concurring in part and dissenting in part:
I concur in the majority opinion to the extent that it affirms the denial of attorneys’ fees in part and denies relief in the form of removing or modifying the dam. Because I disagree with the majority that Corps’ discussion of cumulative impacts failed to take a sufficiently hard look and that the govern*1494ment’s position on cumulative impacts in Marsh I and II was not substantially justified, I dissent.
I
Even assuming, as the majority concludes, that Corps attempts to read our mandate in Marsh II to.o narrowly, I would hold that the extensive environmental analysis provided in EISS-2 was sufficient to satisfy NEPA’s requirement that an agency take a hard look at the environmental consequences of its actions.
ONRC identifies a number of specific respects in which EISS-2 is allegedly inadequate. Broadly stated, they are the failure to disclose and discuss adequately the existing animate and inanimate environment, as NEPA and the regulations require, 40 C.F.R. ■§ 1502.15, which ONRC maintains is indispensable to an adequate discussion of cumulative impacts. ONRC also points to Corps’ failure to disclose and discuss all of the cumulative impacts completely.
A
ONRC asserts that EISS-2 insufficiently describes water quality because it focuses on temperature and turbidity, whereas water quality is also a function of dissolved oxygen levels, pH, coliform levels, algal growths, and bottom or sludge deposits. It argues that Corps failed to comply with its obligation under 40 C.F.R. § 1503.4(a)(3), which provides that supplementing, improving, or modifying analyses is an appropriate response to a comment, by ignoring the comment on the draft EISS-2 by the Oregon Department of Environmental Quality (DEQ) suggesting that discussion of all these factors is necessary to evaluate cumulative effects properly. Corps’ response to the comment indicated that EISS-2 addresses the water quality elements which had been identified as concerns through the previous NEPA processes, i.e., effects of operation of the projects on water temperatures and turbidity in the Rogue River, and affirmed its commitment to develop and implement a water monitoring program to ensure continued compliance with the state water quality standards mentioned by DEQ.
Corps argues that its response suffices under 40 C.F.R. § 1503.4(a)(5), which allows an agency to “[ejxplain why the comments do not warrant further agency response, citing ... reasons which support the agency’s position.” See Marsh II, 832 F.2d at 1488-89 (agency may place opposing views in separate comment and responses section with thoughtful responses). Although it is a close question, I agree. Considering the scoping process turned up no substantial concern about factors other than temperature and turbidity, and DEQ itself issued a State Water Quality Certification for the Elk Creek Dam project, I believe that the detailed discussion in EISS-2 of the importance of water temperature and its effect on the ecological and aquatic environment, the effects of the dam construction on temperature, and the effects of turbidity caused by the Rogue River Basin Project provide enough information about incremental impact of the Elk Creek Dam to comply with our mandate.
B
ONRC argues that EISS-2’s description of the animate environment is inadequate because it fails to note that wild coho salmon are on Oregon’s list of sensitive species; does not describe aquatic invertebrates, which are a major source of food for fish; and neglects to describe how the Rogue River basin’s ecosystems function as a whole. Corps contends that EISS-2 adequately describes existing conditions of the fish and fisheries production and the cumulative effects of the components of the Rogue River basin project upon these fish and fisheries production.
While it might have been preferable for EISS-2 specifically to acknowledge that wild coho are on the list of sensitive species, failing to do so is not fatal, as EISS-2 does discuss the fact that few wild coho are spawning in the upper portions of the basin, including Elk Creek, and that while Elk Creek supports coho salmon, few still spawn there. Further, three comments (by the EPA, the Interior, and ODFW) point out that the wild coho salmon are on the state’s list of sensitive species, ensuring that Corps was aware of this fact when it made its *1495decision. As we recently observed in The Laguna Greenbelt, Inc. v. United States Department of Transportation, 42 F.3d 517, 527 (9th Cir.1994), “relief will not be granted if the decision-maker was otherwise fully informed as to the environmental consequences and NEPA’s goals were met.”
EISS-2 generally discusses the impact of temperature on fish emergence and growth, although it does not specifically mention invertebrate food organisms. This is sufficient, as ONRC’s complaint of insufficiency stems from a comment by the EPA indicating that temperature alterations and changes in the thermal regime could affect the timing of food that is available when the fry emerge, and that comment was limited to the impact of the Conservation Pool Alternative. That Alternative was rejected, and ONRC suggests no other reason why the effect on emergence should have been discussed specifically in terms of invertebrates instead of generally, as the Corps elected to do.
