The Sixth Amendment guarantees the right to a public trial, and “without exception all courts have held that an accused is at the very least entitled to have his friends, relatives and counsel present, no matter with what offense he may be charged.” In re Oliver, 333 U.S. 257, 271-72, 68 S.Ct. 499, 92 L.Ed. 682 (1948). We have repeatedly held that this right extends to the process of jury selection. United States v. Agosto-Vega, 617 F.3d 541, 546 (1st Cir.2010); Owens v. United States, 483 F.3d 48, 61 (1st Cir.2007). Yet when voir dire was conducted in this case, the defendants’ family members and friends were excluded from the courtroom, allegedly due to the Puerto Rico court’s “longstanding practice” of excluding the public from jury selection.
This practice, if it still exists, comes at great cost. The defendants were convicted as members of a sprawling drug conspiracy after a three-month trial. We are mindful that many days of testimony, weeks of diligent juror attention, and months of preparation led to that result. Nevertheless, the trial was doomed before it started. Because the complete exclusion of the public from jury selection infringed the defendants’ Sixth Amendment rights, we are compelled to vacate the defendants’ convictions and remand this case for a new trial. To ensure that holding a second trial will not subject the defendants to double jeopardy, we must also address their argument that the evidence presented at the first trial was insufficient to establish their guilt beyond a reasonable doubt.
I.
Background1
From 2001 to June 2008, a drug marketplace operated twenty-four hours a day, seven days a week at “La Quince,” a street in Altos de Cuba. It was quite an operation. With all the convenience of a supermarket, La Quince offered one-stop shopping for a number of different “brands” of cocaine, heroin, crack cocaine, marijuana, oxycodone and alprazolam.2 Some of the products were colorfully packaged, others had catchy names like “Godzilla” or “Bin Laden,” no doubt intended to inspire brand loyalty in discerning users. The savvy marketers at La Quince even distributed free samples of new drug batches. Perhaps in an attempt to appeal to the youth market, La Quince was located within 1,000 feet of a public school. In short, Walmart had nothing on La Quince.
Like any prosperous retailer, La Quince had a sophisticated supply chain in place to speed inventory from storage to clientele. *300Runners were 'responsible for storing the drugs and supplying them to the sellers as needed. Sellers, as the name implies, sold the drugs to the individual users and gave the proceeds to the runners, who delivered the cash to those at the top of the chain, the owners of each individual brand of drug.
It is the owners who are the subject of this appeal. In September 2008, seventy-four individuals were indicted for their involvement with the drug conspiracy centered in Altos de Cuba. Among them were the appellants: José Negrón-Sostre (“Negrón-Sostre”), Luis Rodríguez-Sostre (“Rodríguez-Sostre”), Josué Pérez-Merca-do (“Pérez-Mercado”), Ramón Maysonet-Soler (“Maysonet-Soler”), and Wilfredo Rosario-Camacho (“Rosario-Camacho”). Each of the appellants was charged with six counts: conspiracy to possess with the intent to distribute narcotics3 within 1,000 feet of a school (Count I); aiding, and abetting in the possession with the intent to distribute heroin, crack cocaine, cocaine, and marijuana (Counts II-V); and conspiracy to carry and use firearms during and in relation to drug-trafficking crimes (Count VI).
A jury trial commenced in January 2010. Three months later, the jury found the appellants guilty of Counts I through V. On Count VI, the gun charge, the jury convicted Rodríguez-Sostre, Maysonet-Soler and Rosario-Camacho, but found Negrón-Sostre and Pérez-Mercado not guilty. The appellants timely appealed, and in March 2011 they moved to supplement the record, alleging that they were denied their right to a public trial when their family members were excluded from the courtroom during jury voir dire.
On July 6, 2011, the district court held, an evidentiary hearing to determine whether the public had been excluded' from the courtroom in violation of the defendants’ Sixth Amendment rights. In December 2011, the district court issued a memorandum and findings of fact, concluding in summary that “[n]o‘ specific evidence was ever presented, '... that demonstrated that [a] supposed long standing district policy of not allowing the public into the courtroom during voir dire was ever followed in this case.”
II.
Discussion
All of the defendants argue that then-convictions should be reversed and then-cases remanded for a new trial because their Sixth Amendment rights were violated because members of the public were excluded during jury voir dire.4 Negrón-Sostre, Rodríguez-Sostre and Maysonet-Soler also challenge the sufficiency of the evidence supporting their convictions on some, but not all, of the charges.5 They make a number of other claims of error, but because they are not relevant in light *301of our ultimate ruling vacating their convictions and remanding for a new trial, we need not reach the remainder of these arguments.6
A.
