Most papers analyzed (86%) fall under the broad research area of ‘Life Sciences’ defined in the Web of Science Collection. As part of this group, papers mainly belong to the fields of ‘Environmental Sciences’ (34.1%), ‘Ecology’ (34.1%), and ‘Biodiversity and Conservation’ (13.9%). Papers in ‘Social Sciences’ represent about 6% of the literature analyzed, dominated by ‘Business and Economics’ (2.6%). Papers related to ‘Technologies’ (2%), ‘Physical Geography’ (1.2%), ‘Oceanography’ (0.9%), and ‘Water resources’ (0.9%) dominate the remaining 8%. About 2.3% of the papers were classified under ‘Other topic’.
The usability of maps for decision-making is the most studied of the four criteria (71.9%, n = 97, Fig. 2), followed by the recommendations for future research for improving ES map operational usability (66.7%, n = 90), the recommendations for effective policy implementation (48.9%, n = 66), and by the integration of stakeholder in the research design (33.3%, n = 45). Only 12.6% of the papers discuss all four criteria, 37% covered two of the four criteria, and about a quarter of papers (respectively 24.4% and 25.9%) cover one or three critera (Table 1, Supplementary Material C).
Figure 3 represents the distances between stakeholder participation levels and decision contexts. Proximity between variables indicate stronger relationships between those variables in the literature analyzed. Three main clusters (A, B, and C) emerge on the map. Cluster A, with the highest density, has ‘academic’ as central variable. Academic has a strong relationship with eight decision contexts (e.g., ‘project evaluation’ and ‘priority setting’) and one participation level (‘no participation’). Clusters B and C have lower densities. B is clustered around ‘consult’ to which are linked two other participation levels (‘inform’ and ‘collaborate’) and the decision context ‘preference assessment’. C clusters ‘involve’ and ‘societal’. Other variables are scattered independently on the plot with strong dissimilarity to other variables. For example, ‘legal and regulatory issues’, ‘awareness’, and ‘reputational and marketing’ present the strongest dissimilarities to other variables. Clusters A and B are highly dissimilar, while B and C remain close.
How were stakeholders participating?
We found 65.2% (n = 88) of studies did not report any stakeholder participation (Table 2, Supplementary Materials C). If 34.8% had stakeholders participating, only 2.2% (n = 3) did so through empowerment, while 16.3% had stakeholders ‘involved’ and 7.4% were ‘consulted’. Involvement was mostly used in expert-scoring and stakeholder workshops, taking advantage of stakeholder knowledge to increase the accuracy of ES maps. Consultations occurred mostly when stakeholders answered surveys on their preferences (Cluster B, Fig. 3). Indeed, the distance map demonstrated a link between ‘preference assessment’, ‘consult’, ‘inform’, and ‘collaborate’. The highest levels of participation (empower and collaborate) show important distances to most decision contexts clustered in group A.
The second most recurring type of participation level was collaboration (8.9%), which mainly took the form of participatory mapping approaches to evaluate policy impacts. None of the studies used ‘Inform’ as a type of stakeholder participation. This could find an explanation in the poor use of ES mapping studies for ‘Awareness’ raising (1.5%).
From the content analysis of the selected papers, we found several authors faced obstacles in co-developing studies with stakeholders due to a lack of understanding of the concepts. This was especially true in countries where ES concepts have been adopted recently. However, we found many studies arguing for the benefits of considering stakeholder preferences and knowledge (34.8%). This includes considering information on the context of the study area in terms of connections humans maintain with their environment and better understanding what benefits populations. It also includes identifying stakeholder networks, their perceptions, interactions between stakeholder groups, as well as identifying conflicts over natural resource exploitation and land use, and developing acceptable policies.
How usable are ES maps for decision-making?
Another general topic that emerged from the literature was the question of usability of ES maps. As defined by Dix et al. (2003), a concept/tool (ES mapping methods and outcomes in our context) can be considered as being usable according to ergonomics principles when it is flexible, learnable, and robust. We used this framework to assess the overall usability of ES maps.
