OPINION

The United States Forest Service (the “Service”) proposed the Smith Creek Project (the “Project”) in the Gallatin National Forest to reduce the risk of severe wildfire, to reduce the risk of insect infestation and disease, and to promote habitat diversity. Sharon Hapner, Alliance for Wild Rockies, and Native Ecosystems Council (collectively “Plaintiffs”) challenged the Project, contending that it violated the National Environmental Policy Act (“NEPA”) and the National Forest Management Act (“NFMA”). After a remand, the district court granted summary judgment to the Service on all of Plaintiffs’ claims.

We affirm the district court in almost all respects. We reverse on only one claim, holding that the Project violates NFMA by failing to comply with the elk-cover requirement contained in the Gallatin National Forest Plan.

I. Background

The Smith Creek Project is located on the west side of the Crazy Mountains, approximately 35 miles north of Livingston, Montana, in the Gallatin National Forest. The Forest is adjacent to the north and west boundaries of Yellowstone National Park.

The Project area is forested with lodge-pole pine, Douglas fir, and to a lesser degree aspen, spruce, and sub-alpine fir. It provides habitat for a number of wildlife species, including resident and migratory herds of elk and Yellowstone cutthroat trout. The area historically has experienced large-scale stand-replacing wildfires that have been important in creating and maintaining the habitat. It has been previously logged, resulting in road construction, skid trail creation, soil disturbance, and riparian damage.

The Project is near residential areas. The largest of these is the Smith Creek subdivision, consisting of about 30 summer and year-round homes. The Service identified a high fire risk to local residents and firefighters due to fuel build-up and limited road access to the subdivision. The Service designed the Project, in part, to reduce likely fire intensity. Other stated purposes of the Project are to improve habitat diversity by maintaining meadow and aspen areas, and to reduce the risk of insect infestations and tree diseases.

The Project would authorize logging on up to 810 acres and prescribed burning on an additional 300 acres. The logging is designed to break up “the vertical and horizontal continuity of vegetation and fuel conditions” in order to reduce the chance of “crown fires” (fire that climbs to, and spreads through, the tops of trees) and to slow the spread of wildfires. In some areas, the logging would remove conifers near aspen trees in order to promote the growth of aspen groves. In other areas, the logging would thin trees so that they are spaced 20 to 50 feet apart. The thinning would leave approximately 300-500 irregularly spaced trees per acre. Current tree densities range up to 3,000 per acre.

Ground-based equipment would be used to cut both large and small diameter trees on up to 435 acres of the Project, while helicopters would be used to help remove timber on 145 acres. The work on the remaining 230 acres would consist of hand-treatment (removal of ladder fuels, limbing of large diameter trees, and thinning of small diameter trees). In order to minimize the Project’s impact on soil quality, ground-based harvest would occur only during the winter and only where there is frozen ground or sufficient snowpack. No new roads would be built, but formerly closed roads would be reopened and used temporarily for the Project.

Some ecosystem restoration activities are part of, or are associated with, the Project. Road maintenance in 2008 improved drainage on parts of two roads in the Project area, thereby improving water quality and habitat for Yellowstone cutthroat trout. That work was completed and is no longer considered part of the Project. The Project includes additional road improvements, dependent on the availability of funds, that would also improve water quality. In addition, five tons per acre of coarse woody debris would be distributed over 4.1 miles of old skid roads within the Project in order to rehabilitate disturbed soil.

In August 2007, the Service issued an environmental assessment (“EA”) proposing three alternatives for the Project. The Service allowed 30 days of public comment, in which Plaintiffs participated. In December 2007, the Service issued a finding of no significant impact and a final decision selecting the third alternative. On July 1, 2008, Plaintiffs filed suit in the United States District Court for the District of Montana claiming that the Project violates NEPA and NFMA.

The district court granted in part and denied in part Plaintiffs’ motion for summary judgment on October 30, 2008. The court granted summary judgment to Plaintiffs on their claim that the Service violated NFMA by failing to map elk habitat as required by the Gallatin Forest Plan, but granted summary judgment to the Service on all other claims. The court issued an injunction against implementation of the Project and remanded to the Service to conduct the necessary mapping.

