Opinion for the Court filed by Circuit Judge SILBERMAN.
SILBERMAN, Circuit Judge:The Michigan Department of Natural Resources (Michigan) petitions for review of a license the Federal Energy Regulatory Commission issued to the Indiana Michigan Power Company (the Company) to operate the Constantine Project, a 94-year-old hydroelectric-generating facility located on the St. Joseph River in Constantine, Michigan. Michigan challenges FERC’s refusal to require license conditions that Michigan sought. Petitioner has waived its objection to FERC’s declination to impose certain conditions — by not raising it in a rehearing before the Commission — and we deny the petition as to the others.
I.
In 1987, FERC determined that the Constantine Project (located on a navigable river of the United States) fell within the Commission’s jurisdiction. See Michigan Power Co., 38 F.E.R.C. ¶ 62,249 (1987). Accordingly, in 1988, the Michigan Power Company, then owner and operator of the project, filed with FERC an application for a hydroelectric license. (The Michigan Power Company subsequently merged into the Company.) Mich-. igan filed a request for license conditions designed to reduce the number of fish trapped in the project’s turbines and to compensate the state for fish killed.1 Michigan *1486designated its request as pursuant to § 10(j) of the Federal Power Act, 16 U.S.C. § 803(j) (1994), which obliges FERC to afford significant deference to recommendations made by state (and federal) fish and wildlife agencies for the “protection, mitigation and enhancement” of fish and wildlife.2
FERC’s Director of the Office of Hydro-power Licensing issued the Company a license for the project on October 20, 1993. The order treated Michigan’s fish protection and mitigation requests as § 10(j) recommendations, but, consistent with the staff-prepared Environmental Assessment accompanying his order, the Director found that fish protection devices at the project were economically infeasible and thus inconsistent with his obligation “to make licensing decisions that represent the best comprehensive use of the waterway.” 65 F.E.R.C. at 64,083; see Federal Power Act § 10(a), 16 U.S.C. § 803(a) (1988). The Director rejected Michigan’s suggestion that the fish killed by the project be valued according to restitution values codified in Michigan state law. Instead, he determined that the Company pay Michigan only the replacement value of fish entrained at the project, which he calculated as $3,880 annually (to be adjusted for inflation).
The Director’s order also included what is called a “bookmark”: an article reserving authority to the Commission to require the Company to set aside funds for the eventual decommissioning of the project. The Commission had only recently issued a Notice of Inquiry inviting comments on the appropriateness of new regulations regarding project decommissioning, and the bookmark was intended to defer decision on Michigan’s request for a decommissioning fund until after FERC adopted a policy. The “bookmark” was not, however, inserted under the auspices of § 10(j); the Environmental Assessment accompanying the Director’s license order stated that Michigan’s decommissioning recommendation was not considered pursuant to § 10(j) because it did not “provide measures for the protection, mitigation of damages to, and enhancement of fish and wildlife resources.” Subsequently, in December of 1994, FERC issued a policy statement (rather than a regulation) indicating that it would resolve decommissioning funding requirements on a case-by-case basis. FERC simultaneously withdrew bookmarks from the project license and from some 57 other licenses, finding that “the records in these cases demonstrate no current need to plan for, or expect, project retirement....”
The Director’s decision is final unless appealed to the full Commission, and, under FERC’s procedures, such an appeal is called — somewhat misleadingly — a rehearing petition. Michigan sought such a “rehearing” after the original order, and it later amended its petition to challenge the Commission’s subsequent order eliminating the bookmark. Michigan disputed the Director’s determination that a “compensatory mitigation” award adequately compensated Michi*1487gan for the loss of fish at the project. The state argued that the Director should have required a “comprehensive assessment of potential fish protection devices,” and he should not have restricted his valuation of the fish losses to their replacement value. The full Commission rejected Michigan’s claim but — and this turn is of central importance to the case — it explicitly disavowed the Director’s treatment of Michigan recommendations for fish protection and compensation as § 10(j) recommendations. It did so because Michigan had phrased its request as calling for a consultant to evaluate “all potential protection devices,” and to design and construct a protection device “if ... determined feasible” (emphasis added), which the Commission construed as calling for pre-li-cense studies which, under the Commission’s regulations,3 are not considered § 10(j) conditions. (It is the Commission’s view, presumably, that a § 10(j) condition must be definitive, but it is not apparent why Michigan’s request for a requirement of the “restitution” value of killed fish as opposed to its replacement value was insufficiently specific.) In any event, the Commission considered Michigan’s objections under § 4(e) and § 10(a) of the Federal Power Act, under which the Commission has broader latitude to balance environmental interests against development interests in promoting the best comprehensive use of a waterway. See 16 U.S.C. §§ 797(e), 803(a) (1988). The Commission held that the replacement value of fish entrained by the project was the proper measure by which to calculate an award to Michigan. The Commission noted that given the relatively low replacement value (which it “rounded up” to $4,000 per year), the Director correctly determined that various proposed protection devices — costing anywhere from $23,400 to $515,000 per year — were economically unsound alternatives. As for the decommissioning fund, FERC, agreeing with its Director, concluded that Michigan had produced no evidence to suggest that the project was financially shaky and that therefore there was no need to consider imposing an obligation on the project to set aside funds for its eventual retirement.
