Review A global synthesis of the impacts of urbanization on bird dawn choruses

Urbanization drives changes in acoustic communication systems in some animal species. Noise and light pollution are among the main urban factors known to disrupt the timing and structure of avian singing behaviour. Despite our understanding of the ways in which urbanization can drive variations in avian acoustic communication, our ability to generalize the underlying causes of such variation and its consequences is still limited. Here, we reviewed the literature focused on the study of avian dawn choruses in urban settings at a global scale. Our ﬁ ndings reveal that avian dawn chorus research has focused on the impact of anthropogenic noise on dawn chorus traits (i.e. timing, peak, song output, song frequencies); relationships between light pollution and chorus timing; the effects of temperature, cloudiness, moonlight and natural light on chorus timing; relationships between nocturnal noise and light, and dawn chorus timing; the effects of chemical pollution and supplementary feeding on dawn chorus activity; and ecological patterns of dawn choruses in soundscapes across urban – non-urban gradients. We identi-ﬁ ed important knowledge gaps in the study of avian dawn choruses in urban settings and thus suggest future research directions, including frameworks (e.g. the urbanization intensity gradient) and consideration of a wider array of urban conditions and variables. Given the complexity of urban settings, we encourage further studies to address the role that all sources of pollution can have on avian acoustic communication at dawn. Addi-tionally, a central question to resolve is whether the function of avian dawn choruses in urban areas differs, and if so how, from non-urban counterparts. Given that most research has been performed across Holarctic cities and towns, studies from tropical and subtropical regions are needed if we aim to understand the phenomenon globally. Finally, studies at the community- and soundscape-level across cities could advance understanding of the way in which urban birds use the acoustic space during the most critical singing time period, dawn.


INTRODUCTION
Sounds are a fundamental property of nature, providing relevant information about the status of biodiversity (Pijanowski et al. 2011). Sounds emitted by animals have become one of the most studied proxies in the assessment and understanding of the ecological impacts of human activities (Slabbekoorn 2018, Burivalova et al. 2019). However, there are still important gaps in our understanding of how the sum of sounds emitted by biological, geophysical and anthropogenic sources (i.e. soundscapes) are affected by landscape transformation and how human-driven forces shape their organization and dynamics (Pijanowski et al. 2011, Krause & Farina 2016.
One important component of soundscapes is animal chorus (i.e. vocalizations emitted by multiple individuals of the same or different species at the same time in response to common diel cues; Burt & Vehrencamp 2005, Catchpole & Slater 2008. Such choruses are common in many wildlife groups, including birds, fish, mammals, insects and shrimps (Farina & Ceraulo 2017). By signalling in choruses, animals communicate by broadcasting important information about their spatial location, territory quality and mate attraction (Kacelnik & Krebs 1982, Burt & Vehrencamp 2005, Greenfield 2005, Schel & Zuberb€ uhler 2012, Farina & Ceraulo 2017.
Recent studies suggest that the disruption of animal chorus timing and organization is a consequence of increasing levels of anthropogenic noise in both terrestrial and aquatic ecosystems (Pijanowski et al. 2011, Krause & Farina 2016, Farina & Ceraulo 2017. However, it is not yet fully understood how anthropogenic activities influence animal chorusing at the population-and community-level across ecosystems worldwide. Urban ecosystems, especially, are characterized by multiple stressors, including chemical pollution, increasing temperature levels, high levels of anthropogenic noise (in both amplitude and frequency domains), artificial light at night (ALAN) and human presence, all of which can affect wildlife, with physiological, ecological and behavioural responses emerging to deal with such stressful conditions (see Isaksson 2018 and references therein). In fact, innovative problem-solving, personality and phenotypic plasticity have been identified as important adjustments to urban life (Gil & Brumm 2014, Isaksson 2018. Evidence of the behavioural plasticity of acoustic phenotypes (i.e. overall temporal, spectral and amplitude characteristics of an acoustic signal; Slabbekoorn 2013) of urban birds has increased in the last decade, due its association with the two dominant drivers in urban environments: anthropogenic noise and artificial light pollution (Slabbekoorn 2013, Shannon et al. 2016, Gaston et al. 2017.
