Guido A. Pronsolino; Betty J. Pronsolino, As Trustees for the Guido A. Pronsolino and Betty J. Pronsolino Trust; the Mendocino County Farm Bureau; the California Farm Bureau Federation; the American Farm Bureau Federation, Plaintiffs-appellants, v. Wayne Nastri, Regional Administrator, United States Environmental Protection Agency Region 9; Christie Whitman, Administrator, United States Environmental Protection Agency; United States Environmental Protection Agency, Defendants-appellees,pacific Coast Federation of Fishermen's Associations, a California Non-profit Corporation; San Francisco Baykeeper, a California Public Benefit Corporation; Association of Metropolitan Sewerage Agencies, Defendants-intervenors-appellees.guido A. Pronsolino, Plaintiff, v. Wayne Nastri,* Regional Administrator, United States Environmental Protection Agency Region 9; Christie Whitman,** Administrator, United States Environmental Protection Agency; United States Environmental Protection Agency, Defendants-appellees,pacific Coast Federation of Fishermen's Associations, a California Non Profit Corporation; San Francisco Baykeeper, a California Public Benefit Corporation, Defendants-intervenors-appellees, v. American Forest & Paper Association; California Forestry Association, Plaintiff-intervenors-appellants, 291 F.3d 1123 (9th Cir. 2002)

The United States Environmental Protection Agency ("EPA") required California to identify the Garcia River as a water body with insufficient pollution controls and, as required for waters so identified, to set so-called "total maximum daily loads" ("TMDLs") — the significance of which we explain later — for pollution entering the river. Appellants challenge the EPA's authority under the Clean Water Act ("CWA" or the "Act") § 303(d), 33 U.S.C. § 1313(d), to apply the pertinent identification and TMDL requirements to the Garcia River. The district court rejected this challenge, and we do as well.

CWA § 303(d) requires the states to identify and compile a list of waters for which certain "effluent limitations" "are not stringent enough" to implement the applicable water quality standards for such waters. § 303(d) (1) (A). Effluent limitations pertain only to point sources of pollution; point sources of pollution are those from a discrete conveyance, such as a pipe or tunnel. Nonpoint sources of pollution are non-discrete sources; sediment run-off from timber harvesting, for example, derives from a nonpoint source. The Garcia River is polluted only by nonpoint sources. Therefore, neither the effluent limitations referenced in § 303(d) nor any other effluent limitations apply to the pollutants entering the Garcia River.

The precise statutory question before us is whether the phrase "are not stringent enough" triggers the identification requirement both for waters as to which effluent limitations apply but do not suffice to attain water quality standards and for waters as to which effluent limitations do not apply at all to the pollution sources impairing the water. We answer this question in the affirmative, a conclusion which triggers the application of the statutory TMDL requirement to waters such as the Garcia River.

Resolution of the statutory interpretation question before us, discrete though it is, "requires a familiarity with the history, the structure, and, alas, the jargon of the federal water pollution laws." Natural Res. Def. Council v. EPA, 915 F.2d 1314, 1316 (9th Cir. 1990). We therefore begin with a brief overview of the Act.

Congress enacted the CWA in 1972, amending earlier federal water pollution laws that had proven ineffective. EPA v. California, 426 U.S. 200, 202, 96 S. Ct. 2022, 48 L. Ed. 2d 578 (1976). Prior to 1972, federal water pollution laws relied on "water quality standards specifying the acceptable levels of pollution in a State's interstate navigable waters as the primary mechanism ... for the control of water pollution." Id. The pre-1972 laws did not, however, provide concrete direction concerning how those standards were to be met in the foreseeable future.

In enacting sweeping revisions to the nation's water pollution laws in 1972, Congress began from the premise that the focus "on the tolerable effects rather than the preventable causes of pollution" constituted a major shortcoming in the pre 1972 laws. Oregon Natural Desert Assoc. v. Dombeck, 172 F.3d 1092, 1096 (9th Cir. 1998) (quoting EPA v. State Water Resources Control Board, 426 U.S. 200, 202-03, 96 S. Ct. 2022, 48 L. Ed. 2d 578 (1976)). The 1972 Act therefore sought to target primarily "the preventable causes of pollution," by emphasizing the use of technological controls. Id.; Oregon Natural Res. Council v. United States Forest Serv., 834 F.2d 842, 849 (9th Cir. 1987).

At the same time, Congress decidedly did not in 1972 give up on the broader goal of attaining acceptable water quality. CWA § 101(a), 33 U.S.C. § 1251(a). Rather, the new statute recognized that even with the application of the mandated technological controls on point source discharges, water bodies still might not meet stateset water quality standards, Natural Res. Def. Council, 915 F.2d at 1316-17,. The 1972 statute therefore put in place mechanisms other than direct federal regulation of point sources, designed to "restore and maintain the chemical, physical, and biological integrity of the Nation's waters." § 101(a).

In so doing, the CWA uses distinctly different methods to control pollution released from point sources and that traceable to nonpoint sources. Oregon Natural Res. Council, 834 F.2d at 849. The Act directly mandates technological controls to limit the pollution point sources may discharge into a body of water. Dombeck, 172 F.3d at 1096. On the other hand, the Act "provides no direct mechanism to control nonpoint source pollution but rather uses the `threat and promise' of federal grants to the states to accomplish this task," id. at 1097 (citations omitted), thereby "recogniz [ing], preserv [ing], and protect [ing] the primary responsibilities and rights of States to prevent, reduce, and eliminate pollution, [and] to plan the development and use ... of land and water resources...." § 101(b).

