Evaluating natural capital performance of urban development through system dynamics: A case study from London

Natural capital plays a central role in urban functioning, reducing flooding, mitigating urban heat island effects, reducing air pollution, and improving urban biodiversity through provision of habitat space. There is also evidence on the role played by blue and green space in improving physical and mental health, reducing the burden on the health care service. Yet from an urban planning and development view, natural capital may be considered a nice to have, but not essential element of urban design; taking up valuable space which could otherwise be used for traditional built environment uses. While urban natural capital is largely recognised as a positive element, its benefits are difficult to measure both in space and time, making its inclusion in urban (re)development difficult to justify. Here, using a London case study and information provided by key stakeholders, we present a system dynamics (SD) modelling framework to assess the natural capital performance of development and aid design evaluation. A headline indicator: Natural Space Performance, is used to evaluate the capacity of natural space to provide ecosystem services, providing a semi-quantitative measure of system wide impacts of change within a combined natural, built and social system. We demonstrate the capacity of the model to explore how combined or individual changes in development design can affect natural capital and the provision of ecosystem services, for example, biodiversity or flood risk. By evaluating natural capital and ecosystem services over time, greater justification for their inclusion in planning and development can be derived, providing support for increased blue and green space within cities, improving urban sustainability and enhancing quality of life. Furthermore, the application of a SD approach captures key interactions between variables over time, showing system evolution while highlighting intervention opportunities.

largely recognised as a positive element, its benefits are difficult to measure both in space and time, making its inclusion in urban (re)development difficult to justify. Here, using a London case study and information provided by key stakeholders, we present a system dynamics (SD) modelling framework to assess the natural capital performance of development and aid design evaluation. A headline indicator: Natural Space Performance, is used to evaluate the capacity of natural space to provide ecosystem services, providing a semi-quantitative measure of system wide impacts of change within a combined natural, built and social system. We demonstrate the capacity of the model to explore how combined or individual changes in development design can affect natural capital and the provision of ecosystem services, for example, biodiversity or flood risk. By evaluating natural capital and ecosystem services over time, greater justification for their inclusion in planning and development can be derived, providing support for increased blue and green space within cities, improving urban sustainability and enhancing quality of life. Furthermore, the application of a SD approach captures key interactions between variables over time, showing system evolution while highlighting intervention opportunities.
Keywords: Natural space performance, Ecosystem Services, System Dynamics, Natural

