Research

INTRODUCTION

Many of the world’s streams, rivers, lakes, and reservoirs are becoming saltier, a process known as the freshwater salinization syndrome (FSS; Kaushal et al. 2019, 2021). The FSS poses a direct threat to critical freshwater ecosystems (Findlay and Kelly 2011, Schuler et al. 2019, Velasco et al. 2019) agricultural productivity (Schwabe et al. 2006, Welle and Mauter 2017), and drinking water supplies (Corsi et al. 2010, Bhide et al. 2021), and undermines many of the United Nations’ Sustainable Development Goals (Flörke et al. 2019). Origins of the FSS include the application of deicers and anti-icers to roads and parking lots in northern climates, salt-laden discharges from sewage treatment plants, mining and energy extraction operations, and agricultural return flows, to name a few (Wilkinson 2005, Steele and Aitkenhead-Peterson 2011, Welle and Mauter 2017, Schwabe et al. 2020, Bhide et al. 2021, Grant et al. 2022, Hintz et al. 2022, Kaushal et al. 2023a).

In many countries, existing water quality regulations are not well suited to manage the FSS, in part because of the ion-, site-, and ecosystem-specific nature of the problem (Cañedo-Argüelles et al. 2013, Schuler et al. 2019, Grant et al. 2022). Few federal regulations presently exist in the United States for individual salt ions (Schuler et al. 2019). Those that do exist include acute and chronic limits for chloride intended to protect aquatic freshwater species, as well as secondary (nonmandatory) guidelines for sulfate, fluoride, and chloride in drinking water (EPA 1988, Schuler et al. 2019, State of Virginia 2019). Every five years, the U.S. Environmental Protection Agency (EPA) publishes a candidate list of unregulated contaminants that may ultimately warrant regulation under the Safe Drinking Water Act. Salt ions such as sodium and sulfate have made the list in prior years (e.g., in 1998) but, at the time, were not determined to present a meaningful opportunity to mitigate health risk and were therefore not regulated (EPA 2003a). This prior determination may reduce the likelihood of federal regulation in future years, even though salinization has continued to worsen (Kaushal et al. 2019, 2021, 2023a,b, Olson 2019, Bhide et al. 2021), and concentrations of specific salt ions that were previously considered low (e.g., 29–60 mg/L of sodium) are increasingly linked to adverse health consequences such as preeclampsia (Thompson et al. 2022).

Given the present challenges associated with regulation, we focus here on locally tailored, stakeholder-driven approaches for addressing the FSS. In urban water systems, stakeholder-driven, collective management represents a radical departure from current strategies for addressing the FSS, which are often source specific and siloed (Kaushal et al. 2018, Bhide et al. 2021), a consequence of the tendency to manage key water subsystems separately (e.g., stormwater, drinking water, wastewater, surface water, and groundwater; Rahaman and Varis 2005). In recent years, urban water practitioners have become increasingly open to integrated management approaches as part of an emerging “One Water” paradigm (Paul et al. 2022). Whether (and how) this openness will translate into new governance arrangements, however, remains unclear.

We argue that Elinor Ostrom’s social-ecological system (SES) framework (Ostrom 2009) provides a roadmap for diagnosing (and perhaps fostering) the conditions under which collective management of the FSS is likely to emerge. Ostrom’s SES contains 10 second-level variables that have previously been shown (in field studies and meta-analyses) to significantly influence the emergence of collective management arrangements (Ostrom 2009, Freeman et al. 2020, Shalsi et al. 2022, Villamayor-Tomas et al. 2022). Work by Grant et al. (2022) calls attention to three of these variables as potential barriers to stakeholder-driven management of the FSS: collective-choice rules, system predictability, and understanding the SES. Here, we focus on one of those variables, understanding the SES, and the ability of novel approaches such as fuzzy cognitive maps (FCMs) both to characterize and to enrich systems knowledge, reducing barriers to collective management.

FCMs are mental models of how a system operates based on its core components (system variables) and the causal links between them (Özesmi and Özesmi 2003). They are often used in the context of participatory modeling as part of a learning process that engages the implicit and explicit knowledge of stakeholders to create formalized and shared representations of reality (Olofsson et al. 2023). This usage makes FCMs well suited for deepening what Emerson et al. (2012) call the “collaborative dynamics” of collaborative governance regimes, which are inclusive of collective management arrangements. Collaborative dynamics involve three interacting components: principled engagement, shared motivation, and capacity for joint action. FCMs are particularly relevant to principled engagement because they can facilitate agreement on the concepts and terminology used to describe problems and lead to the identification of shared interests and objectives (Özesmi and Özesmi 2003, Olofsson et al. 2023). These capabilities complement those of existing deliberative or dialogic approaches for facilitating principled engagement (e.g., soft systems methodology, joint fact-finding; Karl et al. 2007, Checkland and Poulter 2020), particularly with respect to positional bargaining, whereby FCMs help shift stakeholder dialogue from fixed, opposing stances (“my problem definition vs. yours”) to the network of relationships that bridge them (“how our views overlap and interact”; Voinov and Bousquet 2010).

FCMs also show promise as an approach for countering cognitive biases and the limitations of bounded rationality, with some studies suggesting that individual stakeholder FCMs can be aggregated to generate a collective model that captures “the wisdom of the crowd” and is therefore more representative of the real-world SES (Aminpour et al. 2020, 2021, Knox et al. 2023). If true, this property could make collective FCMs a valuable tool for shared learning, particularly in systems in which data are limited or problems are viewed in context with social, cultural, or political elements that can be captured by FCMs but are difficult to quantify explicitly, limiting their representation in traditional model frameworks (Özesmi and Özesmi 2003).

Here, we explore the capacity of FCMs to characterize stakeholder understanding of the SES for freshwater salinization, focusing on (1) their ability to reveal barriers and opportunities for stakeholder-driven, bottom-up management of the FSS (capabilities relevant to principled engagement), and (2) their ability to capture physically meaningful information about the SES when aggregated to represent collective perspectives (capabilities relevant to establishing mutual trust and shared learning). We find that perceptions about individual salt sources, their societal and environmental impacts, and how they might be managed, vary, revealing areas of agreement for which bottom-up management is likely (winter salt sources), as well as divergent views (sewershed salt sources, governance approaches) that could represent barriers to collective action or be leveraged as opportunities for shared learning. Although aggregating FCMs does appear to improve their capacity to capture average biophysical conditions, the range of those conditions are better represented by individual or small-group FCMs. This result suggests that evaluating FCMs at different levels of aggregation (individual through collective) adds value, providing an opportunity for perspective taking (“What conditions do our different perspectives capture?”) and learning. While our emphasis is on the FSS, the process we describe for upscaling individual knowledge to systems-level understanding of an SES is not salinization-specific and could potentially inform management approaches for many emerging challenges.

Field site

The Occoquan reservoir is a drinking water resource for up to one million people in northern Virginia, located at the southern end of the Boston-Washington megaregion, one of the world’s largest (Adler et al. 2020). The reservoir was the nation’s first large-scale experiment in indirect potable reuse, defined as the practice of deliberately introducing highly treated wastewater to surface water or groundwater for potable supply (Grant et al. 2022). Approximately 15% (4.6 × 10⁷ m³/yr) of the reservoir’s average annual inflow is highly treated wastewater from the Upper Occoquan Service Authority (UOSA), and the remaining 85% (7.1 × 10⁸ m³/yr) is baseflow and wet weather runoff from two local streams, Bull Run and the Occoquan River, and ungaged watershed flow (Bhide et al. 2021, Grant et al. 2022).

The primary governance system for managing water quality in the reservoir includes both federal (i.e., Clean Water Act) and state-specific legislation (i.e., the Occoquan Policy), and the associated network of actors that formulate and implement environmental policy and planning in the Occoquan system. The Occoquan Policy, which is codified in the Virginia Administrative Code at 9 VAC 25-410, mandated the construction of the UOSA wastewater treatment plant for two purposes: (1) to improve drinking-water security in the region by providing a high-quality, drought-proof source of water; and (2) to reduce reservoir eutrophication, which, at the time, was attributed to discharges from 11 low-performance treatment plants to the Occoquan reservoir (Bhide et al. 2021, Grant et al. 2022). The Policy also mandated the formation of the Occoquan Watershed Monitoring Program to oversee UOSA’s discharge and monitor reservoir water quality, which has resulted in near continuous monitoring of the system for more than 25 years.