B. The Structure of CWA § 303, 33 U.S.C. § 1313

Section 303 is central to the Act's carrot-and-stick approach to attaining acceptable water quality without direct federal regulation of nonpoint sources of pollution. Entitled "Water Quality Standards and Implementation Plans," the provision begins by spelling out the statutory requirements for water quality standards: "Water quality standards" specify a water body's "designated uses" and "water quality criteria," taking into account the water's "use and value for public water supplies, propagation of fish and wildlife, recreational purposes, and agricultural, industrial, and other purposes...." § 303(c) (2). The states are required to set water quality standards for all waters within their boundaries regardless of the sources of the pollution entering the waters. If a state does not set water quality standards, or if the EPA determines that the state's standards do not meet the requirements of the Act, the EPA promulgates standards for the state. §§ 303(b), (c) (3)-(4).

2. Section 303(d): "Identification of Areas with Insufficient Controls; Maximum Daily Load"¹ 

Section 303(d) (1) (A) requires each state to identify as "areas with insufficient controls" "those waters within its boundaries for which the effluent limitations required by section [301(b) (1) (A)] and section [301(b) (1) (B)] of this title are not stringent enough to implement any water quality standard applicable to such waters." Id. The CWA defines "effluent limitations" as restrictions on pollutants "discharged from point sources." CWA § 502(11), 33 U.S.C. § 1362(11). Section 301(b) (1) (A) mandates application of the "best practicable control technology" effluent limitations for most point source discharges, while § 301(b) (1) (B) mandates application of effluent limitations adopted specifically for secondary treatment at publicly owned treatment works. § 301(b) (1), 33 U.S.C. § 1311(b) (1).

For waters identified pursuant to § 303(d) (1) (A) (the "§ 303(d) (1) list"), the states must establish the "total maximum daily load" ("TMDL") for pollutants identified by the EPA as suitable for TMDL calculation.²  § 303(d) (1) (C). "A TMDL defines the specified maximum amount of a pollutant which can be discharged or `loaded' into the waters at issue from all combined sources." Dioxin/Organochlorine Center v. Clarke, 57 F.3d 1517, 1520 (9th Cir. 1995).³  The TMDL "shall be established at a level necessary to implement the applicable water quality standards...." § 303(d) (1) (C).

Section 303(d) (2), in turn, requires each state to submit its § 303(d) (1) list and TMDLs to the EPA for its approval or disapproval. If the EPA approves the list and TMDLs, the state must incorporate the list and TMDLs into its "continuing planning process," the requirements for which are set forth in § 303(e). § 303(d) (2). If the EPA disapproves either the § 303(d) (1) list or any TMDLs, the EPA must itself put together the missing document or documents. Id. The state then incorporates any EPA-set list or TMDL into the state's continuing planning process. Id.

Each state must also identify all waters not placed on its § 303(d) (1) list (the "§ 303(d) (3) list") and "estimate" TMDLs for pollutants in those waters. § 303(d) (3). There is no requirement that the EPA approve the § 303(d) (3) lists or the TMDLs estimated for those waters. Id.

The EPA in regulations has made more concrete the statutory requirements. Those regulations, in summary, define "water quality limited segment [s]" — those waters that must be included on the § 303(d) (1) list — as " [a]ny segment where it is known that water quality does not meet applicable water quality standards, and/or is not expected to meet applicable water quality standards, even after the application of the technology-based effluent limitations required by sections 301(b) and 306 [, 33 U.S.C. § 1316]." 40 C.F.R. § 130.2(j) (2000). The regulations then divide TMDLs into two types: "load allocations," for nonpoint source pollution, and "wasteload allocations," for point source pollution. § 130.2(g)-(i); see also p. 7919, infra. Under the regulations, states must identify those waters on the § 303(d) (1) lists as "still requiring TMDLs" if any required effluent limitation or other pollution control requirement (including those for nonpoint source pollution) will not bring the water into compliance with water quality standards. § 130.7(b) (2000).⁴ 

The final pertinent section of § 303, § 303(e), requiring each state to have a "continuing planning process," gives some operational force to the prior information-gathering provisions. The EPA may approve a state's continuing planning process only if it "will result in plans for all navigable waters within such State" that include, inter alia, effluent limitations, TMDLs, areawide waste management plans for nonpoint sources of pollution, and plans for "adequate implementation, including schedules of compliance, for revised or new water quality standards." § 303(e) (3).

The upshot of this intricate scheme is that the CWA leaves to the states the responsibility of developing plans to achieve water quality standards if the statutorily-mandated point source controls will not alone suffice, while providing federal funding to aid in the implementation of the state plans. See Dombeck, 172 F.3d at 1097; § 303(e); see also § 319(h), 33 U.S.C. § 1329(h) (providing for grants to states to combat nonpoint source pollution). TMDLs are primarily informational tools that allow the states to proceed from the identification of waters requiring additional planning to the required plans. See Alaska Center for the Environment v. Browner, 20 F.3d 981, 984-85 (9th Cir. 1994). As such, TMDLs serve as a link in an implementation chain that includes federally-regulated point source controls, state or local plans for point and nonpoint source pollution reduction, and assessment of the impact of such measures on water quality, all to the end of attaining water quality goals for the nation's waters.