Capital, London INTRODUCTION
Natural Capital (NC); the stocks of renewable and non-renewable natural resources that benefit people both directly and indirectly, and the flow of ecosystem services (ESS) these provide, have been increasingly recognised for their central role in sustaining economic and social wellbeing, societal resilience and sustainable development (Bateman and Mace 2020; Guerry et al. 2015). As a result, NC has been incorporated into government policy processes (Department for Environment Food and Rural Affairs 2020) and private sector decision making (Natural Capital Coalition, 2016;La Notte et al. 2017;Seppelt et al. 2011). For instance, it has been shown that NC and the flow of ESS significantly contribute to global and national gross domestic product (GDP) (Bradbury et al. 2021;Costanza et al. 2014) and underpin human wellbeing (Dasgupta 2021). ESS are also produced by ecological structures within urban areas (Gutman, 2007;Jansson, 2013;McGranahan, et al., 2005), and with over half of the Earth's population now living in cities, and with rates of urbanisation increasing globally (United Nations 2018), it is increasingly important to understand urban ESS; those that are either directly produced by ecological structures within urban or peri-urban areas, and the human-environment systems they depend upon (Bettencourt and West 2010; within the built and natural environment system. In addition, ESS rarely conform to property, administrative or sectorial boundaries, leading to difficulties in management and regulation. However, Green Infrastructure (GI), the strategically planned network of natural or semi natural areas (European Commission. 2019) and Nature Based Solutions (NBS), approaches to help address societal challenges that involve working with and enhancing nature (Cohen-Sacham et al. 2016; European Commission (EC) 2015; Seddon et al. 2020), are increasingly seen as innovative solutions to transform NC into a source of green growth and sustainable development (Gómez Martín et al. 2020). While there is increasing support for GI and NBS in policy and planning documents, their actual implementation has been slow and examples of high quality NBS is the exception rather than the norm (see Fisher et al. (2020), Jerome et al. (2019) and Matthews et al., (2015)). There have been a number of reasons why this is the case including conflation of GI and NBS with traditional green spaces; the continued siloed approach to policy issues where NBS could play a role; a devaluing of NBS in local planning processes; a lack of consideration of long-term stewardship of GI and uncertainty as to what makes GI successful (Fisher et al., 2020). Yet, GI and NBS (used interchangeably in this paper) present a valid alternative to grey infrastructure for coping with climate-related risks in urban and rural areas alike (Calliari, Staccione, and Mysiak 2019;Frantzeskaki 2019;Giordano et al. 2019;Raymond et al. 2017). National, regional and local priorities in areas such as housing and highways often override environmental concerns (Pluchinotta, Salvia, and Zimmermann 2021). Therefore, making explicit the long term societal and environmental contribution of NC is crucial to its implementation in urban design.
J o u r n a l P r e -p r o o f Journal Pre-proof A comprehensive planning approach has the potential to harmonize human-environment interactions and mitigate the harmful impacts of urbanisation (Puchol-Salort et al. 2021).
Such an approach requires planners and local decision makers to understand and value nature's multiple contributions to the quality of urban life, through a whole systems approach. A recent study of decision makers in local authorities found a need for simple tools that allow incorporation of the importance of nature and green infrastructure into development projects (Pluchinotta, Salvia, et al. 2021). Achieving this presents significant challenges, largely due to framework limitations and a lack of suitable headline indicators for assessing NC performance and measuring trade-offs between multiple interacting elements (Bateman and Mace 2020). Emerging techniques that use outcome-based metrics and incremental management to progressively enhance ecosystem condition, and incorporate diverse stakeholders requirements and opinions across scales, sectors and knowledge systems, show promise, but are under-developed at present (Bateman and Mace 2020).
