Avian Behavior, Ecology, and Evolution

INTRODUCTION

Birdsong has long captured human interest and has been extensively studied in various contexts (Catchpole and Slater 2008). Songs can provide information about the singer and serve as signals of individual condition and intent (Dolby et al. 2005, Searcy and Beecher 2009, Lapierre et al. 2011). Individual birds often have a repertoire of song types, with repertoire size, i.e., the number of song types sung by a bird, and the extent of repertoire sharing with nearby conspecifics varying across species (Kroodsma 1977), populations (Hughes et al. 1998, Peters et al. 2000) and individuals within populations (Kroodsma 1975, Lapierre et al. 2011). Repertoire size and song sharing may, in some cases, influence intrasexual aggression (Beecher et al. 2000, Pitt 2018) and mate choice (Buchanan and Catchpole 1997, Nelson and Poesel 2013). The repertoire-size hypothesis (Beecher and Brenowitz 2005) posits that repertoire size positively reflects singer quality and larger repertoires are preferred by potential mates, whereas the general-sharing hypothesis, which is non-mutually exclusive with the repertoire-size hypothesis, says that song-matching or song-avoidance with nearby conspecifics is facilitated by a large repertoire and offers benefits in competition, as it does in Song Sparrows (Melospiza melodia; Beecher et al. 2000, Beecher 2008), and mating success, as it does in Indigo Buntings (Passerina cyanea; Payne et al. 1988).

Troglodyte wren species show considerable diversity in repertoire size and offer a good model for examining patterns of song-type repertoire use and sharing. Within this group, Rock Wrens (Salpinctes obsoletus) are sexually monomorphic (Benedict et al. 2021) territorial songbirds (Warning and Benedict 2015), with some migratory and some sedentary populations (Benedict et al. 2021). In the breeding season, they typically occupy open and arid habitats with rocky outcrops, cliffs, and/or escarpments, and build their nests in natural rock cavities (Warning and Benedict 2016, Benedict et al. 2021). Rock Wren males defend territories with song and sing with a flexible delivery pattern; song types are usually sung several times interspersed with other song types following no fixed pattern (Kroodsma 1975). Males show large and variable repertoire sizes (Kroodsma 1975, Benedict and Warning 2017). Whereas sympatric Canyon Wrens (Catherpes mexicanus) sing only five to six song types (Benedict et al. 2013), some Rock Wrens sing over 120 song types, with moderate repertoire sharing, e.g., ~42% of their repertoire shared with other birds in one study population (Benedict and Warning 2017). Playback experiments of larger Rock Wren repertoires elicited a greater response in singing by resident males, which suggests that repertoire size plays a role in male-male competition and territory defense (Pitt 2018), supporting the repertoire-size hypothesis. However, contrary to the expectation of the repertoire-size hypothesis that all song types should be sung in equal frequency, Rock Wrens sing some song types in their repertoires much more than others (Benedict and Warning 2017). These results highlight the value of a large repertoire while raising questions about what song types are most likely to be used and shared among individuals at the population level.

Network analysis has been used to examine transition patterns in song sequences of individual birds (Kaluthota et al. 2020) and is an effective method to examine song sharing (Potvin et al. 2019). We, therefore, used network analysis methods to examine song-type sharing among male Rock Wrens in northern Colorado; females of this species do not sing (Benedict et al 2021). We examined recordings of natural broadcast singing of male Rock Wrens in a migratory population after they returned in spring for breeding. In the same population, it was previously found that individual males use song types that maximize transmission in their environment (Benedict and Warning 2017) and that Rock Wrens order their song types in variable ways (Hedley et al. 2018). Detailed patterns of song-type sharing across the population, however, are yet to be examined. We aimed to answer three broad questions:

1. Are popular songs in the population preferred by individual birds?

According to the general-sharing hypothesis, overproducing song types that are sung by many birds in the population, i.e., popular songs (Table 1), may be beneficial by reducing aggression from conspecifics through the dear enemy effect (Briefer et al. 2008) and by allowing mutual assessment (Logue and Forstmeier 2008, Price and Yuan 2011). In this study population, Benedict and Warning (2017) found that individual Rock Wrens preferred different songs, with 10 of 12 birds having unique song types as their most used. Here, we expand on these analyses by asking, first, are most songs sung by many birds and, second, are popular songs sung more or less frequently than rare or unique songs by individual birds? Further, it is possible that not all birds follow the same strategy (Lapierre et al. 2011), so we examined the effect of repertoire size on the tendency to overproduce popular songs.

