O’SCANNLAIN, Circuit Judge:
For the third time within two years we deal with environmental law issues surrounding the construction of an international astrophysical observatory on Mount Graham in southeast Arizona.
I
The background of this dispute is explored at length in this court’s opinion in Mt. Graham Red Squirrel v. Madigan, 954 F.2d 1441, 1443-48 (9th Cir.1992) {Red Squirrel II). An abbreviated account is offered here.
The Mount Graham International Observatory project was first proposed in 1984 by the University of Arizona (“University”) on behalf of an international consortium. The original proposal contemplated construction of the world’s most sophisticated array of telescopes atop Mount Graham, a peak in the Pinaleno range that rises out of the Arizona desert. The proposal immediately engendered opposition from those concerned that the project would greatly magnify threats to the continued survival of the Mount Graham red squirrel. The red squirrel is an endangered species that lives on Mount Graham and exists nowhere else in the world. Once believed to be extinct, a few hundred red squirrels now appear to be extant in the area.
[1570]*1570Mount Graham is part of the Coronado National Forest, administered by the United States Forest Service (“Forest Service”). The University was therefore required to take its proposal in the first instance to the Forest Service. This in time led to the initiation of “formal consultation” between the Forest Service and the Fish and Wildlife Service, pursuant to section 7 of the Endangered Species Act, 16 U.S.C. § 1536. Such consultation is required whenever actions authorized or carried out by a federal agency might pose a threat to the continued existence of an endangered species or have an adverse impact on the species’ habitat. See 16 U.S.C. § 1536(a)(2); 50 C.F.R. § 402.14. Formal consultation culminates in the issuance of a “Biological Opinion,” 50 C.F.R. § 402.02, wherein the Fish and Wildlife Service presents its conclusions regarding the likely effects of the action under consideration. In this case, the Biological Opinion concluded that the plan developed by the University would significantly increase the risks faced by the red squirrel. At the same time, however, the opinion identified three “reasonable and prudent alternatives” to that plan, see id., which, in the judgment of the Fish and Wildlife Service, would avoid the adverse consequences for the red squirrel that were likely to follow from implementation of the University’s original proposal.
Under normal circumstances, the Forest Service would have been required to choose a preferred alternative from among the three options presented. See 50 C.F.R. § 402.15. Four years, however, had already passed, and the University, apparently responding to indications that the other members of the consortium might opt to build the observatory complex in another country rather than endure further delay, had decided to ask Congress to intervene. The result was the enactment, in 1988, of Title VI of the Arizona-Idaho Conservation Act (“AICA”). This legislation cleared the way for the immediate construction of three telescopes on Mount Graham, and provided that an additional four might be built in the future if certain conditions were met. The Act declared that the requirements of section 7 of the Endangered Species Act were to be deemed satisfied with respect to the first phase of the project, thus ending the process of formal consultation, and instructed the Secretary of Agriculture “immediately” to issue, through the Forest Service, a “special use permit” allowing construction to go forward. AICA, sec. 602(a).
The permit was issued in April 1989. Three months later, a coalition of environmental groups led by the Sierra Club brought this suit seeking an injunction to halt construction. The complaint named the Forest Service and the Fish and Wildlife Service as defendants (together with the University,1 the “appellees”), and presented nine claims for relief. The district court granted partial summary judgment for the appellees on seven of those claims and certified an appeal to this court pursuant to Rule 54(b) of the Federal Rules of Civil Procedure. The district court thereafter denied requests for injunctive relief grounded in the two remaining claims, apparently believing it was deprived of jurisdiction to entertain such requests during the pendency of the appeal. Appeal was taken from this denial of relief as well.
After hearing oral arguments, this court issued an Order of Limited Remand instructing the district court to conduct an evidentiary hearing without delay to determine whether injunctive relief was warranted on either of the remaining claims. See Mt. Graham Red Squirrel v. Yeutter, 930 F.2d 703 (9th Cir.1991) (“Red Squirrel /”). Subsequently, we released an opinion that, having made an exhaustive analysis of the statutory language and legislative history of Title VI of the AICA, affirmed the district court’s grant of partial summary judgment. See Red Squirrel II, 954 F.2d at 1450-61.
The present appeal concerns only the two remanded claims. In keeping with the Order of Limited Remand, see Red Squirrel I, [1571]*1571930 F.2d at 705, Sierra Club filed an application for a temporary restraining order halting construction. After holding a limited evidentiary hearing, the district court denied the application. A full evidentiary hearing was then scheduled in connection with plaintiffs request for a preliminary or permanent injunction. Ultimately, however, the parties moved jointly to vacate that hearing, stipulating that they had nothing further to present, and that the district court could decide the remaining claims on the merits based on the record before it. The district court entered judgment for defendants, issuing findings and conclusions on November 20, 1991. This timely appeal followed.
