Gary Richard Whitton v. Secretary, Florida Department of Corrections

• Court: Court of Appeals for the Eleventh Circuit
• Decided: May 6, 2025
• Docket: 23-10786
• Status: Unpublished

Opinion

USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 1 of 72

[DO NOT PUBLISH] In the United States Court of Appeals For the Eleventh Circuit

____________________

No. 23-10786 ____________________

GARY RICHARD WHITTON, Petitioner-Appellant, versus SECRETARY, FLORIDA DEPARTMENT OF CORRECTIONS,

Respondent-Appellee.

Appeal from the United States District Court for the Northern District of Florida D.C. Docket No. 4:15-cv-00200-RH ____________________ USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 2 of 72

2 Opinion of the Court 23-10786

Before ROSENBAUM, NEWSOM, and BRASHER, Circuit Judges. PER CURIAM: In 1992, Petitioner-Appellant Gary Richard Whitton was convicted of murder and, based on a unanimous jury recommen- dation, sentenced to death. The victim was found stabbed to death in a blood-spattered motel room in Destin, Florida. A motel clerk testified that Whitton helped the victim with check-in, and the clerk saw Whitton’s car parked outside the victim’s room later that night. At trial, the State presented the motel clerk’s testimony, blood-spatter evidence from Whitton’s boots, and circumstantial evidence that Whitton had robbed the victim and paid outstanding bills with the money. Plus, two jailhouse informants, Jake Ozio and Kenneth McCullough, testified that Whitton confessed to the mur- der. But years after Whitton’s conviction, both Ozio and McCullough sought to recant their testimony. And postconviction counsel collected additional evidence that could have been pre- sented as mitigation in the penalty phase of Whitton’s trial. That evidence included records, testimony from other family members and childhood acquaintances, and a more comprehensive psycho- logical evaluation. So Whitton filed a petition for state postconvic- tion relief and, after the Florida Supreme Court denied relief, a fed- eral habeas petition. After an evidentiary hearing, the district court denied relief. It granted a certificate of appealability on three claims, which we expanded to include one more (four total). Those claims assert that USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 3 of 72

23-10786 Opinion of the Court 3

(1) the State presented Ozio’s false testimony, in violation of Giglio;1 (2) appellate counsel was constitutionally ineffective for failing to investigate McCullough’s desire to recant; (3) trial counsel was con- stitutionally ineffective at the mitigation phase; and (4) the State improperly commented on Whitton’s invocation of his right to si- lence, in violation of Doyle. 2 After careful consideration of Whitton’s claims, and with the benefit of oral argument, we affirm the district court’s denial of Whitton’s petition for habeas corpus on all grounds. I. BACKGROUND A. Factual Background 1. Criminal Offense, Investigation, and Prosecution On the morning of October 10, 1990, James Maulden 3 was found stabbed to death in his room at the Sun and Sand Motel in Destin, Florida. Whitton v. State (“Whitton I”), 649 So. 2d 861, 863 (Fla. 1994) (per curiam). Maulden’s skull was fractured, and he sus- tained three fatal stab wounds to the chest, as well as stab wounds to his shoulder, cheek, neck, scalp, and back. Id. Blood splatter spanned the floor, furniture, walls, and ceiling of the motel room,

1 Giglio v. United States, 405 U.S. 150 (1972).

2 Doyle v. Ohio, 426 U.S. 610 (1976).

3 The State, as well as certain record documents, refer to the victim as

“Mauldin.” We use “Maulden” because Whitton and the district court use that spelling. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 4 of 72

4 Opinion of the Court 23-10786

consistent with “significant bleeding” and “violent combat” for a thirty-minute period. Id. at 866. A motel clerk told police officers that Maulden had checked in the day before, after arriving in another man’s car. The man had helped Maulden check in because Maulden was intoxicated. Id. Af- ter the man accompanied Maulden to his room, the man left. Id. But the clerk saw someone get into the man’s car, parked near Maulden’s room, later that night. Id. The clerk gave officers the car’s license-plate number. That led them to Whitton’s house in Pensacola. Id. The next day, officers visited Whitton’s home. He invited them inside, and they took Whitton to the police station for ques- tioning. Id. Whitton told the officers he knew Maulden from re- hab. He recounted that he had dropped Maulden off at the motel the previous afternoon. At first, Whitton denied having gone back to Maulden’s room that night. But Whitton later admitted he had returned there. He said that when he had discovered Maulden dead, he had fled in a panic. After three hours of questioning, Whitton invoked his right to remain silent. The officers then incarcerated him in the county jail. They seized his jeans and boots, as well as samples from the floor and seat of his car, to analyze for suspected blood stains. And they found $50 in cash in Whitton’s possession. Latent prints from miscellaneous items in Maulden’s motel room (an ice bucket, a wine bottle, a sandwich wrapper, and a pa- per bag) matched neither Maulden’s nor Whitton’s. But law USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 5 of 72

23-10786 Opinion of the Court 5

enforcement recovered Whitton’s prints from a different bag found near Maulden’s body. Whitton’s boots contained “medium velocity” blood “spat- ter” consistent with “a stabbing or a beating” and were targeted by “forceful bloodshed.” The blood spatter traveled “from top to bot- tom” (not bottom up) inside the boots. This contrasted with Whit- ton’s claim that blood from the motel seeped into his socks from “between his boot soles” and “uppers.” Whitton also said that he took his socks off and threw them out the window of his car and that he cleaned his boots when he got home. Over a year after Whitton’s arrest, the State offered a plea deal to second-degree murder. But the prosecutor withdrew the plea offer and pursued a first-degree murder conviction and death sentence. At trial, the State’s theory was that Whitton stabbed Maulden after robbing him of cash. The State offered the following evidence. On October 9, Maulden asked Whitton to drive him to Maulden’s bank. At the bank, Maulden—who was intoxicated— withdrew his account’s balance of $1,135.88. Then, when Maulden and Whitton arrived at the motel, Whitton wrote down a false li- cense-plate number on the registration. The motel clerk noticed and corrected it. The next day, on October 10, Whitton bought gas in Pensacola and paid his car registration fee and two utility bills, totaling about $228. And blood stains on Whitton’s boots and in his car matched Maulden’s blood type. The motel clerk also testi- fied to having seen Whitton’s car parked near Maulden’s room and USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 6 of 72

6 Opinion of the Court 23-10786

a man get into that car. According to the motel clerk, anywhere from ten minutes to two hours could have elapsed between when he first saw the car back at the motel and when he saw someone enter it and leave. The State also called two key witnesses—Jake Ozio and Ken- neth McCullough 4—who were incarcerated with Whitton at the Walton County Jail. Both testified that Whitton confessed to the murder. Ozio was an eighteen-year-old high-school student from Texas who was on spring break in Florida when he was arrested and jailed for burglary and possession of a short-barrel shotgun. Shortly before Whitton’s trial, Ozio was released on probation, and his firearm charge was reduced to a misdemeanor. At trial, Ozio testified that he overheard Whitton tell McCullough that he had “stabbed the bastard.” For his part, McCullough, who was serving a 15-year sen- tence for 7 felony convictions, corroborated Ozio’s account. He testified that he did not expect or receive favorable treatment for his testimony. But he did admit that he was a “close personal friend” of the prosecutor’s mother. Whitton testified in his own defense. He claimed that he gave a false license-plate number at the motel out of fear that he would be liable for any damage that Maulden caused. Whitton also

4 Portions of the record spell the last name as “McCollough.” We adopt Whit-

ton’s spelling: “McCullough.” USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 7 of 72

23-10786 Opinion of the Court 7

said that after he left the motel, he stopped at Maureen Fitzgerald’s house, and Fitzgerald told him that Maulden’s mother was looking for him. Whitton continued, testifying that he and Fitzgerald tried to call Maulden’s mother but could not reach her. Whitton claimed that he returned to the motel to tell Maulden that his mother was looking for him but fled after he found Maulden dead. He said he didn’t call the police because he was in shock and didn’t want to get involved. As to the bills, Whitton explained that Debra Sims, the former occupant of his house, had recently given him $200 to pay off utility bills incurred during her tenancy. The defense also called Shirley Zeigler, a DNA analyst from the Florida Department of Law Enforcement. Zeigler testified that although the blood on Whitton’s boots was Maulden’s blood type, the DNA matched neither Maulden’s nor Whitton’s. 5 Besides Whitton and Zeigler, the defense called James “Bill” Graham. Graham had interviewed the motel clerk shortly after the murder. Graham testified that the clerk had told him that only ten to fifteen minutes had gone by between his sightings of the car and the man entering it. The jury found Whitton guilty of murder and robbery. Whitton I, 649 So. 2d at 864.

5 The State retested the blood on Whitton’s boots in 2002. The inside of the right boot contained blood from a “mixture of two or more individuals” with the “major donor” matching “the DNA profile of James Maulden.” The test- ing also identified Maulden as a DNA contributor to one area on the left boot and a possible contributor to another area on the right boot. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 8 of 72

8 Opinion of the Court 23-10786

2. The Penalty Phase At the penalty phase, the State called three witnesses: McCullough, a parole officer, and a forensic pathologist. McCullough testified that Whitton said he had to kill Maulden be- cause Whitton was on parole and did not want Maulden testifying against him. The parole officer confirmed that Whitton was on parole for a robbery conviction at the time. The pathologist testi- fied that the initial blows did not leave Maulden unconscious and that Maulden had defensive wounds on his hand and arm. For its part, the defense called a mental-health expert, Dr. James Larson. He testified that Whitton had an IQ of 84 (low to normal), a history of alcohol abuse, and was physically and emo- tionally abused as a child. Whitton’s brother Royal and his aunt Ruth McGuinness testified that Whitton’s parents were alcoholics, and they abused and neglected him. The defense also called Renee Sims and Shirley George, acquaintances of Whitton, to testify that they trusted Whitton with their children and grandchildren. And Dorothy McGuire testified that Whitton had supported her hus- band’s rehabilitation program. The advisory jury unanimously recommended a death sen- tence. The sentencing court imposed the death penalty. In sup- port, it found five aggravating circumstances: (1) Whitton was on parole; (2) Whitton had a previous felony conviction involving the use or threat of violence; (3) Whitton committed the murder for the purpose of avoiding lawful arrest; (4) Whitton committed the murder for pecuniary gain; and (5) the crime was especially USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 9 of 72

23-10786 Opinion of the Court 9

heinous, atrocious, or cruel. The court also found several non-stat- utory mitigating factors. Those included Whitton’s history of child abuse, poverty, and neglect; Whitton’s alcoholism; Whitton’s low IQ and performance at a sixth-grade level; Whitton’s motivation to seek help for his problems; and Whitton’s employment and help of others. On direct appeal, Whitton argued (among other claims) that the State violated his rights under Miranda v. Arizona, 384 U.S. 436 (1966), and its progeny by commenting on Whitton’s invocation of his right to silence during the police interrogation. Whitton I, 649 So. 2d at 864–66. The Florida Supreme Court rejected that argu- ment. It found that there was “no reasonable possibility that the improper comment[s] contributed to Whitton’s conviction.” Id. at 864. It also rejected Whitton’s challenges to the heinous, atrocious, or cruel aggravator and avoiding-arrest aggravator. Id. at 867. So it affirmed Whitton’s conviction and sentence. Id. 3. Postconviction Developments The postconviction record includes several developments that Whitton argues undermine the validity of his conviction. First, during Whitton’s direct appeal, his appellate counsel received two letters from McCullough expressing his desire to re- cant his testimony. In the first, McCullough wrote that he would “sign a sworn statement” saying that (1) the prosecutor “made him a deal to testify”; (2) the prosecutor “told him everything to say” in his testimony; and (3) Whitton “did not teel [sic] him anything about the crime.” Whitton’s trial counsel eventually met with USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 10 of 72

10 Opinion of the Court 23-10786

McCullough. During that meeting, Whitton’s trial counsel appar- ently told McCullough that he “would help in [McCullough’s per- jury] prosecution.” So in his second letter, McCullough wrote that “he would only talk to” Whitton’s “appellate counsel and an inves- tigator.” Despite telling Whitton that she would “pursue” McCullough’s recantation, appellate counsel never contacted McCullough, and he died before the start of Whitton’s habeas pro- ceedings. Second, George Broxson, McCullough’s cellmate, testified that McCullough “made an offer to the state attorney’s office and the sheriff and worked out a deal” to keep his sex-related criminal charges from becoming widely known. Whitton’s habeas counsel later learned McCullough and the prosecutor’s mother, Inez Ad- kinson, were engaged to be married and moved in together after McCullough’s release. Third, in 2000, Ozio swore in an affidavit that his trial testi- mony was false. He affirmed that he did not hear Whitton confess to McCullough. According to Ozio, after officers told him that he would spend five years in Florida prison for his crimes, Ozio made up the confession to try to help himself. Ozio initially agreed to travel to Florida to testify in postconviction proceedings—even making it to the airport—before learning he could face prosecution for perjury because of contradictory statements. The State refused to provide assurances that Ozio would not be prosecuted. Ozio consulted a lawyer and then refused to board his flight, saying, “I do not want to be prosecuted; I’m not coming.” USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 11 of 72

