dissenting:
I respectfully dissent from section II.A of the panel’s opinion. I believe that the EIS for the Winema LRMP was inadequate under NEPA.
The Forest Service itself has emphasized that considering quantity of old growth alone is not sufficient for evaluating species viability and that it is important to consider the size, configuration, and connectivity of old growth in the forest planning process. In a memorandum to Forest Supervisors in Region 6 (the region in which the Winema National Forest is located) which was “intended to help [the Supervisors] focus the diversity discussion in [their] Final Forest Plan and Environmental Impact Statement,” the Regional Forester instructed that “[f]or plant associations and serai stages that are of *531concern, e.g. old growth, hardwood associations, etc., describe the following attributes: [distribution across the forest; [s]pacing within and across watersheds; [p]ateh sizes; [p]atch shapes; [connections between patches.” Similarly, in the Regional Guide (from which the MMRs in the Winema Forest Plan are drawn), the Forest Service stated that:
Once the features of old growth have been examined and described, the descriptions should be used to form the basis of classification of old-growth areas. Classification is needed to give a rational basis for inventory, research, and management. The inventory will provide information on the amount, size, and distribution of remaining old growth, which is essential data for developing research plans and management alternatives.
Forest Service biologists explain that the configuration and connectivity of old growth have significant environmental effects on species viability:
The ecological condition of the stands contributing to total acreage must also be evaluated. Much of the [late stage/old growth] forest is extremely fragmented. Many of the Forest Plan reserves will perpetuate this fragmentation by specifying reserves in fairly small to very small blocks or in long stringers. Although these Forest Plan reserves may meet objectives relative to recreation, visual quality, and stream and soil stability, they do not necessarily provide for effective protection of old-growth dependent species and processes____ The functioning of a [late-stage/old growth] network and a successful conservation strategy for the northern spotted owl and other [late-stage/old growth] associated species are highly dependent on habitat block size, proximity of such blocks to one another, and degree of connectivity between blocks.
Alternatives for Management of Late-Suceessional Forests of the Pacific Northwest, A Report to the U.S. House of Representatives by The Scientific Panel on Late Successional Forest Ecosystems (including Jack Ward Thomas, Chief Research Wildlife Biologist for the USDA Forest Service, Pacific Northwest).
This court has held that a Forest Plan that included no consideration of the configuration of the old growth that was to be protected was unreasonable. Marble Mountain Audubon Soc’y v. Rice, 914 F.2d 179, 182 (9th Cir.1990). In Marble Mountain, the Forest Service completed an EIS assessing a plan to salvage and harvest timber in portions of a drainage area affected by fire. Id. The plaintiffs asserted that the EIS was inadequate because it did not consider the unique value of a particular biological corridor in the area. Id. The court agreed, stating that:
the Forest Service did not take a “hard look” at the impact of the selected salvage and harvest alternative on the Grider drainage biological corridor. Although the FEIS acknowledges that the Grider drainage is a biological corridor, it does not contain a significant discussion of the corridor issue.
Id. at 182.
Here, the inadequacy of the EIS is even more striking than it was in Marble Mountain. First, the Forest Service failed to consider the environmental significance of at least two known biological corridors. One of these is the Yoss Ridge corridor, which already has been the subject of a proposed timber sale. There, the Forest Service has approved timber harvesting which may “cause the Yoss Ridge corridor to be fragmented in several areas.” It did not, however, complete an EIS to evaluate the proposed timber sale. Another corridor that is not protected by MA7 designation and is not discussed in the EIS is between Calimus Butte and Big Wocus Bay.
Moreover, the Forest Service failed to consider in the EIS any of the other issues related to the configuration of actual old growth stands to be protected because it did not have the information to do so. The Forest Service itself states in the FEIS that “[n]o evaluation is made of the distribution of habitat across the Forest or the size of the habitat units.” FEIS at 2-102. Thus, the Forest Service did not consider the issues that were emphasized in its own directives such as “[distribution across the forest; *532[sjpacing within and across watersheds;' [p]atch sizes; [p]atch shapes; [connections between patches.”
