The issue in this case is whether federal agencies adequately followed our environmental laws both procedurally and substantively in approving a road-building project for Stimson Lumber Company (“Stimson”). Stimson sought an easement in order to access its land surrounded by the Colville National Forest. This forest is home to several threatened or endangered species. Once Stimson has access to its land, it will manage it for perpetual logging.
The United States Forest Service (“Forest Service”), in granting the easement to Stimson, was required to complete an Environmental Impact Statement (“EIS”). The Forest Service and the United States Fish and Wildlife Service (“Fish & Wildlife”) had to ensure that granting the easement would not jeopardize the continued existence of any animal species. Both the EIS and the no-jeopardy determination demanded that the Forest Service and Fish & Wildlife contemplate the “cumulative impacts” of the easement on the land and animals in the area. Selkirk Conservation Alliance and other environmental groups (“Selkirk”) contend that the decision to grant the easement was arbitrary and capricious because the agencies failed to consider cumulative impacts and that Fish & Wildlife did not rely on the best information available in determining the likely harm to species. The district court granted summary judgment in favor of Stimson, the Forest Service, and Fish & Wildlife, and dismissed the claims brought by Selkirk challenging the project. We affirm.1
I
A
Stimson owns six parcels of land in the LeClerc Creek watershed in northeast Washington State within the Colville National Forest (“Colville”), approximately 2,240 acres in total. Such parcels of land are called “inholdings.” Five of these parcels are entirely surrounded by -Colville land, and the only reasonable access route to the sixth parcel is over the Colville land. For this reason, Stimson’s predecessor-in-interest, Plum Creek Timber Company, asked the Forest Service to provide access to the inholdings. The Forest Service responded by proposing to grant an easement across Colville land (the “Stimson Project”).2
*949The Forest Service authorized construction of 1.88 miles of new road and reconstruction of 0.81 miles of old road on Forest Service land within the Colville. Once the Stimson Project is completed, Stimson plans to build at least 15.4 miles of inhold-ing road and harvest 1,577 acres on Stimson’s privately owned forest lands accessed by the easement.
The land to be accessed by the Stimson Project lies within the Selkirk Mountains. The Selkirk Mountains straddle the Washington-Idaho border and extend north into the Canadian Rockies. This area hosts approximately 50 grizzly bears and contains about 6 percent of the grizzly-bear-occupied range in the continental United States. In an attempt to monitor and support this grizzly bear population, the Selkirk Mountains are divided into ten Bear Management Units (“BMUs”). The LeClerc BMU is one of these areas. The Interagency Grizzly Bear Committee, which demarked the bear management units, considered that each unit would provide an appropriate area in which to monitor and analyze the bears. The entire Stimson Project and the lands it will reach fall within the LeClerc BMU.
In connection with the Stimson Project, the Forest Service sought formal consultation with Fish & Wildlife in 1993. Fish & Wildlife evaluated the Stimson Project’s impact on threatened or endangered species and created a draft biological opinion in 1994 that found the easements would place some species in jeopardy. The draft opinion stated that “the proposed action will jeopardize the grizzly bear by increasing the potential for direct mortality to grizzly bears due to increased human use of roads and the increased visual access provided by these roads.” The biological opinion was then put on hold while the Forest Service, Fish & Wildlife, and Stimson’s predecessor-in-interest negotiated a multi-party Conservation Agreement intended to mitigate the effects of the Stimson Project. Stimson and the agencies signed a final Conservation Agreement on January 17,1997.
B •
The 1997 Conservation Agreement dictated the terms by which Stimson would manage all of its lands in the LeClerc BMU, not just those lands to be accessed by the Stimson Project. According to Fish & Wildlife, the Agreement “spells out a cooperative management plan to minimize effects to the grizzly bear in the LeClerc BMU.” In particular, the Agreement aims to “minimize displacement of grizzly bears from spring range, to maintain functional female grizzly bear home range in the BMU, and to reduce the potential for human-caused mortality.”
To those ends, the Agreement imposed dozens of requirements on Stimson’s management of its lands in the LeClerc BMU. Stimson agreed to restrict all activities, including harvesting and road building, in spring range areas when bears are out of their dens. That is, Stimson could only harvest those areas when bears are den-ning in the winter. The Agreement also restricted Stimson’s ability to build roads on its privately owned lands, prohibited net gains in open-road densities, and mandated that road construction maintain “visual screening” (e.g., trees) into bear habitat. The Agreement required Stimson to maintain 40 percent “cover” (areas of prime bear habitat) in the LeClerc BMU and provided that all harvest units “be layed [sic] out so that no point in the unit is more than 600 feet from cover.” Finally, the Agreement created an extensive monitoring system. Stimson agreed to monitor “road densities, levels of vehicular use, and seclusion habitat.” Stimson and the agencies agreed that the “monitoring *950results and the Agreement guidelines will be reviewed by the Parties annually ... and the guidelines will be appropriately revised.”
Fish & Wildlife issued a biological opinion on June 20, 1997. Relying heavily on the mitigating effects of the Conservation Agreement in evaluating the Stimson Project’s impact on habitat, the opinion concluded that the Stimson Project would not jeopardize any of the threatened or endangered species in the area. Fish & Wildlife stated that the Agreement mitigated the concerns about the grizzly bears “in several ways: Open road densities will be limited .... there will be no net increase in total road densities, and no net decrease in core [e.g., prime bear habitat] ...” and “[i]mplementation of [the Agreement’s provisions] together will add approximately 6,962 acres of grizzly bear habitat with low levels of motor vehicle access, and provide a large block of available spring habitat.”
The biological opinion concluded that: The proposed project ... is not likely to result in jeopardy to the species [grizzly bears] because:
Implementation of the guidelines in the Agreement should improve grizzly bear occupancy in spring range, especially in and near the Winter Logging Areas; the Agreement results in loss of a large block of core habitat, however measures are provided to improve seasonal secure areas for the grizzly bears, and to minimize motor vehicle use on restricted roads;
monitoring will be conducted....
Thus, Fish & Wildlife believed that the Stimson Project, assuming a fully implemented Conservation Agreement, would not place any species in jeopardy.
In commenting on this opinion, the Forest Service questioned whether the Conservation Agreement’s lowering of bear mortality risk, which focuses mostly on restrictions to spring habitat, would affect bear mortalities in the critical fall season when hunters abound. Despite this skepticism about the ability of the Conservation Agreement’s mitigation measures to lower bear mortality, the Forest Service issued a Finding of No Significant Impact, meaning that the Forest Service would not study the issue further or prepare an EIS. That Finding was appealed administratively and reversed. The Forest Service thereafter developed a Draft EIS.
