Urban refugia sheltering biodiversity across world cities

Over the last years, world cities have emerged as multiple and complex habitats hosting numerous and varied urban refugia for biodiversity. Therefore, the published literature was systematically reviewed to investigate the increasingly acknowledged role of many urban habitats to function as refugia for species. Many plants and animals were reported to colonize a variety of urban environments: Urban forest/grassland (natural/seminatural), City park, Historical park (villa garden, botanical gardens, cemeteries, etc.), Green spaces (abandoned areas, railways, line of trees, etc.), Water body, and Built areas. Refugia were found in over a hundred world cities, especially in medium-to very big-size cities. The number of papers focusing on animals (n=66) were higher than those regarding plants and mushrooms (n=19 and n=1, respectively); however, the number of species recorded within refugia were the highest for plants. Plants exhibited the highest frequency (7 out of 19 papers) in Historical park, while animals in Urban forest and grassland (20 out of 66 papers). In most studies (25.9%), urban refugia were linked to Urban forest/grassland that is terrestrial natural and seminatural sites widespread within or around cities. The 22.3% of studies referred to generic Green spaces of several types interspersed within cities, both public and private. The 14.1% of refugia were found in Built areas (arti�cial). About the 33% of studies reported the presence of species worthy of conservation (rare, endemic, endangered, or protected) for a total of 365 species. The 20% of investigated papers reported the presence of alien species for a total of 879 species. The capability for urban areas to host a huge amount of biodiversity needs to be acknowledged by city planners so that management practices that maintain and support such diversity can be pursued. Recording and monitoring species along with their refugial habitats is fundamental to achieve this goal.


Introduction
The increasing trend of world human population is expected to continue in the future, especially in big urban areas such as metropolis and megalopolis (United Nations 2019).Besides economic growth and social bene ts, this expansion implies the conversion of natural, seminatural, and agricultural areas into urban systems that can seriously deteriorate biodiversity with impacts on ecosystem services, alterations of microclimate and water ows, as well as consumption and erosion of soil (Frank, et al. 2017).
However, the impact caused by urban growth on the environment is going beyond the cities and may drive further future changes at multiple scales (Grimm et al. 2008).In this framework, over the last decades urban landscapes have been seen as socio-ecological systems with a new paradigm when thinking about human-environmental interactions: the co-existence of the ecological and social systems within and across cities (Haase 2021).Consequently, urban areas have become a central node for the long-term functioning of societies and ecosystems with high spatial heterogeneity and different levels of connectivity with surrounding agriculture and/or natural areas (Capotorti et al. 2020).Species can move and persist in the urban matrix despite human activities.
Actually, urban areas host a great diversity of species belonging to different organisms (animal, plants, fungi, etc.), both native and non-native coming from the regions surrounding cities (Lepczyk et al. 2017) or present as relict of past environments (Nagy and Malatinszky 2019).Within urban areas, endemic and rare species and others of conservation interest are increasingly recorded within a variety of anthropogenic habitats (Ives et al. 2016;Toffolo et al. 2021).On the other hand, built environments are optimal habitat for a plethora of non-natives (Shaffer 2018).Therefore, the question whether urban areas can host species of conservation interest and whether this biodiversity can be incorporated into urban planning remained open for a long time (Miller and Hobbs 2002;Kantsa et al. 2013).
The recent COP15 and the Post2020 Global Biodiversity Framework pointed out the need to include cities in conservation strategies (Vlaanderen and Lange 2022; Convention on Biological Diversity-CBD 2022).
Particularly, "to ensure biodiversity-inclusive urban planning, enhancing native biodiversity, ecological connectivity and integrity, and improving human health and well-being and connection to nature" is needed (Convention on Biological Diversity 2022).
Following a conceptual change, cities are emerging across the world as multiple and complex habitats hosting unexpected and widespread refugia for biodiversity where insect pollinators or freshwater species can be found (Chester and Robson 2013;Hill 2021).In this regard, even considering the CBD's targets, urban refugia emerge as a top research priority also when they consists of small habitat remnants (Labadessa and Ancillotto 2023).
Traditionally, the terms refuge(s) or refugium (refugia) has been widely used in the context of climatic change, to describe climatically stable areas within which species survived to past warm or cold climatic stages (i.e.glacial-interglacial periods) under adverse (cold or warm) environmental conditions (Rull 2009;Gentili et al. 2015a).Recently, Monsarrat et al. (2019) have introduced the concept of Anthropocene refugia, referring to areas that provide spatial and temporal protection to species from human activities and that can remain suitable for a given taxonomic unit in the long-term.Speci cally, despite the concept of Anthropocene refugia considers the key role of anthropogenic pressures from which species nd realized or potential shelter, it does not ecologically de ne the characteristics of refuge habitats with stable and favourable environmental conditions for species' persistence, nor the matrix where unfavourable environmental conditions subsist.Therefore, Anthropocene refugia need to be further spatially and ecologically de ned.
Regarding urban areas, the rst likely mention of urban refugia dealt with the role of zoos in the conservation of wildlife within city as well as their role in education and research (Conway 1969) However, over the last decades an increasing number of scienti c works has differently referred to urban refugia (or refuges) (Hall et al. 2017;Hill 2021;Perez et al. 2021), that are locations within cities where biodiversity can nd shelter.For instance, Caccamise et al. (1996) found that well conserved grassland patches within the airport area of Atlantic city (USA) served as refugia for avian species in comparison to habitats outside the airport.In a literature review, Löki, et al. (2019) found that cemeteries and churchyards have a signi cant conservation role within urban areas, as even in strongly altered landscapes, since they frequently function as refugia for the populations of rare and endangered species of plants and animals.
Overall, urban refugia as particular anthropogenic refugia may function in a similar way in comparison to natural refugia, providing critical habitats and then offering new chances for species to live in a context characterized by disturbance and stress (Chester and Robson 2013).
Green areas are expected to provide the major contribution to urban biodiversity and refugia to species.However, cities host a plethora of unplanned and unexplored habitats (e.g.vacant residential properties and abandoned industrial sites) that potentially hold natural remnants embedded in an urban matrix and can support species diversity (Kwok 2018).Therefore, we aimed to perform this literature review to explore the main habitats that can offer protection to species (animal, plant, and other organisms) in urban environments.We wanted to explore main trends relating to urban refugia and discuss if and how they can be preserved and incorporated into conservation and urban planning along with the species they host.

