dissenting in part:
I dissent from the portion of the majority’s discussion of the issues relating to the National Forest Management Act (NFMA), and concur in the court’s reasoning concerning the National Environmental Policy Act (NEPA). Hence I join in full both the judgment and the reasoning of the court as presented in the memorandum disposition filed concurrently with the published opinion. But I regret I cannot fully join the NFMA analysis, and hence I respectfully dissent in part.
Addressing the NFMA issues, that statute explicitly serves up for agency action a catalog of potential competing uses of the forest. The agency administering NFMA, the United States Forest Service (Forest Service), is explicitly to consider managing the forests having in mind environmental considerations but also economic considerations. See Native Ecosystems Council v. Weldon, 697 F.3d 1043, 1056 (9th Cir. 2012) (“Forest plans are designed to manage forest resources by balancing the eon-sideration of environmental and economic factors.”); 16 U.S.C. § 1604. One purpose of the NFMA is to encourage conservation of natural resources in our national forests. But another purpose is to support a timber industry that can provide much needed jobs to those living in forested areas in or near our national forests.
It is obvious that once the agency makes a decision in the normal case and gives more weight, for example, to creating jobs than to protecting trees, the scope of review available for a federal court will be extremely limited under the Administrative Procedures Act. We are to “hold unlawful and set aside agency action, findings, and conclusions found to be arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A). Our review under the Administrative Procedure Act of the agency’s final administrative action is not generally a review of substance or policy, but rather of procedure. As is well-stated by the majority, “we have an Administrative Procedure Act, not an Administrative Policy Act.”
Given that agency action in substance is likely to be upheld when proper procedures are followed, it is all the more important to look carefully at whether the decision of the agency here is procedurally sound. I conclude that it is not.
For one thing, the applicable regulation governing the content of a forest plan, states that viable populations of wildlife in the forests are to be maintained. The NFMA aims to foster diversity of plant and animal communities as part of its mission to meet multiple-use objectives. 18 U.S.C. § 1604(g)(3)(B). The NFMA itself does not say that the wolf species must be *978maintained at specific numbers and as previously stated, the statute sets out competing potential uses of the forest that must be balanced by the responsible federal agency. The majority poses the issue as a choice between wolves and jobs, but it is not so simple as that.
Although the NFMA does not require viability, the agency has a duty to ensure the viability of surrounding wildlife if it incorporates 36 C.F.R. § 219.19 (2000) into its forest plan. See Ecology Ctr. v. Castaneda, 574 F.3d 652, 657 (9th Cir. 2009). And here, the Forest Service did incorporate the regulation in its forest plan by noting that it is directed “to manage wildlife habitat to maintain viable and well distributed populations to ensure continued existence in the planning area.” By so doing, “the forest plan must comply with substantive requirements of the [NFMA] designed to ensure continued diversity of plant and animal communities and the continued viability of wildlife in the forest. ...” Idaho Sporting Cong., Inc. v. Rittenhouse, 305 F.3d 957, 961-62 (9th Cir. 2002). Specifically, 36 C.F.R. § 219.19 requires that “[f]ish and wildlife habitat shall be managed to maintain viable populations of existing native and desired non-native vertebrate species in the planning area.” Our law is clear that an agency must abide by its own regulations. See Nat’l Ass’n of Home Builders v. Norton, 340 F.3d 835, 852 (9th Cir. 2003); see also United States ex rel Accardi v. Shaughnessy, 347 U.S. 260, 266, 74 S.Ct. 499, 98 L.Ed. 681 (1954).
Here, the majority is so preoccupied with the issue whether “sustainability” is a discretionary term imposing no restraint on the Forest Service, that it fails to assess adequately whether the remaining population of the Alexander Archipelago wolf will be viable after the project proceeds. A viable population is “one which has the estimated numbers and distribution of reproductive individuals to insure its continued existence is well distributed in the planning area.” 36 C.F.R. § 219.19 (2000). There is a difference between a viable population and a sustainable population. The latter must contain a larger number of wolves because it is a population “capable of sustaining harvest.” The Forest Service has recognized this distinction: “[M]inimum viable populations for many species may not satisfy the public need for wildlife populations depended upon to meet subsistence and/or sport hunting uses.”
As quoted above, the language of the regulation states in pertinent part that a population of wildlife in the forest shall be maintained, so that is a “viable” population. It does not matter if the term sustainable population is hemmed in by the discretionary words “if possible.” Instead, what is important is that there is a non-discretionary obligation in the Forest Plan for the Forest Service to maintain a “viable” population of the Alexander Archipelago wolf.1
Although the Forest Service concluded that the Forest Plan ensured the viability of the Alexander Archipelago wolf and also concluded that the Big Thorne Project was consistent with the Plan, the Forest Plan and the Big Thorne project do not demonstrate that it will manage old growth habitat in a manner that insures the viability of the wolf in’ well distributed populations throughout the Tongass National Forest. *979The agency is not obligated to provide a sustainable wolf population, but it must ensure viable wolf populations consistent with its substantive obligation under 36 C.F.R. § 219.19. The Forest Plan presently provides no mechanism to ensure wolf population viability.
