Watershed or bank-to-bank? Scales of governance and the geographic definition of Great Lakes Areas of Concern

Introduction

The governance of water quality involves multiple dimensions of geographic scale (Norman et al., 2015). Once treated as an exclusively local problem to be managed by municipalities or other local jurisdictions, over the past two centuries water quality has increasingly become a matter of national and global concern (Vörösmarty et al., 2015). At the same time, and for a variety of reasons, the watershed—also known as the river or drainage basin—has emerged as a widely preferred scale for governing the quality of freshwater resources. However, because the boundaries of watersheds typically stand in tension with conventional political and administrative boundaries, the implementation of governance at the watershed scale has been challenging in practice. The resulting governance of water quality has become a complex, multi-scalar endeavor, placing territorial jurisdictions and ecosystems in complicated relationships with each other.

The Great Lakes Areas of Concern (AOCs) program provides an instructive example of this scalar complexity. The Great Lakes Water Quality Board began naming and studying local “problem areas” a year after the first Great Lakes Water Quality Agreement of 1972 between the United States and Canada (International Joint Commission, 1975); it redesignated these as AOCs in 1981 (Great Lakes Water Quality Board, 1981).¹ At the scale of the Great Lakes Basin, the binational program recognized a set of highly contaminated and ecologically degraded sites as AOCs, each of which would require Remedial Action Plans, or RAPs (Great Lakes Water Quality Agreement, 1987, 2012; Great Lakes Water Quality Board, 1985). The aim of these RAPs is “restoring beneficial uses that have become impaired due to local conditions” at the AOCs (Great Lakes Water Quality Agreement, 2012: 21). Annex 2 of the 1987 amendments to the Great Lakes Water Quality Agreement formalized the “Area of Concern” as the local scale targeted by RAPs. The amendments did not, however, prescribe specific boundaries or spatial extents for the AOCs. For example, should they be defined as watersheds, as sub-watersheds, as watercourse segments, or as some other kind of space? Answering this question was a task undertaken early in the RAP process (Great Lakes Water Quality Board, 1987), identified in the 2012 version of the Agreement in explicitly multi-scalar terms:

For each AOC, the Parties [Canada and the United States], in cooperation and consultation with State and Provincial Governments, Tribal Governments, First Nations, Métis, Municipal Governments, watershed management agencies, other local public agencies, and the Public, shall develop and implement a systematic and comprehensive ecosystem approach to restoring beneficial use (Great Lakes Water Quality Agreement, 2012: 22).

Although the Agreement suggests that the two national Parties hold the ultimate responsibility for defining the boundaries of AOCs, decisions about defining these boundaries—and what it means “to implement a systematic and comprehensive ecosystem approach”—are delegated in practice to other scales of governance, including states, provinces, and local stakeholder groups (Great Lakes Water Quality Board, 1985; Hartig et al, 2020; Jetoo et al., 2015; MacKenzie, 1996). The AOC program thus involves complex relationships among what Cohen (2015) distinguishes as multiple spatial, ecological, and social scales, including the international watershed of the Great Lakes Basin and the numerous sub-watersheds within it, which in turn overlap with the territories of states, provinces, reservations and reserves, and a variety of local jurisdictions.

This article presents findings from a study of the subset of AOCs either located in the United States or straddling the U.S.-Canadian border, focusing on the following questions: how do the definitions of AOC boundaries vary within the United States?; what accounts for this variation?; and, how have scalar definitions of the AOC affected the RAP process? Through an online Qualtrics survey and a series of semi-structured phone interviews, we found that among U.S. and binational AOCs, the spatial and hydrological scales of AOC boundaries are related in important ways with the scale of governance at which decisions about RAPs are made. Although most of these AOCs are defined as “bank-to-bank” stretches of waterways and nearshore areas, several are defined as watersheds or partial watersheds, and a few are hybrids with elements of both definitions. Our interviews suggested that this variation is in part associated with the nature of the problems at specific AOCs, but it is also driven by state-level differences in policy and practice, particularly with respect to the latitude granted to local stakeholder groups to participate in defining AOC boundaries. The interviews also revealed that defining AOC boundaries has in several cases been a matter of controversy and conflict, with some stakeholders arguing for a more expansive definition and others for a narrower definition. Such conflicts have been resolved in different ways, with significant implications for remedial progress.

Our findings contribute not only to scholarship on the governance of the Great Lakes but also to broader conversations about the political dimensions of the watershed as a scale of water governance (e.g., Cohen, 2012; Cohen and Bakker, 2014; Harris and Alatout, 2010; Norman et al., 2015; Swyngedouw, 1999). Our study highlights a case in which the appropriateness of the watershed scale has frequently been a focus of stakeholder debate. In the AOC program, the watershed has not been universally—or even commonly—adopted as the consensus scale of freshwater governance; in effect, it has competed with another local scale also defined in part by natural hydrological boundaries: bank-to-bank segments of watercourses. Unlike the watershed, which includes both water and the land that drains into it, such segments consist only of the water and sediment between river or stream banks. Moreover, we contend that understanding the uneven embrace of the watershed as a scale of water governance requires attending not only to specific governance objectives, but also to variations in the relationships between local and sub-national scales in governance programs. The article begins with a review of recent literature on scale in the governance of water quality and additional background on the Great Lakes Areas of Concern program. After introducing our methods and data, we present and discuss research findings. We conclude with reflections on implications for potential future research on scale and freshwater governance.

Scale and the governance of water quality

The geographic scales at which societies manage and govern problems of water quality and surface water pollution have shifted over time, and they remain a focus of debate. As Vörösmarty et al. (2015: 478) point out, “water problems have been traditionally perceived, understood, and acted on locally,” but institutions of governance increasingly recognize that the causes of freshwater problems extend to extra-local mechanisms. Although local scales remain essential in governing water quality, scientists and policymakers now routinely call for multi-scalar approaches to freshwater governance (e.g., Hering et al., 2015).