II
The AICA makes no specific provision for judicial review of agency decisions taken in pursuance of its requirements. Thus, the scope of review is governed by the Administrative Procedure Act, 5 U.S.C. §§ 701-706 ("APA"). See Tribal Village of Akutan v. Hodel, 869 F.2d 1185, 1193 (9th Cir.), cert. denied, 493 U.S. 873, 110 S.Ct. 204, 107 L.Ed.2d 157 (1989) (looking to APA for standard of review applicable to agency action under Endangered Species Act). Under the APA, an agency’s decisions may be set aside only if "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law." 5 U.S.C. § 706(2)(A). The district court was thus called upon to determine only whether the Forest Service’s decisions challenged by Sierra Club were "based on a consideration of the relevant factors" and did not betray "clear errors in judgment." Marsh v. Oregon Natural Resources Council, 490 U.S. 360, 378, 109 S.Ct. 1851, 1861, 104 L.Ed.2d 377 (1989). Review under this standard is to be "searching and careful," but remains "narrow," and a court is not to substitute its judgment for that of the agency. Id. This is especially appropriate where, as here, the challenged decision implicates substantial agency expertise. See United States v. Alpine Land and Reservoir Co., 887 F.2d 207, 213 (9th Cir.1989) ("Deference to an agency’s technical expertise and experience is particularly warranted with respect to questions involving . . . scientific matters."), cert. denied, 498 U.S. 817, 111 S.Ct. 60, 112 L.Ed.2d 35 (1990).
The district court’s conclusions of law rest on its interpretation of the AICA and are therefore reviewed de novo. See, e.g., United States v. City of Rancho Palos Verdes, 841 F.2d 329, 330 (9th Cir.1988) (reviewing interpretation of the Endangered Species Act de novo). The court’s findings of fact are reviewed for clear error. Securities & Exchange Comm’n v. American Principals Holding, Inc., 962 F.2d 1402, 1405 (9th Cir.), cert. denied, — U.S. —, 113 S.Ct. 210, 121 L.Ed.2d 150 (1992).
Sierra Club suggests that we must conduct a de novo review of the factual questions presented here because the district court simply entered the findings proposed by the appellees without engaging in any independent analysis and without weighing the evidence. We dismiss any such suggestion. The record shows that the court invited both parties to propose findings and conclusions, and that it did not simply adopt those proffered by the appellees. While the court unsurprisingly accepted much of what was suggested by the prevailing parties, it rejected other of their proposals, and incorporated certain of Sierra Club’s proposed findings as well. Moreover, the final findings closely tracked the oral rulings earlier made from the bench in denying Sierra Club’s motion for a temporary restraining order after a full day of hearings on the matter.
Even if we perceived any basis in fact for Sierra Club’s attack upon the trial judge, our standard of review would be unchanged. It is settled law that, “even where the trial judge adopts proposed findings verbatim, the findings are those of the court and may be reversed only if clearly erroneous.” Anderson v. Bessemer City, 470 U.S. 564, 571-73, 105 S.Ct. 1504, 1510-11, 84 L.Ed.2d 518 (1985). See Barnett v. Sea Land Service, Inc., 875 F.2d 741, 745 (9th Cir.1989).
[1572]*1572III
Sierra Club’s principal contention on appeal is that the program implemented by the University and the Forest Service to monitor the effects of construction on the red squirrel fails to satisfy the requirements of the AICA. In support of this contention, Sierra Club presents a highly particularized account of the circumstances surrounding the monitoring program in an effort to show that the appellees have failed to live up to their obligations under the statute. Four matters are singled out for attention: (1) the location of the lower control area adjacent to the lower study area; (2) the suspension of efforts to trap and tag squirrels; (3) the failure to carry out baseline monitoring in the lower control area; and (4) the use of allegedly unsatisfactory procedures in subjecting the monitoring program to peer review.
In order to decide whether the appellees have indeed failed to meet their legal obligations, we must first determine what those obligations are. The law in this instance flows from four sources: the AICA, Reasonable and Prudent Alternative Three of the Biological Opinion (“Alternative Three”), the Special Use Permit, and the Management Plan. An understanding of the nature of these documents, and the interrelationships among them, is required before the legal significance of Sierra Club’s particular factual contentions may be assessed.
The AICA clearly envisions that a program will be put in place to monitor the impact on the red squirrel of the construction effort on Mount Graham. This is evident from section 603(b), which mandates consultation under section 7(a)(2) of the Endangered Species Act regarding the second phase of the observatory project, and requires that “all biological data obtained from monitoring the impact of construction of the first three telescopes upon the Mount Graham red squirrel” be among the things considered in that process. Moreover, the statute, without itself going into detail, offers a clear indication of what sort of monitoring program was contemplated. Section 606 of the AICA provides that “all costs directly associated with ... the biological monitoring program for the Mount Graham red squirrel contained in the terms and conditions of Reasonable and Prudent Alternative Three of the Biological Opinion ... shall be funded by the University of Arizona.” This language clearly conveys the intention of the drafters of the AICA that the monitoring program to be implemented under the statute would be the monitoring program described in Alternative Three.
Turning to this document, however, we find only a little more detail than appears in the AICA regarding the contours of the required monitoring program:
A 10-year study of the red squirrel’s biology and population dynamics, habitat and microclimactic factors would be funded by the Forest Service or the [University]. Information on construction impacts, and techniques would be gathered during the development of the minimum facility [i.e., the first three telescopes] and the responses of red squirrels would be monitored. Interactions between humans and red squirrels would also be studied. If an application to develop additional telescopes ... is subsequently filed, information from these studies could be used in issuance of the ensuing biological opinion.
