Archetypes of social-ecological-technological systems for managing ecological infrastructure

The concept of ecological infrastructure (EI) as a lens for landscape management has the potential to address environmental challenges, such as biodiversity loss and ecosystem degradation, by instrumentalizing Nature’s Contributions to People (NCP). NCPs stems from the Intergovernmental Platform on Biodiversity and Ecosystem Services (IPBES) and refers to the various ways in which the natural world provides benefits, values, and services that directly and indirectly contribute to the well-being, livelihoods, and cultural aspects of human societies. This research explores this potential by proposing an archetype analysis of social-ecological-technological systems (SETS) to manage EI. We derived archetypes using machine learning and clustering on a data-driven SETS framework co-produced with experts in EI management. The archetype analysis was conducted by combining K-means with hierarchical clustering on spatial patterns to generate clusters with similar configurations of social, ecological, and technological subsystems. The approach is illustrated for the canton of Geneva, Switzerland, which experiences high urbanization and ecological pressures. The resulting spatially explicit archetypes of SETS facilitate policy recommendations tailored to multifunctional landscapes, which can be used to derive coherent management strategies for EI. In addition, the approach demonstrates that by taking an integrated landscape approach and engaging with diverse stakeholders, it is possible to develop effective landscape-based management recommendations for promoting the sustainable provision of NCPs and biodiversity within the concept of EI.


