Kurt Meister v. U.S. Department of Agriculture

OPINION KETHLEDGE, Circuit Judge. An agency is not entitled to deference simply because it is an agency. It is true that agencies are more specialized than courts are. But for courts to defer to them, agencies must do more than announce the fact of their comparative advantage; they must actually use it. And that means, among many other things, that the agency must apply — rather than disregard' — the relevant statutory and regulatory criteria. Kurt Meister, a Michigan attorney appearing pro se, argues that the United States Forest Service disregarded the relevant criteria here. Specifically, he claims that the Service failed to comply with several of its own regulations and one federal statute in developing its 2006 management plan for the Huron-Manistee National Forests in Northern Michigan. For the most part, we agree with him; and to that extent we reverse the district court’s entry of judgment in the Service’s favor and remand the case so that the Service may comply with those requirements forthwith. *368 I. A. This case concerns the Service’s management of recreational activities in the Huron-Manistee National Forests (the “Forests” or “Forest”). The Forests occupy about 970,000 acres on each side of the northern one-third of Michigan’s Lower Peninsula. In the east, the Huron National Forest ranges between 12 and 30 miles long from north to south, and stretches 60 miles wide from west to east, reaching the shores of Lake Huron. In the west, the Manistee National Forest is about 75 miles long and 40 miles wide, reaching Lake Michigan near Manistee. The Forests serve a variety of interests, including recreation, timber harvesting, and wildlife habitat. They lie within a two-hour drive of 7.4 million residents of Michigan, and draw visitors from beyond the State. Although technically separate, the Forests have been managed as a single unit since 1945. 1. Every national forest is subject to a “land and resource management plan[.]” 16 U.S.C. § 1604(a). The National Forest Management Act requires each forest’s plan to be revised every fifteen years, id. § 1604(f)(5), but in practice the interval often stretches to twenty. Developing a plan is a formidable process: “The Service must develop its management plans in conjunction with coordinated planning by a specially-designated interdisciplinary team, extensive public participation and comment, and related efforts of other federal agencies, state and local governments, and Indian tribes.” Sierra Club v. Marita, 46 F.3d 606, 609 (7th Cir.1995) (citing 36 C.F.R. §§ 219.4-219.7). The Service’s own regulations prescribe in great detail the procedures the Service must follow in developing a forest plan. Those regulations make clear, as the Service itself states in this appeal, that the Service must balance competing uses of the Forests. See 36 C.F.R. § 219.4(a), (b); 16 U.S.C. § 1604(e)(1). (We cite the 2000 version of the regulations throughout, which is what the Service used in developing the Plan.) Under the National Environmental Policy Act, the Service must also discuss a range of alternative plans and assess the environmental impact of the alternative that it proposes to adopt. 42 U.S.C. § 4332(2)(C)(i), (in). Once the development process is complete, a proposed final plan and final environmental impact statement “are sent to the Regional Forester, who directs one of four national forest regions, for review.” Marita, 46 F.3d at 609. If the Regional Forester approves the plan and the impact statement, he issues both documents along with a Record of Decision explaining his reasoning. The approved plan and impact statement may be appealed to the Chief of the Forest Service, whose ruling becomes a final administrative decision if the Secretary of Agriculture chooses not to review it. 36 C.F.R. §§ 219.10(d), 211.18. 2. The Service issued a management plan for the Forests in 1986. In 2003, the Service published a notice of intent to revise the plan. The Service thereafter held public meetings and solicited public comments as to how to revise the plan. In 2005, the Service published a draft environmental impact statement that described three alternative plans and designated a preferred one. After receiving additional public comments for three months, the Service prepared a final plan (the “Plan”) and a final environmental impact statement for the Forests. In a Record of Decision dated March 2006, the responsible Regional Forester approved both documents. See 71 Fed.Reg. 30399. *369 3. Meister commented on the Plan throughout its development. Those comments reveal fluency with the language of the relevant statutes and regulations; and they explained in considerable detail why Meister thought the Service was not meeting its obligations under the law. His principal comment was that, in developing the Plan, the Service had disregarded certain processes prescribed in its own regulations, so as to favor gun hunters and snowmobile users over other persons — for example, hikers and birdwatchers — who use the Forests for quiet, solitary activities. He also commented that the Service should close more areas of the. Forests to motorized activity than the Service seemed likely to close in the Plan. It appears that the Service disagreed with all of Meister’s comments. Meister thereafter appealed the Plan administratively. The Service Chief issued a decision that upheld the Plan in all but one respect: the Service was prohibited from expanding snowmobiling to all open, unplowed roads in the Forests. The Secretary of Agriculture declined to review the Reviewing Officer’s decision. Meister then filed suit in the district