Although it is fair to say that this language clearly contemplates a substantial effort to determine the effects of construction on the red squirrel, the description of how to conduct that effort remains quite general.
The only detailed statement of what the monitoring program ought to look like is to be found in the Management Plan. The legal status of this document, its ability to bind the various parties to this case, is perhaps the fundamental question raised by this appeal.
The AICA states unequivocally that “the University of Arizona, with the concurrence of [the Forest Service], shall develop and implement a management plan____” AICA, sec. 604(a). The statute envisions the plan as a comprehensive document covering all aspects of “the construction, operation, and maintenance of the Site [i.e., of [1573]*1573the observatory complex], access to the Site, and related support facilities.” AICA, sec. 604(b). Importantly, the statute mandates that the plan be “consistent with the requirements of the Endangered Species Act and with the terms and conditions of Reasonable and Prudent Alternative Three____” AICA, sec. 604(a).
Feature 12 of Alternative Three, in turn, makes a monitoring program a necessary component of any management plan: “Adequate funding for monitoring of red squirrels adjacent to [the] astrophysical development and the associated road systems is required for the life of the astrophysical complex and this monitoring will be considered in development of the management plan” (emphasis supplied). Thus, the AICA, through its incorporation of Alternative Three, makes a monitoring program a statutorily required aspect of the management plan. The University is responsible for drawing up the Plan; the Forest Service is to consult and its approval is a prerequisite to implementation.
IV
The appellees have complied with this requirement, as far as it goes. That is, the Management Plan does incorporate, and the University and the Forest Service have carried out, a monitoring program. It is not disputed, however, that the program as implemented is different in many respects from what was originally envisioned. It is Sierra Club’s contention that the program that is actually in place does not suffice under the AICA.
Two separate grounds appear to be alleged in support of this contention—they have not been clearly distinguished in Sierra Club’s presentations to this court. First, Sierra Club argues that insofar as the monitoring program implemented by the appellees deviates from the requirements of the Management Plan, it is per se violative of the AICA. Second, Sierra Club argues that the monitoring program is not providing and cannot provide “useful” and “unquestionable” data on the squirrel, and is therefore “improper” and “inadequate” under the AICA. We note that these two lines of argument, though nominally alike in alleging that the monitoring program contravenes the AICA, are logically quite independent, and thus must be analyzed separately. We consider each in turn.
A
Sierra Club argues that in deviating from the monitoring provisions in the Management Plan the defendants have committed per se violations of the AICA. The major premise here is left largely implicit. Essentially, Sierra Club takes the position that, once the Forest Service concurred in the Management Plan developed by the University, both parties became bound to carry out the Plan according to its terms. In effect, Sierra Club seeks to enforce the Management Plan against its authors.
To this end, Sierra Club presents a series of comparisons between what the Management Plan “required” and what the defendants did in practice. In order for such comparisons to have any independent legal significance, however, it must first be established that the terms of the Management Plan are binding—more particularly, that they are binding upon the Forest Service, for they are unquestionably enforceable (at the behest of the Forest Service) against the University. We have already determined that the Forest Service is charged by the AICA with responsibility for administering the statute. See Red Squirrel II, 954 F.2d at 1457. Unless the Management Plan is found to be a source of independent legal obligation for the Forest Service, that Plan cannot be “violated” when the agency, exercising its discretion in administering the observatory project to ensure compliance with the AICA, permits the Plan’s monitoring provisions to be abandoned or modified.
One possible source for the obligation Sierra Club seeks to enforce is Feature 5 of Alternative Three. This provision states: “Forest Service will develop construction inspection methods and monitoring that will ensure compliance with the management plan and will provide a mechanism for immediate control of on-site activities” (emphasis supplied). Since Alternative [1574]*1574Three was itself made part of the AICA, this language has the force of statutory law. Arguably, it obligates the Forest Service not to permit any departure from the Management Plan, including the monitoring provisions, once the Plan is in place.
There are, however, several problems with this argument. First, it ignores other indications that Congress did not intend to trammel so severely the discretion exercisable by the Forest Service in overseeing the University’s implementation of the Management Plan. Foremost among these is Feature 20 of Alternative Three, which provides that “[a]s findings are developed from the studies, modifications to the management plan ... will be made as appropriate.” This provides a strong indication that the Forest Service was meant to retain flexibility to institute changes in the Plan, and in particular its provisions concerning the study of the red squirrel, as the desirability of those changes became evident.2
More generally, this argument presumes a rather odd role for an administrative agency to which Congress has committed oversight responsibility in a situation requiring technical expertise. That is, the argument suggests that the Forest Service should, after concurring in the University’s proposed Management Plan, have restricted itself to the role of enforcer of the terms of that Plan. It is difficult to imagine that Congress intended for the Forest Service to bring its scientific knowledge to bear on the question of how to study the red squirrel only once, at the point where relatively little information was available, and then deny it the opportunity to revisit that question as more data were accumulated. Indeed, Feature 20 of Alternative Three suggests exactly the opposite intention.
