Addis Ababa, the capital of Ethiopia, has a population of ~4 million and is one of the fastest growing cities in sub-Saharan Africa (UN-Habitat, 2008). The city is developing large urban extensions and is restructuring and rebuilding existing developed areas, while informal settlements also continue to grow. The city grows at the fringe, thus causing loss of farmland, and at the same time densifies. Moreover, informal settlements have developed along river corridors, which are prone to flooding, and on hillsides. UGI-related goals for Addis Ababa include increasing the (currently low) cover of accessible green space, restoration of river corridors, enhancing food security via urban farming, and improving the quality and functionality of all non-built up areas (Wubneh, 2013). Many of these goals have been laid out explicitly in the city's recent master plan (AACPPO, 2017), though the implementation often falls short of the idealized planning vision due to practical reasons like pressure from developers and other economic/political forces (Herslund et al., 2018). There are currently three public institutions responsible for managing UGI in Addis Ababa. These are (i) Environment and Green Development Commission which is a regulatory body, (ii) River Basins and Green Area Development and Administration Agency which is an operational body and (iii) Farmers and Urban Agriculture Commission, an operational body responsible for providing extension service to urban farmers involved in urban agriculture.

To date there have been a handful of research studies of UGI in the city. Woldegerima et al. (2017) highlighted the potential role of different forests for the provisioning of ES such as carbon sequestration and soil protection. In a study of urban forest planning it was recognized that urban forests provided important functions, but that they are also lacking definite borders and thus in need of re-demarcation (Fetene and Worku, 2013). Within the city, parks act as important temperature regulating resources locally (Feyisa et al., 2014; Teferi and Abraha, 2017). At the same time, the city is undergoing rapid changes, highlighting the need for action to connect and safeguard UGI in order to sustain its heat reducing capacities and other key ES (Teferi and Abraha, 2017; Woldegerima et al., 2017). Several studies have also dealt with the role of UGI in stormwater management (e.g., Herslund et al., 2018; McFarland et al., 2019).

Cincinnati is a Midwestern US city with an urban population of ca. 300,000 inhabitants. The city has a well-established urban park and street tree network, and is one of the top-ranked cities in the U.S. for park accessibility with 82% of residents living within a 10-min walk of a park (Trust for Public Lands, 2020). As a post-industrial city, Cincinnati lost population following WWII, dropping from >500,000 in 1960 to the current level by about 1990, and has recently begun to rebound in population (U.S. Census Bureau, 2020). Substantial areas remain underdeveloped with 7% of the city's area classified as “vacant land”; at the same time, the urban core is experiencing renewed growth accompanied by a re-greening of the city's riverfronts, converting formerly industrial areas to public parks and other community spaces. Pressing environmental and social issues include poor air quality, water pollution primarily from Combined Sewer Overflows (CSOs), and a large income and health inequality, with high infant mortality and low life expectancy particularly among the poor and Black communities (Cincinnati City Planning Commission, 2012). The city's Metropolitan Sewer District (MSD) is currently under a consent decree from the US Environmental Protection Agency to reduce CSOs by 85% - at an estimated cost of >$3,000,000,000. Thus, a primary focus in implementing new UGI has been on reducing stormwater runoff, although co-benefits including economic and ecological benefits are cited in project plans (MSD, 2020). As part of this effort, a wide range of NbS have been implemented including daylighting streams and restoring riparian zones, retention/detention basins, bioswales, rain gardens, rain barrels, previous pavers, and green roofs (www.projectgroundwork.org). The governance and management of UGI in Cincinnati is complex, involving a large number of different societal actors and stakeholders including regional and national government agencies, NGOs, local communities, universities and private landowners as well as more than 30 local municipalities (Shifflett et al., 2019).

There are a growing number of peer-reviewed studies led by researchers from the U.S. EPA office in Cincinnati, focusing largely on the biophysical capacity and economic viability of UGI to reduce stormwater runoff in the city. For instance, the capacity of rain gardens and rain barrels to reduce stormwater runoff and improve local stream ecosystem health has been tested (Shuster and Rhea, 2013; Roy et al., 2014), and a recent hydrologic modeling study pointed to the combined use of UGI and gray infrastructure to address CSO problems (Fu et al., 2019). A life cycle assessment of the implementation of different UGI elements in Cincinnati revealed that residential rain gardens could be a cost-effective approach (Vineyard et al., 2015), while another study compared street tree coverage and associated stormwater runoff benefits among different communities within the city (Berland and Hopton, 2014).