court, challenging the Plan under the Administrative Procedures Act. See 5 U.S.C. § 706. He and the Service each filed motions for summary judgment. The district court granted the Service’s motion, holding in general terms that the Service had complied with the applicable regulations. The district court denied Meister’s motion. This appeal followed. II. A. The Service does not argue before us that Meister lacks standing to assert his claims. But standing and ripeness have proved uncertain in cases like this one. See, e.g., Ohio Forestry Ass’n, Inc. v. Sierra Club, 523 U.S. 726, 118 S.Ct. 1665, 140 L.Ed.2d 921 (1998). So we pause to address those issues here. Standing has three elements. “First, the plaintiff must have suffered an ‘injury in fact’ — an invasion of a legally protected interest which is (a) concrete and particularized, and (b) actual or imminent, not conjectural or hypothetical.” Lujan v. Defenders of Wildlife, 504 U.S. 555, 560, 112 S.Ct. 2130, 119 L.Ed.2d 351 (1992) (internal citations and quotation marks omitted). Second, the injury must be “fairly traceable to the challenged action of the defendant.” Id. (internal alterations omitted). Third, it must be likely that the injury will be “redressed by a favorable decision.” Id. at 561, 112 S.Ct. 2130. The Supreme Court has “held that environmental plaintiffs adequately allege injury in fact when they aver that they use the affected area and are persons for whom the aesthetic and recreational values of the area will be lessened by the challenged activity.” Friends of the Earth, Inc. v. Laidlaw Envtl. Servs. (TOC), Inc., 528 U.S. 167, 183, 120 S.Ct. 693, 145 L.Ed.2d 610 (2000) (internal quotation marks omitted). Here, Meister alleges that he uses the Forests and that the Service’s failures to follow certain procedures in developing the Plan have resulted in more widespread gun hunting and snowmobile usage than there would be had the Service complied with the procedures. Consequently, Meister contends, the Service’s failures have “substantially diminishe[d] Plaintiffs walking, hiking, mountain biking, kayaking, cross country skiing, *370 and snowshoeing experiences” in the Forests. Per Friends of the Earth, that contention states an injury in fact. There remains the question of ripeness. The doctrine seeks “avoidance of premature adjudication”; and, in cases involving challenges to agency action, ripeness depends on whether the “administrative decision has been formalized and its effects felt in a concrete way by the challenging parties.” Ohio Forestry, 523 U.S. at 732-33, 118 S.Ct. 1665 (internal quotation marks omitted). In Ohio Forestry, the Sierra Club sought to challenge provisions of a forest-management plan that designated certain areas as open to logging. The Supreme Court did not question that the aesthetic harm alleged by the Club would amount to an injury in fact. But the Court observed that the Plan itself “does not give anyone a legal right to cut trees, nor does it abolish anyone’s legal authority to object to trees being cut.” Id. at 733, 118 S.Ct. 1665. To the contrary, additional agency action— namely, issuance of a site-specific permit— was required before anyone could engage in the logging that the Sierra Club said would harm its interests. Thus, the Court held, the plan had not yet “inflict[ed] significant practical harm upon the interests that the Sierra Club advance[d.]” Id. Hence the case was not ripe. Meister’s case is different. Unlike logging, the activities about which Meister complains — gun hunting and snowmobile use — do not require further action by the Service before they can occur. To the contrary, they have in fact occurred ever since the Plan’s issuance, with the resultant harms that Meister now alleges. Thus, the Plan itself has harmed him in concrete ways. His claims are ripe. B. We turn next to our standard of review. As an initial matter, we do not defer to the district court’s decision, but instead review the administrative decision as if we were the first reviewing court. Schuck v. Frank, 27 F.3d 194, 197 (6th Cir.1994). Our review of the agency’s decision is governed by the Administrative Procedure Act. The APA’s relevant section provides in relevant part: The reviewing court shall- — • (2) hold unlawful and set aside agency action, findings, and conclusions found to be— (A) arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law; [or] ... (D) without observance of procedure required by law[.] 5 U.S.C. § 706. Meister’s claims arise in part under § 706(2)(D), since he contends that the Service failed to follow certain procedures in implementing the Plan. Under that subsection we review agency action de novo. Coalition for Gov’t Procurement v. Fed. Prison Indus., Inc., 365 F.3d 435, 457 (6th Cir.2004). But even in cases arising under § 706(2)(D), our review as a practical matter is often more deferential than that. The reason is that the question whether a certain procedure is required in a particular circumstance, or whether a certain methodology satisfies the procedure, is often left to the agency’s discretion. See, e.g., Save Our Cumberland Mountains v. Kempthorne, 453 F.3d 334, 339 (6th Cir.2006) (stating that we review an agency’s “decision that an environmental impact statement need not be prepared, under the deferential ‘arbitrary and capricious’ standard”); Hughes River Watershed Conservancy v. Johnson, 165 F.3d 283, 289 (4th Cir.1999) (“[a]gencies are en *371 titled to select their own methodology as long as that methodology is reasonable”). So even in cases arising under § 706(2)(D), the arbitrary-and-capricious standard frequently governs. Meister’s claims also arise to some extent under § 706(2)(A), since he contends that certain aspects of the Plan were arbitrary. As a general matter, agency action is arbitrary or capricious if the agency has relied on factors which Congress has not intended it to consider, entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise. 