This case requires us to navigate the treacherous terrain at the intersection of the federal government’s obligations, on the one hand, to refrain from imposing burdens on the individual’s practice of religion, and, on the other, to protect key aspects of our natural heritage and preserve the culture of Native American tribes. Defendant-Appellee Samuel Ray Wilgus was arrested in June of 1998 for possessing 141 feathers of bald and golden eagles. The Bald and Golden Eagle Protection Act (“Eagle Act”), 16 U.S.C. § 668, prohibits possession of the feathers or parts of eagles, but contains an exception to the ban when the feathers are possessed “for the religious purposes of Indian tribes.” Id. § 668a. The regulations implementing the exception limit its scope to members of federally-recognized tribes only, who are allowed to apply to the government for permits. 50 C.F.R. § 22.22. Wilgus is a follower of a Native American faith, but is not a member of a federally-recognized tribe, nor is he Indian by birth.
Faced with prosecution, Wilgus interposed as a defense the Religious Freedom Restoration Act of 1993, 42 U.S.C. § 2000bb-l (“RFRA”), which prohibits the federal government from substantially burdening the religious freedom of individuals, unless it does so to forward a compelling governmental interest via the least restrictive means. Wilgus argues that the government’s choice to limit legal possession of eagle feathers to members of federally-recognized tribes substantially burdens his religious exercise which, he claims, requires him to possess eagle feathers. The district court did not accept this argument, but this Court, sitting en banc, reversed and ordered a hearing on whether the Eagle Act is the least restrictive means of serving the government’s interests. United States v. Hardman, 297 F.3d 1116, 1135-36 (10th Cir.2002) (en banc).
On remand, the district court conducted a number of hearings, considering both live testimony and documentary evidence, and held that application of the Eagle Act to Wilgus does indeed violate RFRA. United States v. Wilgus, 606 F.Supp.2d 1308, 1334-35 (D.Utah 2009). Exercising jurisdiction under 28 U.S.C. § 1291, we REVERSE.
BACKGROUND
Because of the highly statutory nature of the dispute in this case, we find it most efficient first to review the provisions of the two statutes at issue, then to set forth the factual underpinnings of Wilgus’ conviction, and finally to review the procedural history of this matter.
Statutory Background
The Eagle Act
The Bald and Golden Eagle Protection Act, in combination with the Migratory Bird Treaty Act and the Endangered Spe*1278cies Act, is one of the cornerstones of our nation’s efforts to protect and preserve the bald eagle. The basis of that protection is a blanket ban on possession of any part of a bald or golden eagle:
Whoever ... without being permitted to do so as provided in this subchapter, shall knowingly, or with wanton disregard for the consequences of his act take, possess, sell, purchase, barter, offer to sell, purchase or barter, transport, export or import, at any time or in any manner, any bald eagle commonly known as the American eagle, or any golden eagle, alive or dead, or any part, nest, or egg thereof of the foregoing eagles, or whoever violates any permit or regulation issued pursuant to this subchapter, shall be fined not more than $5,000 or imprisoned not more than one year or both.
16 U.S.C. § 668(a).1 Originally passed in 1940, see Act of June 8, 1940, ch. 278, 54 Stat. 250-51, the Eagle Act was amended in 1962, adding both protection for golden eagles (the young of which are easily confused with bald eagles) and the exception at issue in this case, for possession for the religious purposes of Indian tribes. See Pub.L. 87-884, 76 Stat. 1246; Hardman, 291 F.3d at 1122. The exception reads:
Whenever, after investigation, the Secretary of the Interior shall determine that it is compatible with the preservation of the bald eagle or the golden eagle to permit the taking, possession, and transportation of specimens thereof for the scientific or exhibition purposes of public museums, scientific societies, and zoological parks, or for the religious pmposes of Indian tribes ... he may authorize the taking of such eagles pursuant to regulations which he is hereby authorized to prescribe.
16 U.S.C. § 668a (emphasis added). Therefore, provided that the Secretary of the Interior is satisfied that to do so “is compatible with the preservation” of bald and golden eagles, he may by regulation allow possession of eagle parts “for the religious purposes of Indian tribes.”
The Secretary, acting through the Fish and Wildlife Service (“FWS”) of the Department of the Interior, has exercised this statutory authority and formulated regulations governing eagle permits. See generally 50 C.F.R. Ch. I, pt. 22. These regulations reiterate the broad prohibition on possession without a permit set out in the statute. 50 C.F.R. § 22.11. In implementing the exception to the ban “for the religious purposes of Indian tribes,” the regulations make it clear that only members of federally-recognized tribes may apply for and obtain eagle feather permits: “[w]e will issue a permit only to members of Indian entities recognized and eligible to receive services from the United States Bureau of Indian Affairs listed under 25 U.S.C. § 479a-l engaged in religious activities who satisfy all the issuance criteria of this section.” 50 C.F.R. § 22.22. The Secretary will only grant the permit, however, if “we determine that the taking, possession, or transportation is compatible with the preservation of the bald and golden eagle.” Id. § 22.22(c). Finally, the permits and lawfully-possessed eagle parts “are not transferrable, except such birds or their parts may be handed down from generation to generation or from one Indian to another in accordance with tribal or religious customs.” Id. § 22.22(b)(1).
Eagle carcasses that are recovered by FWS agents or the public at large are sent to the National Eagle Repository, which is housed on the grounds of the Rocky Mountain Arsenal National Wildlife Refuge in *1279Commerce City, Colorado. United States v. Friday, 525 F.3d 938, 944 (10th Cir.2008). The Repository receives the eagle carcasses and the requests for feathers and parts pursuant to permits, then prepares the former in order to fulfill the latter. Native American religious practitioners wishing to obtain eagle parts from the Repository are required to file a request for a permit with the Wildlife Permit Office in their state of residence. Applicants are asked to fill out a form on which they must certify that they are enrolled in a federally-recognized tribe. The FWS forwards those permit requests to the Repository, which then attempts to match the requests with the available supply of eagle parts. As noted below, however, the Repository receives significantly more requests than it has available eagle carcasses, and so applicants for eagle feathers and parts typically must wait a long period of time before those requests are fulfilled.
