OPINION
We decide whether the United States Forest Service’s (Service) 2007 Travel Management Plan for parts of the Lewis and Clark National Forest, including the Middle Fork Judith Wilderness Study Area, violates the Montana Wilderness Study Act of 1977 (Study Act) and the National Environmental Policy Act (NEPA). We hold that nothing in the Study Act, which requires the Service to manage a wilderness study area so as to “maintain” its wilderness character as it existed in 1977, prohibits the Service from exercising its discretion to enhance the wilderness character of a study area. We also hold that NEPA does not require the Service to prepare a supplemental draft environmental impact statement (EIS) where, as here, the final decision makes only minor changes and is qualitatively within the spectrum of the alternatives discussed in the draft EIS. We accordingly reverse the judgment of the district court.
Background
In 2007, the Service issued a revised Travel Management Plan governing recreational motorized and nonmotorized use on 1.1 million acres of the Lewis and Clark National Forest. The area covered by the travel plan encompasses the Little Belt Mountains, the Castle Mountains, the north half of the Crazy Mountains and the 81,000-acre Middle Fork Judith Wilderness Study Area.
The Service’s draft environmental impact statement (DEIS), released in July 2006, considered five summer alternatives and three winter alternatives. The most restrictive summer alternative would have allowed motorized use on 1287 miles of roads and trails. The least restrictive summer alternative would have allowed motorized use on 2262 miles of roads and trails.1 Each of the alternatives also would have permitted motorized vehicles within 300 feet of a road or trail for parking (i.e., accessing dispersed campsites), passing or turning around.
The Service’s final plan, issued in October 2007, adopted summer alternative 5, with several modifications, and winter alternative 2.2 Overall, the plan designated *10401366 miles for motorized recreational use, including 870 miles of routes open year-round and another 496 miles open seasonally.3 The plan also designated about 304 miles for groomed over-snow motorized travel and permitted over-snow, cross-country (i.e., off-road, off-trail) motorized travel on 483,000 acres between December 1 and May 15.
The final plan dropped the 300-foot dispersed camping rule and instead allowed “parking, passing, or turning around ... within the length of the vehicle and attached trailer” (about 70 feet), subject to certain conditions designed to ensure safety and protect vegetation, soil and water. The modified dispersed camping rule also permitted off-road travel beyond the “vehicle length plus trailer” limit to access certain established campsites.
Within the Middle Fork Judith Wilderness Study Area, the plan “emphasize[d] non-motorized recreation,” reducing routes designated for motorized recreational use from 112 miles under the previous travel plan to 38 miles under the new plan. The Service’s record of decision explained this choice as follows:
When Congress passed the Montana Wilderness Study Act, it instructed the agency to maintain the wilderness character of the Middle Fork of the Judith Wilderness Study Act Area (WSA). Managing this area primarily for non-motorized use best accomplishes this goal until Congress decides whether or not the area should be designated as wilderness. Presently there is an abundance of motorized use in this area, some of which is necessary to access private land in the middle of the WSA. To balance the need to provide access to private land, the special “highlight of the summer” trip some of the trails provide for motorized users, with the need to maintain wilderness character, I have eliminated motorized use except for one connected complex of trails (approximately 12 miles) and the road system that accesses the private land.
Nine recreational groups having an interest in motorized recreation subsequently filed suit against the Service, seeking to invalidate the travel plan as violating NEPA and the Study Act. The Montana Wilderness Association intervened as a defendant, and the parties filed cross-motions for summary judgment.
The district court granted the recreational groups’ motion for summary judgment and denied the Service’s cross-motion. The court concluded that the Service failed to comply with NEPA by adopting a final decision that “fell outside the range of alternatives [considered in the DEIS] and made numerous, significant changes to the DEIS” without preparing a supplemental environmental impact statement as required by 40 C.F.R. § 1502.9(c), which states that “[a]gencies ... [s]hall prepare supplements to either draft or final environmental impact statements if ... [t]he agency makes substantial changes in the proposed action that are relevant to environmental concerns.” The court concluded that the final decision departed from the range of alternatives discussed in the DEIS in four areas:
*10411. “The chosen decision reduced total mileage open for motorized travel by nearly thirty percent beyond the most restrictive DEIS alternative.”4
2. The final decision “closed several trails not specified for closure in the DEIS.”
3. The final decision “reduced the snowmobile season short of any DEIS alternative.”5
4. The final decision “scrapped a 300-foot-off-road-travel rule for a much more restrictive ‘vehicle plus trailer length’ area.”
