The role of green roofs as urban habitats for biodiversity modulated by their design: a review

In view of the demographic revolution and the rapid development of urban environments, the installation of green roofs could be a tool to ensure human well-being (e.g. heat island reduction, rainwater management), or to increase urban biodiversity. However, the relationships between biodiversity and green roofs are not yet clear and little research has looked into this. We therefore reviewed studies on the overall biodiversity of green roofs. Our review has shown that there is a lack of knowledge of the biodiversity of green roofs, with recent consideration. We highlighted the importance of green roof contribution, in maintaining urban biodiversity through three lines of research: characterization, modes of use and design. Furthermore, we found that there were very few studies on soil biodiversity on this topic. We concluded that green roof construction guidelines should integrate soil communities into their design and aim to be heterogeneous at roof and landscape level. Future research should focus on the diversification and redundancy of rooftop conditions in the urban matrix. This would increase the area of green habitats and the success of species dispersal in cities.


Introduction
Urbanized areas have become the fastest growing environments in the world (Fischer and Lindenmayer 2007, Seto et al 2011, Lamb and DiLorenzo 2014, leading to many deleterious effects on climate, natural resources and biodiversity (e.g. Pickett et al 2001, Grimm et al 2008, McDonald et al 2008. For example, the urban biological communities of most taxa (e.g. plants, birds) are often less diverse (McKinney 2002(McKinney , 2008 and show greater biotic homogenization compared to other environments (e.g. forest, grassland, agriculture) Lockwood 1999, Lockwood et al 2000). As the urban human population continues to increase, it thus becomes progressively more necessary to include urbanized areas in biological conservation efforts (Marzluff and Rodewald 2008).
The conservation of biodiversity in cities is a concept aimed at making urban ecosystems more resilient to disturbances due to the major role of biodiversity in the provision of ecosystem services (e.g. Isbell et al 2011, Cardinale et al 2012, Pickett et al 2013. In cities, biodiversity can settle in green spaces, in reduced, fragmented and isolated surfaces (Fuller and Gaston 2009). To provide additional vegetated surfaces, buildings appear as a solution to 're-green the grey' (Jim 2004, Francis andLorimer 2011). The installation of green roofs could be a tool for increasing urban biodiversity (Oberndorfer et al 2007).
Green roofs (such as living roof, eco-roof or garden roofs) are either productive or nonproductive, built on flat or sloping surfaces. They consist of an anti-root treated sealing membrane, a drainage layer, a filtering layer, a culture and vegetation medium (Getter andRowe 2006, Oberndorfer et al 2007). According to Adivet (Association des toitures et façades végétales), they are characterized by different types of vegetation depending on the thickness of the growth medium. A distinction is thus made between intensive (>30 cm deep), semiintensive (12-30 cm) and extensive (4-12 cm) roof types (RP TTV). It should be noted that the distinction between these types of green roofs can also vary according to the types of plants, the frequency of maintenance and the cost of installation (US 2011, Hossain et al 2019, Abass et al 2020.
Several recent studies have looked at the biodiversity of green roofs by considering a multitude of organisms ranging from microorganisms to vertebrates and including arthropods (Brenneisen 2006, Colla et al 2009, Braaker et al 2014. Their role as habitat, their integration into the landscape matrix or even their design are all topics covered (Mayrand andClergeau 2018, Partridge et al 2020). However, the relationships between biodiversity and green roofs are not clear yet. If green roofs differ from other urban green spaces located at ground level due to (i) less accessibility for low-dispersal species, (ii) smaller surface area and (iii) specific soil structure, previously manufactured. While some studies have shown that arthropod richness and abundance tends to be lower on green roofs than on ground-level habitats (Tonietto et al 2011, Ksiazek et al 2012, Braaker et al 2017, there is no consensus on these conclusions since Kadas (2006) found an equivalent or an even higher number of species on green roofs compared to those on ground sites. The relationship between green roofs and biodiversity therefore deserves to be systematically examined, as was done within other urban green spaces whose biodiversity was examined (Nielsen et al 2014, Clucas et al 2018, Joimel et al 2022. The objective of this study is therefore to carry out a systematic review of the scientific literature on the whole of the biodiversity of green roofs to (i) elucidate the role of green roofs in supporting biodiversity in the city by focusing on quantitative data, (ii) and explore the factors influencing green roof biodiversity. For this, we address the following questions which will be treated separately in the corpus of the text: (i) do green roofs play a role of habitat for all biodiversity? (ii) What are the differences between green roofs and other urban or rural uses? Here we consider urban uses to be those located within the city limits (e.g. gardens, parks) and rural uses to be natural (e.g. forest) or pseudo-natural spaces (e.g. agricultural), that can be influenced by human activities. (iii) What is the influence of roof design on biodiversity?
This study will also make it possible to identify future avenues of research to be carried out on green roof biodiversity.

