A “Draft of a Draft” or a “Charade:” U.S. Fish & Wildlife Service v. Sierra Club
Sunshine Week, March 15-19, 2021, will soon be upon us. And just in time for its celebrants’ revelries (or perhaps simply virtual conferences and commemorations), the Supreme Court has handed down the major Freedom of Information Act (“FOIA”) decision expected this term, U.S. Fish & Wildlife Service v. Sierra Club, 2021 WL 816352. Alas, the decision furthers government secrecy. (I have previously discussed the Ninth Circuit’s initial opinion in this case in these pages here.)
The case arises out of an Environmental Protection Agency (EPA) proposal made a decade ago. In April 2011, the EPA proposed a rule regarding the design and operation of “cooling water intake structures.” Such structures siphon large volumes of water from natural sources to cool industrial equipment. Not surprisingly, in doing so they frequently trap and kill aquatic wildlife, including endangered species. That brought into play the Endangered Species Act of 1973, which requires federal agencies to consult with the two gatekeeper agencies charged with protecting endangered species, the U. S. Fish and Wildlife Service (FWS) and National Marine Fisheries Service (NMFS), before proceeding with any proposed action that would “adversely affect” a protected species. Slip op. at 1-2. (FWS and NMFS will collectively referred to as “the Services.”)
The Services must produce “biological opinions” declaring whether or not the agency’s proposal will jeopardize the continued existence of threatened or endangered species, opinions termed “jeopardy” or “no jeopardy” opinions. If the Services conclude that the action will cause “jeopardy,” they must propose “reasonable and prudent alternatives” that would avoid harming the threatened species. Once such a “jeopardy” biological opinion is issued, the consulting agency must either implement the Services’ reasonable and prudent alternatives, terminate the action altogether, or seek an exemption from the Endangered Species Committee. Slip op. at 2-3. The regulations provide that the Services can furnish the agency with a draft biological opinion, which they cannot finalize during a 45-day period provided for the agency to review the draft opinion. 50 CFR § 402.14(g)(5). (As we will see below, apparently the Services and consulting agencies can agree upon a shorter time period.)
In practice, the Services rarely need do more than deliver official draft jeopardy opinions to consulting agencies. These draft opinions enable the consulting agency to decide, without receiving a final jeopardy opinion, whether to abandon the proposed action or modify it so as to enable the Services to provide a “no jeopardy” assessment on the new, revised proposed course of action. For example, between 2008 and 2015, FWS engaged in 6,829 formal consultations, but issued only two final jeopardy opinions. See, Brief for Amici Curiae Center for Biological Diversity and Defenders of Wildlife in Support of Respondent 22–23, U.S. Fish & Wildlife Service v. Sierra Club, Dkt. No. 19-547 (July 31, 2020).
EPA’s consultation with the Services’ on its “cooling water intake” proposal appears to have fit that basic pattern. EPA submitted its April 2011 proposal to the Services. Thereafter, “[f]or almost two years, personnel from the agencies ‘met routinely, sometimes more than once a week,’ held ‘multiple conference calls,’ and ‘exchanged thousands of emails,’” Brief for the Petitioners (the Services) 8, U.S. Fish & Wildlife Service v. Sierra Club, Dkt. No. 19-547 (May 21, 2020)(Joint Appendix cites omitted). On November 26, 2013, EPA transmitted its revised proposal, which had undergone OMB review, for the Services to provide an official biological opinion on or before December 6, 2013. In December 2013, the Services sent an unofficial draft jeopardy assessment to EPA.
Even though, as typical in FOIA cases, the factual record consists largely of government officials’ affidavits, there is some dispute regarding the proper characterization of that December’s events.  In Justice Barrett’s recounting, staff members at NMFS and FWS completed a draft jeopardy opinion on December 6 and December 9 respectively, sent the drafts to the relevant decision-makers within each Service, and prepared to circulate the official draft jeopardy opinions to the EPA. Slip op. at 3. But the decision-makers did not approve the drafts, concluding that “more work needed to be done.” Id. In particular, because EPA was still engaged in an internal debate about key elements of the rule, the Services’ decision-makers wanted a better grasp of EPA’s proposal. Id. In this account, it is unclear who directed the forwarding of the unofficial draft jeopardy opinions to EPA, or for what purpose, even though Justice Barrett appears to acknowledge that unofficial draft jeopardy opinions were actually communicated to EPA.