In 1992, California submitted to the EPA a list of waters pursuant to § 303(d) (1) (A). Pursuant to § 303(d) (2), the EPA disapproved California's 1992 list because it omitted seventeen water segments that did not meet the water quality standards set by California for those segments. Sixteen of the seventeen water segments, including the Garcia River, were impaired only by nonpoint sources of pollution.⁵  After California rejected an opportunity to amend its § 303(d) (1) list to include the seventeen sub-standard segments, the EPA, again acting pursuant to § 303(d) (2), established a new § 303(d) (1) list for California, including those segments on it. California retained the seventeen segments on its 1994, 1996, and 1998 § 303(d) (1) lists.

California did not, however, establish TMDLs for the segments added by the EPA. Environmental and fishermen's groups sued the EPA in 1995 to require the EPA to establish TMDLs for the seventeen segments, and in a March 1997 consent decree the EPA agreed to do so. See Pacific Coast Fishermens Assocs. v. Marcus, No. 95-4474. According to the terms of the consent decree, the EPA set March 18, 1998, as the deadline for the establishment of a TMDL for the Garcia River. When California missed the deadline despite having initiated public comment on a draft TMDL and having prepared a draft implementation plan, the EPA established a TMDL for the Garcia River. The EPAs TMDL differed only slightly from the states draft TMDL.

The Garcia River TMDL for sediment is 552 tons per square mile per year, a sixty percent reduction from historical loadings. The TMDL allocates portions of the total yearly load among the following categories of nonpoint source pollution: a) "mass wasting" associated with roads; b) "mass wasting" associated with timber-harvesting; c) erosion related to road surfaces; and d) erosion related to road and skid trail crossings.

In 1960, appellants Betty and Guido Pronsolino purchased approximately 800 acres of heavily logged timber land in the Garcia River watershed. In 1998, after regrowth of the forest, the Pronsolinos applied for a harvesting permit from the California Department of Forestry ("Forestry").

In order to comply with the Garcia River TMDL, Forestry and/or the state's Regional Water Quality Control Board required, among other things, that the Pronsolinos' harvesting permit provide for mitigation of 90% of controllable road-related sediment run-off and contain prohibitions on removing certain trees and on harvesting from mid-October until May 1.⁶  The Pronsolinos' forester estimates that the large tree restriction will cost the Pronsolinos $750,000.

Larry Mailliard, a member of the Mendocino County Farm Bureau, submitted a draft harvesting permit on February 4, 1998, for a portion of his property in the Garcia River watershed. Forestry granted a final version of the permit after incorporation of a 60.3% reduction of sediment loading, a requirement included to comply with the Garcia River TMDL. Mr. Mailliard's forester estimates that the additional restrictions imposed to comply with the Garcia River TMDL will cost Mr. Mailliard $10,602,000.

Bill Barr, another member of the Mendocino County Farm Bureau, also applied for a harvesting permit in 1998 for his property located within the Garcia River watershed. Forestry granted the permit after incorporation of restrictions similar to those included in the Pronsolinos' permit. A forester states that these additional restrictions, included to comply with the TMDL, will cost Mr. Barr at least $962,000.

On August 12, 1999, the Pronsolinos, the Mendocino County Farm Bureau, the California Farm Bureau Federation, and the American Farm Bureau Federation brought this action pursuant to the Administrative Procedure Act, 5 U.S.C. §§ 702, 704, in the District Court for the Northern District of California against the EPA and two of its administrators. The Pronsolinos challenged the EPA's authority to impose TMDLs on rivers polluted only by nonpoint sources of pollution and sought a determination of whether the Act authorized the Garcia River TMDL.

The parties filed cross-motions for summary judgment. On August 6, 2000, the district court entered final judgment in favor of the EPA. The Pronsolinos timely filed this appeal.⁷ 

As this is a summary judgment case, our review of the district court's decision is, of course, de novo. See Oregon Natural Res. Council, 834 F.2d at 844. Harder to answer is the question of the degree of deference we owe the EPA's regulations and decisions interpreting and applying CWA § 303.

The EPA argues that we owe deference to the interpretation of § 303 embodied in its regulations, pursuant to Chevron U.S.A., Inc. v. Natural Res. Def. Council, 467 U.S. 837, 104 S. Ct. 2778, 81 L. Ed. 2d 694 (1984). An agency's statutory interpretation is entitled to Chevron deference if "Congress delegated authority to the agency generally to make rules carrying the force of law, and ... the agency interpretation claiming deference was promulgated in the exercise of that authority." United States v. Mead, 533 U.S. 218, 226-27, 121 S. Ct. 2164, 150 L. Ed. 2d 292 (2001). If Chevron deference applies, we must defer to the agency's interpretation as long as it is reasonably consistent with the statute. Id. at 229, 121 S. Ct. 2164.

