987 F.3d 1248
10th Cir.
2021

UNITED STATES OF AMERICA, Plaintiff - Appellee, v. SHAUN J. SALAZAR, Defendant - Appellant.

No. 19-3217

United States Court of Appeals for the Tenth Circuit

February 16, 2021

PUBLISH. Appeal from the United States District Court for the District of Kansas (D.C. No. 2:08-CR-20084-CM-1). Before HOLMES, SEYMOUR, and MORITZ, Circuit Judges.

Daniel T. Hansmeier, Appellate Chief (Melody Brannon, Federal Public Defender, with him on the briefs), Kansas Federal Public Defender, Kansas City, Kansas, for Defendant-Appellant.

John M. Pellettieri, Attorney, Appellate Section, Criminal Division, Department of Justice, Washington, D.C. (Stephen R. McAllister, United States Attorney, James A. Brown, Assistant United States Attorney, Kansas City, Kansas; Brian A. Benczkowski, Assistant Attorney General, John P. Cronan, Deputy Assistant Attorney General, Criminal Division, Department of Justice, Washington, D.C., with him on the brief), for Plaintiff-Appellee.

MORITZ, Circuit Judge.

Shaun Salazar appeals the district court’s order revoking his term of supervised release and sentencing him to ten months’ imprisonment. He argues that his ten-month prison sentence is illegal because—when combined with his prior 115-month prison term—it exceeds the 120-month statutory maximum for his crime of conviction. We previously rejected this argument in United States v. Robinson, 62 F.3d 1282, where we held “that [18 U.S.C.] § 3583 authorizes the revocation of supervised release even where the resulting incarceration, when combined with the period of time the defendant has already served for his [or her] substantive offense, will exceed the maximum incarceration permissible under the substantive statute.” Id. at 1285 (quoting United States v. Purvis, 940 F.2d 1276, 1279 (9th Cir. 1991)). Because we remain bound by Robinson, we affirm.

Background

In 2010, Salazar pleaded guilty to being a felon in possession of a firearm in violation of 18 U.S.C. § 922(g)(1). Section 922(g)(1), by way of 18 U.S.C. § 924(a)(2), carries a statutory maximum of 120 months in prison. In 2011, the district court sentenced Salazar to 115 months in prison and three years of supervised release. Salazar completed his prison term and began serving his term of supervised release in May 2019. Soon after, a probation officer filed a petition to revoke Salazar’s supervised release, alleging that Salazar violated two conditions of his supervised release by committing battery against his brother and associating with a felon, his girlfriend.

At his revocation hearing, Salazar argued that any term of imprisonment resulting from the revocation of his supervised release could not exceed five months because anything greater would result in a total term of imprisonment that exceeded the 120-month statutory maximum prescribed by § 924(a)(2). The district court rejected this argument, revoked Salazar’s supervised release, and imposed ten months’ imprisonment followed by one year of supervised release.

Salazar appeals.

Analysis

I. Jurisdiction

Before addressing the merits of this appeal, we must be satisfied that we have jurisdiction. See United States v. Vera-Flores, 496 F.3d 1177, 1180 (10th Cir. 2007). Article III of the Constitution limits federal jurisdiction to “[c]ases” or “[c]ontroversies.” U.S. Const. art. III, § 2, cl. 1. In practice, this case-or-controversy requirement means that a party seeking relief must have an actual injury that is likely to be redressed by a favorable judicial decision. Vera-Flores, 496 F.3d at 1180. If a party no longer suffers from a redressable injury, the case becomes moot, and we no longer have jurisdiction. Id. Here, our review of publicly accessible Bureau of Prisons records suggested that Salazar was released from federal custody on or about November 22, 2019. We therefore ordered supplemental briefing from the parties asking whether this case—which challenges the length of Salazar’s prison sentence—is moot.

In response, both Salazar and the government argue that even though Salazar has finished serving his ten-month prison sentence, his case is not moot because he has not yet served his one-year term of supervised release.1 We agree. See Bender v. Williamsport Area Sch. Dist., 475 U.S. 534, 541 (1986) (“[E]very federal appellate court has a special obligation to ‘satisfy itself . . . of its own jurisdiction . . . ,’ even though the parties are prepared to concede it.” (quoting Mitchell v. Maurer, 293 U.S. 237, 244 (1934))).

The general rule in this circuit is that “a defendant’s unexpired term of supervised release, which could be reduced by a favorable appellate decision, is sufficient to defeat a claim of mootness.” United States v. Castro-Rocha, 323 F.3d 846, 847 n.1 (10th Cir. 2003). In Castro-Rocha, the defendant had completed his original 15-month prison sentence but continued to serve his three-year term of supervised release. Id. at 847 & n.1. Success on appeal would have decreased his sentencing range from 15–21 months to 8–14 months. Id. at 847 n.1. The court explained that because under this new sentencing range “the district court could choose to impose a term of imprisonment of less than one year, the district court could also choose to impose a lesser term of supervised release, or no term of supervised release at all.” Id. This possibility of a lesser term of supervised release was sufficient to save the case from mootness.