Literature search and review
In 2022, we conducted a literature search on Scopus (temporal range: 2000-2022) using combinations of keywords including "refugia", "refugium", "refuges", "plant", "animal", "urban".After a rst screening, this process returned 220 titles.Only 149 articles dealt with the investigated subject and were retained for a deeper investigation.Finally, only 85 were considered suitable and processed for the review (Supplementary Table 1).Particularly, we only included the papers considering one or more urban habitat(s) as refugium (refugia), and then as persistence area in a general context of adverse conditions.The presence of the words "refugia", "refugium" and "refuge(s)" was a discriminant for the inclusion in the database.However, we did not include papers that were not clearly carried out in urban environments and/or did not shape a well-de ned association between species and refugium habitats in an adverse urban context.
When organizing the database, we noticed that some articles had a very coarse subdivision of urban habitats, while other articles reported the habitat types more in depth.In our analysis, we considered such aspect to detect possible trends.Therefore, we distinguished the following main habitat categories as generally mentioned in the assessed articles: Green spaces = when a not clear distinction of arti cial or natural/seminatural non built-up areas within city was indicated; Historical park = when refugia were indicated in historical green spaces such as botanical gardens, sacred spaces (close to churches or temples), and cemeteries; City park = when refugia were indicated in green spaces largely accessed by the public and clearly mentioned as "city park"; Urban forest and grassland = when refugia were indicated in natural or seminatural forests, clearings, and grassland within cities; Built area = when refugia were indicated in anthropogenic habitat such as industrial sites, road sides, railways, etc.; Water bodies = when refugia were indicated in water bodies such as ditches, lakes, mires, ponds, streams, and wetlands within cities; Several = when refugia were indicated in more than one type of the previous categories and other ones (e.g.water bodies, city park, forests, sites within airports, etc.).
Other possible urban habitats (allotment and similar gardens, roofs, etc.) not found or mentioned in the selected articles, were not considered since this review was not based on habitat diversity present in urban areas but on the declared recognition of a certain habitat as refugium.Some previous works made the distinction between refuges, existing at ecological temporal and spatial scales, and refugia, existing at evolutionary scales (Keppel et al. 2012).When referring to urban environments, we found that the two terms have been used almost indifferently across the considered literature.Therefore, we preferred to use the terms refugium/refugia that include both the ecological and evolutionary meaning.In fact, in urban environments, new evolutionary events via rapid adaptation of species or local extinction events can occur within short times (Diamond and Martin 2021; Kotze et al. 2022).For instance, Gorton et al. (2018) found rapid adaptive divergence of urban and rural populations of a ruderal species (Ambrosia artemisiifolia) and the greater heterogeneity in urban environments can in uences evolution of species in cites.