Malmö is the third largest city in Sweden, with ~335,000 inhabitants. In 2019, one-third of Malmö's residents had been born outside of Sweden, representing at least 182 different nationalities (Malmö Stad, 2018). Malmö has an industrial background with a focus on heavy industry until the mid-1970s, followed by an economic downturn and decrease in population. Malmö's transition into a post-industrial city has been developing since the mid 1990's but the city still faces ecological, economic and social challenges. Cities in southwestern Sweden have low green space cover as compared to Sweden as a whole (Statistics Sweden, 2019) and Malmö has experienced a gradual decrease in amount of green space per inhabitant over the last 20 years, which might be related to both increased population and continued brownfield and infill development (Malmö Stad, 2021). Malmö has been heavily impacted by recent extreme rain events where substantial surface flooding has hit in particular low lying areas and areas along now-culverted rivers. In general, Malmö has several challenges in relation to urban green infrastructure development as it ranks the lowest in Sweden and in the lower part overall on the ISGlobal ranking of European cities (Pereira Barboza et al., 2021). To this end, the City of Malmö has led several projects to develop urban green space for climate adaptation as well as urban biodiversity and socioeconomic development. In particular, Augustenborg and the western harbor areas have developed as international flagship projects for UGI over the past 20 years, with background information and baseline studies available on environmental performance (Haghighatafshar et al., 2018; Sörensen and Emilsson, 2019).

Malmö has also been included as a case city in multiple research studies focusing on challenges in governance aspects related to UGI including studies broadly covering ES (Wamsler et al., 2014; Schubert et al., 2018) and others focused specifically on stormwater management (Qiao et al., 2019; Wihlborg et al., 2019). The role of UGI for health and recreation has also been demonstrated in several Malmö-based studies, with themes including promoting physical activity (Qvistrom, 2016; Grabalov, 2018), green school yards (Jansson et al., 2018), urban agriculture's role for social cohesion (e.g., Vierikko et al., 2016) and developing health related indicators (Van den Bosch et al., 2016). While both the concepts of ecosystem services and sustainable stormwater management are familiar within the city government, there remain governance related challenges to their wider implementation (Wamsler and Pauleit, 2016; Schubert et al., 2018; Qiao et al., 2019). The governance and management of UGI in Malmö is primarily a concern of the city authorities, but a large amount of urban green spaces are also under the ownership and related management of private and semi-private housing companies.

The aims of the workshops were to (1) establish collaboration with key local expert stakeholders; (2) better understand the current situation and existing challenges to provisioning of ES from UGI; (3) elicit individual and collective visions for future development of UGI in each city. Participants included researchers and stakeholders from the city administration and other public organizations involved in researching, planning and managing urban green spaces and green infrastructure, as well as relevant stakeholders from outside public administration, including private consulting companies and NGO's (Supplementary Table S1).

During the workshop, each participant was asked to complete a questionnaire regarding the current provision of ES from green spaces/green infrastructure in the focal city. The questionnaire (Supplementary Table S2) consisted of a list of 36 ES (Table 2) compiled by browsing various manuals and frameworks for working with ES (MEA, 2005; EU, 2013; Naturvårdsverket, 2018). We aimed for the list of ES to be exhaustive rather than aligning with any particular framing of ES, and no particular spatial scale was specified. For each service the participants were given an option to rate the current provisioning of each ES from the city's UGI on a scale from 1 to 5, where 1 indicated “not provided for at all” and 5 indicated “very well provided for”, and also included an option to rate it “not relevant”. The questionnaire also had space to specify additional services not captured by the list and to rate these, as well as a blank space for additional comments. The questionnaire response rate was 81% (17 of 21 workshop participants) in Addis Ababa, 80% (24 of 30) in Cincinnati, and 100% (11 of 11) in Malmö. The average assessment value was calculated for each ES for each city, as the mean value given by all respondents for the given ES. The actual number of respondents thus varied from 11 to 24 depending on city, which should be adequate to get a trustworthy estimate of expert opinion for each given city (e.g., Elliott et al., 2020). Due to the non-random selection of participants and unequal sample sizes among the different cities, a direct statistical comparison among the results for different cities was not made, but results were visualized for comparison.

A Q-sorting exercise (e.g., Barry and Proops, 1999) was the central interactive part of the workshops, with participants working in small groups (4–6 people each) to generate consensus priorities for future provisioning of ES from UGI. Participants were not given a specific timeframe for the future, other than having a long-term perspective, which may be considered to be 25 years or more (van Notten et al., 2003). Group members were assigned by the organizers in order to include a variety of backgrounds and stakeholder roles within each group, and to avoid multiple representatives from the same stakeholder organization in the same group. To guide group discussion, we used Q-boards (Supplementary Figure S1) with 36 slots shaped in a pyramid and corresponding cards naming the same 36 ES as in the questionnaire (Table 2; Supplementary Table S2). Participants were instructed to negotiate and agree within their group on ranking the ES on a scale of importance from 1 to 7 (1 = least important, 7 = most important) with a varying number of slots by category including only two slots in the “most important” category (Supplementary Figure S1). The aim of this exercise was to establish a negotiated consensus for rankings within each of the smaller groups, followed by an open discussion with all of the groups, primarily focusing on the rationale for choosing the two most important key priorities.

