An expert-based reference list of variables for characterizing and monitoring social-ecological systems

The social-ecological system (SES) approach is fundamental for addressing global change challenges and to developing sustainability science. Over the last two decades, much progress has been made in translating this approach from theory to practice, although the knowledge generated is still sparse and difficult to compare. To better understand how SESs function across time, space, and scales, coordinated, long-term SES research and monitoring strategies under a common analytical framework are needed. For this purpose, the collection of standard datasets is a cornerstone, but we are still far from identifying and agreeing on the common core set of variables that should be used. In this study, based on literature reviews, expert workshops, and researcher perceptions collected through online surveys, we developed a reference list of 60 variables for the characterization and monitoring of SESs. The variables were embedded in a conceptual framework structured in 13 dimensions that were distributed throughout the three main components of the SES: the social system, the ecological system, and the interactions between them. In addition, the variables were prioritized according to relevance and consensus criteria identified in the survey responses. Variable relevance was positively correlated with consensus across respondents. This study brings new perspectives to address existing barriers in operationalizing lists of variables in the study of SESs, such as the applicability for place-based research, the capacity to deal with SES complexity, and the feasibility for long-term monitoring of social-ecological dynamics. This study may constitute a preliminary step to identifying essential variables for SESs. It will contribute toward promoting the systematic collection of data around most meaningful aspects of the SESs and to enhancing comparability across place-based research and long-term monitoring of complex SESs, and therefore, the production of generalizable knowledge.


INTRODUCTION
The social-ecological system (SES) approach arose to formally recognize that human and natural systems are intertwined and interact across nested spatial and temporal scales (Berkes et al. 2000). Currently, the SES approach is widely acknowledged as crucial for addressing global change challenges (Liu et al. 2007, Resilience Alliance 2007, Carpenter et al. 2009) and as a basis for the development of sustainability science (Ostrom 2009, Leslie et al. 2015. It provides new opportunities to understand and manage critical feedbacks between nature and society, which could lead to better ecosystem health, human wellbeing and social equity in the distribution of benefits provided by nature (Collins et al. 2011). However, the complex nature of SESs (Levin et al. 2013) and their heterogeneity across the world challenge place-based social-ecological research (Maass et al. 2016, Norström et al. 2017) and the production of generalizable knowledge from these studies.
Over the past two decades, there has been evident progress in moving the SES approach from theory to practice. First, theoretical studies have defined the general characteristics of SESs, explaining their complexity, dynamics, and emergent properties (e.g., Holling 2001, Berkes et al. 2003, Liu et al. 2007). Second, conceptual frameworks were developed to operationalize the SES concept for place-based research (e.g., Scholz and Binder 2004, Redman et al. 2004, Chapin et al. 2006, Ostrom 2009). Such frameworks have provided lists of variables and components/dimensions of the SES, including the assumed structural relations between these building blocks, usually supported by a graphical representation . Third, the most recent empirical studies have dealt with place-based research through the development of mapping approaches that characterize the diversity of SESs at different spatial scales (e.g., Václavík et al. 2013 or that analyze specific types of SESs at the local scale, e.g., such as fisheries, estuaries, and forest systems (Delgado-Serrano and Ramos 2015, Leslie et al. 2015). Although these empirical studies have provided valuable knowledge on SESs in diverse contexts, it is still difficult to compare and extract general insights from them on how SESs perform over time and across spatial scales (Václavík et al. 2016, Magliocca et al. 2018. Long-term monitoring provides a fundamental basis for understanding the spatiotemporal dynamics of SESs. This has been made explicit in some global research networks, such as the International Long-Term Ecological Research Network (ILTER) and the Program on Ecosystem Change and Society (PECS; https://www.ecologyandsociety.org/vol25/iss3/art1/ Holzer et al. 2018). ILTER includes long-term social-ecological research (LTSER) platforms based on the conceptual model of the SES (Collins et al. 2011). These networks constitute infrastructures for inter-and transdisciplinary research and data collection that aim to produce knowledge for addressing the complex environmental challenges that emerge from naturesociety interactions and to guide sustainability policies , Mirtl et al. 2018. The main goal of PECS research is the integration of place-based and long-term social-ecological knowledge generated from case studies across the world to better understand social-ecological dynamics (Carpenter et al. 2012, Balvanera et al. 2017, Norström et al. 2017). In addition, the World Network of UNESCO Biosphere Reserves introduced the social-ecological approach into protected area management, as well as the need to monitor changes in the biosphere resulting from human-nature interactions . Despite the promising advances in long-term social-ecological monitoring by these networks, one persistent challenge is the harmonization of monitoring protocols to promote cross-site comparability. This would foster more effective interoperability (Vargas et al. 2017) and knowledge generalization from locally driven research initiatives to broader contexts , Magliocca et al. 2018.
The systematic collection of standard datasets is the cornerstone for enhancing our ability to study the spatial patterns of SESs and their trajectories over time . These datasets should be based on a common core set of variables that contribute to fostering a more comprehensive and comparable characterization and monitoring of SESs (Ostrom 2009, Frey 2017. Only a few theoretical studies have dealt with the identification of such common lists of key variables. In this sense, Ostrom (2009) set the most important approach by proposing a list of variables, which were organized in a multilevel nested framework, to understand the sustainability of SESs. Subsequent studies have further developed this list to make it more operational for the empirical study of SESs (e.g., McGinnis and Ostrom 2014, Delgado-Serrano and Ramos 2015, Frey 2017). However, the use of Ostrom's variables in place-based social-ecological research is challenged because of some limitations. For instance, some studies on specific SESs at local scales have reported difficulties in understanding and standardizing the variables and collecting the data (e.g., Basurto et al. 2013, Cox 2014, Delgado-Serrano and Ramos 2015, Leslie et al. 2015. Likely because of these constraints, only a few studies have used this approach for the spatially explicit mapping of SESs , Rocha et al. 2020. To overcome these barriers to operationalization, a standard list of variables should be useful in dealing with the diversity of social-ecological contexts (McGinnis andOstrom 2014, Frey 2017), the complex nature of SESs, and the availability of data (Rocha et al. 2020). Finding a set of variables that meets these requirements will enable the collection of datasets worldwide to enhance place-based research on complex SESs as well as the observation and tracking of long-term trends, encouraging cross-system comparisons.
A promising initiative contributing to the development of core lists of variables to make monitoring of the Earth system comparable across sites is the identification of essential variables (EVs). EVs constitute the minimum set of critical measurements for the study, report, and management of a system and its changes (Reyers et al. 2017, Guerra et al. 2019. Major steps have been taken in the fields of biodiversity (Pereira et al. 2013), climate (Bojinski et al. 2014), and oceans (Constable et al. 2016). However, in transdisciplinary fields, only guidelines have been suggested thus far to identify EVs. Reyers et al. (2017) proposed criteria for the selection of EVs that link socioeconomic and environmental concerns for monitoring sustainable development goals. Guerra et al. (2019) defined a framework for identifying EVs that characterize human-nature dynamics in the context of conservation, and Balvanera et al. (2016) developed a pathway for identifying essential ecosystem service variables. Hence, a widespread consensus on a comprehensive list of EVs for SES monitoring is still lacking, although recent studies have provided valuable insights for identifying relevant variables. For instance, Frey (2017) suggested that in addition to SES sustainability, variables could also inform on other outcomes, such as resilience, social equity, or economic efficiency. Holzer et al. (2018) proposed that indicators collected across LTSER platforms might include qualitative social, political, and economic variables, e.g., sense of place, property ownership, or governance structures, to understand trends in quantitative variables, e.g., population density, ecosystem services, or biodiversity. Additionally, within the LTSER context, Dick et al. (2018) highlighted the importance of collecting social and biophysical data for addressing complex challenges that emerge from nature-society interactions, e.g., climate change, biodiversity loss, or environmental hazards. Additional studies that have developed spatially explicit maps of SESs provide multiple examples of relevant variables from which it is feasible to collect data to characterize SES dynamics (e.g., Alessa et al. 2008, Václavík et al. 2013, Vallejos et al. 2020).
In summary, it is crucial to advance toward an established list of relevant and feasible variables for characterizing and monitoring SESs that can be used in science, policy, and management. Developing such a list could foster a long-term coordinated socialecological monitoring network, allowing the intercomparability of place-based social-ecological research (Redman et al. 2004, Collins et al. 2011, Carpenter et al. 2012, Balvanera et al. 2017 and strengthening the production of generalizable knowledge on SESs across different regions of the world (Frey 2017). To our knowledge, the few integrative lists of SES variables have been built only from Ostrom's (2009) approach, and difficulties have been sometimes reported for their operationalization in empirical research (Delgado-Serrano and Ramos 2015). To progress in the development of a core set of integrative variables, it is important to provide new insights into the fundamental traits to characterize the functioning of SESs, i.e., how the system performs (Jax 2010). For this purpose, it is necessary to compile the variables used in previous studies and to incorporate the assessments of experts working in inter-and transdisciplinary fields (Redman et al. 2004). In this study, we aimed to develop a reference list of prioritized variables for characterizing and monitoring SESs. We provide evidence about the potential most relevant variables based on a comprehensive literature review, an iterative process driven by expert workshops, and researcher perceptions collected through online surveys. Fig. 1. Workflow. The main methodological steps are identified on the left, and their respective results are on the right. The boxes group together the methodological steps to indicate the two main stages of this study: (1) the development of a list of variables structured under a social-ecological system (SES) conceptual framework and (2) the prioritization of the list of variables.

