The Communicative Value of Complex Singing in Passerine Birds

Birdsong is one of the most complex signals in the animal world, as it can consist of many different sounds grouped according to certain rules. Singing acts as a distant signal, indicating, e.g., the species and sex of the singer. However, territorial songbirds also use singing as an interactive social signal during territorial disputes, as well while interacting with females. In these contexts, males vary the type and timing of their songs to convey graded information about their motivational state, and those variations can play a role in communication. In this review, we considered how male songbirds vary their singing in the territorial context. To study such variations, researchers usually simulated territorial intrusion by playing back conspecific singing in territories, including singing modified in a manner necessary for the researcher. For comparison, I considered briefly how singing varies in intersexual context. I focus on the role of singing complexity in communication. Therefore, not all known context-dependent changes in singing are considered, but only those related to “complexity”: the diversity of song/sound types and the transitional patterns of different song/sound types in the course of singing. My review has shown that males change their singing when they detect environmental changes such as the appearance of a female or a competitor as follows: (1) the song rate increases, (2) the syllable rate increases, (3) the song-type switching rate increases, (4) the song-type diversity increases (i.e., the observed repertoire size), and (5) longer and more complex songs are predominantly used. In some species, the song bout organization may also change, but the data are still scarce. Typically, one or more, but not all, of the aforementioned acoustic behaviors have been found in a given songbird species. All these behaviors (tactics) come down to a single strategy, namely, maximizing the acoustic diversity over a short period of time (e.g., several minutes), that is, increasing the number of different song and/or note types. The proximate causes of how the increased acoustic diversity work in the territorial competition context might lie in a sensory or perceptual bias of the receiver. Namely, habituation should occur to repeated presentation of the same song type faster than to presentation of different song types. Therefore, by vocalizing more diversely, males more effectively influence the signal recipient’s behavior.


INTRODUCTION
The song of songbirds (suborder Passeri) consists of elements that are grouped into songs. In some species (e.g., in many Acrocephalus warblers), the element compositions of songs vary greatly: sounds are combined relatively freely. However, in the majority of other species, stereotyped song types consisting of unique sets of elements in a strict sequence can be distinguished. Some species use only one type of song, whereas others have repertoires of several song types. The repertoire sizes of some of them are quite impressive. For example, the singing of one male may include more than 400 types of songs, as in the case of the northern mockingbird Mimus polyglottos (Derrickson, 1987), or more than 200 types of songs in the case of the common nightingale Luscinia megarhynchos (Kipper et al., 2004). Usually, different types of elements, syllables (i.e., combinations of several elements), and/or song types alternate nonaccidentally.
For this reason, it is said that birdsong has a "syntax" (i.e., a set of rules governing the alternation of different types of songs and/or elements).
The singing of passerine birds is one of the most complex signals in the animal world. The complexity of singing has at least three components. First, the repertoire size, i.e., the number of different types of songs and/or elements in ta given individual. Second, the number of "rules" that determine the alternation of elementary acoustic units (types of songs and/or elements) in singing. Third, the configuration of an individual song, mainly the diversity of elements/syllables in it. It is still not entirely clear whether birds use all this complexity in communication or, if so, how they do it. The review will consider specific mechanisms of using complex song signals in communication.
In temperate latitudes, predominantly males sing. In females, singing is less common and more often observed in the tropics and the Southern Hemisphere (Australia is the most well-known in this regard) (Beme and Goretskaya, 2016;Odum et al., 2014). In this review, we will focus on the most studied species inhabiting the temperate latitudes of the Northern Hemisphere. Here, the song can be used in different situations. However, the most frequent case is socalled "self-advertising," or spontaneous singing. We consider a situation in which the male sings "alone"; i.e., other individuals of the same species do not obviously influence his behavior, although they may sing within earshot. Self-advertising is usually aimed at defining the occupied territory (in territorial species) and attracting a female. For this reason, after acquiring a female, males of many species stop singing or considerably reduce their singing activity.
In addition to self-advertising, singing may accompany direct interactions with conspecifics, both males and females. In this case, short-term changes in the structure and organization of singing may play a role in communication. Interactions between males are often aggressive and are associated with territorial competition. Courtship displays are aimed at attracting the attention of females and pair consolidation. In these two types of situations, singing may change, becoming different from spontaneous singing (which, in this case, serves as a control).
In this review, we focus predominantly on the mechanisms of song communication using territorial interactions of males as an example, because such data are most numerous. These interactions in nature can be easily simulated experimentally. To do this, it is necessary to broadcast conspecific singing in the territory of the resident male, including singing modified in a way necessary for the experimenter. Then, the response of the focal male should be recorded and compared with spontaneous singing (control). Most of the data considered below were obtained in such playback experiments (below, we will often note changes in certain singing parameters recorded during or immediately after the playback of conspecific singing). For comparison, we used data on changes in the singing of males noted during their interactions with females.
The communicative significance of certain singing patterns can be considered when at least one of the following conditions is met (Searcy and Beecher, 2009): (1) the pattern appears in a certain context (for example, during territorial confrontations) more often than in other contexts, (2) the magnitude of the change is proportional to the change in behavior (e.g., a change in a given singing parameter becomes more pronounced with increasing aggressiveness of male behavior), and (3) birds respond to appropriately modified singing in a different way than to spontaneous singing (control). Ideally, of course, it is desirable that all these conditions be met; however, in reality, only one or two of them are most often tested.
When the communicative significance of certain structures (including singing) is considered, two terms are used in English-language literature: signal and cue. The term "signal" is used when a given structure (e.g., acoustic) contains some information about the sender (for example, a singing male) and causes an adequate response of the recipient, who "understands" the signal (Scott-Phillips, 2008). If the first condition is met and the second has not been studied, the issue at hand is a potential signal. Changes in singing discussed in this review most often correspond to one of these two concepts (signal or potential signal). It should be noted that the "ethological" concept of the signal here does not fully correspond to the meaning of this term in other sciences.
The term "signal" says nothing about the causes of certain changes in singing. In the review, we will also not consider these causes. There are at least two opinions on this issue. Short-term situation-specific changes in singing may indicate an aggressive motivation of a male (if we are talking about a territorial context) or an increase in general motivation, which is situationally nonspecific (for the hypothesis of the nonspecific activation of behavior, see (Panov, 1978)). This latter point of view is supported by the fact that modifications of singing during "male-male" and "male-female" interactions actually have much in common.
