TORRUELLA, Circuit Judge.

In this consolidated appeal, appellants John Correy ("Correy"), Raymond Nicolai-Cabassa ("Nicolai"), Ramón Flores-Plaza ("Flores"), Angel Casas ("Casas"), José Bonilla-Lugo ("Bonilla"), and Angel Luis Pizarro-Morales ("Pizarro") challenge their sentences imposed after conviction for various drug trafficking activities. The sentences currently on appeal were imposed after a previous remand from this Court, which affirmed the appellants' convictions but remanded their cases for resentencing. See United States v. Casas, 425 F.3d 23 (1st Cir. 2005). After careful consideration, we vacate all appellants' sentences and remand for resentencing consistent with this opinion. I. Background

As is set forth in more detail in our prior opinion, id. at 29, this case arose after Drug Enforcement Administration ("DEA") agents stopped a vehicle leaving the international airport in San Juan, Puerto Rico. After the vehicle was found to contain eighty-one kilograms of cocaine, the DEA exposed a conspiracy which had allegedly transported large amounts of illegal drugs into the United States. Subsequently, by way of superseding indictment on August 8, 1996, the government charged sixty defendants, including appellants, with conspiracy to distribute 1400 grams of heroin and 9445 kilograms of cocaine. In addition to this conspiracy count, the government charged appellant Pizarro with possession with intent to distribute eighty-one kilograms of cocaine. Additionally, Correy and Nicolai were charged with possession with intent to distribute thirty-six kilograms of cocaine, the intentional killing of José Miguel Bianco-Rodriguez, and the intentional killing of Ramón de Jesús-Molina.

Trial of ten of the sixty defendants, including the appellants, commenced on May 12, 1999 and lasted approximately seven months. Numerous witness testified at trial, including two of the leading figures in the conspiracy, Thomas Martinez and Israel Pérez-Delgado. These two individuals pled guilty, received lighter sentences, and testified in detail against their underlings. As detailed below, the import of that testimony is still in dispute.

For background purposes, we will provide a summary of each appellant's alleged role. The government argues that the evidence at trial showed that Correy received narcotics arriving in New York and personally transported drugs from Miami to New York; that Nicolai provided couriers for the operation and received the cocaine once it arrived in New York; that Flores worked at the airport as a janitor and assisted in passing drugs through security; that Casas received money resulting from the sale of drugs, counted it, and stashed it in New York; that Bonilla helped guard drugs and transported money; and that Pizarro coordinated the shipment of drugs through the airport in Puerto Rico.

The jury convicted the six appellants on the conspiracy count, convicted Pizarro of possession with intent to distribute, and acquitted Correy and Nicolai of the murder charges, as well as their charge of possession with intent to distribute. The jury entirely acquitted the remaining four defendants.

Pursuant to order of the First Circuit Judicial Council responding to a backlog of cases, the case was randomly reassigned from the district court judge that presided over the trial to a different district court judge for sentencing. In 2002, the district court imposed sentences on the appellants as follows: Correy received a 480 month sentence; Nicolai received a life sentence; Flores received a 292 month sentence; Casas received a 235 month sentence; Bonilla received a 235 month sentence; and Pizarro received a life sentence. On appeal, this Court affirmed appellants' convictions, affirmed the statutory maximums to which they were subjected, but vacated the sentences imposed on the grounds of Booker error as to some defendants and the failure to timely serve certain defendants with proper Presentence Reports ("PSRs") that made specific drug quantity findings. See id. at 59-65. Specifically, we interpreted an order from the trial judge to require that the Probation Department conduct individualized determinations of drug quantity for each defendant based on the facts presented at trial. Id. at 58, 62-63. We vacated some appellants' sentences based on the Probation Department's failure to do so. Id. Further, in response to the appellants' complaints that the district court erred by relying on trial testimony that was not credible, and considering that the sentencing judge was not the trial judge, we instructed the resentencing court to review the entire record, making its own credibility determinations in order to determine drug quantity. Id. at 64 n. 56.

On remand, sentences were calculated and imposed as follows. The district court found Correy responsible for at least 908 kilograms of cocaine. His base offense level ("BOL") was thus set at 38. He received a two-level enhancement for possession of a firearm, bringing his total offense level ("TOL") to 40. With a criminal history category ("CHC") of VI, Correy's guideline range was 360 months to life. The resentencing court reimposed a 480 month sentence.

The district court found Nicolai responsible for at least 1276 kilograms of cocaine. This finding triggered a BOL of 38. The district court imposed a two-level firearm enhancement and a four-level increase for Nicolai's leadership role. This led to a TOL of 43, since that is the maximum offense level provided for in the guidelines. Nicolai unsuccessfully challenged his CHC of IV. Combined, these levels provided for a guideline sentence of life imprisonment. The district court reimposed a life sentence at the resentencing hearing.

The district court found flores responsible for 3,554 kilograms of cocaine. This finding triggered a BOL of 38. The district court applied a two-point enhancement for abuse of a position of trust, for a TOL of 40. The district court rejected Flores's arguments for a minor role adjustment. Combined with a CHC of I, the guideline range was 292 to 365 months. The district court reimposed a sentence of 292 months.

The district court found Casas responsible for 270 kilograms of cocaine. This finding triggered a BOL of 38. No enhancements applied. Combined with a CHC of I, the guideline range was 235 to 293 months. The district court reimposed a 235 month sentence.

The district court found Bonilla responsible for more than 150 kilograms of cocaine. This finding triggered a BOL of 38. No enhancements applied. Combined with a CHC of I, the guideline range was 235 to 293 months. Defendant sought leniency based on health problems, and the court cited these problems in its decision to reimpose a sentence at the bottom of the guideline range, namely 235 months.

The district court found Pizarro responsible for more than 4,200 kilograms of cocaine. This finding resulted in a BOL of 38. The district court imposed a two-level firearm enhancement and a two-level enhancement for Pizarro's leadership role in the offense. Pizarro's TOL was thus 42. Combined with a CHC of II, Pizarro's guideline range was 360 months to life. The court imposed a 360 month sentence, a decrease from the previously imposed life sentence. The district court explained that the defendant had been "shown as the most contrite, remorseful defendant [of] all of these defendants, and therefore, that's why I give him 360."

The facts underpinning these findings are a primary subject of this appeal and are discussed in more detail, as relevant, below. On appeal, the appellants challenge the drug quantity findings and various enhancements, and assert that the district court did not adequately consider the record or their claims, did not adequately explain their sentences, and afforded presumptive weight to the guideline range. Our analysis focuses on the challenge to the sentencing court's fact-finding procedure.

II. Discussion

A. Issues relevant to all appellants

1. Statutory maximum sentence

Appellants were convicted on the charge of conspiracy to distribute controlled substances in violation of 21 U.S.C. § 841(a). Under that provision, the statutory maximum is increased to life imprisonment where the offense, i.e., the conspiracy, involves "5 kilograms or more of a mixture or substance containing a detectable amount of cocaine. 21 U.S.C. § 841(b)(1)(A)(ii). Several appellants argue that their constitutional rights were violated in that they were exposed to this increased maximum sentence without jury findings on the issue of drug quantity. See Apprendi v. New Jersey, 530 U.S. 466, 490, 120 S.Ct. 2348, 147 L.Ed.2d 435 (2000) ("Other than the fact of a prior conviction, any fact that increases the penalty for a crime beyond the prescribed statutory maximum must be submitted to a jury, and proved beyond a reasonable doubt."). We have already addressed this argument in appellants' challenge to their first sentencing. Casas, 425 F.3d at 65-66. We held that "for Apprendi purposes, it is the drug quantity attributable to the entire conspiracy that determines the statutory maximum" and that any Apprendi error was harmless because the evidence was overwhelming that the entire conspiracy involved more than five kilograms of cocaine. Id. at 66 n. 58. Appellants invite us to reexamine this conclusion and argue that the conspiracy-wide total should not be used in this context. We decline to reconsider our established law.

2. Guideline drug-quantity determinations

Some appellants also question the validity of allowing the sentencing judge to set the guideline range based on fact-finding as to drug quantity, especially after Cunningham v. California, 549 U.S. 270, 127 S.Ct. 856, 166 L.Ed.2d 856 (2007). But our case law remains clear that judges may find facts to establish the guideline range within the applicable statutory maximum. See United States v. Arango, 508 F.3d 34, 42 (1st Cir. 2007) ("We have stated unequivocally that, under the advisory Guidelines, judicial fact-finding on drug quantity is constitutionally permissible. . . ."). Accordingly, we reject this challenge as well.

