Landscape and Urban Planning 171 (2018) 7–17

Contents lists available at ScienceDirect

Landscape and Urban Planning

journal homepage: www.elsevier.com/locate/landurbplan

Research Paper Impact of environment on people’s everyday experiences in Stockholm T ⁎ Karl Samuelssona, , Matteo Giustib, Garry D. Petersonb, Ann Legebyc, S. Anders Brandta, Stephan Barthela,b a Department of Industrial Development, IT and Land Management, University of Gävle, Kungsbäcksvägen 47, 801 76 Gävle, Sweden b Stockholm Resilience Centre, Stockholm University, Sweden c School of Architecture, KTH Royal Institute of Technology, Sweden

ARTICLE INFO ABSTRACT

Keywords: In order to construct urban environments that limit negative impacts for global sustainability while supporting Affordances human wellbeing, there is a need to better understand how features of the environment influence people’s Urban social-ecological systems everyday experiences. We present a novel method for studying this combining accessibility analysis and public Urban ecosystem services participatory GIS (PPGIS). Seven environment features are defined and accessibility to them analysed across Public participatory GIS Stockholm municipality. We estimate the probabilities of positive and negative experiences in places based on Spatial regression these environment features, by using spatial regression to extrapolate from the results of an online PPGIS survey (1784 experiences of 1032 respondents). Six of the seven studied environment features have significant impact on experiential outcome, after accounting for spatial autocorrelation among the data. The results show that number of residents and proximity of nature environments and water, all common quality indicators in urban planning and research, have weak statistically significant effects on people’s experiences. However, areas dominated by large working populations or proximity to major roads have very low rates of positive experiences, while areas with high natural temperature regulating capacities have very high rates, showing that there are considerable qualitative differences within urban environments as well as nature environments. Current urban planning practices need to acknowledge these differences to limit impacts on the biosphere while promoting human wellbeing. We suggest that a good way to start addressing this is through transformation of negatively experienced urban areas through designs that integrate closeness to urbanity with possibilities to have nature experiences on a daily basis.

1. Introduction highlighting the importance of urban inhabitants interacting with nature environments (e.g. Soga et al., 2014). Such interactions provide As the world continues to urbanise, cities need to develop so that possibilities for restoration from stress (van den Berg, Hartig, & Staats, negative impacts for global sustainability are limited (Bren d’Amour 2007), foster psychological connections between urban inhabitants and et al., 2016; Grimm et al., 2008; Kennedy et al., 2015), while wellbeing the biosphere (Soga & Gaston, 2016) and promote physical and mental among urban inhabitants is supported. Much of current thinking in health (Gascon et al., 2015; Mitchell & Popham, 2008). In the words of urban sustainability research and policy promotes compact cities − Hartig and Kahn (2016), “cities designed well, with nature in mind and cities with high citizen density and contained extent. Compact city at hand, can be understood as natural, supportive of both ecosystem development is argued to mitigate climate change impacts by de- integrity and public health”. Here, we refer to this narrative of urban creasing car dependency (Newman, 2006), enabling sustainable modes development as the social-ecological city. These conflicting spatial of transportation (Jabareen, 2006) and requiring less energy-spending paradigms must be reconciled to achieve urban environments that on heating (Kennedy et al., 2015). Moreover, compact cities have fur- support social and ecological sustainability at scales from the local to ther gained favour, as they can promote biodiversity conservation the global. We approach this issue by studying how environment fea- (Soga, Yamaura, Koike, & Gaston, 2014) and ecosystem services (Bren tures emphasised within each paradigm together influence people’s d’Amour et al., 2016; Stott, Soga, Inger, & Gaston, 2015) outside cities. regularly occurring experiences. However, the compact city paradigm has been challenged by research Experiences are a mediating factor between the environment and