A cue is a structure that is useful for the receiver but, unlike a signal, is harmful or useless to the sender (Scott-Phillips, 2008). For example, male greenfinches Chloris chloris infected with coccidia have duller plumage (Horak et al., 2004). In this case, the color of the plumage is a cue, because it can inform other birds that the bird with the dull plumage is not in the best condition. At the same time, it is hardly profitable for the sender to "transmit" this information. Situation-specific changes in singing usually do not fit the definition of "cue." The only exception is, perhaps, the song performance (see the section Singing Tempo and Song Performance).
Thus, the purpose of this review is to consider those relatively short-term (of the order of several minutes) changes in singing that occur during territorial confrontations of males and during courtship for a female. There are no such reviews in Russian-language literature, and the main publications in English are the reviews by Todt and Naguib (2000) and by Searcy and Beecher (2009).
It is worth noting that, in this review, we consider only those changes in singing that are related to "complexity" rather than all known changes in singing. By "complexity," we mean the diversity of song and/or element types used and the rules for combining them in singing. We paid no attention, first, to the frequency parameters of the songs; second, to numerous situation-specific vocalizations (e.g., "aggressive" calls); and, third, to the acoustic patterns that require coor-dinated behavior of two or more birds. The latter include the duets of a male and a female, which are often recorded precisely in a territorial context, as well as the manifestations of consistency in singing two conflicting males (song overlapping (in time), and song matching, that is the phenomenon when a male copies the just sung song of the opponent). Finally, fourth, we did not consider soft songs with a reduced amplitude, which are characteristic of the territorial behavior of a number of species (they are often, though not always, similar in structure to ordinary songs used in spontaneous singing). In the Russianlanguage literature, some information about these situation-specific modifications of singing can be found in the works by Beme and Goretskaya (2013) and Opaev (2012). Thus, our task is to show how, by changing only the structure and organization of spontaneous singing, birds can signal a change in their motivational state.
In addition to the literature data, in this review we used the original materials on a number of species of leaf-warblers (Phylloscopidae), ground babblers (Pellorneidae), and flycatchers (Muscicapidae). These data were collected in 2016-2020 in China (Hunan province, Hupinshan Nature Reserve, 2016), the Amur region (Khingan State Nature Reserve and Khingano-Arkharinsky wildlife sanctuary, 2017-2019) and Vietnam (Dong Nai province, Cattien National Park, 2020).
The Latin names of birds in the review are given primarily according to the synopsis by del Hoyo and Collar (2016).

SONG RATE
In passerine birds, two manners of singing are distinguished-discrete and continuous. In a discrete manner of singing, individual songs (compact acoustic constructions, well separated by pauses from others of the same kind) are clearly expressed (Ivanitsky et al., 2008;Ivanitsky and Marova, 2021). The continuous manner is characterized by disordered rhythmic organization: the durations of both pauses and songs (more importantly) vary widely. Of course, there is no clear boundary between these two manners, since they are two poles of a single continuum. Nevertheless, the singing of most species of songbirds can be attributed without hesitation to the discrete manner. The "song rate" parameter is more often used in relation to them.
The song rate is the average or median number of songs per unit of time (usually per minute). The average or median duration of the pause between songs is a similar index. Often, but not always, these parameters are highly correlated.
The song rate is obviously a labile characteristic that males can rapidly change, thereby signaling a change in their state. However, changes in this parameter in the territorial context are not uniform in differ-ent species: in some species, the song rate increases; in others, it decreases; and, in some species, it does not change.
An increase in the song rate during playback presentation of a recording of conspecific singing was detected, for example, in the red-winged blackbird Agelaius phoeniceus (Searcy and Yasukawa, 1990), the song sparrow Melospiza melodia (Searcy et al., 2000), the black-capped chickadee Poecile atricapillus (Baker et al., 2012), the tree pipit Anthus trivialis (Petrusková et al., 2014), the wood warbler Phylloscopus sibilatrix (Szymkowiak and Kuczyński, 2017), the Claudia's leaf warbler P. claudiae (Kolesnikova, 2020;Opaev et al., 2019), the eastern crowned warbler P. coronatus (Opaev et al., 2019), and in some species of the P. burkii complex (Kolesnikova, 2020;Kolesnikova and Opaev, 2021). At the same time, in the chickadee and wood warbler, the magnitude of the increase in the song rate correlated with the level of aggressiveness of the response. In addition, in the wood warbler, in addition, males responded more actively to the recordings with an increased song rate than to the recordings with a typical rate. We also found a more intense response to recordings with an increased song rate in the Radde's warbler P. schwarzi (Opaev and Shishkina, unpublished data).
However, in a number of species, the song rate during playback of conspecific singing decreases. During playback of conspecific singing, Eurasian blackcaps Sylvia atricapilla usually sing less actively than before (Darolová et al., 2020). The song rate in great tits Parus major is negatively related to the aggressiveness of the response. The tits that reluctantly approach the loudspeaker sing most actively, whereas the active birds flying up to the loudspeaker and flying around a great deal may not sing at all during the playback (Akçay et al., 2020).
Finally, some warblers did not significantly change the song rate in our experiments, for example, the dusky warbler Phylloscopus fuscatus (Opaev et al., 2019), the pale-legged leaf warbler P. tenellipes (Opaev et al., 2019), the large-billed leaf warbler P. magnirostris (Kolesnikova et al., 2019), and the sulfur-breasted warbler P. ricketti (Kolesnikova et al., 2019). It should be noted that these last two species ultimately began to sing significantly more often, though after the end of the playback of the conspecific song (Kolesnikova, 2020). The data on the Radde's warbler are of interest: in this species, the song rate slightly and insignificantly increased during playback, but the effect was much more pronounced after the end of playback. During spontaneous singing, males of this species sang an average of 5.8 songs/min, and the song rate was 10.3 songs/min immediately after playback (Opaev and Kolesnikova, 2019). Something similar was found in song sparrow. Males of this species performed an average of 3.0 songs/min during spontaneous singing, 3.5 songs/min during playback, and 4.4 songs/min after playback (significant differences were found only between spontaneous singing and singing after playback (Searcy et al., 2000)).
