Full opinion text
Reversed, vacated and remanded by published opinion. Judge GREGORY wrote the opinion, in which Judge SHEDD joined. Judge MICHAEL wrote a separate opinion dissenting in part and concurring in part. GREGORY, Circuit Judge: OPINION This appeal concerns a challenge by Plaintiffs-Appellees Ohio Valley Environmental Coalition, the Coal River Mountain Watch, and the West Virginia Highlands Conservancy (hereinafter referred to collectively as “OVEC”) to the U.S. Army Corps of Engineers (“Corps”) issuance of four permits allowing the filling of West Virginia stream waters in conjunction with area surface coal mining operations. Granting judgment for OVEC, the district court rescinded the permits as violations of the Clean Water Act (“CWA”), 33 U.S.C. § 1251 et seq. (2000), the National Environmental Policy Act (“NEPA”), 42 U.S.C. § 4321 et seq. (2000), and the Administrative Procedure Act (“APA”), 5 U.S.C. § 701 et seq. (2000). The court also enjoined all activity under those permits and remanded to the Corps for further proceedings consistent with its order. Separately, in an order dated June 13, 2007, the district court provided declaratory relief to OVEC, holding that the stream segments linking the permitted fills to downstream sediment treatment ponds were “waters of the United States” and that the Corps lacked authority under the CWA to permit discharge from the fills into the stream segments. The Corps now appeals these two orders. For the reasons set forth below, we reverse and vacate the district court’s opinion and order of March 23, 2007, and vacate the district court’s injunction. We also reverse the district court’s June 13, 2007, grant of declaratory relief and we remand for further proceedings consistent with this opinion. I. The mountaintop removal method of surface coal mining, pioneered in West Virginia, involves the blasting of the soil and rock atop a mountain to expose coal deposits below. While mining operations are ongoing, the overburden is hauled or pushed into adjacent valleys. This excavated overburden is known as “spoil.” Once the coal has been extracted, efforts are made to re-contour the mountaintop by replacing the removed overburden, but stability concerns limit the amount of spoil that can be returned to the area. In its natural state, the spoil material is heavily compacted; once excavated, however, the loosening of the rock and soil and incorporation of air causes significant swelling. As a result, large quantities of the blasted material cannot be replaced, and this excess spoil (“overburden”) remains in the valley, creating a “valley fill” that buries intermittent and perennial streams in the process. Water that collects in the fill must be moved out to ensure the fill’s continued stability. Thus, an underdrain system is constructed by placing large boulders up to and above the ordinary high-water mark of the stream. The collected water is then channeled into a treatment pond, where sediment from the runoff is allowed to settle. Sediment ponds usually are constructed in existing streambeds, using earth and rock to create an embankment. After sediments have settled out of the fill runoff, the treated water is discharged from the sediment pond back into existing streams. When practicable, a sediment pond will be constructed in the streambed immediately adjacent to the end (or “toe”) of the fill. But, because West Virginia’s steep, mountainous topography often prevents this kind of positioning, a short stream segment is frequently used to move runoff from the fill downstream to the sediment pond. Once a valley fill is stabilized, the embankments of the sediment pond are removed, and the ponds and the stream segments are restored to their pre-project condition. Much of the impact of a valley fill project is felt by headwater streams. Head-water streams are small streams that form the origin of larger streams or rivers, and may be intermittent or ephemeral. Intermittent streams receive their flow from both surface runoff and groundwater discharge, while ephemeral streams rely on major rain or snow events for their flow. The precise role of headwater streams in overall watershed ecology is a matter of some debate in this litigation, as we discuss more below, but all parties agree that these streams perform important ecological functions. OVEC initiated this challenge in September 2005 in the United States District Court for the Southern District of West Virginia, shortly after the Corps issued an individual valley fill permit and accompanying Combined Decision Document (“CDD”) to the Araeoma Coal Company for the Camp Branch Surface Mine project (“Camp Branch”) under its CWA § 404 authority. As the Corps issued subsequent § 404 permits to West Virginia mining operations, the district court allowed OVEC to amend its complaint several times to include the newly issued permits. In addition to the Camp Branch permit, OVEC’s Third Supplemental Complaint raised challenges to the individual § 404 permits issued to the Elk Run Coal Company for the Black Castle Mine (“Black Castle”), and to Alex Energy, Inc., for the Republic No. 1 and Republic No. 2 Surface Mines (“Republic No. 1” and “Republic No. 2”). The Republic No. 1 challenge was ultimately dismissed on ripeness grounds, but a challenge raised in a separate complaint to the individual permit issued to Independence Coal Company for the Laxare East Surface Mine (“Laxare East”) was consolidated with this proceeding. Each of the affected companies intervened as defendants in the action, as did the West Virginia Coal Association. All together, the four challenged permits authorize the creation of 23 valley fills and 23 sediment ponds, and they impact 68,841 linear feet of intermittent and ephemeral streams, or just over 13 miles. For each of the four permits, the Corps prepared Environmental Assessments that concluded that the permitted activity would not result in significant environmental impacts given planned mitigation measures. On that basis, the Corps issued a “Finding of No Significant Impact” for all four permits. OVEC’s Third Supplemental Complaint charged that the Corps’ issuance of the § 404 fill permits for these mining projects violated both substantive and procedural provisions of the CWA and NEPA, and were “arbitrary, capricious, and an abuse of discretion” under the APA. According to OVEC, the Corps was required under NEPA to prepare an Environmental Impact Statement for each of the projects before issuing a permit, given the significant individual and cumulative adverse effects the projects would have on water quality, aquatic and terrestrial ecosystems and habitats, species survival and diversity, crucial stream functions, forests, and the aesthetic value of the destroyed mountains. Similarly, OVEC claimed that the Corps failed to properly determine the adverse individual and cumulative impacts to the affected aquatic ecosystems in accordance with the CWA and the Corps’ CWA Guidelines. Trial in the case was originally scheduled for June 20, 2006, but on June 16, on the Corps’ motion, the district court remanded the permits to the Corps and stayed the proceedings. Almost a month later, the Corps reissued the permits, but this time with a supplemented administrative record that incorporated new comments from the public and the parties, including the reports prepared by OVEC’s proposed expert witnesses. The district court lifted its stay on July 26, and