A number of modelling platforms have been created to evaluate blue-green infrastructure and the outcomes of development on urban NC and ESS. These include the Natural Capital Planning Tool (NCPT) (Holzinger et al., 2019), InVEST (Hamel et al. 2021) and the Benefits Estimation tool (B£ST) (Horton et al. 2019). NCPT allows the impacts of new or proposed developments on NC or ESS to be assessed, with outputs described through a development impact score. InVEST is suite of modelling tools designed to support city planning by providing information on natural infrastructure and the services it provides to people, complementing more comprehensive planning tools that capture built infrastructure and socioeconomic dimensions (Hamel et al. 2021). B£ST was created to help assess and monetise the J o u r n a l P r e -p r o o f Journal Pre-proof financial, social and environmental benefits of blue-green infrastructure, with a focus on sustainable urban drainage and natural flood management. Such models are extremely useful tools and help address critical gaps in evaluating NC and ESS, in some cases providing outputs which can be further utilised in other modelling frameworks. Yet, while they consider interactions between variables, the representation of system evolution over time is more challenging and results are predominantly provided in a pre-and post-development format, highlighting whether there is any net gain in assessed ESS from existing to new land uses. To fully understand the built and natural environment system and the implications of change, it is important both to assess how different evolve over time (Meadows, 2008). Doing so helps decision makers better understand the trajectory of change, identify tipping points and areas where timely interjections can improve outcomes while helping facilitate discussion among stakeholders.
System dynamics (SD) is an approach for conceptualising, analysing and understanding dynamic complex systems (Sterman 2000). Based on closed chains of relations and feedbacks, SD modelling is well suited to representing the complexity of the integrated built and natural environment (Coletta et al. 2020;Giordano et al. 2019). It provides a useful tool for urban architects, planners, developers and decision makers to identify appropriate design and management strategies while helping policy makers develop sustainable approaches to urban planning (see Hall et al. 2013;Whyte et al. 2020). Using participatory modelling and Group Model Building SD is a well-known tool to allow participation in practice (e.g., Vennix et al., 2016). While SD modelling has been used to explore urban design challenges (e.g., , it is also uniquely suited to evaluate the ESS of a proposed or existing blue or green space. As SD modelling is based on integral J o u r n a l P r e -p r o o f Journal Pre-proof equations to represent variables as stocks and changes in these as flows (Eker et al. 2018) the methodology has potential to help improve the conceptualisation of NC and the influence land use planning decisions can have on it and the services it provides.
Through the use of group modelling exercises, SD modelling also facilitates a participatory approach (Eker et al. 2018), allowing the inclusion of stakeholders views ). Within environmental decision making the importance of stakeholder participation has long been recognised at the international and local level (Reed et al. 2009;UN 1992;UNECE 1998) and has been identified as essential for fair, sustainable (Gokhelashvili 2015;Reed et al. 2009) high quality and durable decisions (Beierle 2002;Fisher, Turner, and Morling 2009;Reed et al. 2009). However, participation is often identified as lacking within the NC and urban ESS approaches (Campbell-Arvai and Lindquist 2021). If the urban ESS concept is to be a useful tool for sustainable urban planning, stakeholders' perceptions of urban ESS should be considered more carefully in research (Luederitz et al. 2015).
Additionally, participatory modelling can empower local stakeholders through the inclusion of their views, concerns and aspirations in the decision making process (Klain and Chan 2012). In this paper, using stakeholder knowledge as a foundation, we develop a SD model representing urban development to explore the potential of the approach in articulating the impacts on and benefits of NC. The novelty of our work lies in using SD to understand and evaluate NC within a complex and changing human and natural environment. The novelty of our work lies in using SD to understand and evaluate NC within a complex and changing human and natural environment. however, our primary goal is a dynamic representation of Urban Natural Capital (UNC), improving our understanding of the trade-offs and opportunities of socio-economic, built and natural environment change.
J o u r n a l P r e -p r o o f