Cook’s (2014, 2015) discussions of historical changes in the governance of freshwater in Ontario and the Great Lakes region of Canada provide useful illustrations of this shift toward multi-scalar approaches. During the 19^th century, water provision and sewage treatment gradually became established as local, municipal-scale responsibilities. However, the emergence in the late 19^th and early 20^th centuries of pollution as a public health threat – typhoid fever from untreated sewage – “confounded the municipal scale,” since both the causes and the capacity to address them exceeded the boundaries of municipalities (Cook 2015: 128). In order to address issues of water quality and quantity that crossed borders, Canada and the United States signed the binational Boundary Waters Treaty of 1909 and subsequently established the International Joint Commission (IJC) to help the two national parties implement the treaty. Although binational cooperation helped solve the typhoid problem by requiring municipal chlorination of drinking water, the uneven implementation of sewage treatment and the rise of industrial pollution in the mid-20^th century gradually made it clear that addressing pollution and deteriorating water quality would require multi-scalar solutions. For example, Canadian municipalities found they could not develop the sewage treatment facilities they were responsible for without financial assistance from the provincial and federal scales. The first Great Lakes Water Quality Agreement of 1972 and its subsequent amendments instituted an even broader multi-scalar approach, establishing roles for Canada and the United States, their provinces and states, and also local scales of jurisdiction. Similar multi-scalar arrangements to address problems of water quality have become increasingly common throughout the world (Moss and Newig, 2010).

Accompanying the rise of multi-scalar approaches to the governance of water quality has been the increasing prominence of the watershed scale. Although there is no single definition of the watershed or basin, this definition from the U.S. Environmental Protection Agency (USEPA, 2018b), encompassing both surface water and land, is typical: “the land area that drains to one stream, lake or river.” The idea of the “drainage basin” as a natural unit took hold among scientists in the 18^th century, and in the 19^th century it became a common basis for water management and planning in industrialized and industrializing countries (Molle, 2009). Although the watershed’s prominence waxed and waned during the 20^th century, since the rise of Integrated Water Resources Management in the 1990s, it has become hegemonic throughout the world as a preferred scale for the management of water resources, water quality, and aquatic ecosystems (Molle, 2009). The Canada-United States border, where the International Joint Commission has increasingly embraced a watershed approach to resolving disputes between the two countries, is typical of this trend (Norman and Bakker, 2009).

The watershed scale has become attractive in freshwater governance for a variety of well-documented reasons. First is the argument that the watershed is a natural hydrological unit, and it therefore makes more sense for the governance of water than artificial territorial jurisdictions (see, e.g., Blomquist and Schlager, 2005; Cohen and Davidson, 2011; Cohen and McCarthy, 2014). Because it is based on natural hydrological processes, the watershed scale is believed by advocates to facilitate integrated, and thus more effective and efficient, treatment of a wide range of issues (Fischhendler and Feitelson, 2005). As Cohen (2015: 25) notes, actors in water governance frequently assume that management at the scale of the watershed enables accounting for “all upstream and downstream factors” in water quality and distribution, provides a more feasible basis for public participation, and facilitates cooperation among different political jurisdictions that overlap with basin boundaries.

A second reason for the appeal of the watershed is its “conceptual malleability” (Cohen, 2012: 2208). In contrast to the idea that watersheds are purely “natural,” scholars argue that they are constructed through a combination of hydrological and social processes (e.g., Cohen, 2012, 2015; Harris and Alatout, 2010; Norman and Bakker, 2009; Swyngedouw, 1999). Watersheds are constructed or reconstructed in both a physical sense – for example, through the building of canals, dams, and infrastructure (Blomquist and Schlager, 2005; Cohen, 2015; Warner, et al., 2014) – and a discursive sense, through “policy discourse, mapping practices, and funding arrangements” (Cohen, 2015: 27). It is in this discursive mode that the social and political constructedness of watersheds becomes most visible. In a Canadian case study, Cohen (2012) attributes the allure of the watershed to its ability to work as a boundary object, or a shared concept that allows different groups to communicate even though they attribute different meanings to it. Specifically, Cohen argues that the malleability of the watershed makes it appealing to scientific, neoliberal, and grassroots epistemic communities, despite their conflicting ideological commitments and objectives.

This malleability, however, creates challenges for implementing the watershed approach in practice (Cohen and Davidson, 2011; Norman and Bakker, 2009). Cohen and Davidson (2011) distill five such challenges, which they attribute to the conflation of watersheds with “participation, integration, decentralization or other governance concepts,” such as Integrated Water Resources Management (Cohen, 2015: 26). First, although the technical definition of a watershed may be straightforward, drawing boundaries for specific watersheds is not, in part because of their dynamic and nested character. This ambiguity can lead to the manipulation of watershed boundaries to achieve political objectives (Warner, et al., 2014), just as invoking the watershed as the most “natural” scale for governance can depoliticize decision-making and threaten the viability of democratic debate (e.g., Cohen and Bakker, 2014; Wester and Warner, 2002). Second, watersheds raise questions of accountability, since they typically do not coincide with electoral or regulatory jurisdictions. Recent research finds that the level of fragmentation of political jurisdictions in watersheds is associated with greater declines in water quality (Kim et al., 2015). Third, as Norman and Bakker (2009) demonstrate, devolving responsibility to watersheds does not necessarily result in meaningful public participation or local empowerment. Their research also shows that watershed-based governance is particularly complex when watershed boundaries intersect with national boundaries; for example, treaties may offer limited roles for actors beyond the government.

The fourth and fifth challenges identified by Cohen and Davidson (2011) are associated with the frequent mismatch between the watershed scale and the scales of either problems or the capacity to address them. Fourth, the watershed may not correspond to the problemshed, defined as a “geographic area that is large enough to encompass the issues but small enough to make implementation feasible” (Griffin, 1999: 509; see also Kneese, 1968). The causes of problems in watersheds, along with the effects of watersheds themselves, often extend beyond the basin boundaries (Cohen and Davidson 2011). For example, because watersheds typically do not correspond neatly to ecoregions (Cohen, 2015; Omernik and Bailey, 1997) or to groundwater flow (Cohen, 2015; Winter et al., 2003), they may be inappropriate for addressing specific ecological or hydrological problems. Fifth, the watershed does not typically match up with “conventional administrative scales,” or what Cohen and Davidson (2011:4) call policysheds (also Davidson and De Loë, 2014). For some governance problems and decisions, such conventional scales may be more appropriate or practical scales to use (Cohen and Davidson, 2011; Muller, 2019). However, federal, state, county, and municipal governments differ in their capacities and willingness to make and implement water quality rules within watersheds (Mancilla Garcia et al., 2019; Wardropper et al., 2015).