a six-day bench trial was held in October 2006. The district court granted judgment in favor of the plaintiffs on March 23, 2007, rescinding the permits, enjoining the Corps and Intervenors from taking any action under those permits, and remanding the permits to the Corps for further proceedings consistent with the court’s order. The district court found, inter alia, that the probable impacts of the valley fills would be significant and adverse under both the CWA and NEPA; that the mitigation plans for each permit were not sufficient to compensate for those adverse impacts; that, in each permit, the Corps improperly limited its scope of review under NEPA to look only at the impact on jurisdictional waters rather than the broader impact of the entire valley fill project; and, finally, that the Corps inadequately evaluated the cumulative impacts of the projects. On June 13, 2007, the district court granted summary judgment to OVEC on a separate claim under which the plaintiffs sought a declaratory judgment that the stream segments running from the valley fill toes to the sediment pond embankments are “waters of the United States,” and that the Corps thus did not have authority to permit the discharge of pollutants into these segments with a CWA § 404 permit. According to the district court, mining operators who wished to discharge runoff from the fill into a stream segment needed to obtain a CWA § 402 permit from the EPA or proper state authority. On September 13, 2007, the district court granted, pursuant to Federal Rule of Civil Procedure 54(b), the Interve-nors’ motion for entry of final judgment on the June 13 order. The Corps and Inter-venors filed timely notices of appeal from both the March 23 and June 13 orders. On appeal, the Corps contends that it is entitled to deference on its determination about the scope of its NEPA analysis and that its findings on individual and cumulative impacts and mitigation were not arbitrary or capricious. The agency further argues that its interpretation of its CWA regulations — treating stream segments and sediment ponds as part of a unitary waste treatment system and thus excepting them from separate CWA § 402 permitting- — was entitled to deference. Inter-venors have raised these same challenges to the district court’s ruling, but also argue that OVEC’s stream segment claim was barred in the first place under principles of res judicata. OVEC has also filed two motions for judicial notice, asking this Court to take notice of five new permits that the Corps has issued since the district court’s orders were entered. II. We review de novo a district court’s findings on an administrative record. See Crutchfield v. County of Hanover, 325 F.3d 211, 217 (4th Cir.2003). This de novo standard applies to questions of both law and fact. See id. Both NEPA and CWA claims are subject to judicial review under the APA, 5 U.S.C. § 706 (2006). For all agency actions, a reviewing court must set aside the action if it is found to be “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A) (2006); Citizens To Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 413-14, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971). III. A complex statutory framework under-girds the regulation of valley fills and associated sediment ponds, and it is this framework that provides the foundation for our opinion. Thus, we begin with a brief overview of the relationship of the four statutes that affect the scope of the Corps’ authority to issue fill permits in connection with mountaintop coal mining operations: the Surface Mining Control and Reclamation Act of 1977 (“SMCRA”), 30 U.S.C. § 1201 et seq. (2000), the CWA, NEPA, and the APA. A. Congress passed SMCRA in 1977 to “establish a nationwide program to protect society and the environment from the adverse effects of surface coal mining operations.” 30 U.S.C. § 1202(a) (2000). Congress also recognized a need, however, to “strike a balance between protection of the environment and agricultural productivity and the Nation’s need for coal as an essential source of energy.” 30 U.S.C. § 1202(f) (2000). In striking this balance, SMCRA utilizes a “cooperative federalism” approach, allocating responsibility for the regulation of surface coal mining among both state and federal agencies. Bragg v. W. Va. Coal Ass’n, 248 F.3d 275, 288 (4th Cir.2001). Under SMCRA, states have “exclusive jurisdiction over the regulation of surface coal mining and reclamation operations” on non-Federal lands, so long as their regulatory program has been approved by the Secretary of the Interior as satisfying the Act’s minimum requirements. 30 U.S.C. § 1253 (2000). Once a state’s SMCRA program has been approved, anyone wishing to engage in surface coal mining operations within the state must first obtain a permit from the state’s regulatory authority. 30 U.S.C. § 1256(a) (2000). In West Virginia, the federally approved regulatory authority is the Department of Environmental Protection (‘WVDEP”). Regulation of the disposal of excess spoil material from surface coal mining operations is within SMCRA’s purview. As part of its environmental protection performance standards, SMCRA requires that all excess spoil material from surface mining operations be disposed of “in a controlled manner ... and in such a way to assure mass stability and to prevent mass movement.” 30 U.S.C. § 1265(b)(22)(A) (2000). The Act clearly contemplates that valley fills will be used in the disposal process. See 30 U.S.C. § 1265(b)(22)(D) (2000) (requiring that, where the disposal area contains “springs, natural water courses, or wet weather seeps ... lateral drains [must be] constructed from the wet areas to the main under-drains in such a manner that filtration of the water into the spoil pile will be prevented.”); Kentuckians for the Commonwealth, Inc. v. Rivenburgh, 317 F.3d 425, 443 (4th Cir.2003) (“[I]t is beyond dispute that SMCRA recognizes the possibility of placing excess spoil material in waters of the United States.... ”). B. An SMCRA permit by itself, however, does not suffice to allow a mine operator to construct a valley fill in conjunction with its mountaintop removal activities. Mining companies must also obtain permits certifying their project’s compliance with the CWA. The CWA aims to “restore and maintain the chemical, physical, and biological integrity of the Nation’s waters” by eliminating “the discharge of pollutants into the navigable waters.” 33 U.S.C. § 1251(a)(2000). In the surface mining context, three sections of the CWA are relevant to the permitting process. First, a mine operator applying for a federal permit under the CWA must comply with CWA § 401, 33 U.S.C. § 1341 (2000), by providing the federal permitting agency with a certification from the proper state authority — in this case, WVDEP— stating that any discharge from the mine site will comply with all applicable water quality standards. Next, the mine operator must obtain a National Pollutant Discharge Elimination System (“NPDES”) permit pursuant to CWA § 402, 33 U.S.C. § 1342 (2000), if their project involves the discharge of a pollutant from a point source within the mining operation into navigable waters. 33 U.S.C. §§ 1342, 1362(12) (2000). The CWA defines “navigable waters” as “the waters of the United States, including the territorial seas.” 33 U.S.C. § 1362(7) (2000). The release of treated waters from sediment ponds back into a stream, for example, require a CWA § 402 NPDES permit. States wishing to administer their own NPDES program must be approved by the Environmental Protection Agency (“EPA”) before they can begin issuing § 402 permits. 