3. Credibility determinations

On appeal, all appellants challenge the sentencing court's drug-quantity determinations and its imposition of certain enhancements. Appellants argue that the resentencing judge was not adequately familiar with the record and so did not make adequate determinations with respect to the credibility of witnesses, especially the key cooperating leaders, Martinez and Pérez. They argue that the sentencing court simply accepted the government's proffered record citations as establishing responsibility for various drug quantities.

The question of drug-quantity determinations was also at issue in the previous appeal. There, appellants leveled broad challenges against the fact-finding done by the sentencing judge. We held that a replacement judge can sufficiently familiarize himself with the record so as to assess credibility. Casas, 425 F.3d at 56. Even though the sentencing judge proceeded with the first sentencing hearings before the trial transcripts were complete, we rejected appellants' earlier arguments that we should automatically afford his findings less deference. Id. We found that the most relevant transcripts requested by appellants and the trial judge were transcribed, that the sentencing judge had reviewed the record pertinent to defendants' sentencing, and that he had cited passages in determining drug quantity. Id. at 56-57. We concluded our analysis of the question of deference due by stating that "the scrutiny with which the record must be examined by a successor judge 'varies with the facts of each case[;] the more the case depends on the credibility, and especially the demeanor, of the witnesses, the more a judge needs to do to become adequately familiar with it.'" Id. at 57 (quoting United States v. Larios, 640 F.2d 938, 943 (9th Cir. 1981)). But, we decided "we need not analyze the merits of appellants' complaints on this issue further because we are remanding their cases for resentencing — at which appellants and the sentencing judge will have access to all of the transcripts." Id.

In remanding the case, we flagged the credibility issue and noted a potential credibility problem with Martinez. Id. at 64 n. 56. We noted that the sentencing judge had rejected these credibility concerns by saving that the jury credited that testimony. Id. We criticized this reasoning, noting that there is no way of knowing exactly what testimony was credited by the jury since there was no special verdict and since the jury did not rule on the overt acts alleged in the conspiracy. Id. Further, we noted that the acquittals on the murder charges showed "it is clear that at least some portion of his testimony was not credited." Id. We emphasized our point, and we assigned a task to the sentencing court upon remand:

We wish to clarify that the jury verdict of guilty did not determine the amount of drugs attributed to each defendant. As a result, the testimony of Martinez as to drug quantity was not necessarily accepted as credible by the jury. Any conclusion as to individual drug quantity should be based on review of the entire record.

Id. Thus, considering our statement that the level of transcript review required of the sentencing judge must depend on the importance of credibility, and our statement that the credibility of Martinez was very much at issue, the sentencing court should have conducted a thorough review of this issue.

For the exact same reasons that the credibility of Martinez may be in doubt, there was also reason to doubt the testimony of the other cooperating leader, Pérez. In fact, we recognized that in post-trial proceedings the trial judge had labeled Pérez's testimony "dubious." Id. at 53-54 (affirming denial of a motion for a new trial as allegedly new evidence would be cumulative as Pérez's credibility was already in doubt). Further, we recognized that Pérez had also testified as to the acquitted murders, offering testimony conflicting with that of Martinez. Id. at 45 ("The defendants charged with the murders were acquitted, demonstrating that the contradictory testimony likely undermined the credibility of both key prosecution witnesses."). Thus, it should have been clear that the credibility of Pérez also needed to be examined by the sentencing judge.

On remand, the appellants developed arguments attacking the credibility and reliability of other witnesses. Though our discussion in the prior opinion was focused on Martinez and Pérez, equally applicable to all witnesses was the rationale for requiring independent credibility assessments by the sentencing judge — namely the fact that the jury's verdict was not determinative of the overt acts. Thus, we do not read our prior opinion as limiting the credibility question to any subset of witnesses, nor does the government so argue. Rather, after our remand, the district court should have made credibility determinations as necessary to resolve the facts in dispute.

A review of the transcripts of the resentencing hearings makes clear that the district court did not heed our instruction, but rather persisted in its view that the jury verdict was controlling. We will discuss the sentencing hearing of Pizarro as an example to illustrate the problem. This hearing was held on September 26, 2006, and was the last of appellants' hearings. Defense counsel made arguments about the credibility of Martinez and Pérez. The court interrupted:

THE COURT: The fact that the jury may not have given credibility to Martinez and Pérez as to the murder of these individuals —

[COUNSEL:] And drugs.

THE COURT: No. No. There you go. Wrong. Nevertheless, they gave credit to the drug quantities. And let me tell you this, your — your point has been addressed and answered by the Court of Appeals in this very case. . . . [I]t says: In the [face] of this evidence, appellants — appellant attempts to attack the credibility of Pérez and Martinez and quibble with the evidence regarding the amounts attributable to them individually. Which is this, now, however, — and I keep quoting — they point to no evidence contradicting the conspiracy-wide drug quantities testified to at trial, nor do they offer an explanation for whether you believe the government witnesses regarding each appellant's activities in furtherance of the conspiracy; yet, discredit testimony regarding the type or quantity of drugs involved in the conspiracy. So — and it says further on — because the evidence is overwhelming and remanded for resentencing the appropriate sentence, I imagine would be life imprisonment, as stated in Section 841(b)(1)(A).

So the fact that the jury didn't believe witnesses to one issue does not mean they have to disbelieve in other issues. In this case, there's no question that the jury found beyond a reasonable doubt that the Defendant was part of the conspiracy, and that this Defendant was guilty of two counts.

The court engaged in no analysis of credibility, and did not otherwise explain its reasons for crediting Pérez and Martínez. The court then made its rulings regarding Pizarro, discussed in more detail below, without conducting any independent credibility analysis.

The district court's statement shows it misunderstood our opinion. As we stated above, we specifically instructed the district court not to defer to the jury on credibility issues. The district court's refusal to follow our instruction, in reliance on another portion of our opinion, was error. We explain. It is true that when finding the Apprendi issue to be harmless error, we stated that many witnesses testified to a number of large shipments, thus easily meeting the five kilogram conspiracy-wide threshold, which triggered the applicable statutory maximum. Casas, 425 F.3d at 66. We continued:

In the face of this evidence, appellants attempt to attack the credibility of Pérez and Martinez and quibble with the evidence regarding the amounts attributable to them individually. However, they point to no evidence contradicting the conspiracy-wide drug quantities testified to at trial, nor do they offer any explanation for why the jury would believe the government witnesses regarding each appellant's activities in furtherance of the conspiracy, yet discredit their testimony regarding the type or quantity of drugs involved in the conspiracy. Because the evidence is overwhelming, on remand for resentencing the appropriate statutory maximum will be life imprisonment as stated in § 841(b)(1)(A).

Id. (emphasis added) (citation omitted). Read in context, this statement is only a statement that we may accept that the jury's verdict was adequate to establish that the conspiracy-wide drug quantity exceeded five kilograms. Put another way, in a conspiracy where almost every alleged shipment involved more than five kilograms, and where several defendants were apprehended with eighty-one kilograms, there was no reason to think the jury could have returned a guilty verdict without concluding that the conspiracy involved at least five kilograms. But this says nothing about the drug quantities foreseeable to each individual, which must be used for purposes of determining each defendant's BOL under the guidelines. See United States v. Laboy, 351 F.3d 578, 582 (1st Cir. 2003); United States v. Sepulveda, 15 F.3d 1161, 1197 (1st Cir. 1993) (stating that "a defendant is not automatically saddled with the full weight of the conspiracy's wrongdoing; rather, a defendant is responsible for drugs he personally handled or anticipated handling, and . . . for drugs involved in additional acts that were reasonably foreseeable by him and were committed in furtherance of the conspiracy"). Our case law makes clear that such a drug-quantity determination is an entirely different inquiry from that of the conspiracy-wide determination. See, e.g., United States v. Colon-Solis, 354 F.3d 101, 103 (1st Cir. 2004) (vacating a sentence, for legal error, where the district court did not make individualized drug findings, even though the conspiracy-wide total was admitted in a plea). Further, as described above, our previous opinion in this case made clear that the individualized drug-quantity issue could not be resolved without assessing credibility. Casas, 425 F.3d at 65 n. 56. Thus, it was error for the district court to forego a credibility analysis related to individualized drug-quantity determinations based on the above quoted language regarding the conspiracy-wide drug quantity.

Considering we review fact-finding on drug-quantity determinations only for clear error, Laboy, 351 F.3d at 582, one could argue that we should consider the district court as having made a positive credibility determination, to which we would afford significant deference. But, where the district court has expressly made clear that it is not conducting a credibility inquiry, it would be disingenuous of us to act otherwise. Clear error review recognizes that a reasonable fact-finder can draw different inferences from evidence, and so, under such review, we tend not to disturb an inference if it has some support in the evidence, even if another inference is possible. See United States v. Santos, 357 F.3d 136, 141 (1st Cir. 2004) ("[I]f there are two plausible views of the record, the sentencing court's choice between them cannot be clearly erroneous."). Affording such deference assumes that the district court actually engaged the factual question and drew an inference. Here, that did not happen. Thus, where the district court categorically refuses to engage in a credibility assessment, we cannot apply deferential review and presume the testimony credible. As such, we will not rely on challenged testimony in determining whether the government met its burden in showing drug quantity or the factual basis for an enhancement.