The Sixth Amendment Right to a Public Trial
The failure to hold a public trial is a structural error that “infect[s] the entire trial process.” Owens, 483 F.3d at 64 (internal quotation marks omitted). Indeed, the Supreme Court has been “pris-tinely clear that the Sixth Amendment right to a public trial extends to the jury voir dire process.” Agosto-Vega, 617 F.3d at 546 (citing Presley v. Georgia, 558 U.S. 209, 213, 130 S.Ct. 721, 175 L.Ed.2d 675 (2010)). That is so because “[j]ury selection is the primary means by which a court may enforce a defendant’s right to be tried by a jury free from ethnic, racial, or political prejudice ... or predisposition about the defendant’s culpability....” Owens, 483 F.3d at 63 (first and second alterations in original) (quoting Gómez v. United States, 490 U.S. 858, 873, 109 S.Ct. 2237, 104 L.Ed.2d 923 (1989)). In Waller v. Georgia, 467 U.S. 39, 48, 104 S.Ct. 2210, 81 L.Ed.2d 31 (1984), the Supreme Court held that, to avoid committing structural error, courtroom closures must be justified by an overriding interest and tailored to be “no broader than necessary to protect that interest.” The defendants contend that the district court erred when the public was unjustifiably excluded from the courtroom during jury voir dire.
Because defendants did not object to the exclusion at the time of trial, “we review only for plain error.”7 United States v. Scott, 564 F.3d 34, 37 (1st Cir.2009). Under plain-error review, the defendants have “the burden of showing (1) that an error occurred (2) which was clear or obvious and which not only (3) affected [their] substantial rights, but also (4) seriously impaired the fairness, integrity, or public reputation of judicial proceedings.” United States v. Almonte-Nuñez, 771 F.3d 84, 89 (1st Cir.2014) (internal quotation marks omitted).
Was there Clear or Obvious Error?
In order to determine whether an error occurred, we must first determine whether, contrary to the district court’s findings, there was, in fact, a courtroom closure. The relevant events of the January 20, 2010 jury selection were the subject of an evidentiary hearing before the same judge who presided over the trial. Following this hearing, the district court essentially determined no courtroom closing had occurred — a finding the defendants challenge on appeal. We review the district court’s findings of fact for clear error.8 Bucci v. United States, 662 F.3d 18, 21 (1st Cir.2011). Under clear error review, we defer to the district court’s findings unless “the record, read as a whole, gives rise to a strong, unyielding belief that a mistake has been made.” *302 United States v. Hughes, 640 F.3d 428, 434 (1st Cir.2011) (internal quotation marks omitted). We will summarize the record of that hearing, as well as the district court’s findings of fact.
Courtroom Security Officer (“CSO”) Carlos Sierra-Medina (“Sierra-Medina”) had been a CSO for nineteen years and was on duty in the courtroom during voir dire. He testified to the regular practice of the court prior to 2010, saying “the tendency was that, because of the space and security, the family will not be allowed” to be in the courtroom during jury selection. Sierra-Medina testified that this practice was followed unless an attorney made a previous special arrangement with the judge; for instance one such arrangement in an earlier case involved bringing additional chairs into the courtroom for family members. He did not recall any change in this “tendency” after the publication of our decision in Agosto-Vega.9
Deputy U.S. Marshal Miguel Portalatin testified that he was in charge of security for the trial and that he “didn’t get any specific instructions from [the judge] as far as closing the door or leaving the door open.” He stated that he never gave any orders to keep people out of the courtroom because “we didn’t have space, so I didn’t have to tell anybody.” Although the door was not locked, and jurors exited freely when released, Portalatin did not recall seeing any family members in the courtroom during jury selection.
Defendant Rosario-Camacho’s sister, Maribel Rodriguez, testified that she attempted to enter the courtroom on the morning of jury selection and was asked by someone, “it could be security officers, marshals, standing in the door,” whether she was a juror. When she replied that she was a family member, she was told “Okay, well, family members are not allowed until we’re finished.” She testified that she told her brother’s attorney that she had not been allowed in the courtroom and he said, “Don’t worry, just wait in here, outside. It’s the procedure, family members are not allowed until we finish with the selection of the jury.”
A friend of Rosario-Camacho’s, Zuheily Otero González, testified that she was present that day, and that Sierra-Medina denied her access to the courtroom when she asked him if she could enter for jury selection. She, too, reported that she spoke with Rosario-Camacho’s attorney and he told her she could not go into the courtroom because “it was a practice carried out in Puerto Rico; that it was a process between the jury, the codefen-dants and thé attorneys.”
