III

In both the consolidated cases before the Court the record contains evidence suggesting the possible existence of a significant nexus according to the principles outlined above. Thus the end result in these cases and many others to be considered by the Corps may be the same as that suggested by the dissent, namely, that the Corps’ assertion of jurisdiction is valid. Given, however, that neither the agency nor the reviewing courts properly considered the issue, a remand is appropriate, in my view, for appli- cation of the controlling legal standard.

Rapanos

As the dissent points out, in Rapanos, No. 04–1034, an expert whom the District Court found “eminently qualified” and “highly credible,” App. to Pet. for Cert. B7, testified that the wetlands were providing “habitat, sediment trapping, nutrient recycling, and flood peak diminution, reduction flow water augmentation.” 4 Tr. 96 (Apr. 5, 1999). Although the expert had “not studied the upstream drainage of these sites” and thus could not assert that the wetlands were performing important pollutant-trapping functions, ibid., he did observe:

“we have a situation in which the flood water attenuation in that water is held on the site in the wetland . . . such that it does not add to flood peak. By the same token it would have some additional water flowing into the rivers during the drier periods, thus, increasing the low water flow. . . . By the same token on all of the sites to the extent that they slow the flow of water off of the site they will also accumulate sediment and thus trap sediment and hold nutrients for use in those wetlands systems later in the season as well.” Id., at 95–96.

In addition, in assessing the hydrology prong of the three-part wetlands test, see supra, at 3–4, the District Court made extensive findings regarding water tables and drainage on the parcels at issue. In applying the Corps’ jurisdictional regulations, the District Court found that each of the wetlands bore surface water connections to tributaries of navigable-in-fact waters.

Much the same evidence should permit the establishment of a significant nexus with navigable-in-fact waters, particularly if supplemented by further evidence about the significance of the tributaries to which the wetlands are connected. The Court of Appeals, however, though recognizing that under SWANCC such a nexus was required for jurisdiction, held that a significant nexus “can be satisfied by the presence of a hydrologic connection.” 376 F. 3d, at 639. Absent some measure of the significance of the connection for downstream water quality, this standard was too uncertain. Under the analysis described earlier, supra, at 22–23, 25, mere hydrologic connection should not suffice in all cases; the connection may be too insubstantial for the hydrologic linkage to establish the required nexus with navigable waters as traditionally understood. In my view this case should be remanded so that the District Court may reconsider the evidence in light of the appropriate standard. See, e.g., Pullman-Standard v. Swint, 456 U. S. 273, 291 (1982) (“When an appellate court discerns that a district court has failed to make a finding because of an erroneous view of the law, the usual rule is that there should be a remand for further proceedings to permit the trial court to make the missing findings”).

Carabell

In Carabell, No. 04–1384, the record also contains evidence bearing on the jurisdictional inquiry. The Corps noted in deciding the administrative appeal that “[b]esides the effects on wildlife habitat and water quality, the [district office] also noted that the project would have a major, long-term detrimental effect on wetlands, flood retention, recreation and conservation and overall ecology,” App. 218a. Similarly, in the district office’s permit evaluation, Corps officers observed:

“The proposed work would destroy/adversely impact an area that retains rainfall and forest nutrients and would replace it with a new source area for runoff pollutants. Pollutants from this area may include lawn fertilizers, herbicides, pesticides, road salt, oil, and grease. These pollutants would then runoff directly into the waterway. . . . Overall, the operation and use of the proposed activity would have a major, long term, negative impact on water quality. The cumulative impacts of numerous such projects would be major and negative as the few remaining wetlands in the area are developed.” Id., at 97a–98a.

The Corps’ evaluation further noted that by “eliminat[ing] the potential ability of the wetland to act as a sediment catch basin,” the proposed project “would contribute to increased runoff and accretion . . . along the drain and further downstream in Auvase Creek.” Id., at 98a. And it observed that increased runoff from the site would likely cause downstream areas to “see an increase in possible flooding magnitude and frequency.” Id., at 99a.