⁎ Corresponding author. E-mail addresses: [email protected] (K. Samuelsson), [email protected] (M. Giusti), [email protected] (G.D. Peterson), [email protected] (A. Legeby), [email protected] (S.A. Brandt), [email protected] (S. Barthel). https://doi.org/10.1016/j.landurbplan.2017.11.009 Received 21 July 2017; Received in revised form 17 November 2017; Accepted 19 November 2017 0169-2046/ © 2017 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/). K. Samuelsson et al. Landscape and Urban Planning 171 (2018) 7–17 wellbeing (Kyttä, Broberg, Haybatollahi, & Schmidt-Thomé, 2016). We through accessibility analyses and 3) an ecological psychology ap- study the relationships between environment and experiences by ap- proach for analysing human-environment relationships, oper- plying affordance theory (Gibson, 1979), both in our quantification of ationalized through public participatory GIS (PPGIS). the physical environment and analysis of experiential outcome. Chemero (2003) defines affordances as “relations between abilities of 2. Methods organisms and features of the environment”.Affordances only emerge when different characteristics of the individual, such as her physical 2.1. Study area abilities, emotions and intentions, are matched in meaningful relations with environment features (Withagen, de Poel, Araújo, & Pepping, Our study area is Stockholm municipality. It forms the centre of the 2012). However, realisations of affordances are probabilistic rather Stockholm urban region. The larger region is home to approximately than deterministic (Altman & Rogoff, 1987). Typically, affordances 1.5 million people, of which around 950,000 live in the municipality. It refer to opportunities or restrictions for behaviour, but there has been a is situated between the Baltic Sea and Lake Mälaren, and has large areas growing understanding of affordances in urban environments as also of nature and water for a city − built-up areas, nature areas and water including opportunities or restrictions related to experiences (Kyttä, bodies each make up roughly one third of the study area. The city’s Broberg, Tzoulas, & Snabb, 2013), and development of environmental inner core consists of mostly compact neighbourhoods with a central attitudes (Marcus, Giusti, & Barthel, 2016). business district and surrounding mixed-use neighbourhoods, whereas Within the compact city narrative, urban environments are argued suburbs include detached housing areas, apartment blocks and more to support wellbeing through opportunities for social and economic recent mixed-use neighbourhoods. The municipality’s population has interactions between people (Dempsey, Brown, & Bramley, 2012; grown by 26% over the last decade, and it faces challenges of main- Legeby, 2013a). Jane Jacobs' (1961) writings were seminal for the taining large connected nature areas while accommodating an in- understanding of how configurative properties of neighbourhoods in- creasing population. fluence possibilities for services, social capital and street life to emerge. Urban space is not simply a setting for social and economic activities, 2.2. Selection of environment features but directly shape them through its configurative properties (Hillier & Hanson, 1984). For example, the preconditions for people to collec- To assess how environment features emphasised within each para- tively share public space influence social segregation (Legeby, 2013b). digm together influence people’s experiences, we chose features whose Configurative properties of neighbourhoods also condition labour impact is well documented in the literature on spatial accessibility of market opportunities (Legeby, Pont, & Marcus, 2015) and possibilities urban resources, urban ecosystem services or environmental psy- for outdoor recreation (Ståhle, 2008; Vries et al., 2007). Consequently, chology. We also ensured that public geographic data for the entire area urban inhabitants’ experiences of their everyday environment are im- was available and that the data had sufficient spatial resolution. We portant indicators of opportunities to access urban resources (Legeby, assessed the accessibility to six features: residential population, 2013b). working population, nature environments, playgrounds and school- Within the social-ecological city narrative, urban nature is argued to yards, water bodies and major roads, as well as local provision of nat- support wellbeing both through direct interaction and processes that ural temperature regulation (Table 1). Temperature regulation here are passively enjoyed. Often termed urban ecosystem services (Bolund refers to the ability that vegetation has to regulate local temperatures, & Hunhammer, 1999), these bene fits are generated within landscapes and capacities depend on volume and type of vegetation, as well as (Andersson, McPhearson et al., 2014) and their supply is influenced by continuous size of vegetated areas (Barthel et al., 2015). urban form (Tratalos, Fuller, Warren, Davies, & Gaston, 2007). For instance, if regulating services, such as temperature regulation, are to 2.3. Quantifying features be enjoyed it is important that they are locally supplied (Andersson, Barthel et al., 2014). In environmental psychology research, a large Accessibility to environment features was mapped across Stockholm body of literature has identified access to nature and water corre- to visualize their spatial patterns. As our definition of accessibility had sponding with psychological restoration (van den Berg et al., 2007), to be relevant for analysing regularly occurring experiences, rather than subjective wellbeing (MacKerron & Mourato, 2013), lower prevalence using administrative boundaries, it was defined as being within 500 m of mental health issues (Gascon et al., 2015), and increased physical from a measurement point. This distance is used in Stockholm muni- activity (Frank, Kerr, Chapman, & Sallis, 2007). On the contrary, reg- cipality’s own planning documents (Stockholms Parkprogram, 2006) ular exposure to major roads is associated with adverse health out- and similar to applications in other studies (Coombes, Jones, & comes (Passchier-Vermeer & Passchier, 2000). In addition to direct ef- Hillsdon, 2010; Kyttä et al., 2016). To analyse our data consistently, we fects, perceptions of one’s environment mediates wellbeing outcomes created a grid with 10 m resolution encompassing all land surfaces in (Gidlöf-Gunnarsson & Öhrström, 2007), stressing that definitions of the study area, and used cell centres as measurement points. This re- spatial accessibility need to be meaningful from a perception and cog- solution provides a level of detail relevant for most urban planning nition perspective. projects, while not being overly computationally taxing to analyse. This study aims to integrate the compact city and social-ecological Cells closer than 500 m from the municipality border were excluded city narratives to create a nuanced understanding of how urban en- from the analysis, as they would suffer from edge effects due to data vironments influence people’s regularly occurring experiences. To do lacking outside the border. this, we quantify relationships between different levels of spatial ac- Accessibility analysis of environment features was done using Place cessibility of a wide range of environment features and people’s ex- Syntax Tool (Ståhle, 2012). The accessibility of all features, but tem- periences. Our primary research question is (i) How does spatial ac- perature regulation, was quantified by measuring the walking distance cessibility to various environment features impact inhabitants’ to them (see the Supplementary Material). Residential and working regularly occurring experiences? To answer this, we also answer (ii) population were calculated as population within 500 m walking dis- How does spatial accessibility to these features differ across Stockholm, tance. Thus, quantifications for these features depend not only on street Sweden? As increasingly called for (Andersson, McPhearson et al., layouts but also citizen density, but for simplicity we refer to these 2014; Kyttä et al., 2016), our method is spatially and experientially quantifications as measuring accessibility. For nature environments, explicit and integrates 1) a social-ecological perspective of nature en- playgrounds, water and major roads, the distance to the closest feature vironments, operationalized through ecosystem service mapping, 2) a was measured with 500 m as the maximum distance considered. The configuration perspective of the built environment, operationalized remaining feature, temperature regulation was quantified as average

8 K. Samuelsson et al. Landscape and Urban Planning 171 (2018) 7–17

Table 1 Environment features analysed and methods of analyses. Data sources and detailed descriptions are provided in Appendix A.

Feature Unit Measurement Description of base layer

Residential population Person Number within walking Residential population per property for 2015 as point data located at the centre points of properties, distance obtained from Stockholm County’s Growth and Regional Planning Administration. Working population Person Number within walking Working population per address point, created from a dataset obtained from Stockholm distance Municipality’s City Planning Administration containing working population per property in 2008 as point data located at the centre points of properties, together with a layer containing address locations as point data. Playgrounds and schoolyards Metre Shortest walking distance Sociotope map, created by Stockholm municipality’s City Planning Administration in 2014, with areas defined as recreational public spaces as polygon data, divided in different categories. Major roads Metre Shortest walking distance The road network of Stockholm county represented as lines, obtained from Geographic Data of Sweden in 2016. Nature environments Metre Shortest walking distance Same map as for playgrounds and schoolyards. Water body Metre Shortest walking distance Polygon data of water surfaces within Stockholm County, obtained from Geographic Data of Sweden in 2016. Natural temperature Scale 1–5 Average within buffer A mapping of temperature regulating capacity produced in 2014 through a multi-criteria method, regulation distance obtained from Stockholm municipality’s City Planning Administration. values within a 500-m buffer distance − as air moves, the utility of outcome. these areas is not dependent on people being able to walk to them. For the experiences containing demographic data, both models were fitted with and without age group and gender terms. A Rao score test and Akaike Information Criterion (AIC) was used to compare models. 2.4. Collecting data on people’s experiences As the inclusion of demographic terms was found to worsen model fits, all experiences were then modelled without demographic data. To analyse the quality of people’s experiences in Stockholm we used Variance inflation factors were used to check for collinearity among data from an online PPGIS survey designed to capture people’s reg- features. As we anticipated experiential outcome to be non-linearly ularly occurring positive and negative experiences in the city. The related to some features, models were tested with quadratic variables in survey, called Where is your Stockholm? (Giusti, Barthel, Samuelsson, addition to linear, and model fits were compared using AIC. To examine Stockholm University, & Stockholm Resilience, 2017), aimed to create a spatial autocorrelation, autocovariates were created from model re- dataset of empirically recorded affordances to be used for several stu- siduals, and new models including the autocovariates were fitted. Any dies. The survey was designed to be easy to use and focused on people’s remaining autocorrelation among residuals was evaluated through experiences. The survey respondents selected locations to record, and Moran’s I tests. The neighbourhood size for autocovariates was chosen whether the experience they had in that location was positive or ne- so that model fit was maximised while residual autocorrelation was gative. After marking the location and positive or negative nature of an statistically insignificant. Frequency and duration of experiences were experience, respondents could provide more data about their experi- graphically investigated in order to analyse qualitative differences be- ence (for full details, see Giusti et al., 2017). For this study, the attri- tween positive and negative experiences. Autocovariate values of ex- butes of frequency and duration of experiences were utilised for ana- periences were interpolated across the grid so that each cell obtained an lysis. There was no lower or upper limit to how many experiences a estimated autocovariate value in addition to feature values. This al- respondent could record. Respondents were also asked to provide socio- lowed us to predict probabilities of experiential outcome for each cell, demographic information: age group, gender and time of living in so that the model could be mapped across the study area. Stockholm. These questions were also optional to answer. For this study, data on age group and gender were used as controls in the sta- tistical modelling. 3. Results The survey collected data from the public over a period of about eight months, from September 21 st 2015 until May 31 st 2016. It was 3.1. Spatial distribution of environment features also featured at the exhibition “Experiment Stockholm” at Färgfabriken, an art hall and policy-practice arena in southern Stockholm, during the The largest residential population within 500 m walking distance is fall of 2015. A Facebook page and Twitter account for the survey were about 18 000 in southern inner city neighbourhoods, whereas the created to spread awareness of its existence and post interest-raising Stockholm median is about 1100 (Fig. 1a). Little more than a percent of comments provided by respondents. Several municipalities within the study area has residential populations above 10,000, whereas al- Stockholm County helped raising awareness of the survey’s existence by most half have numbers below 1000. This indicates that differences spreading information online, on notice boards and in local news- within Stockholm municipality are large and few locations can be de- papers. This method does not produce a representative sample of the scribed to have high residential density. Working populations are even population, but it does provide a large sample of experiences from more spatially concentrated, with the maximum being about 44,000 people engaged in understanding and improving city life in Stockholm. and the median about 200 (Fig. 1b). 1.5% of the study area have working population numbers above 10,000, whereas more than 80% 2.5. Modelling and mapping of experiential outcome have numbers below 1000. Differences in relative sizes of residential and urban populations are particularly noticeable in the inner city, as To estimate the impact of environmental features on the quality of the city centre have the largest working population within 500 m but people’s experiences we used spatial logistic regression models using R almost no residential population, while surrounding inner city neigh- (R Core Team, 2016). Logistic regression was used to relate the pre- bourhoods have a mix of residential and working populations. Outside sence of environmental features to the probability of an outcome. We the inner city, most neighbourhoods have larger residential than logarithmically transformed population values because both social and working populations within 500 m. 13% of the area have both re- spatial properties of cities are related to the power-law scaling of po- sidential and working populations less than 100 and are instead pulations (Bettencourt, 2013). Thus, we assumed each doubling of dominated by nature environments. population to correspond to a similarly sized change in experiential Access to playgrounds and schoolyards is similar in the inner city