2. Is there non-random song sharing between bird repertoires, and what factors influence song sharing?

Song-type sharing might be more similar between immediate neighbors than others (Hill et al. 1999) and, if so, we expected similarity to decrease with increasing distance (Benedict et al. 2013). We constructed networks of song sharing between birds and examined whether some pairs of Rock Wrens shared more or less of their repertoire than expected by chance, hence non-random, and whether distance affected song sharing. Song sharing may be constrained by syntactic rules of song production or might reflect the complexity of repertoires (Benedict and Najar 2019, Searcy et al. 2023). Thus, we examined whether two measures of song production of individual birds affected song sharing in this dataset: the rate of singing songs, i.e., Delivery, and the proportion of song transitions that resulted in a switch in song type, i.e., Switching. We hypothesized that birds that sang at a faster rate and switched songs more often would be able to share more songs with other males.

3. Is there non-random co-occurrence of song types, and does it depend on song characteristics?

In the same population, individual males were found to preferentially sing songs from their personal repertoires that enhanced transmission: longer songs with lower frequencies, lower trill rates, and smaller bandwidths (Benedict and Warning 2017). Song types with these preferred characteristics may co-occur, i.e., such that both are present, within a bird’s repertoire more than expected by chance, hence non-randomly. Accordingly, we examined whether song co-occurrence networks were non-random, and the effects of song duration, low frequency, bandwidth, trill rate, and frequency inflections, i.e., switches from ascending to descending frequency or vice versa (Benedict and Warning 2017), on song popularity. Popular songs may show the same trend as individual song usage and have enhanced transmission. Alternatively, because songs shared by more birds were of lower performance using some song metrics, like trill rate, in other birds (Cardoso and Atwell 2016, Podos et al. 2016), song popularity and performance measures including trill rate and number of frequency inflections may be negatively correlated.

METHODS

Songs were recorded from 12 unbanded Rock Wrens in Larimer County, Colorado, USA, in spring 2013 and spring 2014. All the 12 birds were recorded in arid, exposed, and rocky habitats (Benedict and Warning 2017). Rock Wrens show high site-fidelity and large home ranges. Thus, we were confident about the identity of the singer while recording (Benedict and Warning 2017). Researchers recorded a mean of 1588 ± 418 (SD) songs per individual and categorized songs into 233 types based on visual appraisal of song characteristics, recorded song-type ordering through time, and measured acoustic song characteristics of each song type for every individual that sang it. Rock Wrens sing highly stereotyped songs that consist of repetitions of a single syllable or trill element, making them easy to classify to type visually based on syllable form (Figure 1). See Benedict and Warning (2017) for complete details on song recording and classification methodology; results reported here come from the same dataset, analyzed at the population rather than the individual level.

Popularity of a song type, its usage in individual birds, and the effect of repertoire size

Popular songs in the population may have high or low song usage in individual birds. We plotted the number of song types that were sung by different numbers of birds, and correlated the number of birds that sang a song type, hereafter referred to as the “popularity of a song type,” with the average proportion of usage, i.e., averaged across birds that sung the song (Table 1, Figure 2). If birds overproduced popular songs, we expected a positive correlation. If they instead preferred to sing unique songs, we expected a negative correlation. We also correlated the repertoire size of a bird with, first, the popularity of its top song type and, second, the weighted popularity of that bird’s songs. These terms are detailed in Table 1. We performed Spearman’s correlations in R version 4.2.3 (R Core Team 2023) for these comparisons.

Constructing song sharing and song co-occurrence networks and testing for non-randomness

We calculated three repertoire-level indices between all pairs of the 12 sampled Rock Wrens (Table 1) to measure song sharing between pairs of birds. Index 1, i.e., the Adjusted Jaccard index (Tracy and Baker 1999) = c / ((a + b + c) - d) where c is the number of shared song types, a is the number of song types unique to bird 1, i.e., number of song types in bird 1’s repertoire not shared with bird 2, b is the number of song types unique to bird 2, and d is the (absolute) difference in repertoire sizes of the two birds. Index 2 is a modification of the adjusted Jaccard index, where the songs were weighted by their popularity as 1 - Y_p, with Y_p being the proportion of birds that sang that song type (Potvin et al. 2019), such that rarer songs weighed more than popular songs. These two indices measure similarity; a value of 1 indicates complete sharing and 0 means the birds shared no song types. Finally, Index 3 is the Bray-Curtis dissimilarity index between pairs of birds (Bray and Curtis 1957). Index 3 = 1 - ((2 * C₁₂) / S₁ + S₂), where C₁₂ is the sum of lesser counts among bird 1 and bird 2 for each song type, summed across all song types, S₁ is total number of songs sung by bird 1, and S₂ is total number of songs sung by bird 2, again summed across all song types. Index 3 accounts for the number of times that songs were sung, whereas indices 1 and 2 look at just the presence/absence of song types in a bird’s repertoire, and is an index of dissimilarity, where 0 indicates the birds sang the same songs and in the same relative frequencies.

