Q Methodology is a structured approach for discourse analysis that uses both qualitative and quantitative techniques to reveal dominant discourses, as well as consensus and divergent perspectives using a rank ordering activity and factor analysis (Brown, 1980; Webler et al., 2009; Zabala et al., 2018). A discourse is defined as the way an individual views, or forms conceptions of, the world (Barry and Proops, 1999), and can be elicited by discussion and the rank ordering activity. Q Methodology initially was implemented in psychological research (Watts and Stenner, 2005; Webler et al., 2009), but has more recently been used to study environmental and resource management issues (e.g., Gruber, 2011; Nelson et al., 2022) and stormwater governance (Arik, 2022). Because Q methodology aims to understand the internal frame of reference of individuals (rather than extrapolating to populations), Q studies can be successful with sample sizes between 10 and 40 participants (e.g., Cairns, 2012; Sandbrook et al., 2013) and does not require random sampling (Zabala et al., 2018). Our sample size (n = 21) is within the typical range of Q sort and captured a diversity of experiences (Watts and Stenner, 2005).

We used Q Methodology to investigate perspectives of stormwater experts in the Puget Sound region of Washington, U.S.A. Specifically, we focused on perspectives of experts regarding the efficacy and priority of a diverse suite of potential solutions to address stormwater and associated pollution. The workflow of our Q study consisted of 4 steps: (1) creation of the Q set (or the list of statements to be ranked) and P set (or the study participants); (2) ranking of the Q set (creating a “Q sort”); (3) factor analysis of Q sorts; (4) interpretation of factor analysis (Figure 1) (Brown et al., 1999). Thus in this study, the Q set were stormwater solutions which were ranked and sorted by regional stormwater experts, the P set. To sort the Q Set, participants were tasked with placing the statements into a predetermined semi-normal distribution along a spectrum of lower priority to higher priority, which we refer to as the Q board (Figure 2) (Watts and Stenner, 2005; Webler et al., 2009). As illustrated by Figure 2, participants were able to place a single statement on the Q board as their highest priority, two statements as their second highest priority, and so on. Factor analysis of the completed Q sorts reduced dimensionality of the data and created idealized Q sorts for each group of individuals. Idealized Q sorts disclose common viewpoints regarding stormwater solutions held by group members. Below we provide additional details regarding Q Methodology.

In developing our Q set, we heeded the advice of Stephenson (1953) to be inclusive of a broad range of perspectives. To accomplish this, we hosted a workshop in partnership with The Water 100 Project²—a collaboration between The Nature Conservancy and Puget Sound Partnership that details the 100 most impactful solutions for a clean and resilient Puget Sound. The workshop was attended by experts in stormwater infrastructure, planning, and research. Workshop attendees were employed by tribes, city or state agencies, NGOs, and universities. The workshop took place in Winter of 2021 was conducted virtually due to the COVID pandemic. We used key concepts and sentiments expressed at the workshop about stormwater solutions with language used by the Water 100 Project to create the final Q Set of 29 statements: 13 Green Infrastructure solutions, 8 Gray Infrastructure solutions, and 8 Source Control solutions (Table 1). These concepts and sentiments helped us narrow down our Q set from the 100 solutions listed in the Water 100 Project. The Q Set was developed and finalized prior to participant recruitment for the P Set.

We categorized our Q Set based on both U.S. Environmental Protection Agency (EPA) definitions, and sentiments from participants expressed in our workshop. Green Infrastructure definitions can vary, the EPA describes it as “the range of measures that use plant or soil systems, permeable pavement or other permeable surfaces or substrates, stormwater harvest and reuse, or landscaping to store, infiltrate, or evapotranspirate stormwater and reduce flows to sewer systems or to surface waters”. Individuals who attended the stormwater workshop defined Green Infrastructure more narrowly, focusing on the plant and soil systems mentioned in the EPA description. Therefore, for the purpose of this study we chose to define green infrastructure as solutions that are nature-based (i.e., they incorporate plants and soil into their design.) This excludes solutions such as permeable pavement and stormwater harvesting which we defined as gray infrastructure, because they use composite concrete and rain barrels or cisterns, respectively.

Following the guidance of Webler et al. (2009), we recruited participants for the Q study (the P set) who represented a variety of opinions and perspectives of those involved in stormwater management. Our participants consisted of stormwater experts which we defined as someone who is professionally involved in stormwater policy, science or management. The participant pool consisted of individuals in city, county, and state level of governments, environmental non-profit planners, and private sector green-stormwater infrastructure CEOs. 43% of our participants were government employees and 57% were employed by NGOs, offering us a wide breadth of opinions from main interest groups regarding stormwater management. Many of these individuals participated in county sponsored regional stormwater technical workshops, urban water solutions workshop, and other Puget Sound focused stormwater toxics workshops. We used stratified chain referral (Biernacki and Waldorf, 1981) to enlist participants. Twenty-one individuals participated in the study—a sample size within the typical range of Q studies that meaningfully captures a diversity of views (Watts and Stenner, 2005). Participation was voluntary and no money or goods were exchanged.

