HAMILTON, Circuit Judge.
Jason Austin was convicted of conspiracy to distribute heroin. He appeals only his sentence, challenging the quantity of drugs attributable to him and the court’s finding that he played a leading role in the organization. In particular, he contends that a key witness has been so thoroughly discredited by alleged misstatements and contradictions that it was error for the court to credit his testimony at least in part. We disagree. We also reject Austin’s contention that
Alleyne v. United States,
-U.S.-, 133 S.Ct. 2151, 186 L.Ed.2d 314 (2013), prohibited the judge from basing the Sentencing Guideline calculations on a higher drug quantity than that for which Austin was convicted. We therefore affirm Austin’s sentence.
I.
Factual and Procedural Background
In 2008, defendant Austin was the leader of a street-level, retail drug-trafficking operation. Working primarily from his po
sition at the intersection of Kedzie Avenue and Ohio Street (“Kedzie-Ohio”) in Chicago, Austin received heroin and crack cocaine from suppliers and sold the drugs to users, with the help of many others. At the ground level of the operation were “'pack workers,” who sold the drugs directly to users seven days a week for over twelve hours a day. Money was then passed up to “managers” or “bundle runners,” who in exchange distributed drugs to the pack workers for sales. Drug proceeds were then finally passed from the managers up to Austin as the leader of the Kedzie-Ohio operation.
A double murder early in the morning on August 13, 2008 figures prominently in Austin’s arguments on appeal. Off-duty Chicago Police Detective Robert Soto and Kathryn Romberg were shot and killed while sitting in Soto’s car on the west side of Chicago. The investigation into their murders focused on Austin and his operation. A number of witnesses suggested that Austin was responsible for the murders. He was arrested a few days after the murders, but the case quickly unraveled. Witnesses began to recant or contradict their previous statements. The state blamed the changes on threats and witness-tampering. Austin attributes the changes to statements that had been coerced by the police. Whatever the reasons, within a month the murder case against Austin was dismissed.
' The Federal. Bureau of Investigation and Chicago Police Department then began what became a two-year investigation into Austin and the Kedzie-Ohio operation. The investigation included numerous undercover drug purchases, extensive surveillance, and interviews with informants and co-conspirators. At the end of the investigation in November 2010, over 100 people were arrested. Austin was indicted on federal charges relating to his role in the drug-trafficking conspiracy, particularly for his participation from January 2008 through June 2010. A jury found Austin guilty of five counts of drug distribution and one count of conspiracy to distribute, but only for an amount of less than 100 grams of heroin.
The sentencing hearing took several days. The testimony ranged. from the details of the drug-trafficking operation to Austin’s criminal background and the murders of Detective Soto and Ms. Romberg. A key government witness was Jeffrey Scott, who worked from 2008 to 2009 as a manager and bundle runner for the Ked-zie-Ohio operation and reported to Austin.
Scott’s testimony at trial and during sentencing provided a detailed picture of the Kedzie-Ohio operation. Particularly critical was Scott’s testimony at Austin’s trial regarding the total drug sales of the conspiracy. The district judge found this testimony credible, and it informed much of Austin’s final sentence. Scott testified that the organization’s sales rose from $800 a day in April 2008 to $4000 a day in May, and then higher still in June, July, and August. During that latter period, the Kedzie-Ohio operation experienced a spike in sales to $8000 a day after a man overdosed near the sales location. (Addicts apparently reasoned that heroin strong enough to cause an overdose was a particularly good product.) After Austin’s arrest on August 16, 2008, others took over the operation temporarily and sales tapered off to approximately $1600 a day. No later than May 2009, though, Austin resumed leadership of the Kedzie-Ohio operation.
Scott also provided testimony as to Austin’s alleged role in the murders of Detective Soto and Ms. Romberg. Austin challenges Scott’s credibility as a witness in large part because of this testimony, so it bears review even though it did not direct
ly affect the sentence in this drug ease. Scott testified that in 2008 a disagreement arose between Austin and a rival dealer known as “Quick,” who had gone so far as to shoot at Austin on one occasion. Austin then asked Scott to bring him a gun he had previously given Scott for safekeeping. In the early morning of the Soto and Rom-berg murders, Scott was awoken by his brother Terrance. Jittery and nervous, Terrance told Jeffrey Scott that Austin had just shot Quick and his girlfriend. Later that day, Scott claimed, Austin confided that he had done something terrible: the people he had thought were Quick and his girlfriend turned out to be “a cop and a lady.” Scott understood Austin to be admitting to the murders. Austin insisted that Scott help keep the incident quiet. Police quickly tracked Scott down, however, and interviewed him at the police station.