Defendant Rodriguez-Sostre’s wife testified that a CSO denied her entry into the courtroom on the morning of jury selection. Defendant Maysonet-Soler’s wife testified that she asked her husband’s attorney if she could enter the courtroom and was told that only the lawyers, prosecutors, judge, defendants and potential jurors were allowed inside and that “that’s something that is usual.” She recounted that she tried to peek through the window but that Sierra-Medina “told us not to do that, that that wasn’t allowed.”10
*303After the family members testified, several of the defense attorneys took the stand. Miriam Ramos-Grateroles, Pérez-Mercado’s counsel, testified that she had tried “more than 40” cases and that “it has been a practice, as far as I can remember, that during jury selection, the public, the family members, are not allowed in court.” When family members asked her if they could be present, she told them it was not allowed. She testified that she did not object because “[it] was the standard proceeding in this court.” Alexander Zeno, counsel for Maysonet-Soler, testified that he had been an attorney for ten years and that it was his understanding that “nobody from the public could come into the courtroom” during the jury selection process. Like Ramos-Grateroles, Zeno did not object because it was his understanding that this was a common practice. Ramón Ga-ray-Medina, counsel for Negrón-Sostre, •testified that he has been practicing in the District of Puerto Rico since 1989, and that “[i]t was a matter of general information” that family members were not allowed in the courtroom for voir dire. He did not object because “it was the practice in this district not to allow the family members during jury selection.” 11 Mariangela Tira-do-Vales, Rodriguez-Sostre’s attorney, testified that “in all the jury selections in criminal cases that [she had] had so far in this court,” it was “the usual process” to keep family members outside. Like the other attorneys, Tirado-Vales testified that her failure to object was not the result of a tactical decision, but reflective of her knowledge that it was the “standard operating procedure” of the court to exclude family members from jury selection. Finally, Rosario-Camacho’s attorney, Francisco Dolz Sánchez, testified that he had been practicing in Puerto Rico since 1975 and that, since that time “[everybody was used to the public being excluded during jury selection.” According to Dolz Sán-chez, in the years he’d been practicing, including his “many years” in the Federal Public Defender’s Office, it was standard operating procedure and “nobody objected to that.”
Following the hearing, the district court made the following findings of fact, pertinent to this appeal: 1) approximately seventy-five potential jurors were in attendance that day, taking up all of the available seating; 2) the deputy marshal did not request authority from the court to exclude the public, the doors were not locked, and “the public was not excluded from the courtroom due to a Court order or a determination by the deputy marshal in charge;” 3) family and friends were present in the courthouse, but no members of the public entered the courtroom and those who attempted to look through the windows in the courtroom door were told to step away from the door; 4) neither the court nor the deputy marshal ordered the courtroom closed; 5) none of the attorneys objected to the courtroom closure, and although they all testified that closing the courtroom was standard practice in the district, “[n]o specific evidence was ever presented, however, that demonstrated that this supposed long standing district policy ... was ever followed in this case.” The district court concluded by finding that
the failure of the defendants’ family members to enter the courtroom was *304due to the attorneys informing the family members that they could not enter the courtroom during the jury selection process, but not because of any Court Order or determination by the deputy marshal in charge to exclude the public. Counsel did not object precisely because there was nothing to which object [sic].
In our review of the district court’s findings, we begin by noting that the court did not specifically determine that the courtroom was not closed. Instead, it found that every seat in the courtroom was taken by potential jurors, the courtroom doors were not locked, and “the public was not excluded from the courtroom due to a Court order or a determination by the deputy marshal in charge.” In finding that neither the court nor the deputy marshals had “ordered” the courtroom to be closed, the district court sidestepped the issue of whether the courtroom had in fact been closed despite the absence of any such order. However, the issue of whether there was an actual courtroom closure is key because, “even if the courtroom was closed because of inattention by the judge, courts have expressed concern in the past where a court officer’s unauthorised closure of a courtroom impeded public access.” Owens, 483 F.3d at 63. That the courtroom closure was the result of inaction by the judge, rather than an affirmative order, is not dispositive. Id. (“Whether the closure was intentional or inadvertent is constitutionally irrelevant.”) (quoting Walton v. Briley, 361 F.3d 431, 433 (7th Cir.2004)). What matters is that the public was, barred. Id. (citing Martineau v. Perrin, 601 F.2d 1196, 1200 (1st Cir.1979), for the proposition that the Sixth Amendment is implicated when marshals lock a courtroom without authorization).
