United States v. Flores-De-Jesus ( 2009 )

569 F.3d 8 (2009)

UNITED STATES of America, Appellee,
v.
Saúl FLORES-DE-JESÚS, Daniel Feliciano-Rodríguez, and Rafael Sabino-Morales, Defendants, Appellants.

Nos. 06-2670, 06-2671, 06-2672.

United States Court of Appeals, First Circuit.

Heard October 29, 2008. Decided June 18, 2009.

Edward J. O'Brien, for appellant Flores-de-Jesús.

Luis R. Rivera González, for appellant Feliciano-Rodríguez.

Irma R. Valldejuli, for appellant Sabino-Morales.

Vernon B. Miles, Assistant United States Attorney, with whom Rosa E. Rodriguez-Velez, United States Attorney, Nelson Pérez-Sosa, Assistant United States Attorney, Chief, Appellate Division, and German A. Rieckehoff, Assistant United States Attorney, were on brief, for appellee.

Before TORRUELLA and LIPEZ, Circuit Judges, and DiCLERICO,^[*] District Judge.

LIPEZ, Circuit Judge.

This case requires us to assess the propriety of the government's use of a law enforcement officer as the first witness in a multi-defendant drug prosecution to provide an "overview" of the prosecution's case. While we have condemned aspects of this practice before, most notably in United States v. Casas, 356 F.3d 104, 117 (1st Cir.2004), we must regrettably revisit the overview witness issue in some detail because of the abuse of that practice in this case and others.

Appellants Saúl Flores-de-Jesús, Rafael Sabino-Morales, and Daniel Feliciano-Rodríguez were three of twelve co-defendants charged in a four-count indictment with various crimes related to a drug trafficking enterprise at a public housing facility in Trujillo Alto, Puerto Rico. After a jury trial, all three were convicted of conspiracy to possess with intent to distribute at least five kilograms of cocaine, in violation of 21 U.S.C. § 846. Flores-de-Jesús and Sabino-Morales, but not Feliciano-Rodríguez, were also charged with conspiracy to possess, use, brandish, or carry firearms in relation to a drug trafficking crime, in violation of 18 U.S.C. § 924(c) and (o), but only Flores-de-Jesús was convicted on this count. Flores-de-Jesús, Sabino-Morales, and Feliciano-Rodríguez were sentenced to terms of imprisonment of 235 months, 210 months, and 121 months, respectively.

Two of the appellants, Flores-de-Jesús and Sabino-Morales, assert a number of evidentiary errors and contend that the district court wrongly denied their motions for acquittal pursuant to Rule 29 of the Federal Rules of Criminal Procedure. All three raise distinct objections to their sentences. We affirm the challenged convictions and the sentence imposed upon appellant Feliciano-Rodríguez. However, because we find that the district court improperly applied the manager/supervisor enhancement to defendants Flores-de-Jesús and Sabino-Morales, we vacate their sentences and remand for resentencing.

I.

Appellants were convicted for their participation in a conspiracy that operated out of the Nuestra Señora de la Covadonga public housing development ("Covadonga") in Trujillo Alto, Puerto Rico. Between 1998 and 2004, "the Covadonga drug point operated as a thriving drug marketplace in which different dealers sold various brands, or lines, of crack, cocaine, heroin, and marijuana, with the organized assistance of runners and lookouts equipped with walkie talkies." United States v. Rodríguez, 525 F.3d 85, 93 (1st Cir.2008) (summarizing facts established at the trial of Wilfredo Feliciano Rodríguez, one of the leaders of the drug point).

Appellants and nine other individuals were named in a four-count superseding indictment rendered by a District of Puerto Rico Grand Jury on March 11, 2004. Count One charged defendants with conspiracy to possess with intent to distribute, and conspiracy to distribute, controlled substances in a public housing project, in violation of 21 U.S.C. §§ 841(a)(1) and (b)(1)(A), 846, and 860. The indictment alleged that the conspiracy involved the following drugs: five kilograms or more of cocaine, and/or fifty kilograms of cocaine base (crack), and/or one kilogram or more of a substance containing a detectable amount of heroin, and/or one thousand kilograms or more of marijuana. Flores-de-Jesús and Sabino-Morales were further charged in Count Two of the indictment with conspiracy to use, carry, or possess a firearm in violation of 18 U.S.C. § 924(c)(1)(A) and 924(o).

According to the indictment, Flores-de-Jesús and Sabino-Morales assisted the leaders and organizers of the conspiracy in the administration of the drug-trafficking organization. Flores-de-Jesús and Sabino-Morales were also classified as managers/runners, whose role was to "supervise the operations of the drug points, to obtain the drugs, and to ensure a sufficient supply of drugs to the point so that there would be continual distribution." The managers/runners would "receive packaged narcotics and would deliver them to the sellers for sale at the drug points. Then the managers/runners would collect the proceeds derived from the drug sales." Feliciano-Rodríguez was charged as a seller who "distribute[d] the drugs at various drug points" within Covadonga. At trial, the majority of the government's evidence against appellants was presented through three witnesses: the lead-off witness, Special Agent Anthony Toro Zambrana of the Special Investigations Bureau of the Puerto Rico Department of Justice; confidential informant (CI) Oscar Espada; and cooperating co-conspirator Omar Medina Torres. Importantly, the government introduced surveillance videotapes and photographs of various drug and firearm transactions at the drug point. Also testifying for the government was an agent involved in an undercover drug purchase at the drug point, as well as several forensic chemists. The chemists presented their analyses of drugs obtained during several controlled purchases at Covadonga, as well as those seized by Agent Toro from building 32 of the housing project. The jury convicted all three defendants on the conspiracy charge, but only Flores-de-Jesús was convicted on the charge of using or brandishing a firearm in relation to a drug trafficking crime.^[1]

Feliciano-Rodríguez does not challenge his conviction, but contends that the trial judge failed to correctly compute the amount of drugs individually attributable to him for sentencing purposes. Flores-de-Jesús and Sabino-Morales challenge both their convictions and their sentences on various grounds. Their primary claim is that their convictions must be reversed because the district court improperly admitted the testimony of Agent Toro, whom the government used as a so-called "overview witness." In order to address that contention and determine whether the erroneous admission of portions of Agent Toro's testimony requires a new trial, we first describe the legal principles applicable to overview witnesses. Next, we recount Agent Toro's testimony in significant detail because that detail is crucial to our resolution of the overview witness issue. We then proceed to the harmless error analysis, where we describe the most significant inculpatory evidence against these two defendants. Finally, we briefly address the remaining objections of Flores-de-Jesús and Sabino-Morales to their convictions before turning to the various sentencing issues raised by each of the three appellants. Throughout, the facts are conveyed in the light most favorable to the verdict. United States v. Avilés-Colón, 536 F.3d 1, 8 (1st Cir.2008).

II.

A. The Casas Decision

In Casas, we criticized the government's "misguided use" of a government agent as an overview witness to "map out its case and to describe the role played by individual defendants." 356 F.3d at 117. In that case, DEA Agent Stoothoff described an elaborate drug trafficking organization that he said included the defendants, and he testified that this organization handled "specific massive quantities of cocaine and heroin." Id. at 118. We held that this testimony was "improper," id. at 120-21, and "fatally flawed" for a number of reasons. Id. at 119. First, the agent's testimony "went well beyond his personal knowledge," and he "did not differentiate the testimony that was based on personal knowledge from other sources of information, often hearsay." Id. at 118-19. Second, instead of "present[ing] testimony about the characteristics of large-scale drug organizations in general," he "essentially testified that each of the defendants was guilty of the conspiracy charged." Id. at 119. This was particularly troubling because there was

no indication that Agent Stoothoff's conclusions that the defendants were members of the drug organization were even based on testimony that was eventually presented at trial and could be evaluated by the jury.... In fact, Agent Stoothoff's testimony was likely, at least in part, based on the statements of a witness that the government chose not to call at trial; the record shows that the purported leader of the conspiracy, Israel Perez-Delgado, cooperated with the government and provided information. But Israel Perez-Delgado did not testify. The defendants had no chance to cross-examine him, did not know what he had said to the government, and had no basis to challenge a conclusion drawn from what he had said.

Id.

Casas identified three characteristics of overview testimony that make it "inherently problematic." Id. First, "such testimony raises the very real specter that the jury verdict could be influenced by statements of fact or credibility assessments... not in evidence." Id. Second, "there is... the possibility that later testimony might be different than what the overview witness assumed." Id. at 119-20. Finally, "[o]verview testimony by government agents is especially problematic because juries may place greater weight on evidence perceived to have the imprimatur of the government." Id. at 120.

Unfortunately, prosecutors from the United States Attorney's Office for the District of Puerto Rico have not heeded our warning in Casas. Since that opinion was issued, we have repeatedly been forced to chide the government for its continued improper use of overview witnesses despite our admonitions about this practice.^[2] Indeed, the government itself must now concede that portions of Toro's testimony were "precisely the type of testimony condemned by this Court in United States v. Casas." This statement is eerily familiar. See Rodríguez, 525 F.3d at 96 ("As the Government concedes, Agent Toro's statement ... was precisely the sort of improper overview testimony from a case agent that we have condemned.")^[3] This regrettable history requires us to address once again the problematic nature of overview testimony and to clarify the limited scope of its proper use.