These long-term monitoring data reveal that salinity (e.g., as measured by specific conductance) and concentrations of several salt ions in the reservoir have been increasing over time. The concentration of sodium ions, in particular, occasionally exceeds U.S. EPA guidance on taste and health thresholds for drinking water (EPA 2003b, Bhide et al. 2021). The primary source of sodium ions to the reservoir depends on weather conditions (Bhide et al. 2021); namely, UOSA’s discharges contribute 60–80% of sodium mass during dry weather, and watershed discharges, particularly Bull Run, contribute 40–60% of sodium mass during wet weather. On average, the total daily mass load of sodium to the reservoir is 42,000 kg/day.

METHODS

Fuzzy cognitive map development process

Beginning in summer 2021, 35 stakeholders from in and around the Occoquan system, including water and wastewater utilities (8 total), state and federal governments (8), local cities and counties (6), interstate agencies (5), private corporations, developers, consulting firms or research agencies (5), and environmental nongovernmental organizations (3), participated in a cognitive mapping exercise to construct personalized mental models of freshwater salinization in the Occoquan Reservoir and its tributary streams. All participants were part of the Executive Committee on the Occoquan System (ECOS), which meets quarterly and was formed as part of a U.S. National Science Foundation-funded Growing Convergence Research project geared toward catalyzing stakeholder-driven solutions to freshwater salinization.

Stakeholder mental models were generated via an iterative process (Supplemental Fig. 1 in Appendix 1) that began with co-production of a concept list featuring causes of salinization, consequences of salinization, and actions that might be taken to mitigate salinization. The list was developed using joint fact-finding approaches (Karl et al. 2007), starting with a questionnaire eliciting salt-specific management goals, and culminating in small-group discussions during the first ECOS workshop, where salinization concepts pertaining to these goals were developed and refined.

Following concept list development, one-on-one interviews were conducted to generate individual FCMs of the SES for freshwater salinization using Mental Modeler software (Gray et al. 2013). Each stakeholder’s task was to identify the seven most important causes of salinization, consequences of salinization, and actions that could be taken to mitigate salinization (newly created concepts or concepts drawn from the co-developed list) and establish causal connections between them based on their own understanding of the SES (Aminpour et al. 2020, 2021). All stakeholders were required to include salinization of the Occoquan watershed in their models as the central, organizing concept. Once each FCM was complete, it was validated by conducting dynamic simulations whereby all possible mitigating actions were sequentially perturbed so that their impact on all causally-associated concepts could be viewed. Models were revised in real time to meet stakeholder specifications whenever dynamic simulations resulted in outputs that were perceived to be nonsensical.

Following initial FCM collection, de-identified model results were presented at a second ECOS workshop, allowing stakeholders to comment on model concepts and structure and discuss potential modifications. This discussion initiated a process to streamline model concepts (i.e., concept consolidation and final concept list [Supplemental Figs. 2–4 and Supplemental Tables 1–3, respectively, in Appendix 1]). Following the consolidation process, each stakeholder FCM was recast with updated concepts, and a second round of FCM interviews was conducted to review updated models and revise them as needed. These revised FCMs are our best realization of how stakeholders presently understand freshwater salinization in the Occoquan (see Supplemental Fig. 5 in Appendix 1 for individual FCMs) and served as the basis for all analyses conducted here. Next, we describe how individual FCMs were analyzed (see also Fig. 1).

Statistical characterization of similarities and differences

To identify similarities across stakeholder FCMs (concepts and relationships), we first quantified the fraction of FCMs in which each concept was present. Concepts that were present in > 50% of FCMs were retained as core concepts. Once these concepts were identified, the adjacency matrix of each FCM was used to identify common relationships among them (i.e., relationships present in at least 25% of FCMs) that reflect consistencies in how the SES for freshwater salinization is perceived by stakeholders in the Occoquan.

Concept centrality (specifically, weighted degree centrality; Newman 2004) was used to characterize differences across stakeholder FCMs. The centrality of a concept is considered indicative of its contribution in a mental model (i.e., its importance for system dynamics) because it reflects the number of connections between that concept and other concepts, as well as the strength of those relationships (Özesmi and Özesmi 2004, Aminpour et al. 2020). Weighted degree centrality was calculated from the adjacency matrix of each FCM. Concepts that were not included in a given stakeholder’s model but were present in others were assigned a weighted degree centrality of zero. This process was repeated for all stakeholders, returning a complete set of weighted degree centrality values for each concept and individual. These values were standardized (subtracting the mean and dividing by the standard deviation) prior to performing principal component analysis (PCA) to identify dominant patterns in the perceived importance of salinization concepts across stakeholders (R package FactoMineR; Lê et al. 2008, Vasslides and Jensen 2016). Standardization keeps the analysis focused on the intrinsic structure of the data (patterns in centrality) and mitigates the influence of differences in FCM size (magnitude of centrality; Davis 2002). Only principal components (PCs) that captured significantly more variance than expected due to chance (resampling-based stopping rule, 95% confidence level) were retained (Peres-Neto et al. 2005, Rippy et al. 2017). Hierarchical clustering of these components was used to identify groups of stakeholders with significantly different perspectives about freshwater salinization. We elected to characterize divergent stakeholder groups in this way because it is entirely inductive (i.e., it is independent of our preconceived notions about what might make stakeholder perceptions different). Clustering was performed using Ward’s criterion, followed by k-means consolidation (R package HCPC; Lê et al. 2008). Concepts that were significantly over- or under-represented in a particular stakeholder group were identified using an F-test (P < 0.01 level; Husson et al. 2010).

Fuzzy cognitive map aggregation (group and collective models)

Individual FCMs belonging to each stakeholder group identified above (i.e., each significant cluster) were aggregated to create mental models that best characterize the SES for freshwater salinization perceived by these groups. Aggregation was performed as in Aminpour et al. (2020) by taking the mean of the weighted adjacency matrix for all FCMs within each group, omitting zero values (i.e., only relationships for which weights were provided were included in each average).

A collective model representing all stakeholders’ insights was also created. This model used approaches developed to leverage wisdom of the crowd and reduce biases associated with shared experiences and knowledge (Aminpour et al. 2020, 2021). Addressing these biases is important because the stakeholder pool we interviewed is not balanced (i.e., we have different numbers of individuals across disciplines and practices, which can give certain perspectives more power than others if not corrected for; Lorenz et al. 2011, Navajas et al. 2018). To address this bias, Aminpour et al.’s (2020) two-step aggregation approach was used, whereby the aggregate models for each stakeholder group (i.e., representing significantly different perspectives; see above) were themselves aggregated by taking the median of their respective adjacency matrices, giving each perspective equal weight in the final collective model.

Dynamic simulation (perceived impact of various salt sources)

The dynamic behavior of FCMs was assessed by changing the activation value of one or more model concepts (i.e., salt sources) and quantifying the impact (change relative to baseline) on the central concept “Salinization of the Occoquan” (Özesmi and Özesmi 2004, Aminpour et al. 2020). Simply put, when different causes of salinization are activated in a FCM (i.e., their activation value is set to 1), it triggers a cascade through all causally connected concepts in the network, including the central concept, “Salinization of the Occoquan Watershed”. This procedure allows us to evaluate the relative impact that increasing different salt sources is perceived to have on reservoir salinization for any mental model of the SES. Simulations were performed to evaluate bulk sewershed salt sources (in-home products, industry, water and wastewater treatment), bulk watershed salt sources (winter maintenance chemicals, local geology, agriculture, infrastructure corrosion), and the various subsources that compose them. Simulations were conducted in R using the package fcm (Dikopopoulou and Papageorgiou 2017). Mathematical details are provided in Appendix 1 (Supplemental Methods).

Comparison with biophysical measurements

Perceived sewershed-to-watershed salt ratios from dynamic FCM simulations (see above) were compared to biophysical estimates of sodium loading from these two sources to the Occoquan Reservoir. Details of these estimates are provided in Bhide et al. (2021). Briefly, daily time series of sodium mass load were estimated for UOSA (the sewershed) and Bull Run and the Occoquan River (the watershed) using multiple linear regression. Best-fit models (31–87% variance explained) were used to generate an eight-year synthetic daily time series of sodium concentrations at each location that was combined with measured daily flow using the U.S. Geological Survey package Loadflex, generating an eight-year daily time series of sodium load from each source. A bivariate copula was used to join marginal probability distributions of sodium mass load from each source to cumulative flow across all sources to illustrate how the contribution of different sodium sources to overall sodium mass loading varies with local weather conditions (low flow: 2.55 m³/s, median flow: 6.91 m³/s, high flow: 31.0 m³/s). Probability density functions for the sewershed to watershed sodium ratio, conditioned on flow, were developed using this relationship.