C
Stimson, the Forest Service, and Fish & Wildlife entered into a revised Conservation Agreement on February 1, 2000. As revised, the Agreement creates an ecosystem-based management plan throughout the LeClerc BMU. The 2000 Agreement has many of the same features as the 1997 Agreement. In particular, the Agreement requires Stimson to manage the timber to guarantee tree “cover” for bears every 600 feet, to prevent a net increase in open roads, to limit Stimson’s use of Stimson’s private roads during the spring, and restricts Stimson’s operation of winter logging areas during the times the bears are not in their dens.
D
The Forest Service issued a Final EIS for the Stimson Project in September 2000. The EIS identified six alternatives. For each alternative considered, the EIS evaluated the direct, indirect, and cumulative impacts of the Stimson Project on the following resources (among others): un-roaded areas, old-growth vegetation, soil, water quality, fisheries, caribou, grizzly bear, lynx, bull trout, plant life, road access, range, recreation, fire, and heritage and historic resources.
*951With respect to the geographic scope of the EIS, the Forest Service evaluated cumulative effects by focusing on species-management geographic units. For the grizzly bears, the EIS examined the cumulative effects of the Stimson Project in the context of the LeClerc BMU.
The Forest Service limited the geographic scope of the EIS to the LeClerc BMU, despite a separate Stimson request for access to Stimson lands in the nearby Idaho Panhandle National Forest (“IPNF”). The lands that Stimson will access in the IPNF are just across a ridge-line from the lands that Stimson will access in the LeClerc BMU. Nonetheless, the Forest Service concluded that the IPNF request would not generally affect the cumulative-effects analysis for the Stimson Project in the LeClerc BMU, although the Forest Service did include the IPNF project in its analysis of the cumulative effects on roadless areas. The Forest Service cautioned that expanding the study area for the Stimson Project EIS to consider the IPNF project would skew the analysis of the Stimson Project by improperly spreading out the impact on a larger area.
The Forest Service constrained the geographic scope of the EIS even though it had previously acknowledged that “[m]ost radioed female bears within the Selkirk ecosystem do utilize portions of more than one BMU.” That is, bears do not observe the BMU demarcations, but instead wander across BMUs. The Forest Service reasoned that “[t]his does not, however, negate the value of a BMU for cumulative effects analysis.” The Forest Service maintained the LeClerc BMU as the geographic boundary because BMU “delineation objectives were to include areas with all bear seasonal habitat components and to be large enough to encompass the home range of a female grizzly bear.” The Forest Service, again, was generally concerned that evaluating “cumulative effects on too large of an area can dilute the effects of a proposed activity.” With regard to the specific Stimson projects in the IPNF, the Forest Service concluded that they would not have a cumulative impact because “[t]he two Stimson proposals [Le-Clerc and IPNF] would not result in a connected transportation system and lie within separate watersheds [and] viewsheds.... The ridgeline separates the watersheds, causing hydrological effects to be separate; it also separates the viewsheds, and serves as a boundary line for analysis of wildlife effects.”
The EIS did not reference or discuss the impact of several3 specific Stimson harvests that were to take place inside the LeClerc BMU, although not on land that would be made accessible by the easements granted by the Forest Service. The Forest Service, instead, analyzed the effect of Stimson’s lumber harvest by evaluating the impact of the Conservation Agreement. The Conservation Agreement governs Stimson’s behavior on all of its lands within the LeClerc BMU. When the Forest Service analyzed the effects of the Conservation Agreement, the Forest Service presumed that Stimson would manage all of its lands for perpetual logging. Because the Conservation Agreement universally dictates Stimson’s harvesting activity in the LeClerc BMU, and the EIS assumed Stimson would harvest all of its lands, the Forest Service considered evaluation of the Conservation Agreement to be the primary method of analyzing the cumulative impacts of future Stimson activities.
*952The Forest Service also limited the temporal scope of the EIS. The Forest Service projected the effects of Stimson’s actions for only three years. This time frame was chosen, even though Timothy Bertram — a Forest Service scientist — initially projected the impact of Stimson’s activities over a period of ten years, using average historical levels of harvest. Bertram chose this long-term forecast because “it was apparent that there would be activity beyond the three-year period.” Also, the three-year window was selected in spite of the Conservation Agreement’s five-year term. According to Bertram, the EIS ultimately evaluated the effects for three years because the EIS “was based upon the information we were provided by Stimson, which only provided specific information through the year 2003,” despite the five-year term of the Conservation Agreement.
E
Fish & Wildlife issued a final biological opinion on May 17, 2001. This biological opinion evaluated the impact of the Stimson Project on the threatened and endangered species in the area and concluded that the Stimson Project will not jeopardize these species. The biological opinion conceded that introducing more roads into the LeClerc BMU and facilitating more logging will adversely affect grizzly bears. But the opinion considered the effect of the Conservation Agreement and concluded that, with the Agreement in place, the overall effect of the Stimson Project will not jeopardize the grizzly bears.
The biological opinion concluded that the Agreement sufficiently mitigates the Stimson Project’s effects on the bears, even though biological assessments performed by the Forest Service indicated that the Agreement does not eliminate some of the primary causes of bear mortality and environmental degradation. For example, Fish & Wildlife relied heavily on the Agreement’s restrictions of net gains in “open” roads (those roads open to public use). However, the Agreement does not prevent a net gain in “closed” or private-use roads on Stimson lands, but instead places restrictions on when, where, and how they can be built, maintained, and used. The Forest Service had previously concluded that “closed” roads can be as harmful to bears as “open” roads because it is difficult to keep public motorists, snowmobilers, hunters, and hikers off of “closed” roads. The Forest Service also expressed concerns about the efficacy of restrictions on Stimson’s administrative use of closed roads. The Agreement restricts Stimson’s access to its “closed” roads from April 1 — June 15 so as not to interfere with the critical post-denning period. But Forest Service biologists were concerned that the post-denning period often runs into July, when Stimson is free to use its “closed” roads as it chooses.
Aside from the Forest Service’s concerns about the road restrictions, the Forest Service raised questions about the restrictions on when Stimson may harvest trees. The Agreement prohibits Stimson from logging certain areas during the post-denning period. For the same reasons the Forest Service was concerned that spring road restrictions might be inadequate — the bears often extend their post-denning period into July — the Forest Service commented that these harvesting restrictions are “not likely to adequately minimize displacement of grizzly bears during the critical post den emergence period.”