B. Legal Principles Concerning Overview Witnesses

1. The imprimatur problem

The problem with overview testimony is obvious if the evidence promised by the overview witness never materializes; we repeat our condemnation of the use of overview witnesses in such circumstances. However, even if all of the overview testimony is corroborated during the course of the trial, there is still the problem with such testimony that was identified in Casas: "[o]verview testimony by government agents is especially problematic because juries may place greater weight on evidence perceived to have the imprimatur of the government." 356 F.3d at 120; see also United States v. Garcia, 413 F.3d 201, 214 (2d Cir.2005) (noting that even if the testimony of an overview witness does prove to be "a summary of later-admitted evidence", it is "generally viewed as `improper ... for a party to open its case with an overview witness who summarizes evidence that has not yet been presented to the jury'" and that "[t]he law already provides an adequate vehicle for the government to `help' the jury gain an overview of anticipated evidence as well as a preview of its theory of each defendant's culpability: the opening statement") (quoting 6 Jack B. Weinstein, Weinstein's Federal Evidence § 1006.04[3]); Avilés-Colón, 536 F.3d at 21 n. 13 (expressing concern about the use of a government agent "to endorse the testimony of other witnesses, who testify from personal knowledge about the involvement of the defendant in the conspiracy, and thereby add the imprimatur of the government to those witnesses' testimony"). The overview testimony of a law enforcement official is not simply a repetition (at best) of other evidence. It is also, in effect, an endorsement of the veracity of the testimony that will follow. Garcia, 413 F.3d at 213 (noting that an overview witness did more than simply summarize the trial testimony; in addition, he "told the jury that ... an experienced DEA agent[ ] had determined, based on the total investigation of the charged crimes, that [the defendant] was a culpable member of the conspiracy").

We prohibit this endorsement when it takes the form of a prosecutor affirming a personal belief in the veracity of a government witness. That practice is called vouching, and it is prohibited "because of its potential to shore up a witness's credibility by putting the prestige of the United States behind him and thereby inviting the jury to find guilt on some basis other than the evidence presented at trial." United States v. Vazquez-Botet, 532 F.3d 37, 53 (1st Cir.2008); see also United States v. Perez-Ruiz, 353 F.3d 1, 9 (1st Cir.2003) ("A prosecutor improperly vouches for a witness when she places the prestige of her office behind the government's case by, say, imparting her personal belief in a witness's veracity or implying that the jury should credit the prosecution's evidence simply because the government can be trusted."). In our view, there is no meaningful difference between the endorsement of credibility offered by the government's overview witness and the endorsement offered by the vouching prosecutor. Indeed, Casas itself raised the comparison between overview testimony and vouching by citing to Perez-Ruiz after observing that overview testimony by government agents is especially problematic because of the effect on juries of "evidence perceived to have the imprimatur of the government." Casas, 356 F.3d at 120.

The problematic nature of overview testimony (even when the testimony accurately anticipates the testimony of subsequent witnesses) is further confirmed by the cautious approach that courts have taken to so-called "summary witnesses," who testify at the end of the government's case instead of the beginning. The testimony of a "summary witness" must, by definition, be completely consistent with the other evidence introduced at trial. United States v. Stierhoff, 549 F.3d 19, 28 (1st Cir.2008). Still, courts have allowed such witnesses only in limited circumstances. For example, in United States v. Fullwood, 342 F.3d 409, 413 (5th Cir.2003), the Fifth Circuit endorsed the limited use of summary witnesses in complex cases, but also recognized the inherent dangers of such testimony where an agent had been "allowed, without justification, to simply recap substantial portions of the Government's case-in-chief." Collecting cases, the court identified several of these potential dangers, including the fact that the practice unfairly allowed a prosecution witness merely to repeat the testimony of another, and the possibility that the credibility of the summary witness may be substituted for the credibility of the evidence summarized. Id. at 413-14. Other courts have admitted summary testimony only in the most complicated of cases, and only where the potential prejudice is minimized through the use of procedural safeguards such as limiting instructions and extensive cross-examination. See, e.g., United States v. Ray, 370 F.3d 1039, 1046-47 (10th Cir. 2004)^[4] (twenty-three day drug conspiracy trial with over fifty witnesses); United States v. Johnson, 54 F.3d 1150, 1161 (4th Cir.1995).

The reluctance of courts to allow the government an additional opportunity to present its case in a tidy package at the end of its presentation of evidence, even when the summary evidence is, by definition, completely consistent with the rest of the trial record, confirms that the imprimatur problem with such repetitive testimony is inescapable whether that testimony comes at the beginning or end of the government's case. There is no justification for presenting an overview witness who simply anticipates the testimony of other government witnesses, even if he does so accurately. However, as we shall explain, there may still be value and legitimacy in government overview testimony that serves a different purpose.

2. Summary of the case v. summary of the investigation

There may be value in having a case agent describe the course of his investigation in order to set the stage for the testimony to come about the nature of the conspiracy and the defendants involved. Indeed, in considering a defendant's challenge to the testimony of an IRS agent, we have explicitly distinguished between the kind of "overview" testimony that we described in Casas and an agent's "description of his investigation" into the defendant's activities, which is based on personal knowledge. United States v. Hall, 434 F.3d 42, 57 (1st Cir.2006). The officer's testimony, if properly limited to "constructing the sequence of events in the investigation," could be valuable to "provide background information and to explain how and why the agents even came to be involved with th[e] particular defendant." United States v. Goosby, 523 F.3d 632, 638 (6th Cir.2008) (quotation marks and citation omitted).

Absent a basis in personal knowledge, however, the overview witness may not offer substantive testimony about the nature of the conspiracy or the involvement of particular defendants. When a law enforcement witness "express[es] opinions as to defendants' culpability based on the totality of information gathered in the course of their investigation[ ]," Garcia, 413 F.3d at 211, these conclusory statements often involve impermissible lay opinion testimony, without any basis in personal knowledge, about the role of the defendant in the conspiracy. Id. at 211-13; Fed.R.Evid. 701.

Additionally, such overview testimony may have serious Crawford/Confrontation Clause implications. In Crawford v. Washington, 541 U.S. 36, 124 S.Ct. 1354, 158 L.Ed.2d 177 (2004), the Supreme Court held that a declarant's "testimonial" out-of-court statement is inadmissible under the Confrontation Clause unless "(1) the declarant testifies ... or (2) the defendant had a prior opportunity for cross examination and the declarant is unavailable... or (3) the evidence is admitted for purposes other than establishing the truth of the matter asserted." United States v. Maher, 454 F.3d 13, 19-20 (1st Cir.2006). Therefore, "[p]ost-Crawford, the admission of non-testifying informants' out-of-court testimonial statements, through the testimony of police officers, is a recurring issue in the courts of appeals." Id. at 19. Overview testimony frequently reflects reliance on the statements of non-testifying informants, who implicate the defendants in criminal activity. The problematic nature of such testimony is self-evident. Finally, and more generally, such overview testimony is almost certain to be rife with hearsay. Rodríguez, 525 F.3d at 96 (overview testimony also "constituted improper hearsay testimony"). While the subsequent trial testimony of the out-of-court declarant resolves the Crawford issue, which is based on the Confrontation Clause, it does not cure the hearsay problem more generally.

3. The intersection of overview and expert testimony

Delineating the boundaries of acceptable overview testimony becomes even more complicated when the "overview" witness is also qualified as an expert. Overview witness testimony is normally limited by the rule that lay opinion testimony is admissible only if it is 1) not grounded in scientific or specialized knowledge but "rationally based on the perception of the witness," and 2) helpful to the jury in acquiring a "clear understanding of the witness's testimony or the determination of a fact in issue." Fed.R.Evid. 701. In contrast, the Federal Rules permit experts to provide opinions so long as "(1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and (3) the witness has applied the principles and methods reliably to the facts of the case." Fed.R.Evid. 702. Experts, unlike lay witnesses, may base their opinions on facts or data that need not be admissible, as long as these facts are "of a type reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject." Fed.R.Evid. 703. Facts or data that are otherwise inadmissible are not to be disclosed to the jury unless the court determines that their "probative value in assisting the jury to evaluate the expert's opinion substantially outweighs their prejudicial effect." Id.

The Casas court thus appropriately noted that "expert witnesses have leeway other witnesses do not," although the government had never sought in Casas to qualify Stoothoff as an expert, and, "[m]ore fundamentally, ... Agent Stoothoff's testimony that particular persons were members of the conspiracy was not an appropriate subject for expert testimony. It was not in any way linked to the `specialized knowledge' that Rule 702 requires." Casas, 356 F.3d at 120; see also Johnson, 54 F.3d at 1157 ("Rule [703] does not afford the expert unlimited license to testify or present a chart in a manner that simply summarizes the testimony of others without first relating that testimony to some `specialized knowledge' on the expert's part as required under Rule 702 of the Federal Rules of Evidence."). Thus, as with any other witness, courts cannot allow an overview witness to testify as an "expert" as to matters that are not appropriately the subject of expert testimony.

On the other hand, we have often allowed expert testimony regarding the operation of criminal schemes, "[i]n particular, expert testimony regarding the description of a typical drug network." García-Morales, 382 F.3d at 18-19 (quotation marks and citation omitted). Therefore, we have held that, while portions of an overview witness's testimony were improperly admitted, other sections of the testimony were acceptable in his capacity as an expert qualified to describe the structure and operation of a typical drug conspiracy. Id. at 17-19.

