In its Memorandum Opinion and Order denying a new trial, the district court reviewed the hearing testimony in detail and found that the children's recantations were not credible. The children did not recant until they resumed having contacts with their mothers and grandmother, who did not believe the abuse occurred and told the children they missed the imprisoned men. In these circumstances, the court found, "the combination of the influence from the unsupportive families, contact with the defendants by telephone and letters, being made aware of the lengthy prison sentences given to their uncles and having no outside support [after returning to their homes], pressured the children to recant their truthful testimony about being sexually abused by their uncles." The court found the evidence from Dr. Underwager's interviews not credible. In 1996, D.R.'s mother accompanied T.R. and D.R to the interview with Dr. Underwager. The 1999 interviews took place after the children were returned to their homes. Instead of recanting, R.R. told Dr. Underwager that Uncle Jess did things to her that were not right. The district court found that Dr. Underwager used suggestive questioning and told the children he was there to help get their uncles out of prison. Finally, the court emphasized that the children's trial testimony "is supported by the medical evidence in the case, while their recantations are not." Our prior opinion summarized that powerful medical evidence and bears repeating:

Dr. Kaplan [the pediatrician who examined the children] reported to DSS his medical findings and what the children had said about sexual abuse. J.R. told Dr. Kaplan, "Uncle Jess hurt me," pointing to her left labia; Dr. Kaplan found a recent bruise or contusion consistent with that kind of abuse. L.R. had "a fairly acute injury" on the right side of her labia majora which "really hurt her." R.R. told Dr. Kaplan, "I have a bruise where my uncle put his private spot," and Dr. Kaplan found a sagging vagina and a scar on her anus. Dr. Kaplan found that T.R. had "obvious trauma and contusion ... and very, very much tenderness" on her labia majora; T.R. told him, "Uncle Jess hurt me there." ...

... Dr. Robert Ferrell conducted a colposcopic examination of the five victims. Dr. Ferrell found "very significant" damage to R. R.'s hymenal ring and tearing in her anal area consistent with anal intercourse. He noted a "whole constellation of findings" indicating L.R. had been abused—damage to her hymenal area, furrowing on either side of her vagina, chronic irritation or trauma, and "clue cells" that are "known to be sexually transmitted." To Dr. Ferrell, a scar on J. R.'s hymen where a tear had healed was an "important finding," while T.R.'s "hymenal ring was essentially gone," the entire area was irritated, and she had furrows in her vagina. Infant F.R. had "tearing and scarring of the anal mucosa."

Defendants' medical expert, Dr. Fay, admitted that the reported hymenal scarring on L. R., R. R., and J.R. "certainly ... leads you to think about sexual abuse," and that "a labial injury... is a very significant finding" of abuse. In its rebuttal, the government called Dr. Randall Alexander, a member of the Board of Governors of the National Committee to Prevent Child Abuse. Dr. Alexander testified that it takes considerable force to inflict labial injuries like those exhibited by three of the victims. "It's rare to see one [in young girls] and to see three of them show up is just ... rareness to the third power."

United States v. Rouse, 111 F.3d at 565-66. We rejected a prior attempt to blame these injuries on inter-child sexual activity. See Hubbeling, 288 F.3d at 367. Defendants' attempt at the hearing to explain away the injuries of T.R., J.R., and L.R. as the result of previously unreported sexual abuse by a twelve-year-old male cousin was equally unpersuasive.

We view with suspicion motions for new trial based on the recantation of a material witness because "[t]he stability and finality of verdicts would be greatly disturbed if courts were too ready to entertain testimony from witnesses who have changed their minds, or who claim to have lied at the trial." United States v. Grey Bear, 116 F.3d 349, 350 (8th Cir.1997). This skepticism "is especially applicable in cases of child sexual abuse where recantation is a recurring phenomenon," particularly "when family members are involved and the child has feelings of guilt or the family members seek to influence the child to change her story." United States v. Provost, 969 F.2d 617, 621 (8th Cir.1992), cert. denied, 506 U.S. 1056, 113 S. Ct. 986, 122 L. Ed. 2d 139 (1993).

To receive a new trial, the movant must show that "the newly discovered evidence is of such a nature that, in a new trial, [it] would probably produce an acquittal." United States v. Papajohn, 212 F.3d 1112, 1118 (8th Cir.2000) (quotation omitted). When the claim of newly discovered evidence is based on a recantation, the district court must first determine whether the recantation is credible. In this regard, "the real question ... is not whether the district judge believed the recantation, but how likely the district judge thought a jury at a second trial would be to believe it." Grey Bear, 116 F.3d at 350. Our review of this credibility finding for clear error is extremely deferential. See Grey Bear, 116 F.3d at 351. We review the denial of the new trial motion for a clear abuse of discretion. See Papajohn, 212 F.3d at 1117-18.

After reviewing the record as a whole, we conclude that the district court's credibility findings are not clearly erroneous and the denial of the new trial motion was not a clear abuse of discretion. By the time of the evidentiary hearing, the children had been living with their mothers for at least two years, within walking distance of their grandmother's home. These women never believed the children's accusations, and testified on the defendants' behalf at trial. The children knew their grandmother and mothers missed the defendants. The children saw letters written by the uncles from prison and spoke to the men by telephone. Family members drove the children to interviews by Dr. Underwager, whose stated purpose was to free their uncles from lengthy prison sentences. The district court's finding that the recantations were the product of family pressure and therefore not credible is overwhelmingly supported by this record. Combined with the defendants' failure to refute the powerful medical evidence of abuse at trial, this finding fully justified the court's conclusion "that there is no reasonable probability that the recantations would produce an acquittal if a new trial were held." Accordingly, the district court did not abuse its discretion in denying the defendants' joint motion for a new trial.