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). Applying these standards, we turn to the merits of Meister’s claims. III. A. “It is an elemental principle of administrative law that agencies are bound to follow their own regulations.” Wilson v. Comm’r of Soc. Sec., 378 F.3d 541, 545 (6th Cir.2004). Meister claims that the Forest Service violated three of its own procedural regulations in developing the Plan. 1. Meister first claims that the Forest Service failed to comply with § 219.21(a)(2) of its regulations when it developed the Plan. That subsection provides: (a) Forest planning shall identify— (2) The recreational preferences of user groups and the settings needed to provide quality recreation opportunities!!] The parties disagree about what this provision means. The Forest Service reads “recreational preferences” to mean “activities,” so that this regulation requires “essentially, a demand/supply analysis.” Gov’t Br. at 15. Under this interpretation, the Service must estimate the demand for various activities in the Forest — e.g., camping, cross-country skiing, and snowmobiling — and then assess the quantity of Forest lands “needed” to meet the demand for each activity. Meister disagrees, arguing that a different provision, § 219.21(b), requires a demand-supply analysis. (It does, but of a type not relevant here — namely, for “developed recreational facilities[.]”) Under Meister’s interpretation, the identification required by subsection (a)(2) would lay the foundation for a demand-supply analysis, rather than constitute the analysis itself. Specifically, he reads “recreational preferences of user groups” to mean not the users’ preferred activities themselves, but rather what those users seek to obtain from those activities; as, for example, “cross-country skiers desire a quiet recreation experience.” Meister Reply at 7-8. So in Meister’s view, the Service must determine the kinds of physical settings in which various users would prefer to engage in their activities, rather than merely the settings in which those activities can occur. Thus, for example, the Service should determine that cross-country skiers, for a variety of reasons, do not want their trails located in close proximity to snowmobile trails. Meister’s ultimate concern seems to be that the Service’s interpretation allows it to consider subjectively unsatisfactory settings, rather than the users’ preferred ones, in determining whether an *372 adequate supply of Forest lands exists for a particular activity. We defer to an agency’s plausible interpretation of its own regulation. Auer v. Robbins, 519 U.S. 452, 461, 117 S.Ct. 905, 137 L.Ed.2d 79 (1997). “Recreational preferences,” as used in § 219.21(a)(2), can just as easily be read to mean the activities that users prefer, as it can the physical settings in which they prefer to do them. And though not as easily, “settings needed” can be read to mean the quantity of lands necessary to support an activity. So the Service’s interpretation of this subsection lies within its discretion; and thus the subsection does require essentially a demand-supply analysis. But the question remains: Supply of what? The regulation provides a straightforward answer: The supply of lands “needed to provide quality recreation opportunities!.]” 36 C.F.R. § 219.21(a)(2) (emphasis added). It is not enough, therefore, for the Service merely to identify the supply of lands on which an activity can occur. It must instead identify the supply of lands on which participants in that activity are afforded a “quality recreation opportunity.]” So Meister has a point after all. With this interpretation in hand, we consider the substance of Meister’s claim. He has two complaints. First, he says the Service failed to identify “the settings needed to provide quality recreation opportunities for hikers, backpackers, and cross-country skiers!.]” Meister Br. at 18. The Service responds that a methodology it calls the “Recreation Opportunity Spectrum” (“ROS”) accomplishes that identification for those activities and many more. The ROS classifies forest lands into seven types, depending on their degree of development. The classifications are: Primitive, Semiprimitive Non-motorized, Semiprimitive Motorized, Road-ed Natural, Rural, and Urban. ROS Users Guide at 2. The ROS contains a detailed description of each classification. “The Service is entitled to use its own methodology, unless it is irrational.” Marita, 46 F.3d at 621; see also California v. Watt, 712 F.2d 584, 597 (D.C.Cir.1983). The ROS easily passes that test: It is not an irrational, but a thoughtful methodology for matching settings and activities, among other planning purposes. Suffice it to say that we agree with the Service on this point. But Meister’s other complaint is more problematic. He contends that, as a part of its demand-supply analysis, the Service significantly overestimated snowmobile use and underestimated cross-country ski activity in the Forests. The issue begins with the fact that the Service’s methodology for measuring the intensity of different activities in the Forests — the “National Visitor Use Monitoring” (“NVUM”) survey — indicated that no visitors to the Forests reported snowmobiling or cross-country skiing as the primary purpose of their visit in the year 2000. Recreational Supply and Demand Analysis (Oct. 2004) (“Demand Analysis”) at 