9 K. Samuelsson et al. Landscape and Urban Planning 171 (2018) 7–17

Fig. 1. Maps show the respective spatial distributions of environment features, where cells display values of the surrounding neighbourhood. The maps are color- coded so that more intense colours represent the features being more prominent: for populations they represent more people, for temperature regulation they represent higher capacities and for remaining features they represent smaller walking distances to the feature. Temperature regulation is given on a scale where 1 = “no or little regulating capacity”, 2=“some regulating capacity”,3=“palpable reg- ulating capacity” and 4 = “great regulating capa- city”, according to the definition by Barthel et al. (2015).

and suburbs (Fig. 1c). Almost half of the study area have 500 m walking corresponds to “little or no regulating capacity” in the ecosystem ser- distance or more to the closest playground, housing about a quarter of vice mapping), 70% have values between 1.5 and 2.5 (2 = “some the total population. Major roads have a substantial footprint in regulating capacity”), 22% have values between 2.5 and 3.5 Stockholm, with about 40% of the area being closer than 500 m (3 = “decent regulating capacity”) and 1% have values between 3.5 walking distance of a major road (Fig. 1d). Around 40% of residents live and 4.5 (4 = “great regulating capacity”)(Fig. 1g). No measurements in this area while almost half work in it. were above 4.5 (5 = “excellent regulating capacity”). The lowest va- Roughly a third of the study area consist of nature environments lues are mostly found in the central city, and the highest on the study (Fig. 1e). Larger continuous nature environments are located outside area’s edges (Fig. 1g). the inner city, but large parts of the inner city have smaller nature environments within walking distance. 4% of the area, mostly in the 3.2. Responses to PPGIS survey city centre or in areas dominated by industrial activities or logistics, have walking distances of 500 m or more to the closest nature en- The PPGIS survey collected 1784 experiences within the study area, vironment, containing 2.5% of residents and little more than 10% of from 1032 respondents (Fig. 2). Most of the experiences were positive, working people. Approximately three quarters of the study area is more 1283 experiences (72%), while 501 (28%) were negative. The majority than 500 m walking distance to the closest water body (Fig. 1f), and of experiences were located in the inner city. Outside of the inner city, these areas contain about three quarters of residents and two thirds of some areas had high concentrations of experiences while other were working people. completely void of them. Generally, positive experiences were more Most of the region has some temperature regulating capacity within scattered across the landscape and negative experiences more con- 500 m. 8% of the study area have an average temperature regulating centrated (see Fig. 2). capacity within 500 m radial distance below 1.5 (the value 1 Age and gender data were provided by 451 respondents (44%),

10 K. Samuelsson et al. Landscape and Urban Planning 171 (2018) 7–17

Fig. 2. Visualization of experiences recorded by respondents within the study area. In total, 1013 respondents recorded 1784 experiences. The density of experiences varied greatly, being highest in the inner city and generally lower in suburban areas. Negative experiences were more con- centrated to certain areas, whereas positive experiences were more scat- tered.

together recording 1014 experiences (57%). The most frequent age The variables were tested for nonlinear relationships with experi- group was people aged 25–34, with 322 respondents (32%). Together ential outcome. As can be seen in Table 1, the best fitting model spe- with people aged 35–44 (283 respondents, 28%), they made up the cified residential and working populations and walking distance to majority of the sample. 158 respondents were younger than 25 (16%), major roads and nature environments as non-linear and walking dis- 158 aged 45–54 (16%) and 93 were aged 55 or older (9%). 464 re- tance to playgrounds and water and temperature regulation as linear spondents were men (46%), 527 were women (51%), and 23 preferred predictors (ΔAIC = –55.9 compared to all predictors being linear). not to specify gender (2%). The average number of experiences per Probabilities of experiences being positive decrease with increasing person was 1.73. This number did not vary significantly when com- residential and working populations, respectively. However, whereas paring genders or age groups. the probability stabilises at roughly 75% for residential populations over 1000 people (Fig. 3a), there is a sharp decline for working popu- lations above 1000 people, with less than one in five experiences being 3.3. Impact of environment on experiential outcome positive in areas with the largest working populations (Fig. 3b). Prob- abilities of positive experiences decrease from around 90% when being We estimated spatial logistic regression with a binary outcome of adjacent to water to between 60 and 80% when being a few hundred either positive or negative experiences as a function of environment metres away (Fig. 3e). For nature environments, there is a decline in features. The regression showed no correlation between age group or probability of positive experiences for the first 300 m, from around 90 gender and experiential outcome, and the model was improved by ex- to around 50% (Fig. 3d). A large effect size was observed for distance to cluding these terms (RS(9) = 7.88, p = 0.55; ΔAIC = –10.3). The major roads, with probabilities of an experience being positive ranging spatial model revealed that all studied environment features, except from below 25% at small distances to above 80% at large distances playgrounds, had significant impacts on experiences (Table 2). The (Fig. 3c). Temperature regulating capacity has a positive impact on autocorrelation term was also highly significant. The best fitting model experiential outcome, with probabilities being around 40% in neigh- with insignificant residual autocorrelation was found by including an bourhoods with “no or little regulating capacity ”, but above 90% in autocorrelation matrix based on prior residuals, where all data points neighbourhoods with “decent regulating capacity” (Fig. 3f). within 200 m were given equal weight. Moran’s I tests revealed inclu- sion of this autocorrelation term to reduce the Moran’s I statistic of residuals from 0.104 (p ≪ 0.001) to 0.017 (p = 0.153).