Fish & Wildlife scientists shared many of the concerns expressed by their colleagues at the Forest Service. Fish & Wildlife acknowledged that closed roads are bad for bears, and that the Agreement allows the proliferation of closed roads and does not fully restrict Stimson’s use of *953those roads. Fish & Wildlife also noted that the Agreement will not maintain current amounts of “core” areas. In addition, the Stimson Project will result in several thousand acres of logging, further displacing grizzly bears. Despite all of these concerns, Fish & Wildlife concluded that— although the Stimson Project may adversely affect the grizzlies' — it will. not jeopardize their existence. Again, Fish & Wildlife relied on the Conservation Agreement in reaching this conclusion.4
Fish & Wildlife noted in the 2001 biological opinion (much as it had in its 1997 opinion) that the Agreement would restrict the impact on grizzly bears because: “Open road densities will be limited”; Stimson will be confined to limited use of its closed roads during the post-denning period; “Stimson will avoid commercial use in 4,480 acres of ‘winter logging areas,’ ” thereby “secur[ing] low elevation % spring habitat”; the Agreement “will add approximately 6,962 acres of available spring habitat”; “40 percent minimum cover[will be] maintained within the BMU”; “no point in [a harvest] unit is more than 600 feet from cover”; and Stimson and the Forest Service will monitor the efficacy of the Agreement.
Although the Forest Service and Fish & Wildlife doubted that the Agreement alleviated every negative effect the Stimson Project would have on the grizzly bear, the agencies concluded that the Agreement sufficiently lowered the overall effects of the Stimson Project. Due to these safeguards, Fish & Wildlife concluded that the Stimson Project “is not likely to jeopardize the continued existence of the grizzly bear.”
F
Selkirk filed this action on October 12, 2001, in the District of Oregon, arguing that the Forest Service and Fish & Wildlife violated the National Environmental Policy Act, 42 U.S.C. § 4321 et seq. (“NEPA”), and the Endangered Species Act, 16 U.S.C. § 1536 (“ESA”). Selkirk argued that the Forest Service violated these laws by failing to consider the cumulative impacts of the Stimson Project. Selkirk contended that Fish & Wildlife also violated ESA by not evaluating the cumulative impacts of the project, and by not basing the biological opinion on “the best scientific and commercial data available.”
The parties filed cross-motions for summary judgment on April 22, 2002. The district court denied Selkirk’s summary judgment motion, granted the defendants’ summary judgment motions, and dismissed the case on June 17, 2002. Selkirk filed a notice of appeal on June 25, 2002. We have jurisdiction pursuant to 28 U.S.C. § 1291.
II
We review de novo the district court’s grant of summary judgment. See Hall v. Norton, 266 F.3d 969, 975 (9th Cir.2001). The Administrative Procedure Act governs our review of agency decisions under NEPA and ESA. It mandates that an agency decision may be overturned only where it is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A). Within this narrow review, we cannot substitute our judgment for that of the Forest Service and Fish & Wildlife, but instead must uphold the agency decisions so long *954as the agencies have “considered the relevant factors and articulated a rational connection between the facts found and the choice made.” Washington Crab Producers, Inc. v. Mosbacher, 924 F.2d 1438, 1441 (9th Cir.1990) (quotation marks and citation omitted).
Selkirk argues that the agencies violated federal environmental laws in five different ways: 1) both agencies violated ESA by relying on the Conservation Agreement to mitigate the effects of the Stimson Project; 2) the Forest Service violated NEPA by not considering a wide enough geographic area in its cumulative effects analysis in the EIS; 3) the Forest Service violated NEPA by not considering reasonably foreseeable Stimson activities in its cumulative effects analysis in the EIS; 4) the Forest Service violated NEPA by employing too short of a time frame for the EIS analysis; and 5) Fish & Wildlife violated ESA by not considering reasonably foreseeable Stimson activities in the 2001 biological opinion.
A
Because the terms of the Conservation Agreement are central to the agencies’ decisions, we start by considering Selkirk’s challenge to the agencies’ evaluation of and reliance on the Agreement.
Selkirk argues that the Forest Service and Fish & Wildlife violated ESA by relying on the Conservation Agreement to reduce the risks the Stimson Project poses to grizzly bears. Selkirk reasons that relying on the Conservation Agreement violated ESA’s requirement that the biological opinion and the agency action be based on “the best scientific and commercial data available.” 16 U.S.C. § 1536(a)(2).
The “best scientific and commercial data” requirement prevents the haphazard implementation, “on the basis of speculation or surmise,” of ESA. Bennett v. Spear, 520 U.S. 154, 176, 117 S.Ct. 1154, 137 L.Ed.2d 281 (1997). Selkirk argues that the best available science contradicts the conclusion that the measures contained in the Conservation Agreement sufficiently mitigate the harm caused to the bears by the road-building project. Selkirk contends that it was arbitrary and capricious to rely on the Conservation Agreement to remedy concerns raised in the 1994 determination that the Stimson Project (as it then existed) would jeopardize grizzly bears.
Disputes involving “primarily issues of fact” must be resolved in favor of the expert agency so long as the agency’s decision is based on a reasoned evaluation of the relevant factors. Marsh v. Oregon Natural Resources Council, 490 U.S. 360, 377-78, 109 S.Ct. 1851, 104 L.Ed.2d 377 (1989). Particularly when the analysis “requires a high level of technical expertise,” this Court “must defer to the informed ( discretion of the responsible federal agencies.” Id. at 377, 109 S.Ct. 1851 (quotation marks and citation omitted). As Selkirk concedes in its brief, the agencies’ treatment of the Conservation Agreement may only be subverted by this Court if the Forest Service and Fish & Wildlife acted arbitrarily and capriciously. 5 U.S.C. § 706(2)(A).
We first address whether it was proper for the agencies to rely on the mitigation measures contained in the Conservation Agreement. We will then turn to the particular dispute over whether these mitigation measures, in light of the best scientific data available, sufficiently reduce the harm posed to grizzly bears by the Stimson Project.
As recounted by Fish & Wildlife, “[t]he Agreement spells out a strategy to cooperate and coordinate in management for grizzly bears.” After realizing that the *955proposed action would jeopardize the existence of sensitive species, Stimson’s predecessor-in-interest, the Forest Service, and Fish & Wildlife began a series of meetings designed to lower the risks the project posed for grizzly bears. After a year of consultation, the parties designed an outline of “measures to minimize effects” and sent the outline to grizzly bear researchers for commentary. After another year of discussions, Stimson, the Forest Service, and Fish & Wildlife entered into the 1997 Conservation Agreement that contained a variety of mitigation measures, set forth above.
