Herod v. Guerrero

Opinion

United States Court of Appeals for the Fifth Circuit United States Court of Appeals Fifth Circuit

____________ FILED August 25, 2026

No. 25-40247 ____________ Lyle W. Cayce Clerk

Richard Anthony Herod,

Plaintiff—Appellee,

versus

Eric Guerrero, Director, Texas Department of Criminal Justice, Correctional Institutions Division,

Defendant—Appellant.

______________________________

Appeal from the United States District Court for the Southern District of Texas USDC No. 3:15-CV-338

______________________________

ON PETITION FOR REHEARING Before Richman, Engelhardt, and Wilson, Circuit Judges. * Kurt D. Engelhardt, Circuit Judge: ** Treating the petition for rehearing en banc as a petition for panel rehearing, the petition for panel rehearing is GRANTED. No member of the panel nor judge in regular active service of the court having requested

_____________________

*

Judge Richman concurs in the judgment only.

**

This opinion is not designated for publication. See 5th Cir. R. 47.5.

No. 25-40247

that the court be polled on rehearing en banc (Fed. R. App. and 5th Cir. R. 40), the petition for rehearing en banc is DENIED. We withdraw our prior opinion, Herod v. Guerrero, No. 25-40247, 2026 WL 1419381 (5th Cir. May 20, 2026) (per curiam), and substitute the following.

Appellee Richard Anthony Herod was convicted of aggravated sexual assault and aggravated robbery in Texas state court in 2012. At trial, the prosecution introduced expert testimony interpreting DNA evidence found at the crime scene. The expert testified that Herod could not be excluded as a contributor to the DNA mixture found on one item at the scene—a white t-shirt used to blindfold one of the victims. Several years later, the Texas Department of Public Safety (“DPS”) issued a supplementary report stating that Herod was excluded as a contributor to that DNA mixture under updated interpretation protocols. Finding constitutional errors under Brady v. Maryland, 373 U.S. 83 (1963), and Napue v. Illinois, 360 U.S. 264 (1959), the district court granted Herod federal habeas relief under 28 U.S.C. § 2254. See Herod v. Guerrero, No. 15-0338, 2025 WL 1001609 (S.D. Tex. Apr. 3, 2025). Although Herod has overcome the procedural default of his claims in state court, the district court erroneously granted relief on the merits of his Brady and Napue claims. We REVERSE the district court’s judgment and REMAND for proceedings on the remainder of Herod’s habeas petition.

I.

In January 2010, armed robbers invaded Alissia and Ronnie Gallagher’s Texas City home. While inside the home, one of the robbers secured codeine and cash from Ronnie, bound his hands with zip ties, blindfolded him with a white t-shirt, and then proceeded to sexually assault Alissia in view of the couple’s two children. The state’s evidence included testimony from Alissia, Ronnie, and Herod’s girlfriend, Holly Kelly. The state also introduced expert DNA testimony at trial. Clare Browder, a DPS

3

No. 25-40247

analyst, testified that Herod was excluded as a contributor to DNA mixtures found on almost all the physical evidence left at the scene. Browder used the Combined Probability of Inclusion (“CPI”) method to interpret the DNA mixture. For the white t-shirt, Browder opined that Herod could not be excluded as a contributor. Browder testified to the “probability that you could pick up [a person] with that combination of alleles” as 1 in 87 Caucasians. She explained that the higher the second number, the “less frequent allele combination.” By comparison, Ronnie—who wore the t-shirt blindfold—could not be excluded at 14 of 16 locations on the body of the white t-shirt. For his allele combination, the probabilities of selecting a person at random who could be the contributor were 1 in 9.843 million for Caucasians, 1 in 102.5 million for African Americans, and 1 in 1.786 million for Hispanics.

The defense called its own expert, Melva Ketchum, who agreed that Herod technically could not be excluded from the DNA mixture found on the t-shirt. She also testified that Herod realistically “should be excluded,” given that he was excluded at eight of 16 allele markers. Herod emphasizes that Ketchum’s testimony was impeached by her training in veterinary medicine, running of an unaccredited laboratory, and involvement “in the DNA quest for Bigfoot.”

The jury convicted Herod of aggravated robbery and aggravated sexual assault. Herod, 2025 WL 1001609, at *1. Sentenced to 99 years in prison, Herod appealed his conviction and sentence through the state court system. Id. He then filed an initial state habeas petition in February 2015, which was denied by both the state trial court and the Texas Court of Criminal Appeals (“CCA”). Id.

Herod filed his first federal habeas petition in December 2015. Id.

While that petition was pending, Herod received a letter from the state

4

No. 25-40247

district attorney instructing that recalculation of DNA evidence was available in his case. DPS issued a supplemental report in 2017, explaining that Herod was excluded from all DNA evidence, including the mixture on the white t-shirt. The district court stayed proceedings so Herod could file a second state habeas petition asserting claims based on the DPS report. Id. The state trial court entered findings of fact and conclusions of law rejecting Herod’s claims, and the CCA dismissed the application on procedural grounds. Id.

In January 2020, Herod returned to federal court with an amended habeas petition, raising nine claims for relief. Id. at *2. Herod’s first claim alleged due process violations, encompassing both Brady and Napue arguments. He asserted that the state suppressed (1) the fact that Herod was excluded as a contributor to the DNA on the white t-shirt; and (2) the existence of scientific debate and difficulties surrounding the state’s method of DNA mixture interpretation. He likewise asserted the state’s expert— Browder—falsely testified that “Herod could not be excluded as a contributor of the DNA found on the white shirt,” this was “the strongest language you could use in a mixture case,” and “[t]here’s no difficulty with the statistics.”

