dissenting:
The majority’s denial of a preliminary injunction, like the district court’s before it, rests on two fundamental errors. First, the majority concludes that the Forest Service has no obligation to ensure species viability in the Plumas National Forest despite numerous clear statements to the contrary in the Plumas National Forest Plan. Second, it concludes that the RHT Hazard Tree Marking Guidelines are not binding on the Forest Service despite the fact that the Forest Service itself acknowledges that they are. The district court rested its denial of a preliminary injunction almost entirely on its erroneous conclusions regarding Earth Island’s likelihood of success on the merits. I would therefore grant a temporary injunction and remand to the district court to reconsider Earth Island’s application for a preliminary injunction with the understanding that, as is explained below, Earth Island is likely to prevail on its NFMA challenges to the Moonlight-Wheeler Project.
I.
The Forest Service’s actions in implementing the Moonlight-Wheeler Project violated a substantive obligation, enforceable under the NFMA, to ensure the viability of the Black-backed Woodpecker. To avoid this conclusion, the majority wilfully ignores unambiguous language in the 2004 and 2007 Forest Plan Amendments, which impose a binding obligation to ensure species viability in the Plumas National Forest, as well as our precedent, which makes this obligation applicable to individual projects, such as the Moonlight-Wheeler Project.
A.
The viability requirement that Earth Island alleges the Forest Service violated was originally stated in the first paragraph of the so-called 1982 Rule, 36 C.F.R. § 219.19 (1982). It reads:
Fish and wildlife habitat shall be managed to maintain viable populations of existing native and desired non-native vertebrate species in the planning area. For planning purposes, a viable population shall be regarded as one which has the estimated numbers and distribution of reproductive individuals to insure its continued existence is well distributed in *477the planning area. In order to insure that viable populations will be maintained, habitat must be provided to support, at least, a minimum number of reproductive individuals and that habitat must be well distributed so that those individuals can interact with others in the planning area.
Id. Although the majority is correct that the regulation itself has since been super-ceded, its requirements still apply “to the extent they were incorporated into [a] Forest Plan.” Castaneda, 574 F.3d at 657. The majority appears to conclude that the 1982 Rule was not incorporated into the Plumas National Forest Plan by the 2004 amendments, and that therefore the Forest Service has no obligation to ensure species viability. This conclusion is contrary not only to the district court’s finding that the 1982 Rule was incorporated into the 1993 Forest Plan and the 2001, 2004, and 2007 Forest Plan Amendments, but also to this court’s explicit prior holding that the 2004 amendments did incorporate the 1982 Rule. ER Vol. 1, Tab 1, page 18 (“The Forest Plan and all amendments rely on the 1982 version of the 36 C.F.R. § 219 NFMA regulations.”).1 See Earth Island Inst. v. U.S. Forest Service, 442 F.3d 1147, 1173 (9th Cir.2006) (“As a preliminary matter, we conclude the NFMA regulations promulgated in 1982 apply to the ... 2004 [amendments].”)2 Id. at 1173.
Moreover, even if binding precedent did not foreclose the majority’s conclusions that the 1982 Rule does not apply and there is therefore no viability requirement, these conclusions could not be supported by a sensible reading of the relevant forest plan amendments. The 2004 amendments to the Plumas National Forest Plan, for example, could not have been more clear in incorporating the 1982 Rule and imposing a viability requirement upon the Forest Service. In the Record of Decision implementing the 2004 amendments, the Forest Service’s Regional Forester for the Pacific Southwest Region stated that, “[m]y decision conforms with the 1982 planning regulations (29 CFR 219) that implement the National Forest Management Act.” The document’s following subsection is entitled, “Diversity and Viability Provisions for Fish and Wildlife,” (emphasis added), and concludes by requiring the Forest Service to “provide the fish and wildlife habitat and other ecological conditions necessary to maintain well-distributed viable populations of vertebrate species in the planning area.” (Emphasis added). The majority suggests that this statement “could be read to mean that the Service was to ensure the distribution of the species,” but not their viability. Maj. Op. at 471. In fact, it cannot. To do so, one would first be required to ignore the fact that this statement follows shortly after the amend*478ments’ explicit adoption of the 1982 Rule; one would then have to ignore the fact that the statement falls under a heading pertaining to “Viability Provisions”; and finally, one would have to willfully overlook the presence of the word viable in the statement itself. There is no canon of construction that allows us to ignore words we deem inconvenient. Such a willful misreading is especially unacceptable, where, as here, the context in which the word “viable” appears makes perfectly clear that its inclusion in the amendments was deliberately intended to impose upon the Forest Service a viability requirement consistent with the 1982 Rule.