In assessing the significance of the Agreement, we note that Stimson owns thousands of acres of land within the Le-Clerc BMU, only a fraction of which will be accessed by the Stimson Project. But the Agreement establishes standards of timber management that Stimson must follow on all of its LeClerc lands, not just those lands to be accessed by the Stimson Project. Absent the Agreement, Stimson would be free to harvest timber and build roads on its LeClerc holdings in whatever manner Stimson chose, so long as Stimson complied with state and federal timber regulations. The Agreement thus imposes obligations on Stimson that go far beyond state and federal laws.
Even though the concept of a cooperative Conservation Agreement is attractive, and ought to be encouraged, federal agencies cannot delegate the protection of the environment to public-private accords. Even given the cooperation of private entities, the agencies must vigilantly and independently enforce environmental laws. In this ease, that means that the Forest Service and Fish & Wildlife were still required to make their agency decisions based on “the best scientific and commercial data available.” 16 U.S.C. § 1536(a)(2).
We have previously held that an agency may fulfill its duties under our environmental laws even 'if the agency relies on private mitigation' promises when making environmental assessments. In Friends of Endangered Species, Inc. v. Jantzen, 760 F.2d 976 (9th Cir.1985), a private real estate developer sought to build thousands of residential units and millions of square feet of commercial space on San Bruno Mountain, an area that was inhabited by a type of butterfly that is an endangered species. Id. at 979. The developer, along with local, state, and county officials, created a “Habitat Conservation Plan” to preserve the butterfly. Id. at 980. The Plan restricted how much land could be developed, dedicated land to butterfly habitat, created a permanent habitat conservation and enhancement program, and promised future cooperation among the parties. Id. at 987.
Fish & Wildlife then issued a biological opinion. Id. at 980. The opinion concluded that the development, with the Plan’s restrictions in place, would not jeopardize the butterfly. Id. at 980-81. Fish & Wildlife also issued a Finding of No Significant Impact, which obviated the need to complete a full EIS. Id. at 981. The plaintiff contested the biological opinion and challenged the decision not to complete a full EIS. We rejected these challenges and approved Fish & Wildlife’s methodology. In doing so,' we rejected the plaintiffs argument that Fish & Wildlife failed to use “the best scientific and commercial data available” when it relied on the Plan in reaching its no-jeopardy determination. Id. at 984-85. We also held that Fish & Wildlife properly relied on the Plan when deciding not to draft a full EIS. Id. at 987. We noted that “courts have permitted the effect of mitigation measures to be considered in determining whether preparation of an[EIS] is necessary.” Id.
*956More recently, we approved of agency reliance on mitigation measures in Edwardsen v. Department of Interior, 268 F.3d 781 (9th Cir.2001). In Edwardsen, the plaintiffs challenged an EIS that approved a plan to drill for oil off of the Alaska coast. Id. at 783. In particular, the plaintiffs argued that the EIS failed to adequately discuss the impact of vegetation loss on caribou. Id. at 789-90. We rejected the plaintiffs’ arguments, stating that:
The EIS adequately addresses the effect of pipeline construction on the movement of caribou by noting that pipelines will be elevated to permit the passage of caribou and that the construction of permanent roads along pipelines will be minimized. These mitigation measures make reasonable the EIS’ conclusion that the cumulative effects of the pipelines on caribou would be minor.
Id. at 790.
These prior holdings are sensible. If a Conservation Agreement is in place, then the reviewing agencies ought to consider it when evaluating the impact of the proposed action. It is also relevant to know that the Agreement imposes enforceable obligations on the parties, to assure that the proposed mitigation measures will actually be implemented.5 Accordingly, it was proper for the Forest Service and Fish & Wildlife to consider the Conservation Agreement when evaluating the Stimson Project.
Selkirk’s challenge goes beyond the threshold issue of whether it was appropriate to consider the Conservation Agreement at all. Selkirk also contends that Fish & Wildlife and the Forest Service failed in their duties when they concluded that the Conservation Agreement alleviates the concerns raised in the 1994 draft biological opinion that concluded the Stimson Project jeopardized the grizzly bears. Selkirk’s argument is not frivolous. After all, both Fish & Wildlife and the Forest Service identified serious problems posed by the Stimson Project. These included the devastating effects that roads have on grizzly bear survival and the disruptions caused by harvesting timber in grizzly bear habitat. We hold that Fish & Wildlife and the Forest Service did not violate their duty to rely on the best scientific data available when they reasonably concluded that the effects of road construction and timber harvesting would be sufficiently mitigated by enforcement of the terms in the Conservation Agreement so as not to jeopardize the existence of the species. While another decisionmaker might have reached a contrary result, the agencies conducted a reasonable evaluation of the relevant information and reached a conclusion that, although disputable, was not “arbitrary and capricious.” See Marsh, 490 U.S. at 385, 109 S.Ct. 1851.
In the EIS, the Forest Service recognized the factors contributing to grizzly bear mortality, as identified by undisputed science, and considered at great length the impact of the Conservation Agreement on the primary threats to the grizzly bear. In the portion of the EIS evaluating “Private Land Actions,” the EIS concluded that the Conservation Agreement would “minimize displacement of grizzly bears” and “reduce the possibility of incidental *957take and not jeopardize the continued existence of listed species.” Of the alternatives considered in the EIS, the Forest Service noted that the alternative employing the Conservation Agreement “adds more management direction for Stimson lands.” The “[m]anagement direction from the [Conservation Agreement] for protection of habitat for grizzly bear includes: open road densities, operations and uses, road locations, cover, riparian zones, and security.”6
Later in the EIS, in the section titled “Effects of Actions Likely to Occur on Stimson Lands As a Result of Granting Access,” the Forest Service assumed that future Stimson activity would also be regulated by the Conservation Agreement. This section concluded that “[i]n addition to the cover objectives, ... benefit would be gained on Stimson lands by providing Winter Logging Areas only and restricting activities during the Spring Period.”