Nevertheless, even appropriate expert testimony under Rule 702 "may be excluded under Fed.R.Evid. 403 if its probative value is substantially outweighed by the risk of unfair prejudice it creates." United States v. Lopez-Lopez, 282 F.3d 1, 15 (1st Cir.2002) (quotation marks and citation omitted). While there is no prohibition against a witness testifying as both an expert and a fact witness, "courts must be mindful when the same witness provides both lay and expert testimony," United States v. Upton, 512 F.3d 394, 401 (7th Cir.2008), because of the heightened possibility of undue prejudice. The problem is especially acute where the dual roles of expert and fact witness are filled by a law enforcement official, in part because "the jury may unduly credit the opinion testimony of an investigating officer based on a perception that the expert was privy to facts about the defendant not presented at trial." Id. at 401. See also United States v. Dukagjini, 326 F.3d 45, 53 (2d Cir.2003) (observing that "the use of the case agent as an expert increases the likelihood that inadmissible and prejudicial testimony will be proffered," in part because when the case agent functions as an expert for the government, "the government confers upon him the aura of special reliability and trustworthiness surrounding expert testimony, which ought to caution its use") (quotation marks and citation omitted); United States v. Brown, 776 F.2d 397, 401 n. 6 (2d Cir.1985) (observing that the risk of a jury conflating expert and fact testimony "is increased when the opinion is given by the very officers who were in charge of the investigation") (quotation marks and citation omitted). The law enforcement/overview witness who testifies as both a fact witness and an expert thus exacerbates the imprimatur problem we have noted earlier.

C. Agent Toro's Testimony

Agent Toro was on the witness stand for an entire day of trial. Some of Agent Toro's testimony suffered from the infirmities that have led us to condemn overview testimony in the past. Some of his testimony was acceptable. In the discussion below, we distinguish between the permissible portions of his testimony and those portions that were inappropriate under Casas and its progeny.

1. Background and qualification as an expert

After taking the stand, Agent Toro first explained that his division, the Special Investigations Unit, was responsible for investigating organized crime and governmental corruption in Puerto Rico. Agent Toro testified that he had worked for the Unit for eleven years and three months. During that time period, he had been assigned to a federal agency (the Alcohol, Tobacco and Firearms Agency) for eight and a half years.

Agent Toro's primary job responsibility was conducting investigations related to drug trafficking and weapons, which often involved surveillance, interviewing informers, executing search warrants, seizing contraband, and making arrests. He testified that he had, in the past, worked undercover at various drug points, including in public housing projects. Over the course of his career, he had been involved with more than seventy-five undercover "transactions" (though he did not himself conduct any undercover drug purchases at Covadonga) and had, in the past, been qualified as an expert for two trials involving another public housing project in Puerto Rico. The government tendered Agent Toro as an "expert in sales and distribution of illegal narcotics controlled substances." Although this phrase is ambiguous, the prosecutor apparently intended that Toro would testify as an expert on the structure and typical operation of drug distribution conspiracies at Puerto Rico housing projects.

During the course of voir dire by defense counsel, Agent Toro testified that he had received formal training after being sworn in as an agent through his participation in the Special Investigations Unit Academy, the Puerto Rico Police Academy, several trainings with the DEA, and five trainings at the Institute of Police and Management in Florida. However, he had received no specific training as to "drug quantities or the amount of drugs ... sold in Puerto Rico" or "the movement and amounts of drugs specifically as to the public housing project known as Nuestra Señora de Covadonga." Instead, he presented himself as an expert "based on [his] daily work for the last eleven years and a half."

At sidebar, the defense attorneys expressed reservations about Toro's qualifications as an expert. They protested that Toro should not testify as to the specific drug quantities for which each defendant was responsible, stating that he had no "factual basis to be able to form an opinion as to how much drugs were being sold at a point in Covadonga," much less the amounts attributable to each defendant. Defense counsel argued that Toro had only learned about the case by interviewing informants and therefore his testimony was based on hearsay rather than personal knowledge. The government, for its part, maintained that Agent Toro's experience made him an expert as to the sale and distribution of controlled substances at drug points in general, and also that in other aspects of his testimony (presumably as a fact witness,) he would be testifying about "Covadonga and the drug conspiracy [of] which he ha[d] direct knowledge." The government also repeatedly assured defense counsel and the court that Agent Toro would not attribute specific drug quantities to each defendant, and that he would instead provide "expert" testimony as to "drug distribution methods and so forth, of people involved in this type of operation ... and how the drug point was run, the different roles within the drug point and the quantities sold in Covadonga." The court accepted Agent Toro as an expert over the vigorous objections of all defendants.

2. Covadonga and drug points generally

Drawing on his expertise and testifying as an expert witness,^[5] Agent Toro explained the basic concept of a drug point, noting that drug points in public housing projects are strategically chosen so that the drug dealers may avoid arrest. Next, on the basis of personal knowledge and testifying as a fact witness, he described the layout of Covadonga itself, including the entrances and location of a police precinct at the main entrance. He explained that his office's involvement in the investigation into Covadonga began when he met with the police commander responsible for the area. After that meeting, and based on documents they received from the Puerto Rico police, his Unit decided to embark on an undercover investigation of the drug point, which was located in a common area between buildings 10, 11, and 13 of the housing project. Using an aerial photograph, Agent Toro again described the layout of the project and pointed out the location of the drug point, the vehicle entrance, an additional pedestrian exit, and the police precinct. None of this testimony was problematic.

3. Installation of Oscar Espada and his role in the investigation

Agent Toro then recounted the reasons for the NIE's installation of a confidential informant — Oscar Espada — at the housing project in October 2002. Espada was a professional photographer who was "introduced into the public housing project to become known by and make friendship with the persons involved in the drug point and to collect evidence and information." The government installed him in a second-floor apartment in building 13, which overlooked the suspected drug point. From that location, Espada was able to conduct surveillance of illicit activity in the area. Over the course of seven months in 2002 and 2003, he recorded approximately 70 videotapes documenting the activities at the drug point. Espada also took still photographs of various drug and firearm transactions at Covadonga.

Agent Toro appropriately testified that he had watched Espada's tapes, characterizing them as videos of "people selling and buying drugs. There will be people with firearms at the drug points, people collecting money. People counting money. People delivering drugs." He also permissibly stated, on the basis of his personal involvement in the investigation, that the agency had decided to send in Espada to befriend the individuals involved in the drug trafficking because it was the only way to obtain evidence about the internal operations of the conspiracy. However, when the prosecutor asked Toro a question about how Espada was able to make friends with the conspirators and gain their confidence, defense counsel objected. In response, the court told the jury that "[t]he information is based on the information [Espada] gave [Toro]," and admitted the information conditionally, pending confirmation of Toro's testimony by Espada.

The judge's instruction to the jury missed the point. Toro's testimony was unquestionably hearsay. It unnecessarily anticipated testimony that Espada would give himself. It had the imprimatur problems we have described. The fact that Espada later testified as Toro said he would did not redeem these deficiencies in Toro's testimony.

4. The shootings at Covadonga

The government asked Agent Toro why Espada's videotaping stopped in 2003. Despite defendants' protestations, the court allowed Toro to testify that, following the murder of Luis Osorio, a/k/a/ "Trompi," one of the leaders of the drug conspiracy, there was a wave of violence against those who had been associated with him. Toro explained that Espada had been linked to Trompi and, as a result, the new owners of the drug point had tried to shoot him and then set fire to his apartment. Although the source of Toro's knowledge on these points is unclear, the testimony was not being offered for the truth of the matter asserted, but rather as an explanation for the end of the surveillance. Accordingly, it was not hearsay. Nevertheless, at the defendants' urging, the district court gave three limiting instructions because of Rule 403 concerns.^[6]

5. Roles in the conspiracy

The most troubling part of Agent Toro's testimony involved his conclusions about the roles of the defendants in the conspiracy. He "essentially testified that each of the defendants was guilty of the conspiracy charged." Casas, 356 F.3d at 119. Over defendants' objection, and upon receiving the AUSA's assurance that subsequent witnesses would also testify as to the respective roles of the co-conspirators, the court allowed Toro to identify by name and role in the drug conspiracy twenty-five individuals (including appellants) who, during the course of the video surveillance, had been "involved in drug distribution activities in Covadonga." Toro then circled the three defendants' pictures on a chart that contained the names and photographs of these twenty-five alleged co-conspirators (the photographs were apparently screen captures from the videos), stating that Flores-de-Jesús and Sabino Morales were both sellers and runners, and that Feliciano was a seller.

The government concedes that this testimony — identifying the names and roles of the conspiracy members with the use of the accompanying chart — was "precisely the type of testimony" we condemned in Casas. This concession was unavoidable. Toro stated that he had been able to identify the names and roles of the individuals on the chart using the surveillance videotapes, information from the company that managed the housing project, reports from the Puerto Rico police, and information obtained from CI Espada. This testimony was not based on Agent Toro's personal knowledge.^[7] Most of it was based on inadmissible hearsay.

For example, Toro stated that some of his testimony was based on information gleaned from interviews with CI Espada. It is true that defense counsel later had a chance to cross-examine Espada.^[8] The fact that Espada confirmed much of Toro's testimony with respect to the roles of the various defendants is certainly relevant to the harmless error analysis. Moreover, this confrontation ensures that there is no Crawford problem with this aspect of Agent Toro's testimony. Nevertheless, even if corroborated, the portion of Toro's role testimony that was based on Espada's statements was still hearsay and squarely raises the imprimatur issue discussed above. Moreover, neither Espada nor any other witness corroborated all of Toro's testimony about each co-conspirator and his role. Finally, to the extent that Toro's role testimony was based on police reports and information from the manager of the housing project that were never admitted into evidence, it was improper yet again on hearsay grounds.^[9]

6. Seizure of drugs from Covadonga

Agent Toro then appropriately testified as a fact witness that on July 1, 2003, he seized heroin, marijuana, and crack from a vacant apartment on the first floor of building 32 of the housing project. He later permissibly provided a detailed description of the kinds (and quantities) of drugs and drug paraphenalia that were recovered in the specific seizure in which he had participated; he also identified some of these items, which were introduced by the prosecution as physical evidence, along with several photographs of other materials confiscated during the seizure.

