Plaintiffs Hugh Kern, Leigh Ann Lipscomb, and the Oregon Natural Resources Council Fund (collectively, "ONRC") appeal the district court's grant of summary judgment to defendant United States Bureau of Land Management ("BLM") and defendants-intervenors Douglas Timber Operators, Herbert Lumber Co., and Lone Rock Timber Co. (collectively, "the timber companies"). ONRC's suit involves BLM action in the Coos Bay District, a BLM district along the southwest coast of Oregon.

ONRC contends that the BLM has failed to discharge its obligations under the National Environmental Policy Act ("NEPA"), 42 U.S.C. § 4321 et seq., by twice failing in its decision process to consider adequately the impact of a pathogenic root fungus, Phytophthora lateralis ("the fungus"), on the Port Orford Cedar ("the Cedar"). First, ONRC challenges the adequacy of the Environmental Impact Statement ("EIS") prepared for the Coos Bay Resource Management Plan ("RMP") for the Coos Bay District. Second, ONRC challenges the adequacy of the Environmental Assessment ("EA") prepared for proposed timber sales in the Sandy-Remote Analysis Area within the Coos Bay District. Both ONRC and the defendants moved for summary judgment on ONRC's claims that the EIS and EA were inadequate under NEPA.

The district court dismissed ONRC's challenge to the EIS as unripe without reaching the merits, and rejected ONRC's challenge to the EA on the merits. It found that the EA adequately addressed the impact that the timber sales would have on the spread of the fungus to the Cedar, and it granted summary judgment to the BLM. For the reasons that follow, we reverse the district court's rulings on both the EIS and the EA, and direct that summary judgment be entered for ONRC. We hold that the challenge to the EIS was ripe, and that the EIS is inadequate under NEPA. We also hold that the EA is inadequate under NEPA.

The National Environmental Policy Act has "twin aims. First, it places upon [a federal] agency the obligation to consider every significant aspect of the environmental impact of a proposed action. Second, it ensures that the agency will inform the public that it has indeed considered environmental concerns in its decisionmaking process." Baltimore Gas & Elec. Co. v. Natural Res. Def. Council, Inc., 462 U.S. 87, 97, 103 S. Ct. 2246, 76 L. Ed. 2d 437 (1983) (citation and internal quotation marks omitted). NEPA does not contain substantive environmental standards. Rather, it "establishes `action-forcing' procedures that require agencies to take a `hard look' at environmental consequences." Metcalf v. Daley, 214 F.3d 1135, 1141 (9th Cir.2000); see also Robertson v. Methow Valley Citizens Council, 490 U.S. 332, 348, 109 S. Ct. 1835, 104 L. Ed. 2d 351 (1989).

NEPA requires federal agencies to prepare an EIS prior to taking "major Federal actions significantly affecting the quality" of the environment. 42 U.S.C. § 4332(2)(C).1 Some proposed federal actions categorically require the preparation of an EIS. If the proposed action does not categorically require the preparation of an EIS, the agency must prepare an EA to determine whether the action will have a significant effect on the environment. See 40 C.F.R. § 1501.4(Council on Environmental Quality ("CEQ") regulations implementing NEPA); Metcalf, 214 F.3d at 1142. If the EA reveals that the proposed action will significantly affect the environment, then the agency must prepare an EIS. If the EA reveals no significant effect, the agency may issue a Finding of No Significant Impact ("FONSI"). See 40 C.F.R. §§ 1501.4, 1508.9; see also Metcalf, 214 F.3d at 1142.

B. The Coos Bay Environmental Impact Statement

The Federal Land Policy and Management Act ("FLPMA"), 43 U.S.C. § 1701 et seq., requires the BLM to prepare RMPs for the various districts under its control. See 43 U.S.C. § 1712. By definition, preparation of an RMP is a "major Federal action significantly affecting the quality of the human environment," and so categorically requires preparation of an EIS. See 43 C.F.R. § 1601.0-6(BLM regulations implementing FLPMA). In 1994, the BLM published an EIS for the proposed RMP for the Coos Bay District.

The Coos Bay District is within the geographic range of the Port Orford Cedar. The Cedar is a valuable component of forest ecosystems in southwestern Oregon and northwestern California and is susceptible to infection by the root fungus Phytophthora lateralis. The fungus may be spread in a number of ways, and is usually fatal to infected trees. PL can be transmitted by surface water in streams or ditches. New infections can also occur if soil infested with PL spores is transported to uninfected areas, for example in mud clinging to vehicles, pedestrians, and animals. Human activities that facilitate the spread of the fungus include timber cutting, road construction and maintenance, off-road vehicle use, livestock grazing, and commercial cedar bough and mushroom collection. See Northcoast Envtl. Ctr. v. Glickman, 136 F.3d 660, 662-63 (9th Cir.1998).

The plaintiffs assert that the EIS prepared in connection with the Coos Bay District RMP did not adequately discuss the effect of the fungus on Port Orford Cedar. Although the EIS mentioned the fungus and the Cedar, the EIS's discussion was limited to the statement that the BLM will: Conform all management activities within the range of Port-Orford-cedar to the guidelines described in the BLM Port-Orford-cedar Management Policies to mitigate damage caused by Phytophthora lateralis. Site-specific analysis for projects within the range of Port-Orford-cedar will consider possible effects on the species.

In May 1995, the BLM approved the Coos Bay RMP, supported by the EIS. The RMP now governs projects within the Coos Bay District.