Here, the district court, without finding that the courtroom was closed, blamed the defense attorneys for “informing the family members that they could not enter.” Undeniably, the lawyers were partly at fault.12 However, it would be a misreading of the record to suggest that the lawyers were wholly to blame. Indeed, the court’s finding ignores CSO Sierra-Medina’s testimony that the general practice in Puerto Rico — in other words, the default rule— was to exclude members of the public from voir dire unless the attorneys made “arrangements” — that is, made requests in contravention of the general practice— with the judge beforehand.13 And in this case Sierra-Medina did not recall taking any actions that day that were contrary to that “tendency” to exclude the public. In that way, his testimony serves to corroborate the testimony of those family members who said that Sierra-Medina had turned them away from the door, an assertion he never denied during the hearing. His testimony makes clear that a courtroom closure occurred, accomplished not through locks and direct orders, but through the actions of a CSO familiar with the court’s regular practice.14
*305Our review of the record convinces us that the court clearly erred in not finding that a complete courtroom' closure occurred during jury selection. Moreover, the closure was attributable to court personnel at least as much as to the attorneys. The court clearly erred in finding that the attorneys were wholly responsible for the family members’ exclusion from voir dire. Having found a courtroom closure, we must now determine whether that closure, absent express authorization from the judge, can nevertheless pass constitutional muster.
Although courtroom closures may be justified in some circumstances, these closures “are to be rare and only for cause shown that outweighs the value of openness.” Owens, 488 F.3d at 61 (internal quotation marks omitted). “[C]losure of jury selection to the public for an entire day without meeting the strict requirements of Waller would violate a defendant’s right to a public trial.” Id. at 66. In Bucci, we summarized the Waller test as follows:
(1) the party seeking to close the hearing must advance an overriding interest that is likely to be prejudiced,
(2) the closure must be no broader than necessary to protect that interest,
(3) the trial court must consider reasonable alternatives to closing the proceeding, and
(4) it must make findings adequate to support the closure.
662 F.3d at 22. Given the peculiar posture of this case — where no party affirmatively sought to close the courtroom, and where the district court erroneously found that there was no closure — the Waller test was never applied. Because the courtroom was in fact closed absent the balancing of interests required by Waller, that closure was a clear and obvious error, satisfying the first two prongs of our plain error analysis.
Did the Error Affect the Defendants’ Rights and the Judicial Proceedings?
We now turn to the remaining third and fourth prongs, addressing whether the error affected the defendants’ substantial rights, and whether it “seriously impaired the fairness, integrity, or public reputation of judicial • proceedings.” Almonte-Nuñez, 771 F.3d at 89.
Although a brief, inadvertent closure may be excusable, the exclusion of the public for the entirety of voir dire without meeting the Waller test is a structural error. Agosto-Vega, 617 F.3d at 543. “The category of structural error has been reserved for a very limited class of cases” including “a total withholding of the right to counsel at trial,” and “the specter of a biased judge.” United States v. Padilla, 415 F.3d 211, 219 (1st Cir.2005) (internal quotation marks and citations omitted). Structural errors, as distinguished from trial errors, infect the entire trial process. Id. “Unlike trial rights, structural rights are ‘basic protection^] whose precise effects are unmeasurable.’ ” Owens, 483 F.3d at 64 (quoting Sullivan v. Louisiana, 508 U.S. 275, 281, 113 S.Ct. 2078, 124 L.Ed.2d 182 (1993)). Our precedent is unequivocal; structural error in the form of a denial of-the public trial right prejudices a defendant notwithstanding that the prejudice may be difficult to detect. See id. at 65. *306In Owens, we explored specific ways that such a closure may prejudice a defendant: “It is possible that jurors might have been more forthcoming about biases and past experiences if they had faced the public. It is also possible that [the parties], might have picked a more impartial jury or asked different questions with local citizenry watching.” Id. Those same concerns are at play here where the public, including the family and friends of the defendants, was excluded. Therefore, it is clear on the facts of this case that the third prong has been met.
It remains then, for us to determine whether the error affected the fairness, integrity or public reputation of the proceeding as a whole. Once again, Owens guides our analysis. We stated there that improper courtroom closure “eall[s] into question the fundamental fairness of [the] trial.” Id. “[Structural error transcends the criminal process by depriving a defendant of those basic protections [without which] a criminal trial cannot reliably serve its function as a vehicle for determination of guilt or innocence, and no criminal punishment may be regarded as fundamentally fair.” Padilla, 415 F.3d at 219 (internal quotation marks and citation omitted). Indeed, given the importance of the public trial right, it would be hard to see how the public reputation and integrity of the proceedings would not be compromised in this case.