The Pronsolinos urge an approach at the opposite end of the deference spectrum, asserting that the EPA's interpretation should receive no deference at all because, they maintain, the EPA has inconsistently interpreted § 303(d) and has not included its current interpretation in a regulation that has the force of law. In between Chevron deference and no deference, however, lies another possibility. The Supreme Court in Mead recently clarified that agency interpretations that do not qualify for Chevron deference may nonetheless merit deference pursuant to Skidmore v. Swift & Co., 323 U.S. 134, 65 S. Ct. 161, 89 L. Ed. 124 (1944). 533 U.S. at 237, 121 S. Ct. 2164. Under Skidmore, we defer to the agency's position according to its persuasiveness. Mead, 533 U.S. at 221, 121 S. Ct. 2164. Factors relevant to determining persuasiveness may include the agency's expertise, care, consistency, and formality, as well as the logic of the agency's position. Id. at 228, 121 S. Ct. 2164 (citing Skidmore, 323 U.S. at 139-40, 65 S. Ct. 161). Thus, we must consider whether the EPA's interpretation is due Chevron deference, as the EPA argues; no deference, as the Pronsolinos argue; or, alternatively, Skidmore deference (and, if so, to what extent).

The EPA has the statutory authority to enact a rule carrying the force of law as to the issue at hand. The CWA delegates to the EPA the general rule-making authority necessary for the agency to carry out its functions under the Act. CWA § 501(a), 33 U.S.C. § 1361(a). One of those functions is to approve or disapprove the § 303(d) (1) list and any required TMDLs. § 303(d) (2). So the EPA has the delegated authority to enact regulations carrying the force of law regarding the identification of § 303(d) (1) waters and TMDLs. See Mead, 533 U.S. at 229, 121 S. Ct. 2164.

The Pronsolinos do not contest the EPA's general rule-making authority but maintain that it has not been exercised, because no currently-operative EPA regulation expressly precludes the Pronsolinos' position that §§ 303(d) (1) (A) and (C) do not apply to rivers impaired only by nonpoint source pollution.⁸  The pertinent regulations do, however, reflect the EPA's interpretation — that is, that the statute requires the identification on § 303(d) (1) lists of waters impaired only by nonpoint sources of pollution — and the EPA so reads its regulations.

The EPA regulations pertinent to § 303(d) (1) lists and TMDLs focus on the attainment of water quality standards, whatever the source of any pollution. For instance, the EPA's regulations define TMDLs as the "sum of the individual WLAs [wasteload allocations] for point sources and LAs [load allocations] for nonpoint sources and natural background." 40 C.F.R. § 130.2(i). Section 130.2 also defines a "wasteload allocation" as the "portion of a receiving water's loading capacity that is allocated to one of its existing or future point sources of pollution," § 130.2(h), and a "load allocation" as the "portion of a receiving water's loading capacity that is attributed either to one of its existing or future nonpoint sources of pollution or to natural background sources," § 130.2(g). The load allocation regulation also advises that, if possible, "natural and nonpoint source loads should be distinguished." Id. No reason appears why, under this TMDL definition, the amount of either point source loads or nonpoint source loads cannot be zero. If the wasteload allocation is zero, then the TMDL would cover only the nonpoint sources and natural background sources. So read, the regulation provides that a TMDL can apply where there is no wasteload allocation for point source pollution. See also, e.g., § 130.2 (referencing the establishment of TMDLs for non-point source pollution); 40 C.F.R. § 130.7(c) (1) (ii) (TMDLs must be established for all pollutants that prevent the attainment of water quality standards).

Section 130.7 evinces the same understanding. That regulation directs states to identify those waters listed pursuant to § 303(d) (1) that still require the establishment of TMDLs if:

(i) Technology-based effluent limitations required by sections 301(b), 306, 307, or other sections of the Act;

(ii) More stringent effluent limitations (including prohibitions) required ...; and

(iii) Other pollution control requirements (e.g., best management practices) required by local, State, or Federal authority are not stringent enough to implement any water quality standards ... applicable to such waters.

§ 130.7(b) (1). "Best management practices" pertain to non-point sources of pollution. CWA § 208, 33 U.S.C. § 1288; CWA § 319, 33 U.S.C. § 1329. So, again, § 130.7 does not distinguish between sources of pollution for purposes of applying the TMDL requirement. Instead, control requirements applicable to either type of pollution receive equal treatment in the quest to achieve water quality standards.

Also consistent with application of the § 303(d) (1) listing and TMDL requirements to waters impaired only by nonpoint sources is the regulation addressing water quality standards. Section 130.3 explains that " [s]uch standards serve the dual purposes of establishing the water quality goals for a specific water body and serving as the regulatory basis for establishment of water quality-based treatment controls and strategies beyond the technology-based level of treatment required...." 40 C.F.R. § 130.3. One purpose of water quality standards therefore — and not surprisingly — is to provide federally-approved goals to be achieved both by state controls and by federal strategies other than point-source technology-based limitations. This purpose pertains to waters impaired by both point and nonpoint source pollution. The regulations addressing states' water quality management plans, intended to attain the promulgated water quality standards, confirm this understanding. Such plans must include, among other things, TMDLs, effluent limitations, and "nonpoint source management and control." 40 C.F.R. § 130.6 (emphasis added).

In short, the EPA's regulations concerning § 303(d) (1) lists and TMDLs apply whether a water body receives pollution from point sources only, nonpoint sources only, or a combination of the two. The EPA has issued directives concerning the states' CWA § 303(d) requirements in conformity with this understanding of its regulations. See, e.g., Memorandum from Geoffrey Grubbs, Director, EPA Assessment and Watershed Protection Division, to Water Quality Branch Chiefs and TMDL Coordinators (Aug. 13, 1992) (Section 303(d) (1) (A) "applies equally to segments affected by point sources only, a combination of point and nonpoint sources, and nonpoint sources only."); EPA, National Clarifying Guidance for 1998 State and Territory Clean Water Act Section 303(d) Listing Decisions 6 (1997) ("Consistent with long-standing EPA policy, regulations, and practice, States should include waterbodies impaired by nonpoint sources alone on 1998 section 303(d) (1) (A) lists....").

