Adapting the social-ecological system framework for urban stormwater management : the case of green infrastructure adoption

Stormwater management has long been a critical societal and environmental challenge for communities. An increasing number of municipalities are turning to novel approaches such as green infrastructure to develop more sustainable stormwater management systems. However, there is a need to better understand the technological decision-making processes that lead to specific outcomes within urban stormwater governance systems. We used the social-ecological system (SES) framework to build a classification system for identifying significant variables that influence urban stormwater governance decisions related to green infrastructure adoption. To adapt the framework, we relied on findings from observations at national stormwater meetings in combination with a systematic literature review on influential factors related to green infrastructure adoption. We discuss our revisions to the framework that helped us understand the decision by municipal governments to adopt green infrastructure. Remaining research needs and challenges are discussed regarding the development of an urban stormwater SES framework as a classification tool for knowledge accumulation and synthesis.


INTRODUCTION
The lack of well-integrated urban stormwater management strategies throughout the past century has left a heritage of environmental and social problems that policy-makers continue to deal with today.Municipal stormwater management plans in many developed countries have favored the use of gray infrastructure (e.g., sewer separation projects, deep storage tunnels, and regional treatment facilities).These engineering solutions can be costly, tend to promote centralized subsurface conveyance systems with end-of-pipe treatment, and often take years to complete.Despite major investments in stormwater infrastructure, urban areas continue to experience critical problems in managing water flows, including flooding, surface water impairment, and combined sewer overflows (USEPA 2004, National Research Council 2009, Coles et al. 2012).
Recent advances in stormwater management methods seek to enhance the sustainability of urban water systems.For instance, stormwater systems that include green infrastructure (GI), also known as low impact development, are recognized as a more sustainable approach.GI technologies are designed to protect or restore the natural hydrology of a site, capturing stormwater volume through the use of engineered systems that mimic natural hydrologic systems.Comprehensive GI programs can be implemented for a variety of outcomes, including flood control, surface water quality improvement, and water harvesting, in conjunction with a broad range of additional outcomes such as ecosystem restoration, air quality improvement, and urban heat reduction (Hatt et al. 2004, Villarreal et al. 2004, Walsh et al. 2005, Tzoulas et al. 2007).However, there are potential practical limitations for GI to achieve sustainable outcomes for municipalities, such as a limited capacity for storing and infiltrating stormwater.
The decision to adopt a comprehensive GI program is influenced by a complex array of social and biophysical factors.To explore such complexities, an urban water system can be understood as a social-ecological system (SES), or a collection of dynamic systems that coevolve through interactions among actors, institutions, and water systems, such as source water, groundwater, wastewater, and stormwater (Berkes et al. 1998, Holling andGunderson 2002).The stormwater flows and storage volumes within an urban water SES represent common-pool resources, in that water quality and available storage volumes are diminished as runoff flows through urban environments.These issues prompt the need for public authorities to establish various standards related to the management of stormwater.
A fundamental component of urban stormwater SESs is the role of technology as a critical interface between the social and ecological structures, which allows actors to shape different processes to achieve outcomes in system functioning (Ferguson et al. 2013).Technologies also act as a feedback mechanism between the social and biophysical systems of an SES.Walker et al. (2004) describe the potential of an SES intervention to create a new system when the conditions of an existing system are weakened.Stormwater management systems that are exclusively composed of gray infrastructure may result in urban water system weakening because these technology systems are considered neither sustainable nor sufficiently resilient to accommodate climatic changes, and may result in unforeseen outcomes such as high economic costs and environmental justice issues (Pahl-Wostl 2007, Novotny et al. 2010, Dominguez et al. 2011, Pyke et al. 2011, Wendel et al. 2011, De Sousa et al. 2012).Alternatively, extensive use of GI in stormwater management represents an opportunity for transformational shifts in urban water SESs away from point source solutions to decentralized, systematic techniques that may also bring multiple benefits to communities (Shuster and Garmestani 2015).
There is a need to more easily relate attributes and configurations of urban stormwater SESs to particular outcomes, such as the development of comprehensive GI programs.Several frameworks exist which conceptualize and operationalize SES dynamics, each of which may provide different types of diagnostic insights.Thus, an analyst must be clear about the aim and purpose of any