The majority likewise errs in its conclusion that the 2007 amendments do not impose a viability requirement. It so concludes because, in addressing requirements for Management Indicator Species such as the Black-backed Woodpecker, the Amendments require only “[distribution population monitoring [to] track changes in the distribution of each MIS at the Sierra Nevada scale by monitoring the changes in the presence of the species across a number of sample locations,” and state that “[t]he sole MIS requirement that is applied at the project-level is the assessment of habitat for MIS. Further, there are no monitoring requirements for MIS at the project level.” But this language is simply irrelevant to whether or not the Forest Service is obligated to ensure species viability. The 2007 amendments made clear that the species viability requirement and the MIS monitoring requirement are separate and distinct provisions, and that though the Amendment changed MIS monitoring requirements, it had no effect whatsoever on the Forest Service’s preexisting obligation to ensure viability under the 1982 Rule:
This Amendment does not change the viability requirements. The viability requirements at the planning area scale are described under the first paragraph of the [1982 Rule]; these have already been met in each forest plan, as revised. Forests will continue to ensure that the project-level viability requirements are met: “Provide for adequate fish and wildlife habitat to maintain viable population of existing native vertebrate species” (36 C.F.R. 219.27(a)(6)). This is documented in project-level analysis.... The project-level MIS requirements ... will also be documented in project-level analysis.
FEIS Appendix G at 338. The Final Environmental Impact Statement for the 2007 amendments likewise stated that “[m]anagement for conservation of all species, regardless of whether they are designated as MIS or not, is governed by ... the general viability requirements of the National Forest Management Act implementing regulations.” FEIS at 56 (emphasis added). The 2004 and 2007 amendments impose a viability requirement upon the Forest Service that is distinct from the MIS monitoring requirement, and Forest Service must satisfy both requirements to be in compliance with the NFMA. Although the majority’s extensive discussion of how the Moonlight-Wheeler Project satisfies the 2007 MIS monitoring requirement may well be correct, it is completely beside the point. The question at issue in this case is not whether or not the Forest Service has met its obligations to monitor MIS species, but whether or not it has satisfied the viability requirements so clearly set forth in the 2004 and 2007 amendments.
B.
Our precedents have been deliberately flexible in defining what the Forest Service must do when, as here, a Forest Plan requires it to ensure species viability in the administration of a given national forest. *479See, e.g., The Lands Council v. McNair, 537 F.3d 981, 997 (9th Cir.2008) (en banc) (“To always require a particular type of proof that a project would maintain a species’ population in a specific area would inhibit the Forest Service from conducting projects in the National Forests.”). However, two principles are clear. First, contrary to the majority’s repeated efforts to assert that even if a viability requirement exists it does not apply at the project level, a viability requirement that is applicable at the planning level necessarily applies to and constrains individual projects undertaken in a national forest:
[Cjompliance with NFMA’s forest-wide species viability requirements is relevant to the lawfulness of any individual timber sale. To hold otherwise would permit the Forest Service to don blinders to the overall condition of a national forest each time it approved a sale, quite literally losing sight of the forest for the trees. This would contravene “one of the fundamental purposes of Congress in enacting [NFMA]: that the National Forest System be managed with ‘a systematic interdisciplinary approach,’ by means of ‘one integrated plan for each unit of the National Forest System.’ ” Idaho Sporting Cong. v. Rittenhouse, 305 F.3d 957 (9th Cir.2002) (quoting 16 U.S.C. § 1604).
Neighbors of Cuddy Mountain v. Alexander, 303 F.3d 1059, 1069 (9th Cir.2002); see also McNair, 537 F.3d at 989 (“After a forest plan is developed, all subsequent agency action, including site-specific plans ... must comply with the NFMA and be consistent with the governing forest plan.”).
Second, the NFMA requires that when a viability requirement is included in a Forest Plan, the Forest Service must affirmatively demonstrate that any given project satisfies its obligation to ensure species viability. McNair, 537 F.3d at 992-94. Such a demonstration is to be reviewed deferentially, see id. at 992 (“[W]e defer to the Forest Service as to what evidence is, or is not, necessary to support wildlife viability analyses.”), but, at the least, the Forest Service must:
support its conclusions that a project meets the [viability] requirements of the ... relevant Forest Plan with studies that the [Forest Service], in its expertise, deems reliable. The Forest Service must explain the conclusions it had drawn from its chosen methodology, and the reasons it considers the underlying evidence to be reliable.