As predicted by the acoustic adaptation hypothesis (AAH), differences in habitat structure could shape the acoustic properties of animal communication (Morton 1975). Due to the environmental heterogeneity of urban settings (e.g. differing land-uses, urbanization intensity gradients, spatio-temporal artificial light and noise variations), together with their acoustic complexity (Warren et al. 2006), the AAH has provided a fundamental framework to understand how animals deal with predominant anthropogenic noise by adjusting their acoustic phenotype to increase the effectiveness of their acoustic signals (Slabbekoorn & Peet 2003, Luther & Baptista 2010, Phillips et al. 2020. Some bird species have been shown to deal with anthropogenic noise in cities by adjusting the amplitude, minimum frequency, syntax and timing of songs (Slabbekoorn & Peet 2003, Brumm 2004, Luther & Baptista 2010, Slabbekoorn 2013, Gil et al. 2015, Berm udez-Cuamatzin et al. 2018. Moreover, recent evidence highlights the biological consequences of artificial light for birds, particularly their physiology and behaviour (Gaston et al. 2017). For example, artificial light can affect circadian rhythms through the disruption of melatonin cycles (Dominoni, 2015), leading to earlier dawn choruses in lightpolluted areas (Fuller et al. 2007, Da Silva et al. 2016, Hopkins et al. 2018.
Chorusing (i.e. acoustic display involving three or more members of a social unit, including both males and females) is a widespread, complex behaviour among birds, occurring in at least 1830 species (~18% of all known bird species), including both passerines and non-passerines (Tobias et al. 2016). Among these acoustic displays, dawn choruses are a conspicuous but little understood trait of avian natural history (Catchpole & Slater 2008, Gil & Llusia 2020. Typically, this behaviour is defined as a peak of singing activity performed by different bird species starting around first light, with a gradual decline towards sunrise (Staicer et al. 1996, Catchpole & Slater 2008, Gil & Llusia 2020. This complex phenomenon can also be described at different levels: individuals within species, species within communities (Gil & Llusia 2020) and communities within soundscapes (Farina & Ceraulo 2017). A functional framework based on 12 hypotheses has been proposed to explain the existence of dawn choruses, focused on intrinsic, environmental and social factors (Staicer et al. 1996, Gil & Llusia 2020. Intrinsic factors include the circadian cycles of testosterone, and environmental factors include low predation risk, better acoustic transmission, inefficient foraging at lower ambient light levels, unpredictable overnight conditions and body condition (Kacelnik & Krebs 1982, Staicer et al. 1996, Hutchinson, 2002, Thomas et al. 2002, Brown & Handford 2003, Catchpole & Slater 2008, Gil & Llusia 2020. The role of dawn choruses has been suggested to be mostly social, although there are additional hypotheses (Kacelnik & Krebs 1982, Staicer et al. 1996, Burt & Vehrencamp 2005. In this sense, dawn choruses could be pragmatically assessed as interactive communication networks (i.e. senders both broadcast and eavesdrop information, Burt & Vehrencamp 2005), in which males convey information for mate attraction and mate guarding, territorial ownership, and resolution of social relationships among neighbours (Staicer et al. 1996, Burt & Vehrencamp 2005, Tobias et al. 2016. In this review, we identify the main research topics and general patterns so far addressed in studies of avian dawn choruses in urban settings, and provide new insights for future research directions better to understand the effect of urbanization on dawn choruses. To do this, we combined bibliometrics and systematic mapping tools (Nakagawa et al. 2019) to synthesize the main publication trends across years and the current state of the knowledge on urban bird dawn choruses across the globe.

LITERATURE SEARCH AND ANALYSIS
We performed a systematic search in Web of Science (WoS; all Databases; www.webofscience.c om) including publications from 1966 to 2018 and using the following advanced search Boolean operator string: TS = ((urban* OR city) AND (*bird OR avian OR songbird) AND (dawn OR morning OR awakening OR onset OR sunrise) AND (chorus* OR singing OR song OR call OR calls OR calling OR birdsong OR bird-song OR 'acoustic signalling' OR 'diel patterns' OR 'singing activity' OR 'dawn chorus' OR 'dawn singing' OR 'singing behav*' OR 'first song' OR 'chorus onset' OR 'earlier singing' OR 'chorus composition' OR 'chorus structur*' OR 'chorus duration' OR 'chorus time' OR 'song duration')). After screening the 55 documents retrieved in this search, we kept 17 that met the scope of this review and saved these as a BibTex file for further bibliometric analysis.