Finally, ONRC’s contention that while the parts may be described adequately, the ecosystem as a whole is not, also fails. Corps has undertaken a comprehensive review of the environment of the Rogue River Basin and the components that make up the basin’s ecosystem. ONRC points to no particular respect in which the discussion falls short. Nor did any organization or governmental agency submit any comment during the comment period that suggested that the Corps’ discussion of how the ecosystems function should have been focused differently. This substantially undercuts ONRC’s claim that EISS-2’s discussion of cumulative effects is irretrievably compromised. Additionally, we have previously held that “[ajbsent exceptional circumstances, such belatedly raised issues may not form a basis for reversal of an agency decision.” Havasupai Tribe v. Robertson, 943 F.2d 32, 34 (9th Cir.1991). Given the extensive opportunity for comment that NEPA and CEQ regulations provide, and in the absence of intervening “new” information, I do not see • how Corps’ failure to discuss a topic that was never raised during either the scoping process or the circulation of the draft EIS could be arbitrary and capricious.
I therefore conclude that none of these imperfections, assuming them to be within the mandate on remand, is insufficient to render EISS-2’s discussion of the environment inaccurate or inadequate for providing context for considering cumulative impacts.
C
ONRC contends that Corps’ failure to describe the environment adequately makes the discussion of cumulative impacts meaningless.
First, ONRC cites Corps’ failure to disclose the import of losing Elk Creek’s habitat for the Rogue River’s populations of wild coho and summer st’eelhead. EISS-2 did not, however, fail either to describe the predicament facing those populations or the effect Elk Creek Dam would have. It refers to sixteen years of fisheries evaluations; indicates that there is no evidence that yearling coho increased or decreased post-dam at any site in the Rogue River; advises that few wild coho still spawn in Elk Creek; opines that little effect on wild summer steelhead spawning can be attributed to the upriver project in the Rogue River Basin because the majority of that spawning occurs elsewhere; and acknowledges that fish passage is an area ,.of serious concern.
ONRC also argues that Corps cannot excuse its failure to discuss cumulative impacts on commercial ocean fisheries on the ground that those impacts “cannot be quantified at the present time.” It notes that incomplete information relevant to reasonably foreseeable significant adverse impacts and essential to a reasoned choice among alternatives must be included, provided the costs of obtaining it are not exorbitant, 40 C.F.R. § 1502.22(a), and that an environmental impact statement must evaluate these impacts “based upon theoretical approaches generally accepted in the scientific community.” Id. at § 1502.22(b)(4). Corps analyzed the cumulative impacts on the local economy, but not on commercial ocean fisheries; it did, however, discuss the release of hatchery fish, which it claims may compensate for any economic loss to ocean fisheries of wild runs due to operation of the basin project. ONRC has not *1496shown that Corps’ failure to go further is arbitrary and capricious, particularly in light of the disclosure that few wild coho or spring/fall chinook salmon spawn in the Elk Creek watershed. Therefore, even if Corps is obliged to discuss purely economic effects, a question we need not decide, it sufficiently evaluated the effects of the basin project.
ONRC additionally maintains that EISS-2 inadequately discusses all the significant effects of Lost Creek and Applegate Lakes on flows during flood events. Corps states in EISS-2 that “[f]or long term reservoir studies there are two seasons during the year which are of particular interest: spring filling and summer/fall drawdown. These are the times of year when the effects of reservoirs are most notable, except for the regulation of flood events.” Thus, ONRC contends, EISS-2 should also have discussed more thoroughly the impacts of flood regulation. I cannot conclude that Corps did not take a hard enough look at flood flows in light of EISS-2’s discussion of water control regulation at the dam, how the releases would be regulated to avoid excessive downstream flow, the periodicity of flood events, and the survival of chinook and steelhead eggs during flood control operations.
Finally, ONRC asserts that Corps’ worst omission was its failure to discuss the long-term cumulative impacts of temperature, turbidity, habitat alterations, and hatchery releases on anadromous fish. Its argument rests on EISS-2’s disclosure that “the cumulative effect of Elk Creek [at full pool] and Lost Creek Lakes” would be that spring and fall chinook would emerge earlier than under pre-dam conditions and might suffer from increased mortality as a result. However, this disclosure pertains to the rejected Full Pool Alternative, and ONRC does not suggest how it is material to the No Conservation Pool Alternative. Likewise, ONRC’s argument that Corps cannot avoid discussing the long-term impact of increased water temperatures because they are not yet known is unavailing because the deficiency also pertains only to an alternative rejected by Corps because of its environmental impact. Therefore, information about long-term cumulative impacts on anadromous fish that is not provided in EISS-2 is not “essential to a reasoned choice among alternatives,” 40 C.F.R. § 1502.22(a), and the Corps had no duty to obtain it.
