysis Focal birds were significantly more likely to approach Sonoran, Baja, and Chihuahuan playbacks than confamilial negative control playbacks and showed no difference in approach tendency between Sonoran, Baja, and Chihuahuan playbacks (Tables 4 and 5, Figure 4). We did not find differences between the Sonoran positive controls and confamilial negative controls and therefore did not have sufficient power to detect differences between treatments in the probability of singing (Tables 4 and 5, Figure 4), nor in 3 of the variables representing the strength of response (time near speaker, latency to approach, and latency to sing; Tables 4 and 5, Figure 5). Verdins did, however, sing significantly more songs during Sonoran playbacks than Chihuahuan and confamilial negative control playbacks (Tables 4 and 5, Figure 5) and had a non-significant tendency to sing more songs to Sonoran playbacks than to Baja playbacks (Tables 4 and 5, Figure 5). Table 4. Mean values for behavioral responses to each playback type. . Sonoran . Baja . Chihuahuan . Confamilial . (n = 30) (n = 12) (n = 18) (n = 30) Approach (binomial)a 0.70 0.58 0.61 0.30 Sing (binomial)a 0.67 0.58 0.56 0.50 (n = 21) (n = 7) (n = 11) (n = 9) Latency to 20 m (s)b 113 77 86 64 Time spent 20 m (s)b 221 232 153 228 (n = 20) (n = 7) (n = 10) (n = 15) Latency to sing (s)c 81 145 102 141 Number of songsc 16 9 7 9 . Sonoran . Baja . Chihuahuan . Confamilial . (n = 30) (n = 12) (n = 18) (n = 30) Approach (binomial)a 0.70 0.58 0.61 0.30 Sing (binomial)a 0.67 0.58 0.56 0.50 (n = 21) (n = 7) (n = 11) (n = 9) Latency to 20 m (s)b 113 77 86 64 Time spent 20 m (s)b 221 232 153 228 (n = 20) (n = 7) (n = 10) (n = 15) Latency to sing (s)c 81 145 102 141 Number of songsc 16 9 7 9 aMeans based on all trials. bMeans based on the subset of trials in which a bird approached. cMeans based on the subset of trials in which a bird sang. Open in new tab Table 4. Mean values for behavioral responses to each playback type. . Sonoran . Baja . Chihuahuan . Confamilial . (n = 30) (n = 12) (n = 18) (n = 30) Approach (binomial)a 0.70 0.58 0.61 0.30 Sing (binomial)a 0.67 0.58 0.56 0.50 (n = 21) (n = 7) (n = 11) (n = 9) Latency to 20 m (s)b 113 77 86 64 Time spent 20 m (s)b 221 232 153 228 (n = 20) (n = 7) (n = 10) (n = 15) Latency to sing (s)c 81 145 102 141 Number of songsc 16 9 7 9 . Sonoran . Baja . Chihuahuan . Confamilial . (n = 30) (n = 12) (n = 18) (n = 30) Approach (binomial)a 0.70 0.58 0.61 0.30 Sing (binomial)a 0.67 0.58 0.56 0.50 (n = 21) (n = 7) (n = 11) (n = 9) Latency to 20 m (s)b 113 77 86 64 Time spent 20 m (s)b 221 232 153 228 (n = 20) (n = 7) (n = 10) (n = 15) Latency to sing (s)c 81 145 102 141 Number of songsc 16 9 7 9 aMeans based on all trials. bMeans based on the subset of trials in which a bird approached. cMeans based on the subset of trials in which a bird sang. Open in new tab Table 5. Results of mixed models relating focal bird responses to the fixed effect of playback type. Focal pair ID, playback ID, and recording ID were incorporated into all models as random effects. P-values are for comparisons of the fixed effect levels to Sonoran playbacks (the positive control). Confamilial playbacks represent a heterospecific negative control. P-values ≤ 0.05 are highlighted in bold. . Fixed effect . Estimate . z/t . P . Approach (binomial)a Intercept 0.98 2.07 0.039 Baja –0.68 –0.84 0.399 Chihuahua –0.40 –0.59 0.559 Confamilial –1.97 –2.93 0.003 Sing (binomial)a Intercept 0.83 1.70 0.089 Baja –0.34 –0.42 0.678 Chihuahua –0.62 –0.86 0.388 Confamilial –0.85 –1.33 0.184 Latency to 20 m (s)b Intercept 113.09 6.85 0.000 Baja –32.77 –1.08 0.280 Chihuahua –24.84 –0.98 0.329 Confamilial –43.56 –1.57 0.117 Time spent 20 m (s)b Intercept 222.74 4.91 0.000 Baja 9.69 0.11 0.916 Chihuahua –69.33 –0.89 0.371 Confamilial 4.77 0.06 0.953 Latency to sing (s)c Intercept 83.35 3.84 0.000 Baja 59.87 1.48 0.140 Chihuahua 24.08 0.68 0.500 Confamilial 59.17 1.89 0.059 Number of songsc Intercept 16.35 7.62 0.000 Baja –6.92 –1.64 0.101 Chihuahua –9.85 –2.65 0.008 Confamilial –7.22 –2.20 0.028 . Fixed effect . Estimate . z/t . P . Approach (binomial)a Intercept 0.98 2.07 0.039 Baja –0.68 –0.84 0.399 Chihuahua –0.40 –0.59 0.559 Confamilial –1.97 –2.93 0.003 Sing (binomial)a Intercept 0.83 1.70 0.089 Baja –0.34 –0.42 0.678 Chihuahua –0.62 –0.86 0.388 Confamilial –0.85 –1.33 0.184 Latency to 20 m (s)b Intercept 113.09 6.85 0.000 Baja –32.77 –1.08 0.280 Chihuahua –24.84 –0.98 0.329 Confamilial –43.56 –1.57 0.117 Time spent 20 m (s)b Intercept 222.74 4.91 0.000 Baja 9.69 0.11 0.916 Chihuahua –69.33 –0.89 0.371 Confamilial 4.77 0.06 0.953 Latency to sing (s)c Intercept 83.35 3.84 0.000 Baja 59.87 1.48 0.140 Chihuahua 24.08 0.68 0.500 Confamilial 59.17 1.89 0.059 Number of songsc Intercept 16.35 7.62 0.000 Baja –6.92 –1.64 0.101 Chihuahua –9.85 –2.65 0.008 Confamilial –7.22 –2.20 0.028 aModels based on all trials (n = 90). bModels based on the subset of trials in which a bird approached (n = 48). cModels based on the subset of trials in which a bird sang (n = 52). Open in new tab Table 5. Results of mixed models relating focal bird responses to the fixed effect of playback type. Focal pair ID, playback ID, and recording ID were incorporated into all models as random effects. P-values are for comparisons of the fixed effect levels to Sonoran playbacks (the positive control). Confamilial playbacks represent a heterospecific negative control. P-values ≤ 0.05 are highlighted in bold. . Fixed effect . Estimate . z/t . P . Approach (binomial)a Intercept 0.98 2.07 0.039 Baja –0.68 –0.84 0.399 Chihuahua –0.40 –0.59 0.559 Confamilial –1.97 –2.93 0.003 Sing (binomial)a Intercept 0.83 1.70 0.089 Baja –0.34 –0.42 0.678 Chihuahua –0.62 –0.86 0.388 Confamilial –0.85 –1.33 0.184 Latency to 20 m (s)b Intercept 113.09 6.85 0.000 Baja –32.77 –1.08 0.280 Chihuahua –24.84 –0.98 0.329 Confamilial –43.56 –1.57 0.117 Time spent 20 m (s)b Intercept 222.74 4.91 0.000 Baja 9.69 0.11 0.916 Chihuahua –69.33 –0.89 0.371 Confamilial 4.77 0.06 0.953 Latency to sing (s)c Intercept 83.35 3.84 0.000 Baja 59.87 1.48 0.140 Chihuahua 24.08 0.68 0.500 Confamilial 59.17 1.89 0.059 Number of songsc Intercept 16.35 7.62 0.000 Baja –6.92 –1.64 0.101 Chihuahua –9.85 –2.65 0.008 Confamilial –7.22 –2.20 0.028 . Fixed effect . Estimate . z/t . P . Approach (binomial)a Intercept 0.98 2.07 0.039 Baja –0.68 –0.84 0.399 Chihuahua –0.40 –0.59 0.559 Confamilial –1.97 –2.93 0.003 Sing (binomial)a Intercept 0.83 1.70 0.089 Baja –0.34 –0.42 0.678 Chihuahua –0.62 –0.86 0.388 Confamilial –0.85 –1.33 0.184 Latency to 20 m (s)b Intercept 113.09 6.85 0.000 Baja –32.77 –1.08 0.280 Chihuahua –24.84 –0.98 0.329 Confamilial –43.56 –1.57 0.117 Time spent 20 m (s)b Intercept 222.74 4.91 0.000 Baja 9.69 0.11 0.916 Chihuahua –69.33 –0.89 0.371 Confamilial 4.77 0.06 0.953 Latency to sing (s)c Intercept 83.35 3.84 0.000 Baja 59.87 1.48 0.140 Chihuahua 24.08 0.68 0.500 Confamilial 59.17 1.89 0.059 Number of songsc Intercept 16.35 7.62 0.000 Baja –6.92 –1.64 0.101 Chihuahua –9.85 –2.65 0.008 Confamilial –7.22 –2.20 0.028 aModels based on all trials (n = 90). bModels based on the subset of trials in which a bird approached (n = 48). cModels based on the subset of trials in which a bird sang (n = 52). Open in new tab Figure 4. Open in new tabDownload slide Proportion of birds approaching and singing by playback type. Numbers in bars indicate number of trials. Error bars are binomial standard errors. Asterisk indicates significant difference from other bars. Figure 4. Open in new tabDownload slide Proportion of birds approaching and singing by playback type. Numbers in bars indicate number of trials. Error bars are binomial standard errors. Asterisk indicates significant difference from other bars. Figure 5. Open in new tabDownload slide Forest plot of Cohen’s d effect sizes and 95% confidence intervals comparing strength of response to each playback type to the Sonoran positive control. Figure 5. Open in new tabDownload slide Forest plot of Cohen’s d effect sizes and 95% confidence intervals comparing strength of response to each playback type to the Sonoran positive control. The mean effect sizes (Zr) for Verdins in tests of local vs. foreign treatments aligned well with the mean effect sizes for other species that do not have sympatric confamilials (Figure 6). For movement responses, species without sympatric confamilials tended to have smaller effect sizes (mean Zr = 0.198, n = 6 species) compared to species with sympatric confamilials (mean Zr = 0.321, n = 38 species). Verdin had an even lower mean effect size for movement responses than species without sympatric confamilials (mean Zr = 0.074). For vocal responses, species without sympatric confamilials tended to have larger effect sizes (mean Zr = 0.378, n = 4 species) compared to species with sympatric confamilials (mean Zr = 0.210, n = 21 species). The mean effect size for Verdin vocal responses fell between the means for species with and without sympatric confamilials (mean Zr = 0.290). Figure 6. Open in new tabDownload slide Effect sizes (Zr) for trials comparing response to local and foreign songs for all species included in Parker et al. (2018) (circle-shaped points) and Verdins (diamond-shaped points). Effect sizes are species means binned into movement and vocal responses, and species are divided into those with and without sympatric confamilials. Figure 6. Open in new tabDownload slide Effect sizes (Zr) for trials comparing response to local and foreign songs for all species included in Parker et al. (2018) (circle-shaped points) and Verdins (diamond-shaped points). Effect sizes are species means binned into movement and vocal responses, and species are divided into those with and without sympatric confamilials. DISCUSSION Verdins showed geographic variation in song that corresponded with different desert habitat regions in their range, and they showed discrimination toward foreign conspecific songs based on their vocal behavior, but no discrimination based on their movement behavior. These results suggest that although Verdins can perceive differences between foreign and local songs, they will still approach foreign songs just as readily as they approach local-sounding songs. The geographic variation in Verdin whistle songs mapped well onto the ecologically distinct desert habitat regions within their range. These regions are associated with subspecies of Verdins, or transition zones between subspecies (e.g., the Sierra Madre region), that are differentiated based on subtle plumage variations (Webster 2020). The general concordance between desert habitat, subspecies, and song suggests that there may be some reduction in dispersal between desert habitats that promotes song variation. The different deserts of the Southwest have vastly different plant communities, temperatures, and precipitation regimes (Shreve 1942, Archer and Predick 2008), so ecological differences or barriers such as the Cochise Filter Barrier (between the Sonoran and Chihuahuan deserts) may reduce dispersal between deserts (e.g., Provost et al. 2018). In Verdins, mitochondrial DNA does not show a signal of reduced gene flow between Sonoran and Chihuahuan deserts (Zink et al. 2001), but dispersal may be reduced enough to lead to variation in song despite some gene flow. Additionally, phenotypic clines can exist despite or in discordance with patterns of gene flow (e.g., Greig and Webster 2013, Poesel et al. 2017, Hooper et al. 2019). The variation in song that we documented across the Verdin range was represented reasonably well by the playback exemplars in our experiment, with the exception of 2 playback exemplars from Baja that overlapped in acoustic space with Sonoran songs (Figure 3B). Despite Baja and Chihuahuan playbacks originating from over 900 km away from the focal population, being from different subspecies than the focal population, and being largely acoustically divergent from the positive control Sonoran playbacks, Verdins were equally likely to approach Sonoran, Baja, and Chihuahuan playbacks (Table 5, Figure 4). In contrast, Verdins sang significantly more songs in response to Sonoran playbacks than to Chihuahuan playbacks and had a non-significant trend of singing more in response to Sonoran playbacks than to Baja playbacks (Table 5, Figure 5). Taken together, these results align well with what appears to be typical response patterns for local vs. foreign song treatments in other species that do not have sympatric confamilials (Figure 6). By analyzing the effect sizes presented in Figure 6 in relation to study design, a multitude of life history traits, and evolution history, Parker et al. (2018) found a subtle pattern that species without sympatric confamilials tended to have weaker discrimination between local and foreign song than