To broaden and complement our search, we searched for related literature in Google Scholar (https://scholar.google.com), which focuses on not only peer-reviewed journal publications, but also theses, dissertations, monographs, book chapters and conference publications (Haddaway et al. 2015). The search string we used in Google Scholar (GS) was: 'dawn chorus' AND 'bird' OR 'songbird' OR 'urbanisation' OR 'urbanization', which retrieved 3680 documents. After examining the title of the first 300 retrieved from GS results to exclude those beyond our scope (following suggestions of Haddaway et al. 2015), 127 remained of interest for this review. We screened all studies from WoS and GS for further analysis, considering those which were focused on dawn choruses or singing behaviour at dawn and were performed in urban areas (verifying the geographical coordinates to corroborate the location of the studied locations in doubtful cases). We excluded some studies that simulated urban conditions, such as those focused on artificial light at night and anthropogenic noise in study areas that are not urban (e.g. forest plots) (e.g. Halfwerk & Slabbekoorn 2009, Da Silva et al. 2016, 2017. This added 27 publications to those identified in the Web of Science, giving a total of 44 publications focused on urban bird dawn choruses. We assessed publication trends with the following criteria: publication year, title, subject categories (i.e. noise, ALAN, singing behaviour, noise-ALAN, meteorological factors, soundscape, supplementary feeding, chemical pollution), biogeographical region, country, city, study type (i.e. observational, correlational, experimental), study level (i.e. population, community, soundscape), and framework (i.e. intra-urban, intra-urban greenspaces, periurban greenspaces, urban non-urban gradient, urban-forest contrast). We considered three types of studies: • observational: reporting naturalistic descriptions of singing behaviour at dawn, • correlational: focused on relationships between dawn chorus traits and environmental variables, • experimental: assessing the effect of a controlled variable (i.e. anthropogenic noise, light pollution, food supply), on the dawn chorus with the aim of disentangling the mechanisms behind associations.
We also collated information on the studied bird species in order to assess their phylogenetic representation and biogeographical distribution. For this, we extracted phylogenetic trees for the studied species pool from the Birdtree database (Jetz et al. 2012). We then generated a distribution of 1000 randomly selected permutations of the phylogeny (Hackett backbone), and used 'treeannotator' software (Drummond et al. 2012) to obtain the best supported phylogenetic tree based on a maximum clade credibility phylogeny approach. To assess general global topic trends, we performed a word cloud representation (i.e. an infographic tool to summarize the general trends through the frequency of keywords of a given topic) considering publication titles. Finally, we explored relationships among main topics by exploring the links between author's keywords of each study using word co-occurrence networks ('bibliometrix' R package; Aria & Cuccurullo 2017). This approach allows visualization of the conceptual structure of a given topic based on the links between concepts through keyword co-occurrences (Aria & Cuccurullo 2017). In these networks, the nodes represent the keywords scaled by the number of studies that mentioned them, and the edges represent the co-occurrence of different keywords among studies.

General patterns
Although the study of avian dawn choruses in urban areas began in the 1960s, the number of publications has risen in the last decade ( Fig. 1). In general, studies have focused on: • the impact of anthropogenic noise on song structure and timing (e.g. Bergen & Abs 1997, Dorado-Correa et al. 2016, Marini et al. 2017, Sierro et al. 2017 • relationships between ALAN and dawn chorus onsets (e.g. Bergen & Abs 1997, Miller 2006, Kempenaers et al. 2010, • variation in singing behaviour at dawn related to time of day (e.g. Leopold & Eynon 1961, Shihan 2010, Singh et al. 2019), • the influence of temperature, cloud, moonlight and sunlight on the onset of dawn choruses (Hasan 2010, 2011, Ki et al. 2015, Hasan & Badri 2016, Lee et al. 2017 • relationships between noise, ALAN and dawn chorus timing (Bergen & Abs 1997, Nordt & Klenke 2013), • effects of chemical pollution and supplementary feeding on dawn chorus activity (Cuthill & Macdonald 1990, Gorissen et al. 2005, Saggese et al. 2011. Interestingly, there has been a recent emergence of novel approaches, such as soundscape ecology recognizing avian dawn choruses as a central component of urban soundscapes (Gage et al. 2017, Ross et al. 2018).