The court also granted summary judgment to the recreational groups on their Study Act claim. The court concluded that the Study Act, which directs the Service to administer wilderness study areas “so as to maintain their presently existing wilderness character,” Pub.L. No. 95-150, § 3(a), 91 Stat. 1243 (1977), requires the Service to preserve the wilderness character of a wilderness study area against decline, but prohibits the Service from enhancing the wilderness character of the area. The court said that the Study Act:
directed the Forest [Service] to maintain the wilderness character of Wilderness Study Areas as it existed in 1977. To the extent the wilderness character was there in 1977, it was to be maintained. To the extent the wilderness character was lacking in 1977, it was not to be imposed.
Noting that the final travel plan reduced overall motorized use in the study area from 112 miles to 38 miles, the court concluded that the travel plan “eliminate[d] roughly two-thirds of the previously-available motorized routes” — which the court construed as an impermissible “attempt at enhancement or creation of wilderness character” in the study area.
The court entered judgment setting aside the travel plan in certain respects and remanding to the Service for further action. The Service and the Montana Wilderness Association separately appealed. We consolidated the two appeals, and, after hearing argument, referred the matter for possible mediation. The parties elected not to pursue mediation. The matter is therefore ripe for decision.
Standard of Review
We review de novo a district court’s grant or denial of summary judgment. LaVine v. Blaine Sch. Dist., 257 F.3d 981, 987 (9th Cir.2001). The Study Act and NEPA do not contain their own provisions for judicial review. Accordingly, our review of the Service’s decision under these statutes is governed by the judicial review provisions of the Administrative Procedure Act (APA), 5 U.S.C. §§ 701-06. See City of Sausalito v. O'Neill, 386 F.3d 1186, 1205-06 (9th Cir.2004). Under the APA, agency decisions may be set aside if “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A).
*1042Discussion
I. Montana Wilderness Study Act
The Study Act requires the Service to administer wilderness study areas “so as to maintain their presently existing wilderness character and potential for inclusion in the National Wilderness Preservation System.” Study Act § 3(a). This provision imposes two requirements. First, the Service must administer study areas so as to maintain their wilderness character as it existed in 1977. Second, the Service must administer the areas so as to maintain their potential for designation as wilderness areas — i.e., as part of the National Wilderness Preservation System. The dispute here concerns the first of these requirements.
The recreational groups urge, and the district court concluded, that the obligation to administer study areas so as to “maintain their presently existing wilderness character” prohibits the Service not only from degrading the wilderness character of a study area, but also from enhancing it. They argue that the Service improperly attempted to enhance the study area’s wilderness character by reducing overall motorized use miles in the study area beyond 1977 levels. The Service disputes that interpretation, arguing that the Study Act “creates a floor, not a ceiling, for environmental protection.”