C. The Sandy-Remote Environmental Assessment

In 1996, the BLM proposed timber sales within a subsection of the Coos Bay District known as the "Sandy-Remote Analysis Area." Timber sales do not categorically require preparation of an EIS. An EA covering the Sandy-Remote Analysis Area was prepared in conjunction with the timber sale proposal. The spread of the fungus among Port Orford Cedar was one of the issues identified in the EA, but that issue was eliminated from the analysis. Using language echoing the Coos Bay EIS, the Sandy-Remote EA stated only that "following the guidelines established in the Port Orford Cedar Management Guidelines (BLM 1994) should reduce the spread of the root rot disease and that [following the Guidelines] would not have a significant adverse impacts [sic] to the resources." A FONSI based on the EA for the Sandy-Remote Analysis Area was signed on November 5, 1996. A decision document based on the EA and FONSI was prepared on February 11, 1997. The BLM then entered into eight timber sales based upon the EA and FONSI.

After this suit was filed and after several of the timber sales occurred,2 the BLM revised the Sandy-Remote EA to include Section S, which discussed the impact of the timber sales on the spread of the fungus within the Sandy-Remote Area. Based on the revised EA, the BLM issued a new FONSI on July 14, 1998, finding again that the actions did not constitute a major federal action requiring an EIS. Based on the revised EA and new FONSI, the BLM issued another decision document allowing logging to go forward pursuant to the timber sales that had already occurred.

D. The Port Orford Cedar Management Guidelines

The plaintiffs timely commented on the Coos Bay RMP and its accompanying EIS, on the Sandy-Remote EA, and on the revised Sandy-Remote EA. Plaintiffs' main concerns centered on the fact that both the EIS and the EA referred to a document entitled the "Port Orford Cedar Management Guidelines" ("the Guidelines") but contained no analysis of the impact of the proposed RMP or proposed timber sales on the spread of the fungus and the effect of this spread on the Cedar. The Sandy-Remote EA was revised to respond to the concerns of plaintiffs and others by, inter alia, adding Section S, which provides some analysis of the effect of the proposed timber sales on the fungus and the Cedar. However, the revised EA also continued to refer to and rely on the Guidelines.

The BLM completed the Guidelines in 1994. The Guidelines describe strategies to minimize the spread of the fungus. These strategies have been incorporated by reference into a number of EISs and EAs in addition to the EIS for the Coos Bay RMP and the EAs for the Sandy-Remote Area, but the Guidelines themselves have never been subject to NEPA review. In 1995, a group of environmental organizations, including ONRC, filed suit alleging that the Guidelines were subject to NEPA review and required preparation of an EIS. In 1998, in Northcoast Environmental Center, we held that NEPA review and an EIS were not necessary for the Guidelines. We held first that the Guidelines did not constitute "final agency action" as required by the general review provision of the Administrative Procedure Act. See 136 F.3d at 669-70; see also 5 U.S.C. § 704; Lujan v. Nat'l Wildlife Fed'n, 497 U.S. 871, 882, 110 S. Ct. 3177, 111 L. Ed. 2d 695 (1990). Additionally, we held that the promulgation of the Guidelines did not constitute a "major federal action" as required by § 102(2)(C) of NEPA, 42 U.S.C. § 4332(2)(C). See Northcoast, 136 F.3d at 665.

However, we noted in Northcoast that although the Guidelines in and of themselves were not subject to NEPA, if they were incorporated into a specific agency action they would be subject to NEPA in the context of that action. We noted that "[l]ong-range aims are quite different from concrete plans and specific undertakings such as the ... BLM's Resource Management Plans which the Secretaries have submitted for purposes of environmental analysis under NEPA." Id. at 668 (emphasis added). We also noted that we were not creating a "`catch 22'" situation in which the Guidelines would escape review under NEPA:

There is no reason plaintiffs cannot challenge the sufficiency of an agency EIS when a discrete agency action is called for. The agencies will be unable to shield their [Port Orford Cedar] program from NEPA review because they will not be able to avail themselves of the Council on Environmental Quality's "tiering" provision. 40 C.F.R. § 1502.20.... Furthermore, the Secretaries have stated their intentions to prepare an EIS when they propose to implement particular control strategies with environmental impacts. As we stated in [Salmon River Concerned Citizens v. Robertson, 32 F.3d 1346 (9th Cir.1994)], judicial estoppel will prevent the Secretaries from arguing they have no further duty to consider their [Port Orford Cedar] management policies when site-specific programs are challenged. 32 F.3d at 1357-58. "We assume that government agencies will ... comply with their NEPA obligations in later stages of development." Id. at 1358

136 F.3d at 670 (emphasis added).

A "concrete plan," a "specific undertaking," and a "site-specific program" incorporating the Guidelines, such as we anticipated in Northcoast, are now before us. Plaintiffs challenge the sufficiency under NEPA of both the EIS for the Coos Bay RMP and the revised EA for the eight timber sales within the Sandy-Remote Area, both of which refer to the Guidelines. ONRC argues that the EIS and the revised EA are inadequate under NEPA because rather than analyzing the impact of the fungus on the Port Orford Cedar, they merely refer to the Guidelines, which have never been subjected to NEPA review. ONRC also argues that the revised EA is deficient because it relies in part on the Guidelines, and because its free-standing analysis of the fungus and the Cedar, considered alone, is inadequate.