In light of the current regulations and the agency's understanding of those regulations, as well as the delegated authority of the EPA to interpret the CWA, the EPA's interpretation is entitled to Chevron deference. See Mead, 533 U.S. at 226-27, 121 S. Ct. 2164; see also Auer v. Robbins, 519 U.S. 452, 461, 117 S. Ct. 905, 137 L. Ed. 2d 79 (1997) (stating that an agency's interpretation of its own regulation is "controlling unless plainly erroneous or inconsistent with the regulation") (citations and internal quotation marks omitted).

At the least, however, we owe the agency's interpretation substantial deference under Skidmore. Cf. Mead, 533 U.S. at 237 n. 18, 121 S. Ct. 2164 ("It is, of course, true that the limit of Chevron deference is not marked by a hard-edged rule."). Section 303(d) is one of numerous interwoven components that together make up an intricate statutory scheme addressing technically complex environmental issues. Confronted with an issue dependent upon, and the resolution of which will affect, a complicated, science-driven statute for which the EPA has delegated regulatory authority, we consider the EPA's interpretation of the issue informative. See Mead, 533 U.S. at 234, 121 S. Ct. 2164 (noting the "specialized experience and broader investigations and information available" to agencies and "the value of uniformity in [ ] administrative and judicial understandings of what a national law requires") (citations and internal quotation marks omitted).

Appellants maintain that we should instead ignore the EPA's position, arguing that the Agency has not consistently interpreted the statute. We disagree with this characterization of the EPA's position over the thirty-year period since the enactment of the statute.

The first regulations promulgated after the enactment of the CWA in 1972 quite clearly required the identification on § 303(d) (1) lists of waters polluted only by nonpoint sources. The EPA defined as a "water quality" segment — again, those water bodies to be included on the § 303(d) (1) list, see 43 Fed.Reg. 60662, 60665 (Dec. 28, 1978) — any water "where it is known that water quality does not meet applicable water quality standards and/or is not expected to meet applicable water quality standards even after the application of the effluent limitations required...." 40 C.F.R. § 130.2(o) (1) (1978); id. (1977); id. (1976); 40 C.F.R. § 130.11(d) (1) (1975); id. (1974); id. (1973).⁹  In contrast, the EPA defined as an "effluent limitation" segment — those waters making up the separate § 303(d) (3) list¹⁰  — any water "where it is known that water quality is meeting and will continue to meet applicable water quality standards or where there is adequate demonstration that water quality will meet applicable water quality standards after the application of the effluent limitations required...." 40 C.F.R. § 130.2(o) (2) (1978); id. (1977); id. (1976); 40 C.F.R. § 130.11(d) (2) (1975); id. (1974); id. (1973).¹¹  Thus, if a water segment had not met, or would not soon meet, applicable water quality standards, regardless of the source of pollution, the EPA required its identification pursuant to § 303(d) (1) (A). In other words, the EPA initially interpreted § 303(d) exactly as it does today.¹² 

The Pronsolinos nevertheless contend that the EPA's current interpretation is an invention of the early 1990s. They point out that until that time the EPA did not actively police the requirement that states include on their § 303(d) (1) lists waters polluted only by nonpoint source pollution. While that is true, that agency stance reflected a more general regulatory failure to enforce the § 303(d) requirements, not a failure with regard only to waters impaired by nonpoint sources. Until the early 1990s, the EPA focused its attention almost entirely on the new point source technological controls, to the exclusion of § 303(d) and the TMDL program. See Pronsolino v. Marcus, 91 F. Supp. 2d 1337, 1354 (N.D. Cal. 2000)¹³  (citing United States General Accounting Office, Water Pollution: More EPA Action Needed to Improve the Quality of Heavily Polluted Waters, GAO Report to the Chairman: Subcommittee on Regulation and Business Opportunities Committee on Small Business, House of Representatives (Jan. 1989)); see also id. at 1353-54(describing the history of EPA enforcement action with respect to § 303(d)); Oliver A. Houck, The Clean Water Act TMDL Program: Law, Policy, and Implementation 49-56 (1999) ("The Clean Water Act TMDL Program") (same). We have not found, and the Pronsolinos have not pointed to, any statement by the EPA — either in regulations or otherwise — that is inconsistent with the interpretation the agency now espouses.

In short, Congress entrusted to the EPA the responsibility of approving or disapproving § 303(d) (1) lists, bestowing upon it the discretion that comes with such responsibility; the EPA has specialized experience regarding the CWA which this court lacks; and the agency has consistently interpreted the provisions at issue. We conclude that the EPA's interpretation is one to which we owe substantial Skidmore deference, at the very least. See Mead, 533 U.S. at 227-28, 121 S. Ct. 2164.

In the end, though, it does not much matter in this case whether we review the EPA's position through the Chevron or Skidmore/Mead prism. Under both the more and less rigorous versions of the judicial review standard, the Agency's position is, as the discussion below indicates, more than sufficiently supported by the statutory materials.