We also calculated co-occurrence indices between all pairs of the 233 song types, measuring how often they co-occur in the same bird’s repertoire. Index 4 or the Adjusted Jaccard index (Tracy and Baker 1999) = c / ((a + b + c) - d), where c is the number of birds that sang both song types, a is the number of birds that sang song type 1 but not song type 2, b is the number of birds that sang song type 2 but not song type 1, and d is the (absolute) difference between the number of birds that sang song type 1 and the number of birds that sang song type 2. Index 5 is the Bray-Curtis dissimilarity index between pairs of song types (Bray and Curtis 1957) = 1 - ((2 * C₁₂) / S₁ + S₂), where C₁₂ is the sum of lesser counts among song type 1 and song type 2 for each bird, summed across all birds, S₁ is the total number of times song 1 was sung, and S₂ is the total number of times song 2 was sung, summed across all birds. Index 4 measures similarity and only looked at the presence/absence of one or both song types in a bird’s repertoire. Index 5 is a dissimilarity index that also takes into account the number of times the two song types were sung by different birds (Table 1).

We randomized the pre-network data, i.e., data-stream permutations, creating random repertoires for the 12 birds while keeping the number of birds that sang a song type the same as the observed dataset (Potvin et al. 2019) for all the 233 song types. We created 10,000 permuted datasets, with 10,000 flips per permutation, and compared the coefficient of variation, i.e., CV, of the five observed indices with corresponding permuted datasets. P values for permutations were calculated as 1- (Number of permutations where the observed value was greater than the corresponding permuted value / Number of permutations). If P < 0.05, the real or observed CV value was considered significantly greater than those in the permuted datasets, indicating that the network is non-random, with some pairs of birds sharing more of their songs than other pairs in the song-type sharing networks, and/or some pairs of song types tending to co-occur more than other pairs in the song co-occurrence networks.

All non-random networks were visualized, and network density, i.e., overall connectedness, modularity, i.e., how clustered the network is, and degree of non-random song co-occurrence networks were measured using Gephi version 0.10.1 (Bastian et al. 2009). For the Bray-Curtis dissimilarity indices, i.e., Index 3 and Index 5, we used 1- index to visualize the networks, so that higher values represented higher similarity.

Effects of distance, delivery, and switching on song sharing

Using indices with non-random song sharing between birds, we examined the influence of several factors on song sharing. Distances between the central recording locations of birds were calculated in meters using GPS coordinates. We performed Mantel tests (Mantel 1967) to examine spatial patterns in song sharing. For each bird, we calculated Delivery, i.e., number of songs sung per minute of recording, and Switching, i.e., proportion of song transitions that resulted in a change in song type. These measures were previously found not to correlate with each other or with repertoire size in this dataset (Benedict and Najar 2019). We compared the observed Pearson’s correlation coefficients between the average index of a bird and the bird’s delivery and switching values to 1000 permuted datasets, permuted by node-label permutations, with 1000 flips of node labels in each permutation (Farine 2017). We expected birds with higher Delivery and Switching to have greater song sharing with others and, thus, positive correlations with Index 1 and Index 2, and negative correlations with Index 3. If P < 0.05 (Appendix 1), the observed correlation was considered significant.

Effect of song characteristics on popularity

Using measurements of each song type from all birds, we calculated an average value of (1) song duration in seconds; (2) song low frequency in kHz; (3) song bandwidth in kHz; (4) trill rate in syllable/second; and (5) frequency inflections for each song type. We performed a zero-truncated negative-binomial regression, because our response variable, popularity, was a count that cannot have zero value and was overdispersed, looking at the main effects of the five song characteristics on the popularity of the song type using the VGAM package in R (Yee 2010). If songs are shared across birds because of the same motivation, i.e., of enhancing transmission, as song usage within birds (Benedict and Warning 2017), then we expect to find the same patterns for song sharing, or popularity of the song, and overall song usage. If, on the other hand, popular songs were favored because they were easy to sing then they are predicted to have lower performance than unpopular songs; specifically, we expected popularity to decrease with increasing song duration, increase with increasing low frequency, and decrease with an increase in song bandwidth, trill rates, and with frequency inflections.