We have jurisdiction pursuant to 28 U.S.C. § 1291. The district court's determination on summary judgment that the BLM complied with NEPA is reviewed de novo. Okanogan Highlands Alliance v. Williams, 236 F.3d 468, 471 (9th Cir.2000). Ripeness is a question of law reviewed de novo. Natural Res. Def. Council v. Houston, 146 F.3d 1118, 1131(9th Cir.1998). An agency's threshold decision that certain activities are not subject to NEPA is reviewed for reasonableness. See Northcoast, 136 F.3d at 667. An agency's decision not to prepare an EIS once that agency has prepared an EA is reviewed for abuse of discretion, and will be set aside only if it is "arbitrary and capricious." Marsh v. ONRC, 490 U.S. 360, 376-77, 109 S. Ct. 1851, 104 L. Ed. 2d 377 (1989); see also Okanogan Highlands, 236 F.3d at 471.

II. Ripeness of the Challenge to the Coos Bay EIS

Defendants BLM and the timber companies argue that Ohio Forestry Ass'n v. Sierra Club, 523 U.S. 726, 118 S. Ct. 1665, 140 L. Ed. 2d 921 (1998), precludes ONRC's challenge to the adequacy of the Coos Bay EIS. In Ohio Forestry, the Court held that a challenge under the National Forest Management Act ("NFMA") to an RMP promulgated by the Forest Service was unripe for judicial decision. See id. at 732, 118 S. Ct. 1665. The Court considered "(1) whether delayed review would cause hardship to the plaintiffs; (2) whether judicial intervention would inappropriately interfere with further administrative action; and (3) whether the courts would benefit from further factual development of the issues presented." Id. at 733, 118 S. Ct. 1665; see also Abbott Labs. v. Gardner, 387 U.S. 136, 149, 87 S. Ct. 1507, 18 L. Ed. 2d 681 (1967). First, the Court found no hardship to the plaintiffs in delaying review because the RMP at issue was merely a general plan, rather than an authorization of specific action. That is, the plan itself did not "give anyone a legal right to cut trees," Ohio Forestry, 523 U.S. at 733, 118 S. Ct. 1665, and the plan itself could not harm the plaintiffs. The Court noted that a later challenge to a specific action "might also include a challenge to the lawfulness of the present Plan if (but only if) the present Plan then matters, i.e., if the Plan plays a causal role with respect to the future, then-imminent, harm from logging." Id. at 734, 118 S. Ct. 1665. The Court found that the second and third factors weighed against review as well, because interference with the agency's procedures was premature, and because the issue was presently too abstract for judicial review. See id. at 735-37.

Defendants urge that Ohio Forestry forecloses our review of ONRC's NEPA challenge to the Coos Bay EIS because, like the plan in Ohio Forestry, the challenged plan in this case is an RMP. The defendants argue that the RMP prepared under FLPMA in this case is analogous to the RMP prepared under NFMA in Ohio Forestry. However, this case differs from Ohio Forestry in a critical respect. The plaintiffs in Ohio Forestry alleged that the RMP violated NFMA, the statute that required the preparation of the RMP. They did not allege that the RMP violated NEPA. The plaintiffs in this case, however, do not allege that the RMP violates FLPMA. (FLPMA is the equivalent, in this case, to NFMA in Ohio Forestry.) Rather, they allege that the EIS, prepared in conjunction with the RMP, violates NEPA. Because the plaintiffs here bring a NEPA challenge to an EIS, rather than a NFMA (or a FLPMA) challenge to an RMP, they are able to show an imminence of harm to the plaintiffs and a completeness of action by the agency that the Court held were missing in Ohio Forestry.

A NEPA challenge to an EIS is fundamentally unlike a NFMA (or FLPMA) challenge to an RMP. As the Court explained in Ohio Forestry, "NEPA, unlike the NFMA, simply guarantees a particular procedure.... [A] person with standing who is injured by a failure to comply with the NEPA procedure may complain of that failure at the time the failure takes place, for the claim can never get riper." Id. at 737, 118 S. Ct. 1665. The rights conferred by NEPA are procedural rather than substantive, and plaintiffs allege a procedural rather than a substantive injury. See Methow Valley Citizens, 490 U.S. at 350, 109 S. Ct. 1835 ("NEPA itself does not mandate particular results, but simply prescribes the necessary process."); Muckleshoot Indian Tribe v. USFS, 177 F.3d 800, 814 (9th Cir.1999); Inland Empire Pub. Lands Council v. USFS, 88 F.3d 754, 758 (9th Cir.1996) ("NEPA exists to ensure a process, not to ensure any result.") (emphasis in original). If there was an injury under NEPA, it occurred when the allegedly inadequate EIS was promulgated. That is, any NEPA violation (and any procedural injury) inherent in the promulgation of an inadequate EIS for the Coos Bay RMP have already occurred. See West v. Sec'y of Dep't of Transp., 206 F.3d 920, 930 n. 14 (9th Cir.4737 2000) (noting Ohio Forestry's distinction between a NEPA claim, involving "imminent and certain injury" and a NFMA claim); see also Wilderness Soc'y v. Thomas, 188 F.3d 1130, 1133, n. 1 (9th Cir.1999) (noting that Ohio Forestry had "acknowledged that a forest plan could always be challenged under [NEPA]"). Further, adjudicating ONRC's NEPA claim now will not "inappropriately interfere with further administrative action," see Ohio Forestry, 523 U.S. at 733, 118 S. Ct. 1665, because the BLM has already promulgated the EIS for the Coos Bay RMP. The plaintiffs' claim is therefore ripe for review.