The conditional language in these assessments—“potential ability,” “possible flooding”—could suggest an undue degree of speculation, and a reviewing court must identify substantial evidence supporting the Corps’ claims, see 5 U. S. C. §706(2)(E). Nevertheless, the record does show that factors relevant to the jurisdictional inquiry have already been noted and considered. As in Rapanos, though, the record gives little indication of the quantity and regularity of flow in the adjacent tributaries—a consideration that may be important in assessing the nexus. Also, as in Rapanos, the legal standard applied to the facts was imprecise.

The Court of Appeals, considering the Carabell case after its Rapanos decision, framed the inquiry in terms of whether hydrologic connection is required to establish a significant nexus. The court held that it is not, and that much of its holding is correct. Given the role wetlands play in pollutant filtering, flood control, and runoff storage, it may well be the absence of hydrologic connection (in the sense of interchange of waters) that shows the wetlands’ significance for the aquatic system. In the administrative decision under review, however, the Corps based its jurisdiction solely on the wetlands’ adjacency to the ditch opposite the berm on the property’s edge. As explained earlier, mere adjacency to a tributary of this sort is insufficient; a similar ditch could just as well be located many miles from any navigable-in-fact water and carry only insubstantial flow towards it. A more specific inquiry, based on the significant nexus standard, is therefore necessary. Thus, a remand is again required to permit application of the appropriate legal standard. See, e.g., INS v. Orlando Ventura, 537 U. S. 12, 16 (2002) (per curiam) (“Generally speaking, a court of appeals should remand a case to an agency for decision of a matter that statutes place primarily in agency hands”).

* * *

In these consolidated cases I would vacate the judgments of the Court of Appeals and remand for consideration whether the specific wetlands at issue possess a significant nexus with navigable waters.

Justice Stevens, with whom Justice Souter, Justice Ginsburg, and Justice Breyer join, dissenting.

In 1972, Congress decided to “restore and maintain the chemical, physical, and biological integrity of the Nation’s waters” by passing what we now call the Clean Water Act. 86 Stat. 816, as amended, 33 U. S. C. §1251 et seq . The costs of achieving the Herculean goal of ending water pollution by 1985, see §1251(a), persuaded President Nixon to veto its enactment, but both Houses of Congress voted to override that veto by overwhelming margins. To achieve its goal, Congress prohibited “the discharge of any pollutant”—defined to include “any addition of any pollutant to navigable waters from any point source”—without a permit issued by the Army Corps of Engineers (Army Corps or Corps) or the Environmental Protection Agency (EPA). §§1311(a), 1362(12)(A). Congress further defined “navigable waters” to mean “the waters of the United States.” §1362(7).

The narrow question presented in No. 04–1034 is whether wetlands adjacent to tributaries of traditionally navigable waters are “waters of the United States” subject to the jurisdiction of the Army Corps; the question in No. 04–1384 is whether a manmade berm separating a wetland from the adjacent tributary makes a difference. The broader question is whether regulations that have protected the quality of our waters for decades, that were implicitly approved by Congress, and that have been repeatedly enforced in case after case, must now be revised in light of the creative criticisms voiced by the plurality and Justice Kennedy today. Rejecting more than 30 years of practice by the Army Corps, the plurality disregards the nature of the congressional delegation to the agency and the technical and complex character of the issues at stake. Justice Kennedy similarly fails to defer sufficiently to the Corps, though his approach is far more faithful to our precedents and to principles of statutory interpretation than is the plurality’s.

In my view, the proper analysis is straightforward. The Army Corps has determined that wetlands adjacent to tributaries of traditionally navigable waters preserve the quality of our Nation’s waters by, among other things, providing habitat for aquatic animals, keeping excessive sediment and toxic pollutants out of adjacent waters, and reducing downstream flooding by absorbing water at times of high flow. The Corps’ resulting decision to treat these wetlands as encompassed within the term “waters of the United States” is a quintessential example of the Executive’s reasonable interpretation of a statutory provision. See Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837, 842–845 (1984).

Our unanimous decision in United States v. Riverside Bayview Homes, Inc., 474 U. S. 121 (1985), was faithful to our duty to respect the work product of the Legislative and Executive Branches of our Government. Today’s judicial amendment of the Clean Water Act is not.

I

At each of the three sites at issue in No. 04–1034, the petitioners filled large areas of wetlands without permits, despite being on full notice of the Corps’ regulatory requirements. Because the plurality gives short shrift to the facts of this case—as well as to those of No. 04–1384—I shall discuss them at some length.