Mantel tests were performed using the ecodist package (Goslee and Urban 2007). Data manipulations and permutations were performed in MATLAB version R2023a (The MathWorks Inc., 2023).

RESULTS

Popularity of a song type, its usage in individual birds, and the effect of repertoire size

Most song types were sung by few birds (Average ± SD of popularity: 3.9185 ± 2.5659); frequency distribution in a, Figure 2, with 50, i.e., 21.5%, of the 233 song types sung by one bird and only one song type shared by all the 12 birds. The average proportion of usage was weakly, but significantly, positively correlated with the popularity of the song type; Spearman’s rank correlation: N = 233, R = 0.3322, P < 0.0001 (b, Figure 2). The top songs of all the birds were also sung by other birds, and the repertoire size of a bird was negatively correlated with the popularity of its top song, i.e., Spearman’s rank correlation: N = 12, R = -0.6762, P = 0.0158 (c, Figure 2) and with the weighted popularity of all songs, i.e., Spearman’s rank correlation: N = 12, R = -0.7881, P = 0.0023 (d, Figure 2).

Constructing song sharing and song co-occurrence networks and testing for non-randomness

At the repertoire level between birds, we found moderate and generally consistent levels of song-type sharing; on average, they shared 37.5% to 48.3% of their repertoires with the other 11 birds. We similarly found moderate values using all three indices (Table 2), with all pairs of birds sharing some of their songs. We found that song-type sharing was non-random using Index 1, i.e., presence/absence of unweighted song types (Network in a, Figure 3), and Index 3, i.e., a dissimilarity index that included relative frequencies (Network in b, Figure 3), but not different from random using Index 2, i.e., presence/absence of song types, weighted by their popularity.

Song types co-occurred to different extents, with some pairs sharing no birds in common, and two pairs sharing all of them (Table 2). We found non-random song-type co-occurrence using Index 5, which reflects both the number of birds that sang both song types and the number of times each song type was used by those birds (Network in Figure 4), but not using Index 4, which looks only at presence/absence. The network (Figure 4) with 1- Index 5 had dyads that had no similarity, i.e., 1- Index 5 = 0, but was still a well-connected network, with a density of 0.6706, i.e., 18127 out of 27028 of all possible ties connected, and each song type co-occurring with many other song types (degree or number of connections: Average ± SD: 155.5966 ± 48.3220). The modularity of the network was low, i.e., 0.1920.

Effects of distance, delivery, and switching on song sharing

Distances between the 12 birds varied from 6.1049 meters to 18435.6900 meters (Average ± SD in meters: 3932.9293 ± 5832.6880). There was no significant effect of distance on the repertoire sharing similarity metric Index 1, i.e., Mantel test, Pearson’s correlation, 1000 permutations: N = 12, R = 0.0136, P (two-sided) = 0.9290 (a, Figure 5), or on the repertoire sharing dissimilarity metric Index 3, i.e., Mantel test, Pearson’s correlation, 1000 permutations: N = 12, R = -0.2826, P (two-sided) = 0.1620 (b, Figure 5).

Node-label permutations (1000 permutations) of the network with Index 1 showed a positive correlation between Delivery and average index (c, Figure 5), and no correlation between Switching and average index (d, Figure 5; Appendix 1). Similar permutations of the network with Index 3 showed a negative correlation between Delivery and average index (e, Figure 5), and no correlation between Switching and average index (f, Figure 5; Appendix 1).

Effect of song characteristics on popularity

The zero-truncated negative binomial regression results showed that songs with lower low frequencies and lower trill rates tended to be more popular (Table 3). Frequency inflections, duration and bandwidth had no effect on popularity (Table 3).

DISCUSSION

Rock Wrens in this migratory population shared many song types with each other, but only one song type was shared by all sampled birds. Rock Wrens have large repertoires and, hence, despite sharing many song types, pairs of Rock Wrens only had moderate song-sharing similarity indices. Song sharing was non-random in this population, with some pairs of males sharing more songs than others, but not because of geographic proximity. Small sample size of individual birds and, hence, incomplete sampling of the population are potential limitations of the study. However, the sample size (N = 12 birds) is comparable to other song-sharing networks, e.g., Potvin et al. (2019) constructed a syllable-sharing network with 12 male song sparrows. Further, given that distance did not affect song sharing in our analyses, it is unlikely that the observed trends are due to some individuals being farther away than others. Despite the small sample size of individuals, we found trends in song sharing that were affected by two different measures of song complexity: repertoire size and song delivery. We found non-random song sharing and song co-occurrence using some, but not other, indices.