The Religious Freedom Restoration Act of 1993
In response to the government’s indictment, Wilgus interposed the Religious Freedom Restoration Act of 1993. The genesis of RFRA lies in a protracted exchange between the Supreme Court and Congress over the proper standard to apply when reviewing laws that burden religion. In Employment Division v. Smith, 494 U.S. 872, 879, 110 S.Ct. 1595, 108 L.Ed.2d 876 (1990), the Supreme Court held that neutral laws of general applicability that nonetheless burden the exercise of religion would be subject only to rational-basis scrutiny under the First Amendment, rather than the heightened scrutiny it had applied in previous cases. In response, Congress passed RFRA, which purported “to restore the compelling interest test as set forth in Sherbert v. Verner, 374 U.S. 398, 83 S.Ct. 1790, 10 L.Ed.2d 965 (1963) and Wisconsin v. Yoder, 406 U.S. 205, 92 S.Ct. 1526, 32 L.Ed.2d 15 (1972) and to guarantee its application in all cases where free exercise of religion is substantially burdened.” 42- U.S.C. § 2000bb(b)(l). In City of Boeme v. Flores, 521 U.S. 507, 536, 117 S.Ct. 2157, 138 L.Ed.2d 624 (1997), the Supreme Court held that RFRA could not be constitutionally applied to the states as an exercise of Congress’ power to enforce the Fourteenth Amendment. RFRA can, however, be constitutionally applied to the federal government, as an exercise of Congress’ Article I, Section 8 powers under the Necessary and Proper Clause. See Kikumura v. Hurley, 242 F.3d 950, 959 (10th Cir.2001). Therefore, RFRA provides a statutory claim to individuals whose religious exercise is burdened by the federal government.
RFRA provides that:
(a) In general
Government shall not substantially burden a person’s exercise of religion even if the burden results from a rule of general applicability, except as provided in subsection (b) of this section.
(b) Exception
Government may substantially burden a person’s exercise of religion only if it demonstrates that application of the burden to the person—
(1) is in furtherance of a compelling governmental interest; and
(2) is the least restrictive means of furthering that compelling governmental interest.
42 U.S.C. § 2000bb-l (emphasis added). Thus, in order to make out a defense against a criminal prosecution under RFRA, the defendant must prove that his religious exercise is substantially burdened by the regulation at issue. If he does so, the government may save the provision by establishing that it is the least restrictive means of forwarding a compelling governmental interest. See Friday, 525 F.3d at *1280946 (upholding Eagle Act restrictions on taking eagles against RFRA challenge).
Samuel Ray Wilgus and the Original Offense
Although not born Native American and having been raised a Baptist, Samuel Ray Wilgus turned to Native American religion after his first marriage ended in the mid-1980s. Wilgus, 606 F.Supp.2d at 1311. He moved to Utah, where he lived with enrolled members of the Southern Paiute Nation; Wilgus eventually became a “blood brother” to Wilford Jake, a Paiute. Id. Over the course of his religious training and service, Wilgus was given a number of eagle feathers by Native Americans — both members of the Southern Paiute Nation and other tribes — as gifts. Id. at 1312. Though he received some of the feathers purely as gifts, there is no dispute that at the very least the first feather he received was for religious purposes. Id. at 1312 n. 7.
On June 5, 1998, a speeding truck in which Wilgus was a passenger was pulled over by the Utah Highway Patrol. Id. at 1312. The officer observed what appeared to be drug paraphernalia, but upon searching the back of the truck found 137 eagle feathers that belonged to Wilgus. Id. Upon searching Wilgus’ home, his (second) wife produced four more feathers; when Wilgus could not produce a permit for their possession, all of the feathers were confiscated. Id. Wilgus insisted that he was an adopted member of the Paiute tribe, and thus was entitled to possess the feathers, but the Chairperson of the Tribe pointed out to the authorities that Paiute law “does not permit the adoption of nonNative American persons.”2 Id.
For violation of the Eagle Act, Wilgus pleaded guilty to two misdemeanor counts carrying twelve months’ probation and a special assessment of $50. Id. Wilgus’ plea was conditioned upon his appeal of what he saw as a violation of his religious rights.3 Id. at 1312-13.
First Appeal and This Court’s En Banc Decision in United States v. Hardman
While Wilgus pursued his appeal, two other cases presenting similar issues were before different panels of this court, and when the three panels reached outcomes inconsistent with one another, the court granted rehearing en banc for all three cases. See United States v. Hardman, 260 F.3d 1199 (10th Cir.2001) (en banc) (vacating panel opinions and setting forth issues to be argued on rehearing en banc). In ruling on the three cases, the Hardman court conclusively resolved the majority of the issues underlying Wilgus’ RFRA claim, but left a few issues open for determination on remand. First, the court noted that there was no dispute “that claimants’ beliefs are sincerely held or that the regulations represent a substantial burden upon claimants’ religious beliefs.” Hardman, 297 F.3d at 1126. Thus, the Eagle Act regulations could only be enforceable in the face of RFRA if they advanced a compelling governmental interest via the *1281least restrictive means. 42 U.S.C. § 2000bb-l(b).
The Hardman court had no difficulty determining that the Eagle Act and its implementing regulations potentially advanced two compelling governmental interests: protecting bald and golden eagles, and preserving Native American culture and religion. 297 F.3d at 1127-28. With regard to the first interest, the court noted that it was supported not merely due to the scarcity of the birds, but rather due to the bald eagle’s unique status as one of the enduring symbols of our nation: “[t]he bald eagle would remain our national symbol whether there were 100 eagles or 100,-000 eagles. The government’s interest in preserving the species remains compelling in either situation.” Id. at 1128. This interest applies equally to golden eagles because immature bald and golden eagles are virtually indistinguishable from one another. Id. at 1127 n. 18. As to the second interest, while the court recognized that the government has a compelling interest — arising both from the text of the Constitution and the historical obligations of the government toward the peoples native to this continent — in protecting and preserving Native American culture, it left the precise contours of that interest in this situation undefined. Id. at 1128.
In the cases of defendants Wilgus and Hardman, the en banc court found the factual record insufficient to determine whether the Eagle Act and its regulations are the least restrictive means of forwarding either of the government’s two compelling interests. The court remanded those two cases to the district court to allow the government to develop a record on the issue of least restrictive means. Id. at 1131. In the case of the third defendant, Jesus Saenz, however, the Hardman court determined that the record did not support a finding of least restrictive means. Id. at 1132.
The Facts Presented on Remand
On remand, the district court held that the United States, in Wilgus’ case,4 failed to carry its burden of demonstrating that the Eagle Act was the least restrictive means of advancing the government’s compelling interests. Wilgus, 606 F.Supp.2d at 1335. In reaching this conclusion, the district court considered evidence in the following areas:
Promoting Native American Culture
The district court first attempted to answer whether allowing permits only to members of federally recognized tribes forwards the government’s interest in promoting Native American religion and culture. The court found that it simply could not answer that question either way: “Native American religions are neither hierarchical nor homogenous, and there is considerable disagreement among tribes holding eagle feathers sacred regarding the appropriate role — if any — of persons who are not tribal members in tribal worship.” Id. at 1315. While some tribes welcome non-Native American adherents in their worship, others regard members of other races “playing Indian” as a threat to Native American culture. See id. at 1315-18. With respect to the interest in promoting Native American culture, the court noted that “[w]hatever policy it chooses, the government will have furthered its compelling interest with regard to some tribes and frustrated it with regard to others.” Id. at 1318.