This conclusion is not inconsistent with this Court's recent decision in United States v. Olivero, 552 F.3d 34 (1st Cir. 2009). In that case, the defendant also argued that the sentencing judge had not adequately familiarized himself with the record, since he stated he did not read the entire trial transcript. Id. at 38-39. In rejecting that claim, we stated, "[t]he ultimate test is whether the defendant can show clear error in the factual findings made by the court from whatever sources were consulted." Id. We affirmed, noting that the court had reviewed the evidence contained in the PSR, the prior appellate opinion, and the sentencing memoranda, and could have consulted the record as needed. Id. at 39. In that case, the PSR contained a wiretap transcript showing the defendant discussing drug trafficking. Id. The district court interpreted the transcript as establishing responsibility for a 260 kilogram shipment, but the defendant disputed this interpretation. Id. We specifically stated, "there is no disputed evidence, only competing interpretations of what the. conversations meant" and emphasized our established rule that a court may rely on an undisputed PSR. Id. at 39-40.

The present case is quite different. Our prior opinion remanding the case noted the split jury verdict and instructed the district court to make credibility findings based on the whole record. Casas, 425 F.3d at 64 n. 56. As explained below, the appellants' PSRs relied on allegations from the indictment rather than trial evidence. And, at resentencing, the defendants did attempt to dispute the trial evidence itself by attacking credibility. Thus, the district court could not rely on an undisputed PSR, drawing inferences that we would review for clear error. Instead, it was required to make credibility determinations, a task which it refused to undertake, and which we, therefore, cannot review for clear error.

As we explain below, the record shows that this attitude pervaded the sentencing hearings of all appellants. Nonetheless, we will proceed through the case of each appellant to assess whether the district court's misreading of our last opinion led to prejudicial error at each sentencing hearing. See Fed.R.Crim.P. 52(a); United States v. Nunez-Rodriguez, 92 F.3d 14, 19 (1st Cir. 1996) ("Even if we were to determine that the sentencing court committed error, however, its sentence would not be set aside unless it affected defendant's substantial rights."). Further, where necessary and possible, we will address other legal issues presented that do not depend on credibility.

B. Examination of each appellant's sentencing hearing hand claims on appeal to determine the effect of sentencing court's failure to follow our remand instructions

a. The PSR

Before we turn to the question of credibility, we must confront another problem with Correy's sentencing: the failure to prepare a PSR that complied with this Court's remand orders. This problem requires a digression. Our prior opinion found procedural error in that the Probation Department had not timely prepared a PSR for Bonilla which identified the specific drug amounts attributable to him based on the evidence presented at trial, even after the trial judge had requested such a PSR. Casas, 425 F.3d at 57-59. Such a PSR is important, since, as we explained above, a guideline drug-quantity determination must be made on the basis of quantity individually foreseeable to each defendant.

In response to the trial judge's order to base the PSR on trial testimony, the Probation Department prepared a new PSR but failed to serve it on Bonilla in a timely manner. Id. at 58. We concluded this failure to serve such a PSR on Bonilla within the time-limits set forth in Rule 32(e) to be prejudicial. Id. at 59. We noted that this is a complex case and that one purpose of the PSR is "so that parties do not have to comb through voluminous trial transcripts to anticipate remaining factual and legal issues." Id. We further interpreted the trial judge's order as extending to require such a PSR for all appellants. Id. at 62-63. The failure to prepare a proper PSR also led us to find prejudicial procedural error as to Casas and Correy. Id. at 63-64. We did not reach this issue as to the remaining appellants since their sentences were vacated based on preserved Booker error.

On the present appeal, Correy claims that his PSR at resentencing still simply included allegations copied from the indictment and did not include a individualized drug-quantity determination based on the evidence at trial, as we required in our remand order.

The government argues that the PSR adequately showed that Correy was responsible for more than 150 kilograms of cocaine. We disagree. Our review of the PSR shows that the document persists, just as in the previous version, in simply copying allegations from the indictment. Incredibly, Correy's resentencing proceedings in 2006 used the exact same PSR as was prepared in 2000, except with certain dates and ages handwritten over the old numbers.

This is highly inadequate. Rule 32(d) requires that the presentence report calculate the defendant's offense level. Such an offense level is of course dependent on the drug-quantity determinations. We need not state a per se rule on what level of information a PSR must generally include in order to support a drug-quantity determination. However, in this case, the answer is clear. Our last opinion applied the trial judge's order that the PSR be based on trial testimony to all appellants. Id. at 62-63. Thus, in this case, it is insufficient for the PSR to simply recite the allegations in the indictment in order to justify a conclusion that each individual appellant is responsible for more than 150 kilograms of cocaine.

The Probation Departments's failure to comply with this directive is not excusable. In addition to the fact that we explicitly ordered compliance, other considerations emphasize the need for a thorough PSR in this case. Each defendant facing testimony from Martinez or Pérez has challenged their credibility in disputing the drug-quantity determinations. As explained above, the jury had clearly disbelieved some of their testimony, and the jury's verdict did not establish the truth of each overt act alleged. Id. at 64 n. 56. Thus, we instructed the sentencing judge to weigh the trial testimony on remand. A PSR which merely recites allegations from the indictment adds nothing to that process. This should have been clear from our last opinion:

The offense conduct in the PSR was based solely on the overt acts alleged in the indictment as to the entire conspiracy. However, the jury did not issue a special verdict or rule on the overt acts alleged in the indictment. The PSR thus contained no findings as to drug type or quantities attributable to Bonilla based on evidence in the record.

Id. at 58 n. 45.

Further the issue is preserved on appeal. At defendant Correy's resentencing hearing, held on June 9, 2006, defense counsel objected that the PSR contained only the allegations in the indictment, in contravention of our instructions in remanding the case. The sentencing court told defense counsel the error was corrected because the PSR had been served in a timely manner. Of course, it is not enough that some PSR be served in a timely manner. Rather, what must be served in a timely manner is a PSR which conforms to the requirements imposed both by the trial judge and our remand order. The resentencing court also said it was up to the court, not the Probation Department, to determine drug quantity: "In that matter, I take issue with [the trial judge], because it is the trial judge or judge in the record, as I've done, [who has] to determine what amount is attributable to each defendant, not the Probation Officer." While it is true that determining the drug quantity is ultimately the district court's responsibility, our previous remand made clear that Correy was entitled to a PSR which provides assistance to the court and notice to the defendant of the evidence to be used to determine drug quantity. In this case, we specifically ordered that the PSR include references to the trial record. It did not. This failure to comply with our order justifies vacatur and remand, just as it did when we last considered this case in 2005.

The other appellants' PSRs suffer the same infirmity. But these appellants have not explicitly challenged the preparation of their PSRs on appeal. Nonetheless, all defendants challenge the district court's fact-finding process. Our discussion of each other appellant's hearing will show how the district court's error was compounded by the fact that sentencing hearings were permitted to proceed using PSRs that were not prepared in accordance with our specific order.

b. Credibility and drug quantity

Though the failure to prepare a proper PSR could alone support vacatur, we proceed to analyze Correy's underlying claim about credibility determinations. Performing this analysis will reveal that the PSR defect was not a harmless technical error. Further, this analysis will allow us to specifically address other claims raised by Correy on appeal.

At Correy's resentencing hearing, the court acknowledged that it had the responsibility to determine what drug quantity was attributable to Correy. The court also considered the applicability of a weapons enhancement. The government read an excerpt from the testimony of José Vélez-Román ("Vélez") implicating Correy in the possession of weapons. Then Correy him-self, who was acting "pro se with the assistance of counsel," addressed the relationship between both issues and the question of credibility:

THE DEFENDANT: Thank you. Now he made allegations of weapons for me for an enhancement. As a Judge, do you find anything wrong with his accusations?

THE COURT: He's not making an accusation. He's making a statement. He read from the transcript. There's record support that you were in control of weapons.

THE DEFENDANT: Have you read the entire transcript, Your Honor? THE COURT: I read whatever is important to this case.

THE DEFENDANT: The Appellate Court has told you to familiarize your-self.

THE COURT: I went and checked specific transcripts that appear to this case, because I don't have to make any decision as to guilt or innocence. I have to determine how much of these drugs are attributable to you, and that's the issue.

THE DEFENDANT: Now, when another witness contradicts that testimony, that's not important?