Although few song types were sung by many birds (a, Figure 2), highly shared, i.e., popular, songs tended to be sung frequently by Rock Wren males in this population. Observed patterns of non-randomness in repertoire sharing are primarily driven by the sharing of popular songs, because song sharing was not different from random when we controlled for the popularity of the song in Index 2. These results suggest that birds in this population are singing the hits rather than emphasizing the unique parts of their repertoires. However, all males did not sing popular songs to the same extent; males with smaller repertoires preferred singing more popular songs (Figure 2). Thus, whereas Rock Wren repertoire size seems to affect intrasexual competition and assessment (Pitt 2018), in line with the repertoire-size hypothesis, we find an overall tendency to overproduce popular songs, with birds singing the popular songs in their repertoire more often than the rare songs (b, Figure 2), supporting the general-sharing hypothesis. Lapierre et al. (2011) found that older, more competitive Song Sparrow males sang popular songs but we found the reverse, with the presumably less competitive males (Pitt 2018) singing the popular songs more than the competitive ones. Males preferentially singing popular local songs may face less aggression from other males (Beecher et al. 2000), making this strategy beneficial for males with smaller repertoires. Alternatively, males with smaller repertoires may possibly compensate for their repertoire-size disadvantage by singing popular songs that might be preferred by females, because females in other species are known to prefer males who sing locally common songs (Danner et al. 2011, Nelson and Poesel 2013). If female Rock Wrens prefer novel songs, males with larger repertoires could be singing preferred songs, but in Rock Wrens there is little evidence that females use song-repertoire sizes to choose mates (Pitt 2018). It would be interesting to examine the role of singing locally popular and rare/unique songs on direct measures of female choice.

We found non-random song-type sharing by Rock Wren males in two out of three indices, but song sharing was not strongly influenced by distance in this dataset. The lack of a clear relationship between distance and song sharing in this study may result from this being a migratory population or from frequent local movements (Benedict et al. 2021). It also raises questions about how age might affect repertoire size and usage of popular songs (Lapierre et al. 2011). Little is known about where and when Rock Wrens learn songs, because they do not seem to learn typically localized types, with most songs sung by only one bird (a, Figure 2). Further, spatial patterns of song sharing may vary at different spatial scales (Koetz et al. 2007, Rodríguez-Fuentes et al. 2022) and the sampled distances may not have been at the appropriate scale to capture the relationship.

We tested whether song-delivery patterns covaried with song-type sharing and found that one measure, delivery rate, was positively related to the average song-sharing index. This could result if birds singing at faster rates sing more songs in total, and consequently share more of their songs. However, we also saw this correlation with the adjusted Jaccard index that only looks at the presence and absence of song types. Therefore, we suggest that higher rates of singing and more song sharing may be related indicators of male quality, allowing individuals to sing locally popular song types that are shared with other males, being that during countersinging, male Rock Wrens have been known to song-match (Kroodsma 1975), which could then mediate male-male competition or female choice. Singing rates may be a signal of aggressive intent during male-male interactions (Szymkowiak and Kuczyński 2017), and male Rock Wrens sing at a greater rate in response to intruders (Pitt 2018). Thus, higher delivery of broadcast songs might directly play a role in competition, and/or affect it through influencing song sharing. Surprisingly, song-switching rate, which is also a signal in male-male interactions in some species (Deoniziak and Osiejuk 2020), was not related to song sharing in our study.

Groups of Rock Wren song types did not markedly co-occur with each other more than with other types. Although a song co-occurrence network using the Bray-Curtis index was non-random (Figure 4), it was well-connected and not strongly modular or clustered. Consistent with the low modularity, we did not find strong relationships between song co-occurrence and song characteristics (Appendix 2). This result suggests that a few birds were not responsible for all of the song types in the population with particular acoustic properties, e.g., wide bandwidths or rapid trill rates. Rock Wren song types with lower low frequencies and lower trill rates were popular across birds, similar to what was found for within-bird song-type usage in Benedict and Warning (2017). Given the weak, positive correlation between the popularity of a song and its usage (b, Figure 2), it is unsurprising that we found similar trends at the population (this analysis) and individual (Benedict and Warning, 2017) levels. Low frequencies in a song type might be a potential signal of male body size and fighting ability (Benedict et al. 2012, Linhart and Fuchs 2015). By this metric, popular songs show higher performance. In contrast, higher trill rates are considered a signal of higher male quality in many species (Ballentine et al. 2004, Illes et al. 2006), and using trill rates as the metric, popular songs showed poorer performance. Thus, popular songs may have better transmission and carry variable performance information. Unlike the individual-level analysis in Benedict and Warning (2017), we did not find any significant relationship between popularity and song-type duration or bandwidth. This could be due to the small sample size for comparisons across birds, i.e., 12 birds, compared to within-bird song usage, i.e., hundreds of songs per bird. Alternatively, given that we find individual-level variation in song usage, it is possible that combining song-type usage across birds resulted in the non-significant relationships, highlighting that within-individual song use patterns do not always scale up to produce parallel population-level patterns. Thus, although individual birds prefer to sing the longer songs within their repertoires (Benedict and Warning 2017), those are not the most highly shared song types at the population level. More detailed studies on the relationship between song-type usage, popularity, and characteristics in different habitat types will help us understand whether birds of any species consistently preferentially share broadcast songs that enhance transmission.