Sierra Club predominantly relies on a rather different argument to establish that the Management Plan cannot be modified without violating the AICA. Sierra Club maintains that the monitoring provisions of the Management Plan acquired binding force by virtue of being incorporated into the Special Use Authorization issued by the Forest Service to the University. This argument proceeds as follows.
A special use authorization is, as the name implies, a document that evidences permission to use land administered by the Forest Service in a certain manner, while establishing the terms and conditions of that use. Here, the AICA expressly directed the Forest Service to issue a special use authorization to the University so that the observatory project could proceed. AICA, sec. 604(a). At the same time, the statute mandated that “the Management Plan shall be included in any Special Use Authorization issued ... to the University____” AICA, sec. 604(c). Sierra Club essentially argues that, because Congress expressly provided that the Management Plan be included in the Special Use Authorization, the Plan is constitutive of the “use” authorized. Any departure from the Management Plan, Sierra Club contends, therefore amounts to a change in the “use” to which the land covered by the authorization is being put.
Sierra Club then points out that Forest Service regulations provide that the holder of a special use authorization must “file a new or amended application ... to cover new, changed, or additional use(s) or area.” 36 C.F.R. § 251.61(a). The University was [1575]*1575thus obligated, says Sierra Club, to file an application to cover the new “uses” that arose when the monitoring provisions written into the Management Plan were modified in practice. Similarly, the Forest Service was obligated to follow its own procedures for evaluating applications to modify special uses.
Sierra Club argues that the failure of the defendants to follow these procedures when implementing changes to the Management Plan establishes that the Management Plan was never lawfully “changed,” but has instead been repeatedly “violated.” The conclusion we are invited to draw from this is that the will of Congress as expressed in the AICA has been transgressed.
This argument is meritless. Particularly suspect is Sierra Club’s contention that changing a given aspect of the monitoring program suffices to change the University’s “use” of the land covered by the permit. On its face, the special use permit states that it “is issued for the purpose of: Scientific and Astrophysical Research.” This is surely the most natural understanding of the “use” authorized by the permit.3
In addition, it is difficult to believe that Congress intended to force the Forest Service to resort to the procedures for approving new uses every time a change in the monitoring program (or, indeed, any aspect of the observatory project) was found to be desirable. Such a construction of congressional intent is inconsistent with the flexible approach to the study of the red squirrel contemplated by Feature 20 of Alternative Three, as well as with the broad discretion necessarily conferred upon the agency by the AICA’s lack of specificity in defining its monitoring requirements. A better explanation of Congress’s decision to require the incorporation of the Management Plan in the Special Use Authorization is that it intended thereby to enhance the Forest Service’s authority over the University. Building the Management Plan into the permit gave the agency power to revoke its authorization if ever the University departed from the Plan in a fashion not approved by the Forest Service. And, indeed, the special use permit itself announces that “compliance with the requirements of” the Management Plan “is made an express condition of this permit.”
In sum, Sierra Club fails to demonstrate that the Management Plan creates any obligations with respect to monitoring that tie the hands of the Forest Service. That certain aspects of the monitoring program have been changed from what the Management Plan originally envisioned is thus of no independent légal significance. Only the AICA, Alternative Three, and the general obligation of an agency to refrain from action that is arbitrary and capricious in implementing legislation can be asserted as grounds for challenging the Forest Service’s decisions in connection with the squirrel monitoring program. Simply put, deviations from the Management Plan do not, in and of themselves, constitute violations of the AICA.
B
The other theme running through Sierra Club’s briefs is that the monitoring program is “inadequate.” In staking out this position, Sierra Club relies heavily on certain comments made in this court’s Order of Limited Remand. In particular, it relies on our observation that “an effective monitoring program is ... a necessary prerequisite to carrying out the mandate of section 603(b) of the statute regarding the second phase of construction. A monitoring program that yields questionable information is of little value in determining the effects of construction on the Mount Graham red squirrel.” Red Squirrel I, 930 F.2d at 705. This language, according to Sierra Club, “settles” the standards we are to apply [1576]*1576here, and is conclusive as to the relevant inquiry on this appeal. We are told that the question before us, finally, is whether or not the monitoring program implemented by appellees is “yield[ing] questionable data of little value,” because we ourselves have said that the AICA requires no less.
We must reject any suggestion that the observations made in our earlier remand order carried the significance Sierra Club assigns to them. Our comments did not purport to “settle” the question of the scope or nature of review applicable to this appeal, nor could they do so, since that question was not before us then.4 More to the point, if used to frame the issue on appeal, this language plainly misstates the proper judicial inquiry. Whether the monitoring program is “adequate,” “proper,” or “effective,” whether it yields “useless” or “questionable” data that is not of “genuine value”—these are questions that a reviewing court is understood to be ill-equipped to decide. See, e.g., Marsh v. Oregon Natural Resources Council, 490 U.S. 360, 378, 109 S.Ct. 1851, 1861, 104 L.Ed.2d 377 (1989). It is reasonable to assume that Congress, in enacting the AICA, intended that an adequate monitoring program, rather than an inadequate one, should be implemented. But this does not mean that Congress intended the federal courts to be the arbiters of what constitutes “adequacy” in squirrel monitoring. Such judgments require technical expertise the courts do not possess, and they are committed by the AICA to the sound discretion of the Forest Service. Judicial review therefore involves neither more nor less than an assessment of whether the Forest Service “conducted a reasoned evaluation of the relevant information and reached a decision that, although perhaps disputable, was not ‘arbitrary or capricious.’ ” Marsh, 490 U.S. at 378, 109 S.Ct. at 1861.