In November 2008, the Service issued an EA containing elk-mapping information and solicited public comment. On March 6, 2009, the Service issued a second finding of no significant impact and final decision re-approving the Project.

Plaintiffs filed a new complaint on June 5, 2009, challenging the March 2009 decision and raising the same claims addressed in the earlier case. Rather than opening a new case, the district court issued an order re-opening the original case and consolidating it with the newly filed case. The court explained that it “retains jurisdiction over an injunction, even in the absence of an express statement to that effect.” The court granted the Service’s motion to dismiss all duplicative claims from the earlier decision, and noted that all claims raised in the first and second complaint would be available on appeal. On October 8, 2009, the district court held that the Service had complied with its remand order, granted summary judgment to the Service, and dissolved the injunction.

Plaintiffs appealed the district court’s dismissal of their claims, including those disposed of by the court’s order of October 30, 2008. The parties agree that Plaintiffs’ appeal is timely as to all claims. See Alsea Valley Alliance v. Dep’t of Commerce, 358 F.3d 1181, 1184-86 (9th Cir.2004).

II. Discussion

We review de novo the district court’s grant of summary judgment. See United States v. City of Tacoma, 832 F.3d 574, 578 (9th Cir.2003). We review the Service’s actions for compliance with NEPA and NFMA under the Administrative Procedure Act (“APA”). See Native Ecosystems Council v. U.S. Forest Serv., 428 F.3d 1233 , 1238 (9th Cir.2005). Under the APA, the Service’s action may be set aside if it is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706 (2)(A). A decision is arbitrary and capricious if the Service

Lands Council v. McNair, 537 F.3d 981, 987 (9th Cir.2008) (en banc) (citation and internal quotation marks omitted). “Although the ultimate scope [of our review] may be narrow, the depth must be sufficient for us to be able to comprehend the agency’s handling of the evidence cited or relied upon.” Nw. Goal, for Alternatives to Pesticides v. EPA, 544 F.3d 1043 , 1052 n. 7 (9th Cir.2008) (citation omitted).

A. Plaintiffs’ Claims Under NEPA

[3] NEPA is a purely procedural statute, intended to protect the environment by fostering informed agency decision-making. See California ex. rel. Lockyer v. U.S. Dep’t of Agric., 575 F.3d 999, 1012 (9th Cir.2009). NEPA “does not mandate particular results, but simply provides the necessary process to ensure that federal agencies take a hard look at the environmental consequences of their actions.” High Sierra Hikers Ass’n v. Blackwell, 390 F.3d 630, 639 (9th Cir.2004) (internal quotation marks omitted). NEPA requires agencies to prepare a detailed environmental impact statement (“EIS”) for all “major Federal actions significantly affecting the quality of the human environment.” 42 U.S.C. § 4332 (C). “As a preliminary step, the agency may prepare an Environmental Assessment (‘EA’) to determine whether the environmental impact of the proposed action is significant enough to warrant an EIS.” High Sierra Hikers, 390 F.3d at 639 -40 (citing 40 C.F.R. § 1508.9 ). An EA must include “brief discussions” of the need for the proposal, of reasonable alternatives, and of the anticipated environmental impacts. See 40 C.F.R. § 1508.9 (b).

1. Effect on Wildfires

The Service’s EA cites the reduction of the risk of wildfires to local residents as a primary purpose of the Project. Plaintiffs argue that the Service violated NEPA by failing to address scientific debate concerning whether forest thinning actually reduces wildfire intensity. A failure in an EA to “discuss and consider” evidence contrary to the Service’s position suggests that the Service “did not take the requisite ‘hard look’ at the environmental consequences” of its proposed action. Blue Mountains Biodiversity Project v. Blackwood, 161 F.3d 1208, 1213 (9th Cir.1998).