In any event, the unofficial draft opinion, evidencing the Services considered position on the April 2011 proposal, was apparently the catalyst EPA needed to withdraw its proposal, on which a final jeopardy assessment would surely issue, and substantially revise the proposal so as to receive a no jeopardy assessment. Though the Services were prepared to send official versions the draft jeopardy opinions, EPA asked the Services to refrain from doing so. Brief for Respondent 14, U.S. Fish & Wildlife Service v. Sierra Club, Dkt. No. 19-547 (July 27, 2020)(citing the declarations of Samuel D. Rauch, III and Gary Frazer, at the time NMFS’ Acting Assistant Administrator and FWS’s Assistant Director for Ecological Services, respectively). The EPA’s strategy worked, and its revised proposal received a joint “no jeopardy” opinion from the Services.
The draft jeopardy opinion regarding EPA’s initial proposal, and several other related documents, were at issue in the U.S. Fish & Wildlife v. Sierra Club: the Sierra Club requested the documents pursuant to FOIA; the Services’ sought to withhold them pursuant to Exemption 5’s deliberative process exception. The Ninth Circuit ruled that the December draft jeopardy opinions, and most of the related documents, fell outside the deliberative process exception and must be produced. Sierra Club v. U.S. Fish & Wildlife Service, 925 F.3d 1000 (2019).
The Opinion of the Court
Justice Barrett authored the opinion for the seven-Justice majority. She began by noting that documents could be withheld under the deliberative process privilege if they were both pre-decisional and deliberative, but not if they embodied an agency’s final decision. Slip op. at 6 (citing NLRB v. Sears, Roebuck & Co., 421 U. S. 132, 150-152 (1975)), But Justice Barrett went further than this now-standard boilerplate language. Citing a decision Justice Kavanaugh authored when on the D.C. Circuit, National Security Archive v. CIA, 752 F.3d 460, 463 (D.C. Cir. 2014), Justice Barrett noted that “finality” is not merely a question of chronology. Slip op. at 6. In particular, a document does not qualify as final simply because nothing else follows it.” Id. Sometimes “a proposal dies on the vine;” thus, what matters “is not whether a document is last in line, but whether it communicates a policy on which the agency has settled.” Id.
To decide whether a document communicates an agency’s settled position, courts must consider whether the agency treats the document as its final view on the matter. In other words, does the document reflect “the ‘consummation’ of the agency’s decisionmaking process” rather than a “merely tentative” position? Or does the document “leave agency decisionmakers ‘free to change their minds.’” Slip op. at 6-7.
Justice Barrett concluded that the draft biological opinions at issue were not final, but merely reflected the Services’ preliminary views regarding the likely effect of the EPA’s proposed rule on endangered species. She noted that the Services had labelled the documents “drafts,” a denomination suggesting that the documents were subject to “feedback and change.” Slip op. at 7.
The administrative context confirmed that the opinions were subject to change. The regulation requiring the Services to provide a draft jeopardy opinion to the consulting agency, and guaranteeing that agency a fixed period before the Service could issue a final opinion, was particularly significant. The regulation “specifically contemplates further review by the agency after receipt of the draft, and with it, the possibility of changes” to the Services’ work. Moreover, the agreement between the Services and the EPA allowed for the possibility of post-circulation changes. To Justice Barrett, “[t]he logical inference [wa]s that the Services expected the EPA to provide comments that they might incorporate into the final opinion.” Slip op. at 7-8.
Justice Barrett then addressed the Sierra Club’s assertions the documents at issue “were actually intended to give the EPA a sneak peek at a conclusion that the Services had already reached,” and had the “operative effect” contemplated by the statute. In particular, the Endangered Species Act’s congressional progenitors had envisioned that the Services’ draft jeopardy opinions would prod a consulting agency like the EPA to revise its proposed action to avoid issuance of a final jeopardy opinion officially binding the it. Slip op. at 8. Acknowledging that prior opinions had “identified a decision’s ‘real operative effect’ as an indication of its finality,” Justice Barrett explained that those cases focused on the legal, not the practical, “consequences that flow from an agency’s action.” Indeed, she noted, “many documents short of a draft biological opinion could prompt an agency” proposing an action to alter its approach, and yet would be covered by the privilege nevertheless. To hold that any document that had such a decisive, practical effect fell outside the privilege’s scope would “gut” the privilege. Slip op. at 8-9.