THE COURT: Well, I have record support. I have to find record support to determine whether — how much of that conspiracy, how much drugs are attributable to you. That's the issue. THE DEFENDANT: Wouldn't it be important to know if there was testimony to contradict?

THE COURT: Well . . . Mr. Correy, I'm going to show you something. In this case, the leader of this conspiracy was Israel Pérez Delgado, and you knew him, and you interacted with him and he was the leader of this conspiracy, which in a way shows that you knew more about the conspiracy than other people; you were acquainted with the extent of the conspiracy. Now in the testimony of Israel Pérez Delgado . . .

[The court proceeded to read from the record testimony excerpts from P6rez and Martinez.]

So it appears that this defendant committed an offense involving more than 150 kilos of cocaine. That's what it takes to meet the threshold of the sentence. Let me finish. I want to put this clearly on the record.

So the total amount of cocaine that can be directly attributable to this defendant John Correy is 908 kilograms. But let me go further.

Your sentence must be based on the 12,000 kilos of cocaine that can be attributable to the conspiracy as a whole, about 11,000 to 12,000 kilos. But it is enough if I find that you can be attributable to 908 kilograms. But I have to consider the whole conspiracy. That's the underlying conduct. That's what the Court of Appeals said. For the whole sentencing, this is the whole conspiracy you look at, and from that, saw how much you were responsible or reasonably attributable to you.

I'll hear you.

THE DEFENDANT: For the record, what you just read to me word for word is the Government's Sentencing Memorandum.

THE COURT: Yes, I did.

THE DEFENDANT: You asked the Government to provide this for you for sentencing purposes.

THE COURT: It's in the transcript.

THE DEFENDANT: But the Appellate Court wants you to familiarize yourself with the record, not have the Government provide you with the answer.

THE COURT: You have to find the truth. The truth is in the transcript, and I have to address myself to that; to the transcript that supports this finding.

THE DEFENDANT: That's why it's important for you to read the entire transcript, Your Honor, very important.

THE COURT: Let me say this: The Court of Appeals sent this case back for me to show you why I should give you a lesser sentence.

Why are you entitled, if you're entitled, to a lesser sentence? That's the whole issue in this case. Because the appeal was affirmed, your conviction was affirmed. So I don't want to go over anything that the Court of Appeals already ruled. Because I'm bound by that and you're bound by that.

The Court says, send the case back for you to show me — and I'll give you the opportunity to show me — why should you receive a lesser sentence under the advisory guidelines under the Booker regime, and that's the issue.

THE DEFENDANT: There's important issues in the transcript. He told you José Velez said I was present and doing things. They presented Thomas Martinez as truth. Now they take you to look at José Velez, and without you knowing the record, you don't know that Thomas sat there and contradicted that.

[PROSECUTOR]: With all due respect, these are all matters of credibility. The jury heard all the evidence and heard all the witnesses, and they believed it, and the Court of Appeals has already affirmed all of those issues.

The only issue pending before this Court is the issue of the sentencing based on an assessment given by a preponderance of the evidence, and that is what Booker Fanfan decision is based on, and that's what the Court of Appeals wants you to assess. That's the only issue before the Court.

[The Defendant and Defense counsel continued to argue the issue.]

THE COURT: Let me say this: With all those credibility issues, the jury found you guilty, and the Court of Appeals addresses those issues and says although you were acquitted of murdering — the jury acquitted you on that. So it was a very discriminating jury. They just went through the whole evidence and said well, he's not guilty of murdering this one individual, but he's definitely guilty by a unanimous verdict of drugs.

THE DEFENDANT: They also found me not guilty of the drug count also. To say if they didn't find me guilty of the 36 keys, that they now would find me guilty of 150 keys —

THE COURT: You had your chance before the jury. Now it's the judge who has to do this finding based upon the record.

THE DEFENDANT: For the record, why do you give Israel Pérez Delgado's testimony credibility?

THE COURT: Because I found that has not been contradicted, what I just read on the record.

You still keep on trying to deviate the issue to the transcript. You passed that stage. I'm . . . telling you now for the fourth time, giving you all the opportunity in the world, give you all day to show me why you should get a lesser sentence.

Because the Court of Appeals didn't say you're free to go, walk away. You're still guilty, and the issue is the amount of drugs. Show me why you should receive a lesser sentence under the new discretionary guidelines.

. . . .

THE DEFENDANT: A credibility issue is a very heavy thing to weigh here. That's why the transcripts are very important. You want to base me on what they told you Israel Pérez said. They already called him dubious, his credibility was. They said you must look beyond Thomas Martinez. They said it here.

THE COURT: I'm making this finding more clearly. Regardless of that testimony, there's testimony that is uncontradicted that you were close to Israel Pérez Delgado, who was the leader; that you were with him on these occasions, that you were with Ralph Casas, who was another leader. In other words, I can also find, regardless of the transcript, regardless of the specific amounts, because of the extent of the conspiracy, the Court has to determine the extent of the conspiracy; also has to determine the interaction between the defendant and the leaders of the conspiracy.

And this conspiracy was a sophisticated drug conspiracy based principally in Puerto Rico, transporting large quantities of cocaine and heroin between September '92 and March '95. That's the extent of the conspiracy. The leader of the organization was Israel Pérez Delgado, who according to the indictment had 60 subordinates. This was a 60-defendant indictment. He coordinated working together the members of the organization transporting drugs from Puerto Rico to the Dominican Republic and Panama into Miami International for ultimate distribution in New York.

In total, the organization smuggled approximately 9,000 kilograms of cocaine and approximately 1400 grams of heroin, and this is taken from 356 [F.3d 104] at page 119. That's the first appeal. That's the first appeal. There were two appeals because the appeal [sic] was severed.

By doing that, this is what's called related conduct, by saying that the extent of the conspiracy for three years, that you were acquainted with the leaders, so it is reasonable to presume that you can be even held accountable for the whole conspiracy, but I think what I said before, I think I said you were accountable for 900 kilograms.

I'll hear you again. Show me why you should get a lesser sentence. This is the last time I'm giving you this opportunity.

THE DEFENDANT: Again, in that Sentencing Memorandum that you read word for word, I was acquitted of 36 keys. You still came and used them.

THE COURT: The Government is asking me to find you for the amount of — that's not correct. The Government says you were responsible for —

[PROSECUTOR]: May we address it?

THE COURT: — for 908 kilos, and I say that — I could find for that, but I also could find 840 kilos. Anything beyond 150 is important. Whether it's 500 or 600 or 800 doesn't matter. It doesn't affect the sentence.

And this is well, well beyond and above 150 kilos isn't it?

Anything else?

THE DEFENDANT: Did you read my motion in opposition of the Government's Sentencing Memorandum?

THE COURT: Yes, I read it.

THE DEFENDANT: Did you check the trial transcripts about my objections?

THE COURT: Counsel, you're not here to ask questions from me. You're here to respond to my questions.

(Emphasis added). The court later concluded that "The record is clear that [Correy] easily handled or could reasonably foresee that the amounts of this conspiracy was more than 150."

This lengthy and unusual exchange confirms that the sentencing court was under the same misapprehension of its responsibilities at Correy's hearing as it was at Pizarro's hearing. At his hearing, Correy correctly read our prior opinion as requiring the sentencing judge to conduct an independent credibility assessment. The government and the court misread our opinion and continued to rely on the jury's verdict. Of course, as we explained above, the jury verdict showed a credibility problem, and did not resolve credibility nor make any findings with respect to overt acts. Thus, Correy attempted to direct the court's attention to contradictions in support of his credibility argument. But the court refused to consider Correy's arguments.

As we explained above, we will not use the clear error standard of review to assume that these witnesses are credible where the court declined to make such a finding. Thus, we must, only for the purposes of the present appeal, disregard their testimony. On appeal, the government has provided us with the record cites it alleges support Correy's drug-quantity determinations. These citations point only to the testimony of Pérez, Martinez, and Velez — in fact, all but 12 kilograms depends solely on the testimony of the leaders Pérez and Martinez. Since we cannot assume these witnesses were credible in the absence of a district court finding on the issue, we must vacate Correy's sentence and remand for resentencing.