The facts related to the 230-acre Salzburg site are illustrative. In 1988, John Rapanos asked the Michigan Department of Natural Resources (MDNR) to inspect the site “in order to discuss with him the feasibility of building a shopping center there.” App. to Pet. for Cert. in No. 04– 1034, p. B15. An MDNR inspector informed Rapanos that the land probably included wetlands that were “waters of the United States” and sent him an application for a permit under §404 of the Act.[1] Rapanos then hired a wetland consultant, Dr. Frederick Goff. After Dr. Goff concluded that the land did in fact contain many acres of wetlands, “Rapanos threatened to ‘destroy’ Dr. Goff if he did not destroy the wetland report, and refused to pay Dr. Goff unless and until he complied.” Ibid. In the meantime, without applying for a permit, Rapanos hired construction companies to do $350,000 worth of work clearing the land, filling in low spots, and draining subsurface water. After Rapanos prevented MDNR inspectors from visiting the site, ignored an MDNR cease-and-desist letter, and refused to obey an administrative compliance order issued by the EPA, the matter was referred to the Department of Justice. In the civil case now before us, the District Court found that Rapanos unlawfully filled 22 acres of wetlands.

Rapanos and his wife engaged in similar behavior at the Hines Road and Pine River sites. Without applying for §404 permits, they hired construction companies to per- form extensive clearing and filling activities. They continued these activities even after receiving EPA administrative compliance orders directing them to cease the work immediately. They ultimately spent $158,000 at the 275-acre Hines Road site, filling 17 of its existing 64 acres of wetlands. At the 200-acre Pine River site, they spent $463,000 and filled 15 of its 49 acres of wetlands.

Prior to their destruction, the wetlands at all three sites had surface connections to tributaries of traditionally navigable waters. The Salzburg wetlands connected to a drain that flows into a creek that flows into the navigable Kawkawlin River. The Hines Road wetlands connected to a drain that flows into the navigable Tittabawassee River. And the Pine River wetlands connected with the Pine River, which flows into Lake Huron.

At trial, the Government put on a wetland expert, Dr. Daniel Willard, whom the trial court found “eminently qualified” and “highly credible.” Id., at B7. Dr. Willard testified that the wetlands at these three sites provided ecological functions in terms of “habitat, sediment trapping, nutrient recycling, and flood peak diminution.” 4 Tr. 96 (Apr. 5, 1999).[2] He explained:

“[G]enerally for all of the . . . sites we have a situation in which the flood water attenuation in that water is held on the site in the wetland . . . such that it does not add to flood peak. By the same token it would have some additional water flowing into the rivers during the drier periods, thus, increasing low water flow.

...

“By the same token on all of the sites to the extent that they slow the flow of water of the site they will also accumulate sediment and thus trap sediment and hold nutrients for use in those wetland systems later in the season as well.” Id., at 95–96.

The District Court found that the wetlands at all three sites were covered by the Clean Water Act and that the Rapanoses had violated the Act by destroying them without permits. The Sixth Circuit unanimously affirmed. 376 F. 3d 629 (2004).

The facts of No. 04–1384 are less dramatic. The petitioners in that case own a 20-acre tract of land, of which 16 acres are wetlands, located in Macomb County a mile from Lake St. Clair. These wetlands border a ditch that flows into a drain that flows into a creek that flows into Lake St. Clair. A 4-foot-wide manmade berm separates the wetlands from the ditch; thus water rarely if ever passes from wetlands to ditch or vice versa.