33 U.S.C. § 1342(c) (2000). West Virginia has had an EPA-approved § 402 program since 1982. See Approval of West Virginia’s NPDES Program, 47 Fed. Reg. 22, 363 (May 24,1982). Finally, and most importantly for the purposes of this litigation, surface mining projects that intend to dispose of excess spoil from their mining operations in jurisdictional waters must obtain a CWA § 404, 33 U.S.C. § 1344 (2000), permit from the Corps. Section 404 permits allow “the discharge of dredged or fill material into the navigable waters at specified disposal sites.” 33 U.S.C.A. § 1344(a) (2008). The Corps uses § 404 permits to authorize the fill activity itself, as well as the construction of downstream sediment ponds. In issuing § 404 permits, the Corps follows the § 404(b)(1) Guidelines (“CWA Guidelines”) promulgated by the Environmental Protection Agency (“EPA”) pursuant to 33 U.S.C. § 1344(b)(1) (2008), and incorporated by the Corps into its own regulations. See 40 C.F.R. pt. 230 (2008); 33 C.F.R. § 320.2(f) (2008). The Guidelines prohibit discharges that “will cause or contribute to significant degradation of the waters of the United States.” 40 C.F.R. § 230.10(c) (2008). A discharge contributes to significant degradation if it has “[significantly adverse effects” on human health or welfare, on aquatic life and other wildlife dependent on aquatic ecosystems, on aquatic ecosystem diversity, productivity, and stability, or on recreational, aesthetic, and economic values. Id. The Corps’ § 404 permit evaluation process must also include a public interest review component, in which “[t]he benefits which reasonably may be expected to accrue from the proposal must be balanced against its reasonably foreseeable detriments.” 33 C.F.R. § 320.4(a)(1) (2008). The Corps’ decision to issue a permit “should reflect the national concern for both protection and utilization of important resources.” Id. Ultimately, the § 404 permitting process requires extensive review and coordination with numerous federal and state agencies, as well as significant consideration of the public interest. C. Under NEPA, federal agencies must take a “hard look” at the potential environmental consequences of their actions. Robertson v. Methow Valley Citizens Council, 490 U.S. 332, 350, 109 S.Ct. 1835, 104 L.Ed.2d 351 (1989). Because NEPA is a procedural and not a results-driven statute, even agency action with adverse environmental effects can be NEPA-compliant so long as the agency has considered those effects and determined that competing policy values outweigh those costs. Id. NEPA requires only that federal agencies prepare an Environmental Impact Statement for “major Federal actions significantly affecting the quality of the human environment.” 42 U.S.C. § 4332(2)(C) (2000). Significance is determined by evaluating both the context of the action and the intensity, or severity, of the impact. 40 C.F.R. § 1508.27 (2008). Where it is not readily discernible how significant the environmental effects of a proposed action will be, federal agencies may prepare an Environmental Assessment (“EA”). 40 C.F.R. § 1501.4(b) (2008). An EA is a “concise public document ... that serves to ... [bjriefiy provide sufficient evidence and analysis for determining whether to prepare an [EIS] or a finding of no significant impact [(“FONSI”)].” 40 C.F.R. § 1508.9(a)(1) (2008); see also 33 C.F.R. §§ 230.10-230.11 (2008) (explaining the Corps’ requirements for an EA). Even where an EA determines that a proposed action will have a significant environmental impact, an agency may avoid issuing an EIS where it finds that mitigating measures can be taken to reduce the environmental impact of the project below the level of significance. Roanoke River Basin Ass’n v. Hudson, 940 F.2d 58, 62 (4th Cir.1991). In these situations, the agency can issue a “so-called mitigated FONSI.” Spiller v. White, 352 F.3d 235, 241 (5th Cir.2003) (internal quotations omitted). D. Claims challenging federal agency action under the CWA and NEPA are subject to judicial review under the APA. 5 U.S.C. § 702 (2006); Holy Cross Wilderness Fund v. Madigan, 960 F.2d 1515, 1521 (10th Cir.1992). In issuing the § 404 permits challenged here, the Corps was engaged in informal (“notice and comment”) rule-making. 33 U.S.C. § 1344(a) (2000); Ohio Valley Envtl. Coal. v. Bulen, 429 F.3d 493, 496 (4th Cir.2005). Such informal rulemaking, done pursuant to Section 4 of the APA, 5 U.S.C. § 553 (2006), must be reviewed under Section 10 of the APA, 5 U.S.C. § 706(2) (2006). Ethyl Corp. v. EPA, 541 F.2d 1, 34 (D.C.Cir.1976). Section 10 of the APA establishes that, as a general rule, “agency action, findings, and conclusions” will be set aside only when they are “found to be ... arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706(2) (2000); Citizens To Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 413-14, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971). Review under this standard is highly deferential, with a presumption in favor of finding the agency action valid. Natural Res. Def. Council, Inc. v. EPA 16 F.3d 1395, 1400 (4th Cir.1993). Especially in matters involving not just simple findings of fact but complex predictions based on special expertise, “a reviewing court must generally be at its most deferential.” Baltimore Gas & Elec. Co. v. Natural Res. Def. Council, 462 U.S. 87, 103, 103 S.Ct. 2246, 76 L.Ed.2d 437 (1983). In determining whether agency action was arbitrary or capricious, the court must consider whether the agency considered the relevant factors and whether a clear error of judgment was made. Citizens To Preserve Overton Park, 401 U.S. at 416, 91 S.Ct. 814. “Although this inquiry into the facts is to be searching and careful, the ultimate standard of review is a narrow one. The court is not empowered to substitute its judgment for that of the agency.” Id. Deference is due where the agency has examined the relevant data and provided an explanation of its decision that includes “a ‘rational connection between the facts found and the choice made.’ ” Motor Vehicle Mfrs. Ass’n v. State Farm Mutual Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983) (quoting Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 168, 83 S.Ct. 239, 9 L.Ed.2d 207 (1962)); accord Ohio River Valley Envtl. Coal, Inc. v. Kempthorne, 473 F.3d 94, 102-03 (4th Cir.2006). The “arbitrary and capricious” standard is not meant to reduce judicial review to a “rubber-stamp” of agency action. Ethyl Corp., 541 F.2d at 34. While the standard of review is narrow, the court must nonetheless engage in a “searching and careful” inquiry of the record. Citizens To Preserve Overton Park, 401 U.S. at 416, 91 S.Ct. 814. But, this scrutiny of the record is meant primarily “to educate the court” so that it can “understand enough about the problem confronting the agency to comprehend the meaning of the evidence relied upon and the evidence discarded; the questions addressed by the agency and those bypassed; the choices open to the agency and those made.” Ethyl Corp., 541 F.2d at 36. IV. With this statutory guidance in mind, we turn now to the substance of this appeal, first taking up the issues on appeal concerning the district court’s March 23, 2007, opinion and order. A. The Corps and Intervenors (collectively “Appellants”) claim on appeal that the Corps’ decision about the scope of its NEPA analysis for each of these permits was entitled to deference as a reasonable interpretation of its own regulations. The district court found that the Corps acted contrary to its regulations by limiting the scope of its NEPA analysis to the impact of the filling of jurisdictional waters and by not looking at the larger environmental impacts of the valley fill as a whole. Agreeing with the district court, OVEC argues that the Corps’ NEPA analysis should have considered all environmental impacts caused by the fill, including the impacts to the upland valleys where the fills will be located. The Corps counters that it reasonably determined that, under its regulations, its jurisdictional reach was limited to the affected waters and adjacent riparian areas and that this determination is entitled to deference. Because we are asked here to review the Corps’ interpretation of its own regulations, our review is cabined to assessing the reasonableness of that interpretation. This kind of review is highly deferential, with the agency’s interpretation “controlling unless plainly erroneous or inconsistent with the regulation.” Auer v. Robbins, 519 U.S. 452, 461, 117 S.Ct. 905, 137 L.Ed.2d 79 (1997) (internal quotations omitted); see also Bowles v. Seminole Rock & Sand Co., 325 U.S. 410, 413-14, 65 S.Ct. 1215, 89 L.Ed. 1700 (1945); Kentuckians for the Commonwealth v. Rivenburgh, 317 F.3d 425, 439 (4th Cir.2003) (noting that, when reviewing an agency’s interpretation of its own regulation, “[t]he reviewing court does not have much leeway”). In applying this principle, also known as “Auer deference” or “Seminole Rock deference,” we must first determine whether the regulation itself is unambiguous; if so, its plain language controls. See Christensen v. Harris County, 529 U.S. 576, 588, 120 S.Ct. 1655, 146 L.Ed.2d 621 (2000); United States v. Deaton, 332 F.3d 698, 709 (4th Cir.2003). If ambiguous, however, Auer/Seminole Rock deference is applied. See Christensen, 529 U.S. at 588, 120 S.Ct. 1655; Deaton, 332 F.3d at 709. NEPA requires federal agencies to take a “hard look” at the environmental consequences of their actions, but the statute does not specify how an agency should determine the scope of its NEPA analysis. Wetlands Action Network v. United States Army Corps of Eng’rs, 222 F.3d 1105, 1115 (9th Cir.2000). The Corps’ implementing regulations, however, specify that the proper scope of analysis for NEPA review is “to address the impacts of the specific activity requiring a DA [Department of the Army] permit and those portions of the entire project over which the [Corps] district engineer has sufficient control and responsibility to warrant Federal review.” 33 C.F.R. pt. 325, App. B, § 7(b)(1) (2008). OVEC’s challenge to the scope of the Corps’ NEPA review rests largely on its misapprehension of what constitutes the “specific activity” requiring a permit. The Corps’ regulations are unambiguous in requiring a district engineer to address the impacts of the “specific activity requiring a DA [Department of the Army] permit” in its NEPA analysis. Id. According to OVEC, the Corps’ § 404 permit is a permit for the entire valley fill, down to the last shovelful of dirt at the edge of the valley. But § 404 is itself unambiguous about what the Corps is authorized to permit under the CWA: the Corps “may issue permits, after notice and opportunity for public hearings for the discharge of dredged or fill material into the navigable waters at specified disposal sites.” 33 U.S.C. § 1344(a) (2000) (emphasis added). The specific activity that the Corps is permitting when it issues a § 404 permit is nothing more than the filling of jurisdictional waters for the purpose of creating an underdrain system for the larger valley fill. In fact, the Corps has no legal authority to prevent the placement of fill material in areas outside of the waters of the United States. All other fill activity falls under the exclusive jurisdiction of the WVDEP, as the federally approved state SMCRA regulatory authority. Of course, even if the “specific activity” being permitted under CWA § 404 is the filling of valley streams, the Corps could still be required under NEPA to consider larger impacts of the broader valley fill project if the Corps is found to have “sufficient control and responsibility to warrant Federal review.” 33 C.F.R. pt. 325, App. B, § 7(b)(1) (2008). In eases where the permitted activity is only one part of a larger project, the regulations specify that the Corps has “control and responsibility for portions of the project beyond the limits of Corps jurisdiction where the Federal involvement is sufficient to turn an essentially private action into a Federal action. These are eases where the environmental consequences of the larger project are essentially products of the Corps permit action.” 33 C.F.R. pt. 325, App. B, § 7(b)(2) (2008) (emphasis added). The regulations go on to suggest several factors to be considered in making this determination, including: (i) Whether or not the regulated activity comprises “merely a link” in a corridor type project (e.g., a transportation or utility transmission project). (ii) Whether there are aspects of the upland facility in the immediate vicinity of the regulated activity which affect the location and configuration of the regulated activity. (iii) The extent to which the entire project will be within Corps jurisdiction. (iv) The extent of cumulative Federal control and responsibility. Id. OVEC’s argument that the Corps has sufficient control and responsibility over the larger valley fill to warrant its consideration of the environmental impacts of the entire valley fill project has some intuitive appeal. As OVEC points out, “[t]he Corps could not seriously contend that, if the § 404 permit for the stream-covering portions of the fill were denied, the applicants could build the remainder of the fills with a cutout around the streams. That could be dangerous. The valley fill is designed for stability as an integral unit.” (Appellees’ Br. at 35 n. 2.) Undoubtedly, obtaining a § 404 permit is a “small but necessary” component of the overall upland project. United States Army Corps of Engineers, Combined Decision Document for the Camp Branch Surface Mine Project 4 (July 6, 2006) [hereinafter Camp Branch CDD ]; see also United States Army Corps of Engineers, Combined Decision Document for the Black Castle Surface Mine Project 6 (July 18, 2006) [hereinafter Black Castle CDD ]; United States Army Corps of Engineers, Combined Decision Document for the Laxare East Surface Mine Project 7 (July 18, 2006) [hereinafter Laxare East CDD ]. But the fact that the Corps’ § 404 permit is central to the success of the valley-filling process does not itself give the Corps “control and responsibility” over the entire fill. See Wetlands Action Network v. U.S. Army Corps of Eng’rs, 222 F.3d 1105, 1116-17 (9th Cir.2000) (observing that the fact .that construction of a development project covering hundreds of acres was dependent on a Corps § 404 permit to fill sixteen acres of wetlands did not suffice to make the Corps responsible for including the entire project in the scope of its NEPA analysis). The Corps’ jurisdiction under CWA § 404 is limited to the narrow issue of the filling of jurisdictional waters. To say that the Corps has a level of control and responsibility over the entire valley fill project such that “the environmental consequences of the larger project are essentially products of the Corps permit action,” 33 C.F.R. pt. 325, App. B, § 7(b)(2) (2008), is to effectively read out of the equation the elaborate, congressionally mandated schema for the permitting of surface mining operations prescribed by SMCRA. Under SMCRA, the state of West Virginia has “exclusive jurisdiction over the regulation of surface coal mining and reclamation operations.” 