The Service is correct. The Study Act plainly mandates preservation of a base level, but does not prohibit enhancing the area’s wilderness character above that level. Webster’s defines “maintain” as “to keep in a state of repair, efficiency, or validity” and as to “preserve from failure or decline.” Webster’s Third New International Dictionary 1362 (2002). Other dictionaries confirm this meaning. See Owasso Indep. Sch. Dish No. 1-011 v. Falvo, 534 U.S. 426, 433, 122 S.Ct. 934, 151 L.Ed.2d 896 (2002) (“The ordinary meaning of the word ‘maintain’ is ‘to keep in existence or continuance; preserve; retain.’ ” (quoting Random House Dictionary of the English Language 1160 (2d ed.1987))); American Heritage Dictionary of the English Language 1055 (4th ed.2000) (defining “maintain” as “To keep up or carry on; continue,” and as “To keep in an existing state; preserve or retain”); Oxford English Dictionary (online version June 2011) (defining “maintain” to include “To keep up, preserve, cause to continue in being (a state of things, a condition, an activity, etc.); to keep vigorous, effective, or unimpaired; to guard from loss or deterioration.”). In sum, the Study Act simply requires the Service to preserve a study area’s wilderness character against decline. Enhancement of wilderness character is fully consistent with the Study Act’s mandate, although the Study Act does not require it.6
This meaning is confirmed by the purposes of the Study Act. One of the Act’s express aims is to preserve a study area’s “wilderness character” throughout the study period. The Study Act does not define the term “wilderness character,” but the parties agree that it borrows a definition of wilderness from the Wilder*1043ness Act, Pub.L. No. 88-577, 78 Stat. 890 (1964) (codified at 16 U.S.C. § 1131(c)).7 The Wilderness Act defines “wilderness as an area that has, among other things, ‘outstanding opportunities for solitude or a primitive and unconfined type of recreation.’ ” Greater Yellowstone Coal. v. Timchak, No. CV-06-04-E-BLW, 2006 WL 3386731, at *2 (D.Idaho Nov. 21, 2006) (applying the Wyoming Wilderness Act of 1984, which contains language identical to the Study Act) (quoting 16 U.S.C. § 1131(c)).8 The Study Act accordingly “requires the Forest Service to administer [wilderness study areas] to maintain” overall wilderness character, including “opportunities for solitude or primitive and confined recreation[,] that existed there in [1977],” until the area is either designated as a wilderness area or removed from the Study Act. Id. at *3; see also id. at *3-*6 (overturning the Service’s decision permitting increased heli-skiing in the Palisades Wilderness Study Area where the Service failed to show that increased helicopter use would not diminish current users’ available opportunities for solitude compared to 1984 levels). The Service can accomplish this purpose — providing current users with opportunities for solitude comparable to those that existed in 1977 — when the Service either preserves against decline or enhances wilderness character.
The Study Act’s other express aim is to maintain a study area’s “potential for inclusion in the National Wilderness Preservation System.” Study Act § 3(a). Once again, the Service acts consistently with this objective when it either preserves against decline or enhances the wilderness protection of the area. Preserving motorized recreational uses, by contrast, does nothing to maintain the area’s potential for wilderness designation.
The Study Act’s legislative history also shows that enhancement is permitted. The Senate report explains that the Study Act “directs the Secretary to administer the proposed study areas so as not to diminish their presently existing wilderness character and potential.” S.Rep. No. 95-163, at 1 (1977) (emphasis added); see also id. at 2 (stating that the study areas are “to be managed by the Secretary so as not to diminish their presently existing wilderness character and potential” (emphasis added)). The choice of the word “diminish” reveals that Congress intended to protect wilderness character from decline rather than to prevent enhancement.
*1044The recreational groups point out, correctly, that Congress appears to have contemplated that existing recreational activities, including motorized uses, could continue during the study period, so long as those activities did not diminish wilderness character, undermine a study area’s potential for wilderness designation or conflict with the Service’s overall forest management objectives. See S.Rep. No. 95-163, at 2 (1977) (explaining that the “language regarding wilderness character and potential was added by the committee ... to assure continued enjoyment of the areas by those recreationists whose pursuits will not, in the judgment of the Secretary, preclude potential wilderness designation for the areas”); H.R.Rep. No. 95-620 (1977), at 4 (“The use of off-road vehicles, while generally prohibited in designated wilderness areas, is entirely appropriate in wilderness study areas .... ”). Congress did not, however, mandate that motorized recreational levels be maintained.9 And Congress made clear that the Service was free to reduce motorized use levels when carrying out its general obligations to manage national forests — as it has done here. See id. (“Nothing in [the Study Act] will prohibit the use of off-road vehicles, unless the normal Forest Service planning process and travel planning process, which applies to all national forest lands, determines off-road vehicle use to be inappropriate in a given area.”).10
We therefore hold that nothing in the Study Act prohibits the Service from enhancing the wilderness character of a wilderness study area. The district court’s decision that the travel plan violates the Study Act is accordingly reversed.11
II. National Environmental Policy Act
The district court also concluded that the Service violated NEPA by adopting restrictions on motorized use that “fell outside the range of alternatives” considered in the DEIS and by making “numerous, significant changes to the DEIS” without preparing a supplemental draft environmental impact statement, as required by 40 C.F.R. § 1502.9(c)(1)(i). The Service challenges that ruling on appeal. We ordinarily review the Service’s decision not to prepare a supplemental EIS under the arbitrary or capricious standard. Cf. Friends of the Clearwater v. Dombeck, 222 F.3d 552, 556-57 (9th Cir.2000) (applying 40 C.F.R. § 1502.9(e)(1)(ii)). We reverse.