23-10786 Opinion of the Court 11

Whitton sought to compel Ozio’s testimony through Wash- ington state court, where Ozio lived. The postconviction court cer- tified that Ozio was a crucial out-of-state witness. But the Wash- ington court declined to compel Ozio to travel to Florida. Instead, the Washington court concluded that the perjury charges against Ozio constituted undue hardship under state law. The court agreed to oversee a deposition or remote testimony. But at the State’s re- quest, the postconviction court excluded any out-of-state testi- mony. The court reasoned that allowing such a deposition to pro- ceed would thwart Florida’s perjury laws. But Ozio’s codefendant, Kevin Wallace, did testify in state postconviction proceedings. He corroborated Ozio’s sworn affida- vit’s account of their dealings with officers. Wallace explained that Ozio was somehow responsible for their lenient treatment in Wal- ton County. In response, Clayton Adkinson, the State prosecutor, testi- fied. Adkinson denied making any deals or promises for specific testimony. And he said that he “would be surprised” if any officers encouraged Ozio or McCullough to elicit a confession. Also in the postconviction proceedings, a cab driver testified that Maulden, intoxicated and still drinking, asked him for help finding a sex worker. After Maulden pulled out a large roll of bills to pay the fare, the cab driver warned him to be careful carrying so much cash around. The cab driver said he told Maulden, “[Y]ou’re just making yourself a target.” Another postconviction witness USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 12 of 72

12 Opinion of the Court 23-10786

added that Maulden told her he had been robbed by a sex worker days before his murder. B. Procedural History 1. State Postconviction Proceedings In 1997, Whitton filed a petition for state postconviction re- lief. See FLA. R. CRIM. P. 3.851. He later filed three amendments to that petition. Whitton’s last amendment contained 29 claims, and he filed it in November 2004. 6 As relevant here, Whitton claimed that (1) Ozio and McCullough’s false testimony violated his Gi- glio/due-process rights; (2) appellate counsel was ineffective by fail- ing to investigate McCullough’s desire to recant his false testimony; and (3) trial counsel was ineffective at the penalty phase. Whitton v. State (“Whitton IV”), 161 So. 3d 314, 322–34 (Fla. 2014) (per cu- riam). The state postconviction court summarily denied eleven claims and conducted an evidentiary hearing on the remaining eighteen claims. Id. at 321–22. To support his penalty-phase mitigation arguments, Whit- ton presented the report of Dr. George Woods, a neuropsychiatrist. Dr. Woods diagnosed Whitton with Fetal Alcohol Spectrum

6 Also, in summary dismissals, the Florida Supreme Court twice declined to

intervene to prevent the State from re-testing Whitton’s boots. See, e.g., Whit- ton v. State (“Whitton II”), 824 So. 2d 171 (Fla. 2002) (denying mandamus to prevent retesting); Whitton v. State (“Whitton III”), 838 So. 2d 560 (Fla. 2003) (dismissing nonfinal-order appeal to prevent retesting). USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 13 of 72

23-10786 Opinion of the Court 13

Disorder, brain damage from left frontal-lobe impairment, alco- holism, and Post-Traumatic Stress Disorder. Whitton also called two of his childhood teachers, Max Bo- vee and Irene Erickson. They described the abuse and neglect to which Whitton’s parents subjected him and his siblings. Erickson testified that the neglect Whitton endured was the worst she had seen in 28 years of teaching. Similarly, Whitton called his aunt and uncle, Kathy Robinson and Charles Jesmer, and foster parent, Lois Langworthy. They also described the child abuse and neglect, as well as Whitton’s brother Michael’s “mysterious” death. Beyond these witnesses, Whitton presented affidavits corroborating the abuse from other teachers, siblings, and friends. The state postconviction court denied relief. And the Flor- ida Supreme Court affirmed. Whitton IV, 161 So. 3d at 322. Beginning with the Giglio claim, the court found that Whit- ton “failed to demonstrate that” Ozio’s testimony “was false” be- cause “Ozio refused to testify and Whitton did not seek to have Ozio’s affidavit submitted into evidence.” Id. at 323. And “even if Ozio’s trial testimony was false, Whitton ha[d] not demonstrated that the State was aware that Ozio intended to present false testi- mony,” the court said. Id. at 324. Second, the Florida Supreme Court held that Whitton’s ap- pellate counsel was not constitutionally ineffective in failing to fur- ther investigate McCullough’s recantation. See id. at 334. It deter- mined that counsel did not perform deficiently, and Whitton could USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 14 of 72

14 Opinion of the Court 23-10786

not establish prejudice given the “overwhelming evidence against” him. Id. Third and finally, as to Whitton’s ineffective-mitigation claim, the Florida Supreme Court concluded that the additional ev- idence was “cumulative,” and Whitton could not establish preju- dice. See id. at 332. The court reasoned that “trial counsel is not deficient simply because postconviction counsel could find a more favorable expert.” Id. at 333. Whitton filed a successive state habeas petition based on Hurst v. Florida, 577 U.S. 92 (2016). The state court denied relief, and the Florida Supreme Court affirmed. Whitton v. State (“Whitton V”), 238 So. 3d 724, 725 (Fla. 2018) (per curiam), cert. denied, Whit- ton v. Florida, 139 S. Ct. 328 (2018). Whitton then sought federal habeas relief. 2. Federal Habeas Proceedings In April 2015, Whitton filed a federal habeas petition. In that petition, he reiterated several claims. Among them were the fol- lowing: (1) the Giglio claim regarding Ozio’s false testimony; (2) the Strickland claim as to appellate counsel’s failure to investigate McCullough’s recantation; (3) the Strickland claim as to mitigation; and (4) the Miranda/Doyle claim related to the State’s comments on Whitton’s silence. In 2016, the district court stayed proceedings until the Florida Supreme Court decided whether Hurst applied ret- roactively on collateral review and until Whitton exhausted his Hurst claims in state court. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 15 of 72

23-10786 Opinion of the Court 15

A few years later, in March 2021, the district court lifted the stay, held oral argument on the Giglio claim, and entered an order allowing limited discovery to obtain Ozio’s testimony. After Ozio’s deposition, the district court ordered additional briefing and oral argument. It then ordered an evidentiary hearing. In doing so, it found that the Florida Supreme Court’s resolution of Whitton’s Ozio-related Giglio claim was unreasonable under § 2254(d) and that Whitton did not fail to develop the factual basis in state court under § 2254(e)(2). In September 2022, the district court held the evidentiary hearing. Ozio testified remotely. He said that the officers assigned to Whitton’s case took Ozio out of jail several times, including to the pawn shops where Ozio had sold his stolen jewelry. The offic- ers mentioned the five-year mandatory minimum on the firearm charge. But they also mentioned Whitton’s case. Ozio believed that he could avoid prison, but only if he “completely cooper- ate[d]” with the officers on Whitton’s case. Based on the officers’ “continuous” conversations, Ozio “picked up most everything that they wanted to hear . . . and just regurgitated it back in a different way.” But, Ozio testified, he “really didn’t have any information to give until [he] overheard it from them.” Ozio testified that, in fact, he had never heard Whitton testify to stabbing anyone. Once Ozio agreed to testify against Whitton, Ozio contin- ued, “it was a night and day change.” The officers moved Ozio to a new cell and gave him special privileges. And they took him from the jail additional times and provided him with alcohol and USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 16 of 72

16 Opinion of the Court 23-10786

marijuana. In fact, the officers even took him to “a conjugal visit with a girl that [he] had never met before.” Plus, Ozio and Wallace abruptly received probation, and they were released and allowed to return home to Texas. As for Ozio’s testimony at Whitton’s trial, Ozio explained how that occurred. Ozio said that before he testified at Whitton’s murder trial, he met with a State representative for “three or four hours of grilling.” The representative was “to the point” and told Ozio, “This is what I need. No, don’t say this. No, that’s not the direction we’re going.” Ozio “was told not to mention” his juvenile record. The representative also told Ozio not to testify to a “knife detail” to which Ozio had alluded in his previous deposition and which apparently came from another case that Ozio had “mixed up.”7 Ozio testified that the prosecutor was “telling [him] to give testimony that was untrue” and “knew pointblank that [he] wasn’t telling the truth.” Ozio was “scared” and “looking for an out” from

7 In his 1992 pretrial deposition, Ozio testified that he heard Whitton mention

a knife that was located in a ravine or a canal behind Whitton’s parents’ house in Pensacola. But Whitton’s parents lived in New York, not Pensacola. The “knife detail” apparently came from another Walton County capital case pend- ing at the time, Suggs v. State, No. 1990-CF-338. Whitton filed a motion for judicial notice of the state-court docket and filings in Suggs. In 2022, Suggs filed a state postconviction motion citing newly discovered evidence that Clay- ton Adkinson (also Whitton’s prosecutor) had coached jailhouse informants’ trial testimony. See Suggs v. State, No. 1990-CF-338, Doc. 194 (Walton Cnty. Cir. Ct. May 16, 2022). The informants also claimed that Sheriff McMillan offered them favorable treatment in exchange for their testimony. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 17 of 72

23-10786 Opinion of the Court 17

the mandatory minimum. And, Ozio testified, he was never served with conditions of probation; it was “almost like it didn’t exist.” Ozio confirmed that he would have testified consistently with his affidavit at the state postconviction hearing; he wanted to clear his conscience. But Ozio did not travel to Florida because he was concerned about being arrested and incarcerated for a proba- tion violation. Importantly, Ozio acknowledged that neither the officers nor the prosecutor explicitly “promise[d]” him anything in exchange for his testimony; it was only “implied.” He did testify, however, that prosecutors were aware of his juvenile record. Whitton called two other witnesses: Ozio’s former attorney, Mark Davis, and a former probation officer, Troy Miller. Davis tes- tified that Ozio was sentenced “below the guidelines” range and received a “good” outcome given the charges he faced. Miller tes- tified that Ozio’s presentence investigation report (“PSI”) and sen- tencing scoresheet reflected neither Ozio’s juvenile record nor his arrest for “terroriz[ing]” his family. The State called several former officers—Fred Mann, Glenn Adkinson, Brad Trusty, Bill Chapman, Steve Sunday, Allen Cotton, John Peaden, and former Sheriff Quinn McMillan. They testified that they had minimal memory of Whitton’s case but would not have made improper deals or encouraged a witness to lie at trial. Still, Trusty testified that, “[o]n occasion,” he had “gone to the state attorney’s office on behalf of inmates to determine if they could have some sort of break on their case for information.” The offic- ers also testified that conjugal visits, alcohol, and marijuana were USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 18 of 72

18 Opinion of the Court 23-10786

not allowed, even for trustees, the officers’ label for cooperating in- mates. Former prosecutor Clayton Adkinson died before the hear- ing. On November 30, 2022, the district court denied relief but granted a COA on three claims: (1) the Giglio claim about Ozio’s false testimony; (2) the Strickland claim as to mitigation; and (3) the Miranda/Doyle claim related to the State’s comments on Whitton’s silence. We expanded the COA to include Whitton’s Strickland claim on appellate counsel’s failure to investigate McCullough’s re- cantation. II. STANDARDS OF REVIEW We review a district court’s denial of habeas relief de novo. McNair v. Campbell, 416 F.3d 1291, 1297 (11th Cir. 2005). We also review mixed questions of law and fact de novo. Pye v. Warden, Georgia Diagnostic Prison, 50 F.4th 1025, 1034 (11th Cir. 2022) (en banc). But we review district courts’ factual findings for clear error. Campbell, 416 F.3d at 1297. That is a highly deferential standard of review. “A finding that is plausible in light of the full record—even if another is equally or more so—must govern.” In re Wagner, 115 F.4th 1296, 1305 (11th Cir. 2024) (cleaned up). III. DISCUSSION Beyond these traditional standards of review, the Antiterror- ism and Effective Death Penalty Act of 1996 (“AEDPA”) further lim- its our role in reviewing state criminal convictions. A federal court “shall not” grant habeas relief on a “claim that was adjudicated on the merits in State court proceedings” unless the State decision (1) USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 19 of 72

23-10786 Opinion of the Court 19

“was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States” or (2) “was based on an unreasonable determination of the facts in light of the evidence presented.” 28 U.S.C. §2254(d); see generally Williams v. Taylor, 529 U.S. 362 (2000). As to the first prong of § 2254(d), a “state court decision is contrary to clearly established federal law if the state court either arrives at a conclusion opposite to that reached by the Supreme Court on a question of law or decides a case differently than the Supreme Court has on a set of materially indistinguishable facts.” Davis v. Sellers, 940 F.3d 1175, 1185 (11th Cir. 2019) (cleaned up). A state court decision is an “unreasonable application of Supreme Court precedent if it identifies the correct governing legal principle from the Supreme Court’s decisions but unreasonably applies that principle to the facts of the petitioner’s case.” Id. (cleaned up). This is a high standard—the “state court’s ruling” must be “so lacking in justification that there was an error well understood and compre- hended in existing law beyond any possibility for fair-minded disa- greement.” Harrington v. Richter, 562 U.S. 86, 103 (2011). The second prong of § 2254(d) “requires that we accord the state trial court substantial deference” with respect to its factual de- terminations. Brumfield v. Cain, 576 U.S. 305, 314 (2015). We can’t grant relief “even if some of the state court’s individual factual findings were erroneous—so long as the decision, taken as a whole, doesn’t constitute an unreasonable determination of the facts and isn’t based on any such determination.” Pye, 50 F.4th at 1035 USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 20 of 72