Though the government has not raised this argument on appeal, the sentencing court also appeared to find that Correy was responsible for much more than 908 kilograms, by virtue of his membership in a large, long-running conspiracy, and by virtue of his familiarity with the conspiracy leaders. As detailed below, the sentencing judge recited similar general findings as to the other defendants. We cannot affirm on such a basis. First, the sentencing court's findings do not reveal to what extent its conclusion of familiarity is based on testimony of Martinez and Pérez. Second, such a finding is too generic and goes against our instruction to base individualized drug finding on a review of the record. It may well be that Correy could foresee a much larger drug quantity than he was specifically involved in handling, but such a foreseeability conclusion cannot merely be based on general findings about the conspiracy. See United States v. Cruz-Rodriguez, 541 F.3d 19, 32 (1st Cir. 2008) ("When making the individualized finding of drug quantity responsibility, the court must not automatically shift the quantity attributable to the conspiracy as a whole to the defendant."); United States v. Willis, 49 F.3d 1271, 1274 (7th Cir. 1995) ("[I]t is highly questionable to leap from one person's knowledge that the organization is big to knowledge of its full scope. The district judge must take a closer look at this subject."). For example, the fact that the conspiracy involved many members and spanned several cities could just as easily limit Correy's foreseeability, rather than expand it. Thus, a finding that the quantity of drugs smuggled by the entire conspiracy was foreseeable to Correy must detail Correy's level of involvement so as to explain why the nature of the conspiracy or his relationship with the leaders of the conspiracy showed he could foresee a given quantity of drugs. For these reasons, Correy's sentence is vacated, and his case remanded for resentencing.

2. Nicolai

a. Credibility and drug quantity

Nicolai has framed his drug quantity arguments on appeal using procedural arguments, developing detailed Apprendi arguments, which we rejected above. While he does not directly argue that the district court judge erred in its credibility determinations, he does argue that the effect of the Apprendi error was to take the question of the credibility of Martinez and Pérez out of the hands of the jury. According to Nicolai, the questionable credibility of Pérez and Martinez, combined with the fact that the sentencing judge did not make any independent factual findings, shows that the alleged Apprendi error was harmful. Essentially Nicolai is arguing that credibility as to drug quantity findings should have been put to the jury, and that the failure to do so was harmful since the judge did not make any independent determinations. While we disagree with Nicolai that the guidelines drug-quantity question must be submitted to the jury, we agree with his basic argument that he was deprived of a chance to have credibility issues determined. We will not penalize him for raising his credibility arguments in the frame of an Apprendi challenge.

Further, Nicolai himself made credibility arguments to the sentencing judge in connection with the drug quantity, and his counsel raised the issue with respect to the weapons enhancement. In response, the court did not explain its credibility rulings, but rather cited to the clear error standard of review, apparently implying the court believed it would be affirmed regardless of its ruling. The court later cited snippets of trial testimony to support its conclusions, but provided no credibility findings. Thus, we treat Nicolai as having raised this argument before the district court and on appeal.

Next we evaluate whether this error was harmless with respect to each issue on appeal, the drug-quantity determination, the leadership enhancement, and the weapons enhancement.

Although Nicolai himself admitted at his sentencing hearing that he was involved in "big quantities of drugs," we cannot conclude that the district court's refusal to make credibility findings was harmless. In response to Nicolai's arguments about drug quantity, the government responds with a list of record cites supporting the sentencing court's drug-quantity determination. The number of incidents and quantity of drugs is staggering, but all but two of these quantities are based on the testimony of Martinez or Pérez. The two remaining quantities total only twenty kilograms. As explained above, we cannot assume the credibility of Martinez or Pérez. Similarly, as explained above, though the sentencing court also found that Nicolai could likely foresee many more kilograms being distributed because of his close relationship, or in the words of the sentencing court, "propinquity," with the conspiracy leaders and the sophistication of the conspiracy, we cannot affirm on this basis.

Though Nicolai has not directly raised the issue on appeal, we also note that he was denied, over his objection at sentencing, a PSR based on trial evidence as required by this Court, thus making meaningful fact-finding on drug quantity all the more difficult. Thus, Nicolai's guideline drug-quantity determination must be vacated and recalculated by the district court after a PSR has been prepared in compliance with our last opinion and after the sentencing court has made credibility determinations.

b. Leadership enhancement

Nicolai also argues that the district court erred by imposing a four-level enhancement on him for his leadership role in the conspiracy. Nicolai objects that the sentencing court found him to be a "manager" or "supervisor," but then applied the four-level enhancement for organizers and leaders rather than the three-level enhancement for managers and supervisors. Compare U.S.S.G. § 3B1.1(a) with id. § 3B1.1(b). The government argues that the sentencing court simply misspoke and intended the higher enhancement. But, we need not decide whether the sentencing court misapplied the guideline, since this highly specific factual determination also depends on credibility and should be reassessed on remand. We have said, "[a] role adjustment determination is 'heavily driven by the facts,' and 'absent a mistake of law, battles over a defendant's status . . . will almost always be won or lost in the district court.'" Olivero, 552 F.3d at 41 (quoting United States v. Sanchez, 354 F.3d 70, 74 (1st Cir. 2004)). As we have explained, here there was an error of law.

This error was not harmless. While the government points to other sources of record support for this finding besides the testimony of Pérez and Martinez, their testimony was clearly important in conducting this fact-specific inquiry. Thus, on remand the district court must also assess the factual disputes relevant to this issue.

c. Firearms enhancement

In contrast, the applicability of the firearms enhancement can be resolved now, without reference to the credibility of a challenged witness. "The sentencing guidelines provide for a two-level enhancement in drug offenses when a defendant possesses a dangerous weapon, including a firearm." United States v. Thongsophaporn, 503 F.3d 51, 58 (1st Cir. 2007) (citing U.S.S.G. § 2D1.1(b)(D). Nicolai again attacks the credibility of some witnesses' testimony at trial which links him to firearm possession. But, he does not contest the fact adduced in the PSR that he pled guilty to a state drug offense after he was arrested with a large quantity of controlled substances and a fully loaded handgun. In this case, that fact is enough to apply the enhancement. "According to the [guideline] commentary, '[t]he adjustment should be applied if the weapon was present, unless it is clearly improbable that the weapon was connected with the offense.'" Id. (quoting U.S.S.G. § 2D1.1(b)(1), cmt. n. 3). "For example, the enhancement would not be applied if the defendant, arrested at his residence, had an unloaded hunting rifle in the closet." U.S.S.G. § 2D1.1(b)(1), cmt. n. 3. But, "'when the weapon's location makes it readily available to protect either the participants themselves during the commission of the illegal activity or the drugs and cash involved in the drug business, there will be sufficient evidence to connect the weapons to the offense conduct.'" Thongsophaporn, 503 F.3d at 58 (quoting United States v. Corcimiglia, 967 F.2d 724, 727 (1st Cir. 1992)). Nicolai has raised no argument that the weapon he carried at the time of his arrest, together with drugs, was for use in anything other than drug trafficking. Thus, our precedent supports application of this enhancement. This particular enhancement can be applied on remand without regard to credibility determinations.

d. Criminal History Category

Nicolai's only rejoinder to the conclusion regarding the firearms enhancement over-laps with his next argument. He argues that the 1994 incident in which he was arrested with drugs and a firearm, to which he pled guilty in 1996, must be viewed consistently as either part of the charged conspiracy or not part of the charged conspiracy. If it is part of the conspiracy, it may be used for the weapons enhancement but not for his criminal history category determination, and vice versa. See U.S.S.G. § 4A1.2, cmt. n. 1; see also United States v. Cyr, 337 F.3d 96, 99 n. 1 (1st Cir. 2003) ("Under this application note, a prior sentence deemed to be relevant conduct (i.e., part of the res gestae of the instant offense) [under U.S.S.G. § 1B1.3] is disregarded for purposes of calculating the defendant's criminal history."). But, Nicolai's position that the government has double counted the offense is simply factually wrong. Nicolai's PSR noted this prior offense, but awarded zero criminal history points for this arrest, reasoning that it "is considered as part of the overt acts of the instant offense." Neither Nicolai nor the government contests this conclusion. And with good reason: the 1996 superseding indictment in this case charged Nicolai with conspiracy to possess with intent to distribute drugs spanning the time from September 1992 to March 1995. Thus, it was proper to use the offense, drug possession in 1994, as relevant conduct, and not in computing criminal history.

Nicolai also challenges another prior offense that did contribute one point to his criminal history score. Without this point, his CHC would have been III instead of IV. According to the PSR, the offense involved a January 1, 1994 arrest for possession of a forged instrument. Nicolai pled guilty to a misdemeanor offense on January 7, 1994, and received a $200 fine.

Nicolai asserts that this incident stems from his arrest at LaGuardia Airport with a large sum of cash and a forged driver's license while on his way to Puerto Rico. He argues that no charges were brought against him because of the ongoing investigation in this case. Nicolai's appellate brief asserts these facts are drawn from a DEA report, but fails to cite to the record. Nicolai's sentencing memorandum to the district court also gives no indication that this report is in the record. Accordingly, we will not consider these factual assertions.

The PSR simply states the crime, the punishment, and that "[d]etails as to this case were not available." It is true, as Nicolai argues, that the offense occurred during the conspiracy alleged in this case. But, it is also true, as the government argues, that the offense, bearing a forged instrument, may bear little relation to the activities charged in this case.