Our holding comports with the Seventh Circuit's decision in Heartwood, Inc. v. USFS, 230 F.3d 947, 952-53 (7th Cir.2000). In Heartwood, the plaintiffs challenged the Forest Service's failure to prepare either an EIS or an EA prior to promulgating categorical exclusions from NEPA for certain classes of Forest Service actions. Id. at 950-51. The Forest Service contended that the plaintiffs' claim was not ripe under Ohio Forestry because it did not pertain to authorization of a "specific project," but the Seventh Circuit disagreed. Id. at 952. Noting that the Supreme Court in Ohio Forestry had explicitly distinguished a NEPA claim from a NFMA claim, the court explained that "a plaintiff clearly has standing to sue where there is a concrete injury underlying the procedural default even if the plan is not implemented immediately." Id. at 953 (emphasis added) (citation and internal quotation marks omitted). Therefore, the Heartwood plaintiffs were not barred by Ohio Forestry from asserting their NEPA claim, because they "need not wait to challenge a specific project when their grievance is with an overall plan." Id. at 953.

ONRC argues that the EIS for the Coos Bay RMP is inadequate in that it merely refers to the Guidelines, which have never been analyzed in an EIS. The Coos Bay EIS makes no attempt itself to analyze the effect of the fungus on the Port Orford Cedar.

In reviewing the adequacy of an EIS, we employ a "rule of reason to determine whether the EIS contains a `reasonably thorough discussion of the significant aspects of probable environmental consequences.' [Or. Natural Res. Council v. Lowe, 109 F.3d 521, 526 (9th Cir.1997)]. Under this standard, review consists only of insuring that the agency took a `hard look.' Id." Neighbors of Cuddy Mountain v. USFS, 137 F.3d 1372, 1376 (9th Cir. 1998). "The rule of reason analysis and the review for an abuse of discretion are essentially the same." Id. We examine the adequacy of the EIS using an objective good faith standard. See Coalition for Canyon Pres. v. Bowers, 632 F.2d 774, 782 (9th Cir.1980).

As a threshold matter, defendants argue that NEPA did not obligate the BLM to undertake a detailed environmental analysis of the fungus and the Port Orford Cedar when it prepared the EIS for the Coos Bay RMP. The defendants argue that the EIS merely analyzes the resource management plan, rather than a specific action, and that analysis of the fungus and the Cedar is not required until the time a federal agency makes an irreversible and irretrievable commitment of resources. We disagree.

Federal regulations require preparation of an EIS in conjunction with the preparation of any RMP. See 43 C.F.R. § 1601.0-6 ("Approval of a resource management plan is considered a major Federal action significantly affecting the quality of the human environment. The environmental analysis of alternatives and the proposed plan shall be accomplished as part of the resource management planning process ..."). An agency may not avoid an obligation to analyze in an EIS environmental consequences that foreseeably arise from an RMP merely by saying that the consequences are unclear or will be analyzed later when an EA is prepared for a site-specific program proposed pursuant to the RMP. "[T]he purpose of an [EIS] is to evaluate the possibilities in light of current and contemplated plans and to produce an informed estimate of the environmental consequences.... Drafting an [EIS] necessarily involves some degree of forecasting." City of Davis v. Coleman, 521 F.2d 661, 676 (9th Cir.1975) (emphasis added). If an agency were able to defer analysis discussion of environmental consequences in an RMP, based on a promise to perform a comparable analysis in connection with later site-specific projects, no environmental consequences would ever need to be addressed in an EIS at the RMP level if comparable consequences might arise, but on a smaller scale, from a later site-specific action proposed pursuant to the RMP.

Once an agency has an obligation to prepare an EIS, the scope of its analysis of environmental consequences in that EIS must be appropriate to the action in question. NEPA is not designed to postpone analysis of an environmental consequence to the last possible moment. Rather, it is designed to require such analysis as soon as it can reasonably be done. See Save Our Ecosystems v. Clark, 747 F.2d 1240, 1246 n. 9 (9th Cir.1984) ("Reasonable forecasting and speculation is ... implicit in NEPA, and we must reject any attempt by agencies to shirk their responsibilities under NEPA by labeling any and all discussion of future environmental effects as `crystal ball inquiry,'" quoting Scientists' Inst. for Pub. Info., Inc. v. Atomic Energy Comm'n, 481 F.2d 1079, 1092 (D.C.Cir.1973)). If it is reasonably possible to analyze the environmental consequences in an EIS for an RMP, the agency is required to perform that analysis. The EIS analysis may be more general than a subsequent EA analysis, and it may turn out that a particular environmental consequence must be analyzed in both the EIS and the EA. But an earlier EIS analysis will not have been wasted effort, for it will guide the EA analysis and, to the extent appropriate, permit "tiering" by the EA to the EIS in order to avoid wasteful duplication.

It is clear that the EIS for the Coos Bay RMP should have included an analysis of the likely impact of the RMP on the fungus and the Port Orford Cedar. This environmental problem was readily apparent at the time the EIS was prepared. (Indeed, it was apparent several years before, as evidenced by the earlier preparation of the Guidelines.) The RMP contained enough specifics to permit productive analysis of the fungus and the Cedar, including proposals for alternative ways of dealing with the problem. We therefore inquire whether the analysis in the EIS for the Coos Bay RMP is adequate under NEPA.

ONRC argues that the EIS is inadequate because its only discussion of the fungus and the Cedar is contained in a brief reference to the Guidelines, and because the EIS illegally "tiers" to the Guidelines. Tiering, or avoiding detailed discussion by referring to another document containing the required discussion, is expressly permitted by federal regulation:

Agencies are encouraged to tier their environmental impact statements to eliminate repetitive discussions of the same issues and to focus on the actual issues ripe for decision at each level of environmental review. Whenever a broad environmental impact statement has been prepared (such as a program or policy statement) and a subsequent statement or environmental assessment is then prepared on an action included within the entire program or policy (such as a site specific action) the subsequent statement or environmental assessment need only summarize the issues discussed in the broader statement and incorporate discussions from the broader statement by reference and shall concentrate on the issues specific to the subsequent action.