Most of the reviewed studies have focused on relationships between urbanization intensity and dawn chorus at the population-level (80%), whereas the community-and soundscape-levels have remained largely unexplored (Fig. 2). Moreover, 59% of the studies focused on ecological aspects (e.g. comparing variations on dawn chorus behaviour between urban and adjacent non-urban areas) and 41% performed bioacoustics assessments (e.g. evaluating how birds dealt with masking issues). Most studies were correlational (64%), assessing relationships between traffic noise, light pollution and urban intensity with dawn chorus onset (Miller 2006, Shihan 2010, Sierro et al. 2017, song output (Bergen & Abs 1997, Singh et al. 2011, Marini et al. 2017) and song frequency adjustments (Ripmeester et al. 2010, Potvin & Parris 2012, Rios-Chelen et al. 2013), but did not control for potentially confounding factors, such as differences in noise and light pollution, or variation between urban and non-urban birds (Da . A minority of studies were experimental (20%), focusing on the effects of traffic noise on chorus Most research has been conducted across the Palearctic and Nearctic regions, especially the USA, Germany, Canada and Spain (Table 1). However, there was also a high representation of studies from India ( Fig. 2) and recent studies from Argentina, Brazil, Palestine, Saudi Arabia and South Korea (Fig. 2). Studies were performed in 51 urban centres (Table 1), five of which lead in the number of publications (n = 3): Kelowna and Kamloops (Canada), Brisbane (Australia), Haridwar (India) and Mexico City (Mexico). The size of the urban settings where studies have been performed ranges from small villages (e.g. Kfar Ruppin, Israel) to megacities, such as Delhi (India). However, most studies have been performed in small-to medium-sized cities (74.5%), followed by large cities (17.7%) and megacities (7.8%) (i.e. Delhi, Mexico City, Bogotá).
The most frequent framework followed in urban bird dawn chorus studies has been to compare dawn chorus traits (i.e. timing, peak, song output, song frequencies) between urban bird populations or communities and their non-urban (usually forest) counterparts ( Fig. 2; Table 1; Leopold & Eynon 1961, Bergen & Abs 1997, Miller 2006, Potvin et al. 2011, Sierro et al. 2017, Alquezar 2018. Peri-urban and intra-urban greenspaces have also received some attention, in particular through experimental studies of the effects of supplementary feeding and ALAN on dawn chorus timing (Cuthill & Macdonald 1990, Kempenaers et al. 2010, Saggese et al. 2011 However, in contrast to general urban bird literature (see Marzluff 2017 and references therein), there are few studies based on urban-rural (or 'non-urban') gradients. Although urbanization gradients have been a widely used in urban ecology (McDonnell & Pickett 1990), their use in studies of avian dawn choruses is relatively recent and is focused at the non-urban end of such gradients. Some studies have assessed avian acoustic activity across urbanization gradients, suggesting an inverse correlation between the acoustic diversity of avian dawn choruses and urbanization intensity (e.g. Ross et al. 2018). Moreover, urban-non-urban gradient studies have also provided experimental evidence of the effects of ALAN and anthropogenic

Topic trends
Word clouds based on publication titles indicated the most representative key subjects such as dawn chorus singing behaviour, anthropogenic noise, artificial light pollution and the effects of urbanization on song frequency adjustments (Fig. 4). However, the co-word network analysis added further information on the relationships between topics in the study of urban dawn choruses. Author keywords showed a tightly connected network with five main research clusters and four peripheral ones that were not connected with the core (Fig. 5). The core cluster (depicted in purple, Fig. 5) comprises the influence of anthropogenic drivers, such as noise and light pollution, on temporal shifts of dawn choruses (Bergen & Abs 1997, Miller 2006, Gil et al. 2015, as well as variations in dawn song rates (Dominoni et al. 2016) and dawn song repertoire sizes (Trimboli 2010). The second cluster (depicted in red, Fig. 5) focuses on the behavioural plasticity of territorial songs (Laiolo 2011a(Laiolo , 2011b, variations in song length and the role of learning to deal with fluctuating noise levels (LaZerte et al. 2015(LaZerte et al. , 2016. The third cluster (depicted in orange, . Finally, the peripheral clusters include studies describing the ecological patterns of soundscapes across urban-non-urban gradients (Gage et al. 2017, Ross et al. 2018, the impact of ALAN on fitness through mate selection (Kempenaers et al. 2010), behavioural descriptions of singing activity at dawn (Pizo & Silva 2001) and the impacts of chemical pollution on dawn song repertoire size (Gorissen et al. 2005).