30 U.S.C. § 1253 (2000). Congress clearly contemplated that the regulation of the disposal of excess spoil and the creation of valley fills fall under the SMCRA rubric. See 30 U.S.C. § 1265(b)(22)(D) (2000) (requiring that lateral drains be constructed where a spoil disposal area contains “springs, natural water courses or wet weather seeps”); Kentuckians for the Commonwealth, Inc. v. Rivenburgh, 317 F.3d 425, 443 (4th Cir.2003) (“[I]t is beyond dispute that SMCRA recognizes the possibility of placing excess spoil material in waters of the United States.... ”). As part of its federally approved SMCRA regulatory program, the WVDEP surface mine permitting process examines “[ejvery detail of the manner in which a coal mining operation is to be conducted .... including] the plan for disposal of excess spoil for surface ... mining opera-tions_” (Br. for the W. Va. Dep’t of Commerce and the W. Va. Dep’t of Envtl. Prot. as Amici Curiae Supporting Appellants at 13.) As the Corps explains in its permits, “the social and environmental impacts associated with surface coal mining and reclamation operations are appropriately analyzed by WVDEP in this context before that agency decides whether to permit the mining operation under SMCRA.” Camp Branch CDD 4; Black Castle CDD 6; Laxare East CDD 7; United States Army Corps of Engineers, Combined Decision Document for the Republic No. 2 Surface Mine Project 6 (July 6, 2006) [hereinafter Republic No. 2 CDD ]. A SMCRA permit applicant must provide detailed information about possible environmental consequences of the proposed operations, as well as assurances that damage to the site will be prevented or minimized during mining and substantially repaired after mining has come to an end. The WVDEP must ensure compliance with SMCRA’s environmental protection performance standards. See 30 U.S.C. §§ 1257,1260,1265 (2000). If the Corps, by issuing a § 404 permit, can turn a valley fill project “into a Federal action,” 33 C.F.R. pt. 325, App. B, § 7(b)(2) (2008), the WVDEP’s regulation of the fill process becomes at best duplica-tive, and, at worst, meaningless. NEPA plainly is not intended to require duplication of work by state and federal agencies. See 40 C.F.R. § 1506.2(b) (“Agencies shall cooperate with State and local agencies to the fullest extent possible to reduce duplication between NEPA and State and local requirements....”). The Corps’ general regulatory approach echoes this sentiment. See 33 C.F.R. § 320.1(a)(5) (2008) (“The Corps believes that state and federal regulatory programs should complement rather than duplicate one another.”); 33 C.F.R. § 337.1 (2008) (noting that, in issuing public notice for projects involving the discharge of fill material into jurisdictional waters, “[district engineers are encouraged to develop procedures to avoid unnecessary duplication of state agency procedures”). SMCRA also calls for a coordinated and non-duplicative approach to environmental review. See 30 U.S.C. § 1253(a)(6) (2000) (requiring that a state SMCRA program establish “for the purposes of avoiding duplication, ... a process for coordinating the review and issuance of permits for surface coal mining and reclamation operations with any other Federal or State permit process applicable to the proposed operations”). While SMCRA’s provisions should not be construed as “superseding, amending, modifying, or repealing” the requirements of NEPA or the CWA, 30 U.S.C. § 1292(a) (2000), neither should NEPA be construed to require the Corps to essentially federalize an environmental review process that has already been delegated to federally approved state programs. See Wetlands Action Network, 222 F.3d at 1117 (noting, in support of its finding that the Corps’ NEPA analysis for a wetlands-filling permit need not include the effects of the larger development project, that state regulations control the design of the project and that the larger project was already subject to extensive state environmental review); Sylvester v. U.S. Army Corps of Eng’rs, 884 F.2d 394, 401 (9th Cir.1989) (“We, finally, draw comfort from the fact that ordinary notions of efficiency suggest a federal environmental review should not duplicate competently performed state environmental analyses.”). In Department of Transportation v. Public Citizen, 541 U.S. 752, 767, 124 S.Ct. 2204, 159 L.Ed.2d 60 (2004), the Supreme Court rejected the idea that “an agency’s action is considered a cause of an environmental effect [for purposes of NEPA] even when the agency has no authority to prevent the effect.” The Court instructed that proximate causation, rather than “but for” causation, was the relevant measure of the causal relationship between the agency action and the environmental effects. 541 U.S. at 767, 124 S.Ct. 2204. In engaging in this proximate cause analysis, “ ‘courts must look to the underlying policies or legislative intent in order to draw a manageable line between those causal changes that may make an actor responsible for an effect and those that do not.’ ” Id. (quoting Metro. Edison Co. v. People Against Nuclear Energy, 460 U.S. 766, 774 n. 7, 103 S.Ct. 1556, 75 L.Ed.2d 534 (1983)). But for the Corps’ § 404 permit, a valley fill could not be built; yet it is WVDEP, and not the Corps, that has “control and responsibility” over all aspects of the valley fill projects beyond the filling of jurisdictional waters. Thus, under the plain language of the regulation, activity beyond the filling of jurisdictional waters is not within the Corps’ “control and responsibility” because upland environmental effects are “not essentially a product of Corps action,” 33 C.F.R. pt. 325, App. B, § 7(b)(2) (2008). Even if we credit OVEC’s arguments regarding the Corps’ control and responsibility over the greater valley fill project as a plausible construction of the Corps’ regulation, we must still deem the regulation “ambiguous,” and the Corps’ interpretation would be entitled to deference as long as it is not “plainly erroneous or inconsistent with the regulation.” Auer, 519 U.S. at 461, 117 S.Ct. 905 (internal quotations omitted); see also Seminole Rock, 325 U.S. at 413-14, 65 S.Ct. 1215; Kentuckians for the Commonwealth, 317 F.3d at 439. In the case of each of the challenged permits, the Corps’ engineers reasonably determined that a scope of NEPA analysis extending beyond the Corps’ limited jurisdiction to include environmental effects on upland areas would encroach on the regulatory authority of WVDEP, which administers the state’s SMCRA program and is responsible for determining the social and environmental impacts associated with surface mining operations. Camp Branch CDD 6-7; Black Castle CDD 5-7; Laxare East CDD 6-8; Republic No. 2 CDD 4-7. Thus the Corps did not act arbitrarily or capriciously in determining the scope of its NEPA analysis. B. The Corps next challenges the district court’s finding that the Corps failed to adequately support its mitigated FONSIs under NEPA and its findings of no significant degradation to waters of the United States under the CWA. The Corps takes issue with three areas in which the district court found the Corps’ findings to be lacking: (1) the Corps’ CWA analysis of the impact of the permitted fills on the structure and function of affected streams; (2) the sufficiency of the proposed mitigation measures for purposes of CWA and NEPA compliance; and (3) the adequacy of the Corps’ NEPA and CWA assessments of cumulative impacts of the proposed fills. In finding fault with the Corps’ conclusions, the district court failed to heed the admonition of Citizens of Overton Park that, in reviewing agency action, “[t]he court is not empowered to substitute its judgment for that of the agency,” 401 U.S. at 416, 91 S.Ct. 814. Focusing now on the administrative record before us, and viewing the Corps’ findings through the lens of arbitrary and capricious review, we cannot say that its findings regarding stream structure and function, mitigation, or cumulative impacts were an “abuse of discretion” or “not in accordance with law,” 5 U.S.C. § 706(2) (2000). 