Section 303(d) (1) (A) requires listing and calculation of TMDLs for "those waters within [the state's] boundaries for which the effluent limitations required by section [301(b) (1) (A)] and section [301(b) (1) (B)] of this title are not stringent enough to implement any water quality standard applicable to such waters." § 303(d) (emphasis added). The precise statutory question before us is whether, as the Pronsolinos maintain, the term "not stringent enough to implement ... water quality standard [s]" as used in § 303(d) (1) (A) must be interpreted to mean both that application of effluent limitations will not achieve water quality standards and that the waters at issue are subject to effluent limitations. As only waters with point source pollution are subject to effluent limitations, such an interpretation would exclude from the § 303(d) (1) listing and TMDL requirements waters impaired only by nonpoint sources of pollution.

The EPA, as noted, interprets "not stringent enough to implement ... water quality standard [s]" to mean "not adequate" or "not sufficient ... to implement any water quality standard," and does not read the statute as implicitly containing a limitation to waters initially covered by effluent limitations. According to the EPA, if the use of effluent limitations will not implement applicable water quality standards, the water falls within § 303(d) (1) (A) regardless of whether it is point or nonpoint sources, or a combination of the two, that continue to pollute the water.

Whether or not the appellants' suggested interpretation is entirely implausible, it is at least considerably weaker than the EPA's competing construction. The Pronsolinos' version necessarily relies upon: (1) understanding "stringent enough" to mean "strict enough" rather than "thorough going enough" or "adequate" or "sufficient";¹⁴  and (2) reading the phrase "not stringent enough" in isolation, rather than with reference to the stated goal of implementing "any water quality standard applicable to such waters." Where the answer to the question "not stringent enough for what?" is "to implement any [applicable] water quality standard," the meaning of "stringent" should be determined by looking forward to the broad goal to be attained, not backwards at the inadequate effluent limitations. One might comment, for example, about a teacher that her standards requiring good spelling were not stringent enough to assure good writing, as her students still used bad grammar and poor logic. Based on the language of the contested phrase alone, then, the more sensible conclusion is that the § 303(d) (1) list must contain any waters for which the particular effluent limitations will not be adequate to attain the statute's water quality goals.

Placing the phrase in its statutory context supports this conclusion. Section 303(d) begins with the requirement that each state "identify those waters within its boundaries...." § 303(d) (1) (A). So the statute's starting point for the listing project is a compilation of each and every navigable water within the state. Then, only those waters that will attain water quality standards after application of the new point source technology are excluded from the § 303(d) (1) list, leaving all those waters for which that technology will not "implement any water quality standard applicable to such waters." § 303(d) (1) (A); see American Wildlands v. Browner, 260 F.3d 1192, 1194 (10th Cir. 2001) (" [E]ach state is required to identify all of the waters within its borders not meeting water quality standards and establish [TMDLs] for those waters.") (citing § 303(d)); Pronsolino, 91 F. Supp. 2d at 1347. The alternative construction, in contrast, would begin with a subset of all the state's waterways, those that have point sources subject to effluent limitations, and would result in a list containing only a subset of that subset — those waters as to which the applicable effluent limitations are not adequate to attain water quality standards.

The Pronsolinos' contention to the contrary notwithstanding, no such odd reading of the statute is necessary in order to give meaning to the phrase "for which the effluent limitations required by section [301(b) (1) (A)] and section [301(b) (1) (B)]... are not stringent enough." The EPA interprets § 303(d) (1) (A) to require the identification of any waters not meeting water quality standards only if specified effluent limitations would not achieve those standards. 40 C.F.R. § 130.2(j). If the pertinent effluent limitations would, if implemented, achieve the water quality standards but are not in place yet, there need be no listing and no TMDL calculation. Id.

So construed, the meaning of the statute is different than it would be were the language recast to state only that "Each State shall identify those waters within its boundaries ... [not meeting] any water quality standard applicable to such waters." Under the EPA's construction, the reference to effluent limitations reflects Congress' intent that the EPA focus initially on implementing effluent limitations and only later avert its attention to water quality standards. See e.g., 1 Legislative History 171("The Administrator should assign secondary priority to [§ 303] to the extent limited manpower and funding may require a choice between a water quality standards process and early and effective implementation of the effluent limitation permit program." (statement of Sen. Muskie, principal author of the CWA and the Chair of the Senate's Public Works Committee)); see also Environmental Def. Fund, Inc. v. Costle, 657 F.2d 275, 279 (D.C. Cir. 1981) (The 1972 CWA "assigned secondary priority to the [water quality] standards and placed primary emphasis upon both a point source discharge permit program and federal technology-based effluent limitations....").¹⁵ 

Given all these language considerations, it is not surprising that the only time this court addressed the reach of § 303(d) (1) (A), it rejected a reading of § 303(d) (1) (A) similar to the one the Pronsolinos now proffer. In Dioxin, 57 F.3d at 1526-27, the plaintiffs argued that the phrase "not stringent enough" prohibited the EPA from listing under § 303(d) (1) (A) and establishing TMDLs for toxic pollutants, until after the implementation and proven failure of § 301(b) (1) (A) "best practicable technology" effluent limitations. Toxic pollutants, however, are not subject to "best practicable technology" controls,¹⁶  but to more demanding "best available technology," precisely because of their toxicity. Id.