RESULTS

Seventeen key knowledge concepts

Seventeen concepts were common across stakeholder FCMs (i.e., present in > 50% of models; Fig. 2). Frequently mentioned causes of salinization included population and economic development in the watershed (51% of models); watershed imperviousness (71% of models); the application of deicers and anti-icers to roads and parking lots during winter (henceforth referred to as “winter maintenance activities”) by public agencies (100% of models), private companies (91% of models), and residential communities (86% of models); industrial discharges to reservoir tributaries and the sanitary sewer system (74% of models); and down-the-drain disposal of products used in homes, which ultimately flow through UOSA into the reservoir (57% of models). Frequently mentioned impacts of freshwater salinization included ecosystem health (94% of models); sustainability, reliability, and safety of the drinking water supply (89% of models); longevity of urban infrastructure such as pipes, roads, and bridges (66% of models); and the continued practice of indirect potable reuse (51% of models). Frequently mentioned actions that could be taken to address salinization included the creation and adoption of best management practices such as processes, technologies, and training for winter salt application and storage (89% of models); guidance documents for winter maintenance activities (71% of models); imposition of ion-specific regulations such as total maximum daily loads (63% of models); public education focused on reducing salt pollution (74% of models); and the adoption of low-salt products for indoor and outdoor use (63% of models). Across all FCMs, only one concept was considered important by everyone: salts added by winter maintenance activities in the public sector.

Common associations between concepts (i.e., that appeared in > 25% of stakeholder models) are also illustrated in Fig. 2. Not surprisingly, stakeholders frequently linked winter maintenance activities (by public agencies, private companies, and residential communities) to elevated impervious area in the watershed (as a primary driver), as well as to best management practices, guidance documents, and the adoption of low-salt alternative deicers (as potential actions that could be taken to reduce salinization; blue connections in Fig. 2). Regulation, public education campaigns, and controls on population and economic development were frequently mentioned as ways to reduce salt loads from both the watershed (e.g., winter maintenance activities) and sewershed (e.g., down-the-drain disposal of salt-containing products; green connections in Fig. 2).

Four stakeholder groups with significantly different perceptions of the social-ecological system

PCA and hierarchical cluster analysis were used to identify groups of stakeholders with significantly different perceptions of the SES for freshwater salinization that could represent barriers to collective action or opportunities for shared learning. Only two PCs were significantly different than random, collectively explaining ~34% of the variance in centrality of salinization concepts across mental models (Fig. 3). The first PC (PC1, 19.6% variance explained) captures the perceived contribution of winter maintenance activities to salinization of the Occoquan Reservoir and actions that could be taken to address them (positive PC1: winter maintenance is emphasized, negative PC1: winter maintenance is deemphasized). The second PC (PC2, 14.2% variance explained) captures the perceived role of the sewershed in reservoir salinization, specifically whether salt discharges from UOSA contribute (positive PC2) or do not contribute (negative PC2) to the problem.

Hierarchical clustering with respect to these PCs yields four statistically distinct groups of FCMs. Group 1 includes the largest number of stakeholders (20 of 35; orange circles in Fig. 3) and is dominated by employees from public utilities and city or county government agencies. Group 2 (11 of 35; teal circles in Fig. 3) is dominated by employees from federal, state, or interstate agencies. Group 3 (yellow circles in Fig. 3) consists of two individuals from an interstate agency and one from a private corporation. Group 4 (blue circle in Fig. 3) consists of a single individual from a state or federal agency. All groups excepting Group 4 contain stakeholders with different professional backgrounds, and all professional backgrounds are represented in more than one group (pie charts in Fig. 3), suggesting that differences in how the SES is perceived are not profession specific.

Group-specific social-ecological system networks

Individual mental models belonging to each of the groups were aggregated (see Methods) to highlight salinization concepts common to each group. The aggregate model for Group 1 contains the most salinization concepts (C) and relationships (R; 39 C, 177 R; Fig. 4a), followed by the aggregate models for Group 3 (30 C, 127 R; Fig. 4b), Group 2 (29 C, 78 R; Fig. 4c), and Group 4 (19 C, 52 R; Fig. 4d).

The Group 1 aggregate model deemphasizes two causes of salinization (salts added by winter maintenance chemicals in private commercial and residential sectors), two factors affected by salinization (infrastructure longevity and ecosystem health), and one mitigating action (development of regulatory frameworks for salt; Fig. 4a). Group 3 emphasizes the importance of guidance documents for managing salt as well as all concepts deemphasized in Group 1 except ecosystem health (Fig. 4c). Group 4 is similar to Group 3 in its emphasis on salts added by winter maintenance activities in the residential sector. However, it also emphasizes salts contributed by local geology, education and outreach programs (both for the general public and for management), ecosystem health, and the sustainability, reliability, and safety of drinking water supply, concepts that are not emphasized by other groups (Fig. 4d). Group 2 is unique in its emphasis on environmental justice, adoption of low-salt technologies for water and wastewater treatment, and three causes of salinization (salts added by indoor products discharged to UOSA’s sewershed, salts added as part of drinking water treatment, and salts released by urban infrastructure corrosion; Fig. 4b).

Focusing on the causes of salinization that each group felt are important, it becomes clear that Groups 1, 3, and 4 are more watershed focused, differing mainly in their degree of emphasis (Fig. 4c,d) or de-emphasis (Fig. 4a) on private sector use of deicers and anti-icers (see quotes Q13 and 14 for context, Table 1). In contrast, Group 2 is more sewershed focused (Fig. 4b). Stakeholder narratives (collected during FCM interviews) suggest that Group 2’s sewershed focus may reflect concerns about the prevalence of salt-containing products available for in-home use (Q18, Table 1) and the quantities that are used (Q19, Table 1), as well as the cyclical nature of indirect potable reuse in the Occoquan, wherein salt in surface water enters drinking water systems, moves through homes and businesses into the wastewater collection system, and from there passes through wastewater treatment plants back into surface waters (Q17, Table 1).

Notably, the preferred mode of governance of the Occoquan SES also differed by stakeholder group, with top-down regulatory action being advocated by Group 3 and collective action being advocated by Group 1 (Fig. 4a,c). Stakeholder narratives suggest that preferences for regulatory frameworks are driven by prior experience with this governance model (Q4, Table 1), cynicism about the likelihood of success of other models (Q5, Table 1), and perceptions about who needs to be regulated, for example, industry, where regulation is a standard approach (Q3, Table 1). Preferences for non-regulatory modes of governance appear to reflect strong advocacy for stakeholder-driven, bottom-up approaches (Q1, Table 1) and concerns about the feasibility of regulating certain salt sources, specifically, in-home products (Q2, Table 1).

Perceived contribution of different salt sources

To evaluate the perceived contribution of different salt sources to salinization of the Occoquan, dynamic simulations of various salt-loading scenarios were performed using individual stakeholder FCMs, the four aggregated group models, and a single cross-group “collective model” (Fig. 4e) prepared by taking the median of the aggregated group models (see Methods). Dynamic simulations illustrate that salt sources in the watershed (such as winter maintenance activities) were perceived to contribute more to reservoir salinization than salt sources in the sewershed (dark grey vs. black bars in Fig. 5a). This pattern is most evident for the Group 3 model and the average taken across all individual mental models, followed by the Group 4 model, and the collective model. Only models for Groups 1 and 2 predicted that salt contributions from the sewershed and the watershed were comparable.

The relative importance of specific salt subsources (i.e., sources that together comprise total sewershed or total watershed salt loads) also differed by FCM. Typically, salts added by winter maintenance activities had the most impact on salinization, followed by in-home products and industry, other watershed sources (i.e., local geology, agriculture, and infrastructure corrosion), and salts added by water and wastewater treatment (Fig. 5a). An extreme version of this pattern is evident in the Group 4 model, in which salts added by water and wastewater treatment are entirely absent (Fig. 5a). The Group 1 model differs from the other models in that salts added by in-home products and industry were perceived as having the most impact on salinization. The Group 2 model also differs from other models, with water and wastewater treatment perceived to contribute more salts to the reservoir than do local geology, infrastructure corrosion, or agriculture.