Petitioners applied for a permit to fill most of these wetlands with 57,500 cubic yards of material. They intended to build a 112-unit condominium development on the site. After inspecting the site and considering comments from, among others, the Water Quality Unit of the Macomb County Prosecutor’s Office (which urged the Corps to deny the permit because “[t]he loss of this high quality wetland area would have an unacceptable adverse effect on wildlife, water quality, and conservation of wetlands resources,” App. in No. 04–1384, p. 79a), the Corps denied the permit. Id., at 84a–126a. As summarized in a letter sent to petitioners, reasons for denial included: “

Your parcel is primarily a forested wetland that provides valuable seasonal habitat for aquatic organisms and year round habitat for terrestrial organisms. Additionally, the site provides water storage functions that, if destroyed, could result in an increased risk of erosion and degradation of water quality in the Sutherland-Oemig Drain, Auvase Creek, and Lake St. Clair. The minimization of impacts to these wetlands is important for conservation and the overall ecology of the region. Because the project development area is a forested wetland, the proposed project would destroy the resources in such a manner that they would not soon recover from impacts of the discharges. The extent of impacts in the project area when considered both individually and cumulatively would be unacceptable and contrary to the public interest.” Id., at 127a–128a.

As in No. 04–1034, the unanimous judgment of the District and Circuit Judges was that the Corps has jurisdiction over this wetland because it is adjacent to a tributary of traditionally navigable waters. 391 F. 3d 704 (CA6 2004). The Solicitor General defends both judgments.

II

Our unanimous opinion in Riverside Bayview squarely controls these cases. There, we evaluated the validity of the very same regulations at issue today. These regulations interpret “waters of the United States” to cover all traditionally navigable waters; tributaries of these waters; and wetlands adjacent to traditionally navigable waters or their tributaries. 33 CFR §§328.3(a)(1), (5), and (7) (2005); §§323.2(a)(1), (5), and (7) (1985). Although the particular wetland at issue in Riverside Bayview abutted a navigable creek, we framed the question presented as whether the Clean Water Act “authorizes the Corps to require landowners to obtain permits from the Corps before discharging fill material into wetlands adjacent to navigable bodies of water and their tributaries .” 474 U. S., at 123 (emphasis added).[3]

We held that, pursuant to our decision in Chevron,

“our review is limited to the question whether it is reasonable, in light of the language, policies, and legislative history of the Act for the Corps to exercise jurisdiction over wetlands adjacent to but not regularly flooded by rivers, streams, and other hydrographic features more conventionally identifiable as ‘waters.’” 474 U. S., at 131.

Applying this standard, we held that the Corps’ decision to interpret “waters of the United States” as encompassing such wetlands was permissible. We recognized the practical difficulties in drawing clean lines between land and water, id., at 132, and deferred to the Corps’ judgment that treating adjacent wetlands as “waters” would advance the “congressional concern for protection of water quality and aquatic ecosystems,” id., at 133.

Contrary to the plurality’s revisionist reading today, ante, at 21–24, 28–29, Riverside Bayview nowhere implied that our approval of “adjacent” wetlands was contingent upon an understanding that “adjacent” means having a “continuous surface connection” between the wetland and its neighboring creek, ante, at 23. Instead, we acknowledged that the Corps defined “adjacent” as including wetlands “‘that form the border of or are in reasonable proximity to other waters’” and found that the Corps reasonably concluded that adjacent wetlands are part of the waters of the United States. 474 U. S., at 134 (quoting 42 Fed. Reg. 37128 (1977)). Indeed, we explicitly acknowledged that the Corps’ jurisdictional determination was reasonable even though

“not every adjacent wetland is of great importance to the environment of adjoining bodies of water. . . . If it is reasonable for the Corps to conclude that in the majority of cases, adjacent wetlands have significant effects on water quality and the ecosystem, its definition can stand. That the definition may include some wetlands that are not significantly intertwined with the ecosystem of adjacent waterways is of little moment, for where it appears that a wetland covered by the Corps’ definition is in fact lacking in importance to the aquatic environment . . . the Corps may always allow development of the wetland for other uses simply by issuing a permit.” 474 U. S., at 135, n. 9.

In closing, we emphasized that the scope of the Corps’ asserted jurisdiction over wetlands had been specifically brought to Congress’ attention in 1977, that Congress had rejected an amendment that would have narrowed that jurisdiction, and that even proponents of the amendment would not have removed wetlands altogether from the definition of “waters of the United States.” Id., at 135– 139.