11. “[B]ased upon discussions with Dr. Daniel Stynes, Professor, Michigan State University!,]” however, and the “professional judgment of the Huron-Manistee’s staff and management,” the Service adjusted its estimate of annual snowmobile visits from zero to 120,000 for the year 2000 and to 138,000 for the year 2010. Final Environmental Impact Statement (March 2006) (“Final Statement”) at III-337, 287; Demand Analysis at 21. But the Service made no parallel adjustment for cross-country visitors, notwithstanding its recognition that demand “for this activity is projected to increase approximately 50 percent over the next 50 years!,]” a rate “faster than the rate of population *373 growth.” Demand, Analysis at 30, 23. The estimates of cross-country visitors, for the years 2000 to 2050, were left at zero. Meister says this treatment of the numbers amounts to “everything imaginable short of making up the data about snowmobile use.” Meister Br. at 17 n. 4. The Service responds that it “documented its methodology” in the Demand Analysis and that “[c]ommunications with the Michigan State researcher and internal discussions about his opinions were also well-documented in the administrative record.” Gov’t Br. at 22. The whole episode debases the coinage of agency deference. We begin with the discussions with Dr. Stynes. So far as the record is concerned, the discussions are in fact nothing more than an single email exchange comprising several pages. The exchange is informal — Dr. Stynes himself says at the outset that “I’m not ready to present formal estimates for a peer review” — and the exchange actually focuses almost entirely on Dr. Stynes’ belief that the total number of annual visits (as opposed to just snowmobile visits) to the Forests should be adjusted from 1.3 million (the number suggested by the NVUM) to 3 million. A Forest Service Social Scientist, Dr. Donald English, expressed serious concerns about that proposed adjustment, stating that “I reiterate that I think it would require a fair bit of work to justify using some other visitation estimate other than the NVUM estimate.” So far as we can tell, no further work was done — -another Service official said simply to choose either number, “[jjust make sure it’s something defensible” — and the Service chose the estimate of 3 million. But the validity of that number is not before us today. What is before us is the snowmobile number, which as noted above was adjusted from zero to 120,000 for the year 2000. According to the Service’s Demand Analysis, “Stynes estimated that snowmobile visits on the National Forest were approximately 120,000” for that year. Demand Analysis at 11. But that characterization is unfair to Dr. Stynes: He made no such estimate. What he does say — in a passage comprising three sentences — is that the NVUM survey was surely wrong in estimating zero snowmobile visits to the Forest, that an earlier State study estimated about 1.2 million snowmobile user-days in the entire northern lower peninsula of Michigan in 1996, and that if “10% of the activity is on the Forests, we get 120,000 [National Forests] snowmobile visits.” There is nothing more. Those three sentences, so far as we can tell from the record, are the entire basis of the Service’s adjustment of the snowmobile estimate from zero to 120,000. So we see two serious problems with the Service’s estimates of snowmobile and cross-country visitors. The first is that there is scarcely any basis for the snowmobile estimate. Behind all of the Service’s invocations of communications, discussions, and professional judgment, there lie only those three sentences. Those sentences themselves contain not an estimate, but a hypothetical. The record otherwise contains not a hint of professional judgment on the Service’s part as to why the 120,000 number, as opposed to some other number, is reasonable. We mean no criticism of Dr. Stynes, since in this respect his email was put to a use that he appears not to have contemplated. But we do say that the nation’s forests deserve better than this as the basis for significant decisions regarding their management. The second problem is the disparate treatment of the cross-country estimate. Most, if not all, of the Service’s rationales for boosting the snowmobile numbers generally (as opposed to pegging them at 120,-000 specifically) — namely, “abnormally low *374 snowfall” during the survey year, high gas prices, and that participants in the activity typically enter the Forests via trails rather than at developed sites where the survey forms were located — apply fully to cross-country visitors. Meister quite reasonably asks why snowmobile visits received an upward adjustment, but cross-country visits did not. The Service responds that there was “[insufficient data” from which to make an estimate of cross-country visitors. Given the data found sufficient for the snowmobile estimate, that response is hardly persuasive. And given that the Demand Analysis contains reliable (and likely peer-reviewed) data showing the participation rate in cross-country skiing to be about half that for snowmobiling in the parts of the nation near the Forests, see Demand Analysis at 23, it would seem that once the Service had a snowmobile estimate in hand, a cross-country estimate would soon follow. But there is none: So far as the tables in the Demand Analysis are concerned, the Forests have seen their last cross-country skier. Deference must be earned. Kadia v. Gonzales, 501 F.3d 817, 821 (7th Cir.2007). It was not earned in the two respects that we discuss here. The Service’s estimates of snowmobile and cross-country visitors to the Forests are entirely arbitrary. And — under the Service’s own interpretation of § 219.21(a)(2)' — those estimates are integral to the demand-supply analysis required under that subsection. To that extent, the Service’s issuance of the Plan was arbitrary and without observance of procedure required by law. 5 U.S.C. § 706(2)(A), (D). 