Introduction

Urbanization is a key anthropogenic driver of accelerated biodiversity loss (Aronson et al. 2014). It causes profound alterations in local environmental systems, including ecological, climatic, hydrological and biogeochemical systems, which ultimately affects the suitability of habitats used by species (Folke et al. 1997; Grimm et al. 2008; Kaye et al. 2006). Moreover, the impact of urban areas extends beyond their physical footprint, as surrounding ecosystems are drawn upon for resources and ecosystem services (Folke et al. 1997), while urban waste and pollutants spread into surrounding ecosystems (Grimm et al. 2008). The modification and destruction of habitat associated with this process leads to changes in both the abundance and richness of species in urban and surrounding environments (Aronson et al. 2014). The threats posed by urbanization to biodiversity are expected to intensify as cities expand to support an increasingly urban-centric human population (Angel et al. 2011; Seto et al. 2012). Management and planning interventions are key in preventing urbanization driven species extinctions and fostering urban biodiversity, particularly through the strategic planning of cities to preserve more remnant vegetation (Hahs et al. 2009), the establishment and maintenance of urban greenspaces to provide more habitat and connectivity in the urban matrix (Threlfall et al. 2017) and the prevention of urban sprawl from converting more natural habitat (Sushinsky et al. 2013).

The motivation to conserve the biodiversity within urban ecosystems is increasing as their global footprint expands in areas that are often fertile, have high primary productivity and are relatively biodiverse (Dearborn and Kark 2010; Luck 2007a, 2007b). As a result, despite the proportionally small global land area occupied by urban areas, and a traditional view that they mainly contain common (especially invasive) species and are barriers to movements and dispersal (Chace and Walsh 2006), urban environments support a significant number of threatened species, with some being recognized as threatened species hotspots (Ives et al. 2016). Research into the impact of urban expansion on threatened species has been rapidly increasing, particularly for birds (Evans et al. 2009; Marzluff 2017) This focus on birds is likely associated with their status as sensitive ecological indicators (Browder et al. 2002; Canterbury et al. 2000; Croci et al. 2008), their high public value, both intrinsically and as a connection to nature (Cox and Gaston 2016), as well as being easy to detect, identify and monitor (Baillie 1991; Chace and Walsh 2006).

Despite the challenges associated with living in human-dominated landscapes, approximately 75% of bird families and 20% of all bird species are found in urban environments (Aronson et al. 2014), often in abundance (Chace and Walsh 2006). Relatively high species richness has been observed in urban parklands and peri-urban areas in particular (Catterall et al. 2010; McKinney 2008). However, urban environments often have a homogenizing effect on bird biodiversity (McKinney 2006; McKinney and Lockwood 1999), with a subset of species, often coined ‘urban exploiters’ (Blair 1996), predominating the avifauna of many cities. Many urban exploiters are from a small suite of introduced species, resulting in urban avifaunal assemblages becoming increasingly similar at a global level (Chace and Walsh 2006).

Although well studied in urban environments, the focal topics of research on urban birds are not well understood. Studies in urban ornithology have shifted away from exploring temporal and spatial trends and patterns, toward determining the processes that dictate patterns of bird occupancy in urban environments (Marzluff 2017). The cause of this shift may be linked to the expectation that such studies should highlight potential practical applications to meet journal submission criteria (Miller 2019). Miller (2019), however, highlights that this link is usually established through the assertion that understanding the ecology of species can inform the management of urban habitats, rather than providing concrete conservation recommendations. This suggests that there may be a somewhat superficial link between these process-orientated studies and conservation application, resulting in a potential deficit in research that has direct conservation implications.