Introduction
Ongoing biodiversity loss and ecosystem degradation have unprecedented economic and societal consequences (Costanza et al 2014, IPBES 2022).A means to address some of these challenges is to protect and restore natural habitats to sustain the delivery of Nature's Contributions to People (NCPs), i.e. the benefits that people derive from ecosystems, such as food, clean water, and cultural experiences (Díaz et al 2018).NCPs and ecosystem services are nested or interrelated terms, with NCPs encompassing and extending the scope of the ecosystem service concept (Harrison et al 2019) by aiming to be more inclusive and creating space for different viewpoints (Hill et al 2021).Ecological infrastructure (EI) has emerged as a promising planning concept that aims to create and maintain a network of interconnected natural and semi-natural areas that provide various NCPs and support biodiversity (European Commission 2014, Li et al 2017, Sun et al 2020, Perschke et al 2023).As a counterpart or complement to built infrastructure (Cumming et al 2017), the concept highlights that nature's 'infrastructure' is essential to support human well-being and socio-economic development (Perschke et al 2023) by enhancing the quantity, quality, and connectivity of green spaces while promoting well-being, and improving living, working, and recreational environments (Sun et al 2020).
For example, wetlands as a specific type of EI, are of recreational importance, provide local cooling, contribute to water quality, and store excess water during heavy rains and release it slowly, reducing the risk of flooding downstream (Childers et al 2019).
Converting wetlands to agriculture or urban development can lead to changes in the drainage system, enhancing the risk of downstream flooding (Childers et al 2019).This example illustrates that, functionally, EI provides various NCPs that represent essential prerequisites for protecting settlements and facilitating positive social interactions in urbanized areas (Li et al 2017).In terms of space, EI functions as an interconnected ecological space that spans various scales, forming the foundational spatial structure that upholds ecological processes and safeguards natural landscapes.When it comes to infrastructure, EI acts as a life-support system at national, regional and local scales (Sun et al 2020).For example, wetlands act as natural buffers, absorbing excess water during storms, and thus, support and protect surrounding built infrastructure (Li et al 2017).EI is frequently used interchangeably with the term green infrastructure, which is often applied within urban contexts (Matsler et al 2021).However, while the concept has been put on high-level political agendas, such as the United Nations Agenda 2030 (United Nations 2015) and the draft post-2020 Global Biodiversity Framework (Xu et al 2021), the transition from EI as a theoretical concept towards realizing its implementation presents several challenges (Perschke et al 2023).
From a practical perspective, because EI encompasses a mosaic of landscape types for fostering biodiversity and NCPs (Andersson et al 2014), its implementation necessarily requires transdisciplinary collaboration.There are three aspects for which this is particularly pertinent: (1) to plan and negotiate the physical networked structure of the EI to connect landscape elements across scales and ensure multifunctionality (Lennon 2015, Johns 2019); (2) to determine management strategies for biodiversity and NCPs that are holistic with respect to the interdependent systems that are encapsulated by the EI (Perschke et al 2023), and (3) to recognize that EI is also inherently intertwined with the social and technological systems that shape landscapes, as evidenced by its clear linkage to livelihoods and human wellbeing (Sun et al 2020).One way to navigate the challenge of collaboration in EI implementation is to draw upon the concept of the integrated landscape approach formalized in the field of land system science (Freeman et al 2015).The essence of this approach is to integrate the diverse interests and dynamics of stakeholders to enhance coordination in environmental management activities (Sayer et al 2013).Framing the implementation of EI through the lens of an integrated landscape approach thus offers a promising foundation for reconciling conservation, agriculture, and other land uses.It allows for the consideration of the spatial connectivity of EI at a large scale and the interconnectedness of its various components beyond administrative boundaries (Dennis et al 2018).By taking into account the potentials of different landscape types to provide multiple NCPs, landscape managers can develop and implement strategies that support EI as a whole, rather than focusing on individual components in isolation (Goodwin et al 2022).This approach helps ensure the protection and management of the natural environment in a manner that meets human and ecological needs (Sayer et al 2013).
The complexity of landscape mosaics, characterized by diverse land-use patterns with varying implications for the supply and demand of NCPs along the urban-rural continuum, poses challenges for effective EI management and decision-making (Andersson et al 2014).To address this issue, scholars have proposed the use of archetype analysis to cluster similar recurring landscape patterns (Oberlack et al 2019) and multifunctional areas (Karrasch et al 2019).These archetypal patterns, when observed at different scales, may exhibit variations, and their applicability depends on the consistency of underlying relationships across these scales (Piemontese et al 2022).Archetypes are often conceptualized as representations of recurring land systems (Václavík et al 2013), which can be constructed based on either purely biophysical indicators or by combining the demands of the social system with environmental conditions, as exemplified in this study.Archetypes can be used for informing policies that are specifically tailored to the distinctive characteristics of individual land systems, which proves to be more effective compared to implementing measures in a homogeneous manner (Oberlack et al 2023).Archetype analysis has been used for mapping social-ecological systems to research poverty alleviation and food security issues (Rocha et al 2020); for identifying the human-nature connectedness in social-ecological systems to map areas that face sustainability challenges (Pacheco-Romero et al 2021); and provides a promising approach for facilitating transdisciplinary collaboration to plan and implement EI.
In this article, we show how a spatially explicit archetype analysis approach of the landscape can help define management strategies for EI.We take a socialecological-technological system (SETS) approach to adequately acknowledge the complexity of urbanized landscapes as a co-constituted and co-emergent nexus of social, ecological, and technological systems (Ahlborg et al 2019, McPhearson et al 2022).We combine machine learning techniques with expert knowledge in a highly urbanized area in Switzerland, the canton of Geneva, which is experiencing a strong population growth with high demand for the goods and services provided by EI.Our work extends previous efforts of archetypal mapping of social-ecological systems (Sietz et al 2017, Rocha et al 2020, Pacheco-Romero et al 2021, Goodwin et al 2022) by including high-resolution input data and a new framing which differentiates technological alongside social and ecological variables.We close by discussing how a better understanding of social, ecological, and technological aspects is required to integrate the implementation and management of EI into spatial planning.

Case study
The canton of Geneva is the westernmost canton of the Swiss Confederation (figure 1), and Geneva is the second largest city with the second highest population density (2050.8inhabitants km −2 ) in Switzerland (BFS 2021b).The canton of Geneva consists of a dense urban center (13%) surrounded by peri-urban settlements and agricultural areas (45%), while forest areas (11%) tend to be located towards the periphery of the canton (Honeck et al 2020).The proportion of forest in Geneva is significantly lower than the Swiss average (BFS 2021b).The 14.8% increase in settlement and urban areas (residential, transportation, industrial and commercial, recreational areas and cemeteries, other building and special urban areas (BFS 2021a)) over 24 years has resulted in a total settlement and urban area of 94.3 km 2 or 33.4% of the entire canton (BFS 2021b).The increase in settlement and urban areas over the last decades creates pressure on the remaining areas of natural land cover which exhibit increasing fragmentation (Honeck et al 2020).This trend has led to a concerted effort by various partners to assess the EI for the canton of Geneva.In this regard, the GE-21 initiative was established as a network of experts with the aim of promoting and enhancing biodiversity and NCPs to improve the well-being of the inhabitants of Geneva and its region (Honeck et al 2021).