2. Meister also claims that, in developing the Plan, the Service did not coordinate with Michigan’s recreational planning as required by § 219.21(e). That subsection provides that the Service’s planning efforts shall be coordinated to the extent feasible with present and proposed recreation activities of local and State land use or outdoor recreation plans, particularly the State Comprehensive Outdoor Recreation Plan, and recreation opportunities already present and available on other public and private lands, with the aim of reducing duplication in meeting recreation demands. 36 C.F.R. § 219.21(e). In plainer English, the subsection requires the Service, “to the extent feasible, to coordinate with the State in eliminating duplicate recreation opportunities on state and national forests.” Gov’t Br. at 16. Plainer still: The acreage already dedicated to a particular activity on state land should be considered in deciding how much federal land to allocate to the activity. Meister claims that the Service failed to coordinate in this manner with respect to gun hunting and snowmobiling. The first step in this coordination, it would seem, would be simply to determine how much land is already available for these activities on state lands. The Service has not done that. In its Answer to Meister’s Complaint, the Forest Service admitted' — long after the Plan had been approved — that it “lack[s] sufficient knowledge or information to respond to the allegations concerning firearm hunting on state forest lands managed by the State of Miehigan[.]” The Service made essentially the same admission with respect to snowmobiling on state lands. It is hard to see how the Service could even attempt to reduce duplication of hunting and snowmobiling opportunities on state lands without that basic information. Meister also seeks to show actual duplication as a means of proving that the Service made no attempt to reduce it. Specifically, he notes — and the Service *375 does not dispute — that “[hjunting is permitted on approximately 3.8 million acres of state forest land and over 340,000 acres of state game and wildlife areas in the State of Michigan[,J” and that “there are more than 6,500 miles of snowmobile trails in the State[.J” Meister Br. at 14, 15. He further notes that hunting is permitted virtually everywhere in the Forests, save perhaps the parking lots- — -meaning that the Forests’ duplication of state-land hunting opportunities, to the extent there is any within the meaning of the rule, literally could not be greater. And he asks specifically why it was not feasible to reduce duplication by prohibiting gun hunting (as opposed to bow hunting, which is largely silent, and to which Meister seems not to object) and snowmobiling in at least the Primitive and Semiprimitive Nonmotorized areas of the Forest. Those areas, as defined by the ROS, are supposed to bring an “extremely high probability” and a “high probability,” respectively, “of experiencing isolation from the sights and sounds of humans.” ROS Users Guide, Table 1 (“Experience Characterization”). Firearm and snowmobile use, as the Service itself recognizes, involve high-intensity noises created by humans. Primitive and Semiprimitive areas together comprise less than 66,000 acres of the 970,000 acres occupied by the Forests, or just 6.75% of the Forests’ acreage. See Final Statement at III-275. Meister therefore contends that the Service’s failure to restrict these activities in even these areas- — particularly in light of the circa 4.1 million acres open to hunting on state land, and the 6,500 miles of snowmobile trails there — is strong evidence that the Service did not comply with the coordination procedures mandated by § 219.21(e) so far as gun hunting and snowmobiling are concerned. The Service’s response is insubstantial. First, it says that it uses the State Comprehensive Outdoor Recreation Plan as a tool to monitor recreation demand. But that only means the Service has done half of what its regulation requires. Section 219.21(e), as shown above, requires consideration of the extent to which the supply of state lands already meets that demand. And the Service admits it lacks knowledge on that point. Second, the Service again tells us that it consulted with Dr. Stynes, and that its “communications” with him and “internal discussions” regarding his opinions are “well-documented.” Gov’t Br. at 25. We are referred to the same email exchange. The exchange has even less to say about duplication (that is, nothing) than it does about snowmobile visits. Finally, the Service cites a letter from the Michigan Department of Natural Resources, in which the State says that the Plan’s increases in Semiprimitive Nonmotorized and Semiprimitive Motorized areas “work[ ] to meet the demand” for those areas and “may supplement existing inter-agency recreational efforts that address recreation demand in a cooperative manner (i.e. ORV trails, pathways, campground placement).” Final Statement at J-157-58. The State’s comments speak well of the Service’s decision to increase Semiprimitive acreage in the Forests. But they simply have nothing to do with whether hunting and snowmobiling opportunities on Forest lands duplicate those on state lands. Thus, with respect to hunting and snowmobiling, there is no evidence that the Service complied with § 219.21(e), and ample evidence that it did not. In that respect as well, therefore, the Plan’s approval was “without observance of procedure required by law.” 5 U.S.C. § 706(2)(D). 