A place‐based participatory mapping approach for assessing cultural ecosystem services in urban green space

1School of Biological Sciences, University of Southampton, Southampton, UK; 2School of Biological Sciences, Royal Holloway, University of London, Egham, UK; 3Environment Agency, Canal Walk, Romsey, UK; 4Department of Plant Sciences, University of California, Davis, CA, USA; 5United Nations Environment Programme World Conservation Monitoring Centre, Cambridge, UK; 6School of Geography and Environmental Science, University of Southampton, Southampton, UK and 7Conservation Science Group, Department of Zoology, University of Cambridge, Cambridge, UK


| INTRODUC TI ON
Green spaces are crucial features in urban environments, providing local communities access to a variety of benefits from ecosystem services, derived from natural processes (Dickinson & Hobbs, 2017;Maller et al., 2009;MEA, 2005). While research into the use, conservation and management of ecosystem services has proliferated in the last 20 years and much of the focus has been on more easily quantified services such as provisioning (e.g. food, raw materials) and regulating services (e.g. air and water quality, pollination; Bennett, Peterson, & Gordon, 2009;Costanza et al., 1997;Nelson et al., 2009), whereas cultural ecosystem services have received less attention (Chan, Satterfield, & Goldstein, 2012;Queiroz et al., 2017;Tew, Simmons, & Sutherland, 2019). Cultural Ecosystem Services (CES) can be defined as the benefits arising from a mutualistic relationship of culture, heritage and the environment, including cultural identity and heritage, spirituality and religion, inspiration and aesthetic appreciation as well as more marketable recreation and tourism benefits (MEA, 2005).
While urban green spaces provide a rich variety of biological and anthropogenic ecosystem services, research has indicated that access to green space has significant positive effects on social cohesion, cultural memory, health and well-being benefits associated with the residents' lived experience in the urban areas (Cox, Shanahan, Hudson, Fuller, & Gaston, 2018;Hartig et al., 2011;Pereira et al., 2013;. The development of connections to nature in an age of rapid urbanisation is increasingly important for supporting local peoples' physical and mental well-being and enhancing quality of life (Cox et al., 2018;Maas, Verheij, Groenewegen, Vries, & Spreeuwenberg, 2006;Shanahan et al., 2017). However, evidence suggests that connectedness to nature is diminishing, especially in urbanizing regions, leading to an ongoing 'extinction of experience' of the natural world and the cultural benefits that accompany it (Soga, Gaston, Koyanagi, Kurisu, & Hanaki, 2016;Soga, Gaston, Yamaura, Kurisu, & Hanaki, 2016). The so-called 'nature-deficit disorder' has been found to incur negative effects on human health, well-being and overall happiness, with indirect impacts on conservation interest (Alfonso, Zorondo-Rodríguez, & Simonetti, 2016;Louv, 2005). The links between connectedness to nature, well-being and conservation exemplify the often unnoticed importance of urban green spaces in our everyday lives and the need for quantifying CES provided by green space for better management of urban land (Raymond, Giusti, & Barthel, 2018).
Management efforts to maintain and improve urban green spaces are often carried out to enhance the visitors' experience, with increasing interest in the effect on ecosystem service provision for the benefit of local communities. Rapid assessment of ecosystem services is required to make efficient, balanced and effective management decisions at a site-scale, especially under economic and time pressures. However, CES is particularly challenging to measure using traditional techniques as it departs from the original linear ecosystem service framework (Chan et al., 2011).
While market-based valuations are insufficient for capturing the full range of social values, qualitative methods also often lack robust statistical analysis or a spatial element (Milcu, Hanspach, Abson, & Fischer, 2013). To solve the problem of quantifying and mapping CES on a fine scale, there has been significant interest in creating novel methods. This includes the use of social media photographs with geo-tagging (Guerrero, Møller, Olafsson, & Snizek, 2016;Oteros-Rozas, Martin-Lopez, Fagerholm, Bieling, & Plieninger, 2018;Richards & Friess, 2015), and participatory methods such as participatory geographic information systems (PPGIS) techniques (Rall, Hansen, & Pauleit, 2019;Tew et al., 2019;Viirret, Raatikainen, Fagerholm, Käyhkö, & Vihervaara, 2019), as well as and experiential methods, designed to collect the personal intensity of CES experience on a spatial plane (Teff-Seker & Orenstein, 2018). A move towards spatial, participatory techniques for practical CES assessment framework will allow land managers to match cultural values to landscape characteristics and assess how they are impacted by different management options.
Our study aims to explore management options to control the levels of the River Itchen at its tidal confluence within Riverside Park in Southampton, UK The flow of the Itchen into the estuary is currently controlled by a sluice gate which has reached a critical condition and is no longer economically viable to maintain. Working alongside the Environment Agency, we aimed to provide a rapid comparative assessment of the CES for Southampton's Riverside Park, comparing the current state and a proposed alternative management option. The proposed alternative management option centres around the decommissioning of the sluice gate and conservation of the significant local cultural, economic and environmental importance of Riverside Park and the River Itchen as a chalk stream habitat. To achieve this, we adopted the Toolkit for Ecosystem Service Site-based Assessment (TESSA) for rapid assessment of cultural ecosystem service provision by the park in its current state and in its most plausible alternative state under a different management option (Peh et al., 2013(Peh et al., , 2017. TESSA was selected over alternative tools as a low-cost written step-by-step approach with a dedicated section focussed on CES assessment which was specificially designed for comparing the impact of contrasting scenarios at the site scale (Neugarten et al., 2018). Following several previous CES studies (Fagerholm, Käyhkö, Ndumbaro, & Khamis, 2012;Fish et al., 2016;Klain & Chan, 2012 complements the language, values and knowledge of the local population to gain perspective and context of their perceptions of both states. Participation of local residents was essential given the social importance of the park to them, fulfilling the previous recommendations for iterative frameworks for participatory engagement Dickinson & Hobbs, 2017).