II.
As should be apparent, the question whether the Commission legitimately treated Michigan’s recommendations as falling outside of § 10(j) and therefore not entitled to the deference that section carries nor requiring the specific finding FERC must make before rejecting such recommendations (that they are “inconsistent with the purposes and requirements” of the Act or other provisions of law) is a weighty one. Unfortunately for Michigan (and other intervenors and amici), Michigan did not preserve that question for our review.
The Federal Power Act precludes a reviewing court from considering an “objection to the order of the Commission” unless the petitioner has urged that objection before the Commission in a petition for rehearing, or unless reasonable grounds excuse its failure to do so. Federal Power Act § 313(b), 16 U.S.C. § 825Z(b) (1988). We have emphasized repeatedly that we must construe strictly this “ ‘express statutory limitation[ ] on the jurisdiction of the courts.’ ” Town of Norwood v. FERC, 906 F.2d 772, 774 (D.C.Cir.1990) (quoting Tennessee Gas Pipeline Co. v. FERC, 871 F.2d 1099, 1107 (D.C.Cir.1989)); see also Platte River Whooping Crane Critical Habitat Maintenance Trust v. FERC, 962 F.2d 27, 34-35 (D.C.Cir.1992) (Platte River II); Malta Irrigation Dist. v. FERC, 955 F.2d 59, 65 (D.C.Cir.1992); Platte River Whooping Crane Critical Habitat Maintenance Trust v. FERC, 876 F.2d 109, 112-13 (D.C.Cir.1989) (Platte River I).
The Director’s first order in this case, as we noted, made clear that petitioner’s request that decommissioning funds be set aside in the event this project should prove some day to be no longer economically viable was not regarded as a § 10(j) recommendation. When FERC decided to remove the “bookmark” from the project license, ef-*1488feetively denying Michigan’s recommendation for a decommissioning trust fund for the project, Michigan did challenge that decision in a petition for rehearing. But it did so only on the grounds that the Commission’s substantive decision was arbitrary and capricious and “otherwise unlawful.” Michigan did not, in its petition for rehearing, make the specific objection it now makes before this court: that FERC was obliged to consider petitioner’s decommissioning recommendation under § 10(j). Michigan contends that it “implicitly raised the § 10(j) issue” “in light of Michigan’s numerous requests before the Director that its decommissioning recommendation be treated as a 10(j) recommendation.” (Emphasis added). Suffice to say that an argument “implicit” in prior requests before the Commission’s staff does not satisfy the strict standard of § 313(b). We are therefore barred from considering that objection under FPA § 318(b). See, e.g., Platte River II, 962 F.2d at 34-35; Town of Norwood, 906 F.2d at 774.
Michigan’s fish and wildlife recommendations, on the other hand, were considered by the Director as § 10(j) recommendations and rejected as such. It was only after Michigan appealed to the Commission that FERC disavowed that determination. The question is, then, whether Michigan was obliged to seek a “rehearing” before the Commission (which would have been the first rehearing before the- actual Commission) to challenge that legal determination. Michigan argues no because it claims that it had already raised the issue before the Commission by simply referring to its recommendations in its pleadings and rehearing petition as § 10(j) recommendations. That will not do. Michigan simply never presented to the Commission its argument that FERC’s interpretation of its regulation is contrary to the statute, or even that FERC was legally obliged to treat the recommendations as pursuant to § 10(j).