1. The Corps’ CWA Guidelines require the Corps to “[d]etermine the nature and degree of effect that the proposed discharge will have, both individually and cumulatively, on the structure and function of the aquatic ecosystem and organisms.” 40 C.F.R. § 230.11(e) (2008). The Guidelines do not expressly define the phrase “function of the aquatic ecosystem,” but they do identify a variety of factors the Corps’ should consider, including “potential changes in substrate characteristics and elevation, water or substrate chemistry, nutrients, currents, circulation, fluctuation, and salinity, on the recolonization and existence of indigenous aquatic organisms or communities.” 40 C.F.R. § 230.11(e) (2008). In February 1990, the Corps and EPA developed a Memorandum of Agreement to make clear what kind of functional analysis the Corps was required to conduct. Memorandum of Agreement Between the Environmental Protection Agency and the Department of the Army Concerning the Determination of Mitigation Under the Clean Water Act Section 404(b)(1) Guidelines, 55 Fed.Reg. 9210 (Mar. 12, 1990) [hereinafter “MOA”]. The MOA sets out a common approach for evaluating stream function, calling for functional values to be assessed “by applying aquatic site assessment techniques generally recognized by experts in the field and/or the best professional judgment of Federal and State agency representatives, provided such assessments fully consider ecological functions included in the Guidelines.” Id. In 2002, the Corps issued a Regulatory Guidance Letter that called upon Corps district offices “when possible ” to “use functional assessments by qualified professionals to determine impacts and compensatory mitigation requirements.” United States Army Corps of Engineers Regulatory Guidance Letter, No. 02-2 (Dec. 24, 2002) [hereinafter “RGL 02-02”]. RGL 02-02 provided that assessment techniques should be “generally accepted by experts or the best professional judgment” of federal and state agency representatives. Id. The Corps currently does not have a functional assessment protocol in place for use in West Virginia, though it is in the process of developing one. As a result, the Corps relies on the best professional judgment of its staff to assess aquatic impacts and potential mitigation measures. This generally means assessing stream structure as a surrogate for function. OVEC argues that the plain language of the Guidelines requires that the Corps analyze both the structural and the functional effects of fill permits on the affected streams. In their view, this means that a full functional assessment is required. The Corps’ failure to complete such an assessment, OVEC continues, is inconsistent with the plain language of its regulation, and the Corps’ substitute method of using its “best professional judgment” was arbitrary and capricious because it lacked any objective standards. Appellants argue that because the CWA Guidelines provide no definition of “function” or any specific methodology for evaluating function, the Corps’ interpretation and implementation of the regulation is entitled to deference. According to Appellants, the methodologies used by the Corps were an effective surrogate for functional assessment, and these techniques are compliant with the Guidelines, the MOA, and RGL 02-02. The district court agreed with Appellants that a functional assessment was not required and that the Corps was entitled to deference on how to measure stream structure and function. It further found that the Corps was entitled to use its “best professional judgment” in accordance with the MOA and RGL 02-02 in evaluating functional loss. Nonetheless, the court concluded that, even under a “best professional judgment” standard, the Corps was obligated, and failed, to fully assess all ecological functions; to take a “hard look” at the evidence; and, to provide a reasoned basis for its conclusions. Based on our review of the administrative record, however, we cannot say that the Corps’ assessments of stream functions in the challenged permits were arbitrary and capricious. Contrary to OVEC’s position that the CWA Guidelines mandate a full functional assessment, the Guidelines in fact offer no definition of the word “function” or any explanation of how “structure” and “function” are to be assessed. The MOA and RGL 02-02 attempt to fill this gap by encouraging use of a functional assessment but allowing Corps engineers to use their best professional judgment when such an assessment is not possible. An agency’s interpretation of its own regulations is due significant deference, Kentuckians for the Commonwealth Inc., 317 F.3d at 439, and the MOA/RGL 02-02 approach does not appear plainly erroneous or inconsistent with the Guidelines. In this case, the Corps, using its best professional judgment, used stream structure as a surrogate for assessing stream function. Taking Black Castle as an example, the Corps used detailed measurements provided by Intervenors on the benthic macroinvertebrate population to draw conclusions about the level of stream function at the proposed fill sites. Black Castle CDD 102 (“Biological measurements (metrics) represent elements of the structure and function of the bottom-dwelling macroinvertebrate assemblage ... Such a measure of structure and function of the biota ... is an appropriate indicator of ecological quality, the integrity of soil and water chemistry, geological processes, and land use changes.”). The Corps also used the EPA’s Rapid Bioassessment Protocol for Use in Streams and Wadeable Rivers (“RBP”) to assess aquatic habitat, it used the EPA-developed West Virginia Stream Condition Index (“WV-SCI”) for measuring the types and proportions of benthic insects, and it followed the EPA’s field manual for measuring the ecological condition of streams. OVEC identifies nutrient cycling as one of the factors the Corps is instructed, but failed, to consider under 40 C.F.R. § 230.11(e) (2008). The Corps’ CDDs themselves acknowledge this shortcoming, noting that the effects of filling ephemeral streams on nutrient cycling are difficult to measure and that there is a lack of consensus among the relevant agencies about how best to collect quantitative evidence regarding these functions. To compensate for these effects, however, the Corps’ permitting decisions call for limiting impacts