Scott did not initially share with police what his brother had told him. But after watching a video of his brother confess to his role in the shootings, Scott told the police what his brother had told him the night of the murders. A few days later, after Austin’s arrest, Austin’s brother Charles approached Scott, angry that Scott had implicated Austin in the murders. Charles tried to choke Scott, leading to a brief scuffle. Charles left Scott with instructions to fix the situation. He continued to threaten Scott in the days to come. Finally, Charles ordered Scott to recant his statement to the police and to claim that his previous statement implicating Austin was made as a result of beatings by police. Scott claimed that he worked with Austin’s lawyer to prepare a statement — at points being instructed to “exaggerate it and ... make it seem[ ] like it was torture.” Scott signed the statement, which he later claimed he knew to be false.
The district judge’s explanation of Austin’s sentence took into account the extensive testimony available, including Scott’s conflicting statements, to determine Austin’s proper offense level. Judge Lefkow’s written decision contained a detailed explanation for her estimate of the drug quantity sold by Austin’s operation. From the beginning of April to mid-May 2008, the judge credited Scott’s trial testimony of $800 in sales per day for the organization, yielding 0.7 kilograms of heroin sold for that period.
For mid-May to June 2008, again crediting Scott’s testimony, the judge estimated that sales were approximately $4000 per day, yielding 3.6 kilograms for that period. For July through the August murders, daily sales were estimated to have had a peak of $8000 and a conservative $5000 daily average, as supported by Scott and testimony by other conspirators, Kevin Terry, Jr. and Troy Davis. This yielded 4.5 kilograms for that period.
Even though there was evidence that Austin may have continued to sell drugs from August 2008 to May 2009, the judge, relying in part on Scott’s testimony, conservatively held Austin responsible for no heroin sales during this period. According to Scott, the arrests in August 2008 led to lower sales under new manager Kenneth Terry. After Terry was arrested later that year on heroin charges, the Kedzie-Ohio operation moved in December 2008 or January 2009 to a new location under the management of Kevin Terry, Jr. In May 2009, Scott testified, Jason Austin decided to move the operation back to Kedzie and Ohio. Adopting well-known
marketing tactics from legitimate businesses, Austin attracted customers during the move by offering a “pass-out,” giving free heroin to buyers to promote the operation’s return. Video recordings showed that Austin may have been involved with drug sales before he resumed leadership in May 2009. Given the testimony of Austin and Scott to the contrary, however, the judge took the conservative approach of attributing no drug quantity to Austin for those months.
For the remainder of 2009 until the supply ran dry in August, the judge used Scott’s testimony in conjunction with other government evidence to estimate 1.6 kilograms sold. Based on these conservative estimates, the judge found that Austin was responsible for 10.7 kilograms. The sum of 0.7 + 3.6 + 4.5 + 0 + 1.6 is 10.4 kilograms, not 10.7, but this minor discrepancy does not affect the court’s finding that Austin was responsible for over 10 kilograms of heroin and therefore subject to a base offénse level of 36 under the 2008 Sentencing Guidelines.
Other findings raised Austin’s guideline offense level. The court found: that Austin had possessed a firearm in furtherance of the conspiracy, adding two levels; that he was a leader of the conspiracy, adding four more; and that minors had been used in furtherance of the drug conspiracy, adding two more levels, to 44, meaning that the Guidelines’ highest level of 43 applied. Due to two previous convictions, one for delivery of a controlled substance and one for aggravated battery, Austin was considered a career offender with a criminal history of Category VI. Offense level-43 and criminal history Category VI is as high as the Sentencing Guidelines go, recommending a sentence of life in prison, though none of the offenses of conviction authorized a life sentence. In the end, the district court sentenced Austin to a term of thirty-five years in prison, including thirty years for the conspiracy conviction, followed by ten years of supervised release.
II.
Analysis
On appeal Austin argues that his sentence was based on two errors in calculating the guideline offense level: drug quantity and aggravating role. Austin argues there were two errors in calculating the drug quantity. First, he insists that the district judge incorrectly determined the drug quantity attributable to him by erroneously relying on the testimony of Jeffrey Scott. Second, Austin argues that the judge could not lawfully have found him responsible for a higher drug quantity than the quantity for which he was convicted. Finally, Austin argues that in determining his role as a leader or organizer of the conspiracy, the court erred again by relying on Scott’s testimony. We find no error.
A.