Urban refugia over the world
The 27.7% of the selected papers were published between 1994 and 2010, the remnant 73.3% were published between 2011 and 2022.Most of the studies considering urban refugia included medium size to very big-size cities considering the number of inhabitants (recovered from Wikipedia) all over the world, for a total of about 80 cities and some widespread urbanized areas (Fig. 1).In particular, ve papers considered megalopolis with more than 10 million of inhabitants (Hangzhou two times, Tokyo, Guangzhou,Chongqing); 39 papers included metropolis with 1 to 9.9 million of inhabitants (e.g.Berlin, Prague, Milan, Buenos Aires, Rio de Janeiro, New York, Tokyo, Cape Town, and Sidney); 22 works included medium cities with 100 to 999 thousand inhabitants (e.g.León, Bologna, Zurich, Canberra, and Denver); ten papers described refugia within small cities from 10 to 99 thousand inhabitants (e.g.Atlantic city, České Budějovice, and Matias Barbosa).Finally, nine papers, includes more than one urban areas from about 1 to hundreds of thousands of inhabitants.America (36 works: 24 in N-America and 12 in S-America) and Europe (32 works) were the most represented continents, while Africa accounted for only one paper (Fig. 1).USA was the county exhibiting the highest number of papers (18) followed by Brazil, Czech Republic and Germany (6 works each).

Organisms within urban refugia
With regard to life kingdoms, 65 works included animals (Animalia) belonging to several groups (28 insects, 7 other arthropods, 1 shes, 5 amphibians, 9 birds, 10 reptiles and 13 mammals; Fig. 2), 19 woks described plants (Plantae; of which 18 referred to higher plants, 1 to algae) and one work referred to mushrooms (Fungi); nally, another work described both plants and animals.Among animals, insects were the most reported group followed by mammals (Fig. 2).

Sampling effort, number of taxa and species characteristics
The number of samplings across studies ranged between 1 and 2900 (median = 27) while the number of taxa detected within refugia ranged between 1 and 663 (median = 19).Single species' studies were 25: 22 focused on animals, 2 on plants and 1 on mushroom.Overall, the studies on plants reported a higher number of taxa (median = 166) than those on animals (median = 10), since numerous studies on animal focused on single species (Kruskal-Wallis test: H-chi 2 = 18,24; p<0.001;Fig. 3).On the other hand, the number of samplings did not show signi cant differences between plant and animals.About 33% of the papers reported the presence of species worthy of conservation (rare, endemic, endangered or protected) for a total of 365 species; the 51.7% of these papers referred to animals, 44.8% to plants and 3.5% to The 20% of the investigated papers reported the presence of alien species for a total of 879 species; the 66,6% of papers referred to alien plants, 33.4% to alien animals.

Habitats acting as urban refugia
In most studies (25.9%), the urban refugia were linked to terrestrial natural and seminatural sites, such as forests, grasslands, and scrublands (Fig. 4).The 22.3% of studies referred to generic "Green spaces" of several types interspersed within cities, both public and private.City and Historical parks (including botanical garden and sacred spaces) were speci cally mentioned in 8.2% and 10.6%, respectively.The 14.1% of refugia were found in urban built arti cial areas (residential, roads, industrial, etc.), while 3.6% indicated the presence of shelter for species in both semi-natural and built areas.Finally, the 15.3% of the studies were performed in Water bodies both natural and arti cial (wetlands, mires, ponds, lakes, etc.).
Among unusual habitats falling in target categories of habitat, species found shelter within the following locations (Supplementary Table  The habitats with the highest number of species in a single article was found in City park (n=663).
Always concerning habitats considering the median difference (H-chi 2 = 12.77, p < 0.05), Built areas exhibited the lowest value in comparison to the other main habitats (signi cantly in comparison to Urban forest/grassland, Historical park and Several).
The frequency of refugia for animals and plant species was different across the main urban habitats (Chi 2 : 21.8; df=6; p < 0.01).Particularly, plants exhibited the highest frequency (7 out of 20 papers) in Historical park, while animal in Urban forest and grassland (20 out of 66 papers).On the other hand, no trend was found when linking the groups of organisms and the main urban habitats, with the exception of higher plants that were mainly found in Green spaces and Historical park.
Most of the main habitat types were reported to support species worthy of conservation (Fig. 5): City park, Green spaces, Historical parks, Natural forest and grassland, and Water body.On the other hand, almost the half of the alien species referred to Green spaces (47.1%).