20 Opinion of the Court 23-10786

(internal quotation marks omitted). And “a determination of a fac- tual issue made by a State court shall be presumed to be correct,” rebuttable by a petitioner only upon “clear and convincing evi- dence.” 28 U.S.C. § 2254(e)(1). Also, when state-court decisions rest on independently dis- positive reasons, all “are due AEDPA deference.” Hammond v. Hall, 586 F.3d 1289, 1332 (11th Cir. 2009). And habeas relief “is due to be rejected” unless every independent reason contravenes or unrea- sonably applies Court precedent. Id. We may go beyond the state court’s “particular justifications” for reaching its holding and “con- sider additional rationales that support the state court’s” decision. Pye, 50 F.4th at 1038. Or we may skip altogether the question whether the state court acted unreasonably—and thus whether AEDPA’s deferential standard of review applies—if the petitioner “cannot show prejudice under de novo review, the more favorable standard of review.” Berghuis v. Thompkins, 560 U.S. 370, 390 (2010). We apply these principles to the four claims in Whitton’s Certificate of Appealability: (1) the State presented Ozio’s false tes- timony in violation of Giglio; (2) appellate counsel was constitu- tionally ineffective for failing to investigate McCullough’s desire to recant; (3) trial counsel was constitutionally ineffective at the miti- gation phase; and (4) the State improperly commented on Whit- ton’s invocation of his right to silence in violation of Doyle. A. Whitton’s Giglio claims fail because he has not shown that the State knew or should have known that its prosecutors USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 21 of 72

23-10786 Opinion of the Court 21

knowingly elicited false testimony and because he cannot estab- lish harmful error. Giglio error is a type of Brady error. It happens when the evidence the government failed to disclose shows that “the prose- cution’s case included perjured testimony and that the prosecution knew, or should have known, of the perjury.” United States v. Stein, 846 F.3d 1135, 1147 (11th Cir. 2017) (quoting Ford v. Hall, 546 F.3d 1326, 1331 (11th Cir. 2008)); see also Giglio v. United States, 405 U.S. 150, 153 (1972); Brady v. Maryland, 373 U.S. 83, 87 (1963). To prove his Giglio claim, Whitton must show two things: “(1) the prosecutor knowingly used perjured testimony or failed to correct what he subsequently learned was false testimony; and (2) such use was ma- terial i.e., that there is any reasonable likelihood that the false testi- mony could have affected the judgment.” Stein, 846 F.3d at 1147 (quoting Ford, 546 F.3d at 1331–32). Giglio’s traditional “materiality standard is more defense-friendly than Brady’s,” requiring the State to show that “the false testimony was harmless beyond a reasona- ble doubt.” Guzman v. Sec’y, Dep’t of Corr., 663 F.3d 1336, 1348 (11th Cir. 2011) (cleaned up). But on collateral review, a petitioner must meet the more onerous standard in Brecht v. Abrahamson, 507 U.S. 619 (1993). Un- der this standard, we may grant relief “only if the federal court has grave doubt about whether a trial error of federal law had substan- tial and injurious effect or influence in determining the jury’s ver- dict.” Davis v. Ayala, 576 U.S. 257, 267–68 (2015) (internal quotation marks omitted). There must be “more than a reasonable possibility USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 22 of 72

22 Opinion of the Court 23-10786

that the error was harmful.” Id. at 268 (internal quotation marks omitted). If we “cannot say, with fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the judgment was not substantially swayed by the er- ror,” we “must conclude that the error was not harmless.” Granda v. United States, 990 F.3d 1272, 1293 (11th Cir. 2021) (quoting Trepal v. Sec’y, Fla. Dep’t of Corr., 684 F.3d 1088, 1114 (11th Cir. 2012)). So “Brecht can prevent a petitioner from obtaining habeas relief even if he can show that, were he raising a Giglio claim in the first in- stance on direct appeal before a state appellate court, he would be entitled to relief.” Rodriguez v. Sec’y, Fla. Dep’t of Corr., 756 F.3d 1277, 1302 (11th Cir. 2014). The Supreme Court has explained that we apply such an onerous standard because states should “not be put to the arduous task of retrying a defendant based on mere spec- ulation that the defendant was prejudiced by trial error.” Ayala, 576 U.S. at 268 (cleaned up). Whitton argues the State violated Giglio when it introduced three false pieces of testimony that prejudiced the outcome of his case: (1) that Ozio overheard Whitton confess to “stabb[ing] the bastard”; (2) that Ozio received no benefits from the State in ex- change for his testimony; and (3) that Ozio had no criminal history. And Whitton challenges the Florida Supreme Court’s disposition of his Giglio claim as both an unreasonable application of Giglio, see 28 U.S.C. § 2254(d)(1), and based on an unreasonable determination of the facts, see id. § 2254(d)(2). He claims the Florida Supreme USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 23 of 72

23-10786 Opinion of the Court 23

Court erred in six ways—two legal and four factual. And he argues we should not afford AEDPA deference to the Florida Supreme Court’s decision. But for two reasons, we need not consider his arguments that the Florida Supreme Court acted unreasonably. First, most of Whitton’s claims do not survive de novo review. See Berghuis, 560 U.S. at 390. And second, we can’t say that the Florida Supreme Court unreasonably concluded that, even without both Ozio’s and McCullough’s confessions, the State marshalled “overwhelming ev- idence against Whitton.” Whitton IV, 161 So. 3d at 334. In other words, the alleged errors did not prejudice Whitton. 1. Whitton failed to establish a material Giglio violation. We take each of the alleged Giglio violations in turn: (1) the State knew or should have known Ozio fabricated Whitton’s con- fession; (2) the State knew or should have known Ozio lied about not receiving favorable treatment for his cooperation; and (3) the State knew or should have known that Ozio had a criminal history. i. Whitton failed to show that the State knew or should have known that Ozio falsely testified about Whit- ton’s purported confession. We assume without deciding that Ozio falsely testified that he overheard Whitton’s confession that he killed Maulden, as the district court found. Whitton then bears the burden of showing “the prosecutor knowingly used [that] perjured testimony or failed to correct what he subsequently learned was false testimony.” Stein, 846 F.3d at 1147 (quoting Ford, 546 F.3d at 1332). USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 24 of 72

24 Opinion of the Court 23-10786

The Florida Supreme Court concluded that Whitton failed to meet that burden. That is, it determined he failed to show “that the State was aware that Ozio intended to present false testimony.” Whitton IV, 161 So. 3d at 324. The district court reached the same conclusion. It explained that “the state did not know, and had no way of knowing,” that Ozio’s “testimony was false” and that, at the time Ozio testified, there was not “a high probability that” his tes- timony “was false.” Whitton disputes these findings and conclusions. He claims the district court found only that the prosecution team lacked ac- tual knowledge and did not resolve the issue of constructive knowledge, which the record supported. In Whitton’s view, “[t]he only plausible inference” from Ozio’s postconviction testimony is that the prosecutors elicited his trial testimony with “reckless dis- regard for [the alleged confession’s] falsity.” But neither our precedent nor the facts support Whitton’s position. To understand why, we must quickly review some basic mens rea principles. Criminal law typically employs four levels of mens rea: purposefulness, knowledge, recklessness, and negligence. MODEL PENAL CODE § 2.02(2) (Am. L. Inst. 1962). As for knowledge, its routine application is self-explanatory: a person ac- tually knows a relevant fact. “By contrast, a reckless [individual] is one who merely knows of a substantial and unjustified risk” that a relevant fact is true. Glob.-Tech Appliances, Inc. v. SEB S.A., 563 U.S. 754, 770 (2011) (internal citations omitted). USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 25 of 72

23-10786 Opinion of the Court 25

And “a negligent [individual] is one who should have known of a similar risk but, in fact, did not,” id., because the individual deviated from the relevant standard of care that a reasonable per- son would observe, see MODEL PENAL CODE § 2.02(2)(d). That said, this articulation of the negligence standard spurs the parties’ dispute because it resembles the Giglio standard—knew or should have known, Stein, 846 F.3d at 1147—that we’ve re- counted. And Whitton argues that we’ve endorsed a mens rea standard akin to negligence: “constructive knowledge based on red flags to falsity.” The State argues otherwise. It asserts our “should have known” language refers to actual knowledge that we impute as a matter of law to the prosecutors, not a duty to resolve the po- tential falsity of each witness’s testimony. Both parties are partially correct. But ultimately, under our precedent, Whitton failed to show the State “should have known” that Ozio’s testimony was false. Whitton is correct that our “should have known” standard may apply to more than just cases of imputed actual knowledge. We impute state agents’ actual knowledge to the prosecutor be- cause “the individual prosecutor has a duty to learn of any favora- ble evidence known to the others acting on the government’s be- half in the case, including the police.” Kyles v. Whitley, 514 U.S. 419, 437 (1995). So we deem a prosecutor’s “constructive knowledge” to be as extensive as any actual knowledge a prosecutor would have acquired had she adequately performed her relevant duties, see United States v. Robinson, 627 F.3d 941, 951–52 (4th Cir. 2010) USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 26 of 72

26 Opinion of the Court 23-10786

(describing the “duty to learn” as “constructive knowledge”); United States v. Avellino, 136 F.3d 249, 255 (2d Cir. 1998) (same), such as studying all relevant case files, see Glossip v. Oklahoma, 145 S. Ct. 612, 627 (2025) (explaining “the prosecution knew Sneed’s state- ments were false as he testified to them” in part because the “pros- ecution almost certainly had access to Sneed’s medical file,” which would have disproved his testimony). But the State is correct about the dispositive legal question: a prosecutor does not have the duty to ensure the accuracy of a witness’s testimony when the prosecutor does not “believe[] the testimony to be false or perjured.” United States v. Brown, 634 F.2d 819, 827 (5th Cir. 1981) (adding “it is not enough that the testimony is challenged by another witness or is inconsistent with prior state- ments”); see United States v. Noriega, 117 F.3d 1206, 1221 (11th Cir. 1997) (explaining “willful blindness . . . might establish constructive knowledge”); see also Glob.-Tech Appliances, 563 U.S. at 770 (noting “willful blindness . . . surpasses recklessness and negligence” be- cause “the defendant must take deliberate actions to avoid learning of ” a relevant “fact”).8 We’ve been clear that prosecutors are not guilty of miscon- duct simply by adducing evidence that is not completely uncontra- dicted by other evidence. For instance, we’ve explained that “a prior statement that is merely inconsistent with a government

8 All decisions the Fifth Circuit issued by the close of business on September

30, 1981, are binding precedent in this Court. Bonner v. City of Prichard, 661 F.2d 1206, 1207 (11th Cir. 1981) (en banc). USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 27 of 72

23-10786 Opinion of the Court 27

witness’s testimony is insufficient to establish prosecutorial mis- conduct.” United States v. McNair, 605 F.3d 1152, 1208 (11th Cir. 2010); see also United States v. Michael, 17 F.3d 1383, 1385 (11th Cir. 1994) (“We refuse to impute knowledge of falsity to the prosecutor where a key government witness’ testimony is in conflict with an- other’s statement or testimony.”). And we’ve held that “witnesses’ capacity to remember events” or witnesses’ “possible motives and individual relationships with” other relevant characters “may be considered as factors in weighing their testimony.” United States v. Lopez, 985 F.2d 520, 524 (11th Cir. 1993). But the fact that witnesses remember “incidents and participants differently” and tell “differ- ent stories falls far short of establishing that the government had knowledge of false testimony being presented to the jury.” Id. Those considerations go to “credibility,” which the “defense” may challenge in its “case.” Id. On the other hand, the prosecution does have a duty to dis- close impeachment “evidence affecting” witnesses’ “credibility.” Giglio, 405 U.S. at 154. Unless the prosecution—or someone from whom the prosecution has a “duty to learn”—has knowledge of testimony’s falsity, then the government may present that evidence so long as it has disclosed to the defense relevant impeachment ma- terial. See Ventura v. Att’y Gen., Fla., 419 F.3d 1269, 1276 (11th Cir. 2005) (explaining Giglio error occurs when “undisclosed evidence demonstrates that the prosecution’s case included perjured testi- mony” (emphasis added) (citation omitted)). The jury may then judge the veracity of the alleged false testimony. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 28 of 72