The limitations of the watershed have generated interest in other local scales of governance, including conventional administrative jurisdictions, as alternatives. For example, another possible way of scaling water governance is to focus exclusively on watercourses, exclusive of the land area that drains into them, such as in the United Nations Convention of the Law of Non-Navigational Uses of International Watercourses (Cohen, 2015; Savenije and van der Zaag, 2000). The conception of the problemshed, which does not start with the assumption that the causes or solutions to problems should be coterminous with watersheds or any other specific scale, provides another alternative (Mollinga et al., 2007; Muller, 2019; Woodhouse and Muller, 2017). In the context of indigenous territories, recent scholarship has proposed alternatives to the watershed scale based on indigenous knowledge and sensitive to culturally specific relationships with land and resources (Sarna-Wojcicki et al., 2019). In short, despite the widespread uptake of the watershed scale, it is not universally accepted as the appropriate local or regional scale for all situations of water quality governance, and there is growing recognition that governance should take place at multiple scales corresponding to different needs and functions (e.g., Muller, 2019).

The introduction of new local scales of governance, whether the watershed or an alternative scale, is inextricably connected with relationships among other scales. As McCarthy (2005: 738) notes:

the delineation or elimination of any particular scale as an arena, locale, place, or so on is always done relative to other scales and the relationships among them, and necessarily introduces changes into their ordering and hierarchies.

Because watersheds or other local and regional scales of water quality governance overlap and intersect with existing scalar hierarchies and patterns of spatial jurisdiction, they carry implications for how different scales—for instance, the municipal and the sub-national—relate with each other. In addition, the responsibilities and powers assigned to various scales are always defined and constrained in relation to other scales. As we will discuss below, in the United States both the introduction of the AOC as a new scale of governance and the question of whether it should correspond to the watershed scale or a different scale are imbricated with the relationships between the scales of state-level agencies and local stakeholder groups.

Scale and the Great Lakes Areas of Concern

The challenge of assigning the governance of water pollution to appropriate geographic scales has featured prominently in the history of the Great Lakes AOC program. The 1972 Great Lakes Water Quality Agreement constituted an important first step toward multi-scalar governance, establishing roles for the binational, national, and subnational (provincial and state) scales (Cook, 2015). Later, the Great Lakes Water Quality Board (1985) recommended delegating RAPs to states and provinces, and in the United States, the states subsequently committed to developing RAPs for the AOCs (Hartig and Zarull, 1992). The 1987 Amendments to the Great Lakes Water Quality Agreement formalized the AOC process. The Amendments did not explicitly spell out specific roles for local scales, but as noted above, the 2012 version of the Agreement identified several local scales—including tribal/First Nations governments, municipal governments, and watershed management agencies—as essential to the process of restoring AOCs. According to the International Joint Commission (2019: np) website,

A key lesson learned [since the inception of the AOC program] is that the more the local community’s residents, governments and leaders are involved, the more success they will have in developing and funding an effective plan that will reap the benefits desired. Public advisory committees, councils and other groups are essential, especially in binational locations where multiple communities are involved.

However, neither the 1987 nor the 2012 version of the Agreement prescribes how the division of labor between state/provincial and local scales should be implemented in practice, which means that there is variation in the delegation of responsibility for key decisions in the RAP process, such as the defining the geographic extent of AOCs themselves.

Annex 2 of the 1987 amendments to the Great Lakes Water Quality Agreement also instituted two scales to target for remediation of pollution and ecological degradation. The first, established by the Great Lakes Water Quality Board in 1981 to replace the “problem area,” was the AOC, which would be the scale corresponding to the RAP process recommended by the same Board in 1985 (Great Lakes Water Quality Board, 1981, 1985). The second was the scale of the individual lake, which is the scale addressed by Lakewide Management Plans (LaMPs). Again, the Annex does not assign the AOC to a specific local scale, such as the watershed or municipality; instead, it defines the AOC more ambiguously, as “a geographic area that fails to meet the General or Specific Objectives of the Agreement where such failure has caused or is likely to cause impairment of beneficial use or of the area’s ability to support aquatic life” (Great Lakes Water Quality Agreement, 1987: 24).

Annex 2 lists 14 specific beneficial use impairments (BUIs), and the objective of each RAP is to identify and address all pertinent impairments for an AOC so that the Area can be delisted. The majority continue to undergo remediation. There was little dedicated funding for the program in its early years, and several AOCs faced limited information and a lack of intergovernmental cooperation (Botts and Muldoon, 2005). When several states discontinued support for the program, activity at most of the AOCs all but ceased in the mid-1990s (Botts and Muldoon, 2005), except for activities associated with other programs, such as the Superfund remediation in the US.

Since the turn of the millennium, however, the situation for AOCs and RAPs has changed. First, in the United States the Great Lakes Legacy Act of 2002 authorized enough funding to support projects to clean up contaminated sediment at many of the AOCs (Tuchman et al., 2018) and it was followed up by the creation of the Great Lakes Interagency Task Force. Even more significant was the Great Lakes Restoration Initiative (GLRI), introduced in 2010 (USEPA, 2019a). While the Legacy Act has provided over $300 million of federal money for AOC projects, the GLRI has provided over $2 billion, about a third of which has gone to AOCs specifically, addressing not just contaminated sediment but also a wider array of problems (USEPA, 2017). As a result, RAP activity and progress increased significantly in the 2010s, especially in the United States (Hartig et al., 2018; Hartig et al., 2020).

Figure 1 shows the geographic distribution of AOCs. Of the 43 AOCs, 31 are in the US and twelve are in Canada, and five (St. Marys River, St. Clair River, Detroit River, Niagara River, and St. Lawrence River) are designated as binational. Within the United States, AOCs are distributed unevenly among the states (Table 1). Unsurprisingly, given its proportion of the Great Lakes coastline, Michigan has the most AOCs of any of the U.S. states. New York, Ohio, and Wisconsin also have multiple AOCs, while Illinois, Indiana, Minnesota, and Pennsylvania have only one each. Three of the binational AOCs (St Marys River, St Clair River, and Detroit River) are in Michigan, and two (Niagara River and St Lawrence River) are in New York. As of 2019, only four AOCs in the United States had been delisted: Oswego River (July 2006), Presque Isle Bay (February 2013), Deer Lake (October 2014), and White Lake (October 2014) (USEPA 2019b). Several others have met the qualifications for newer categories of progress: in the United States, “management actions completed,” and in Canada, “in recovery.” These designations indicate that the needed work to restore the beneficial uses has been completed, but there is not yet enough information to remove the BUI from the list of impairments.