Rock Wrens, with their large and variable repertoire sizes, present an interesting study system to examine song sharing. Individual Rock Wrens can have popular and rare songs in their repertoire and use them differently, and pairs of Rock Wrens can share their repertoires to different extents. Network analysis offers a holistic way to examine how those repertoires are shared across a population. As expected for a species with large repertoires, there was variation in song popularity, song sharing between pairs of males, and individual song-use patterns. Males with the smallest repertoire sizes were most likely to sing popular song types, prompting questions about the relative benefits of singing locally popular songs. Song popularity covaried with song form and song-sharing rates covaried with other measures of song behavior, such as delivery, providing insight into how song traits and song preferences of individual birds scale up to create population-wide patterns.

LITERATURE CITED

Animal Behavior Society and the Association for the Study of Animal Behavior (ABS/ASAB). 2012. Guidelines for the treatment of animals in behavioural research and teaching. Animal Behaviour 83:301-309. https://doi.org/10.1016/j.anbehav.2011.10.031

Ballentine, B., J. Hyman, and S. Nowicki. 2004. Vocal performance influences female response to male bird song: an experimental test. Behavioral Ecology 15:163-168. https://doi.org/10.1093/beheco/arg090

Bastian, M., S. Heymann, and M. Jacomy. 2009. Gephi: an open source software for exploring and manipulating networks. Proceedings of the International AAAI Conference on Web and Social Media 3:361-362. https://doi.org/10.1609/icwsm.v3i1.13937

Beecher, M. D. 2008. Function and mechanisms of song learning in song sparrows. Advances in the Study of Behavior 38:167-225. https://doi.org/10.1016/S0065-3454(08)00004-1

Beecher, M. D., and E. A. Brenowitz. 2005. Functional aspects of song learning in songbirds. Trends in Ecology & Evolution 20:143-149. https://doi.org/10.1016/j.tree.2005.01.004

Beecher, M. D., S. E. Campbell, and J. C. Nordby. 2000. Territory tenure in song sparrows is related to song sharing with neighbours, but not to repertoire size. Animal Behaviour 59:29-37. https://doi.org/10.1006/anbe.1999.1304

Benedict, L., and N. A. Najar. 2019. Are commonly used metrics of bird song complexity concordant? Auk: Ornithological Advances 136:uky008. https://doi.org/10.1093/auk/uky008

Benedict, L., A. Rose, and N. Warning. 2012. Canyon Wrens alter their songs in response to territorial challenges. Animal Behaviour 84:1463-1467. https://doi.org/10.1016/j.anbehav.2012.09.017

Benedict, L., A. Rose, and N. Warning. 2013. Small song repertoires and high rate of song-type sharing among Canyon Wrens. Condor 115:874-881. https://doi.org/10.1525/cond.2013.130025

Benedict, L., and N. Warning. 2017. Rock Wrens preferentially use song types that improve long distance signal transmission during natural singing bouts. Journal of Avian Biology 48:1254-1262. https://doi.org/10.1111/jav.01357

Benedict, L., N. Warning, N. A. Najar, S. G. Pitt, P. E. Lowther, D. E. Kroodsma, and G. H. Farley. 2021. Rock Wren (Salpinctes obsoletus), version 2.0. In P. G. Rodewald and B. K. Keeney, editors. Birds of the World. Cornell Lab of Ornithology, Ithaca, New York, USA. https://doi.org/10.2173/bow.rocwre.02

Bray, J. R., and J. T. Curtis. 1957. An ordination of the upland forest communities of southern Wisconsin. Ecological Monographs 27:325-349. https://doi.org/10.2307/1942268