V
Having identified the relevant inquiry, we may now evaluate Sierra Club’s factual allegations. Sierra Club alleges that the monitoring program is deficient with respect to: (1) the location of the lower control area; (2) the trapping and tagging of squirrels; (3) baseline monitoring; and (4) peer review procedures. On all these points, the district court found that the Forest Service had acted reasonably under the circumstances.5 These findings are conclusive of this appeal unless clearly erroneous.
In a situation like this, an appellate court’s search for clear error is unlikely to prove fruitful. At bottom, “this case is a classic example of a factual dispute the resolution of which implicates substantial agency expertise.” Marsh, 490 U.S. at 376, 109 S.Ct. at 1860. For every weakness in the monitoring program identified by Sierra Club and attested by its expert witnesses, appellees point to strengths and present corroborative testimony of their own. “When specialists express conflicting views, an agency must have discretion to rely on the reasonable opinions of its own qualified experts even if, as an original matter, a court might find contrary views more persuasive.” Id. at 378, 109 S.Ct. at 1861.
Location of the Lower Control Area. The Management Plan describes the basic design of the monitoring program as follows:
Red squirrels living near the astrophysical complex and its associated roads will be monitored intensively during the construction period and periodically from that time on as long as this facility re[1577]*1577mains on the mountain. Such monitoring will be designed to assess the impact of the construction, as well as the presence and operation of the astrophysical complex, on these squirrels. This assessment will be accomplished by a comparison of the behavior and other characteristics of the squirrels in a study area (surrounding the complex and its access roads) with those occurring in a control area____ The control area will have conditions similar to those existing in the study area, have an adequate number of middens to constitute a statistically significant sample, and will be sufficiently distant from the study area to prevent any disturbance from direct effects of construction and operation of the Observatory.
Initially, as this language indicates, monitoring was to encompass only one study area and one control area. Appropriate sites were chosen and the monitoring effort was begun. University biologists soon recognized, however, that the lower part of the study area had certain unique characteristics—a significant number of squirrel middens that had not been previously identified, a “mixed conifer habitat”—that served to distinguish it from the upper part. It was therefore decided that the study area would be split in two and a second control area added to the program, one having characteristics more in common with the newly created lower study area. A site was chosen and the lower control area established immediately adjacent to the lower study area. This decision was implemented in conjunction with the Forest Service.
Sierra Club argues that because the two areas share a common border, the data amassed in comparing the one with the other are likely to prove questionable. Two reasons are advanced: (1) the effects of construction (the variable being studied) are being felt in both the study and control areas; and (2) squirrels can move back and forth between the study and control areas. The district court heard testimony that these two factors would combine to frustrate efforts to maintain two distinct populations of squirrels, and to ensure that in general each was exposed to comparable stimuli while only the squirrels in the study area were exposed to the effects of construction. Thus, the court was told, a meaningful comparison would not be possible between squirrels that were exposed to the effects of construction and squirrels that were not.6
Appellees do not deny that it is not optimal to have a control area adjacent to a study area. They do deny, however, that the data obtained from monitoring squirrels here will be questionable. Moreover, they argue that sound scientific considerations guided the choice of the lower control area. Foremost among these were its similarity to the lower study area with respect to habitat characteristics and density of squirrels, and its accessibility. The appellees also presented evidence in the district court that the lower control area was insulated from direct construction effects, that studies of other potential control areas were carried out before the actual area was designated, and that utilization of particular techniques of analysis would help to reduce any bias caused by the contiguity of the lower study and control areas.
The above is by no means an exhaustive summary of the arguments of the parties and the evidence before the district court. It is enough, however, to permit us to conclude that the district court committed no error in finding that “[t]he establishment of control and study areas was reasonable under all the circumstances.” Given the presence of apparently reasonable expert opinion on both sides of the question whether the placement of the lower control area compromised the monitoring program, we could arrive at no other conclusion.
Trapping and Tagging. The Management Plan provides that, “[although sever[1578]*1578al years may be required for this to be accomplished, using procedures approved by the Forest Service, the red squirrels in both the study and control areas will be live trapped, tagged, and marked for observational recognition.” In the period immediately preceding the start of construction, two squirrels were successfully trapped and tagged. However, a squirrel then died of stress during handling while University personnel were attempting to tag it. Trapping was suspended by the Forest Service, which reviewed the matter and promulgated revised trapping and tagging procedures. Even so, a second squirrel died during tagging a short time later. The Forest Service, after consultation with the University and the Fish and Wildlife Service, suspended the trapping and tagging indefinitely, until improvement in the squirrels’ food supply made them better able to withstand the stress of handling.