This case is different from those in which courts have identified significant controversies as to the efficacy of the Service’s proposed methods. In Sierra Club v. Eubanks, 335 F.Supp.2d 1070, 1074, 1077-78 (E.D.Cal.2004), the Service failed to respond to scientific studies showing that its activities would increase fire risk and, instead, relied on studies that did not actually support its position. In Sierra Club v. Bosworth, 199 F.Supp.2d 971, 979-80 (N.D.Cal.2002), the literature review that accompanied the EIS included a report that called into question the Service’s methodology, but that the EIS failed to disclose or analyze.

In this case, the EA acknowledges the limits of the benefits that would be provided by the Project. The EA does not claim that the Project would eliminate wildfires in the area altogether, but merely that it would reduce potential fire severity, in particular crown fires. The EA explains that limiting crown fires would enhance firefighter and public safety by reducing the average rate of fire spread from 1 to 3 miles per hour to 0.1 to 0.5 miles per hour. The Service’s risk reduction calculations are supported by studies conducted in other regions, as well as by extensive modeling.

We therefore affirm the district court’s grant of summary judgment to the Service on this claim.

2. Global Warming

Plaintiffs argue that the Service violated NEPA by failing to discuss global warming in the EA. The Service’s decision to implement a project is arbitrary and capricious under NEPA if an EA or EIS “entirely failed to consider an important aspect of the problem.” Lands Council, 537 F.3d at 987 . However, the Service is only required to focus on the issues “that are truly significant to the action in question.” 40 C.F.R. § 1500.1 (b). Also, “[ijmpacts shall be discussed in proportion to their significance.” 40 C.F.R. § 1502.2 (b).

Plaintiffs point out that global warming has been recognized by courts as an issue of national importance. See Massachusetts v. EPA, 549 U.S. 497, 521 , 127 S.Ct. 1438 , 167 L.Ed.2d 248 (2007); Ctr. for Biological Diversity v. Nat’l Highway Traffic Safety Admin., 538 F.3d 1172, 1221-24 (9th Cir.2008). They also point out that the Deputy Chief for the National Forest System has issued a guidance document directing the Service to incorporate climate change analysis into its evaluations of projects. That guidance document suggests, for example, that a qualitative discussion of climate change would be necessary in an EA for a proposal to underburn 30,000 acres of ponderosa pine stands. It states, however, that proposals require no discussion if they are of a “minor scale [so] that the direct effects would be meaningless.” The Project involves a relatively small amount of land and it will thin rather than clear cut trees. Further, we note that the Service addressed comments regarding climate change in its December 2007 notice of final decision.

We therefore conclude that the EA adequately considered the Project’s impact on global warming in proportion to its significance.

3. Soil Quality

The EA acknowledges that the Project would cause detrimental soil disturbance, particularly through ground-based logging. Plaintiffs argue that the Service violated NEPA because it failed to analyze soil disturbance mitigation measures sufficiently in the EA, and because it should have prepared an EIS evaluating cumulative soil disturbance. We reject both arguments.

The Project seeks to minimize soil disturbance by cutting most of the acreage during the winter such that all ground-based harvest activities using tractors or skidders would be done over snow or frozen ground. The Project would use helicopters on much of the remaining acreage. In the areas subject to ground-based logging, the Project would implement soil restoration, placing five tons of woody debris per acre on old skid roads. The Service provided analytical support for its claim that placing woody debris on old skid trails would reduce or eliminate detrimental soil conditions. The Service’s EA, relying on an expert study, explains that the method would be “beneficial for restoration of soil productivity,” and that it would prevent further soil disturbance from unauthorized ATV use and from weed infestation. It further explains:

The EA’s explanation is not the “perfunctory description of mitigating measures” provided in Neighbors of Cuddy Mountain v. U.S. Forest Service, 137 F.3d 1372 , 1380 (9th Cir.1998). The Service’s analysis in the EA is consistent with other agency analysis, unlike in League of Wilderness Defenders v. Forsgren, 309 F.3d 1181, 1192 (9th Cir.2002). Further, the EA relies on applicable Best Management Practices (“BMPs”), including winter harvesting and placement of woody debris, that are appropriate for this project. See Envtl. Prot. Info. Ctr. v. U.S. Forest Serv., 451 F.3d 1005, 1015-16 (9th Cir.2006) (approving of reliance on specifically identified BMPs); Cf. Blue Mountains Biodiversity Project, 161 F.3d at 1214 (BMPs not appropriate for specific circumstances of project).