Id. at 994. In McNair, for example, the Forest Service supported its conclusion that a project in the Idaho Panhandle National Forest provided for the Flammulated Owl’s viability by citing to several scientific studies pertaining to the owl’s habitat preferences and conducting on-the-ground analysis of the owl in an area adjacent to the project area. Id. at 994. That “relatively sparse” showing, we stated, “approaches the limits of our deference.” Id. at 995.
In implementing the Moonlight-Wheeler Project, the Forest Service failed to meet even these minimal obligations. The agency did not reach a conclusion that the Project would ensure species viability, let alone did it support such a conclusion, as the NFMA requires, with studies and methodological explanations sufficient to establish that its actions were not arbitrary or capricious. The agency argues that it satisfied its obligations through analysis that it undertook in fulfillment of the Forest Plan’s unrelated requirement that it “assess wildlife habitat,” and the majority asserts that it need not demonstrate project-level species viability at all. Where the Forest Service seeks to use habitat as a proxy for species viability it *480“must describe the quantity and quality of habitat that is necessary to sustain the viability of the species in question and explain its methodology for measuring this habitat.” McNair, 537 F.3d at 998. The Forest Service has not done so in this case, and its conclusion that the project would, despite undoing some habitat gains from the Moonlight fires, “still support an upward trend in [Black-backed Woodpecker] population,” tell us nothing about whether the project interferes with the Woodpecker’s viability. The Forest Service’s essentially non-existent conclusions regarding species viability thus fall far short of the conclusions that “approach[ed] the limits of our deference” in McNair. Id. at 995.
The Forest Service’s decision to simply ignore a binding viability requirement in the Plumas National Forest Plan violates the NFMA. See 16 U.S.C. § 1604(i) (“Resource plans and permits, contracts, and other instruments for the use and occupancy of National Forest System lands shall be consistent with the land management plans.”). Earth Island has thus shown that it is extremely likely to prevail on the merits of its NFMA claim. The district court was wrong to conclude otherwise, and because its application of the preliminary injunction test turned predominantly on this erroneous conclusion, I would grant an injunction pending appeal and issue a limited remand to allow the district court to reconsider Earth Island’s request for a preliminary injunction. The majority is simply wrong when it says that species viability on a project level is not required.
II.
The majority further errs in its conclusion that the Forest Service was not bound by the RHT Guidelines for identifying and marking hazardous trees. The majority, like the district court, concludes that the RHT Guidelines are not enforceable. It reaches this conclusion based on the presence of certain non-mandatory phrases in the Guidelines, as well as its conclusion that the Guidelines were not promulgated pursuant to a specific congressional grant of authority. However, both of these arguments for the non-enforceability of the Guidelines are erroneous.
First, the fact that the Guidelines use non-mandatory phrases to describe their general objectives, does not, as the majority suggests, render them unenforceable. See Maj. Op. at 473-74. The majority emphasizes two such uses of non-mandatory phrases — one is a reference to “[t]he spirit of these guidelines”; the other states that the “balance [struck by the Guidelines] aims to retain healthy forest cover” — and somehow arrives at the conclusion that the Guidelines were intended only to provide “internal guidance and parameters” for the marking of hazardous trees. Id. In fact, notwithstanding such general statements, the guidelines are predominantly phrased in mandatory language, and impose specific requirements on which trees may be marked for removal. See, e.g., ER II, Tab 16 at 1-2 (“Mark for removal any hazard tree capable of falling on a road or facility that meets the following criteria....”).
Second, the majority’s conclusion that Earth Island “fails to show that [the Guidelines] were promulgated pursuant to a specific grant of congressional authority” is incorrect. The 2004 amendments to the Plumas National Forest Plan state that the Forest service must “[u]se the best available information for identifying dead and dying trees for salvage purposes as developed by the Pacific Southwest Region Forest Health Protection Staff,” and the Forest Service concedes that “the Plumas Forest Plan, as amended, contemplates that salvage projects will use the RHT Guidelines.” Moreover, a portion of the *481guidelines was explicitly incorporated into the Moonlight-Wheeler Project. The Forest Service thus clearly adopted the Guidelines pursuant to a congressional grant of authority, namely, its authority under the NFMA to develop and execute Forest Plans. See 16 U.S.C. § 1604(e) (“In developing, maintaining, and revising plans for units of the National Forest System pursuant to this section, the Secretary shall assure that such plans ... determine forest management systems, harvesting levels, and procedures in the light of all of the uses set forth in subsection (c)(1) of this section.”) (emphasis added).