We performed the interviews and Q sort using virtual video conferencing. After receiving consent from 20 of the 21 participants, we recorded and later transcribed the interviews to capture explanations and the thought process used by participants during the exercise. Prior to conducting the Q sort, we asked each participant basic demographic questions and to self-identify as a scientist, practitioner, manager or something else. Participants were also asked to describe their area of stormwater expertise.

Next, we guided each participant through the Q sort using the online platform, Q Method Software. Participants were given the prompt, “Consider each of the stormwater solutions identified in each of the 29 statements and rank them as lower, high, and highest priority for implementation”. We chose to use the language “lower, high, and highest” rather than “lower, medium, and high” because pilot Q sorts revealed that stormwater experts tended to prioritize solutions as low or high, and thus the use of “highest” forced individuals to identify those solutions that should receive highest priority. Experts first performed a preliminary sort where each of the stormwater solutions were ranked as lower, high, or highest priority. This was followed by a detailed sort where participants placed the pre-sorted statements onto a Q board where −4 represented their lowest ranked and +4 represented their highest ranked solutions (Figure 2). Statements placed in the same column were treated as the same rank (i.e., all statements in the −2 column were of the same prioritization) (Figure 2). Throughout the exercise, we prompted participants to explain their rationale for their sorting choices—the details of which were transcribed from the recordings.

Following the completion of the Q sort activity, we asked participants a series of questions to get them to articulate the ecological or human outcomes that they considered during their sort. We also asked whether there was anything especially challenging during this process. The average total duration of the Q sort activity and interview was 42 min.

Q sorts (n = 21) were analyzed using the Q Method Software³ (Lutfallah and Buchanan, 2019). We used Centroid Extraction (CE) factor analysis to reduce the data into factors which were then subjected to varimax rotation to associate each individual with only one factor. Based on the amount of variability explained, inspection of scree plots, eigenvalues, and the interpretability and theoretical significance of the factors (Brown et al., 1999; Watts and Stenner, 2005), we extracted three factors (Appendix 1: Supplementary Figure A1).

The factors represent idealized Q sorts reflecting the dominant discourse of the group. Each statement has a z-score which represents the weighted average of the scores that similar respondents gave to the statement (Zabala et al., 2018), as well as an integer that represents where the statement was placed in the idealized Q Sorts. Each individual participant also has a factor loading score indicating how closely they associated with the idealized Q sorts (Zabala et al., 2018). Our analysis also identified distinguishing and consensus statements—statements that are statistically different (p ≤ 0.05) or similar to other perspectives, respectively. We interpreted discourses by comparing the ranking of statements among factors, overlapping distinguishing and consensus statements among the factors, and qualitative analysis of the transcribed interviews.

In addition to identifying statements that distinguish each factor, we identified consensus statements—statements that were common across discourses. Out of the 29 statements, four were identified as statistical points of consensus among all three discourses (Table A3). These four consensus statements were all types of green infrastructure: Green roofs and walls, Increase voluntary buyouts of repeatedly flooded properties, Implement groundwater recharge areas, and Floodable parks and flex space (Table 1, Table A3). Of the four consensus statements, Floodable parks and flex space ranked the highest while Green roofs and walls ranked the lowest (Figure 3, Table A3). It is important to note that all four consensus statements fell in the moderate priority range. No statements falling in the extreme ends of the Q Board were agreed upon across the three discourses.

An important focus of the emergent discourses was the distinct needs of urban and rural locales. Urban areas often dominate the discussion surrounding stormwater management because of the volume of surface runoff from impervious surfaces and larger populations compared to rural areas (Goonetilleke et al., 2005; Walsh et al., 2005; Cousins, 2017); however, there is still the need for stormwater solutions in rural areas. Priorities for stormwater management differ between urban and rural areas because of differences in land use practices (Paule-Mercado et al., 2017). In rural areas there is less impervious surface, lower population density and associated traffic pollution than urban regions (Messager et al., 2021), but rural, agricultural landscapes generate nutrients and pesticides in stormwater runoff that need to be addressed (Zhang and Zhang, 2011). These differences in pollution sources require different stormwater solutions. Solutions such as street sweeping that can effectively remove pollutants from roadways (e.g., Järlskog et al., 2020) do not have the same impact in rural areas compared to urban areas because there are fewer cars contributing to vehicle pollution. Thus, stormwater solutions favored by experts with urban expertise may not translate well to more rural areas. This point was emphasized by one of our interviewees:

In contrast to urban areas, rural areas focus on implementation of roadside bioswales to effectively treat roadway stormwater runoff (AECOM, 2020). While both street sweeping and roadside bioswales remove roadway contaminants, street sweeping focuses on vehicle related contaminants including PAHs, inhalable particles, and metals (Järlskog et al., 2020) while roadside vegetated bioswales have been utilized to remove pesticides in agricultural runoff (Cooper et al., 2004; Zhang and Zhang, 2011).