Notably, a reduced term of supervised release need not be a guaranteed result of success on appeal—the mere possibility of a reduced term of supervised release is enough to maintain a live controversy. In Castro-Rocha, for instance, if the defendant had been successful on appeal, the district court on remand could nevertheless have chosen the high end of the newly applicable sentencing range and imposed a 14-month prison sentence, which is more than one year (and, indeed, is only one month shorter than his original sentence). See id. In so doing, it could further have chosen to impose the same three-year term of supervised release. See id. Thus, Castro-Rocha’s success on the merits of his appeal would not guarantee him a shorter term of supervised release; such relief was certainly possible, but it remained within the district court’s discretion. See id. And that discretion was enough to maintain a live controversy. See id.; see also United States v. Montgomery, 550 F.3d 1229, 1231 n.1 (10th Cir. 2008) (concluding that “sentencing appeal [wa]s not moot because [defendant’s] unexpired term of supervised release potentially could be reduced if we were to render a ruling favorable to him on his upward departure challenge” (emphasis added)); United States v. Westover, 435 F.3d 1273, 1277 (10th Cir. 2006) (finding it “sufficient to prevent this appeal from being moot” that district court on remand could “potentially shorten[] the term [of supervised release] or eliminat[e] it altogether” (emphasis added)); cf. United States v. Fields, 823 F. App’x 587, 590 (10th Cir. 2020) (unpublished) (finding sentencing appeal moot despite unexpired supervised-release term because length of that term was mandated by statute and thus could not be shortened or eliminated on remand).

Here, although Salazar has served his prison sentence, he has not yet served his term of supervised release. And critically, a favorable appellate decision could potentially reduce his term of supervised release: If we were to grant Salazar the relief he seeks and remand for resentencing, the district court “could . . . choose to impose a lesser term of supervised release, or no term of supervised release at all.” Castro-Rocha, 323 F.3d at 847 n.1; see also § 3583(h) (providing that “[w]hen a term of supervised release is revoked and the defendant is required to serve a term of imprisonment, the court may include a requirement that the defendant be placed on a term of supervised release after imprisonment” (emphasis added)).

As the government suggests, neither United States v. Meyers, 200 F.3d 715 (10th Cir. 2000), nor Rhodes v. Judiscak, 676 F.3d 931 (10th Cir. 2012), require a different result. Meyers is not on point because it found moot an appeal by a defendant who was “out of prison, under no further terms of probation or supervised release.” 200 F.3d at 718. Thus, the defendant there had no continuing injury for purposes of Article III. See Vera-Flores, 496 F.3d at 1180–81 (explaining that defendant on supervised release satisfies Article III “because the defendant’s liberty is affected by ongoing obligations to comply with supervised release conditions and restrictions”; finding appeal moot despite unexpired term of supervised release because defendant had been deported and therefore was not subject to conditions of supervised release). In Rhodes, on the other hand, the defendant could “assert an actual injury” because “he remain[ed] subject to supervised release.” 676 F.3d at 933. But because Rhodes was a habeas case, the court concluded that the defendant’s injury was not redressable. The court explained that in this circuit, habeas courts lack jurisdiction to shorten a term of supervised release—so a favorable appellate decision could not reduce the defendant’s term of supervised release. Id. In other words, even if the appellate court were to grant relief on Rhodes’s claim of a too-long prison sentence, the district court would have no power to modify the defendant’s term of supervised release.2 See id. Such is not the case here.3

In summary, although Salazar has served his prison sentence, he has not yet served his term of supervised release. And a favorable appellate decision could potentially reduce that term of supervised release. Thus, Salazar’s case is not moot.

II. Merits

Having concluded that Salazar’s case presents a live controversy, we turn to the merits. Salazar argues that the district court imposed an illegal sentence following the revocation of his supervised release. A subsection of the supervised-release statute, § 3583(e)(3), enables a court, after finding that the defendant violated the terms of supervised release, to “revoke a term of supervised release[] and require the defendant to serve in prison all or part of the term of supervised release authorized by statute for the offense that resulted in such term of supervised release.” It also sets the maximum term of reimprisonment based on the classification of the crime of conviction. For example, here, Salazar’s crime of conviction is a class C felony, so § 3583(e)(3) limits any term of imprisonment imposed after revocation to no more than two years. See 18 U.S.C. § 3559(a)(3) (classifying felonies with maximum sentences between ten and 25 years as class C); § 924(a)(2) (establishing ten-year maximum prison sentence for being felon in possession). And Salazar’s ten-month prison sentence falls well within the two-year maximum established in § 3583(e)(3).