Journal Pre-proof
In Section 2 we provide a description of the study area; an urban space containing a significant amount of NC which is currently undergoing redevelopment. This is followed by an overview of the collaborative investigation of the Thamesmead area. Section 3 describes the translation of this knowledge into a SD model capable of evaluation the performance of natural space. The model is then applied to the study area in Section 4 where a number of different scenarios are explored to demonstrate the efficacy of the approach to support the understanding of how NC and ESS in urban environments may be affected by new development.

The study area: Thamesmead Waterfront Development
Originally

Collaborative investigation of the Thamesmead blue-green and built environment
SD modelling processes can include both qualitative/conceptual and quantitative/numerical modelling phases (e.g. Pagano et al. 2019). Within the Thamesmead case study, a participatory qualitative modelling process was carried out to bring together organisational and institutional stakeholders, including developers, regulatory bodies, NGO's and local Government, to jointly scope the focus of several quantitative SD models around the built and natural environment and sustainability. Specifically, the qualitative modelling phase aimed (i) to collaboratively identify a shared concern (namely a shared formulation of a "problem" which serves as a representation of the different concerns and stakes carried by the different stakeholders, see Ostanello and Tsoukiàs, 1993;Pluchinotta et al., 2019); (ii) to build a number of Causal Loop Diagrams (CLD) around the identified shared concern in order to gather knowledge on the system and to capture different perceptions from each stakeholder group. The identified shared concern was: how best to sustain and increase the quality of Built/Blue/Green space to ensure long term stewardship. Between 10 and 15 stakeholders participated in each workshop. A detailed description of the participatory qualitative modelling process is described in (Pluchinotta, Salvia, et al. 2021).
At the end of the participatory qualitative phase, as described in ,) stakeholders identified NC and Natural Space Performance (NSP) as one of the key issues and a priority to investigate, via a voting poll and a group discussion. Afterwards, several J o u r n a l P r e -p r o o f modelling sessions between academic experts were held for the creation of the CLD presented in Figure 2 and the related SD model, which focused on changes in the quality of natural space from a NC viewpoint. The model can evaluate the capacity of natural space within the study area to provide ESS over time in relation to change.
The modellers developed the main structure, equations and parameter values of the SD model mainly using the information gathered from the scientific literature and technical reports (Please see Appendix A of this paper for further details). Experts in hydrology (5), NC (2) and urban environment (2) also shared their knowledge and supported the main modellers during the process. The modelling meetings were mainly between 1 or 2 experts and the modelers and focussed on discussing and improving specific sectors of the model.
Following the completion of the model prototype, an additional stakeholder workshop was held with key stakeholders (namely members of a social-environmental NGOs working in the area with specific technical knowledge), during which the model structure and operation was described and validated. While participants suggested alternative equations, which could be used for variable change calculation, the overall structure of the framework was deemed to be accurate. The validation activity lasted two hours and involved a request for feedback on the overall structure of the model and focused on specific items of the model, for example, the type of land, the dynamics of change in developable land and the idea of the space performance indicator. In addition, during the validation activity, the stakeholders highlighted the usefulness of the framework, both for system understanding and information dissemination.
Therefore, the SD model presented was developed using both qualitative and quantitative information obtained through a combination of discussions with key stakeholders and J o u r n a l P r e -p r o o f subject experts. Information includes relevant variables for UNC and GI functioning in the study area, for example local population, the area of built and natural space, biodiversity, access to space and rainfall runoff.
A CLD describing the framework is presented in Figure 2. The balancing loops B2 and B3, show how the 'biodiversity performance' and the related 'natural space performance' is highly dependent on both the quantity of 'developed land' and 'natural space'. Similarly, B4 and B5 link the 'hydrological performance' with land type. The CLD includes an implicit link between 'natural space performance' and the desire to move to a particular area. The current version of the SD model does not include this aspect due to lack of robust data for parametrization. Moreover, the hydrological link between the built environment and run off was also included in the CLD for completeness, but not in the SD model, as

Population (expected and actual)
Change in population is a primary driver of housing demand and represent one of the main reasons behind the Thamesmead Waterfront Development project. While this modelling framework has the capacity to directly model changes in population through the number of people moving into and out of an area, birth and death rates, we instead use the projected population (Askew 2018), in order to explore the outcomes of planned and potential future J o u r n a l P r e -p r o o f scenarios. There are many factors which influence the number of people moving into and out of a location, including the quality of its blue and green space. Data shows that properties less than 100 metres from green space are on average £2,500 more expensive than those greater than 500 metres from green spaces (ONS, 2019). Desirability is also influenced by the quality of the natural space offer which includes the ecological quality and the condition of the space, its maintenance, safety and aesthetics. Therefore, the model accounts for the feedback between the performance of the natural space and its attractiveness and how that influences the number of people moving into and out of a location. However, in the current implementation of the model the latter is not used as an input; in other words, actual occupancy of the houses built does not feature. explicitly included in the model.

Built Space
This is the area of land in hectares either available for development (Developable Land) or has already been converted to developed land (Built Area). An initial Developable Land value is set at the beginning of the model run based on-site specific information. During the model run, this value can increase by re-zoning land from the other land use types, predominantly natural space, though the model can also consider the impracticality of developing some types of land, for example blue space or wetlands. The parameters for which this occurs are set by the modeller and depend on local policy and development plans. Changes in the amount of developable land is driven by the demand; typically, through an increase in population and the need to develop housing and associated infrastructure. The type of development depends on the local needs and characteristics.