Archetype assessment
The process of mapping archetypes involves identifying the appropriate scale of analysis, dividing the study area into discrete units, and classifying these units according to their association with specific archetypes (Cullum et al 2017).We build on the framework of Piemontese et al (2022) to structure the required steps in our archetype assessment.The archetype assessment includes four phases (1) design, (2) analysis, (3) stakeholder engagement, and (4) application phase, illustrated graphically in figure 2 and explained in the following paragraphs.The archetype assessment is designed to be iterative and after completion of the four phases, the process should begin again at the first phase.However, due to time constraints and availability of the stakeholders, only a single iteration of the four phases was carried out in the frame of this study.

Design
Based on stakeholders' responses from semistructured interviews about relevant datasets to represent social, ecological, or technological services in Geneva, we conducted a systematic and comprehensive search for spatial data to represent the SETS subsystems by combining the search terms 'Geneva' AND 'social' OR 'economic' , 'ecological' OR 'biophysical' , 'technological' OR 'infrastructural' .We searched (1) the Swiss Federal Geoportal, (2) the Geneva Land Information System (SITG), and (3) scientific publications that fulfilled requirements for a spatial resolution of at least 100 m and a temporal validity of 2009 or newer.This resulted in the consideration of 12 indicators, which we categorized into the three SETS subsystems (social, ecological, and technological), as shown in table 1.We defined the social subsystem to include socio-demographic variables for which we utilized indicators describing population (SFSO 2022a), employment density (SFSO 2022b) and artificial lighting (Ranzoni et al 2019) as indicators of human activity.We designed the ecological subsystem to encompass exclusively the data available on NCPs (Honeck et al 2020).The choice of the NCP framework was informed by the engagement of stakeholders with diverse perspectives.Compared to the ecosystem service concept, the NCP approach aims to foster inclusivity by accommodating various viewpoints (Hill et al 2021).Our primary objective was to understand the impact of these NCPs on human well-being.Consequently, we deliberately omitted other biophysical indicators indirectly incorporated into the NCP models to prevent potential confounding factors.The technological subsystem includes infrastructure and other human-engineered amenities (Keeler et al 2019), represented by traffic noise and data layers on road and rail networks (FOEN 2009, Swisstopo 2011).All data layers were rasterized to a cell size of 25 m.

Analysis
The appropriate numbers of clusters for each subsystem were selected based on the silhouette statistic (Rousseeuw 1987) obtained using the R-package NbClust (Charrad et al 2014) (appendix A).The data layers were normalized and a principal component analysis (PCA) was conducted using the Rpackage raster (Hijmans and van Etten 2016).PCA was applied separately to each subsystem to ensure non-collinearity among the variables (Schirmer and Axhausen 2019, Chen et al 2022), and loading plots were generated to assess the interrelationships and identify potential sources of multicollinearity (appendix B).In the subsequent step, we selected principal components of each subsystem (appendix B) with eigenvalues exceeding one (Schirmer and Axhausen 2019) and used varimax rotation to estimate their impact on clustering results (Joshi et al 2022).The rotated principal component (RC) scores were then employed as inputs for the Kmeans clustering algorithm (Lloyd 1982), following the methodology proposed by Joshi et al (2022).We   (in review) and was modeled using a constrained cellular automata approach under business-as-usual assumptions regarding changes in the drivers and rates of LULC change.Assuming that there will not be significant changes in the way EI is managed, we used the LULC predictions for 2060 as a proxy to represent the expected change in each archetype.We further analyzed the current habitat distribution (Price et al 2021) in the archetypes according to the habitat classification of Switzerland (Delarze et al 2015) by looking at the average percentage cover of habitats within each archetype.Lastly, the distribution of biodiversity, represented by the proportion of species richness (Honeck et al 2020) and the values of ecological connectivity based on the red deer (Cervus elaphus) using circuit theory (Honeck et al 2020) were normalized and analyzed across all archetypes by calculating the average percentages within each archetype (appendix C).
To compare the archetypes, we normalized the input rasters and determined how high each archetype scores in each subsystem (social, ecological, and technological) and ranked highest to lowest performing archetypes within each subsystem.The results were plotted as a three-dimensional scatterplot of SETS (figure 4).We then conducted pairwise Spearman's rank correlation tests to identify correlations between the social, ecological, and technological subsystems (Dytham 2011).The results of these analyses were used to describe and characterize the archetypes to inform the stakeholder engagement step.While the stakeholder inputs significantly informed the entire research process, the specific nomenclature of the archetypes was determined by the research team.This approach was undertaken to ensure consistency and coherence in the representation of the identified clusters.