Disregarding the importance of Riverside Bayview, the plurality relies heavily on the Court’s subsequent opinion in Solid Waste Agency of Northern Cook Cty. v. Army Corps of Engineers, 531 U. S. 159 (2001) (SWANCC) . In stark contrast to Riverside Bayview, however, SWANCC had nothing to say about wetlands, let alone about wetlands adjacent to traditionally navigable waters or their tributaries. Instead, SWANCC dealt with a question specifically reserved by Riverside Bayview, see n. 3, supra, namely, the Corps’ jurisdiction over isolated waters— “‘waters that are not part of a tributary system to interstate waters or to navigable waters of the United States, the degradation or destruction of which could affect interstate commerce.’” 531 U. S., at 168–169 (quoting 33 CFR §323.2(a)(5) (1978); emphasis added); see also 531 U. S., at 163 (citing 33 CFR §328.2(a)(3) (1999), which is the later regulatory equivalent to §323.2(a)(5) (1978)). At issue in SWANCC was “an abandoned sand and gravel pit . . . which provide[d] habitat for migratory birds” and contained a few pools of “nonnavigable, isolated, intrastate waters.” 531 U. S., at 162, 166. The Corps had asserted jurisdiction over the gravel pit under its 1986 Migratory Bird Rule, which treated isolated waters as within its jurisdiction if migratory birds depended upon these waters. The Court rejected this jurisdictional basis since these isolated pools, unlike the wetlands at issue in Riverside Bayview, had no “significant nexus” to traditionally navigable waters. 531 U. S., at 167. In the process, the Court distinguished Riverside Bayview ’s reliance on Congress’ decision to leave the Corps’ regulations alone when it amended the Act in 1977, since “‘[i]n both Chambers, debate on the proposals to narrow the definition of navigable waters centered largely on the issue of wetlands preservation’” rather than on the Corps’ jurisdiction over truly isolated waters. 531 U. S., at 170 (quoting 474 U. S., at 136).[4]

Unlike SWANCC and like Riverside Bayview, the cases before us today concern wetlands that are adjacent to “navigable bodies of water [or] their tributaries,” 474 U. S., at 123. Specifically, these wetlands abut tributaries of traditionally navigable waters. As we recognized in Riverside Bayview, the Corps has concluded that such wetlands play important roles in maintaining the quality of their adjacent waters, see id., at 134–135, and consequently in the waters downstream. Among other things, wetlands can offer “nesting, spawning, rearing and resting sites for aquatic or land species”; “serve as valuable storage areas for storm and flood waters”; and provide “significant water purification functions.” 33 CFR §320.4(b)(2) (2005); 474 U. S., at 134–135. These values are hardly “ independent ” ecological considerations as the plurality would have it, ante, at 23—instead, they are integral to the “chemical, physical, and biological integrity of the Nation’s waters,” 33 U. S. C. §1251(a). Given that wetlands serve these important water quality roles and given the ambiguity inherent in the phrase “waters of the United States,” the Corps has reasonably interpreted its jurisdiction to cover non-isolated wetlands. See 474 U. S., at 131–135.[5]

This conclusion is further confirmed by Congress’ deliberate acquiescence in the Corps’ regulations in 1977. Id., at 136. Both Chambers conducted extensive debates about the Corps’ regulatory jurisdiction over wetlands, rejected efforts to limit this jurisdiction, and appropriated funds for a “‘National Wetlands Inventory’” to help the States “‘in the development and operation of programs under this Act.’” Id., at 135–139 (quoting 33 U. S. C. §1288(i)(2)). We found these facts significant in Riverside Bayview, see 474 U. S., at 135–139, as we acknowledged in SWANCC . See 531 U. S., at 170–171 (noting that “ [b]eyond Congress’ desire to regulate wetlands adjacent to ‘navigable waters,’ respondents point us to no persuasive evidence” of congressional acquiescence (emphasis added)).

The Corps’ exercise of jurisdiction is reasonable even though not every wetland adjacent to a traditionally navigable water or its tributary will perform all (or perhaps any) of the water quality functions generally associated with wetlands. Riverside Bayview made clear that jurisdiction does not depend on a wetland-by-wetland inquiry. 474 U. S., at 135, n. 9. Instead, it is enough that wetlands adjacent to tributaries generally have a significant nexus to the watershed’s water quality. If a particular wetland is “not significantly intertwined with the ecosystem of adjacent waterways,” then the Corps may allow its development “simply by issuing a permit.” Ibid. [6] Accordingly, for purposes of the Corps’ jurisdiction it is of no significance that the wetlands in No. 04–1034 serve flood control and sediment sink functions, but may not do much to trap other pollutants, supra, at 4–5, and n. 2, or that the wetland in No. 04–1328 keeps excess water from Lake St. Clair but may not trap sediment, see supra, at 5–6.