Research criteria
The research on green roof biodiversity was conducted on 15 November 2021 in Web of Sciences (WoS) using all the databases suggested. Different keywords were chosen from a sample of 34 articles dealing with both biodiversity and green roofs. The keywords selected were as follows: (« green roof * » or rooftop * or greenroof * ) and (biodiversity * or plant * or fauna or wildlife). In total, 2392 articles were retrieved as search results. A first selection was made based on the titles and abstracts, or even the whole article, in order to identify the articles that really fall within the topic i.e. the study of biodiversity on green roofs. In addition, we considered ten articles which were not included in WoS, they were found using Google Scholar on the same subject. In total, 154 articles were found to be relevant for our study.

Data collection
For each of the 154 articles selected, the following data were extracted based on titles and abstracts: • Title of the result • Authors, year, and journal of publication • Type of article (research, journal, book, book chapter) • Geographical location of the study, classified by country and continent • Species/Taxa examined, status of taxa (fauna, flora) • Compartment of species examined (above ground, below ground) • Quantified biodiversity (yes, no, cannot be determined) • Main conclusion of the study.

Bibliometric analysis
Our bibliometric analysis revealed an increase in the consideration of urban biodiversity in scientific studies from 1995 (figure 1). This is all the more true for the biodiversity of green roofs, since 99% of the publications on this subject only appeared after 1995, with a notable increase in the 2010s. Representing less than 1% of the publications on urban biodiversity, studies on roofs developed especially after 2013. However, the share of studies on green roof biodiversity remains rare and represents less than 2.5% of the total number of publications on urban biodiversity, illustrating a relative lack of knowledge on this subject. This fits in with the evolutionary history of green roofs. Since it was only during the 20th century that the creators of modern architecture (Le Corbusier, Alvar Aalto and Frank Lloyd) began to implement green roofs in their design to integrate the natural within the construction (Abass et al 2020). The development in Germany, some thirty years ago, of very light substrates (or growing media), mixing mineral matter and organic matter (Jim 2017), allowed green roofs to develop significantly. This technology then spread to Europe, especially in Germanic and Scandinavian countries, then to North America and to some Asian countries.  Focusing on the 154 articles selected for our study, we noted 137 studies in simple search, 15 reviews and 2 meta-analyses. From a geographical point of view, the majority of these studies were carried out in the countries of the northern hemisphere (figure 2) with about 36% in North America and Europe, followed by 23% in Asia. Only four studies described roofs in different countries; all being located in Europe (Bubnova et al 2012, Van Mechelen et al 2015, Nicolaisen et al 2017, Lönnqvist et al 2021. Moreover, no study on this subject was referenced in Africa. This is quite common, since according to previous reviews (e.g. Guilland et al 2018, Joimel et al 2022, there are few, if any, studies concerning urban biodiversity on this continent either for lack of laboratories or of financing. Furthermore, with its moderately developed urban environments compared to those of other continents (Europe, North America, Asia) (Appolloni et al 2021), it is not surprising to see the lack of studies on green roofs in Africa.