Michigan alternatively claims that it should be excused from presenting the statutory interpretation issue to the Commission because it had “reasonable grounds” not to do so. It is argued that too formalistic an interpretation of the jurisdictional exhaustion provision would permit FERC to engage in repeated shifts of its decision rationale, thus obliging a party to continually petition for rehearing and frustrating judicial review. We have recognized the difficulty in that scenario. In Town of Norwood, we “generally agreo[d]” that the Federal Power Act does not require “an endless cycle of rehearing applications.” 906 F.2d at 775. But in that case, in which the issue presented was whether a petitioner who had sought one rehearing of a full commission decision was obliged to seek another, we held that the petitioner was so obliged because the Commission had substantially changed its position in its rehearing order; indeed, it had reversed field and switched the result in the proceeding. Our case is a fortiori to Town of Norwood because we have only one Commission decision. It is only because FERC’s rules label an appeal from the Director as a “rehearing petition” that petitioner is able to claim that § 313(b) requires it to seek multiple rehearing petitions.4
Michigan also claims that it had reasonable grounds for not seeking rehearing because FERC sua sponte adopted its § 10(j) conclusion. Presumably, Michigan’s point is that this means that FERC must have actually considered whether the fish protection recommendation was a § 10(j) recommendation, and come to the conclusion that it was not. But unless we are to assume that when FERC acts sua sponte, it manifests an imperviousness to reasoned argument that it erred, petitioner’s argument cuts against it. That the Commission reached its conclusion without the benefit of argument *1489on the rather subtle question whether Michigan’s recommendations are properly treated under § 10(j) makes it more important, rather than less, that the nature of petitioner’s objection be presented to FERC before review is sought here.. Michigan raised that objection for the first time here, when it could easily have sought rehearing by the Commission. See Tennessee Gas Pipeline Co. v. FERC, 871 F.2d at 1110. Because it failed to do so, we lack jurisdiction to consider the objection.
III.
Although FERC is not obligated to defend the validity of the project license according to § 10(j), the license must still satisfy the other standards of the Federal Power Act. Section 4(e) mandates that, in considering whether to issue a license, the Commission must, in addition to power and development purposes, give “equal consideration to the purposes of energy conservation, the protection, mitigation of damage to, and enhancement of, fish and wildlife ..., the protection of recreational opportunities, and the preservation of other aspects of environmental quality.” 16 U.S.C. § 797(e). And § 10(a) requires that FERC determine that all licenses it issues “ “will be best adapted to a comprehensive plan for,’ inter alia, the protection of wildlife.” Platte River II, 962 F.2d at 32 (quoting 16 U.S.C. § 803(a)). Our review of the Commission’s decisions in this regard is deferential and “narrowly circumscribed”: “so long as its decision is supported by substantial evidence in the record and reached by reasoned decisionmaking, we will deny a petition for review.” Mine Reclamation Corp. v. FERC, 30 F.3d 1519, 1524 (D.C.Cir.1994) (internal quotation marks and citations omitted); see Federal Power Act § 313(b), 16 U.S.C. § 825i(b).
Michigan argues at the outset that the Commission did not engage in the requisite reasoned decisionmaking because it did not distinguish a prior decision in Consumers Power Co., 68 F.E.R.C. ¶ 61,077 (1994), which involved the licensing of 11 hydropower projects in Michigan. It is, of course, axiomatic that an agency adjudication must either be consistent with prior adjudications or offer a reasoned basis for its departure from precedent. Michigan emphasizes that Consumers Power found both that total funding of $5 million for fish protection devices at the 11 projects was “fair, reasonable and in the public interest,” id. at 61,377, and that there are “multiple benefits in pre-financing of project retirement,” id. at 61,382. Since the license order for the project differs so significantly from that approved in Consumers Power, and since the Commission did not explain why, Michigan contends that FERC’s decision here constitutes an unexplained deviation from precedent and is therefore arbitrary and capricious. See Bangor Hydro-Elec. Co. v. FERC, 78 F.3d 659, 663 n. 3 (D.C.Cir.1996) (explaining that the Federal Power Act’s substantial evidence test is an application of the arbitrary and capricious test to factual findings).5
The Commission responds that it was not required to distinguish Consumers Power because it was merely an approval of a settlement agreement, and such approvals do not establish precedent. For its part, Michigan asserts that a settlement that operates as a “decision on the merits as to the nonsignatories” does indeed count as precedent, and that Consumers Power is such a ease since several intervenors were not parties to the settlement. We think FERC has much the better of the argument. Consumers Power went to great lengths to emphasize that, particularly with respect to pre-financing of project retirement, it did not necessarily approve the approach taken by the parties. See 68 F.E.R.C. at 61,382 (“The Settlement does not seek Commission approval of the use of trust funds, and the Commission is not either approving or disapproving that approach.”). And although Michigan is correct to point out that a settlement that is forced onto a party is effectively converted into a decision “on the merits,” see, e.g., Mobil Oil Corp. v. FPC, 417 U.S. 283, 314, 94 S.Ct. 2328, 2348-49, 41 *1490L.Ed.2d 72 (1974),6 this is surely not such a case. Eight intervenors did not sign the settlement agreement in Consumers Power, but, we are told, neither did they register any objections to it. The Commission thus viewed the settlement as “uncontested,” and it reviewed it only to determine whether it was “fair and reasonable and in the public interest and ... supported by the record.” 68 F.E.R.C. at 61,372. We have previously admonished FERC for attempting to use uncontested settlements as precedent in later cases. See Office of Consumers’ Counsel v. FERC, 783 F.2d 206, 235 (D.C.Cir.1986) (“‘[i]t is the essence of a settlement agreement that the participants can agree on an end result without necessarily agreeing upon the reasoning, data, analyses or principles which led to the agreement’”) (quoting General Public Utilities Corp., 13 F.E.R.C. ¶ 61,002, at 61,006 (1980)); Alabama Power Co. v. FPC, 450 F.2d 716, 722 n. 12 (D.C.Cir.1971) (noting that “the elementary principles” of a settlement — that a company may settle for less than it could have obtained from litigation and that a court may approve a settlement different “from what it would have ordered” — have “vitality for agency proceedings”). The converse follows: if FERC cannot use uncontested settlements as precedent, neither can its adversaries.