Evidence of the Numbers of Non-Native American Adherents
To attempt to establish the number of non-Native American adherents of Native American religions, the government relied *1282on combined data from the Census and from the General Social Survey (G.S.S.). The government’s expert sociologist calculated that approximately 33% of those claiming to adhere to Native American religion on the G.S.S. were, in fact, not of Native American descent. Wilgus, 606 F.Supp.2d at 1319. Based on the statistics, the sociologist further predicted that there were 30,590 Native Americans practicing Native American religions who could apply to the Repository for eagle feathers and parts. Id. at 1320. This number is far greater than the capacity of the Repository, but it is also far greater than the actual number of annual applications for eagle permits. The district court eventually discounted both the figures and the government’s expert testimony on this issue. Id. at 1328-30. The United States did not appeal the district court’s decision to discount the expert testimony.
Evidence of Adherents of Afro-Caribbean Religions and Eagle Parts
The government also put forward data on the numbers of practitioners of Afro-Caribbean religions, such as Santería, in which eagle feathers are also sacred. According to the government’s expert, there may be as many as one million practitioners of Santería who might apply for eagle feather permits, if the permitting process were opened up to non-Native Americans. The government also put forward evidence that modern “eclectic” approaches to religion have led many adherents to adopt aspects of Native American religion into a broader patois of spiritual beliefs, often identified as “New Age.” (Apl’t App at 96-97.)
Statistical Data on Eagle Populations
Eagles are particularly vulnerable to hunting, and small changes in the number of breeding adults can have drastic impacts on the overall health of the species. Bald eagles breed only after they reach the age of six years; they produce on
average only two eggs per year, of which only one will fledge. Wilgus, 606 F.Supp.2d at 1321. Of the birds that do fledge, only 50% will survive to maturity. Id. Because both the male and female parents of eagle chicks participate in raising young, losing one member of the pair makes it significantly less likely that chicks will survive. Id. Therefore, the loss of adult eagles has a proportionally greater impact on the species as a whole than in other birds. Id. “The loss of an adult breeding pair may be equivalent to the loss of 10-15 fledgling eagles due to the long-term loss of reproductive capability. On an annual basis, that would be ... all the reproduction of bald eagles for the State of Kansas for a year.” Id. at 1322 (quotation omitted). Nevertheless, efforts at conservation have increased the number of eagles surviving in the wild. In 2007, the Department of the Interior officially removed the bald eagle from the list of threatened species protected under the Endangered Species Act, 16 U.S.C. §§ 1531-1544.
Supply and Demand of Eagle Feathers and Parts
In 2004, the Repository received 1,822 requests for eagles and eagle parts. Wilgus, 606 F.Supp.2d at 1321. During that same year it received 1,647 eagles. Id. There is greater demand for golden eagles, but more of the eagles received by the Repository are bald eagles. Id. The waiting period for a whole bald eagle is two- and-one-half years; for an adult golden eagle the wait is three-and-one-half to four years; for an immature golden eagle the wait is up to four-and-one-half-years. Id. According to a brochure about the Repository produced by FWS, 95% of all requests are for whole golden eagles, and “[cjurrently there are over 4000 people on the waiting list for approximately 900 eagles the Repository receives each year.” (Apl’t *1283App. at 132.) As the district court found, “it is clear the Repository system is already vulnerable to any significant increase in demand.” Wilgus, 606 F.Supp.2d at 1321. While the number of eagles received by the Repository has risen on a year-to-year basis, it is unlikely that this will continue. Efforts at eliminating the most obvious causes of eagle deaths (such as power lines) have led to a reduction in the number of eagle deaths most likely to lead to a carcass being sent to the Repository. Id. at 1322-23. “[T]here is no significant untapped source of birds not already being sent to the Repository.” Id. at 1323 (quotation omitted).
Further complicating the supply-and-demand issue is the black market for eagle feathers and parts. Several FWS agents testified to the existence of a thriving black market in eagle feathers, driven in part by the “pow-wow circuit,” in which Native Americans compete in traditional performances. (See, e.g., Apl’t App at 86-87.) Participants compete for cash prizes, and the more decorated with eagle feathers a competitor’s costume, the more likely he is to win. Other sources of demand in the black market include Native Americans who need the feathers for religious purposes but who are stymied by the delays in the permitting system, and nonNative American collectors of Native American art and artifacts. No government witness, however, provided hard data on the volume of traffic in black-market eagle parts.
FWS agents also testified that a blanket ban on possession — with an exception only for members of recognized tribes for religious purposes — is the only effective way to enforce the statute. “At present,” the district court noted, “[FWS] agents can be fairly certain that a person who is not Native American who possesses eagle feathers does so illegally.” Wilgus, 606 F.Supp.2d at 1324-25. FWS agents predict that without a clear method for determining who is not eligible, efforts to enforce the ban will be mired, particularly given the lack of resources available for FWS enforcement. Id. at 1325. And the eagle feathers distributed by the Repository cannot be marked as legitimate; to do so would destroy the religious value to many Native Americans. Id. at 1324. There is, therefore, no way for FWS to tie particular feathers to particular permits or permit holders.
The District Court’s Conclusion
After reviewing the evidence, the district court held that the United States had not established that the limitation on permit applications to members of federally-recognized tribes was the least restrictive means of implementing the Eagle Act.
First, while it acknowledged the Hard-man court’s conclusion that the number of living eagles does not affect the compelling nature of the government’s interest in protecting them, the district court concluded that the delisting of the bald eagle “ma[de] it more difficult to show that measures impinging on religious practices which are justified — at least in part — as essential to the protection of eagles are the least restrictive means of ... protecting eagles.” Wilgus, 606 F.Supp.2d at 1327. Though the government argued that opening the permitting process to non-Native Americans would create overwhelming demand that would threaten eagle populations, the district court rejected that argument as speculative and based on questionable expert witness testimony. Id. at 1328. The district court also rejected the evidence of the numbers of practitioners of Afro-Caribbean religions, as those individuals would not be “identically situated” to Wilgus, as required by the Hardman en banc opinion. Wilgus, 606 F.Supp.2d at 1330-31.