| Study area
Riverside Park (hereafter 'the park') comprises a 32-hectare public green space situated along the River Itchen, found within the greater Bitterne Park area in Southampton, Hampshire (50.933853, −1.374796), UK. The park contains diverse public facilities (playing fields, children playground, carpark, etc.), and is managed by Southampton City Council. As a chalk stream, the River Itchen is designated as a Special Area of Conservation (SCA) and a Site of Special Scientific Interest (SSSI), recognized as an important freshwater habitat for many rare species. The River Itchen is of great cultural importance owing to its aesthetic quality with 'gin-clear waters' (Vallings, 1885), diverse wildlife and iconic local heritage accessible via the Itchen Navigation Trail. The Trail follows the Itchen for 17 km from the headwaters through its termination in Riverside Park. Angling rights upon the approximately 1 km section of the River Itchen are managed by the UK Environment Agency (EA), a governmental organization that is responsible for conservation and the environment. The River Itchen is amongst the world's most famous dry fly-fishing rivers, a sport which is thought to have originated in Hampshire's chalk streams, thanks to the clean, clear waters and abundant with wild trout and salmon, adding significant heritage and tourism value (Greene, 1936).
The imminent failure of a major sluice gate at the confluence of the Itchen Navigation and Solent Estuary threatens the park and the river ecosystem (Environment Agency, 2018). Should the gate fail, the lower Itchen would become tidal, resulting in a 2-m drop in river levels at low tide. Local ecology would transition from a biodiverse freshwater system to intertidal marine silt beds and the current and historical views of the Itchen would be significantly transformed. As the sluice gate is expensive to replace and maintain, there is debate as to whether the structure should be decommissioned. One option would involve replacing the old sluice gate completely; and another would involve blocking off the existing sluice gate and creating a new white-water canoe course with a dedicated salmon pass incorporated in the design. Potential improvements at the park are being co-ordinated by the EA and Groundwork South, a charity organization working towards sustainable communities. Sluice gate replacement to maintain the current state of the river and park would incur significant financial cost, hence this study compares the CES provided by current state of the park with another plausible and cost-effective 'alternative state'. In this alternative scenario, the current sluice gate is permanently closed and approximately 90% of the current river flow is diverted through a white-water canoeing facility through the grounds of a sports complex, maintaining current river levels and aiding salmon passage. Additionally, the old course of the Itchen will be unblocked; and the currently culverted section would be 'daylighted' to create a new stream into the estuary with improved paths and landscaping options (see Appendices S1 and S2 for maps/images).