We therefore conclude that the EA sufficiently explained that its mitigation measures would minimize, and compensate for, any soil disturbance from the Project.

B. Plaintiffs’ Claims under NFMA

NFMA imposes procedural and substantive requirements on the Service’s management of national forests. Proeedurally, NFMA requires the Service to develop and maintain forest resource management plans. See Ecology Ctr. v. Castaneda, 574 F.3d 652, 656 (9th Cir.2009) (citing 16 U.S.C. § 1604 (a)). After a forest plan is adopted, all subsequent agency actions must comply with that plan. Id. The Gallatin National Forest Plan (“Gallatin Plan” or “Plan”) is the relevant forest plan in this case. Substantively, NFMA requires that the forest plans adopted by the Service provide certain protections, such as protection of forest habitat and diversity of wildlife. See 16 U.S.C. § 1604 (g)(3).

1. Soil Quality

Plaintiffs argue that the Project violates NFMA’s prohibition on timber harvesting that irreversibly damages soil. Under NFMA, the Service may harvest timber from national forests only where “soil, slope, or other watershed conditions will not be irreversibly damaged.” 16 U.S.C. § 1604 (g)(3)(E)®. Because the Gallatin Plan does not itself provide specific standards to implement this requirement, the Service applies soil standards developed for the Northern Region as a whole. The Northern Region Soil Quality Standard prohibits logging that results in more than 15% detrimental soil disturbance in affected areas. In areas where more than 15% detrimental soil conditions already exist from prior activities, the standard allows harvesting only if it does not result in a net increase in soil disturbance.

Due to previous logging, 408 acres of the Project already exceed 15% detrimental soil disturbance. On these acres, and on an additional 28 acres, the Project would employ ground-based harvesting, resulting in an additional 3.5% soil disturbance on average. This represents the equivalent of an additional 15 acres of disturbance. The Service estimates that approximately 15 acres of land would be treated by its woody debris restoration method, thus resulting in no net increase in soil disturbance in the areas.

Plaintiffs respond that it is merely speculative that the Service will follow through on the soil-disturbance mitigation measures described in the EA. The Service has not yet set aside funding for all of its proposed mitigation activities, and the Service has failed to follow through on mitigation in past projects. However, the woody debris restoration method is one of the least expensive proposed mitigation measures, costing only $12,300 over a five-year period. Most of the unfunded mitigation methods in the Project are road treatments that are unrelated to soil disturbance. Lack of funding therefore does not appear to be a serious problem for the soil restoration plans.

We therefore conclude that it was not arbitrary and capricious for the Service to conclude that its mitigation measures will ensure that the Project does not violate the Northern Region soil standards.

2. Old Growth Indicator Species

The Gallatin Plan requires that designated management indicator species (“MIS”) “be monitored to determine population changes.” The designated MIS for the habitat impacted by the Project are the northern goshawk and pine marten. Plaintiffs argue that the Service violated the Gallatin Plan, and therefore NFMA, by failing to monitor northern goshawk and pine marten. They also argue that the Service violated NFMA by not accurately determining the effect of the Project on those species. We reject both arguments.

The EA identifies a number of efforts to monitor the populations of both the northern goshawk and the pine marten in the region. Montana conducts annual studies of pine marten population trends that look at snow tracks of the birds. The studies have detected an average of 75 pine martens per 100 “transect miles” in southwest Montana over a ten-year period. A recent Service study on species viability in the Gallatin Forest concluded that habitat to support the pine marten is abundant in the Project area. The Service has also conducted habitat surveys and population monitoring for northern goshawks in the Gallatin Forest. The Service relied in its EA on several independent surveys concluding that goshawk viability is not a concern. One 2005 study involved a “systematic random survey” that “showed that the goshawk is relatively common and well-distributed in the Northern Region.” Another study from 2005 concluded that “short-term viability of the goshawk in the Northern Region is not an issue.”