*1284The district court then considered the potentially less restrictive alternative offered by Wilgus. It rejected the alternative of allowing Native Americans to give their eagle feathers to whomever they want as essentially unenforceable. Id. at 1332. A second alternative — simply allowing non-Native Americans who practice Native American religion to apply to the Repository for permits — was, in the district court’s view, less restrictive while still serving the government’s interests. Id. at 1333. The court felt that requiring FWS enforcement officers to ask for a permit whether the potential violator was Native American or not would be a de minimis burden, and it dismissed the government’s concerns regarding a black market as speculative. Id. Finally, it acknowledged that its holding would be, in essence, an “experiment” that did entail potentially unforeseen risks. Id. at 1334. The court concluded, however, that RFRA dictated the result. Id.
DISCUSSION
I. Standard of Review
In Hardman, we expressly left open the question of what standard of review would apply to the district court’s conclusion on remand regarding RFRA’s least-restrietive-means inquiry. 297 F.3d at 1130 (“[D]ue to the dearth of factual findings on the records before us, we need not address that question today.”). In a subsequent case involving RFRA and the Eagle Act, however, we held that whether the government has chosen the least restrictive means to advance its compelling interest is a question of law that we review de novo. Friday, 525 F.3d at 949 (“We now conclude, as other circuits have, that both prongs of RFRA’s strict scrutiny test are legal questions.”).
The Friday court recognized, however, that factual findings may be embedded in a district court’s least-restric-
tive-means conclusion. Id. While such findings would normally be reviewable only for clear error, see Hardman, 297 F.3d at 1120, the Friday court held that, in the RFRA context, some facts are constitutional facts subject to our “independent examination.” Friday, 525 F.3d at 949 (quoting Citizens for Peace in Space v. City of Colo. Springs, 477 F.3d 1212, 1219 (10th Cir.2007), quoting in turn Bose Corp. v. Consumers Union of U.S., Inc., 466 U.S. 485, 499, 104 S.Ct. 1949, 80 L.Ed.2d 502 (1984)). We are thus obliged “ ‘to make an independent examination of the whole record in order to make sure that the judgment does not constitute’ ” too great an intrusion on religious expression. Citizens for Peace in Space, 477 F.3d at 1219 (quoting Bose, 466 U.S. at 499, 104 S.Ct. 1949). This standard, however, “applies only to ‘constitutional facts’ and not to the basic historical facts upon which the claim is grounded, which are subject to the usual ‘clearly erroneous’ standard of review.” Friday, 525 F.3d at 950. And district court determinations of witness credibility, even in the constitutional facts context, still demand our “due regard.” Bose, 466 U.S. at 499-500,104 S.Ct. 1949.
II. RFRA Analysis
As noted above, the Hardman en banc court determined that: (1) Wilgus’ religious exercise was substantially burdened by the Eagle Act’s limitation of permits to members of federally-recognized tribes, and (2) that the government had two compelling interests in enacting its regulatory scheme. It now falls to us to consider those compelling interests, and to determine whether the existing regulatory structure is the least restrictive means of forwarding those interests.
A. Compelling Governmental Interests
Before delving into the question of least restrictive means, we must briefly revisit *1285the nature of the government’s two compelling interests at issue in this case, to ensure that we view the evidence produced by the government on remand through the proper lens. As discussed in Hardman, we identified two compelling governmental interests that might support the burden imposed by the Eagle Act on Wilgus’ religious exercise: the interest in protecting eagles and the interest in fostering and preserving Native American religion and culture. 297 F.3d at 1127-29.
The first interest is straightforward: the government has a compelling interest in protecting the bald eagle as our national symbol, and the golden eagle, as its survival and the survival of the bald eagle are intimately intertwined. The removal of the bald eagle from the list of species protected under the Endangered Species Act does not render this interest a nullity; we observed in Hardman that “whether there [are] 100 eagles or 100,000 eagles,” the government’s interest in protecting them remains compelling. 297 F.3d at 1128. We did note, however, that a more robust eagle population could be factored into the least restrictive means analysis. Thus, the fewer eagles there are in the wild, the more restrictive a regulatory scheme might be and still pass muster under RFRA, and vice versa. Id. (“What might change depending on the number of birds existing is the scope of a program that we would accept as being narrowly tailored as the least restrictive means of achieving [the] interest.”).
The second governmental interest — protecting and fostering Native American culture and religion — is considerably more complex. At the outset, while the Hardman court did not precisely define the contours of this interest, the opinion’s language strongly suggests that the interest found compelling arises from the federal government’s obligations, springing from history and from the text of the Constitution, to federally-recognized Indian tribes. For example, the very first time this interest is mentioned in the Hardman opinion, the court noted that “[a]long with Congress’s power to ‘regulate Commerce ... with the Indian Tribes’ [in article I, section 8 of the Constitution] comes an obligation of trust to protect the rights and interests of federally recognized tribes and to promote their self-determination.” 297 F.3d at 1128 (emphasis added).
The Hardman opinion also mentioned a second aspect to this governmental interest, “preserving Native American culture in-and-of-[itself],” upon which the government could adduce evidence on remand. Id. at 1129. This could be read as a general warrant to the government to protect “Native American culture” as a thing separate and apart from those Indian tribes to whom the government owes a trust obligation. In response to a question from Judge Murphy in concurrence regarding whether the Hardman majority was in fact endorsing such an interest, the majority responded, “[w]e agree that simply guaranteeing that Native American cultural practices are somehow a part of our modern culture is too cramped a view of the compelling interest, and we wish to dispose of any notion that such a view is advanced by this opinion, or by Supreme Court jurisprudence.” Id. at 1134 n. 23.
Based on our examination of Hardman and governing Supreme Court precedent, we conclude that it is the former formulation of this second governmental interest— protection of the culture of federally-recognized Indian tribes — that controls in this case. We so conclude for a number of reasons. First, the federally-recognized tribes formulation of the interest is consistent with the language of Hardman. The first time this interest appears in the Hardman opinion, it is prefaced by an acknowledgement of Congress’ “obligation *1286of trust to protect the rights and interests of federally-recognized tribes and to promote their self-determination.” Id. at 1128 (emphasis added). Based on this power — the power to protect federally-recognized tribes — the Hardman court had “little trouble finding a compelling interest in protecting Indian cultures from extinction, growing from government’s historical obligation to respect Native American sovereignty and to protect Native American culture.” Id. (quotation omitted). But, again, that compelling interest, in the Hardman court’s own words, arose from the government’s power to protect federally-recognized tribes. To be sure, the opinion does refer to the other formulation of the interest — “preserving Native American culture and religion in-and-of-themselves” — but it does not foreclose the government relying upon either version. Id. at 1129. Thus we elect to focus on the federally-recognized-tribes facet of the interest, as consistent with Hardman.