Next, petitioner contends that the Commission acted illegally in declining to require the Company to set aside funds for a future decommissioning of the project. The Environmental Assessment that accompanied the Director’s licensing order pointed out that, at that stage, “Michigan ha[d] provided no persuasive evidence, indeed no evidence at all, to support its recommendation” regarding project retirement funding. In other words, petitioner has not shown that the project is economically vulnerable and therefore faces the prospect of decommissioning. Michigan relies only on the assumption, or hypothesis, that any such project has a limited useful life; accordingly the state wishes to avoid the risk that it will be saddled with the cost of removing the facilities at some point in time if the owner is financially unable to accomplish the task. The state claims that this has happened before, although apparently not regarding a project within FERC’s jurisdiction. The Commission, in fact, reserved authority to require the Company “to conduct studies, make financial provisions or otherwise make reasonable provisions for de-cpmmissioning of the project.” If the project were to appear shaky some time in the future FERC could therefore act to prepare for decommissioning. Michigan’s reliance on only a theoretical risk that the Company or a successor could use the bankruptcy courts to avoid any obligation to FERC is not enough to convince us that the Commission’s policy choice is an unreasonable one.
We turn finally to petitioner’s claim that FERC’s decision not to require the Company to install fish protection devices and instead to require it to pay Michigan $4,000 annually is arbitrary and capricious. Michigan does not. question the actual calculation of replacement value so much as it challenges the Commission’s refusal to take into account values in addition to replacement value.7 That is, Michigan argues that FERC ignored the value of the fish killed to anglers as well as the value Michigan residents place on. the continued existence of the fishery in question and on the opportunity to *1491use that fishery in the future. And if FERC had contemplated additional measures of loss when it determined the value of the fish killed by the project, it would have concluded that certain fish protection devices which seemed so expensive when compared to the replacement value were in reality quite reasonable when compared to the “actual” value. With respect to the value of the fish killed to anglers, however, the Commission explicitly noted that less than seven percent of fish entrained at the project were game fish, and of these the vast majority were young-of-the-year. FERC, moreover, found that the relevant fish population had co-existed with the hydroelectric project for over 90 years and that continued existence of the fishery did not appear to be in jeopardy.
Michigan argues that our decision in State of Ohio v. Department of Interior, 880 F.2d 432 (D.C.Cir.1989) (per curiam), precludes FERC’s reliance on replacement value. In State of Ohio, we reviewed regulations promulgated by the Department of Interior pursuant to the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA), 42 U.S.C. § 9651(c) (1994). We held invalid the Secretary’s regulation that limited “damages for injury to, destruction of, or loss of natural resources,” 42 U.S.C. § 9607(a)(4)(C), because it established ‘“a strong presumption in favor of market price and appraisal methodologies.’ ” 880 F.2d at 462-63 (quoting 51 Fed.Reg. 27,720 (1986)). We reasoned:
While it is not irrational to look to market price as one factor in determining the use value of a resource, it is unreasonable to view market price as the exclusive factor, or even the predominant one. From the bald eagle to the blue whale and snail darter, natural resources have values that are not fully captured by the market system.