species with sympatric confamilials/congeners. They also found that across all species movement responses tended to have larger effect sizes than vocal responses. Verdin, therefore, are unusual compared to most species because they approach foreign songs, but they are not unusual compared to species without sympatric confamilials. One difference in our study compared to the studies used in Parker et al. (2018) and presented in Figure 6 is that our positive control Sonoran playbacks were not recorded from the focal population, so are not representative of true local songs. Although we cannot determine if Verdins respond even more strongly to songs from the focal population because we did not test these, we are confident that our Sonoran playbacks represent songs quantitatively similar to focal population songs; the mean acoustic divergence of Sonoran songs from the focal population was almost equal to the mean acoustic divergence present within the focal population (Table 1). Additionally, the Sonoran song pc1 and pc2 scores almost completely overlap with the scores from the focal population (Figure 3B). Using Sonoran playbacks originating from non-local populations as a positive control had the advantage of being unfamiliar, so any difference in response to Sonoran vs. treatment playbacks was based on acoustic structure rather than prior experience of the focal birds. How can we reconcile the Verdins’ discrimination in vocal response with their lack of discrimination in movement response? The answer may be that different behaviors indicate different things about an individual’s motivation to respond (e.g., de Kort et al. 2009) or the tactics an individual uses to respond (e.g., Nowicki et al. 2002). It may be that for Verdins, approaching represents a useful behavioral response for a wide range of signals (e.g., investigation), but increasing song rate represents a useful behavior for a narrower subset of local-sounding signals (e.g., aggression). Although we cannot determine the precise functions and corresponding motivations of the behaviors Verdins exhibited in this study other than to assume they are related to territory defense, it is not unusual to have variation and inconsistency when multiple behavioral responses are measured in playback studies (McGregor 1992). Approaching foreign conspecific songs may not be costly for Verdins if the lack of confamilials in North America eliminates any risk of responding to similar-sounding heterospecifics. An example of such selective pressure release is found in singing mice (Scotinomys teguina), in which allopatric populations have stronger behavioral responses to heterospecifics than sympatric populations because allopatric populations are spared agonistic character displacement (Pasch et al. 2017). Another example is seen in African tinkerbirds, in which allopatric populations have stronger behavioral responses to heterospecifics than sympatric populations because sympatric populations have experienced character displacement and selection to avoid interactions (Kirschel et al. 2009). In these examples, a lack of exposure to divergent signals, and a lack of selective pressure to avoid them, seems to lead to equal responses to divergent and local signals (see also Dingle et al. 2010). Alternatively, approaching foreign conspecific songs may be useful for Verdins if they tend to encounter unfamiliar songs during their lifetime and benefit from responding to them. For example, in Black-throated Blue Warblers (Setophaga caerulescens) southern populations tend to respond to northern songs, but northern populations do not respond to southern songs. One hypothesized reason for this is that southern populations are exposed to northern songs during migration, but not vice versa (Colbeck et al. 2010). Another example is in Dickcissels, which do not discriminate between local and foreign songs. Males are exposed to foreign songs when winter flocks aggregate, and males that do not conform to the local song style will settle and defend territories in the breeding season (Parra et al. 2017). In these examples, exposure to divergent signals, and selective pressure to respond to them, seems to lead to equal responses to divergent and local signals. Yet another reason for equal responses to foreign and local signals is illustrated in Grasshopper Sparrows (Ammodramus savannarum), which do not distinguish between foreign and local songs, presumably because they improvise rather than imitate their songs and therefore do not have strict discrimination thresholds based on familiarity (Soha et al. 2016). We do not have any reason to think that Verdins improvise their songs rather than learn them, but song acquisition has not been studied in Verdins so we cannot rule out this hypothesis. There are numerous other examples of species both recognizing and responding to divergent songs from sympatric heterospecifics, for example, if they compete for resources, interfere with mating, or cooperate in resource/predator defense (Martin and Martin 2001, Sedlacek et al. 2006, Drury et al. 2015, Reif et al. 2015, Johnson et al. 2018). Ultimately, the criteria for responding to divergent songs, either conspecific or heterospecific, are diverse and appear to depend upon the interactions (or lack thereof) of the individuals involved and the costs and benefits of responding. This work provides an example of a species that shows geographic variation in song and behavioral responses consistent with some level of recognition of unfamiliar, acoustically divergent, and geographically distant conspecific songs. When considering the large number of playback studies showing reduced approach responses to foreign conspecific song (Parker et al. 2018), we might consider Verdins to be on the permissive end of the spectrum. Our single-species study does not give us the inference power to determine if the cause of the permissiveness in approach behavior is because Verdins have no sympatric confamilials, but it offers a case study that is consistent with this hypothesis and encourages future comparative work that explores other hypotheses such as the likelihood of encountering unfamiliar songs in an individual’s lifetime. Future studies that assess geographic variation in song and conspecific recognition in more species with a variety of life histories and evolutionary trajectories will be valuable because the questions we discuss here must ultimately be answered in a comparative framework. ACKNOWLEDGMENTS Thank you to J. Hunter Reed, Liam Berigan, Alex Weibe, Allison Johnson, Joe Cacioppo, Daniel Hooper, Sara Kaiser, Dylan Meyer, Samantha Symon, Ivy Sandquist, and Isaac Krone for help collecting data in the field. Thank you to Chrissy Kondrat-Smith, Peter Holm, and the staff of Organ Pipe Cactus National Monument for logistical support in the field and for organizing permits. Thank you to Irby Lovette, David Bonter, Anne Marie Johnson, Chelsea Benson, and Holly Faulkner for being supportive of this work. Funding statement: Thank you to the Blanksteen Foundation, the Cornell Lab of Ornithology, and the University of Chicago for funding that made this work possible. Ethics statement: All work in this study was evaluated and approved by the Cornell University Animal Care and Use Committee and National Park Service Institutional Animal Care and Use Committee (Protocol #2015-0052). The study was conducted under appropriate Federal (Permit #23245), State (License #SP770320 and #SP735654), and National Park (Study #ORPI-00021) permits. Author contributions: E.I.G. conceived the idea, design, and hypotheses, E.I.G. and E.C.L. supervised research, E.I.G., M.L.W., and E.K. collected data, performed experiments, refined methodology and hypotheses, E.I.G. wrote the paper and analyzed the data, E.I.G., M.L.W., E.K., and E.C.L. revised and edited the paper. Data depository: All data used in this paper are included as Supplementary Material and available at Greig et al. (2021). LITERATURE CITED Allen , F. H . ( 1919 ). The evolution of bird song . The Auk 36 : 528 – 536 . Google Scholar OpenURL Placeholder Text WorldCat Amézquita , A. , S. V. Flechas, A. P. Lima, H. Gasser, and W. Hödl ( 2011 ). Acoustic interference and recognition space within a complex assemblage of dendrobatid frogs . Proceedings of the National Academy of Sciences USA 108 : 17058 – 17063 . Google Scholar OpenURL Placeholder Text WorldCat Archer , S. R. , and K. I. Predick ( 2008 ). Climate change and ecosystems of the Southwestern United States . Rangelands 30 : 23 – 28 . Google Scholar OpenURL Placeholder Text WorldCat Bates , D. , M. Maechler, B. Bolker, and S. Walker ( 2015 ). Fitting linear mixed-effects models using lme4 . Journal of Statistical Software 67 : 1 – 48 . Google Scholar OpenURL Placeholder Text WorldCat Bioacoustics Research Program ( 2004 ). Raven Pro: Interactive Sound Analysis Software (version 1.5) . The Cornell Lab of Ornithology , Ithaca, NY, USA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Bird Studies Canada and NABCI ( 2014 ). Bird Conservation Regions. Bird Studies Canada, North American Bird Conservation Initiative . https://nabci-us.org/resources/bird-conservation-regions-map/ Colbeck , G. J. , T. S. Sillett, and M. S. Webster ( 2010 ). Asymmetric discrimination of geographical variation in song in a migratory passerine . Animal Behaviour 80 : 311 – 318 . Google Scholar OpenURL Placeholder Text WorldCat Dingle , C. , J. W. Poelstra, W. Halfwerk, D. M. Brinkhuizen, and H. Slabbekoorn ( 2010 ). Asymmetric response patterns to subspecies-specific song differences in allopatry and parapatry in the gray-breasted wood-wren . Evolution 64 : 3537 – 3548 . Google Scholar OpenURL Placeholder Text WorldCat Drury , J. P. , K. W. Okamoto, C. N. Anderson, and G. F. Grether ( 2015 ). Reproductive interference explains persistence of aggression between species . Proceedings of the Royal Society B 282 : 20142256 . Google Scholar OpenURL Placeholder Text WorldCat Freeman , B. G. , and G. A. Montgomery ( 2017 ). Using song playback experiments to measure species recognition between geographically isolated populations: A comparison with acoustic trait analysis . The Auk: Ornithological Advances 134 : 857 – 870 . Google Scholar OpenURL Placeholder Text WorldCat Greig , E. I. , E. Kinnebrew, M. L. Witynski, and E. C. Larsen ( 2021 ). Data from: A desert songbird with no confamilials in the Western Hemisphere (Verdin, Auriparus flaviceps) investigates divergent conspecific songs . Ornithology 138 : 1 – 13 . doi:10.5061/dryad.xksn02vfq Google Scholar OpenURL Placeholder Text WorldCat Greig , E. I. , and M. S. Webster ( 2013 ). Spatial decoupling of song and plumage generates novel phenotypes between two avian subspecies . Behavioral Ecology 24 : 1004 – 1013 . Google Scholar OpenURL Placeholder Text WorldCat Grether , G. F. , N. Losin, C. N. Anderson, and K. Okamoto ( 2009 ). The role of interspecific interference competition in character displacement and the evolution of competitor recognition . Biological Reviews of the Cambridge Philosophical Society 84 : 617 – 635 . Google Scholar OpenURL Placeholder Text WorldCat Hamao , S . ( 2016 ). Asymmetric response to song dialects among bird populations: The effect of sympatric related species . Animal Behaviour 119 : 143 – 150 . Google Scholar OpenURL Placeholder Text WorldCat Hooper , D. M. , S. C. Griffith, and T. D. Price ( 2019 ). Sex chromosome inversions enforce reproductive isolation across an avian hybrid zone . Molecular Ecology 28 : 1246 – 1262 . Google Scholar OpenURL Placeholder Text WorldCat Irwin , D. E. , and T. Price ( 1999 ). Sexual imprinting, learning and speciation . Heredity 82 : 347 – 354 . Google Scholar OpenURL Placeholder Text WorldCat Johnson , A. E. , C. Masco, and S. Pruett-Jones ( 2018 ). Song recognition and heterospecific associations between 2 fairy-wren species (Maluridae) . Behavioral Ecology 29 : 821 – 832 . Google Scholar OpenURL Placeholder Text WorldCat Kirschel , A. N. , D. T. Blumstein, and T. B. Smith ( 2009 ). Character displacement of song and morphology in African tinkerbirds . Proceedings of the National Academy of Sciences USA 106 : 8256 – 8261 . Google Scholar OpenURL Placeholder Text WorldCat de Kort , S. R. , E. R. Eldermire, E. R. Cramer, and S. L. Vehrencamp ( 2009 ). The deterrent effect of bird song in territory defense . Behavioral Ecology: Official Journal of the International Society for Behavioral Ecology 20 : 200 – 206 . Google Scholar OpenURL Placeholder Text WorldCat Marler , P . ( 1958 ). Bird songs and mate selection . Animal Behaviour 6 : 348 – 367 . Google Scholar OpenURL Placeholder Text WorldCat Martin , P. R. , and T. E. Martin ( 2001 ). Behavioral interactions between coexisting species: Song playback experiments with wood warblers . Ecology 