Research axes on the biodiversity of green roofs
The 154 articles were classified according to three main categories: characterization, modes of use and design. Examination of these various publications on green roof biodiversity revealed that 17.5% of scientific articles focused on characterization, 17% on modes of use and 52.5% on the various aspects linked to design. Some items found themselves in two categories simultaneously, i.e. 4.5% for use/characterization as well as for use/design and approximately 4% for characterization/design (figure 3).

Characterization of the biodiversity by green roofs: biodiversity support
Many studies (n = 27) have characterized the biodiversity of green roofs, focusing for more than half on extensive roofs (59%). The other studies did not consider it useful to provide information on the type of green roof (41%).
In general, studies mainly use taxonomic diversity and abundance as metrics to quantify biodiversity. For example, in their study, Partridge et al (2020) collected over 15 000 arthropod individuals from 16 different taxa and one dominant bat species (Lasirus borealis), all living on a green roof in New York City. Other authors studying bees visiting a green roof in downtown Toronto, showed a differentiation in the number of individuals according to the taxonomic groups (e.g. Lasioglossum sp., Apis sp., Bombus sp.) identified (MacIvor et al 2015). Only a few studies (n = 3) focused on functional diversity and these also investigated taxonomic diversity. Finally, two studies assessed the survival or presence/absence index (Dvorak and Volder 2013, Marttinen et al 2020). Furthermore, it is important to note that 37% of the studies (n = 10) do not use an index (table 1).
Regarding the biodiversity studied on green roofs (table 2), it is mainly the aerial compartment (81%), and more specifically plants (55%) (e.g. Bubnova These different studies have made it possible to assess the capacity of green roofs to serve as potential habitats for some organisms (e.g. plants, bees, birds, arthropods) of biodiversity. In order to be a viable habitat, green roofs must correspond to a place or location where an organism lives, with sufficient resources allowing it to feed and reproduce. Davis (1960) exploring the formal distinctions between the concepts of habitat and environment also concluded that the emphasis is on the location as well as on the conditions. The notion of resources, defined as 'various forms of energy and matter potentially or actively useful to organisms' is central to, but often neglected or omitted from, the concept of habitat (Davis 1960). These resources must be available to the organism beyond its minimum needs (Davis 1960). The habitat therefore constitutes an ecological entity including species and communities, as well as their biotic and abiotic environment. Across the 27 articles, it appears that roofs form a viable habitat for many species such as plants (e.g. sedum), arthropods (e.g. moths, springtails), or vertebrates (e.g. bats). It is specifically by providing a viable habitat for arthropods or for plants that other organisms such as bats can find food resources in the context of urbanized environments (Pearce andWalters 2012, Partridge et al 2020). It sometimes seems that some species, especially bees, favor vegetation (Hofmann and Renner 2018). Conversely, when these plant and invertebrate communities are depleted, the role of green roofs for other species is no longer ensured (Rumble and Gange 2013). Several authors have concluded that promoting soil communities in green roofs would create sustainable habitats and maximize gains in urban biodiversity (e.g. Rumble andGange 2013, Williams et al 2014).
If green roofs seem to be a viable habitat, the majority of species present on green roofs are often generalist species (e.g. mosses, insects), which are able to withstand difficult environmental conditions (e.g. sun, wind).

Modes of use
Studies comparing modes of use mainly focused on the differences between: (i) types of roofs (green vs. bare, n = 8), (ii) types of green roofs (extensive, productive, intensive, and various plant structures, n = 5), (iii) green roofs vs. urban habitats (at ground level, n = 23) or natural habitats (retention basin, n = 1) or pseudo-natural habitats (agricultural environment, n = 1).

Comparison between 'types of roofs'
The studies (n = 8) that compared the effects of green roofs on biodiversity came to different conclusions, including for the same taxonomic group (e.g. birds, bats) (table 3). (i) The majority (62.5%) of the studies demonstrated higher biodiversity on green roofs than on bare roofs for a diversity of taxonomic groups (e.g. arthropods, gastropods, birds). For example, Wooster et al (2022) found that green roofs supported four times more birds, over seven times more arthropods, and twice as many gastropods than non-green roofs. The same trends were shown for bat activity (Pearce and Walters 2012). (ii) Twenty-five percent of the studies did not however observe any differences between a green roof and a bare roof (table 3). This is the case, for example, of the similar avian use of green roofs to that of bare roofs in Chicago (Washburn et al 2016). (iii) Finally, only 12.5% of the studies demonstrated a higher biodiversity on non-green roofs (see Wong and Jim 2016).