28 Opinion of the Court 23-10786

But Whitton does not argue that the prosecution failed to disclose relevant impeachment material. Instead, Whitton argues the prosecution knew Whitton did not confess to Ozio when Ozio testified to the opposite effect. And under our precedent governing false-testimony claims, the statements Ozio made before and dur- ing trial—which Whitton claims include inconsistencies as to the timing, content, and language of the alleged confession—do not give rise to such a Giglio violation. That’s so because Whitton hasn’t shown that either the prosecutors or any relevant state offi- cial actually knew Ozio fabricated Whitton’s confession when he testified to it. Cf. Glossip, 145 S. Ct. at 627 (finding a due-process violation where “the prosecution knew Sneed’s statements were false as he testified to them”). To be sure, Whitton argues the prosecution knew Ozio’s tes- timony was false. But he bases his argument on the following four circumstances: (a) Ozio had a motive to fabricate a confession to obtain favorable treatment, (b) officers, in discussing Whitton’s case with Ozio, influenced Ozio’s understanding of Whitton’s case, (c) Ozio included in his recounting of Whitton’s confession a false knife detail that actually occurred in another case Clayton Adkin- son was also prosecuting, and (d) the prosecution coached Ozio to testify to information Ozio said was false. And the district court rejected those arguments and found that “the circumstances did not show conclusively or with a high probability that” Ozio’s testi- mony “was false.” We can’t say that finding was clearly erroneous. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 29 of 72

23-10786 Opinion of the Court 29

As to Ozio’s motive to provide false testimony, knowledge of a motive to falsify testimony does not equal knowledge of false testimony. Witnesses routinely testify on behalf of the govern- ment after pleading guilty—even after pleading to lesser charges or receiving lenient sentences. But that fact does not mean a witness’s testimony is false, much less make a prosecutor guilty of miscon- duct by eliciting it. Rather, a witness’s motive for his testimony is a bias the defense may explore on cross-examination. See, e.g., United States v. Thayer, 204 F.3d 1352, 1355 (11th Cir. 2000) (per cu- riam). And here, the district court found that no officer or prose- cutor offered Ozio favorable treatment in return for specific testi- mony, like Whitton’s confession. So the district court did not clearly err in discounting Ozio’s motive to obtain beneficial treat- ment in finding the prosecutors did not have actual knowledge that Ozio testified falsely. Nor did the district court clearly err in finding that the state officers who accompanied Ozio to pawn shops to collect stolen property did not know Ozio’s testimony was false. That is a “plau- sible” finding on this record. In re Wagner, 115 F.4th at 1305. Although the officers could have influenced Ozio’s recollec- tion of Whitton’s alleged crime in their conversations on the way to the pawn shops, this record does not establish that the officers knew Ozio was lying about Whitton’s confession. As the district court found, “[n]o officer told Mr. Ozio anything the officer did not believe was true,” and “[n]o officer suggested to Mr. Ozio that he should provide false testimony.” USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 30 of 72

30 Opinion of the Court 23-10786

Nor did Ozio’s conversations with the officers appear to in- fluence the substance of his testimony. In fact, Ozio could not dis- tinguish among which information he gathered from his conversa- tion with the officers who drove him to the pawn shops, from cor- rectional officers in the jail, or from other detainees. The only de- tail about Whitton’s case that Ozio could link to his car ride to the pawn shop was about the blood on Whitton’s clothes. But Ozio testified that he also overheard that detail from Whitton while in jail. Plus, Whitton’s confession—the “I stabbed the bastard” com- ment—did not originate in Ozio’s ride-along conversations. So the record does not reveal that the officers knew of anything that con- tradicted the core of Ozio’s testimony about Whitton’s confession. For similar reasons, neither the false knife detail in Ozio’s testimony nor the prosecution’s preparation with Ozio for his tes- timony establish that the prosecution knew Ozio was lying, either. Ozio recounted that, in preparing him to testify at Whitton’s trial, the prosecution frequently “correct[ed] [him] about dates and . . . about [the] sequence of events.” And when Ozio brought up his recollection of Whitton tossing a knife into a ravine, the prosecution suggested that Ozio not volunteer that detail. But Ozio also explained to the district court that the prose- cution never asked Ozio to testify to false information, and the team never instructed Ozio to lie if asked directly about the knife. So the district court found that the prosecution reasonably could have believed that Ozio “heard talk of this around the jail and mixed up the cases.” USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 31 of 72

23-10786 Opinion of the Court 31

The district court had a plausible basis for finding that the prosecution did not know Ozio’s testimony was false. Knowledge of testimony that is inconsistent with other evidence and testi- mony—and witness preparation de-emphasizing those inconsisten- cies—is not the same thing as knowledge that a witness testified falsely. See Jacobs v. Singletary, 952 F.2d 1282, 1287 (11th Cir. 1992) (explaining “the fact that state detectives failed to corroborate her testimony after interviewing three other cellmates . . . rendered [her] testimony only less credible, not incredible”); cf. Alcorta v. Texas, 355 U.S. 28, 31 (1957) (finding a due-process violation where a prosecutor elicited testimony that “gave the jury the false impres- sion” that a witness’s relationship with the defendant’s wife was that of casual friendship although the prosecutor “knew” it was, in fact, sexual).9

9 This is not to say prosecutors should attempt to correct or influence a witness

to testify to something other than what the witness believes to be true, even if the prosecutors believe they are aligning a witness’s inaccurate testimony to the truth. They shouldn’t. See In re Eldridge, 82 N.Y. 161, 171 (1880) (“[An attorney’s] duty is to extract the facts from the witness, not to pour them into him; to learn what the witness does know, not to teach him what he ought to know.”); Geders v. United States, 425 U.S. 80, 90 n.3 (1976) (“An attorney must respect the important ethical distinction between discussing testimony and seeking improperly to influence it.”). And we repeat the district court’s ad- monition against such behavior. Indeed, that misconduct may jeopardize the validity of a conviction. See, e.g., Kyles, 514 U.S. at 443 (concluding disclosure of a witness’s original statement, which contained “adjustments” at trial, would have “destroy[ed] confidence in [his] story and rais[ed] a substantial im- plication that the prosecutor had coached him to give it”). But for the reasons USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 32 of 72

32 Opinion of the Court 23-10786

Finally, even considering each of these facts together, they do not show that the district court clearly erred in concluding the prosecution neither knew nor should have known that Ozio’s testi- mony was false. Ozio confirmed that no member of the prosecu- tion team suggested that he testify that Whitton confessed to the crime. And Ozio was consistent that Whitton confessed to the murder. Plus, at the time, McCullough corroborated that Whitton confessed to the murder. So the prosecutors could have believed that Whitton confessed to the murder, even if they were unsure Ozio had a firm grasp of some relevant details about Whitton’s crime. Whitton failed to establish his first alleged Giglio violation because he didn’t establish that the prosecution team knew or should have known Ozio falsified Whitton’s confession when the prosecution elicited the confession at trial. ii. Whitton failed to show the State agreed to a deal that Ozio testify against Whitton in exchange for benefits, and he failed to show that the State adduced material, false testimony when it allowed Ozio to disclaim his subjective anticipation of benefits from the State. Whitton next argues the State failed to correct Ozio’s testi- mony about the benefits he received from the State in exchange for his testimony against Whitton.

we’ve explained, on this record, it does not establish the asserted Giglio viola- tions. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 33 of 72

23-10786 Opinion of the Court 33

To recap, at trial, Ozio testified that state officers did not “promise or offer [him] any help for [his] statement.” Ozio further testified that he “never at any time” thought that testifying against Whitton “was going to help [him] in any way” and that he “didn’t think [he’d] be going to prison, [that] being [his] first offense.” But in his postconviction affidavit and testimony, Ozio recounted the various benefits he received from the State in exchange from his testimony: his “trustee” privileges, reduced misdemeanor charges, a sentence of probation, and even a conjugal visit and access to al- cohol and marijuana. Ozio testified at his postconviction deposition that he was “fear[ful] of [the mandatory minimum] charge,” and he said that officers “portrayed it as a situation that [he] probably wasn’t going to get out of.” That said, at the federal evidentiary hearing, Ozio acknowledged that neither the officers nor the prosecutor explicitly “promise[d]” him anything in exchange for his testimony; any promise was “implied,” Ozio clarified, because the officers “in- formed [him] that the only way [he] was going to get any actual help was to completely cooperate” regarding “Whitton’s case.” These facts prompt two theories of potential Giglio viola- tions. First, the State could have presented false testimony had the State and Ozio reached some form of agreement. And second, the prosecutors could have presented false testimony if they knew Ozio was testifying with the hope of receiving benefits from the State. The district court rejected both those theories. And we find neither error in its conclusions nor clear error in its findings. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 34 of 72

34 Opinion of the Court 23-10786

Starting first with the theory that the State and Ozio reached an agreement about leniency and other benefits in exchange for Ozio’s testimony against Whitton, the district court found that Ozio “testified truthfully that he was promised nothing in exchange for the false testimony.” So it concluded no Giglio violation had occurred. The district court reasoned that Ozio “hoped to obtain favorable treatment, and in fact he did obtain favorable treatment,” but “a unilateral hope is much different from an offer, promise, or mutual agreement.” Whitton argues the district court erred because Giglio “ap- plies not only to undisclosed explicit promises from the State in ex- change for testimony, but also to undisclosed implicit agreements or understandings.” And that is true. We’ve reasoned, “Giglio does not require that the word ‘promise’ . . . must be specifically em- ployed.” Brown v. Wainwright, 785 F.2d 1457, 1464–65 (11th Cir. 1986). But a Giglio violation contemplates at least a “bilateral agree- ment” or an offer “subject to acceptance.” Id.; see Alderman v. Zant, 22 F.3d 1541, 1554–55 (11th Cir. 1994) (concluding the Giglio “rule states that there must be a full disclosure of any agreements” but “does not address nor require the disclosure of all factors which may motivate a witness to cooperate”); cf. Haber v. Wainwright, 756 F.2d 1520, 1524 (11th Cir. 1985) (remanding for further factfinding on whether a prosecutor’s “advice . . . amounted to a grant of im- munity or otherwise constituted an agreement or understanding between Brandt and the state”). USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 35 of 72

23-10786 Opinion of the Court 35

Here, though, the district court found no such agreement— a finding we can’t say is clearly erroneous. The only testimony that calls into question the district court’s finding is Ozio’s assertion that state officers, on pawn-shop rides, informed Ozio “the only way [he] was going to get any actual help was to completely cooperate.” But we’ve upheld district courts’ findings that similar statements did not amount to an implicit, mutual agreement. In Traver v. Hopper, for instance, a prosecutor stated, “any co- operation [the witness] gave us and if he told the truth in this mat- ter would be taken into consideration,” and the prosecutor subjec- tively believed (but did not convey) that the witness would not be tried for capital murder if the witness testified for the prosecution. 169 F.3d 710, 716 (11th Cir. 1999). Under those facts, we accepted the district court’s finding—“because it [was] not clearly errone- ous”—“that whatever exchange may have taken place between” the prosecutor and the witness “did not ripen into a sufficiently definite agreement” that would require disclosure under Giglio. Id. at 717. “Some promises, agreements, or understandings do not need to be disclosed,” we added, “because they are too ambiguous, or too loose or are of too marginal a benefit to the witness to count.” Id. And confirmation that a witness’s testimony would be “taken into consideration” was “too preliminary and ambiguous to demand disclosure.” Id.; see also Depree v. Thomas, 946 F.2d 784, 797–98 (11th Cir. 1991) (holding promise to “take care” of witness does not re- quire disclosure). USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 36 of 72

36 Opinion of the Court 23-10786

So here, although the officers who drove Ozio to the pawn shops gave Ozio the impression that they wished for him to provide any information he possessed implicating Whitton, the district court did not clearly err in finding those impressions did not “ripen into a sufficiently definite agreement before” Whitton’s trial. Traver, 169 F.3d at 717. To be sure, it’s plausible that Whitton was “merely trying to cooperate in hopes of improving [his] bargaining position later.” Id. But under our binding precedent, Ozio’s “hope[s]” that his testimony against Whitton “would result in more favorable treatment” do “not convert ambiguous statements by the State”—like impressions, as the district court put it—“into prom- ises that [he] would, in fact, receive more favorable treatment.” Depree, 946 F.2d at 798; see, e.g., id. at 797 (discussing a promise to “take care” of the witness); Traver, 169 F.3d at 716 (addressing pros- ecutor’s “consideration” of a witness’s cooperation). Indeed, even though Ozio received favorable treatment after the fact, and even if the State subjectively intended to give favorable treatment to Ozio after the fact, under our precedent, those con- siderations do not transform infirm statements about leniency or benefits into real agreements within the scope of Giglio. See, e.g., Hays v. Alabama, 85 F.3d 1492, 1499 (11th Cir. 1996) (rejecting argu- ment that Alabama’s failure to prosecute a witness after giving fa- vorable testimony amounted to proof of an agreement with the State); Traver, 169 F.3d at 716–17 (upholding finding of no agree- ment even though prosecutor “believed” the witness “would not be tried for capital murder if ” the witness “testified for the prose- cution”). USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 37 of 72