Drug Quantity
To apply the Sentencing Guidelines, the district court determined that Austin was responsible for more than 10 kilograms of heroin. See U.S.S.G. § 2D1.1(c). A district court’s calculation of drug quantity is a factual determination that we review only for clear error.
United States v. Clark,
538 F.3d 803, 812 (7th Cir.2008). At sentencing, the government must prove by a preponderance of the evidence the quantity of drugs attributable to a defendant.
United States v. Longstreet,
669 F.3d 834, 836 (7th Cir.2012), citing
United States v. Krasinski,
545 F.3d 546, 551 (7th Cir.2008). In a drug conspiracy, a defendant is responsible under U.S.S.G. § 1B1.3 for “all reasonably foreseeable acts and omissions of others in furtherance of the jointly undertaken criminal activity.”
United States v. Soto-Pie-
dra,
525 F.3d 527, 531 (7th Cir.2008). Accordingly, each conspirator is responsible for both the drug quantities directly attributable to him and amounts involved in reasonably foreseeable dealings by co-conspirators.
United States v. Turner,
604 F.3d 381, 385 (7th Cir.2010), quoting
United States v. Acosta,
534 F.3d 574, 585 (7th Cir.2008).
A district court may base its sentence only on information with “sufficient indicia of reliability to support its probable accuracy.” U.S.S.G. § 6A1.3(a). But as we have often said, determining drug quantities attributable to defendants is “not an exact science.”
United States v. Sewell,
780 F.3d 839, 849 (7th Cir.2015). A district court may make reasonable estimates of drug quantity based on evidence in the record.
Id.,
citing
Acosta,
534 F.3d at 582.
1.
Scott’s Credibility
Relying upon co-conspirator testimony, video records, and evidence of Austin’s violent behavior toward rival drug sellers, the district court determined that Austin controlled the drug sales occurring at Ked-zie and Ohio for much of 2008 and 2009. Austin therefore knew that drug sales were occurring at that site in furtherance of the conspiracy, making him responsible for the quantity of all drugs sold by his co-conspirators during that time. The district judge said that determining “an exact number is impossible,” but she concluded that Austin was responsible for more than ten kilograms of heroin sold at the location during his time in control. In doing so, the judge drew upon evidence of controlled buys, seizures, and co-conspirator testimony and admissions. On appeal, Austin challenges whether the testimony of one of these co-conspirators — Jeffrey Scott — was credible. Austin argues that Scott is simply so untrustworthy that the judge could not reasonably have credited his testimony.
“Determining witness credibility is especially within the province of the district court and ‘can virtually never be clear error.’ ”
Longstreet,
669 F.3d at 837, quoting
United States v. Clark,
538 F.3d at 813; see also
United States v. Johnson,
342 F.3d 731, 735 (7th Cir.2003) (granting “exceptional deference to a sentencing judge’s credibility determinations”). The district court is best situated to make credibility determinations in light of the totality of the evidence, including the witness’s statements and behavior, other witness statements, and further corroborating or contrary evidence. See
United States v. Contreras,
249 F.3d 595, 602 (7th Cir.2001).
A district court may rely upon ambiguous testimony to estimate drug quantity. See, e.g.,
United States v. Cross,
430 F.3d 406, 410 (7th Cir.2005). A district court may even consider testimony that is “totally uncorroborated and comes from an admitted liar, convicted felon, or large scale drug-dealing, paid government informant,”
United States v. Clark,
538 F.3d at 813, quoting
United States v. White,
360 F.3d 718, 720 (7th Cir.2004), as long as the court evaluates the evidence carefully.
Austin first argues that Scott’s testimony as to the drug quantity was so inconsistent and so thoroughly unreliable that crediting it was clear error despite the deferential appellate review of credibility findings. His argument is based on Scott’s trial testimony about dividing the money from heroin sales. Each bundle of heroin sold at retail brought in $960. Austin characterizes Scott’s testimony as follows: the supplier received $700 per bundle (leaving $260), then Austin received one out of every three bundles as payment,
for an average of $320 per bundle (leaving a shortage of $60), then the storehouse received an average of $192 per bundle (raising the shortage to $252), then the runners and pack workers received $100 and $160 per bundle, respectively (leaving a shortfall of $512 per bundle). Thus, Austin contends, because Scott’s testimony about the money was mathematically impossible, the judge clearly erred by relying on his testimony to estimate sales volume.