Seemingly alarmed by the costs involved, the plurality shies away from Riverside Bayview ’s recognition that jurisdiction is not a case-by-case affair. I do not agree with the plurality’s assumption that the costs of preserving wetlands are unduly high. It is true that the cost of §404 permits are high for those who must obtain them[7]— but these costs amount to only a small fraction of 1% of the $760 billion spent each year on private and public construction and development activity. Sunding & Zilberman 80. More significant than the plurality’s exaggerated concern about costs, however, is the fact that its omission of any discussion of the benefits that the regulations at issue have produced sheds a revelatory light on the quality (and indeed the impartiality) of its cost-benefit analysis.[8] The importance of wetlands for water quality is hard to overstate. See, e.g., U. S. Congress, Office of Technology Assessment, Wetlands: Their Use and Regulation, OTA–206, pp. 43–61 (Mar. 1984), http://govinfo.library.unt.edu/ota/Ota_4/DATA/1984/8433.PDF (hereinafter OTA) (describing wetlands’ role in floodpeak reduction, shoreline protection, ground water recharge, trapping of suspended sediment, filtering of toxic pollutants, and protection of fish and wildlife). See also ante, at 20 (Kennedy, J., concurring in judgment). Unsurprisingly, the Corps’ approach has the overwhelming endorsement of numerous amici curiae, including 33 States and the county in which the property in No. 04–1384 is located.

In final analysis, however, concerns about the appropriateness of the Corps’ 30-year implementation of the Clean Water Act should be addressed to Congress or the Corps rather than to the Judiciary. Whether the benefits of particular conservation measures outweigh their costs is a classic question of public policy that should not be answered by appointed judges. The fact that large investments are required to finance large developments merely means that those who are most adversely affected by the Corps’ permitting decisions are persons who have the ability to communicate effectively with their representatives. Unless and until they succeed in convincing Congress (or the Corps) that clean water is less important today than it was in the 1970’s, we continue to owe deference to regulations satisfying the “evident breadth of congressional concern for protection of water quality and aquatic ecosystems” that all of the Justices on the Court in 1985 recognized in Riverside Bayview, 474 U. S., at 133.

III

Even setting aside the plurality’s dramatic departure from our reasoning and holding in Riverside Bayview, its creative opinion is utterly unpersuasive. The plurality imposes two novel conditions on the exercise of the Corps’ jurisdiction that can only muddy the jurisdictional waters. As Justice Kennedy observes, “these limitations . . . are without support in the language and purposes of the Act or in our cases interpreting it.” Ante, at 11 (opinion concurring in judgment). The impropriety of crafting these new conditions is highlighted by the fact that no party or amicus has suggested either of them.[9]

First, ignoring the importance of preserving jurisdiction over water beds that are periodically dry, the plurality imposes a requirement that only tributaries with the “relatively permanent” presence of water fall within the Corps’ jurisdiction. Ante, at 13–14. Under the plurality’s view, then, the Corps can regulate polluters who dump dredge into a stream that flows year round but may not be able to regulate polluters who dump into a neighboring stream that flows for only 290 days of the year—even if the dredge in this second stream would have the same effect on downstream waters as the dredge in the year-round one. Ante, at 14, n. 5.[10]