Many papers discussed the flexibility in the application of their ES mapping methodology for decision-making. Indeed, the broad range of decision contexts ES maps are used for are a testimony to the flexibility of ES mapping. Most papers addressing the usability of ES concepts and their mapped outcomes (n = 67) fall under the broad class of ‘Priority Setting’ (24.4%, n = 33). Those use ES maps for prioritizing areas (e.g., ecologically vulnerable areas, recreational areas, areas of conflicts over ES access) by defining hotspots of ES where several ES are provided on a given area. These papers argue using ES maps for priority setting allows framing problems and monitoring changes in ES delivery to define appropriate management strategies and actions (e.g., spatially target areas for environmental restoration), or test the impact of alternative policies. However, 50.4% of papers made no clear statement on the decision context associated with the produced maps (n = 69) other than academic purposes. ‘Policy evaluation’, ‘Instrument design’, ‘Preference assessment’, and ‘Societal purposes’, dominate the second group of decision contexts (ranging from 11.1% to 11.9%). Policy instruments designed with the ES maps produced in the papers were diverse: e.g., setting up monitoring systems; developing incentive mechanisms, such as compensation payments; or, designing mitigation measures to recover environmental damage. In support of policy evaluation, the papers essentially produced ES maps to assess the performance of instruments and policy measures aimed at improving ES provision. Flexibility is also discussed in the papers through method substitution. Indeed, methods for ES mapping are considered by the papers as substitutable depending on data availability (e.g., using publicly available data in data-scarce regions) and the problem to solve. In sum, ES mapping practices are claimed to be highly customizable to fit specific needs.
Learnability is defined in this paper as the ease with which new users can rapidly achieve optimal performance when using something new to them. We refer to optimal performance as the production of consistent and predictable results with accepted standards. Overall, learnability as a component of usability is not well defined in the papers collected. On the contrary, the papers discuss more the rigorous and complex nature of ES mapping along with difficulty to produce consistent and accurate results. To overcome the low levels of predictability and consistency, studies recommend using ES maps combined with other spatial information (e.g., maps of biodiversity, migration routes, vulnerable species) to guide decision-making efficiently. Moreover, learnability is increased when users are allowed to learn from previous experiences, a parameter that is not discussed by any of the papers. Therefore, although 71.9% of papers make statements of the usability of their results for decision-making, few are these drawing explicitly on the ease of learning the methodology and replicating it while producing consistent results.
We finally considered robustness as the combination of responsiveness, replicability, and accuracy. Overall, robustness is rarely documented in the context of ES mapping studies, demonstrating a need to understand better the conditions in which ES maps provide robust information for decision-making. First, the use of indicators sensitive to change is required to obtain responsive mapping results. Responsiveness of ES indicators used for mapping is essential for the adoption of ES maps. If maps are out-of-date and are not sensitive to changes, the information provided by the maps could misguide decision-making, especially if map uncertainties are not clearly stated. The papers analyzed highlight the difficulty of developing highly sensitive indicators because it requires a good understanding of the drivers of change. However, they point out the importance of developing indicators sensitive to external impacts to guide decision-makers in the adaptation of management measures in accordance with pressures. We found the methods used in most papers could be difficult to apply to other regions or for other problems to solve, or failed to mention the results’ potential to do so. However, a small share of the papers analyzed claimed their methods are replicable in other locations (e.g., replication from an island to another) or consistent with other ES studies on the same territory. Nevertheless, papers generally failed to state clearly the uncertainties of their results. Map uncertainties find their sources in data inputs and methods used for the quantification of ES. For example, Wang et al. (2018) tested the influence of using different quantification methods and found important variations in the accuracy levels of mapping outcomes, which greatly influenced their reliability.
Key policy recommendations
Half the papers made policy recommendations (48.9%, n = 66), most of them without using participatory approaches (62.1%, n = 41). The main decision contexts for which these policy recommendations were made covered academic purposes (18.5%, n = 25) and priority setting (16.3%, n = 22). In a lesser proportion, decision contexts also covered preference assessment (8.9%, n = 12), policy evaluation (7.4%, n = 10), and societal analysis (6.7%, n = 9). From the analysis of the content of the papers, we identified four types of policy recommendations:
Recommendation 1: harmonize semantics to allow cross-sectoral frameworks for ES mapping
Papers argue for the development of cross-sectoral multidisciplinary frameworks and indicators, as ES mapping can contribute to a broad range of policy questions, and one ES map can answer several policy objectives. This could lead to cost-saving approaches to policy strategy and design. However, these papers identified the fragmentation of government bodies (e.g., the separation between agriculture and water departments) as hindering the integration of ES concepts in multidisciplinary strategies. A solution would be to create semantic links between policy tools of different government bodies and create a link between these tools and ES tools.