The accounting in Scott’s trial testimony was in fact quite possible. Austin mischar-acterizes Scott’s statements. According to Scott’s testimony, every bundle of heroin contained 8 packs, and each pack contained 12 individual bags of heroin. A total bundle would return $960 in proceeds. Pack workers would keep $20 per pack sold, or $160 per bundle, and pass the “rest of the money” — the remaining $800 — to bundle runners like Jeffrey Scott. Scott testified that he would then receive $100 for every bundle sold for his work as a bundle runner, leaving $700 in remaining • proceeds to distribute. Finally, Scott would turn over the remainder of the -proceeds to
either
Austin, as the leader of the operation, or Kenny Bell, the supplier. Bell would be paid twice as much as Austin. Bell would keep the proceeds from ten bundles ($7000), and then Austin would keep the proceeds from the next five bundles ($3500), and so on. Averaged per bundle then, pack workers would receive $160, bundle runners $100, Bell about $467 per bundle ($7000 / 15), and Austin about $233 per bundle ($3500 / 15), totaling $960.
Austin’s interpretation of Scott’s testimony arrives at an impossible result by adopting the less probable version of the order of payment. Under his understanding, Austin would have been paid his one-third share before the pack workers or bundle runners were paid. If Austin had really taken the entire share one third of the time, there would have been no money left for other workers. Scott’s testimony, however, also suggests a more reasonable interpretation: that Austin and Bell were paid after the lower-level members had already been paid, making the payment division mathematically possible. The judge did not clearly err by deeming credible Scott’s testimony on these points.
Moreover, even if Austin were correct that Scott’s testimony on how profits were divided was impossible, it was not clear error to find Scott credible on his testimony on overall sales figures. In determining the reliability of Scott’s testimony concerning gross sales volume, and therefore drug quantity, the judge made clear that she found him “credible in this respect.” She went on to note that she would specifically address Jeffrey Scott’s credibility further and later compared Scott’s testimony to other witness testimony to test its reliability. The judge made a narrow credibility determination on the question of sales figures. Her quantity estimates were quite conservative and not clearly erroneous.
Austin also challenges Scott’s drug quantity testimony by pointing to inconsistencies in his statements regarding the murders of Detective Soto and Ms. Rom-berg. Austin focuses on Scott’s testimony about the make and color of the vehicle driven by the shooter. Scott’s account was different from his brother’s, which was different in turn from a dying declaration by Detective Soto. Because all of these accounts of the vehicle cannot be true, and because Scott’s testimony regarding a number of other issues differed from that of other witnesses, Austin concludes that Scott’s testimony is so unreliable that it was clear error to credit his testimony on drug quantities.
To reiterate, though, the law gives the sentencing judge the .responsibility to de
termine the credibility of a testifying witness on a particular issue, even in light of previous misconduct or misstatements. It is not unusual for witnesses in drug cases to have all sorts of credibility problems. And any witness may have been unable earlier to remember details accurately, had reasons to bias his testimony, or just outright lied, but his testimony on a separate drug quantity issue may still be reasonably deemed credible. See, e.g.,
United States v. Clark,
538 F.3d at 812-13. The judge carefully considered Scott’s credibility and limited her credibility determination to the drug quantity issue, as was proper. Again, we find no clear error. Also, the judge’s guideline "calculations did not depend at all on whether Austin played any role in the murders of Detective Soto and Romberg. Austin earned the highest level of the Guidelines based solely on the drug conspiracy and his criminal history.
2.
Acquitted Conduct
The fact that the jury found Austin guilty of conspiring to distribute no more than 100 grams of heroin did not bar the court from finding, for guideline purposes, that Austin was responsible for more than 10 kilograms of heroin. It is well established that a sentencing court may treat conduct for which the defendant has been acquitted as relevant conduct for guideline purposes.
United States v. Watts,
519 U.S. 148, 154, 117 S.Ct. 633, 136 L.Ed.2d 554 (1997);
United States v. Gonzalez,
765 F.3d 732, 738 (7th Cir.2014). An acquittal means that a charge has not been proved beyond a reasonable doubt, but a sentencing court may rely on facts established by a preponderance of the evidence.
Gonzalez,
765 F.3d at 738, quoting
United States v. Home,
474 F.3d 1004, 1006 (7th Cir.2007). For purposes of calculating drug quantity, the offense level is determined by the amount of drugs attributable to the defendant during his entire course of relevant conduct, “not simply the amount associated with the particular offenses of conviction.”
United States v. Redmond,
667 F.3d 863, 875 (7th Cir.2012).
The Supreme Court’s decision in
Alleyne v. United States,
570 U.S.-, 133 S.Ct. 2151, 186 L.Ed.2d 314 (2013), did not change these. principles of guideline sentencing.
Alleyne
held that facts that increase mandatory minimum sentences are elements of a crime and therefore must be admitted or found beyond a reasonable doubt by a jury.