And that conclusion is not altered by our acknowledgment of the role defense counsel contributed to the closure. Although it is disturbing to us that this practice passed without objection by those who seemingly accepted it as lawful status quo in Puerto Rico, we have no reason to believe the attorneys made a tactical decision not to object (in order to “sandbag” the court into creating a reversible error). On the contrary, it is apparent, given the testimony of Sierra-Medina and the defense attorneys, as well as the statement made by the prosecutor at sidebar (see fn. 12), that the practice of excluding the public from voir dire was alive and well in Puerto Rico long past the point when Owens made clear that it was unacceptable. It is a practice which however it got started, could only have been sustained and implemented by the court, not by defense attorneys. We reiterate, the ultimate responsibility of avoiding “even the appearance that our nation’s courtrooms are closed or inaccessible to the public” lies with the judge. Scott, 564 F.3d at 39. “We commend to the sound judgment of the district court the responsibility, in the first instance, of ensuring both openness and order, and above all, preserving the defendant’s constitutional right to a public trial.” Id. The district court’s failure in this instance to properly police the public’s access to defendants’ jury voir dire substantially impaired'the fairness of the trial proceedings. We find that the fourth prong has been met.
Summing up, the closure of the courtroom during the entirety of voir dire was a plain and obvious error that, as a structural error, affected the defendants’ substantial rights and seriously impaired the fairness, integrity, or public reputation of the proceedings. Our precedent compels us to find that the structural error in this case was plain error. Accordingly, we vacate the defendants’ convictions and remand their cases for a new trial. Agosto-Vega, 617 F.3d at 543.
B.
Sufficiency of the Evidence
Because the defendants will now have a new trial on the same charges, “to prevent an allegation that they will be subjected to double jeopardy in violation of the Fifth Amendment by reason of this retrial, it is incumbent upon us to address [their] con*307tentions that the government failed to present sufficient evidence at the first trial” to sustain their convictions. Id.
We review challenges to the sufficiency of the evidence de novo, “considering all the evidence, direct and circumstantial, in the light most favorable to' the prosecution, drawing all reasonable inferences consistent with the verdict, and avoiding credibility judgments, to determine whether a rational jury could have found the defendants] guilty beyond a reasonable doubt.” Id. at 548 (internal quotation marks omitted). “Testimony from even just one witness can support a conviction.” United States v. Alejandro-Montañez, 778 F.3d 352, 357 (1st Cir.2015) (internal quotation marks omitted).
Further Background15
During the three-month trial, the jury heard testimony from several cooperating witnesses who were involved in the drug operation, including two runners and a seller. Alfredo Sierra-García (“Sierra-García”) testified that at the age of thirteen he became a “lookout” at La Quince— a drug point where several owners sold drugs at the same location. According to Sierra-García, La Quince featured something for every drug consumer: heroin, cocaine, marijuana, and alprazolam were all on offer. He described how the drugs were packaged with different brightly-colored papers that corresponded to a variety of brand names and owners. He testified that the lookouts were necessary to prevent the losses the owners would suffer if police arrested a seller and confiscated the drugs. In that capacity, he was paid to warn all of the sellers if police were approaching.
Sierra-García described a hierarchy at La Quince consisting of owners, sellers, runners, and packagers, with the owners atop the pecking order, functioning as “bosses.” The La Quince organization also offered opportunities for advancement; Sierra-García later became a runner for Negrón-Sostre, who owned a brand of heroin known as Arco Iris. According to Sierra-García, Negrón-Sostre held two positions in the heroin department; not only was he an owner, but he also worked as a runner for Pérez-Mercado, the owner of the Regalito brand of heroin.
Sierra-García testified that drug brand ownership in La Quince was something of a family business, with Maysonet-Soler having inherited his interest from his mother. According to Sierra-García, May-soneNSoler owned the brand of cocaine known as Green, or Osito.
The jury also heard testimony from Xiomara Rosado-Pabón (“Rosado-Pa-bón”), who worked as a runner for Rodriguez-Sostre. She testified that Rodriguez-Sostre owned the Lexus brand of heroin. Jesus Robles-Santana (“Robles-Santana”), a seller, described the variety of products available at La Quince, recalling that fifteen or sixteen different brands of heroin were available there at any one time. Employee turnover was high in this cut-throat business; many of the individuals Robles-Santana identified as sellers are now deceased. Others survived and thrived. According to Robles-Santana, by 2007, Rosario-Camacho controlled all of the crack cocaine and marijuana at La Quince.