Stakeholder engagement
As a first step, we conducted semi-structured interviews with participants with a scientific background in ecology to ensure the validity of the data.In a second step, a stakeholder workshop was conducted in a world café format (Löhr et al 2020) with 11 participants from academia and practice in the fields of biology, environmental science, spatial planning, and agricultural planning.Participants were shown posters that included maps of all archetypes and displayed data on current and future LULC, habitat, biodiversity, connectivity, NCP distribution, as well as photos and aerial images of each archetype.Participants were asked to discuss management recommendations for each archetype, which actors to involve, and what policies are needed to support EI implementation in Geneva.

Application
In the final step, we structured the results and findings of the stakeholder engagement and summarized the relevant actors, policies, and management recommendations for EI in the different archetypes (table 2).The actors named by stakeholders were assigned to a category of (1) national stakeholders; (2) cantonal stakeholders; (3) municipal stakeholders; (4) research; (5) non-governmental organizations; or (6) business community adapted from Slätmo et al (2019).The proposed policies were structured into (1) legal and regulatory instruments; (2) rights-based instruments and customary norms; (3) economic and financial instruments; and (4) social and cultural instruments according to IPBES (2022).

SETS archetypes for Geneva
The interaction of social, ecological, and technological systems, incorporating insights from stakeholders and guided by statistical methods (appendix A), results in seven distinguishable archetypes with characteristic features in Geneva, as shown in figure 3.These SETS archetypes characterize different landscape patterns, such as the densely built environment with urban greenery (Concrete Jungle), surrounded by settlements with gardens close to the city (Urban Fringe).Around these settlements, we observed a transition to agriculture (Transition Fields) and a higher proportion of intensive agriculture in the outskirts (Working Fields).Additionally, there were identifiable areas with vineyards and crops (Cultivated Oasis), while agricultural areas surrounding forests (Buffer Fields) can be distinguished from forests located in the peripheral areas of Geneva (Green Frontier).
The three-dimensional scatter plot representing the performance of the archetypes across the SETS subsystems (figure 4) suggests an underlying relationship between the respective SETS subsystems that follows a gradient that runs from the Concrete Jungle archetype in the upper left corner to the lower right corner in the foreground with the Green Frontier.A clustering of the Concrete Jungle, Urban Fringe, and Transition Fields archetypes can be seen, as these archetypes typically have a high percentage of residential development with a consequently corresponding low ranking in the ecological subsystem, while the Green Frontier exhibits the opposite pattern.The other archetypes show a more balanced interaction between the social, ecological, and technological subsystems, with a notable result being that the Cultivated Oasis archetype is ranked highest in the ecological subsystem, while the Working Fields archetype is ranked second lowest (figure 4).
The results of the Spearman's correlation analysis with a 95% confidence interval demonstrates that the social and technological subsystems are positively correlated (R = 0.82, p = 0.024 * ) while the ecological subsystem is negatively correlated with both social (R = −0.93,p = 0.003 * * ) and technological (R = −0.75,p = 0.052) subsystems (appendix D).This indicates that there is a disconnect between the ecological subsystem and the other subsystems in Geneva.This result could be expected, as social and technological services are centered around heavily modified urbanized areas while ecological services are centered around natural areas.
To understand the SETS of Geneva better and to comprehend how to define suitable management options, we further explored the distribution of the social, ecological and technological indicators (figure 5) as well as the distribution of biodiversity within the archetypes (figure 6).Analyzing the distribution of biodiversity (figure 6) and NCPs (figure 5) across the archetypes confirms that lowest biodiversity and NCP values can be observed in the Concrete Jungle followed by the Working Fields, whereas the highest biodiversity levels are reached in the Green Frontier and the highest NCP levels are provided by the Cultivated Oasis.The Green Frontier contributes substantially to microclimate and air quality regulation as well as carbon sequestration, but plays no role in pollination, whereas, the Cultivated Oasis exhibits the highest values of the NCPs, including water quality, pollination, and erosion control.