40 C.F.R. § 1502.20.

However, tiering to a document that has not itself been subject to NEPA review is not permitted, for it circumvents the purpose of NEPA. While NEPA empowers neither the plaintiffs nor this court to second-guess the BLM's management decisions, it does require the BLM to articulate, publicly and in detail, the reasons for and likely effects of those management decisions, and to allow public comment on that articulation. See Methow Valley Citizens, 490 U.S. at 349, 109 S. Ct. 1835 ("Publication of an EIS ... gives the public the assurance that the agency has indeed considered environmental concerns in its decisionmaking process."). Although the Guidelines may contain a detailed analysis of the impact of the fungus on the Port Orford Cedar, the BLM is not excused from its responsibility under NEPA to perform an analysis of the effects of the fungus on the Cedar in an EIS specifically addressed to the Coos Bay RMP. See 42 U.S.C. § 4332(2)(C)(i), (iii).

We specifically noted in Northcoast that the Guidelines would not be indefinitely shielded from NEPA review: "Although CEQ procedures allow agencies to incorporate by reference certain materials to cut down on the bulk of an EIS, they cannot `tier' their site-specific EISs to the broader [Port Orford Cedar] program where the program itself has not been subject to NEPA procedures." 136 F.3d at 670. We cautioned additionally that "judicial estoppel will prevent the Secretaries from arguing they have no further duty to consider their [Port Orford Cedar] management policies when site-specific programs are challenged." Id. We now hold, as we warned in Northcoast that we would, that the EIS for the Coos Bay RMP may not tier to the Guidelines, for which an EIS has never been prepared.

Because the Coos Bay EIS may not tier to the Guidelines, its adequacy depends on the analysis contained in the EIS itself. The sum total of the analysis in the EIS is the statement that the BLM will:

Conform all management activities within the range of Port-Orford-cedar to the guidelines described in the BLM Port-Orford-cedar Management Policies to mitigate damage caused by Phytophthora lateralis. Site-specific analysis for projects within the range of Port-Orford-cedar will consider possible effects on the species.

This two-sentence statement is obviously inadequate. The first sentence is an impermissible attempt to tier the analysis to the Guidelines. The second sentence is not an analysis, but rather a promise of a later site-specific analysis to be performed in connection with specific projects "within the range of the Port-Orford-cedar." The revised EA for the Sandy-Remote Analysis Area is such a site-specific analysis. The adequacy of that EA has also been challenged by ONRC. We now turn to that question.

IV. Adequacy of the Sandy-Remote Environmental Assessment

The district court held that the revised EA adequately analyzed the environmental impacts of the proposed timber sales in the Sandy-Remote Area. The original EA contained no analysis of the fungus on the Port Orford Cedar. Indeed, it specifically excluded such analysis. Section S of the revised EA, which was prepared only after the timber sales had commenced and this litigation was initiated, does analyze the impact of the timber sales on the fungus and the Cedar, but it is the only part of the EA or revised EA to do so.

On appeal, ONRC attacks the revised EA, including Section S, on three grounds. First, ONRC claims the revised EA was not timely because it was prepared after the timber sales were completed. Second, ONRC claims the revised EA improperly tiers to the deficient EIS for the Coos Bay RMP and the Guidelines. Finally, ONRC claims that Section S of the revised EA fails to address the cumulative impact of the Sandy-Remote timber sales and other "reasonably foreseeable future actions" on the Cedar. Because we agree with ONRC on the second and third grounds, we express no opinion as to the first. Even assuming that the analysis contained in Section S was timely, the revised EA is nevertheless inadequate under NEPA.

The revised EA for the Sandy-Remote Area impermissibly attempts to tier both to the EIS for the Coos Bay RMP and to the Guidelines. As just discussed, the EIS for the Coos Bay RMP is inadequate because it, too, impermissibly attempts to tier to the Guidelines. The revised EA is therefore inadequate to the extent that it attempts to tier to the EIS. The same is true of the revised EA's attempt to tier to the Guidelines. As the district court correctly stated:

[D]efendant attempts to utilize the [Port Orford Cedar] guidelines, which were never subject to NEPA review, by asserting that the EA did not tier to the guidelines but merely referenced them as[a] shorthand way of describing a set of potential mitigation measures. However, NEPA requires more than a shorthand reference to consider environmental impacts of an agency's mitigation measures. To the extent that the Coos Bay RMP played a role with respect to the timber sales, it suffers from the same infirmity.

Kern v. U.S. Bureau of Land Mgmt., 38 F. Supp. 2d 1174, 1182 n. 3 (D.Or.1999) (citation omitted).

Because the revised EA for the Sandy-Remote Area may not tier to either the EIS or the Guidelines, we examine whether the revised EA, standing alone, satisfies the requirements of NEPA. Section S of the revised Sandy-Remote EA analyzes the impact only of the specific timber sales proposed for the Sandy-Remote Area, and it analyzes the impact of that agency action only in that Area. It performs no analysis of impacts outside the Area. Most important, it performs no cumulative impact analysis of "reasonably foreseeable future actions" outside the Sandy-Remote Area that, in combination with the Sandy-Remote timber sales, could constitute "collectively significant actions... over a period of time." 40 C.F.R. § 1508.7. Because the revised Sandy-Remote EA does not perform such an analysis, we hold that it is inadequate under NEPA.