3. Meister’s remaining claim under the Management Act is that the Service failed *376 to comply with 36 C.F.R. § 219.21(g). That regulation provides: Off-road vehicle use shall be planned and implemented to protect land and other resources, promote public safety, and minimize conflicts with other uses of the National Forest System lands. Forest planning shall evaluate the potential effects of vehicle use off roads and, on the basis of the requirements of 36 CFR part 295 of this chapter, classify areas and trails of National Forest System lands as to whether or not off-road vehicle use may be permitted. This regulation plainly contemplates that off-road vehicle use may conflict with other interests and uses of the Forests. In Meister’s view, the regulation means that, “if there is a conflict between off-road vehicles and other uses, off-road vehicles lose.” Meister Br. at 19. The Service disagrees. We agree with the Service: The regulation requires it only to “minimize conflicts” between off-road vehicles and other uses, not to eliminate them. But the mandate to minimize remains. Meister contends it was not met. He is concerned about snowmobiles in particular; and the basis of his claim is that— as the Service itself states — snowmobile activities “have continued to occur in or near several existing semiprimitive nonmotorized management areas.” Final Statement at III — 313. Those areas, as noted above, are supposed to be characterized by “a large probability of isolation from the sights and sounds of others.” Id. at III-312. Therein lies the conflict; and Meister suggests that its elimination was feasible enough to fall within the mandate to minimize. The Service offers several responses. First, it says that “the forest plan is not an appropriate vehicle for making individualized trail closure decisions.” Gov’t Br. at 26. That is true so far as it goes; once the Service decides that an area should generally be open to off-road vehicle usage, the decision whether to close trails within it can properly be made on an individualized basis at the field level. But § 219.21(g) by its plain terms imposes the bulk of its obligations at the planning stage; and one such obligation, as we read the regulation, is to determine whether certain classes of areas and trails ought to be altogether off-limits to off-road vehicle use. Meister says that one such class are trails that the Service itself admits are “in or near” Semi-primitive Nonmotorized areas. That claim is properly presented at the Plan level. The Service’s other responses go to the merits. One is that “[s]ome of the nonconformities result because the snowmobile trail coincides with a county road that the Service does not have jurisdiction to close.” Gov’t Br. at 27. That is a good answer as to those trails. Less good is that other trails (which we call “pre-designation trails”) “existed prior to their designation or proposal for designation.” Final Statement at III-313. The argument seems to be that these trails were used by off-road vehicles before being designated for that use, and thus should be grandfathered in. But elsewhere in the Plan documents, the Service repeatedly emphasizes the importance of limiting off-road vehicle usage to “designated routes only.” Record of Decision at 10. So off-road vehicle use on undesignated trails is a bad thing; and thus it hardly seems rational to say that a trail should remain open to off-road vehicles simply because the trail was used by them before its designation. And it is likewise incoherent to say — as a basis for keeping other trails (“club trails”) open to off-road vehicles — that “[s]nowmobile clubs use motorized equipment to groom snowmobile trails on roads within several of the current semiprimitive nonmotorized management areas.” Final Statement at III- *377 313. The mere fact of a nonconformity is no reason to allow it to continue. There conceivably might be reasons for keeping pre-designation and club trails open to off-road vehicle usage — though presumably the reasons should be good ones, given the Service’s own recognition that these uses are “nonconformities.” If the Service were to articulate good reasons, we would defer to them. As the record now stands with respect to these trails, however, the Service has not “articulate[d] 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., 463 U.S. at 43, 103 S.Ct. 2856 (internal quotation marks omitted). Thus, as matters now stand, the Service has not complied with § 219.21(g)’s mandate to minimize conflicts between off-road vehicle use and other uses and interests of the Forests. To that extent, the Plan’s approval was arbitrary and without observance of procedure required by law. 5 U.S.C. § 706(2)(A), (D). B. Meister’s remaining two claims arise under the National Environmental Policy Act. The Act “serves procedural rather than substantive goals.” Save Our Cumberland Mountains, 453 F.3d at 338. The Act has a specialized standard of review for arbitrariness: “In deciding whether the agency acted arbitrarily, we will not substitute our own judgment” for that of the agency, “but we will insist that the agency has, in fact, adequately studied the issue and taken a hard look at the environmental consequences of its decision.” Id. at 339 (internal quotation marks omitted). The Act requires, among other things, that an agency “study, develop, and describe appropriate alternatives to recommended courses of action in any proposal which involves unresolved conflicts concerning alternative uses of available resources.” 