The above data suggest that the song rate is, in fact, a quite universal way to modify singing in a territorial context in passerine birds (although this does not mean that it is characteristic of all species without exception). However, in some species, the song rate is associated with a kind of tradeoff with agonistic and/or aggressive behavior: the more actively the male is looking for an intruder, the less often he will sing. It is clear that, for a male sitting on a branch and for a male actively flying back and forth in search for an imaginary opponent, the song rate will be different, with other things being equal. Therefore, in a number of species, the song rate during playback does not increase but, conversely, decreases.
The following facts support the above hypothesis. First, in a number of species, the song rate increases most significantly immediately after the completion of playback, rather than during it, when the male, as a rule, no longer flies very actively in search of the intruder. This is characteristic of the large-billed leaf warbler, the sulfur-breasted warbler, the Radde's warbler, and the song sparrow. Second, individuals most aggressively responding to playback may not sing at all in attempt to find and attack the intruder as soon as possible (the great tit).

SINGING TEMPO AND SONG PERFORMANCE
In addition to the song rate, birds can increase the rate of emission of syllables (singing tempo). This is the ratio of the number of syllables to the duration of the song (this is how the singing tempo is usually calculated), or the number of syllables per unit of time. The parameter is well studied on the example of trilled songs, or trills. Trill is a sequence of identical short syllables following one after another with minimal pauses (for example, see Fig. 1a). Such series of identical elements in avian bioacoustics are called "homotypic series" (according to Hultsch and Todt, 1996).
It was noticed that the trill rate (tempo) is inversely proportional to the width of the frequency range of syllables that constitute the trill, with this distribution being triangular (Fig. 1b): the "fast" trills (a greater number of syllables per unit of time) are formed mainly by narrowband syllables, whereas the "slow" trills may consist of both narrowband and wideband syllables. Such dependence was found in a number of species (Podos, 1997(Podos, , 2001Ballentine et al., 2004;Illes et al., 2006;Cardoso et al., 2007;Cramer and Price, 2007;Goodwin and Podos, 2014;Wilson et al., 2014). It is assumed that this is due to the physical limitations of the vocal organ (syrinx), due to which birds cannot simultaneously maximize the trill rate and the width of the frequency range above a certain limit. Thus, this is an example of song-performance limitations. "Song performance" indicates how difficult it is for an animal to produce syllables with certain characteristics (Lambrechts, 1996;Podos, 1997).
The tradeoff between the bandwidth and trill rate suggests that individual and/or situation-specific variability in song performance may play a role in communication. In Fig. 1b, the dotted line shows the statistical estimate of the "upper limit of capabilities": most of the songs are located below and to the left of this line. The idea of a number of authors is that, the closer a given song is to the "upper limit of capabilities," the more difficult it is for a male to sing it (Lambrechts, 1996;Ballentine et al., 2004;Cramer and Price, 2007;Illes et al., 2006; for review, see Podos and Sung, 2020). Therefore, such "high-performance" songs are characteristic of high-quality males. It is believed that females and other males can "read" this information.
Indeed, in the swamp sparrow Melospiza georgiana, larger and older (i.e., presumably higher-quality) males sing, on average, with a higher performance  (Ballentine, 2009). However, in the dark-eyed junco Junco hyemalis, song performance was not associated with the quality and/or age of the male (Cardoso et al., 2012). Recently, the postulate that there is a correlation between song performance and male quality has been seriously criticized (Kroodsma, 2017): differences in song performance between males can be well explained by random factors of the learning process (e.g., the way that a suitable male teacher sings) without involving ideas about individuals having different qualities.
With regard to communication, one consequence of the idea that there is a correlation between the quality of a male and his song performance is as follows. In key situations (territorial conflicts or interactions with a female), males should try to sing songs that are as "high-performance" as possible to show themselves of best and, in the case of spontaneous singing, given the difficulties with such production, do it less often. Indeed, playback experiments showed that swamp sparrows, in response to playback of a conspecific song, increased their song performance (DuBois et al., 2009;Liu et al., 2018); however, the response to the presentation of the recording depended on the performance of songs in it (Moseley et al., 2013). Data for some other species are not so unambiguous. In response to the playback of conspecific singing, banded wren Thryophilus pleurostictus males increased trill rate, but reduced their frequency range; i.e., as a result, they did not try to maximize their song performance (Vehrencamp et al., 2013).
In 2018-2019, in the Amur region, the author of the review studied the song behavior of the pale-legged leaf warbler using playback experiments. The song of this species is a trill (see Fig. 1), and the repertoire of each male consists of up to 11 types of songs. Different songs of the same individual vary in performance. It turned out that the characteristics of different types of songs are not associated with the peculiarities of their use: males of this species, first, did not use predominantly the "high-performance" songs in a territorial context and, second, did not use the "low-performance" and "high-performance" songs with different frequency. Thus, the existing physical restrictions on the performance of songs (see Fig. 1b) apparently do not play a role in the communication of this species, although they certainly affect the species-specific characteristics of singing.
At the same time, comparison of the same types of songs recorded during spontaneous singing and in response to playback gave interesting results. It turned out that the trill rate in the second case was higher. This increase was noticeable, but very small: on average, by only 2.7 ± 2.0% in a given male. It should be noted that the song performance did not change depending on the context (Opaev, 2021).
Thus, it is doubtful that song performance itself plays a significant role in bird communication. However, the increase in the trill rate looks like a very real situational modification in some species. In some cases, this may lead to an increase in song performance. Unfortunately, not all authors analyze trill rate along with song performance. In an increase in the frequency of the emission of syllables, one can easily see an analogy with an increase in the song rate. Both patterns lead to an increase in the number of syllables produced per unit of time. Moreover, it is quite possible that trills are preferred for these purposes in a territorial context.
Interesting data were obtained for the tree pipit. This species can sing both when perched and in lekking flight, with songs in these two situations differing in a number of parameters (Panov et al., 2006;Petrusková et al., 2008). Perch songs last an average of 2.4 s and consist mainly of alternating homotypic series (trills) of identical elements averaging 6.7 elements per song (Petrusková et al., 2008). Among the series, "loud trills" (one to three types in the repertoire of each male), sequences of longer loud syllables, can be distinguished ( Fig. 2; Petrusková et al., 2014). It was found that, in response to playback of conspecific singing, tree pipits increase the proportion of songs containing the "loud trills" on average from 52 to 72%, and the tempo of such trills significantly increases (Petrusková et al., 2014).