Id. at 462-63 (citations omitted). State of Ohio focused on the impracticality of calculating a damage amount for an endangered species by looking exclusively to the market price. We made this clear when we emphasized that the challenged regulations “would dictate a use value for fur seals of $15 per seal, corresponding to the market price for the seal’s pelt.” Id. at 463 (citation omitted). The regulations at issue in State of Ohio, in contrast to the order at issue here, could not speak in terms of “replacement value,” since replacing an endangered species is simply not a viable concept. Here, on the other hand, there is no suggestion that the fish entrained by the project are of a species that is endangered, or that the fishery itself is in any danger due to the project. Requiring the Company to pay only the cost of replacing the fish killed by the project is thus reasonable.
‡ ‡ ‡ ‡ ‡ *
Accordingly, we deny the petition.
. Michigan recommended that the Company, "after consultation with and with approval of the resource agencies, develop a turbine entrainment and mortality plan, including contracting a qualified consultant to evaluate all potential protection devices to prevent fish losses at the project, determine the technical and economic feasibility of potential protection devices, and design and construct a protection device, if determined to be feasible. If no protection device is determined to be feasible at the project, Michigan recommend[ed] that [the Company] pay the annual *1486restitution value” of fish killed by the project to Michigan by October 1 of each year. 65 F.E.R.C. ¶ 62,063, at 64,108 (1993).
. Section 10(j) reads:
(1) That in order to adequately and equitably protect, mitigate damages to, and enhance, fish and wildlife ... affected by the development, operation, and management of the project, each license issued under this subchapter shall include conditions for such protection, mitigation, and enhancement. Subject to paragraph (2), such conditions shall be based on recommendations received pursuant to the Fish and Wildlife Coordination Act (16 U.S.C. 661 et seq.) from the National Marine Fisheries Service, the United States Fish and Wildlife Service, and State fish and wildlife agencies.
(2) Whenever the Commission believes that any recommendation referred to in paragraph (1) may be inconsistent with the purposes and requirements of this subchapter or other applicable law, the Commission and the agencies referred to in paragraph (1) shall attempt to resolve any such inconsistency, giving due weight to the recommendations, expertise, and statutory responsibilities of such agencies. If, after such attempt, the Commission does not adopt in whole or in part a recommendation of any such agency, the Commission shall publish each of the following findings (together with a statement of the basis for each of the findings):
(A) A finding that adoption of such recommendation is inconsistent with the purposes and requirements of this subchapter or with other applicable provisions of law.
(B) A finding that the conditions selected by the Commission comply with the requirements of paragraph (1).
16 U.S.C. § 803(j) (emphases added).
. "A ‘fish and wildlife' recommendation includes a request that cannot be completed prior to licensing, but does not include ... a request for additional pre-licensing studies or analysis.... ” 18 C.F.R. § 4.30(b)(9)(ii) (1996).
. We also expressed concern over the possibility of an endless cycle of rehearings in Southern Natural Gas Co. v. FERC, 877 F.2d 1066 (D.C.Cir.1989), in which we held that the judicial review provision of the Natural Gas Act did not require a petition for rehearing of a FERC rehearing order that made no change in the prior result but that "merely supplie[d] a new improved rationale.” Id. at 1073; see Natural Gas Act § 19(b), 15 U.S.C. § 717r(b). Southern Natural Gas was a case in which petitioner, prior to seeking review in the court of appeals, had twice pressed its cause before the Commission. See id. at 1072. And unlike in Town of Norwood, FERC did not fundamentally change its decision.
. This argument also relies on the premise— unarticulated here and unproven before the Commission — that the project is so similar to the projects licensed in Consumers Power that the principles of the latter case, if precedential, should apply to this one.
. We have previously indicated that FERO. need only review a contested settlement in a remedial proceeding to determine whether it is "reasonable and appropriate.” See Laclede Gas Co. v. FERC, 997 F.2d 936, 944-45 (D.C.Cir.1993). This view is not without its detractors. See, e.g., id. at 948-49 (Sentelle, J., concurring) (contending that FERC must ensure that the contested settlement is supported by substantial evidence).
. Indeed, Commission staff adopted Michigan’s proposed methodology in calculating the number of fish entrained at the project. The Company advocated the use of a geometric mean to derive an annual entrainment rate from data that it believed did not follow a normal distribution. Michigan favored an arithmetic mean because, in its view, the data were too few to determine whether they were normally distributed. The Environmental Assessment defends the choice of an arithmetic mean because the outliers that a geometric mean would de-emphasize in this context are actually high entrainment occasions that would occur at various times within a typical year and should thus be weighted equally with other samples.