Sierra Club argues that the biologists who testified regarding monitoring were unanimously of the view that “some sort of tagging or marking” is “critical” to allow meaningful data to be obtained from the monitoring program. The appellees, for their part, admit that marking or tagging would be “useful” in monitoring the squirrel middens. They maintain, however, that Sierra Club overstates the benefits to be realized by tagging. In support, appellees point to expert testimony stating that important information about individual squirrel behavior was obtained even without tagging, as the squirrels could be differentiated, albeit with more difficulty, on the basis of sex characteristics and individual animal markings.
Moreover, appellees point to other evidence that the decision to suspend trapping and tagging was a reasonable one. They note that it was the considered judgment of their biologists that trapping the squirrel was an “impossible” task under then-existing conditions, making the decision to suspend trapping a prudent step that avoided the risk of harming other squirrels. In addition, they insist that in the face of the congressional intent expressed in the AICA that construction on the first three telescopes go forward “immediately,” it would have been “untenable” and a breach of the Forest Service’s duty under the statute to halt construction when trapping and tagging were suspended.
This court ordinarily defers to an agency’s interpretation of its responsibilities under a statute. See Citizens for Clean Air v. EPA, 959 F.2d 839, 844 (9th Cir.1992) (only “administrative constructions which are contrary to clear congressional intent” will be rejected). The district court found that it was “reasonable” to suspend trapping and tagging, and that, “[although trapping and tagging might be desirable, the current lack of such is not fatal to the overall monitoring plan____” These findings do not appear to be clearly erroneous.7
C
Baseline Monitoring. One component of the monitoring program looked to the collection of data pre-construction, in order to assist in identifying construction impacts. The Management Plan thus provided that “[sjtarting at least one month prior to tree felling and excavation, the nearby middens will be monitored to establish baseline information.” Sierra Club directs our attention to the undisputed fact that the lower control area was not monitored for a full thirty days. It does not, however, indicate how this fact shows that the monitoring program is unacceptable. Appellees note that the adequacy of baseline monitoring of the “nearby middens” and upper control area is not challenged. They explain that the reason for the limited baseline monitoring of the lower control area was the fact that it was only established after baseline monitoring of the other areas had already begun. They adduce expert opinion to the effect that the inability to monitor for a full thirty days was not harmful to the monitoring results.
The district court found that “an adequate thirty day baseline monitoring program was carried out” with respect to the [1579]*1579original control area and that “[t]he addition of another control area ... did not require further baseline monitoring prior to construction.” These conclusions do not betray clear error.
D
Peer Review. Sierra Club charges that the monitoring program was not subjected to proper peer review before its adoption. The shortcomings alleged seem to boil down to the fact that only two (rather than three) reviewers were used, both of whom suffered from conflicts of interest.
On the first point, there was testimony in the district court that an “adequate” peer review generally involves a minimum of three reviewers. As for the second, Sierra Club complains that one of the peer reviewers was an employee of the Fish and Wildlife Service, while the other was a former college professor of the Forest Service biologist with responsibility for overseeing the monitoring program. As a result, says Sierra Club, neither peer reviewer could be counted upon to conduct an objective appraisal of the proposed monitoring program.
Appellees introduced testimony from the University’s Vice President for Research to the effect that two peer reviewers is a “typical” number. They also produced a letter written by a Sierra Club official which stated that the two individuals ultimately selected to carry out the peer review were considered by the organization to be “exceptionally qualified” to be on the peer review committee. Finally, appellees sought to rebut the allegations of conflict of interest by clarifying the relationships of the reviewers to Forest Service personnel.
The district court found that the peer review was “adequate and reasonable under the circumstances,” that there was no conflict of interest, that the use of two reviewers “was both proper and within the bounds of accepted procedure within the scientific community,” and that there was no evidence the peer reviewers “lacked either the expertise or the impartiality necessary to peer review the plan.” These findings are not clearly erroneous.
In the end Sierra Club can point to nothing done by the University and permitted by the Forest Service that was not “based on a consideration of the relevant factors,” nor can it demonstrate that the Forest Service has been guilty of any “clear errors in judgment.” Marsh, 490 U.S. at 378, 109 S.Ct. at 1861. We conclude that the Forest Service has been neither arbitrary nor capricious in any of its actions with respect to the monitoring program. Sierra Club’s claim that the program violates the AICA therefore fails.
VI
Sierra Club argues that road closures required by the AICA have not been properly implemented by the Forest Service.
Feature 17 of Alternative Three requires that all roads beyond the junction of Swift Trail and FR 507 “be closed year-round to all motorized vehicles except those officially authorized.” Appellees admit that some of those roads remain open, albeit on a seasonal basis only. However, the Forest Service has interpreted the language quoted above as granting it discretion to “officially authorize” any vehicles to use these roads when there is reason to do so. This apparently strained interpretation becomes more plausible when note is taken of Feature 15 of Alternative Three, which provides that “Forest Road 669 will be gated closed to all but official vehicles beyond the Observatory boundary” (emphasis supplied). The distinction between vehicles that are “official” and those that are “officially authorized” supports the Forest Service’s interpretation of its authority.
