The number of people that live and work in an urban environment is expected to increase further in the coming decades1. Consequently, urbanisation has become a major cause of biodiversity decline, both within city borders, due to the conversion of natural habitats into urban land uses12, and outside cities, due to indirect impacts, such as the need to produce food for city dwellers14. Even though the biodiversity in cities is generally lower than in the natural habitat replaced by the urban environment, a surprisingly high number of species can occur in cities2–6. Cities have been shown to provide important habitat for threatened species, particularly in biodiversity hotspots15, providing a case for urban species conservation for the sake of biodiversity itself. In addition, urban nature is the nature that city dwellers can experience in their day-to-day life12,13. An increasing body of evidence suggests that nature has multiple positive effects on urban inhabitants16,17, including the provisioning of ecosystem services18 such as regulating climatic conditions or mitigating extreme events such as heavy rainfall or heat waves19, as well as direct positive effects on human health and well-being20, including a reduced risk for allergies, stress reduction and providing a sense of place16,17,21. Today, there is widespread agreement on the need to safeguard and increase biodiversity in cities7–9,22, which, however, requires a better understanding of what drives biodiversity in the urban environment3,11.
Human societies are often disconnected from nature23, to the extent that many consider cities, the place where civilised humans live, to be functionally different from the ‘wilderness’, which is the place of most other animals24. This separation between people and nature has also influenced ecology. For a long time, there was little interest in studying species in cities, i.e., in the ‘unnatural’ environment. Detailed assessments of plants and animals in the urban environment started only in the 1980s3,25–27, often using transects from outside a city to the city centre or comparing species compositions between urban and non-urban areas28–31. These comparisons have shown that urban communities differ from those outside the cities so that certain traits and taxa are overrepresented32, e.g. the Fringillidae (finches) and Columbidae (pigeons) among birds33. However, clear differences only emerged when extra-urban communities were compared to communities from the inner city (high proportion of impervious surfaces) but not compared to communities from urban areas with a lower fraction of impervious surfaces (e.g., suburban areas). Such results suggest that urban communities are not homogenous and that properties of the particular site in the urban environment determine what species can occur there34,35. Consequently, studies have begun comparing sites within cities to understand why certain sites hold higher biodiversity than others35. Studies comparing different land uses within cities, or focussing on particular habitats such as gardens3,36,37, found that various features such as total vegetation, vegetation structure, or management, but also the presence of potential habitats in the surroundings affect the diversity of a site in the urban environment. In a large meta-analysis, the size of the green area within a city, the presence of corridors, and vegetation structure were important to explain local species richness for several taxa38. Importantly, taxa can differ in their responses to the features of urban sites39. Thus, while the overall greenness (vegetation cover) of a site has been shown to affect the diversity of birds, bats and pollinators40, greenness is comprised of a variety of components, which might affect individual taxa differently. Investigating the effects of individual site features other than overall greenness may allow a more detailed understanding of why certain taxa are common on certain sites.
Urban green spaces comprise a wide range of different forms ranging from large remnants of natural habitats and biotopes, parks, and urban productive sites (forests, gardens, farms) to small private gardens, street trees and flower beds. Remnants of natural habitat and large historic green spaces such as the ‘Castle Park Nymphenburg’ in Munich41 or the Central Park in New York42 are known to be hotspots of biodiversity and are often targets for biodiversity conservation43,44. In contrast, in the built-up area of the city, i.e. the area that is developed primarily for the needs of humans, biodiversity conservation is not a main concern, as it competes with other priorities of city planning, such as economic growth, transport infrastructure and housing development45. In the built-up area, however, backyards, gardens, neighbourhood common areas or vacant lots have also been shown to be important habitats for a variety of species46–48. Yet, small public green spaces in the built-up area are less intensively studied, so their contribution to urban biodiversity and which of their features affect different taxa remains to be investigated.