Natural Space
We designate Natural Space as all land types which are not part of the built or developable land. It may also include reclaimed undevelopable land, which cannot be developed on but can also provide ESS. In our study area this represents a reclaimed landfill with the capacity to provide additional natural space for residents. Within the modelling framework the natural space area can be reduced and converted into Developable Land and Built Area, representing the spatial impacts of development on natural space area. Each natural land type, including grassland, woodland, wetland and bluespace, is treated separately. Their This calculation is undertaken individually for each land use type. by an increased number of extreme rainfall events due to changes in climate. This is accounted for within the framework by including a hydrological performance metric.
Based on a methodology described by Whitford et al. (2001), the approach accounts for land use type, its permeability and generation of run-off: Where Pe is run off, P is precipitation (the model uses average annual rainfall), S is the maximum retention of the area where the greater the S value, the smaller the run-off. The value of S in mm, is given by: Where CN is the curve number which describes the land type and conditions. Fully impermeable and water surfaces have a CN value of 100. In this case S will be 0, whereas when the CN < 100, S will be positive. The curve number values in this application are taken from the USDA Hydrologic Soil-Cover Complexes (2004)

Access
Here defined as the physical and social capacity to engage with blue and green space, Health England 2020). Accessibility is the mean of proximity, transport, facilities and amenities and physical obstacles which are described in the following sections and in the Appendix 1 section on Natural Space Performance.

Proximity
The distance of natural space from the homes of potential users is a major determinant of use; two thirds of visits to green space in the UK occur within two miles of the home (Office of National Statistics, Natural England 2018). For the purposes of this model, we define proximity as the distance of green space from the home, following the approach used by

J o u r n a l P r e -p r o o f
Public Health England (2020). This is reflected in the model through a user assigned value between 0.1 and 1, representing the average distance from the home to natural space.
Green or blue space located less than 2 minutes' walk of the home is deemed excellent and given a value of 1. A value of 0.1 denotes greenspace located a walk equal or greater than 40 minutes from the home; the average time it takes to walk two miles.

Transport
Transport infrastructure is an additional variable which determines the usability of public natural space. Public transport includes buses and trains, the number of stops and how regular the service is. Private transport includes parking facilities for cars and bikes, their quality, safety and cost if relevant. This is captured in the model through a user assigned value between 0.1 and 1, where 0.1 represents a scenario with little to no public transport or parking options, and 1 where excellent transport facilities exist.

Availability of facilities
Public amenities include the presence of toilets and washroom facilities, cafes, playgrounds and play areas, benches and seating areas. The presence of particular types of amenities are crucial to the use of space by certain groups of people and their absence will significantly reduce the likelihood of the space being used. It should also be noted that the type of amenities within a park are location specific -public space design will be different for a green space adjacent to a large city office block than a suburban area with young families considering the different users of the space; their needs and how physical obstacles may inhibit safe and regular use. This included access for wheelchair users. These input parameters will vary depending on the case study and are qualitative scores that stakeholders and experts have agreed upon.

Natural Space Performance (NSP)
The assessment of natural resource and ecosystem service sustainability is critical to system understanding and a key part of evidence generation for improved decision making. A number of approaches already exist, for example, the UK's NC Accounts which estimate exchange prices that are directly comparable to GDP. However, whether or not society is on a sustainable trajectory is best accounted for as the aggregate of all NC assets (Bateman and Mace 2020). In order to fully assess the state of human-natural environment interactions it J o u r n a l P r e -p r o o f is important to consider both the stocks of natural assets and the flow of ES, thereby including resource sustainability in decision making; a factor which can be missed using simple flow based assessments (Bateman and Mace 2020). Almost all natural resources are limited in some way; if a decision is made to change the stock or flow of a natural resource or eco-system service, it can reduce the possibility of further utilisation, generating an opportunity cost that may or may not be known when decisions are made (Bateman and Mace 2020).
Following Yun et al. (2017), we consider ESS as a portfolio of assets. The overall performance of the portfolio depends on the performance of the underlying assets which are subsequently influenced by their interactions. These changes in portfolio performance provide an attractive headline index for ecosystem based management, regardless of whether ecosystem wealth is ultimately included in a broader wealth index (Yun et al. 2017). In this paper we develop an evaluation approach where key variables, such as biodiversity, hydrology and access (as described throughout Section 3) are combined and normalised. These values provide metrics, or headline indicators, which can be used to measure the performance of part of (e.g., changes in green or blue space hydrological performance) or the entire (e.g. the urban natural space in the study area) system in relation to change. These metrics provide a useful indication of system behaviour; however, our primary goal is a dynamic representation of UNC, improving our understandings of the trade-offs and opportunities of socio-economic, built and natural environment change.
Creating such a composite indicator, however, is not trivial, as the reduction from three dimensions to one necessarily involves choices to be made. We based our choices on a list  Table 1.
J o u r n a l P r e -p r o o f