Defendants argue that the prosecution suppressed materially favorable evidence from the defense in violation of Brady v. Maryland, 373 U.S. 83, 83 S. Ct. 1194, 10 L. Ed. 2d 215 (1963). First, they claim that the government suppressed notes taken by the children's foster parent, Donna Jordan, relying on an FBI agent's pretrial 302 Report saying, "Donna Jordan ... made notes when the children have told her things and those notes will be made available at a later date." Defendants argue they never received these notes, despite the prosecutor's statement at trial that any notes would have been produced as "302 type material." This claim was not presented to the district court. The defendants failed to prove that the notes ever existed. Moreover, Jordan testified that the children never denied that abuse occurred, in which case her notes would have corroborated the claims of abuse. Thus, defendants can only speculate that the notes might have contained material exculpatory information. On this record, defendants failed to establish a Brady violation, much less plain error by the district court. See, e.g., United States v. Keltner, 147 F.3d 662, 673 (8th Cir.), cert. denied, 525 U.S. 1032, 119 S. Ct. 574, 142 L. Ed. 2d 478 (1998).

Second, defendants argue that the government suppressed its knowledge that the children were testifying falsely at trial. The district court rejected this contention based on its finding that the children's recantations were not credible. We agree. A finding that the children did not testify falsely at trial refutes a claim that the government knew the testimony was false. See United States v. Zuno-Arce, 339 F.3d 886, 891 (9th Cir.2003), cert. denied, 540 U.S. 1208, 124 S. Ct. 1483, 158 L. Ed. 2d 132 (2004). Finally, defendants argue that the government became aware of the children's post-trial recantations before the defense, and its suppression of this information allowed "a significant level of unfairness ... to seep into these proceedings." Any knowledge gained by the prosecution after the trial is irrelevant to a Brady claim. See United States v. Kern, 12 F.3d 122, 126 (8th Cir.1993). In any event, the record neither provides a factual basis for this assertion nor establishes any prejudicial unfairness.

Defendants argue that the district court erred in permitting Eva Cheney, the children's court-appointed guardian ad litem, to testify at the hearing on the motion for new trial. Defendants contend that attorney Cheney's testimony without a waiver by the children violated the attorney-client privilege because Cheney told them she was their lawyer and served as their lawyer during the trial. The district court overruled this objection on the ground that "the Court's intention was to create a guardianship for purposes of the trial .... As far as the Court was concerned then and now, there was no attorney/client relationship established." Whether an attorney/client relationship existed is a finding of fact we review for clear error. See State v. Catch The Bear, 352 N.W.2d 640, 645-46 (S.D.1984). On this record, the district court's finding was not clearly erroneous. In addition, defendants fail to identify any testimony by Ms. Cheney that disclosed a confidential communication protected by the privilege.

Defendants further argue that Ms. Cheney's testimony violated 18 U.S.C. § 3509(h)(2), which provides in relevant part that "[a] guardian ad litem shall not be compelled to testify in any court action or proceeding concerning any information or opinion received from the child in the course of serving as a guardian ad litem." This ground was not asserted in the district court, so our review is for plain error. See Revels v. Vincenz, 382 F.3d 870, 877 (8th Cir.2004). The district court appointed Ms. Cheney guardian ad litem pursuant to 18 U.S.C. § 3509(h)(1), a statute that authorizes such an appointment "to protect the best interests of the child." Though Cheney was subpoenaed to testify at the evidentiary hearing, defendants fail to identify any portion of her testimony that was "compelled" within the meaning of § 3509(h)(2), nor have they established that her testimony was against "the best interests of the child[ren]." Thus, there was no plain error.

Defendants argue that the district court erred when it refused to consider the results of a September 1999 polygraph test tending to support the hearing testimony of D.R., a male child family member, that his trial testimony consisted of made-up lies about what happened. The district court held an evidentiary hearing to determine whether this scientific evidence meets the reliability standards of Daubert v. Merrell Dow Pharmaceuticals, Inc. 509 U.S. 579, 113 S. Ct. 2786, 125 L. Ed. 2d 469 (1993).3 Weighing the conflicting testimony of the polygraph examiner and the government's polygraph expert, the court found that the test did not meet the standards of any "accepted polygraph testing procedure," that the circumstances surrounding the examination "further undermine its reliability," and therefore that "the polygraph evidence in this case is not reliable enough to determine the truthfulness of D.R.'s testimony."

On appeal, defendants argue at length that the court misapplied the Daubert standards as they relate to polygraph testing. We do not believe the district court abused its discretion in declining to consider the polygraph evidence for the reasons stated. But in any event, the court was the ultimate fact-finder regarding the credibility of D.R.'s testimony at the evidentiary hearing. The court learned the results of the polygraph testing at the Daubert hearing and found it unreliable. Thus, the court's Daubert ruling was at most harmless error—had the court admitted the polygraph results under Daubert, this unreliable evidence would not have altered the court's finding, based on hearing D.R.'s live testimony at the evidentiary hearing, that the child's recantation was not credible.

Finally, defendants argue that alleged cumulative errors, including the contentions raised and rejected on direct appeal, see Rouse, 111 F.3d at 566-73, warrant a new trial. This contention is without merit. There was no error at these post-trial motion proceedings, and we decline to reconsider our earlier rulings. Accordingly, the district court's February 10, 2004 order denying defendants' joint motion for a new trial is affirmed.