Here we investigate how different features of urban squares affect the diversity of different taxa living in the squares. We focus on squares as a core element of cities that serve as significant public spaces where city dwellers have gathered since ancient times49. We define urban squares as open spaces in a city without housing that are publicly accessible50. We selected 103 squares (Extended Data Fig.1) in Munich, Germany, out of a candidate list of 814 open spaces. We used stratified random sampling while accounting for square size, the distribution across the distance to the city center51, and the presence of trees (methods). We then characterised squares by a number of different features. The greenness of the squares was calculated as the proportion of a square’s area that was vegetated (Normalized Difference Vegetation Index – NDVI values ≥ 0.2). Square size, position in the city, proportion of sealed areas, number of streets boarding the square, the presence of water and night-time illumination (Artificial Light at Night – ALAN, methods) were assessed from different digital sources. Using on-site visits, we measured the planted vegetation on each square as flower beds, lawns (grass-dominated low vegetation), shrubs and trees. For trees, the number of tree species, the number of trees and individual tree size (DBH) were registered. Old trees (DBH ≥ 60 cm) were counted separately. For shrubs, we estimated volume, and for lawns and flower beds, we estimated the area on the square. As additional covariates for flower-visiting insects, we quantified species richness and abundance of flowers around sampling points. Human usage of squares and the presence of domestic animals was quantified by counting humans on several days and times of the days using photographs. Because connectivity and availability of habitat in the surrounding area have been shown to be important for biodiversity in urban environments38, we also calculated the percentage of green area in a buffer of 1000 m around the border of each square. For many features, we calculated both the total area and the proportion of the square they occupied. We calculated a correlation matrix for all square features and used the variance inflation factor (Extended Data Fig. 2) to select a final set of 19 independent square features for analyses (Table 1, methods). For example, absolute measures of vegetation (areas and abundances) were strongly correlated to one another and to square size. In a second step, we quantified the abundance and richness of birds (separately considering the feral pigeon), pollinators, and other arthropods, the richness and activity of bats, the activity of small mammals, pest small mammal species (mice and rats), and the richness of mosses (including liverworts) and spontaneous herbaceous vegetation on each square, using taxon-specific sampling methods (methods). As a measure of overall diversity, we calculated a multidiversity index, summing the scaled richness of all taxonomic groups (methods).
Table 1. | Summary of features and biodiversity of 103 public squares in Munich. Square features were assessed using maps and other digital resources and on-site visits. The biodiversity of different taxa was assessed using taxa-specific methods. Name, unit, and summary statistics for each variable.
Square features
|
Variable
|
Mean±sd
|
Min
|
Max
|
Surrounding greenness
|
0.44±0.13
|
0.174
|
0.757
|
Square size (ha)
|
0.92±0.98
|
0.09
|
6.71
|
Distance from centre (km)
|
4.26±2.43
|
0
|
10.88
|
Proportion of lawn
|
0.21±0.16
|
0
|
0.66
|
Shrub volume (m3)
|
0.17±0.28
|
0
|
1.46
|
Area of flowers (m2)
|
18.04±67.61
|
0
|
460
|
Tree richness
|
6.92±5.96
|
0
|
25
|
Tree density
|
0.64±0.47
|
0
|
2.52
|
Median DBH (cm)
|
27.92±13.44
|
0
|
76
|
Variability of DBH
|
0.42±0.22
|
0
|
1.07
|
Old tree abundance
|
4.23±9.72
|
0
|
65
|
Old tree proportion
|
0.07±0.13
|
0
|
1
|
Number of people
|
18.69±32.97
|
0
|
262
|
Number of streets
|
4.16±2.09
|
0
|
11
|
ALAN (grey value)
|
34825.19±25701.5
|
5912
|
178530
|
Water (binary)
|
NA
|
0
|
1
|
Flower abundance
|
190.59±326.97
|
0
|
2831.333
|
Flower richness
|
2.5±2.03
|
0
|
10.333
|
Number of pets
|
1.1±0.84
|
0
|
4.286
|
Biodiversity
|
Variable
|
Mean±sd
|
Min
|
Max
|
Multidiversity
|
0.59±0.13
|
0.23
|
0.85
|
Arthropod abundance
|
359.53±491.09
|
1
|
2427
|
Arthropod richness
|
11.23±3.79
|
1
|
17
|
Bat activity
|
157.09±128.26
|
4
|
545
|
Bat richness
|
5.91±1.5
|
3
|
9
|
Pigeon abundance
|
11.92±28.45
|
0
|
143
|
Bird abundance
|
71.67±53.16
|
0
|
292
|
Bird richness
|
10.23±4.61
|
0
|
21
|
Moss richness
|
19.58±7.85
|
1
|
38
|
Pollinator abundance
|
27.51±16.94
|
0
|
78
|
Pollinator richness
|
4.98±1.54
|
0
|
8
|
Pest mammal activity
|
0.12±0.2
|
0
|
0.84
|
Small mammal activity
|
0.14±0.18
|
0
|
0.81
|
Vegetation richness
|
47.25±15.74
|
6
|
82
|
Squares differed greatly in biodiversity, and the taxon richness of the most diverse square was almost eight times the richness of the biodiversity-poorest square (Table 1, Fig. 1a). Some species occurred on all or almost all squares, such as the plants Plantago major and Taraxacum officinale (all squares), the silver moss Bryum argenteum (93 squares) and the bird species carrion crow Corvus corone (92 squares) and great tit Parus major (94 squares). As a first test for drivers of the biodiversity of squares, we fitted linear models for the effect of greenness. Greenness on the squares positively affected both diversity and abundance/activity of all taxa, except for the abundance of feral pigeons, bat activity and richness, and pest mammal activity (Fig. 1c). For taxa where both abundance (or activity) and richness were measured, the relationship to richness was stronger (Extended Data Table 1). Multidiversity was most strongly affected by greenness (F1,101=79.01; p<0.001, Fig. 1b). These results emphasise the important role of planted vegetation for urban diversity, despite these green infrastructures being created mostly with human use in mind. Thus, the greening of cities benefits not only humans through better provisioning of ecosystem services19, but also other organisms, which in turn will enhance human-nature interactions20,52. Both the scatter around the regression lines and the differences in slopes between greenness and abundance/diversity of individual taxa suggest, however, that greenness is a coarse measure of the suitability of squares for organisms and that individual square features may offer better explanations where certain taxa occur.