In evaluating the effect of the Project on MIS populations, the Service appropriately relied on management of the species’ habitat as a proxy for management of the species themselves. In Lands Council, 537 F.3d at 996 , we approved “of the Forest Service’s use of the amount of suitable habitat for a particular species as a proxy for the viability of that species.” We concluded that such a method was appropriate even where the Service is taking actions that “will disturb some suitable habitat.” Id. at 999 .

We review the Service’s application of the proxy approach under the arbitrary and capricious standard. Id. at 997-98 . The Service “must both 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.” Id. at 998 . We conclude that the Service’s reliance on habitat as proxy in this case was not arbitrary and capricious. In the EA, the Service adequately demonstrated a “knowledge of what quality and quantity of habitat is necessary to support the species,” and its “method for measuring the existing amount of habitat [was] reasonably reliable and accurate.” Lands Council, 537 F.3d at 998-99 (quotation omitted). The EA describes in detail the habitat necessary for the viability of both the pine marten and the goshawk. It adequately describes existing habitat in the Gallatin Forest and the Project area, as well as the Project’s limited effect on the habitat’s ability to sustain MIS. Further, the EA provides for substantial mitigation measures to ensure that a nest is not disturbed in the unlikely event that one is located during implementation of the Project. The Service reasonably concluded that sufficient habitat currently exists for both species and that the Project would have “[n]o direct effects on the pine marten,” and “little, if any, direct affect [sic] on goshawks.”

Plaintiffs nonetheless argue that the Service’s methodology for measuring existing habitat was flawed because the Service failed to adequately field-verify old growth stands. The Service conducted extensive examinations of stands in the field in the 1980s, as well as additional subsequent field verification. The Service is not required to use on-the-ground verification so long as alternative methodologies are reliable. “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.” Lands Council, 537 F.3d at 997 . The Service used multiple databases containing old growth stand information and multiple methods to verify its old growth information in the Project area. This analysis led the Service to predict a range of between 21.7% and 29.3% old growth in Gallatin Forest, with a 90% confidence interval, which places the old growth area well above the 10% minimum required by the Gallatin Plan.

We therefore conclude that the Service’s conclusion that the Project would comply with the Gallatin Plan’s monitoring requirements is not arbitrary and capricious.

3. Yellowstone Cutthroat Trout

The Gallatin Plan requires that “[h]abitat that is essential for [Yellowstone cutthroat trout] be managed to maintain the[ ] species.” As the Service recognized in the EA, streams in the Project area are “extremely important to conservation and recovery of the species.” Yellowstone cutthroat trout in the area are at risk from non-native fish and from “habitat degradation due to roads and timber harvest.” In particular, the Service noted that “any increase in sediment yield from this proposal would perpetuate degraded spawning habitat conditions.” Plaintiffs argue that the Project would violate the Gallatin Plan by increasing sediment levels in the streams due to ground disturbance from harvesting and increased road use. We disagree.

The Project, along with its associated road improvements, would reduce rather than increase long-term sediment levels. Plaintiffs argue that some of the improvements in sediment levels will result from pre-Project road improvements that have already been completed, and therefore are no longer considered part of the Project. But even putting those improvements aside, the Project would reduce long-term levels through other road improvements. Plaintiffs argue that these improvements are not yet fully funded, and therefore not guaranteed. But Plaintiffs fail to show any likelihood of increased long-term sediment levels from the Project even if the road improvements are not completed. We therefore find that the Service was justified in concluding that the Project does not present a threat to cutthroat trout through increases in long-term sediment levels.

The Service has also adequately taken short-term sediment increases into account. We have held agency decisions to be unlawful where they focused only on long-term effects, ignoring short-term effects that jeopardize species. See Pac. Coast Fed’n of Fishermen’s Ass’ns v. U.S. Bureau of Reclamation, 426 F.3d 1082, 1092, 1095 (9th Cir.2005) (the agency failed to provide analysis showing that “the coho would receive sufficient protection against jeopardy under the proposed plan”); Nat’l Wildlife Fed’n v. Nat’l Marine Fisheries Serv., 524 F.3d 917, 935 (9th Cir.2008) (proposal at issue “would have significant negative impacts on each affected species’ critical habitat” and the agency analysis failed to “demonstrate that these impacts would not affect the fishes’ survival and recovery, in light of their short life-cycles and current extremely poor habitat conditions”). But the Project incorporates a number of mitigation measures intended to limit short-term increase in sediment levels. All ground-based harvest activities using tractors or skidders would be done over snow or frozen ground, limiting soil disturbance that might eventually reach the streams; the Project would limit harvest activities within 100 feet of any stream; and the Project would leave untouched those trees identified as most likely to provide stability to soil around streams.