The Forest Service offers two explanations for why roads have been kept open above the Swift Trail-FR 507 junction. First, it argues that the terms of the AICA itself require the Forest Service to continue to permit access to a Bible school and summer homes that are located on Mount Graham. See AICA, sec. 605(a) (“Special Use Authorizations now in effect for the ... Summer Home Tract area and the ... Bi[1580]*1580ble School ... shall continue”). In addition, the Forest Service says that it has permitted traffic to flow to some existing campgrounds and recreation areas on Mount Graham that are outside of the squirrel’s protected habitat, in an attempt to lessen the “pressure” on the squirrel in the more “environmentally sensitive” areas in which it dwells. Sierra Club does not contradict these explanations.
It therefore appears that the Forest Service’s road closure decisions have not violated the AICA.
VII
Sierra Club alleges violations of section 9 of the Endangered Species Act (“ESA”), which prohibits “incidental takings” of endangered species. 16 U.S.C. § 1538. A “taking” is defined by statute to include harm or harassment of any kind, including, of course, killing. “Incidental” takings are those that “result from, but are not the purpose of, carrying out an otherwise lawful action.” 50 C.F.R. § 402.02. Under section 7 of the ESA, however, limited takings may be permitted if they are incorporated into the “terms and conditions” of a Reasonable and Prudent Alternative drawn up in connection with the issuance of a Biological Opinion.
Here, Alternative Three permits the “incidental taking” of up to six squirrels per year. Sierra Club does not allege that this limit has been exceeded. Instead, it argues that the two squirrels that died during trapping and tagging were taken “outside of required procedures,” and are therefore not covered by the permission built into Alternative Three pursuant to section 7 of the ESA. As such, Sierra Club maintains, these squirrel mortalities constituted prohibited takings under section 9 of the ESA.
We reject this argument. To begin, the assertion that when these squirrels died their handlers were not using “required procedures” misstates the record before us. In both cases, the handlers were authorized to trap and to tag squirrels and were using procedures approved by the Forest Service when the deaths occurred. In stating that “required procedures” were not followed Sierra Club appears primarily to mean that “accepted tagging procedures” were being ignored. The “accepted” procedures to which Sierra Club adverts are those of Dr. Halvorsen, a Fish and Wildlife Service biologist, who uses a paper cone to avoid handling squirrels with his bare hands in an effort to reduce the stress to which they are subjected. After the first squirrel mortality, Dr. Halvorsen was invited to demonstrate his technique to University personnel. The second squirrel died only a few days later. Its handler, Dr. Young, the University official in charge of the monitoring program, used his bare hands in tagging the squirrel.' However, Dr. Young himself is, as Sierra Club notes in another context, “an emerging squirrel expert” with extensive experience in trapping and tagging squirrels. Whether the squirrels that were killed were being improperly handled thus depends on the resolution of a difference of opinion between battling experts.. We do not intercede in such battles.
All this is somewhat beside the point, however. Even if both squirrels were mishandled, their deaths would still fall within the incidental take allowance established by Alternative Three. These were takings that were “not the purpose of ... an otherwise lawful action,” and thus unquestionably incidental takings within the meaning of the ESA. Sierra Club points to nothing in the statute that distinguishes between incidental takings that result from the application of “improper” procedures and their opposite. Thus, the two deaths were covered by the incidental take limit, and removed from the ambit of section 9 of the ESA by the terms of Alternative Three.
VIII
Finally, we address the University’s challenge to Sierra Club’s standing to seek judicial review of the AICA. Although logically it is among the first of our concerns, we have deferred discussion of this issue until now in order that the facts on which we base our conclusion might come into [1581]*1581better focus through consideration of the merits of the case.
Sierra Club suggests that this court has already implicitly rejected the contention that it has no standing to assert the claims we consider here. It draws attention to our previous Order of Limited Remand, which allegedly reflects a conclusion that Sierra Club had standing to challenge the Forest Service’s decisions with respect to monitoring and road closures. There we stated: “The monitoring allegations were properly before the district court ... when it refused to act because it believed it lacked jurisdiction” while the appeal of its grant of summary judgment was pending. Red Squirrel I, 930 F.2d at 705 (emphasis supplied). Since we are always mindful of our obligation to raise the issue of standing sua sponte, City of South Lake Tahoe v. California Tahoe Regional Planning Agency, 625 F.2d 231, 233 (9th Cir.), cert. denied, 449 U.S. 1039, 101 S.Ct. 619, 66 L.Ed.2d 502 (1980), this language conceivably could support a determination that doubts about standing have already been considered and resolved in this litigation. See Red Squirrel II, 954 F.2d at 1448 n. 13 (observing that the standing of “the first-named party” to sue had not been raised “and, notwithstanding our normal rules,” would not be discussed). In any event, because the question of standing trenches upon our very jurisdiction to hear this case, we deem it appropriate to undertake a thorough review of the question rather than to rely on uncertain inferences regarding the import of our past deliberations.