Developing a comprehensive list of social-ecological system variables
The list of variables for characterizing and monitoring SESs was developed in four steps (Fig. 1). First, we performed a literature review to search for candidate variables. We also identified candidate conceptual frameworks to structure the list of variables and to depict the relationships among them. We searched Scopus for journal articles and book chapters with the following terms in their titles, keywords, or abstracts: "soci*-ecological system*" and ("map*" or "framework"). Then, we followed a "snowballing" approach (see van Oudenhoven et al. 2018) to identify additional papers that explicitly developed SES maps, SES conceptual frameworks, or were pivotal for understanding SES functioning (Appendix 1). From this search, we registered all variables and conceptual frameworks that were empirically used or theoretically introduced to characterize SESs. Second, we organized an initial workshop (November 2015) with experts on Earth system dynamics (carbon, water, energy, nutrient cycling) and sustainability science (ecosystem services, transdisciplinarity, translational ecology; see participants in Appendix 2) to develop a preliminary list of variables structured under an integrative conceptual framework. Experts analyzed the candidate variables and selected the most suitable framework. The variables were classified into a nested scheme of three SES components, and there were multiple dimensions within these components. Third, to complete the list of variables and to validate the structure of the dimensions and components, we conducted a preliminary online survey targeted at researchers with experience in SES science (August-December 2016; see acknowledgments). The survey (Appendix 3) introduced the list of variables classified into the dimensions and components and asked respondents to score each variable from 0 to 5 according to its relevance for characterizing and monitoring SESs. Scientists were also encouraged to suggest the addition or deletion of variables and to provide any other comments. These scores, suggestions, and comments were analyzed during a second scientific workshop (January 2017; see participants in Appendix 2) to improve the set of variables and dimensions. We then launched a final online survey (January-May 2017; Appendix 4) that was distributed to a new group of researchers with similar expertise in SES science (see acknowledgments). As in the preliminary survey, they were asked to score each variable from 0 to 5 and to provide comments and suggestions.
to each variable. The consensus was estimated as the difference between the maximum standard deviation of the scores found throughout the 149 variables and the standard deviation of the score for each variable (low differences indicated low consensus and high differences, high consensus). Then, the variables were separately ranked according to their percentile for relevance and consensus and grouped into five categories (four levels of priority and one nonpriority). Priority level 1 (top priority) included variables with relevance and consensus above the 90 th percentile; level 2 included variables between the 75 th and 90 th percentiles; level 3 included variables with relevance above the 75 th percentile but consensus between the 50 th and 75 th percentiles and vice versa; and finally, level 4 included variables with relevance and consensus between the 50 th and 75 th percentiles. The nonpriority category included variables with relevance and consensus below the 50 th percentile. Finally, to assess potential biases and gaps in the list of variables, we analyzed the additional suggestions and comments provided by researchers in both surveys (Fig. 1). This analysis was performed by annotating key words and organizing them through generalization in a conceptual map. We identified recurrent key words (addressed five or more times by respondents) as "featured topics."