Potentially exacerbating this deficit is that research effort is not consistently distributed among species. There appears to be a collective bias towards species that are iconic or popular (Martín-López et al. 2009), belong to certain taxonomic groups, are larger, accessible (Ducatez and Lefebvre 2014; Yarwood et al. 2019), and have significant ecological roles such as keystone species (Mace et al. 2007). As a result, a significant proportion of species receive very little conservation research attention, even as many common species experience substantial declines (Gaston and Fuller 2008). Common species, although often overlooked, play important ecological roles as foundational species, particularly in urban areas (Gaston and Fuller 2008). The combination of these biases may culminate in an inequitable representation of species in the research literature, a dearth of certain focal themes in urban ornithology and an overall lack of focus on practical applications to achieve conservation outcomes, all of which can compromise efforts to conserve biodiversity through the retention and enrichment of urban ecosystems. Given the potential risk, it is crucial to better understand how birds are being studied in urban environments.

In the present study, we conducted a systematic quantitative review of the literature to assess how urban birds are studied in Australia. When compared to many other cities globally, Australian cities are young and sprawling, with lower human density and higher levels of remnant vegetation (Hall 2010; Lonsdale and Fuller 2014). Due to an increasing population, Australia’s urban areas are growing rapidly, both through the infill of inner cities and the expansion of outer suburbs (Cresswell and Murphy 2016). Like many other urban bird communities globally, the intensification of urbanization poses a threat to Australia’s highly diverse avian community, which represents approximately 10% of global bird biodiversity (Chapman 2009), particularly as many of Australia’s threatened species overlap with urban settings (Ives et al. 2016). Australia has a substantive urban bird literature with a marked increase recently in the number of publications published per year (Marzluff 2017). In Australia, along with the USA, Canada and Europe, urban bird studies are becoming increasingly centered on mechanistic themes (i.e., why species are reacting to urban environments) rather than diagnostic themes (i.e., assessing how species are responding to the threat of urbanization) (Marzluff 2017). We hypothesized that biases in research effort will be multifaceted, with a synergy between the uneven representation of species, a lack of diversity in the types of studies that species are included in, and a lack of highly specific conservation studies resulting in a disparity between how common and threatened species are studied. Our analysis aimed to identify potential biases in urban bird research which may be resulting in deficiencies in urban bird conservation. Specifically, we aimed to identify biases in the representation of species in the literature, the areas of research that are most prominent, and whether there are species characteristics associated with differences in research effort.

Methods

Literature search and dataset compilation

We completed a search of the literature to identify studies that focus on birds and occur in Australian Significant Urban Areas (SUAs), which are areas containing towns and cities with a population of 10,000 or more people (Australian Bureau of Statistics 2016). The search and dataset compilation process was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement (Moher et al. 2009) and the systematic quantitative literature review protocol described by Pickering and Byrne (2014).

The search was conducted through two online scientific journal repositories, Clarivate’s Web of Science and Elsevier’s Scopus, using the following combination of keywords: (bird* OR avian OR avifauna) AND (urban* OR suburb* OR city OR cities) AND (Australia* OR Queensland OR “New South Wales” OR Victoria* OR “Northern Territory” OR Tasmania*). Studies were restricted to English language, peer-reviewed, primary research articles.

A total of 505 studies were identified by the search after removing duplicates (Fig. 1). Only studies that included complete species lists, occurred in a SUA and included a substantial urban or suburban element (including parkland, urban edges/remnants < 100 ha in area) were retained in the dataset. A forward and backward citation search was completed to ensure thorough coverage of the literature. This involved inspecting the cited and citing references of a random selection of studies from the database using either Web of Science or Scopus, for articles meeting the inclusion criteria. Studies which met the criteria were subsequently included in the database. This process was repeated until 20% of studies from the original database had been searched.