23-10786 Opinion of the Court 37

So as to the first theory of a potential Giglio violation, Whit- ton failed to prove the existence of an implicit agreement that would make the State guilty of misconduct. As for Whitton’s second theory that the State knew Ozio co- operated with the hope of receiving favorable treatment but per- mitted him to testify otherwise—specifically, that the State allowed Ozio to falsely state he didn’t think testifying against Whitton “was going to help [him] in any way”—the district court rejected it. It found that the prosecution “did not know” Ozio’s “testimony” con- cerning his anticipation of benefits “was false—only Mr. Ozio had personal knowledge of his own motives.” Whitton contests that finding as clearly erroneous. He says the ride-along officers impressed upon Ozio that he could benefit from testifying against Whitton, and the State prosecutor, Adkin- son, knew of the “condition” of probation placed on Ozio to “re- turn to testify truthfully in [the] Whitton case.” We agree that sug- gests the State knew Ozio testified with the hope of receiving fa- vorable treatment. But even so, on this record, it does not amount to a material Giglio violation. Defense counsel cross-examined Ozio about his reduced charges and sentence of probation, as well as his desire to “help” himself. Whitton’s counsel emphasized that before Ozio reported to Florida authorities that Whitton confessed, Ozio faced numer- ous charges, some of which included mandatory minimum sen- tences of five years each. Then, after Ozio implicated Whitton, the defense emphasized, the State allowed Ozio to plead guilty to lesser USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 38 of 72

38 Opinion of the Court 23-10786

charges that did not include any mandatory minimum sentence. And ultimately, in closing, the prosecution admitted that “perhaps [Ozio] received a break in regards to probation versus the maxi- mum of three and a half that he could have gotten in prison,” even though it argued that Ozio received leniency because he was a first- time offender. So the jury had the facts to infer that the prosecution and Ozio reached some type of an agreement in exchange for his testimony. And Ozio’s refusal to acknowledge an agreement, or even a selfish motivation on his part to testify against Whitton, could have further undermined his credibility. As the district court put it, “[a] cooperating witness’s denial of any desire for favorable treatment rarely helps the proponent of the testimony.” So the testimony Ozio gave at trial was, at a minimum, just as persuasive, if not less persuasive, than it would have been had he candidly explained he hoped the prosecution would look favorably upon his reporting of Whitton’s confession. At bottom, defense counsel’s impeachment of Ozio on his motivation to testify makes any related Giglio viola- tion immaterial. See McNair, 605 F.3d at 1211 (finding no reversable error, in part, because the jury was aware the witness “had made a plea deal and that the government’s assessment of his cooperation would impact his eventual sentence”). USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 39 of 72

23-10786 Opinion of the Court 39

iii. Whitton failed to show Ozio’s false testimony that he had no criminal history would have had a substantial and injurious effect or influence on the jury’s verdict. Lastly, Whitton argues the State failed to correct Ozio’s tes- timony on cross-examination that he had “never been arrested be- fore.” That testimony was false: Ozio’s juvenile records show he was charged with assault with bodily injury against his father, ter- roristic threats against his mother, and at least one other burglary. The State also had those records, as the district court found. And Whitton adds, the State capitalized on the false testimony by por- traying Ozio as a young kid with no criminal past. See DeMarco v. United States, 928 F.2d 1074, 1077 (11th Cir. 1991) (“[T]he prosecu- tor’s argument to the jury capitalizing on the perjured testimony reinforced the deception of the use of false testimony and thereby contributed to the deprivation of due process.”). 10 Still, the district court concluded that any such Giglio viola- tion would not have made a difference because, in Florida, juvenile criminal records are generally inadmissible to attack a witness’s credibility. See Fla. Stat. § 90.610(1)(b) (1977); Benedit v. State, 575 So. 2d 236, 237 (Fla. 3d DCA 1991). This conclusion was erroneous. There is no indication the trial court would have used Section 90.610 to bar the use of Ozio’s juvenile records after Ozio opened the door to such evidence. Florida courts admit juvenile records

10 The State also contends that Whitton failed to exhaust this claim. But be- cause the claim fails on the merits, we do not address the State’s exhaustion arguments. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 40 of 72

40 Opinion of the Court 23-10786

when a “witness, by volunteering that he had never been in trouble before, . . . open[s] the door to [prior] conviction[s], which would not have otherwise been admissible.” Cullen v. State, 920 So. 2d 1155, 1156 (Fla. 4th DCA 2006); see also Mosley v. State, 739 So.2d 672, 677 (Fla. 4th DCA 1999) (collecting cases where a witness was impeached after minimizing his criminal past). And impeachment concerning Ozio’s juvenile record would have undermined his credibility: it would have shown Ozio per- jured himself. It also may have solidified the inference that the prosecution struck a deal with Ozio by eliminating the prosecu- tion’s argument that Ozio received leniency because he was a first- time offender.11 Although the prosecution admitted that Ozio caught “a break” by receiving a lenient sentence, it attempted to brush off suggestions of any deals by proffering the applicable sen- tencing guidelines for a first-time offender like Ozio. Had the jury learned Ozio was not a first-time offender, the prosecution could not have made such an argument. And the jury would have known that Ozio lied on the stand. Such a combination of impeachment

11 Potentially, the pattern of prosecutorial misconduct could have caused the

jury to doubt the credibility of the prosecutors themselves. See Brecht, 507 U.S. at 638 n.9 (addressing the “unusual case” where “a pattern of prosecutorial misconduct, might so infect the integrity of the proceeding as to warrant the grant of habeas relief, even if it did not substantially influence the jury’s ver- dict”). But Whitton references this issue in a single-sentence footnote, so we do not address it. See Sapuppo v. Allstate Floridian Ins. Co., 739 F.3d 678, 681 (11th Cir. 2014) (“We have long held that an appellant abandons a claim when he either makes only passing references to it or raises it in a perfunctory man- ner without supporting arguments and authority.”). USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 41 of 72

23-10786 Opinion of the Court 41

by perjury and a deal for favorable treatment, we’ve held, can de- stroy a witness’s credibility. See, e.g., United States v. Sanfilippo, 564 F.2d 176, 178 (5th Cir. 1977) (explaining a propensity for dishonesty, combined with evidence of promised favorable treatment, “might well have” caused the jury to conclude the witness “had fabricated testimony in order to protect himself against another criminal prosecution”); accord Brown, 785 F.2d at 1466; cf. Ventura, 419 F.3d at 1292 & n.9. So we cannot affirm the district court’s decision on its find- ing of harmless error. Still, we cannot reverse the district court, either, because we must affirm on the other basis the State ad- vances, see Lucas v. W.W. Grainger, Inc., 257 F.3d 1249, 1256 (11th Cir. 2001)—that Ozio’s testimony, overall, was immaterial to the jury’s verdict. See McNair, 605 F.3d at 1208 (“When a government lawyer elicits false testimony that goes to a witness’s credibility, we will consider it sufficiently material to warrant a new trial only when the estimate of the truthfulness and reliability of the given witness may well be determinative of guilt or innocence.” (citation and in- ternal quotation marks omitted)). For the reasons we discuss in the next section, we owe deference to the Florida Supreme Court’s de- termination that, even without Whitton’s confession, the State marshalled overwhelming evidence that would have caused the jury to reach the same result. 2. Assuming Whitton established Giglio violations that materially undermined the veracity of Ozio’s testi- mony, the Florida Supreme Court did not USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 42 of 72

42 Opinion of the Court 23-10786

unreasonably determine that overwhelming evi- dence supported the jury’s conviction. Under AEDPA, we owe deference to a state court’s prejudice determinations. See Pye, 50 F.4th at 1041–42. So to pierce AEDPA’s veil, Whitton must show the state court’s determination that he did not suffer prejudice was an unreasonable application of, or con- trary to, federal law. See id. In other words, even if we may disagree with the state court’s conclusion and would have reached a differ- ent one had we taken the first look, we cannot overturn the state court’s decision unless it was “so obviously wrong that its error lies beyond any possibility for fairminded disagreement.” Id. (quoting Shinn v. Kayer, 592 U.S. 111, 118 (2020)); see also Harrington, 562 U.S. at 101. The Florida Supreme Court considered the weight of the ev- idence against Whitton, albeit in a roundabout way. It did not di- rectly address prejudice flowing from Ozio’s testimony when it considered Whitton’s Giglio claims. See Whitton IV, 161 So. 3d at 323–24 (addressing falsity of Ozio’s testimony and the State’s knowledge). But in assessing whether a jury could sustain Whit- ton’s conviction without McCullough’s testimony, the court de- scribed the weight of the evidence against Whitton, independent of both confessions, and found it to be “overwhelming.” The court said, Further, Whitton cannot establish prejudice. As dis- cussed above, recantations are not credible. Without McCollough’s testimony, Ozio’s testimony would still USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 43 of 72

23-10786 Opinion of the Court 43

have provided the jury with evidence that Whitton admitted to murdering Mauldin. That, coupled with the overwhelming evidence against Whitton, makes it ex- tremely unlikely that McCollough’s recantation would have changed the outcome of the trial. Id. at 334 (emphasis added). Unless unreasonable, the Florida Su- preme Court’s conclusion that the non-confession evidence was “overwhelming” binds us and is dispositive of the prejudice ques- tion here. We address this legal conclusion and then delve into the reasonableness of the Florida Supreme Court’s holding that the ev- idence against Whitton, independent of the confession, was “over- whelming.” First, the Florida Supreme Court’s conclusion that, even without considering the confessions, there was still “overwhelming evidence against Whitton” binds us. That’s so because AEDPA re- quires us to review “the specific reasons given by the state court and defer[] to those reasons if they are reasonable.” Wilson v. Sellers, 584 U.S. 122, 125 (2018). Plus, to ignore the Florida Supreme Court’s determination that the State presented “overwhelming ev- idence” against Whitton would bypass our rationale in Pye that we should not accept readings of AEDPA that “strictly limit[]” a state court “to the particular justifications” it “memorialized in its writ- ten opinion.” 50 F.4th at 1039. As we’ve made clear, we owe defer- ence to a state court’s implicit yet reasonable findings and conclu- sions. See Blankenship v. Hall, 542 F.3d 1253, 1272 (11th Cir. 2008) (explaining “implicit findings of fact are entitled to deference under USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 44 of 72

44 Opinion of the Court 23-10786

§ 2254(d) to the same extent as explicit findings of fact”). And here, they are reasonable. Second, a lack-of-prejudice holding immediately follows from the Florida Supreme Court’s conclusion that, even without Ozio’s and McCullough’s testimony that Whitton confessed, the State still introduced “overwhelming evidence against Whitton.” We’ve repeated that errors are harmless “under the Brecht standard where . . . other evidence of guilt is overwhelming.” Mansfield v. Sec’y, Dep’t of Corr., 679 F.3d 1301, 1313 (11th Cir. 2012); accord Tre- pal, 684 F.3d at 1114; Prevatte v. French, 547 F.3d 1300, 1304 (11th Cir. 2008); Grossman v. McDonough, 466 F.3d 1325, 1340 (11th Cir. 2006); see also Lamb v. Jernigan, 683 F.2d 1332, 1342 (11th Cir. 1982); Cape v. Francis, 741 F.2d 1287, 1294–95 (11th Cir. 1984); Davis v. Kemp, 752 F.2d 1515, 1521 (11th Cir. 1985); Tucker v. Kemp, 762 F.2d 1496, 1501 (11th Cir. 1985); United States v. Harriston, 329 F.3d 779, 789 (11th Cir. 2003). So if the Florida Supreme Court reasonably determined that overwhelming evidence supported Whitton’s conviction, we must give deference to its ultimate decision to reject Whitton’s ha- beas petition. See Pye, 50 F.4th at 1042 (explaining “some debatable calls as to the weight” of certain “pieces of evidence” did not ren- der unreasonable the “ultimate decision to deny relief ”). Third, the Florida Supreme Court reasonably determined that the record contained “overwhelming evidence” of Whitton’s guilt. The undisputed facts place Whitton at the scene of the crime the night that Maulden was killed, and evidence suggests Whitton committed the murder: the motel clerk saw Whitton’s car parked USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 45 of 72

23-10786 Opinion of the Court 45

at the motel, and Whitton admitted he was at the motel; Whitton’s boots and his car were stained with blood that, after later retesting, matched Maulden’s DNA; the blood stains on and inside Whitton’s boots are consistent with “a stabbing or a beating”; Whitton could not explain the downward blood spatter on the inside of his boots; and Whitton’s car contained a power and gas receipt and a car wash ticket for 2:37 a.m. October 10, 1990, the night of the murder. To be sure, portions of the State’s case rely on credibility de- terminations on which the jury could easily side with Whitton. For instance, the State posits that Whitton murdered Maulden because Whitton was behind on his bills, was about to lose electricity, and knew that Maulden had over $1,000 cash. Yet Whitton testified that he received money from the prior occupant, Debra Sims, because the bills had been incurred during her tenancy. And the State was neither able to recover Maulden’s cash nor prove what Whitton spent that money on. Similarly, as to the State’s arguments that Whitton did not report the death of his “pretty close” friend and that Whitton told several lies, 12 Whitton argues that the jury could have believed his allegedly untruthful and irrational actions were driven by panic and