In sum, I would conclude that the more than 1100 pages of report and appendices in EISS-2 are appropriately “analytic rather than encyclopedic.” 40 C.F.R. § 1502.2. Its “form, content and preparation” are sufficient to “foster both informed decision-making and informed public participation.” State of California v. Block, 690 F.2d 753, 761 (9th Cir.1982). While EISS-1 failed to pass muster for lack of discussion of cumulative impacts, EISS-2 provides an expansive discussion that should suffice. Whether selected areas should have been plumbed deeper or analyzed differently is a question requiring the agency’s environmental expertise; therefore, we may only disturb the agency’s decisions if it is arbitrary and capricious. Overall, I cannot say that any of the deficiencies asserted by ONRC show that Corps failed its NEPA obligation to take a “hard look” at the cumulative impact of the basin project. Accordingly, I would hold that the district court did not err in dismissing the action.1
II
Finding that the government’s position on sufficiency of Corps’ cumulative impacts analysis was substantially justified, the district court denied ONRC’s request for attorneys fees under EAJA. While I agree with the majority that this is a difficult issue, I disagree with its result. In light of the difficulty of the question and the abuse of discretion standard of review, the district court should be affirmed as to its denial of attorneys’ fees.
A
The district court focused on whether the Corps could reasonably have thought it had adequately complied with the obligation set out in 40 C.F.R. § 1508.7, to consider the *1497cumulative impacts of proposed actions, on which it was the prevailing party in Marsh II and Marsh III. The court noted that EISS-1 described the cumulative impacts on the physical, biological, and human elements of the environment, paying particular attention to water quality, fish production (including detrimental effects on anadromous fish), and fishing; and that the Corps had considered the existing conditions (including the already completed portion of the Rogue River Basin Project) and anticipated future consequences with and without the Elk Creek project. As to the two flaws Marsh II cited as representative examples of the general defectiveness of EISS-1, the district court explained that, despite the fact that the supplemental statement did not itself discuss cumulative impact on turbidity, the Corps could reasonably have believed that its reference to 1974 and 1979 Water Quality Studies, which provided a state of the art analysis of turbidity contributed by the Lost Creek Dam, adequately addressed the cumulative impact of turbidity from both dams. (In this connection the district court found that Corps could reasonably have believed it was unnecessary to discuss turbidity contributions from the Ap-plegate Dam because of geographical remoteness.) On the question of water quality, the district court found that the 1974 and 1979 Water Quality Studies included discussion of the effects of Lost Creek Dam on downstream water quality in relation to the effects of Elk Creek Dam that a reasonable person could think sufficed. Based on these findings, the court concluded that reasonable minds could differ on whether the Corps was correct to assert that it had taken a sufficiently hard look at cumulative impacts to go forward with the project; and that its litigation position was reasonable.
There is no dispute that ONRC was the prevailing party on the cumulative impacts issue in Marsh II and on remand, when the district court enjoined further construction of the dam at its behest. Marsh III, 677 F.Supp. at 1077-78. However, ONRC argues that the district court erred in failing to recognize that the government had to justify its position on the .mitigation of impacts and the uncertainty issues on which it contends the Ninth Circuit was not completely reversed in Marsh IV, as well as on cumulative impacts. In addition, ONRC urges that the district court erred by taking into account its own, overturned position on the merits; by concluding that the government argued “forcefully and well”; and by misinterpreting the scope of Marsh IPs remand. Finally, it contends that the district court’s finding that the government’s position was substantially justified was based on clearly erroneous findings of fact.