Fig. 1
figure 1

PRISMA diagram showing the systematic methods used to search, collate and filter articles for review

Full size image

We allocated each study a binary score for a set of defined variables of interest. Variables included species inclusion, study scope, conservation implications and study themes (Table 1). Species taxonomy followed The BirdLife Australia Working List of Australian Birds (BirdLife Australia 2019). For each species, we calculated a study count for each study variable, study level specificity for each study that the species was included in and mean specificity across all studies in which that species occurred. Specificity of species i in study j was calculated as the proportion of the total species count in that study Nj that a species represents:

$$Specificity_{i,j} = \frac{1}{{N_{j} }}$$
Table 1 Description of variables included in the scoring database
Full size table

Species characteristics

For each species included in at least one study, data were collated on characteristics of interest: threat status was based on the Australian IUCN Red List Status reported for each species in the BirdLife Australia Working List of Australian Bird (BirdLife Australia 2019) summarized into three categories, threatened, non-threatened and introduced. Species order was also based on the BirdLife Australia (2019) taxonomy. Species range sizes, in the form of the number of 1° degree grids occupied and the maximum 1° degree grid reporting rates, were obtained from Barrett et al. (2003). The average body mass for each species was from Garnett et al. (2015) and log-transformed. Species were classified as either migratory or resident based on data compiled in Yarwood et al. (2019) and BirdLife International IUCN Red List Fact Sheets (BirdLife International 2020). Each species also had a time value calculated representing the number of years the species occurred in the literature (from the first appearance to the last) and the overall study count.

Modelling

We conducted a series of generalized linear models (GLMs) using the statistical software R version 4.0.3 (R Core Team 2020) to determine which species characteristics were significantly associated with species inclusion (i.e., presence) in different study themes. Specifically, Gamma GLMs were used to model the number of studies per species and the species specificity of the research as a function of the species level variables. In addition, Binomial GLMs were used to model which species characteristics were predictors of species presence in each study type. We minimized the risk of overfitting by systematically removing predictors to minimize the Akaike information criterion (AIC). For categorical variables, we selected the following reference groups: Passeriformes were selected as the reference order as it was the largest group, Non-Threatened species were selected as the reference for IUCN Red List Status and Resident species were selected as the reference for species movement.

Results

A total of 519 species (55% of all species occurring in Australia) and 18 orders (86% of all orders occurring in Australia) were represented in the 242 studies included in the final dataset. Low conservation context studies (undefined and adjacent) were more common than high conservation context studies (direct and applied), representing 75% of all studies. Applied conservation studies were the least common conservation context, representing 5.4% of studies. The majority of studies had a narrow scope, focussing on specific species, represented by a total of 65% of studies. Study scope varied with conservation context, with studies with broad scope (location specific) being more likely to have high conservation context than studies with narrower species scope. Studies examining trends or changes in landscapes were the most common thematically, followed by studies of behaviour (Table 2). The number of species occuring in each study theme and species characteristic group was highly variable (Table 3).

Table 2 Summary of the number of studies and number of species in each category
Full size table
Table 3 Summary of results for all species characteristics (excluding order) by study group
Full size table

Predictors of number of studies per species

The species occurring in the highest number of studies was the Australian magpie (Gymnorhina tibicen; N = 74 studies), followed by the noisy miner (Manorina melanocephala) and the common myna (Acridotheres tristis) with N = 68 for both. The data was skewed toward lower study counts for species, with 178 species occurring in only one study, and a median of three studies per species (see Supplementary Material A for full species list). Coraciiformes (kingfishers, bee-eaters and allies) were the order with the highest mean study count, followed by Psittaciformes (parrots) and Cuculiformes (cuckoos).

The model incorporating species characteristics as a predictor of the number of studies per species showed that there were significant differences between species based on their characteristics (Table 4). Migratory and vagrant species had significantly fewer studies than resident species. The number of studies per species also significantly differed between different orders, as parrots (Psittaciformes) occurred in significantly more studies than the reference group, the passerines (Passeriformes), while shorebirds (Charadriiformes), large waterbirds (Pelecaniformes), grebes (Podicipediformes), waterfowl (Anseriformes) and owls (Strigiformes) occurred in significantly fewer studies relative to passerines. There was also a small positive effect of range, the amount of time that the species had been mentioned in the literature and average species body mass, on the total number of studies per species.