Scenario 1 (S1): Approximation of proposed building design (Baseline)
In this scenario, we increase built space area through implementation of a development, broadly in line with the proposed urban design for the Thamesmead Waterfront Development plan. For the purposes of this scenario, no changes are made to the natural space parameters, allowing us to directly explore the impacts of development on NSP.

Scenario 2 (S2): High density building design
In S2, the development footprint is reduced through implementation of high-rise buildings.
Under this scenario no additional changes are made to the natural space.

Scenario 3 (S3): Low density building design
This scenario explores the model's ability to investigate the impacts of urban sprawl on natural space performance. The overall built area footprint is increased, requiring a significant reduction in natural space area. No additional changes are made to the natural space.

Scenario 4 (S4): Proposed building design with green roofs
Here we explore how the implementation of green roofs can affect the overall NSP, providing additional ESS including biodiversity, improved hydrological performance and greater natural space access.

Scenario 6 (S6): Integrated scenario
A combined implementation of parameters outlined in S4 and S5 are used in S6 to explore how natural space performance can be improved through the provision of NBS's to reduce flood risk, increase biodiversity and provide additional high quality green space for residents.   J o u r n a l P r e -p r o o f

Results
The results of all scenarios are presented in Figure 5. for S5 increase from 0.604 to 0.646. This is slightly higher than S1 and S4 due to the increased capacity to reduce flood risk.

Scenario 6 (S6): Integrated scenario
S6 follows a similar building configuration to S1, S4 and S5 resulting in approximately 19ha of built space. The implementation of green roofs during S6 also sees an increase in natural area from 61.5 to 78.9ha. This change supports an increase in biodiversity; from 0.710 to 0.733 and is also reflected in hydrological performance; increasing from 0.437 to 0.568 due to both green roof implementation and the adaption of natural space to reduce rainfall runoff. Access values increase during S6 from 0.711 to 0.799. The implementation of green roofs as part of S6 leads to an increased capacity to provide ESS through additional natural area, leading to an increase in NSP from 0.604 to 0.693.

DISCUSSION
This paper explores how SD can improve understanding of the complex built and natural environment, demonstrating, through a stakeholder informed SD model, its capacity to help planners, designers and developers reduce negative development impacts while maximising the provision of ESS from urban NC. A key benefit of using a SD model is its ability to explore the outcomes of a variety of design options, considering interaction and feedbacks between different key variables over time. Through a series of scenarios, we use the model to J o u r n a l P r e -p r o o f investigate the impacts of high-and low-density housing design, the implementation of green roofs and the use of NBS for flood risk reduction. These scenarios allow us to explore the dynamics of the urban human-natural environment, and how change in one element of design can propagate throughout the entire system. Despite a lack of qualitative and quantitative data, this model is well suited to NC evaluation and urban design, providing useful insights to study area development. This was supported by stakeholders who highlighted the usefulness of the framework, both for system understanding and information dissemination.
The S4 model outputs which describe the hydrological performance of the study area show that the addition of green roofs can play a role in the reduction of flood risk, though less than benefits which could occur by adaption of green space, through changes in vegetation type and corresponding CN for flood risk reduction (S5). However, focusing on just one ESS masks additional benefits which are provided. The inclusion of green roofs creates additional natural space, leading to improvements in biodiversity, and accessibility for residents. Here, the importance of taking a portfolio approach (Yun et al. 2017), where the overall result depends on the performance of the underlying assets and their interactions, becomes clear. The NSP of S4 is similar to that of S5, and while S5 benefits are as a result of targeted changes to improve hydrological performance, improvements in S4 are due to the increase in green space and the corresponding increases in biodiversity, accessibility and hydrological performance. It also supports the idea that we should not focus on improving a single ESS or addressing a single challenge. Without adequate evaluation, NBS are likely to appear less cost effective than traditional grey infrastructure and therefore less attractive to developers and planners (Mell et al. 2013). However, the model and results highlight how J o u r n a l P r e -p r o o f NC provides multiple benefits, including reduced flood risk and increased biodiversity, which should also be considered when designing or adapting urban space.
Globally, urban flooding is a growing problem (Fiori and Volpi 2020), exacerbated by an increase in impermeable area which reduces the natural storage capacity and retention abilities of land to slow the movement of water to rivers and other water bodies. This is particularly evident during extreme rainfall events which are becoming more common under climate change (Li et al. 2020). Flood risk was highlighted as a concern among stakeholders and in the context of climate change, the ability of the development to deal with more frequent extreme rainfall events is required. Both S5 and S6 highlight the positive implication of NBS for flood risk reduction, helping balance the impact of development and the increase in impermeable area, as seen in S1 and S3.