Id.
at 2158. But the Court’s opinion was, by its own terms, narrow.
Alleyne
continued to endorse broad sentencing discretion,’ noting that such discretion is not unlawful even if its exercise depends on facts found by the judge.
Id.
at 2163. Accordingly, while statutory drug quantity issues affecting maximum and minimum sentences must be submitted to a jury, the scope of the narcotics conspiracy and drug responsibility under the advisory Sentencing Guidelines is subject to judicial fact-finding during sentencing. In the wake of
Alleyne,
we have repeatedly considered the type of challenge Austin is making here, and we have repeatedly held that
Alleyne
does not apply to such challenges to drug quantity determinations under the Sentencing
Guidelines. See, e.g.,
United States v. Garrett,
757 F.3d 560, 574-75 (7th Cir.2014);
United States v. Hernandez,
731 F.3d 666, 672 (7th Cir.2013).
Austin finally argues that the district court erred by stepping outside the bounds of U.S.S.G. § 1B1.3, which instructs sentencing courts to consider conduct that was part of the offense of conviction or the same course of conduct or common scheme or plan. Because Austin’s conviction was for conspiracy to distribute less than 100 grams of heroin, Austin contends, it is impossible for him to be sentenced on the basis of responsibility for more than 100 grams. Sections 1B1.3 and 3D1.2 have been interpreted together, however, to provide “that a district court
must
increase a defendant’s base offense level to account for ‘relevant conduct,’ which includes drugs from any acts that ‘were part of the same course of conduct or common scheme or plan’ as the convicted offense, regardless of whether the defendant was charged with or convicted of carrying out those acts.”
United States v. Duarte,
950 F.2d 1255, 1263 (7th Cir.1991) (emphasis added), quoting
United States v. Franklin,
902 F.2d 501, 504 (7th Cir.1990). The judge did not err by determining for purposes of the Guidelines that Austin was responsible for more than 10 kilograms of heroin.
B.
Leader or Organizer of Conspiracy
Finally, the district court also did not err in finding that Austin had earned a four-level aggravating role enhancement under U.S.S.G. § 3Bl.l(a). According to Austin, the district court “cherry-picked” among unreliable and inaccurate testimony about his role in the conspiracy. This objection was not raised during sentencing, however, and the district judge reasonably described Austin’s leadership role as “undisputed.” We therefore review this sentencing finding only for plain error, which is a demanding standard.
United States v. Butler, 111
F.3d 382, 387-88 (7th Cir.2015) (failure to object during sentencing to guideline calculation forfeited issue and warranted plain-error review). To demonstrate plain error, the defendant must show: “(1) an error or defect that (2) is clear or obvious and (3) affects the defendant’s substantial rights.”
Id.
Even then, the appellate court has discretion to correct the error if it seriously affected the fairness, integrity, or public reputation of the judicial proceedings, but it need not do so.
Id.
A defendant is subject to a four-level aggravating role enhancement where he was the organizer or leader of a criminal activity that involved five or more participants or was otherwise extensive. U.S.S.G. § 3Bl.l(a). In determining a defendant’s role in a criminal organization, the district court should consider a number of factors, including “the nature of the defendant’s participation in the offense, his claimed right to a larger share of the fruits of the crime, his degree of participation in planning or organizing the offense, and his degree of control and authority exercised over other participants.”
United States v. Reynolds,
714 F.3d 1039, 1043 (7th Cir.2013).
There is no dispute that the Ked-zie-Ohio organization involved more than five persons. And there is ample evidence that Austin was the leader of the organization under § 3Bl.l(a). Numerous co-conspirators identified Austin in their plea agreements, statements to police, and testimony as maintaining control of the organization. The judge specifically determined this testimony to be credible. Further, Austin received a larger share of the drug profits than any other member of the organization. He used threats of violence against competitors and rivals in
the drug-trafficking business to defend the operation. He had decision-making authority over the organization, determining the operation’s direction on strategy and marketing. In the face of such substantial evidence, Austin’s protests that Jeffrey Scott’s testimony is unreliable fall far short of showing plain error.
In the absence of some compelling evidence showing decisively that the aggravating role enhancement was wrong, this plain-error challenge is particularly weak. Without an objection from Austin on the issue, the government had no reason to waste the court’s time building an even more extensive record on the issue. A defendant cannot show plain error by merely pointing to possible gaps in evidence on an issue that was not disputed in the trial court.
For the foregoing reasons, Austin’s sentence is AFFIRMED.