dissenting:
The majority opinion resolves this case by adopting a new rule: “when deciding whether an exception to CAFA removal applies, ... [i]f the applicability of an exception is not shown with reasonable certainty, federal jurisdiction should be retained.” The law of our circuit does not require such a presumption in favor of federal jurisdiction, nor should it. Moreover, if the facts of this case do not satisfy the local controversy exception “with reasonable certainty,” then the majority’s presumption is a strong one indeed. Because I agree with the district court’s determination that the local controversy exception required remand of this case to state court, I respectfully dissent.
Among other limitations, CAFA’s local controversy exception applies only when “greater than two-thirds of the members of all proposed plaintiff classes in the aggregate are citizens of the State in which the action was originally filed.” 28 U.S.C. § 1332(d)(4)(A)(i)(I). Arbuckle, the party seeking remand, bears the burden to “prove the statutory citizenship requirement by a preponderance of the evidence.” Preston v. Tenet Healthsystem Mem’l Med. Ctr., Inc. (Preston I), 485 F.3d 793, 797 (5th Cir.2007). As the majority opinion explains, whether the local controversy exception is satisfied depends on how the proposed class is defined. Under the “narrow class definition” (current mineral interest owners), Arbuckle has adduced sufficient evidence that greater than two-thirds of the proposed class members are citizens of Texas, where the case was originally filed.1 Under the “broad class definition” (current and former mineral interest owners since 2004), it has not.
In discerning the proposed class, we must be guided by two principles. First, “the application of the local controversy exception depends on the pleadings at the time the class action is removed.” Cedar Lodge Plantation, L.L.C. v. CSHV Fairway View I, L.L.C., 768 F.3d 425, 426 (5th Cir.2014). Second, when CAFA jurisdiction depends on an analysis of state court pleadings, we read the pleadings as the state court would read them. See Brand v. Transport Serv. Co. of Ill., 445 F.3d 801, 803 (5th Cir.2006) (holding, for purposes of CAFA’s requirement that a lawsuit have been “commenced” in state court on or after CAFA’s enactment date, that “when an action is commenced in state court is determined based on the state’s own rules of procedure,” not Federal Rule of Civil Procedure 3).
Because state law governs our construction of the pleadings, a word on Texas’s pleading rules is in order. When a plaintiff files a petition in Texas state court, the sufficiency of that petition is not judged *344against the federal pleading standard codified in Federal Rule of Civil Procedure 8(a)(2) and expounded in Ashcroft v. Iqbal, 556 U.S. 662, 129 S.Ct. 1937, 173 L.Ed.2d 868 (2009). Rather, the petition must simply be “sufficient to give fair notice of the claim involved.” Tex.R. Civ. P. 47(a). In applying the fair notice standard, Texas courts “consider the petition in its entirety and construe it as favorably as possible” to the pleader. Ealey v. Ins. Co. of N. Am., 660 S.W.2d 50, 52 (Tex.1983). Granted, we are deciding between two interpretations of Arbuckle’s petition rather than judging its sufficiency. But our interpretive process must be informed by that backdrop. Moreover, Texas appellate courts tasked with discerning a proposed class from a trial court record do not constrain themselves to a formalistic examination of the portion of the petition that purports to define the class. See Intratex Gas Co. v. Beeson, 22 S.W.3d 398, 408 (Tex.2000) (declining to adopt an alternate class definition on appeal after rejecting the definition certified by the trial court because, even after reviewing “the pleadings and the record of the class-certification proceedings,” “the parameters of the proposed new class are not easily identified”). With that context in mind, our task is to examine the pleadings as a whole and decide between the two definitions of the proposed class. The district court, after briefing and a hearing, concluded that “the totality of the pleading makes it clear that [Arbuckle was] talking about current owners.” I agree.