Table 4 A summary of all models used in this study, including formulae and significant predictors
Full size table

Predictors of species study specificity

The mean specificity value (the proportion of the total number of species in a study that a species represents) highlighted that four species exclusively occurred in species specific studies (with a mean specificity value of (1): the fairy tern (Sternula nereis), the little penguin (Eudyptula minor), the pied imperial pigeon (Ducula bicolor) and the far eastern curlew (Numenius madagascariensis). Most species, however, had low mean specificity, with the mean value across all species being 0.04. Using the sum of all specificity values for a species (across all the studies that it was included in) highlighted which species occurred in a high number of highly specific studies. The common myna and the Australian white ibis (Threskiornis moluccus) had the highest summed specificity scores (see Supplementary Material A for all specificity scores).

This model also showed significant differences among species (Table 4). Threatened species were significantly more likely to occur in more specific studies compared to non-threatened and introduced species. Body mass and species reporting rate were also demonstrated to have a positive effect on species study specificity, with heavier species and species that have a higher reporting rate occurring in more specific studies.

Predictors of species presence in different study scopes

A total of 75 species were represented in the 156 species specific studies, with the common myna occurring most frequently in this study type (N = 17). There was a total of 61 studies that included all species detected in a landscape, representing the majority of species (N = 510). Modelling showed significant differences in study inclusion based on species characteristics (Table 4). In species specific studies, threatened species were significantly more likely to be present compared to introduced and non-threatened species. By contrast, in less specific studies, threatened species were significantly less likely to be present compared to non-threatened and introduced species. Species specific studies were also more likely to include species that are not migratory or vagrant and had more literature (higher study count and duration in the literature). Occurrence in species specific studies also varied by order, with Accipitriformes (birds of prey), Anseriformes (waterfowl), Caprimulgiformes (nightjars and frogmouths), Charadriiformes (shorebirds), Galliformes (landfowl), Gruiformes (coots, cranes and rails) and Strigiformes (owls) being significantly more likely to occur in species specific studies relative to passerines, which were the reference group for model comparison.

Predictors of species presence in different conservation contexts

Applied conservation was the least frequent conservation context (N = 13) followed by direct conservation studies (N = 44). The 55 conservation adjacent studies represented nearly every species included in this analysis, thus was excluded from modelling. Studies with undefined conservation context were the most frequent (N = 130).

Modelling showed that the likelihood of being included in applied conservation studies was significantly higher (relative to the reference group, the Passeriformes) for Accipitriformes (bird of prey), Caprimulgiformes (nightjars and frogmouth) and Strigiformes (owls), and significantly lower for Anseriformes (waterfowl) (Table 4). Charadriiformes (shorebirds) were the only order to significantly differ from Passeriformes in direct conservation studies, with them significantly more likely to be included. Species study count had a positive correlation with study inclusion in both applied and undefined conservation studies.

Predictors of species presence in different study themes

Studies exploring changes in landscapes spatially and temporally were the most common type of study, with 170 studies (70.3%) falling into this category, which contained all possible species. Given that all species were present in the landscape category, a binomial GLM was not applied to this category.

Introduced species were significantly more likely to occur in management studies than both non-threatened and threatened species, while both threatened and introduced species were more likely to be included in resource related studies compared to non-threatened species. Inclusion in resource related studies was also influenced by the average mass of the species, with a positive correlation between mass and study inclusion. Average body mass was also positively correlated with inclusion in behavioural studies. Migratory and vagrant species were more likely to appear in behavioural studies than non-migratory. The only significant predictor in physiology related studies was their total study count, suggesting a preference toward more commonly studied species. Management and human-wildlife interaction studies were more likely to include heavier species, as well as species that have occurred in the literature for longer.