Policy and management recommendations for EI
The workshops with stakeholders revealed possible actors, policies, and recommendations for the management of EI within each archetype as shown in table 2. In all archetypes, cantonal stakeholders were identified as crucial to plan for and implement EI.The cantonal stakeholders were proposed to work closely with the national and municipal stakeholders and alongside affected business communities or   non-governmental organizations.The policy setting required for the implementation of EI was seen as a mix of economic and financial instruments (e.g.compensation or taxes) with legal and regulatory instruments (e.g.new laws or control of compliance) in most archetypes.More information describing the archetypes can be found in the appendix E.

Discussion
This paper is the first to spatially operationalize the SETS concept by using machine learning and co-production with experts to derive SETS archetypes aiming to prompt the planning of EI implementation and policies for management.To operationalize the The findings presented in this study show that the ecological subsystem negatively correlates with the social and technological subsystems, revealing a notable disconnect between the ecological subsystem and the social and technological subsystems in the canton of Geneva.This indicates that the ecological subsystem has not reached its potential of being integrated into the built infrastructure.We argue that to bridge the gap and to alleviate the disconnect from the ecological to the social and technological subsystems, it is necessary to restore and integrate nature into the built environment, particularly in the Concrete Jungle and Urban Fringe archetypes.To provide the broad range of demands of NCPs and biodiversity to residents, designing and managing multifunctional EI in urban settings should be prioritized (Jones et al 2022).Stakeholders identified several potential strategies for incorporating nature into the Concrete Jungle, such as the selection and introduction of appropriate species, the creation of large parks, the planting of trees, the reduction of individual mobility, the provision of food production gardens, and the reduction of impermeable surfaces.These suggestions are supported by the findings of Hegetschweiler et al (2017) who conducted a review of publications assessing the linkage of demand and supply of green spaces and found that size and shape of green spaces as well as the habitat, structural and species diversity are crucial for the use and benefit of nature in the urban environment.Enhancing NCPs can be achieved without compromising biodiversity, for example, Belaire et al (2022) found a positive correlation between tree species richness, native tree percentage, and carbon sequestration in urban green spaces.These promising results call for more systematic evaluations of the potential trade-offs and co-benefits between NCPs and biodiversity (Lanzas et al 2019).
Another leverage point for bridging the gap between ecological and other subsystems is by minimizing potential conflicts between environmental conservation and other land use (Honeck et al 2020).In the agriculturally dominated archetypes, stakeholders suggested strategies such as extensification of agriculture, increasing pollination, planting hedgerows, and revitalizing rivers.The relevance and interconnectedness of these suggestions are supported by literature (Bokusheva et al 2022).For example, maintaining adequate resources in the agricultural landscape is essential for sustaining wild pollinators, highlighting the dependence of pollination services on small landscape elements, including hedgerows or forest patches (Bokusheva et al 2022).These resources encompass nesting habitats, such as tree cavities or suitable soil substrate, as well as floral resources in the form of pollen and nectar (van Berkel et al 2018).Such habitats would be of significant importance in the Working Fields archetype, which is characterized by intensive agriculture and has correspondingly low biodiversity values.However, to mitigate conflicts, measures aimed at safeguarding and reviving small landscape elements need to be co-produced and implemented through policies that allow for compensation for actors of competing land uses.
The effectiveness of such policies is enhanced when they are specifically tailored to distinctive characteristics of individual land systems instead of homogeneously implemented (Oberlack et al 2023).The implementation and management of EI should avoid a one-size-fits-all approach and instead needs to account for regional heterogeneity of actors and landscapes.Indeed, we show that the diverse types of landscapes in Geneva require the identification of spatial patterns reflecting social, ecological, and technological systems in the data.However, the majority of planning authorities that use an EI planning approach have maintained conventional administrative structures (Lennon 2015).The utilization of small-scale spatial parceling or communal and cantonal borders as management zones for EI neglects the fact that they are not ideal for managing EI (Honeck et al 2020), which may perpetuate a silo approach to planning (Lennon 2015).To ensure the sustainability of EI, it is imperative to adopt a multi-sectoral, transdisciplinary landscape approach to their management.We therefore propose a delineation of management zones for EI which is co-produced with experts (Chambers et al 2021) in a transdisciplinary collaboration and anchored on landscape-based data-driven analysis (Dennis et al 2018).
While generalization has benefits for refining management strategies, it is important to acknowledge that this process inevitably results in the loss of context-specific and location-specific details.It is crucial to recognize that archetypes identified at a    et al 2022).Furthermore, the selection of the number of archetypes significantly influences the outcomes (Joshi et al 2022).In our case, stakeholders recommended a range of four to eight archetypes for the canton of Geneva to guide the determination of the appropriate number of archetypes (Piemontese et al 2022), with the exact number of archetypes derived from statistical analysis of the data.Given that the number of archetypes is case-specific, there is no standardized procedure for its determination.This underscores the need for attention to detail and further research on this topic.
