dissenting.
After the district court’s dismissal of this case was reversed by a divided panel in Cottier v. City of Martin, 445 F.3d 1113 (8th Cir.2006) (Cottier I), the district court on remand ruled that the City of Martin was liable for diluting the voting strength of Native Americans, in violation of § 2 of the Voting Rights Act of 1965. Cottier v. City of Martin, 466 F.Supp.2d 1175 (D.S.D.2006). As a remedy, the district court ordered the City to adopt a municipal government structure that is not authorized by South Dakota law, and to implement an at-large, cumulative voting scheme designed to result in the election of a candidate preferred by Native American voters in each election cycle. Cottier v. City of Martin, 475 F.Supp.2d 932, 936 (D.S.D.2007).
I would reverse the judgment of the district court. To the extent the plaintiffs made a sufficient showing to justify a finding of liability for vote dilution under § 2, this showing required proof that Native Americans were “sufficiently large and compact to constitute a majority in a single-member district.” Thornburg v. Gingles, 478 U.S. 30, 50, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986). If the record supports that conclusion (the so-called first Gingles precondition), then it was unnecessary for the district court to impose a remedy contrary to South Dakota law. The appropriate remedy would have been to create a district (or ward) in which Native Americans constitute a majority. The district court concluded, however, that it “could not draw a three-ward plan that contains an effective majority of Indian voters in one ward.” Cottier, 475 F.Supp.2d at 938. If this conclusion is correct, then there was no liability for vote dilution in the first place, because the first Gingles precondition was not satisfied. Accordingly, I would remand the case to the district court with directions either to revisit the finding of liability and dismiss the case if the plaintiffs have failed to meet the first Gin-gles precondition, or to implement a remedy that conforms to South Dakota law by creating a ward in which Native Americans constitute a majority.
The Supreme Court in Gingles established three preconditions to a finding of liability under § 2. The first precondition is that the minority group is large enough and geographically compact enough that it would be a majority in a single-member district. 478 U.S. at 50, 106 S.Ct. 2752. *746The Court explained that the reason for this precondition is to ensure that minority voters “possess the potential to elect representatives in the absence of the challenged structure or practice.” Id. at 50 n. 17, 106 S.Ct. 2752 (emphasis in original). If minority voters are not a sufficiently large and compact group to constitute a majority in a single-member district — “the smallest political unit from which representatives are elected” — then it follows that they have no potential to elect a representative of their choice, and there is no principled basis to find that their voting strength has been “diluted.” Id.
“To establish the first Gingles precondition, a plaintiff must demonstrate a proper and workable remedy exists.” Cottier I, 445 F.3d at 1117. In Stabler v. County of Thurston, 129 F.3d 1015 (8th Cir.1997), this court recognized that where the plaintiffs fail to present a workable remedy in which “Native Americans are geographically compact to form an effective voting majority in a single-member district,” they do not meet the first precondition of Gingles, and they fail to establish liability under § 2. Id. at 1025. Stabler thus echoed the conclusion of other circuits that “[t]he inquiries into remedy and liability ... cannot be separated,” and that “[a] district court must determine as part of the Gingles threshold inquiry whether it can fashion a permissible remedy in the particular context of the challenged system.” Nipper v. Smith, 39 F.3d 1494, 1530-31 (11th Cir.1994) (en banc); id. at 1547 (Edmondson, J., concurring in part and concurring in result); accord Sanchez v. Colorado, 97 F.3d 1303, 1311 (10th Cir.1996). As the Supreme Court put it in Growe v. Emison, 507 U.S. 25, 113 S.Ct. 1075, 122 L.Ed.2d 388 (1993), “[ujnless these points are established, there neither has been a wrong nor can be a remedy.” Id. at 40-41, 113 S.Ct. 1075.
In Cottier I, the district court found that the first precondition was satisfied, 445 F.3d at 1117, and a panel of this court, applying a clearly erroneous standard of review, id. at 1116, affirmed the district court’s finding. Id. at 1118. After this court reversed the district court’s finding that the third precondition was not established, id. at 1122, the district court on remand found that under the totality of the circumstances, the plaintiffs had established vote dilution that violated § 2. 466 F.Supp.2d at 1199. When it turned to the remedy, however, the district court effectively backtracked on its previous conclusion that the plaintiffs had established the first Gingles precondition. At that point, the court concluded that “[rjedistricting Martin’s three-ward system will not remedy the § 2 violation,” cited an expert who “could not create a three-ward plan in Martin that contained an effective majority of Indians,” and found that it “could not draw a three-ward plan that contains an effective majority of Indian voters in one ward.” 475 F.Supp.2d at 938. Once the court found in its remedial analysis that Native Americans were not geographically compact enough to form an effective voting majority in a single-member ward, Stabler dictated a corresponding finding that the first Gingles precondition was not met, and that the plaintiffs failed to establish a § 2 violation. Stabler, 129 F.3d at 1025; see also Nipper, 39 F.3d at 1533 (“The absence of an available remedy is not only relevant at the remedial stage of the litigation, but also precludes, under the totality of the circumstances inquiry, a finding of liability.”).