These examples highlight a common theme found in all three of our discourses; the specific stormwater intervention should be selected with specific goals in mind. There is no “one size fits all” solution that can be implemented in any situation to provide immediate and effective benefits to water quality. The context-specific nature of stormwater solutions can make prioritization decisions difficult, even for stormwater experts, and motivates the need to expose underlying assumptions and unconscious biases.

Our results reveal that even stormwater experts with many years of experience have difficulty choosing which solutions to prioritize. All stormwater solutions are important for different reasons and can be more or less efficient in different settings. The difficulty is in deciding which solutions to implement, where to implement it, and for what purpose or goal in mind. Source controls, such as pharmaceutical management and household best practices, prevent specific pollutants that have known adverse effects from entering the environment. However, while source control can be effective, study participants diverged in their opinions of this class of solutions because there are many different compounds stemming from different sources that vary widely in their impacts (Pitt et al., 1999; Davis et al., 2009). For instance, the tire rubber-derived compound 6PPD-quinone has been demonstrated to be acutely toxic in coho salmon (O. kisutch) (Tian et al., 2021) resulting in mortality of adult fish prior to spawning. However, the impact of this compound is heterogenous, and its toxicity may be less in some other species (e.g., McIntyre et al., 2021; Brinkmann et al., 2022). Given its ubiquity and toxicity (Tian et al., 2022), it is rational to remove this pollutant at its source; however, with more than 5,000 different unique pollutants entering Puget Sound annually (Peter et al., 2018; Saifur and Gardner, 2021; Tian et al., 2022), some of the participants in our study noted that it can be difficult to decide on which pollutant to focus on first and why.

How do managers decide between solutions that provide known ecological protection versus solutions that protect humans from environmental contaminant exposure and improve human wellbeing? In this study we found two divergent views on this question; “Green Action Now” prioritized stormwater solutions with nature-based components because they have co-benefits related to human health and environmental justice, while “First Things First” prioritized solutions that prevented contaminants with adverse ecological and human health effects from entering the environment. There is no objectively correct answer or viewpoint because both achieve stormwater quality improvements that provide ecological and human health benefits. The ways in which solutions go about achieving these outcomes is where the discourses surrounding this decision diverge.

One of the tensions that arose in our interviews of stormwater experts was between “proven” solutions such as gray infrastructure and source control vs. non-traditional, nature-based green infrastructure solutions. A typical perspective by experts who favored green infrastructure is expressed by this interviewee:

This participant and the “Green Action Now” group felt that green infrastructure solutions would result in the greatest overall ecosystem improvement because of the broad ranging benefits provided by natural systems and ecosystem services (Coutts and Hahn, 2015). Thus, the group felt that green infrastructure should be higher priority than solutions focusing specifically on the quality of the stormwater. These experts acknowledge that while solutions like industrial area source control are necessary and important to include in stormwater management plans, it is important to look beyond the immediate outcome and take into account objectives that may be further removed from the solution.

Another participant takes a different stance in favor of source control and gray infrastructure solutions stating:

This participant put more value on the knowledge that a solution is going to produce known and measurable results rather than risk implementing a nature-based solution that may not be as effective at doing the same thing for stormwater. Some stormwater managers believe that evidence supporting the efficiency of green infrastructure and its longevity is insufficient to advocate for its implementation (Copeland, 2014). The question therefore becomes, do the co-benefits associated with green infrastructure outweigh their potential inefficiencies or should gray infrastructures and source control solutions that produce known measurable effects without co-benefits be prioritized? More research comparing the stormwater mitigation capabilities of green infrastructure to traditional source control and gray infrastructure solutions could help to answer this question.