Second, the federally-recognized-tribes formulation of the interest is more consistent with the Eagle Act itself. The language of the statute instructs the Secretary of the Interior to issue permits for possession of eagle parts — when it is “compatible with the preservation” of the species — “for the religious purposes of Indian tribes.” 16 U.S.C. § 668a (emphasis added). Congress could easily have worded the section differently; the exception could have read, for example, “for the purposes of Native American religion.” Such wording would have been a clear indication to courts that Congress saw itself as protecting and promoting Native American religion per se, rather than as expressed in the religion of federally-recognized tribal members. But Congress specifically chose to tie the exception to “Indian tribes,” rather than individual practitioners. From this we infer that Congress saw the statutory exception not as protecting Native American religion qua religion, but rather as working to preserve the culture and religion of federally-recognized tribes. We will thus evaluate the statute under the same interest that the statutory language leads us to believe Congress relied upon in passing it.
Third, this formulation of the governmental interest is consistent with the Supreme Court’s longstanding interpretation of the federal government’s relationship with Native American tribes. In Morton v. Mancari, 417 U.S. 535, 537-39, 94 S.Ct. 2474, 41 L.Ed.2d 290 (1974), the Supreme Court considered an equal protection attack on a provision of the 1934 Indian Reorganization Act that gave Native Americans preference for employment in the Bureau of Indian Affairs. The Court began by noting that Congress has “plenary power” to legislate concerning the tribes, springing from both the Indian Commerce Clause of Article I of the Constitution, U.S. Const, art. I, § 8, cl. 3, and the treaty power of Article II, id. art. II, § 2, cl. 2. Morton, 417 U.S. at 551-52, 94 S.Ct. 2474. The Court noted that, as a consequence of the forcible seizure of Indian lands by the United States, “the United States assumed the duty of furnishing ... protection [to the Native Americans], and with it the authority to do all that was required to perform that obligation.” Id. at 552, 94 S.Ct. 2474 (quoting Bd. of County Comm’rs v. Seber, 318 U.S. 705, 715, 63 S.Ct. 920, 87 L.Ed. 1094 (1943)).
Pursuant to this obligation to the tribes, Congress was empowered to “single out for special treatment a constituency of tribal Indians.” Id. (emphasis added). The employment preference before the Court did not draw a distinction based on race, but rather one based on political constituency: “The preference, as applied, is granted to Indians not as a discrete racial group, but, rather, as members of quasi-sovereign tribal entities.” Id. at 554, *128794 S.Ct. 2474. The preference was “political rather than racial, in nature.” Id. at 553 n. 24, 94 S.Ct. 2474. Thus the preference did not offend the equal protection component of the Due Process Clause of the Fifth Amendment. The Court has returned to this justification — that tribes are political units — to justify enactments treating Native American tribes differently than other groups. See, e.g., United States v. Antelope, 430 U.S. 641, 645-47, 97 S.Ct. 1395, 51 L.Ed.2d 701 (1977) (upholding Major Crimes Act provisions relating to Native Americans); Fisher v. Dist. Ct., 424 U.S. 382, 390-91, 96 S.Ct. 943, 47 L.Ed.2d 106 (1976) (per curiam) (affirming tribal court jurisdiction over adoption proceedings); cf. Rice v. Cayetano, 528 U.S. 495, 518-20, 120 S.Ct. 1044, 145 L.Ed.2d 1007 (2000) (reaffirming the core holding of Morton, but holding it inapplicable to Fifteenth Amendment challenges to state voting legislation). See generally Cohen’s Handbook of Federal Indian Law § 14.03[2][b] (Nell Jessup Newton et al. eds., 2005) [hereinafter Cohen ].
Similarly, here, even though we are not considering an equal protection or due process attack on the Eagle Act, we note that the language of the exception to the possession ban in the Eagle Act refers specifically to “the religious purposes of Indian tribes.” 16 U.S.C. § 668a. The Act thus draws a distinction between Native Americans and non-Native Americans based on the “quasi-sovereign” status of the tribes. Cf. Morton, 417 U.S. at 554, 94 S.Ct. 2474. Morton itself characterized Congress’ power over Indian affairs in terms of the tribes: “Resolution of the instant issue turns on ... the plenary power of Congress, based on a history of treaties and the assumption of a ‘guardian-ward’ status, to legislate on behalf of federally recognized Indian tribes.” Id. at 551, 94 S.Ct. 2474 (emphasis added). Thus, by adopting the federally-recognized tribes version of the compelling governmental interest in this case, we situate ourselves in the very heartland of federal power, as recognized by the Supreme Court in its Morton line of cases.
Fourth and finally, if we were instead to adopt the general “protection of Native American religion” iteration of the compelling governmental interest, we believe that we would risk running afoul of the Establishment Clause of the First Amendment to the Constitution. The Establishment Clause commands that “Congress shall make no law respecting an establishment of religion.” U.S. Const, amend. I. The Supreme Court has repeatedly held that the “First Amendment mandates governmental neutrality between religion and religion, and between religion and nonreligion.” Epperson v. Arkansas, 393 U.S. 97, 104, 89 S.Ct. 266, 21 L.Ed.2d 228 (1968). “When the government acts with the ostensible and predominant purpose of advancing religion, it violates that central Establishment Clause value of official religious neutrality, there being no neutrality when the government’s ostensible object is to take sides.” McCreary County, Ky. v. Am. Civil Liberties Union of Ky., 545 U.S. 844, 860, 125 S.Ct. 2722, 162 L.Ed.2d 729 (2005). If we were to hold that the federal government has a compelling interest in fostering Native American culture generally by providing special exceptions to criminal laws for Native American religious practices, we are concerned this might run up against this principle.
By adopting the federally-recognized tribes version of the interest, however, we remain on safe ground, based on the Supreme Court’s conclusion that federally-recognized tribes are political — rather than religious or racial — in nature. See Morton, 417 U.S. at 554, 94 S.Ct. 2474. As long as the federal government takes action toward federally-recognized tribes as *1288political entities, supported by the specific provisions of the Constitution that grant Congress exceptional powers vis-á-vis such tribes, the government can avoid unconstitutionally favoring one religion over another.5 See Cohen § 14.03[2][c][iii] (noting that equal protection-based claims by non-Indians challenging religious exemptions for Indians have been rejected under the Morton reasoning).