The strongest evidence for the narrow class definition appears in paragraph 14 of Arbuckle’s petition: “Plaintiff and all class members are currently mineral interest owners (and in almost all instances also owners of the surface estate) in Johnson and Tarrant Counties, Texas.” (emphasis added). That provision unambiguously limits the proposed class to current owners. The only evidence for the broad class definition appears in paragraph 23 and is much less clear:
Plaintiff seeks and requests the certification of a class ... comprised of the following: All non-excluded persons or entities, in Johnson and Tarrant Counties, Texas, who are, or were, since 2001, purchasers of property, including mineral interests, at foreclosure, of valid, pri- or-recorded mortgages on properties, or owners who took title by, through or under such a purchaser, from which Chesapeake and Total produced gas and other minerals....
(emphasis added). Focusing on the use of both present and past tense, defendants-appellants and the majority conclude that paragraph 23 proposes a class encompassing both current and former mineral interest owners.
That reading of paragraph 23 is not obvious. The “are, or were, since 2004” language immediately precedes a description of “purchasers,” not owners, and reading the paragraph to describe persons who “are, or were, since 2004, ... owners,” as defendants-appellants would do,2 requires breezing through five commas in addition to the ones I have reproduced. The majority opinion does not undertake that splicing exercise but nevertheless concludes that both current and former owners are covered. Under the majority’s reading of paragraph 23, the group of “[a]ll non-excluded persons or entities, in John'son and Tarrant Counties, Texas, who are, or were, since 2004, purchasers of property, including mineral interests,” necessari*345ly encompasses “all” non-excluded purchasers, including those who no longer own the mineral interests. But even that reading leaves awkwardness. Why does the provision frame itself around “persons or entities in Johnson and Tarrant Counties, Texas,” rather than property in those counties?3 One explanation is that paragraph 23 simply catalogs the ways that a current owner might have acquired his mineral interest: by purchasing the interest at a foreclosure sale at some point since 2004, or by acquiring it from somebody else who did so. A contorted reading, perhaps, but no reading of paragraph 23 leaves all of its parts intact. The majority opinion may not find paragraph 23 ambiguous, but I do.
Fortunately, we resolve ambiguities at the level of the petition as a whole rather than within any one paragraph. Though paragraph 23 is ambiguous when read alone, paragraph 14’s crystal-clear limitation of the class to current owners illuminates the petition’s true class definition: the narrow one. In giving effect to paragraph 14, we need attach little significance to its placement in the petition’s “facts” section rather than among the “class action allegations.” Texas pleading rules “reject[] any requirement that a sharp, intelligible, and consistent distinction be drawn between ‘evidentiary allegations,’ allegations of ‘ultimate facts,’ and allegations of ‘legal conclusions.’ ” 2 Roy W. McDonald & Elaine A. Grafton Carlson, Texas Civil Practice § 7:4(a), Westlaw (database updated Dec. 2015); see also Tex. R. Civ. P. 45(b) (“That an allegation be evidentiary or be of legal conclusion shall not be grounds for objection when fair notice to the opponent is given by the allegations as a whole.... ”). The district court chalked paragraph 23 up to “pleading error,” but we need not even go that far; paragraph 23 is ambiguous, and the petition as a whole clarifies that the proposed class includes only current mineral interest owners.
Even if the majority opinion is correct that paragraph 23 unambiguously supports the broad definition and that paragraphs 14 and 23 are consequently in direct conflict, other provisions of the petition — to which we must look in resolving any conflict — support only the narrow definition. Paragraph 40 provides that “Plaintiff,- and the Class, is the rightful owners of the mineral estates of the Properties by virtue of a valid foreclosure action.” (emphasis added). The use of the present tense implies current ownership, as the majority opinion concedes. Shortly thereafter, paragraph 42 alleges that “multiple acts of trespass by Defendants have caused, and continue to cause damage to Plaintiff and the Class.” An act of trespass could “continue to cause damage” only to a current mineral interest owner. The majority opinion dismisses paragraph 42 on the grounds that it could be read to allege ongoing harms against some class members but not others. But that is not the most natural reading, and so paragraph 42 lends at least some credence the narrow class definition.