Figure 1: U.S. Great Lakes Areas of Concern (Courtesy of USEPA).

This map illustrates the location of AOCs, the AOCs where management actions have been completed, and delisted AOCs.

Table 1:

Distribution of United States AOCs

State	Total number of AOCs	Binational AOCs	Number of delisted AOCs (as of 2019)
Illinois	1	0	0
Indiana	1	0	0
Michigan	13	3	2
Minnesota/Wisconsin*	1	0	0
New York	6	2	1
Ohio	4	0	0
Pennsylvania	1	0	1
Wisconsin (only)*	4	0	0

^*Minnesota’s only AOC is shared with Wisconsin.

Methods and data

The research for this article, which investigates variation in how AOCs are defined and bounded, draws on multiple sources of data. First, using a list provided by the USEPA’s Great Lakes National Program Office and Great Lakes Commission, we administered an online survey to RAP coordinators and citizen, community, or public advisory committee leaders at each of the 31 United States and binational AOCs. For the purposes of this study, we did not include the other Canadian AOCs. Second, we followed up the survey with recorded, semi-structured telephone interviews with a subset of survey participants who indicated their willingness and availability. Our survey generated 48 responses from 24 of the 31 AOCs, including delisted ones, and we received two or more responses from 12 AOCs. Our 18 follow-up interviews included an equal number of RAP coordinators (9) and Community Advisory Committee members or chairs (9) from 14 different AOCs.

The importance of AOC definition was initially indicated through the text response of a few survey respondents to a question about the geographic dimensions of the remedial process. However, this article focuses primarily on the interviews, in which we asked specifically about the definition of the AOCs. We analyzed the interview data using content analysis and grounded theory, through which emergent themes were generated from the data (Glaser, 2002; Hseih and Shannon, 2005) and organized with NVivo qualitative data analysis software (QSR International, 2016). We focus specifically in this article on how respondents discussed the AOC boundaries. To supplement the data from the surveys and interviews, we consulted RAP documents, other documents from the International Joint Commission and the Great Lakes Water Quality Board, and USEPA’s AOC maps.

The findings of both our surveys and our interviews are limited by our lack of access to representatives from all 31 AOCs in the USA. To address these limitations, we also interviewed Canadian coordinators from two of the binational AOCs, and in a few cases, we conducted more than one interview from the same AOC. This approach allowed us to focus on the process of how individual AOCs negotiate the process of boundary definition in a specific federal-state-local government structure. The only state not represented among our interviews was Indiana, which has only one AOC.

Scaling and bounding Great Lakes Areas of Concern

Differences among AOC boundaries in the United States

To illustrate the range of ways that AOCs are scaled and bounded, let us first consider two extreme examples. On one end of the spectrum is Oswego River in New York (Figure 2), which in 2006 became the first AOC to be delisted in the United States. This is one of the smallest AOCs, with an area of just more than 1 km² (about a half square mile). It consists of a short stretch of the river downstream of the Varick power dam and a small segment of Lake Ontario going out to the breakwall (USEPA 2018a). It includes only the watercourse, not the land that drains into it. The AOC is entirely within the city of Oswego, with a population of roughly 17,000, and before its delisting it had only four of the 14 possible beneficial use impairments. In stark contrast, the Cuyahoga River AOC in Ohio (Figure 3) has nine beneficial use impairments and “includes the lower 46.5 river miles – approximately half the full length – of the river, and approximately 500 square miles of watershed draining either to the river or to eleven miles of Lake Erie shoreline adjacent to the river mouth” (Cuyahoga River Restoration, 2015: 5). Due to its larger size, the Cuyahoga also features far greater jurisdictional complexity; it intersects with 45 political subdivisions, including the municipal boundaries of Cleveland (est. pop. 385,000) and Akron (est. pop. 197,000), along with numerous smaller municipalities, counties, and other jurisdictional territories within and beyond their metropolitan areas. As these two examples show, the size, scope, and relationship with watersheds vary considerably among AOCs.

Figure 2. Map of Oswego River Area of Concern (Courtesy of USEPA).

The extent of the AOC is represented by orange hash marks and an outline that ends at the river bank and nearshore area of Lake Ontario.

Figure 3. Map of the Cuyahoga River AOC (Courtesy of USEPA).

The extent of the AOC is represented in orange. The extent of the AOC goes beyond the banks of the Cuyahoga River to include the tributaries and the watershed.

Table 2 shows the ways that U.S. and binational AOCs are geographically demarcated vary both within and among states. Most AOCs (19 out of 31, or 61%) are currently defined as bank-to-bank segments of rivers and nearshore areas within the Great Lakes (see Supplemental Material). However, several AOCs have territories that comprise both water and land; some of these (7 out of 31, or 23%) are defined—although sometimes ambiguously—as watersheds or partial watershed areas, and the remaining five have definitions we characterize as “hybrids.” Some of these consist of both an AOC and an extended or expanded area of some kind, and others have AOC definitions that encompass at least some land area in addition to the affected watercourses. In each of the states with multiple AOCs—Michigan, New York, Ohio, and Wisconsin—there is, in effect, a “default” approach and one or more exceptions to this approach. In Michigan, New York, and Wisconsin, the default definition is bank-to-bank, although Michigan has more than one AOC defined in watershed terms. The only state with multiple AOCs in which the majority of AOCs are now defined as watersheds or partial watersheds is Ohio, although initially these were defined in bank-to-bank terms as well. As for the states with only one AOC each (Illinois, Indiana, Pennsylvania, and Minnesota, which shares its only AOC with Wisconsin), they have defined their sole AOCs in terms of watersheds, partial watersheds, or hybrids. One important caveat is that the definitions of several AOCs have changed over time, and again, in some cases their official delineations are ambiguous (see Supplemental Material).