| Cultural ecosystem service assessment
The key principles behind our data collection primarily follow TESSA protocols for CES assessment and build on previous participatory mapping techniques (Ingwall-King et al., 2017). A summary of the TESSA rapid assessment process was adapted to suit the study aims specifically ( Figure 1). We employed a public participation geographic information system (PPGIS) approach, using an inductive free-listing questionnaire in which participants were encouraged to map cultural benefits for both the current state and an envisioned alternative state onto ordinance survey (OS) maps for later spatial analysis

| Online questionnaire survey
We carried out an online questionnaire survey from May to July 2017 to gain a wider insight of the cultural benefits provided by the current state of the park; and to supplement the findings of the experimental freelisting exercise. The online questionnaire was created on Southampton University's iSurvey platform, consisting of multiple choice and openended questions. The questionnaire was designed to extract information on the users' visitation frequency, favourite features, primary use and choice between the current and alternative states. Groundwork South and the EA used social media such as Facebook and Twitter to advertise this online survey which could be accessed via their website. The attendants of the River Itchen consultation events could also take part in the survey using small electronic tablets provided by Groundwork South.

| Scoping and mapping exercises
We conducted a preliminary scoping exercise by piloting a free-listing method with nine residents who attended a local stakeholder meeting on 22 March 2017. We first introduced the participants to the concept of ecosystem services in layman's terms. Then we gave each participant a free-listing form and an up-to-date topographical OS map with land-use, public access and leisure points of interest highlighted for spatial context. We encouraged them -by referring to the OS map -to consider which benefits they 'appreciated' about the park in its current state. The benefits under consideration should include both tangible CES (such as recreation and heritage) and intangible services (such as spiritual and aesthetic value), as well as the practices they engaged in (such as angling, water sports and walking).
Free-listing and mapping exercises were further piloted using a second group of stakeholders at the University of Southampton. A total of nine staff and students took part in this trial, enabling the evaluation of methods for feasibility, improvement of free-listing F I G U R E 1 Flow diagram of the rapid cultural ecosystem service assessment process adapted from the Toolkit for Ecosystem Service Site-based Assessment toolkit (Peh et al., 2013), breaking down the five key steps into main questions/objectives and the specific steps taken in this study questionnaire and further refinement of the primer and discussion style. Neutral, unbiased language was key for the explanation of the data collection process and the alternative state of the park so as to achieve objectivity among the respondents.
Both preliminary free-listing exercises used local terminology, vocabulary and historical context for the discussion of the park's geography, primary land use and culture. This was helpful for maintaining a focus on the park for this site-based study (Johansson, Pedersen, & Weisner, 2019). We used the data derived from these exercises to classify the park into key zones based on the land cover and usage ( Figure 2). The use of zones within the park allowed us to identify pattern of intensity, richness and diversity in terms of CES appreciation and provision, and map CES 'hotspots' and 'coldspots' (areas of relative high/ 'hot' and low/ 'cold' CES provision) within the park.

| Free-listing exercises
We performed the full free-listing exercises at the River Itchen To ensure consistency, we asked all participants to think of themselves as an individual as well as a part of the wider community and individually list all aspects, uses and areas of the park that they currently appreciate in their own words; and then mark their entries on an OS map. Continuous routes generated by activities such as walking, jogging and cycling could not have a pinpoint location and they were marked by the respondents' preferred points along that route as mentioned in their free-listing.
To introduce the participants to the proposed alternative state, we used artistic impressions and map overlays together with oral explanation. Given this information, we asked the stakeholders to envision their use and appreciation of the counterfactual and carry out another free-listing and mapping exercise (see Appendix S3). We encouraged the participants to consider the park as a whole under the alternative state to prevent erroneous focus towards the new structures and changes. Finally, to check that our sample was representative of the population of the area, we compared the age profile and gender ratio of our respondents with the demographic information of Southampton City and three wards local to Riverside park obtained from the National 2011 census (https ://www.ons.gov.uk/ censu s/2011c ensus ).