7. Testimony about stash houses

Stepping back from the specific raid at Covadonga, Toro properly explained to the jury, based on "[his] training and experience as a law enforcement officer," (i.e., in his expert capacity), that large quantities of drugs intended for sale at a drug point are often stored in a remote location so that if the police arrived at the location where the drugs were sold, they would only be able to seize a small quantity of the product. However, Toro then added that there were other stash houses in Covadonga where drugs were kept (besides the location from which he had seized the drugs). Upon defendants' objection, the government inquired as to the basis for his knowledge of the stash houses. Although Toro responded that he had obtained this information from "[r]eports from the Puerto Rico Police and information from the informant," thereby exposing the impropriety of his testimony on this topic, the court did not strike the testimony from the record. This ruling was erroneous.

8. Drug quantities and drug sales at Covadonga

The prosecutor asked Agent Toro this question: "Based on the investigation and based on all of your training and experience as a law enforcement officer, approximately how much cocaine was being sold in Covadonga between the years 2002 until [the end of the investigation in 2003]?" Toro responded with estimates of the amounts of cocaine, crack, heroin, and marijuana sold monthly during that time period, stating that his estimates had been arrived at "using the videos and interview of witnesses and reports." The government concedes that this drug quantity testimony was impermissible under Casas.

In Casas, part of the overview testimony we found inappropriate was testimony "that the organization handled specific massive quantities of cocaine and heroin." 356 F.3d at 118. That statement may account for the government's concession that Agent Toro's drug quantity estimates were inappropriate. Agent Stoothoff was not testifying as an expert in Casas, and, based on Agent Toro's testimony that he had arrived at his drug quantity estimation based on, inter alia, witness interviews and other reports, it appears that Agent Toro's testimony as to drug quantity may not have been offered in his expert capacity but rather as a fact witness. In any event, in light of the government's concession, this is not an issue that we need decide here. Moreover, while we accept the government's concession that Toro's drug quantity testimony was improper under these circumstances, we do not imply that a properly qualified expert may never offer valid drug quantity testimony where that testimony comports with the other requirements for expert testimony under the Federal Rules. Additionally, we note that because Agent Toro's drug quantity testimony appears to have been based at least in part on interviews with individuals who did not testify at trial, such testimony also has the potential to raise Crawford problems.

D. Summary

Generally, and based on the principles we described in Part II.B., supra, Agent Toro's testimony was permissible to the extent that he was testifying either 1) as a case agent describing the course of the investigation and events in which he had personally participated, or 2) as an expert whose testimony provided background and context on drug conspiracies and distribution in public housing projects in Puerto Rico. More specifically, in his capacity as a case agent and fact witness, the district court properly allowed Agent Toro to testify to: the general layout and other specific descriptions of Covadonga and the course of his unit's investigation of the drug point, including the use of Oscar Espada as an informant and his own visits to the project and seizure of drugs from the property. As an expert, Toro properly described the operation of drug points generally, including the various "roles" typically involved in an intricate drug conspiracy and the practice of storing drugs intended for sale, as opposed to immediate distribution, at a remote location from the drug point itself.

In contrast, the government concedes that Toro's testimony "identifying the names and roles of the conspiracy members, the accompanying chart, and his estimates of the amount of cocaine sold monthly at Covadonga^[10] — was precisely the type of testimony condemned by this Court in United States v. Casas." Specifically, all of Toro's testimony about the role of the defendants in the conspiracy that was based on information gathered from police reports, other documents not introduced into evidence, and interviews with CI Espada, cooperating co-conspirator Medina Torres, or other individuals, was hearsay and inappropriate overview testimony. Even though some of this objectionable testimony was later repeated by other witnesses, thereby obviating any potential Crawford problem, the repetition did not validate the admission of the overview testimony. Later repetition of testimony by a witness with first-hand knowledge does not eliminate the hearsay nature of the initial overview testimony. And the imprimatur problem remains a barrier to such overview testimony, even if it is repeated — a problem that is exacerbated by the agent's dual role as both a fact witness and an expert. Finally, the repetition is problematic in itself.

As we have already noted, the law enforcement officer providing the overview testimony is essentially endorsing the testimony of other witnesses in what can only be viewed as an attempt by the government to bolster the credibility of those later witnesses. From the government's point of view, that enhancement effort may seem particularly critical in these drug cases where so much of the government's case relies on the often problematic testimony of confidential informants with unsavory pasts or cooperating co-defendants with myriad credibility problems. Although we do not minimize these problems for the government, we cannot condone the use of overview witnesses in the impermissible ways we have described here to overcome those problems. We made that view clear in Casas. We restate that view here with renewed emphasis.

Henceforth, there can be no justification in this circuit for the government's repetition of the errors that we have identified here in the use of overview testimony in criminal cases. If somehow prosecutors in the U.S. Attorney's Office in Puerto Rico did not get the message before about the dangers of such testimony, they should surely get it now. And they should draw no comfort from the fact that the harmless error analysis we are now required to undertake saves these convictions despite the misuse of overview testimony. The Supreme Court has made clear that we may not vacate appellants' convictions simply out of indignation at the repeated failure of the U.S. Attorney's office in Puerto Rico to abide by the strictures on the use of overview testimony. That is, the harmless error inquiry may "not be avoided by an assertion of supervisory power" to "discipline the prosecutors of [a particular] jurisdiction." United States v. Hasting, 461 U.S. 499, 505, 103 S.Ct. 1974, 76 L.Ed.2d 96 (1983).^[11] Nevertheless, if prosecutors fail to heed our guidance in the future, they may be referred for sanctions to the Department of Justice or to the appropriate Puerto Rico attorney disciplinary body. See id. at 506 n. 5, 103 S.Ct. 1974 (stating that such disciplinary sanctions would be more appropriate remedies for prosecutorial misconduct where the error is harmless). Such conduct will also subject government counsel to disciplinary actions before this court. See Rules of Attorney Disciplinary Enforcement for the Court of Appeals for the First Circuit, Rule IV (2002) (governing our authority to sanction an attorney admitted to practice before this court for misconduct).

E. Harmless Error Analysis

We must now undertake a harmless error analysis in order to determine whether the erroneous admission of the overview evidence requires the reversal of appellants' convictions.^[12]

The admission of improper testimony is harmless if it is "highly probable that the error did not influence the verdict." Casas, 356 F.3d at 121; see also United States v. Garcia-Ortiz, 528 F.3d 74, 80 (1st Cir.2008). "The government, not the defendants, bears the burden of establishing harmlessness." United States v. Rodriguez-Marrero, 390 F.3d 1, 18 (1st Cir.2004) (quotation marks and citation omitted). "There is no bright-line rule" as to how to properly conduct this inquiry; instead, the

harmlessness determination demands a panoramic, case-specific inquiry considering, among other things, the centrality of the tainted material, its uniqueness, its prejudicial impact, the uses to which it was put during the trial, the relative strengths of the parties' cases, and any telltales that furnish clues to the likelihood that the error affected the factfinder's resolution of a material issue.

United States v. Sepúlveda, 15 F.3d 1161, 1182 (1st Cir.1993). Accordingly, it is impossible to assess whether the improper admission of portions of Agent Toro's testimony was truly harmless without engaging in a detailed analysis of the evidence against each defendant apart from Toro's improper testimony. In so doing, we must be careful not to conflate the standard for harmlessness with a sufficiency standard. Here, the appropriate inquiry is not whether a reasonable jury could have looked at the evidence and reasonably found defendants guilty, but rather whether it is "highly probable" that the error did not influence the verdict.

1. Flores-de-Jesús

Flores-de-Jesús was the only defendant to be convicted of both the drug conspiracy and the firearms conspiracy charges. We turn first to the evidence regarding the drug conspiracy count.

Trial testimony established that in 2002, the leader of the drug point, Bebe La Rubia, was murdered after a power struggle, and the point was taken over by Luis Osorio, a/k/a "Trompi," and Cristian Villegas, a/k/a "Casi." In 2003, Trompi himself was murdered and Casi was driven out of the drug point. At this point, Wilfredo Feliciano Rodríguez ("Wilfre") and Alex Trujillo assumed control of the drug point, which they maintained until the organization was dismantled by a series of arrests in 2004.

Confidential informant Espada testified that two or three days after he had moved into Covadonga in approximately October 2002, he met Flores-de-Jesús in the stairwell of his building, where Flores-de-Jesús was selling drugs. Espada told him that he was a professional photographer who was unemployed and had lost his equipment, and was badly in need of money to support his family. Flores-de-Jesús promised to help Espada, and Trompi, one of the leaders of the drug point at the time, eventually gave Flores-de-Jesús the money to lend Espada so that he could buy a camera. With his new camera, Espada would take photographs of some of the individuals associated with the drug point and then print those pictures on T-shirts for them. Flores-de-Jesús was the first person Espada contacted about these photographs, one of which was introduced into evidence by the government.

As part of his undercover operation, Espada purchased fifty packages of cocaine from "Casi," who was also one of the leaders of the drug point at that time (the government later played a videotape of this transaction). Espada also testified that he had seen all three defendants associating with Casi, at times when Casi was "distributing substances and weapons." Espada stated that Flores-de-Jesús and the others working at the drug point took no measures to hide their activities and that drug-related activity was obvious from the first day he moved into the project. This account was later corroborated by the testimony of cooperating co-conspirator Omar Medina Torres, who stated that drugs were being sold at the point 24 hours a day, "in the open air."