Table 2:

Definitions of United States AOCs

State	Watershed or partial watershed	Bank-to-bank + nearshore areas	Hybrid/special/ambiguous
Illinois	0	0	1*
Indiana	1*	0	0
Michigan	2*	10	1*
Minnesota/Wisconsin*	1	0	0
New York	0	4	2*
Ohio	3*	1	0
Pennsylvania	0	0	1*
Wisconsin (only)*	0	4	0
Total	7	19	5

^*Minnesota’s only AOC is co-managed with Wisconsin. See Supplemental Material for other explanations.

Understanding variation in AOC scales and boundaries: drivers

We considered two broad categories of explanation for the variation in how AOCs are scaled and bounded: physical conditions and policy/governance decisions. The simplest explanation would be that the scale and boundaries of each individual AOC are determined by the area’s physical conditions. Based on Table 2 above, this would suggest that a bank-to-bank definition is adequate for most conditions of contamination and degradation, but in a minority of cases physical conditions require a watershed, partial watershed, or hybrid definition. Indeed, our interviews and reviews of documents indicated that physical conditions played an important role in defining AOCs. However, the interviews suggested that the scale of the AOC—defined in terms of size, boundaries, and relationship with watershed or river reach scales—is also related closely to policy decisions or governance processes at federal, sub-national (state, provincial, tribal/First Nations), and various local scales. Specifically, bank-to-bank definitions are the default of both the main federal agency involved (the U.S. EPA) and most U.S. states in the AOC program. However, some states have permitted AOCs to be defined in watershed or hybrid terms, usually as a reflection of the priorities and preferences of local stakeholders; moreover, in one state with multiple AOCs (Ohio), watershed-based AOCs have become the rule rather than the exception. In some cases, according to our interviewees, physical conditions were secondary to policy orientations and decisions at the state level, including how much latitude the state gives to local stakeholders, typically represented by public advisory committees (PACs) or citizen/community advisory committees (CACs).

Federal and state priorities and interests

Some interviewees suggested that USEPA preferred bank-to-bank definitions. A participant involved with the Rochester Embayment AOC suggested that the AOC program’s initial focus on the watershed had largely faded:

I think the directive originally was pretty clear, at least to the county from … the state. Everybody was concerned with the whole watershed concept then. (Interview 10).

Despite the earlier interest in the watershed, most AOCs were defined in bank-to-bank terms, in part because the latter lend themselves to meeting management goals. As the same interviewee described it, the USEPA

looks at it almost like an accountant: we’re going to get this many BUIs delisted, this many removed this year, and we’re going to delist this many AOCs by this year. And that’s the kind of thing that people in the government want to see (Interview 10).

This description is consistent with the Great Lakes Interagency Task Force’s (2014: 7) Great Lakes Restoration Initiative Action Plans, which include annual targets for the removal of beneficial use impairments and the implementation of “management plans necessary for delisting.” The interviewee went on to characterize the approach of defining AOCs as small bank-to-bank segments in the terms of Total Quality Management (TQM):

it’s sort of the TQM principle: you give the customer exactly what the customer wants, but you don’t give the customer more than what they want (Interview 10)

To go beyond the narrowly defined delisting targets for AOCs—and to expand AOCs to encompass the full range of watershed issues—would be to do the latter.

A few of our interviewees discussed the preference for bank-to-bank definitions at the state level. In Wisconsin, for example, all AOCs are defined in bank-to-bank terms except the St. Louis River AOC, which is administered jointly with Minnesota. A respondent representing one of Wisconsin’s AOCs remarked:

In some AOCs they look at the management actions in the entire watershed… It looks more holistically at a watershed. So that even though the AOC may go halfway between the estuary and the headwaters of the watershed, for instance, they know that what you do in the headwaters impacts downstream. Here you don’t look at that. Here you look at only what is happening within the boundaries, one bank-to-bank. I guess there is a little bit of wiggle room there, and I don’t know what that wiggle room is because I don’t know that it is completely defined (Interview 18).

As described in the next section, in at least one case, Wisconsin’s state-level preference for bank-to-bank definitions generated controversy.

The one exception to the rule in Wisconsin is the St. Louis River AOC, which the state shares with the state of Minnesota; this one is defined in watershed terms:

We definitely have watershed delineation here on our AOC boundaries… And within our AOC boundary as it is defined, for sure if that is bank to bank or stopped at the dam, maybe it would be a different story there (Interview 8; also see Supplemental Material).

An interview participant from this AOC noted that the watershed delineation allowed the RAP to pursue a wider range of projects than other AOCs, but acknowledged that getting funding for these projects required extra justification:

We do oftentimes have to make a good case for including watershed-type projects in the AOC funding requests and project proposals because oftentimes AOCs are not having to go that far out from their bank to bank projects (Interview 8).

The case of Ohio shows, however, that state-level approaches to defining AOCs do not necessarily align with the orientations of U.S. EPA or other states. Here, three of the four AOCs are now defined—again, with some ambiguity—as watersheds or partial watersheds (see Supplemental Material). As an interviewee from Cuyahoga AOC described it:

We’re that way [defined in watershed terms], Maumee is that way, and … Black River started out only as bank to bank, the mouth and last couple of miles where the most industrial section was. They expanded their area of concern to include the whole watershed, or almost – maybe not the whole watershed. Because we recognize that we weren’t going to get anywhere in the main stems if we didn’t address the stuff coming out of the sub-watersheds, out of the tributaries (Interview 11).

More than one interviewee suggested that the predominance of watershed-based AOCs in Ohio reflected both the preferences of local stakeholders and the willingness of the relevant state agencies—including Ohio Environmental Protection Agency (Ohio EPA), Ohio Lake Erie Commission, and Ohio Department of Natural Resources—to accept the watershed scale. As a participant from Cuyahoga noted, Ohio EPA supported the local stakeholder committee’s request to expand the AOC and “agreed that we had to work up in the headwaters…and not just the main stem” (Int 11). However, one Ohio-based interviewee expressed skepticism about the approach:

Some of these have, you know, we got three that are total watersheds. You know, I don’t know what their thinking was when doing that, but they’ve got watersheds upstream that are pretty pristine, and I don’t know why they’re part of an Area of Concern, unless they were trying to get funding to do stuff on it (Interview 7).