In conducting its own grizzly bear analysis, Fish & Wildlife also identified and considered at length the primary causes of grizzly bear mortality in its section on grizzly bear status. Fish & Wildlife considered the efficacy of the Conservation Agreement in mitigating those threats to grizzly bears. Regarding the issue of road restrictions, Fish & Wildlife determined that under the Conservation Agreement, “[o]pen road densities will be limited[,] ... [n]ewly constructed roads will be closed to public motorized access,” and that “Stimson and [the Forest Service] will limit access on several restricted roads ....” As for the winter logging restrictions, Fish & Wildlife determined that “Stimson will avoid commercial use in 4,480 acres of ‘winter logging. areas,’ ” which will “provide a large block of available spring habitat” that “is- a benefit to the grizzly bear.”
While Selkirk may advocate additional restrictions on Stimson’s activities, Fish & Wildlife provided the requisite' rational connection between the best available science identifying threats to the grizzly bear population and its decision that the Conservation Agreement sufficiently mitigated those threats: “[The Agreement] will help decrease adverse effects to the grizzly bear by providing space and isolation ... limiting activities especially during the spring season ... limiting road use, and by monitoring to ensure effectiveness.” See also Pac. Coast Fed’n of Fishermen’s Ass’ns v. Nat’l Marine Fisheries Serv., 265 F.3d 1028, 1034 (9th Cir.2001) (“Essentially, we must ask whether the agency considered the relevant factors and articulated a rational connection between the facts found and the choice made.”) (quotation marks and citation omitted).
The conclusion of Fish & Wildlife and the Forest Service that the Conservation Agreement would not wipe out all effects of the Stimson Project does not, on its own, allow us to overturn the decision to grant the easement. The agencies performed a credible task: they identified the most troublesome problems (roads and harvesting in sensitive areas), realized the magnitude of those problems, and then determined that mitigation measures contained in the Conservation Agreement would lower the threats to the grizzlies enough that the Stimson Project would not place the existence of the species in jeopardy. While Selkirk advances a contradictory conclusion, we cannot say that the agencies’ evaluations of the Conservation Agreement, and their overall decisions, were not based on “the best scientific and *958commercial data available.” See also Greenpeace Action v. Franklin, 14 F.3d 1324, 1333 (9th Cir.1992) (“To set aside the Service’s determination in this case would require us to decide that the views of Greenpeace’s experts have more merit than those of the Service’s experts, a position we are unqualified to take.”).
B
We next consider Selkirk’s argument that the Forest Service violated NEPA. NEPA requires federal agencies to prepare an EIS for all “major Federal actions significantly affecting the quality of the human environment.” 42 U.S.C. § 4332(2)(c). Every EIS must consider the cumulative impacts of the actions evaluated. 40 C.F.R. § 1508.25(a)(2). Federal regulations define “cumulative impact” as the “incremental impact of [an] action when added to other past, present, and reasonably foreseeable future actions regardless of what agency ... or person undertakes such other actions.” Id. § 1508.7.
“We review an EIS under a rule of reason to determine whether it contains a ‘reasonably thorough discussion of probable environmental consequences.’ ” Edwardsen, 268 F.3d at 784-85 (quoting Neighbors of Cuddy Mountain v. United States Forest Serv., 137 F.3d 1372, 1376 (9th Cir.1998)). “In our review, we must not substitute our judgment for that of the agency.” Id. at 785.
1
Selkirk argues that the Forest Service violated NEPA by impermissibly constraining the geographic scope of the EIS. Selkirk contends that the Forest Service acted arbitrarily and capriciously by not fully considering the cumulative impact of a project proposed by Stimson in the IPNF, which borders the LeClerc BMU.
The task of selecting the geographic boundaries of an EIS requires a complicated analysis of several factors, such as the scope of the project considered, the features of the land, and the types of species in the area. “[Determination of the extent and effect of these factors, and particularly identification of the geographic area within which they may occur, is a task assigned to the special competency of the appropriate agencies.” Kleppe v. Sierra Club, 427 U.S. 390, 414, 96 S.Ct. 2718, 49 L.Ed.2d 576 (1976). Despite this deference given to the Forest Service, the agency must at least have considered cumulative effects in creating the boundaries of its analysis. “Federal regulations do not explicitly require an EIS to include a discussion of cumulative impacts, but they do direct agencies to consider cumulative impacts in determining the scope of an EIS.” Kern v. Bureau of Land Management, 284 F.3d 1062, 1076 (9th Cir.2002) (internal quotation marks and citations omitted).
The Forest Service considered the IPNF project, determined that it would not create “additional, cumulative effects,” and therefore did not include the IPNF in the EIS’s analysis area. In fact, the cumulative effects section of the EIS starts by contemplating what to do with the IPNF project. The EIS states that
Because of topography there would be no additional, cumulative effects from the [IPNF] access proposal. The two Stimson access proposals would not result in a connected transportation system and lie within separate watersheds, viewsheds, and management areas for proposed, threatened and endangered species. The ridgeline separates the watersheds, causing hydrological effects to be separate; it also separates the viewsheds, and serves as a boundary line *959for analysis of wildlife effects. The transportation system of roads would not be connected by the two proposals.
Furthermore, when the Forest Service was determining the scope of the EIS, a wildlife biologist expressed concern that including the IPNF project in the EIS would be an “arbitrary increase! ],” and that “the magnitude of the effects would actually appear to be less as they would be spread over a larger area.” These evaluations of the impact of the IPNF projects adequately discharge the Forest Service’s obligations under NEPA.
Our analysis is supported by our recent decision in Neighbors of Cuddy Mountain v. Alexander, 308 F.3d 1059 (9th Cir.2002). In Neighbors, the plaintiffs challenged the EIS for a timber sale on the ground that the EIS only analyzed cumulative effects on a portion of the forest. The opinion reasoned that “under NEPA we defer to an agency’s determination of the scope of its cumulative effects review. Given our standard of review, we conclude that the Forest Service took the requisite ‘hard look’ at the environmental effects of the sale before approving it. That is all NEPA demands.” Id. (citations omitted). Like in Neighbors, the Forest Service in this case took a “hard look” at the activity in the IPNF and made a considered judgment that the EIS would be a more accurate document if it did not consider the IPNF activity in the EIS’s cumulative impacts analysis.