Cooperating co-conspirator Omar Medina Torres stated that he had known Flores-de-Jesús for most of his life. According to Medina, Flores-de-Jesús started selling drugs at Covadonga in 2000, and continued until the time he was arrested. Flores-de-Jesús would almost always work the shift from 2:00 pm until 9:00 pm, the same shift that Medina worked. When he first began selling drugs, Flores-de-Jesús sold "black-vial" crack cocaine, $5 "clown" cocaine (the clown symbol identified the seller), $20 cocaine, and $5 heroin. These sales were for an individual nicknamed "Surdo." During this period, Medina estimated that Flores-de-Jesús would sell 200-300 baggies of clown cocaine and maybe five baggies of the $20 cocaine during one shift.

Medina stated that, over the history of the drug point, Flores-de-Jesús sold for several different individuals who changed with the leadership of the drug point: "Surdo" was first, then "Trompi," and finally Wilfre Feliciano. During the latter part of his career, when he was selling for Feliciano, Flores-de-Jesús sold pink-cap crack, black-vial crack, $5 cocaine, and $5 heroin. Medina estimated that Flores-de-Jesús could sometimes sell 250-300 of the black vials of crack, the same amount of pink-cap crack, and 50-150 packages of heroin per shift.

Medina also characterized Flores-de-Jesús as a runner and a manager. He initially served this role for "Trompi," and ended up working as a runner/manager for Alex Trujillo and Feliciano when they took over the drug point. As a runner, Flores-de-Jesús would stay at the point to see that everything was "running OK." He would go get more drugs if the supplies at the point were running low, and he would collect the money from the point and deliver it to whoever was the owner at the time, first to Trompi, and later to Alex Trujillo and Wilfre Feliciano. Medina knew this was so because Flores-de-Jesús often worked during shifts when Medina was selling drugs at the point. Medina stated that, although he was the best seller at the drug point, Flores-de-Jesús was the second-best, and would sell about the same quantities of drugs per shift as Medina (Medina had already provided extensive testimony on exactly how much this was).

There were also several videotapes introduced into evidence that corroborated the testimony of Medina and Espada. A video taken on December 10, 2002, showed Flores-de-Jesús selling crack and cocaine and holding bags of drugs, along with Medina and some of the other alleged members of the conspiracy. Another recording, dated December 16, 2002, showed Flores-de-Jesús selling drugs, including cocaine and heroin, near Medina and several other co-conspirators. Medina testified at trial that he was upset on the tape because Flores-de-Jesús was "moving in on him."^[13] A video taken December 26, 2002, showed Flores-de-Jesús with two other conspirators "balancing out the shift," i.e., sorting and distributing the money earned in a particular shift. Later on that same day, another runner was seen delivering cocaine to Flores-de-Jesús, who was selling. Another recording dated December 30, 2002, again showed Flores-de-Jesús selling cocaine. Medina, Flores-de-Jesús, and Casi all appeared in a January 3, 2003 video in which Flores-de-Jesús was conducting a drug sale. Finally, Flores-de-Jesús was also seen selling drugs and holding a walkie-talkie in a video dated February 12, 2003.

We contrast the "substantial evidence of [a]ppellant's guilt" that we have described, which included not only the testimony of two different witnesses, but numerous video recordings of Flores-de-Jesús conducting drug transactions over the course of several months, with the "defense's case [which] was relatively weak, largely limited to cross-examination of the prosecution's witnesses." Rodríguez, 525 F.3d at 97. None of the cross-examination yielded any exculpatory evidence with respect to Flores-de-Jesús (or, for that matter, any of the defendants). There were no witnesses presented to contradict the government's evidence respecting defendants' involvement in the conspiracy.

Nevertheless, we acknowledge that the evidence against Flores-de-Jesús came primarily from a confidential informant and a cooperating co-defendant. As we have mentioned, testimony from such individuals is often problematic. Moreover, because Agent Toro's repetition of such testimony was interspersed with proper expert testimony, the risk of undue prejudice was higher than it would have been if he were simply a fact witness for the government. Therefore, the bolstering of the government's case by the improper overview testimony of Agent Toro was a significant part of this trial. It squarely implicates the imprimatur problem that we noted in Casas and that we discussed at length above.

Although the issue is close, we conclude that the harmless error analysis saves the government's convictions. The videotapes played for the jury clearly depicted appellant's interactions with members of the alleged conspiracy, as well as his involvement with illegal narcotics. The videotapes corroborate the testimony of the confidential informant and the cooperating co-conspirator. That corroboration does not depend on any endorsement of their testimony by Agent Toro through his overview testimony. Nor did the government, in closing, refer to Agent Toro's improper testimony or use it to try to prove their case. In short, on these facts, we conclude that it is highly probable that the improper assertion by Agent Toro — that Flores-de-Jesús was a manager and a runner in the Covadonga drug conspiracy — was "the same determination that the jury would have drawn in the absence of the inadmissible testimony." Casas, 356 F.3d at 122.

There is also enough evidence supporting Flores-de-Jesús's conviction on the firearms conspiracy charge that we find it "highly probable" that the erroneous admission of Agent Toro's testimony did not influence the verdict on this count either. Espada testified that he occasionally photographed some of the co-conspirators with weapons. In the photograph of Flores-de-Jesús and three other alleged members of the drug conspiracy that was introduced into evidence, one of the other individuals was holding a weapon. Espada also testified that all three defendants often saw "Casi" carrying weapons at the point. Additionally, Medina stated that he frequently saw individuals at the drug point carrying weapons, and that Flores-de-Jesús was often present at these times. He explained that "there are always going to be weapons at a drug point, always ... that's a drug point." More specifically, Medina stated that he himself would carry a weapon, a .38 caliber, while selling drugs at the point, and that all three defendants had seen him carrying it before. He said that not only the so-called "enforcers" carry weapons, but "nearly all of [the conspirators] would [go around] with them" in order to protect the drug point. Finally, a videotape dated December 16, 2002 showed Flores-de-Jesús with a radio (perhaps looking out for police), and then, more importantly, loading a magazine with a bullet. An April 10, 2003 video showed one of the drug conspiracy's "enforcers" with a pistol; Flores-de-Jesús was present in the frame.^[14] Because none of Agent Toro's improper testimony specifically pertained to whether Flores-de-Jesús conspired with others to use a firearm in connection with the drug conspiracy, it is highly probable that the error did not influence the verdict. Indeed, when asked on cross examination whether he had ever seen Flores-de-Jesús with a weapon on his person, Toro explicitly stated that he had not.

2. Sabino-Morales

Although Rafael Sabino-Morales was charged with both the drug conspiracy and firearms conspiracy counts, he was convicted only of the former. Because the witnesses occasionally made general statements about all three of the defendants at once, some of the evidence against Sabino-Morales overlaps with the evidence we recounted above with respect to Flores-de-Jesús.

We first recall Espada's testimony that all three defendants had associated with "Casi," one of the leaders of the conspiracy, while he was "distributing substances and weapons." Espada had also linked Juan Carlos, the individual who had tried to kill Espada, with all three defendants.

Medina testified that he had known Sabino-Morales for almost his whole life. Medina had seen Sabino-Morales "working the table" a few times, decking cocaine. Sabino-Morales began this work when Bebe la Rubia was still one of the leaders of the drug point. On one occasion during that time period, Sabino-Morales packaged the drugs in his own home. Medina also testified that when Wilfre Feliciano and Alex Trujillo took over the point in 2003, everyone was scared because of the recent power struggles, and no one wanted to run drugs. At that point, Sabino-Morales began to act as a runner and, in fact, because of the tense atmosphere, he was the only runner at the point for some period of time.

As a runner, Sabino-Morales would pick up money from the sellers, including Medina himself, and bring new supplies of drugs to the point. Sabino-Morales would run drugs and money to and from the drug point two to three times a day, every day. On several occasions, Medina saw the leader of the point, Feliciano, in Sabino-Morales's apartment.

The video evidence against Sabino-Morales was not as overwhelming as the recorded evidence against Flores-de-Jesús, but it provided significant corroboration of the witness testimony. A video showed that on December 26, 2002, Sabino-Morales gave a seller marijuana. On April 10, 2003, conspiracy leaders Wilfre Feliciano and Alex Trujillo appeared in a video at the drug point, and several minutes later, Sabino-Morales appeared on the tape, collecting money, and, later, holding a bag of cocaine. Once again, therefore, we conclude that it is highly probably that the admission of Toro's improper overview testimony did not influence the verdict against Sabino-Morales.

However, unlike Flores-de-Jesús, Sabino-Morales also asserts another, more specific, objection related to Agent Toro's overview testimony. In addition to his general objections about the impropriety of the overview testimony, Sabino-Morales further challenges his conviction by arguing that the district court erred by admitting into evidence the drugs that Agent Toro identified as the ones he seized in building 32. Sabino-Morales asserts that Agent Toro's testimony was the only evidence linking these drugs to the conspiracy, and that this was an insufficient foundation for their admission into evidence.

There was no error in the admission of this testimony. During Medina's testimony, he identified the drugs seized in building 32 by their packaging as those sold at the Covadonga drug point with which he was involved, thereby providing a sufficient, if belated, basis for their admission. Specifically, Agent Toro stated that heroin seized in building 32 was decked in aluminum foil, that marijuana was sold in plastic baggies pressure-sealed with symbols of marijuana leaves, and that crack was packaged in vials with different colored lids. Medina confirmed that this was the manner in which all of these drugs were packaged at Covadonga. Additionally, Medina was even shown samples of the various drugs that Agent Toro had previously identified as those that he had seized in Building 32. Medina specifically identified the "weed baggies" and vials of crack as the kind used at the drug point, and also stated that the aluminum packaging containing the heroin was the same kind used at the drug point, although he stated that a different color had been used. Accordingly, there was no error in admitting these drugs, and we reject Sabino-Morales's challenge to his conviction on this basis.

III.