Rather, Justice Barrett reiterated, the privilege’s applicability does not turn on whether the “drafts provoked a response from the EPA but [on] whether the Services treated them as final.” Slip op. at 9 (emphasis added). Though the Sierra Club had characterized the drafts as “polished documents lacking only an autopen signature”, Justice Barrett reiterated that the draft opinions “had been prepared by lower-level staff and sent to the Services’ decisionmakers” and that those decisionmakers had neither approved the drafts nor forwarded them to the EPA. She characterized each draft opinion as merely a “draft of a draft.” Slip op. at 10. That “the staff recommendations [had] prove[n] to be the last word within the Services about the 2013 version of the EPA’s proposed rule” was of no moment. The recommendations were not last because they had been finalized, but because they had died on the vine. Id. In her view, “[f]urther consultation with the Services prompted the EPA to alter key features of its 2013 proposal”, and thus “there was never a need for the Services to render a definitive judgment about it.” Id.
Justice Barrett also responded to the Sierra Club’s dire warning that upholding the agencies’ position would allow agencies to withhold final opinions under the deliberative process privilege by simply stamping them “draft.” She considered such fears unwarranted,observing that the finality inquiry “is a functional rather than formal” one. Id. More particularly, whenever the evidence establishes that an agency has hidden a functionally final decision in draft form, the deliberative process privilege will not apply. But, she explained, the Services, had not engaged in such a “charade.” Slip op. at 10-11.
Agreeing that function should take precedence over form, slip op. at 1, Justice Breyer dissented, for himself and Justice Sotomayor. He viewed the Services’ draft biological opinions as reflecting “final” decisions regarding the “jeopardy” the EPA’s initial proposal would have caused, and thus outside the deliberative process privilege. He enumerated five features of draft biological opinions that justified his conclusion.
First, the typical draft biological opinion is a “final” document with respect to its content, distinguishing it from “a draft of a draft.” Only if further deliberation about the draft’s content is not merely possible, but likely, would a draft biological opinion qualify as merely “a draft of a draft.” Though the Services could change their minds about a draft biological opinion, they could also do so with regard to a final biological opinion – theoretically they could withdraw such an opinion and issue a new one. In his view “the mere possibility of a future change does not alter the finality, or the final effect, of the original document.” Slip op at 2.
Second, final jeopardy opinions and draft jeopardy opinions serve the same functions within the administrative process, and have substantially the same effect upon the EPA. Both explain the Services’ findings. And both constrain EPA, leaving it a very limited range of choices. Simply put, transmitting the draft biological opinion to the EPA merely allows the EPA to make its choice before a final biological opinion issues. Slip op. at 2-4.
Third, “agency practice shows that the [d]raft [b]iological [o]pinion, not the [f]inal [b]iological [o]pinion, is the document that informs [consulting agencies] of the Services’ conclusions about jeopardy and alternatives and triggers within [the consulting agencies] the process of deciding how to respond.” Citing the Center for Biological Diversity and Defenders of Wildlife amicus brief, Justice Breyer noted the rarity of the issuance of a final jeopardy opinion.
Fourth, permitting release of draft biological opinions is “unlikely to chill frank discussion within the Services” because the Services’ staff already know that these drafts may well be made public. Opinions that relate to private entities’ application for a permit or some other agency action must be made available to the private entity, making deliberative process doctrine protection unavailable. Even with respect to other matters, consulting agencies have a long history of publicly disclosing draft biological opinions. Slip op. at 4-5.
Fifth, the legal consequences flowing from the Services’ completion of a draft biological opinion are identical to those that flow from a final biological opinion. Slip op. at 6-7.
Justice Breyer found the facts surrounding the biological opinion documents and their transmittal to EPA sufficiently murky to justify a remand for the courts below to engage in the “fact-intensive” inquiry regarding whether the material the Services transmitted to EPA were really works in progress or, instead, complete but for a signature. Slip op. at 6.
The majority opinion is quite formal, even while denying its own formality. Justice Barrett largely does not focus on the facts suggesting that the documents the Services transmitted to EPA were not intended to be the official draft jeopardy opinions endorsed by the relevant decision-makers. A focus on such facts might have made the opinion more cogent, but at the cost of severely limiting its significance.