Rather than revisit its conclusion on liability, however, the district court proceeded to implement an unprecedented remedy — a plan that is “not a municipal government structure authorized by South Dakota law.” 475 F.Supp.2d at 938. Adopting the plaintiffs’ Plan C, the *747court ordered that the City abolish its al-dermanic wards and adopt “an at-large voting scheme using cumulative voting.” Id. at 936. Under this scheme, the City will have “six city council members, elections for which will be held in a 3-3 staggered cycle.” Id. “Pursuant to the cumulative voting scheme, each voter will receive three votes,” and “[e]ach voter can cast one, all, or any whole number of his or her votes for any one or more of the candidates.” Id. “The winners of the election will be the three candidates that receive the highest number of votes.” Id. As the district court recognized, however, South Dakota law authorizes only two municipal government structures: (1) a common council form in which the council “consists of the mayor elected at large and two aldermen elected from and by the voters of each ward of the municipality,” with staggered terms so that “two aldermen are not up for reelection in the same year” (i.e., the City’s present form), S.D. Codified Laws § 9-8-4 (emphasis added); or (2) a commission form in which “the board of commissioners shall consist of the mayor and two or four commissioners elected at large.” Id. § 9-9-1 (emphasis added). The district court imposed a hybrid scheme, not authorized by South Dakota law, under which six city council members are to be elected at large, under a cumulative voting scheme.
Neither the majority nor the district court cites any precedent in which a federal court has ordered a municipality to adopt a governmental structure that is unauthorized by state law as a remedy under § 2 of the Voting Rights Act. The district court, in its decision reviewed in Cottier I, concluded that it did not have authority to order this remedy. (Order, R. Doc. 371, at 21 n. 4). The district court cited Cane v. Worcester County, 35 F.3d 921 (4th Cir.1994), which reversed a district court’s de-cisión to adopt an at-large cumulative voting scheme that abolished residency districts, because the court failed to give due deference to contrary legislative judgments about an appropriate voting plan. Id. at 927-29; see also Cane v. Worcester County, Nos. 95-1122, 95-1688, 1995 WL 371008, at *3 (4th Cir. June 16, 1995) (directing the district court to adopt a different plan that came “closer to satisfying all of the legislative goals expressed by the County”). The district court in this case reversed course on remand, however, because it felt bound by dicta from a footnote in Cottier I, which said “it appears that plaintiffs’ third plan would be a viable option.” 445 F.3d at 1123 n. 7. The majority on this appeal provides no independent analysis, relying exclusively on the footnote in Cottier I. Ante, at 744-45. For their part, the plaintiffs assert that the district court “plainly erred” by refusing to deem viable both plaintiffs’ Plan A, which divided the city into three dual-member wards, and thus conformed to South Dakota’s common council form of government, see S.D. Codified Laws § 9-8-4, and Plan B, which divided the city into six single-member wards, but the plaintiffs maintain that the district court nonetheless had discretion to adopt Plan C. (Appellees’ Br. 46).
The Supreme Court has declared that federal courts in voting rights cases “should follow the policies of the preferences of the State, as expressed in statutory and constitutional provisions or in the ... plans proposed by the state legislature, whenever adherence to state policy does not detract from the requirements of the Federal Constitution.” White v. Weiser, 412 U.S. 783, 795, 93 S.Ct. 2348, 37 L.Ed.2d 335 (1973); see also Whitcomb v. Chavis, 403 U.S. 124, 160-61, 91 S.Ct. 1858, 29 L.Ed.2d 363 (1971) (holding that a district court erred in fashioning a plan *748that rejected state policy choices “without solid constitutional or equitable grounds for doing so”). Thus, at least two circuits have opined that district courts applying § 2 “must to the greatest extent possible give effect to the legislative policy judgments underlying the current electoral scheme or the legally unacceptable remedy offered by the legislative body.” Cane, 35 F.3d at 928; accord Harper v. City of Chi. Heights, 223 F.3d 593, 602 (7th Cir.2000); see also Nipper, 39 F.3d at 1531 (“Implicit in [the] first Gingles requirement is a limitation on the ability of a federal court to abolish a particular form of government and to use its imagination to fashion a new system.”). These observations are bolstered by the Supreme Court’s decision in Holder v. Hall, 512 U.S. 874, 114 S.Ct. 2581, 129 L.Ed.2d 687 (1994), which established that the size of a governing body is not subject to a vote dilution challenge under § 2. “Holder also confirms that, from the inception of a section 2 case, the existence of a workable remedy within the confines of the state’s system of government is critical to the success of a vote dilution claim.” Nipper, 39 F.3d at 1533.