Comparison between 'types of green roofs'
Comparing the different typologies of green roofs, the majority of the studies found that the biodiversity was lower on roofs with extensive vegetation (plants in Droz et al (2021); arthropods in Madre et al (2013)) (table 4).
In general, a high diversity of plant species on roofs increases the biological diversity of other species (Kyrö et al 2020). This observation is also demonstrated in the literature in natural areas (e.g. Perez et al 2013, Henneron et al 2016, Salmon 2018. Madre et al (2013) confirmed this for green roofs by demonstrating a significantly higher richness and abundance of most arthropod taxa on roofs with a more complex vegetation. However, on extensive roofs, often highlighted for their lower biodiversity, the vegetation is generally composed solely of succulent plant species (e.g. moss and sedum). Therefore, plant communities are more diverse on intensive roofs than on extensive roofs (Droz et al 2021). Table 3. Representation of the studies comparing the biodiversity between green roofs and other habitats (non-green roofs, urban green spaces on the ground, natural green spaces, agricultural environments). The signs '+' , '−' or ' = ' which precede the letters mean that the biodiversity (taxa) of green roofs is 'higher' , 'lower' or 'similar' compared to that of other habitats. The letters are associated with the corresponding studies. Multi-taxa = invertebrates, birds, bats, plants.  These variations in the structure of biodiversity within green roofs are also observed at the level of species composition, including in studies that show no effect on the richness or abundance between the two types of roofs (Joimel et al 2018). Extensive roofs thus tend to host generalist species, tolerant to xerothermophilic conditions, which are favored by these types of green roofs (Madre et al 2013, Joimel et al 2018. Conversely, on intensive roofs, the species are rather specialists with requirements in relation to the microclimatic conditions which must be favorable to their development. Besides, in comparison to extensive roofs, the plant diversity on intensive roofs can be more sensitive to local environment variables (soil texture, size and fertilizer use) (Droz et al 2021). We also notice there can be differences in colonisation depending on the different roof types due to design/uses. Joimel et al (2018) showed that a way for colonising these urban habitats for collembola could be wind dispersion.
A new type of classification of green roofs could help to: (i) bridge the gap between practitioner knowledge and ecological knowledge and (ii) predict patterns which favor biodiversity installation focusing on the surrounding landscape green roofs built rather than on the thickness of their substrates.

Comparison 'green roofs vs. urban habitats and/or natural or pseudo-natural habitats'
Publications on the comparison between green roofs and other urban habitats show that: (i) 75% of the studies showed a lower biodiversity on green roofs than that of urban habitats (at ground level) (table 3). For example, recently, Domínguez et al (2020) showed that species richness, abundance, and arthropod predation rates were significantly higher in ground-level habitats than on green roofs. In addition, plant-pollinator relationships may be satisfactory on green roofs, even though bee numbers and diversity are lower on green roofs compared to ground habitats (Ksiazek et al 2012). (ii) Only 25% of studies demonstrated a higher biodiversity on green roofs. Thus, it was shown that the root zones of green roofs can have an abundant soil microbial community which, in some cases, may be more diverse and numerous than the communities found in brownfield areas (Molineux et al 2015). Also in their review, Hofmann and Renner (2018) showed that the percentage of cavity-nesting bees on rooftops was higher than in the surrounding soils.
Indeed, the conditions on green roofs are very distinct from those of ground habitats. . The absence of direct solar radiation on the ground, less trampling, and the addition of litter or compost lead to more organic matter and higher moisture content in the soil of the parks than on the roofs (Sarah et al 2015).
Regarding the comparison with natural and pseudo-natural (agricultural) environments, the two rare studies on the subject have different conclusions. If Filazzola et al (2019) demonstrated a green roof biodiversity comparable to that of natural habitats; Nicolaisen et al (2017) demonstrated a higher biodiversity on green roofs than on agricultural habitats. Indeed, rooftop fungal communities were more diverse than in rapeseed fields, likely reflecting greater mixing of air from a range of microenvironments for rooftop sites.
Green roofs, like other urban green spaces, can have ecological significance in cities by attracting and supporting a greater taxonomic diversity which can then add important functional capacities to previously impoverished spaces Clark 2015, Wooster et al 2022).