Instead, Justice Barrett appears to be discussing the status of official draft jeopardy assessments under the Endangered Species Act’s unique statutory scheme. In doing so, she focuses on the abstract possibility of the Services’ changing its mind, not how realistic such a prospect is in actuality. Moreover, as an analysis of draft jeopardy opinions in general, Justice Barrett’s opinion fails to grapple with the core problem: that draft jeopardy opinions provide a means by which to confront a consulting agency with the choice of deciding whether it will adhere to the Services’ conclusions, conclusions that will become legally binding if the agency insists on proceeding without revising its approach. The Services’ draft opinions seem to operate exactly as final opinions were designed to operate, putting the agency to the choice of deciding to proceed against the views of the Services, or modifying their proposed actions.
Justice Barrett’s “finality” test also appears to introduce an element of subjectivity into the inquiry. A court must now determine whether the agency head, or the signer of the document, is contemplating the possibility of some further change to the document. However, Justice Barrett wisely limits this inquiry to an assessment of objective indicia of the agency’s amenability to change. This assessment of an agency’s amenability to changing its position will cause little trouble if the agency contemplates releasing the document to the public. In such a case, the document would presumably be final only when released to the public (though in such situations FOIA really doesn’t matter, at least in terms of access to the chronologically final documents). The inquiry is more difficult to pursue if the document is one the agency does not intend to make public but one which nevertheless is intended to constrain another actor’s choices, whether the actor is located within the agency or without. An example of a category of documents fitting this description are Office of Legal Counsel (OLC) opinions to agencies that OLC never plans to publicly disclose.
If the test of finality is to be “subjective,” the required “subjective assessment” should be one more directly relevant to the deliberative process privilege, namely whether release of the document will likely chill subordinate or agency head candor. In the case of draft jeopardy opinions generally (even if perhaps not in the potentially unique circumstances of the case before the court), the prospect of the release of the draft jeopardy assessment will hardly chill the candor of anyone at the Service. Rather, the Services’ staff members and decision-makers will know that the draft will be subject to intense scrutiny by the consulting agency and could become the source of an inter-agency controversy that could prove damaging to the agency.
And to the extent that the risk of public disclosure of draft jeopardy assessments makes agencies cautious in rendering draft biological opinions, that concern is salutary, at least that is the judgment underlying FOIA. When agencies makes and explains official decisions, they should do so with the realization that at least the decision-makers should be publicly accountable for their actions. As Louis Brandeis observed before his appointment to the Supreme Court, “[s]unlight is said to be the best of disinfectants; electric light the most efficient policeman.” Louis D. Brandeis What Publicity Can Do, HARPER’S WEEKLY (1913) (later reprinted in LOUIS D. BRANDEIS, OTHER PEOPLE’S MONEY: AND HOW THE BANKERS USE IT (1914)).
Finding that most draft biological statements are not exempt from FOIA, and thus subject to FOIA requests, would certainly seem to further FOIA’s core purpose, namely enabling the public to inform itself about what their Government “is up to.” U.S. v. Reporters Committee for Freedom of the Press, 489 U.S. 749, 772-73 (1989). Congress entrusted the Services with an unusual gatekeeping role with regard to the Endangered Species Act’s implementation. The public can hardly monitor the Services’ performance of such a relatively formal, congressionally mandated gatekeeping role in preserving endangered species without being able to obtain 99+ percent of the Services’ “jeopardy” assessments. But because draft jeopardy opinions have routinely proven successful at forcing consulting agencies to modify their plans, those critical documents, a core part of the Services’ work, are shielded from public view.
In addition, perhaps the reason consulting agencies so often disclose the draft biological opinions, see slip op. at 5 (Breyer, J., dissenting) may stem from the challenge consulting agencies face in honestly accounting for their own change in plans. A consulting agency may find it difficult explain its changed approach to an issue without revealing that it did not really change its own views, but acceded to compulsion from an external source, a draft jeopardy opinion issued by one or both of the Services reflecting their contrary views on the matter.
The Supreme Court’s major FOIA cases in the 2000’s have exhibited an increasing devotion to textualism. However, in some respects exemption differs from many of the other exemptions, because it incorporates common law (i.e., non-statutorily-defined) privileges. The hallmark of common law development of legal doctrine is its potential for flexibility and adaptability. Thus, interpreting the deliberative process privilege by considering its purposes and its practical effects is less controversial than taking the same approach for statutorily-defined FOIA exemptions.