We have no difficulty concluding that Sierra Club has standing here.8 Standing is asserted under section 10(a) of the Administrative Procedure Act (“APA”), which provides that “a person ... adversely affected or aggrieved by agency action within the meaning of a relevant statute, is entitled to judicial review thereof.” 5 U.S.C. § 702. We have recently noted that, “[i]n cases arising under the APA, the standing requirement ... mean[s] that plaintiffs must show the challenged action has caused them injury in fact and that the injury is to an interest arguably within the zone of interests to be protected or regulated by the statutes that the agencies were claimed to have violated.” Idaho Conservation League v. Mumma, 956 F.2d 1508, 1514 n. 12 (9th Cir.1992) (internal punctuation and citations omitted).9
In its complaint, Sierra Club alleged that “the organizations [bringing suit] and their members10 derive scientific, recreational, and aesthetic benefit and enjoyment from the existence in the wild of the Mt. Graham Red Squirrel.” Sierra Club supports its claim to standing by arguing that these interests would be irreparably harmed if the squirrel were permitted to become extinct. The logic of such a position is hard to dispute. Moreover, there is no question that harm to these kinds of interests is sufficient to make out injury in fact under current standing doctrine. As recently as last year the Supreme Court stated, “the desire to use or observe an animal species is undeniably a cognizable interest for purposes of standing.” Lujan v. Defenders of Wildlife, — U.S. —, —, 112 S.Ct. [1582]*15822130, 2137, 119 L.Ed.2d 351 (1992).11 We therefore conclude that Sierra Club has made a sufficient showing of injury in fact to be entitled to standing here.
The University nonetheless argues that the injury of which Sierra Club complains is to an interest that is not within the “zone of interests” protected by the AICA. According to the. University, the sole purpose behind the enactment of Title VI of the AICA was to allow the observatory project to go forward at once, thereby eliminating the delays that might be caused by judicial challenges like this one. The interests asserted by Sierra Club in this litigation, we are told, are not among those sought to be served by permitting construction of the first three telescopes, and, indeed, are flatly inconsistent with those interests.
We are not persuaded. In the first place, we reject the University’s characterization of the “purpose” behind the AICA. The AICA surely was intended to forestall any further challenges to the first phase of the observatory project based on section 7 of the Endangered Species Act (“ESA”)— indeed, we held as much in Red Squirrel II. But the statute cannot be said to have thereby foreclosed all judicial challenges on any grounds whatsoever.
Moreover, the University's argument ignores other purposes plainly intended to be served by the AICA. For example, one purpose of the statute was to set up a monitoring program to collect data on the red squirrel, which data is required by statute to be considered during consultation under the ESA with respect to the second phase of the observatory project. Had Congress been unconcerned with the fate of the squirrel, it could easily have refrained from requiring such a monitoring program. Similarly, Congress need not have incorporated into the statute the terms of Alternative Three, which expressly provide that the management plan should “govern the construction and operation of the astrophysical complex and the associated road systems in ways least likely to adversely affect the squirrel____” Furthermore, the monitoring provisions of the AICA, as the University itself points out, derive in part from the ESA requirement that the effects of incidental takings be monitored, which requirement is surely aimed at preventing the extinction of endangered species.
Far from being antithetical to Congress’s purposes, Sierra Club, in seeking to ensure that the Forest Service lives up to its responsibilities with respect to the monitoring program, may well be acting in furtherance of one of them. We have concluded that Sierra Club is wrong that the Forest Service has fallen short of the mark; but that does not foreclose Sierra Club’s right to make the case.
For the same reasons, we believe that the University’s reliance on the Supreme Court’s decision in Clarke v. Security Industry Ass’n, 479 U.S. 388, 107 S.Ct. 750, 93 L.Ed.2d 757 (1987), is misplaced. Clarke states that a plaintiff does not meet the “zone of interests” test for standing “if the plaintiff’s interests are so marginally related to or inconsistent with the purposes implicit in the statute that it cannot reasonably be assumed that Congress intended to permit the suit.” 479 U.S. at 399, 107 S.Ct. at 757.
The University, citing Clarke, argues that Sierra Club’s interests are wholly inconsistent with the purposes of the AICA. Even if it were so, this is only half an argument. Under Clarke, we would have to conclude not only that Sierra Club’s interest is inconsistent with the purposes of the AICA, but also that this inconsistency is so fundamental as to make it impossible [1583]*1583to believe that Congress intended to permit Sierra Club to bring suit. See Clarke, 479 U.S. at 399, 107 S.Ct. at 757 (“at bottom the reviewability question turns on congressional intent, and all indicators helpful in discerning that intent must be weighed”). The University does not directly assert that Congress truly intended to foreclose review, nor would such a position have much to recommend it. Had Congress really wanted to prevent Sierra Club from continuing this battle in the courts, as it must have expected Sierra Club would attempt to do, it could easily have tailored this unique piece of legislation to that end. That judicial review is not expressly forbidden by the AICA is, under the particular circumstances of this case, substantial evidence that it was not intended implicitly to be foreclosed.
We therefore conclude that the University’s challenge to Sierra Club’s standing is without merit.
IX
The Sierra Club has standing to bring this suit. On the merits, however, the judgment in favor of the government and the University of Arizona is
AFFIRMED.