Design
The effects that the design has on green roof were widely studied in extensive roofs (90% of publications). Biodiversity can be influenced by different factors relating to the design of green roofs: a mix of plant species, plant cover, substrates, irrigation, age (or aging of the roofs), surface area, height, maintenance and surrounding landscape (table 5). Some studies have simultaneously crossed some of these factors. Almost all of the studies describe plants (table 5).

Effects of plant species
The vegetation, whether in terms of diversity, as we have already discussed, percentage cover or flowering, will affect other species. First, an increase in the area of vegetation cover seems to favorably affect other species, such as arthropods (Salman and Blaustein 2018). Several studies (n = 9) found a positive effect on the biodiversity when there was a mix of plant diversity (e.g. fungi, arthropods, plants) (table 5). Hoch et al (2019) thus demonstrated that the relative abundance of mycorrhizal fungi was higher on roofs with a mixed vegetation and observed higher pathogen loads on roofs with sedum. Some effects may be linked to flowering times (Benvenuti 2014).
Overall, incorporating functional diversity, particularly varied growth forms, increases the diversity of green roofs, potentially improving the resilience and performance of green roof systems over the long term (Heim and Lundholm 2014, van Mechelen et al 2015).

Effects of substrates
Publications on substrates have highlighted their effects on biodiversity depending on their (i) composition and/or their (ii) thickness (table 5).
Generally, the soils which develop on green roofs are referred to as Technosols (e.g. World Reference Base for Soil Resources-WRBSR, Group I.W 2006, Lehmann 2006. They are composed of various technogenic materials, i.e. artifacts, generally marked in their composition by human activity (e.g. urban waste, building materials, industrial by-products) (Hiller 2000, El Khalil et al 2008, Grard et al 2020. Regarding the composition of substrates, several authors have highlighted the major role of organic matter, which has positive effects on biodiversity (Nagase and Dunnett 2011, Chen et al 2018, 2021. It is not only about increasing the substrate OM content but also about reaching the optimal level. Some authors mention adding values of 10% OM (Nagase and Dunnett 2011) especially thanks to the application of biochar (Chen et al 2018) in order to obtain an optimal growth of cultivated plant species and a significant increase in microbial biomass.
Structure also plays a role. Coarse particle substrates induce higher mortality rates in most introduced plant species, as well as lower vegetation cover compared to treatments with fine particles (Toland et al 2014).
As for the effects of substrate depth on biodiversity, Molineux et al (2015) suggested that increasing substrate depth improved plant establishment, but this effect was not consistent across substrates.
Once established, technosols will undergo much faster pedogenesis than natural soils. They are above Table 5. Representation of the studies on the effects of design parameters of green roofs influencing biodiversity. The signs '+' , '−' or '0' mean the 'positive' , 'negative' or 'neutral' effect of a design parameter on the biodiversity (taxa) considered.   (Scalenghe andFerraris 2009, Scholtus et al 2009). This evolution will induce a strong evolution of the profiles over 3 years, with rapid changes in the number and characteristics of the horizons (Séré et al 2010). Chemical (decarbonation) and physical (aggregation) weathering occur through processes similar to those that occur in natural soils. We can expect that these rapid modifications in the structure and composition of the constructed technosols will affect the local pedoclimatic conditions, which depend on the future of biodiversity in these environments.