In the following sections, we summarize the major findings regarding the effects on urban bird dawn choruses of anthropogenic noise, light pollution, chemical pollution, meteorological factors and supplementary feeding, and the implications of these for soundscape ecology.

Anthropogenic noise
The effects of noise pollution on avian urban ecology have been well studied (e.g. Slabbekoorn 2013, Halfwerk et al. 2018. Urban soundscapes are characterized by high levels of low-frequency anthropogenic noise (2 kHz). Some bird species respond by adjusting song frequency, in particular increasing minimum frequency to avoid acoustic masking (Slabbekoorn & Peet 2003, Ripmeester et al. 2010, Potvin et al. 2011, Rivera-Caceres et al. 2011, Rios-Chelen et al. 2013. This behavioural response has been supported by further experimental work suggesting rapid acoustic adjustments in song structure to intermittent traffic noise (Arroyo-Sol ıs et al. 2013, Berm udez-Cuamatzin et al. 2018, Halfwerk et al. 2018. Changes in the timing of dawn singing routines have also been recorded (Bergen & Abs 1997, Fuller et al. 2007, Nordt & Klenke 2013, Dominoni et al. 2016. In general, urban birds tend to advance their dawn chorus onset and chorus peak times due to increasing levels of traffic noise and pedestrian activity during rush hours (Bergen & Abs 1997, Warren et al. 2006  Urban birds can also change the song repertoire used during the dawn chorus. For instance, birds in noisier urban settings produce both slower (Potvin et al. 2011) and longer songs (Rivera-Caceres et al. 2011, Marini et al. 2017, and sing more than their non-urban counterparts (Trimboli 2010). Moreover, urbanization can also result in changes to the syntax of dawn songs . For example, urban Mountain Chickadees have shown higher vocal output (i.e. number of songs in a given period of time) compared with non-urban males, but also can shift to using more songs, whereas males in quiet areas produce more calls , Marini et al. 2017.
In accordance with the predictions of the AAH (Morton 1975), studies have also focused on understanding how birds increase signal transmission in urban environments through acoustic plasticity of territorial dawn songs. There is good evidence that urban birds in noisier territories sing at a higher pitch than their counterparts living in quieter locations (Potvin et al. 2011, Laiolo 2011a, 2011b, Rios-Chelen et al. 2013, LaZerte et al. 2016. However, such short-term adjustments can also be related to previous noise experience of individual birds (Ripmeester et al. 2010, LaZerte et al. 2015. For example, Black-capped Chickadees Poecile atricapillus living in noisy areas increase the pitch of their territorial songs in response to experimental noise, suggesting that prior experience and learning processes can influence immediate adjustments to fluctuating noise levels (LaZerte et al. 2016). Studies of the role of noise pollution on dawn chorus behaviour have focused on airports due to the opportunity they provide to disentangle the influence of higher levels of noise in the absence of other urban-related confounding factors. Results from such studies have provided important evidence of changes in dawn chorus timing driven by intermittent and predictable high noise levels in a broad range of frequencies (Gil et al. 2015, Sierro et al. 2017. In general, birds living near airports show earlier dawn chorus onsets to avoid peak noise periods in the early morning (Gil & Brumm 2014, Dominoni et al. 2016, Sierro et al. 2017. However, the magnitude of such change can vary among species due to differential tolerance to noise levels; some species can be either earlier or later singers (Gil et al. 2015, Dominoni et al. 2016, although tropical birds seem to be less Figure 5. Co-word network for urban dawn chorus studies showing the most frequently studied topics via author keywords (n = 50). Some keywords were omitted to avoid overlapping terms. Node size represents keyword frequency and links the frequency of keyword co-occurrence. Colours represent clusters of keywords with a high degree of connectedness affected by exposure to airport noise (Alquezar 2018). For example, when considering studies from temperate regions, nearly half of the studied species have earlier chorus onsets (Gil et al. 2015, Dominoni et al. 2016, whereas in tropical sites, only two of 15 studied species showed significant advances in their dawn chorus onset (Alquezar 2018).