82 : 207 – 218 . Google Scholar OpenURL Placeholder Text WorldCat Maruvka , Y. E. , N. M. Shnerb, D. A. Kessler, and R. E. Ricklefs ( 2013 ). Model for macroevolutionary dynamics . Proceedings of the National Academy of Sciences USA 110 : E2460 – E2469 . Google Scholar OpenURL Placeholder Text WorldCat McCreedy , C. , and C. van Riper , III ( 2015 ). Drought-cased delay in nesting of Sonoran Desert birds and its facilitation of parasite- and predator-mediated variation in reproductive success . The Auk 132 : 235 – 247 . Google Scholar OpenURL Placeholder Text WorldCat McGregor , P. K . ( 1992 ). Quantifying responses to playback: One, many, or composite multivariate measures? In Playback and Studies of Animal Communication ( P. K. McGregor, Editor). Plenum Press , New York, NY, USA . pp. 79 – 96 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Nakagawa , S. , and I. C. Cuthill ( 2007 ). Effect size, confidence interval and statistical significance: A practical guide for biologists . Biological Reviews of the Cambridge Philosophical Society 82 : 591 – 605 . Google Scholar OpenURL Placeholder Text WorldCat Nowicki , S. , W. A. Searcy, T. Kruegar, and M. Hughes ( 2002 ). Individual variation in response to simulated territorial challenges among territory-holding song sparrows . Journal of Avian Biology 33 : 253 – 259 . Google Scholar OpenURL Placeholder Text WorldCat Nychka , D. , R. Furrer, J. Paige, and S. Sain ( 2015 ). Fields: Tools for spatial data. R package version 9.0 . https://github.com/NCAR/Fields Parker , T. H. , E. I. Greig, S. Nakagawa, M. Parra, and A. C. Dalisio ( 2018 ). Subspecies status and methods explain strength of response to local versus foreign song by oscine birds in meta-analysis . Animal Behaviour 142 : 1 – 17 . Google Scholar OpenURL Placeholder Text WorldCat Parra , M. , A. C. Dalisio, W. E. Jensen, and T. H. Parker ( 2017 ). Male territorial aggression does not drive conformity to local vocal culture in a passerine bird . Ethology 123 : 800 – 810 . Google Scholar OpenURL Placeholder Text WorldCat Pasch , B. , R. Sanford, and S. M. Phelps ( 2017 ). Agonistic character displacement in social cognition of advertisement signals . Animal Cognition 20 : 267 – 273 . Google Scholar OpenURL Placeholder Text WorldCat Podos , J. , and P. S. Warren ( 2007 ). The evolution of geographic variation in birdsong . Advances in the Study of Behavior 37 : 403 – 458 . Google Scholar OpenURL Placeholder Text WorldCat Poesel , A. , A. C. Fried, L. Miller, H. L. Gibbs, J. A. Soha, D. A. Nelson ( 2017 ). High levels of gene flow among song dialect populations of the Puget Sound White-crowned Sparrow . Ethology 123 : 581 – 592 . Google Scholar OpenURL Placeholder Text WorldCat Provost , K. L. , W. M. Mauck , 3rd, and B. T. Smith ( 2018 ). Genomic divergence in allopatric Northern Cardinals of the North American warm deserts is linked to behavioral differentiation . Ecology and Evolution 8 : 12456 – 12478 . Google Scholar OpenURL Placeholder Text WorldCat R Core Team ( 2017 ). R: A language and environment for statistical computing . R Foundation for Statistical Computing , Vienna, Austria. https://www.R-project.org/ . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Reif , J. , M. Jiran, R. Reifova, J. Vokurkova, P. Dolata, A. Petrusek, and T. Petruskova ( 2015 ). Interspecific territoriality in two songbird species: Potential role of song convergence in male aggressive interactions . Animal Behaviour 104 : 131 – 136 . Google Scholar OpenURL Placeholder Text WorldCat Rothstein , S. I. , and R. C. Fleischer ( 1987 ). Vocal dialects and their possible relation to honest status signaling in the Brown-headed Cowbird . The Condor 89 : 1 – 23 . Google Scholar OpenURL Placeholder Text WorldCat Searcy , W. A. , S. Nowicki, M. Hughes, and S. Peters ( 2002 ). Geographic song discrimination in relation to dispersal distances in song sparrows . The American Naturalist 159 : 221 – 230 . Google Scholar OpenURL Placeholder Text WorldCat Sedlacek , O. , B. Cikanova, and R. Fuchs ( 2006 ). Heterospecific rival recognition in the Black Redstart (Phoenicurus ochruros) . Ornis Fennica 83 : 153 – 161 . Google Scholar OpenURL Placeholder Text WorldCat Shreve , F . ( 1942 ). The desert vegetation of North America . Botanical Review 8 : 195 – 246 . Google Scholar OpenURL Placeholder Text WorldCat Soha , J. A. , A. Poesel, D. A. Nelson, and B. Lohr ( 2016 ). Non-salient geographic variation in birdsong in a species that learns by improvisation . Ethology 122 : 343 – 353 . Google Scholar OpenURL Placeholder Text WorldCat Symes , L. B . ( 2014 ). Community composition affects the shape of mate response functions . Evolution 68 : 2005 – 2013 . Google Scholar OpenURL Placeholder Text WorldCat Taft , B . ( 2011 ). The role of dawn song in Tree Swallows and its place in the diversity of oscine song learning. Ph.D. dissertation, University of Massachusetts , Amherst, MA, USA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Taylor , W. K . ( 1971 ). Breeding biology study of Verdin, Auriparus flaviceps (Sundevall) in Arizona . American Midland Naturalist 85 : 289 – 328 . Google Scholar OpenURL Placeholder Text WorldCat Turcokova , L. , V. Pavel, B. Chutny, A. Petrusek, and T. Petruskova ( 2011 ). Differential response of males of a subarctic population of Bluethroat Luscinia svecica svecica to playbacks of their own and foreign subspecies . Journal of Ornithology 152 : 975 – 982 . Google Scholar OpenURL Placeholder Text WorldCat Venables , W. N. , and B. D. Ripley ( 2002 ). Modern Applied Statistics with S , fourth edition. Springer , New York, NY, USA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Webster , M. D . ( 2020 ). Verdin (Auriparus flaviceps), version 1.0 . In Birds of the World ( A. F. Poole and F. B. Gill, Editors). Cornell Lab of Ornithology , Ithaca, NY, USA . https://doi.org/10.2173/bow.verdin.01 Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Weir , J. T. , and T. D. Price ( 2019 ). Song playbacks demonstrate slower evolution of song discrimination in birds from Amazonia than from temperate North America . PLoS Biology 17 : e3000478 . Google Scholar OpenURL Placeholder Text WorldCat Wiley , R. H . ( 1994 ). Errors, exaggeration, and deception in animal communication. In Behavioral Mechanisms in Ecology ( L. Real, Editor). University of Chicago Press , Chicago, IL, USA . pp. 157 – 189 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Wilkins , M. R. , N. Seddon, and R. J. Safran ( 2013 ). Evolutionary divergence in acoustic signals: Causes and consequences . Trends in Ecology & Evolution 28 : 156 – 166 . Google Scholar OpenURL Placeholder Text WorldCat Zink , R. M. , A. E. Kessen, T. V. Line, and R. C. Blackwell-Rago ( 2001 ). Comparative phylogeography of some aridland bird species . The Condor 103 : 1 – 10 . Google Scholar OpenURL Placeholder Text WorldCat Copyright © American Ornithological Society 2021. All rights reserved. For permissions, e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Ornithology Oxford University Press

A desert songbird with no confamilials in the Western Hemisphere (Verdin, Auriparus flaviceps) investigates divergent conspecific songs

Ornithology , Volume Advance Article – May 8, 2021

Loading next page...
 
/lp/oxford-university-press/a-desert-songbird-with-no-confamilials-in-the-western-hemisphere-rzAEw6GdxS
Copyright
Copyright © 2021 American Ornithological Society
eISSN
2732-4613
DOI
10.1093/ornithology/ukab032
Publisher site
See Article on Publisher Site

Abstract

Abstract Most birds that show geographic variation in their songs discriminate between local and foreign songs, which may help them avoid unnecessary conflicts with vagrant individuals or similar-sounding congeners. However, some species respond equally to foreign and local songs, which may be useful if foreign individuals present territorial threats or if there are no sympatric congeners to avoid. Species without sympatric congeners are not commonly tested in playback studies, but they offer an opportunity to see how song variation and recognition unfolds when the pressure to avoid similar congeners is absent. Here, we use Verdins (Auriparus flaviceps), a monotypic genus of songbird with no confamilials in North America, to explore song variation and recognition in a species living without close relatives. We assessed geographic variation in song across the Verdin range and conducted a playback experiment using exemplars from 2 acoustically divergent and geographically distant regions as treatments. We found significant geographic variation in song that mapped well onto ecologically distinct desert regions. We found that Verdins had stronger vocal responses to local-sounding songs, but had equal movement responses to local-sounding and foreign songs. These results are similar to results found in other species without sympatric congeners and provide an example of a species that investigates acoustically divergent conspecific songs, despite recognizing salient differences in those songs. RESUMEN La mayoría de las aves que muestran variación geográfica en sus cantos discriminan entre los cantos locales y extranjeros, lo que puede ayudarlas a evitar conflictos innecesarios con individuos vagabundos o con congéneres que suenan similares. Sin embargo, algunas especies responden de mismo modo a los cantos extranjeros y locales, lo que puede ser útil si los individuos extranjeros representan una amenaza territorial o si no hay congéneres simpátricos a evitar. Las especies sin congéneres simpátricos no son comúnmente evaluadas en estudios de playback, pero ellas ofrecen una oportunidad para ver cómo se revela la variación y el reconocimiento del canto cuando la presión para evitar a congéneres similares está ausente. En este estudio usamos a Auriparus flaviceps, un género monotípico de ave canora sin familiares en Norteamérica, para explorar la variación y el reconocimiento del canto en una especie que vive sin parientes cercanos. Evaluamos la variación geográfica en el canto a través del rango de esta especie y realizamos un experimento de playback usando ejemplares prevenientes de dos regiones acústicamente divergentes y geográficamente distantes como tratamientos. Encontramos variación geográfica significativa en el canto, que se correspondió bien con las regiones desérticas ecológicamente distintas. Encontramos que A. flaviceps tuvo respuestas vocales más fuertes a los cantos que suenan como locales, pero tuvo las mismas respuestas de movimiento a los cantos que suenan como locales o extranjeros. Estos resultados son similares a los encontrados en otras especies sin congéneres simpátricos y brindan un ejemplo de una especie que indaga los cantos de individuos conespecíficos acústicamente divergentes, a pesar de reconocer diferencias marcadas en esos cantos. Lay Summary Many songbirds have geographic variation in their songs. Most species prefer songs that sound local, or familiar, but some will respond to foreign songs just as readily as they respond to local songs. This may be more likely if there are no similar-sounding relatives around, and therefore no risk of mistaken identity. Few species without close relatives have been studied, so we study one that is also a desert specialist, the Verdin (Auriparus flaviceps). We found that Verdins have geographic variation in song that mapped well onto large desert habitat regions, and they sing less to playbacks of songs from distant regions. However, despite recognizing salient differences in foreign songs, they are just as likely to approach foreign songs as local-sounding songs, as are other species that live without close relatives. When close relatives are absent, birds may be more permissive in what they consider a song worth investigating. INTRODUCTION Bird song is an important signal for mate and competitor recognition (Marler 1958), and understanding how song and recognition evolves has captured the attention of biologists for over a century (Allen 1919). Many species show a pattern of stronger response to acoustically similar song phenotypes and weaker response to dissimilar phenotypes (Podos and Warren 2007, Wilkins et al. 2013, Freeman and Montgomery 2017, Parker et al. 2018, Weir and Price 2019). This is reasonable because recognition of conspecifics is important for mate acquisition and resource defense, but responses to unimportant signals, such as signals from congeners, could be a waste of time and energy (Wiley 1994, Amézquita et al. 2011, Symes 2014, Hamao 2016). Even within species, more geographically proximate populations tend to sound more similar (e.g., may have the same dialect) and tend to receive stronger responses in territorial playbacks than more geographically distant populations (Podos and Warren 2007, Parker et al. 2018). Stronger