Briefer, E., T. Aubin, K. Lehongre, and F. Rybak. 2008. How to identify dear enemies: the group signature in the complex song of the Skylark Alauda arvensis. Journal of Experimental Biology 211:317-326. https://doi.org/10.1242/jeb.013359

Buchanan, K. L., and C. K. Catchpole. 1997. Female choice in the Sedge Warbler Acrocephalus schoenobaenus multiple cues from song and territory quality. Proceedings of the Royal Society B-Biological Sciences 264:21-526. https://doi.org/10.1098/rspb.1997.0074

Cardoso, G. C., and J. W. Atwell. 2016. Shared songs are of lower performance in the Dark-eyed Junco. Royal Society Open Science 3:160341. https://doi.org/10.1098/rsos.160341

Catchpole, C. K., and P. J. B. Slater. 2008. Bird song: biological themes and variation. Cambridge University Press, Cambridge, UK. https://doi.org/10.1017/CBO9780511754791

Danner, J. E., R. M. Danner, F. Bonier, P. R. Martin, T. W. Small, and I. T. Moore. 2011. Female, but not male, tropical sparrows respond more strongly to the local song dialect: implications for population divergence. American Naturalist 178:53-63. https://doi.org/10.1086/660283

Deoniziak, K., and T. S. Osiejuk. 2020. Song-type switching rate in the chaffinch carries a message during simulated intrusion. Behavioral Ecology and Sociobiology 74:1-11. https://doi.org/10.1007/s00265-020-2825-2

Dolby, A. S., C. E. Clarkson, E. T. Haas, J. K. Miller, L. E. Havens, and B. K. Cox. 2005. Do song-phrase production rate and song versatility honestly communicate male parental quality in the Gray Catbird? Journal of Field Ornithology 76:287-292. https://doi.org/10.1648/0273-8570-76.3.287

Fair, J. M., and J. Jones, editors. 2010. Guidelines to the use of wild birds in research. Ornithological Council, Washington, D.C., USA.

Farine, D. R. 2017. A guide to null models for animal social network analysis. Methods in Ecology and Evolution 8:1309-1320. https://doi.org/10.1111/2041-210X.12772

Goslee, S. C., and D. L. Urban. 2007. The ecodist package for dissimilarity-based analysis of ecological data. Journal of Statistical Software 22:1-19. https://doi.org/10.18637/jss.v022.i07

Hedley, R. W., D. M. Logue, L. Benedict, and D. J. Mennill. 2018. Assessing the similarity of song-type transitions among birds: evidence for interspecies variation. Animal Behaviour 140:161-170. https://doi.org/10.1016/j.anbehav.2018.04.008

Hill, C. E., S. E. Campbell, J. C. Nordby, J. M. Burt, and M. D. Beecher. 1999. Song sharing in two populations of Song Sparrows (Melospiza melodia). Behavioral Ecology and Sociobiology 46:341-349. https://doi.org/10.1007/s002650050628

Hughes, M., S. Nowicki, W. A. Searcy, and S. Peters. 1998. Song-type sharing in song sparrows: implications for repertoire function and song learning. Behavioral Ecology and Sociobiology 42:437-446. https://doi.org/10.1007/s002650050458

Illes, A. E., M. L. Hall, and S. L. Vehrencamp. 2006. Vocal performance influences male receiver response in the banded wren. Proceedings of the Royal Society B-Biological Sciences 273:1907-1912. https://doi.org/10.1098/rspb.2006.3535

Kaluthota, C., D. Logue, and D. Rendall. 2020. Conventional and network analyses of song organization and complexity in Northern House Wrens (Troglodytes aedon parkmanii). Journal of Field Ornithology 91:375-392. https://doi.org/10.1111/jofo.12347

Koetz, A. H., D. A. Westcott, and B. C. Congdon. 2007. Spatial pattern of song element sharing and its implications for song learning in the Chowchilla, Orthonyx spaldingii. Animal Behaviour 74:1019-1028. https://doi.org/10.1016/j.anbehav.2006.11.035

Kroodsma, D. E. 1975. Song patterning in the Rock Wren. Condor 77:294-303. https://doi.org/10.2307/1366225

Kroodsma, D. E. 1977. Correlates of song organization among North American wrens. American Naturalist 111:995-1008. https://doi.org/10.1086/283228

Lapierre, J. M., D. J. Mennill, and E. A. MacDougall-Shackleton. 2011. Spatial and age-related variation in use of locally common song elements in dawn singing of Song Sparrows Melospiza melodia: old males sing the hits. Behavioral Ecology and Sociobiology 65:2149-2160. https://doi.org/10.1007/s00265-011-1223-1

Linhart, P. and R. Fuchs. 2015. Song pitch indicates body size and correlates with males’ response to playback in a songbird. Animal Behaviour 103:91-98. https://doi.org/10.1016/j.anbehav.2015.01.038

Logue, D. M., and W. Forstmeier. 2008. Constrained performance in a communication network: implications for the function of song-type matching and for the evolution of multiple ornaments. American Naturalist 172:34-41. https://doi.org/10.1086/587849

Mantel, N. 1967. The detection of disease clustering and a generalized regression approach. Cancer Research 27:209-220.