Determining whether Nicolai's prior offense is related conduct under U.S.S.G. § 1B1.3 or a severable and distinct offense requires a "'a fact-specific inquiry that involves more than just a consideration of the elements of the two offenses. Factors such as the temporal and geographical proximity of the two offenses, common victims, and a common criminal plan or intent also must be considered.'" United States v. Collazo-Aponte, 216 F.3d 163, 203 (1st Cir. 2000) (quoting United States v. Beddow, 957 F.2d 1330, 1338 (6th Cir. 1992)), vacated on other grounds by 532 U.S. 1036, 121 S.Ct. 1996, 149 L.Ed.2d 1000 (2001). The record is undeveloped regarding the circumstances surrounding Nicolai's arrest, and the district court made no specific findings on this issue. Thus, we presently have no basis for determining the relevant facts, including whether Nicolai was acting with the purpose of furthering the conspiracy when arrested on January 1, 1994. It is true that "[o]nce the government establishes the existence of a prior conviction, the burden shifts to the defendant to show that the earlier conviction was constitutionally infirm or other wise inappropriate for consideration." United States v. Serrano-Beauvaix, 400 F.3d 50, 54 (1st Cir. 2005) (quoting United States v. Barbour, 393 F.3d 82, 93 (1st Cir. 2004)). Nonetheless, in this case, since the relevant conduct inquiry is "fact-specific" and because we are already remanding the case for credibility determinations which may affect the relevant factual context, we decline to address this issue at this time and leave the issue for the resentencing court to address.

3. Flores

a. Credibility and drug quantity

Flores's sentencing hearing on September 8, 2006, was also affected by the sentencing court's refusal to consider credibility arguments. At that hearing, Flores's counsel noted our ruling that the drug-quantity determination was not resolved by the jury verdict and that the sentencing judge must make such a ruling on the basis of the entire record. Flores's counsel then advanced an argument that the sentencing court should read the record as not implicating Flores in testimony from Hector Martinez-Medina ("Martinez-Medina") about moving 3,500 kilograms through the airport. This particular quantity was essential to the finding that Flores was responsible for more than 150 kilogram of cocaine. The only other quantities on which the government relied were two twelve-kilogram quantities testified to by Velez and one thirty-kilogram quantity testified to by Martinez-Medina. In each of these latter cases, the testimony reflected that quantities of cocaine were delivered to Flores to bring through the airport, using his position there. At sentencing, Flores's counsel also attempted to attack the credibility of Velez, since he was unable to identify Flores at trial though he purported to deal directly with him. In response to these arguments attacking the government's drug-quantity evidence, the sentencing court stated:

The jury could have believed or disbelieved [the] witness, but apparently they did believe the witnesses because they found them guilty. In fact, there might be some discrepancy, but that's up to the jury. However, when I take a look at the transcripts and see this Defendant's participation, the amount of drugs that is attributed to him now, I think that the 38 Base Offense Level is sustained.

The sentencing court proceeded to cite a number of incriminating portions of the record, including the discussion about moving 3,500 kilograms. The district court used this incident in its drug finding without addressing Flores's argument regarding that incident.

On appeal, Flores renews his challenge to the credibility of Velez. He also argues that Martinez-Medina's testimony was confusing and unclear, especially considering that Martinez-Medina had no personal knowledge as to the weights of actual amounts moved.

As we found above, the district court's refusal to make credibility determinations goes directly against our recognition that the jury verdict did not pass upon the overt acts alleged in the indictment, which could not alone be used to establish drug quantity. Casas, 425 F.3d at 64 n. 56. The sentencing judge should have heard Flores on this issue and then made an affirmative finding on the issue. Flores's attack on these witnesses' testimony may be more attenuated than the pervading issue of the credibility of Pérez and Martinez, but we still believe he was entitled to consideration of the issue.

As with other defendants, the sentencing judge suggested that much more contraband could be attributed to Flores because of the sophistication of the conspiracy and Flores's "relationship with people who were important players in the conspiracy." As explained above, we cannot affirm on this basis. Accordingly, we vacate his sentence and remand for proceedings consistent with this finding and the below clarifications to the other outstanding issues.

b. Foreseeability of drug quantity

Flores argues that he should not have been held responsible for the 3500-kilogram quantity, even if the testimony of Martinez-Medina is accepted, because this drug quantity was not within the scope of his agreement. Flores argues that his participation was limited to moving bags through the airport, and asserts that 3500 kilograms of cocaine would have been physically difficult to move in that manner. He argues that no evidence shows he agreed to move that much. Flores argues that the evidence showed that he soon became no longer able to assist, and the conspiracy turned to another individual to get drugs through the airport. The government does not respond to this argument.

"[A] defendant may be held responsible for drug quantities involved in his 'relevant conduct.'" Laboy, 351 F.3d at 582 (citing U.S.S.G. § 1B1.3). "Such conduct may include a defendant's own acts or the acts of others: the sentencing guidelines provide responsibility for the acts of others in the case of 'jointly undertaken criminal activity', which includes any 'criminal plan, scheme, endeavor or enterprise undertaken by defendant in concert with others.'" Id. (quoting U.S.S.G. § lB1.3(a)(1)(B)). In a drug conspiracy, this includes, "'all reasonably foreseeable quantities of contraband that were within the scope of the criminal activity that he jointly undertook.'" Id. (quoting U.S.S.G. § 1B1.3, cmt. n. 2). But, "mere knowledge of another's activity is not enough to show liability." Id. at 582-83. Rather, "[t]he defendant is only responsible for foreseeable conduct within the scope of his own explicit or implicit agreement." Id. at 583. "Thus, the scope of relevant conduct is 'not necessarily the same as the scope of the entire conspiracy.'" Id. (quoting U.S.S.G. § 1B1.3, cmt. n. 2).

Flores argues that his situation resembles an illustration in the guideline commentary, which states that a defendant who agrees to assist in a single shipment of contraband is not liable for other shipments made by his co-conspirators. U.S.S.G. § 1B1.3, cmt. n. 2, illus. (c)(3). Another illustration explains that when a girlfriend of a drug dealer agrees to make a single delivery for him, but knows about his ongoing activity, the girlfriend is liable only for the single delivery. Id. at illus. (c)(5). Flores argues that he similarly was only involved in moving certain suitcases through the airport in San Juan, and did not undertake any broader criminal activity. Flores says that the testimony regarding the remaining 3,500 kilograms shows, at most, mere knowledge.

But, it is not entirely clear that Flores's involvement was limited to simply assisting on certain shipments. The district court did not directly address this issue, but did suggest that Flores's role was more involved when it suggested a high drug quantity could be attributed to him because of "his relationship with people who were important players in the conspiracy." As we have explained above, we cannot affirm on the basis of this broad statement, since its factual underpinnings are unclear. Rather, the district court should have determined the scope of Flores's specific agreement. U.S.S.G. § 1B1.3, cmt. n. 2 ("In order to determine the defendant's accountability . . ., the court must first determine the scope of the criminal activity the particular defendant agreed to jointly undertake. . . ."). While the above-quoted testimony suggests Flores did not ultimately assist in moving the 3500 kilograms, it is not clear as to whether Flores had agreed to assist in the broader conspiracy. Considering that we are remanding for further fact-finding, we need not categorically rule whether the 3500 kilogram quantity may be imputed to Flores simply on the basis of the relevant testimony from Martinez-Medina. See United States v. Valencia-Lucena, 988 F.2d 228, 235 (1st Cir. 1993) (stating that "[t]he district court's failure to more fully state the evidence upon which it based its finding of foreseeability . . . at the sentencing stage has frustrated this court's appellate task" and concluding "[w]e make no comment on whether the record supports a finding of foreseeability on the preponderance of the evidence; this task is for the district court."); United States v. Garcia, 954 F.2d 12, 16 (1st Cir. 1992) (stating "a sentencing court's determination that drug transactions did, or did not, form part of the same course of conduct as the counts of conviction is a predominantly factual finding"). On remand, the district court must determine not only whether the testimony implicating Flores for 3500 kilograms is credible, but also whether, based on the entire record, Flores's involvement was limited to bypassing airport security in certain narrow instances, or rather whether he had undertaken or agreed to the broader conspiracy's goals of transporting large quantities of cocaine through Puerto Rico.

c. Abuse of Position of Trust

Flores's janitorial position at the airport in San Juan allowed him to enter and leave the airport without inspection, thus facilitating his role in transporting cocaine through airport security. On this basis, the district court imposed an enhancement for abuse of a position of public or private trust. See U.S.S.G. § 3B1.3. Flores challenges this enhancement. He did not challenge the imposition of this enhancement at resentencing, so plain error review applies. "For us to vacate a sentence under the plain error standard, appellant bears the burden of establishing that '(1) an error occurred; (2) the error was clear and obvious; (3) the error affected the defendant's substantial rights; and (4) the error impaired the fairness, integrity, or public reputation of the judicial proceedings.'" United States v. Gonzalez-Castillo, 562 F.3d 80, 82 (1st Cir. 2009) (quoting United States v. Mangual-Garcia, 505 F.3d 1, 15 (1st Cir. 2007)).