Nevertheless, on appeal, Salazar argues that the maximum two-year term specified in § 3583(e)(3) is eclipsed by the maximum term of imprisonment for his crime of conviction; in other words, he contends that the original term of imprisonment plus any term of reimprisonment imposed following revocation of supervised release can never exceed the maximum term permitted by the statute of conviction. Thus, he maintains that his ten-month term of imprisonment following the revocation of his supervised release is illegal because—when aggregated with his prior 115-month prison term—it exceeds the 120-month statutory maximum for his crime of conviction.

We rejected this same argument in Robinson, 62 F.3d 1282. There, proceeding under § 3583(e), the district court revoked the defendant’s supervised release and ordered him to serve one year in prison, even though the defendant had already served the statutory maximum prison sentence for his crime of conviction. Robinson, 62 F.3d at 1283–84. On appeal, we rejected the defendant’s argument that “because he had served the maximum . . . prison term provided in the statute under which he was convicted, the [district court] had no authority to impose the additional sentence for imprisonment under the supervised[-]release statute.” Id. at 1283 (citation omitted). Reasoning that “supervised release is a separate part of the original sentence,” we held “that § 3583 authorizes the revocation of supervised release even where the resulting incarceration, when combined with the period of time the defendant has already served for his substantive offense, will exceed the maximum incarceration permissible under the substantive statute.” Id. at 1285–86 (quoting Purvis, 940 F.2d at 1279).

Because Salazar seeks relief based on the same argument that we rejected in Robinson, he necessarily asks this panel to overrule Robinson. See United States v. Brooks, 751 F.3d 1204, 1209 (10th Cir. 2014). Our ability to do so is limited. A three-judge panel may overrule a precedent without en banc consideration in light of a statutory change or intervening Supreme Court precedent. See id.; United States v. Jones, 818 F.3d 1091, 1100 (10th Cir. 2016). This is true even if the intervening Supreme Court case is not directly on point: “The question . . . is not whether an intervening Supreme Court case is on all fours with our precedent, but rather whether the subsequent Supreme Court decision contradicts or invalidates our prior analysis.” Brooks, 751 F.3d at 1209–10 (emphasis omitted); see also United States v. Bettcher, 911 F.3d 1040, 1046–47 (10th Cir. 2018) (overruling decision of prior panel where “our reasoning . . . lost viability after” intervening Supreme Court precedent). Thus, we may overrule Robinson if subsequent controlling law undermined its reasoning.

Recognizing as much, Salazar asserts that we can and should overrule Robinson because of a statutory change in § 3583(e) and several intervening Supreme Court cases: Johnson v. United States, 529 U.S. 694 (2000), Apprendi v. New Jersey, 530 U.S. 466 (2000), and United States v. Haymond, 139 S. Ct. 2369 (2019). Our review is de novo. Brooks, 751 F.3d at 1209 (noting de novo review over whether to overrule precedent); United States v. LeCompte, 800 F.3d 1209, 1215 (10th Cir. 2015) (noting de novo review of “[l]egal questions relating to the revocation of supervised release”). We consider each of Salazar’s arguments in turn.

A. The 1994 Amendment to § 3583(e)(3)

Salazar first asserts that the 1994 amendment to § 3583(e)(3) justifies overruling Robinson. See Jones, 818 F.3d at 1100 (“We may depart from precedent without en banc review when an amendment to an applicable rule or statute creates a new standard.”). Before the amendment, § 3583(e)(3) permitted courts to “revoke a term of supervised release[] and require the person to serve in prison all or part of the term of supervised release.” § 3583(e)(3) (Supp. V 1993) (emphasis added). After the amendment, the statute permitted courts to “revoke a term of supervised release[] and require the defendant to serve in prison all or part of the term of supervised

release authorized by statute for the offense that resulted in such term of supervised release.” § 3583(e)(3) (1994) (emphasis added). According to Salazar, this new language “unlink[ed] reimprisonment from the term of supervised release imposed at sentencing” and instead “link[ed] it to the term of supervised release ‘authorized by statute for the offense that resulted in the term of supervised release.’” Aplt. Br. 15 (emphasis omitted) (quoting § 3583(e)(3) (1994)); see also United States v. Lamirand, 669 F.3d 1091, 1096–97 (10th Cir. 2012) (explaining that 1994 amendment changed “the reference point for determining the maximum post[]revocation terms of imprisonment” from “the originally imposed supervised-released term” to “the statute authorizing supervised-release terms”).