METHODS
The identification of attributes associated with GI adoption in municipal urban stormwater programs included several phases of data collection and analysis (Fig. 1).Exploratory work began with observations at GI summits in 2013 and 2014, in which delegates from U.S. municipalities were invited to discuss their respective community's GI programs.Extensive field notes from both meetings were coded line-by-line to identify factors that affected decisions to adopt municipal GI programs.The resulting codes were grouped into general categories of attributes that emerged during the analysis process.These categories were then incorporated into the SES framework, using first-and secondtier modifications, as suggested by McGinnis and Ostrom (2014), Epstein et al. (2013), and Vogt et al. (2015), as the initial framework.
Another stage of data collection included a literature review of original research efforts related to the adoption and implementation of GI in urban stormwater systems.Green infrastructure, green stormwater infrastructure (GSI), low impact design (LID), and best management practices (BMPs) are among the terms used for various suites of urban stormwater management technologies.We refer to GI, GSI, and LID technologies are "GI" because these terms are often used synonymously (Fletcher et al. 2014).Searches were carried out using Scopus, Web of Knowledge, and Google Scholar.Key words included in the literature review were "green infrastructure," "low impact development," "stormwater," and "municipal."Searches were conducted for studies published between 2000 and 2015.In total, 135 articles, theses, and reports were reviewed for their relevance to factors affecting the adoption and implementation of GI technologies for municipal programs.Reasons for exclusion included a study focus on adoption of water systems other than stormwater (e.g., drinking water, wastewater), or an exclusive focus on GI technology design attributes outside the context of municipal stormwater management program implementation (e.g., experimental findings).Studies were not excluded on the basis of study design, the scale or primary design goal of stormwater technologies discussed, nor the geographical location of the study; however, most studies reviewed were based in the United States or Australia.This process resulted in 83 studies that met the criteria, and thus formed the basis of the review.Qualitative document analysis techniques were used to identify factors that influence municipal GI programs in each of the collected studies.These methods often involve the development of a "protocol," which is tested on each unit of analysis and revised based on the quality and likely efficiency of the results (Altheide et al. 2008).The SES framework adapted in the initial research phase served as a beginning protocol that consisted of identified attributes related to GI adoption.After analyzing each study of the literature review, new findings were organized within the protocol.After all studies were analyzed, each study was reviewed a second time to test the protocol.This process resulted in the addition or redefining of second-tier SES framework attributes and the development of new third-, fourth-, and fifth-tier attributes presented in Table 1.Working definitions were developed for each attribute and are included in Appendix 1, along with at least three citations of illustrative studies collected in the literature review for the highest tier of each nested attribute added to the SES framework.Listed citations for each attribute are not presented as definitive authoritative sources nor as a comprehensive listing of all studies in which the attribute was identified.Rather, they represent examples of how scholars have applied the concept in other studies.

RESULTS
The SES framework organizes system attributes into nested tiers.The first-tier attributes of the SES framework, as defined for an urban stormwater management system, are summarized in Fig. 2.  (2014), Epstein et al. (2013), andVogt et al. (2015) that are specific for green infrastructure adoption in urban stormwater social-ecological systems are noted with italic font.The resource system (RS) is defined as an urban stormwater system; i.e., the system of water flows that results from wet weather.Multiple sets of resource units (RU) can be defined within an urban stormwater system, such as units of stormwater or the storage volumes available for stormwater throughout the system.The governance system (GS) includes the sets of rules agreed upon by national, state, and local organizations for managing urban stormwater.The actors (A) category includes individuals and groups that interact with the urban stormwater system.Multiple categories of actors can be defined, including http://www.ecologyandsociety.org/vol21/iss4/art19/individuals and groups that are involved in rule-making processes, and property owners that are affected by stormwater management decisions.Attributes from each of these categories provide inputs to action situations, where interactions (I) among actors transform these inputs into various outcomes, which can be measured by outcome criteria (O).Additional influences flow between the focal SES attributes and related ecosystems (ECO); ecological rules (ER); and social, economic, and political settings (S).