Nelson, D. A., and A. Poesel. 2013. Song sharing correlates with lifetime social pairing success but not territory tenure in the Puget Sound White-crowned Sparrow. Behavioral Ecology and Sociobiology 67:993-1000. https://doi.org/10.1007/s00265-013-1525-6

Payne, R. B., L. L. Payne, and S. M. Doehlert. 1988. Biological and cultural success of song memes in Indigo Buntings. Ecology 69:104-117. https://doi.org/10.2307/1943165

Peters, S., W. A. Searcy, M. D. Beecher, and S. Nowicki. 2000. Geographic variation in the organization of song sparrow repertoires. Auk 117:936-942. https://doi.org/10.1093/auk/117.4.936

Pitt, S. G. 2018. Why sing so many songs? Testing the function of song type repertoires in Rock Wrens using playback experiments and behavioral observations. Thesis. University of Northern Colorado, Greeley, Colorado, USA. https://digscholarship.unco.edu/dissertations/510

Podos, J., D. L. Moseley, S. E. Goodwin, J. McClure, B. N. Taft, A. V. Strauss, C. Rega-Brodsky, and D. C. Lahti. 2016. A fine-scale, broadly applicable index of vocal performance: frequency excursion. Animal Behaviour 116:203-212. https://doi.org/10.1016/j.anbehav.2016.03.036

Potvin, D. A., K. Strickland, E. A. MacDougall-Shackleton, J. W. Slade, and C. H. Frère. 2019. Applying network analysis to birdsong research. Animal Behaviour 154:95-109. https://doi.org/10.1016/j.anbehav.2019.06.012

Price, J. J., and D. H. Yuan. 2011. Song-type sharing and matching in a bird with very large song repertoires, the tropical mockingbird. Behaviour 148:673-689. https://doi.org/10.1163/000579511X573908

R Core Team. 2023. R: A Language and Environment for Statistical Computing, version 4.2.3. R Foundation, Vienna, Austria. http://www.R-project.org

Rodríguez-Fuentes, J., C. D. Cadena, and J. E. Avendaño. 2022. Vocal behavior and microgeographic variation in song types in a neotropical songbird. Journal of Ornithology 163:105-119. https://doi.org/10.1007/s10336-021-01930-9

Searcy, W. A., and M. D. Beecher. 2009. Song as an aggressive signal in songbirds. Animal Behaviour 78:1281-1292. https://doi.org/10.1016/j.anbehav.2009.08.011

Searcy, W. A., L. M. Chronister, and S. Nowicki. 2023. Syntactic rules predict song type matching in a songbird. Behavioral Ecology and Sociobiology 77:12. https://doi.org/10.1007/s00265-022-03286-3

Szymkowiak, J., and L. Kuczyński. 2017. Song rate as a signal of male aggressiveness during territorial contests in the wood warbler. Journal of Avian Biology 48:275-283. https://doi.org/10.1111/jav.00969

The MathWorks Inc. 2023. MATLAB, Version 9.14.0 (R2023a). Natick, Massachusetts, USA. https://www.mathworks.com

Tracy, T. T., and M. C. Baker. 1999. Geographic variation in syllables of House Finch songs. Auk 116:666-676. https://doi.org/10.2307/4089328

Warning, N., and L. Benedict. 2015. Overlapping home ranges and microhabitat partitioning among Canyon Wrens (Catherpes mexicanus) and Rock Wrens (Salpinctes obsoletus). Wilson Journal of Ornithology 127:395-401. https://doi.org/10.1676/14-170.1

Warning, N., and L. Benedict. 2016. Facultative nest modification by Rock Wrens (Salpinctes obsoletus). Avian Biology Research 9:58-65. https://doi.org/10.3184/175815516X14447553481564

Yee, T. W. 2010. The VGAM Package for categorical data analysis. Journal of Statistical Software 32:1-34. https://doi.org/10.18637/jss.v032.i10