On the whole, then, the federally-reeognized-tribes version of the compelling governmental interest is a better fit with Hardman, Supreme Court precedent, the Eagle Act itself, and the Supreme Court’s Establishment Clause jurisprudence. We will therefore evaluate the validity of the Eagle Act under RFRA with reference to two compelling governmental interests: protecting bald and golden eagles, and fostering the culture and religion of federally-recognized Indian tribes.
B. Least Restrictive Means
Now that we have a clear view of the government’s compelling interests in this case, we turn to the question of whether the Eagle Act permitting structure is the means of forwarding those interests that is least restrictive of Wilgus’ religious rights. Before we tackle that question, however, we must first precisely define the nature of the least restrictive means inquiry. The legislative history of RFRA reveals that, in answering the least restrictive means question, courts are to “look to free exercise cases decided prior to Smith.” Religious Freedom Restoration Act of 1993, S. Rep. 103-111 (1993), reprinted in 1993 U.S.C.C.A.N. 1892, 1898. Further, in Sherbert, while the Supreme Court did not use the words “least restrictive means,” it did observe that the government in that case would have to “demonstrate that no alternative forms of regulation would [serve the compelling interest] without infringing First Amendment rights.” 374 U.S. at 407, 83 S.Ct. 1790. And in Hardman, we pointed out that “ ‘least restrictive means’ is a severe form of the more commonly used ‘narrowly tailored’ test.” 297 F.3d at 1130.
But the notion that the government must prove that it has used the “least restrictive means,” or that “no alternative forms of regulation” would suffice to serve its interests, is an odd creature. On its face, it requires the government to prove a negative — -that no matter how long one were to sit and think about the question, one could never come up with an alternative regulation that adequately serves the compelling interest while imposing a lesser burden on religion. This is a formidable task. While the government is sometimes required to prove a negative, see, e.g., United States v. Prentiss, 206 F.3d 960, 971 (10th Cir.2000) (noting that some federal criminal statutes, such as immigration laws forbidding re-entry into the United States without permission, require the *1289government to prove a negative), courts have recognized that to do so is inherently difficult, see Elkins v. United States, 364 U.S. 206, 218, 80 S.Ct. 1437, 4 L.Ed.2d 1669 (1960) (“[A]s a practical matter it is never easy to prove a negative.”). In the abstract, such a thing can never be proven conclusively; the ingenuity of the human mind, especially if freed from the practical constraints of policymaking and politics, is infinite.
Thus, a number of courts that have considered the least restrictive means question in the context of incarcerated worshippers have held that the government should not be required “to refute every conceivable option in order to satisfy the least restrictive means prong of RFRA.” Hamilton v. Schriro, 74 F.3d 1545, 1556 (8th Cir.1996) (characterizing such a requirement as a “herculean burden”); accord Fowler v. Crawford, 534 F.3d 931, 940 (8th Cir.2008) (considering the identical language of the Religious Land Use and Institutionalized Persons Act (“RLUIPA”)); Spratt v. R.I. Dep’t of Corr., 482 F.3d 33, 41 n. 11 (1st Cir.2007) (RLUIPA claim); May v. Baldwin, 109 F.3d 557, 563 (9th Cir.1997) (RFRA claim). Not requiring the government to do the impossible — refute each and every conceivable alternative regulation scheme — ensures that scrutiny of federal laws under RFRA is not “strict in theory, but fatal in fact.” Fullilove v. Klutznick, 448 U.S. 448, 507, 100 S.Ct. 2758, 65 L.Ed.2d 902 (1980) (Powell, J., concurring) (addressing strict scrutiny under the Equal Protection Clause of the Constitution). Thus the government’s burden is two-fold: it must support its choice of regulation, and it must refute the alternative schemes offered by the challenger, but it must do both through the evidence presented in the record.
The theoretical difficulty inherent in the least-restrictive-means inquiry is doubly dangerous to an appellate court sitting in review of a cold record. The task of deciding whether a particular regulatory framework is the least restrictive — -out of all conceivable — means of achieving a goal virtually begs a judge to go on a fishing expedition in his or her own mind without tethering the inquiry to the evidence in the record. It is incumbent upon us, therefore, to limit ourselves to consideration of the alternative regulation schemes proffered by the parties, and supported in the record. A statute that asks whether a regulation is the least restrictive means of achieving an end is not an open-ended invitation to the judicial imagination.
At the same time, however, we have an obligation to ensure that the record supports the conclusion that the government’s chosen method of regulation is least restrictive and that none of the proffered alternative schemes would be less restrictive while still satisfactorily advancing the compelling governmental interests. Therefore, we review the evidence adduced by the district court as part of our de novo consideration of whether the Eagle Act regulations are the least restrictive means of forwarding the government’s two compelling interests.
C. The Existing Regulatory Scheme Does Balance and Advance Both of the Government’s Compelling Interests in the Least Restrictive Manner and Neither of the Alternatives Proposed Satisfy the Government’s Compelling Interest in Striking an Appropriate Balance Between its Two Compelling Interests of Protecting Bald Eagles and Preserving the Culture and Religion of Native American Tribes
Unfortunately, the task before us is not as simple as identifying the compel*1290ling governmental interests and determining how well Congress’ chosen means serve those interests. The two compelling governmental interests in this case — eagle protection and providing for the religions needs of members of federally-recognized tribes — are, to at least some degree, in conflict.6 As the Hardman court noted, “there is a great deal of interplay between the two compelling interests and ... the ultimate test is whether the balance between the two interests struck by the government is achieved by the least restrictive means available.” 297 F.3d at 1134. Therefore, we must determine whether the possession ban exception for members of federally-recognized tribes adequately balances the two compelling interests against one another, and whether any of the proffered alternative schemes would also adequately strike such a balance, but would do so in a manner less restrictive of Wilgus’ religious exercise.
The district court considered two alternative means that the government could have chosen that might have been less restrictive of Wilgus’ religious rights than the current regime: opening the Repository permitting process to all sincere practitioners of Native American religion, regardless of them tribal membership or lack thereof, and allowing tribal members who lawfully possess eagle parts to give those parts as gifts to non-tribal-members who are nevertheless sincere practitioners of Native American religion. Wilgus proffered the latter option, while the district court presented the former.7 Our task, as outlined by the Hardman court, is to decide, based on the evidence adduced in the district court, whether either of these two proffered means adequately balance the government’s two competing interests in a manner less restrictive to Wilgus’ religious needs, and whether the evidence presented supports the government’s chosen regulatory scheme as least restrictive.