Flores-de-Jesús also argues that his convictions should be set aside because he was unfairly prejudiced by the admission into evidence of certain testimony about several murders and other violent episodes at Covadonga. Flores-de-Jesús claims that the erroneous admission of this testimony cannot be considered harmless.

The challenged evidence of violence at the housing project came from several witnesses. We have already recounted Agent Toro's testimony citing the shootings at Covadonga and the attempt on Espada's life as the reasons that the video surveillance was terminated. See supra II.C.4 and n. 5. We see no error in the admission of this testimony. However, CI Espada himself also told the jury about the attempt on his life in April 2003. As he described these events in detail, including how he had been forced to flee the housing complex with his wife and child, he became visibly emotional. Espada further stated that the individual who had tried to kill him was associated with the defendants as a lookout for the point. This individual also distributed drugs. Additionally, Medina was asked about another individual ("Fredito") who used to be at Covadonga but was no longer there because he had been gunned down.

Espada's testimony about the shootings and attempts on his life was cumulative and more detailed than necessary to explain why the video surveillance had ended, a fact which had already been established. The admission of that detail was erroneous. Medina's testimony regarding Fredito was also erroneously admitted. The government's brief merely notes that this testimony regarding Fredito "does not implicate the defendants in any way," but makes no further attempt to explain the relevance of this testimony. In the absence of any apparent relevance, it was unfairly prejudicial within the meaning of Rule 403. See Fed.R.Evid. 403. The court did provide a limiting instruction after Espada's testimony, stating that "[t]he jury will take note of the fact that there was an attempt on his life, but it is in no means to be attributed to the defendants on trial." The court gave a similar limiting instruction after Medina's testimony: "you have heard testimony concerning a murder here or someone being gunned down. Under no circumstances are you to impute that to any of the three defendants in this case."

In considering the significance of these errors, we note again the substantial evidence against Flores-de-Jesús that we recounted in the course of our analysis of the improper overview testimony. In addition to that substantial evidence, we note the court's repeated curative instructions, and the fact that the government, in closing, did not even mention this testimony, let alone dwell on it. Under these circumstances, we conclude that it is highly probable that this erroneously admitted testimony did not influence the jury's verdict against Flores-de-Jesús. Hence, the error was harmless.^[15]

IV.

Having affirmed the convictions of Flores-de-Jesús and Sabino-Morales, we now address each appellant's sentencing challenges in turn.

A. Flores-de-Jesús

The Presentence Investigation Report (PSR) for Flores-de-Jesús, the only defendant convicted of both the drug conspiracy and the firearms conspiracy charges, grouped the two counts together for a combined base offense level of 32. See U.S.S.G. § 3D1.2(c) (providing for such grouping where one of the defendant's convictions "embodies conduct that is treated as a specific offense characteristic in, or other adjustment to, the guideline applicable to another of the counts"); U.S.S.G. § 2D1.1 (c)(4) (setting a base offense level of 32 for a conspiracy to possess with intent to distribute at least five but less than fifteen kilograms of cocaine). The PSR then added a two-point enhancement for the possession of a firearm in furtherance of the conspiracy, see U.S.S.G. § 2D1.1(b)(1), and three points for appellant's role as a supervisor/manager of criminal activity involving five or more participants, see U.S.S.G. § 3B1.1(b). The total offense level was thus 37, which, for someone in appellant's criminal history category of II, yielded a guidelines sentencing range (GSR) of 235-293 months.

Flores-de-Jesús timely objected to both enhancements. At the sentencing hearing, the district court stated that it found "the three level enhancement [ ] justified on the basis of the evidence presented in this case" and, regarding the two level firearm enhancement, similarly stated that "from what I have seen here, from the evidence, I believe the enhancement is justified." The court sentenced Flores-de-Jesús to 235 months imprisonment, the minimum sentence within the GSR, as well as five years of supervised release on count one and three years on count two, to be served concurrently. On appeal, Flores-de-Jesús argues that the court improperly applied both the three-level upward adjustment for his role as a manger/supervisor in the drug conspiracy and the two-level firearms enhancement.

For all of appellants' sentencing challenges, we review the district court's factual determinations for clear error. United States v. Pierre, 484 F.3d 75, 88 (1st Cir.2007). "We will not find clear error unless on the entire evidence we are left with the definite and firm conviction that a mistake has been committed." United States v. Arbour, 559 F.3d 50, 53 (1st Cir.2009) (internal quotation marks and citations omitted). However, we review questions of law in sentencing determinations de novo. United States v. Ramos-Paulino, 488 F.3d 459, 463 (1st Cir. 2007). "A question about whether the evidence is sufficient to support a particular guideline determination is a question of law and, therefore, engenders de novo review." Id.

1. The Manager/Supervisor Enhancement

U.S.S.G. § 3B1.1(b) prescribes a three-level enhancement to the base offense level if "the defendant was a manager or supervisor ... and the criminal activity involved five or more participants or was otherwise extensive." Although the terms "manager" and "supervisor" are not defined in the Sentencing Guidelines, the terms are described in U.S.S.G. § 3B1.1 cmt.4 as involving

the exercise of decision making authority, the nature of participation in the commission of the offense, the recruitment of accomplices, the claimed right to a larger share of the fruits of the crime, the degree of participation in planning or organizing the offense, the nature and scope of the illegal activity, and the degree of control and authority exercised over others.

U.S.S.G. § 3B1.1 cmt.4; see also United States v. Soto-Beniquez, 356 F.3d 1, 54 (1st Cir.2004). Evidence of the defendant's role in the conspiracy "may be wholly circumstantial," and need only show that he "exercised authority or control over another participant on one occasion." García-Morales, 382 F.3d at 19-20. "It is not enough, however, that the defendant merely controlled, organized, or managed criminal activities; rather, he must instead control, organize, or manage criminal actors." United States v. Ofray-Campos, 534 F.3d 1, 40 (1st Cir.2008) (emphasis added).

Flores-de-Jesús argues that there is no evidence that he ever exercised control over other conspirators, since his role was "simply to deliver drugs to the point and to collect the money and to sell." The government contends that the record shows that Flores-de-Jesús also "oversaw" the drug point to make sure the operation was running smoothly, and that this oversight included collecting the money from the sellers and restocking the point when the drugs sold out. The government claims that this establishes the requisite supervisory or managerial role. We disagree.

The district court did not make specific findings of fact pertaining to whether the manager/supervisor enhancement should be applied to Flores-de-Jesús. Nor does the PSR contain specific findings of fact on this topic; instead, it merely characterizes appellant as a "manager/runner" whose job was to "supervise the operation of the drug point, to obtain the drugs, and to ensure a sufficient supply of drugs to the point so that there would be continual distribution." Thus, the PSR, like the government's brief, merely asserts a connection between drug runners and "oversight" of the drug point. This link is not borne out by the record.

Omar Medina Torres, the sole valid source of the "role" testimony at trial, was asked: "Other than simply selling drugs, did Saul Flores de Jesús have any other roles or jobs on the drug point?" Medina answered that appellant was also a runner and a manager. The questioning continued:

Q: And as a runner, what was his function or what were his duties?

A: To stay at the point, to see that everything was running okay. If anything was needed, to go get it. To collect the money....

Q: Who ended up as the manager?

A: "Saulito" Flores.

Q: And as the manager, what did "Saulito" do? What were his responsibilities?

A: Picking up the money and taking more drugs to the drug point.

Q: So were these jobs or roles interchangeable, the roles of runners and managers?

A: The thing is that you all — in more decent words, you all say "manager." But we others, we guys from the housing project, we say "runners."

Finally, cross examination yielded the following exchange:

Q: You stated in your direct testimony that you equate — that you understand that the Government equates the term "manager" to that of what you referred to as "runner."

A: Yes.

Q: That is your understanding.

A: Yes, as a runner.

Q: And the runner's job is simply to go take drugs to the point whenever their supply is running down, as well as pick up monies that would be there to take it back to wherever.

A: Yes.

Thus, even though Medina used the word "manager" at trial to characterize appellant's role in the conspiracy, he used this term interchangeably with "runner," and his particular choice of language is not a proper basis for the enhancement. As a runner, Flores-de-Jesús unquestionably played an essential role in the drug trafficking operation. But keeping the drug point well-stocked and collecting the proceeds to deliver to the point's owners or leaders is insufficient to establish the requisite control over another criminal actor that our case law requires. See, e.g., Ofray-Campos, 534 F.3d at 40; Ramos-Paulino, 488 F.3d at 464. "[T]he record is devoid of any evidence to show that [Flores-de-Jesús] exercised control over any individual ... [or] oversaw their activities." Ofray-Campos, 534 F.3d at 40. Although there is certainly evidence that Flores-de-Jesús was deeply involved in the operation of the drug point and that he worked with other individuals to carry out this purpose, there was nothing to show that these individuals were his "subordinate[s] in the chain of command" or that he oversaw their activities. Ramos-Paulino, 488 F.3d at 464. Any argument that Flores-de-Jesús "qualified for the upward enhancement on the basis of his control over the property, assets or activities of a criminal organization is squarely foreclosed" by our precedent. Ofray-Campos, 534 F.3d at 41 (internal quotation marks and citation omitted). The application of this enhancement was an error.

2. The Firearms Enhancement

We find no merit in Flores-de-Jesús's other challenge to his sentence. In order to justify an enhancement under U.S.S.G. § 2D1.1(b)(1), the government must prove by a preponderance of the evidence that "it was reasonably foreseeable that a co-conspirator would possess a gun in furtherance of the criminal activity." Casas, 356 F.3d at 129. The evidence presented at trial included a photograph and a video of Flores-de-Jesús with co-conspirators holding weapons, a video of Flores-de-Jesús himself loading a magazine with a bullet, and Medina's testimony that he, like other co-conspirators, would consistently carry firearms at the drug point in order to protect it, and that all three appellants had seen at least Medina with a weapon. This evidence is more than sufficient to sustain the imposition of the § 2D1.1(b)(1) enhancement. We therefore find no error in the application of the firearm enhancement.