territorial responses to local-sounding songs may be useful for many reasons, for example, if local individuals pose the greatest threat (Turcokova et al. 2011), if local individuals make the best mates (Rothstein and Fleischer 1987), or if local-sounding individuals are the highest quality (Searcy et al. 2002). Having a preference for the familiar need not even be adaptive and may simply be a byproduct of individual experience (Irwin and Price 1999). Even though many species show stronger responses to local-sounding songs, there are exceptions that highlight how response may depend upon circumstance. For example, Dickcissels (Spiza americana) respond equally to conspecific local and foreign songs, either because they tend to encounter foreign individuals in territorial interactions or because, as a monotypic genus, they face no selection to avoid similar-sounding congeners (Parra et al. 2017). Populations of Japanese Tits (Parus minor) and Varied Tits (Sittiparus varius) that live in isolation from other members of the family Paridae respond equally to local and foreign conspecific songs, presumably because they are released from selection to avoid competitive interactions with heterospecifics (Hamao 2016). Work in frogs, insects, and birds supports the idea that the likelihood of encountering phenotypically similar heterospecifics (e.g., congeners or confamilials) may influence responses to divergent conspecifics (Grether et al. 2009, Amézquita et al. 2011, Symes 2014, Parker et al. 2018). However, in birds, studies examining conspecific recognition in species without sympatric confamilials are relatively uncommon (Parker et al. 2018). This discrepancy is expected given that polytypic genera are more common than monotypic genera (Maruvka et al. 2013), but monotypic genera offer an opportunity to see how conspecific recognition unfolds when there is no pressure to avoid costly heterospecific interactions. We add to the literature on species without sympatric confamilials by assessing geographic variation in song and response to divergent conspecific signals in Verdins (Auriparus flaviceps), a species that is in a monotypic genus and is the only member of its family (Remizidae) in the Western Hemisphere. Verdins are an interesting study species not only because of their isolated evolutionary history, but also because of their unusual ecology as a non-migratory desert specialist (Webster 2020). Their range encompasses all 3 hot desert habitats in North America (Sonoran, Chihuahuan, and Mohave), and subspecies have been described based on subtle plumage variations, but no one has yet described geographic variation in their song (Webster 2020). We leveraged the recordings of Verdins from 2 large public audio collections to describe geographic variation in song, and we conducted a playback experiment with treatments from 2 acoustically divergent and geographically distant geographic regions to assess response to divergent conspecific songs in a central focal population. Finally, we assessed how Verdin responses compare to those of other species by qualitatively comparing the effect sizes from this study to effect sizes from a meta-analysis of 45 species with and without congeners (Parker et al. 2018). METHODS Study Species and System Verdins are small insectivorous passerines that live in desert habitat throughout the Southwestern United States and Mexico. They are non-migratory, though study populations have a relatively large turnover of individuals between breeding seasons, potentially because of high dispersal and/or low survival (Taylor 1971, Webster 2020, E. I. Greig personal observation). They breed and roost in ball-shaped nests suspended from the tips of vegetation such as paloverde (Parkinsonia florida) branches (Taylor 1971, McCreedy and van Riper 2015). Verdins have a variety of vocalizations that are involved in intra-pair communication, mate attraction, and territory defense (Taylor 1971, Webster 2020). The “whistle” song, which is the focus of this study, is described as what is considered a typical song in oscines and is given often by males and sometimes by females (Taylor 1971). Although the function of the whistle song is uncertain, it is likely associated with territory maintenance or nesting behavior because it is used primarily during the breeding season and in the dawn chorus (Taylor 1971, E. I. Greig personal observation). We choose the whistle song as the focus for this study because it is the vocalization most analogous to what is usually considered a “song” in birds, based on context and acoustic structure (i.e., it is relatively loud and elaborate compared to most other Verdin vocalizations). Whistle songs are typically composed of 2–4 notes that are similar in structure (Figure 1). Although repertoire size has not been specifically quantified for Verdins, song variation within individuals seems minimal (E. I. Greig personal observation). An individual might sing a 2-note song and a 3-note song, but the structure of the notes within those songs appears consistent over time (e.g., male #119 in Figure 1). Figure 1. Open in new tabDownload slide Spectrograms of songs from the focal population (top row) and songs used for the playback experiment (lower 3 rows) showing representative geographic variation between regions. Dotted lines separate songs. Letters indicate playback type (S = Sonoran, B = Baja, C = Chihuahuan, OW = Old World confamilial), and numbers indicate playback exemplar ID (lower 3 rows) or individual ID (top row). Figure 1. Open in new tabDownload slide Spectrograms of songs from the focal population (top row) and songs used for the playback experiment (lower 3 rows) showing representative geographic variation between regions. Dotted lines separate songs. Letters indicate playback type (S = Sonoran, B = Baja, C = Chihuahuan, OW = Old World confamilial), and numbers indicate playback exemplar ID (lower 3 rows) or individual ID (top row). We studied a population of Verdins at Organ Pipe Cactus National Monument (ORPI) in Pima County, Arizona (31.942851°N, –112.813883°W, WGS 84). We located and color-banded nesting birds on 30 territories during the breeding seasons of March–April 2016 and 2017 (n = 18 territories in 2016; 14 with both members of the pair banded, 1 with the female only banded, and 3 with neither member of the pair banded and n = 12 territories in 2017; 7 with both members of the pair banded, 1 with the female only banded, and 4 with banded females, but in which the males were raising nestlings at other nests). Although we were not able to band every individual, we were sure that at least the female member of the pair was different on every territory because territories had active nests. We included the 4 territories in 2017 that had females raising nestlings alone because females engage in territory defense (E. I. Greig personal observation), so there was no a priori reason to exclude them. Acoustic Analysis To quantify the range of geographic variation of Verdin whistle songs and the acoustic divergence of playbacks relative to songs from the focal population, we conducted an acoustic analysis on a collection of songs from the focal population (n = 154 songs from 21 individuals recorded at ORPI in March 2015 and April 2019) and all available Verdin songs from the Macaulay Library (Cornell Lab of Ornithology, Ithaca, New York, n = 222 songs from 48 individuals), Xeno-canto (www.xeno-canto.org, n = 267 songs from 52 individuals) and colleagues (Lazarus Pomara, n = 12 songs from 3 individuals). Details of all recordings, including accession numbers, are in Supplementary Material Recording Metadata. We excluded songs that were a single note because they were not representative of typical whistle songs (n = 21 out of an original 676 identified songs). The final collection of 655 local and non-local whistle songs included the conspecific exemplars used for our playback experiment (n = 17 songs from 8 individuals; Figure 1). The collection of recordings spanned a variety of geographic regions, and we classified all songs as belonging to 1 of 4 geographic regions informed by the Bird Conservation Regions (BCR) (Bird Studies Canada and NABCI 2014); Sonoran Desert (BCR region 33, associated with the acaciarum subspecies, n = 232 songs plus 154 songs from the focal population), Sierra Madre Occidental (BCR region 34, a transition region between acaciarum and ornatus subspecies, n = 83 songs), Chihuahuan Desert (BCR regions 35 and 36, associated with the ornatus subspecies, n = 57 songs), and Baja California Sur (BCR regions 40 and 42, associated with the lamprocephalus subspecies, n = 129 songs). We grouped songs from the adjacent BCR regions 35/36 and 40/42 because recordings were sparse from each individual region. We measured the frequency contour of each note with an equal number of points (acoustic landmarks), which allowed us to assess note shape as well as calculate standard acoustic features such as length and high/low frequency (Taft 2011). This method is ideal for simple, tonal sounds such as Verdin whistle song notes because they can be represented well by a single frequency contour (Figure 2). Considering only the fundamental frequency contour of each note also minimizes any impact of recording quality on song variation, and when we tested for an effect of recording source (Macaulay Library, Xeno-Canto and personal collection) on song structure we did not find an effect. We used spectrograms digitized in Raven 1.5 (Bioacoustics Research Program 2004) with 16-bit sample format, discrete Fourier transform (DFT) = 512 samples, frequency resolution = 124 Hz, time resolution = 11.6 ms, frame overlap = 50%. We manually selected each note (n = 1,883 notes) to ensure we were measuring the target vocalizations, rather than background noise or other vocalizing birds. We described every note using landmarks that measured the approximate peak frequency of the note at 24 points in time along the note. To create these landmarks in Raven, we used the Split All Selection Borders function to create 24 “selections” for each note, and we used the Center Frequency measurement for each selection to extract frequency values at which the energy within the selection was equally divided above and below the value (this approximates the peak frequency contour of the note; Figure 2). We used the Center Time measurement for each selection to extract 24 time values along each note that we then scaled from the beginning of the note. Figure 2. Open in new tabDownload slide Example of a song note with acoustic landmarks made using the Split All Selection Borders function in Raven. Points represent the locations within each section where the Center Time measurement intersects with the Center Frequency measurement. This example shows 10 landmarks for clarity, but 24 were taken on each note for analysis. Figure 2. Open in new tabDownload slide Example of a song note with acoustic landmarks made using the Split All Selection Borders function in Raven. Points represent the locations within each section where the Center Time measurement intersects with the Center Frequency measurement. This example shows 10 landmarks for clarity, but 24 were taken on each note for analysis. We used R v. 3.3.3 (R Core Team 2017) for all calculations with acoustic landmarks. We used the landmarks to characterize song length, high frequency, and low frequency. We also used the landmarks to characterize note shape by running a principal component analysis on the centered and scaled landmarks for all 1,883 notes using the princomp function. The first 3 principal components from this analysis described 89.9% of the variation in notes. Because Verdin whistle songs are a series of repeated, similar notes, for every song we calculated the mean of the first 3 principal component scores across notes. The resulting mean scores (note shape pc1, pc2, and pc3) became our song-level metrics of note shape. To characterize song divergence relative to the focal population, we ran a second principal component analysis on the 6 song-level measurements (song length, song high frequency, song low frequency, note shape pc1, pc2, and pc3; Table 1) for all 655 songs. The first 3 principal components from this analysis described 80% of the variation in songs (Table 2). We used these 3 principal components to calculate a measure of acoustic divergence. We define acoustic divergence as the Euclidean distance in 3-dimensional acoustic space between every non-focal population song (n = 501 songs) and the point representing the mean song-level principal component scores of the focal population (based on n = 154 songs from ORPI). We calculated the geographic distance between every non-focal population song and the focal population using the rdist.earth function