For singing devoid of trills, an increase in the syllable rate may be also characteristic. The common chiffchaff Phylloscopus collybita shortened the interval between successive syllables in response to the playback of species singing. The magnitude of this parameter returned to the previous values (characteristic of spontaneous singing) immediately after the end of playback. In addition, males responded less actively to the playback of artificially slowed down singing (Linhart et al., 2013). "Fast" songs (songs with an increased syllable emission rate) can also be observed during interaction with a female. An example is the singing of the male lesser gray shrike Lanius minor during his expressive "dance" with bows near the female (Panov, 2008; observations of the author of the review). This perfectly illustrates the above idea that the changes in singing observed in the territorial context may actually express the general excitation of the male, which is not specific for a given situation, rather than his having aggressive intentions. SONG-TYPE SWITCHING RATE As was discussed in the Introduction, the singing of the majority of passerine species consists of stereotyped song types. Let us denote each song type with a capital letter (A, B, C, …). The simplest rules for arranging them into a song sequence are described by two concepts-eventual variety and immediate variety (Hartshorne, 1973). In the case of eventual variety, singing is composed of series, each of which is a multiple repetition of the same type of song (AAABBBB-CCDDDDD…). In the case of immediate variety, the bird avoids repeating the song that has just been performed (ABCDE…). The lack of variety (AAAA…) can also be observed in cases in which the male has only one type of song in his repertoire. It is convenient to characterize the tendency of a bird to immediate or eventual variety by the song-type switching rate parameter.
The song-type switching rate describes how often the bird switches between different types of songs (this is the ratio of the number of transitions between songs of different types to the total number of transitions). This parameter varies in the range 0 < "song-type switching rate" ≤ 1. The song-type switching rate is low in the case of eventual variety and high in the case of immediate variety. In the absence of variety, it makes no sense to talk about a song-type switching rate.
The song-type switching rate often changes in the territorial context. However, there are many species in which it does not change significantly. These are often the birds with the most pronounced immediate variety, i.e., those that normally change the song type after each song (ABCD…). The Radde's warbler is an example (Opaev and Kolesnikova, 2019).
In a recent study, we compared the change in the song-type switching rate in an experiment with the magnitude of this parameter during spontaneous singing (Kolesnikova and Opaev, 2021). Three very closely related species from the Phylloscopus (Seicercus) burkii s.l. complex of golden-spectacled warblers were included in the analysis: the Bianchi's warbler P. valentini, the Marten's warbler P. omeiensis, and the graycrowned warbler P. tephrocephalus. They live sympatrically, often symbiotopically, in the mountains of Central China (Opaev and Kolesnikova, 2019). The organization of singing of different species differs quantitatively, but not qualitatively. Significant differences in the organization of singing can also be observed between males of the same species (Opaev, 2016). It turned out that the higher the song-type switching rate in a male during spontaneous singing, the less significantly he changes (increases) this parameter in the experiment. This pattern was revealed both within the same species and between different species (i.e., the males of the species that changes the song-type switching rate during spontaneous singing less frequently are more likely to increase the song rate). In other words, in this case, the tendency for immediate variety reduces the amplitude of the situational variation in the song-type switching rate (Fig. 3).
Species the variety of which is eventual or intermediate between eventual and immediate variation are more likely to change the song-type switching rate depending on the context. However, it is difficult to distinguish a common pattern: in the same situation, the song rate in some species increases, whereas it decreases in others.
Song sparrows significantly increased the songtype switching rate during the playback and returned to the level of this parameter characteristic of spontaneous singing after the playback. The mean values of song-type switching rate during spontaneous singing and during and after playback were 0.09, 0.31, and 0.06, respectively (Searcy et al., 2000). The higher the song-type switching rate during the experiment, the more aggressively the male behaved (Kramer et al., 1985).  The northern cardinal Cardinalis cardinalis also increases the song-type switching rate in a territorial context. The repertoire of each male of this species includes 8-12 types of songs performed according to eventual variety. During confrontations with a neighboring male, the song-type switching rate increases due to a reduction in the number of repetitions of identical songs (Ritchison, 1988).
An increase in the song-type switching rate also plays a role in interaction with females. Male dunnocks Prunella modularis switch the song type in the period of fertility of their partner approximately ten times more often than in other times (Langmore, 1997). Male red-winged blackbirds increased the song-type switching rate in response to the presentation of a decoy female in the "invitation to copulation" posture from 0.08-0.09 to 0.21-0.29. Conversely, the demonstration of a live male in a cage caused a very small but significant decrease in song-type switching rate, from 0.11-0.15 to 0.07-0.10 (Searcy and Yasukawa, 1990).
The common chaffinch Fringilla coelebs normally also reduce the song-type switching rate during the playback of the conspecific song. In this species, a differentiated response to recordings differing in the song-type switching rate was revealed. In Poland, males were presented with one of two records with song-type switching rates of 0.06-0.29. The males responded more actively to the first recording: they spent more time near the loudspeaker, sang less actively, but produced more calls (Deoniziak and Osiejuk, 2020).
It is still obscure why some species increase the song-type switching rate while others decrease it. To explain this phenomenon, the following idea was proposed. It is assumed that species that normally sing with eventual variety usually increase the song-type switching rate, and those that tend towards immediate variety reduce the song-type switching rate (Vehrencamp, 2000;Collins, 2004). However, this possible pattern has many exceptions. For example, chaffinch sings with eventual variety (Riebel and Slater, 1999); however, contrary to expectations, this species reduces the song-type switching rate in territorial competition context.
Thus, the pattern of situation-specific changes in the song-type switching rate in different species may vary. The causes for this variability are not completely clear. Nevertheless, the general consensus is that, when the environment changes, the song-type switching rate usually increases rather than decreases (Searcy and Beecher, 2009). Note that, as shown above, an increase in the song-type switching rate was revealed not only in the territorial context, but also during interaction with the female.

SONG REPERTOIRE SIZE AND SONG-TYPE DIVERSITY
Repertoire size is the number of different types of songs or types of syllables in a given male. In this review, we will focus mainly on the repertoire of stereotyped songs.