Selkirk relies on two other cases recently decided by this Court. The first case is Native Ecosystems Council v. Dombeck, 304 F.3d 886 (9th Cir.2002), which also dealt with the choice of a Bear Management Unit as the sole area of analysis. Native Ecosystems held that the Forest Service violated ESA when it failed to consider the impact of a nearby sheep grazing operation when conducting its ESA cumulative effects analysis for a proposed timber sale in a national forest bordering Yellowstone National Park. The sheep grazing unit neighboring the proposed timber sale area was a known “population sink,” meaning that the Forest Service was aware that the grazing unit was an area in which “[t]he majority of known and probable deaths of [area] grizzlies are clustered.” Id. at 902. This “population sink” was a little more than a mile from the proposed timber sale, but was not considered by the Forest Service in its cumulative effects analysis because it was outside the Bear Management Subunit. Id. at 901-02. We stated that “the Management Subunit may well be a proper proxy for the project’s action area, but one cannot tell from the administrative record. An agency must provide support for its choice of analysis area and must show that it considered the relevant factors, and the Forest Service failed to do so here.” Id. at 902 (citation omitted).
In Idaho Sporting Congress, Inc. v. Rittenhouse, 305 F.3d 957 (9th Cir.2002), the Forest Service chose the “home range” of various species to defíne the geographic scope of the EIS, despite a 1996 Forest Service report that mandated that “the habitat needs of these species must be evaluated at a landscape scale.” Id. at 973 (emphasis in original). We held that the Forest Service violated NEPA by confining the EIS to the smaller “home range” and not considering the entire “landscape scale” “without justifying its decision.” Id. at 974.
Unlike Native Ecosystems and Idaho Sporting Congress, the Forest Service in this case did “provide support” for and “justify” its decision to exclude the IPNF area from the EIS analysis. The agency concluded that the IPNF activity implicated a different watershed with different topography, and that including the *960IPNF area in the analysis might have skewed the results. The EIS acknowledged the existence of the IPNF project and stated why the project should not be included in the EIS. “NEPA exists to ensure a process, not to ensure any [particular] result.” Inland Empire Public Lands Council v. United States Forest Serv., 88 F.3d 754, 758 (9th Cir.1996) (emphasis in original). The Forest Service followed an appropriate process.
Selkirk argues that Native Ecosystems is similar to this case, despite the justification of the EIS’s geographic scope in this case that was missing in Native Eco systems. Selkirk contends that Native Ecosystems establishes that the Forest Service must account for how the Stimson Project affects the “wandering bear.” Stimson maintains that this accounting must go beyond the LeClerc BMU because, as the Forest Service acknowledges, bears move between BMUs. Although Native Ecosystems appears to be factually similar to this case, upon closer examination there are some significant differences. In Native Ecosystems, the Forest Service did not analyze the impact of a sheep grazing unit; this unit was acknowledged as a center for bear deaths. In this case, Selkirk objects to the Forest Service’s decision not to evaluate the effects of Stimson access requests in the IPNF. But there is no indication that the Stimson access request in the IPNF is a “population sink.” The activities “ignored” in this case and Native Ecosystems are substantially different. Even though the Forest Service recognizes that bears may wander out of the LeClerc BMU, wandering into the IPNF is not wandering into the kind of place where most area bears die, which was the situation in Native Ecosystems.
It is also significant that the IPNF project into which bears may wander had not yet been approved when the EIS was drafted, while the sheep grazing “population sink” already existed in Native Ecosystems. The Forest Service need not always consider “all proposed actions in an appropriate region before approving any of the projects.” Kleppe, 427 U.S. at 414 n. 26, 96 S.Ct. 2718.
Even with the acknowledgment that bears wander out of BMUs, the selection of the LeClerc BMU was not arbitrary and capricious. It would be absurd to think that grizzly bears in the wild confine themselves to a discrete area as if they were in a zoo. The Interagency Grizzly Bear Committee established the BMUs as the proper areas for grizzly bear analysis because the BMUs are the “approximate size of annual home ranges of an adult female grizzly bear.” This committee of experts was aware of the wandering nature of bears when they selected the BMUs for analysis areas, as shown by the committee’s statement that “[the analysis areas] are not intended to be the actual home ranges of known adult female grizzly bears.”
It was not unreasonable for the Forest Service to limit its analysis to the BMU in which the Stimson Project would take place. The Forest Service is allowed to consider “practical considerations of feasibility” in its selection of a geographic scope for an EIS. Kleppe, 427 U.S. at 412, 96 S.Ct. 2718. Three articulable reasons supported the agency decision: (1) the selection of one BMU made sense based on the geographic features contained therein; (2) expanding the analysis area would dilute the effects of the proposed project; and (3) the bear’s likely wandering area was not a known “population sink” like that in Native Ecosystems. We cannot say the decision to limit the scope of analysis to the LeClerc BMU was unreasonable.
*9612
Selkirk next argues that the Forest Service violated NEPA by failing to consider the impact of several future Stimson projects in the LeClerc BMU. NEPA obligates the Forest Service to consider “past, present, and reasonably foreseeable future actions” in its cumulative effects analysis. 40 C.F.R. § 1508.7. NEPA requires that an EIS contain a “meaningful analysis of the cumulative impacts” for future projects. City of Carmel-By-The-Sea v. Dep’t of Transp., 123 F.3d 1142, 1161 (9th Cir.1997).
Selkirk maintains that the EIS did not sufficiently describe, locate, or analyze seventeen road-building and logging projects identified by Stimson in state forest practices applications. We disagree.
The EIS contains a section on “Effects of Actions Likely to Occur on Stimson Lands As a Result of Granting Access.” This section states that “[information on the activities planned on Stimson lands [includes] ... inventory forest type size class maps, color coded maps óf proposed 5-year harvest plans, [and] current State approved Forest Practices Applications in the [LeClerc BMU].... These documents are included in the analysis files for this EIS.” With specific regard to the grizzly bear habitat, the EIS considered “other harvest related activity within the” Le-Clerc BMU when analyzing the cumulative effects of the Stimson Project.
The Forest Service included in the EIS’s administrative record most of the forest practices applications. The Forest Service requested and received from Stimson on April 26, 1999, “[c]opies of all currently active State approved Forest Practices Applications.” The Forest Service received additional applications on September 21, 1999, October 13, 1999, November 4, 1999, and March 8, 2000. In short, the Forest Service sought and received updated Washington State forest practices applications during the entire time it was drafting the EIS.
Selkirk complains that the Forest Service did not list the location of all the harvest projects that Stimson had planned. The EIS detailed the locations of areas Stimson planned to harvest when the EIS was drafted, however, and this detail is part of a careful consideration of the effects of these harvests. The EIS contemplated that future Stimson harvests throughout the LeClerc BMU would result in the loss of more than 4,000 acres of core habitat. This analysis referred to a map of planned Stimson projects, which provided locations beyond those directly related to the easements considered by the EIS.