Discussion and Conclusion
Urban refugia has been acknowledged by 85 works according to our literature review.Considering the habitat-scale (i.e.wide-ranging categories) we analysed, they can be de ned as habitats or sites within cities supporting the survival of one or more species in the general context of unfavourable environmental conditions of disturbance and stress present in urbanized matrixes.In other worlds, urban refugia consist of anthropogenic (mainly(, seminatural, and natural (sometimes) spaces with no to moderate disturbance and su cient resources for species to survive within the urban matrix.
An important aspect to consider in this de nition concerns alien species that are very common in cities (e.g. over 35% in Vojík et al. 2020) and, in some cases, are among listed taxa within refugia.In turns some woody plants within city parks or green spaces can function as refugia for other species, especially animals.Invasive species are no or scarcely considered in the papers considering urban refugia, probably due to peculiar conditions of cities that include harsh and stressful environments for animal and plant life.
This review enhances the knowledge about the biodiversity in urban spaces within cities of different size, from megalopolis and metropolis (most studies) to small cities.The literature search indicated the increase in articles regarding urban refugia over time.Particularly, it is highlighted how recent research on urban biodiversity has changed perspective on the biological value of species inhabiting cities and on the importance of habitats widespread within built environments and the built matrix itself.Many places within cities have the ability or the potential to provide refugia for numerous native species.Therefore, these species can nd shelter in urban landscapes holding a plethora of unique, unusual, and complex habitats in locations which are absent or rare in semi-natural and rural landscapes surrounding cities.It has been emphasised that the exceptional environmental conditions existing in cities have speciesspeci c impacts on species; these can range from negative to neutral or positive, depending on each species life-history and adaptation ability (Sol et al. 2014;Spotswood et al. 2021).Depending on the geographic context where cities are situated, different environmental conditions between cities and their surrounding landscapes create unique gradients in biotic and abiotic factors (Spotswood et al. 2021).Mainly, this unicity derives from the highly arti cial context within which such habitats are included, the urban matrix, and from their isolation with respect to analogous habitats outside the cities.
From our review, it appears that natural and seminatural habitats within cities offer the highest number of refugia to species.Indeed, as already stressed by other authors (e.g.Pregitzer et al. 2020), near-natural areas of urban forests can provide more bene ts than designed landscapes in conserving of native species and contrasting invasive ones.There is also evidence that forest patches widespread in cities harbour valuable rare, endemic, protected, and endangered species.Subsequently, they may hold a signi cant potential for the preservation of biodiversity from the gene to the ecosystem level (see Alvey 2006).Besides their habitat function and the role in protecting biodiversity, urban forests provide multiple ecosystem services like health bene ts to citizens, the regulation of water resources, and help in mitigating climate change (Berglihn and Gómez-Baggethun 2021).Although less mentioned than forests as refugia, grasslands have a bene cial role for biodiversity and ecosystem functioning within cities, as observed by Onandia et al. (2019).
Like forested areas, city parks have been frequently mentioned as biodiversity refugia across the investigated literature.However, city parks have a different role in comparison to natural and seminatural landscapes.Indeed, even if they can host a similar number of species than natural areas, there are differences in the species composition, in the traits selected and in the community structure (Banaszak-Cibicka et al. 2018).For instance, Chen and Cheng (2022) reported that the network of city parks promotes interactions between metacommunities of different animals like birds, reptiles, frogs, and butter ies.In any case, many land-use practices (e.g.mowing) and environmental factors (e.g.proximity to water bodies) have an effect on biodiversity in city parks in terms of specie composition, vegetation structure, and ecosystem services (Talal and Santelmann, 2019).
Water bodies includes multiple near-natural habitats suitable to serve as urban refugia especially for amphibian and aquatic or hygrophilous plants.Water bodies can have the function of stepping stone refuges, which assist species movement across aquatic landscapes, providing connectivity between natural or seminatural wetlands and running waters (Chester and Robson 2013).In urban areas, aquatic habitats, and the species they host (e.g.invertebrates), are severely affected by the water quality and the hydrological process of urban wetlands.For this reason, they need frequent management activities and are subject to restoration actions (Hale et al. 2019).