Recommendation 2: use ES maps in support of existing policies relying on mapping outcomes
Papers highlight windows of opportunity for the use of ES mapping in policies relying heavily on spatially explicit information. We found ES policy implications for European strategies, such as the EU Water Framework Directive (WFD) and the Marine Spatial Planning (MSP), received special attention. Papers proposing this recommendation establish a path between the ecological functions assessed in these strategies with ES supply relying on these functions (e.g., the link between habitat provision and species richness). Papers argue policies could use ES maps with other spatial information to enhance the general benefits of strategic planning, highlighting the need to better balance the use of biophysical metrics (dominant use) with other metrics such as socio-cultural components, often neglected in these policies. For example, Keeler et al. (2019) argue the general lack of inclusion of both biophysical and socioeconomic data in decision-making leads adversely to either decisions made for societies or decisions made for the environment. It is suggested to ponder economic, social, and environmental aspects when including ES maps in planning strategies to combine ecosystem protection with sustainable socioeconomic development.
Recommendation 3: use ES maps to support sustainable urban planning
In the context of growing global urbanization, ES assessments for urban landscapes are widely discussed in the papers reviewed. Due to their spatially explicit nature, policies dealing with landscapes should rely on maps to support policy-making and to assess their effects on land-use changes. However, only 7.4% of the papers analyzed in this study proposed policy recommendations for the purpose of policy evaluation. Land-use change constitutes a major threat to ES supply. The papers dealing with urban planning suggest making use of ES mapping to design policies for two main purposes: First, for green corridors’ development, to maintain the flow of ES delivery in urban areas while ensuring equal distribution of urban ES to the populations, and second, for the development of land-sparing’ strategies, to avoid chaotic urban development while increasing the resilience of existing natural areas. We found a common obstacle to the implementation of ES maps for urban planning. There is an agreement on the difficulty to define the spatial boundaries of ES, as ES delivery do not end with administrative boundaries. The authors suggest not limiting ES assessments to small areas, but instead broadening the extent of studies. Papers encourage policy-makers to: (1) consider as many ES types as possible and (2) to understand trade-offs and synergies between ES and pressures affecting urban areas by mapping ES not only within urban area boundaries.
Recommendation 4: build capacity to promote locally relevant policy-making
While top-down approaches are needed to enforce measures, organize human activities, and preserve natural areas, bottom-up approaches are also key to designing policies more adaptive to local realities. Papers argue the usefulness of bottom-up approaches to fit decisions to populations’ needs for ES (e.g., dependency on natural resources). These recommendations link directly to stakeholder participation, which is believed to foster interest and acceptance of decisions while developing a shared knowledge system, a ‘mutual learning phase’. Policy regulations should therefore attempt to connect ES supply and demand at the local level to improve populations' access to ES. Two key solutions to build capacity were discussed in the papers: (1) policy-maker training and (2) technical guidance proposed by communities of practice.