The Conservation Agreement plays a significant role here, as well. To the extent that consideration of specific forest prac- ■ tices applications would duplicate the thorough consideration of the operational plan set forth in the Conservation Agreement, the Forest Service made a reasoned decision to exclude detailed descriptions of those applications. The EIS presumed that Stimson would manage all of its lands for perpetual logging, and the Agreement regulates Stimson’s behavior on all of these lands. As noted earlier, we place great emphasis on Stimson’s obligations to honor its word given in the Agreement. As long as Stimson’s behavior accords with the Agreement, the Forest Service’s thorough consideration of the effects of the Agreement reduces the need to list, map, and discuss every pending Stimson harvest plan. Given that the EIS listed and considered most pending forest practices applications and evaluated the effects of the Agreement, we conclude that the EIS provided a “ ‘reasonably thorough discussion of probable environmental consequences.’ ” Edwardsen, 268 F.3d at 784-85 (quoting *962Neighbors of Cuddy Mountain, 137 F.3d at 1376).
3
Selkirk’s final NEPA-based challenge is that the EIS used an inadequate temporal scope. NEPA requires that an EIS engage in reasonable forecasting. Because “ ‘speculation is ... implicit in NEPA, [ ] we must reject any attempt by agencies to shirk their responsibilities under NEPA by labeling any and all discussion of future environmental effects as “crystal ball inquiry.” ’ ” Kern, 284 F.3d at 1072 (quoting Scientists’ Inst. for Public Information, Inc. v. Atomic Energy Comm’n, 481 F.2d 1079, 1092 (D.C.Cir.1973)). Selkirk argues that the Forest Service violated NEPA by only considering the effects of the Stimson Project through 2003, a three-year period.
The selection of the scope of an EIS is a delicate choice and one that should be entrusted to the expertise of the deciding agency. Cf. Kleppe, 427 U.S. at 414, 96 S.Ct. 2718. NEPA does not impose a requirement that the Forest Service analyze impacts for any particular length of time. In reviewing the EIS, we ask only whether the Forest Service “considered the relevant factors and articulated a rational connection between the facts found and the choice made.” Washington Crab Producers, 924 F.2d at 1441 (quotation marks and citation omitted).
Stimson, in its brief and at oral argument, advanced the only argument in defense of the three-year analysis period. Stimson argued that the Forest Service’s temporal decision was based on the constantly changing nature of the multitude of federal and state regulations applicable to Stimson’s activities. The Washington State forest management rules were instituted on an “emergency” basis while the EIS was being drafted. Permanent rules will replace them. The federal forest practices rules are also in a state of flux. The specific management guidelines for individual animal species are also evolving. For example, temporary rules were imposed in January 1999 for grizzly bear recovery area management, with a proviso that those rules will change when the “Selkirk/Cabineb-Yaak Grizzly Bear Subcommittee determines a need to modify this direction.” Stimson therefore argues that the Forest Service acted reasonably when it limited its analysis to the time period for which it was certain of the rules that would govern Stimson’s behavior.
The problem with Stimson’s argument (and it was only Stimson’s; the government never articulated a defense of the three-year period) is that the Forest Service appears to have had two other choices for temporal analysis. The Forest Service could have analyzed the effects of the Stimson Project for a ten-year period. Indeed, this is the time period initially selected by the Forest Service’s wildlife biologist, who used historical harvest averages to project timber activity for ten years. The Forest Service also could have selected a five-year window. The Conservation Agreement, which governs much of Stimson’s behavior, runs for five years. Even if the Forest Service did not know exactly what Stimson was going to cut in years four and five, and even if the Forest Service did not know what forest practices rules would apply in years four and five, the Conservation Agreement gives a pretty good idea of what Stimson would be doing in those years and how it would do it.
These problems with the three-year window make this an extremely close issue. But, in the end, we cannot say that the Forest Service acted arbitrarily and capriciously in selecting the three-year period. The regulations that are subject to change *963are critical to evaluating Stimson’s harvesting, and their change would affect analysis of that harvesting. The Washington State forest management rules govern the smallest details of harvesting. The EIS contains a list of “Assumptions Used for Activities on Stimson’s Lands Based on [Washington] DNR regulations.” These assumptions include such fine detail as:
No harvest is permitted within the inner zone unless the basal area of conifer and hardwoods for trees greater than 6 inches dbh is: Greater than 130 or less than 90 square feet per acre on medium site indexes
Harvest must leave at least 50 trees per acre AND a basal area of 90 square feet per acre on medium site indexes as follows: The 21 largest trees must be left. The remaining 29 trees must be greater than or equal to 10 inches dbh.
At least 20 tons per acre of down wood present on the site before harvest must be left as follows: A minimum of 8 pieces greater than 16 inches diameter and 20 feet in length ....
These regulations, and others equally detailed and important, were all imposed on a temporary basis and were set to expire after three years. The restrictions imposed on Stimson by state regulations are critical to understanding how Stimson’s harvest will affect the environment. For example, the regulations could change from requiring fifty trees per acre to remain to requiring only five trees; such a change would affect the Forest Service’s evaluation of the project.
Not only was the Forest Service’s certainty of these vital regulations limited to three years, the EIS “was based upon the information ... provided by Stimson, which only provided specific information through the year 2003.” Although we do not mean to suggest that a private actor should be allowed to limit the reach of an EIS by providing only limited information about its intentions, it was significant that the Forest Service received concrete Stimson plans only for a three-year period.
A ten-year study may have been preferable in this case. Or even a five-year study. But the three-year study chosen by the Forest Service was not unreasonable. Although the Forest Service had some information for ten years, and some more information for five years, it had the most information for the next three years. In this situation, the three-year period of analysis did not prevent the EIS from setting forth a “ ‘reasonably thorough discussion of the significant aspects of probable environmental consequences.’ ” Kern, 284 F.3d at 1071 (quoting Oregon Natural Resources Council v. Lowe, 109 F.3d 521, 526 (9th Cir.1997)). Moreover, any new information is required undér the Conservation Agreement to be analyzed and the Agreement revised, if necessary. The enforcement of this provision is particularly vital in light of the less-than-ideal three-year period of analysis in the EIS.