Variables and dimensions to guide the characterization and monitoring of SESs
We developed a list of 149 variables structured in 13 dimensions within the three components of the SESs: the social system, the ecological system, and their interactions (Table A5.1, Appendix 5). We selected the Resilience Alliance conceptual framework (Resilience Alliance 2007) in the first workshop as the most pragmatic and illustrative framework to depict the structural relations among the dimensions and to guide more coordinated SES characterization and monitoring (Fig. 2). In the social system, three dimensions (human population dynamics, wellbeing and development, and governance) containing 36 variables were identified. In the ecological system, five dimensions (organic carbon dynamics, water dynamics, nutrient cycling, surface energy balance, and disturbance regime) containing 51 variables were identified. In the interactions between nature and people, five dimensions (ecosystem service supply, ecosystem disservice supply, ecosystem service demand, human actions on the environment, and social-ecological coupling) containing 62 variables were identified. The featured topics derived from the researchers' comments in the preliminary online survey that guided the development of the list of variables and dimensions are shown in Fig. A6.1, Appendix 6, as well as in the conceptual map in Appendix 7.

Prioritization of social-ecological variables based on scientist scoring
The analysis of the final survey revealed a significant positive linear relationship (n = 149; r = 0.82; p-value < 0.001) between the average relevance for characterizing and monitoring SESs obtained for each variable and the consensus observed across respondents (Fig. 3). A positive slope lower than one (m = 0.33; p-value < 0.001; root-mean-square error = 0.12) indicated that relevance increased faster than consensus. By applying the prioritization thresholds, 60 variables were considered relevant because they were included at one of the four priority levels (Table   1). Ten variables were included under priority level 1 (highest priority), representing the dimensions of nutrient cycling, disturbance regime (ecological system component), ecosystem service supply, human actions on the environment, and socialecological coupling (interaction component). Sixteen variables were considered at priority level 2, adding new dimensions such as well-being and development, governance (social system), water dynamics (ecological system), and ecosystem service demand (interaction component). Twenty-two variables constituted priority level 3, incorporating the dimensions human population dynamics (social system), organic carbon dynamics, and surface energy balance (ecological system). Finally, level 4 (lowest priority) added 12 variables, two of them belonging to the dimension of ecosystem disservice supply (interaction component). Thus, the prioritized variables represented all 13 dimensions proposed to characterize SES functioning, though we found it remarkable that no variables in the social system component reached priority level 1, reaching level 2 at the highest. Overall, 25% of the variables assessed for the social system were prioritized, 24% in the ecological system, and 48% for the interaction component. To explore in detail the relevance and consensus obtained for each variable, see Figs. A6.2 to A6.14 in Appendix 6 and Appendix 8.

Fig. 2.
Conceptual framework to guide the characterization and monitoring of social-ecological systems (SESs). The framework is structured in three components (social system, ecological system, and interactions between them) and 13 dimensions of SES functioning (modified from Resilience Alliance 2007).

Additional comments from the respondents
The analysis of respondents' comments and suggestions in the final survey allowed us to identify 14 featured topics indicating potential biases and gaps in the list of variables ( Fig. 4 and Appendix 7). In the social system, several researchers emphasized the importance of "social equity" and "living conditions" to characterize the well-being and development dimension. In the ecological system, "biodiversity" was the most featured topic, which was considered the foundation for explaining the supply of provisioning, regulating, and cultural ecosystem services. Respondents also argued that the water dynamics dimension should be mainly based on the characterization of the "water balance," with some additional variables concerning water and soil salinity and seasonality. Within the interactions, the importance of measuring the "strength of links between people and nature" was the most addressed topic. Within this scope, other related featured topics were "resource consumption patterns," the . Relevance and consensus obtained by variables for characterizing and monitoring social-ecological systems (SESs) in the final survey. Relevance was evaluated as the mean of the scores assigned by experts to each variable. The consensus was estimated as the difference between the maximum standard deviation of the scores found throughout the 149 variables and the standard deviation of the score for each variable (low differences indicated low consensus and high differences, high consensus). Squares, circles, and plus signs identify the variables belonging to the social system, ecological system, and interaction components, respectively. Horizontal and vertical lines represent the 25 th , 50 th , 75 th , and 90 th percentiles of relevance and consensus. Boxes over the grid illustrate the clustering of the variables by priority levels. The red box (priority level 1) includes those variables with relevance and consensus above the 90 th percentile; the green box (level 2) includes those variables with both values between the 75 th and 90 th percentiles; the yellow box (level 3) includes those with relevance above the 75 th percentile but consensus between the 50 th and 75 th percentiles and vice versa; and the blue box (level 4) includes variables with relevance and consensus between the 50 th and 75 th percentiles. At the bottom right of the figure, the equation of the regression line, the significance of the line slope (p-value) and the rootmean-square error (RMSE) are indicated, as are the number of variables (n), the Pearson's correlation coefficient (r), and its significance (p-value).
"cultural value of nature," "cultural ecosystem service demand," "local ecological knowledge," and the "beneficial human actions on the environment." Other highlighted issues were transversal to the three SES components. Some researchers argued that all "variables should reflect the underlying processes and functions" occurring in SESs, instead of outcomes or symptoms of their functioning. In addition, the need to consider more variables related to "energy fluxes" as indicators of system complexity was also suggested. Finally, researchers also stated that variable relevance might be "context-dependent" and that SES complexity makes it "difficult to assess some variables." An extended version of Fig. 4 with the whole list of topics is available in Fig. A6.15, Appendix 6.