Limitations and Future Work:
While this model is designed to explore the dynamic links and feedbacks between the natural and built environment, and particularly the impacts of development on the capacity of the natural space to provide ESS, it is, like all models, an approximation of reality and is J o u r n a l P r e -p r o o f not free of biases from experts, stakeholders and modelers. The interactions between all variables have not been fully accounted for, including those between access and biodiversity, or biodiversity and hydrology. Future versions of the model will explore these additional links, along with how they interact with socio-economic variables, including income, ethnicity, age and gender. This model could be further developed to investigate the role played by blue and green space in improving physical and mental health -an increasingly important benefit on which there is limited understanding.
As discussed in previous sections, weighting can be applied to key variables to emphasise their relative importance. Data to inform these variables is typically obtained through stakeholder discussion. Due to the limited qualitative and quantitative information available from stakeholders, it was not deemed appropriate to assign differential weighting to all key variables, however, this can be varied easily within the framework by the user. This model includes key study area variables as highlighted by stakeholders and expert opinion. Such data sources have well known limitations (see Vennix (1999) for further details), yet present a key component in system understanding and model development. This data has been further supported with information obtained through scientific literature, published reports and environmental data sets. The framework is designed, for those with some SD modelling experience, to be fully adaptable and extensible and has the capacity to be used in collaboration with stakeholders to explore their priorities, thus facilitating a participatory approach to urban design which can be applied across a range of settings, and spatial and temporal scales. In addition, during the validation activity, the stakeholders highlighted the usefulness of the framework, both for system understanding and information dissemination. Further research is needed to explore potential synergies between SD and J o u r n a l P r e -p r o o f other approaches to support decision-making, (such as Multi-Criteria Decision Analysis). However, through system understanding and the inclusion of stakeholder highlighted key management components, this framework, allows users to weigh and prioritise development and management decisions to help achieve critical socio-environmental targets.

CONCLUSIONS
In this paper we describe the development and application of a system dynamics model to quantify and evaluate the impact of urban design on the multiple benefits and co-benefits provided by UNC and associated ecosystem services. This model provides a tool to help address a significant gap in practice, policy and research by spatially and temporally integrating the human, built and natural environment systems where key links and feedbacks are considered and represented. We have demonstrated that using a SD approach enables a more holistic ESS assessment, evaluating NC performance and allowing differing scenarios for development to be explored simultaneously. Through taking a whole system approach, this model helps identify the negative impacts of development while allowing the user to propose and assess alternative solutions. Stakeholder feedback on the framework emphasised its usefulness in triggering discussion and informing decision making. We introduce a Natural Space Performance metric; a composite of the performance outputs of key variables represented in a single headline indicator showing the propagation of change. We apply the framework to a London case study, comparing different plausible alternative permutations of the development plan. Model outputs highlight where potential J o u r n a l P r e -p r o o f improvements could be made, leading to increased green space and a reduction in flood risk. This framework helps articulate and explore the many interconnected effects of development on NC over time, allowing users to weigh and prioritise decisions, helping achieve socio-environmental targets while addressing housing and natural environment requirements.