To test the effects of individual square features on urban biodiversity, we fitted separate random forest models for multidiversity and each taxon’s richness and abundance/activity (methods). Random forest models, which excluded overall greenness, explained, on average, almost twice as much variance in the response variables compared to models with greenness only (greenness models r2=0.239 Extended Data Table 1, random forest models r2=0.433, Extended Data Table 2), confirming the hypothesis that the individual components of urban green explain biodiversity better than overall greenness. Our analysis found that each taxon was affected by several square features (Fig. 2). To rank variables by the strength of their effects on biodiversity, we calculated importance scores (percent increase in Mean Square Error (MSE), methods) for each square feature. The proportion of green in the 1000 m buffer around the squares positively affected only some of the taxa, including bat richness. However, for most taxa, local square features were more important for their abundance and diversity. The proportion of lawn, shrub volume and tree density affected most biodiversity variables (Fig. 2), and these three variables were also strongly correlated with a square’s overall greenness.
The proportion of lawn on a square affected most taxa positively and was the most important explanatory variable for multidiversity, moss richness, arthropod abundance, arthropod richness, and bird richness (Fig. 2). Lawns in the city are generally highly managed and therefore only occasionally allow taller plants to flower or set seeds; hence they do not allow insect species that feed on shoots, flowers and flowerheads to complete their life cycles. However, they do provide habitat for soil-dwelling organisms and species living close to the surface, and hence also provide food for arthropod-feeding guilds, including birds, predatory arthropods, mammals such as the hedgehog53 and some bats, e.g. Pipistrellus kuhlii and P. pipistrellus 54, both common in our study. We found lawns to have the highest abundance and second-highest richness of arthropods and the highest richness of spontaneous vegetation.
Several tree-related variables positively affected the biodiversity of the squares, in particular, tree richness and tree density (Fig. 2). Trees offer the highest vegetation biomass in cities, and tree species differ in the resources they provide to different organisms. For example, large native trees can provide breeding habitat for birds and are associated with higher activity of bats55 and some tree species, such as oak, have been shown to be highly important in providing resources for birds, regardless of being native or non-native56. In our study, there were up to 25 species of trees on a square, whereby Norway maple (Acer platanoides; n=1179) and Small-leaved Linden (Tilia cordata; n=1044) were the most abundant and occurred on 62 and 58 squares, respectively. Interestingly, the proportion of old trees was not an important variable, despite old trees being disproportionally important for biodiversity worldwide57. There are a couple of potential reasons for this. First, our definition of old trees (dbh > 60cm) includes trees that are <100 years old, i.e. they are still relatively young; among all squares, there were only 17 trees with a dbh>100 cm. Second, the management of old trees in highly accessible urban areas dictates the premature removal of the tree or management of deadwood for public safety58, and this deadwood is one of the most important features of old trees for many species59.
Shrubs were the third most important feature on a square predictive for overall greenness, and shrubs also affected biodiversity. Strong effects were found for the abundance and richness of birds and small mammals. Like trees, shrubs offer biomass, structure, resources and habitats in a city55. These can result from the shrubs themselves (e.g., nesting sites in the shrub, flowers, and berries) or from shielding areas inside or below the shrubs from access or management (e.g. for spontaneous herbaceous vegetation). While not demonstrated in our results, shrubs, in particular native ones60, have been shown to be highly important for native insects61, an important food source for birds and small mammals.