B. Sabino-Morales

After finding appellants guilty on the conspiracy count, the jury filled out a special verdict form indicating its finding that the drug conspiracy as a whole involved 5 kilograms or more of cocaine. On January 9, 2006, the district court ordered a United States Probation Officer to determine the specific amount of drugs attributable to each individual defendant in accordance with this court's decision in United States v. Colón-Solís, 354 F.3d 101 (1st Cir.2004). After reviewing the trial transcripts, the probation officer recommended that Sabino-Morales be held responsible for 86.68 kilograms of cocaine, which would have placed him at a base offense level of 36.^[16] However, on October 6, 2006, the district court ruled that, for sentencing purposes, the amount of drugs it would attribute to each defendant would be the five kilograms found by the jury. Accordingly, the PSR for Sabino-Morales started with a base offense level of 32. He then received the same enhancements for firearms (two levels) and his role as a manager/supervisor (three levels) as Flores-de-Jesús, yielding a total offense level of 37. Because he had no criminal record, however, the GSR was 210-262 months, which was lower than the calculated range for Flores-de-Jesús. The district court sentenced Sabino-Morales at the lowest end of the sentencing range (210 months), to be followed by five years of supervised release.

Sabino-Morales now raises the same argument as Flores-de-Jesús with respect to the manager/supervisor enhancement. He also claims that the district court failed to adequately compute the correct amount of drugs individually attributable to him and that the sentence he received was disproportionate to the offense charged and to the sentences imposed upon other defendants. We take each argument in turn.

1. The Manager/Supervisor Enhancement

Our analysis of Sabino-Morales's challenge to the imposition of this enhancement is no different than his co-defendant's. At Sabino-Morales's sentencing hearing, the court's only specific findings regarding this enhancement focused on his role as a runner. The court was

satisfied from what it has heard and from what appears in the trial transcripts of this case, and that has been presented to this court that certainly it is justified. It believes that a runner has a supervisory role, otherwise he would not be able to comply with the duties of a runner.

Thus, the court based its decision to impose the enhancement on its understanding of the general responsibilities of drug runners. As we have already explained, the description of runners offered by Medina-Torres is insufficient to warrant an enhancement that is designed for individuals who exercise control over other members of the organization. The record reveals no other basis on which Sabino-Morales may have warranted this enhancement. Accordingly, the enhancement was wrongly applied.

2. The Drug Quantity Computation

"In determining drug quantity for purposes of calculating a defendant's base offense level under the Guidelines, the sentencing court may attribute to the defendant `all reasonably foreseeable quantities of contraband that were within the scope of the criminal activity that he jointly undertook.'" United States v. Jones, 523 F.3d 31, 41 (1st Cir.2008) (quoting U.S.S.G. § 1B1.3 cmt. n. 2(ii)). Absent "an individualized finding as to drug amounts attributable to, or foreseeable by" a particular defendant, a court may not automatically assume that he is responsible for the total amount attributable to the conspiracy as a whole, Colón-Solís, 354 F.3d at 102-03. However, "a drug dealer who engages in criminal activity with others to further their collective interests may be held liable for the quantities of drugs sold by his partners, if those sales were a reasonably foreseeable consequence of the jointly undertaken actions." United States v. Laboy, 351 F.3d 578, 582 (1st Cir.2003). The government must prove drug quantity by a preponderance of the evidence, and we "will uphold the sentencing court's estimate as long as it is reasoned and finds support in the record." Jones, 523 F.3d at 41.

Sabino-Morales argues that the "most the evidence showed at trial was that defendant-appellant was a late-joining conspirator, who worked sporadically and was a runner only for a certain period of time during 2003." He claims that because there was no evidence that he continued running drugs into 2004, it was improper for the district court to include 2004 in the calculations, and that his sentence reflected the court's improper attribution to him individually of the drug amount for which the entire conspiracy was responsible.

This argument is unavailing. In the first place, the district court obviously understood the need for individualized drug quantity determinations pursuant to Colón-Solís, because it asked the probation office to produce reports with respect to each defendant. Moreover, at a sentencing proceeding involving counsel for all defendants, the court heard extensive arguments challenging these determinations, and explained its position that the evidence amply supported the fact that each defendant could have reasonably foreseen that the conspiracy would move at least five kilograms of cocaine. This determination is supported by the record.

The probation officer's report was based on Medina's testimony. Medina testified that prior to 2003, Sabino-Morales would sometimes work the table at Covadonga, decking drugs. In 2003 (on one occasion Medina characterized the date as "mid-2003"), Sabino-Morales began to act as a runner for Wilfre Feliciano. Indeed, Sabino-Morales was the conspiracy's only runner for at least some period of time, meaning that during that time period he was running drugs "every single day." According to Medina, runners would bring fresh supplies of drugs to the point up to five times a day. The cocaine would arrive in packages of one-hundred small baggies. The report also noted the expert testimony of a forensic chemist who stated that it would take approximately 1,759 of the small baggies of cocaine to equal 1 kilogram.

The probation officer's report on Sabino-Morales calculated that he was responsible for a total of 86.68 kilograms of cocaine.^[17] As we have described, however, the district court did not adopt the report's calculation. Instead, it used only the more conservative five kilogram quantity to sentence Sabino-Morales. The record easily justifies that quantity determination, even on the most conservative assumptions.

If one credits Sabino-Morales's assertion that he worked as a runner only during 2003, and assumes that Medina was correct that Sabino-Morales only began working as a runner in the middle of that year, and assumes further that Sabino-Morales worked for only one day a week during that time period and not "every single day," these factors would still lead to the conclusion that, during his time as a drug runner, Sabino-Morales personally handled 7.39 kilograms^[18] of cocaine. Even this estimate, which certainly gives Sabino-Morales the benefit of the doubt, is more than the five kilograms used by the district court for sentencing purposes. And this figure does not even include the drugs foreseeably handled by other members of the conspiracy. Accordingly, we find no error in the district court's determination as to drug quantity with respect to Sabino-Morales.

3. Proportionality

Sabino-Morales contends that the sentence imposed by the district court is unreasonable under 18 U.S.C. § 3553(a)(6) in light of certain sentencing disparities between himself and what he claims are similarly situated defendants. Specifically, Sabino-Morales points to the sentences received by the leader of the drug point, Cristian Villegas (135 months), his assistant (108 months), an enforcer and four sellers (70 months each), and two more co-defendants with unspecified roles (57 months each). Sabino-Morales mechanically compares his 210-month sentence with those of the other defendants, claiming that the lower sentences imposed upon individuals who were more "culpabl[e], in need of much more rehabilitation and [who posed] much more danger to the community" make the disparities "outrageous and untenable."

We disagree. The key difference between Sabino-Morales and the co-defendants he cites is that all of them pled guilty. A defendant who chooses to enter into a plea bargain is not similarly situated to a defendant who contests the charges against him. See, e.g., United States v. Rodríguez, 162 F.3d 135, 152 (1st Cir.1998) (plea bargains lead to "sentencing disparity for the defendants who chose to put the government to its burden in proving its case. Nevertheless, the law allows the government to do this, even if it results in sentences of such disparity as would strike many as unfair"). We do not think it unreasonable that Sabino-Morales was sentenced to significantly more time than other members of the conspiracy who chose to plead guilty when his sentence was at the lowest end of a GSR that was arrived at using an extremely conservative estimate of drug quantity.

C. Feliciano-Rodríguez

The PSR for Feliciano-Rodríguez began with a base offense level of 32. The PSR once again noted that, although the probation officer had attributed 17.74 kilograms of cocaine to Feliciano-Rodríguez individually, a quantity which would have carried a base offense level of 34, the court had ruled that it would use the five kilogram quantity for sentencing purposes. The PSR then recommended the same two-level firearm enhancement that his co-defendants had received, but also a two-level decrease pursuant to U.S.S.G. § 3B1.2(b) because his participation in the conspiracy was minor. Because of several prior convictions, and because Feliciano-Rodríguez was on probation when he committed the offense, he fell into a criminal history category of III. Accordingly, the PSR computed a GSR of 151-188 months.

Feliciano-Rodríguez then moved for a downward departure based on diminished capacity pursuant to U.S.S.G. § 5K2.13. In support of this motion, he presented the results of several psychiatric evaluations. The court determined that a two-level downward departure was warranted under the circumstances, and, accordingly, sentenced Feliciano-Rodríguez based on a total offense level of 32 and a criminal history category of II. The court once again sentenced appellant to the lower end of the guidelines sentencing range, which was 121-151 months. He was also sentenced to five years of supervised release.

Feliciano-Rodríguez's sole claim on appeal is one he shares with Sabino-Morales: namely, that the district court failed to adequately compute the correct amount of drugs individually attributable to him. He argues that the evidence does not support the district court's decision to attribute five kilograms of cocaine to him for sentencing purposes. Once again, this argument fails.