| Data analysis
We categorized each respondent into nine stakeholder groups (angler, cyclist, dog walker, family/community use, green-space, nature, swimmer, walker and water sports) based on participants' primary use of the park. We considered each free-listing entry of each person as a response. Each response was transcribed, assigned to its associated zone of the park and encoded to its associated cultural benefit type as in Table 1. For both current and alternative states of the park, we counted the number of responses for each zone of the park, group of stakeholders and type of cultural benefits. We used Mann-Whitney tests to compare the number of responses between male and female users for the current state of the park; and make same-gender groups comparison between the two states of the park. Furthermore, we used Kruskal-Wallace tests to compare the number of responses across different age groups (18-30, 31-50, 51-65 and 65+) for both states of the park separately. We also used Chi-square test to determine if there was an association between an age group and a particular state of the park.
As an inductive survey, participants were encouraged to list all the CES 'appreciations' in their own words. Sentiment analysis was used to examine the qualitative response terms, aggregated at the synonym level, associated with the current state, as well as the alternative state of the park. We transcribed and uploaded the qualitative free-listing comments into NVivo Pro (version 11) which enabled us to analyse each transcription through the creation of nodes according to key concept, in this case a service type or zone.
By combining the transcripts for all participants for the current F I G U R E 2 Geographic information systems plot of Riverside Park study area indicating zones named according to land use as determined by local stakeholders in the first pilot study and alternative states, we could directly compare word frequency and sentiment for each state. Key words, stemmed words and synonyms were examined through word frequency counts, and complex word-trees divulged the context in which each word was used. The data from both states were compared for noteworthy differences in the language used, while giving crucial context to the corresponding spatial data. Most significantly, NVivo enabled quantitative analysis of text sentiment using an inbuilt scoring system. Each word containing sentiment has a predetermined sentimental score based on a Likert-style scale from 'very negative' to 'very positive' which depends on word placement and modifiers within the surrounding sentence. The creation of word clouds -a visual means of mind and concept mapping with viral structuresalso enabled qualitative inspection of the data for key differences between the two opposing states. We also read and considered holistically the responses to obtain the integrated perceptions of the contrasting states. As all responses were written in the respondents' own words, it was important to consider the wider context of their perception from a less systematic perspective.
Finally, we employed RStudio (version 1.0.143, specifically packages 'ggplot2' and 'reshape2'; Wickham, 2007Wickham, , 2009) to create confusion matrices of CES versus zone to illustrate: (a) the difference between the current and alternative states of the park in terms of the normalized response number (expressed in percentage) across different CES for each zone; (b) the difference between the two states of the park in terms of normalized response number (percentage) across different zones for each CES; and (c) the degree of positive sentiment (expressed in a ratio of positive sentiment score to negative sentiment score calculated within NVivo) of each CES in each zone for each state of the park. We used Google Maps to obtain the GPS coordinates of every response mapped onto the OS maps. We then created heat maps using these co-ordinate data to visualize response density.

| Online questionnaire survey
A total of 587 respondents completed our online questionnaire survey. Nearly 50% of respondents visited the park at least once weekly, and over 96% at least once monthly. The top features of the park favoured by the users are River Itchen (93%), the open space (52%) and the flora and fauna (40%); while the activities the users mostly engaged are walking (59%), water sports (48%), relaxation/unwinding (36%), sitting by the river (35%) and wildlife spotting (33%). Finally, the alternative state of the park was favoured as the primary choice for 63% of respondents compared to the current state (Table 2).

| Free-listing exercises
In total, 51 participants gave 273 free-listing responses for the current state and 233 responses for the alternative state (Table 3), each mapped onto corresponding OS maps. Respondent age ranged between 18 and 76 years old (Table 3)

| Qualitative free-listed text analyses
It is imperative to also investigate the free-listing responses themselves, as sentiment alone is difficult to analyse systematically without the context in which specific terms were specified. Word