42 U.S.C. § 4332(2)(E). Meister argues that there were unresolved conflicts between different recreational uses of the Forests here, and that the Service should have, but did not, consider his proposed alternatives to resolve them. 1. Meister’s first claim is based upon a conflict between the high-intensity noise generated by gun hunting and snowmobile use, on the one hand, and the quiet recreational activities that Primitive and Semi-primitive Nonmotorized areas are supposed to support, on the other. He argues that the Service should have, but did not, consider the alternative of closing these areas to gun hunting and snowmobiling as a means of resolving this conflict. Meister proposed this alternative during the administrative process, but the Service rejected it out of hand. “As a general matter, the range of alternatives that must be discussed under the National Environmental Policy Act is a matter within an agency’s discretion.” Save Our Cumberland Mountains, 453 F.3d at 342 (internal quotation marks omitted). Thus, an “agency may apply a ‘rule of reason’ in this area and discuss only ‘reasonable’ alternatives to the proposed action.” Id. at 346. The question, then, is whether the Service acted within its discretion in rejecting Meister’s alternative without any discussion. That rejection was based on a series of factual and legal errors. The principal factual error is that, during both the administrative and now the legal process, the Service has treated Meister’s proposed restriction on hunting as something it obviously is not: Namely, a proposal that the Service impose “[a] complete ban on hunt *378 ing in the Forests[.]” Gov’t Br. at 31; see also May 27, 2007 letter from Forest Supervisor Leanne Marten to Kurt Meister (“it is my determination that a general closure” of the Forests to hunting “is not justifiable”); Final Statement at J-93 (same). All but one paragraph of the government’s briefing of this claim is dedicated to ruling out such a complete ban. (The Service did accurately characterize Meister’s proposal in its decision rejecting this part of his administrative appeal, but in doing so merely adopted Supervisor Marten’s complete-ban rationale.) But a complete ban is not what Meister proposes. He proposes that gun hunting and snowmobiling be prohibited only in Primitive and Semiprimitive Nonmotorized areas (the “challenged areas”), which as noted above comprise only 6.75% of the Forests. See, e.g., Meister Br. at 21 (arguing that the Service “impermissibly failed to consider alternatives to a general closure such as only closing Wilderness and Semi-Primitive Non-Motorized areas to hunting and snowmobiling”); Feb. 28, 2003 letter from Kurt Meister to James Schuler (stating that gun hunting and snowmobiling “should be banned in all Primitive and Semi-Primitive Non-Motorized areas”). And the Service has compounded (no pun intended) its error by overlooking that Meister proposes to restrict only gun hunting, not bow hunting. These aspects of Meister’s proposal are hardly abstruse; and they narrow the proposal’s scope both qualitatively (so that it concerns only gun hunting rather than bow) and quantitatively (so that it concerns only 6.75% of the Forests’ acreage rather than 100%). An agency cannot reject a proposed alternative by treating it as something it is not. To exercise discretion with respect to a proposed alternative, the agency must treat the alternative on its own terms. The Service has not done that here. In its brief to this court, the Service does briefly discuss the more narrow closure (although it still overlooks the gun versus bow distinction). In doing so the Service makes different contentions with respect to snowmobiles and gun hunting. As to the former, the Service essentially contends that Congress did not require it to close the challenged areas to snowmobiling. But Congress did not prohibit such a closure either; and the mere fact that Congress left the adoption of that alternative to the agency’s judgment does not mean that the agency was free not to consider the alternative at all. The National Environmental Policy Act and its regulations say precisely the contrary. As to gun hunting, the Service contends that Congressional policy favors leaving the entire Forest open to hunting, so that even the narrow closure that Meister proposes is beyond its authority. Specifically, the Service notes that the Multiple-Use Sustained-Yield Act of 1960, 16 U.S.C. § 528, the Wilderness Act, 16 U.S.C. § 1133(d)(7), and the Federal Land Policy and Management Act, 43 U.S.C. § 1732(b), each make clear that they do not affect “the jurisdiction or responsibilities of the several States with respect to wildlife and fish on the national forests.” 16 U.S.C. § 528. But there are clear exceptions to that rule. The Federal Land Policy and Management Act expressly provides that the Service “may designate areas of public land and of lands in the National Forest System where, and establish periods when, no hunting or fishing will be permitted for reasons of public safety, administration, or compliance with provisions of applicable law.” 43 U.S.C. § 1732(b) (emphasis added). The Service’s own guidelines echo the point: *379 Hunting, fishing, and trapping of fish and wildlife and associated practices on National Forest System lands are subject to State fish and wildlife laws and regulations, unless one or both of the following apply: 1. State fish and wildlife laws and regulations conflict with Federal laws; or 2. State laws and regulations would permit activities that conflict with land and resource management responsibilities of the Forest Service or that are inconsistent with direction