Discussion

This is the first study to explore the distribution of research effort among species and themes in Australian urban ornithology. We found research effort and theme to be highly variable between species and groups of species, with a complex interplay of characteristics predicting inclusion of species in studies of particular types. We found that only a small proportion of studies fell into the applied and direct conservation groups, indicating either a low collective priority to implement conservation action in urban environments, a lack of impetus to include implemented urban conservation actions in the scientific literature or, as suggested by Miller (2019), a tendency to create somewhat superficial links between ecological findings and conservation implications. We also found that while many apparently common species were included frequently in the literature (i.e., were present in many studies), this does not equate to comprehensive representation across study themes, with many common species only occurring specifically in weakly conservation linked research. These findings suggest that there are deficiencies in knowledge pertaining to how species are responding to the threat of urbanization. It is crucial to avoid such complacency so that declines in urban species are not missed in this era of rapid change (Seto et al. 2012).

Some general trends in terms of what predicts species inclusion in particular study types were evident. Our results are consistent with the prediction that threatened and non-threatened species would be represented differently in the research, particularly in terms of research scale and context. Threat status influenced the likelihood of being included in studies, with threatened species significantly more likely to occur in species specific studies than non-threatened and introduced species and to have high specificity scores, illustrating their prioritization in the literature. Threatened species were also more likely to occur in studies about resources when compared to non-threatened species. Conversely, threatened species were less likely to occur in studies that were looking at broad suites of species, such as studies that are focussed on all species in an environment, and in management related studies, or studies with undefined conservation contexts. Many common species are classified as Least Concern on the IUCN Red List due to their broad range size and their presence in modified habitats (Birdlife International 2018). However, this may not be indicative of how individual species are faring in densifying urban environments, especially with evidence highlighting a broad scale decline in bird populations, even for common species (Rosenberg et al. 2019). While this may not be a concern for the broader population, the loss of common species, which play an important role in species assemblages (Gaston and Fuller 2008) may be detrimental to urban ecosystems more broadly. There were few significant predictors of a species occurring in applied conservation studies, which was likely a result of the limited number of studies in this category. Over half of the applied studies focussed on a broad set of species, exploring landscape level conservation (i.e., improving abundance and richness of a locale) which further influenced the model. One clear signal was that applied conservation studies seemed to be predominated by raptors, as Accipitriformes and Strigiformes were both significantly more likely to be included in this study type than the reference group, the Passeriformes. This may stem from the fact they are a charismatic group of species that are generally very distinct, or from the general appeal of vertebrate predators (Sergio et al. 2006). In terms of direct conservation studies, Charadriiformes (shorebirds) were the only group significantly more likely to be included compared to Passeriformes. This is likely linked to the fact that Australia’s urban areas are predominantly located on the coast (Australian Bureau of Statistics 2018), and as such there is often an overlap between human populations and shorebird habitats (Antos et al. 2007). Many shorebirds are international migrants, which exposes them to an additional suite of threats and means that they are often subject to internationally coordinated conservation efforts, which potentially explains their significantly higher inclusion in direct conservation studies.

Considering the species that have the highest number of studies, these were all species that occur commonly in urban environments and are easily identifiable. These species occurred in so many studies axiomatically because they are common, and thus appear regularly in landscape level studies. The commonness of these species also means that they are easily accessible and thus are used as test cases for a broader suite of study types. As an example, the species that appeared in the highest number of studies were all common species that are often negatively perceived: the Australian magpie is well-known for aggressive interactions with humans during their breeding season, the noisy miner, an aggressive honeyeater that thrives in moderately disturbed environments and are a major contributor to species richness declines in these environments (Maron et al. 2013; Parsons et al. 2006; Piper and Catterall 2003) and the common myna, a widespread invasive species thought to outcompete native species in urban environments (Pell and Tidemann 1997). While many of the common species’ high study counts could be attributed to being included in many broad scale studies, common mynas and noisy miners were particularly prevalent in species specific studies and behavioural studies, while occurring in surprisingly few applied management studies (noisy miners N = 0, common mynas N = 2). This suggests a lack of action pertaining to the direct management of introduced or pest species. These findings may also link to Marzluff’s (2017) conclusions that research focus in urban ornithology has shifted toward why certain species are successful in urban environments, as the species identified as threats to Australian urban assemblages are usually studied with regards to their behaviour rather than on management.