| Mapping cultural appreciation and sentiment hotspots
Zones 1 (Woodmill and salmon pool), 2 (Upper Itchen path), 9 (Lower path) and 10 (Playing field) were identified as cultural appreciation hotspots of the park, with the highest number of mapped CES responses (Table 4). However, under the alternative state, zones 9 (Lower path -11.6% decrease) and 10 (Playing fields -13.4% decrease) saw the most dramatic declines in cultural service responses; even though these areas were not affected by proposed management plans (Table 4). Zones 2 (Upper Itchen path), 3 (pitch and putt grass), key areas of proposed renovation also declined by 7.6% and 5.1% respectively. By contrast, zone 1 (Woodmill and salmon pool) remained a hotspot, increasing number of responses by 13.3%, while 4 (Old stream) and 7 (Culvert/path) also emerged as hotspots for cultural importance, as numbers of CES responses increased by 9.6 and 14.0% respectively (Table 5).
Using NVivo, sentiment analysis revealed a decrease (an average decline of 14.3%) in positive sentiment from the current to the alternative state (Table 6)  Note: Using the NVivo built-in sentiment scoring system, red highlighted text denotes negative sentiment and green highlighted text denotes positive sentiment in the context of the sentence modifiers and overall structure.  groups: written step-by-step tools (e.g. TESSA, PA-BAT) and computer-based modelling (e.g. ARIES, InVEST). TESSA was chosen for this study as it is a free and publicly available tool, enabling a wide range of users to investigate the provision of local ES without formal training in GIS or modelling techniques. The toolkit includes a suite of eight biological and anthropocentric ecosystem services (including CES) allowing multidisciplinary site-based analysis using one cohesive framework (Peh et al., 2017). However, the use of any single tool may dictate the perspective of the study and the definitions by which ecosystem services are valued. The PPGIS approach driven by the TESSA framework enabled the capture of individual perceptions and values; as participants were able to convey their personal connection to the natural environment in their own words, a method that could apply to any language (Menzel & Teng, 2010;Rall et al., 2019). This method is especially significant in the context of spatially specific co-production of CES which was described as a 'co-dominating triangular relationship' of interlinking preference, virtue and principal-based methods, highlighting the need for place-based methods (Chan et al., 2011;Fish et al., 2016). However, while inductive and participatory methods in- As human co-produced values, CES assessment should always consider the 'mind of the observer' as well as their interactions with the biophysical environment Ives et al., 2017;Scholte, Teeffelen, & Verburg, 2015). However, while this could be remedied using predefined label methods such as stated preference, this undermines the purpose of an inductive approach, exposing revealed preference as a more informative measure of respondent perception (Dallimer et al., 2014). Future study should therefore seek to explore and analyse locally specific definitions and perceptions of CES and the links to local greenspace value.

| D ISCUSS I ON
Understanding historical context, predisposed opinion and communication barriers as well as the key factors motivating personal needs, beliefs and perceptions of the local environment would help to contextualise the beliefs of every respondent and understand the spatially explicit demands for CES (Chenail, 2011;Creswell & Miller, 2000).

| CON CLUS IONS
In a rapidly urbanizing world, urban green spaces are critical for the provision of multiple ecosystem services on a global scale (Green et al., 2016). The provision of CES plays an especially important role in the promotion of community cohesion, health and well-being. Yet, CES are often overlooked in landscape management assessments due to their perceived intangibility (Scholte et al., 2015). Through this study, we used a variety of participatory, inductive, spatial and statistical methods to provide guidance as to the best course of action for future development of an important urban green space in Southampton, UK. Analysis of CES provision in the park's current state provided a critical baseline, outlining the present value of specific areas of the park in terms of CES and disclosed personal perceptions of key, local stakeholders. Comparison with results regarding the proposed, alternative state revealed spatially specific insights into areas of the park where conflict between multiple users with competing demand for non-substitutable services reside. In this case, cultural benefits that are inevitably affected by the development could potentially be recreated in areas that are considered as cultural 'coldspots'. However, in the wider context, this study provides further evidence that the combined use of participatory methods and spatial mapping of CES can provide land managers and decision-makers with the information required to find an optimal compromise between stakeholder groups, so as to maximize CES, health and well-being benefits for the local community. The future use of the CES assessment methods outlined in this study, in conjunction with the assessment of other types of ecosystem services can encompass multiple biocentric and economic aspects of ecosystem service assessment, providing an effective and holistic approach to landscape management.

ACK N OWLED G EM ENTS
This work was funded by the University of Southampton, and the Environment Agency. The authors acknowledge and thank the Environment Agency for its support of this work, both financial an in-kind. The work is that of the authors and is independent from the Agency: it does not represent the Agency's policies or opin-

ions. KSHP was supported by Southampton University's Institute for Life Sciences Fellowship and JB by Vitacress Conservation
Trust. We also thank Groundwork South for their support and consultation.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R S ' CO NTR I B UTI O N S
K.S.-H.P., J.L.S. and L.J. conceived the research. All authors designed the study. L.J. collected the data and completed the analysis, L.J.
wrote the paper. All authors commented on the analysis and presentation of data, and revised the earlier drafts.

DATA AVA I L A B I L I T Y S TAT E M E N T
Due to the personal nature of the demographic information and free-listing statements collected for this study are fully anonymized.