Benchmarking perceptions with biophysical measurements

The ratio of sodium mass load from the sewershed (UOSA) vs. watershed (Bull Run + Occoquan River) spans three orders of magnitude during low, median, and high flow conditions, from 0.02 (the watershed contributes 46-times more sodium to the reservoir than the sewershed) to 17.3 (the sewershed contributes 17-times more sodium to the reservoir than the watershed; Fig. 5b). Sodium loading to the reservoir is dominated by outflow from UOSA during low flow conditions (i.e., during dry weather; mean sewershed to watershed ratio: 2.6) and by the watershed during high flow conditions (mean ratio: 0.20). During median flow conditions, the contribution of these two sources is relatively even (mean ratio of 0.75; Fig. 5b).

These biophysical estimates were compared to stakeholder perceptions about the relative importance of sewershed and watershed salt sources estimated using our dynamic FCM simulations. Only two sewershed-to-watershed salt ratios from individual stakeholder FCMs (small grey dots in Fig. 5b) fell outside the estimated range (i.e., 94% of individual FCMs are consistent with estimated salt loading from the sewershed and watershed). Most simulations conducted using individual FCMs (21) are consistent with median flow conditions (i.e., fall under the black distribution in Fig. 5b), followed by high flow conditions (11) and low flow conditions (1; Fig. 5b). When individual mental models are averaged or aggregated, the resultant sewershed to watershed ratios are generally consistent with median flow conditions in the Occoquan reservoir (note the positioning of Groups 1 through 4, the collective FCM, and the mean of individual stakeholder FCMs in Fig. 5b). Of these models, the collective FCM appears to best approximate the average sewershed-to-watershed sodium ratio during median flow (black vertical line in Fig. 5b).

DISCUSSION

It is often challenging to identify key drivers, impacts, and potential outcomes of different management approaches for emerging environmental problems where the SES is not well characterized. Our results suggests that FCMs can be an effective tool for: (1) characterizing individual and collective models of the SES; (2) articulating common (Fig. 2) and divergent (Figs. 3 and 4) viewpoints that could be opportunities for collaboration or shared learning experiences, respectively; and (3) comparing SES perceptions to biophysical estimates (Fig. 5), furthering the ability to discuss understanding in context with biophysical conditions. This utility makes FCMs a useful and much needed tool for rapidly coalescing knowledge about emerging environmental challenges and generating systems-level information that can guide future research and collective management efforts.

Areas of agreement and disagreement: implications for collective management

Mode of salt governance

Stakeholder FCMs reveal different perspectives about the importance of top-down regulatory control of salinization vs. bottom-up, voluntary measures. These stances may seem opposing, with regulation appearing to obviate the need for voluntary collective action to manage salt, but this is not really the case. Polycentric, collaborative governance arrangements can emerge through bottom-up collective action or state-centered policy processes, either by design (e.g., intentional decentralization) or institutional layering, as new policy venues and rules are gradually introduced (Langridge and Ansell 2024). In the Occoquan, many existing collaborations between utilities, counties, and other agencies are actually rooted in policy (e.g., the Occoquan Policy), which created new institutions, a compliance monitoring program, and the multijurisdictional financial arrangements to support it (i.e., there are elements of polycentric governance of water quality in the Occoquan that were instituted top-down; Grant et al. 2022). Although the Policy only addresses regulated contaminants, which many salt ions are not (Schuler et al. 2019), the collaborative arrangements that formed under the Policy have been leveraged to create working groups on salt and other unregulated contaminants, an evolution in line with voluntary, collective action (Langridge and Ansell 2024).

Certain salt-related activities are still catalyzed by regulation. For instance, the 2017 total maximum daily load for chloride in Accotink Creek, located near the Occoquan watershed in Fairfax County, initiated a wave of collaborative activities centered around the development of a salt management strategy toolkit for Northern Virginia (ICPRB and VA DEQ 2020) and may have informed the pro-regulation stance evident in certain stakeholder FCMs (e.g., group 3; Fig. 3c). Other salt-related activities had no regulatory catalyst, including participation in ECOS as part of this Growing Convergence Research Project and the decision to finance a network of continuous monitoring stations for specific conductance (a measure of salinity) in the Occoquan Reservoir. If stakeholders were to co-develop voluntary standards for sodium, as advocated in Group 1 FCMs (see Table 1, Q1), it would be another example of a bottom-up measure. A range of sodium ion standards exist across U.S. states (Schuler et al. 2019) that, in context with health and toxicity studies (Calabrese and Tuthill 1981, Hallenbeck et al. 1981, Welty 1986, Corsi et al. 2010, Khan et al. 2011, Cooper et al. 2014, Clements and Kotalik 2016, Schuler and Relyea 2018; Canadian Environmental Protection Act, 1999: https://www.canada.ca/en/environment-climate-change/services/canadian-environmental-protection-act-registry/publications/canadian-environmental-protection-act-1999.html), could potentially inform the development of local criteria for the Occoquan. FCMs could be used to identify critical stakeholders and decision-makers who need to be involved in collaborative decision-making about these criteria, supporting a stakeholder or issues analysis (Susskind et al. 1999). They might also be combined with other stakeholder assessment processes (see the Society of Professionals in Dispute Resolution 1997), leading to recommendations regarding: (1) institutional, financial, or other constraints that could limit collaborative problem-solving efforts, and (2) the circumstances under which key parties would agree to participate in those efforts.

Salt sources

From the standpoint of collective management, winter maintenance chemicals appear to be low-hanging fruit, consistent with the attention they are paid in the salinization literature (Godwin et al. 2003, Kaushal et al. 2005, Kelly 2008, Corsi et al. 2015). Co-development of the salt management strategy toolkit for Northern Virginia (ICPRB and VA DEQ 2020) appears to have played an outsized role in shaping stakeholder perspectives about winter maintenance chemicals. Many of the concepts central to the salt management strategy toolkit (e.g., winter best management practices, low-salt products such as alternative deicers, and public education and outreach; ICPRB and VA DEQ 2020) appear frequently in stakeholder FCMs, illustrating the salt management strategy toolkit’s contribution to collective learning and shared understanding of freshwater salinization in the region (see Appendix 1 for additional details about the salt management strategy toolkit).

Variability in stakeholder FCMs with respect to the sewershed suggests that the sewershed represents a new frontier for shared learning rather than an area that is already ripe for collective action. Attention to the sewershed has been increasing in the literature, with a variety of studies identifying significant wastewater contributions to freshwater salinization, even in regions where deicing is prevalent (Novotny et al. 2009, Bhide et al. 2021, Overbo et al. 2021). Factoring wastewater into the freshwater salinization story has important environmental and public health implications given the contribution of wastewater to overall stream flow in the United States (at least 10% of the flow in > 25% of U.S. streams is wastewater effluent; Rice and Westerhoff 2017) and the important role treated wastewater is projected to play as a future water source due to water supply shortfalls (GAO 2014, EPA 2019, Martin and Via 2020). Following multiple rounds of FCM interviews and small group FCM discussions, one stakeholder noted, “My greatest knowledge growth or conceptual change [has been] around the wastewater... now I see there are these personal care products and human excretion and industrial [inputs] too...” [stakeholder 7]. This observation suggests that, as much as the sewershed is presently contentious, it is also a frontier for shared learning that participatory modeling tools like FCM might help to facilitate. Next, we address how this learning might occur, focusing on “cheap talk” and perspective-taking.

Fuzzy cognitive maps as tools for facilitating shared learning

Encouraging “cheap talk”

When aggregated to represent collective perceptions, FCMs have the potential to become co-produced boundary objects that facilitate communication across stakeholders from different social worlds (Leigh Star 2010, Black 2013, Huang et al. 2018). In this mode, they may serve as an organizing framework for “cheap talk”, i.e., frequent, ordinary, communication that is neither costly nor binding (Farrell and Rabin 1996, Wilson and Sell 1997), about areas where stakeholders disagree. Cheap talk can be a valuable tool for shared learning. It has been shown to foster group identity, facilitate cooperation, and help individuals negotiate jointly beneficial solutions when perspectives differ (Sally 1995, Ostrom 2010a,b), as is the case here with the sewershed. Effective cheap talk, however, requires trust; conversations can be nonbinding, but they should not be dishonest (Wilson and Sell 1997). The collective trust and group identity the Occoquan’s stakeholders have built though past successful collaborations like the salt management strategy toolkit and other initiatives bodes well for the effectiveness of cheap talk in this system (Kramer and Tyler 1996, Nagendra and Ostrom 2014, Bodin 2017; Canadian Environmental Protection Act, 1999: https://www.canada.ca/en/environment-climate-change/services/canadian-environmental-protection-act-registry/publications/canadian-environmental-protection-act-1999.html). As one stakeholder said, “I, several times, have expressed my confidence in all of our colleagues, and here this morning I hear it. Each organization, for want of better words, upping their game [to manage salt]” [4]. Comments like this illustrate that while stakeholders may not always agree about salt sources, they share a commitment to addressing salinization, making the emergence of collective action a real possibility.