implemented in the package fields (Nychka et al. 2015). All acoustic data are available in Supplementary Material Acoustic Data and Greig et al. (2021). Table 1. Mean values for acoustic characters of songs from the focal population and each geographical region. Acoustic divergence is the Euclidean 3-dimensional distance between each song in the first 3 song-level principal component scores relative to the mean scores of the focal population. . Focal . Sonoran . Sierra Madre . Baja . Chihuahuan . . (n = 154) . (n = 232) . (n = 83) . (n = 129) . (n = 57) . Song length (s) 0.66 0.70 0.61 0.53 0.75 High frequency (Hz) 5,579 5,403 5,165 6,478 4,558 Low frequency (Hz) 3,590 3,610 3,911 3,454 3,520 Note shape pc1 –4.49 –0.67 3.37 2.90 0.61 Note shape pc2 –1.77 0.28 0.59 –0.82 3.27 Note shape pc3 0.19 –0.12 –0.56 2.18 –3.49 Acoustic divergence 1.48 1.85 2.94 2.89 3.48 . Focal . Sonoran . Sierra Madre . Baja . Chihuahuan . . (n = 154) . (n = 232) . (n = 83) . (n = 129) . (n = 57) . Song length (s) 0.66 0.70 0.61 0.53 0.75 High frequency (Hz) 5,579 5,403 5,165 6,478 4,558 Low frequency (Hz) 3,590 3,610 3,911 3,454 3,520 Note shape pc1 –4.49 –0.67 3.37 2.90 0.61 Note shape pc2 –1.77 0.28 0.59 –0.82 3.27 Note shape pc3 0.19 –0.12 –0.56 2.18 –3.49 Acoustic divergence 1.48 1.85 2.94 2.89 3.48 Open in new tab Table 1. Mean values for acoustic characters of songs from the focal population and each geographical region. Acoustic divergence is the Euclidean 3-dimensional distance between each song in the first 3 song-level principal component scores relative to the mean scores of the focal population. . Focal . Sonoran . Sierra Madre . Baja . Chihuahuan . . (n = 154) . (n = 232) . (n = 83) . (n = 129) . (n = 57) . Song length (s) 0.66 0.70 0.61 0.53 0.75 High frequency (Hz) 5,579 5,403 5,165 6,478 4,558 Low frequency (Hz) 3,590 3,610 3,911 3,454 3,520 Note shape pc1 –4.49 –0.67 3.37 2.90 0.61 Note shape pc2 –1.77 0.28 0.59 –0.82 3.27 Note shape pc3 0.19 –0.12 –0.56 2.18 –3.49 Acoustic divergence 1.48 1.85 2.94 2.89 3.48 . Focal . Sonoran . Sierra Madre . Baja . Chihuahuan . . (n = 154) . (n = 232) . (n = 83) . (n = 129) . (n = 57) . Song length (s) 0.66 0.70 0.61 0.53 0.75 High frequency (Hz) 5,579 5,403 5,165 6,478 4,558 Low frequency (Hz) 3,590 3,610 3,911 3,454 3,520 Note shape pc1 –4.49 –0.67 3.37 2.90 0.61 Note shape pc2 –1.77 0.28 0.59 –0.82 3.27 Note shape pc3 0.19 –0.12 –0.56 2.18 –3.49 Acoustic divergence 1.48 1.85 2.94 2.89 3.48 Open in new tab Table 2. Correlations of song variables with the first 3 principal components of the song-level principal component analysis. . pc1 . pc2 . pc3 . Eigenvalue 1.46 1.16 1.16 Cumulative variation 0.36 0.58 0.80 Song length (s) 0.28 –0.34 0.35 High frequency (Hz) –0.64 0.00 0.08 Low frequency (Hz) 0.21 0.73 0.24 Note shape pc1 –0.08 0.48 –0.59 Note shape pc2 0.26 –0.32 –0.69 Note shape pc3 –0.63 –0.10 –0.05 . pc1 . pc2 . pc3 . Eigenvalue 1.46 1.16 1.16 Cumulative variation 0.36 0.58 0.80 Song length (s) 0.28 –0.34 0.35 High frequency (Hz) –0.64 0.00 0.08 Low frequency (Hz) 0.21 0.73 0.24 Note shape pc1 –0.08 0.48 –0.59 Note shape pc2 0.26 –0.32 –0.69 Note shape pc3 –0.63 –0.10 –0.05 Open in new tab Table 2. Correlations of song variables with the first 3 principal components of the song-level principal component analysis. . pc1 . pc2 . pc3 . Eigenvalue 1.46 1.16 1.16 Cumulative variation 0.36 0.58 0.80 Song length (s) 0.28 –0.34 0.35 High frequency (Hz) –0.64 0.00 0.08 Low frequency (Hz) 0.21 0.73 0.24 Note shape pc1 –0.08 0.48 –0.59 Note shape pc2 0.26 –0.32 –0.69 Note shape pc3 –0.63 –0.10 –0.05 . pc1 . pc2 . pc3 . Eigenvalue 1.46 1.16 1.16 Cumulative variation 0.36 0.58 0.80 Song length (s) 0.28 –0.34 0.35 High frequency (Hz) –0.64 0.00 0.08 Low frequency (Hz) 0.21 0.73 0.24 Note shape pc1 –0.08 0.48 –0.59 Note shape pc2 0.26 –0.32 –0.69 Note shape pc3 –0.63 –0.10 –0.05 Open in new tab Playback Experiment To assess the response of birds from the focal population to geographically and acoustically distant conspecific songs, we conducted a playback experiment from March 26 to April 3, 2016 and from March 27 to March 31, 2017. Focal pairs were nesting and generally responsive to playbacks at that time in the season. We used an experimental design in which each focal territory received each of 3 playback types: (1) a positive control song recorded from the Sonoran region within the range of 160–480 km from the study site and from the same subspecies as the focal population (acaciarum, n = 9 exemplars; Figure 1 and 3A); (2) a treatment song recorded from the Chihuahuan region or Baja region, 960–1,220 km from the playback site and from a different subspecies as the focal population (ornatus or lamprocephalus, n = 8 exemplars; Figure 1 and 3A); and (3) a negative control heterospecific song recorded from an Old World confamilial Remizid (Remiz pendulinus, R. consobrinus, Anthoscopus caroli, A. minutus or A. musculus, n = 9 exemplars; Figure 1). We chose exemplar recordings from the Macaulay Library and Xeno-canto based on the location of the recording and the clarity of the spectrograms (details of exemplar recordings in Supplementary Material Recording Metadata). We standardized the recordings in Raven to be the same sample rate, bit-depth, and amplitude for playback (16-bit and 44 kHz). We used an amplified speaker (Pignose Legendary 7-100, Pignose-Gorilla, Las Vegas, Nevada, USA; frequency response: 0.1–12.0 kHz) for playback. We tested the amplitude of playbacks in the field using a sound-pressure level meter (model number 33-2050, Radio Shack Corporation, Fort Worth, Texas, USA), set at C-weighting, fast response (~89.0 dB at 1 m for all playbacks). This amplitude was similar to natural Verdin song levels. Figure 3. Open in new tabDownload slide (A) Map of geographic distribution of song recordings (small dots) and playback exemplars (large labeled dots) overlayed on the range of Verdin (gray) and the Bird Conservation Regions (black lines) and colored by our geographic region classification (yellow = Sonoran, orange = Sierra Madre, blue = Baja, pink = Chihuahuan). Black dot shows the focal population site. (B) Song-level principal component scores 1 and 2 for every song recording and playback exemplar colored by geographic region. Black dots are recordings from the focal population. Sierra Madre songs are not labeled with larger dots because they were not used in the playback experiment. Figure 3. Open in new tabDownload slide (A) Map of geographic distribution of song recordings (small dots) and playback exemplars (large labeled dots) overlayed on the range of Verdin (gray) and the Bird Conservation Regions (black lines) and colored by our geographic region classification (yellow = Sonoran, orange = Sierra Madre, blue = Baja, pink = Chihuahuan). Black dot shows the focal population site. (B) Song-level principal component scores 1 and 2 for every song recording and playback exemplar colored by geographic region. Black dots are recordings from the focal population. Sierra Madre songs are not labeled with larger dots because they were not used in the playback experiment. We presented all 3 playback types to every focal territory (n = 30 territories) so that we could account for between-individual variation in response. We presented playbacks to focal territories on separate days and we balanced the order of playback presentations so that each playback type (i.e., positive control, treatment, and negative control) was equally represented in every order position (i.e., first, second, or third). A trial consisted of 1 or 2 observers arriving at a focal territory, placing the playback speaker at the base of a bush ~15–30 m from the focal birds’ nest, concealing themselves behind nearby vegetation ~20 m from the speaker, and turning on the playback. The playback consisted of a single exemplar song repeated every 10 s for 5 min, during which the territory was monitored. We chose a 5-min playback period because we began each playback irrespective of the location of the focal birds, and it often took several minutes for focal birds to notice the playback (mean latency to approach positive control songs was 113 s). This means there was some probability of false-negative responses, in which birds did not respond because they were not within the range of the speaker during the playback, but this was equally possible across playback types and thus is accounted for in our experimental design. Additionally, there was some probability of false-positive responses because some birds may have come to the area of their nests (which was near the speaker) irrespective of the playback being presented, but again this was equally possible across playback types and thus is accounted for in our experimental design. The observers recorded all focal bird vocal behavior and dictated all movement behavior possible using a digital recorder (Marantz PMD 661 solid-state digital recorder at 96 kHz sampling rate, 24-bit depth, D&M Professional, Itasca, Illinois, USA combined with ME66 shotgun microphone capsules and K6 power modules, Sennheiser Electronic Corporation, Old Lyme, Connecticut, USA; frequency response: 0.04–20.0 kHz). From these dictated recordings of the trials, we later transcribed the following response variables: (1) whether or not at least 1 member of the pair approached within 20 m of the speaker, (2) the latency of the first member of the pair approaching within 20 m of the speaker, (3) the total time that at least 1 member of the pair spent within 20 m of the speaker, (4) whether or not at least 1 member of the pair sang a whistle song, (5) the latency of the first member of the pair to sing a whistle song, and (6) the total number of whistle songs given by both members of the pair. We continued to observe territories for 5 min after the playback had stopped, but for our analysis, we only use responses that began within the 5-min playback period. We combined male and female responses in our analysis for 2 reasons. First, male and female Verdins have similar enough plumage that seeing color bands on birds sexed in the hand is the only way to be certain of the sex of individuals in the field. Observers were not always able to identify color bands because focal birds would sometimes sing from locations out of sight (n = 31 sing instances) or approach for a period of time too fleeting to see their bands (n = 15 approach instances). By considering responses blind to the sex of the focal birds we could treat all trials equally, despite variation in observer success identifying color bands and variation in whether or not a focal pair had been banded. We have high confidence that responses were from the focal pair rather than from intruding birds because we never observed more than 2 birds respond simultaneously and all birds that we were able to identify by their bands were focal pair members. Second, because every focal territory received all 3 trial types, any between-territory variation in male/female responsiveness (e.g., focal territories where only a female was present or focal territories at different stages of the nesting cycle) was accounted for in mixed models. All playback data are available in Supplementary Material Playback Data and Greig et al. (2021). Statistical Analysis We used R v. 3.3.3. for all statistical analyses. To assess how well the 6 song-level measurements categorized all songs (n = 655) as originating from 1 of the 4 different geographical regions, we used a linear discriminant function analysis implemented in the package MASS (Venables and Ripley 2002). We standardized all song measurements using the scale function before analysis. To assess the importance of geographic region and geographic distance from the focal population on acoustic divergence (Euclidean distance) for the 501 non-focal population songs, we used a model-selection approach with linear mixed models implemented with the lmer function in the package lme4 (Bates et al. 2015). We used the Akaike information criterion of ΔAIC < 2 indicating equivalent models. We excluded the 154 songs from the focal population from this analysis because their geographic distance was, by definition, zero for every song. We included bird ID as a random effect. We initially included recording sources (Macaulay Library, Xeno-canto or personal collection) as a fixed effect, but it was not a significant predictor. Therefore, we removed it from final models for simplicity. To determine if the playback exemplars from different geographic regions were significantly different from one another in song-level principal component scores and acoustic divergence from the focal population, we used non-parametric pairwise Wilcoxon rank-sum tests implemented with the pairwise.wilcox.test function. To assess differences in responses to the playbacks, we