But as the government points out, the 1994 amendment preceded Robinson, and Salazar’s argument—that the amendment undermines Robinson’s analysis—is precluded by the language of Robinson itself. There, we specifically recognized the statutory amendment and noted that we “s[aw] no substantive difference in the language [of the amended statute] . . . that would impact on the issue submitted.” Robinson, 62 F.3d at 1284 n.2. Despite Robinson’s explicit recognition of the amendment, Salazar argues that the court in Robinson nevertheless analyzed the prior version of the statute because the court “omitt[ed] the ‘authorized by statute’ language” in its subsequent discussion. Rep. Br. 22 (quoting § 3583(e)(3) (1994)). On the contrary, we view the absence of this language as consistent with Robinson’s conclusion that the amended language made no substantive difference to its analysis.

Because Robinson based its holding on the amended statute, the 1994 amendment provides no basis to overrule Robinson. Cf. Jones, 818 F.3d at 1100 (departing from precedent based on subsequent statutory amendment).

B. Johnson, 529 U.S. 694

Next, Salazar argues that the Supreme Court’s decision in Johnson undermines our reasoning in Robinson. In Johnson, the defendant violated the terms of his supervised release, and the district court imposed a term of reimprisonment followed by another term of supervised release. 529 U.S. at 698. The defendant argued that the second term of supervised release violated the Ex Post Facto Clause because he was sentenced for his original crime of conviction before Congress enacted § 3583(h), which provides explicit authority to impose an additional term of supervised release following revocation of the initial term of supervised release and subsequent reimprisonment. Id. The Sixth Circuit found that the additional term of supervised release was not ex post facto because “revocation of supervised release was punishment for [the defendant’s] violation of the conditions of supervised release, which occurred after” Congress enacted § 3583(h). Id. at 698–99.

The Supreme Court affirmed, but for different reasons. Id. at 713. In rejecting the Sixth Circuit’s rationale, the Court pointed out that treating revocation of supervised release as punishment for the violation of the conditions of supervised release would raise (1) Sixth Amendment issues because the “violative conduct . . . need only be found by a judge under a preponderance of the evidence standard” and (2) Double Jeopardy Clause issues if a defendant’s violative conduct results in both

revocation of supervised release and independent criminal prosecution. Id. at 700. The Court noted that “[t]reating postrevocation sanctions as part of the penalty for the initial offense . . . (as most courts have done)[] avoids these difficulties.” Id. And it “therefore attribute[d] postrevocation penalties to the original conviction.”4 Id. at 701.

Citing this language, Salazar asserts that “Johnson adopted an aggregation approach to imprisonment upon revocation.” Aplt. Br. 15. In other words, Salazar argues that under Johnson’s “attribut[ion of] postrevocation penalties to the original conviction,” 529 U.S. at 701, “a term of imprisonment imposed upon revocation of supervised release aggregates with the term of imprisonment imposed for the offense of conviction” and “[t]his aggregate term can never exceed the statutory maximum term of imprisonment . . . provided for in the statute of conviction,” Rep. Br. 9. But nothing in Johnson states or even suggests that a term of imprisonment and a term of

reimprisonment must be aggregated. Indeed, as the government notes, the Court in Johnson pointed out that under § 3583(e)(3), “the gravity of the initial offense determines the maximum term of reimprisonment” without mentioning the maximum term of the statute of conviction. 529 U.S. at 708. Because Johnson did not adopt or endorse an aggregation approach, we reject Salazar’s argument that Johnson undermined the logic of Robinson.

Relatedly, Salazar contends that Robinson relies on the “now-discredited view”—discredited in Johnson, specifically—“that revocation penalties are punishments for violating supervised release.” Aplt. Br. 18; see also United States v. Collins, 859 F.3d 1207, 1216 (10th Cir. 2017) (noting that “the penalty for violating terms of supervised release ‘relate[s] to the original offense’” (alteration in original) (quoting Johnson, 529 U.S. at 701)). But Salazar mischaracterizes the relevant statement in Robinson. Robinson merely stated that “supervised release is a separate part of the original sentence.” 62 F.3d at 1286. That concept is distinct from the expressly disapproved proposition that “revocation of supervised release ‘imposes punishment’” for violating the conditions of supervised release. Johnson, 529 U.S. at 699–700 (quoting United States v. Page, 131 F.3d 1173, 1176 (6th Cir. 1997)). Instead, as the government asserts, Johnson’s statement that “postrevocation penalties relate to the original offense,” 529 U.S. at 701 (emphasis added), is compatible with Robinson’s statement that “supervised release is a separate part of the original sentence,” 62 F.3d at 1286 (emphasis added). As such, we reject Salazar’s contention that Robinson does not survive Johnson on this basis.