DISCUSSION
With this study, we contributed to the identification of a common core set of relevant variables for the study and monitoring of SESs by providing a reference list of 60 variables, which were structured in 13 dimensions of SES functioning embedded in the social, ecological, and interaction components of the SES (Fig. 2). The use of such a nested framework contributes to understanding the relationships among variables, aims to maintain the holistic approach in the study of SESs, and promotes transdisciplinary communication by acting as a boundary object (Ostrom 2009, van Oudenhoven et al. 2018. The variables were classified into four levels of priority according to researcher consensus on their relevance ( Fig. 3 and Table 1) to facilitate their adaptation to the data availability, context, and sociopolitical needs. The prioritization revealed the crucial role that socialecological interactions have in characterizing SES complexity (Liu et al. 2007, Carpenter et al. 2009) but also showed that all the dimensions of social-ecological functioning are necessary to disentangle SES dynamics (Table 1). In general, the development of reference lists of variables is an emerging need in sustainability research to foster the collection of structured, long-term, coordinated core datasets across SESs (Frey 2017, Holzer et al. Table 1. List of prioritized variables for characterizing and monitoring social-ecological systems (SESs). The list is structured into 13 dimensions across the three components of a SES (see Fig. 2). Priority level 1 includes variables with relevance and consensus above the 90th percentile; level 2 includes variables with both values between the 75th and 90th percentile; level 3 contains those variables whose relevance was above the 75th percentile and consensus between the 50th and 75th percentiles and vice versa; and finally, level 4 includes those variables with relevance and consensus between the 50th and 75th percentiles. An extended version of this · Cropland production (P) · Livestock production (P) · Surface and groundwater sources for drinking (P) · Hydrological cycle and water flow maintenance (R) · Surface and groundwater sources for nondrinking purposes (P) · Local climate regulation (R) · Pest and disease control (R) · Pollination and seed dispersal (R) · Chemical conditions maintenance of freshwaters and salt waters (R) Ecosystem disservice supply 2 · Abiotic-economic (e.g., droughts, fires) · Bioeconomic (e.g., biological invasions) Ecosystem service demand · Appropriation of land for agriculture · Energy use level · Water use level · Water use for irrigated crops · Material use level · Human appropriation of net primary production (HANPP) Human actions on the environment · Land cover/land use change · Land use intensity · Eutrophication of water bodies · Land protection · Pollution · Soil erosion · Anthropogenic water management · Net CO2 flux · Territorial connectivity Social-ecological coupling · Local natural capital dependence · Access to natural and seminatural areas · Biocapacity · Import/export rates of agricultural products · Renewable energy use Fig. 4. Featured topics (addressed by five or more respondents in different dimensions) related to potential biases and gaps in the list of variables identified from comments and suggestions in the final survey. Black, white, and gray bars represent the social system, ecological system, and interaction components, respectively, while striped bars reflect issues that are transversal to the whole conceptual framework. (See also these topics in the conceptual map of Appendix 7).

Insights to address existing barriers in SES research
The list of variables presented in this study offered new perspectives for addressing the main barriers, i.e., applicability to place-based research, representativeness of SES complexity, and feasibility for monitoring, detected in operationalizing existing lists to assess SESs (e.g., Ostrom 2009, McGinnis and Ostrom 2014, Delgado-Serrano and Ramos 2015, Frey 2017. First, regarding their applicability for place-based research, according to van Oudenhoven et al. (2018), variables not only need to be credible, i.e., scientifically sound based on expert judgment, scientific literature, and a conceptual framework, but also practically feasible for collection. For instance, Ostrom's list of variables, which was conceived to diagnose the sustainability of SESs (Ostrom 2009), has sometimes been considered too abstract and general to characterize concrete systems (Cox 2014, Delgado-Serrano and Ramos 2015, Hinkel et al. 2015, Leslie et al. 2015. To overcome such limitations, we emphasized the selection of variables easily derivable from primary data that have been used in previous research for the spatially explicit mapping of SESs (Appendix 1; Table A5.3, Appendix 5). In addition, the list of variables and the conceptual framework must offer certain flexibility to be adapted to the diversity of contexts and scales of analysis and to data availability (McGinnis and Ostrom 2014). The Ostrom SES framework presents a hierarchical structure at different levels (tiers), with variables and subvariables that could be adapted depending on the type of SES (Delgado-Serrano and Ramos 2015) but that lack any guidance on their relevance. In our study, we not only hierarchically structured the variables under the dimensions and components of SESs but also distributed them into priority levels according to their agreed relevance for characterizing SESs. By doing so, we provide guidance for adapting variable selection according to the research context while retaining consistency regarding the relevance and representativeness of variables across SES dimensions.
Second, regarding their representativeness of SES complexity, variables not only need to provide information on the different "pieces" of the system but also must help to understand the linkages among such "pieces" (Ostrom 2009). To achieve this goal, embedding variables within a nested conceptual framework helps to organize them across components and hierarchical levels while depicting the structural relationships between them (Frey 2017, Ostrom 2009, McGinnis and Ostrom 2014. For instance, Ostrom's SES framework uses an anthropocentric perspective of SESs, where variables that are supposed to focus on the ecological subsystem also have a social origin or reflect the interaction between humans and nature (Binder et al. 2013). However, if most variables make sense only if humans exist, it implies that there exists an unbalanced representation among the social, ecological, and interaction variables, which is acknowledged as a key principle for addressing SES complexity (Liu et al. 2007, Resilience Alliance 2007, Reyers et al. 2017. Our proposal provides a scheme that categorizes all variables into 13 expertvalidated dimensions embedded into the three key components of a SES, i.e., social system, ecological system, and interactions. The variables for characterizing the ecological system followed an "ecocentric" perspective (sensu Binder et al. 2013) and were structured into five dimensions, where the system and its processes were analyzed independently of their links to humans. For the social system, our variables focused on understanding human population dynamics, well-being and development, and governance dimensions without considering ecological processes. Finally, for the interactions between humans and nature, similar to Ostrom (2009), our variables addressed the reciprocity between the social and ecological systems (Binder et al. 2013). However, we suggested a more detailed structure for the variables, which we divided into five dimensions, depending on the type and direction of the interactions: (a) from the ecological to the social system (ecosystem service and disservice supply), (b) from the social to the ecological system (ecosystem service demand and human actions on the environment), and (c) bidirectionally between the social and the ecological system (social-ecological coupling). We recognize that relying on a single framework might be unrealistic, but understanding and generalizing the complexity of SESs requires common hierarchical analytical structures that comprehensively integrate the multiple dimensions and components of SESs (Reyers et al. 2017, Magliocca et al. 2018. Third, regarding the feasibility of the variables for long-term monitoring (van Oudenhoven et al. 2018), our list facilitates SES characterization at the system level, i.e., it focuses on the macrolevels according to Binder et al. (2013) to integrate properties of the SES components as a whole. Aggregated variables at the system level have been clearly more used to characterize, map, and track SESs than variables collected at the individual level, i.e., variables focused on the microlevels according to Binder et al. (2013) to measure properties of the SES individual building blocks, e.g., plant, animal, individual producer, user, or consumer (see examples in Table A5.3). In fact, even those SES mapping strategies based on Ostrom's framework, which combines both system-and individual-level perspectives, i.e., macro-and microlevels according to Binder et al. (2013), have https://www.ecologyandsociety.org/vol25/iss3/art1/ only used system level metrics (e.g., , Rocha et al. 2020. Several studies show that system-level characterizations can better inform on social-ecological processes from local to global scales (e.g., Václavík et al. 2013, Levers et al. 2018, Vallejos et al. 2020 and could help to overcome current limitations to upscale place-based research for the coproduction of generalizable knowledge on SES (Balvanera et al. 2017).