*1291Before examining the three regulatory-options before us, however, we should briefly restate the factual predicates underlying this examination, as developed on remand. First, while the numbers of bald and golden eagles in the wild is on the rise — as the delisting of the bald eagle from the Endangered Species Act list attests — this rise in eagle populations is not likely to translate into an equivalent increase in the numbers of eagle feathers and parts available for distribution from the Repository. Based on the evidence before it, the district court concluded that “more eagles in the wild will not immediately translate into more feathers available for distribution at the Repository.” Wilgus, 606 F.Supp.2d at 1322. The vast majority of deceased birds turned in to the Repository die from electrocution or collision with a vehicle. Id. at 1323. But those are the very causes of eagle mortality that conservationists have managed to reduce in recent years, leading in part to the rebound in eagle populations. Id. And while the Repository staff has made successful “efforts to ensure that those who find eagle carcasses know where to send them,” those very efforts to raise awareness have also led to an increase in applications for eagle parts from eligible Native Americans. Id. Taken together, this evidence shows “that there is no significant untapped source of birds not already being sent to the Repository.” Id. (quotation omitted). We must take the current level of supply of eagle parts, then, as a given.
Second, the demand for eagle feathers and parts by tribal members for their religious practices already greatly outstrips the supply available through the Repository. As noted above, in 2004 the Repository received many more requests for eagle parts than it could accommodate. Even with the current restriction of permits to members of federally-recognized tribes, the waiting period to receive a whole bald eagle is two and one-half years; for a whole golden eagle it is three-and-a-half to four years; and a tribal worshipper seeking a whole immature golden eagle must wait four to four-and-one-half years. Id. at 1321. Loose feathers can be provided more quickly (though still after a wait of between three and six months), but in the words of the district court, “the Repository system is already vulnerable to any significant increase in demand.” Id.
Third, the district court found it extremely difficult to estimate the number of non-Native American adherents of Native American religions that might be eager to sign up for permits. The government attempted to prove the number via “a combination of census data and data from the General Social Survey (G.S.S.),” but given that neither measure is designed to capture the kind of data necessary to answer the question at issue here, the district court was unconvinced by the estimates provided by the government’s expert sociologist. Id. at 1319, 1329. Another government witness testified that “no one has studied this question [of the number of non-Native American adherents of Native American religions] using a reliable statistical methodology.” Id. at 1328 (quotation omitted). The district court also considered evidence of the numbers of practitioners of Afro-Caribbean religions who might also require eagle feathers, and noted that there may be as many as one million of such practitioners in the United States.
Fourth, testimony by FWS agents before the district court revealed that there is a thriving black market for eagle feathers and parts. FWS Special Agent Kevin Ellis testified that the black market is driven by three factors. First, many individuals take part in the so-called “powwow circuit,” in which cash prizes are offered for the best performers of traditional Native American dances. The traditional garb for these dances frequently includes *1292eagle feathers, and “persons who kill eagles ... for their feathers, approach winners of pow-wows and attempt to sell them feathers as they know cash for purchases is readily available.” (Apl’t App. at 87 [Affidavit of Kevin Ellis].) Second, “[m]any Native Americans who would normally obtain feathers from legal sources ... turn to the black market due to frustration in obtaining feathers legally (i.e. short supply and/or the multi year waiting list at the USFWS Eagle Repository).” (Id.) Third, the black market is driven by “collectors (mostly non Native Americans) who wish to own Native American cultural items.” (Id.) Such collectors “are often very well organized, and funded, and use ‘runners’ who regularly visit Reservations to seek out items to purchase for many thousands of dollars.” (Id. at 88.)
Finally, the district court heard testimony regarding some of the enforcement difficulties facing FWS agents. Eagle feathers that come legally from the Repository cannot be marked as legitimate, because to do so would destroy their religious value. Wilgus, 606 F.Supp.2d at 1324. Therefore, one of the few tools FWS has at its disposal to distinguish between lawful and unlawful possession is the distinction between members and non-members of federally-recognized tribes.8
Option 1: Opening the Permitting Process to All Adherents of Native American Religion
The district court held that the government could broaden permit eligibility to all sincere adherents of Native American religion, and thus alleviate the burden on Wilgus’ religious exercise. As a threshold matter, this solution would unquestionably lessen the burden upon Wilgus’ religious exercise, because he would be as able as any other follower of Native American faiths to apply to the Repository for eagle parts. It goes almost without saying, though, that he would then be subject to the same delays in receiving eagle feathers from the Repository that members of federally-recognized tribes must currently endure. The manner in which he obtained his feathers — gifts from Native American religious practitioners — would still be prohibited under the Eagle Act.
The question, thus, is how well this alternative forwards the government’s two compelling interests. It is self-evident that the net effect of this change would be to increase the number of permit applications received by the Repository. We are sympathetic with the difficulty faced by the district court in arriving at a number for how many additional applicants, situated similarly to Wilgus, might be waiting out there to apply for permits. But at the very least, Wilgus and his prior co-defendants Saenz and Hardman would have applied for permits, and we have no difficulty inferring from the evidence presented on remand that a non-trivial number of additional applicants would appear, if the rules were changed.
This increase in applications would likely not affect the government’s compelling interest in eagle protection. After all, the Eagle Act itself does not allow the Secretary of the Interior to issue any possession permits unless to do so would be “compatible with the preservation of the bald eagle or the golden eagle.” 16 U.S.C. § 668a. Thus, no matter how many religious prac*1293titioners apply for permits, and no matter how long the waiting list for eagle parts becomes, the number of permits granted may never, by law, rise to a level that threatens the preservation of the species. It is true that, as wait times at the Repository grow, more Native Americans may opt out of the legitimate system and turn to the black market, but we see demand for eagle feathers to be essentially a zero-sum game. Some tribal members may turn to the black market, but at the same time some non-members who would have relied upon the black market would turn to the Repository and the permitting system. And without hard data, we cannot say conclusively whether the one group would outnumber the other. Therefore, under the text of § 668a, we do not feel that this option would impact the governmental interest in protecting eagles.