Policies for conservation, sustainability and land management often seek to achieve multiple objectives (e.g., Feng et al., 2004; Stagnari et al., 2017; Gardali et al., 2021). Multiple Benefit Conservation is an emerging approach in conservation that measures success by evaluating outcomes against multiple predetermined ecological and societal objectives (Gardali et al., 2021). Conservation efforts that promote and focus on multiple benefits can generate opportunities for inclusivity of those with diverse cultures, values and worldviews (Gould et al., 2018; Gardali et al., 2021). Categorizing stormwater solutions as Multiple Benefit Conservation is yet to be discussed in depth, but the inclusion of more holistic goals and outcomes for stormwater solutions is gaining traction (Wang et al., 2016; Jessup et al., 2021). Stormwater solutions that benefit receiving waters but also produce multiple benefits that address other ecological and human wellbeing outcomes may provide opportunities for additional federal, state, and local funding opportunities. Indeed, policies that benefit human health, or environmental justice may have larger budgets than those only promoting green infrastructure stormwater action (Copeland, 2014). Instead of focusing on stormwater as the center for the solution, what if these solutions are framed with the intent of promoting human and community wellbeing and have co-benefits to stormwater quality and quantity? This is an example of diversifying the funding and resource allocation for stormwater solutions to be implemented. Currently there is no federal funding specifically toward green infrastructure projects, typically smaller municipalities offer incentives for private landowners to implement green infrastructures (Kirschbaum and Lowry, 2012; Copeland, 2014). Our Q sort transcribed interviews revealed that financial and political feasibility are common factors in stormwater management decision making.

Green infrastructure solutions offer a plethora of benefits to human health, community strength, and climate change mitigation (Copeland, 2014; Jessup et al., 2021). Solutions that increase green space such as floodable parks, tree planting, and neighborhood stormwater facilities not only aid in stormwater filtration and carbon uptake, but they also aid in physical and mental human health (Mackenzie and McIntyre, 2018). Increased greening is also linked to improvements in standardized test scores, concentration in children with ADHD, hospital recovery time, and physical activity in residents (Mackenzie and McIntyre, 2018). Not only does green infrastructure implementation provide benefits to biodiversity, increased green space has also been linked to lowered crime rates and an increased sense of community (Dunn, 2010). One participant from “Green Action Now” stated the following about the link between green space and community strength:

Studies have shown that neighborhoods with higher occupancy of people of color and lower socioeconomic status have less access to green space (Wolch et al., 2014; Voelkel et al., 2018), more exposure to environmental hazards and pollutants (Adamkiewicz et al., 2011; Tessum et al., 2019; Schell et al., 2020), and increased incidence of negative effects associated with urban heat islands (Jenerette et al., 2011; Jesdale et al., 2013; Mitchell and Chakraborty, 2015; Schell et al., 2020). Additionally, access to safe greenspace in low income and non-white communities is lower than in more affluent white neighborhoods (Day, 2006; Williams et al., 2020). There is potential to alleviate the uneven distribution of stressors to these disadvantaged communities by implementing nature-based green infrastructure stormwater solutions. Solutions such as floodable parks, green roofs and walls, and tree planting can increase green space, remove impervious surfaces and increase urban canopy cover thereby reducing the urban heat island effect and increasing carbon sequestration (Dunn, 2010; Foster et al., 2011; Block et al., 2012; Balany et al., 2020; Jessup et al., 2021). These solutions are designed to reduce stormwater peak flows and reduce contaminants by natural infiltration but can also alleviate climate-related inequalities to disadvantaged communities. The implementation of specific stormwater solutions that have a multiple benefit conservation framework can simultaneously benefit stormwater, human health, and promote environmental justice by providing underrepresented low income and non-white communities with the multiple benefits associated with green stormwater infrastructures.

Q studies, such as this one, should not be extrapolated since the small sample size and non-random participant selection mean perspectives found in this study may not encompass all perspectives and worldviews (Brown et al., 1999). However, Q methodology can highlight minority perspectives within a group (Watts and Stenner, 2005); this is useful in uplifting underrepresented voices in the stormwater realm. In our work this is particularly evident for those participants who linked solutions to human health and wellbeing. Future studies focused on individuals from different sectors (e.g., human health, urban planning, transportation) or other roles (e.g., policy makers and elected officials) may provide additional insight about prioritization of stormwater solutions. Such additional perspectives could help provide further insight about tradeoffs and the multiple benefits associated with green infrastructure (Andersson et al., 2014; Copeland, 2014; Coutts and Hahn, 2015).

Solving the stormwater problem is clearly a difficult task, with obstacles at all levels of implementation from designers and land managers to policy decision makers (Qiao et al., 2018). Embracing diverse discourses will help to spark discussion, clarify foundational areas of disagreement, and can sharpen focus on the shared goals for stormwater mitigation. While there are many perspectives and opinions on specific solutions, the ultimate goal for stormwater management is shared: equitably improve water quality for both nature and people. Given the rapid urbanization of the globe and the accompanying increase in stormwater contamination, the integrity of coastal ecosystems depends on learning from diverse perspectives, and then, as one participant urged, “doing something.”