The EA explains that any small short-term increases in sediment levels that nonetheless occur despite the Service’s mitigation efforts would not threaten the viability of the Yellowstone cutthroat trout. Current sediment levels vary by location. In the most severely affected areas, sediment levels are predicted to be 26% above natural levels over the short term. According to Service studies, such levels permit a stream to provide for a population of cutthroat trout 90% the size of its inherent capability.

We therefore conclude that the Service was not arbitrary and capricious in concluding that the Project would not threaten Yellowstone cutthroat trout through either long-term or short-term increases in sediment levels.

4. Road Density

Prior to 2006, the Gallatin Plan limited road density in the forest in order to protect elk habitat. The Plan ensured at least 70% elk effective cover, limiting road density to 0.75 miles/square mile. It did so based on the recommendations of scientists who stated that “[rjoads are undoubtedly the most significant consideration on elk summer range.” The road density restriction was intended to meet the NFMA requirement that the Service “provide for diversity of plant and animal communities.” 16 U.S.C. § 1604 (g)(3)(B); see also 36 C.F.R. § 219.19 (2000) (“Fish and wildlife habitat shall be managed to maintain viable populations.”).

In 2006, the Gallatin Plan was amended to remove the road density restriction. The Forest Supervisor concluded that the restriction “was not scientifically supportable or logical” and that it caused unjustifiable problems. The Supervisor explained that the standard compelled more timber harvests in some areas than was desired, led to disagreements over its application, was ineffective in protecting elk, and was already violated in many areas. The Plan amendment placed all road development decisions within the Travel Management Plan and the Service’s discretion. The current road density in the Project area is 2.1 miles of road/square mile, which well exceeds the previous 0.75 limitation.

Plaintiffs argue that the 2006 amendment to the Gallatin Plan violated NFMA. They claim that the Project therefore violates NFMA because it would be implemented subject to an unlawful plan. The Service responds that the 2006 amendment did not violate NFMA because it was justified by new scientific data and by the circumstances prevailing in the Gallatin National Forest.

We need not decide whether the 2006 amendment violated NFMA because the Project would reduce, not increase, long-term road density in the area. We consider challenges to the lawfulness of a forest plan only to the extent that the contested portion of the plan “plays a causal role with respect to the [Project].” Ohio Forestry Ass’n v. Sierra Club, 523 U.S. 726, 734 , 118 S.Ct. 1665 , 140 L.Ed.2d 921 (1998); see also Neighbors of Cuddy Mountain v. Alexander, 303 F.3d 1059, 1067 (9th Cir.2002). The Project would use only existing roads, some of which would be permanently closed and rehabilitated at the end of the Project. Because road density in the Project area would not be increased as a result of the 2006 amendment, we have no occasion to evaluate its lawfulness.

5. Elk Cover

Plaintiffs’ single meritorious argument on appeal concerns the Gallatin Plan’s elk-cover requirement. The Gallatin Forest Plan requires that the Service “[m]aintain at least two thirds of the hiding cover associated with key habitat components over time. Subsequent timber sale activity will be allowed after regeneration provides hiding cover.” Elk are designated in the Plan as an indicator species of the Gallatin National Forest, for which two-thirds hiding cover must be maintained. Plaintiffs argue that the Project violates the Plan because it would reduce elk cover to under two thirds. We agree.