Encouraging perspective-taking

Nearly all stakeholder mental models captured sewershed-to-watershed sodium ratios that fell within the observed range for the Occoquan system during low, median, or high flow conditions (Fig. 5b), which indicates that stakeholder understanding of this dimension of the SES, although variable, was generally high. Because system variability validates a range of possible perspectives on the relative importance of salt sources, it has the potential to focus future discussions on why certain models capture specific conditions well, rather than the identification of a single “best” model. In effect, it may encourage a form of perspective-taking (trying on alternative models and considering their validity), which has been shown to increase collaborative behavior in collectively managed natural resource systems (Wald et al. 2017). Perspective-taking has proven especially beneficial in high-variability systems where management becomes less a search for optimal solutions and more an ongoing process of learning and negotiation (Pahl-Wostl and Hare 2004, Innes and Booher 2018). This may make it particularly appropriate for fostering shared understanding in systems like the Occoquan, where variability in the resource (sodium ion concentrations in the reservoir) informs variability in perceptions of the SES.

Benefits and limitations of fuzzy cognitive maps for understanding the social-ecological system

Several features of FCMs, beyond their ability to identify critical stakeholders who should be involved in collaborative decision-making or facilitate cheap talk and perspective-taking, make them well suited for characterizing SESs and using that knowledge to support collective management of emerging environmental challenges. A subset of these capabilities (and potential limitations) are detailed here. First, the FCM development process is relatively straightforward and engaging, bringing a diversity of perspectives to the table (Özesmi and Özesmi 2004). This characteristic is evident in the fraction of stakeholders willing to develop FCMs in this study (90% of those we contacted) and the time invested in model building (50 min on average for the initial FCM exercise, followed by an additional 1.5 h on average in subsequent meetings to refine concepts, revise models, and validate final FCMs; see Appendix 1).

The iterative nature of FCM development in this study (i.e., the inclusion of review and revision steps) is not standard for the field but appears to have increased the legitimacy of FCMs in the eyes of stakeholders: “The process the team has done, and going through and revising them [the FCMs] and getting more input and going through and revising them again; I think the legitimacy is the best you can possibly do” [10]. This finding is something to consider when using FCMs as a tool for co-producing new knowledge about the form or structure of SESs, where fully incorporating the values and perspectives of different actors involved in resource management is a top priority (Hailu and Tolossa 2020). The time commitment required for cyclical co-production of FCMs, however, can also be a limitation, particularly if the pool of possible actors is large. In such instances, alternative approaches for eliciting FCMs may be more appropriate, for example, group model-building (Martin and Via 2020, Shahvi et al. 2021, Olofsson et al. 2023) or use of survey instruments in lieu of one-on-one interviews to identify concepts and their interrelationships (Özesmi and Özesmi 2004, Nozari et al. 2021, Aminpour et al. 2022).

Another strength of FCMs for understanding SESs is their applicability in systems where information is limited (i.e., where data supporting certain relationships is hard to come by or only available to select experts; Özesmi and Özesmi 2004, Gray et al. 2019). This property can allow stakeholders to make connections they feel are valid, but where the underlying mechanisms are unclear (i.e., “I don’t know enough about groundwater, but I’m imagining that it could be directly impacted” [28]). However, it can also become a point of methodological contention, potentially undermining trust (i.e., “I think there is a danger that some of this process is driven entirely by opinions” [9]). Addressing this issue is challenging and may require recurrent discussions of FCM results and approaches in the attempt to acknowledge/resolve methodological concerns and remind stakeholders that FCMs are not intended to be a substitute for physical models. Rather, they are a way to provide a broader systems perspective that incorporates elements such as governance that are typically outside the scope of physical models, facilitating shared learning (Özesmi and Özesmi 2003, Olofsson et al. 2023). Exploring alternative modes of FCM collection that explicitly account for uncertainty (e.g., fuzzy grey cognitive maps or intuitionistic fuzzy cognitive maps; Salmeron 2010, Papageorgiou and Iakovidis 2013) may also help to address stakeholder concerns, providing the means to identify regions of the SES where less is known and the potential for learning is highest.

A final feature that makes FCMs valuable for understanding SESs (both for freshwater salinization and other emerging environmental problems) is their ease of incorporation into the cycles of learning and assessment characteristic of collective management (i.e., each new mental model can build upon the last, creating a dynamic SES; Pahl-Wostl and Hare 2004). This dynamicity is important because SES understanding (particularly of emerging problems) is expected to evolve over time, making recurrent evaluation important (Carley 1997, Schlüter et al. 2014). Our future work will leverage and explore this functionality, improving the ability to treat second-level variables such as understanding the SES as fully dynamic, and expanding knowledge of the role that understanding plays in catalyzing collective management of emerging environmental problems such as inland freshwater salinization.

CONCLUSIONS

We have demonstrated that FCMs can be an effective tool for characterizing stakeholder understanding of freshwater salinization, including key knowledge concepts and how those concepts are linked together into complex SES networks. We illustrate that FCMs can: (1) capture physically meaningful information about the SES (e.g., reproducing realistic sewershed vs. watershed salt source balances), and (2) reveal differences in the perceived importance of individual salt subsources, their societal and environmental impacts, and how they might be managed, that can be used to diagnose areas of agreement where bottom-up management is likely (winter salt sources), as well as divergent views (sewershed salt sources, governance approaches) that could represent barriers to collective action or opportunities for shared learning. The process we describe for upscaling individual knowledge to systems-level understanding is not salt specific and can be used to inform management approaches for other emerging challenges.

LITERATURE CITED

Adler, P., R. Florida, and M. Hartt. 2020. Mega regions and pandemics. Tijdschrift voor Economische en Sociale Geografie 111(3):465-481. https://doi.org/10.1111/tesg.12449

Aminpour, P., S. A. Gray, M. W. Beck, K. L. Furman, I. Tsakiri, R. K. Gittman, J. H. Grabowski, J. Helgeson, L. Josephs, M. Ruth, and S. B. Scyphers. 2022. Urbanized knowledge syndrome—erosion of diversity and systems thinking in urbanites’ mental models. npj Urban Sustainability 2(1):11. https://doi.org/10.1038/s42949-022-00054-0

Aminpour, P., S. A. Gray, A. J. Jetter, J. E. Introne, A. Singer, and R. Arlinghaus. 2020. Wisdom of stakeholder crowds in complex social-ecological systems. Nature Sustainability 3(3):191-199. https://doi.org/10.1038/s41893-019-0467-z

Aminpour, P., S. A. Gray, A. Singer, S. B. Scyphers, A. J. Jetter, R. Jordan, R. Murphy Jr., and J. H. Grabowski. 2021. The diversity bonus in pooling local knowledge about complex problems. Proceedings of the National Academy of Sciences 118(5):e2016887118. https://doi.org/10.1073/pnas.2016887118

Bhide, S. V., S. B. Grant, E. A. Parker, M. A. Rippy, A. N. Godrej, S. Kaushal, G. Prelewicz, N. Saji, S. Curtis, P. Vikesland, A. Maile-Moskowitz, M. Edwards, K. G. Lopez, T. A. Birkland, and T. Schenk. 2021. Addressing the contribution of indirect potable reuse to inland freshwater salinization. Nature Sustainability 4(8):699-707. https://doi.org/10.1038/s41893-021-00713-7

Black, L. J. 2013. When visuals are boundary objects in system dynamics work. System Dynamics Review 29(2):70-86. https://doi.org/10.1002/sdr.1496

Bodin, Ö. 2017. Collaborative environmental governance: achieving collective action in social-ecological systems. Science 357(6352):eaan1114. https://doi.org/10.1126/science.aan1114

Calabrese, E. J., and R. W. Tuthill. 1981. The influence of elevated levels of sodium in drinking water on elementary and high school students in Massachusetts. Science of the Total Environment 18:117-133. https://doi.org/10.1016/S0048-9697(81)80054-X