An absolute assessment of the size and composition of the repertoire of a given male is associated with certain difficulties. First of all, in a given period of time (an hour, a day, a month, or even a year), a bird may not reproduce the entire set of syllables/songs stored in its memory (Shishkina and Opaev, 2020;Geberzahn and Hultsch, 2003;Kipper and Kiefer, 2010). In addition, the repertoire may depend on age, and this dependence is not always linear (Kipper and Kiefer, 2010). For example, in the whinchat Saxicola rubetra, the repertoire size increases most noticeably between the second and third calendar years of life and, then, almost does not change (Vaytina and Shitikov, 2019).
For this reason, researchers usually consider the observed repertoire size-the number of different types of songs in the analyzed recording. The observed repertoire size is an estimation index. There are two more similar indices. The song-type diversity is the number of song types sung in some fixed, but short, period of time (for example, 1 min). The song-type versatility is the number or proportion of different types of songs in a short sequence of a fixed number of songs (for example, the number of different types in ten consecutive songs (Kroodsma and Vemer, 1978)). In different studies, one or another of the three considered repertoire estimates is usually used. In general, these metrics describe the same phenomenon: how diverse singing is over a short period of time (for example, during playback of conspecific singing).
The study of the communicative function of the repertoire size is interesting for understanding the role of complex singing in the life of birds. We have already said above that the repertoires of males of some species can be very large. What is it for? Is it used in communication between males? Oddly enough, there is a great deal more data on this issue. The reason is that the appearance and development of complex songs in evolution has traditionally been associated with sexual selection, namely, with the fact that females prefer complex and diverse singing (Catchpole and Slater, 2008). It was believed that, for the regulation of relations between neighboring males, simple singing is more convenient (Catchpole, 1980), because it is easier for a male to remember and, then, recognize neighbors by singing, allowing timely responding to strangers. For this reason, the role of song repertoires in communication between males has not been sufficiently studied. However, the evolution of complex and diverse singing due to differentiated selection by females has recently been called into question (Garamszegi and Møller, 2004;Byers and Kroodsma, 2009;Cardoso and Hu, 2011). Some authors postulate that song complexity plays an important role in the territorial behavior of males (Byers and Kroodsma, 2009).
The first person to think about the role of song repertoires in territorial behavior may have been John Krebs. He and his colleagues studied the great tit in the vicinity of Oxford (United Kingdom). The repertoire of each male of this species contains two to eight stereotyped types of songs. Krebs et al. (1978) performed several experiments in February-March 1976-1977 according to the following scheme. In one morning, all the great tits dwelling in a small forest area (several pairs) were caught. Then, this area was divided into three territories. In one territory, singing with only one song type was played back; in another, singing included several types of songs (i.e., the typical singing of a great tit). The third territory was used as control (playback-free territory). There were enough nonterritorial great tits (or tits with bad territories) in the vicinity. Therefore, pairs of these birds began to occupy the empty territories the very next day. The control "silent" territory was occupied first, the territory where singing included only one song type was occupied next, and the territory where diverse singing was played back was occupied last (Fig. 4). A study of redwinged blackbirds with similar results was performed several years later in the United States (Yasukawa, 1981).
There is also evidence that, in a territorial context, males of a number of species use more types of songs than when they sing spontaneously. Males of the grass wren Cistothorus platensis sang more variedly when another male vocalized nearby (Kroodsma and Verner, 1978). In the banded wren, the males whose singing was most diverse stayed near the loudspeaker longer (Vehrencamp et al., 2007). Claudia's leaf warblers significantly increased the repertoire size in response to presentation of a conspecific song. The median number of song types used in spontaneous singing was 5.1 before playback (about 5 min) and 11.1 during playback (5 min) (Opaev et al., 2019). Apparently, a similar trend is characteristic of the chestnut-sided warbler Setophaga pensylvanica. Males of this species have an average of eight types of songs in their repertoire. In ordinary singing, only two types are used more often, and the rest can rarely be heard. The proportion of such rare types significantly increased in singing in response to the playback of a conspecific song (Byers, 2017).
In our experiments in Vietnam in 2020, the songtype diversity in the Indochinese blue flycatcher Cyornis sumatrensis increased. Each male of this species has a set of stereotyped song types that make up his repertoire. In response to experimental playback, the observed repertoire size increased. The most striking changes, in comparison with the control recording before the playback, were observed on the phonogram immediately after, rather than during, the playback. The observed repertoire size averaged 5.4 song types during spontaneous singing (5 min of recording), 4.6 song types during the playback of a conspecific song (5 min), and 7.5 song types immediately after the playback (5 min).
In the bobolink Dolichonyx oryzivorus, the songtype diversity increased when territorial males were presented with a caged female. However, their songtype diversity did not change if they were presented with another male in a similar way (Ammer and Capp, 1999). The males of another species from the same family Icteridae, the red-winged blackbirds, behaved in a similar way. Males used more song types compared with the control period (4.7 vs. 3.6) when they were presented with a decoy female in the "invitation to copulate" position. However, their song-type diversity did not change if a cage with a live male redwinged blackbird was placed on the resident's territory (Searcy and Yasukawa, 1990).
Thus, in most cases, song-type diversity increases in response to a change in the external environment, when a female or a competitor appears (Searcy and Beecher, 2009). However, there are exceptions: the northern mockingbird uses fewer song types than in other situations when patrolling the boundaries of the   , 1987). Finally, there are many species in which certain metrics of song diversity do not change in the territorial context. These are, for example, the rock wren Salpinctes obsoletus (Pitt, 2018) and some species of warblers studied by the author of the review and colleagues (Opaev et al., 2019;Kolesnikova and Opaev, 2019). It should be noted that there are many more species in which the song-type diversity was just not considered. A natural question arises as to does an increase in the song-type diversity play a role in communication.
To answer it, it is necessary to compare the response to ordinary and "diverse" singing. Such data are scarce and remain contradictory.
In the cases described, experiments with repertoire of syllables were performed. Experiments with the presentation of recordings contrasting in the song-types diversity were performed with song sparrow and rock wren. The repertoire of each male of the first species includes, on average, nine song types. Experiments showed that responses to recordings with one or more song types did not differ (Searcy, 1983;Stoddard et al., 1988). The individual repertoires of rock wrens are wider and include 50-120 song types. Males of this species were presented with playback of recordings with diverse (25 different song types per 50 songs in the recording) and more monotonous (five song types per 50 songs) singing. During playbacks of diverse singing, these wrens sang more actively, performing more songs per minute (Pitt, 2018).