Effects of irrigation
Roofs, with their impermeable surfaces, are simple and unobtrusive hydrological compartments in which water inflows and outflows can be monitored precisely. Ideally, precision water management would use irrigation to maintain soils at higher average moisture levels for satisfactory vegetation development. During storms, this higher antecedent moisture causes irrigated soils to shed more water than they would otherwise (Harada et al 2018). For example, regarding the effects of irrigation on green roof biodiversity, studies have shown that a low irrigation regime is enough to see satisfactory plant growth (Kokkinou et al 2016, Paraskevopoulou et al 2020. However, it is clear that these results were observed on roofs with extensive vegetation and that the plants that live there are adapted to xerothermophilic conditions (Madre et al 2014), a low irrigation regime is enough for their growth.

Effects of roof ageing
The effects of roof aging on the biodiversity of green roofs are difficult to measure because the study of roofs is an emerging research subject.
However, a study evaluating the effect of roof age on biodiversity indicated that Collembola density was slightly higher on old roofs (built between 1990 and 1994) compared to young roofs (built between 1998 and 1999) (Schrader and Böning 2006).
On the dynamics, studies are often confined to limited time stages: 1 year (summer 2002) for Schrader and Böning (2006), while Bubnova et al (2012) conducted their study over 2 years (2011 and 2012). They noticed a dynamic evolution in species composition and in projective vegetation cover surface, in the absence of maintenance. Thus, over time, plants of the sedum genus are oppressed and replaced by moss. These studies indicate that over time, the environment is probably more stable in the substrates due to the advanced formation of the soil, the improvement of Collembola niche occupation and to the establishment of undemanding plants. However, studies on the evolution of soils/technosols are rare and it is difficult to really conclude on the orientation of their pedogenesis. Grard et al (2020) thus showed a high fertility in technosols with nutrient stocks after five years of cultivation. Maintaining this fertility would potentially lead to positive effects on biodiversity support.

Combined effects of different layouts and practices
Studies on the modulation of several design features (e.g. plant species, surface area and roof age) to promote biodiversity show positive effects.
Fabian et al (2021) showed that plant species richness, roof canopy cover, substrate depth, roof area and age, and isolation in terms of height and canopy cover at the landscape scale, have beneficial effects on arthropod communities. More specifically, these authors indicate that: (i) the composition of arthropod community on green roofs was affected by their area and by the vegetation cover in the landscape. (ii) Green roof area was positively correlated to total species richness, species richness of most functional feeding groups, and total arthropod abundance. (iii) Green roofs with greater plant richness, mostly spontaneous, and less isolated (both vertically and horizontally) favored entomophagous arthropods. (iv) Other variables, such as age and ground cover of green roofs, were also important for herbivores, predators, parasitoids and detritivores.
In addition, other design features (e.g. substrates, irrigation, height of buildings) put forward by some studies to promote biodiversity showed divergent effects. For example, in a study by MacIvor et al (2013), ground cover and biomass of grasses and herbaceous plants were significantly higher in organic-based growing media, with no effect of additional irrigation. Surprisingly, other studies showed that plant growth was not dependent on water type or substrate type (Solodar et al 2018). On the other hand, the mixture of diversified plants could be more advantageous than a monoculture in terms of greater survivability and higher visual index in dry conditions (Nagase and Dunnett 2010). Thus, planting extensive green roofs with a mix of plant species can ensure the survival of certain species and maintain cover and biomass when additional irrigation is stopped to save water, or during extreme drought (MacIvor et al 2013).
Other design features involving the interaction of substrate depth and building height showed positive effects on the biodiversity of green roofs. Indeed, increasing the depth of the substrate and the amount of shade (coming from a taller building) can promote the richness and diversity of plant species (van der Kolk et al 2020). Shade probably works by reducing water stress, while, increasing substrate depth improves plant diversity due to the addition of native nutrients and seeds. These authors recommend taking these two aspects into account when designing green roofs.