For another thing, the agency’s rationale and reasoning process is too summary and conclusory, and fails to deal with the entire problem before it. When the agency opts to protect jobs rather than wolves, it would normally be allowed to make that choice. But the Supreme Court in State Farm elaborated on when 'we should consider an agency’s action to be arbitrary and capricious. The Court stated that “the agency must examine the relevant data and articulate a satisfactory explanation for its action including a ‘rational connection between the facts found and the choice made.’ ” Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983) (quoting Burlington Truck Lines v. United States, 371 U.S. 156, 168, 83 S.Ct. 239, 9 L.Ed.2d 207 (1962)).
That problem is presented here. The agency’s assessment of the plight of the wolves in the Tongass National Forest is in my view inadequate.2 In the mid 1990s, 250-350 wolves were thought to inhabit the Alexander Archipelago. The Big Thorne project area, located on the Prince of Wales Island, had sufficient habitat to support 45-50 wolves, making up three separate packs and a portion of a fourth pack. A project, funded primarily by the Forest Service, estimated that by fall 2012, only about 29 wolves and only two remaining packs remained in the Big Thorne project area. By spring of 2013, researchers could only account for a mere six to seven wolves left in the project area. Another researcher observed that “numbers seem to indicate that the population of wolves in the central portion of Prince of Wales Island is approaching zero.”
The Alexander Archipelago wolf and the Sitka black-tailed deer, as the Forest Service explained, are closely interrelated. Survival of the wolves depends on survival of the deer on which the wolves feed, which in turn depends on maintenance of the old growth forest habitat on which the deer depends. Deer populations depend on a sufficient quantity, distribution, and quality of winter old-growth habitat. It is inescapable that logging reduces this habitat. The Forest Service, aware that wolves depend on deer for their survival, and that those deer to survive depend on old growth forest habitat, has stressed that maintaining sufficient deer habitat capability, and thus greater number of deer, was “the most important factor” in sustaining wolf viability in the Tongass National Forest. In fact, the destruction of deer habitat capability is, as the agency has explained, “the most important factor limiting wolf viability.” Yet the Forest Plan does not support deer habitat capability numbers, and nothing in the record supports the Forest Service’s contention that viable wolf populations will remain relying on the other two prongs of habitat reserves and wolf mortality management.
Perhaps it cannot be said that the agency could never conclude on the limited evidence that a group of animals will or will not survive as a consequence of forest activities. But here the record is too sparse for definitive conclusion, and what evidence exists in the record shows drastieal*980ly decreasing wolf population over the past decades and a project that diminishes deer habitat capability, further threatening the wolves.
I would remand for more proceedings before the district court in which the Forest Service could better explain its reasoning and position relating to threatened habitat needed by the wolves that will be lost if the pending project is approved. I urge the majority to reassess its conclusions, remand for more development of the record and pertinent conclusions, and also give guidance (1) as to the interpretation of the term “viable” in its context; and also (2) on when and under what circumstances the agency would be free to modify its restrictive regulation on viable populations.
Hence I respectfully dissent in part. I would vacate the decision of the Forest Service and remand for its further proceedings, which at a minimum need to include both a thorough assessment of viability of the Alexander Archipelago wolf if the project proceeds, and an explanation of its reasoning sufficient to satisfy State Farm. Perhaps the agency can satisfactorily explain itself. But explain it must do for the decision to be procedurally correct. And who knows, perhaps in going through the rigorous explanation of whether the Alexander Archipelago wolf will continue to exist in viable numbers, the agency may learn more, which could lead it to take some different action more favorable to wolf and human communities. At a minimum, the agency, by explaining itself, will help to fully inform the public about the issues presented in balancing wolf survival, adequacy of deer habitat, and impact on jobs and pertinent communities caused by the final agency action.
. The majority argues that our precedent rejects the idea that the Forest Service needs to identify a "mechanism” to secure viability, and has rejected the idea that viability must be assessed in terms of population size, trends, or dynamics. But the problem here is not that the agency didn't use a particular "mechanism” to assess viability, the problem is rather that the agency just did not really grapple with assessing whether current wolf populations are viable. Nor does the majority.
. The majority criticizes my specific discussion of the Prince of Wales Island, because it only covers less than 10% of the Tongass National Forest. I do so because the island includes unique, temperate old-growth rainforest habitat, and because deer populations depend on a sufficient quantity, distribution, and quality of winter old-growth habitat.