ONRC’s appeal turns on whether the district court abused its discretion in finding that the government’s position was “justified to a degree that could satisfy a reasonable person.” Pierce v. Underwood, 487 U.S. 552, 565, 108 S.Ct. 2541, 2550, 101 L.Ed.2d 490 (1988). Although this question is never easy to resolve, I do not believe the panel can fairly say that the district court’s reasoning lacks support in the record.2
B •
ONRC contends that the district court erroneously concluded that Corps argued “forcefully and well” and that no adverse precedent clearly foreclosed its litigation position because precedent clearly established the Corps’ duty to discuss cumulative impacts.3 However, the government has never *1498argued that it had no such duty; its position (with which we disagreed in Marsh II) was that Corps adequately considered the cumulative impacts in its report. This case is therefore unlike ONRC v. Madigan, 980 F.2d 1330 (9th Cir.1992), upon which ONRC and the majority rely. In Madigan, the underlying appeal ruled that there was a clear statutory duty to issue regulations; on the attorneys’ fees appeal, therefore, both the district court and our court were bound by that holding. Here, we are bound by our earlier determination that EISS-1 did not take a hard enough look at cumulative impact, but since we did not hold that Corps had breached a clear duty, neither this court nor the district court is foreclosed from considering whether Corps’ position was substantially justified. The district court accordingly did not err in concluding that in light of the earlier studies to which EISS-1 referred, Corps’ position was “justified to a degree that could satisfy a reasonable person.” Pierce, 487 U.S. at 565, 108 S.Ct. at 2541.
The majority overreads Marsh IV, asserting that “[o]ur mandate that ‘Corps should consider the impact of Elk Creek Dam in conjunction with Lost Creek Dam and, if appropriate, Applegate River Dam,’ 832 F.2d at 1498, reflects our conclusion that the Corps had not addressed the cumulative impacts at all,” and that Marsh II held that “Corps completely omitted an area of analysis clearly required.” Maj. op. at 1492. The analysis of Marsh IV was limited to “[Plaintiffs .. two examples of the Corps failure to consider the cumulative impacts.” 832 F.2d at 1498. It is apparent from the discussion that the Marsh IV panel not only did not conclude that “the Corps had not addressed cumulative impacts at all, ” but that it never even considered the issue. Rather, the panel agreed that the two examples showed deficiencies in that analysis, and ordered that the Corps correct deficiencies in the cumulative analysis. This minimal statement is insufficient to bind the hands of the district court in determining whether Corps’ position was substantially justified.4
C
ONRC submits that the district court clearly erred in finding the Corps’ position substantially justified based on the 1974 and 1979 studies, because neither study addressed the cumulative impact of turbidity on fish production (although they did address the cumulative impact of turbidity on water quality). It suffices to quote the Supreme Court’s opinion in Marsh IV: “Based on these [1974 and 1979] studies, the FEISS predicted that changes in the ‘turbidity regime’ would not have any major effect on fish production, but that the combined effect of the Lost Creek and Elk Creek Dams on the turbidity of the Rogue River might, on occasion, impair fishing.” 490 U.S. at 365-66,109 S.Ct. at 1855. Whether or not the Corps’ report adequately addressed these cumulative impacts was a question on which reasonable minds could differ; I cannot, therefore, say that the district court abused its discretion in determining that the Corps’ position was substantially justified. Accordingly, I would affirm its denial of attorneys’ fees.
Ill
While I concur in the denial of the extraordinary relief sought by ONRC, I would deny the relief on the basis that no violation has been shown, and thus no remedy is warranted.
. Because I consider the environmental evaluation adequate, I do not discuss the appropriateness of the order to prepare a new supplement.
. I concur in the majority's discussion of the victories allegedly left intact by the Supreme Court's decision in Marsh TV.
. The majority chooses not to address ONRC's contention that the district court erred in considering its own earlier ruling. The district court was not in error for two reasons. First, the court's opinion acknowledges that its initial ruling is not dispositive. While a district court’s original agreement with a government position does not conclusively demonstrate the position was reasonable, Thomas v. Peterson 841 F.2d 332, 336 (9th Cir.1988), the failure of a government position to prevail does not establish a presumption that the position itself was unreasonable, Kali v. Bowen, 854 F.2d 329, 334 (9th Cir.1988); Pierce, 487 U.S. at 569, 108 S.Ct. at 2552 (“Conceivably, the Government could take a position that is not substantially justified, yet win; even more likely, it could take a position that is substantially justified, yet lose"). In any event, the court's opinion relies on what EISS-1 described and the extent to which it, in conjunction with the 1974 and 1979 Water Quality Stan*1498dards, addressed cumulative effects — not on its conclusion in Marsh I.
. The majority asserts that "the district court appears to believe that the Corps was required to prove substantial justification only with regard to the specific deficiencies identified by this court.” Maj. op. at 1492. However, the district court explicitly noted that we identified the two flaws as representative examples of general defectiveness, addressed the specific claimed examples of deficiency, and was “satisfied that the remaining deficiencies in the 1980 EISS are of a similar nature.” I do not, therefore, believe that the majority’s assessment of how the district court perceived its task is well founded.