Because the RHT Guidelines were incorporated into the Plumas National Forest Plan, the district court erred in concluding that they are unenforceable. This erroneous conclusion kept the district court from deciding whether the factual foundation of Earth Island’s allegations regarding violations of the Guidelines was sufficiently strong to warrant an injunction under the test set forth in Winter v. Natural Res. Def. Council, 555 U.S. 7, 129 S.Ct. 365, 374, 172 L.Ed.2d 249 (2008).3 Its denial of a preliminary injunction regarding the tree marking guidelines — and the majority’s affirmation of that denial — was thus based entirely on the faulty legal conclusion that the RHT Guidelines do not impose any enforceable obligations upon the Forest Service. I would therefore remand this issue to the district court for further proceedings to determine whether, given that the Guidelines create legally binding obligations, Earth Island has alleged facts sufficient to satisfy the Winter test and, accordingly, to warrant the issuance of a preliminary injunction.
III.
The district court failed to address the last three factors of the Winter test— likelihood of irreparable injury, the balance of the harms, and the public interest — as questions distinct from Earth Island’s likelihood of success on the merits. Rather, it addressed each of the three remaining preliminary injunction factors as if they were settled by its finding that Earth Island was unlikely to succeed on the merits of its claim.4 Had the district court analyzed these three factors without reference to its erroneous conclusions regarding Earth Island’s likelihood of success, it may well have been compelled to conclude that a preliminary injunction was appropriate.
Regarding whether the harm suffered by Earth Island is irreparable, the Supreme Court has stated that because “[e]nvironmental injury, by its very nature, can seldom be adequately remedied by money damages and is often permanent or at least of long duration [it is] irreparable.” Amoco Production Co. v. Gambell, 480 U.S. 531, 545, 107 S.Ct. 1396, 94 L.Ed.2d 542 (1987). We have thus repeatedly recognized that the irreversible environmen*482tal effects of logging activity suffice to establish “irreparable harm” for purposes of obtaining a preliminary injunction. See, e.g., Neighbors of Cuddy Mountain v. United States Forest Service, 137 F.3d 1372, 1382 (9th Cir.1998); City of Tenakee Springs v. Clough, 915 F.2d 1308, 1314 (9th Cir.1990).5 Here, Earth Island has alleged that if the snag forest habitat created by the 2007 fires is logged, it will be permanently lost as potential habitat for the Black-backed Woodpecker and other species. The district court expressed skepticism that this logging project would ultimately precipitate the extinction of the Black-backed Woodpecker in the Sierra Nevadas, a question that is difficult to answer given the shoddiness of the Forest Service’s analysis of that issue. But Earth Island need not be required to prove with certainty that the project will lead to the Woodpecker’s extinction to establish that it will be irreparably harmed. There is no dispute that the project will destroy thousands of acres of rare habitat that is critical for the Woodpecker’s survival. As such, there is more than a likelihood— there is an absolute certainty — that Earth Island was and will continue to be irreparably harmed by the denial of an injunction.
The district court likewise committed serious errors in its analysis of the public interest and the balancing of the equities. First, as in its analysis of irreparable harm, the district court’s treatment of these factors rested largely on its erroneous conclusion that Earth Island was unlikely to prevail on the merits. It stated, “While the court must seriously consider the potential harm to the environment caused by the Project, where plaintiff has not made the requisite showing on the merits which, in turn, undermines the likelihood of irreparable injury, the balance of equities cannot be found in plaintiffs favor.” (Emphasis added).
Second, the district court conducted its analysis as though Earth Island had sought an injunction broader than the one it expressly requested. On the same day it filed its motion for a preliminary injunction, Earth Island filed a proposed preliminary injunction order. The proposed order explicitly states that the Forest Service “may fell and leave trees or remove naturally fallen trees to the side of the roadway under emergency circumstances pursuant to 36 C.F.R. § 220.4(b)(1)”, but may not remove any such trees from the project area. Id. The cited regulation allows “actions necessary to control the immediate impacts of the emergency and ... urgently needed to mitigate harm to life, property, or important natural or cultural resources.”