Another aspect to consider is the choice of datasets for inclusion in archetype analysis.Highresolution mapping is rarely found amongst archetype analyses (Yang et al 2023).In our study, we focused on high-resolution datasets, all of which were available at a 25 m cell size, except for employment density, which we resampled from 100 m to 25 m.This resampling introduced abrupt transitions due to the lack of gradual changes from high to lowdensity areas.Given its importance to stakeholders, we included this layer in our analysis.Stakeholders also provided input on other datasets to include, such as land use intensity (social), agricultural production, cultural ecosystem services, tranquility and wilderness maps, temperature, precipitation (ecological), and building density, wildlife crossings, wind turbines/photovoltaic, and other energy infrastructure (technological).We deliberately excluded biophysical indicators, like precipitation or temperature, which are indirectly utilized in the NCP models, to prevent potential confounding effects.As for the remaining indicators, these datasets could have added to the depth of the analysis.However, we were unable to obtain them at a suitable resolution, and the decision was made to not include them.
This analysis employs a distinct framing that explicitly separates social and technological aspects.In contrast, some similar studies have integrated technological aspects into the social subsystem (Rocha et al 2020, Pacheco-Romero et al 2021, Yang et al 2023).We chose to distinguish between social and technological subsystems to emphasize the significance of human-engineered elements like infrastructure, buildings, and other amenities prevalent in Geneva's urban landscapes.The supply and demand for NCPs are notably influenced by patterns of settlement development, particularly the presence of existing built infrastructure such as stormwater and sanitary sewers, or railway systems (Keeler et al 2019).This social-ecological-technological approach to EI helps underscore the substantial impact that built infrastructure has on ecological systems (Andersson et al 2022), and how efforts to influence NCPs are shaped by social, ecological, and technological factors (Keeler et al 2019).It is worth noting that not all indicators strictly belong to a single subsystem; for example, artificial lighting could be interpreted as part of the social or technological subsystem.Based on the literature, we categorized artificial lighting as a demographic and economic indicator (Ranzoni et al 2019) and integrated it into the social subsystem.
Several studies have explored the potential of various clustering algorithms to identify archetypes (Rocha et al 2020, Pacheco-Romero et al 2021, Beckmann et al 2022, Joshi et al 2022).The spatial patterns clustering we applied here relies on a spatial signature of sub-areas representing a single spatial scale and its results can be influenced by resolution and edge effects.Machine learning techniques such as convolutional neural networks could be beneficial in this regard as they can identify complex patterns in continuous data (e.g.satellite imagery) that do not require pre-processing (van Strien and Grêt-Regamey 2022).These techniques to extract landscape patterns from continuous input layers could reveal more subtle patterns which spatial patterns clustering might not detect.Leveraging these methods to read landscape patterns over time from satellite data may enhance our understanding of the underlying processes and lead to more specific support for management and decision-making processes.Such landscape patterns trajectories over time also show newly emerging risks and opportunities, which are crucial for the design and selection of management policies (Oberlack et al 2023) in the context of EI.
To pursue a successful policy tailoring in EI is dependent on broadly accepted archetypes.The acceptance of archetypes is key to generating actionable knowledge (Piemontese et al 2022).In order to achieve acceptance, archetypes should demonstrate internal coherence (Jones et al 2022) as well as ownership over and commitment to the archetypes, best accomplished through participatory and transdisciplinary research efforts (Piemontese et al 2022).While archetypes are useful in identifying management recommendations, they rely on a counterpart for allocating specific locations of ecological interventions, which is crucial for addressing environmental challenges (Jones et al 2022).To allocate interventions effectively, spatially explicit bluegreen-infrastructure maps (Donati et al 2022) and spatial conservation prioritization maps (Lanzas et al 2019, Honeck et al 2020) could additionally be used in the stakeholder process.This would foster actionable knowledge, allowing to pin-point management recommendations to specific locations in a coproduction process with land management actors.
The objective of archetype analysis is to identify essential components that facilitate knowledge transfer across different locations (Oberlack et al 2019).Generalizing findings from place-based research can be challenging due to various factors unique to the study system (Václavík et al 2016).The broad and synthesized insights obtained from archetype analysis in Geneva are transferable to other areas with similar SETS-configurations.In Geneva, as in many urbanizing landscapes in the world, nearly all archetypes are susceptible to substantial LULC shifts towards built-up areas, resulting in increased LULC pressure and higher demand for NCPs (Lanzas et al 2019).Consequently, there is a need for further comprehensive research towards management recommendations for EI to avoid biodiversity loss and meet the rising demand for NCPs.Therefore conducting such an analysis at various locations to compare the archetypes and define more general policy recommendations encompassing various landscapes, even across national boundaries to extend their applicability to unexplored systems (Václavík et al 2016) would be of great interest.