Finally, paragraph 43 of the petition requests injunctive relief as to the entire class, which makes sense only if the class is composed entirely of current mineral interest owners. The majority opinion posits that paragraph 43 does not preclude a class composed of both current and former owners because a mismatch between the class definition and the relief sought is not “a reason to divest a federal court of *346jurisdiction,” even if the mismatch will eventually require the creation of subclasses. But that logic falters here, where federal jurisdiction depends on discerning the true class definition and the narrow option cures the mismatch, making sense of paragraph 43’s request for injunctive relief as to the entire class. In sum, examining the petition as a whole- does not reveal “competing contextual clues” that collectively shed no light on the conflict between narrow and broad class definitions in paragraphs 14 and 23. The context clues all support the narrow definition — current mineral interest owners.
The majority opinion is most troubling when it enshrines a presumption against the local controversy exception in this circuit’s precedent. Quoting our previous decision in Opelousas General Hospital Authority v. FairPay Solutions, Inc., 655 F.3d 358 (5th Cir.2011), the majority observes that “other circuits ‘recognize that the exception is intended to be narrow, with all doubts resolved in favor of exercising jurisdiction over the case.’ ” It is important to note that Opelousas did not adopt, rely on, or further discuss that “all doubts” rule beyond noting that “[ojther courts” have adopted it. 655 F.3d at 360 (citing Evans v. Walter Indus., Inc., 449 F.3d 1159, 1163 (11th Cir.2006); Westerfeld v. Indep. Processing, LLC., 621 F.3d 819, 822 (8th Cir.2010)). Nevertheless, today’s majority opinion “adopts th[at] general approach,” ultimately deciding the case in accordance with “the need to resolve lingering doubts in favor of exercising federal jurisdiction.” Per the majority opinion, “[i]f the applicability of an exception is not shown with reasonable certainty, federal jurisdiction should be retained.” This is a course we should not take.
First, we have previously held that an exception to CAFA jurisdiction was satisfied despite some uncertainty. In Preston v. Tenet Healthsystem Mem. Med. Ctr., Inc. (Preston II ),4 we endorsed a district court’s “credible estimate” as to the size of the proposed class for purposes of evaluating the citizenship requirement in CAFA’s discretionary jurisdiction exception, noting that “it is unnecessary for the district court to ... determin[e] the exact class size to an empirical certainty.” 485 F.3d 804, 820-22 (5th Cir.2007); see also id. at 817 (approving district court’s “extrapolation” that affidavits of some proposed class members “were probably representative of many other proposed class members”); Preston I, 485 F.3d at 801 (holding that a district court must make a “credible estimate” as to the citizenship of proposed class members for purposes of the local controversy exception). The majority opinion’s requirement that CAFA exceptions be proven “with reasonable certainty” is difficult to reconcile with these prior holdings.
Second, of the “other courts” referenced by Opelousas that resolve “all doubts” in favor of federal jurisdiction, neither’s reasoning is persuasive. The Eleventh Circuit’s analysis contains a critical error. The Evans case reached its conclusion that “all doubts” regarding the local controversy exception should be “resolved ‘in favor of exercising jurisdiction over the case’ ” by quoting a portion of CAFA’s legislative history addressing the amount-in-contro*347versy requirement, not the local controversy exception. See 449 F.Bd at 1163 (quoting S.Rep. No. 109-14 at 42 (2005)). This is an error that we should not reproduce, regardless of the propriety of consulting the relevant portions of a statute’s legislative history. The Eighth Circuit’s analysis is also unpersuasive: the Westerfeld case makes too much of the plaintiffs evidentia-ry burden to prove the local controversy exception. 621 F.3d at 822. The local controversy exception is surely narrow, and the party seeking remand has the burden of establishing its application. But that does not mean that “all doubts”— whether evidentiary, legal, existential, etc. — must be resolved in favor of federal jurisdiction. Rather, Arbuckle’s burden is purely an evidentiary one, the standard of proof is a mere preponderance of the evidence, and the parties agree that the state court petition is the only piece of evidence relevant to discerning the proposed class. Arbuckle having produced that petition, it is our task to determine the petition’s legal significance without the aid of any doubt-resolving presumption in favor of federal jurisdiction. The Supreme Court has recognized that “no antiremoval presumption attends cases invoking CAFA,” Dart Cherokee Basin Operating Co., LLC v. Owens, - U.S. -, 135 S.Ct. 547, 554, 190 L.Ed.2d 495 (2014), but we should not go further and announce a pro-removal presumption, whether for CAFA as a whole or as to the local controversy exception.