used mixed models implemented in the package lme4. We included focal pair ID, playback ID (because we presented playbacks more than once), and recording ID (because we used different songs from the same recording for some playbacks) as random effects in all models. We initially included playback order and year as fixed effects in models, but they were not significant. Therefore, we removed them from the final models for simplicity. We assessed the probability of approaching and singing as binomial response variables for all playback trials (n = 90 trials). For these binary responses, we used generalized linear mixed logistic regression models, using binomial distributions and logit-link functions. For the subset of trials in which a bird approached (n = 48 trials), we assessed the strength of approach behavior based on latency to approach within 20 m and time spent within 20 m of the speaker. For the subset of trials in which a bird sang (n = 52 trials), we assessed the strength of singing behavior based on latency to first song and number of songs. For models assessing strength of response, we used linear mixed models. We evaluated effect sizes using Cohen’s d and 95% confidence intervals for the strength of response variables following Nakagawa and Cuthill (2007). To put our playback results for Verdin in the context of previous studies on other species, we calculated Fisher’s transformation of the correlation coefficients (Zr) for all response variables following Parker et al. (2018). We binned Verdin response variables into movement or vocal response types and calculated a mean Zr value for each type so we could compare Verdin Zr values to the equivalent values calculated for 45 species in Parker et al. (2018), using the data deposited in Open Science Framework doi:10.17605/OSF.IO/W2MVP. We did not conduct statistical analyses on Zr values because a formal analysis of this dataset is the focus of Parker et al. (2018) and is beyond the scope of this paper. In their analysis, Parker et al. (2018) found that movement responses tended to have larger effect sizes than vocal responses, and that species living without confamilials tended to have smaller effect sizes than species living with congeners or confamilials. This provided us with a framework for qualitative comparison of Verdin Zr values in relation to the Zr values of the 45 species included in Parker et al. (2018). RESULTS Acoustic Analysis Songs were distinguishable as originating from 1 of the 4 geographical regions with 86% accuracy based on the linear discriminant function analysis with jackknifed validation (Table 3). The likelihood of being correctly assigned was high (~90%) for Sonoran, Chihuahuan, and Baja songs, but lower for Sierra Madre songs (54% classified correctly). Sierra Madre songs were most often misclassified as Sonoran songs (76% of misclassifications). Of the 17 songs used for playbacks, 70% were classified correctly: 5/5 Chihuahuan playbacks, 6/9 Sonoran playbacks, and 1/3 Baja playbacks. The 2 misclassified Baja playbacks were classified as Sonoran and the 3 misclassified Sonoran playbacks were classified as Sierra Madre (1) or Baja (2). These misclassifications are consistent with the overlap in song characteristics evident in Figure 3B. Table 3. Coefficients of linear discriminants from the discriminant function analysis. . ld1 . ld2 . ld3 . Proportion of trace 0.68 0.18 0.14 Song length (s) –0.22 0.15 –0.06 High frequency (Hz) 0.49 –0.41 –1.16 Low frequency (Hz) –0.11 0.26 0.96 Note shape pc1 0.51 –1.18 0.23 Note shape pc2 –0.39 –0.20 –0.14 Note shape pc3 1.10 0.77 1.18 . ld1 . ld2 . ld3 . Proportion of trace 0.68 0.18 0.14 Song length (s) –0.22 0.15 –0.06 High frequency (Hz) 0.49 –0.41 –1.16 Low frequency (Hz) –0.11 0.26 0.96 Note shape pc1 0.51 –1.18 0.23 Note shape pc2 –0.39 –0.20 –0.14 Note shape pc3 1.10 0.77 1.18 Open in new tab Table 3. Coefficients of linear discriminants from the discriminant function analysis. . ld1 . ld2 . ld3 . Proportion of trace 0.68 0.18 0.14 Song length (s) –0.22 0.15 –0.06 High frequency (Hz) 0.49 –0.41 –1.16 Low frequency (Hz) –0.11 0.26 0.96 Note shape pc1 0.51 –1.18 0.23 Note shape pc2 –0.39 –0.20 –0.14 Note shape pc3 1.10 0.77 1.18 . ld1 . ld2 . ld3 . Proportion of trace 0.68 0.18 0.14 Song length (s) –0.22 0.15 –0.06 High frequency (Hz) 0.49 –0.41 –1.16 Low frequency (Hz) –0.11 0.26 0.96 Note shape pc1 0.51 –1.18 0.23 Note shape pc2 –0.39 –0.20 –0.14 Note shape pc3 1.10 0.77 1.18 Open in new tab Including geographic region as a fixed effect significantly improved the fit of linear mixed models explaining acoustic divergence (Euclidean distance) from the focal population (full model df = 7, AIC = 770.9 vs. model without geographic region df = 4, AIC = 778.2, χ 2 = 14.10, P = 0.003) but including geographic distance from focal population did not (full model df = 7, AIC = 770.9 vs. model without geographic distance df = 6, AIC = 769.8, χ 2 = 1.72, P = 0.189). After correcting for multiple comparisons using the Benjamini-Hochberg adjustment, Baja (n = 3) and Chihuahuan (n = 5) playbacks were significantly different from Sonoran (n = 9) playbacks in at least 1 principal component score (song pc1: Sonoran mean = 0.082, Chihuahuan mean = 2.85, Wilcoxon P = 0.003; song pc3: Sonoran mean = 0.483, Baja mean = –0.854, Wilcoxon P = 0.027). The acoustic divergence (Euclidean distance) of Baja playbacks was not significantly different from the acoustic divergence of Sonoran playbacks (Sonoran mean = 1.86, Baja mean = 2.37, Wilcoxon P = 0.209), but the acoustic divergence of Chihuahuan playbacks was significantly greater than the acoustic divergence of Sonoran playbacks (Chihuahuan mean = 3.50, Wilcoxon P = 0.036). Playback Analysis Focal birds were significantly more likely to approach Sonoran, Baja, and Chihuahuan playbacks than confamilial negative control playbacks and showed no difference in approach tendency between Sonoran, Baja, and Chihuahuan playbacks (Tables 4 and 5, Figure 4). We did not find differences between the Sonoran positive controls and confamilial negative controls and therefore did not have sufficient power to detect differences between treatments in the probability of singing (Tables 4 and 5, Figure 4), nor in 3 of the variables representing the strength of response (time near speaker, latency to approach, and latency to sing; Tables 4 and 5, Figure 5). Verdins did, however, sing significantly more songs during Sonoran playbacks than Chihuahuan and confamilial negative control playbacks (Tables 4 and 5, Figure 5) and had a non-significant tendency to sing more songs to Sonoran playbacks than to Baja playbacks (Tables 4 and 5, Figure 5). Table 4. Mean values for behavioral responses to each playback type. . Sonoran . Baja . Chihuahuan . Confamilial . (n = 30) (n = 12) (n = 18) (n = 30) Approach (binomial)a 0.70 0.58 0.61 0.30 Sing (binomial)a 0.67 0.58 0.56 0.50 (n = 21) (n = 7) (n = 11) (n = 9) Latency to 20 m (s)b 113 77 86 64 Time spent 20 m (s)b 221 232 153 228 (n = 20) (n = 7) (n = 10) (n = 15) Latency to sing (s)c 81 145 102 141 Number of songsc 16 9 7 9 . Sonoran . Baja . Chihuahuan . Confamilial . (n = 30) (n = 12) (n = 18) (n = 30) Approach (binomial)a 0.70 0.58 0.61 0.30 Sing (binomial)a 0.67 0.58 0.56 0.50 (n = 21) (n = 7) (n = 11) (n = 9) Latency to 20 m (s)b 113 77 86 64 Time spent 20 m (s)b 221 232 153 228 (n = 20) (n = 7) (n = 10) (n = 15) Latency to sing (s)c 81 145 102 141 Number of songsc 16 9 7 9 aMeans based on all trials. bMeans based on the subset of trials in which a bird approached. cMeans based on the subset of trials in which a bird sang. Open in new tab Table 4. Mean values for behavioral responses to each playback type. . Sonoran . Baja . Chihuahuan . Confamilial . (n = 30) (n = 12) (n = 18) (n = 30) Approach (binomial)a 0.70 0.58 0.61 0.30 Sing (binomial)a 0.67 0.58 0.56 0.50 (n = 21) (n = 7) (n = 11) (n = 9) Latency to 20 m (s)b 113 77 86 64 Time spent 20 m (s)b 221 232 153 228 (n = 20) (n = 7) (n = 10) (n = 15) Latency to sing (s)c 81 145 102 141 Number of songsc 16 9 7 9 . Sonoran . Baja . Chihuahuan . Confamilial . (n = 30) (n = 12) (n = 18) (n = 30) Approach (binomial)a 0.70 0.58 0.61 0.30 Sing (binomial)a 0.67 0.58 0.56 0.50 (n = 21) (n = 7) (n = 11) (n = 9) Latency to 20 m (s)b 113 77 86 64 Time spent 20 m (s)b 221 232 153 228 (n = 20) (n = 7) (n = 10) (n = 15) Latency to sing (s)c 81 145 102 141 Number of songsc 16 9 7 9 aMeans based on all trials. bMeans based on the subset of trials in which a bird approached. cMeans based on the subset of trials in which a bird sang. Open in new tab Table 5. Results of mixed models relating focal bird responses to the fixed effect of playback type. Focal pair ID, playback ID, and recording ID were incorporated into all models as random effects. P-values are for comparisons of the fixed effect levels to Sonoran playbacks (the positive control). Confamilial playbacks represent a heterospecific negative control. P-values ≤ 0.05 are highlighted in bold. . Fixed effect . Estimate . z/t . P . Approach (binomial)a Intercept 0.98 2.07 0.039 Baja –0.68 –0.84 0.399 Chihuahua –0.40 –0.59 0.559 Confamilial –1.97 –2.93 0.003 Sing (binomial)a Intercept 0.83 1.70 0.089 Baja –0.34 –0.42 0.678 Chihuahua –0.62 –0.86 0.388 Confamilial –0.85 –1.33 0.184 Latency to 20 m (s)b Intercept 113.09 6.85 0.000 Baja –32.77 –1.08 0.280 Chihuahua –24.84 –0.98 0.329 Confamilial –43.56 –1.57 0.117 Time spent 20 m (s)b Intercept 222.74 4.91 0.000 Baja 9.69 0.11 0.916 Chihuahua –69.33 –0.89 0.371 Confamilial 4.77 0.06 0.953 Latency to sing (s)c Intercept 83.35 3.84 0.000 Baja 59.87 1.48 0.140 Chihuahua 24.08 0.68 0.500 Confamilial 59.17 1.89 0.059 Number of songsc Intercept 16.35 7.62 0.000 Baja –6.92 –1.64 0.101 Chihuahua –9.85 –2.65 0.008 Confamilial –7.22 –2.20 0.028 . Fixed effect . Estimate . z/t . P . Approach (binomial)a Intercept 0.98 2.07 0.039 Baja –0.68 –0.84 0.399 Chihuahua –0.40 –0.59 0.559 Confamilial –1.97 –2.93 0.003 Sing (binomial)a Intercept 0.83 1.70 0.089 Baja –0.34 –0.42 0.678 Chihuahua –0.62 –0.86 0.388 Confamilial –0.85 –1.33 0.184 Latency to 20 m (s)b Intercept 113.09 6.85 0.000 Baja –32.77 –1.08 0.280 Chihuahua –24.84 –0.98 0.329 Confamilial –43.56 –1.57 0.117 Time spent 20 m (s)b Intercept 222.74 4.91 0.000 Baja 9.69 0.11 0.916 Chihuahua –69.33 –0.89 0.371 Confamilial 4.77 0.06 0.953 Latency to sing (s)c Intercept 83.35 3.84 0.000 Baja 59.87 1.48 0.140 Chihuahua 24.08 0.68 0.500 Confamilial 59.17 1.89 0.059 Number of songsc Intercept 16.35 7.62 0.000 Baja –6.92 –1.64 0.101 Chihuahua –9.85 –2.65 0.008 Confamilial –7.22 –2.20 0.028 aModels based on all trials (n = 90). bModels based on the subset of trials in which a bird approached (n = 48). cModels based on the subset of trials in which a bird sang (n = 52). Open in new tab Table 5. Results of mixed models relating focal bird responses to the fixed effect of playback type. Focal pair ID, playback ID, and recording ID were incorporated into all models as random effects. P-values are for comparisons of the fixed effect levels to Sonoran playbacks (the positive control). Confamilial playbacks represent a heterospecific negative control. P-values ≤ 0.05 are highlighted in bold. . Fixed effect . Estimate . z/t . P . Approach (binomial)a Intercept 0.98 2.07 0.039 Baja –0.68 –0.84 0.399 Chihuahua –0.40 –0.59 0.559 Confamilial –1.97 –2.93 0.003 Sing (binomial)a Intercept 0.83 1.70 0.089 Baja –0.34 –0.42 0.678 Chihuahua –0.62 –0.86 0.388 Confamilial –0.85 –1.33 0.184 Latency to 20 m (s)b Intercept 113.09 6.85 0.000 Baja –32.77 –1.08 0.280 Chihuahua –24.84 –0.98 0.329 Confamilial –43.56 –1.57 0.117 Time spent 20 m (s)b Intercept 222.74 4.91 0.000 Baja 9.69 0.11 0.916 Chihuahua –69.33 –0.89 0.371 Confamilial 4.77 0.06 0.953 Latency to sing (s)c Intercept 83.35 3.84 0.000 Baja 59.87 1.48 0.140 Chihuahua 24.08 0.68 0.500 Confamilial 59.17 1.89 0.059 Number of songsc Intercept 16.35 7.62 0.000 Baja –6.92 –1.64 0.101 Chihuahua –9.85 –2.65 0.008 Confamilial –7.22 –2.20 0.028 . Fixed effect . Estimate . z/t . P . Approach (binomial)a Intercept 0.98 2.07 0.039 Baja –0.68 –0.84 0.399 Chihuahua –0.40 –0.59 0.559 Confamilial –1.97 –2.93 0.003 Sing (binomial)a Intercept 0.83 1.70 0.089 Baja –0.34 –0.42 0.678 Chihuahua –0.62 –0.86 0.388 Confamilial –0.85 –1.33 0.184 Latency to 20 m (s)b Intercept 113.09 6.85 0.000 Baja –32.77 –1.08 0.280 Chihuahua –24.84 –0.98 0.329 Confamilial –43.56 –1.57 0.117 Time spent 20 m (s)b Intercept 222.74 4.91 0.000 Baja 9.69 0.11 0.916 Chihuahua –69.33 –0.89 0.371 Confamilial 4.77 0.06 0.953 Latency to sing (s)c Intercept 83.35 3.84 0.000 Baja 59.87 1.48 0.140 Chihuahua 24.08 0.68 0.500 Confamilial 59.17 1.89 0.059 Number of songsc Intercept 16.35 7.62 0.000 Baja –6.92 –1.64 0.101 Chihuahua –9.85 –2.65 0.008 Confamilial –7.22 –2.20 0.028 aModels based on all