Our conclusion that Johnson did not undermine Robinson is further bolstered by several analogous, if not precisely on point, out-of-circuit cases cited by the government. First, in United States v. Cenna, 448 F.3d 1279 (11th Cir. 2006), the Eleventh Circuit upheld the defendant’s sentence of the statutory maximum sentence plus a term of supervised release. Cenna, 448 F.3d at 1280. In doing so, it explicitly rejected the defendant’s Johnson-based argument that her sentence violated the statutory maximum sentence for her crime of conviction “because any imprisonment given for violating supervised release would result in a greater period of incarceration than permitted by the statute of conviction.” Id. The court pointed out that “the settled law pre-Johnson was that a court may impose the maximum term of imprisonment under the statute of conviction and a term of supervised release, because supervised release is an independent part of a defendant’s sentence.” Id. (emphasis added). And in declining to alter this settled law, the court noted that the defendant had “not pointed to any case from any circuit that supports [the] argument that the reasoning in Johnson mandates a finding that her sentence is illegal.”5 Id. at 1281. Although Cenna involved a direct appeal from a defendant’s original sentence and not reimprisonment following the revocation of supervised release, its rationale applies equally here.

Notably, Cenna relied in part on United States v. Wirth, 250 F.3d 165 (2d Cir. 2001). See id. There, the Second Circuit concluded that “[n]othing in the Johnson Court’s retroactiv[ity] discussion compels us to depart from the well-settled rule that punishment for a violation of supervised release, when combined with punishment for the original offense, may exceed the statutory maximum for the underlying substantive offense.” Wirth, 250 F.3d at 170 n.3. Salazar points out that this statement was dictum because the holding in Wirth turned on the district court’s prior error in modifying rather than terminating the defendant’s term of supervised release. See id. at 167. Although we don’t disagree with Salazar’s characterization of Wirth’s footnote as dictum, the footnote nevertheless supports our reading of Johnson.6

Additionally, we note that the Third Circuit—albeit in an unpublished decision—recently affirmed a defendant’s postrevocation prison sentence even though his aggregate prison terms exceeded the statutory maximum for his crime of conviction. See United States v. Cook, 775 F. App’x 44, 48–49 (3d Cir. 2019) (unpublished). In rejecting the defendant’s Johnson-based aggregation argument, the

Cook court explained that, under Johnson, “[s]upervised release, and penalties for violating its terms, are attributable to the original offense.” Id. at 47–48. But the court added that “it does not follow that the term of supervised release (or imprisonment for violating its terms) is limited by the original offense’s maximum sentence. While supervised release attaches to the original conviction, a separate statute governs its mechanics and outlines penalties that may result when its conditions are violated.” Id. at 47. Thus, “because § 3583(e)(3)—rather than the underlying statute [for the crime of conviction]—provides the relevant limitation on revocation imprisonment, a defendant who has served the statutory maximum sentence may face additional imprisonment for violating the terms of supervised release.” Id. at 49.

These out-of-circuit cases consistently support our conclusion here: that Johnson is limited to its circumstances and does not disturb Robinson’s holding that a prison sentence following the revocation of supervised release, when combined with the prison term for the crime of conviction, may exceed the statutory maximum prison sentence for the crime of conviction.

C. Apprendi, 530 U.S. 466, and Haymond, 139 S. Ct. 2369

Next, Salazar argues that Robinson is no longer good law after Apprendi and Haymond. Because his arguments based on these two cases overlap, we begin by summarizing the cases before turning to Salazar’s arguments.

In Apprendi, the Supreme Court held that “[o]ther than the fact of a prior conviction, any fact that increases the penalty for a crime beyond the prescribed

statutory maximum must be submitted to a jury[] and proved beyond a reasonable doubt.” 530 U.S. at 490. The Supreme Court later extended this reasoning to any fact that increases a statutory minimum sentence. See Alleyne v. United States, 570 U.S. 99, 103 (2013).

In Haymond, the Supreme Court wrestled with the impact of Apprendi and Alleyne on a portion of the supervised-release statute, § 3583(k). See 139 S. Ct. at 2378–79 (plurality opinion); id. at 2385–86 (Breyer, J., concurring); id. at 2386–87 (Alito, J., dissenting). Section 3583(k) provides that if “a judge finds by a preponderance of the evidence that a defendant on supervised release committed one of several enumerated offenses,” then “the judge must impose an additional prison term of at least five years and up to life without regard to the length of the prison term authorized for the defendant’s initial crime of conviction.” Id. at 2374 (plurality opinion). The Haymond plurality accordingly reasoned that § 3583(k) violated Alleyne, explaining that “any ‘increase in a defendant’s authorized punishment contingent on the finding of a fact’ requires a jury and proof beyond a reasonable doubt ‘no matter’ what the government chooses to call the exercise.” Id. at 2379 (quoting Ring v. Arizona, 536 U.S. 584, 602 (2002)).