Potential biases and gaps in the list of variables
The analysis of the researchers' comments revealed potential conceptual biases introduced by the proposed framework during the construction of the list of variables (Fig. 4). In the interaction component, a majority of comments highlighted that sociocultural values and identities might be underrepresented and that the variables addressing the "strength of the links between people and nature" and the "cultural value of nature" could be enhanced, for instance, by incorporating the variable "local ecological knowledge." However, interestingly, cultural ecosystem service variables (following the categories of the Common International Classification of Ecosystem Services, CICES; Haines-Young and Potschin 2013) were not prioritized by researchers during the survey (Table A5.2, Appendix 5; Appendix 8). Although these findings may seem contradictory, they align with new insights into the nature's contributions to people (NCP) paradigm  and the plurality of values associated with these contributions (UNEP 2015, Pascual et al. 2017. Under the new NCP paradigm, culture plays a central role in defining all links between people and nature . Thus, further lists of SES variables should expand the ecosystem service supply dimension by giving culture and traditional/indigenous knowledge a more transversal role across ecosystem services categories, beyond the independent cultural category of CICES and the Millennium Assessment (MA 2005). Furthermore, enhancing the characterization of the cultural contexts and identities goes further for the instrumental values of ecosystem services and NCP by incorporating those values that emerge from individual and collective relationships of humans with nature (Chan et al. 2018). To address these "relational values," new variables, such as sense of belonging, responsibility toward nature, or maintenance of traditions (Chan et al. 2016), may be added to the list. In the ecological system component, the explicit role of biodiversity might also be underrepresented because many comments suggested the addition of more biodiversity variables or of a whole biodiversity dimension within this component. Given the role of biodiversity in SESs as the natural capital that supports social metabolism (Costanza et al. 1997) and the biocentric conservationist tradition (Mace 2014), we agree that biodiversity could be explicitly named in the framework. However, we initially excluded the structural and compositional biodiversity facets because of their slower response to disturbances compared to functional variables (McNaughton et al. 1989, Milchunas andLauenroth 1995). Instead, we focused on the functional aspects of biodiversity at the ecosystem level, such as the candidates to become essential biodiversity variables for the ecosystem function class (e.g., Pereira et al. 2013, Pettorelli et al. 2018. We are also aware of additional sources of potential methodological biases. On the one hand, the way that the variables were sorted in our framework during the survey could have influenced respondents in assigning priority levels. By displaying the variables sorted into dimensions, we aimed to facilitate the completion of the survey. We are aware that a random display or other sorting could have led to different variable scores. However, this impact may have been low because there was no significant correlation between the priority scores and variable order in the online survey. On the other hand, because the field of expertise of most respondents was sustainability science and ecology (Appendix 9), the social variables might have received lower scores than expected. Indeed, the social variables never reached the highest priority level (level 1; Table A5.2, Appendix 5) despite their importance for human well-being and for explaining the form and intensity of human-nature interactions, e.g., education and population density, respectively Ramankutty 2008, Hamann et al. 2016). Most inter-and transdisciplinary efforts in social-ecology and sustainability science come from ecology (Lowe et al. 2009, Holzer et al. 2019), but a wide range of perspectives still exist among ecologists for integrating concepts and methods from social science. This disparity of perspectives might be because some researchers consider ecology as a basic science that studies wild nature (where people are only the "ecological audience"), others see it as an instrument for guiding ecosystem and species management (treating people as "ecological agents"), and still others view it as a discipline that considers human societies to be integrated in ecosystems (people as "ecological subjects/objects"; Lowe et al. 2009, Mace 2014). Indeed, these perceptions of ecology have been evidenced throughout the development and implementation of the longterm social-ecological monitoring network, which mainly originated from ecological monitoring and research. Despite the adoption of a new social-ecological paradigm, the network continues to monitor primarily ecological processes, although it is progressing toward incorporating economic and social data and conducting more germane transdisciplinary research , Angelstam et al. 2019. In our study, the potential coexistence of these three perceptions among the surveyed researchers could be the basis of the lack of consensus around the most relevant social variables. This highlights the need to strengthen cooperation between natural and social scientists and experts to lead to a truly integrated approach for long-term socialecological research . Finally, many scientists have reported difficulties in scoring the variables without considering a specific SES, arguing that variable relevance is context dependent. Although biodiversity, climate, oceans, or sustainable development goal variables may have more evident global perspectives, this is not easily applicable to SES variables given the place-based nature of SES research (Carpenter et al. 2012). All these potential biases should be considered when using our list of variables and formally analyzing them in future assessments.