To find this arbitrary distinction compelled by the statute, the plurality cites a dictionary for a proposition that it does not contain. The dictionary treats “streams” as “waters” but has nothing to say about whether streams must contain water year round to qualify as “streams.” Ante, at 13–14, and n. 6 (citing Webster’s New International Dictionary 2493 (2d ed. 1954) (hereinafter Webster’s Second), as defining stream as a “‘current or course of water or other fluid, flowing on the earth’”). From this, the plurality somehow deduces that streams can never be intermittent or ephemeral (i.e., flowing for only part of the year). Ante, at 13–15, and nn. 5–6. But common sense and common usage demonstrate that intermittent streams, like perennial streams, are still streams.[11] See, e.g., U. S. Dept. of Interior, U. S. Geological Survey, Topographic Map Symbols 3 (2005), http://erg.usgs.gov/isb/pubs/booklets/symbols/(identifying symbols for “[p]erennial stream” and “[i]ntermittent stream,” as well as for “[p]erennial river” and “[i]ntermittent river”). This was true well before the passage of the Act in 1972. E.g., Webster’s Third New International Dictionary 1180 (1961) (hereinafter Webster’s Third) (linking “intermittent” with “stream”). Indeed, we ourselves have used the term “intermittent stream” as far back as 1932. Harrisonville v. W. S. Dickey Clay Mfg. Co., 289 U. S. 334, 335 (1933). Needless to say, Justice Brandeis’ use of the term in a unanimous opinion should not be dismissed as merely a “useful oxymor[on],” ante, at 15, n. 6 (plurality opinion).

The plurality attempts to bolster its arbitrary jurisdictional line by citing two tangential statutory provisions and two inapplicable canons of construction. None comes close to showing that Congress directly spoke to whether “waters” requires the relatively permanent presence of water.

The first provision relied on by the plurality—the definition of “point source” in 33 U. S. C. §1362(14)—has no conceivable bearing on whether permanent tributaries should be treated differently from intermittent ones, since “pipe[s], ditch[es], channel[s], tunnel[s], conduit[s], [and] well[s]” can all hold water permanently as well as inter-mittently.[12] The second provision is §1251(b), which announces a congressional policy to “recognize, preserve, and protect the primary responsibilities and rights of States” to prevent pollution, to plan development, and to consult with the EPA. Under statutory additions made in 1977 when Congress considered and declined to alter the Corps’ interpretation of its broad regulatory jurisdiction, the States may run their own §404 programs. §§1344(g)–(h). As modified, §1251(b) specifically recognizes this role for the States as part of their primary responsibility for preventing water pollution. Even focusing only on the Act as it stood between 1972 and 1977, but see International Paper Co. v. Ouellette, 479 U. S. 481, 489–490 (1987) (interpreting §1251(b) in light of the 1977 additions), broad exercise of jurisdiction by the Corps still left the States with ample rights and responsibilities. See S. D. Warren Co. v. Maine Bd. of Environmental Protection, 547 U. S. __, __ (2006) (slip op., at 14–15). States had the power to impose tougher water pollution standards than required by the Act, §1370, and to prevent the Corps and the EPA from issuing permits, §1341(a)(1)—not to mention nearly exclusive responsibility for containing pollution from nonpoint sources.

The two canons of construction relied on by the plurality similarly fail to overcome the deference owed to the Corps. First, the plurality claims that concerns about intruding on state power to regulate land use compel the conclusion that the phrase “waters of the United States” does not cover intermittent streams. As we have recognized, however, Congress found it “‘essential that discharge of pollutants be controlled at the source,’” Riverside Bayview, 474 U. S., at 133 (quoting S. Rep. No. 92–414, p. 77 (1972)), and the Corps can define “waters” broadly to accomplish this aim. Second, the plurality suggests that the canon of constitutional avoidance applies because the Corps’ approach might exceed the limits of our Commerce Clause authority. Setting aside whether such a concern was proper in SWANCC, 531 U. S., at 173; but see id., at 192–196 (Stevens, J., dissenting), it is plainly not warranted here. The wetlands in these cases are not “isolated” but instead are adjacent to tributaries of traditionally navigable waters and play important roles in the watershed, such as keeping water out of the tributaries or absorbing water from the tributaries. “There is no constitutional reason why Congress cannot, under the commerce power, treat the watersheds as a key to flood control on navigable streams and their tributaries.” Oklahoma ex rel. Phillips v. Guy F. Atkinson Co., 313 U. S. 508, 525 (1941).