Table 2 Odds ratios between quintiles for predictors of experiential outcome for the non-spatial and the spatial model.

Non-spatial model Spatial model

Feature Interquintile change Q1 Q2 Q3 Q4 Q5 p-value Q1 Q2 Q3 Q4 Q5 p-value

Residential population Factor 6.96 increase 0.63 0.82 1.07 1.41 1.84 0.003 0.68 0.88 1.13 1.47 1.90 0.005 Working population Factor 8.42 increase 2.28 1.38 0.84 0.51 0.31 ≪0.001 1.70 1.20 0.85 0.60 0.42 0.001 Playgrounds 100 m distance 0.88 0.88 0.88 0.88 0.88 0.024 NS Major roads 100 m distance 2.70 2.17 1.75 1.40 1.13 ≪0.001 2.78 2.15 1.67 1.30 1.00 ≪0.001 Nature environments 100 m distance 0.56 0.72 0.92 1.18 1.51 ≪0.001 0.66 0.79 0.94 1.12 1.34 0.004 Water 100 m distance NS 0.86 0.86 0.86 0.86 0.86 0.003 Natural temperature regulation 0.53 score units 1.64 1.64 1.64 1.64 1.64 0.003 1.72 1.72 1.72 1.72 1.72 0.001 Autocorrelation matrix ≪0.001 AIC 1567.1 1501.1 Residual Moran’s I 0.104 (p ≪ 0.001) 0.017 (p = 0.153) Rao score 62.327 (p ≪ 0.001)

11 K. Samuelsson et al. Landscape and Urban Planning 171 (2018) 7–17

Fig. 3. Relationships between environment features and the probability of an experience being positive. The graphs show predicted probability relationships of every 10th percentile when all other features than the one graphed are at their mean value for the graphed variable’s percentile +/− 5. Thus, these diagrams reflect the spatial distribution of features in Stockholm and cannot be generalised to other contexts. Probabilities are given as percentages on y-axes. The lines represent predicted probabilities and the coloured areas 95% confidence intervals of predicted probabilities. X-axes show number of people, distance to feature or temperature regulating capacity on a 5-step scale (maximum value = 3.63). Distance to residential and working populations and walking distance to major roads and nature environments were fitted with quadratic terms, while remaining features were fitted as linear predictors.

3.4. Impact of environment on frequency and duration of experiences 3.5. Mapping of experiential outcome across Stockholm

We analysed the frequency and duration of experiences in different The probabilities of positive or negative experiences estimated by environments, using the 1083 (61%) of the experiences that contained the spatial regression model were mapped across the study area based data on these attributes. Positive and negative experiences were con- on the study area’s environment features. Fig. 5 shows estimated siderably different (Fig. 4). Many positive experiences either last less probabilities of cells providing positive rather than negative experi- than an hour and occur weekly or last a few hours and occur monthly. ences. Predicted outcomes range between 3% and 99% chance that an In contrast, many negative experiences last only for a moment, but experience is positive. With the mean being about 80%, positive ex- occur several times per week. Taken together with the fact that the periences are more probable than negative in most neighbourhoods. spatial concentration of these negative experiences is highest in the city Neighbourhoods with high probabilities of negative experiences are centre and near major roads, it suggests a connection to people’s daily nearest to major roads, while other features have little impact. The commute or other parts of their work routine. major exception to this pattern is the city centre, which is instead dominated by large working populations with little presence of other features.

Fig. 4. Frequency and duration of positive (left) and negative (right) experiences, respectively. Areas of the circles are proportional to the number of experiences sharing the same frequency and duration. Positive experiences vary more, lasting from only for a moment up to a whole day and occurring from daily to quarterly. By comparison, many negative experiences last only for a moment but occur several times per week.

12 K. Samuelsson et al. Landscape and Urban Planning 171 (2018) 7–17

Fig. 5. Map of predicted probabilities that if an ex- perience was to occur at the location, it is positive rather than negative, as modelled on environment features through spatial regression. Transparency of cells reflect experience density of the surrounding neighbourhood, in order to give an impression of how likely an experience is to occur in that cell. The borders of the study area are 500 m within the bor- ders of Stockholm municipality, and the dashed lines demarcate the inner city.

The highest probabilities for positive experiences are found in areas Mourato, 2013). The importance of nature and water is further in- dominated by adjacency to water and nature environments and high dicated by the fact that respondents were not asked to record any temperature regulating capacity. Furthermore, positive experiences are certain kinds of experiences, but only that they are regularly occurring. more likely in relatively less densely occupied areas than in ones close A surprise was that natural temperature regulation strongly predicts to large residential or working populations. Except for areas close to positive experiences (see Fig. 3). This is surprising because, except for major roads, suburban neighbourhoods have slightly higher prob- direct shading by trees during daytime, the bio-physical processes of abilities of creating positive experiences (in the range of 80–90%) than temperature regulation are often not perceived by people. This finding central neighbourhoods (in the range of 70–80%). However, inner city raises further questions; why is temperature regulation making a dif- neighbourhoods display more variation over short distances (see ference for experiences? Voluminous deciduous trees are effective Fig. 5). temperature regulators (Lehmann, Mathey, Rößler, Bräuer, & Goldberg, 2014) and are weighted heavily in the analysed GIS-layer (Barthel 4. Discussion et al., 2015). Such large trees can also possess aesthetic qualities, which may be a complementary reason for this surprisingly strong correlation In Stockholm, environment features have significant impact on ex- with positive experiences. As our analysis was limited to only positive periential outcome, after accounting for spatial autocorrelation among contra negative experiences, future research might build on the method the data. Number of residents and proximity of nature environments presented here to further explore what the qualities of experiences of and water, all common quality indicators in urban planning and re- nature or water are that ultimately lead to wellbeing. The ability of search, have weak statistically significant effects on people’s experi- PPGIS methods to unveil more detailed qualitative information about ences. Very low rates of positive experiences occur in areas with large experiences has been demonstrated (Kyttä et al., 2013). We foresee that working populations or in proximity to major roads, while areas with coupling such information with affordance-based descriptions of the high natural temperature regulating capacities have very high rates. physical environment will increase understanding of how urban nature These findings suggest that in terms of experiential outcome, urban can support human wellbeing. environments should not be simply thought of as compact or green. The spatial context that urban nature exists in is important for how Such sweeping categorisations overlooks substantial differences ex- it is perceived. Positive experiences are more often located in nature isting within both urban and nature environments. environments accessible by walking. On the other hand, probabilities of positive experiences are higher in remote nature environments. We 4.1. Nature’s impact on experiences could not determine how spatial accessibility of nature impacts what kind of nature experience people have, but we speculate that nature Our analysis aligns with other work to demonstrate the beneficial embedded in built-up areas provides everyday experiences whereas effects of nature and water on urban inhabitants’ experiences. This remote nature can provide more intense although more rare experi- ’ finding is aligned with recent research on restoration from stress (van ences. Nature experiences are important for the formation of people s den Berg et al., 2007), health (Mitchell & Popham, 2008), social relationship with the natural world (Hartig & Kahn, 2016). Giusti, learning (Colding & Barthel, 2013) and happiness (MacKerron & Barthel, and Marcus (2014) found in a study on kindergarten-children