The probation officer's report recounted Medina's testimony that Feliciano-Rodríguez first became involved in the conspiracy in 1998. He then acted as a runner for "Toño" until 2001. At this time, he ran crack daily. Starting in 2001, Medina stated that Feliciano-Rodríguez would sell "any [drug] that might be available" at the point once or twice a week until he was arrested in early 2004. Medina testified that Feliciano-Rodríguez would sell for 24-hour shifts on Sundays. He also stated that during one seven-hour shift, a person might sell 200-300 small bags of cocaine. The probation officer's report used this testimony. Taking account only of Feliciano-Rodríguez's time as a seller, and not a runner, the officer arrived at an individualized drug quantity of 17.74 kilograms of cocaine for Feliciano-Rodríguez.^[19]

Once again, however, the district court used the more conservative estimate of five kilograms, which was certainly a reasonable amount to attribute to him. Even assuming that he only sold for one day a week for one year, this would still put the amount he himself sold over the five kilograms used by the sentencing court. Once again, this figure does not even take into account the amount that Feliciano-Rodríguez could have reasonably foreseen his co-conspirators selling. We therefore reject Feliciano-Rodríguez's challenge to the sentence imposed on him by the district court.

V.

For the foregoing reasons, we affirm the convictions of both Flores-de-Jesús and Sabino-Morales, but vacate their sentences and remand for resentencing due to the district court's erroneous imposition of the manager/supervisor enhancement. However, on remand, we leave open the possibility that the government can cite evidence in the record not relied upon here, nor apparent to us from our own review of the record, that would justify the application of the manager/supervisor enhancement. See Ramos-Paulino, 488 F.3d at 465. With respect to Sabino-Morales, we also direct the court, in accordance with footnote 1 of this opinion, to enter an amended judgment that accurately reflects his conviction on Count One of the indictment only. Feliciano-Rodríguez's sentence is affirmed.

So ordered.

NOTES

[*] Of the District of New Hampshire, sitting by designation.

[1] Although the jury verdict states that Sabino-Morales was found guilty only on Count One (the drug conspiracy charge) and not Count Two (the firearms conspiracy charge), the judgment actually entered against him states that he was convicted on both counts. In January 2008, counsel on behalf of Sabino-Morales moved to correct his sentence because the court had imposed a special monetary assessment of $200, suggesting that it was imposed for both counts. The government agreed that it was a clerical error that should be corrected. On February 11, 2008, the district court granted Sabino-Morales's motion and stated that "[a]n amended judgment shall be issued." To date, an amended judgment has not been docketed. Accordingly, the district court is directed to enter the appropriate amended judgment as to Sabino-Morales.

[2] See Avilés-Colón, 536 F.3d at 21 n. 13 ("It is troubling to us that the government's use of the overview testimony indicates an unawareness of our decision in Casas."); Rodríguez, 525 F.3d at 95 ("This court on several occasions has strongly cautioned the Government against the practice of having a case agent make conclusory statements about a defendant's culpability at the beginning of the prosecution's case, before any supporting evidence has been offered."); see also United States v. Casas, 425 F.3d 23, 50-51 (1st Cir. 2005) ("Casas II") (criticizing government for improper use of overview testimony, though trial was held before Casas was published); United States v. García-Morales, 382 F.3d 12, 16-17 (1st Cir.2004) (same).

[3] The behavior of the U.S. Attorney's Office in Puerto Rico is particularly troubling because it is not the first time that numerous warnings from this court have gone unheeded. See United States v. Martínez-Medina, 279 F.3d 105, 127-28 & n. 12 (1st Cir.2002) (Torruella, J., concurring) (describing our repeated admonishments, made over the course of fifteen years, to prosecutors in Puerto Rico concerning improper statements in closing arguments).

[4] Although Ray was reversed and remanded in the aftermath of United States v. Booker, 543 U.S. 220, 125 S.Ct. 738, 160 L.Ed.2d 621 (2005), see Ray v. United States, 543 U.S. 1109, 125 S.Ct. 995, 160 L.Ed.2d 1035 (2005), the Tenth Circuit later reinstated all non-sentencing portions of its original opinion. United States v. Ray, 147 Fed.Appx. 32, 34 (10th Cir.2005).

[5] While Agent Toro was qualified as an expert on the typical operation of drug conspiracies, see García-Morales, 382 F.3d at 18-19, we are not suggesting that only an expert could properly provide similar testimony. See, e.g., United States v. Santiago, 560 F.3d 62, 66 (1st Cir.2009) (observing that testimony about coded language in drug transactions may be admissible as either lay or expert testimony); United States v. Pinillos-Prieto, 419 F.3d 61, 71 (1st Cir.2005) (holding that case agent's testimony regarding his experience with general operation of drug conspiracies was permissible lay opinion testimony under Rule 701).

[6] The court first stated that Toro's testimony was meant to explain why the surveillance was stopped, and not to imply that "any of the[ ] three defendants were connected with any of those events;" therefore, the jury was to draw no adverse inferences as to the defendants. Its second and third limiting instructions were essentially identical.

[7] Because of the government's concession with respect to this testimony, and because Agent Toro did not identify which portions of his testimony, if any, were based solely on the surveillance tapes, we need not determine whether a law enforcement officer's personal viewing of surveillance footage would ever be a permissible basis for such testimony.

[8] This situation contrasts with Casas, where we specifically noted that Agent Stoothoff's testimony was likely based on the testimony of a cooperating co-conspirator whom the government chose not to call at trial. Casas, 356 F.3d at 119. This scenario was particularly problematic because "t[]he defendants had no chance to cross-examine him, did not know what he had said to the government, and had no basis to challenge a conclusion drawn from what he had said." Id.

[9] Just before Toro began listing the co-conspirators and identifying their roles, the trial court asked the government whether it planned to introduce evidence to corroborate his testimony. After the government answered in the affirmative, the court stated to the jury that "[Toro's testimony] will be taken by the jury as something [that was] related to him." We cannot be sure what the trial judge was trying to convey to the jury with this statement. Perhaps he was reminding the jury that Toro obtained the information and conclusions to which he was about to testify from other sources, and thus that this evidence was not based on personal knowledge. Accordingly, the jury could have inferred that they should not give the testimony as much weight as they otherwise would have given it. Whatever the purpose of the judge's instruction, it does not avoid the error in allowing Toro to give a preview of the testimony of other witnesses.

[10] We reiterate that we do not foreclose the possibility that a properly qualified expert could offer appropriate expert testimony as to drug quantity.

[11] "Supervisory power to reverse a conviction is not needed as a remedy when the error to which it is addressed is harmless since by definition, the conviction would have been obtained notwithstanding the asserted error." Id. at 506, 103 S.Ct. 1974.

[12] Because Flores-de-Jesús and Sabino-Morales properly preserved their claims of error by repeatedly objecting to Agent Toro's testimony, there is no question that our review is for harmless error and not plain error.

[13] Medina testified that at this time he felt that Flores-de-Jesús was his competition, and that although they sold drugs "belonging" to different individuals (with the different "lines" of drugs designated by the packaging), they were both working for the same individual, Casi.

[14] In view of the evidence we have described, we summarily reject Flores-de-Jesús's claim that he is entitled to a judgment of acquittal on the firearms conspiracy charge pursuant to Federal Rule of Criminal Procedure 29. "[C]hallenges to the sufficiency of the evidence and to the denial of the motion for judgment[ ] of acquittal raise a single issue," United States v. Morillo, 158 F.3d 18, 22 (1st Cir.1998), which we review de novo. United States v. Thompson, 449 F.3d 267, 275 (1st Cir.2006). We inquire "whether the evidence presented at trial, together with all reasonable inferences and viewed in the light most favorable to the verdict, would allow a rational jury to establish the defendant's guilt beyond a reasonable doubt." United States v. Combs, 555 F.3d 60, 65 (1st Cir.2009). In this case, there is no question that the evidence we have recounted was sufficient to permit a reasonable jury to conclude that Flores-de-Jesús was guilty of the charged offense. He was convicted of conspiring to use, carry, or possess a firearm in furtherance of a drug trafficking crime in violation of 18 U.S.C. § 924(o). Therefore, the jury did not even need to find that Flores-de-Jesús himself ever used or possessed a firearm in furtherance of the drug conspiracy. It would be sufficient to find that he was part of an agreement to do so. In light of the testimony we have described, there was ample evidence to allow the jury to reasonably infer that Flores-de-Jesús had agreed with "fellow members of the Covadonga drug market to possess guns in order to provide protection and enforcement for their drug trade." Rodríguez, 525 F.3d at 104; see also id. at 104-05 (jury could have inferred such an agreement from Medina's testimony that "drug owners and sellers regularly carried guns at Covadonga while dealing in drugs" even absent any direct evidence "showing an agreement to possess weapons between [defendant] and four [other] individuals" named in Count Two of the indictment).

[15] We have noted that a combination of multiple errors may have a cumulative effect that is not harmless. See, e.g., United States v. Meserve, 271 F.3d 314, 332 (1st Cir.2001). However, the defendants have not made this argument on appeal; therefore, without suggesting that the argument would be meritorious, we do not consider it.

[16] A similar computation was made with respect to Flores-de-Jesús. However, he does not challenge the court's drug quantity determination.

[17] The PSR described the following calculation: assuming that Sabino-Morales delivered 100 bags of cocaine at least 5 times per day, 6 days a week, he would deliver 3,000 bags of cocaine per week. Multiplying this by the 52 weeks in 2003 would yield 156,000 bags of cocaine. Dividing the 156,000 baggies by the chemist's estimate of bags/kilogram, this yields the 86.68 kilogram amount.

[18] Assuming five deliveries of 100 baggies a day, once a week, this would mean 500 baggies a week. Multiplied by 26 weeks in 2003, this equals 13,000 baggies. Divided by 1,759 (the number of baggies in a kilogram), this yield the conservative estimate of 7.39 kilograms of cocaine.

[19] The report assumed that Feliciano-Rodríguez sold only on Sundays, selling 200 bags of cocaine each time. In a year, he would have sold 10,400 bags, and in three, 31,200. Dividing 31,200 by 1759 bags per kilogram yielded 17.74 kilograms of cocaine.