Interestingly, no increase in song diversity in a territorial context was shown for any of these six species. In other words, it remains obscure whether the repertoire size plays even a potential role in direct interactions between birds. At the same time, it is known that some of these species practice other ways of expressing territorial aggression. Common blackbirds produce a distinctive twitter song during playback (Hesler et al., 2011), chiffchaffs increase the singing tempo (Linhart et al., 2013;Sierro et al., 2020), and rock wrens increase song rate (Pitt, 2018).
A completely different situation is observed for the pale-legged leaf warbler, which the author of the review studied in the Amur region. In this species, two singing modes can be conditionally distinguished. In the first case (stereotyped singing, or S-singing), the male monotonously repeats the only type of song, whereas, in the second case (variable singing, or V-singing), he alternates songs of from four to ten different types. S and V-singing do not differ in the structure of songs (examples of songs of this type are shown in Fig. 1). The differences between these modes consist only in the observed repertoire size. When a male sings spontaneously, he usually (but not always) uses S-singing. If he is presented with a playback with conspecific singing, he usually switches to the V-mode. Thus, in response to the playback of conspecific singing, pale-legged leaf warbler increases the song-type diversity (unlike the six species considered above), remaining unchanged other acoustic parameters. In addition, some males showed a peculiar behavior: they produce a crackling when flying.
In experiments, we compared whether the responses to monotonous S-singing (one song type) and more diverse V-singing (the presented recording included five song types) differed. It was found that the number of flights with crackles was significantly higher in response to the V-singing playback. Thus, song-type diversity plays a role in the communication of this species. We hypothesize that song-type diversity also plays a significant role in many other birds. Further research should clarify this issue.

SONG BOUT ORGANIZATION: ORDER IN ALTERNATION OF DIFFERENT SONG TYPES
To date, ample data on the organization of bird singing have been accumulated. It was found that, in most cases, the sequence of acoustic units (for example, song types) during singing, on the one hand, is not strictly determined and, on the other hand, is not random (Kershenbaum et al., 2016). Two main types of syntactic organization of singing were identified (Ivanitskii, 2015;Ivanitskii and Marova, 2021).
In the case of the linear syntax (Fig. 5b), the order of performance of vocal components is more or less strictly fixed. This means that a given song type determines the next type. This variant of the organization of singing "in its pure form" has so far been described in a few species of birds, for example, in the Radde's warbler (Ivanitskii et al., 2011;Opaev and Kolesnikova, 2019;Shishkina and Opaev, 2020;Ivanitskii et al., 2012) and the gray-crowned warbler (Opaev, 2016), as well as the thrush nightingale Luscinia luscinia (Ivanitskii et al., 2017). Interestingly, in most of these species, the rigid linear syntax is characteristic of many, but not all, males of this population without exception.
In the case of the combinatorial syntax (Fig. 5a), one or more clusters of mutually associated songs are distinguished, within which the order of their performance may change; i.e., each song type may have a different continuation in different cases. Recombina-tions-arbitrary variations in the order of performance of different songs within their certain set, or clusterare often observed (Ivanitskii, 2015).
When studying the organization of singing, researchers are mainly interested in general specific features, whereas intraspecific variability (e.g., individual, situational, or age-related) remains insufficiently studied.
In the gray-crowned warbler, a noticeable individual variability in the organization of spontaneous singing was revealed (Opaev, 2016). The repertoire of each male of this species includes 18 to 48 (on average, 36) discrete stereotyped types of songs, each lasting about 1 s. The pause between successive songs is 4-9 s. When singing, males adhere to immediate variety. In some individuals, the alternation of different types of songs is subject to a linear syntax: each given type of song is usually followed by another strictly defined type (for example, male 6 in Fig. 6). The organization of singing of other males rather corresponds to the combinatorial syntax (male 9 in Fig. 6). Ways of singing organization that are transitional between these extreme variants have also been identified (males 7 and 3 in Fig. 6).
A similar individual variability in the song bout organization was also detected in the Radde's warbler (Opaev and Kolesnikova, 2019;Shishkina and Opaev, 2020). The causes for the considered variation in these two species are not known. Interesting data in this regard were obtained in a long-term study of the collared flycatcher Ficedula albicollis in Hungary. This species has no stereotyped songs. When singing, males "build" songs from the repertoire of syllables at their disposal being guided by some not very strict rules. It turned out that, with age, the orderliness of singing increases: males begin to perform predominantly certain "favorite" combinations of syllables, i.e., the freedom to combine original elements (syllables) decreases (Zsebök et al., 2021).
Thus, theoretically, song-bout organization may play a role in communication. In some species of warblers, we have indeed found situational variation of the corresponding parameters. Males of Bianchi's warbler, during the playback of the conspecific song, began to alternate the stereotyped types of songs in a significantly more orderly manner as compared to the spontaneous singing. The syntax of their song bout became more linear (Kolesnikova and Opaev, 2021) (Fig. 7). We identified the same pattern in Claudia's leaf warbler (Opaev et al., 2019). However, the Radde's warbler, on the contrary, sang in a less orderly manner immediately after the playback (Opaev and Kolesnikova, 2019). It should be noted that a number of other warblers with extensive repertoires (e.g., the dusky warbler) did not change the song bout organization in our experiments (Opaev et al., 2019;Kolesnikova, 2020).
Obviously, the variability of the organization of singing in a territorial context requires additional research. It also remains unclear whether the birds themselves can differentiate singing with different syntax (linear and combinatorial). Such data, with a few exceptions, are practically missing. In 2016, graycrowned warbler males in China were presented with recordings of singing contrasting in syntax (linear or combinatorial). No differences in the response to these two recordings were revealed (Kolesnikova, 2020).  It should be noted that, other things being equal, linear syntax maximizes the song-type diversity over short periods of time (Opaev, 2016). Consider a sequence of 10 or 20 songs sung in the continuousvariation mode. The observed repertoire size, or the song-type diversity, in such a sequence will be greater the more linear its organization is. Indeed, in this case, the repetition of the same types of songs (for example, at the beginning and at the end of the sequence) will be excluded, provided that the male's repertoire allows this (i.e., if it includes more than 10 or 20 types of songs). Therefore, it can be said that increasing the singing linearity to a certain extent maximizes the diversity of acoustic production over short periods of time (i.e., the syntax itself may not be important). This is consistent with the increase in the diversity of singing in the territorial context discussed in the previous section.