Cañedo-Argüelles, M., B. J. Kefford, C. Piscart, N. Prat, R. B. Schäfer, and C.-J. Schulz. 2013. Salinisation of rivers: an urgent ecological issue. Environmental Pollution 173:157-167. https://doi.org/10.1016/j.envpol.2012.10.011

Carley, K. M. 1997. Organizational adaptation. Annals of Operations Research 75:25-47. https://doi.org/10.1023/A:1018963630536

Checkland, P., and J. Poulter. 2020. Soft systems methodology. Pages 201-253 in M. Reynolds and S. Holwell, editors. Systems approaches to making change: a practical guide. Second edition. Springer, London, UK. https://doi.org/10.1007/978-1-4471-7472-1_5

Clements, W. H., and C. Kotalik. 2016. Effects of major ions on natural benthic communities: an experimental assessment of the US Environmental Protection Agency aquatic life benchmark for conductivity. Freshwater Science 35(1):126-138. https://doi.org/10.1086/685085

Cooper, C. A., P. M. Mayer, and B. R. Faulkner. 2014. Effects of road salts on groundwater and surface water dynamics of sodium and chloride in an urban restored stream. Biogeochemistry 121(1):149-166. https://doi.org/10.1007/s10533-014-9968-z

Corsi, S. R., L. A. De Cicco, M. A. Lutz, and R. M. Hirsch. 2015. River chloride trends in snow-affected urban watersheds: increasing concentrations outpace urban growth rate and are common among all seasons. Science of the Total Environment 508:488-497. https://doi.org/10.1016/j.scitotenv.2014.12.012

Corsi, S. R., D. J. Graczyk, S. W. Geis, N. L. Booth, and K. D. Richards. 2010. A fresh look at road salt: aquatic toxicity and water-quality impacts on local, regional, and national scales. Environmental Science and Technology 44(19):7376-7382. https://doi.org/10.1021/es101333u

Davis, J. C. 2002. Statistics and data analysis in geology. Third edition. Wiley, Hoboken, New Jersey, USA.

Dikopopoulou, Z., and E. Papageorgiou. 2017. fcm: inference of fuzzy cognitive maps (FCMs). R package version 0.1.3. https://cran.r-project.org/package=fcm

Emerson, K., T. Nabatchi, and S. Balogh. 2012. An integrative framework for collaborative governance. Journal of Public Administration Research and Theory 22(1):1-29. https://doi.org/10.1093/jopart/mur011

Farrell, J., and M. Rabin. 1996. Cheap talk. Journal of Economic Perspectives 10(3):103-118. https://doi.org/10.1257/jep.10.3.103

Findlay, S. E. G., and V. R. Kelly. 2011. Emerging indirect and long-term road salt effects on ecosystems. Annals of the New York Academy of Sciences 1223(1):58-68. https://doi.org/10.1111/j.1749-6632.2010.05942.x

Flörke, M., I. Bärlund, M. T. H. van Vliet, A. F. Bouwman, and Y. Wada. 2019. Analysing trade-offs between SDGs related to water quality using salinity as a marker. Current Opinion in Environmental Sustainability 36:96-104. https://doi.org/10.1016/j.cosust.2018.10.005

Freeman, J., J. A. Baggio, and T. R. Coyle. 2020. Social and general intelligence improves collective action in a common pool resource system. Proceedings of the National Academy of Sciences 117(14):7712-7718. https://doi.org/10.1073/pnas.1915824117

Godwin, K. S., S. D. Hafner, and M. F. Buff. 2003. Long-term trends in sodium and chloride in the Mohawk River, New York: the effect of fifty years of road-salt application. Environmental Pollution 124(2):273-281. https://doi.org/10.1016/S0269-7491(02)00481-5

Grant, S. B., M. A. Rippy, T. A. Birkland, T. Schenk, K. Rowles, S. Misra, P. Aminpour, S. Kaushal, P. Vikesland, E. Berglund, J. D. Gomez-Velez, E. R. Hotchkiss, G. Perez, H. X. Zhang, K. Armstrong, S. V. Bhide, L. Krauss, C. Maas, K. Mendoza, C. Shipman, Y. Zhang, and Y. Zhong. 2022. Can common pool resource theory catalyze stakeholder-driven solutions to the freshwater salinization syndrome? Environmental Science and Technology 56(19):13517-13527. https://doi.org/10.1021/acs.est.2c01555

Gray, S. A., S. Gray, L. J. Cox, and S. Henly-Shepard. 2013. Mental modeler: a fuzzy-logic cognitive mapping modeling tool for adaptive environmental management. Pages 965-973 in Proceedings of the 2013 IEEE 46th Hawaii International Conference on System Sciences, Wailea, HI, USA. IEEE, Piscataway, New Jersey, USA. https://doi.org/10.1109/HICSS.2013.399

Gray, S., E. J. Sterling, P. Aminpour, L. Goralnik, A. Singer, C. Wei, S. Akabas, R. C. Jordan, P. J. Giabbanelli, J. Hodbod, E. Betley, and P. Norris. 2019. Assessing (social-ecological) systems thinking by evaluating cognitive maps. Sustainability 11(20):5753. https://doi.org/10.3390/su11205753

Hailu, R., and D. Tolossa. 2020. Multi-stakeholder platforms: institutional options to achieve water security in the awash basin of Ethiopia. World Development Perspectives 19:100213. https://doi.org/10.1016/j.wdp.2020.100213

Hallenbeck, W. H., G. R. Brenniman, and R. J. Anderson. 1981. High sodium in drinking water and its effect on blood pressure. American Journal of Epidemiology 114(6):817-826. https://doi.org/10.1093/oxfordjournals.aje.a113252

Hintz, W. D., L. Fay, and R. A. Relyea. 2022. Road salts, human safety, and the rising salinity of our fresh waters. Frontiers in Ecology and the Environment 20(1):22-30. https://doi.org/10.1002/fee.2433

Huang, J., C. E. Hmelo-Silver, R. Jordan, S. Gray, T. Frensley, G. Newman, and M. J. Stern. 2018. Scientific discourse of citizen scientists: models as a boundary object for collaborative problem solving. Computers in Human Behavior 87:480-492. https://doi.org/10.1016/j.chb.2018.04.004

Husson, F., J. Josse, and J. Pagès. 2010. Principal component methods - hierarchical clustering - partitional clustering: Why would we need to choose for visualizing data? Technical Report. Agrocampus-Ouest, Rennes-Angers, France. http://factominer.free.fr/more/HCPC_husson_josse.pdf

Innes, J. E., and D. E. Booher. 2018. Planning with complexity: an introduction to collaborative rationality for public policy. Second edition. Routledge, London, UK. https://doi.org/10.4324/9781315147949

Interstate Commission on the Potomac River Basin (ICPRB) and Virginia Department of Environmental Quality (VA DEQ). 2020. Virginia salt management strategy (SaMS): a toolkit to reduce the environmental impacts of winter maintenance practices. Interstate Commission on the Potomac River Basin, Rockville, Maryland, USA. https://www.novaregion.org/DocumentCenter/View/13054/SaMS_Toolkit_final?bidId=

Karl, H. A., L. E. Susskind, and K. H. Wallace. 2007. A dialogue, not a diatribe: effective integration of science and policy through joint fact finding. Environment: Science and Policy for Sustainable Development 49(1):20-34. https://doi.org/10.3200/ENVT.49.1.20-34

Kaushal, S. S., P. M. Groffman, G. E. Likens, K. T. Belt, W. P. Stack, V. R. Kelly, L. E. Band, and G. T. Fisher. 2005. Increased salinization of fresh water in the northeastern United States. Proceedings of the National Academy of Sciences 102(38):13517-13520. https://doi.org/10.1073/pnas.0506414102

Kaushal, S. S., G. E. Likens, P. M. Mayer, R. R. Shatkay, S. A. Shelton, S. B. Grant, R. M. Utz, A. M. Yaculak, C. M. Maas, J. E. Reimer, S. V. Bhide, J. T. Malin, and M. A. Rippy. 2023a. The anthropogenic salt cycle. Nature Reviews Earth and Environment 4(11):770-784. https://doi.org/10.1038/s43017-023-00485-y

Kaushal, S. S., G. E. Likens, M. L. Pace, S. Haq, K. L. Wood, J. G. Galella, C. Morel, T. R. Doody, B. Wessel, P. Kortelainen, A. Räike, V. Skinner, R. Utz, and N. Jaworski. 2019. Novel ‘chemical cocktails’ in inland waters are a consequence of the freshwater salinization syndrome. Philosophical Transactions of the Royal Society B 374(1764):20180017. https://doi.org/10.1098/rstb.2018.0017