Funding for projects has in fact often featured in stakeholder discussions about AOC boundaries. For example, the minutes from a meeting of the Black River AOC Advisory Committee includes funding as one of the considerations discussed in response to a proposal to change the AOC’s boundaries: “If resized, the upper watershed won’t be able to leverage AOC funds for projects, but there are other funding sources available” (Black River Area of Concern Advisory Committee 2015).

Preferences of local stakeholders

Although interviewees from Ohio suggested that site-specific physical conditions played roles in determining AOC boundaries, their narratives also proposed that local stakeholder preferences were the main drivers. For example, for the Ashtabula AOC, more than one respondent suggested that a small bank-to-bank territory made more sense because of the nature of contamination at the site:

the issue wasn’t the upstream watershed. You know, there are stormwater issues and junk like that, but that was not the cause of the pollution. The pollution was basically coming out of one small stream down in the harbor area (Interview 7).

Another characterized the Ashtabula AOC as “small, short, and really just the harbor” and also pointed out that “it’s not being used as an industrial ship channel anymore” (Interview 11). The same interviewee also framed the watershed approach at Cuyahoga as driven in part by the nature of the problem:

Because problems in the river basically begin at Akron and continue up to Cleveland and the mouth at Lake Erie, the original Area of Concern had to include the portions of the watershed starting in Akron and coming up this way, which is only half of the whole river (Interview 11).

But on the other, this interviewee pointed out that the initial stakeholder committees at this AOC were “heavily populated by metropolitan planning agencies” from Cleveland and Akron, implying that there were also strong local interests in defining the AOC more generously (Interview 11). Similarly, a participant described the decision to define Ashtabula in bank-to-bank terms as based not only on the limited scope of the pollution, but also on a desire among local stakeholders to keep the scope of work manageable: “We did not want to dissipate our energy on looking at watershed stuff…. We would never have gotten anything done if we had dissipated our energy” (Interview 7). A respondent from Michigan, the only other state with more than one AOC defined in watershed terms, also described a process in which local stakeholder groups drove the decisions:

I was involved when Michigan’s boundaries were set, and basically the [US] EPA sent us maps, and then we discussed them with the different Public Advisory Councils as to what they wanted to see. And some of them, like the Clinton, wanted the whole watershed. That was the Public Advisory Council’s choice. Other ones are very narrowly defined: like for Torch Lake, it’s Torch Lake and its immediate environs as the AOC boundary (Interview 16).

As the cases described in the next section suggest, Ohio and Michigan were exceptional in the leeway that they offered to local stakeholder groups in defining the scale and boundaries of AOCs.

Scaling and bounding AOCs: controversies in the RAP process

Our interviews revealed that in many cases, the definition of AOC boundaries has been a source of long and sometimes ongoing contention and dispute in the RAP process. One common narrative we heard—although with considerable variation—was that local stakeholders wanted to define the AOC as watersheds or in similarly broad terms, but faced resistance from state agencies that aimed to confine AOCs to smaller bank-to-bank areas. One survey respondent summarized this conflict in the following terms: “Gov’t entities view the AOC as bank to bank, where stakeholders view it as the ‘watershed’ - or more expansive than ‘Bank to Bank.’” However, in some cases local groups came over time to accept or even embrace the bank-to-bank definition, and in at least one case, a local stakeholder group adopted a narrow bank-to-bank definition that the state agency found too limited. Overall, the interviews suggested that even if there is broad agreement among managers and stakeholders that the watershed scale is important for addressing problems in Great Lakes tributaries, there is not universal consensus about the appropriate relationship between the watershed scale and the scale of each of the specific AOCs.

One example of a case in which stakeholders sought a broader definition than the state is the Lower Green Bay and Fox River AOC in Wisconsin, where the state Department of Natural Resources resisted efforts to define the AOC in watershed terms. This AOC has attracted considerable attention because its initial Remedial Action Plan was one of the most thoroughgoing in its efforts to adopt an explicit ecosystem approach (Harris et al., 1987), and the state’s bank-to-bank definition was the source of “quite a debate” (Interview 12). According to one interviewee, prior to the designation of the AOC as a relatively small bank-to-bank area, there was an “earlier effort [that] was really more ecosystem focused, and we had a much larger area,” encompassing the southern half of Green Bay and “the whole watershed of the Lower Fox River… as well as some of the other tributaries coming in” (Interview 12). However, according to the same respondent, key state officials became concerned about the mounting costs of Wisconsin’s remedial action plans, and

[the Wisconsin Department of Natural Resources] really wanted to scale back their, the geographic area… There was an administrative decision made, I guess, in Madison to scale back our AOC boundary to just the very southern end of Green Bay and the last 7 miles of the river leading up to it….There was absolutely no ecological reason for that, other than it was a matter of trying to minimize the amount of effort that was going to be needed to write this plan (Interview 12).

Even though “a number of people protested, the decision was made and so that’s how our boundary got drawn” (Interview 12).

Eventually the Wisconsin Department of Natural Resources “did accept that they had to expand the boundary,” at least unofficially, because there were PCBs in sediment along the entire length of the Lower Fox River (Interview 12). Still, the smaller bank-to-bank area remains the official extent of the AOC (see Supplemental Material). The respondent described the boundary as “a hindrance” to complete restoration, since “they can’t spend money outside the boundary of the AOC,” and characterized the current focus on removing beneficial use impairments inside the smaller area as “somewhat politically driven” (Interview 12). From the perspective of the interviewee, then, the bank-to-bank definition not only prioritizes quicker delisting over thorough remediation and restoration, but also all but abandons the integrated ecosystem approach that a watershed definition would presumably offer. The interviewee noted: “most of us…still remind them frequently that…if we are actually going to address nutrient and sediment loadings to the system, we had to look at the entire river system” (Interview 12).