Justice Breyer authored a short concurrence. Id. at 2385. Because he shared the dissent’s concern about the “potentially destabilizing consequences” of “transplant[ing] the Apprendi line of cases to the supervised-release context,” he declined to expressly rely on Alleyne. Id. Nevertheless, he agreed with the plurality

that § 3583(k) was unconstitutional, concluding that the provision was “less like ordinary revocation and more like punishment for a new offense.” Id. at 2386.

Salazar first argues that Apprendi undermines Robinson. According to Salazar, Apprendi forbids an increase in a statutory maximum sentence based on judge-found facts, and Robinson permits just that: a prison sentence that exceeds the statutory maximum for the offense of conviction based on judge-found facts at a revocation hearing. But the government rightly contends that binding circuit precedent forecloses Salazar’s argument. In particular, it points out that in United States v. Cordova, 461 F.3d 1184 (10th Cir. 2006), this court explicitly held that Apprendi does not impact § 3583(e)’s application. In Cordova, we explained that “[i]t is well settled that supervised release is ‘part of the penalty for the initial offense’ and that ‘once the original sentence has been imposed in a criminal case, further proceedings with respect to that sentence [have not been] subject to Sixth Amendment protections.’” Cordova, 461 F.3d at 1186 (second alteration in original) (citation omitted) (first quoting Johnson, 529 U.S. at 700; and then quoting Work, 409 F.3d at 491); see also United States v. McIntosh, 630 F.3d 699, 702–03 (7th Cir. 2011) (stating that “Apprendi does not apply to a sentence imposed under § 3583 following the revocation of a supervised release”; rejecting argument that Seventh Circuit’s equivalent to Robinson was “no longer controlling because it was decided before Apprendi’s release”).

Salazar distinguishes Cordova on the basis that it did not involve a term of reimprisonment that, when combined with the initial term of imprisonment, would

exceed the statutory minimum for the crime of conviction. And he relies on Haymond to assert that Cordova cannot stand for the broad principle that Apprendi has no role to play in revocation hearings. Yet Haymond doesn’t offer Salazar the relief he seeks from Cordova’s holding.

The plurality in Haymond did rely on Alleyne, which is part of the Apprendi line of cases, to conclude that § 3583(k) was unconstitutional. 139 S. Ct. at 2378–79. But Justice Breyer’s concurrence refused to go so far; he agreed that § 3583(k) was unconstitutional, but he “would not transplant the Apprendi line of cases to the supervised-release context.” Id. at 2385–86. And Justice Breyer’s concurrence—the narrowest ground supporting the judgment—represents the Court’s holding. See Marks v. United States, 430 U.S. 188, 193 (1977) (explaining that narrowest ground supporting judgment provides controlling rule); Haymond, 139 S. Ct. at 2386 (Alito, J., dissenting) (stating that Justice Breyer’s concurrence contains “today’s holding”); United States v. Watters, 947 F.3d 493, 497 (8th Cir. 2020) (“Justice Breyer’s opinion is the narrower opinion[] and therefore controls.”); United States v. Ewing, 829 F. App’x 325, 329 (10th Cir. 2020) (unpublished) (noting “that Justice Breyer’s opinion controls”).

Moreover, even the plurality in Haymond explicitly disclaimed any ruling as to Apprendi’s impact on § 3583(e): “[W]e do not pass judgment one way or the other on § 3583(e)’s consistency with Apprendi.” Id. at 2382 n.7; see also id. at 2383–84

(emphasizing that its “decision [was] limited to § 3583(k)”).7 Thus, even if the plurality provided the controlling rule, Salazar’s argument would fail.

Other circuits have also rejected the argument that Haymond undermines prior holdings that Apprendi has no role to play in supervised-release proceedings. See United States v. Coston, 964 F.3d 289, 296 (4th Cir. 2020) (“[G]iven that no majority of the Supreme Court endorsed the application of Alleyne in the supervised[-]release context, we remain bound by this [c]ourt’s prior decision that it does not.”), cert. denied, 2021 WL 161125 (Jan. 19, 2021); United States v. Doka, 955 F.3d 290, 296 (2d Cir. 2020) (“Haymond did not undermine our clear precedent on the constitutionality of § 3583(e)(3).”); United States v. Eagle Chasing, 965 F.3d 647, 650–51 (8th Cir.) (concluding that Haymond did not undermine prior precedent holding Apprendi inapplicable to revocation proceedings), cert. denied, 141 S. Ct. 575 (2020); United States v. Cameron, 808 F. App’x 1020, 1021 (11th Cir.) (unpublished) (same), cert. denied, 2020 WL 7132576 (Dec. 7, 2020).