C
Finally, Selkirk argues that Fish & Wildlife violated ESA by not considering future Stimson activities in the 2001 biological opinion. ESA requires that the Forest Service “shall, in consultation with [Fish & Wildlife], insure that any action authorized, funded, or carried out by [the Forest Service] ... is not likely to jeopardize the continued existence of any endangered species or threatened species.” 16 U.S.C. § 1536(a)(2). Fish & Wildlife must review the information prepared by the Forest Service and author a biological opinion indicating whether the Forest Service action jeopardizes any species. 16 U.S.C. § 1536(b)(3)(A). Fish & Wildlife must “[e]valuate the effects of the action *964and cumulative effects on the listed species” when rendering the biological opinion. 50 C.F.R. § 402.14(g)(3). Selkirk contends that Fish & Wildlife did not adequately consider the cumulative effects of the Stimson Project because Fish & Wildlife’s biological opinion did not analyze the forest practices applications filed by Stimson. In addition, Selkirk argues that the failure to adequately assess forest practices applications for Stimson projects involving a total of 14.55 miles of new road and 4,462 acres of logging in the LeClerc BMU violates the requirement to use the best available commercial data. See 16 U.S.C. § 1536(a)(2).
As in the other challenges raised by Selkirk, we evaluate Fish & Wildlife’s biological opinion to see if it was arbitrary and capricious. We defer to Fish & Wildlife’s no-jeopardy determination, especially where, “resolution of[the] dispute involves primarily issues of fact.” Marsh, 490 U.S. at 377, 109 S.Ct. 1851.
Fish & Wildlife, may employ any method that adequately considers cumulative impacts. ESA does not impose a requirement that Fish & Wildlife list, detail, and discuss each and every forest practices application, so long as Fish & Wildlife employs a device, that considers the cumulative impacts of future Stimson activities. Here, Fish & Wildlife used the Conservation Agreement as such a device.
Unlike the forest practices applications, which do not cover the bulk of harvests likely to occur on Stimson land, the Conservation Agreement governs Stimson’s behavior on all of its land in the LeClerc BMU. Consideration of the Conservation Agreement-mandated behaviors was therefore an evaluation of the best information as to Stimson’s future actions. Fish & Wildlife, however, was still required to evaluate the effects of the actions and cumulative effects on the listed species.
Selkirk argues that Fish & Wildlife failed to provide any useful analysis of the Stimson Project’s impacts on grizzly bears. In making this assertion, Selkirk overlooks the extensive analysis devoted to the effects of the Conservation Agreement. Fish & Wildlife devoted nearly all of the biological opinion to addressing the impacts of the mitigation measures contained in the Conservation Agreement. The first page of the biological opinion states that the Conservation Agreement “is incorporated as part of the proposed action. It addresses measures necessary to minimize adverse impacts to bull trout, woodland caribou, and grizzly bear.”
The biological opinion goes on to discuss the Conservation Agreement’s requirements related to bull trout, caribou, and, relevant to this case, grizzly bear (“open road density ... will not exceed one mi/ mi2 during the non-denning period,” “no net increase in roads open to public motorized use,” “provide more secure low elevation habitat,” “leave visual screening between roads,” “a minimum of 40% of all land in the BMU is maintained in cover,” “no point in the unit is more than 600 feet from cover,” “new information gained from monitoring and research, conducted either within or outside the BMU, will be reviewed on an annual or more frequent basis, as necessary, to determine if changes in management direction are appropriate”).
Fish & Wildlife’s assessment of the Conservation Agreement drove the conclusions of the biological opinion. The biological opinion concluded that the “access project is not likely to jeopardize the continued existence of the grizzly bear” because of four requirements imposed by the Conservation Agreement: (1) no net increase in open road density, (2) winter logging areas and security areas “will result in more secure low elevation spring habitat than is *965presently available,” (3) adequate cover will be maintained, and (4) extensive monitoring will maintain the effectiveness of the Conservation Agreement. In choosing to analyze the impacts of Stimson’s planned actions through the Conservation Agreement, Fish & Wildlife adequately considered the environmental impact of future Stimson timber harvesting.
Further, we cannot say that Fish & Wildlife did not use the best commercial data available. Stimson must conduct all its activities on its lands within the Le-Clerc BMU in accordance with the terms of the Conservation Agreement; the Conservation Agreement assumed that Stimson would use its lands for perpetual logging. Thus, even if Fish & Wildlife did not consider several specific forest practices applications, the Conservation Agreement governs Stimson’s relevant commercial activities, and it was not arbitrary and capricious for Fish & Wildlife to use the Agreement as the best commercial data available. Finally, we again note that the Conservation Agreement provides for revision based upon the accumulation of better data, both scientific and commercial, which will provide continuing assessment of the environmental context of Stimson’s actions.
Ill
Our environmental laws must balance the oft-competing statutory policies of environmental protection and private property rights. NEPA and ESA attempt to strike that balance by requiring federal agencies to follow rigorous procedures when taking action that will impact the environment. Here, the Forest Service and Fish & Wildlife adequately followed those procedures. For the reasons we have discussed, it was not unreasonable for the agencies to rely in part on the Conservation Agreement in conducting their biological and environmental assessments. Such an agreement, the product of a constructive public-private cooperation for the protection of the environment, can be an important source of relevant data and a guarantee of future behavior designed to mitigate adverse environmental impacts. In this case, we are persuaded that the Conservation Agreement qualifies as a reasonable source of data, and will be enforced to deliver on its promises of mitigation.
The district court’s summary judgment against Selkirk is AFFIRMED.
. By order entered on October 9, 2002, we affirmed the district court and vacated our temporary stay that had prevented Stimson from beginning roadwork to access its landlocked property within the National Forest. This opinion explains the rationale for that decision.
. The project was officially called the "ANIL-CA Access Easement,” so named after the federal statute that mandates the Forest Service to provide private landowners with access to their property, 16 U.S.C. § 3210(a), but for purposes of clarity will be referred to here as the "Stimson Project.”
. The parties dispute how many specific harvesting plans were not detailed in the EIS, but the EIS appears to have left out at least six specific harvesting proposals.
. The biological opinion incorporated the provisions of the Conservation Agreement into the terms and conditions of the Incidental Take Section, thus making Stimson’s compli-anee with the Agreement mandatory if Stimson wishes to avoid liability for the unauthorized taking of endangered and threatened species.
. No party to this appeal has suggested that the Conservation Agreement is not enforceable; indeed, they have all implicitly assumed Stimson is contractually and legally bound to implement the agreed mitigation measures and that the government agencies intend to enforce Stimson’s compliance. We fully expect that Stimson and the government agencies will fulfill their obligations; and Selkirk is undoubtedly committed to assuring that they do.
. In addition to evaluating the effect of the Conservation Agreement on grizzly bears, the EIS considered the Conservation Agreement’s impact on woodland caribou, bull trout, stream protection, and future monitoring.