To address the biases highlighted in this study, the source of the biases first needs to be identified, particularly the lack of applied conservation studies. Deficits in the scientific literature may be caused by a disconnect between academics and practitioners, leaving conservation planning and monitoring only reported in the grey literature. It may be beneficial to bridge this gap and include implemented conservation actions in scientific literature, providing an opportunity for more collaboration between academics and practitioners to improve conservation outcomes (Miller 2019). Such change would likely need to be supported by journal publishers, particularly through the encouragement of reporting applied conservation action by academics and practitioners alike. However, if these biases exist because of a lack of impetus to conserve urban ecosystems, the solution likely lies in the advocacy of urban ecosystem conservation, ideally leading to increased attention among conservation scientists and practitioners and much needed funding.

As these findings illustrate, urban bird research is highly diverse, which makes it difficult to summarize without generalization. Applying systematic methods and using clear categories reduced the risk of subjectivity and facilitates repeatability. Through the use of these methods, this study provides insights into deficits in urban bird conservation research broadly, even through the lens of Australia’s unique urban history, ecosystem and bird life (Hall 2010; Lonsdale and Fuller 2014; Schodde 2006). Australian cities are generally more sprawling, lower density and retain more (albeit fragmented) vegetation compared to older cities on other continents (Hall 2010; Lonsdale and Fuller 2014). These low-density areas (particularly suburbs) typically support high species richness (Catterall, Cousin, Piper and Johnson). Australia is also one of the few countries to have urban vegetation assemblies that significantly differ from surrounding local native floral assemblies (Aronson 2014), which has been noted to further alter Australian urban bird assemblages (White et al. 2005). As such Australian urban ornithological studies likely include a greater diversity of species than elsewhere globally. Additionally, while introduced species are playing a significant role in the homogenization of urban assemblages in many cities globally (McKinney 2006), the strong focus on introduced species in Australian studies may relate to Australia having one of the highest numbers of exotic bird species in urban areas in the world (Aronson 2014). Despite these unique attributes, Australia shares similarities with other countries including the USA, Canada and Europe regarding the way that urban birds are studied, with all of these countries shifting toward a focus on mechanisms of change in urban bird communities (Marzluff 2017). Furthermore, the tendency to neglect conservation in urban environments seems to be widespread; it is likely that biases related to conservation context will be similar across continents.

Urbanization poses an enormous challenge for conservation, leading to the current rapid biodiversity loss (Aronson 2014). Understanding how to foster biodiversity in urban areas is crucial in preventing more extreme losses in an increasingly urbanized world. To achieve this, a strong focus on understanding how species fare in urban environments is necessary to allow the timely implementation of conservation actions that improve urban biodiversity. Common species play an important role in ecosystem structure and function, and as such it is important that they are represented in conservation research so that further biodiversity loss can be mitigated. While many common species are well represented in the literature in terms of the number of studies, this study challenges the idea that this equates to adequate representation, as the way that species are covered in the literature varies considerably and conservation scientists should be mindful not to be complacent in an era of rapid change. This study also draws attention to the low proportion of urban bird studies with direct or applied conservation outcomes. This indicates a clear need for future studies to further explore biases in the way that birds and other taxa are studied in urban environments and highlights a lack of conservation application or conservation scientific publications in relation to urban areas. Replicating this study on other continents would be a valuable way to assess whether the biases highlighted in this study are broadly applicable, or whether the differences in the way that urban ornithology is studied across different countries perpetuates disparate biases.