NEPA requires the government to prepare an EIS for any proposed federal action “significantly affecting the quality of *1045the human environment.” 42 U.S.C. § 4332(2)(C). The EIS must address, among other things, “the environmental impact of the proposed action” and “any adverse environmental effects which cannot be avoided should the proposal be implemented.” Id. § 4332(2)(C)(i)-(ii). Effects that may be relevant to the EIS include ecological impacts, such as “effects on natural resources and on the components, structures, and functioning of affected ecosystems,” as well as the “aesthetic, historic, cultural, economic, social, or health” effects of the proposed action. 40 C.F.R. § 1508.8(b). The EIS must “[r]igorously explore and objectively evaluate all reasonable alternatives.” Id. § 1502.14(a).
An agency can modify a proposed action in light of public comments received in response to a draft EIS. See id. § 1503.4(a). “[Ajgencies must have some flexibility to modify alternatives canvassed in the draft EIS to reflect public input.” California v. Block, 690 F.2d 753, 771 (9th Cir.1982). If the final action departs substantially from the alternatives described in the draft EIS, however, a supplemental draft EIS is required: “Agencies ... [sjhall prepare supplements to either draft or final environmental impact statements if [t]he agency makes substantial changes in the proposed action that are relevant to environmental concerns.... ” 40 C.F.R. § 1502.9(c) (emphasis added).
Section 1502.9(c) does not define the terms “substantial changes” and “relevant to environmental concerns.” The Council for Environmental Quality (CEQ), however, has published guidance on when changes to a proposed action will require preparation of a supplemental EIS. The CEQ guidance provides that supplementation is not required when two requirements are satisfied: (1) the new alternative is a “minor variation of one of the alternatives discussed in the draft EIS,” and (2) the new alternative is “qualitatively within the spectrum of alternatives that were discussed in the draft [EIS].” Forty Most Asked Questions Concerning CEQ’s National Environmental Policy Act Regulations [hereinafter “Forty Questions”], 46 Fed.Reg. 18,026, 18,035 (Mar. 23, 1981) (emphasis added).
The First, Eighth and Tenth Circuits have adopted this CEQ guidance as a framework for applying § 1502.9(c)(1)(i). See New Mexico ex rel. Richardson v. Bureau of Land Mgmt., 565 F.3d 683, 705 & n. 25 (10th Cir.2009); In re Operation of Missouri River Sys. Litig., 516 F.3d 688, 693 (8th Cir.2008); Dubois v. U.S. Dep’t of Agric., 102 F.3d 1273, 1292 (1st Cir.1996).12 We now join them in doing so.
Here, the district court pointed to four changes in the Service’s decision that the court concluded required supplementation of the DEIS. We address each in turn.