No statutory text defines the deliberative process exception, nor imposes obstacles limiting judicial elaboration of the doctrine. The common law nature of the deliberative process privilege certainly allows the court to distinguish “finality” for purposes of judicial review from “finality” for purposes of the deliberative process privilege. And it certainly allow a court to distinguish situations when an agency project dies on the vine because the agency changes its mind from instances where another agency intentionally moots completion of the process by voluntarily acquiescing to the agency’s position.
Finally, as Dan Farber and Anne Joseph O’Connell have noted, there are a variety of interrelationships between agencies, characterized by varying degrees of formality. Daniel A. Farber, Anne Joseph O’Connell, Agencies as Adversaries, 105 Calif. L. Rev. 1375 (2017). The relationship between the Services and consulting agencies is a fairly formal one; the Services hold the power to require agencies to alter plans that the Services conclude have an adverse effect on endangered species. It is a statutory regime involving on agency’s issuance of formal opinions that another agency must either formally accept or seek authorization to override. Other intra-agency relationships are much less formal.
Thus, U.S. Fish and Wildlife v. Sierra Club could be a harbinger of future cases in which courts will have to apply FOIA to multi-agency processes. In such multi-agency processes, each agency’s official expression of its position could be viewed separately from the entire deliberative process among government agencies. Or each agency’s determination of its own could be viewed merely as a part of the larger deliberative process involving multiple government agencies. When is one approach more appropriate than the other in terms of the application of the deliberative process privilege incorporated by FOIA? I discussed this question in my January 2, 2019 blogpost on this case here.
In any event, to the devotees of transparency out there, best wishes for an enjoyable Sunshine Week!
 The process is more extensively discussed in Jacob W. Malcom & Ya-Wei Li, Data Contradict Common Perceptions about a Controversial Provision of the US. Endangered Species Act, 112 PNAS 15844-15849 (Dec. 29, 2015), cited by Biological Diversity and Defenders of Wildlife in their joint amicus brief..
 The Ninth Circuit described the process as follows:
NMFS completed its draft jeopardy opinion on December 6, 2013 and FWS completed its draft jeopardy opinion on December 9, 2013, both for transmission to the EPA. The Services sent the EPA portions of its December 2013 draft jeopardy opinions, but never formally transmitted them in their entirety.
On December 12, 2013, the FWS Deputy Solicitor called and emailed the EPA General Counsel to “touch base . . . about transmitting a document to EPA.” He also emailed “the current draft RPAs” [“reasonable and prudent alternatives”] to the EPA that same day. On December 17, 2013, the NMFS sent a “Revised Combined NMFS and FWS RPA” to the EPA. The Services have further indicated in their briefing that they also provided other unspecified portions of the draft jeopardy opinions to the EPA.
Sierra Club v. U.S. Fish & Wildlife Service, 925 F.3d 1000, 1008 (9th Cir. 2019). For a timeline of the relevant events as well as a full listing of the documents at issue see the Ninth Circuit’s decision. 925 F.3d at 1008-1010.
 Interestingly, in the direct challenge to the rule, the Second Circuit held that Sierra Club could not obtain the documents via civil discovery because they were protected by the deliberative process privilege. Cooling Water Intake Structure Coal. v. EPA, 905 F.3d 49, 65 n.9 (2d Cir. 2018).
 National Security Archive involved five-volume official history of the CIA that was abandoned after three volumes, even though a draft of the fourth volume had been released, and a draft of fifth had been completed by the CIA’s consultant. The CIA changed its own mind regarding whether to see the project through to completion.
Moreover, National Security Archive adopts a particularly expansive view of the scope and absoluteness of the deliberative process privilege, 752 F.3d 463 (“[a] privilege contingent on later events—such as whether the draft ultimately evolved into a final agency position—would be an uncertain privilege, and as the Supreme Court has said, an uncertain privilege is ‘little better than no privilege at all’”). Interestingly, the FOIA Improvement Act of 2016, Pub. L. 114-185, 130 Stat. 538, overruled part of the holding in National Security Archive, establishing a 25-year limit on the applicability of the deliberative process privilege to a agency records. In addition, the Act’s incorporation of the “foreseeable harm” standard for discretionary release of protected documents seem at odds with (and perhaps flatly contradicts) then-Judge Kavanaugh’s discussion in National Security Archive, 752 F.3d 463-64.