Light pollution
ALAN is a global ecological change that has received research attention in recent years due to its interference with circadian rhythms and the selective pressures it imposes on organisms in cities (Hopkins et al. 2018). In fact, ALAN has been suggested to be an important threat to nocturnal migratory bird species, as well as a driver of avian behavioural change and reproductive success in cities (Kempenaers et al. 2010, Isaksson 2018. Field experiments suggest that light pollution can disrupt sleep (Raap et al. 2015), as well as the daily singing routines of birds (Da Silva et al. 2016. For example, forest songbirds adjust the timing of their dawn choruses to the experimental manipulation of artificial light levels (Da Silva et al. 2016). However, a recent study showed that experimental illumination of a forest using lights of different colours had no effect on the dawn chorus timing of 14 temperate songbird species (Da . These studies were performed outside urban areas to avoid confounding urbanization effects such as built and vegetation covers, noise pollution, increased temperatures, and altered food supply (Da .
Correlative studies of ALAN have provided evidence of earlier dawn chorus onsets in light-polluted urban sites (Miller 2006, Shihan 2010, Nordt & Klenke 2013, Ki & Cho 2014, which also interact synergistically with noise pollution (Bergen & Abs 1997, Nordt & Klenke 2013, Dominoni et al. 2020a. Given that songbirds are capable of adjusting their dawn chorus onsets to variations of both natural and artificial light levels (Leopold & Enyon 1961, Da Silva et al. 2016, Lee et al. 2017, the effects of ALAN are expected to be strong in cities at high latitudes where the daily photoperiod varies across the year (Da . The effects of ALAN are also expected to affect earlier-singing songbirds (Edenborg 2018). Interestingly, the few recent studies performed in tropical cities have found no relationships between dawn chorus onsets and ALAN at the population-(Dorado-Correa et al.

2016) or community-level (Mar ın-G omez &
MacGregor-Fors 2019), suggesting that ALAN does not seem to represent a driver of the singing routines of tropical urban birds.

Chemical pollution
Although chemical pollution is an important influence on urban avian ecology (Isaksson 2018), we found only one study exploring its impact on chorus onset and song output (Gorissen et al. 2005). This study found that Great Tits exposed to higher levels of heavy metal pollution showed smaller repertoire size and lower song rate during their dawn chorus. Environmental pollution may therefore influence song production by constraining available time and energy budgets for vocal activity (Gorissen et al. 2005).

Meteorological factors
Evidence from Middle Eastern cities suggests that time changes in dawn choruses (i.e. earlier onset) are associated with higher environmental temperatures (Hasan 2010). Other meteorological variables, such as higher wind, precipitation and cloudiness, and lower temperatures have also been suggested to delay dawn chorus onsets (Leopold & Enyon 1961, Nordt & Klenke 2013, Hasan & Badri 2016. For instance, longer days, together with high temperatures during summer, can lead to earlier House Sparrow dawn chorus onsets (Hasan & Badri 2016), and Nordt and Klenke (2013) suggest that cloud cover in cities could act as a reflector that can amplify the sky luminance produced by city lighting and thus increase the effects of ALAN on advancing the dawn chorus onset of Common Blackbirds. This suggestion shows how the interaction between environmental and anthropogenic variables may affect avian dawn chorus timing.
Another important environmental factor related to avian dawn choruses is lunar phase (York et al. 2014). Findings of non-urban studies suggest that birds show earlier dawn chorus with moonlight presence at dawn (i.e. full or three-quarter moon; Bruni et al. 2014, York et al. 2014). Yet this effect varies among species, and seasonally. For example, American Robins sing much earlier during moonlight at the beginning of spring but not during summer (Leopold & Enyon 1961). However, other studies have shown no correlation between dawn chorus onset and lunar phase or cloud cover in urban settings (Ki & Cho 2014, Ki et al. 2015. Perhaps the most important meteorological variable related to dawn chorus timing is variation of light intensity at dawn. As light levels increase before sunrise, bird species join dawn choruses in a specific time window due to differences in species' sensitivity to light (Thomas et al. 2002, Berg et al. 2006, Chen et al. 2015. For example, a recent study performed in urban greenspaces of Los Angeles (CA, USA) showed that light levels could influence the timing of dawn chorus activity (Lee et al. 2017).