These authorities indicate strongly that a federal court may disregard state law in remedying a violation of § 2 only as a last resort. The plaintiffs here essentially concede an abuse of discretion by urging that the district court could have adopted Plan A or Plan B as a remedy for the purported violation. If the plaintiffs have satisfied the first Gingles precondition that Native American voters constitute a geographically compact majority in a single-member ward, then a “proper and workable remedy” presumably would have been to create such a ward, and it was an abuse of discretion for the district court to impose a municipal government structure that contravenes state law, rather than a remedy that follows the policy preferences of the State.
The problem of overriding state law concerning the structure of municipal government is compounded by the district court’s imposition of a cumulative voting scheme. The plaintiffs cite no authority that cumulative voting in municipal elections is authorized by South Dakota law, and the relevant statute indicates that it is not. See S.D. Codified Laws § 9-13-21 (“If more than one member of the governing body is to be elected, the ballot shall contain instructions as to how many candidates for the governing body are to be voted for.”). As others have noted, moreover, “the effect and, proponents would argue, the strength of cumulative voting” is that it will achieve proportional representation of protected classes. Cousin v. Sundquist, 145 F.3d 818, 829 (6th Cir.1998). Because “[t]he imposition of cumulative voting is thus meant to achieve an end not contemplated in the Voting Rights Act,” id., its appropriateness as a remedy for a violation of § 2 is dubious. See id. (“[W]e feel that cumulative voting, the remedy ordered by the district court in this case, is an inappropriate remedy for a Section 2 claim ....”); Cane, 35 F.3d at 928 (holding that district court abused its discretion in adopting a cumulative voting scheme). The dicta in Cottier I did not address the tension between the specific prohibition on use of the Voting Rights Act to achieve proportional representation, 42 U.S.C. § 1973(b), and the natural effect (and perhaps design) of a cumulative voting scheme.9 The only authority offered *749by the plaintiffs in support of cumulative voting is the statement of Justice Thomas in Holder, that “nothing in our present understanding of the Voting Rights Act places a principled limit on the authority of federal courts that would prevent them from instituting a system of cumulative voting as a remedy under § 2,” 512 U.S. at 910, 114 S.Ct. 2581 (Thomas, J., concurring in judgment) — a statement that read in context can hardly be taken as a ringing endorsement of the remedy. See id. at 908-14 (observing that changes such as cumulative voting or a system using transferable votes “may seem radical departures from the electoral systems with which we are most familiar,” and “may be unwanted by the people in the several States who purposely have adopted dis-tricting systems in their electoral laws,” and urging that the Court’s interpretation of the Voting Rights Act should be systematically reexamined).
In summary, this case in several respects represents a remarkable assertion of authority by a federal court of appeals under § 2 of the Voting Rights Act:
• The panel majority in Cottier I declared clearly erroneous the district court’s refusal to give decisive weight to unreliable exit poll data and other equivocal election data, in the face of Supreme Court guidance that vote dilution claims are “peculiarly dependent upon the facts of each case,” and require “an intensely local appraisal of the design and impact of the contested electoral mechanisms.” Gingles, 478 U.S. at 79, 106 S.Ct. 2752 (internal quotations omitted).
• The majority opinion on this appeal affirms a determination of liability under § 2, despite the district court’s finding that it “could not draw a three-ward plan that contains an effective majority of Indian voters in one ward,” 475 F.Supp.2d at 938, thus generating an intra-circuit conflict with Stabler, 129 F.3d at 1025, and an inter-circuit conflict with cases holding that the first Gingles precondition means that the inquiries into remedy and liability cannot be separated. See Sanchez, 97 F.3d at 1311; Nipper, 39 F.3d at 1530-31.
• The majority opinion on this appeal, applying dicta from Cottier I, upholds as a remedy under § 2 the imposition of a municipal government structure that is not authorized by state law, and becomes the only published authority to approve the imposition of cumulative voting as a remedy for a violation of § 2.
I respectfully dissent.
. Cottier I distinguished the Fourth Circuit's decision in Cane on the ground that the plan in that case "completely eliminated elected positions,” 445 F.3d at 1123 n. 7, presumably by abolishing residency districts and "imposing a plan that would permit all five seats on the Board to be filled by individuals living in the same general area of the County.” Cane, *74935 F.3d at 929. The remedy adopted by the district court in this case, however, similarly abolishes the wards in Martin, and provides that all six council members will be elected at large.