Features not related to the overall greenness of a square also affected several taxa. Square size, i.e. total area, influenced the richness of spontaneous vegetation and mosses and the density of birds and small mammals, supporting the importance of urban patch size as found in previous studies38. Larger areas generally provide more and more diverse local habitats (i.e. the species-area hypothesis62) and this may also hold for squares. Distance from the city centre negatively affected pollinator abundance, pest mammal and bat activity, and positively moss, small mammals and bat richness. Mosses, for example, are sensitive to the effects of urbanization, such as the heat island effect63 and air pollution64,65, both of which increase closer to city centres. While other taxa, such as pest mammals, are strongly associated with human activity and the built-up areas of cities66,67. Artificial light at night had negative effects on small mammal activity, bat activity and richness (i.e. the nocturnal taxa) but not on other taxa. While under certain circumstances, some bat and small mammal species may be able to exploit opportunities created by ALAN68,69, ALAN is agreed to negatively impact biodiversity70. The number of humans visiting a square negatively affected a number of taxa and affected only feral pigeon abundance and pest mammal activity positively. The presence of humans has been shown to have negative effects, even on highly urbanized species71,72, while synanthropic species, such as the pest mammals in our study, rely almost entirely on the resources created by humans. Water sources were rare on the squares (n=18 squares), and the presence of water was not selected as important for any of the taxa. While all species need water, there are likely to be other local sources of water than the fountains or other small sources observed on the squares.
Some taxa, in particular birds and small mammals, were affected by a larger number of square features, while others, such as the richness of spontaneous vegetation and arthropods, only by a limited number of variables. While several taxa were strongly affected by the greenness features (shrubs, trees, and lawns), birds were most influenced by several other green features. Several studies (e.g. 55,73) demonstrate how having a variety of vegetation structures can support a greater diversity of urban birds. In contrast, small mammals were the least affected by the greenness variables of a square but were negatively influenced by the number of people on squares and ALAN, both indicative of disturbances. Similarly, bats showed a limited, negative effect of the greenness variables with the strongest influence from disturbance variables. This suggests that for some taxa, either local green is not a factor, or some additional feature is more important, for example, connectivity (demonstrated to be important for common urban bats (e.g. Pipistrellus pipistrellus)74). We demonstrate that important features differ markedly between taxa. This indicates that there is not one set of features which is optimal for urban biodiversity. Different squares with different features, heterogeneity, and diversity in square features will sustain higher biodiversity at local and city scale.
While our study shows the potential of designing for both humans and biodiversity, there are also trade-offs and limits resulting from the different associations between features and taxa. For example, only some species benefit from increased tree density, and if high tree density implies a low proportion of open grassy areas, other species will be less likely to occur (Fig. 2). Further, lighting (ALAN) required to make people feel comfortable and safe during the night will negatively impact light sensitive species such as small mammals and bats. There are also limits to the simultaneous use of squares by both wildlife and humans, as indicated by the negative effect of human abundance on the abundance and richness of several taxa.
Our findings add to the growing body of evidence that urban green benefits biodiversity in general. Thus, the replacement of sealed surfaces with grass or other vegetation by urban planners will result in immediate biodiversity benefits75. Extending beyond the general “green in cities begets biodiversity”, we showed that several different square features are important for the biodiversity of taxa ranging from mosses and higher plants to arthropods, birds and small mammals. This is important, as in the discussion of creating a green infrastructure for cities, ‘green’ is often not qualified, suggesting that it does not matter what is grown. Although we often used coarse categories (trees, shrubs, lawns and flower beds), our study found striking differences in the effect of these different components on the biodiversity of the square. It is likely that describing the urban form in more detail, in particular the planted vegetation will unravel further relationships between planted vegetation and the associated biodiversity. Embracing these different greens in urban planning and design by creating diverse and heterogeneous sites will benefit various taxa, thus contributing to maintaining and increasing citywide biodiversity. As our study was restricted to Munich, we recommend that comparable studies are carried out in other regions to determine how local biodiversity responds to different components of urban green. Further, we only consider the current state of the square. As cities move to be more resilient and energy efficient, there is the danger that renovation of structures results in the loss of breeding and roosting sites. On the other hand, the modernisation of cities also offers the chance for a multifunctional and multispecies design, accounting for the needs of local wildlife. Beyond improving biodiversity, urban squares with varying features could also provide essential ecosystem services that may aid in mitigating the effects of climate change, such as reducing floods and providing local cooling76.
Our study emphasises that in the city, plants and wildlife not only occur in remnants of natural habitats or large parks but also in the built-up area of the city. Our analysis showed that the way humans design a public urban square, a structure made for human use and central to urban life, strongly influences the biodiversity cohabiting these squares. The square features included in our study were planned and implemented by humans for humans; all other organisms, including spontaneous vegetation, assembled on the square as a consequence of this design. The strong effects we found of this design on the biodiversity of urban squares emphasise that urban planners and landscape architects have an important role to play in creating biodiverse cities where human-nature interactions are possible.