13 K. Samuelsson et al. Landscape and Urban Planning 171 (2018) 7–17 in Stockholm that nature experiences in daily habits contribute to the mediating factor for wellbeing (Kyttä et al., 2016). These two ap- formation of environmental attitudes. Such attitudes are in turn strong proaches are fundamentally related, and both are supported by our predictors of pro-environmental behaviour, much needed for sustain- results. We believe combining accessibility analysis and PPGIS allows a ability transformations in the near future (Abson et al., 2017). Enabling better analysis of how people experience cities than either method in daily spatial access requires that nature is intertwined with the urban isolation. This research demonstrates the ability of this method to fabric. It has been frequently argued that such nature may be more bridge spatial accessibility to environment features and everyday ex- ecologically fragmented (e.g. Stott et al., 2015). However, our analysis periences, but much research remains to be done to develop a fully shows that in Stockholm, walking distances to nature environments integrated framework for interpreting and analysing how urban en- seem to often be unrelated to their level of fragmentation. We suggest vironments influence human wellbeing. For instance, the PPGIS ana- that rather than assuming trade-offs, researchers and designers should lysed here did not survey indifferent or ambiguous experiences of focus on empirically investigating and designing spatial circumstances places. Such experiences are undoubtedly a central part of urban life, under which nature environments are accessible and at the same time and tenable to investigate through PPGIS or similar methods. large and continuous. It is beyond the scope of this study to establish what sizes and shapes of nature areas are preferable for achieving this, 4.4. Diversity of experience opportunities as social-ecological urban design and we speculate that different geographical contexts provide different principle opportunities to do so. Diversity is a central principle of resilience (Biggs, Schlüter, & 4.2. Residential and working populations’ impact on experiences Schoon, 2015). Resilience has in turn been proposed as a guiding me- taphor for moving towards social and ecological sustainability in cities Our results show that residential and working populations have (Ahern, 2011; Pickett, Cadenasso, & Grove, 2004). In ecosystems, re- markedly different impacts on experiential outcome. Based on this silience is reliant on functional and response diversity, meaning that finding, we suggest that no single indicator of density of built en- elements perform different functions, but also respond to change in vironments is adequate to describe how people experience them. Non- different ways, allowing a portfolio of functions to be maintained de- physical factors, such as attitudes towards dense environments, have spite fluctuations and shocks (Elmqvist et al., 2003). Similar thinking previously been highlighted as important for how people relate to can be applied to urban functions (Marcus & Colding, 2014). Resilient density (e.g. Howley, Scott, & Redmond, 2009). However, our results urban systems, in terms of retail-markets, street life and amenities, show that even from a purely spatial perspective, no one neighbour- seem to emerge in neighbourhoods where closeness to large popula- hood typology outperforms all others in producing positive experiences tions and many plots exist in combination (Marcus, 2010). Thus, al- in Stockholm. Our conclusion is that at least residential and workplace though urban scholars and ecosystem scholars apply similar under- density, and potentially other indicators not included here, need to be standings of resilience, the dichotomy of social and ecological remains combined to understand how different kinds of dense environments unresolved in urban design. We suggest that diversity of affordances influence experiences. can be a novel way to overcome this dichotomy of language, addressing Our identification of primarily nonlinear relationships between resilience in cities from an experiential perspective. Transformations of density of the built environment and experiential outcome supports places characterised by many negative experiences can potentially be previous research (Kyttä et al., 2016). Our findings appear to support guided by this design principle, with the goal being accessibility to the contrast between environmental contexts in experiential outcome of affordances of urbanity combined with accessibility to affordances of different densities reported by Kyttä et al. (2016), but rather than ob- significant nature experiences. For example, experiential outcome in serving a strict division between suburbs and inner city, we observe a Stockholm’s inner-city business district might be enhanced if some difference between more negatively perceived neighbourhoods domi- workplaces were converted to apartments, convenient walkways to nated by workplaces and more positively perceived mixed-use neigh- water and nature were emphasised through design elements and car bourhoods. The inner-city business district contains a disproportionate traffic flows were limited, rather than workplaces being further densi- amount of negative experiences. As many of these are brief but occur fied as is the current case. often, we think that they arise from many people sharing a limited In this study, we found little evidence of synergistic effects between space, especially during their work commute, without choosing so. features on experimental outcome, and little evidence of saturation. With few living in the business district, most people working here are This suggests that the impacts of environment features can be analysed likely reliable on public transportation or cars for commuting. Also, this separately, and that there are substantial opportunities to improve is one of few dense areas in Stockholm where walking distance to people’s experiences in the city. Overall, we find this promising for nature environments often exceed 500 m, possibly adding to negative exploring affordances as a guiding principle in urban design, as clever experiential outcomes by limiting possibilities for restoration from spatial designs can potentially improve the diversity of affordances stress. Future research might try to establish whether the deciding within walking distance. However, combining regulating ecosystem factor for experiential outcome in high density work place neighbour- services with other features on the neighbourhood scale can be parti- hoods is crowded commutes, limited access to nature, lack of vibrancy cularly challenging. Neighbourhood averages of temperature regulating outside work hours or other factors. capacity are generally low because this feature is more directly de- pendent on large areas producing it than the others, corroborating 4.3. An integrated method for spatial and experiential analysis previously reported results (Stott et al., 2015). Strategies for better providing regulating ecosystem services in cities need to consider 1) This GIS-study presents a novel spatially and experientially explicit density of built environments, so that few nature areas are transformed, approach that uses the concept of affordances to describe environment 2) morphology of built environments, so that they share long edges features and analyse their impact on people’s reoccurring experiences. with nature areas and many people are close to them, and 3) micro- The methodological innovation is combining accessibility analysis with scale design elements within built environments, such as voluminous an online PPGIS survey. These methods have been applied to study the trees, forest patches and biotope roofs, that aid larger nature areas’ urban environment in relation to wellbeing in different ways. The service production. spatial configuration approach enable understanding how the en- In order for diversity of affordances to be realized as a design vironment interacts with development of attitudes (Marcus et al., 2016) principle, we identify two critical components: 1) Many different en- and how ‘use value’ correspond to proximity (Ståhle, 2008). The PPGIS vironment features need to be considered simultaneously. In this study, approach enable understanding how experiential outcome acts as a seven features were selected, and significant effects found for six of