Kaushal, S. S., G. E. Likens, M. L. Pace, J. E. Reimer, C. M. Maas, J. G. Galella, R. M. Utz, S. Duan, J. R. Kryger, A. M. Yaculak, W. L. Boger, N. W. Bailey, S. Haq, K. L. Wood, B. M. Wessel, C. E. Park, D. C. Collison, B. Y. I. Aisin, T. M. Gedeon, S. K. Chaudhary, J. Widmer, C. R. Blackwood, C. M. Bolster, M. L. Devilbiss, D. L. Garrison, S. Halevi, G. Q. Kese, E. K. Quach, C. M. P. Rogelio, M. L. Tan, H. J. S. Wald, and S. A. Woglo. 2021. Freshwater salinization syndrome: from emerging global problem to managing risks. Biogeochemistry 154(2):255-292. https://doi.org/10.1007/s10533-021-00784-w

Kaushal, S. S., G. E. Likens, M. L. Pace, R. M. Utz, S. Haq, J. Gorman, and M. Grese. 2018. Freshwater salinization syndrome on a continental scale. Proceedings of the National Academy of Sciences 115(4):E574-E583. https://doi.org/10.1073/pnas.1711234115

Kaushal, S. S., P. M. Mayer, G. E. Likens, J. E. Reimer, C. M. Maas, M. A. Rippy, S. B. Grant, I. Hart, R. M. Utz, R. R. Shatkay, B. M. Wessel, C. E. Maietta, M. L. Pace, S. Duan, W. L. Boger, A. M. Yaculak, J. G. Galella, K. L. Wood, C. J. Morel, W. Nguyen, S. E. C. Querubin, R. A. Sukert, A. Lowien, A. W. Houde, A. Roussel, A. J. Houston, A. Cacopardo, C. Ho, H. Talbot-Wendlandt, J. M. Widmer, J. Slagle, J. A. Bader, J. H. Chong, J. Wollney, J. Kim, L. Shepherd, M. T. Wilfong, M. Houlihan, N. Sedghi, R. Butcher, S. Chaudhary, and W. D. Becker. 2023b. Five state factors control progressive stages of freshwater salinization syndrome. Limnology and Oceanography Letters 8(1):190-211. https://doi.org/10.1002/lol2.10248

Kelly, W. R. 2008. Long-term trends in chloride concentrations in shallow aquifers near Chicago. Groundwater 46(5):772-781. https://doi.org/10.1111/j.1745-6584.2008.00466.x

Khan, S. S., A. Mukherjee, and N. Chandrasekaran. 2011. Impact of exopolysaccharides on the stability of silver nanoparticles in water. Water Research 45(16):5184-5190. https://doi.org/10.1016/j.watres.2011.07.024

Knox, C. B., S. Gray, M. Zareei, C. Wentworth, P. Aminpour, R. V. Wallace, J. Hodbod, and N. Brugnone. 2023. Modeling complex problems by harnessing the collective intelligence of local experts: new approaches in fuzzy cognitive mapping. Collective Intelligence 2(4). https://doi.org/10.1177/26339137231203582

Kramer, R. M., and T. R. Tyler, editors. 1996. Trust in organizations: frontiers of theory and research. Sage, Thousand Oaks, California, USA. https://doi.org/10.4135/9781452243610

Langridge, R., and C. K. Ansell. 2024. How did we get here? The evolution of a polycentric system of groundwater governance. Ecology and Society 29(1):34. https://doi.org/10.5751/ES-14830-290134

Lê, S., J. Josse, and F. Husson. 2008. FactoMineR: an R package for multivariate analysis. Journal of Statistical Software 25(1):1-18. https://doi.org/10.18637/jss.v025.i01

Leigh Star, S. 2010. This is not a boundary object: reflections on the origin of a concept. Science, Technology, and Human Values 35(5):601-617. https://doi.org/10.1177/0162243910377624

Lorenz, J., H. Rauhut, F. Schweitzer, and D. Helbing. 2011. How social influence can undermine the wisdom of crowd effect. Proceedings of the National Academy of Sciences 108(22):9020-9025. https://doi.org/10.1073/pnas.1008636108

Martin, B., and S. Via. 2020. Integrating water reuse into the US water supply portfolio. Journal of the American Water Works Association 112(1):8-14. https://doi.org/10.1002/awwa.1426

Nagendra, H., and E. Ostrom. 2014. Applying the social-ecological system framework to the diagnosis of urban lake commons in Bangalore, India. Ecology and Society 19(2):67. https://doi.org/10.5751/ES-06582-190267

Navajas, J., T. Niella, G. Garbulsky, B. Bahrami, and M. Sigman. 2018. Aggregated knowledge from a small number of debates outperforms the wisdom of large crowds. Nature Human Behaviour 2:126-132. https://doi.org/10.1038/s41562-017-0273-4

Newman, M. E. J. 2004. Analysis of weighted networks. Physical Review E 70(5):056131. https://doi.org/10.1103/PhysRevE.70.056131

Novotny, E. V., A. R. Sander, O. Mohseni, and H. G. Stefan. 2009. Chloride ion transport and mass balance in a metropolitan area using road salt. Water Resources Research 45(12):W12410. https://doi.org/10.1029/2009WR008141

Nozari, M. A., A. S. Ghadikolaei, K. Govindan, and V. Akbari. 2021. Analysis of the sharing economy effect on sustainability in the transportation sector using fuzzy cognitive mapping. Journal of Cleaner Production 311:127331. https://doi.org/10.1016/j.jclepro.2021.127331

Olofsson, K. L., P. P. Selvakumar, K. Peach, M. Leon-Corwin, S. A. Stormer, K. Gupta, N. Carlson, and M. Sibley. 2023. Effective stakeholder engagement in environmental problem-solving through group model building: an Oklahoma case study. Environmental Challenges 13:100755. https://doi.org/10.1016/j.envc.2023.100755

Olson, J. R. 2019. Predicting combined effects of land use and climate change on river and stream salinity. Philosophical Transactions of the Royal Society B 374(1764):20180005. https://doi.org/10.1098/rstb.2018.0005

Ostrom, E. 2009. A general framework for analyzing sustainability of social-ecological systems. Science 325(5939):419-422. https://doi.org/10.1126/science.1172133

Ostrom, E. 2010a. Analyzing collective action. Agricultural Economics 41(S1):155-166. https://doi.org/10.1111/j.1574-0862.2010.00497.x

Ostrom, E. 2010b. Beyond markets and states: polycentric governance of complex economic systems. American Economic Review 100(3):641-672. https://doi.org/10.1257/aer.100.3.641

Overbo, A., S. Heger, and J. Gulliver. 2021. Evaluation of chloride contributions from major point and nonpoint sources in a northern U.S. state. Science of the Total Environment 764:144179. https://doi.org/10.1016/j.scitotenv.2020.144179

Özesmi, U., and S. Özesmi. 2003. A participatory approach to ecosystem conservation: fuzzy cognitive maps and stakeholder group analysis in Uluabat Lake, Turkey. Environmental Management 31(4):518-531. https://doi.org/10.1007/s00267-002-2841-1

Özesmi, U., and S. L. Özesmi. 2004. Ecological models based on people’s knowledge: a multi-step fuzzy cognitive mapping approach. Ecological Modelling 176(1-2):43-64. https://doi.org/10.1016/j.ecolmodel.2003.10.027

Pahl-Wostl, C., and M. Hare. 2004. Processes of social learning in integrated resources management. Journal of Community and Applied Social Psychology 14(3):193-206. https://doi.org/10.1002/casp.774

Papageorgiou, E. I., and D. K. Iakovidis. 2013. Intuitionistic fuzzy cognitive maps. IEEE Transactions on Fuzzy Systems 21(2):342-354. https://doi.org/10.1109/TFUZZ.2012.2214224

Paul, R., P. Mukheibir, and S. Kenway. 2022. Transitioning to a one water paradigm in developing countries: an institutional framework to implement distributed recycled water systems. SSRN preprint. http://dx.doi.org/10.2139/ssrn.4122219

Peres-Neto, P. R., D. A. Jackson, and K. M. Somers. 2005. How many principal components? Stopping rules for determining the number of non-trivial axes revisited. Computational Statistics and Data Analysis 49(4):974-997. https://doi.org/10.1016/j.csda.2004.06.015