The court in Dioxin held that the EPA acted within its statutory authority in setting TMDLs for toxic pollutants, even though the effluent limitations referenced by § 303(d) (1) (A) did not apply to those pollutants. Id. at 1528. The court explained that, since best practical technology effluent limitations do not apply to toxic pollutants, those limitations are, as a matter of law, "not stringent enough" to achieve water quality standards. Id. In other words, Dioxin read § 303(d) (1) (A) as applying to all waters in the state, not only to the subset covered by certain kinds of effluent controls, and it understood "not stringent enough" to mean "not adequate for" or "inapplicable to."

Nothing in § 303(d) (1) (A) distinguishes the treatment of point sources and nonpoint sources as such; the only reference is to the "effluent limitations required by" § 301(b) (1). So if the effluent limitations required by § 301(b) (1) are "as a matter of law" "not stringent enough" to achieve the applicable water quality standards for waters impaired by point sources not subject to those requirements, then they are also "not stringent enough" to achieve applicable water quality standards for other waters not subject to those requirements, in this instance because they are impacted only by nonpoint sources. Additionally, the Dioxin court, applying Chevron deference, upheld the EPA's interpretation of § 303(d) "as requiring TMDLs where existing pollution controls will not lead to attainment of water standards," id. at 1527; see also 40 C.F.R. § 130.7(b), a holding that directly encompasses waters polluted only by nonpoint sources.

The Pronsolinos' objection to this view of § 303(d), and of Dioxin, is, in essence, that the CWA as a whole distinguishes between the regulatory schemes applicable to point and non-point sources, so we must assume such a distinction in applying §§ 303(d) (1) (A) and (C). We would hesitate in any case to read into a discrete statutory provision something that is not there because it is contained elsewhere in the statute. But here, the premise is wrong: There is no such general division throughout the CWA.

Point sources are treated differently from nonpoint sources for many purposes under the statute, but not all. In particular, there is no such distinction with regard to the basic purpose for which the § 303(d) list and TMDLs are compiled, the eventual attainment of state-defined water quality standards. Water quality standards reflect a state's designated uses for a water body and do not depend in any way upon the source of pollution. See § 303(a)-(c).

Nor is there any other basis for inferring from the structure of the Act an implicit limitation in §§ 303(d) (1) (A) and (C). The statutory subsection requiring water quality segment identification and TMDLs, § 303(d), appears in the section entitled "Water Quality Standards and Implementation Plans," not in the immediately preceding section, CWA § 302, 33 U.S.C. § 1312, entitled "Water Quality Related Effluent Limitations." So the section heading does not suggest any limitation to waters subject to effluent limitations. Porter v. Nussle, 534 U.S. 516, 122 S. Ct. 983, 990, 152 L. Ed. 2d 12 (2002) (" [T]he title of a statute and the heading of a section are tools available for the resolution of a doubt about the meaning of a statute.") (citation omitted).

Additionally, § 303(d) follows the subsections setting forth the requirements for water quality standards, § 303(a)-(c) — which, as noted above, apply without regard to the source of pollution — and precedes the "continuing planning process" subsection, § 303(e), which applies broadly as well. Thus, § 303(d) is structurally part of a set of provisions governing an interrelated goal-setting, information-gathering, and planning process that, unlike many other aspects of the CWA, applies without regard to the source of pollution.

True, there are, as the Pronsolinos point out, two sections of the statute as amended, § 208 and § 319, that set requirements exclusively for nonpoint sources of pollution. But the structural inference we are asked to draw from those specialized sections — that no other provisions of the Act set requirements for waters polluted by nonpoint sources — simply does not follow. Absent some irreconcilable contradiction between the requirements contained in §§ 208 and 319, on the one hand, and the listing and TMDL requirements of § 303(d), on the other, both apply.

There is no such contradiction. Section 208 provides for federal grants to encourage the development of state "areawide waste treatment management plans" for areas with substantial water quality problems, § 208(a), (f), and requires that those plans include a process for identifying and controlling nonpoint source pollution "to the extent feasible." § 208(b) (2) (F). Section 319, added to the CWA in 1987, directs states to adopt "nonpoint source management programs"; provides grants for nonpoint source pollution reduction; and requires states to submit a report to the EPA that "identifies those navigable waters within the State which, without additional action to control nonpoint sources of pollution, cannot reasonably be expected to attain or maintain applicable water quality standards or the goals and requirements of this chapter." § 319(a) (1) (A). This report must also describe state programs for reducing nonpoint source pollution and the process "to reduce, to the maximum extent practicable, the level of pollution" resulting from particular categories of nonpoint source pollution. § 319(a) (1) (C), (D).

The CWA is replete with multiple listing and planning requirements applicable to the same waterways (quite confusingly so, indeed), so no inference can be drawn from the overlap alone. See, e.g., § 208(b); § 303(d) (1) (A), (d) (1) (B), (d) (3), (e); CWA § 304(l), 33 U.S.C. § 1314(l); CWA § 314, 33 U.S.C. § 1324(a); § 319(a). Nor are we willing to draw the more discrete inference that the § 303(d) listing and TMDL requirements cannot apply to nonpoint source pollutants because the planning requirements imposed by § 208 and § 319 are qualified ones — "to the extent feasible" and "to the maximum extent practicable" — while the § 303(d) requirements are unbending. For one thing, the water quality standards set under § 303 are functional and may permit more pollution than it is "feasible" or "practicable" to eliminate, depending upon the intended use of a particular waterway. For another, with or without TMDLs, the § 303(e) plans for attaining water quality standards must, without qualification, account for elimination of nonpoint source pollution to the extent necessary to meet those standards. § 303(e) (3) (F).