A. Overall Motorized Use Mileage
The district court concluded that the Service was required to prepare a supplemental draft EIS because the final decision “reduced total mileage open for motorized travel by nearly thirty percent beyond the most restrictive DEIS alternative.” According to the court, the most restrictive alternative considered in the *1046DEIS (summer alternative 4) permitted motorized use on 1951 miles, and the least restrictive DEIS alternative (summer alternative 1) permitted motorized use on 3036 miles. The court found that the final decision, which allowed motorized use on just 1366 miles, fell outside the range of alternatives considered in the DEIS, summarizing the figures as follows:
Total Miles Designated by Vehicle Type
Passenger hxk. ATV Motorcycle Total
Summer 1 1523 514 260 739 3036
Summer 4 955 397 262 337 1951
Decision 741 178 222 225 1366
The Service argues that the DEIS total mileage figures relied on by the district court are inappropriate for comparison to the 1366 mileage figure for the final plan because they double count route miles that are open to several motorized uses. As we have noted, under the travel plan “lower-level” vehicles can use routes designated for “higher-level” vehicles. Thus, for example, a route that is open to both motorcycle and ATV use is counted twice in the mileage totals relied on by the district court. The Service has offered numbers that avoid double counting, and these numbers show that the 1366 total motorized route miles permitted in the final decision fall within the range of alternatives discussed in the DEIS:
Total Miles Designated for Motorized Use
Summer 1 2262
Summer 3 1774
Summer 4 1287
Summer 5 1441
Decision 1366
The recreational groups offer no credible reason to doubt either the accuracy of the Service’s mileage figures, which are supported by the administrative record, or the Service’s assertion that the numbers relied on by the district court are inappropriate for comparison to the 1366 figure because they reflect double counting. We therefore credit the Service’s mileage figures. As a consequence, the overall motorized use miles authorized by the travel plan are within the range of alternatives included in the DEIS. The district court’s finding of a NEPA violation therefore inadvertently relied on a mistaken premise.13
B. Trail Closures Not Specified in the DEIS
The district court faulted the Service for not preparing a supplemental draft EIS because the final decision included several trail closures that were not included in any *1047of the alternatives discussed in the DEIS. Neither the district court nor the recreational groups, however, explain why these modifications were “substantial changes ... relevant to environmental concerns.” 40 C.F.R. § 1502.9(c)(l)(i). They appear to have been “minor variation[s]” that were “qualitatively within the spectrum of alternatives that were discussed in the draft [EIS].” Forty Questions, supra, 46 Fed.Reg. at 18,035. Accordingly, no supplemental draft EIS was required.
C. Modification of the End Date of the Snowmobile Season
The district court concluded that the Service was required to prepare a supplemental draft EIS because the Service changed the end of the snowmobile season from May 15 in the DEIS to May 1 in the final decision. The Service restored the May 15 date during the appeals phase of the administrative proceedings. Any continued objection to this change is therefore moot.
D. Modification of the Dispersed Camping Rule
Finally, the district court concluded that the Service violated NEPA because the final decision included a modified dispersed camping rule that was not discussed in a supplemental draft EIS. The DEIS proposed permitting off-road driving in a 300-foot corridor on either side of roads and trails for parking, passing and turning around. In the final decision, the Service dropped that proposal in favor of a plan allowing “parking, passing, or turning around ... within the length of the vehicle and attached trailer,” subject to certain conditions. The plan also permitted off-road travel beyond the “vehicle length plus trailer” limit to access certain established dispersed campsites. The Service explained its decision as follows:
Several respondents were concerned about the possibility of a 600-foot wide roaded corridor resulting from the 300 foot dispersed camping rule. I did not find specific information in the Final EIS (FEIS) that indicates this was happening on the ground, however, in too many cases, users are developing roads or trails out of dispersed camp sites and extending the area covered by the 300 foot rule. Once a trail or road is created, another user may travel on that newly created road or trail and camp 300 feet beyond the newly developed road or trail, allowing the road or trail to continue to grow. My decision will reduce the creation of new trails out of dispersed camp sites by prohibiting travel off designated routes to a campsite, while still allowing access to continue to the majority of existing dispersed campsites.
The Service says the district court’s decision requiring supplementation was wrong for two reasons. We reject the Service’s first argument but agree with its second.
The Service argues that it was not required to prepare a supplemental draft EIS because the changes to the dispersed camping rule would have only lessened environmental impacts in comparison to the alternatives discussed in the DEIS.14 The Service contends that a change in a proposed action that only lessens environmental impacts is, as a categorical matter, *1048not a change that is “relevant to environmental concerns” for purposes of § 1502.9(c)(l)(i).