Rahaman, M. M., and O. Varis. 2005. Integrated water resources management: evolution, prospects and future challenges. Sustainability: Science, Practice and Policy 1(1):15-21. https://doi.org/10.1080/15487733.2005.11907961

Rice, J., and P. Westerhoff. 2017. High levels of endocrine pollutants in US streams during low flow due to insufficient wastewater dilution. Nature Geoscience 10(8):587-591. https://doi.org/10.1038/ngeo2984

Rippy, M. A., A. Deletic, J. Black, R. Aryal, J.-L. Lampard, J. Y.-M. Tang, D. McCarthy, P. Kolotelo, J. Sidhu, and W. Gernjak. 2017. Pesticide occurrence and spatio-temporal variability in urban run-off across Australia. Water Research 115:245-255. https://doi.org/10.1016/j.watres.2017.03.010

Sally, D. 1995. Conversation and cooperation in social dilemmas: a meta-analysis of experiments from 1958 to 1992. Rationality and Society 7(1):58-92. https://doi.org/10.1177/1043463195007001004

Salmeron, J. L. 2010. Modelling grey uncertainty with fuzzy grey cognitive maps. Expert Systems with Applications 37(12):7581-7588. https://doi.org/10.1016/j.eswa.2010.04.085

Schlüter, M., J. Hinkel, P. W. G. Bots, and R. Arlinghaus. 2014. Application of the SES framework for model-based analysis of the dynamics of social-ecological systems. Ecology and Society 19(1):36. https://doi.org/10.5751/ES-05782-190136

Schuler, M. S., M. Cañedo-Argüelles, W. D. Hintz, B. Dyack, S. Birk, and R. A. Relyea. 2019. Regulations are needed to protect freshwater ecosystems from salinization. Philosophical Transactions of the Royal Society B 374(1764):20180019. https://doi.org/10.1098/rstb.2018.0019

Schuler, M. S., and R. A. Relyea. 2018. A review of the combined threats of road salts and heavy metals to freshwater systems. Bioscience 68(5):327-335. https://doi.org/10.1093/biosci/biy018

Schwabe, K. A., I. Kan, and K. C. Knapp. 2006. Drainwater management for salinity mitigation in irrigated agriculture. American Journal of Agricultural Economics 88(1):133-149. https://doi.org/10.1111/j.1467-8276.2006.00843.x

Schwabe, K., M. Nemati, R. Amin, Q. Tran, and D. Jassby. 2020. Unintended consequences of water conservation on the use of treated municipal wastewater. Nature Sustainability 3(8):628-635. https://doi.org/10.1038/s41893-020-0529-2

Shahvi, S., P.-E. Mellander, P. Jordan, and O. Fenton. 2021. A fuzzy cognitive map method for integrated and participatory water governance and indicators affecting drinking water supplies. Science of the Total Environment 750:142193. https://doi.org/10.1016/j.scitotenv.2020.142193

Shalsi, S., C. M. Ordens, A. Curtis, and C. T. Simmons. 2022. Coming together: insights from an Australian example of collective action to co-manage groundwater. Journal of Hydrology 608:127658. https://doi.org/10.1016/j.jhydrol.2022.127658

Society of Professionals in Dispute Resolution. 1997. Best practices for government agencies: guidelines for using collaborative agreement-seeking processes. Report and Recommendations of the SPIDR Environment/Public Disputes Sector Critical Issues Committee. Society of Professionals in Dispute Resolution, Washington, D.C., USA. https://readingroom.law.gsu.edu/cgi/viewcontent.cgi?article=1008&context=seedgrant

State of Virginia. 2019. 9VAC25-260-140. Criteria for surface water. State of Virginia, Richmond, Virginia, USA. https://law.lis.virginia.gov/admincode/title9/agency25/chapter260/section140/

Steele, M. K., and J. A. Aitkenhead-Peterson. 2011. Long-term sodium and chloride surface water exports from the Dallas/Fort Worth region. Science of the Total Environment 409(16):3021-3032. https://doi.org/10.1016/j.scitotenv.2011.04.015

Susskind, L., S. McKearnen, and J. Thomas-Larmer, editors. 1999. The consensus building handbook: a comprehensive guide to reaching agreement. Sage, Thousand Oaks, California, USA. https://doi.org/10.4135/9781452231389

Thompson, D. A., D. M. Cwiertny, H. A. Davis, A. Grant, D. Land, S. J. Landsteiner, D. E. Latta, S. K. Hunter, M. P. Jones, H.-J. Lehmler, M. K. Santillan, and D. A. Santillan. 2022. Sodium concentrations in municipal drinking water are associated with an increased risk of preeclampsia. Environmental Advances 9:100306. https://doi.org/10.1016/j.envadv.2022.100306

U.S. Environmental Protection Agency (EPA). 1988. National recommended water quality criteria - aquatic life criteria table. U.S. Environmental Protection Agency, Washington, D.C., USA. https://www.epa.gov/wqc/national-recommended-water-quality-criteria-aquatic-life-criteria-table

U.S. Environmental Protection Agency (EPA). 2003a. Contaminant candidate list regulatory determination support document for sodium. EPA-815-R-03-15. U.S. Environmental Protection Agency, Washington, D.C., USA. https://www.epa.gov/sites/default/files/2014-09/documents/support_cc1_sodium_ccl_regdet.pdf

U.S. Environmental Protection Agency (EPA). 2003b. Drinking water advisory: consumer acceptability advice and health effects analysis on sodium. EPA 822-R-03-006. U.S. Environmental Protection Agency, Washington, D.C., USA. https://www.epa.gov/sites/default/files/2014-09/documents/support_cc1_sodium_dwreport.pdf

U.S. Environmental Protection Agency (EPA). 2019. Draft national water reuse action plan. U.S. Environmental Protection Agency, Washington, D.C., USA. https://www.epa.gov/waterreuse/draft-national-water-reuse-action-plan

U.S. Government Accountability Office (GAO). 2014. Freshwater: supply concerns continue, and uncertainties complicate planning. GAO-14-430. U.S. Government Accountability Office, Washington, D.C., USA. https://www.gao.gov/assets/gao-14-430.pdf

Vasslides, J. M., and O. P. Jensen. 2016. Fuzzy cognitive mapping in support of integrated ecosystem assessments: developing a shared conceptual model among stakeholders. Journal of Environmental Management 166:348-356. https://doi.org/10.1016/j.jenvman.2015.10.038

Velasco, J., C. Gutiérrez-Cánovas, M. Botella-Cruz, D. Sánchez-Fernández, P. Arribas, J. A. Carbonell, A. Millán, and S. Pallarés. 2019. Effects of salinity changes on aquatic organisms in a multiple stressor context. Philosophical Transactions of the Royal Society B 374:20180011. https://doi.org/10.1098/rstb.2018.0011

Villamayor-Tomas, S., G. García-López, and G. D’Alisa. 2022. Social movements and commons: in theory and in practice. Ecological Economics 194:107328. https://doi.org/10.1016/j.ecolecon.2021.107328

Voinov, A., and F. Bousquet. 2010. Modelling with stakeholders. Environmental Modelling and Software 25(11):1268-1281. https://doi.org/10.1016/j.envsoft.2010.03.007

Wald, D. M., E. A. Segal, E. W. Johnston, and A. Vinze. 2017. Understanding the influence of power and empathic perspective-taking on collaborative natural resource management. Journal of Environmental Management 199:201-210. https://doi.org/10.1016/j.jenvman.2017.05.030

Welle, P. D., and M. S. Mauter. 2017. High-resolution model for estimating the economic and policy implications of agricultural soil salinization in California. Environmental Research Letters 12(9):094010. https://doi.org/10.1088/1748-9326/aa848e

Welty, T. K. 1986. Effects of exposure to salty drinking water in an Arizona community: cardiovascular mortality, hypertension prevalence, and relationships between blood pressure and sodium intake. JAMA 255(5):622-626. https://doi.org/10.1001/jama.1986.03370050064022

Wilkinson, B. H. 2005. Humans as geologic agents: a deep-time perspective. Geology 33(3):161-164. https://doi.org/10.1130/G21108.1

Wilson, R. K., and J. Sell. 1997. “Liar, liar...”: cheap talk and reputation in repeated public goods settings. Journal of Conflict Resolution 41(5):695-717. https://doi.org/10.1177/0022002797041005005