Another example of this scenario, although with a different outcome, is the Waukegan Harbor AOC in Illinois. One respondent pointed out that one of the first jobs of the Citizens Advisory Group was to “hammer out the boundary,” which the interviewee characterized as “a contentious issue all along” (Interview 15). The state of Illinois had proposed a small bank-to-bank area of concern approximately coterminous with the Outboard Marine Superfund site, a hazardous waste site under the jurisdiction of the federal Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA). However, local citizens and stakeholders advocated for a larger AOC; the interviewee noted that the “citizens and the groups that got together and agreed to form Waukegan Area of Concern - they would not agree until they had the Extended Area of Concern added on” (Interview 15). The latter—which U.S. EPA maps refer to as the “Expanded Study Area”—is not an entire watershed or even an entire sub-watershed, but it does encompass a broader area and includes not just the water but also the land that drains into it.² Such hybrid definitions are unusual in the AOC program, but unlike Wisconsin, Illinois has only one AOC, and thus no need to develop a policy that would apply consistently to multiple sites. The respondent characterized the boundaries of the smaller area proposed by the state as reflective of the interests of government agencies in getting a quick win: “very simple, you could take the Outboard Marine site and get it down… make it manageable, get it done, get it delisted and get it out of the way” (Interview 15). In contrast, the interviewee made the case that the expanded boundary was actually the more pragmatic choice: “It actually has helped make it much more effective to be able to treat it as a whole… I personally think it is much easier to handle because you can look at it from a watershed perspective” (Interview 15).

Although other interviewees described debates over their boundaries that were similar to those at the Lower Green Bay and Waukegan AOCs, some also emphasized that defining AOCs in bank-to-bank terms does not necessarily mean ignoring upstream problems within the watershed. According to a respondent from Rochester Embayment AOC in New York, for example, the local stakeholder group initially protested as the AOC became confined to a small bank-to-bank area and a small area of land adjacent to part of the Genesee River. They were especially concerned that the AOC definition omitted Irondequoit Bay, which had for decades been discussed as just as much of a problem as the embayment. Still, even though this meant that Irondequoit Bay has been left out of the Great Lakes Legacy Act and other AOC-specific funding opportunities, the interviewee contended that narrowing the focus and emphasizing targets for delisting has actually helped move things along in the embayment: “That works, as far as getting people moving. Because you can look at it, see things in manageable and measurable bits to accomplish” (Interview 10). In addition, he suggested that a watershed approach to the Rochester Embayment AOC would have created a paralyzing task. The Genesee River is about 140 miles long, covering 2,373 square miles of land area, with another 5,000 plus miles of tributary rivers and streams (New York Department of Environmental Conservation, 2019). As our respondent put it, if

… you go in there and you start to try and define habitat projects necessary for resolving this beneficial use impairment at the end of the river, and in the lake… that’s one of those jobs where you could stop working because the magnitude of the job would just make you, like, sort of short circuit (Interview 10).

Finally, he emphasized that shrinking the AOC did not mean neglecting the watershed entirely. As he put it: “… it’s not that things go away. It’s that they go to different lists” (Interview 10). In other words, watershed problems that might have been placed in the AOC and addressed through the RAP had shifted to other management programs or legal authorities, such as the Lakewide Management Plans (LaMP), Superfund (CERCLA), the Resource Conservation and Recovery Act (RCRA), or others.

We heard similar comments from a respondent from the Presque Isle Bay AOC in Pennsylvania, one of the very few delisted AOCs. Here, too, the boundary had been a source of controversy, but for different reasons. As the interviewee put it, with respect to the local stakeholder group, “we’ve always talked about extending to the whole Lake Erie watershed instead of just Presque Isle Bay” (Interview 9). He described early meetings about the boundaries as contentious:

We argued about… exactly what is the AOC? … Just how far upstream do you go? … Or how far downstream, or how far into the lake…? And so that was a pretty tough meeting, where we had to finally come up with the boundaries (Interview 9).

One problem that ultimately led to a narrow definition was that the AOC abutted a popular state park, and some local stakeholders contended that a watershed-based definition would “destroy tourism” (Interview 9). However, the narrow boundaries were a source of frustration to other stakeholders in the local stakeholder group, many of whom contended that addressing problems in Presque Isle Bay would require attending to issues in the broader Lake Erie watershed:

One of the things we had… trouble selling… to the other PAC [Public Advisory Council] members in the community was… as long as we were an AOC, any money we got through the state or federal government had to be focused on the Presque Isle Bay and our two BUIs: sediment and fish tumors (Interview 9).

As the AOC approached delisting, there was considerable fear that the PAC would run out of steam, and that sources of funding for other problems they wanted to address would disappear.

However, the narrowly defined Presque Isle Bay AOC may have ultimately freed up local stakeholders to tackle watershed problems earlier and more effectively than they otherwise would have. According to the interviewee:

as we delisted…, we moved right into the watersheds. And there was actually never any discussion about dissolving the PAC because we’ve delisted. … In fact, we were more encouraged, and had … a large number of people attend our meetings, once we started moving more into the watershed… Now that we’ve delisted, our main focus is not only on the Bay, but the whole Lake Erie watershed… So it actually was a good thing to be delisted (Interview 9).

Stakeholder participation has increased, and delisting has enabled the PAC, in the interviewee’s words, to “focus our energies and apply for grants to do projects in the watershed” (Interview 9). In contrast to the interviewees who maintained that addressing downstream problems in AOCs required defining the areas in terms of watersheds or watershed ecosystems, this interview suggested that a narrower designation that enabled the AOC to be delisted more quickly was what was in fact enabling work in the larger watershed to begin.

In at least one case, the roles of the more common narrative were reversed, with the state agency calling for a broader definition of the AOC and a local stakeholder group arguing for a narrow bank-to-bank definition. At Torch Lake AOC in the Upper Peninsula of Michigan, one interviewee described how the local stakeholder group initially became dominated by real estate and tourism stakeholders who wanted the AOC defined as bank-to-bank in hopes that it would be delisted quickly (Interview 16). However, the AOC coincides with a very complex and challenging hazardous waste site in the federal Superfund/CERCLA program; the Michigan Department of Environmental Quality³ cautioned against too-rapidly delisting and emphasized the need to attend to watershed issues before delisting can take place. The definition—bank-to-bank, but encompassing a small area of land in the immediate environs—remains the official one there, although in this case the preference was driven less by the management objectives of government agencies than a set of highly localized political-economic interests. Some of the tension is evident in the U.S. EPA’s boundary map; the main map shows the AOC as defined in official documents, but the legend for the inset map also describes the watershed as the AOC.⁴

Discussion and conclusion

Based on our interviews, we developed a basic conceptual model to represent the apparent major drivers of decisions about how to scale and bound individual AOCs in the United States (Figure 4). The Great Lakes Water Quality Agreement provides the binational legal basis for designating AOCs, and the biophysical properties of contaminated waterways are essential foundations as well. However, neither the Agreement nor biophysical conditions have directly dictated the scale or boundaries of individual AOCs. In practice in the United States, federal and state agencies have proposed AOC definitions, with varying levels of control delegated to local stakeholder groups. Although the preference for smaller bank-to-bank AOCs has prevailed in most cases, some states have allowed specific AOCs to be defined as larger watersheds or partial watersheds. Either definition has the potential to generate controversies, which in a couple of cases have led to hybrid or compromise solutions that combine elements of both.