Fig. 2.
First tiers of social-ecological system framework for an urban stormwater social-ecological system (adapted from Ostrom [2007] and Epstein et al. [2013]).
Table 1 summarizes the changes made to the SES framework.A detailed summary of the modifications, along with working definitions and illustrative references, are provided in Appendix 1.Because the study focus is only on changes related to resource management programs, the findings led to detailed expansions of multiple governance system and actor attributes.Attributes for RU, ECO, ER, and S were not modified beyond second-tier changes suggested by McGinnis and Ostrom et al. (2014) and Vogt et al. (2015), though many of these attributes have direct and important effects on the design of municipal stormwater management programs.Additional studies on implementing various technological designs may result in a more detailed account for influential attributes in these categories.
Multiple third-, fourth-, and fifth-tier variables were added to describe various attributes of stormwater management technologies that are available to actors within the SES (A9), such as research support (A9.2), associated costs (A9.3), and perceptions of particular technologies (A9.4).The addition of third-, fourth-, and fifth-tier variables related to humanconstructed facilities (RS4) designates both the types and functionalities of existing and potential stormwater infrastructure.
A notable factor related to the construction of GI technologies is the availability of suitable locations for potential facilities (RS4.1.1),which is often associated with other factors such as local soil characteristics (RS8.1)(Shuster et al. 2014).Additional tiers allow for a detailed account of the assortment of resources and rules used by organizations to manage GI technologies.

DISCUSSION
In the broadest sense, integration of GI into an urban stormwater management system can be understood as the development of human-constructed facilities (RS4) across diffuse locations (RS4.1) using available technologies (A9) to alter the storage characteristics of an urban stormwater system (RS8).In developing this SES framework, additional third-, fourth-, and fifth-tier variables were needed to account for complex arrangements of social and biophysical factors that affect GI implementation.Operational-choice rules (GS6.1),such as ordinances, funding schemes, and comprehensive management plans, were found to be among the most complex factors.These rules are often further complicated by related SES regulations (GS6.1.4),such as zoning, building codes, and demolition practices (Lassiter 2007, Carter and Fowler 2008, Shuster et al. 2014).These related regulations are often managed by separate organizations, which may create barriers to GI implementation if the regulations are prohibitive.Property-rights systems that include prior-appropriation doctrines (GS7.1.1)can limit the choices of GI technologies (e.g., rainwater collection systems for some communities in the western United States) (Jensen 2008, Salkin 2009).
Funding was found to be among the most frequently cited barriers to GI (Godwin et al. 2008, Roy et al. 2008, Brown et al. 2009, Earles et al. 2009, Ruppert and Clark 2009, Stockwell 2009, Clean Water America Alliance 2011), most often in reference to the limited economic resources of enforcement organizations (GS5.1.1.2) and a lack of information on the cost-effectiveness of GI (A9.3).In the studies reviewed, stormwater management programs were enforced primarily by public organizations that selected stormwater management technologies to meet outcome criteria in a cost-effective manner.Environmental services associated with GI (A9.2.1.2),such as reducing urban heat island effects or promoting recreational opportunities, were cited as drivers for adoption when these benefits were quantifiable (Nowacek et al. 2003, Madden 2010).This suggests that it is difficult to maintain clear institutional boundaries when assessing the market and nonmarket value of GI because there may be additional benefits that GI can bring to a community beyond stormwater management.
The financial concerns of enforcement organizations are complicated by the design of effective stormwater utility funding schemes (GS6.1.2).Many funding schemes are predicated on the extent of total impervious area of urban land parcels because this metric has frequently been used to predict levels of surface water impairments due to stormwater runoff (Booth andJackson 1997, Parikh et al. 2005).However, studies suggest that the subset of impervious surfaces that route runoff directly to surface waters via sewer pipes, known as directly connected impervious area or effective impervious area, may be responsible for most surface http://www.ecologyandsociety.org/vol21/iss4/art19/water impairments due to urbanization (Brabec et al. 2002, Walsh 2004, Walsh et al. 2005, Roy and Shuster 2009).Thus, stormwater utility funding schemes based on total impervious area rather than effective impervious area may not lead to desired SES outcomes.Additional limitations of utility funding schemes may develop if financial credits for GI are calculated as a one-time credit based on the initial installation without including ongoing performance and maintenance criteria, or if residential property owners are not included in financial incentive programs (Parikh et al. 2005).(Melosi 1999, Smith et al. 1999).Urban water infrastructure choices may also lead to unforeseen consequences over long periods.For example, combined sewer systems were once deemed to be the most appropriate choice for urban settings due to factors such as cost-effectiveness and availability of water courses for overflow disposal (Tarr 1979).These decisions have left a legacy of water pollution problems for many communities, as combined sewer overflows continue to impair surface waters and create human health hazards (USEPA 2004, Donovan et al. 2008, Gooré Bi et al. 2015).By developing a comprehensive categorization of technological attributes within an SES framework, policy-makers will be better equipped to make wellinformed decisions concerning technology selection for desired urban water SES outcomes.
Though additional characterizations were not added within several second-tier categories, such as resource units (RU) and outcome criteria (O), attributes in these categories have important implications for stormwater management technology decisions.For instance, stormwater management plans are traditionally designed according to the spatial and temporal distribution of stormwater flows in an urban area (RU7), which will be affected by changes in local precipitation patterns (RU2).The spatial and temporal distribution of stormwater volumes within an urban setting places clear boundaries on which technologies should be considered and where they should be situated in an urban setting (Askarizadeh et al. 2015).Additionally, the criteria used to select stormwater management technologies, such as relative costeffectiveness or ecological performance measures, will often strongly influence enforcement officials' decision-making processes (Flynn et al. 2014).Expansion of these attribute categories may be necessary when considering research questions related to the design of specific stormwater technologies or the influence of particular outcome criteria.
Some limitations of the modified framework attributes should be noted.Because several programs reviewed in the literature are in early phases of development, some SES framework attributes are likely relevant to only nascent GI implementation.However, an analysis of GI technologies in urban stormwater SESs over longer timescales may result in other variables having a greater effect (Brown et al. 2013).Much of the research we reviewed relies on case study methods such as the solicitation of particular actors' perceptions.Thus, some factors listed may pertain to specific actors or institutions, such as engineering firms, municipal officials, developers, or community residents.Additional studies can provide further insights into the possibility of shared, complementary interactions among actors within specific situations that result in the development of successful GI programs.It is also important to note that while the literature review was not restricted to studies from particular geographic locations, most studies were based in the United States or Australia, which prescribe similar stormwater governance structures.Researchers who use the revised SES framework in studies of community-based stormwater governance regimes may need to add more detailed characterizations of particular attributes (such as property-rights systems or collective-choice rules), or may need to omit others (such as particular operational rules).