Supplementary feeding
Much evidence suggests that food supply is a main influence on the ecology of urban birds (Amrhein 2014). Among other effects, supplementary feeding can enhance territoriality and increase song output (Cuthill & Macdonald 1990, Amhrein 2014. Consequently, social hierarchies can be influenced by body mass differences between dominant and subordinate birds, particularly at dawn (Grava et al. 2009). Two studies from peri-urban greenspaces have explored the effects of supplementary feeding on dawn chorus onset and song structure. One of them showed that Common Blackbirds sang earlier at dawn and had greater song output when supplemented with food, due to the influence of energetic requirements on male phenotypic condition (Cuthill & Macdonald 1990). In contrast, another study showed the inverse pattern for Great Tits, with individuals under continuous food supply conditions showing delayed dawn chorus onsets (Saggese et al. 2011). This unexpected result was explained by the presence of predators at feeding stations and the suboptimal energetic quality of the food supply, as predicted by formal models of daily singing and foraging routines in birds (Hutchinson 2002).

Soundscape ecology
Recent advances in soundscape ecology have highlighted the value of avian dawn and dusk choruses because their predominance in multiple soundscapes around the world and the information included in them can be used to explore the impacts of anthropogenic global driving forces on biodiversity (Krause & Farina 2016). Studies have suggested that avian dawn choruses are important indicators of soundscape structure across urbanization gradients (Gage et al. 2017, Ross et al. 2018. For example, a study from peri-urban Brisbane (Australia) showed that avian dawn choruses can be identified by abrupt changes of energy power, as revealed by acoustic indices (Gage et al. 2017). Moreover, a recent study that explored the impact of urbanization on soundscape patterns in Okinawa (Japan) also suggests a relationship between some acoustic indices and temporal variations in avian choruses across urban gradients (Ross et al. 2018).

FUTURE DIRECTIONS AND CONCLUSIONS
In this review we have synthesized the available knowledge on the impacts of urbanization on an emergent property of the behaviour of many individual birds of multiple speciesthe dawn chorus. Beyond previous reviews on the topic outside urban settings that provide general explanations of why birds sing at dawn (Staicer et al. 1996, Catchpole & Slater 2008, Gil & Llusia 2020, our findings show short-term adjustments in the timing, song frequency, song repertoire and syntax in response to urban noise; earlier dawn chorus onsets in light-polluted urban sites; smaller repertoire size and lower song rate in sites exposed to chemical pollution; and increase of song output and delayed chorus onsets because of supplementary feeding. Our results also shed light on important future research directions.
Due to the biological importance of dawn choruses in territorial signalling and reproductive success (Staicer et al. 1996), most research has focused on understanding how anthropogenic noise affects the structure and temporal traits of songs emitted during this critical period. As result, most of the reviewed studies suggest significant changes in different song traits at dawn. Additionally, although correlative and experimental studies have shown how birds increase the minimum frequency of dawn songs to increase sound transmission in noisier urban environments (Ripmeester et al. 2010, Potvin et al. 2011, Potvin & Parris 2012, Rios-Chelen et al. 2013, the mechanisms behind such patterns are not well understood. Moreover, not all bird species respond in a similar fashion to urban noise. Short-term and nonadaptive changes (e.g. acoustic plasticity) are well documented (e.g. Ripmeester et al. 2010, Arroyo-Sol ıs et al. 2013, Gil et al. 2014, whereas evolutionary responses, such as cultural evolution of song, have received less attention (e.g. Luther & Baptista 2010, Potvin et al. 2011, Potvin & Parris 2012. Thus, we encourage future studies to assess the consequences of those adjustments on individual fitness. In fact, whether acoustic phenotype adjustments driven by urbanization are adaptive or not remains a controversial issue in urban avian ecology (Nemeth & Brumm 2010, Slabbekoorn et al. 2012, Gil & Brumm 2014.