12 The State argues it proved Whitton lied about the following: that Whitton

did not know how much money Maulden had despite helping Maulden with- draw his money; that Whitton tried to contact Maulden’s mother when she never received a call; that Whitton told the police he did not go back to the hotel when he in fact did; and that Whitton lied to his boss about why he did not show up to work. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 46 of 72

46 Opinion of the Court 23-10786

fear. And he urges, that’s a credibility question that is front and center after we discount Ozio’s and McCullough’s testimony. Then, as to the supposed persuasiveness of the overall case, Whitton highlights the state prosecutors’ conduct. He notes that after discovering Whitton’s purported confessions, the State with- drew a plea deal to “stand mute . . . with a guidelines sentence” to second-degree murder and robbery. Instead, the State sought ei- ther life imprisonment or the death penalty. And after learning of Zeigler’s potentially exculpatory DNA testimony, the prosecution team engineered what the district court called an “inexcusable at- tempt to block” it through verbal and physical assault. Had the State viewed the evidence against Whitton as “overwhelming,” he suggests, the prosecution would not have acted the way it did. A jury may well find some of these points to be persuasive. But we cannot say that the state court was “so obviously wrong,” Pye, 50 F.4th at 1042, in concluding the non-confession evidence against Whitton is “overwhelming.” In particular, Whitton can’t rebut the State’s blood-spatter evidence. The inside of Whitton’s boots contained blood matching the victim’s type and were the “target for forceful bloodshed” and “medium velocity” blood spat- ter consistent with “a stabbing or a beating.” Evidence also sug- gested that the blood traveled from the top of Whitton’s boots to the bottom, directly contradicting Whitton’s testimony that he was in Maulden’s room for only a minute before fleeing. Notably, the State’s expert rebutted Whitton’s theory that blood seeped into his boots through their bottoms by testifying that some of the USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 47 of 72

23-10786 Opinion of the Court 47

downward blood spatter had dried before Whitton put his boots back on; Whitton could not have been wearing his boots when “the blood spatter occurred on the interior of ” them. Whitton responds that the State could not explain the lim- ited amounts of blood on Whitton clothes, boots, and in his car, considering the extremely bloody scene—officers reported that blood covered at least three walls. To this, Whitton adds that the DNA evidence matched a third party, not Maulden or Whitton. But the State introduced expert testimony that effective washing can eliminate traces of blood, as well as evidence that Whitton at- tempted to wash away the blood he collected in Maulden’s motel room. Indeed, Whitton purchased a car wash ticket, dated 2:37 a.m. on the night of the murder. And an officer reported that Whit- ton said he washed his boots, although Whitton disclaims making such a statement. Also, after retesting the DNA on Whitton’s boots, see, e.g., Whitton II, 824 So. 2d at 171; Whitton III, 838 So. 2d at 560, the State confirmed that the inside of Whitton’s right boot contained blood from a “mixture of two or more individuals” with the “major do- nor” matching “the DNA profile of James Maulden.” In short, the blood-spatter evidence ties Whitton directly and firmly to Maulden’s murder. So we can’t say it was unreasonable for the State court to conclude there was “overwhelming evidence against Whitton.” Whitton IV, 161 So. 3d at 334. Nor does Supreme Court precedent preclude the prejudice finding the State made on these facts. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 48 of 72

48 Opinion of the Court 23-10786

Rather, even if the “call[]” is ultimately “debatable,” it’s within the realm of fair-minded disagreement, Pye, 50 F.4th at 1042, that blood spatter evidence—which has no explanation other than one that is consistent with the defendant’s guilt—will surely convince a jury of the defendant’s guilt, see Brecht, 507 U.S. at 638. See, e.g., State v. Perkins, 1999 WL 334974, at *3 (Ohio Ct. App. May 28, 1999) (hold- ing “blood spattered” on defendant’s “T-shirt and jeans,” as well as DNA testing proving the spatter matched the victim, “presented overwhelming evidence of ” the defendant’s guilt). Because the state court’s prejudice determination “wasn’t contrary to or an unreasonable application of the Supreme Court’s precedents, based on an unreasonable determination of the facts, or ‘so obviously wrong that its error lies beyond any possibility for fairminded disagreement,’” AEDPA requires us to defer to its “weighing of the evidence” and “cumulative-prejudice conclu- sion.” Pye, 50 F.4th at 1056 (quoting Shinn, 592 U.S. at 118); see also Harrington, 562 U.S. at 102 (“If this standard is difficult to meet, that is because it was meant to be.”). B. Whitton’s appellate counsel was not prejudicially ineffective. We next turn to Whitton’s claim that his appellate counsel was ineffective for failing to investigate McCullough’s recantation. The same standards that apply to trial counsel under Strick- land also govern claims of ineffective assistance of appellate coun- sel. Brooks v. Comm’r, Ala. Dep’t of Corr., 719 F.3d 1292, 1300 (11th Cir. 2003). To prevail on a claim of ineffective assistance of counsel, a petitioner must show that his counsel “made errors so serious USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 49 of 72

23-10786 Opinion of the Court 49

that counsel was not functioning as the ‘counsel’ guaranteed the defendant by the Sixth Amendment” and that the deficient perfor- mance prejudiced the defendant by depriving him of a fair trial with a reliable result. Raheem v. GDCP Warden, 995 F.3d 895, 908 (11th Cir. 2021) (quoting Strickland v. Washington, 466 U.S. 668, 687 (1984)). An ineffective-assistance showing involves two prongs: per- formance and prejudice. Under this framework, a petitioner must show both that (1) counsel’s representation was objectively unrea- sonable, and (2) a reasonable probability exists that, without coun- sel’s unprofessional errors, the proceeding’s result would have dif- fered. Id. We evaluate an attorney’s performance by whether it was reasonable “under prevailing professional norms,” and we judge that “on the facts of the particular case.” Strickland, 466 U.S. at 688, 690. In performing our review, we apply “a strong presump- tion that counsel’s conduct falls within the wide range of reasona- ble professional assistance.” Id. at 689. And when we perform a Strickland analysis within the AEDPA framework, which requires us to defer to state rulings that are not unreasonable, our deference to the state court’s determina- tion that counsel has performed adequately is “doubly deferential.” Dunn v. Reeves, 594 U.S. 731, 739 (2021) (quoting Burt v. Titlow, 571 U.S. 12, 15 (2013)); see also Raulerson v. Warden, 928 F.3d 987, 996–97 (11th Cir. 2019) (“Because Strickland provides a most deferential standard for assessing the performance of counsel, when we USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 50 of 72

50 Opinion of the Court 23-10786

combine it with the extra layer of deference that § 2254 provides, the result is double deference.” (cleaned up)). Whitton argues that his appellate counsel, Paula Saunders, was constitutionally ineffective in failing to reasonably investigate McCullough’s recantation and press for a new trial. But we do not decide the issue of deficient performance because, in any case, Whitton has failed to show that the Florida Supreme Court’s prej- udice determination was unreasonable. To establish prejudice based on failure to file a motion, a pe- titioner must show both that (1) the motion was “meritorious” and (2) a reasonable probability exists that the outcome of the proceed- ing would have been different if the motion had been granted. Tay- lor v. Sec’y, Fla. Dep’t of Corr., 64 F.4th 1264, 1271 (11th Cir. 2023). Whitton argues that, had McCullough submitted an affida- vit or otherwise recanted his testimony, Saunders could have moved to remand the case for a new trial. Even if the result of the direct appeal would have been unchanged, he adds, Saunders’s fail- ure to investigate “caused irreparable harm to” Whitton’s postcon- viction proceedings. In particular, he points to the fact that the Florida Supreme Court found McCullough’s recantation unreliable because it was never memorialized. See Whitton IV, 161 So. 3d at 323. What’s more, Whitton continues, the state trial court “based its finding” of the avoid-arrest aggravator entirely on McCullough’s testimony. Whitton I, 649 So. 2d at 867. So had Whitton received a new trial, he posits, it is reasonably possible that the trial court would have found one less aggravator at sentencing. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 51 of 72

23-10786 Opinion of the Court 51

The State responds that it is objectively unlikely the Florida Supreme Court would have granted any motion to relinquish its jurisdiction and remand for a hearing on McCullough’s affidavit. Rather, it argues, Whitton’s prejudice argument rests on a chain of “speculative events”: (1) the Florida Supreme Court would have had to grant Whitton’s motion to relinquish jurisdiction to pursue the McCullough-recantation issue; (2) McCullough would have had to give a sworn statement recanting and maintain that recan- tation at the hearing despite his perjury fears; (3) the trial court would have had to find McCullough’s recantation credible; and (4) the Florida Supreme Court would have had to reverse Whitton’s judgment on direct appeal. To be sure, we don’t know whether a court would find McCullough’s recantation to be credible, especially given his exten- sive criminal history. And McCullough may have refused to testify out of fear of a perjury prosecution. But we assume for the sake of argument that there is a reasonable probability that the first three “speculative events” would have occurred. Even so, though, given the substantial evidence against Whitton—the blood spatter, the motel clerk’s testimony, Whitton’s own admission that he returned to the motel, and other circum- stantial evidence—we can’t say the Florida Supreme Court unrea- sonably found that McCullough’s recantation was insufficient to warrant a new trial. As we’ve covered, the Florida Supreme Court concluded Whitton could not establish prejudice on his Strickland claim because (1) “recantations are not credible”; (2) “[w]ithout USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 52 of 72

52 Opinion of the Court 23-10786

McCollough’s testimony, Ozio’s testimony would still have pro- vided the jury with evidence that Whitton admitted to murdering Mauldin”; and (3) Ozio’s testimony, “coupled with the overwhelm- ing evidence against Whitton, makes it extremely unlikely that McCollough’s recantation would have changed the outcome of the trial.” Whitton IV, 161 So. 3d at 334. And for the same reasons we explained in the last section, we can’t say the Florida Supreme Court’s prejudice determination was unreasonable. Next, we address Whitton’s contention that Saunders’s fail- ure to investigate and memorialize McCullough’s recantation prej- udiced Whitton’s postconviction proceedings. For this claim, Whitton trains his arguments on the wrong proceeding. The rele- vant inquiry is whether Saunders’s deficient performance preju- diced Whitton’s appeal, not his future postconviction proceedings. See Heath v. Jones, 941 F.2d 1126, 1132 (11th Cir. 1991) (holding prej- udice is shown if “the neglected claim would have a reasonable probability of success on appeal” (emphasis added)); cf. also Purvis v. Crosby, 451 F.3d 734, 739 (11th Cir. 2006) (explaining “when the claimed error of counsel occurred at the guilt stage of a trial (in- stead of on appeal) we are to gauge prejudice against the outcome of the trial: whether there is a reasonable probability of a different result at trial, not on appeal”). True, trial and appellate counsel in death-penalty cases should act with an eye towards postconviction litigation. As a practical matter, though, a holding that the preju- dice prong may be satisfied through an adverse impact on future postconviction litigation would expand the scope of cognizable USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 53 of 72

23-10786 Opinion of the Court 53

Strickland claims. So we cannot say the Florida Supreme Court ran afoul of clearly established federal law. See 28 U.S.C. § 2254(d)(1). And in any event, even if McCullough had signed a sworn statement and testified in the postconviction proceedings, a post- conviction court may well have found that recantation not to be credible. Or a post-conviction court may well have decided any Gi- glio claim the way the district court decided the Ozio-related Giglio claim here. In other words, we don’t find a reasonable probability that, based on McCullough’s recantation alone, Whitton would have obtained postconviction relief. Even taking McCullough’s recantation and Ozio’s recanta- tion together, we can’t say that “there is a reasonable probability that . . . the result of the [postconviction] proceeding would have been different.” Strickland, 466 U.S. at 694. That’s so because, as we’ve discussed, the Florida courts determined that otherwise “overwhelming evidence [existed] against Whitton.” Whitton IV, 161 So. 3d at 334; cf. also Green v. Sec’y, Dep’t of Corr., 28 F.4th 1089, 1151 (11th Cir. 2022) (concluding “it would not be unreasonable for a jury to credit these witnesses’ original testimony and discredit their new versions, just as the [state court] did”). And as to the avoid-arrest aggravator, the sentencing court found four other aggravators: (1) Whitton was on parole; (2) Whit- ton had a previous felony conviction involving the use or threat of violence; (3) the murder was committed for pecuniary gain; and (4) the crime was especially heinous, atrocious, or cruel. Based on these other findings and this record, we don’t see a reasonable USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 54 of 72