to channels that do not sustain long periods of flow and for establishing a riparian buffer around mitigation sites. In fact, in each of its CDDs, the Corps provides its complete findings under 40 C.F.R. § 230.11(e) (2008), including a section on “Physical and Chemical Characteristics of the Aquatic Ecosystem,” which covers substrate characteristics, water quality, current patterns and water circulation, water fluctuations, and salinity gradients; and a section on “Biological Characteristics of the Aquatic Ecosystem,” which covers threatened and endangered species and their habitat, aquatic organisms in the food web, and other wildlife. Camp Branch CDD 10-16; Black Castle CDD 13-31; Laxare East CDD 15-28; Republic No. 2 CDD 15-19. The Corps is entitled to use its best professional judgment for assessing the structure and function of the affected aquatic ecosystem, and its CDDs address the required considerations under the Guidelines, 40 C.F.R. § 230.11(e) (2008). Thus, these findings were not inconsistent with the Corps’ regulations and cannot be characterized as arbitrary, capricious, or otherwise not in accordance with the law. The district court placed great weight on the Appellees’ expert testimony at trial in finding that the Corps’ functional evaluation was lacking. We acknowledge the importance of extra-record evidence in NEPA cases to inform the court about environmental factors that the agency may not have considered. While review of agency action is typically limited to the administrative record that was available to the agency at the time of its decision, Camp v. Pitts, 411 U.S. 138, 142, 93 S.Ct. 1241, 36 L.Ed.2d 106 (1973) (per curiam), a NEPA suit is inherently a challenge to the adequacy of the administrative record, see County of Suffolk v. Sec’y of the Interior, 562 F.2d 1368, 1384 (2d Cir.1977). That is why, in the NEPA context, “courts generally have been willing to look outside the record when assessing the adequacy of an EIS or a determination that no EIS is necessary.” Webb v. Gorsuch, 699 F.2d 157, 159 n. 2 (4th Cir.1983) (citing County of Suffolk, 562 F.2d at 1384). Such consideration of extra-record evidence in a NEPA case does not, however, give courts license to simply substitute the judgment of plaintiffs experts for that of the agency’s experts. Hughes River Watershed Conservancy v. Johnson, 165 F.3d 283, 289-90 (4th Cir.1999). “Agencies are entitled to select their own methodology as long as that methodology is reasonable,” and we must defer to such agency choices. Id. at 289; see also Native Ecosystems Council v. United States Forest Serv., 428 F.3d 1233, 1244 (9th Cir. 2005) (finding, in the context of a NEPA challenge, that because the Forest Service had provided a “thorough and reasoned explanation” for its position, the court would not “take sides in a battle of the experts” (internal quotations omitted)); Spiller v. White, 352 F.3d 235, 244 (5th Cir.2003) (same). Having found that the Corps was not obligated to engage in a full functional assessment, it is not our place to dictate how the Corps should go about assessing stream functions and losses. In matters involving complex predictions based on special expertise, “a reviewing court must generally be at its most deferential.” Baltimore Gas & Elec. Co. v. Natural Res. Def. Council, 462 U.S. 87, 103, 103 S.Ct. 2246, 76 L.Ed.2d 437 (1983). When presented with conflicting evidence, courts must generally defer to the agency evaluation because “an agency must have discretion to rely on the reasonable opinions of its own qualified experts even if, as an original matter, a court might find contrary views more persuasive.” Marsh v. Or. Natural Res. Council, 490 U.S. 360, 378, 109 S.Ct. 1851, 104 L.Ed.2d 377 (1989). The CDDs issued with each permit include substantial analysis and explanation about the Corps’ impact findings. These determinations are within the agency’s special expertise and were based on Corps staffs “best professional judgment.” As such, the Corps cannot be said to have acted arbitrarily or capriciously. 2. OVEC next questions the sufficiency of the mitigation plans contained in the CDDs for each of the challenged permits. OVEC charges that the proposed mitigation measures are insufficient both to satisfy the Corps’ requirements under the CWA and to justify the issuance of a mitigated FONSI in lieu of a full EIS under NEPA. Under the Corps’ CWA Guidelines, a § 404 permit cannot issue “unless appropriate and practicable steps have been taken which will minimize potential adverse impacts of the discharge [of fill material] on the aquatic ecosystem.” 40 C.F.R. § 230.10(d) (2008). In their MOA of 1990, the EPA and Corps make “no overall net loss” the goal of the § 404 regulatory program, and agree that mitigation has three components: avoidance, minimization, and compensatory mitigation. 55 Fed.Reg. 9210 (Mar. 12, 1990). Avoidance is defined as the selection of the least environmentally damaging practical alternative. Id. Minimization is achieved through practicable project modifications and permit conditions that minimize adverse impacts. Id. Finally, compensatory mitigation is used where appropriate to compensate for unavoidable adverse impacts after all avoidance and minimization measures have been taken. Id. Compensatory mitigation can include the restoration of existing wetlands or the creation of new wetlands, and is to be done as close to the discharge site as possible (“on-site mitigation”). Id. Where on-site mitigation is not possible, off-site mitigation is permitted but should 'take place in the same geographic area if possible. Id. The MOA specifically directs that the functional values lost should be carefully considered when determining compensatory mitigation, and that, generally, in-kind mitigation should be used. Id. Noting the continued uncertainty of success in wetland creation, the MOA further instructs that restoration options should be considered before creation options. Id. The mitigation measures specified for the four challenged fill projects include stream enhancement, stream restoration, and stream creation. Each of the proposed compensatory mitigation plans would, according to the Corps, lead to no net loss of habitat. Much of OVEC’s concern over the proposed compensatory mitigation plans focuses on the potential impacts on head-water streams. As noted earlier, the role of headwater streams in downstream ecology is a matter of some debate in the scientific community and among the parties to this litigation. According to OVEC’s experts, headwater streams function uniquely in stream ecology, collecting and transporting organic matter to nourish aquatic life downstream, allowing higher levels of nutrient uptake than perennial streams, and serving as a habitat for a variety of benthic organisms. Intervenors, on the other hand, offered expert testimony that, while headwater streams are “very important” to stream ecology (J.A. 4345), downstream waters could still maintain a healthy benthic community even when headwater streams were filled, as long as the water quality below the fill remained good. Another expert for the Intervenors further testified that, in fact, ephemeral streams “will not provide as much benefit as downstream reaches,” because their ephemeral nature does not allow them to “be giving the same type of value and processes as the one that’s functioning all the time.” (J.A. 4381.) The Corps, meanwhile, seems to take the