Even if CAFA does require that courts “resolve lingering doubts” against application of the local controversy exception, considerations of federalism and comity are not jettisoned entirely in the CAFA context. See Hood ex rel. Mississippi v. JP Morgan Chase & Co., 737 F.3d 78, 84-85 (5th Cir.2013); 14AA Charles Alan Wright, et al., Federal Practice & Procedure § 3705.1, Westlaw (database updated Apr. 2015). Only after fully exhausting the judicial toolkit with which we typically confront a difficult interpretive problem should we resign to having a “doubt” that must be resolved in favor of federal jurisdiction.5 This is not a case in which we have “no basis” to pick one interpretation or the other, as the majority opinion suggests. Because the petition lends more support to the narrow class definition when read as a whole, we should adopt that definition as the district court did. Arbuckle has established the local controversy exception “with reasonable certainty,” if such certitude is indeed required. If this case nevertheless presents the sort of “lingering doubt” that must be resolved in favor of federal jurisdiction, then the majority opinion has, in effect, adopted a broad-reaching presumption against the local controversy exception that extends beyond what our court’s and the Supreme Court’s precedents require.
Reading Arbuckle’s Texas petition as a whole, I would conclude that the narrow class definition is the correct one and that Arbuckle has consequently met its burden to show that two-thirds of the proposed class members are Texas citizens so as to trigger the local controversy exception.6 I *348would affirm the district court’s order remanding this case back to state court. Accordingly, I respectfully dissent.
. Along with the majority opinion, I would hold that defendants-appellants have waived any challenge .to Arbuckle's proof of citizenship under the narrow class definition.
. Defendants-appellants advanced this reading in both of their briefs but abandoned it at oral argument.
. This language plausibly limits the proposed class to citizens of Johnson and Tarrant Counties, a class definition under which the two-thirds requirement would be met on the face of the petition. But Arbuckle does not argue that its proposed class is thus limited.
. Preston I and Preston II, though consolidated and decided by the same panel on the same day, were separate cases against different medical facilities concerning injuries suffered in connection with Hurricane Katrina. Preston I, 485 F.3d at 796 n. 1; Preston II, 485 F.3d at 809 n. 1. Though Preston II was decided under CAFA’s discretionary jurisdiction exception, § 1332(d)(3), rather than the local controversy exception, the court clarified that “the sufficiency of the evidence necessary to satisfy the citizenship requirements remains consistent” between the two. 485 F.3d at 812.
. This proposition mirrors the approach courts take when applying the familiar Chevron framework for reviewing administrative agencies’ statutory interpretations; only after "employing traditional tools of statutory construction” to determine whether "Congress had an intention on the precise question at issue” does a court defer to a reasonable agency interpretation. Chevron, U.S.A., Inc. v. Nat. Res. Defense Council, 467 U.S. 837, 843 n. 9, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984).
. The remaining requirements of the local controversy exception, save one, are uncontested on appeal. The one area of dispute is the requirement that there be an in-state defendant "whose alleged conduct forms a sig*348nificant basis for the claims asserted.” 28 U.S.C. § 1332(d)(4)(A)(i)(II)(bb). I do not analyze this issue in full because the majority does not reach it, but I would hold that defendant Total E & P USA, Inc., a Texas citizen, satisfies the "significant basis” requirement. See Williams v. Homeland Ins. Co. of N.Y., 657 F.3d 287, 291-92 (5th Cir.2011).