In determining the scope of the required NEPA analysis, an agency must consider not only the proposed action, but also three types of related actions — "connected actions," "similar actions," and "cumulative actions." 40 C.F.R. § 1508.25(a). "Cumulative actions" are those "which when viewed with other proposed actions have cumulatively significant impacts." Id. § 1508.25(a)(2). In determining the significance of a proposed action, an agency must consider

Whether the action is related to other actions with individually insignificant but cumulatively significant impacts. Significance exists if it is reasonable to anticipate a cumulatively significant impact on the environment. Significance cannot be avoided by terming an action temporary or by breaking it down into small component parts.

40 C.F.R. § 1508.27(b)(7). See also Churchill County v. Norton, 276 F.3d 1060, 1072 (9th Cir.2001). The regulations define "cumulative impact" as:

the impact on the environment which results from the incremental impact of the action when added to other past, present, and reasonably foreseeable future actions regardless of what agency (Federal or non-Federal) or person undertakes such other actions. Cumulative impacts can result from individually minor but collectively significant actions taking place over a period of time.

Id. (emphasis added).

Consideration of cumulative impacts requires "some quantified or detailed information; ... [g]eneral statements about `possible' effects and `some risk' do not constitute a `hard look' absent a justification regarding why more definitive information could not be provided." Neighbors of Cuddy Mountain, 137 F.3d at 1379-80. The cumulative impact analysis must be more than perfunctory; it must provide a "useful analysis of the cumulative impacts of past, present, and future projects." Muckleshoot Indian Tribe, 177 F.3d at 810. Finally, cumulative impact analysis must be timely. It is not appropriate to defer consideration of cumulative impacts to a future date when meaningful consideration can be given now. See Neighbors of Cuddy Mountain, 137 F.3d at 1380; City of Tenakee Springs, 915 F.2d at 1312-13. When an agency's determination of what are "reasonably foreseeable future actions" and appropriate "component parts" is "`fully informed and well-considered,'" we will defer to that determination. Blue Mountains Biodiversity Project v. Blackwood, 161 F.3d 1208, 1211 (9th Cir.1998) (quoting Save the Yaak Comm. v. Block, 840 F.2d 714, 717 (9th Cir.1988)). But we "need not forgive a `clear error in judgment.'" Id. Federal regulations "do not explicitly require an EIS to include a discussion of cumulative impacts," Edwardsen v. United States Dep't of Interior, 268 F.3d 781, 786(9th Cir.2001), but they do "direct[] agencies to consider cumulative impacts in determining the scope of an EIS." Id., citing 40 C.F.R. § 1508.25(c)(3). We have interpreted the regulations to require that the EIS consider the cumulative impact of the proposed action. See Edwardsen, 268 F.3d at 786-90; see also Neighbors of Cuddy Mountain, 137 F.3d at 1378("[W]here `several actions have a cumulative ... environmental effect, this consequence must be considered in an EIS.' City of Tenakee Springs v. Clough, 915 F.2d 1308, 1312 (9th Cir.1990).").

Similarly, NEPA regulations contain only a brief description of the requirements for an EA, and do not specifically mention cumulative impact analysis. They state that an EA must:

(1) Briefly provide sufficient evidence and analysis for determining whether to prepare an environmental impact statement or a finding of no significant impact; (2) Aid an agency's compliance with the Act when no environmental impact statement is necessary; (3) Facilitate preparation of a statement when one is necessary.

40 C.F.R. § 1508.9(a). An EA "[s]hall include brief discussions of the need for the proposal, of alternatives as required by section 102(2)(E) [of NEPA], of the environmental impacts of the proposed action and alternatives, and a listing of agencies and persons consulted." Id. § 1508.9(b).

We have held that an EA may be deficient if it fails to include a cumulative impact analysis or to tier to an EIS that has conducted such an analysis. See Hall v. Norton, 266 F.3d 969, 978 (9th Cir.2001); Blue Mountains Biodiversity Project, 161 F.3d at 1214; Idaho Sporting Cong. v. Thomas, 137 F.3d 1146, 1152 (9th Cir.1998). Other circuits have also recognized the requirement that, in appropriate cases, an EA must include a cumulative impact analysis. See, e.g. Soc'y Hill Towers Owners' Ass'n v. Rendell, 210 F.3d 168, 180(3d Cir.2000) ("[I]f the cumulative impact of a given project and other planned projects is significant, an applicant can not simply prepare an EA for its project, issue a FONSI, and ignore the overall impact of the project ..."); Newton County Wildlife Ass'n v. Rogers, 141 F.3d 803, 809 (8th Cir.1998) (holding that an EA adequately addressed cumulative impacts where it covered a timber sale involving 1,871 acres but considered environmental impacts on 26,699 acres). The importance of analyzing cumulative impacts in EAs is apparent when we consider the number of EAs that are prepared. The Council on Environmental Quality noted in a recent report that "in a typical year, 45,000 EAs are prepared compared to 450 EISs.... Given that so many more EAs are prepared than EISs, adequate consideration of cumulative effects requires that EAs address them fully." Council on Environmental Quality, Considering Cumulative Effects Under the National Environmental Policy Act at 4, Jan. 1997, also available at http://ceq.eh.doe.gov/nepa/ccenepa/ccenepa.htm (last visited Feb. 26, 2002) (emphasis added).