When discussing the role of the song bout organization in territorial behavior, the phenomenon called "song advancing" should be mentioned. It was described using the Cassin's vireo Vireo cassinii as an example . Shortly before this, a similar pattern was revealed in experiments with the common nightingale (Weiss et al., 2014). Imagine that a male prefers to sing all or some of the songs from his   repertoire in a certain sequence, for example, A → B → C → D → E → F. If this male is presented with a playback of corresponding types of songs, he will respond to them as if he had sung them himself. For example, having heard a type A song, he will continue, as he usually does, with song B. He will answer to a E song heard from the loudspeaker with an F song. This is song advancing.
What role may song advancing play in communication between males? The fact is that males of many bird species have shared song types, which are identical or very similar in their structure in different individuals (Beecher and Brenowitz, 2005). Moreover, males of some species can also sing shared songs in the same sequence (Payne, 1979;Grießmann and Naguib, 2002;Hedley et al., 2018). For example, different male common nightingales in Moscow can sing shared song sequences including up to five different types of songs, which accounts for a significant part of the repertoire, which in different individuals consists of 7-17 types of songs (Ivanitskii et al., 2013(Ivanitskii et al., , 2017. If two interacting males have shared song sequences, the "song advancing" phenomenon may play some (still unclear) role in their interaction .

SONG COMPLEXITY: DURATION, DIVERSITY OF SYLLABLES, AND SYNTAX
The complexity of a single song, as well as of a song bout, has two components: the diversity and number of syllables in the song (the structural complexity of the song) and the number of rules that determine the alternation of these syllables (the complexity of the organization of the song).
First of all, it should be noted that the song structure is largely a species-specific characteristic. This means that males of a given species usually group individual syllables into a song according to certain rules. For example, in the song of the European robin Erithacus rubecula, phrases consisting of high-and low-pitched syllables naturally alternate (Bremond, 1968). Such rules can be called the "species syntax" of a song. There are an ample amount of data confirming that birds respond differently to artificially modified songs as compared to the typical songs organized according to the species syntax. For example, the California thrashers Toxostoma redivivum responded differently to singing in which the sequence of phrases was atypical for this population (Taylor et al., 2017). Similar results were obtained in studies of the field sparrow Spizella pusilla (Nelson, 1988), the swamp sparrow (Balaban, 1988), the Eurasian wren Nannus (Troglodytes) troglodytes (Holland et al., 2000), and the Eurasian skylark Alauda arvensis (Briefer et al., 2013).
However, the abilities of birds in terms of "understanding" the song syntax should not be overestimated. An interesting experiment was performed with the zebra finches Taeniopygia guttata and three other species. The repertoire of each male of this species includes only one motif (song type) consisting of several syllables performed in a certain sequence. In the experiments, the birds were trained to respond to the difference between sequentially (with an interval of 250 ms) presented series of five syllables (i.e., "tunes"). If the bird "noticed" the difference, it pecked at a certain mark (Fig. 8, top). It was found that birds are bad at differentiating changes in the sequence of syllables. If, after playback of a typical A-B-C-D-E motif (Fig. 8), a motif with a modified syntax (С-E-D-A-B in Fig. 8) is included, the bird may not notice it. However, the structure of the syllables themselves is much more important. Finches almost always pecked at the desired mark when they heard a song motive with even one syllable artificially inverted (mirrored) on the computer. Domestic canaries Serinus canaria domestica and Bengalese finches Lonchura striata domestica in similar experiments behaved similarly. In contrast, budgerigars Melopsittacus undulatus detected the sequences with altered syntax as successfully as the sequences containing inverted syllables (Lawson et al., 2018;Fishbein et al., 2019). Thus, the species syntax of a song may allow for some variation, which may play a role in communication.
In a territorial context, as well as when courting a female, the structural complexity of the song tends to increase or remain unchanged. The latter is characteristic of very many species whose songs are stereotyped and, therefore, change very little in the course of singing in principle. An example is many warblers and sparrows that have repertoires consisting of stereotyped types of songs (song sparrow, dusky warbler, pale-legged leaf-warbler, crowned warblers, etc.).
The structural singing complexity has two components-the duration of the song and the diversity of syllables in it, which are difficult to separate. The fact is that, in many cases, longer songs are at the same time more diverse in terms of the number of types of syllables.
The singing of the Eurasian skylark, which often sings during flight, is continuous. In France, researchers studied how the spontaneous singing of this species differs from what is observed after the playback of a conspecific song. In this study, 40-s segments of both singing patterns were analyzed. It was found that the singing observed in the experiment was more diverse. The average number of different types of syllables in a 40-s fragment was 124 ± 35, and that in the spontaneous singing was 110 ± 29 (Geberzahn and Aubin, 2014). In this case, it was the diversity of the song that mattered.
In great tits, the variability of song duration was studied. A typical song of this species consists of rhythmically repeating identical syllables, one or two elements in each. The number of syllables in a song may vary. It was found that the males that most actively responded to the playback of a conspecific BIOLOGY BULLETIN Vol. 49 No. 10 2022 OPAEV song sang longer songs with a greater number of syllables (McGregor and Horn, 1992). In red-winged blackbird males, the number of syllables in songs (and, accordingly, the duration of the song itself) increased when males interacted (Ritchison, 1988). For a number of species, the role of an increase in the structural complexity of the song in territorial behavior was shown experimentally. Males of some species responded more actively to the playback of more complex and long songs with a greater number of syllables and/or types of syllables. Such data are available for the common whitethroat (Balsby and Dabelsteen, 2001), chiffchaff (Linhart et al., 2012), wood warbler (Goretskaya, 2013), common chaffinch (Leitão et al., 2006), and tui Prosthemadera novaeseelandiae (Hill et al., 2018).