La Quince offered not only a vast selection, but convenient shopping hours as well. The drug point operated on a twenty-four hour schedule, with two shifts for the sale of cocaine, heroin and other drugs starting at 6:00 a.m. and 6:00 p.m. Shifts *308for crack cocaine, however, began at 7:00 a.m. and 7:00 p.m. These separate shifts were tailored to serve that particular market because, according to Sierra-García, crack users “would come in at six.” Rather than allow potential sales to slip through the cracks, as it were, the normal shift was extended another hour to cater to the 6:00 a.m. rush.
La Quince was a streamlined model of efficiency. During their shifts, according to Sierra-García and Rosado-Pabón, sellers were allowed to peddle brands from different owners simultaneously. Similarly, runners carried drugs for multiple owners.
There was always something new to tempt the shoppers at La Quince, and debut products received savvy marketing support. Sierra-García recalled seeing Rodríguez-Sostre, Negrón-Sostre, and Maysonet-Soler distributing samples of new drug batches at the drug point.
The management style at La Quince was similarly hands-on. Robles-Santana testified that he saw defendants Rodríguez-Sostre, Negrón-Sostre, Rosario-Camacho, and Maysonet-Soler at the drug point regularly, and that the men met and had discussions there. But even the best-managed business occasionally has personnel problems. According to Sierra-García, a dispute arose at La Quince over the sale of cocaine, leading Maysonel — Soler to shoot at Rodriguez-Sostre’s sister. Concerned that this incident could impact business, several owners held a meeting “to fix the problem.” Maysonet-Soler, Rodríguez-Sostre, and Rosario-Camacho were present at this meeting where peace was restored when the owners agreed that Rodrí-guez-Sostre would be allowed to “finish off some bundles” of his inventory at the drug point before Maysonet-Soler would be allowed to take over.
Defendants Negrón-Sostre, Rodríguez-Sostre, and Maysonet-Soler have each challenged the sufficiency of the evidence of some, but not all, of the charges against them. Because the evidence to convict each defendant was largely the same, we take defendants’ challenges count by count.16
Count I — Conspiracy to Distribute Drugs
All of the defendants were convicted of conspiracy to possess and distribute drugs in violation of 21 U.S.C. §§ 841(a)(1), 846, and 860, but only Rodrí-guez-Sostre and Maysonet-Soler have challenged the sufficiency of the evidence supporting this charge. To sustain a conviction, “the government must establish that (1) a conspiracy existed; (2) the defendants] had knowledge of the conspiracy; and (3) the defendants] knowingly and voluntarily participated in the conspiracy.” United States v. Diaz-Arias, 717 F.3d 1, 20 (1st Cir.2013) (internal quotation marks omitted). The agreement to conspire does not need to be express and its existence may be proven by circumstantial evidence. United States v. Lizardo, 445 F.3d 73, 81 (1st Cir.2006). “[E]ach coconspirator need not know of or have contact with all other members, nor must they know all of the details of the conspiracy or participate in every act in furtherance of it.” United *309 States v. Martínez-Medina, 279 F.3d 105, 113 (1st Cir.2002).
Rodríguez-Sostre and Mayso-net-Soler argue that the evidence showed multiple conspiracies, but did not support a single, over-arching conspiracy, and thus a variance existed between the evidence presented and the charge. We look to the totality of the evidence to determine whether it supports the existence of a single conspiracy. United States v. Mangual-Santiago, 562 F.3d 411, 421 (1st Cir.2009). In evaluating the evidence, we consider the following factors: “(1) the existence of a common purpose, e.g., the distribution of drugs; (2) interdependence of various elements in the overall plan; and (3) overlap among the participants.” United States v. Rivera Calderón, 578 F.3d 78, 89 (1st Cir.2009).
The first factor, common goal or purpose, is “broadly drawn.” United States v. Portela, 167 F.3d 687, 695 n. 3 (1st Cir.1999). We have previously found that having “an interest in furthering the distribution” of drugs is sufficient evidence of a common goal. Id. at 695. Here, there was ample testimonial evidence that Maysonet-Soler and Rodríguez-Sostre were “member[s] of a large drug distribution network that had the common purpose of selling drugs for profit.” Rivera Calderon, 578 F.3d at 89. Although Mayso-net-Soler owned brands of cocaine, and Rodríguez-Sostre heroin, they both shared an interest in furthering the distribution of drugs at the La Quince drug point.