The timing of dawn choruses is affected not only by urban-related stressors (anthropogenic noise, ALAN) and meteorological factors, but also potentially by the abundance of conspecific neighbours (Hodgson et al. 2018, Stuart et al. 2019, perceived predation risk (Santema et al. 2019), body condition (Kacelnik & Krebs 1982) and transmission properties of the urban environment (Phillips et al. 2020). Predation pressure has been shown to increase with urbanization (Fischer et al. 2012); yet, potential influences of predation on dawn chorus timing remain untested. Additionally, although supplementary feeding is widespread in cities and can alter avian body condition (Amrhein 2014, Phillips et al. 2018, Baverstock et al. 2019, the effects of feeders on dawn chorus timing remain controversial, so that a larger sample of studies of its impacts would be valuable. Given the multifunctional nature of the dawn chorus, experimental studies combining controlled noise exposure, experimental light conditions and food provisioning treatments could shed light on the mechanisms related to dawn chorus adjustments (Arroyo-Sol ıs et al. 2013, Marini et al. 2017). For instance, there is a pressing need to understand the mechanisms behind the identified patterns, such as physiological mechanisms (i.e. how does artificial light pollution affect avian circadian rhythms across latitudes? e.g. , behavioural mechanisms (i.e. do short-term responses, phenotypic plasticity or cultural evolution drive dawn chorus timing or structure? e.g. Gil & Brumm 2014) and ecological mechanisms (i.e. how do bird communities from heavily urbanized sites use the acoustic space differently from non-urban communities? e.g. Mar ın-G omez et al. 2020).
Tackling geographical bias is also crucial to developing further an understanding of the patterns and processes of urban dawn choruses. Currently, the main body of knowledge comes from studies conducted in the Palearctic and Nearctic regions, with substantial information only for a few species. Therefore, more studies from a more representative sample of regions and species are needed to understand the factors driving dawn chorus adjustments.
We consider that assessing avian dawn choruses in cities at the community-level could provide important insights into the use of the acoustic space by bird communities in different urbanization conditions, and in particular, how urbanization could change the structure of dawn choruses and how exotic species use the acoustic space and potentially limit the signalling of native species. For example, a recent study in a neotropical city compared the structure of the avian dawn choruses at the community-level in both intra-urban and peri-urban areas (Mar ın-G omez et al. 2020). The findings of that study suggested a modular structure of avian dawn choruses in peri-urban areas (i.e. groups of co-occurring bird species singing at different times through the morning) supporting the temporal partitioning of the acoustic space, but the loss of the temporal order of dawn choruses at intra-urban areas related to the depauperization of the avian community together with the dominance of the acoustic space by invasive species (Mar ın-G omez et al. 2020). This study calls for the need to consider the influence of different sources of noise in urban settings and its influence on the partitioning of the acoustic space.
Although we identified studies from cities of differing sizes, there is an under-representation of evidence from megacities, contrasting with a preponderance of data from megacities in other arenas of urban ecology. Currently, most studies are based on urban-non-urban contrasts, with few considering urbanization intensity gradients (e.g. Marini et al. 2017, Ross et al. 2018. Thus, to understand acoustic phenotype adjustments to urbanization, studies should include survey sites across urbanization intensity gradients. This would better represent the differing acoustic transmission traits across cities (e.g. Phillips et al. 2020) by considering their environmental heterogeneity. Thus, future studies should consider urbanization intensity gradient frameworks to explore the influence of a wider array of urbanization conditions on avian dawn singing routines.
We were surprised to find no study assessing relationships between increasing temperature in cities (i.e. urban heat islands) and singing behaviour and sound transmission across urban to nonurban gradients. High temperatures have been shown to correspond to earlier dawn chorus onsets (e.g. Hasan 2010, Hasan & Badri 2016, but the associated mechanisms remain unknown. In some cities from Saudi Arabia and Palestine, for instance, where daily temperatures typically exceed 30°C, House Sparrow singing routines start earlier to avoid higher temperatures during mornings (Hasan & Badri 2016). High temperatures can also advance the breeding season and stimulate earlier dawn chorus onsets (Dominoni et al. 2020b), but they may also constrain sound transmission by increasing the absorption of lower frequencies (Farina 2013). Under climate change, and with the increasing role of urban heat islands, we suggest that high temperatures, together with noise and artificial light pollution, may disrupt the circadian rhythms of urban birds and change their singing routines.
Finally, we consider that attention should importantly be paid to the recent boom of soundscape ecology studies and the use of passive acoustic monitoring as a tool to evaluate biodiversity through soundscapes (Farina & Caeraulo 2017). Such technologies and procedures will allow the generation of a significant amount of data that would allow us better to understand the influence of urbanization on avian choruses at all levels of organization at larger scales. As bird dawn choruses are a fundamental component of urban soundscapes and provide multiple cultural services, we forecast an increase in their importance in urban ecological studies, as well as management and planning strategies directed toward fostering sustainable and healthier cities