In this instance, the government concedes plain error. We agree and commend the government for its forthrightness. Application note 1 to U.S.S.G. § 3B1.3 defines "position of public or private trust" as a position "characterized by professional or managerial discretion" and provides that the enhancement does not apply "in the case of an embezzlement or theft by an ordinary bank teller or hotel clerk because such positions are not characterized by the above-described factors." We have said that it was error for a district court to impose the enhancement simply because a defendant's job enabled her to commit the crime. United States v. Reccko, 151 F.3d 29, 32 (1st Cir. 1998) (reversing imposition of the enhancement on a switchboard operator at police headquarters). Rather, the sentencing court must first ask whether the defendant's position was a position of trust. Id. Like the switchboard operator in Reccko, Flores's job enabled him to commit the crime, but did not afford him any "professional or managerial discretion." See United States v. Parrilla Román, 485 F.3d 185, 190-92 (1st Cir. 2007) (applying Reccko to reach the conclusion that airport fleet service clerks did not occupy a position of trust). Though Parrilla Román post-dates sentencing in this case, its rule was clear from Reccko, so we find the sentencing court committed error that was plain. The resulting increase in Flores's guideline range clearly impacted his substantial rights unfairly. Therefore, as to Flores, we would independently vacate his sentence on this ground. Thus, on remand we order that this enhancement not be applied.

d. Minor Role Adjustment

"[T]he sentencing court's determination of a defendant's role in the offense is necessarily fact-specific." Santos, 357 F.3d at 142. Such a feature normally leads us to deferential review. Id. But, here, where we have identified a need for reexamination of the factual context in light of appellants' credibility arguments, this "fact-specific" feature counsels that we leave the applicability of this enhancement to be addressed on remand.

4. Casas

At Casas's sentencing hearing on September 12, 2006, defense counsel and the defendant himself attacked the credibility of Martinez. The sentencing court made explicit the same erroneous conclusion that it need not question credibility: "The fact that the Court of Appeals mentioned Martinez may or may not be credible, I think is answered of [sic] the same opinion." The sentencing court then proceeded to quote the above-quoted provision of our previous Casas opinion pertaining to the conspiracy-wide drug finding. Unfortunately, the government also contributed to this misunderstanding, when, at Casas's sentencing hearing, it said,

With regard to credibility, credibility was an issue for the jury. The jury did believe all the evidence in its totality as presented. And also the Court of Appeals has set the case for resentencing for the amounts of narcotics specifically. In light of Booker and Fanfan, not specifically to address credibility issues.

The court accepted this view of the remand, and stated later in the hearing,

Now, I am — again, am repeating that the Court of Appeals sent this case back for one purpose only — for only one issue — and that is to determine whether Angel Casas can convince the Court to impose a lesser sentence than the Guidelines now provide. That's the issue. Give me a specific reason why he should get a lesser sentence.

As we explain above, this is not a fair reading of our previous opinion, which called for the district court to conduct credibility determinations in light of the entire record before proceeding to resentence in an advisory guideline regime. Even after these statements, defense counsel tried once more to attack the credibility of Martinez, and once more the court interrupted and repeated the portion of our prior opinion concerning whether the Apprendi error was harmless.

This refusal to make a credibility determination was not harmless. The only record evidence offered by the government at the resentencing hearing, and now on appeal, that implicates Casas in specific drug quantities came from the testimony of Martinez. As Casas emphasizes in the present appeal, some of this testimony implicated individuals who were entirely acquitted by the jury. Of course, this fact does not preclude the sentencing court from deciding to use the evidence. See, e.g., United States v. Marti-Lon, 524 F.3d 295, 302 (1st Cir. 2008) (affirming sentencing judge's use of acquitted conduct where sentencing judge made explicit finding by preponderance of the evidence). But here, the record is clear that the sentencing court failed to make an independent determination that Martinez's testimony was credible.

The sentencing court also based its finding on reasoning that Casas "could reasonably have foreseen the amount of drugs involved in the conspiracy" because the organization was "sophisticated" and because of the "propinquity" between Casas and the leaders. The court gave evidence of propinquity by citing Martinez's testimony. But, this testimony was subject to Casas's credibility attack. Thus, the court's refusal to consider credibility was also not harmless as to its generic foreseeability findings, and we must vacate Casas's sentence and remand for resentencing.

This error was again exacerbated by the failure to prepare a PSR that complied with our prior orders. At resentencing, Casas also challenged his PSR's use of overt acts from the indictment. In response to his objection, the Probation Office filed an addendum expressing its position that the drug amount determination was based on the record of trial as reflected in the government's sentencing memorandum. But this is not what we ordered. The PSR should not incorporate a party's sentencing memorandum by reference, but rather, make its own drug-quantity determination based on the evidence, which, in this case, we specifically required to be based on the trial testimony

On remand, Casas will have the benefit of a PSR based on trial evidence, so that he may prepare meaningful credibility arguments, which will be evaluated and resolved by the district court. 5. Bonilla

Bonilla does not directly raise a credibility issue on appeal, rather he frames a similar argument as a challenge to the adequacy of the district court's explanation, required under 18 U.S.C. § 3553(c). Specifically, he argues that the sentencing court "failed to articulate it's [sic] reasoning how a Base Level Offense of 38 was derived" and "adopted the conclusion reached by the Probation Officer, as memorialized in the Presentence Report." Bonilla argues this was error because "the quantity of drugs attributed to him was highly disputed at the resentencing hearing." This is essentially the argument we have approved above: we cannot review the district court's credibility determinations because the district court did not explicitly make or express such determinations. Just as with Nicolai, we will not ignore Bonilla's argument simply because he has framed his credibility attack as an attack on the court's sentencing procedure.

The transcript of Bonilla's sentencing, held on September 12, 2006, clearly shows the sentencing judge deferring to the supposed credibility assessment of the jury. Defense counsel attacked the jury verdict as based on "junk from the snitches," and the court responded

I don't think we can rehash the facts presented to the jury. The jury found the Defendant guilty of conspiracy to possess and distribute huge amounts of drugs. The Court [of] Appeals affirmed the convictions, and now sends the case back to me just to determine what amount of those 9,445 kilos are attributable to your client.

Later, when the defendant was attacking the version of events presented at trial, the court said, "Mr. Bonilla, this is not the time to argue that. That was the — that was to be argued before the jury, and the jury — heard by the jury, and the jury found you guilty." The court made its finding on the drug quantity issue after hearing a recitation of incriminating portions of the record from the government and without explaining its credibility determinations. Thus, the transcript shows what was made explicit in the other hearings: that the sentencing court felt its task was to determine drug quantity under the assumption that all trial testimony was true. This is contrary to our instruction, explained above, that the jury verdict did not establish the individual drug quantities. This error was not harmless, since the only drug quantity evidence cited by the sentencing court, or by the government on appeal, came from Martinez. Thus, we vacate Bonilla's sentence and remand for resentencing.

Even though our prior opinion indicated that a new PSR was prepared for Bonilla's first sentencing, the PSR used in the 2006 resentencing simply copies the overt acts from the indictment. In response to an objection from the defendant, the Probation Department indicated that the government sentencing memorandum supported its conclusion that Bonilla was responsible for more than 150 kilograms, and then left "this matter to the sound discretion of the Court." As we explained above, this is not what our remand order required. At Bonilla's third sentencing hearing, he should have the benefit of a PSR referencing trial testimony showing specific facts which underpin the Probation Department's determination regarding the drug quantities for which Casas is responsible, as the trial judge ordered almost nine years ago. 6. Pizarro

a. The PSR

At Pizarro's resentencing hearing, counsel objected that the PSR was basically identical to the prior PSR, was a rehashing of the indictment, and did not comply with the trial judge's order. Pizarro did not raise this argument on appeal, but attempted to incorporate all co-appellants' arguments by reference. As we stated in the last appeal, "'Adoption by reference . . . cannot occur in a vacuum; to be meaningful, the arguments adopted must be readily transferrable from the proponent's case to the adopter's case.'" Casas, 425 F.3d at 30 n. 2 (quoting United States v. David, 940 F.2d 722, 737 (1st Cir. 1991)). Pizarro argues that the arguments are readily transferrable because of the factual overlap of the underlying conspiracy count. We rejected this argument last time on appeal as inadequate. Id. We continue to subscribe to the position, and have not addressed other attempts by appellants' to incorporate arguments by reference. But, in the limited case of the failure to prepare proper PSRs, it is clear that all of the co-appellants' arguments are readily transferrable. We applied the trial judge's order to each appellant, and probation failed to prepare a proper PSR as to each. This is a global procedural failing. Thus, we consider Pizarro's claim on appeal.