We agree with the Service up to a point. That a modified alternative only lessens environmental impacts may tend to show that the new alternative is a “minor variation of one of the alternatives discussed in the draft EIS” and is “qualitatively within the spectrum of alternatives that were discussed in the draft [EIS].” Forty Questions, supra, 46 Fed.Reg. at 18,035. See, e.g., Sierra Club v. Van Antwerp, 526 F.3d 1353, 1360 (11th Cir.2008) (“When the change to the proposed action is a ‘minimizing measure,’ ... the agency ‘is not automatically required to redo the entire environmental analysis’ ... because a minimizing measure’s effects on the environment will usually fall within the scope of the original NEPA analysis” (quoting Sierra Club v. U.S. Army Corps of Eng’rs, 295 F.3d 1209, 1221 (11th Cir.2002))); Friends of the Bow v. Thompson, 124 F.3d 1210, 1218-19 (10th Cir.1997) (“Although we are not prepared to say that a reduction in the environmental impact of an action can never trigger a requirement to prepare a supplemental [environmental assessment], we believe that a reduction in environmental impact is less likely to be considered a substantial change relevant to environmental concerns than would be an increase in the environmental impact.”).
A new alternative, however, may lessen environmental impacts and yet fall outside the range of alternatives discussed in a draft EIS. Supplementation may be required, for example, when modifications to a proposed action, although lessening environmental impacts, also alter the overall cost-benefit analysis of the proposed action. In Massachusetts v. Watt, 716 F.2d 946, 948-49 (1st Cir.1983) (Breyer, J.), for instance, the court required preparation of a supplemental analysis when the government lowered its estimate of the benefits expected to be gained from granting oil drilling leases in the North Atlantic Ocean. The court reasoned that additional analysis was required because the adverse environmental consequences of the action — although diminished — might no longer be justified in light of the drastically reduced expectation of economic benefit. See id.
The Service relies on the Third Circuit’s decision in South Trenton Residents Against 29 v. Federal Highway Administration, 176 F.3d 658, 664-66 (3d Cir.1999). That case, however, involved a Department of Transportation (DOT) regulation that expressly omits a duty to prepare a supplemental EIS when a new alternative lessens environmental impacts. The DOT regulation states that “a supplemental EIS will not be necessary where ... [t]he changes to the proposed action ... result in a lessening of adverse environmental impacts evaluated in the EIS without causing other environmental impacts that are significant and were not evaluated in the EIS.” 23 C.F.R. § 771.130(b). Section 1502.9(c) contains no similar language. The government’s reliance on South Trenton Residents is therefore misplaced.
Although we disagree with the Service that modifications that lessen impacts never require supplementation, we nonetheless agree that the modifications to the dispersed camping rule here did not require preparation of a supplemental draft EIS. We hold that the final decision was a “minor variation” and “qualitatively within the spectrum of alternatives” discussed in the DEIS. Forty Questions, supra, 46 Fed. Reg. at 18,035.
We reach this conclusion for several reasons. First, the dispersed camping rule is a secondary rather than a primary aspect of the overall travel plan. Second, the modifications are relatively minor — scaling back the 300-foot limit to approximately 70 *1049feet. Third, there is very little reason to believe the modified travel plan will have environmental impacts that the agency has not already considered. The Service modified the rule to eliminate the risk that users would develop roads or trails out of dispersed camp sites. The change eliminates that adverse impact, leaving only the impacts caused by parking, passing and turning around that have already been fully considered. Although the modified dispersed camping rule has the potential to concentrate motorized travel in a smaller area, which could theoretically pose different or additional impacts relative to the 300-foot rule, there does not appear to have been much actual use beyond the 70-foot perimeter even under the original rule. Finally, there is no indication that the modifications alter the overall cost-benefit analysis of the proposed action.
The recreational groups argue that supplementation is required in light of the Tenth and First Circuits’ decisions in New Mexico ex rel. Richardson v. Bureau of Land Mgmt. (“BLM”), 565 F.3d 683 (10th Cir.2009), and Dubois v. U.S. Department of Agriculture, 102 F.3d 1273 (1st Cir. 1996). BLM and Dubois, however, involved substantial modifications that went to the heart of the proposed action and posed new and previously unconsidered environmental questions. They are therefore readily distinguishable from the dispersed camping modifications adopted here.15
Conclusion
We hold that the Service’s 2007 travel plan conforms to the Study Act and NEPA. The judgment of the district court is accordingly reversed.
REVERSED.