However, the second governmental interest — the fostering and protection of the culture and religion of federally-recognized tribes — could be drastically impacted if we were to adopt the district court’s alternative. Members of federally-recognized tribes already face long wait times for eagle parts from the Repository, and adding applicants in Wilgus’ position would make those wait times longer. Thus, the district court’s alternative would in fact harm the very population that § 668a was designed to help. And it would do so in order to help non-tribal practitioners of Native American religion, a group not encompassed within the compelling governmental interest supporting the Eagle Act. Therefore, this option would not forward the government’s compelling interest in protecting the religion and culture of federally-recognized Indian tribes.9
This option also has the additional disadvantage of presenting significantly greater enforcement problems than the current regulatory scheme. Under the district court’s proposed amendment to the regulatory scheme, if an FWS agent catches a suspect in possession of eagle feathers, that suspect can then claim that he is a sincere follower of Native American religion. The unlucky FWS agent would then be cast in the role of “religion cop,” and would be forced to decide whether the suspect is being truthful about his religious beliefs, or is in fact a black marketeer using Native American religion as a smokescreen. “At present,” the district court noted, “[FWS] agents can be fairly certain that a person who is not Native American who possesses eagle feathers does so illegally.” Wilgus, 606 F.Supp.2d at 1324-25. The district court’s option would eliminate even that admittedly imperfect gauge of the legality of an individual’s possession, and replace it with a religious litmus test that it is highly unlikely rational FWS agents would even attempt to apply. (See Apl’t App. at 63 [Affidavit of FWS Special Agent Lucinda Schroeder] (“I have never questioned the religious claim made for the use of eagle feathers due to the great difficulty in ever successfully doing so.”).)
On balance, then, we hold that the district court’s preferred alternative regula*1294tion scheme — opening the permitting process to all sincere practitioners of Native American religion — would fail to advance the government’s compelling interest in preserving the religion and culture of federally-recognized Indian tribes. It would, in fact, do that interest a disservice. We therefore reject it.
Option 2: Allowing Members of Tribes to Give Feathers to Non-Members Who Practice Native American Religion
Next, Wilgus urges that we craft an exception to the Eagle Act’s possession ban that would, in essence, be tailored specifically to his situation, and would allow non-Native Americans such as himself who practice Native American religion to possess eagle feathers when given them as gifts from tribal members with permits. He argues that such transfer should be regarded as one under 50 C.F.R. § 22.22(b)(1), which allows feathers to be “handed down from generation to generation or from one Indian to another in accordance with tribal or religious customs.” As noted above, Wilgus has conceded that he is not an adopted member of the Paiute tribe, and so his alternative would require us to decide that a transfer “in accordance with tribal or religious customs” could encompass transfer to a nonNative American. Like the previous option, this would entirely remove the burden upon Wilgus’ religious exercise imposed by the Eagle Act. After all, Wilgus received at least one of the feathers at issue here from a member of a federally-recognized tribe as part of his religious training, and the government does not contest the sincerity of Wilgus’ religious beliefs. Therefore, the validity of this option also turns on how well it advances and protects the government’s compelling interests.
As with the first option, we cannot see that this scheme would harm or help the government’s interest in eagle protection. No more permits would be issued than under current law, and even the demand-shifting effects noted above seem less likely to occur under this more narrow exception. Once again, though, the problem lies in the governmental interest in protecting the religion and culture of federally-recognized tribes. The individuals who would gain by this change in law, such as Wilgus himself, are by definition those individuals not encompassed by the governmental interest. And allowing feathers to be disseminated legally beyond the federally-recognized tribes would, by necessity, reduce the number of feathers available to those tribes, given the sharply limited overall supply of feathers. This effect could well be exacerbated by the demand pressures exerted by the black market. The FWS already believes that feathers that are lawfully dispensed via permits from the Repository sometimes enter the black market. Adopting Wilgus’ proposed possession loophole for those who legitimately receive feathers as gifts will increase pressure on tribal members who lawfully possess feathers to “gift” them into the black market, once black marketeers become aware of the new easily-invoked “sincere practitioner” shield. As the district court recognized, such a scheme would “create well-nigh insuperable obstacles to enforcement of any restrictions on possession of eagle feathers.” 10 Wilgus, 606 F.Supp.2d at 1333. *1295Thus, again, it seems likely that feathers and eagle parts will be diverted away from members of federally-recognized tribes, the very people that the governmental interest protects.
Wilgus’ proffered alternative regulatory scheme, thus, fails to advance the government’s compelling interest in protecting and fostering the religion and culture of federally-recognized Indian tribes. We therefore reject it as well.
Option 3: The Existing Regulatory Scheme
Finally, we consider the government’s currently existing regulatory framework. As acknowledged by the government and noted by the Hardman court, the limitation of permits to members of federally-recognized tribes does substantially burden Wilgus’ religious exercise. We hold that it does so, however, as the least restrictive means of forwarding the government’s two compelling interests.
As to eagle protection, the existing regulatory framework protects eagles via the “compatible with the preservation” language of 16 U.S.C. § 668a. We are left, then, once again with the compelling governmental interest in protecting and preserving the religion and culture of federally-recognized Indian tribes. And it should be obvious by now how the existing permit regulations do so. By allowing only members of federally-recognized tribes an essential though otherwise prohibited commodity (eagle feathers and parts), the United States ensures that those tribes are able to continue to practice their traditional culture to the greatest extent possible. And by limiting the permitting process to only members of those recognized tribes, the United States does its best to guarantee that those tribes, which share a unique and constitutionally-protected relationship with the federal government, will receive as much of a very scarce resource (eagle feathers and parts) as possible.
Thus, we hold that the existing scheme for issuing eagle feather possession permits and enforcing the Eagle Act is the least restrictive means of forwarding the government’s compelling interests. We would also be remiss if we failed to note that, in upholding the Eagle Act provisions against a RFRA challenge by non-Native American practitioners of Native American religion, we join the two other circuit courts to have considered the question. See United States v. Antoine, 318 F.3d 919, 924 (9th Cir.2003); Gibson v. Babbitt, 223 F.3d 1256, 1258 (11th Cir.2000) (per curiam).11
CONCLUSION
While RFRA embodies a crucial policy severely limiting the government’s ability *1296to interfere with the ability of citizens to worship as they please, the Eagle Act represents equally important, opposing principles of stewardship, both- of natural resources and of the tribal religious cultures to which the United States has unique obligations. The government thus faces the unenviable task of navigating between this particular Scylla and Charybdis.
We are sensitive to the sincerity of Wilgus’ religious beliefs, and we do not question either the authenticity or the weight of his religious experience among Native Americans. We recognize that this litigation has now been pending for more than a decade, and that both sides have put forward many meritorious arguments and reams of evidence. The district court performed yeoman’s service in sorting through that evidence in an attempt to determine whether the Eagle Act permitting requirements are the least restrictive means of forwarding the government’s dual compelling interests. The district court concluded in the negative, and we respect the work that went into it, but we cannot agree. We are convinced that, in light of the options before the federal government, the regulations at issue are the least restrictive means available to advance its compelling interests. We therefore REVERSE the conclusion of the district court to the contrary and hold that Wilgus’ conviction did not violate RFRA.