Toward the definition of essential variables for social-ecological systems
The development of essential variables (EVs) that harmonize global observation networks is a priority for tracking changes and coordinating monitoring efforts (e.g., Pereira et al. 2013, Bojinski et al. 2014, Constable et al. 2016. Despite the call from https://www.ecologyandsociety.org/vol25/iss3/art1/ sustainability science to extend this systemic thinking to areas of interaction between the social and the biophysical domains, building a list of essential social-ecological system variables is still needed (Reyers et al. 2017). The set of dimensions and variables developed here can contribute to creating a common structure to study SESs and to starting to work toward such essential variables. Because the variables and dimensions were based on consensual expert knowledge, their credibility, salience, and feasibility were reaffirmed (van Oudenhoven et al. 2018). In addition, fundamental steps in EV development were followed in the codesign process (Reyers et al. 2017): (1) adoption, through an expert-driven process, of a conceptual model of SESs functioning, representing the social and ecological systems as well as the interactions between them; (2) identification of the broad categories and disaggregated inputs of candidate variables; (3) refining and prioritization of variables based on the consensus on their relevance; and all this by means of (4) an iterative procedure fed by scientific expert knowledge obtained from workshops and online surveys. However, given the preliminary nature of our exercise, further work is needed to build a global consensus around a set of EVs for the study of SESs. For instance, new surveys should address the potential biases and limitations outlined above, for instance (1) by explicitly considering the role of biodiversity and of relational values about NCP; (2) by having a greater and more balanced number of respondents (particularly the inclusion of social scientists); and (3) by reporting on the most frequently relevant variables in relation to specific place-based social-ecological contexts.
To further develop EVs for SESs, finding common aspects and variables among the existing lists could also help to establish a baseline. Some variables suggested in Ostrom's (2009) and Frey's (2017) lists were also relevant in our study. The most common aspects were found for the interaction component. For instance, the harvesting variable on Ostrom's list was related to human appropriation of net primary production, material use, water use, or energy use on our list. Similarly, pollution patterns on Ostrom's list were related to eutrophication of water or net CO 2 flux on our list; constructed facilities on Ostrom's list and accessibility on Frey's list were related to territorial connectivity, access to natural areas, or anthropogenic water management on our list; and importance of resources on Ostrom's list and dependency on resources on Frey's list with dependence on local natural capital on our list. In the social system, economic development and socioeconomic attributes (Ostrom 2009) were associated with poverty, educational level, or social equity variables on our list, and number of actors (Ostrom 2009) with population density. Similarly, governance-related variables, such as conflicts and political stability, were included on both Ostrom's list and our list, while Frey (2017) considered conflict management as a crucial aspect for the stability of rule systems and resource use. In the ecological system, Ostrom's (2009), Frey's (2017, and our list converged on including climate characteristics and primary productivity or the regeneration rate of resources.
In addition, some of our prioritized variables from the ecological and interaction components of SESs are related to six of the nine major environmental challenges listed in the planetary boundaries framework (Rockström et al. 2009, Steffen et al. 2015. For instance, the monitoring of net solar radiation and net CO 2 flux could provide information to assess "climate change" and "atmospheric aerosol loading"; information on biological invasions, pest outbreak occurrence, and ecosystem composition by plant functional types to assess "changes in biosphere integrity"; measuring nitrogen deposition and eutrophication of water to evaluate interferences with "biogeochemical flows"; the appropriation of land for agriculture and land use intensity for "land-system change"; and finally, water use level and water use for irrigated crops to assess "freshwater use." From a general perspective, additional steps should be given to foster the institutionalization of the development and implementation of essential SES variables (see Pereira et al. 2013, Bojinski et al. 2014, Constable et al. 2016, Reyers et al. 2017. As a first step, the compliance of the variables with the criteria to be considered essential should be thoroughly checked, for instance, to be (i) state variables, sensitive for long-term monitoring of changes; (ii) representative for the system level, between primary observations and indicators; (iii) flexible to adapt to multiple monitoring programs; and (iv) feasible to observe and derive and to be scaled to meet local, regional or subglobal needs. Second, consensus should be built and coordinated to align the development of the variable list with research and policy needs by setting an open platform for scientist, policy maker, and stakeholder cooperation. Third, the learning loop should be optimized to refine and stabilize the list of EVs by establishing a transparent process with specific targets and time lines to plan the development of the list and track the updates. Finally, to increase the global efficiency of Earth monitoring systems, the interconnection of the EVs that may emerge from our list with other sets of EVs (for biodiversity, climate, oceans, etc.) should be coordinated.