In the reply brief it filed with the district court in support of its request for a preliminary injunction, Earth Island emphasized the safety exception contained in its proposed order, and added that “[i]f this Court is convinced that the allowances made by Plaintiff in [its] Proposed Order are not sufficient to safeguard the public from harm, .... the Court could merely identify some segment of the trees marked for cut as hazards to be logged, while enjoining the rest of the project.” Yet the *483district court, in its analysis of the public interest and the balance of the equities, explicitly relied on evidence that “a real safety risk exists to the public as a result of hazardous trees,” — a risk that the Forest Service would retain the authority to address even if the requested preliminary injunction were issued. The district court thus violated the requirement that “[w]hen deciding whether to issue a narrowly tailored injunction, district courts must assess the harms pertaining to injunctive relief in the context of that narrow injunction.” Sierra Forest Legacy v. Rey, 577 F.3d 1015, 1022 (9th Cir.2009). Had the district court evaluated the narrow injunction Earth Island requested and thus not incorrectly concluded that an injunction would pose a threat to the public safety, it would have been left with only two bases for concluding that the balance of the equities and public interest support an injunction: economic harm to the community and government, and loss of the “benefits of reforestation.” The latter of these is clearly not a legitimate basis for denying a preliminary injunction — even if reforestation is not occurring naturally, there is no reason why the steps necessary to ensure reforestation cannot be undertaken following a decision on the merits.
The district court’s denial of an injunction was thus flawed in a number of respects, as is the majority’s affirmation of that denial. It failed to recognize Earth Island’s strong likelihood of success on the merits as well as the irreparable harm that would befall Earth Island absent an injunction. Likewise, its treatment of the public interest and the balance of the equities rested largely on the court’s flawed analysis of the merits of the case, and failed to account for the narrow scope of the injunction Earth Island requested. I therefore must disagree strongly with the majority. I believe that Earth Island is entitled to an injunction pending appeal, as well as a limited remand to allow the district court to consider whether, in light of its likelihood of success on the merits, the irreparable nature of its foreseeable harm, and the narrow scope of the injunction it requested, it is entitled to a preliminary injunction.
. Neither party appears to dispute that the 2001 amendments incorporated the 1982 Rule's viability requirement. As the Record of Decision for those amendments stated:
My decision conforms with the 1982 planning regulations (36 CFR 219) that implement the National Forest Management Act. These regulations were recently changed.... Transition language within the new rule permits plan revisions and amendments, such as the amendments that are part of my decision, to be completed under the 1982 procedures.
The document then proceeds to discuss the amendments’ compliance with the 1982 Rule. Thus, the only question is whether either the 2004 or 2007 amendments eliminated the Forest Service’s preexisting obligation to ensure species viability consistent with the 1982 Rule.
. Earth Island addressed a project in the El Dorado National Forest. Like Plumas National Forest, El Dorado is governed by the 2001 Sierra Nevada Framework Plan and the 2004 amendments to that framework. See id. at 1154.
. While the district court expressed significant skepticism toward several of Earth Island's arguments regarding the Guidelines, it did so in addressing Earth Island’s NEPA claims. The NEPA claims that the district court addressed are distinct from the claims at issue in Earth Island’s NFMA claim. Whereas the former address the validity of the Guidelines and whether the Forest Service adequately responded to Earth Island's comments on the subject, the latter address whether or not the Forest Service actually adhered to the Guidelines. See Dist. Ct. Op. at 20.
. See, e.g., Dist. Ct. Op. at 63 ("Considering the court has found that it is unlikely plaintiff can demonstrate any clear error in judgment by the [Forest Service] in rendering its decision on the Project’s impacts to wildlife and the soils and watersheds, the court cannot find that plaintiff has shown a likelihood of irreparable harm.”).
. The district court concluded that such cases no longer had precedential value following Winter v. Natural Res. Def. Council, 555 U.S. 7, 129 S.Ct. 365, 172 L.Ed.2d 249 (2008). Winter held that a party seeking a preliminary injunction must show a likelihood of irreparable harm, overruling Ninth Circuit precedent requiring only a showing of a possibility of irreparable harm. Id. at 375-76. However, this holding has no bearing on whether a certain category of alleged harm is in fact irreparable. These precedents thus remain valid to the extent they make clear that the sort of environmental harm that Earth Island anticipates in this case is irreparable in nature.