Figure 4.

Conceptual Model.

One implication of this research is that in the United States, decisions about scaling and bounding AOCs have been considerably dependent on the relationship between federal, state (sub-national), and local scales of governance. In this case, the possibility for defining AOCs as watersheds, as opposed to the default bank-to-bank, evidently depended heavily on decisions at the state scale about how much latitude local advisory councils could have in defining AOCs. This resonates with McCarthy’s (2005: 738, emphasis in original) point—responding to Brenner’s (2001) argument for more carefully distinguishing scale as a category of analysis—that “[p]olitics at or about a given scale are inseparable from politics concerning relationships among scales.” It also suggests that analyzing the uptake of the watershed scale requires attention to the specific ways in which scales of governance are related in different programs and situations. For example, to what degree are decisions about whether to select the watershed scale or a different scale centralized or decentralized? Moreover, how do conflicts and controversies over scale and boundary determination vary with degree of decentralization?

Since most AOCs in the United States are defined in bank-to-bank terms, the case of the US Great Lakes AOC program provides a prominent but also complex exception to the dominance of the watershed scale in freshwater governance. On the one hand, none of our interviewees denied the importance of addressing issues at the watershed scale. On the other, they disagreed about whether the watershed scale is the most appropriate for the AOC program or for individual AOCs. Some interviewees argued that watershed-based AOCs are more practical, because they accommodate projects that address conditions upstream or on land. They emphasized that removing beneficial use impairments in bank-to-bank AOCs is often difficult or even impossible without addressing upstream causes. One survey respondent, for example, described the following as a “geographic barrier” to remediation: “many problems originate in watershed not included in AOC.” In contrast, others argued that smaller bank-to-bank AOCs facilitate more rapid delisting and allow remediation efforts and funding to shift more quickly to issues in the broader watershed, typically in programs “on different lists.” The latter contended that a bank-to-bank definition makes it easier than a watershed definition to focus attention and resources, especially in large, complex watersheds or AOCs that encompassed more than one watershed. For example, another survey respondent described the challenges of addressing an AOC that overlapped with multiple watersheds: “many watersheds in one AOC, not connected or sub-watersheds of one larger watershed so problems addressed in one watershed do not improve issues in a different, parallel watershed.”

To some degree, disputes over the appropriate scale of the AOC resonate with what Cohen and Bakker (2014: 132) identify as the “displacement of crisis” associated with the “eco-scalar fix,” defined as

a process of rescaling and reorganizing governance as a strategy of either internalizing or externalizing socio-environmental externalities, or both, and thereby displacing conflicts and crises, often through the construction of (purportedly ‘natural’) ecological scales, which simultaneously depoliticize and repoliticize governance.

To borrow terms from Cohen and Bakker (2014: 133), the disagreement about whether to define AOCs in watershed or bank-to-bank terms reflects a tension between strategies of internalization—”managing cumulative environmental effects in a singular spatial unit”—and strategies of externalization, or shifting responsibility from one governance program or entity to another. However, while Cohen and Bakker focus on the political implications of rescaling governance to the watershed scale, Great Lakes AOCs present a case of competition between two purportedly natural boundaries: the watershed and the banks of waterways. Here, advocates for the watershed scale emphasize internalizing downstream problems and upstream causes within the administrative unit of the AOC, while proponents of the bank-to-bank scale call for externalizing upstream issues to other programs or authorities. Although in some cases the watershed may indeed be a “technocratic” solution that forecloses democratic decision-making, the experience of the AOCs suggests that this is not always the case; the watershed is often the preferred scale of local citizen and stakeholder groups. Also, some of our interviews illustrate impacts of scaling AOCs on the accumulation of capital, for example, for tourism and real estate interests or for upstream agricultural users. However, our interviews suggest that other drivers, including conflicting perspectives on the scope of the AOC program and the most effective and efficient ways to achieve environmental objectives, are typically more important in the scaling and bounding of AOCs.

It is clear that the Great Lakes AOC Program represents one in which the adoption of watershed boundaries has not been implemented “as an unquestionable scale at which good water governance must take place” (Cohen and Davidson 2011: 8). But as for the question of when the watershed is the appropriate scale for water governance, the AOC program raises more questions than it answers. Cohen and Davidson (2011: 9) suggest that watersheds might be the most appropriate in at least two situations: “in cases where there is a hydrologically bounded issue that guides the selection of boundary choice, as well as in cases where the foundations of good water governance are in place in advance of re-scaling.” Although we cannot speak to whether the foundations of good water governance were in place, the problem in the case of many AOCs, as our interviews show, is that there has often been disagreement about how and where beneficial use impairments are “hydrologically bounded.” Finally, Cohen and Davidson (2011: 9) also suggest that “in cases where there is a similarly bounded challenge, clarity about why a particular set of boundaries were chosen may help to avert the challenges of boundary delineation and watershed asymmetries with problem-sheds.” Within the Great Lakes AOC program, clarity alone evidently did not always avert such challenges, given fundamental disagreements about what problems and activities should fall within the program’s scope and about how best to address beneficial use impairments.

Our analysis suggests several potential directions for future research, focused on both Great Lakes AOCs and elsewhere. Although contrasting the US experience with Canada’s was beyond the scope of this research, a richer understanding of boundary and scale definition in the governance of water quality would benefit from comparative analysis across countries. It would also benefit from comparison with similar programs in other parts of the world, particularly within basins that overlap national and sub-national borders. More broadly, we suggest research that examines cases in which watersheds are, in effect, “in competition” with other potential scales for water governance – including those in which other scales and boundaries have prevailed. Such investigations will help enrich our comprehension not only of how the “naturalness” of watersheds succeeds or fails in providing a justification for rescaling in particular situations, but also of how the selection of watersheds relates to how the power to make decisions about boundaries is itself distributed across scales.