54 Opinion of the Court 23-10786

probability that the sentencing jury and court would have declined to impose a death sentence in the absence of the avoid-arrest ag- gravator. The Florida Supreme Court, after all, has said that the heinous, atrocious, and cruel aggravator “is among the weighti- est . . . in the statutory scheme.” Rigterink v. State, 66 So. 3d 866, 900 (Fla. 2011). Having rejected each of Whitton’s prejudice arguments, we cannot conclude that the Florida Supreme Court’s prejudice deter- mination was an unreasonable application of Strickland. See 28 U.S.C. § 2254(d)(1). We thus affirm the district court’s denial of habeas relief on Whitton’s claim that Saunders was constitutionally ineffective on appeal. C. Whitton’s trial counsel was not prejudicially ineffective in mar- shalling mitigatory evidence in the penalty phase. Whitton next argues that his trial counsel was constitution- ally ineffective at the penalty and mitigation phase of trial. Penalty-phase counsel has an “obligation to conduct a thor- ough investigation of the defendant’s background” for potential mitigation evidence. Williams, 529 U.S. at 396. When we deter- mine the constitutional adequacy of counsel’s performance, we look to the practices a typical, adequate counsel undertakes “in pre- paring for the sentencing phase of a capital trial.” Rompilla v. Beard, 545 U.S. 374, 380 (2005). Defense counsel’s main goal is to counter the State’s alleged aggravating circumstances with “evidence in mitigation.” Id. at 381. Still, to perform adequately, counsel need not “investigate every conceivable line of mitigating evidence no USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 55 of 72

23-10786 Opinion of the Court 55

matter how unlikely the effort would be to assist the defendant at sentencing.” Wiins v. Smith, 539 U.S. 510, 533 (2003). As for prejudice, we consider “whether there is a reasonable probability that, absent the errors, the sentencer . . . would have concluded that the balance of aggravating and mitigating circum- stances did not warrant death.” Strickland, 466 U.S. at 695. Preju- dice may occur when newly adduced descriptions and details about the defendant’s “depth of abuse . . . far exceeded what the jury was told.” Johnson v. Sec’y, DOC, 643 F.3d 907, 936 (11th Cir. 2011). We agree with Whitton that counsel could have presented a more compelling mitigation case here. But even so, we can’t say that the Florida Supreme Court’s no-prejudice determination was an unreasonable application of Strickland. To explain why, we first recount the relevant mitigation evidence from trial and from post- conviction proceedings. Then we discuss the Florida Supreme Court’s application of Strickland’s prejudice principles. 1. Additional mitigation evidence from postconviction proceedings James Tongue handled Whitton’s penalty-phase defense. Tongue had worked at the public defender’s office for less than three years and had only second-chaired a capital case. He was in private practice beforehand and had never handled a felony crimi- nal case. In postconviction proceedings, he testified that his miti- gation strategy was to humanize Whitton and focus on his present good deeds. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 56 of 72

56 Opinion of the Court 23-10786

According to his notes, Tongue spoke to three family mem- bers, all on the same day, but did not attempt to contact any of Whitton’s other (seven) siblings. Tongue did not find it necessary to travel to New York (where Whitton grew up) to speak to Whit- ton’s family members. He did not contact Whitton’s teachers, fos- ter parents, aunts and uncles (besides his aunt Ruth), or other child- hood acquaintances. Nor did he request welfare or foster-care rec- ords, records of Whitton’s brother’s death, or other records from Whitton’s childhood. At the penalty phase of trial, Whitton’s brother Royal testi- fied that their parents drank daily, someone was whipped daily, their mother Dot would put their heads in the toilet, and their fa- ther Roy threw Whitton into a wall once. Royal said the kids tore plaster off the walls of their room to burn for heat. The children slept on the floor or in a bed in one large room with a broken win- dow. And sometimes there was snow on the bed. Royal testified that Whitton was “lucky” because Royal got most of the abuse in the family. Whitton’s aunt Ruth testified at trial that she never saw Whitton’s father abuse him. And though she testified to specific incidents of Whitton’s mother’s abuse, only one involved Whitton directly: When Whitton was four or five, his mother forced him to sit outside in the snow in wet clothing rather than come into the house. Ruth also testified that Whitton’s mother fed her babies paregoric and wine and would often slap, hit, or throw her children. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 57 of 72

23-10786 Opinion of the Court 57

Valerie Brookins, Whitton’s sister, attended the trial. But the defense did not call her as a witness. Instead, the defense told her, “We don’t need you.” Yet Tongue testified that had a family mem- ber with potentially helpful testimony attended the trial, he would have called that family member. Tongue also called a mitigation expert, Dr. Larson. Dr. Lar- son spoke to Whitton for only two hours. His primary task was to evaluate Whitton for competency. Dr. Larson believed that he could not opine on non-statutory mitigators because Whitton “de- nie[d] any wrongdoing relative to the charges.” After Whitton’s conviction, two of Whitton’s siblings swore in affidavits that they believed Whitton’s mother killed their brother Michael by beating him to death, based on stories from other relatives. Dr. LeRoy Riddick, a forensic pathologist who re- viewed Michael’s autopsy, testified at the postconviction hearing that Michael “died from blunt force injuries” and “child abuse.” Other Whitton siblings confirmed that their father beat his children with “belts, branches, and fists—anything [he] could get his hands on.” In his affidavit, Timothy Whitton recounted a time his father “threw [him] head first through a wall,” after which he had to wear an eye patch. The children were starved and in con- stant fear of their parents. Their father sexually abused Valerie, and their uncle sexually abused several of the Whitton siblings. Neighbors “describe[d] the Whitton family as the worst fam- ily in terms of loving their children, taking care of their children, providing for their children that they had ever seen.” The house USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 58 of 72

58 Opinion of the Court 23-10786

had no indoor plumbing, and the beds were constantly urine- soaked. A former babysitter, who reported the parents to the De- partment of Social Services, recounted “mice and rats running around.” Whitton’s aunt recounted the “filthy” and “vermin in- fested” living conditions and lack of health care, even when the children fell ill. The children were teased for poor hygiene and lack of clothing. The family moved closer to Whitton’s great aunt for help with childcare after Michael’s death. The parents continued to drink and abuse their children. For example, Dot would sit them in a plastic chair with “wrought iron handles . . . about . . . [the] length of a love seat. . . . [A]nd she would tie them on there with rope.” She would also make “them kneel on a five-pound bag of sugar for hours or mak[e] them stand in the corner with a pound brick of butter in each hand with their arms extended.” Whitton’s parents separated, and his father requested that the children be put in foster care. By that point, the children were essentially unsupervised; they would ask neighbors to “use the phone frequently in the evenings to call local bars looking for their mother.” Dot did not cooperate with the foster-care workers but continued “gallivanting and le[aving] her children alone.” Roy “abandoned his job and left,” for which the court placed him under a child-support order. The children bounced around from place to place. Lois Langworthy, Roy’s former co-worker, took in the children. At the time, Whitton was “dirty and smelly and virtually without USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 59 of 72

23-10786 Opinion of the Court 59

clothes.” But under Langworthy’s care, Whitton “began to fit in” and “to improve in school,” even attending church. Frustrating these improvements, Roy would pick up Whitton and take him to a “booze joint,” providing Whitton with alcohol. As for Whitton’s cognitive functioning, his elementary- school principal, Max Bovee, testified that the school nurse and teachers were concerned about Whitton’s motor skills and atten- tion-deficit problems. Bovee opined that Whitton had brain dam- age and cognitive dysfunction based on his lack of coordination, learning difficulties, and hyperactivity. Whitton’s postconviction expert, Dr. Woods, diagnosed Whitton with Fetal Alcohol Spectrum Disorder, left frontal lobe impairment, alcoholism, and Post-Traumatic Stress Disorder. Dr. Woods opined that Whitton “clearly has impairments of memory sequencing and [in his] ability to weigh and deliberate,” indicative of “brain damage.” In Dr. Woods’s view, the “documented deprav- ity” that Whitton experienced “impairs one cognitively as well as the soul.” Based on his review of Whitton, Dr. Woods testified that, “to a reasonable degree of neuropsychiatric certainty,” Whitton qualified for two statutory mitigators: (1) a “substantially impaired” capacity to conform his conduct to the law and (2) an “extreme mental or emotional disturbance.” Dr. Woods further opined that Dr. Larson had not diagnosed Whitton with these impairments be- cause he lacked the necessary social and medical information. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 60 of 72

60 Opinion of the Court 23-10786

Another neuropsychologist, Dr. Barry Crown, opined that Whitton “has a significant neuropsychological impairment impact- ing language-based critical thinking, information storage and re- trieval.” 2. The Florida Supreme Court reasonably determined that missed mitigating evidence did not prejudice Whitton’s penalty-phase case. Whitton argues that Tongue “had no viable strategic reason for not delving into Petitioner’s childhood and ensuing trauma” to the extent that postconviction counsel did. And, he claims, a rea- sonable probability exists that he would have received a different sentence if the mitigation evidence uncovered postconviction had been before the court at sentencing. The State argues that we cannot reach the additional records and affidavits under AEDPA because they were not admitted as substantive evidence; rather, “they were only admitted for the lim- ited purpose of showing what Dr. Woods relied on in reaching his conclusions.” They cannot, in the State’s view, support an ineffec- tive-assistance claim. And even if we can consider it as substantive evidence, the State asserts, the additional lay-mitigation evidence was cumulative. The State also argues that the expert claim is “a non-cognizable claim of ineffective-assistance-of-expert.” The Florida Supreme Court characterized both the child- hood and mental-health-expert evidence as “cumulative” to that presented at the penalty phase, so it found that Whitton could not establish prejudice. Whitton IV, 161 So. 3d at 332. In the district USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 61 of 72

23-10786 Opinion of the Court 61

court’s view, this finding was not unreasonable because Tongue “presented a substantial mitigation case” and “[h]e made his point.” We affirm that holding. To make a prejudice determination under Strickland, courts must “evaluate the totality of the available mitigating evidence— both that adduced at trial, and the evidence adduced in the habeas proceeding in reweighing it against the evidence in aggravation.” Williams, 529 U.S. at 397–98. As for our role, we must ensure the Florida Supreme Court did not unreasonably apply federal law, as the Supreme Court of the United States has clearly established it. See Pye, 50 F.4th at 1041–42. After considering all the evidence, we conclude that Whitton’s case is materially distinguishable from the cases in which we and the Supreme Court have found ineffective penalty-phase assistance and on which Whitton relies—Williams, Wiins, Rompilla, Sears, and Porter. We begin with Williams. There, penalty-phase counsel en- tirely “failed to conduct an investigation that would have uncovered extensive records graphically describing Williams’ nightmarish childhood.” 529 U.S. at 395. This failure, the Court held, was con- stitutionally deficient performance. See id. at 396. And as for prej- udice, the Court found that the “graphic description of Williams’ childhood, filled with abuse and privation, or the reality that he was ‘borderline mentally retarded,’ might well have influenced the jury’s appraisal of his moral culpability.” Id. at 398. Here, by con- trast, counsel did not altogether fail to undertake a mitigation in- vestigation. And although the presented mitigating evidence was USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 62 of 72

62 Opinion of the Court 23-10786

certainly not as comprehensive as Whitton’s postconviction evi- dence was, the jury did hear “graphic description[s],” see id., of Whitton’s childhood abuse. Similarly, in Wiins v. Smith, the Court found that “[c]oun- sel’s decision not to expand their [mitigation] investigation beyond the PSI and the [Department of Social Services] records” was con- stitutionally deficient. 539 U.S. 510, 524 (2003). That failure was prejudicial, the Supreme Court concluded, because the “sentencing jury heard only one significant mitigating factor”—“that Wiggins had no prior convictions.” Id. at 537. And had “the jury been able to place petitioner’s excruciating life history on the mitigating side of the scale,” it reasoned, there was “a reasonable probability that at least one juror would have struck a different balance.” Id. But unlike in Wiins, Whitton’s penalty-phase jury unanimously rec- ommended a death sentence even after hearing Royal’s and Ruth’s testimony about his child abuse. And Whitton’s sentencing court credited that testimony, as well as several other mitigating factors; Whitton was not limited to just “one” theory of mitigation, as Wig- gins was. Given the several aggravators and grisly details of the murder, and the fact that the jury was already aware of Whitton’s troubled childhood (albeit in less detail), this case differs from Wig- gins. And in Rompilla, the Court found that penalty-phase counsel was constitutionally deficient for failing to investigate the defend- ant’s criminal history, on which the State intended to rely as an ag- gravator. See 545 U.S. at 390. It was “uncontested” that review of USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 63 of 72

23-10786 Opinion of the Court 63

the defendant’s criminal record would have triggered “a range of mitigation leads that no other source had opened up,” including information about “Rompilla’s childhood and mental health” that differed “from anything defense counsel had seen or heard.” Id. at 390. But here, and unlike in Rompilla, Tongue’s penalty-phase evi- dence did not paint a “benign” picture of Whitton’s “upbringing and mental capacity.” Id. at 391. Nor did Tongue make only a “few naked pleas for mercy,” id. at 393; he presented a mitigation case related to Whitton’s upbringing. And he showed that Whitton had had a difficult childhood, complete with abuse and deprivation. Rompilla does not help Whitton here—if anything, it supports the State’s position. Porter v. McCollum, is even further afield. 558 U.S. 30 (2009) (per curiam). There, penalty-phase counsel “failed to uncover and present any evidence of [the defendant’s] mental health or mental impairment, his family background, or his military service.” Id. at 40. That deficient performance, the Court concluded, was prejudi- cial because the “judge and jury at Porter’s original sentencing heard almost nothing that would humanize Porter or allow them to accurately gauge his moral culpability.” Id. at 41; cf. Ferrell v. Hall, 640 F.3d 1199, 1236 (11th Cir. 2011) (finding prejudice where “the jury heard absolutely nothing about the substantial mitigating evi- dence”); Hardwick v. Sec’y, Fla. Dep’t of Corr., 803 F.3d 541, 559 (11th Cir. 2015) (finding prejudice where “jury and the trial judge, how- ever, heard none of ” the “powerful mitigating evidence,” which “enabled the prosecutor to emphasize repeatedly in closing argu- ments that there were no mitigating circumstances in Hardwick's USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 64 of 72