We next consider interdependence. Interdependence was at the very heart of La Quince- — a highly-organized drug supermarket where owners worked cooperatively to maximize profits. “Interdependence exists where the activities of one aspect of the scheme are necessary or advantageous to the success of another aspect of the scheme.” Id. (internal quotation marks omitted). In Rivera Calderón, evidence that the participants “actively worked with each other to protect the drug points from threats,” and held meetings to discuss security at the drug point, demonstrated interdependence. Id at 90. Here, Sierra-García testified that lookouts were hired to warn all of the participants of the approach of law enforcement, and in so doing protected all of the owners from losses. There was also testimony that Robles-Santana saw Rodríguez-Sostre, May-sonet-Soler, and other owners talking at the drug point “daily.”
“Interdependency [is] also demonstrated by the various rules established by the participants in the conspiracy.” Id. La Quince ran like clockwork in twelve-hour shifts, with a separate schedule for the sale of crack cocaine. It is reasonable to infer that the shift system did not spring fully-formed by happenstance, but rather was devised by the owners for the benefit of them all.
“[K]nown interdependence makes it reasonable to speak of a tacit understanding between the distributor and others upon whose unlawful acts the distributor knows his own success likely depends.” Portela, 167 F.3d at 695 (alterations in original) (internal quotation marks omitted). “[E]videnee of an individual participant’s understanding of the interdependence of' the co-conspirators’ activities is ... often the best evidence [ ] of tacit agreement between the individual and his co-conspirators.” Id. The jury, heard testimony that sellers sold drugs for multiple owners, and that this was permitted by the owners. From this evidence it would be reasonable for the jury to infer that the owners worked cooperatively to ensure the success of the drug point. Further evidence of this cooperation was the testimony that several owners met to resolve the problem that arose after Maysonet-Soler shot at Rodriguez-Sostre’s sister. And perhaps the strongest evidence of this co*310operation is the fact that a solution was reached at that meeting, implying that peaceful commerce for all at La Quince was more important to the participants than the individual goals of either of the feuding owners.
Finally, the evidence demonstrated substantial overlap among the participants, with Negrón-Sostre wearing two hats, working as a runner for Pérez-Mer-cado, while at the same time owning his own brand of heroin. Negrón-Sostre was not the only multi-tasker; runners and sellers worked for multiple owners, and lookouts worked for the benefit of all. The highly-organized nature of the shift system and the meetings between the owners all suffice to show overlap.
The defendants argue, however, that there were many groups selling drugs independently, and the fact that there were different brands of drugs was indicative of independent lines of supply. Further, they assert that because the owners did not share profits, and there was “no central figure” in control, La Quince played host to multiple conspiracies, rather than a single conspiracy. According to the defendants, the only thing they had in common was the location in which they peddled their wares.
There is no requirement that the members of the conspiracy share profits, or answer to a single boss. There is abundant evidence, however, that the owners worked cooperatively to maintain security and negotiate disputes in order to maximize their own profits. “The fact that every defendant did not participate in every transaction necessary to fulfill the aim of their agreement does not transform a continuing plan into multiple conspiracies.” Id. at 696 (internal quotation marks omitted).
The defendants suggest that “this case may be best understood if we think of the coconspirators as owners of a supermarket that sold different products.” In United States v. Dellosantos, 649 F.3d 109, 121 (1st Cir.2011), we applied the supermarket simile to determine that multiple conspiracies existed where cocaine from one “chain” of three suppliers, and marijuana from another chain of two different suppliers, was sold by a single owner. In that case, we found that the members of the distinct chains were part of separate conspiracies, and the fact that their products were sold by a single “supermarket” owner did not make “the members of the two separate chains overall business partners.” Id. Here, unlike the defendants in Dello-santos, Rodríguez-Sostre and Maysonet-Soler not only owned the brands, but they ran the supermarket. A supermarket is an enterprise that offers one-stop shopping for a number of different products and brands. It is the very consolidation of all of those brands within one convenient location — with security, ample inventory, and staffed by personnel in regular shifts— that makes a supermarket profitable. It is the fact that the defendant drug-brand owners also ran the drug “supermarket” that evidences the conspiracy.
“Ultimately, while the analysis of common goals, interdependence, and overlap is useful for resolving challenges to the sufficiency of the evidence on appeal, this court has looked beyond any such lists of factors to the totality of the evidence in determining whether there is factual support for a finding of a single conspiracy.” Portela, 167 F.3d at 696 (internal quotation marks omitted). Our review of the record reveals ample evidence to support a single conspiracy.17 Moreover, the same evidence dem*311onstrates the defendants’ knowledge of and voluntary participation in that conspiracy.