Pizarro's PSR similarly relies on the overt acts alleged in the indictment and is inadequate for the reasons we explained above. As explained below, this procedural error is not harmless, but contributes to our decision to vacate Pizarro's sentence and remand for resentencing.

b. Drug quantity

Much of Pizarro's appellate brief appears to be copied directly from Nicolai's appellate brief. Included in this appropriation is the section of Nicolai's brief arguing that an Apprendi error occurred, and that the error was harmful because of the court's failure to make credibility determinations as to Pérez and Martinez. This argument does apply to Pizarro, whose drug quantity testimony was based, to some extent, on these two individuals' testimony. Thus, Pizarro's copied brief has successfully raised the credibility issue on appeal.

As we explained above, the district court's refusal to make credibility determinations was apparent at Pizarro's sentencing. Whether this error was harmless is a much closer question than for any other appellant. At the hearing, the court ruled:

There's no doubt in my mind, not at all, not a shadow of a doubt, that by a preponderance of the evidence that this Defendant handled in excess of 150 kilos of cocaine. Not only that, he — I attribute to him at least 4,200 kilos of cocaine, based on the record.

The court also supported this ruling by concluding that Pizarro could foresee much more because it was "wide conspiracy" and because of Pizarro's "propinquity" or "close relationship" with the leaders. For the reasons described above, we do not see this finding of "propinquity" as sufficient to support the district court's ultimate quantity determination.

The specific quantities in the record raise a more difficult question. First of all, the jury did convict Pizarro on the separate charge of possession of eighty-one kilograms. At sentencing, and on appeal, the government attributes to Pizarro a number of large quantities based on the testimony of Pérez, Martinez, Velez, and Martinez-Medina. At Pizarro's sentencing hearing, his counsel challenged the credibility of Pérez and Martinez as "contradictory" and "not credible." He also noted inconsistencies in the testimony of Velez. The district court refused to make these credibility determinations, so we will not use this testimony on appeal.

But, there is still the matter of Martinez-Medina's testimony that Pizarro had told Flores he had 3500 kilograms of cocaine to move. Counsel has not challenged Martinez-Medina's credibility on appeal. At the sentencing hearing, counsel only argued that Martinez-Medina's testimony did not show that 3500 kilograms had actually been moved. Counsel emphasized the technical difficulty in moving that much weight, but admitted that there were many other couriers.

This argument is somewhat unconvincing. If we assume Martinez-Medina's unchallenged testimony is credible, the court could reasonably find that Pizarro told Flores that Pizarro intended to transport 3500 kilograms of cocaine. From this finding, a court could reasonably infer that such a quantity was reasonably foreseeable to Pizarro and within the scope of Pizarro's agreement (even if not foreseeable to Flores). That Martinez-Medina did not actually see this much actual cocaine is irrelevant; the court could infer that Martinez-Medina was not aware of all the shipments which Pizarro foresaw. Thus, in this particular instance, our recent decision in Olivero, discussed supra at Section II.A.3, is more on point. 552 F.3d at 39-40. ("Rule 32(i)(3)(A) is explicit that the court may accept any undisputed portion of the PSR as a finding of fact. If the facts plausibly support competing inferences, as here, a sentencing court cannot clearly err in choosing one." (footnote omitted)).

As such, in normal circumstances, we would not call such an inference clear error. But, in this case, considering the paucity of factual findings regarding these shipments, and considering that this quantity was addressed only in the government's sentencing memorandum, and not in the PSR, and considering the district court's general refusal to consider credibility arguments, we are not convinced that Pizarro was given a full and fair opportunity to challenge the drug quantity evidence presented against him. Put another way, we have no reason to believe that the sentencing court actually drew the above-identified permissible inference, rather than improperly relying on the jury's verdict, as it did in every other defendant's case. Pizarro should have an opportunity to argue against such an inference before a fact-finder willing to consider his argument.

Thus, in this case, we will not affirm the drug-quantity determination based on a single report from Martinez-Medina that Pizarro mentioned 3500 kilograms where that report was not included in the PSR, where the judge did not make any specific findings regarding this quantity, and where the judge generally announced his refusal to question such facts. Rather, as with the other defendants, the sentencing court should, on remand, provide Pizarro with a PSR which identifies specific drug quantities. Armed with this PSR, Pizarro will be given a genuine opportunity to argue that the testimony was incredible, that it does not support the incriminating inference, and/or that it was not foreseeable. These findings will receive consideration by the sentencing judge, who will pass on each one, rather than rely on the jury's verdict.

c. Firearm enhancement

Pizarro also challenges the imposition of a firearm enhancement. This problem raises a similar situation with respect to the question of harmless error.

The government cites testimony from Martinez and Velez which shows Pizarro in possession of a firearm. It also cites to the PSR to attempt to implicate Pizarro in the use of firearms in relation to the acquitted murders. For the reasons explained above, we will not consider this evidence.

There nonetheless remains some basis for believing Pizarro could foresee the use of weapons. Pizarro did not seriously dispute that co-conspirators carried weapons. And, as we described above, Pizarro was convicted in connection with eighty-one kilograms. Considering the use of weapons by co-conspirators and Pizarro's role with respect to a large quantity of drugs, our case law permits the inference that the conspiracy's use of weapons would be foreseeable to Pizarro. Specifically, we have said, "[b]ecause firearms are considered 'common tools of the drug trade,' a codefendant's possession of a dangerous weapon 'is foreseeable to a defendant with reason to believe that their collaborative criminal venture includes an exchange of controlled substances for a large amount of cash.'" Thongsophaporn, 503 F.3d at 58 (quoting United States v. Bianco, 922 F.2d 910, 912 (1st Cir. 1991)). Even stripped of reliance on challenged testimony, the sentencing court might have fairly drawn this inference.

But the record gives no indication that the sentencing court actually drew such an inference. Rather, the record gives every indication that the sentencing court relied on the jury's verdict and even highly disputed acquitted conduct, without making any independent credibility determination. It is our law that "[i]f a district court makes an erroneous factual finding under the sentencing guidelines, yet there is enough evidence to support the alternative explanation for the court's finding, the error would be harmless and there would be no reason to remand to the district court when the result will be the same." United States v. Rodríguez, 525 F.3d 85, 108-09 (1st Cir. 2008) (emphasis added) (internal quotation marks deleted). Here, there was no supported alternative explanation. Considering the totality of the record, we are left with a "definite and firm conviction" that the district court improperly relied on the jury's verdict and challenged testimony. Cf. Cruz-Rodriguez, 541 F.3d at 32 n. 9 (stating, in the sentencing context, "[w]e 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.'" (quoting United States v. Brown, 298 F.3d 120, 122 (1st Cir. 2002))). Pizarro should have a real opportunity to challenge this inference by arguing the question of foreseeability to a fact-finder willing to consider his arguments. Thus, on remand, he will have a PSR which identifies the trial transcripts portions which support any conclusion that he possessed weapons or that weapon possession by co-conspirators was foreseeable to him. He will be able to make factual arguments attacking credibility and foreseeability, which the court will consider and resolve.

d. Leadership enhancement

Pizarro also challenges the two-point leadership enhancement he was given by the district court. As explained above, see supra Section II.B.2.b, we leave this fact-specific issue to be resolved on remand.

C. Remaining issues

Correy, Nicolai, Flores, Casas, and Bonilla also assert that their sentences were substantively unreasonable, especially in light of the lower sentences given to Pérez and Martinez. Since we are remanding for a new sentencing hearing, we need not reach this challenge. See United States v. Vidal-Reyes, 562 F.3d 43, 48 (1st Cir. 2009). Some defendants also raise various procedural issues and charge that the district court gave presumptive weight to the guidelines and failed to explain its sentence. Since there will be a new sentencing hearing, we also need not address these alleged procedural defects in the last hearing.

III. Conclusion

All appellants' sentences are vacated , and their cases remanded for resentencing consistent with this opinion. We take no view as to the ultimate sentences that should be imposed on remand. Nor does this opinion express any view on how the district court should resolve the appellants' credibility challenges. Cf. Rodríguez, 525 F.3d at 108 (cataloging situations where we have affirmed district courts' decisions to credit questionable testimony). We only insist that the sentencing court consider and resolve these credibility issues, as well as the other factual issues identified herein.

Vacated and Remanded.