Key research recommendations
Research recommendations were made by over half the papers analyzed (66.7%, n = 90), for the most part without using participatory approaches (67.8%, n = 61). Nearly half (43.3%, n = 39) of these recommendations are given in a decision context. Priority setting (16.3%, n = 22) was the main decision context (other than academic purposes) for which these research recommendations were made. In a lesser proportion, decision contexts also covered ‘Societal’ analysis (6.7%, n = 9), ‘Policy evaluation’, ‘Instrument design’, and ‘Preference assessment’ (each 5.9%, n = 8). From the analysis of the papers’ content, we identified four types of research recommendations:
Recommendation 1: ensure transparency on mapping processes and related uncertainties
ES maps can be perceived as being risky to use in support of decision-making due to the large uncertainties associated with their content. Communication on ES maps uncertainties is a very common recommendation made to researchers by the studies reviewed. The absence of rigorous quantitative methods for assessing uncertainties constrains the usefulness and legitimacy of resulting maps. ES study should highlight biases, generalizations, and uncertainties to determine whether uncertainty levels are too important to be used for sound decision-making or not. Using uncertain results could increase the risk of taking harmful decisions for the environment, waste limited public resources, and create conflicts between researchers and practitioners. An obstacle to reducing the gap between theoretical knowledge and its practical use for decision- and policy-making is the disciplinary fragmentation of ES studies. The solutions exposed in the papers encompassed: (1) unifying methods for ES mapping; (2) requiring researchers to make a clear statement on results’ quantitative or qualitative uncertainties; and (3) developing standards to achieve interoperability among varying ES tools.
Recommendation 2: select carefully the appropriate spatial scale for ES mapping
Most papers collected identified the inconsistency of spatial scales as the primary barrier for understanding and predicting ES trends. This was the case, since a given ES on a territory can be represented at different scales (e.g., quantification of ES per grid cell, per municipality, per watershed), depending on the problem identified. The papers debate on the dominant use of coarse spatial resolutions for ES mapping based on the attribution of an ES quantity per Land Use Land Cover (LULC, e.g., Corine Land Cover). This proxy-based approach follows the assumption that if one LULC class provides an ES, the supply level is considered constant within the class, no matter the internal variability of the ecosystem (e.g., level of degradation). This is considered as leading to generalization errors and unreliable outputs for local decision-making. Moreover, the inconsistency in approaches and scales limits the comparison of ES patterns and scientific findings from different authors. Some authors suggest using models based on proxies, which are time-efficient and low-cost approaches, but propose to refine these units to better account for their spatial variations (e.g., account for geodiversity, soil ES, or parcel data) and to produce more reliable results at a scale understandable and usable for decision-makers.
Recommendation 3: investigate ES interactions and links with human well-being
The mapping of ES mostly relies on the visualization of ES supply but less on the mapping of ES demand and capacity due to a persistent theoretical gap in ES research. A clear distinction between these components is needed to understand the flow of ES, how and where they benefit people, and to avoid conflicts over ES access. Another research gap discussed is the poor understanding of the relationship that connects ecosystem conditions with human well-being. As ES concepts are by nature positioned in a transdisciplinary context, studies should draw special attention to linking ecosystem characteristics with socio-cultural components. Four research needs were put forward in the papers. First developing different sets of explaining variables to map ES supply and demand. Second, mapping ES delivery and ES demand separately and provide recommendations on how to connect ES delivery with beneficiaries. Third, creating an explicit path linking ecosystem conditions with human well-being and communicating this link to raise awareness and improve the uptake of ES maps for decision-making. Fourth, developing metrics for ES quantification sensitive to human actions to demonstrate the importance of this link.
Recommendation 4: monitor ES trends over time to assess policy performance
The last type of research recommendation emerging for the content analysis of the papers is the need to plan better for the future, based on temporal analyses. If ES maps are to be valuable for priority setting, policy design, policy, and project evaluation (e.g., understanding the effects of past decisions on the actual state of ecosystems), future research must focus on monitoring ES and predicting their trends. ES maps often produce a static vision of ES, which does not allow for sustainable planning. If policies are expected to be adaptive to change, research outcomes must provide guidance by keeping track of trends in ES delivery, and spot synergies and trade-offs between ES in relation to different scenarios. Linking ES trends with management strategies and policy implementation can lead to pointing out which actions will generate ES gains. Five future research needs on temporal analysis are put forward in the papers. First, baselines should be set for ecosystem’s condition against which ES changes could be monitored over time. Second, ecosystem drivers for change and historical landscapes should be better understood to determine how and where land-use changes will affect ecosystems, resulting in a common set of indicators highly sensitive to changes. Third, ecological thresholds and tipping points should be considered in spatial models. Fourth, plausible scenarios should be developed, consistent and relevant to policy questions. Finally, decision and strategies should be confronted with projected ES trends for project and policy evaluation.