Moreover, they have even done so in the same factual circumstances presented here, where a defendant’s aggregate time in prison exceeded the statutory maximum for the crime of conviction. See United States v. Seighman, 966 F.3d 237, 244–45 (3d Cir. 2020) (explaining that “Justice Breyer’s refusal to ‘transplant the Apprendi line of cases to the supervised-release context’ forecloses” Apprendi-based aggregation argument; noting that “Justice Breyer’s opinion is consistent with our own precedent, where we have rejected” aggregation arguments (quoting Haymond, 139 S. Ct. at 2385 (Breyer, J., concurring))); United States v. Patterson, 829 F. App’x 917, 918, 920–21 (11th Cir. 2020) (unpublished) (concluding that Haymond did not undermine precedent holding Apprendi inapplicable to supervised-release proceedings, even where defendant’s total imprisonment exceeded statutory maximum for crime of conviction). We have held the same, albeit in an unpublished decision. Ewing, 829 F. App’x at 329–30 (noting that defendant failed to “present[] any binding authority holding that Apprendi applies to revocation proceedings even when, as here, the initial and post[]revocation sentences add up to a term that exceeds the statutory maximum term for the crime of conviction”). Thus, Haymond’s limited ruling about “an unusual provision” of the supervised-release statute does not impact our prior holding in Cordova that Apprendi does not apply to standard revocation proceedings under § 3583(e)—even when a defendant’s aggregate time in prison exceeds the

statutory maximum sentence for the crime of conviction. 139 S. Ct. at 2383 (plurality opinion).8

In sum, contrary to Salazar’s arguments, neither Apprendi nor Haymond represent intervening authority that undermines Robinson. Robinson opined that “supervised release is a separate part of the original sentence.” 62 F.3d at 1286 (emphasis added). And the Haymond plurality similarly explained that the “defendant receives a term of supervised release thanks to [the] initial offense, and whether that release is later revoked or sustained, it constitutes a part of the final sentence for [the] crime.” 139 S. Ct. at 2380. The differing result in Haymond arose because the provision at issue there was not part of the final sentence for the initial crime. See id. But the provision at issue here, § 3583(e)(3), ties the term of reimprisonment to the crime of conviction, and therefore the ranges for each part of a sentence—the initial sentence, supervised release, and any reimprisonment if that release is violated—are all fixed by the jury’s initial determination. As such, findings at a revocation hearing

do not “increase[] the penalty for a crime beyond the prescribed statutory maximum.” Apprendi, 530 U.S. at 490. Thus, neither Apprendi nor Haymond disturb our holding in Robinson that § 3583(e)(3) authorizes the revocation of a defendant’s supervised release and reimprisonment even if the resulting incarceration, when combined with the time the defendant already served in prison for his substantive offense, exceeds the statutory maximum term of imprisonment for the substantive offense.9

Conclusion

Because Robinson remains controlling precedent, the district court did not err in imposing a ten-month prison sentence after revoking Salazar’s term of supervised release, even though his aggregate time in prison—125 months—exceeded the 120-month statutory maximum for his original crime of conviction.

Affirmed.