trials (n = 90). bModels based on the subset of trials in which a bird approached (n = 48). cModels based on the subset of trials in which a bird sang (n = 52). Open in new tab Figure 4. Open in new tabDownload slide Proportion of birds approaching and singing by playback type. Numbers in bars indicate number of trials. Error bars are binomial standard errors. Asterisk indicates significant difference from other bars. Figure 4. Open in new tabDownload slide Proportion of birds approaching and singing by playback type. Numbers in bars indicate number of trials. Error bars are binomial standard errors. Asterisk indicates significant difference from other bars. Figure 5. Open in new tabDownload slide Forest plot of Cohen’s d effect sizes and 95% confidence intervals comparing strength of response to each playback type to the Sonoran positive control. Figure 5. Open in new tabDownload slide Forest plot of Cohen’s d effect sizes and 95% confidence intervals comparing strength of response to each playback type to the Sonoran positive control. The mean effect sizes (Zr) for Verdins in tests of local vs. foreign treatments aligned well with the mean effect sizes for other species that do not have sympatric confamilials (Figure 6). For movement responses, species without sympatric confamilials tended to have smaller effect sizes (mean Zr = 0.198, n = 6 species) compared to species with sympatric confamilials (mean Zr = 0.321, n = 38 species). Verdin had an even lower mean effect size for movement responses than species without sympatric confamilials (mean Zr = 0.074). For vocal responses, species without sympatric confamilials tended to have larger effect sizes (mean Zr = 0.378, n = 4 species) compared to species with sympatric confamilials (mean Zr = 0.210, n = 21 species). The mean effect size for Verdin vocal responses fell between the means for species with and without sympatric confamilials (mean Zr = 0.290). Figure 6. Open in new tabDownload slide Effect sizes (Zr) for trials comparing response to local and foreign songs for all species included in Parker et al. (2018) (circle-shaped points) and Verdins (diamond-shaped points). Effect sizes are species means binned into movement and vocal responses, and species are divided into those with and without sympatric confamilials. Figure 6. Open in new tabDownload slide Effect sizes (Zr) for trials comparing response to local and foreign songs for all species included in Parker et al. (2018) (circle-shaped points) and Verdins (diamond-shaped points). Effect sizes are species means binned into movement and vocal responses, and species are divided into those with and without sympatric confamilials. DISCUSSION Verdins showed geographic variation in song that corresponded with different desert habitat regions in their range, and they showed discrimination toward foreign conspecific songs based on their vocal behavior, but no discrimination based on their movement behavior. These results suggest that although Verdins can perceive differences between foreign and local songs, they will still approach foreign songs just as readily as they approach local-sounding songs. The geographic variation in Verdin whistle songs mapped well onto the ecologically distinct desert habitat regions within their range. These regions are associated with subspecies of Verdins, or transition zones between subspecies (e.g., the Sierra Madre region), that are differentiated based on subtle plumage variations (Webster 2020). The general concordance between desert habitat, subspecies, and song suggests that there may be some reduction in dispersal between desert habitats that promotes song variation. The different deserts of the Southwest have vastly different plant communities, temperatures, and precipitation regimes (Shreve 1942, Archer and Predick 2008), so ecological differences or barriers such as the Cochise Filter Barrier (between the Sonoran and Chihuahuan deserts) may reduce dispersal between deserts (e.g., Provost et al. 2018). In Verdins, mitochondrial DNA does not show a signal of reduced gene flow between Sonoran and Chihuahuan deserts (Zink et al. 2001), but dispersal may be reduced enough to lead to variation in song despite some gene flow. Additionally, phenotypic clines can exist despite or in discordance with patterns of gene flow (e.g., Greig and Webster 2013, Poesel et al. 2017, Hooper et al. 2019). The variation in song that we documented across the Verdin range was represented reasonably well by the playback exemplars in our experiment, with the exception of 2 playback exemplars from Baja that overlapped in acoustic space with Sonoran songs (Figure 3B). Despite Baja and Chihuahuan playbacks originating from over 900 km away from the focal population, being from different subspecies than the focal population, and being largely acoustically divergent from the positive control Sonoran playbacks, Verdins were equally likely to approach Sonoran, Baja, and Chihuahuan playbacks (Table 5, Figure 4). In contrast, Verdins sang significantly more songs in response to Sonoran playbacks than to Chihuahuan playbacks and had a non-significant trend of singing more in response to Sonoran playbacks than to Baja playbacks (Table 5, Figure 5). Taken together, these results align well with what appears to be typical response patterns for local vs. foreign song treatments in other species that do not have sympatric confamilials (Figure 6). By analyzing the effect sizes presented in Figure 6 in relation to study design, a multitude of life history traits, and evolution history, Parker et al. (2018) found a subtle pattern that species without sympatric confamilials tended to have weaker discrimination between local and foreign song than species with sympatric confamilials/congeners. They also found that across all species movement responses tended to have larger effect sizes than vocal responses. Verdin, therefore, are unusual compared to most species because they approach foreign songs, but they are not unusual compared to species without sympatric confamilials. One difference in our study compared to the studies used in Parker et al. (2018) and presented in Figure 6 is that our positive control Sonoran playbacks were not recorded from the focal population, so are not representative of true local songs. Although we cannot determine if Verdins respond even more strongly to songs from the focal population because we did not test these, we are confident that our Sonoran playbacks represent songs quantitatively similar to focal population songs; the mean acoustic divergence of Sonoran songs from the focal population was almost equal to the mean acoustic divergence present within the focal population (Table 1). Additionally, the Sonoran song pc1 and pc2 scores almost completely overlap with the scores from the focal population (Figure 3B). Using Sonoran playbacks originating from non-local populations as a positive control had the advantage of being unfamiliar, so any difference in response to Sonoran vs. treatment playbacks was based on acoustic structure rather than prior experience of the focal birds. How can we reconcile the Verdins’ discrimination in vocal response with their lack of discrimination in movement response? The answer may be that different behaviors indicate different things about an individual’s motivation to respond (e.g., de Kort et al. 2009) or the tactics an individual uses to respond (e.g., Nowicki et al. 2002). It may be that for Verdins, approaching represents a useful behavioral response for a wide range of signals (e.g., investigation), but increasing song rate represents a useful behavior for a narrower subset of local-sounding signals (e.g., aggression). Although we cannot determine the precise functions and corresponding motivations of the behaviors Verdins exhibited in this study other than to assume they are related to territory defense, it is not unusual to have variation and inconsistency when multiple behavioral responses are measured in playback studies (McGregor 1992). Approaching foreign conspecific songs may not be costly for Verdins if the lack of confamilials in North America eliminates any risk of responding to similar-sounding heterospecifics. An example of such selective pressure release is found in singing mice (Scotinomys teguina), in which allopatric populations have stronger behavioral responses to heterospecifics than sympatric populations because allopatric populations are spared agonistic character displacement (Pasch et al. 2017). Another example is seen in African tinkerbirds, in which allopatric populations have stronger behavioral responses to heterospecifics than sympatric populations because sympatric populations have experienced character displacement and selection to avoid interactions (Kirschel et al. 2009). In these examples, a lack of exposure to divergent signals, and a lack of selective pressure to avoid them, seems to lead to equal responses to divergent and local signals (see also Dingle et al. 2010). Alternatively, approaching foreign conspecific songs may be useful for Verdins if they tend to encounter unfamiliar songs during their lifetime and benefit from responding to them. For example, in Black-throated Blue Warblers (Setophaga caerulescens) southern populations tend to respond to northern songs, but northern populations do not respond to southern songs. One hypothesized reason for this is that southern populations are exposed to northern songs during migration, but not vice versa (Colbeck et al. 2010). Another example is in Dickcissels, which do not discriminate between local and foreign songs. Males are exposed to foreign songs when winter flocks aggregate, and males that do not conform to the local song style will settle and defend territories in the breeding season (Parra et al. 2017). In these examples, exposure to divergent signals, and selective pressure to respond to them, seems to lead to equal responses to divergent and local signals. Yet another reason for equal responses to foreign and local signals is illustrated in Grasshopper Sparrows (Ammodramus savannarum), which do not distinguish between foreign and local songs, presumably because they improvise rather than imitate their songs and therefore do not have strict discrimination thresholds based on familiarity (Soha et al. 2016). We do not have any reason to think that Verdins improvise their songs rather than learn them, but song acquisition has not been studied in Verdins so we cannot rule out this hypothesis. There are numerous other examples of species both recognizing and responding to divergent songs from sympatric heterospecifics, for example, if they compete for resources, interfere with mating, or cooperate in resource/predator defense (Martin and Martin 2001, Sedlacek et al. 2006, Drury et al. 2015, Reif et al. 2015, Johnson et al. 2018). Ultimately, the criteria for responding to divergent songs, either conspecific or heterospecific, are diverse and appear to depend upon the interactions (or lack thereof) of the individuals involved and the costs and benefits of responding. This work provides an example of a species that shows geographic variation in song and behavioral responses consistent with some level of recognition of unfamiliar, acoustically divergent, and geographically distant conspecific songs. When considering the large number of playback studies showing reduced approach responses to foreign conspecific song (Parker et al. 2018), we might consider Verdins to be on the permissive end of the spectrum. Our single-species study does not give us the inference power to determine if the cause of the permissiveness in approach behavior is because Verdins have no sympatric confamilials, but it offers a case study that is consistent with this hypothesis and encourages future comparative work that explores other hypotheses such as the likelihood of encountering unfamiliar songs in an individual’s lifetime. Future studies that assess geographic variation in song and conspecific recognition in more species with a variety of life histories and evolutionary trajectories will be valuable because the questions we discuss here must ultimately be answered in a comparative framework. ACKNOWLEDGMENTS Thank you to J. Hunter Reed, Liam Berigan, Alex Weibe, Allison Johnson, Joe Cacioppo, Daniel Hooper, Sara Kaiser, Dylan Meyer, Samantha Symon, Ivy Sandquist, and Isaac Krone for help collecting data in the field. Thank you to Chrissy Kondrat-Smith, Peter Holm, and the staff of Organ Pipe Cactus National Monument for logistical support in the field and for organizing permits. Thank you to Irby Lovette, David Bonter, Anne Marie Johnson, Chelsea Benson, and Holly Faulkner for being supportive of this work. Funding statement: Thank you to the Blanksteen Foundation, the Cornell Lab of