Discussion/conclusion on the research axes
We reviewed the data available in the scientific literature on the biodiversity of green roofs. These data allowed us to show three main axes evaluating the biodiversity of green roofs: characterization, modes of use and design.
Characterization studies have provided information on the ability of green roofs to serve as habitat for some groups of organisms (e.g. plants, bees, birds, arthropods). However, to our knowledge there are not enough data on other organisms (e.g. soil invertebrates) to generalize the ecological role of green roofs in maintaining urban biodiversity.
The studies regarding modes of use have contributed to our ability to assess the categories of roofs that are more biodiversity friendly. Green roofs or types of green roofs with deeper substrates (e.g. intensive roofs) and those with more complex plant structures seem to be the most suitable.
The design studies allowed us to assess several factors (e.g. height, surface area depth of substrate) influencing biodiversity. Taking these factors into account when designing green roofs makes it possible to quantify its biodiversity. Overall, studies advocate installing green roofs more and more in cities to mitigate the negative effects of urbanization and preserve one of the many ecosystem services, such as biodiversity.

Biodiversity services
Currently, the interest for studies on biodiversity lies mainly because of its key role in the provision of a range of ecosystem services. These services range from aesthetic, cultural and recreational values to goods that have direct use value and enhance many other ecosystem services on which humans depend (Bulte et al 2005, Mertz et al 2007. Beyond the characterization of biodiversity, it is interesting to evaluate the services offered by biodiversity to green roofs. Only few studies have shown an interest in plant species as the main factor. For example, in a study conducted in Shanghai (China), the presence of Poa pratensis, Lolium perenne and Agrostis stolonifera on roofs led to a reduction in runoff (Li et al 2018). In general, incorporating annuals into Sedum-based green roofs reduces not only CO 2 concentrations (Klein et al 2017) but also heat island effects (Chow et al 2019). In addition, other studies highlighted the services provided by living invertebrates on green roofs (MacIvor and Ksiazek 2015). This is the case for example of (i) pollination for plant reproduction and yield of cultivated crops, (ii) pest control to reduce damage to green roof vegetation, (iii) decomposition to retain the organic matter and nutrient cycling in the substrate, and (iv) contribution to food webs of species such as birds which visit green roofs.

Research perspectives
We analyzed the scientific literature available on green roof biodiversity. Green roofs seem to serve as habitat for some organisms and this depended on the taxonomic groups but also on the design of the green roofs. Thus, it would be necessary to prioritize the actions to be carried out in favor of biodiversity thanks to green roofs. A point of attention must be made on the risk of biotic homogenization by favoring only a few specific species.
Today's cities, designed for the well-being of city dwellers, must adapt and take into account the needs of biodiversity. For this, the development of roof biourbanisms requires a better knowledge of both the characteristics of roofs influencing biodiversity and of their landscape integration to ensure ecological continuities.
Regarding the knowledge acquired on green roof biodiversity, we have identified several factors which influence biodiversity and which are poorly taken into account in the literature: (i) taking into account the surface. A large disparity in size is observed between the different roofs. This may explain the variability in the results obtained on the capacity of the roofs to serve as habitats. (ii) Integrating both aboveground and belowground biodiversity indicators. Soil biodiversity is often understudied even though it is highly sensitive to different types of soil. Plants are the most studied taxon. (iii) Considering time. Currently, there is very little data on green roof population dynamics. However, the soils of green roofs are subject to very rapid pedogenesis which can potentially influence their role in supporting biodiversity depending on the age of the roof. (iv) Studying new roofs which combine different functions. For example, the combination of green roofs with photovoltaic or solar (PV) panels is believed to provide synergistic benefits. The panel is cooled down by the presence of vegetation, and thus produces more electricity, while PV improves the conditions for vegetation growth and increases abiotic heterogeneity, resulting in greater plant diversity.
Concerning bio-urbanism, our research proposal focuses mainly on the integration of green roofs, at the landscape scale, in the ecological continuities of the city. The distance between rooftops and other urban or peri-urban green spaces is therefore a major research issue. In addition, in the case of green roofs, the height is another criterion to be taken into account. Indeed, the height of buildings strongly influences the diversity of species regardless of their degree of mobility ( Finally, the links between green walls and green roofs could act in concert for the movement of organisms in the city and the synergies between these two habitats remain unresolved.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).