CONCLUSION
The development of reference lists of variables is an emerging need in sustainability research to foster the systematic collection of comprehensive and coordinated datasets of SESs and to enhance our ability to study SESs across time and space. These lists of variables structured under a conceptual framework provide a common language that facilitates comparisons and the generalization of knowledge from empirical studies. Although the development of such lists in specific fields of Earth systems (climate, biodiversity, oceans) has progressed significantly in recent years, integrative approaches for SESs are still scarce. With this study, we contributed to the identification of a common core set of variables for the characterization and monitoring of SESs. Our 60-variable list gathered relevant traits and processes of the SES from scientific literature reviews and expert knowledge. This list was embedded in a framework of 13 dimensions across the three key components of the SES (social system, ecological system, and the interactions between them) to help maintain an integrative approach when working with SESs. In addition, variables were classified into priority levels to provide more flexibility in their application to place-based research. Throughout this process, new insights have arisen that could contribute to overcoming existing barriers in the operationalization of lists of variables in the study of SESs, such as the applicability to place-based research, the capacity to deal with SES complexity, or the feasibility for long-term monitoring of social-ecological dynamics. Our list of variables may constitute a preliminary step in the direction of identifying essential variables for SESs, whose further development will provide an opportunity to boost the https://www.ecologyandsociety.org/vol25/iss3/art1/ long-term social-ecological research network. This could strengthen our capacity to respond to global change challenges, extend systemic thinking to the field of human-nature interactions, and foster sustainability sciences through more efficient operationalization of the social-ecological approach.

Essential variables to describe the functioning of Social-Ecological Systems Introduction
We aim to integrate biophysical and social processes to produce a functional characterization and mapping of social-ecological systems at the regional scale and landscape level. This survey aims to agree on a set of 'Essential Social-Ecological Functional Variables' (ESEFVs) to be used in such

Participating Institutions
Appendix 3. Preliminary online survey process.
A list of candidate variables is structured in three 'Components' of the social-ecological system (Social System, Ecosystem and Interactions) and each Component into several 'Functional Dimensions' (dimensions of the social system functioning, dimensions of ecosystem functioning, and dimensions of the interactions between the social system and the ecosystems). Possible indicators are shown in some cases only to exemplify, but the answers should focus on the variables (whatever the indicator is).

Personal data (optional)
In any case, your answers will be treated as confidential 8. In your opinion, which variables that describe human population dynamics are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 14. Do you consider evapotranspiration as essential to characterize social-ecological systems functioning? Please, punctuate this variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 18. In your opinion, which variables that describe nutrient cycling are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 22. In your opinion, which variables that describe provisioning services supply are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila.
No essential 1 2 3 4 5 Agricultural production Livestock production Wild plants, algae and their outputs for food Wild animals and their outputs for food Surface and ground water sources for drinking Surface and ground water sources for non-drinking purposes Fibres and other materials from plants, algae and animals for direct use or processing Biomass-based energy sources 23. In your opinion, which variables that describe regulation & maintenance services supply are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 24. In your opinion, which variables that describe cultural services supply are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 26. In your opinion, which variables that describe ecosystem disservices supply are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 28. In your opinion, which variables that describe the human capture of ecosystem goods and services are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 30. In your opinion, which variables that describe the human pressure on environment are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 32. In your opinion, which variables that describe the degree of connection of a community to its local environment are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila.

Participating Institutions
Appendix 4. Final online survey The list of candidate variables is structured in three 'Components' of the social-ecological system (Social System, Ecosystem and Interactions) and each Component into several 'Functional Dimensions' (dimensions of the social system functioning, dimensions of ecosystem functioning, and dimensions of the interactions between the social system and the ecosystem). Possible indicators are shown in some cases only to exemplify, but the answers should focus on the variables.
We ask you to punctuate each variable according to its relevance to characterize the functioning of social-ecological systems. A key aspect to deal with is the issue of context-dependence. We are aware of the difficulties to assess the relevance of proposed variables without bearing in mind any specific social-ecological system. However, we call for a common effort to identify those variables that better explain the differences among social-ecological systems across the world.
We consider as essential those variables that encompass and integrate critical processes to characterize the functioning of social-ecological systems. They should be coherent and appropriate for comparing across social-ecological systems diversity. Spatially, these variables aim to target the ecosystem level and the human community level. Ideally, they should be viable for regional or global implementation in monitoring programs, regional land-use planning, and sustainability and resilience assessment. Our final goal is to integrate both biophysical and social processes to produce a functional characterization and mapping of social-ecological systems at the regional scale and landscape level.
Please, feel free to visit the webpage of the E&SEFT Project: "Ecosystem & Socio-Ecosystem Functional Types: integrating biophysical and social functions to characterize and map the ecosystems of the Anthropocene" (http://functionaltypes.caescg.org/) to know more about project goals, scientists involved, and other partners. In this webpage you can also learn more about the variables included in this survey (selection process, definitions, etc.).
*Important: if you are viewing this survey through your mobile phone, we recommend that you use it in horizontal position for better visualization.

Personal data (optional)
In any case, your answers will be treated as confidential 8. In your opinion, which variables that describe human population dynamics are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 10. In your opinion, which variables that describe human well-being and development are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 12. In your opinion, which variables that describe regional governance are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila.
No essential 1 2 3 4 5 Institutional diversity (degree of polycentrism and nesting level in government, with efficient horizontal and vertical coordination) Agenda effectiveness (degree in which the agenda is adequately formulated and assessed to achieve specific goals and have a popular understanding) Stakeholders participation in decision making (degree of stakeholders inclusiveness, with an adequate leadership arrangement and commitment to group and purpose) Internal capacity (degree of sufficiency of resources -money, information and expertise, authority and legitimacy-to achieve success on a specific goal) External capacity (skills and reach of the government to connect toat both the national and international levels-and secure external resources to support regional goals) Implementation experience (level of experience addressing regional goals and degree of institutionalization of these experience in policies and processes) Political stability Corruption level Current conflicts (e.g.: armed conflicts, political violence) Candidate variables from 2 to 6 have been included following Foster & Barnes (2012) proposal of indicators for regional governance. 14. In your opinion, which variables that describe carbon dynamics are essential to characterize social-ecological systems functioning? Please, punctuate this variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 16. In your opinion, which variables that describe water dynamics are essential to characterize social-ecological systems functioning? Please, punctuate this variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 18. In your opinion, which variables that describe surface energy balance are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 20. In your opinion, which variables that describe nutrient cycling are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 22. In your opinion, which variables that describe disturance regime are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 24. In your opinion, which variables that describe provisioning services supply are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 26. In your opinion, which variables that describe cultural services supply are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 28. In your opinion, which variables that describe ecosystem disservices supply are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila.