We have previously indicated, in cases similar to this one, that cumulative impact analysis is appropriate at the EA level. In Hall v. Norton, the BLM proposed to exchange 4,975 acres of land it owned in the Las Vegas Valley so that a private developer could build approximately 11,200 new homes on the land. The BLM prepared an EA that analyzed the pollution that would result from development on the land, but that failed to analyze the additional pollution that might result if other land it owned in the Las Vegas Valley were also exchanged. Plaintiff had alleged in his complaint that the EA had failed to consider cumulative impacts, but the district court had dismissed the complaint without considering the argument. We noted that the BLM owned an additional 57,000 acres of land in the Las Vegas Valley. Those lands had already been "`identified for disposal' [by the BLM], but the EA did not attempt to quantify the cumulative emissions from potential development on these lands. NEPA requires that an agency consider cumulative impacts of an action and of foreseeable related actions." 266 F.3d at 978. We remanded to the district court for consideration of the plaintiff's cumulative impact argument.

In Blue Mountains Biodiversity Project, the Forest Service planned to conduct a number of "salvage sales" of timber after wildfires had burned large areas in eastern Oregon. The Tower Fire, the largest of the fires, had burned a 140 square-mile area in the Umatilla National Forest, and the Forest Service planned five salvage sales in that area. It prepared no overall EIS, but did prepare an EA for one of the five sales. Plaintiff challenged the failure to prepare an overall EIS for all the salvage sales contemplated in the area burned by the Tower Fire, and it challenged the adequacy of the EA for the particular sale.

We held, first, that an EIS was required for the contemplated salvage sales in the entire Tower Fire area. We held, second, that the EA prepared for the single sale was inadequate because it failed to do a cumulative impact analysis of all the reasonably foreseeable sales:

[Plaintiff] alleges that the ... EA fails to address, or even mention, three of the four other salvage sales proposed for the Tower Fire area. Relying on the Forest Service's own acknowledgment that the five sales are part of a coordinated Tower Fire recovery strategy, [plaintiff] argues that the Forest Service should have evaluated the cumulative impacts of these sales in a single EIS. We agree.

161 F.3d at 1215.

We have also recognized the importance of cumulative impact analysis of agency actions in other contexts. In Pacific Coast Federation of Fishermen's Ass'n v. National Marine Fisheries Service, 265 F.3d 1028 (9th Cir.2001), plaintiffs challenged the adequacy of four biological opinions issued by the National Marine Fisheries Service ("NMFS"). The opinions concluded that 23 timber sales in the Umpqua River Basin in Oregon were not likely to jeopardize the continued existence of two species of cutthroat trout and coho salmon. We held that the opinions violated the "arbitrary and capricious" and "clear error in judgment" standard of the Administrative Procedure Act, 5 U.S.C. § 706(2)(A), because they failed to consider the cumulative effect of the proposed timber sales:

[T]he record contains no proof that the cumulative effect of site specific degradation was considered in reaching a no jeopardy opinion at the regional watershed level.... [The agency's] disregard of projects with a relatively small area of impact but that carried a high risk of degradation when multiplied by many projects and continued over a long time period is the major flaw in NMFS study.... If in fact [the NMFS] disregards [the effects of individual projects] as "localized" when they can have significant aggregate effects, it acts arbitrarily and capriciously.

Id. at 1036-37.

NEPA requires that the BLM analyze the impact of reasonably foreseeable timber sales, and of other "reasonably foreseeable future actions," on the fungus and the Cedar. We have already held that the BLM must conduct such an analysis in the EIS for the RMP for the entire Coos Bay District. We now hold that the BLM must also conduct such an analysis in any EA for this site-specific project within that District. We do not hold that no EA for a site-specific project can ever be adequate if the project is undertaken pursuant to an RMP for which there is an inadequate EIS. Rather, we hold that the impacts, including the cumulative impacts, of the site-specific project must be fully analyzed in any EA for that project. If, as is the case here, there is no analysis in the EIS, the scope of the required analysis in the EA is correspondingly increased.

In the case now before us, the entire analysis of the impact of the timber sales in the Sandy-Remote Area is contained in Section S of the revised EA. Section S purports to be a cumulative impact analysis, but it is limited to an analysis of the impacts of the timber sales solely within the Sandy-Remote Area. (Section S makes clear its limited scope when it states that "[t]he Sandy-Remote watershed is a reasonable landscape scale for considering the cumulative impacts of[the fungus] on [the Port Orford Cedar].")

Section S makes no attempt to analyze possible spill-over effects of the Sandy-Remote timber sales on the spread of the fungus outside the Sandy-Remote Area. For example, it proposes closure of some roads, as well as restriction of some roads to summer use (when there will be less mud), but it makes no attempt to estimate the likelihood that, despite such measures, mud containing the fungus will cling to logging trucks that drive out of the Sandy-Remote watershed into other areas. More important, Section S makes no attempt to analyze the cumulative impacts of the Sandy-Remote timber sales when combined with "reasonably foreseeable future actions" (such as other timber sales) outside the Sandy-Remote Area. Section S has thus not performed a cumulative impact analysis as that term is used in the regulations and in our case law.