Numerous species have been reported to nd refugia in built areas offering physical structures analogous to their natural habitat.Man-made structures, like ancient walls, roads, or industrial sites (waste deposits), can often supply new chances for species to utilise urban spaces (Fenu et al. 2016).These areas cannot be comparable to natural environments but may still have suitable habitat characteristics for some species.In any case, it is well known that biodiversity in built areas is signi cantly lower than in other urban green areas or semi-natural areas (Kondratyeva et al. 2020;Toffolo et al. 2021).
On the other hand, historical parks, and green spaces (botanical gardens, sacred spaces, cemeteries, etc.) have a peculiar refugial function for higher plants (e.g., Jaganmohan et al. 2018).They can be halfway between man-made (built) and other green areas previously mentioned.Urban green spaces refer to a variety of habitat types ranging from remnant patches of native vegetation to roundabouts and private gardens have been frequently cited to function as urban refugia.The ecological function of green spaces is widely recognized both for urban planning and management of natural remnants in cities as well as for the conservation of biodiversity and restoration activities (Lepczyk et al. 2017).
Overall, all kinds of refugia emphasizes their potential for the conservation of species in urbanized areas including species worthy of conservation that in our synthesis were found in 33% of studies.Indeed, as stressed by Aronson et al. (2014), despite the low density in green spaces and species, cities can host rare, endemic, protected and endangered native species, thus providing prospects for biodiversity conservation efforts and education activities for citizens.In this direction, Ives et al. (2016) highlighted that cities are becoming hotspot for threatened species and that the species assemblages of individual cities are distinct from those of other ones.In any case, considering very recent studies on urban evolution in cities (Diamond and Martin 2021), also common or alien species can take on a different value or role in urban versus other environments.In this regard, species' populations living in cities can have different mutation rates, shifts in methylation patterns, phenotype and trait selection, and tness in comparison to rural populations (Diamond et al. 2018;Sepp et al. 2020; Diamond and Martin 2021, and cited references within).In other words, populations of plant and animal living in urban refugia may undergo speciation as frequently observed in climatic refugia (Gentili et al. 2015b).Indeed, all evolutionary forces such as isolation, selection/adaptation, mutation etc., may act in urban landscapes and especially in refugia.
Despite the presence of a peculiar biodiversity, all kind of urban spaces are continuously under pressure due to urban expansion, construction plans or incorrect environmental restoration (Boulton et al. 2020).Consequently, due to the fast changes to which cities are subject, urban refugia and the species that they hold can be undoubtedly prone to constant human pressure and extinction.However, the expected declines of biodiversity with rapid urban growth may be stopped if anthropogenic, semi-natural, and natural urban refugia are considered in urban and landscape plannings (Alvey 2006;Fingland et al. 2022;Hall et al. 2017;Salinitro et al. 2018), making citizens involved and aware of role of refugia in saving the biodiversity.Urban refugia should be acknowledged, monitored, and managed by practitioners (city planners and conservationists, ecologists, foresters, etc.) to improve their capacity to support the species they host and promote further biodiversity.Management activities should promote the increase of biodiversity in all aspects of urban green spaces, from trees along streets to city parks and near-natural forests.Mainly efforts should be directed towards conservation and restoration of native vegetation across all kinds of urban landscapes.For instance, some possible action should envisage the following activities: -listing of refugia and biodiversity they hold in each city; -planning and implementing a correct management of forest and grassland vegetation (selective clearcuttings, mowing, etc.), to maintain the habitat functionality over the seasons (Chollet et al. 2018;Pregitzer et al. 2019).
For the future, research assessing the record of new urban refugia, their link with other habitat outside cities, and speciation potential of isolated populations they host would be welcome.

-
continuous monitoring of biodiversity in the refugia also involving citizen (Callaghanet al. 2020); -creation of green infrastructures (Hostetler et al. 2021); -improving of ecological corridors both within cities and with peri-urban and rural areas in terms of quality and connectivity (Huang et al. 2021);

Figure 1 World
Figure 1

Figure 2 Groups
Figure 2