14 K. Samuelsson et al. Landscape and Urban Planning 171 (2018) 7–17 them. However, many other features not considered here are also im- close by international comparisons, and we hypothesise that the effect portant. 2) How these features are related spatially should be con- of these features will be even more pronounced in environments where sidered in a way that is meaningful from a human perception and they are rarer. On the other hand, although evidence of benefits of cognition perspective. For urban planning to support human wellbeing, nature experiences are generally consistent across the planet, cultural environments must be analysed and shaped with tools that identify such aspects might influence what size an area need to encompass before mechanisms. We chose 500 m as a general distance for analysing fea- being considered nature or from what distance it is considered acces- ture accessibility. This is a coarse representation of how people relate to sible. Lastly, since the 1960s workplaces in Stockholm are very con- these features in reality. For example, we found nature environments to centrated in the landscape and cities with different planning traditions impact experiences up to a distance of about 300 m. Coincidently, the might not display the same strong relationship between workplaces and residuals from the non-spatial regression were autocorrelated up to a negative experiences. 300-m distance, indicating that this is around the distance within which experiential outcome is often more similar. Nevertheless, other features 5. Conclusion might impact experiences on distances greater than 500 m. All lines of work urgently need to contribute with measures that 4.5. Avenues for future research mitigate climate change and halt biodiversity loss. A key role is held by urban scholars as the number of urban inhabitants are prospected to A long history of urbanism studies and a rapidly increasing interest grow with 2.5 billion people by 2050 (United Nations, 2014). However, in PPGIS suggests a wide variety of possible approaches to future work. measures aimed at mitigating these issues will not contribute to real We believe that three major avenues offer great potential for future sustainability if they simultaneously erode the potential of urban en- research: how physical space shapes experiences and ecosystem func- vironments to support wellbeing. The conflicting spatial paradigms tioning, the temporal dynamics of affordances, and new approaches to outlined herein emphasise what Chan et al. (2016) term “the unhelpful public participation in planning. dichotomy of sustaining either human well-being or nature for its own First, research could better address how physical space structures sake”. Our study identifies merits with both narratives in understanding both experiential outcome and ecological functioning. For example, we how experiences of urban environments support or hinder wellbeing. have not introduced network analysis here, despite it being common- However, we also identify qualitative differences both within what is place in urbanity studies to understand pedestrian behaviour (e.g. usually considered urban and nature environments, supporting the Hillier, Penn, Hanson, Grajewski, & Xu, 1993) and urban landscape notion of this dichotomy truly being unhelpful for sustainable urban ecology to understand ecological functioning (e.g. Andersson & Bodin, development. Instead, attention should be focused towards transfor- 2009). More importantly, as ecosystem services obtained by urban mation of urban areas with many negative experiences, as captured by nature environments depend on their scale, situation in larger ecolo- PPGIS studies, with designs integrating closeness to mixed-use urbanity gical networks, and accessibility for people, we envision the coupling of with possibilities of daily nature experiences. Such efforts can create a these factors to increase knowledge about spatial forms of constructed narrative of social-ecological urbanism, accommodating qualities of and nature environments that generate urbanity, provide access to compact as well as social-ecological cities. More widely, our results nature environments and uphold supply of regulating ecosystem ser- demonstrate the usefulness of the concept of affordances in spatial vices. analysis of urban environments, and we provide an innovative metho- Second, how do affordances develop over time? To understand how dological toolbox to apply it. We expect similar studies in other regions the environment influences wellbeing, it is crucial to understand how to encounter both similarities with and differences from our results, relations between people and places change. Place attachment and its increasing knowledge about the complex interplay between environ- emergence has long been central in urban studies (Dempsey, Bramley, ment and wellbeing in different urban settings from various parts of the Power, & Brown, 2009). Comparably, continued stewardship of urban planet. nature often results in social learning and memory-making (Barthel, Folke, & Colding, 2010). As of lately, there has been an increased in- Acknowledgements terest in coupling affordance theory with sense of place scholarship (Raymond, Kyttä, & Stedman, 2017). For example, cultural ecosystem We acknowledge the contribution of Lars Marcus from Chalmers services can be understood through an affordance framework as University of Technology. We are also grateful for the support from emergent and meaningful human-environment relations evolving over Stockholm Resilience Centre. We thank University of Gävle and Formas time (Raymond, Giusti, & Barthel, 2017). We foresee that processes (project ZEUS dnr.nb. 2016-01193) for funding Karl Samuelsson and such as place attachment development or social learning can be em- Stephan Barthel and Formas (project 2011-75) for funding Ann Legeby. pirically investigated with spatially explicit affordance-based methods to identify environments allowing for such processes over time. Appendix A. Supplementary data Third, how can PPGIS studies alter public participation in planning, design and management? In Helsinki, Finland, PPGIS was used in the Supplementary data associated with this article can be found, in the development of the current masterplan, with planners finding it valu- online version, at http://dx.doi.org/10.1016/j.landurbplan.2017.11. able (Kahila-Tani, Broberg, Kyttä, & Tyger, 2015). However, how PPGIS 009. can impact urban social-ecological systems remains an open question. Experiential knowledge is useful for sustainable management and References planning of urban social-ecological systems (Barthel et al., 2010; Colding, Lundberg, & Folke, 2006). PPGIS can potentially improve such Abson, D. J., Fischer, J., Leventon, J., Newig, J., Schomerus, T., Vilsmaier, U., & Lang, D. management and planning, as data is spatially and experientially pre- J. (2017). Leverage points for sustainability transformation. Ambio, 46(1), 30–39. http://dx.doi.org/10.1007/s13280-016-0800-y. cise and knowledge can be captured in great quantities and in- Ahern, J. (2011). From fail-safe to safe-to-fail: Sustainability and resilience in the new corporated much earlier in planning processes than is currently done, at urban world. Landscape and Urban Planning, 100(4), 341–343. http://dx.doi.org/10. least in Sweden. 1016/j.landurbplan.2011.02.021. Altman, I., & Rogoff, B. (1987). World views in psychology: Trait, interactional, organismic, In addition, studies of other cities applying the method presented and transactional perspectives. Handbook of Environmental Psychology7–40. https://doi. here will increase knowledge of how experiences in urban environ- org/citeulike-article-id:4045038. ments relate to geographical, cultural and planning aspects, respec- Andersson, E., & Bodin, Ö. (2009). Practical tool for landscape planning? An empirical tively. For example, in Stockholm nature or water is in most places investigation of network based models of habitat fragmentation. Ecography, 32(June