Counts II-V — Aiding and Abetting
Regarding tlm remaining counts, Negrón-Sostre, Rodriguez-Sostre, and Maysonet-Soler all sound a similar theme. Negrón-Sostre concedes that there was sufficient evidence to convict him of Count II (heroin), but he argues that there was no evidence that he assisted or intended to assist in the possession with intent to distribute crack cocaine, cocaine, or marijuana (Counts III, IV, and V, respectively). Rodriguez-Sostre challenges Counts III (crack cocaine) and V (marijuana), contending that no evidence was presented that he aided and abetted any of his co-defendants in the possession with intent to distribute these . substances. Maysonet-Soler argues that, while there “may have been evidence that [he] agreed with other persons to sell cocaine,” the evidence failed to show that he aided and abetted the possession with intent to distribute heroin, crack cocaine, and marijuana (Counts II, III, and V).
In essence, the defendants argue that the evidence did not connect each of them to any of the illicit drugs other than their own brands. MaysoneNSoler further argues that co-operating witness Rosado-Pabón didn’t tie him to the others, and mere knowledge of his co-defendants’ drug trafficking (and his presence during sales) is insufficent to prove he aided and abetted that trafficking. Because the required showing is the same under each count, we will analyze them together.
It is a well-settled principle of aiding and abetting liability that if the government proves the elements of a crime charged by proof beyond a reasonable doubt, a defendant may be held indirectly responsible as an aider and abettor if he “associated himself with the venture ... participated in it as something that he wished to bring about, and ... sought by his actions to make the venture succeed.” United States v. Lugo Guerrero, 524 F.3d 5, 13 (1st Cir.2008). The government can satisfy its burden by demonstrating “that the defendant consciously shared the principal’s knowledge of the underlying criminal act, and intended to help the principal.” United States v. Bristol-Mártir, 570 F.3d 29, 39 (1st Cir.2009) (internal quotation marks omitted). It is not necessary to prove that a defendant had “[k]nowledge of the particular controlled substance being imported or distributed ... intent to distribute can be inferred from the quantity of drugs involved.” Id. (first alteration in original) (internal quotation marks omitted).
As discussed in the previous section, there was ample testimony that Neg-rón-Sostre, Rodriguez-Sostre, and Mayso-neNSoler associated themselves with the venture of operating a drug “supermarket” at La Quince. Robles-Santana testified that he saw these defendants at La Quince regularly, and that the defendants met and had discussions there. Further, Sierra-García described seeing the defendants handing out samples of new drug batches at the drug point. The government “is entitled to rely, even exclusively, on circumstantial evidence to prove its case, and the proof need not exclude every reasonable hypothesis of innocence, provided the record as a whole supports a conclusion of guilt beyond a reasonable doubt.” Lugo-Guerrero, 524 F.3d at 13 (internal quotation marks omitted). Although the defendants’ mere presence in La Quince is not sufficient, standing alone, to prove that *312they aided and abetted the possession with intent to distribute every type of drug sold there, their regular, ongoing presence and interaction with each other is certainly strong circumstantial evidence that they associated themselves with the venture.
La Quince was a highly-organized operation that ran 24/7 and provided seemingly all of the illicit substances its clientele might desire. Lookouts, much like store security, served to protect all owners from losses — not from shoplifters, but from law enforcement. Runners supplied multiple sellers, and sellers simultaneously sold brands from several owners, much like warehouse operators and sales clerks. All of these workers were organized in strict twelve-hour shifts. This level of coordination would not have been possible without the participation of the defendants with an intent to ensure the success of the venture. Not only did the owners cooperate by allowing their runners and sellers to work for different owners at the same time, but when necessary, they met to resolve a dispute that might have threatened the profitability of the enterprise.
It is apparent that each of the defendants consciously shared knowledge of the criminal design of the La Quince drug point, and worked together to ensure its success; “[k]nowledge of the particular controlled substance being ... distributed is not necessary.” Bristol-Martir, 570 F.3d at 39 (first alteration in original) (internal quotation marks omitted). Accordingly, there was sufficient evidence to establish that each of the defendants aided and abetted each of the others in the possession with intent to distribute all of the types of drugs charged.
III.
Conclusion
There was sufficient evidence to sustain the defendants’ convictions, however, the closure of the courtroom during jury selection was a structural error that requires us to vacate their convictions and remand for a new trial.