Ornithology, and the University of Chicago for funding that made this work possible. Ethics statement: All work in this study was evaluated and approved by the Cornell University Animal Care and Use Committee and National Park Service Institutional Animal Care and Use Committee (Protocol #2015-0052). The study was conducted under appropriate Federal (Permit #23245), State (License #SP770320 and #SP735654), and National Park (Study #ORPI-00021) permits. Author contributions: E.I.G. conceived the idea, design, and hypotheses, E.I.G. and E.C.L. supervised research, E.I.G., M.L.W., and E.K. collected data, performed experiments, refined methodology and hypotheses, E.I.G. wrote the paper and analyzed the data, E.I.G., M.L.W., E.K., and E.C.L. revised and edited the paper. Data depository: All data used in this paper are included as Supplementary Material and available at Greig et al. (2021). LITERATURE CITED Allen , F. H . ( 1919 ). The evolution of bird song . The Auk 36 : 528 – 536 . Google Scholar OpenURL Placeholder Text WorldCat Amézquita , A. , S. V. Flechas, A. P. Lima, H. Gasser, and W. Hödl ( 2011 ). Acoustic interference and recognition space within a complex assemblage of dendrobatid frogs . Proceedings of the National Academy of Sciences USA 108 : 17058 – 17063 . Google Scholar OpenURL Placeholder Text WorldCat Archer , S. R. , and K. I. Predick ( 2008 ). Climate change and ecosystems of the Southwestern United States . Rangelands 30 : 23 – 28 . Google Scholar OpenURL Placeholder Text WorldCat Bates , D. , M. Maechler, B. Bolker, and S. Walker ( 2015 ). Fitting linear mixed-effects models using lme4 . Journal of Statistical Software 67 : 1 – 48 . Google Scholar OpenURL Placeholder Text WorldCat Bioacoustics Research Program ( 2004 ). Raven Pro: Interactive Sound Analysis Software (version 1.5) . The Cornell Lab of Ornithology , Ithaca, NY, USA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Bird Studies Canada and NABCI ( 2014 ). Bird Conservation Regions. Bird Studies Canada, North American Bird Conservation Initiative . https://nabci-us.org/resources/bird-conservation-regions-map/ Colbeck , G. J. , T. S. Sillett, and M. S. Webster ( 2010 ). Asymmetric discrimination of geographical variation in song in a migratory passerine . Animal Behaviour 80 : 311 – 318 . Google Scholar OpenURL Placeholder Text WorldCat Dingle , C. , J. W. Poelstra, W. Halfwerk, D. M. Brinkhuizen, and H. Slabbekoorn ( 2010 ). Asymmetric response patterns to subspecies-specific song differences in allopatry and parapatry in the gray-breasted wood-wren . Evolution 64 : 3537 – 3548 . Google Scholar OpenURL Placeholder Text WorldCat Drury , J. P. , K. W. Okamoto, C. N. Anderson, and G. F. Grether ( 2015 ). Reproductive interference explains persistence of aggression between species . Proceedings of the Royal Society B 282 : 20142256 . Google Scholar OpenURL Placeholder Text WorldCat Freeman , B. G. , and G. A. Montgomery ( 2017 ). Using song playback experiments to measure species recognition between geographically isolated populations: A comparison with acoustic trait analysis . The Auk: Ornithological Advances 134 : 857 – 870 . Google Scholar OpenURL Placeholder Text WorldCat Greig , E. I. , E. Kinnebrew, M. L. Witynski, and E. C. Larsen ( 2021 ). Data from: A desert songbird with no confamilials in the Western Hemisphere (Verdin, Auriparus flaviceps) investigates divergent conspecific songs . Ornithology 138 : 1 – 13 . doi:10.5061/dryad.xksn02vfq Google Scholar OpenURL Placeholder Text WorldCat Greig , E. I. , and M. S. Webster ( 2013 ). Spatial decoupling of song and plumage generates novel phenotypes between two avian subspecies . Behavioral Ecology 24 : 1004 – 1013 . Google Scholar OpenURL Placeholder Text WorldCat Grether , G. F. , N. Losin, C. N. Anderson, and K. Okamoto ( 2009 ). The role of interspecific interference competition in character displacement and the evolution of competitor recognition . Biological Reviews of the Cambridge Philosophical Society 84 : 617 – 635 . Google Scholar OpenURL Placeholder Text WorldCat Hamao , S . ( 2016 ). Asymmetric response to song dialects among bird populations: The effect of sympatric related species . Animal Behaviour 119 : 143 – 150 . Google Scholar OpenURL Placeholder Text WorldCat Hooper , D. M. , S. C. Griffith, and T. D. Price ( 2019 ). Sex chromosome inversions enforce reproductive isolation across an avian hybrid zone . Molecular Ecology 28 : 1246 – 1262 . Google Scholar OpenURL Placeholder Text WorldCat Irwin , D. E. , and T. Price ( 1999 ). Sexual imprinting, learning and speciation . Heredity 82 : 347 – 354 . Google Scholar OpenURL Placeholder Text WorldCat Johnson , A. E. , C. Masco, and S. Pruett-Jones ( 2018 ). Song recognition and heterospecific associations between 2 fairy-wren species (Maluridae) . Behavioral Ecology 29 : 821 – 832 . Google Scholar OpenURL Placeholder Text WorldCat Kirschel , A. N. , D. T. Blumstein, and T. B. Smith ( 2009 ). Character displacement of song and morphology in African tinkerbirds . Proceedings of the National Academy of Sciences USA 106 : 8256 – 8261 . Google Scholar OpenURL Placeholder Text WorldCat de Kort , S. R. , E. R. Eldermire, E. R. Cramer, and S. L. Vehrencamp ( 2009 ). The deterrent effect of bird song in territory defense . Behavioral Ecology: Official Journal of the International Society for Behavioral Ecology 20 : 200 – 206 . Google Scholar OpenURL Placeholder Text WorldCat Marler , P . ( 1958 ). Bird songs and mate selection . Animal Behaviour 6 : 348 – 367 . Google Scholar OpenURL Placeholder Text WorldCat Martin , P. R. , and T. E. Martin ( 2001 ). Behavioral interactions between coexisting species: Song playback experiments with wood warblers . Ecology 82 : 207 – 218 . Google Scholar OpenURL Placeholder Text WorldCat Maruvka , Y. E. , N. M. Shnerb, D. A. Kessler, and R. E. Ricklefs ( 2013 ). Model for macroevolutionary dynamics . Proceedings of the National Academy of Sciences USA 110 : E2460 – E2469 . Google Scholar OpenURL Placeholder Text WorldCat McCreedy , C. , and C. van Riper , III ( 2015 ). Drought-cased delay in nesting of Sonoran Desert birds and its facilitation of parasite- and predator-mediated variation in reproductive success . The Auk 132 : 235 – 247 . Google Scholar OpenURL Placeholder Text WorldCat McGregor , P. K . ( 1992 ). Quantifying responses to playback: One, many, or composite multivariate measures? In Playback and Studies of Animal Communication ( P. K. McGregor, Editor). Plenum Press , New York, NY, USA . pp. 79 – 96 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Nakagawa , S. , and I. C. Cuthill ( 2007 ). Effect size, confidence interval and statistical significance: A practical guide for biologists . Biological Reviews of the Cambridge Philosophical Society 82 : 591 – 605 . Google Scholar OpenURL Placeholder Text WorldCat Nowicki , S. , W. A. Searcy, T. Kruegar, and M. Hughes ( 2002 ). Individual variation in response to simulated territorial challenges among territory-holding song sparrows . Journal of Avian Biology 33 : 253 – 259 . Google Scholar OpenURL Placeholder Text WorldCat Nychka , D. , R. Furrer, J. Paige, and S. Sain ( 2015 ). Fields: Tools for spatial data. R package version 9.0 . https://github.com/NCAR/Fields Parker , T. H. , E. I. Greig, S. Nakagawa, M. Parra, and A. C. Dalisio ( 2018 ). Subspecies status and methods explain strength of response to local versus foreign song by oscine birds in meta-analysis . Animal Behaviour 142 : 1 – 17 . Google Scholar OpenURL Placeholder Text WorldCat Parra , M. , A. C. Dalisio, W. E. Jensen, and T. H. Parker ( 2017 ). Male territorial aggression does not drive conformity to local vocal culture in a passerine bird . Ethology 123 : 800 – 810 . Google Scholar OpenURL Placeholder Text WorldCat Pasch , B. , R. Sanford, and S. M. Phelps ( 2017 ). Agonistic character displacement in social cognition of advertisement signals . Animal Cognition 20 : 267 – 273 . Google Scholar OpenURL Placeholder Text WorldCat Podos , J. , and P. S. Warren ( 2007 ). The evolution of geographic variation in birdsong . Advances in the Study of Behavior 37 : 403 – 458 . Google Scholar OpenURL Placeholder Text WorldCat Poesel , A. , A. C. Fried, L. Miller, H. L. Gibbs, J. A. Soha, D. A. Nelson ( 2017 ). High levels of gene flow among song dialect populations of the Puget Sound White-crowned Sparrow . Ethology 123 : 581 – 592 . Google Scholar OpenURL Placeholder Text WorldCat Provost , K. L. , W. M. Mauck , 3rd, and B. T. Smith ( 2018 ). Genomic divergence in allopatric Northern Cardinals of the North American warm deserts is linked to behavioral differentiation . Ecology and Evolution 8 : 12456 – 12478 . Google Scholar OpenURL Placeholder Text WorldCat R Core Team ( 2017 ). R: A language and environment for statistical computing . R Foundation for Statistical Computing , Vienna, Austria. https://www.R-project.org/ . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Reif , J. , M. Jiran, R. Reifova, J. Vokurkova, P. Dolata, A. Petrusek, and T. Petruskova ( 2015 ). Interspecific territoriality in two songbird species: Potential role of song convergence in male aggressive interactions . Animal Behaviour 104 : 131 – 136 . Google Scholar OpenURL Placeholder Text WorldCat Rothstein , S. I. , and R. C. Fleischer ( 1987 ). Vocal dialects and their possible relation to honest status signaling in the Brown-headed Cowbird . The Condor 89 : 1 – 23 . Google Scholar OpenURL Placeholder Text WorldCat Searcy , W. A. , S. Nowicki, M. Hughes, and S. Peters ( 2002 ). Geographic song discrimination in relation to dispersal distances in song sparrows . The American Naturalist 159 : 221 – 230 . Google Scholar OpenURL Placeholder Text WorldCat Sedlacek , O. , B. Cikanova, and R. Fuchs ( 2006 ). Heterospecific rival recognition in the Black Redstart (Phoenicurus ochruros) . Ornis Fennica 83 : 153 – 161 . Google Scholar OpenURL Placeholder Text WorldCat Shreve , F . ( 1942 ). The desert vegetation of North America . Botanical Review 8 : 195 – 246 . Google Scholar OpenURL Placeholder Text WorldCat Soha , J. A. , A. Poesel, D. A. Nelson, and B. Lohr ( 2016 ). Non-salient geographic variation in birdsong in a species that learns by improvisation . Ethology 122 : 343 – 353 . Google Scholar OpenURL Placeholder Text WorldCat Symes , L. B . ( 2014 ). Community composition affects the shape of mate response functions . Evolution 68 : 2005 – 2013 . Google Scholar OpenURL Placeholder Text WorldCat Taft , B . ( 2011 ). The role of dawn song in Tree Swallows and its place in the diversity of oscine song learning. Ph.D. dissertation, University of Massachusetts , Amherst, MA, USA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Taylor , W. K . ( 1971 ). Breeding biology study of Verdin, Auriparus flaviceps (Sundevall) in Arizona . American Midland Naturalist 85 : 289 – 328 . Google Scholar OpenURL Placeholder Text WorldCat Turcokova , L. , V. Pavel, B. Chutny, A. Petrusek, and T. Petruskova ( 2011 ). Differential response of males of a subarctic population of Bluethroat Luscinia svecica svecica to playbacks of their own and foreign subspecies . Journal of Ornithology 152 : 975 – 982 . Google Scholar OpenURL Placeholder Text WorldCat Venables , W. N. , and B. D. Ripley ( 2002 ). Modern Applied Statistics with S , fourth edition. Springer , New York, NY, USA . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Webster , M. D . ( 2020 ). Verdin (Auriparus flaviceps), version 1.0 . In Birds of the World ( A. F. Poole and F. B. Gill, Editors). Cornell Lab of Ornithology , Ithaca, NY, USA . https://doi.org/10.2173/bow.verdin.01 Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Weir , J. T. , and T. D. Price ( 2019 ). Song playbacks demonstrate slower evolution of song discrimination in birds from Amazonia than from temperate North America . PLoS Biology 17 : e3000478 . Google Scholar OpenURL Placeholder Text WorldCat Wiley , R. H . ( 1994 ). Errors, exaggeration, and deception in animal communication. In Behavioral Mechanisms in Ecology ( L. Real, Editor). University of Chicago Press , Chicago, IL, USA . pp. 157 – 189 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Wilkins , M. R. , N. Seddon, and R. J. Safran ( 2013 ). Evolutionary divergence in acoustic signals: Causes and consequences . Trends in Ecology & Evolution 28 : 156 – 166 . Google Scholar OpenURL Placeholder Text WorldCat Zink , R. M. , A. E. Kessen, T. V. Line, and R. C. Blackwell-Rago ( 2001 ). Comparative phylogeography of some aridland bird species . The Condor 103 : 1 – 10 . Google Scholar OpenURL Placeholder Text WorldCat Copyright © American Ornithological Society 2021. All rights reserved. For permissions, e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

OrnithologyOxford University Press

Published: May 8, 2021

References