15 K. Samuelsson et al. Landscape and Urban Planning 171 (2018) 7–17

2008), 123–132. http://dx.doi.org/10.1111/j.1600-0587.2008.05435.x. Househttp://dx.doi.org/10.2307/794509. Andersson, E., Barthel, S., Borgström, S., Colding, J., Elmqvist, T., Folke, C., & Gren, Å. Kahila-Tani, M., Broberg, A., Kyttä, M., & Tyger, T. (2015). Let the citizens map—Public (2014). Reconnecting cities to the biosphere: Stewardship of green infrastructure and participation GIS as a planning support system in the helsinki master plan process. urban ecosystem services. Ambio, 43(4), 445–453. http://dx.doi.org/10.1007/ Planning Practice & Research, 7459(2), 1–20. http://dx.doi.org/10.1080/02697459. s13280-014-0506-y. 2015.1104203. Andersson, E., McPhearson, T., Kremer, P., Gomez-Baggethun, E., Haase, D., Tuvendal, Kennedy, C. A., Stewart, I., Facchini, A., Cersosimo, I., Mele, R., Chen, B., ... Sahin, A. D. M., & Wurster, D. (2014). Scale and context dependence of ecosystem service pro- (2015). Energy and material flows of megacities. Proceedings of the National Academy viding units. Ecosystem Services, 12(2003), 1–8. http://dx.doi.org/10.1016/j.ecoser. of Sciences of the United States of America, 112(19), 5985–5990. http://dx.doi.org/10. 2014.08.001. 1073/pnas.1504315112. Barthel, S., Folke, C., & Colding, J. (2010). Social-ecological memory in urban gardens- Kyttä, M., Broberg, A., Tzoulas, T., & Snabb, K. (2013). Towards contextually sensitive Retaining the capacity for management of ecosystem services. Global Environmental urban densification: Location-based softGIS knowledge revealing perceived re- Change, 20, 255–265. http://dx.doi.org/10.1016/j.gloenvcha.2010.01.001. sidential environmental quality. Landscape and Urban Planning, 113,30–46. http:// Barthel, S., Koffman, A., Bovin, M., Lundqvist, E., Campbell, E., & Tuvendal, M. (2015). dx.doi.org/10.1016/j.landurbplan.2013.01.008. Kartläggning och analys av ekosystemtjänster i Stockholms stad. Stockholm: Calluna AB. Kyttä, M., Broberg, A., Haybatollahi, M., & Schmidt-Thomé, K. (2016). Urban happiness: Bettencourt, L. M. A. (2013). The origins of scaling in cities. Science, 340(6139), Context-sensitive study of the social sustainability of urban settings. Environment and 1438–1441. http://dx.doi.org/10.1126/science.1235823. Planning B: Planning and Design, 43,34–57. http://dx.doi.org/10.1177/ Biggs, R., Schlüter, M., & Schoon, M. L. (Eds.). (2015). Principles for building resilience: 0265813515600121. Sustaining ecosystem services in social-ecological systems. Cambridge: Cambridge Legeby, A., Pont, M. B., & Marcus, L. (2015). Streets for co-presence? Mapping potentials. University Press. In K. Karimi, L. Vaughan, K. Sailer, G. Palaiologou, & T. Bolton (Eds.). Proceedings of Bolund, P., & Hunhammer, S. (1999). Ecosystem services in urban areas. Ecological the 10th international space syntax symposium (pp. 1–18). London: Space Syntax Economics, 29(29), 293–301. http://dx.doi.org/10.1016/S0921-8009(99)00013-0. Laboratory, The Bartlett School of Architecture, University College London. Bren d’Amour, C., Reitsma, F., Baiocchi, G., Barthel, S., Güneralp, B., Erb, K.-H., ... Seto, Legeby, A. (2013a). Configuration and co-presence: The underpinnings of job opportu- K. C. (2016). Future urban land expansion and implications for global croplands. nities. In Y. O. Kim, H. T. Park, & K. W. Seo (Eds.). Proceedings of the 9th international Proceedings of the National Academy of Sciences, 201606036. http://dx.doi.org/10. space syntax symposium. Seoul: Sejong University. 1073/pnas.1606036114. Legeby, A. (2013b). Patterns of co-presence: Spatial configuration and social segregation. KTH Chan, K. M. A., Balvanera, P., Benessaiah, K., Chapman, M., Díaz, S., Gómez-Baggethun, Royal Institute of Technology. Retrieved from http://www.diva-portal.org/smash/ E., ... Turner, N. (2016). Opinion: Why protect nature? Rethinking values and the get/diva2:662753/FULLTEXT01.pdf. environment. Proceedings of the National Academy of Sciences, 113(6), 1462–1465. Lehmann, I., Mathey, J., Rößler, S., Bräuer, A., & Goldberg, V. (2014). Urban vegetation http://dx.doi.org/10.1073/pnas.1525002113. structure types as a methodological approach for identifying ecosystem services – Chemero, A. (2003). An outline of a theory of affordances. Ecological Psychology, 15(2), Application to the analysis of micro-climatic effects. Ecological Indicators, 42,58–72. 181–195. http://dx.doi.org/10.1207/S15326969ECO1502_5. http://dx.doi.org/10.1016/j.ecolind.2014.02.036. Colding, J., & Barthel, S. (2013). The potential of Urban Green Commons in the resilience MacKerron, G., & Mourato, S. (2013). Happiness is greater in natural environments. building of cities. Ecological Economics, 86, 156–166. http://dx.doi.org/10.1016/j. Global Environmental Change, 23(5), 992–1000. http://dx.doi.org/10.1016/j. ecolecon.2012.10.016. gloenvcha.2013.03.010. Colding, J., Lundberg, J., & Folke, C. (2006). Incorporating green-area user groups in Marcus, L., & Colding, J. (2014). Toward an integrated theory of spatial morphology and urban ecosystem management. Ambio, 35(5), 237–244. resilient urban systems. Ecology and Society, 19(4), http://dx.doi.org/10.5751/ES- Coombes, E., Jones, A. P., & Hillsdon, M. (2010). The relationship of physical activity and 06939-190455. overweight to objectively measured green space accessibility and use. Social Science Marcus, L., Giusti, M., & Barthel, S. (2016). Cognitive affordances in sustainable ur- and Medicine, 70(6), 816–822. http://dx.doi.org/10.1016/j.socscimed.2009.11.020. banism: Contributions of space syntax and spatial cognition. Journal of Urban Design, Dempsey, N., Bramley, G., Power, S., & Brown, C. (2009). The social dimension of sus- 4809(June), http://dx.doi.org/10.1080/13574809.2016.1184565. tainable development: Defining urban social sustainability. Sustainable Development, Marcus, L. (2010). Spatial capital: A proposal for an extension of space syntax into a more 19(5), 289–300. http://dx.doi.org/10.1002/sd.417. general urban morphology. The Journal of Space Syntax, 1(1), 30–40. Retrieved from Dempsey, N., Brown, C., & Bramley, G. (2012). The key to sustainable urban development http://joss.bartlett.ucl.ac.uk/index.php/joss/article/viewFile/12/pdf_5. in UK cities? The influence of density on social sustainability. Progress in Planning, Mitchell, R., & Popham, F. (2008). Effect of exposure to natural environment on health 77(3), 89–141. http://dx.doi.org/10.1016/j.progress.2012.01.001. inequalities: An observational population study. Lancet, 372(9650), 1655–1660. Elmqvist, T., Folke, C., Nyström, M., Peterson, G., Bengtsson, J., Walker, B., & Norberg, J. http://dx.doi.org/10.1016/S0140-6736(08)61689-X. (2003). Response diversity, ecosystem change, and resilience. Frontiers in Ecology and Newman, P. (2006). The environmental impact of cities. Environment and Urbanization, the Environment, 1(9), 488–494. http://dx.doi.org/10.1890/1540-9295(2003) 18(2), 275–295. http://dx.doi.org/10.1177/0956247806069599. 