In preparing the EA, the Service did not measure elk cover according to the definition provided in the Gallatin Plan. The Gallatin Plan defines elk cover as “[v]egetation, primarily trees, capable of hiding 90 percent of an elk seen from a distance of 200 feet or less.” In the EA, the Service relied on two separate measurements of elk cover. One calculation of cover was based on the current prevalence of various tree classifications in the Project area. The Service measured 70-90% elk cover under this definition but does not explain what percent cover this translates to under the Plan definition. The other calculation suggested 62% elk cover under a canopy cover definition. According to a table in the Helena National Forest Plan, 60% elk cover as measured under the canopy cover definition translates to 42% elk cover as measured under the Gallatin Plan definition. The Project therefore violates the Gallatin Plan’s two-thirds elk-cover requirement. The Service’s failure to measure elk cover as defined by the Gallatin Plan renders us “unable to determine from the record that the agency is complying with the forest plan standard.” Native Ecosystems Council v. U.S. Forest Serv., 418 F.3d 953, 962 (9th Cir.2005).

The Service argues that the Gallatin Forest Plan does not require that elk cover exceed 67% at all times, but only that any Service action retain two thirds of then-existing elk cover. The Service claims, in other words, that the Plan prohibits only timber sales that would reduce now-existing elk cover by more than 33%. The Service bases its argument on the wording of the Plan, which requires the Service to maintain “two thirds of the hid ing cover” as opposed to merely “two thirds hiding cover.” Because the Project would reduce now-existing elk cover by less than 33% under any measure, the Service claims that the Project complies with the Gallatin Plan.

“Agencies are entitled to deference to their interpretation of their own regulations, including Forest Plans.” Native Ecosystems Council, 418 F.3d at 960 . “[W]e have effectively treated forest plan directives as equivalent to federal regulations adopted under the APA, deferring to the Forest Service’s interpretation of plan directives that are susceptible to more than one meaning unless the interpretation is plainly erroneous or inconsistent with the directive.” Siskiyou Reg’l Educ. Project v. U.S. Forest Serv., 565 F.3d 545 , 554-55 & n. 9 (9th Cir.2009) (citing Auer v. Robbins, 519 U.S. 452, 461 , 117 S.Ct. 905 , 137 L.Ed.2d 79 (1997)). But the Service’s interpretation of the elk-cover requirement in the Plan is plainly erroneous. The Plan requires that the two-thirds cover be maintained “over time.” It further provides that “[sjubsequent timber sale activity will be allowed after regeneration provides hiding cover.” The Service’s interpretation would allow iterative Service actions to whittle elk cover down to nearly nothing so long as each individual action removed only 33% of then-existing cover.

Alternatively, the Service argues that even if the Project violates the Gallatin Plan’s elk-cover requirement, the error is harmless given the large elk populations in the Project area. But “[i]t is well-settled that the Forest Service’s failure to comply with the provisions of a Forest Plan is a violation of NFMA.” Native Ecosystems Council, 418 F.3d at 961 . “If the Forest Service thinks any provision ... of the Plan is no longer relevant, the agency should propose amendments to the ... Plan altering its standards, in a process complying with NEPA and NFMA.” Id. Although current elk populations may meet or exceed Montana objectives, those objectives cannot replace federal management objectives. The Service’s own research scientists have written, in guidelines for elk management, “Reducing habitat effectiveness should never be considered as a means of controlling elk populations. A population over target is not a Forest Service habitat problem.”

We therefore conclude that the Service has violated the Gallatin Plan, and NFMA, by not ensuring that the Project complies with the current Gallatin Plan elk-cover requirement. We remand to the Service to remedy this error.

Conclusion

We affirm the district court’s grant of summary judgment to the Service in almost all respects. We remand, however, because the Project fails to comply with the Gallatin Plan’s elk-cover requirements.

AFFIRMED in part, REVERSED in part, and REMANDED. Each side to bear its own costs.

. Plaintiffs also argue that the Project violates NFMA due to soil disturbance. We discuss and reject that argument below. See infra Part II.B.l. . Plaintiffs also claim that the Service violated NEPA by failing to take a “hard look” at the Project's effects on MIS. Plaintiffs’ NEPA claim raises no issues not included in its NFMA claim, so we do not analyze the NEPA claim separately. Cf. Lands Council, 537 F.3d at 994-96 .