Also troubling is the fact that the inventory revealed that many areas in MA7 contained no old growth at all, while other old growth stands were located in management areas devoted to objectives such as timber harvesting. As a result, the plan may protect less old growth than is needed to meet MMRs while leaving other old growth stands available for logging. This court has held that NEPA “imposes a duty on federal agencies to gather information and do independent research when missing information is important, significant, or essential to a reasoned choice among alternatives.” Oregon Envtl. Council v. Kunzman, 817 F.2d 484, 495 (9th Cir.1987) (citations omitted). Here, the Forest Service did not wait for the old growth inventory results that soon would be available in order to assess their implications for the Forest Plan. As a result, the Forest Service was unable to address in the FEIS issues related to the actual location of old growth stands to be protected. In this respect, the FEIS for the Winema LRMP was inadequate.
The defendants’ arguments to the contrary fail. First, the defendants argue that the specific location of old growth.does not need to be included in a broad, programmatic document such as a Forest Plan because no logging will be approved until site-specific assessments are performed. This argument is inconsistent with the Forest Service’s own directives saying that data from old growth inventory is essential for planning. It is also inconsistent with Idaho Conservation League v. Mumma, in which this court held that, although the Forest Service would have to take further action under NEPA before authorizing site-specific development, “the initial plan and wilderness recommendation represent important decisions.” 956 F.2d 1508, 1516 (9th Cir.1992). “Indeed,” the court went on to say, “short of assuming that Congress imposed useless procedural safeguards, and that wilderness designation is a superfluous step, we must assume that the management plan plays some, if not a critical, part in subsequent decisions.” Id.
Defendants assert that under Douglas County v. Babbitt, 48 F.3d 1495 (9th Cir. 1995), cert. denied, — U.S. -, 116 S.Ct. 698, 133 L.Ed.2d 655 (1996), the administrative delineation of habitat for wildlife does not cause significant environmental effects that require an EIS. Douglas is distinguishable because it concerns the delineation of critical habitat under the Endangered Species Act and therefore, ESA procedural requirements rather than NEPA requirements apply. Id. at 1503. In contrast to NEPA, under the ESA the Secretary of Interior must designate any area without which the species at issue would become extinct. Id. Therefore, the Secretary “has no discretion to consider the environmental impact of his or her actions.” Id. In this case, in contrast, the Forest Service is under no such obligation and has considerably more discretion with respect to the designation of habitat. As a result, the holding of Douglas does not apply.
Moreover, site-specific assessment is not likely to address the broad issues associated with the forest-wide configuration of old growth. Rather, it is the purpose of the Forest Plan EIS to assess the overall.network to determine whether it will ensure species viability. Finally, it is unlikely' that the Forest Service will complete EIS’s at the site-specific level when approving timber sales. Several timber sales in areas containing inventoried old growth have already been approved and EAs rather than EIS’s have been issued. See Finding of No Significant Impact and EA for Yoss Ridge (1991) (approving a timber sale in the Yoss Ridge area without an EIS).
The defendants also devote a great deal of discussion to the adequacy of the 1981 Timber Inventory, on which the Forest Service relied when it selected the original MA7 sites to meet the MMRs. This argument is irrelevant because the Forest Service has conceded that it did not know where particular old growth stands were located when it developed the Forest Plan. It was for this reason that the Forest Service conducted an old growth inventory in order to complete Amendment 3. Even assuming, however, that the 1981 Timber Inventory was ade*533quate for selecting MA7 locations, the Forest Service has not explained why it did not address issues related to old growth network configuration in the EIS. Because the Forest Service did not address these issues, it is impossible for this court to be sure that the agency took the required “hard look” at the environmental impact of the Plan. See Steamboaters v. F.E.R.C., 759 F.2d 1382, 1393 (9th Cir.1985).
I conclude that the Forest Service violated NEPA by relying on the 1981 Timber Inventory and by ignoring issues such as stand size, configuration, and connectivity in the EIS for the Winema LRMP. I would reverse the district court’s decision with respect to NEPA compliance and remand with instructions to issue an injunction prohibiting timber sales of the old growth that was identified in the 1990 inventory until the Forest Service addresses in an EIS the environmental impact of selecting particular stands for preservation, including edge effects, fragmentation, and the possible implications of biological corridors. Such a holding would not oblige the Forest Service to complete new inventories; it merely would require it to consider the important issues associated with the actual location of old growth reserves, taking into account inventory data that is already available.