 Of course, some documents that have such a decisive practical effect, forcing the agency to change its approach despite not having changed its mind (due to some external constraint imposed by another agency), would often fall outside the privilege because such documents would almost certainly have to be incorporated by reference in the constrained agency’s final decision.
 It would have been more accurate to say that EPA’s initial proposal had died on the vine. And indeed, that had happened because EPA realized that the Services’ would not change their position, reflected in the unofficial draft jeopardy opinions communicated to EPA, that EPA’s initial proposal would jeopardize endangered species unless certain ameliorative actions were taken.
 Such a course might often be impractical or inadvisable for other reasons. In some circumstances, agencies may be compelled to issue official final documents to comply with statutory mandates or for the policies encoded in the document to have legal effect. Moreover, policies issued as drafts are more susceptible to “repeal” or revision by a succeeding political administration.
 Ironically, though left unmentioned by Justice Breyer, the Ninth Circuit had issued a final opinion in Sierra Club v. U.S. Fish & Wildlife, only to withdraw it an issue a modified final opinion (or, if you wish, a final final opinion).
 I have discussed OLC opinions extensively in the context of FOIA’s affirmative disclosure obligations. Making Soup from a Single Oyster? CREW v. DOJ and the Obligation to Publish Office of Legal Counsel Opinions (three part series), YALE J. ON REG.: NOTICE & COMMENT (May 13, 16, 21, 2019), accessible here, here, and here; “Office of Legal Counsel Opinions: Advice or Law?” Administrative & Regulatory Law News, Vol. 44 No.4 (Summer 2019).
 The request at issue predates the FOIA Improvement Act of 2016. Presumably, for requests that post-date that Act, the agency will have to conduct a foreseeable harm analysis that will presumably involve making a realistic assessment of whether the release of such documents will chill the candor of subordinates, and not merely assume a record will do so just because it falls within the deliberative process privilege’s parameters.
 Th Court quoted EPA v. Mink, 410 U.S. 73, 105 (1973)(Douglas, J., dissenting), which in turn quoting Henry Steele Commager, The Defeat of America, THE NEW YORK REVIEW OF BOOKS, Oct. 5, 1972, at 7.
 This is in some ways the mirror image of the Department of Commerce’s problem in Department of Commerce v. New York, 588 U.S. ___, 139 S. Ct. 2551 (2019) where the Department sought to claim that its decision to include a citizenship question on the census form originally externally, from the Civil Rights Division of the Department of Justice, rather than internally. Id. at 2575. As the Court explained
Altogether, the evidence tells a story that does not match the explanation the Secretary gave for his decision. In the Secretary’s telling, Commerce was simply acting on a routine data request from another agency. Yet the materials before us indicate that Commerce went to great lengths to elicit the request from DOJ (or any other willing agency).
 Food Marketing Institute v. Argus Leader Media, —U.S. —, 139 S.Ct. 2356 (2019) (narrowing the definition of confidential documents for purposes of Exemption 4); Milner v. Department of the Navy, 562 U.S. 562, 569-76 (2011) (rejecting the “high-2” Exemption); FCC v. AT&T, 562 U.S. 397 (2011)(holding that corporations cannot assert an interest in personal privacy under Exemption 7(c)); Department of Interior v. Klamath Water Users Protective Assn, 532 U.S. 1, 12 (2001)(questioning the “consultant’s corollary” under Exemption 5).
 See, e.g., Hunton & Williams LLP v. U.S. Environmental Protection Agency, 248 F.Supp.3d 220 (D.D.C. 2017). There the FOIA requester sought the Army Corps of Engineers’ final draft of its Approved Jurisdiction Determination under the Clean Water Act. EPA invoked its authority to take jurisdiction from the Corps and make the Determination itself, and as of the date of the FOIA request had not issued a Determination of its own. The Judge observed that a “deliberative process can span between two different agencies,” noting that such arrangements are “particularly common when one agency serves a secondary function to a supervising agency.” Id. at 247, Because EPA retained jurisdiction to make CWA determinations, the Corps’ submissions to EPA were a part of EPA’s deliberative process. Id. “[R]evealing the Corps’ reasoning could damage the open dialogue between the two collaborating agencies.” Id. at 248.