The various reporting requirements that apply to nonpoint source pollution are no more impermissibly redundant than are the planning requirements. Congress specifically provided that in preparing the § 319 report, states may rely on information from § 303(e), which incorporates the TMDLs. § 319(a) (2). Moreover, states must produce a § 319 report only once, but must update the § 303(d) (1) list periodically. § 319; § 303(d) (2). Also, the § 319 report requires the identification of a plan to reduce nonpoint source pollution, without regard to the attainment of water quality standards, while the plans generated using the § 303(d) (1) lists and TMDLs are guided by the goal of achieving those standards. § 319; § 303(d), (e).

Essentially, § 319 encourages the states to institute an approach to the elimination of nonpoint source pollution similar to the federally-mandated effluent controls contained in the CWA, while § 303 encompasses a water quality based approach applicable to all sources of water pollution. As various sections of the Act encourage different, and complementary, state schemes for cleaning up nonpoint source pollution in the nation's waterways, there is no basis for reading any of those sections — including § 303(d) — out of the statute.

There is one final aspect of the Act's structure that bears consideration because it supports the EPA's interpretation of § 303(d): The list required by § 303(d) (1) (A) requires that waters be listed if they are impaired by a combination of point sources and nonpoint sources; the language admits of no other reading. Section 303(d) (1) (C), in turn, directs that TMDLs "shall be established at a level necessary to implement the applicable water quality standards...." Id. (emphasis added). So, at least in blended waters, TMDLs must be calculated with regard to nonpoint sources of pollution; otherwise, it would be impossible "to implement the applicable water quality standards," which do not differentiate sources of pollution. This court has so recognized. Browner, 20 F.3d at 985 ("Congress and the EPA have already determined that establishing TMDLs is an effective tool for achieving water quality standards in waters impacted by non-point source pollution.").

Nothing in the statutory structure — or purpose — suggests that Congress meant to distinguish, as to § 303(d) (1) lists and TMDLs, between waters with one insignificant point source and substantial nonpoint source pollution and waters with only nonpoint source pollution. Such a distinction would, for no apparent reason, require the states or the EPA to monitor waters to determine whether a point source had been added or removed, and to adjust the § 303(d) (1) list and establish TMDLs accordingly. There is no statutory basis for concluding that Congress intended such an irrational regime.

Looking at the statute as a whole, we conclude that the EPA's interpretation of § 303(d) is not only entirely reasonable but considerably more convincing than the one offered by the plaintiffs in this case.¹⁷ 

The Pronsolinos finally contend that, by establishing TMDLs for waters impaired only by nonpoint source pollution, the EPA has upset the balance of federal-state control established in the CWA by intruding into the states' traditional control over land use. See Solid Waste Agency of Northern Cook County v. United States Army Corps of Eng'rs, 531 U.S. 159, 172-73, 121 S. Ct. 675, 148 L. Ed. 2d 576 (2001). That is not the case.

The Garcia River TMDL identifies the maximum load of pollutants that can enter the Garcia River from certain broad categories of nonpoint sources if the river is to attain water quality standards. It does not specify the load of pollutants that may be received from particular parcels of land or describe what measures the state should take to implement the TMDL. Instead, the TMDL expressly recognizes that "implementation and monitoring" "are state responsibilities" and notes that, for this reason, the EPA did not include implementation or monitoring plans within the TMDL.¹⁸  EPA, Garcia River Sediment Total Maximum Daily Load 43(Mar. 16, 1998).

Moreover, § 303(e) requires — separately from the § 303(d) (1) listing and TMDL requirements — that each state include in its continuing planning process "adequate implementation, including schedules of compliance, for revised or new water quality standards" "for all navigable waters within such State." § 303(e) (3). The Garcia River TMDL thus serves as an informational tool for the creation of the state's implementation plan, independently — and explicitly — required by Congress.

California chose both if and how it would implement the Garcia River TMDL. States must implement TMDLs only to the extent that they seek to avoid losing federal grant money; there is no pertinent statutory provision otherwise requiring implementation of § 303 plans or providing for their enforcement. See CWA § 309, 33 U.S.C. § 1319; CWA § 505, 33 U.S.C. § 1365.¹⁹ 

Finally, it is worth noting that the arguments that the Pronsolinos raise here would apply equally to nonpoint source pollution controls for blended waters. Yet, as discussed above, Congress definitely required that the states or the EPA establish TMDLs for all pollutants in waters on § 303(d) (1) lists, including blended waters.

We conclude that the Pronsolinos' federalism basis for reading § 303 against its own words and structure is unfounded.

For all the reasons we have surveyed, the CWA is best read to include in the § 303(d) (1) listing and TMDLs requirements waters impaired only by nonpoint sources of pollution. Moreover, to the extent the statute is ambiguous — which is not very much — the substantial deference we owe the EPA's interpretation, under either Chevron or Skidmore, requires that we uphold the agency's more than reasonable interpretation. We therefore hold that the EPA did not exceed its statutory authority in identifying the Garcia River pursuant to § 303(d) (1) (A) and establishing the Garcia River TMDL, even though the river is polluted only by nonpoint sources of pollution.

The decision of the district court is AFFIRMED.