CONCLUSION
We developed a modified SES framework to recognize the combinations of influential variables related to the development of municipal urban stormwater management programs that feature extensive use of GI technologies.The modifications made to the SES framework revealed the need for additional attribute tiers related to variables such as available technologies, actor characterizations, and operational-choice rules.Our findings demonstrate that affecting change in the built structure of urban stormwater systems involves multiple interacting attributes of the actors and governance systems within an SES.
The framework we developed should be interpreted as a flexible, proposed framework rather than a definitive set of variables that will be relevant in all urban stormwater SES cases.Other studies highlight qualities of particular attributes within adapted SES frameworks to explore dynamic interactions and outcomes of interest (Fleischman et al. 2010, Basurto et al. 2013, Nagendra and Ostrom 2014, Leslie et. al 2015, Partelow and Boda 2015).
The revised framework we presented highlights key factors of GI adoption that can be further explored using various theories and models to assess outcomes of interest related to urban stormwater SESs seeking to adopt GI technologies (Flynn et al. 2014).Tiers may be added or omitted to accommodate particular theories and research questions.
There is a need to explore the specific, contextual factors affecting the decision to adopt particular management approaches in urban stormwater SESs.The growing popularity of GI systems across municipalities carries a risk that these technologies will be perceived as a panacea for stormwater management (Ostrom 2007).However, there continues to be a need for a more sophisticated quantitative understanding of how GI technologies bring out particular SES outcomes.Neither a fully green nor entirely gray infrastructure approach to stormwater management will likely be optimal at any location.Instead, long-term solutions must be built around improved knowledge of factors influencing water quantity and quality in urban areas, and leveraging the services and capacities of both gray and green infrastructure.Such understanding should include the consideration of the unique characteristics of a particular urban water SES.