Notes

1
Salazar explains that “[d]uring the pendency of this appeal, [he] has been in pretrial custody in [a] subsequent federal case.” Aplt. Supp. Br. 2. Thus, he “has been in continuous custody since June 2019.” Id. at 3. And “[b]ecause he has been in continuous custody, he has not yet begun to serve the 12-month term of supervised release imposed in this case.” Id. The government agrees.
2
Additionally, the Rhodes court concluded that the possibility of a separate district court granting discretionary relief under § 3583(e)(1) was too remote to be a collateral consequence of success in the defendant’s habeas proceeding. 676 F.3d at 935.
3
We also agree with the government that United States v. Miller, 891 F.3d 1220 (10th Cir. 2018), does not command a different result. There, citing only Meyers and Rhodes rather than Castro-Rocha, Montgomery, or Westover, the court found a sentencing appeal moot without acknowledging the defendant’s unexpired term of supervised release and without analyzing whether the district court could reduce or eliminate that term on remand. Miller, 891 F.3d at 1225, 1242. To the extent that Miller is inconsistent with Castro-Rocha, Montgomery, and Westover, “the earlier decision[s] control[].” United States v. Reese, 745 F.3d 1075, 1083 (10th Cir. 2014).
4
The Court further held that because Congress provided no clear intent to apply § 3583(h) retroactively, that subsection did not apply “and the ex post facto question does not arise.” Johnson, 529 U.S. at 701–02 (italics omitted). Instead, the Court framed the question as “whether § 3583(e)(3)”—rather than subsection (h)—“permitted imposition of supervised release following a recommitment.” Id. at 702–03. The Court answered this question affirmatively, ruling that despite the absence of an express provision allowing a court to impose a term of supervised release after revoking supervised release and imposing reimprisonment, the district court nevertheless retained such authority. Id. at 704, 712–13. In so holding, the Court expressly abrogated United States v. Rockwell, 984 F.2d 1112 (10th Cir. 1993). Johnson, 529 U.S. at 698 n.2, 712–13; see also United States v. Garfinkle, 261 F.3d 1030, 1032 (10th Cir. 2001) (recognizing that Johnson overruled Rockwell). Salazar cites Garfinkle to support his position that Johnson undermined Robinson, but Garfinkle stands only for the unremarkable proposition that the Supreme Court can expressly overrule Tenth Circuit precedent. It does not establish that Johnson also undermined Robinson.
5
Salazar contends that this statement is dictum. But the determination that the sentence was not illegal was necessary to affirm the defendant’s sentence, so it was not dictum. See Tuttle v. United States (In re Tuttle), 291 F.3d 1238, 1242 (10th Cir. 2002) (explaining that dicta are statements “not necessarily involved nor essential to determination of the case” (quoting Rohrbaugh v. Celotex Corp., 53 F.3d 1181, 1184 (10th Cir. 1995))).
6
The government also cites United States v. Work, 409 F.3d 484 (1st Cir. 2005). There, the First Circuit rejected the defendant’s argument that when a revocation of supervised release “leads to additional imprisonment above and beyond the top of the original [United States Sentencing G]uideline[s] . . . range, the facts underlying the revocation must be proven to a jury beyond a reasonable doubt.” Work, 409 F.3d at 486. In so doing, the court explained that “a federal criminal sentence [need not] be aggregated for all purposes” because a “sentence contains distinct aspects.” Id. at 489. Although Work was issued after Johnson, its rationale is consistent with Robinson. But because Work did not consider Johnson, it offers little guidance on the question we face here—whether Johnson contradicts or invalidates our analysis in Robinson.
7
The plurality did speculate that “§ 3583(e)(3) [could] turn[] out to raise Sixth Amendment issues in a small set of cases” where “combining a defendant’s initial and post[]revocation sentences issued under § 3583(e) will . . . yield a term of imprisonment that exceeds the statutory maximum term of imprisonment the jury has authorized for the original crime of conviction.” Haymond, 139 S. Ct. at 2384 (emphasis added). Salazar’s case, of course, is one such case. Justice Alito suggested in his dissent in Haymond that the plurality’s contemplative comments regarding § 3583(e) may have been “carefully crafted” to lay the groundwork for a later decision “much broader [in] scope.” 139 S. Ct. at 2386 (emphasis omitted). That may well be. But in light of the plurality’s explicit language limiting its decision to § 3583(k) and refusing to consider § 3583(e)’s consistency with Apprendi, this statement does not undermine our holding in Robinson. 139 S. Ct. at 2382 n.7, 2383. Moreover, Justice Breyer’s controlling concurrence stepped back even further, refusing to join the plurality’s reliance on Alleyne. See id. at 2385–86.
8
Salazar’s citation to United States v. Rodriguez, 945 F.3d 1245 (10th Cir. 2019), does not alter our conclusion. There, we noted that “[t]he right to a jury trial does not apply in a supervised[-]release revocation hearing where the maximum sentence ‘could not exceed the remaining balance of the term of imprisonment already authorized by the [original conviction].’” Rodriguez, 945 F.3d at 1250 n.5 (second alteration in original) (quoting Haymond, 139 S. Ct. at 2377). Not only was this statement dictum, and thus not controlling, but Rodriguez quoted Haymond out of context. In the relevant passage, the Court merely explained why parole and probation proceedings historically did not implicate the Fifth or Sixth Amendments: “[T]he prison sentence a judge or parole board could impose for a parole or probation violation normally could not exceed the remaining balance of the term of imprisonment already authorized by the jury’s verdict.” 139 S. Ct. at 2377. Thus, Rodriguez does not assist Salazar’s argument here.
9
We note that the government argues that “[e]ven if this [c]ourt were writing on a blank slate and were free to reconsider its precedent, Apprendi does not place into doubt the legality of the [ten]-month postrevocation penalty the district court imposed here.” Aplee. Br. 44. Because we have concluded that we are not “writing on a blank slate,” we will not speculate as to whether, if we were free to reconsider Robinson, we would arrive at the same conclusion.