in forest plans. Forest Service Manual § 2643.1 (emphasis added). Gun hunting is inconsistent with the “direction in forest plans” as set forth in the ROS descriptions of the challenged areas, since those areas are supposed to present little chance of encountering noise by humans. The Service cannot expect us to defer to its ROS descriptions when they support its decision (which we have done above), but then to disregard those same descriptions when they conflict with its decision. And the descriptions conflict with the decision here. Meister has cited precisely that conflict throughout the administrative and now legal process. The question whether the Service’s own specific descriptions of the challenged areas, on the one hand, should prevail over the State’s much more general rules permitting gun hunting, on the other, is one that warrants a careful answer under the National Environmental Policy Act. See Save Our Cumberland Mountains, 453 F.3d at 347. Meister’s alternative likewise warrants consideration on grounds of compliance with applicable law. 43 U.S.C. § 1732(b). The Service, to its credit, forthrightly recognizes in its brief that it “has been charged by Congress with balancing a variety of competing interests in its land use planning.” Gov’t Br. at 28 (citing 16 U.S.C. § 1604(e)(1) and 36 C.F.R. § 219.1(a), (b)). But striking a balance typically involves some give on each side. We think that should be especially true for participants in activities that do not conform to the area descriptions in the ROS. One might at least expect them not to have the run of the areas. And so, in striking a balance between competing uses of the Forests, one might expect the Service seriously to consider whether, say, birdwatchers in fall should be able to enjoy their pastime, in 6.75% of the Forests, without ducking for the occasional gunshot. Or whether, in some corner of the Forest— especially an ostensibly “nonmotorized” one — a snowshoer should be able to walk a trail without hearing the whine of snowmobile engines. The Service is charged with balancing competing uses of the Forests, rather than favoring one or two uses above all others. And if that balance requires closure of certain areas to certain activities, Congress has granted the Service that authority. There is no lawful policy that ties the Service’s hands in this regard. “[A]n alternative within the ambit of an existing standard ... generally may not be abandoned without any consideration whatsoever.” Save Our Cumberland Mountains, 453 F.3d at 347 (internal quotation marks omitted). The Service was incorrect to conclude that Meister’s proposed alternative fell outside the ambit of the relevant standards here. It seems more likely to us that the Service’s decision not to balance these competing uses, and to disregard its own ROS descriptions, is what fell outside the relevant standards. Meister’s proposed alternative deserves serious consideration under the National Environmental Policy Act. *380 2. Meister’s final claim is that the Service should have considered designating substantially more areas as Semiprimitive Nonmotorized than it did in the Plan. Unlike Meister’s other proposed alternative, the Service did consider something akin to this one, but concluded that the alternative was not viable. It had rational reasons for that conclusion. See, e.g., Final Statement at III-313 (“[a]fter reviewing such things as Recreation Opportunity Spectrum, road densities, old growth and the management and ownership of adjoining private lands, no new areas were identified for consideration as semiprimitive management areas”). We defer to those reasons, and reject this claim. IV. We summarize our holdings today. First, the Service’s estimates of snowmobile and cross-country visitors to the Forests are arbitrary. Thus, the Service has not complied with § 219.21(a)(2)’s requirement of a demand-supply analysis. Second, the Service has not complied with the requirement that it coordinate its recreational planning with that of the State of Michigan with the aim (to the extent feasible) of “reducing duplication in meeting recreation demands” with respect to gun hunting and snowmobiling. 36 C.F.R. § 219.21(e) (2000). Third, the Service’s reasons for keeping pre-designation and club trails open to snowmobile use are arbitrary. Thus, the Service has not complied with § 219.21(g)’s mandate to minimize conflicts between off-road vehicle use and other uses and interests of the Forests. Fourth, the Service violated the National Environmental Policy Act when it failed to consider whether to close Primitive and Semiprimitive Nonmotorized areas to gun hunting and snowmobile use, as Meister has proposed. Each of these failures was material to the Plan’s development. To that extent, the Plan’s approval was arbitrary or without observance of procedures required by law. Given that holding, we have authority to “set aside” the Plan. See 5 U.S.C. § 706(2)(A), (D). We choose not to exercise that authority today, but instead grant the Service a reasonable time to adopt a plan that complies with the law. Ninety days from the date of our mandate seems to us ample time for that compliance. The district court may extend that period upon some showing that the court finds compelling; but in any event the Service shall comply forthwith. The district court’s judgment is reversed with respect to the claims summarized in Part IV of this opinion. Meister is entitled to judgment on those claims to the extent described in that Part; and the claims are remanded to the district court for further proceedings consistent with this opinion. The district court’s judgment is otherwise affirmed.