The role of complex singing in communication with females is also well studied. As we have already noted, the formation of complex songs in the evolution is conventionally associated with the preferences of females for them in comparison with the simpler songs (Catchpole and Slater, 2008). One of the first studies of this kind was performed on European warbler species. It was found that, among the six species studied, the most complex singing (maximum repertoires) is characteristic of monogamous species, whereas the singing of polygamous species is simpler (Catchpole, 1980). The author assumed that, in polygamous warblers, when making their choice, females are guided mainly by the quality of the territory of the male (so-called "resource-dependent polygyny"). Therefore, the most important function of song in these species is the regulation of relationships between neighboring males. Simple songs here, as it was then believed, are better, because they help males to recognize each other. The complex song of monogamous warblers is needed only to attract a female. A similar relationship between the social system and the singing complexity was also revealed in Icteridae (Kroodsma, 1977) and in some Emberizidae (Catchpole and McGregor, 1985). However, an opposite picture was observed in the North American Troglodytidae: a more complex song is characteristic of polygynous species (Kroodsma, 1977).
The preference for complex singing by females was also shown experimentally. In a number of species, they responded more actively to the playback of more complex (including more complex than typical) songs. This was shown, for example, for the sedge warbler (Catchpole et al., 1984), zebra finch (Collins, 1999), song sparrow (Searcy and Marler, 1984), and redwinged blackbird (Searcy, 1988).
Data on the role of song-organization complexity are very scarce. An interesting work was performed with the house finch Haemorhous mexicanus (Ciaburri and Williams, 2019). A song of this species is a more or less variable sequence of syllables lasting several seconds. Due to such variation, it is not possible to distinguish the stereotyped types of songs (Fig. 9), and the repertoire of each male includes 30-65 types of syllables. The authors of the cited work compared the singing of males in three contexts: during spontaneous singing, when moving after a female, and during vocal duels with neighboring males. It was found that the number of types of syllables in one song does not differ between contexts. However, the songs in the presence of a female were approximately twice as long due to the duplication of individual syllables or their combinations. The most interesting observation was the fact that, in the presence of a conspecific (both male and female), the birds combined syllables more freely than during spontaneous singing: the number of types of transitions between different syllables was greater when following a female and confronting a neighbor (Fig. 9). This, in turn, suggests that the successive songs, in general, differed to a greater extent than songs in spontaneous singing; therefore, the singing diversity was greater.
Thus, longer and/or more diverse songs are generally more characteristic of direct interactions between birds than of spontaneous singing. The playback of such songs in some cases causes a more vivid behavioral response in both males and females.

MAXIMIZING DIVERSITY AS A GENERAL PRINCIPLE OF SINGING MODIFICATION
This review showed that, when the external environment changes (when a female or competitor appears), the following changes are most often observed in the singing of songbirds (in a particular species, usually one or more, but not all at once): (1) the song rate increases, (2) the syllable rate increases, (3) the song-type switching rate increases, (4) the song-type diversity increases (i.e., the observed repertoire size increases), and (5) longer and more complex songs are predominantly used. In some species, the song-bout organization may also change. These last changes, in our opinion, are important not in themselves, but because they lead to an increase in the song-type diversity over short periods of time (Opaev, 2016).
In our opinion, all the five tactics of changes in singing come down to a single strategy. It is the maximization of acoustic diversity, i.e., an increase in the number of different types of songs and/or syllables per unit of time (for example, per 1 min). Obviously, the number and/or variety of syllables per 1 min will increase with the implementation of tactics (1)-(5). Maximization of diversity is a general principle for using complex signals (such as singing) in communication. When the social context changes or when signaling an increase in the level of his excitation, the male "tries" to perform as many different songs and/or syllables per unit of time as possible (i.e., to maximize diversity). The idea of the relationship of the song-type diversity with context is not new. For example, back in 1938, Chapman was so impressed by the singing of the black-bellied wren Pheugopedius fasciatoventris on the Island of Barro Colorado (Panama) that he devoted almost an entire chapter in one of his books on the fauna of the island to this bird (Chapman, 1938). Chapman wrote that, when excited, the bird (he was talking about observations of one male) sang more variedly. When nothing special happened in the area, the song-type diversity was lower.
An increase in the variety and complexity of vocalizations with an increase in the emotionality of the context is probably not unique to birds. Such data, though scarce, are also known for mammals. For example, in the rock hyraxes Procavia capensis, the sequences of syllables provoked by other individuals (both conspecifics and potential predators) are longer and contain more syllables compared to the "spontaneous" sequences (Demartsev et al., 2014). The humpback whale Megaptera novaeangliae males begin to change the tune in their "songs" more often when they hear the vocalizations of another male (Cholewiak et al., 2018).
The meaning (proximal mechanism) of increasing diversity may be to increase the time of habituation to the stimulus. It seems obvious that, during multiple repetition of the same signal, the animal sooner or later gets used to it and stops to pay attention to it. The idea is that habituation to a complex signal (such as diverse and constantly changing singing) will take more time as compared to simple and monotonous singing. Hartshorne was the first to draw attention to this possibility. He assumed that the meaning of switching between different types of songs may lie precisely in increasing the time for the recipient to get used to such a stimulus (Hartshorne, 1956). Later, it was shown in several works that, indeed, in the case of a monotonous repetition of a single type of sound, habituation to it occurs faster if there are more types of syllables in the sequence (Ryan, 1998;Collins, 1999;Flower et al., 2014). This means that, by vocalizing more diversely, birds more efficiently affect the recipients of the signal (competitors or mate partners).

ACKNOWLEDGMENTS
The author thanks, first of all, his colleagues who participated in collecting the original data used in the review: A.I. Antonov, Yu.A. Kolesnikova, A.S. Rubtsov, and E.M. Shishkina. The author is grateful to Meisi Liu, Kang Zujie, and Shurong Tian for assistance in organizing and conducting field research in the Hupinshan Nature Reserve (Hunan province, China); in the Khingan Reserve (Amur region), to A.I. Antonov and V.A. Kastrikin; and in Vietnam (Dong Nai province, Cattien National Park), to V.V. Rozhnov, A.N. Kuznetsov, I.V. Palko, and S.S. Gogoleva.

FUNDING
This study was supported by the Russian Foundation for Basic Research, project no. 20-14-50017.

COMPLIANCE WITH ETHICAL STANDARDS
The author declares that he has no conflicts of interest. This article does not contain any studies involving animals or human participants performed by the author.