For each city, the average values for perceived current ES provisioning (using the data gathered from the individual questionnaires) and the future priorities (from the Q-sorting group exercise) were reported for each ES, after rescaling the data from each of these two exercises to express them on the same scale (from 0 to 4). A third category called apparent urgency was also calculated for each ES, defined as the difference between perceived current provision and future priority after rescaling. High values for urgency indicate ES for which future priority is high while current provision is low. Negative values for urgency indicate ES for which current provision outstrips the future priority. Values near neutral for urgency indicate that the assessment of current provision and future priority were similar—note that this can happen either with two high values (high priority for future but also high current ES provision) or two low values, or anything in between. Thus, the analysis of apparent urgency is only meaningful in conjunction with analysis of the ES provision and priorities, and is most useful for identifying the most urgent high priority ES. In our analysis we highlight ES which have an apparent urgency value of >0.4, which corresponds to a value >10% of the total potential range of values.

We conducted four expert interviews in each of the three cities: three individual, semi-structured interviews and one group interview with 3–4 participants where a semi-structured discussion was facilitated with a picture of the Cascade Model of ES (TEEB, 2010) as reference-point. These interviews served to develop a more full understanding of the context and rationales behind the responses to the questionnaire and Q-sorting exercise. The interviewees were selected among the workshop participants and in dialogue with the main stakeholder collaborators in each city. We aimed to represent expertise from different organizations, and the respondents included municipal planners and managers, academics, and representatives from local and regional special interest groups (Supplementary Table S3). All interviewees received both oral and written information about the study and had the chance to ask questions before deciding on participation. All interviewees provided written consent for voluntary participation in the study.

The individual interviews followed a semi-structured approach, using a thematic and dynamic question guide (Kvale and Brinkmann, 2009) shown in Supplementary Table S4. Each interview took about 45–60 min and was conducted by members of the research team in the days following the workshop. All interviews were recorded and verbatim transcribed. Following transcription, interviews were analyzed independently by two members of the research team. The analysis was primarily exploratory, to develop in-depth understandings of the interviewees NbS, UGI and ES perceptions. Initially, the analysis focused on three main aspects: (1) Concrete examples of NbS explicitly connected to ES provision, (2) Rationales for the current UGI situation/strategies and potential future developments, (3) Challenges to the current UGI situation/strategies and potential future developments. In the analysis, emphasis is placed on key themes that occurred in at least two of the three interviews for each city.

The group interview was undertaken with a group of 3–4 local expert participants in the days following the initial workshop. Participants were different from the individual interviewees, to help broaden the perspective and allow for new insights. The interviews followed a semi-structured approach organized around the Cascade Model for ES (Figure 1) as a framework. Two NbS were placed in focus, in two consecutive discussion sessions of about 60 min each: street trees and mid-sized city parks, both of which are ubiquitous and important NbS as parts of the overall UGI in each of the three cities and most others worldwide. Participants were asked to envision the specific NbS in their city, then start by listing and describing the ES provided, and gradually proceeding to discuss other aspects related to ES provisioning. During those interviews, when biophysical aspects of NbS were discussed, the micro-scale was often in focus—for instance, the ways in which traits like crown shape of different specific tree species affected their ability to provide shade. The results of these interviews were summarized both as Cascade model diagrams illustrating the identified factors and barriers, and as a list of challenges to ES provision, organized by domain (biophysical, societal, or governance-related).

Local expert stakeholders perceived all four ES categories (cultural, provisioning, regulating and supporting ES) as being more or less equally provided at a low-intermediate level in Addis Ababa, while in both Cincinnati and Malmö, cultural ES were seen as being more well-provided than either regulating or supporting ES, while provisioning ES were seen as only minimally provided (Figure 2A).

In general, these patterns were reflected in the responses for individual ES (Figure 3), with for instance Cincinnati scoring particularly high on the cultural ES “physical recreation,” “relaxation,” and “socializing informally.” There were also some clear among-city differences, for instance spiritual experience and religious values are seen as well-provided by UGI in Addis Ababa but not in the other two cities; while green spaces are seen as providing arenas for group activities only in Cincinnati and Malmö, but not Addis Ababa. Individual regulating ES were generally assessed at similar intermediate-low levels for all three cities, though Malmö was seen as somewhat more successful providing storm water management than the other two cities. In contrast, the assessment of the current availability of provisioning services varied widely among cities, with Addis > Cincinnati > Malmö. In Addis Ababa, provision by the urban landscape and associated vegetation of raw materials for building, fuels for household use, and food were noted as particularly prevalent (Figure 3).