Resource Units (RU)
Characteristics of the units extracted from a resource system, which can then be consumed or used as an input in production or exchanged for other goods or services.

Resource unit mobility
Ability for resource units to move throughout the resource system R U2

Growth or replacement rate
Absolute or relative descriptions of changes in quantities (x) of resource units over time (t)

Number of units
Amount of individual resource units in resource system R U6

Distinctive markings
Characteristics that can be identified in resource units and affect actors' behavior toward them

Spatial and temporal distribution
Allocation patterns of resource units across a geographic area in a particular time period

Resource systems (RS)
The biophysical system from which resource units are extracted and through which the levels of the focal resource are regenerated by natural dynamic processes McGinnis (2011) † RS1 Sector Characteristic(s) of a resource system that distinguishes it from other resource systems Ostrom (2007) † RS2

Clarity of system boundaries
Biophysical characteristics that make feasible for actors to determine where the resource system starts or ends

Economic resources
Funds available to an organization that are used for the creation, operation and maintenance of the stormwater management program.Funds may be generated through a variety of means such as a variety of taxes, service charges, exactions, assessments, grants, loans, and bonds.

Operationalchoice rules
Set of regulations or principles governing the implementation of practical decisions by individuals authorized or allowed to take these actions, often as a result of collective choice processes

Stormwater ordinances and regulations
Sets forth public policies directly related to drainage, flood control, and water quality aspects of stormwater, as well as the legal framework for permitting implementation of the controls.

Technical basis
Performance standards, design criteria and information provided by rule-making organizations to assist designers in complying with ordinances and regulations.

Related regulations
Sets forth public policies that affect the implementation of decisions related to stormwater management (e.g., zoning codes, building codes).

Collective-choice rules
Set of regulations or principles governing institution creation and policy decisionmaking by actors who are authorized (or allowed) to do so, often as a result of constitutional-choice processes

Constitutionalchoice rules
Set of regulations or principles governing the processes though which collectivechoice stormwater management procedures are defined and legitimized, often resulting in a state or federal guideline or law

Types of knowledge
Types of knowledge actors use to understand SES A 7. 1.1

Traditional ecological knowledge
Degree to which actors make use of the cumulative body of knowledge, practices and beliefs evolving by adaptive processes and handed down through generations by cultural transmissions about the relationship of living beings (including humans) with one another and with their environment

Local ecological knowledge
Degree to which actors make use of knowledge and beliefs held by a specific group of people related to their environment acquired over the lifetime of individual generations Olsson and Folke  (2001)

Fig. 1 .
Fig. 1.Sequence of data collection and analysis.(SESF: socialecological system framework) Regulations or principles that specify the values of the working components of an action situation, each of which has emerged as the outcome of interactions in an adjacent action situation at a different level of analysis or arena of choice.

Table 1 .
Modified second-through fifth-tier attributes of the urban stormwater social-ecological system (SES) framework.Factors modified from McGinnis and Ostrom Technological attributes are described in both the social and ecological domains of the SES framework.While it has been argued that there is no need to create a separate technological domain (McGinnis andOstrom 2014), we demonstrate a need to more fully develop robust descriptions of technological attributes within urban stormwater SESs because these attributes act as key feedback mechanisms between the social and ecological domains.Historically, technological innovations in urban water SESs have been shown to bring about desired social and ecological regime shifts, such as a reduction in water-borne illness and a decrease in the frequency of algal blooms due to eutrophic states of receiving waters

Table 1 :
http://www.ecologyandsociety.org/vol21/iss4/art19/Modified social-ecological system framework for urban stormwater systems with working definitions and references.Modified framework for green infrastructure adoption in urban stormwater social-ecological systems.References provided as working definitions and illustrative examples from the literature.
Reference for attribute definition ‡ Reference that provides illustration of example of attribute's relationship to green infrastructure adoption in urban stormwater SESs