64 Opinion of the Court 23-10786

case”); Daniel v. Comm’r, Ala. Dep’t of Corr., 822 F.3d 1248, 1254 (11th Cir. 2016) (similar). But here, Tongue did not fail to “present any evidence,” Por- ter, 558 U.S. at 40, of Whitton’s abusive childhood. And although additional mitigation evidence could have painted a more sympa- thetic picture, the sentencing court heard from multiple witnesses who attempted to “humanize,” id. at 41, Whitton. Most recently, in Sears v. Upton, the Court found both defi- cient performance and prejudice where “evidence relating to [the defendant’s] cognitive impairments and childhood difficulties was not brought to light at the time he was sentenced to death.” 561 U.S. 945, 946 (2010) (per curiam). Penalty-phase counsel “presented evidence describing [the defendant’s] childhood as stable, loving, and essentially without incident,” but the opposite was true. Id. at 947–48. Here, though, Tongue did no such thing. He portrayed Whitton’s childhood as difficult. The sentencing court explicitly afforded substantial weight to Whitton’s background mitigation. And Whitton’s postconvic- tion evidence, while certainly compelling, told a “more detailed version of the same story told at trial,” not an entirely new story. Holsey v. Warden, Georgia Diagnostic Prison, 694 F.3d 1230, 1261–64, 1266 (11th Cir. 2012) (finding no prejudice where “the basic story of his troubled, abusive childhood was nonetheless known to the sentencing court,” even if postconviction counsel presented “more details” and “different examples” (cleaned up)); Cullen v. Pinholster, 563 U.S. 170, 200 (2011) (same, where the ‘“new’ evidence,” USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 65 of 72

23-10786 Opinion of the Court 65

including school records and testimony from additional siblings as to child abuse, “largely duplicated the mitigation evidence at trial”); Ferguson v. Comm’r, Ala. Dep’t of Corr., 69 F.4th 1243, 1261 (11th Cir. 2023) (“While more mitigation witnesses could have presented more details or different examples of these unfortunate aspects of [the defendant’s] life, these aspects were nonetheless known to the sentencing jury and judge.”). Not only that, but we have “repeat- edly held that even extensive mitigating evidence wouldn’t have been reasonably likely to change the outcome of sentencing in light of a particularly heinous crime and significant aggravating factors.” Pye, 50 F.4th at 1049 (collecting cases). And that was the case here. To be sure, “counsel’s effort to present some mitigation evi- dence” does not “foreclose an inquiry into whether a facially defi- cient mitigation investigation might have prejudiced the defend- ant.” Sears, 561 U.S. at 955. But a defendant is not prejudiced just because counsel could have presented more mitigation evidence. And although we may not have necessarily characterized the pow- erful and disturbing evidence uncovered in postconviction as “cu- mulative,” the Florida Supreme Court’s determination to that ef- fect was not an unreasonable application of Strickland. See 28 U.S.C. § 2254(d)(1). As the district court found, Tongue “made his point,” and the Sixth Amendment does not require him to have made it as persuasively or comprehensively as possible. Cf. King v. Warden, Ga. Diagnostic Prison, 69 F.4th 856, 875 (11th Cir. 2023) (ex- plaining “counsel is not ineffective whenever more witnesses could have been called”). USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 66 of 72

66 Opinion of the Court 23-10786

As to Whitton’s claim that Dr. Larson was inadequate, Whit- ton conflates ineffective assistance of counsel with ineffective assis- tance of an expert. True, we have recognized the “particularly crit- ical interrelation between expert psychiatric assistance and mini- mally effective representation of counsel.” United States v. Fessel, 531 F.2d 1278, 1279 (5th Cir. 1979). But Fessel involved counsel’s failure to obtain expert psychiatric testimony altogether, not an ar- gument that the expert’s report was not sufficiently comprehen- sive. And because evidence that Whitton had some mental defects was already before the jury, we cannot conclude that Tongue’s fail- ure to obtain a more comprehensive psychological evaluation was prejudicial. Cf. Thornell v. Jones, 602 U.S. 154, 166–71 (2024) (finding no prejudice where state court already received testimony that the defendant “suffers from a major mental illness” and has “cognitive impairment,” even if not in the same level of detail, in light of the “weighty aggravating circumstances present”); Sochor v. Sec’y Dep’t of Corr., 685 F.3d 1016, 1031 (11th Cir. 2012) (same, where petitioner “fail[ed] to explain why the judge and jury would have been any more likely to accept such a theory when the underlying mental health issue” was one diagnosis as opposed to another). In sum, we affirm the district court’s conclusion that the Florida Supreme Court’s application of Strickland was not unrea- sonable. USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 67 of 72

23-10786 Opinion of the Court 67

D. Whitton cannot show any Doyle violations prejudiced the out- come of his trial. Finally, Whitton argues that three references to his invoca- tion of his right to remain silent violated his Doyle rights and were not harmless. Those references occurred during (1) Officer Cot- ton’s testimony, (2) Whitton’s testimony, and (3) the prosecutor’s closing argument. First, Officer Cotton’s testimony contained the following ex- change: A: [. . .] And at 0515 a.m., he decided, once we were getting much closer to what we felt was the truth and we were tightening down on him being at the murder scene, he decided he did not want to talk to us any- more. Q: Did he tell you he went to the police at midnight on the 2nd (sic) and said, “Hey, I found a body in a room; my friend is dead.”? A: No, sir, he did not. Second, during Whiton’s testimony, the prosecutor asked, Q: And what did you do when you told him, what did you do after that when you told him you didn’t go back there? A: I told him -- Q: You didn’t say anymore, did you? USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 68 of 72

68 Opinion of the Court 23-10786

A: Excuse me? Q: You didn’t say anymore then, did you? A: No, I did tell him after awhile that I did go back there. Q: And when you told him that, then you didn’t say nothing else. A: No, sir. Defense counsel did not object to either line of questioning. And third, in his closing statement at trial, the prosecutor said the following: But in the last part of that interview, before the de- fendant says, “I’m not talking to you anymore,” he tells him, “I went back over there, I walked in, and I saw my friend dead, and I left.” Then he doesn’t say anything else. He realizes at that point, “Uh-oh.” Defense counsel objected and moved for a mistrial based on those “improper comments” as to Whitton’s invocation of his right to remain silent. The court denied the defense’s motion for a mis- trial but offered to give a curative instruction. Defense counsel de- clined so as not to “call[] attention to exactly what he’s done” and “compound[] the problem.” The court did, however, instruct the jury at the start of trial that it could not “draw any inference of guilt” from “the exercise of a defendant’s right to remain silent.” USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 69 of 72

23-10786 Opinion of the Court 69

On direct appeal, the Florida Supreme Court concluded that there was “no reasonable possibility that the improper comment[s] contributed to Whitton’s conviction,” especially given the “sub- stantial amount of permissible evidence that conclusively proves Whitton’s guilt.” Whitton I, 649 So. 2d at 864–66. On federal ha- beas review, the district court held that this finding was not unrea- sonable given “the overall strength of the evidence as well as what Mr. Whitton said before invoking his right to remain silent.” Whitton argues that this conclusion was based on an unrea- sonable determination of the facts. See 28 U.S.C. § 2254(d)(2). In response, the State contends, he has abandoned any challenge un- der § 2254(d)(1). Also, in the State’s view, because defense counsel refused a curative instruction at trial, “law and justice do not permit [Whitton] to take advantage of his refusal in federal court decades later.” But regardless of whether § 2254(d)(1) or (d)(2) is the oper- ative provision, or whether the State’s “law and justice” argument has merit, Whitton’s challenge fails because he cannot prove preju- dice. To briefly recap the relevant law, a prosecutor cannot use a defendant’s invocation of his right to remain silent to impeach his exculpatory testimony. See Doyle v. Ohio, 426 U.S. 610, 619 (1986). The Doyle rule “rests on the fundamental unfairness of implicitly assuring a suspect that his silence will not be used against him and then using his silence to impeach an explanation subsequently of- fered at trial.” Wainwright v. Greenfield, 474 U.S. 284, 291 (1986) (ci- tation and internal quotation marks omitted). But Doyle errors are USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 70 of 72

70 Opinion of the Court 23-10786

subject to Brecht’s harmlessness inquiry. See, e.g., Hill v. Turpin, 135 F.3d 1411, 1416 (11th Cir. 1998). On the one hand, “we have declined to find Doyle error harmless in those cases where the prosecutor returned repeatedly to the defendant’s post-Miranda silence throughout trial to impeach a plausible exculpatory story offered by the defendant.” Id. at 1417. For example, in Hill, “on four separate occasions during [the defendant’s] trial, the prosecution brought to the jury’s attention Hill’s post-Miranda silence and request for counsel,” in violation of the court’s order in limine. Id. at 1414. The court admonished the prosecutor and gave two curative instructions. See id. at 1415. But the prosecutor continued to reference the defendant’s invocation of his Miranda rights, though his “remarks during closing argument [were] somewhat ambiguous.” See id. at 1415 n.4. We held that the Doyle violations were not harmless given “the importance of Hill’s credibility to his defense, the repeated and deliberate nature of the prosecution’s Doyle violations, and the significant weaknesses in the state’s case against Hill.” Id. at 1416–17; see also United States v. Ten- orio, 69 F.3d 1103, 1107 (11th Cir. 1995) (Doyle error not harmless where the defendant’s “silence was the touchstone of the govern- ment’s case-in-chief, its cross-examination of the defendant, and its closing argument” and “could reasonably have been the basis for the [jury’s] guilty verdict”). On the other hand, “we have repeatedly held Doyle error harmless where the violation consisted of only a single reference to the defendant’s post-Miranda silence during the course of a trial USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 71 of 72

23-10786 Opinion of the Court 71

at which the government’s evidence was otherwise overwhelm- ing.” Hill, 135 F.3d at 1417 (first citing United States v. Gabay, 923 F.2d 1536, 1541 (11th Cir. 1991); then citing United States v. Ruz–Sal- azar, 764 F.2d 1433, 1437 (11th Cir. 1985); and then citing Sullivan v. Alabama, 666 F.2d 478, 485 (11th Cir. 1982)); see also Prevatte v. French, 547 F.3d 1300, 1305 (11th Cir. 2008) (finding Doyle error to be harm- less where we were “overwhelmed by the evidence of [the defend- ant’s] guilt, separate and apart from any evidence of his post-arrest silence”); United States v. Miller, 255 F.3d 1282, 1286 (11th Cir. 2001) (same, where improper questioning “took only moments of the trial” and “the government’s evidence of guilt was strong”). That includes cases where the Doyle violation was ‘“isolated’ or ‘uninten- tional’ or promptly addressed by a curative instruction from the trial court.” Hill, 135 F.3d at 1417. Whitton contends the Doyle violations here “were repeated throughout the trial, and were used to show both consciousness of guilt and to impeach Petitioner’s plausible explanation for not ini- tially wanting to admit he returned to the motel.” But we conclude this case is closer to Brecht itself, where the Supreme Court found the Doyle errors to be harmless because they “were infrequent, comprising less than two pages of the 900–page trial transcript,” and “the State’s evidence of guilt was, if not over- whelming, certainly weighty.” 507 U.S. at 639. Just as in Brecht, Whitton cites three comments throughout the entire trial, contrasted with the substantial evidence of his guilt: the blood spatter on his boots, the motel clerk’s testimony, his own USCA11 Case: 23-10786 Document: 50-1 Date Filed: 05/06/2025 Page: 72 of 72

72 Opinion of the Court 23-10786

admission that he returned to the motel (and originally lied about it to the police), and other circumstantial evidence. (That is, of course, ignoring Ozio and McCullough’s testimony that Whitton confessed to the murder). Although the prosecutor’s “Uh-oh” comment was certainly improper, it was five lines in 30 pages of closing argument. And defense counsel declined a curative instruc- tion—a declination that was a deliberate, strategic decision to not call more attention to the matter. Given this case’s resemblance to Brecht, we conclude that Whitton cannot meet his burden to show that the Florida Supreme Court’s harmless-error conclusion was based on an unreasonable determination of the facts or the law. See 28 U.S.C. § 2254(d)(1)–(2). We thus affirm the district court’s denial of relief on Whitton’s Doyle claim. IV. CONCLUSION Accordingly, we affirm the district court’s denial of Davis’s federal habeas petition. AFFIRMED.