Conclusion
EI as a concept is a promising countermeasure to the negative impacts of urbanization and subsequent environmental deterioration.However, the practical implementation of EI is challenging because there is a lack of formalized approaches to facilitate the transdisciplinary collaboration that it requires to be successful.This research has contributed to addressing this challenge by operationalizing the concept of SETS combined with data-driven archetype analysis and co-production methods to bring together relevant experts to develop spatially tailored, comprehensive, and coherent management options to support effective EI implementation.This approach has clear potential to be expanded and refined across other case studies to assist in better planning for people and nature while simultaneously promoting biodiversity and the provision of NCPs.The Urban Fringe archetype represents the boundary or edge between urban and rural landscapes and provides significantly higher contributions of NCPs and scores significantly higher for biodiversity than the Concrete Jungle (figures 5 and 6).In terms of LULC, it is expected to see a very high increase in settlement and urban areas, as well as a decrease in intensive agriculture and grasslands (appendix C5).To react to this settlement pressure, stakeholders recommended that cantonal agencies in cooperation with non-governmental organizations should apply economic and financial instruments or legal and regulatory instruments to limit dense hedgerows and incorporate permeable walls or fences to enable connectivity.Furthermore, more microstructures for biodiversity are seen necessary and making use of synergies that promote slow mobility, connectivity, and microclimate as well as NCP provision play an important role in the Urban Fringe.Finally, installing green roofs and energy panels, improving light and safety, quality densification, and prioritize public transport through spatial planning were mentioned as important steps to manage EI in the Urban Fringe (table 2).