001[0488:RDECAR]2.0.CO;2. Passchier-Vermeer, W., & Passchier, W. F. (2000). Noise exposure and public health. Frank, L., Kerr, J., Chapman, J., & Sallis, J. (2007). Urban form relationships with walk Environmental Health Perspectives, 108(6 Suppl), 123–131. http://dx.doi.org/10. trip frequency and distance among youth. American Journal of Health Promotion, 21(4 2307/3454637. Suppl), 305–311. http://dx.doi.org/10.4278/0890-1171-21.4S.305. Pickett, S. T. A., Cadenasso, M. L., & Grove, J. M. (2004). Resilient cities: Meaning, Gascon, M., Mas, M. T., Martínez, D., Dadvand, P., Forns, J., Plasència, A., & models, and metaphor for integrating the ecological, socio-economic, and planning Nieuwenhuijsen, M. J. (2015). Mental health benefits of long-term exposure to re- realms. Landscape and Urban Planning, 69(4), 369–384. http://dx.doi.org/10.1016/j. sidential green and blue spaces: A systematic review. International Journal of landurbplan.2003.10.035. Environmental Research and Public Health, 12(4), 4354–4379. http://dx.doi.org/10. R Core Team (2016). R: A language and environment for statistical computing. Vienna, 3390/ijerph120404354. Austria: R Foundation for Statistical Computinghttp://dx.doi.org/10.1007/978-3- Gibson, J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin. 540-74686-7. Gidlöf-Gunnarsson, A., & Öhrström, E. (2007). Noise and well-being in urban residential Raymond, C. M., Giusti, M., & Barthel, S. (2017). An embodied perspective on the co- environments: The potential role of perceived availability to nearby green areas. production of cultural ecosystem services: Toward embodied ecosystems. Journal of Landscape and Urban Planning, 83(2–3), 115–126. http://dx.doi.org/10.1016/j. Environmental Planning and Management, 568(June), 1–9. http://dx.doi.org/10.1016/ landurbplan.2007.03.003. j.aqpro.2013.07.003. Giusti, M., Barthel, S., & Marcus, L. (2014). Nature routines and affinity with the bio- Raymond, C. M., Kyttä, M., & Stedman, R. (2017). Sense of place, fast and slow: The sphere: A case study of preschool children in stockholm. Children, Youth and potential contributions of affordance theory to sense of place. Frontiers in Psychology, Environments, 24(3), 16–42. http://dx.doi.org/10.7721/chilyoutenvi.24.3.0016. 8(September), 1674. http://dx.doi.org/10.3389/fpsyg.2017.01674. Giusti, M., Barthel, S., Samuelsson, K., Stockholm University, & Stockholm Resilience Soga, M., & Gaston, K. J. (2016). Extinction of experience: The loss of human-nature Centre (2017). Where is your Stockholm? A Public Participatory GIS study to unfold interactions. Frontiers in Ecology and the Environment, 14(2), 94–101. http://dx.doi. positive and negative experiences in the landscape of Stockholm. Swedish National Data org/10.1002/fee.1225. Servicehttp://dx.doi.org/10.5878/002916. Soga, M., Yamaura, Y., Koike, S., & Gaston, K. J. (2014). Land sharing vs. land sparing: Grimm, N. B., Faeth, S. H., Golubiewski, N. E., Redman, C. L., Wu, J., Bai, X., & Briggs, J. Does the compact city reconcile urban development and biodiversity conservation? M. (2008). Global change and the ecology of cities. Science, 319(5864), 756–760. Journal of Applied Ecology, 51(5), 1378–1386. http://dx.doi.org/10.1111/1365-2664. http://dx.doi.org/10.1126/science.1150195. 12280. Hartig, T., & Kahn, P. H. J. (2016). Living in cities, naturally. Science, 352(6288), Ståhle, A. (2008). Compact sprawl: Exploring public open space and contradictions in urban 938–940. http://dx.doi.org/10.1126/science.aaf3759. density. Royal Institute of Technology. Retrieved from http://www.diva-portal.org/ Hillier, B., & Hanson, J. (1984). The social logic of space. Cambridge: Cambridge University smash/get/diva2:37326/FULLTEXT01.pdf. Press. Ståhle, A. (2012). Place syntax tool (PST). In A. Hull, C. Silva, & L. Bertolini (Eds.). Hillier, B., Penn, A., Hanson, J., Grajewski, T., & Xu, J. (1993). Natural movement: Or, Accessibility instruments for planning practice (pp. 173–178). COST Office. configuration and attraction in urban pedestrian movement. Environment and Stockholms Parkprogram (2006). Stockholm. Retrieved from www.stockholm.se/ Planning B: Planning and Design, 20(1), 29–66. http://dx.doi.org/10.1068/b200029. PageFiles/64633/Parkprogr_kap1-5.pdf. Howley, P., Scott, M., & Redmond, D. (2009). Sustainability versus liveability: An ex- Stott, I., Soga, M., Inger, R., & Gaston, K. J. (2015). Land sparing is crucial for urban amination of neighbourhood satisfaction. Journal of Environmental Planning and ecosystem services. Frontiers in Ecology and the Environment, 13(7), 387–393. http:// Management, 52(6), 847–864. http://dx.doi.org/10.1080/09640560903083798. dx.doi.org/10.1890/140286. Jabareen, Y. R. (2006). Sustainable urban forms: Their typologies, models, and concepts. Tratalos, J., Fuller, R. A., Warren, P. H., Davies, R. G., & Gaston, K. J. (2007). Urban form, Journal of Planning Education and Research, 26(1), 38–52. http://dx.doi.org/10.1177/ biodiversity potential and ecosystem services. Landscape and Urban Planning, 83, 0739456X05285119. 308–317. http://dx.doi.org/10.1016/j.landurbplan.2007.05.003. Jacobs, J. (1961). The death and life of great american cities. New York: Random United Nations (2014). World urbanization prospects: The 2014 revision, highlights (ST/ESA/

16 K. Samuelsson et al. Landscape and Urban Planning 171 (2018) 7–17

SER.A/352). http://dx.doi.org/10.4054/DemRes.2005.12.9. Promotion, 21(4 Suppl), 312–316. http://dx.doi.org/10.4278/0890-1171-21.4s.312. van den Berg, A. E., Hartig, T., & Staats, H. (2007). Preference for nature in urbanized Withagen, R., de Poel, H. J., Araújo, D., & Pepping, G.-J. (2012). Affordances can invite societies: Stress, restoration, and the pursuit of sustainability. Journal of Social Issues, behavior: Reconsidering the relationship between affordances and agency. New Ideas 63(1), 79–96. http://dx.doi.org/10.1111/j.1540-4560.2007.00497.x. in Psychology, 30(2), 250–258. http://dx.doi.org/10.1016/j.newideapsych.2011.12. de Vries, S. I., Bakker, I., & van Mechelen, W. (2007). Determinants of activity-friendly 003. neighborhoods for children: Results from the SPACE study. American Journal of Health

17