Would you add/modify any variable of governance to better describe social-ecological
No essential 1 2 3 4 5 Bio-economic (e.g.: biological invasions, agricultural and fisheries pests and diseases incidence, red tydes) Abiotic-economic (e.g.: droughts and fires occurrence, siltation, leaching of nutrients) Bio-health (e.g.: human diseases incidence from pathogens, allergens) Abiotic-health (e.g.: flood and storm events occurrence ) Bio-cultural (e.g.: bird droppings on outdoor sculptures, tree roots cracking pavements) Abiotic-cultural (e.g.: soil erosion rates, mud/landslide scar events, unpleasant odours from rotting organic matter) It is noted that this candidate variables express the incidence of different kinds of harmful events. For simplicity, they have been classified according to their origin and primary dimension of human wellbeing affected, following  approach.
29. Would you add/modify any variable of ecosystem disservices supply to better describe social-ecological systems functioning? Please specify: Dimension 3c. Ecosystem services demand (You are in: Component 3. Interactions) 30. In your opinion, which variables that describe the human capture of ecosystem goods and services are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 32. In your opinion, which variables that describe the human actions on the environment are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. 34. In your opinion, which variables that describe the degree of connection of a community to its local environment are essential to characterize social-ecological systems functioning? Please, punctuate each variable according to its relevance for being considered as 'Essential Social-Ecological Functional Variable' (from 1 "less essential" to 5 "more essential") Marca solo un óvalo por fila. Cropland production (P) Cropland production (P) Livestock production (P) Livestock production (P) Surface and groundwater sources for drinking (P) Surface and groundwater sources for drinking (P) Surface and ground water sources for nondrinking purposes (P) Surface and ground water sources for nondrinking purposes (P) Biomass-based energy sources (P)

Biomass-based energy sources (P)
Fibres and other materials from plants, algae and animals for direct use or processing ( with examples and explanations). The list is structured into 13 dimensions across the three components of a SES (Fig. 2 in the paper). Priority level 1 (top priority) includes variables with relevance and consensus above the 90 th percentile; level 2 includes variables between the 75 th and 90 th percentiles; level 3 includes variables with relevance above the 75 th percentile but consensus between the 50 th and 75 th percentiles and vice versa; and finally, level 4 includes variables with relevance and consensus between the 50 th and 75 th percentiles. The nonpriority category includes variables with relevance and consensus below the 50 th percentile. Agenda effectiveness (degree in which the agenda is adequately formulated and assessed to achieve specific goals and have a popular understanding) 1 External capacity (skills and reach of the government to connect to -at both the national and international levelsand secure external resources to support regional goals) 1 Implementation experience (level of experience addressing regional goals and degree of institutionalization of these experience in policies and processes) 1 Institutional diversity (degree of polycentrism and nesting level in government, with efficient horizontal and vertical coordination) Internal capacity (degree of sufficiency of resources -money, information and expertise, authority and legitimacy-to achieve success on a specific goal) Stakeholders participation in decision making (degree of stakeholder's inclusiveness, with an adequate leadership arrangement and commitment to group and purpose) Weight of traditional (vs. intensive) agricultural sector in the economy † P = provisioning services; R = regulating services; C = cultural services.
In this paper, ecosystem disservices are defined as "the ecosystem generated functions, processes and attributes that result in perceived or actual negative impacts on human wellbeing." We based on  classification to distinguish among 6 categories of ecosystem disservices, according to their origin (biological or abiotic) and the nature of their impacts on human wellbeing (economic; physical and mental health and safety; aesthetics and culture): bio-economic, abiotic-economic, bio-health, abiotic-health, bio-cultural, abiotic-cultural. Examples of ecosystem disservices for each category are include in the Table above.  Figures A6.2 to A6.14. Detail view of the relationship between average relevance and consensus obtained by the variables belonging to each dimension of social-ecological system functioning. Relevance was evaluated as the mean of the scores assigned by experts to each variable. The consensus was estimated as the difference between the maximum standard deviation of the scores found throughout the 149 variables and the standard deviation of the score for each variable (low differences indicated low consensus and high differences, high consensus). Horizontal and vertical lines represent the 25th, 50th, 75 th and 90 th percentiles of relevance and consensus for the whole set of variables belonging to the 13 dimensions of social-ecological functioning. Boxes over the grid illustrate the clustering of the variables by priority levels. The red box (priority level 1) includes those variables with relevance and consensus above the 90 th percentile; the green box (level 2) includes those variables with both values between the 75 th and 90 th percentiles; the yellow box (level 3) includes those with relevance above the 75 th percentile but consensus between the 50 th and 75 th percentiles and vice versa; and the blue box (level 4) includes variables with relevance and consensus between the 50 th and 75 th percentiles. At the bottom right of each figure, the equation of the regression line, the significance of the line slope (p-value) and the root-mean-square error (RMSE) are indicated, as are the number of variables (n), the Spearman's correlation coefficient (r) and its significance (pvalue).  Figure A6.15. Extended version of Figure 4 in the manuscript. Featured topics addressed by respondents related to potential biases and gaps in the list of variables identified from comments and suggestions in the final survey. Black, white and gray bars represent the social system, ecological system and interaction components, respectively, while stripped bars reflect issues that are transversal to the whole conceptual framework. (See also these topics in the conceptual map of Appendix 7). Appendix 7. Conceptual map with keywords annotated from comments and suggestions provided by respondents in both surveys (zoom in to see in detail). The concepts are shown hierarchically interlinked, and structured into dimensions across the three components of a social-ecological system (social system, ecological system and interactions between them). We used this conceptual map to improve the preliminary list of variables and dimensions after the preliminary survey, and to assess the potential biases and gaps in the improved list of variables after the final survey.