Most importantly, the plurality disregards the fundamental significance of the Clean Water Act. As then-Justice Rehnquist explained when writing for the Court in 1981, the Act was “not merely another law” but rather was “viewed by Congress as a ‘total restructuring’ and ‘complete rewriting’ of the existing water pollution legislation.” Milwaukee v. Illinois, 451 U. S. 304, 317. “Congress’ intent in enacting the [Act] was clearly to establish an all-encompassing program of water pollution regulation,” and “the most casual perusal of the legislative history demonstrates that . . . views on the comprehensive nature of the legislation were practically universal.” Id., at 318, and n. 12; see also 531 U. S., at 177–181 (Stevens, J., dissenting). The Corps has concluded that it must regulate pollutants at the time they enter ditches or streams with ordinary high-water marks—whether perennial, intermittent, or ephemeral—in order to properly control water pollution. 65 Fed. Reg. 12823 (2000). Because there is ambiguity in the phrase “waters of the United States” and because interpreting it broadly to cover such ditches and streams advances the purpose of the Act, the Corps’ approach should command our deference . Intermittent streams can carry pollutants just as perennial streams can, and their regulation may prove as important for flood control purposes. The inclusion of all identifiable tributaries that ultimately drain into large bodies of water within the mantle of federal protection is surely wise.

The plurality’s second statutory invention is as arbitrary as its first. Trivializing the significance of changing conditions in wetlands environments, the plurality imposes a separate requirement that “the wetland has a continuous surface connection” with its abutting waterway such that it is “difficult to determine where the ‘water’ ends and the ‘wetland’ begins.” Ante, at 24. An “intermittent, physically remote hydrologic connection” between the wetland and other waters is not enough. Ibid. Under this view, wetlands that border traditionally navigable waters or their tributaries and perform the essential function of soaking up overflow waters during hurricane season— thus reducing flooding downstream—can be filled in by developers with impunity, as long as the wetlands lack a surface connection with the adjacent waterway the rest of the year.

The plurality begins reasonably enough by recognizing that the Corps may appropriately regulate all wetlands “‘adjacent to’” other waters. Ante, at 21. This recognition is wise, since the statutory text clearly accepts this standard. Title 33 U. S. C. §1344(g)(1), added in 1977, includes “adjacent wetlands” in its description of “waters” and thus “expressly stated that the term ‘waters’ included adjacent wetlands.” Riverside Bayview, 474 U. S., at 138. While this may not “conclusively determine the construction to be placed on the use of the term ‘waters’ elsewhere in the Act . . ., in light of the fact that the various provisions of the Act should be read in pari materia, it does at least suggest strongly that the term ‘waters’ as used in the Act does not necessarily exclude ‘wetlands.’” Id., at 138, n. 11.

The plurality goes on, however, to define “‘adjacent to’” as meaning “with a continuous surface connection to” other water. Ante, at 21–24. It is unclear how the plurality reached this conclusion, though it plainly neglected to consult a dictionary. Even its preferred Webster’s Second defines the term as “[l]ying near, close, or contiguous; neighboring; bordering on” and acknowledges that “[o]bjects are Adjacent when they lie close to each other, but not necessarily in actual contact. ” Webster’s Second 32 (emphasis added); see also Webster’s Third 26. In any event, the proper question is not how the plurality would define “adjacent,” but whether the Corps’ definition is reasonable.

The Corps defines “adjacent” as “bordering, contiguous, or neighboring,” and specifies that “[w]etlands separated from other waters of the United States by man-made dikes or barriers, natural river berms, beach dunes and the like are ‘adjacent wetlands.’” 33 CFR §328.3(c) (2005). This definition is plainly reasonable, both on its face and in terms of the purposes of the Act. While wetlands that are physically separated from other waters may perform less valuable functions, this is a matter for the Corps to evaluate in its permitting decisions. We made this clear in Riverside Bayview, 474 U. S., at 135, n. 9—which did not impose the plurality’s new requirement despite an absence of evidence that the wetland at issue had the sort of continuous surface connection required by the plurality today. See supra, at 7; see also ante, at 15–17 (Kennedy, J., concurring in judgment) (observing that the plurality’s requirement is inconsistent with Riverside Bayview). And as the facts of No. 04–1384 demonstrate, wetland separated by a berm from adjacent tributaries may still prove important to downstream water quality. Moreover, Congress was on notice of the Corps’ definition of “adjacent” when it amended the Act in 1977 and added 33 U. S. C. §1344(g)(1). See 42 Fed. Reg. 37129 (1977).