NbS/UGI emphasis from individual expert stakeholder interviews varied from city to city, though with some common to all, particularly street trees, city parks, and river corridors/blue spaces (Table 3). For Addis Ababa a primary focus was on developing neighborhood parks, while Cincinnati stakeholders emphasized integrating existing parks as part of a larger-scale watershed management strategy. In Malmö, common examples included green corridors along roads and paths connecting parks from city center to periphery. Interviewees from Malmö also emphasized the relevance of green roofs and walls, as well as the city's blue spaces such as coast, canals, ponds, and streams. Urban agriculture was additionally highlighted by the interviewees from Addis Ababa.

There was variation in ES emphasis among the cities (Table 3), but also common threads and a clear recognition of the multifunctionality of individual NbS, particularly for street trees and city parks. Examples from street trees in Addis Ababa included local climate regulation with emphasis on shade and heat regulation, but also to create a sense of place. In Malmö and Cincinnati local climate regulation and walking/cycling possibilities provided by street trees and green corridors were primarily emphasized. In Malmö, emphasis was also placed on recreational and aesthetic elements, while in Cincinnati, emphasis was placed on integrating street trees with bioswales, in order to better manage the city's stormwater and flooding issues. For city parks, across the three cities, cultural ES (socializing and group activities, relaxation, physical recreation, mental and physical health, aesthetic value, and sense of place) and regulating ES (regulation of air quality, noise, stormwater runoff, local climate, and habitat provision) were primarily emphasized.

For Addis Ababa, each of the four different ES categories were prioritized by stakeholders about equally for the future (Figure 2B). However, the top individual ES priorities (Figure 3) centered on provision of food and clean water for both household-use and irrigation. Simultaneously, the stakeholders saw issues with food security and the importance of urban agriculture for feeding the current population of Addis Ababa as an essential service provided by UGI. Biodiversity and “awareness and understanding of nature” were also prioritized.

For Malmö and Cincinnati, regulating and supporting ES were given by far the highest priority for the future, then cultural ES, and lastly provisioning ES (Figure 2B). For Cincinnati, “Stormwater management” emerged as a clear top priority, with “adaptation to climate change,” “mental and physical health,” and “flood control” also very highly ranked (Figure 3). The strong focus on UGI for stormwater management pervaded the discussions of for example street trees, river corridors, and governance-regimes and incentives, while a broader perspective including more potential ES also emerged. For Malmö, “Adaptation to climate change” emerged as the most commonly selected priority, with stakeholders noting that this ES could encompass a range of other services, and thus be a good strategic pick to garner multiple ES. Also prioritized in Malmö were “mental and physical health,” “biodiversity,” “habitats for species,” and “stormwater management” (Figure 3).

For each city, apparent urgency was estimated for each individual ES by taking the normalized difference between future priority and perceived current provision. We highlight those ES with particularly high urgency values, which we interpret as having high perceived potential to help address pressing urban challenges through further expansion/development of UGI (Figure 3; See Supplementary Table S2 for numerical values). The results were also summarized by ES category (Figure 2C).

For all three cities, a majority of regulating and supporting ES were identified as having a high degree of urgency (Figure 3). Although most provisioning ES were seen as adequate currently for all three cities, a particularly strong urgency was identified for the provision of freshwater for irrigation, freshwater for household use, and food in Addis Ababa. The same three ES had high perceived urgency in Malmö as well, with provision of food highlighted for Cincinnati as well. Finally, awareness and understanding of nature and mental and physical health emerged as important cultural ES with high apparent urgency for all three cities (Figure 3).

Based on the expert interviews, challenges to UGI optimization were identified in all three domains (biophysical, social, governance) in each of the cities (Table 4), and were most often related to gaps in governance, structural pressures and lack of ecological vison and awareness.

During the group interviews focused on urban parks and street trees, many challenges to UGI optimization were highlighted by the expert stakeholders (Table 4), and these challenges could be depicted as barriers between the different parts of the ES cascade model (e.g., Figure 4). For instance, in Addis Ababa many of the existing parks are closed to the general public except during limited hours, and require entrance fees. This introduces inequity and prevents certain groups of people from realizing the benefits of the parks, even though the green spaces may be physically providing the desired functions (space for recreation and sport, shading, cooling, beautification, etc.). This challenge can best be depicted as a barrier between the “Ecosystem service” and “Benefit” boxes in the Cascade Model diagram (Figure 4).