In the absence of an EIS analyzing the impact of reasonably foreseeable future timber sales within the Coos Bay District under the proposed RMP, we hold that it was arbitrary and capricious, and a clear error in judgment, for the BLM not to include in the revised EA for the Sandy-Remote Area an analysis of the cumulative impacts of such sales within that District. At a minimum, the BLM is required to provide such an analysis in the EA. If such an analysis is not made, it would be easy to underestimate the cumulative impacts of the timber sales in the Sandy-Remote Area, and of other reasonably foreseeable future actions, on the spread of the fungus and the welfare of the Cedar. Such a restricted analysis would impermissibly subject the decisionmaking process contemplated by NEPA to "the tyranny of small decisions." Considering Cumulative Effects, at 1. As we have previously said, NEPA does not allow an approach that "would permit dividing a project into multiple `actions,' each of which individually has an insignificant environmental impact, but which collectively have a substantial impact." Thomas v. Peterson, 753 F.2d 754, 758 (9th Cir.1985); see also Churchill County, 276 F.3d at 1076("an agency [can]not escape the existence of a comprehensive program with cumulative environmental effects by `disingenuously describing it as only an amalgamation of unrelated smaller projects'") (citation omitted).

It is thus clear that the BLM should have included reasonably foreseeable timber sales under the currently proposed RMP for the Coos Bay District in a cumulative impact analysis. It is less clear whether the BLM should have included other actions in such an analysis. Such other actions might include BLM timber sales outside the Coos Bay District but within the geographic range of the Port Orford Cedar; timber sales by other federal agencies, or timber harvesting by private parties, outside the Coos Bay District but within the range of the Cedar; or any other actions with an impact on the fungus and the Cedar. If such actions are "reasonably foreseeable future actions" within the meaning of § 1508.7, the regulations required that they be included in a cumulative impact analysis.

We say this to make clear how far the obligation of the BLM extends under the regulations. However, because at this point the BLM has conducted no cumulative impact analysis whatsoever of timber sales or other actions outside the Sandy-Remote Area, we do not know whether there are any actions, beyond the timber sales that are within the scope of the RMP, that would qualify as "reasonably foreseeable future actions" within the meaning of the regulations. It is for the BLM to make that judgment in the first instance. Once it has made that judgment, we will not interfere unless it has acted arbitrarily and capriciously or has made a clear error in judgment. On the record that is now before us, we intimate no opinion on whether there are or may be any other reasonably foreseeable future actions that the BLM must consider.

We reverse the judgment of the district court finding the challenge to the Coos Bay EIS unripe, and hold that the EIS is inadequate under NEPA. We therefore remand with instructions that summary judgment be entered for ONRC on the adequacy of the EIS. We also reverse the district court's grant of summary judgment to the defendants on the challenge to the Sandy-Remote EA, and remand with instructions that summary judgment be entered for ONRC on the adequacy of the EA.

REVERSED and REMANDED.

GRABER, Circuit Judge, concurring in part and dissenting in part:

I respectfully dissent from Part IV(B) of the majority opinion. In my view, the BLM permissibly limited its EA, including the cumulative impact component, to the Sandy-Remote Analysis Area.

Our standard of review is a limited one. We are to decide only whether the agency's solution is arbitrary or capricious. Morongo Band of Mission Indians v. FAA, 161 F.3d 569, 573 (9th Cir.1998); 5 U.S.C. § 706(2)(A). We must consider "`whether the decision was based on a consideration of the relevant factors and whether there has been clear error of judgment.'" Alaska Ctr. for the Env't v. United States Forest Serv., 189 F.3d 851, 859 (9th Cir.1999) (quoting Marsh v. Or. Natural Res. Council, 490 U.S. 360, 378, 109 S. Ct. 1851, 104 L. Ed. 2d 377 (1989)). Accordingly, it is not up to us to impose the "best" way to analyze the problem, but only to determine whether the BLM's decision in this instance to limit its examination of cumulative impacts to one watershed is supported by the record and is neither arbitrary nor capricious.

The record supports the BLM's decision to limit the EA to analyzing the Sandy-Remote Area, which comprises about 24,000 acres and represents between 0.7% and 1.4% of the range of the Port Orford Cedar. The fungus that affects the Cedar is transported naturally by streams and soil erosion, so that most of its natural spread is confined to a given watershed.1 The risk from the proposed action is that logging could contribute to the spread of the fungus by transporting infected soil from motor vehicles along roads. The BLM has determined that it can close all the presently uninfected roads within the watershed, thus sealing off the uninfected areas from those that are infected. That fact makes the watershed a geographically and scientifically coherent choice. See Pac. Coast Fed'n of Fishermen's Ass'ns v. Nat'l Marine Fisheries Serv., 265 F.3d 1028, 1036-37 (9th Cir.2001) (holding that cumulative effects in a particular context under the Endangered Species Act must be considered at the watershed level). Within the watershed, the EA does consider cumulative effects.

To be sure, a different case with a different record may require a different result, as Hall v. Norton, 266 F.3d 969, 978 (9th Cir.2001), did. NEPA prohibits an agency from breaking up a large project into smaller parts to avoid designating the project as a "`major federal action.'" Churchill County v. Norton, 276 F.3d 1060, 1076 (9th Cir.2001) (quoting Nat'l Wildlife Fed'n v. Appalachian Reg'l Comm'n, 677 F.2d 883, 890 (D.C.Cir.1981)). Similarly, an agency cannot avoid a "significant" impact by breaking one overall action into small components. 40 C.F.R. § 1508.27(b)(7). But the record here does not reflect that the BLM has done either of those artificial things or has engaged in geographic manipulation or avoidance.

For these reasons, I do not agree that Section S of the revised Sandy-Remote Area EA is insufficient for failing to account for the cumulative impact of proposed timber sales in other watersheds and I would affirm the district court in that regard. Therefore, I dissent from Part IV(B), but concur in the remainder of the opinion.