position that, whatever the functional uniqueness of headwater streams, nothing in NEPA, the CWA, or the Corps’ regulations prevents them from allowing mitigation of headwater stream destruction through enhancement, restoration, or creation of a downstream perennial system. The district court, again relying heavily on the trial testimony of OVEC experts, concluded that the Corps had failed to fully assess the impacts of destroying headwater streams. Taking OVEC’s view of the unique role of headwater streams, the district court found that the mitigation plans failed to explain how a valley fill’s destruction of headwater streams could be compensated for simply by the creation, enhancement, or replacement of an equal or greater length of some other stream type. The court further suggested that the Corps’ failure to conduct a full functional assessment meant that it ignored a number of critical headwater stream values in its evaluation of adverse impacts, and therefore the mitigation plans could not possibly be adequate to offset adverse impacts. The Corps defends the mitigation plans by arguing that nothing in the CWA Guidelines requires compensatory mitigation measures that precisely replicate the functions of the impacted streams. Having reviewed the Guidelines, this Court concludes that, whatever the role of head-water streams in overall watershed ecology, the Corps is not required to differentiate between headwater and other stream types in the determination of mitigation measures. In reaching this conclusion, we look to the Corps’ guidance in RGL 02-02, which provides that “[districts should require compensatory mitigation projects for streams to replace stream functions where sufficient functional assessment is feasible. However, where functional assessment is not practical, mitigation projects for streams should generally replace linear feet of stream on a one-to-one basis.” As we have already noted, a full functional assessment protocol is not yet available to the Corps, and the Corps is thus entitled to use its best professional judgment to assess structural and functional losses for purposes of the Guidelines. Similarly, RGL 02-02 advises that, where a full functional assessment is not feasible, the only compensatory mitigation measure the Corps must require in a permitting decision is stream replacement on a one-to-one basis. Nothing in the Corps’ CWA guidance requires that only in-kind, on-site mitigation measures be used. By this standard, the Corps’ permitting decisions have exceeded the mitigation requirements by creating mitigation plans involving greater than one-to-one replacement schemes. The Camp Branch project, for example, involved 15,514 linear feet of direct impacts, and the permit requires mitigation of 43,306 feet. The Corps’ guidance does instruct that “functional values lost by the resource to be impacted must be considered” in developing a mitigation plan. MOA, 55 Fed.Reg. 9210 (Mar. 12, 1990). But the guidance also provides that compensatory mitigation must be “practicable.” Id. The Corps’ guidance documents indicate that, in “determining ‘practicability,’ Districts will consider the availability of suitable locations, constructability, overall costs, technical requirements, and logistics.” RGL 02-02. “In certain circumstances of regions of the country, on-site compensatory mitigation opportunities are limited,” and the Corps must look instead to other compensatory options. Thomas F. Caver, Deputy Dir. of Civil Works, U.S. Army Corps of Eng’rs, Internal Guidance on Mitigation for Impacts to Aquatic Resources from Surface Coal Mining (May 7, 2004). In other circumstances, the stream functions being lost on-site may be “ubiquitous in the watershed,” while “wetland functions are rare or degraded.” Id. In such a situation, “it may be appropriate to replace lost stream functions with wetlands functions.” Id. Thus, where on-site or in-kind functional mitigation is not practicable or even ecologically preferable, the Corps’ guidance allows compensation plans that employ off-site or out-of-kind mitigation based on improvements to the overall aquatic health of the watershed. For example, in the case of the Laxare East and Black Castle permits, the Corp’s mitigation plans aimed to improve the water quality of already severely distressed streams in portions of the Laurel Creek watershed. Similarly, in the case of the Republic No. 2 mine, the mitigation plan was designed to address stability issues along the Long Branch tributary, based on an assessment that “this section of stream could contribute to improved aquatic habitat and ultimately aquatic diversity .... by attempting to replace the chemical, hy-drologic, and geomorphic functions of the impacted channels.” Republic No. 2 CDD 13. Each of these mitigation plans accords with the holistic watershed approach called for in the Corps’ guidance documents. Moreover, each of the mitigation plans for the challenged permits included requirements for continued monitoring of the efficacy of the mitigation measures, in some cases for as much as 10 years. Each permit also contains detailed special conditions that impose numerous performance standards to measure and ensure the success of mitigation. OVEC also takes issue with the use of stream creation as a mitigation measure. The Camp Branch and Laxare East plans both employ stream creation as a significant component of their compensatory mitigation schemes. Under these plans,' sediment ditches used during mining to collect runoff, control drainage, and collect sediment, will be converted into new stream channels. OVEC’s experts have questioned these stream creation proposals, calling them scientifically untested and unsound. The Corps’ support for its claim that the proposed stream creation measures have good potential for success is admittedly limited. The Corps cites one example of stream creation in a mining area in Kentucky as well as an Ohio State University study on the potential for enhancing the natural ecology of drainage ditches. See Camp Branch CDD 44; Laxare East CDD 98-101. However, the novelty of a mitigation measure alone cannot be the basis of our decision to discredit it. When an agency is called upon to make complex predictions within its area of special expertise, a reviewing court must be at its most deferential. Baltimore Gas & Elec. Co. v. Natural Res. Def. Council, 462 U.S. 87, 103, 103 S.Ct. 2246, 76 L.Ed.2d 437 (1983); see also Marsh v. Or. Natural Res. Council, 490 U.S. 360, 378, 109 S.Ct. 1851, 104 L.Ed.2d 377 (1989) (“When specialists express conflicting views, an agency must have discretion to rely on the reasonable opinions of its own qualified experts even if, as an original matter, a court might find contrary views more persuasive.”). The Corps admits that “[t]ime is required with any new, scientifically based development as well as monitoring and evaluation to show the success and/or failures of the project.” Camp Branch CDD (Supplement) 3. And, the monitoring plans in place for Camp Branch and Lax-are East allow the Corps to reevaluate their efficacy determinations as the stream creation projects progress. Because the mitigation measures reflect the Corps’ determinations of the most appropriate and practicable means of compensating for anticipated impacts and losses of value, we cannot say that the Corps’ conclusion that compensatory mitigation would offset the adverse effects of the All activity was arbitra