Ecological subsystem
The Transition Fields archetype materializes the dynamic nature of areas between the urban and rural environments.Transition Fields encompass a sense of change and transformation, as these fields are undergoing a shift from rural agriculture, crops, and grasslands towards more peri-urban development with expanded settlement and urban areas (appendix C5).This interplay requires a balancing of the needs of agriculture, urban development, and environmental conservation.Transition Fields contribute slightly higher values to NCPs and biodiversity than the Urban Fringe (figures 5 and 6).Stakeholders recommend having cantonal stakeholders and municipal stakeholders collaborate on economic and financial instruments as well as legal and regulatory instruments to create or restore connectivity by limiting barriers, planting hedgerows, revitalizing rivers, and making use of wetland potential.Additionally, increasing biodiversity by measures to protect soil, extensification of agriculture, plant trees and engage in agroforestry and protect NCPs by limiting disturbance, barriers, and settlement expansion are central to managing EI in the Transition Fields (table 2).
The Working Fields archetype contains productive nature, playing an important role in providing food and other resources for the local community and beyond.These lands are actively managed and tended to by farmers.In terms of LULC, it is expected to see a high increase in settlement and urban areas and a sharp decrease in intensive agriculture and grassland (appendix C5).However, this archetype also comprises the potential negative impact of intensive agriculture on biodiversity and NCPs.The biodiversity values and NCPs in the Working Fields are the second lowest following the Concrete Jungle (figures 5 and 6).Stakeholders recommend to involving national, cantonal, municipal, as well as business communities to implement legal and regulatory instruments in order to promote connectivity by planting hedgerows, revitalizing rivers, and creating bridges and underpasses.Another aspect of importance is the extensification of agriculture and creation of biodiversity hubs to protect and improve biodiversity.Lastly, stakeholders emphasized the choice of crops through spatial planning measures to manage EI in the Working Fields (table 2).
The Cultivated Oasis archetype highlights the coexistence of productive agriculture and valuable ecological features.In terms of LULC, it is expected to see an increase in settlement, as well as a decrease in permanent crops and grasslands (appendix C5).The Cultivated Oasis shows high biodiversity values and is responsible for the largest share of NCPs, particularly contributing to the NCPs water quality and erosion control (figures 5 and 6).Stakeholders recommend to including national, cantonal, and business communities to implement legal and regulatory instruments and social and cultural instruments to improve and maintain pinch points, hedges, diversification of extensive cultures and long-term herbaceous vegetation.Additionally, species prioritization, grass strips, and limiting pesticide use are recommended to increase biodiversity and heritage value.Due to the high number of vineyards, enhancing erosion and nutrient control and increasing pollination as well as planning for climate change and limiting urban expansion are considered the main plans of action for the EI in the Cultivated Oasis (table 2).
The Buffer Fields archetype plays an important role in buffering and protecting the forested areas from the impacts of agricultural activities.In terms of LULC, it is expected to see an increase in settlement, as well as a decrease in grasslands (appendix C5).
Buffer Fields show relatively high levels of biodiversity and NCPs (figures 5 and 6).Stakeholders recommend incorporating cantonal and municipal actors to apply economic and financial instruments as well as legal and regulatory instruments to connect habitats by limiting barriers and planting hedgerows and revitalizing rivers.The increase of biodiversity could be accomplished by making use of wetland potential and extensification of agriculture.Finally, fostering NCPs through the protection of soil, planting trees and agroforestry, and limiting disturbance and settlement expansion are seen important steps for EI in the Buffer Fields (table 2).
The Green Frontier archetype entails areas bordering between different natural areas and emphasizes the ecological importance of the forested areas.The Green Frontier is expected to see only small changes in LULC (appendix C5).The Green Frontier contains the largest share of biodiversity and high NCP values as it contributes largely to air quality regulation, carbon sequestration and microclimate regulation due to the vast amount of forests in this archetype.However, for the same reason almost no contribution to pollination is recognizable in the Green Frontier (figures 5 and 6).For the Green Frontier, stakeholders recommended that cantonal stakeholders and business communities work out and enforce economic and financial instruments and legal and regulatory instruments to improve connectivity between patches and enrichen biodiversity with tree habitats, wood on site, ponds, and wetlands, and select climate change resistant species.Furthermore, the allocation of recreational pressures to forest parks with a maintenance of use plays an important role for the Green Frontier to continue to provide high levels of biodiversity and NCPs.The thermoregulatory role of water cycle, limiting urban expansion and extensification of agriculture are crucial for the Green Frontier to fulfill its functions like supporting biodiversity, microclimate regulation, or carbon sequestration (table 2).

Figure 1 .
Figure 1.Overview map of the canton of Geneva, located in western Switzerland.The habitat map shows nine main habitat types based on Price et al (2021).Adapted from Price et al (2021).© EnviDat 2022.CC BY-SA 3.0.

Figure 2 .
Figure 2. Archetype assessment sequencing phases of design, analysis, stakeholder engagement and application iteratively to define management strategies for ecological infrastructure.

Figure 3 .
Figure 3.A map delineating the seven archetypes of social-ecological-technological systems (SETS) identified in the canton of Geneva, Switzerland.Satellite Imagery Maps Data: Google, © CNES/Astrium, Maxar Technologies.

Figure 4 .
Figure 4. Three-dimensional scatterplot showing the ranking of the archetypes based on their performances within the SETS (social, ecological, and technological) subsystems (i.e. 1 = highest performing, 7 = lowest performing).

Figure 5 .
Figure 5. Radar plots of the archetypes of Geneva showing the normalized mean average score per indicator grouped into the social, ecological, and technological subsystems.

Figure 6 .
Figure 6.Distribution of biodiversity within the archetypes (normalized from 0 to 100).
particular scale may have to be re-evaluated when applied at another scale (Piemontese et al 2022).Utilizing nested archetypes (Sietz et al 2017, Yang et al 2023) offers a practical approach to implementing a multi-scale strategy, thereby enhancing the potential for transferring effective sustainability solutions between different geographical contexts (Piemontese

Table 1 .
Indicators used for the archetype assessment of the social-ecological-technological system (SETS) in the canton of Geneva, Switzerland.The indicators are grouped by the three SETS subsystems (social-economic, ecological-biophysical, and technological-infrastructural).

Table 2 .
Summary of the relevant actors, policies and management recommendations suggested by stakeholders across the archetypes.