The district court concluded that Herod had overcome the procedural default of his Brady and Napue claims. It reached the merits, holding Herod was entitled to § 2254 relief on both fronts. Id. at *22, 27. “In reviewing a grant of habeas relief, we review issues of law de novo and findings of fact for clear error.” Prible v. Lumpkin, 43 F.4th 501, 513 (5th Cir. 2022). And we apply “the same standards to the state court’s decision as did the district court.” Grace v. Hooper, 123 F.4th 800, 804 (5th Cir. 2024) (internal quotation marks and citation omitted).

5

No. 25-40247

II.

Under the Antiterrorism and Effective Death Penalty Act (“AEDPA”), the federal courts “shall entertain an application for a writ of habeas corpus” for a person in custody under a state court judgment “only on the ground that he is in custody in violation of the Constitution or laws . . . of the United States.” 28 U.S.C. § 2254(a). AEDPA ordinarily requires highly deferential review of the state court’s habeas decision. See id. § 2254(d). These deferential review standards do not apply, though, when the claims were “not adjudicated on the merits in the state court.” Nelson v. Davis, 952 F.3d 651, 658 (5th Cir. 2020) (citation modified). In that case—a procedural default—federal courts cannot review the claims “absent a showing of cause and prejudice to excuse the default.” Prible, 43 F.4th at 513.

Cause is “some objective factor external to the defense” that “impeded counsel’s efforts to raise the claim in state court.” Id. at 513 (quoting Canales v. Stephens, 765 F.3d 551, 562 (5th Cir. 2014)). Prejudice means that “the errors ‘worked to [the petitioner’s] actual and substantial disadvantage, infecting his entire trial with error of constitutional dimensions.’” Id. at 514 (quoting Smith v. Quarterman, 515 F.3d 392, 403 (5th Cir. 2008)). The “possibility of prejudice” is insufficient. Hughes v. Quarterman, 530 F.3d 336, 341 (5th Cir. 2008) (citation modified). A petitioner must show “pervasive actual prejudice,” meaning that he “was denied fundamental fairness at trial.” Murray v. Carrier, 477 U.S. 478, 494 (1986) (citation modified). Brady violations “can provide cause and prejudice” to overcome the procedural bar. Prible, 43 F.4th at 514 (citation modified). Brady’s suppression element tracks cause, while its materiality element tracks prejudice. Id.

6

No. 25-40247

The parties here agree that Herod’s claims are procedurally defaulted.

After all, the CCA dismissed Herod’s second habeas petition on state procedural grounds. See Ex parte Torres, 943 S.W.2d 469, 472 (Tex. Crim. App. 1997) (en banc) (“In our writ jurisprudence, a ‘denial’ signifies that we addressed and rejected the merits of a particular claim while a ‘dismissal’ means that we declined to consider the claim for reasons unrelated to the claim’s merits.”). And at all stages of this case, the state has conceded cause. Prejudice, however, it vigorously disputes. The district court ultimately agreed with Herod, concluding “Browder’s false testimony . . . worked to Herod’s actual and substantial disadvantage, infecting his entire trial with error of constitutional dimensions.” Herod, 2025 WL 1001609, at *21. We review the district court’s cause-and-prejudice conclusion de novo. Prible, 43 F.4th at 513.

Using the Brady-materiality-as-prejudice proxy, we evaluate the materiality of allegedly suppressed evidence “collectively, not item by item.” Floyd v. Vannoy, 894 F.3d 143, 162 (5th Cir. 2018) (per curiam) (citation modified) (assessing Brady materiality). This is “not a sufficiency of the evidence test.” United States v. Brumfield, 89 F.4th 506, 516 (5th Cir. 2023) (citation modified). Instead, we question whether Herod “received a fair trial”—one “resulting in a verdict worthy of confidence in the absence of the favorable evidence.” LaCaze v. Warden La. Corr. Inst. for Women, 645 F.3d 728, 736 (5th Cir. 2011) (citation modified). Disclosing the “fact” that Herod was actually excluded as a contributor to the mixture on the white t- shirt, the existence of scientific debate surrounding Browder’s DNA interpretation methods, and that perhaps there was some “difficulty with the statistics” presents a “reasonable probability” that “the result of the proceeding would have been different.” Miller v. Dretke, 404 F.3d 908, 914 (5th Cir. 2005) (citation modified) (concerning Brady materiality).

7

No. 25-40247

For one, the DNA mixture interpretation was the only physical evidence tying Herod to the crime scene. Underscoring its importance, the state emphasized Herod’s DNA in both its opening and closing statements. It explained that “DNA is better than a fingerprint. It’s your genetic fingerprint. And the defendant’s DNA is absolutely right there on that blindfold.” Although the defense attempted to minimize the DNA testimony, the prosecution rebutted those arguments, reflecting just how crucial the DNA evidence was to the prosecution’s trial strategy. See Kyles v. Whitley, 514 U.S. 419, 444 (1995) (relying on prosecution’s framing of the evidence in closing arguments to emphasize the “likely damage” from not disclosing certain witness statements); Banks v. Dretke, 540 U.S. 668, 700– 03 (2004) (explaining that the “stress placed by the prosecution” on uncorroborated testimony by paid informant whose status was suppressed contributes to materiality finding).

For another, the evidence of Herod’s guilt was not as “overwhelming” as the state makes it out to be. The state emphasizes Ronnie’s, Alissia’s, and Kelly’s testimony. But, as the district court explained, the “non-DNA evidence . . . was challenged at trial and is further weakened by evidence in the state habeas record.” Herod, 2025 WL 1001609, at *21. At trial, Ronnie identified Herod as one of the robbers, and Alissia identified the “white male” as her assailant. Holly Kelly testified that she drove Herod and another individual from Santa Fe, Texas, to Texas City, Texas, on the night of the crime, watched them take “beanies” and handguns out of the vehicle, and return later with codeine and several hundred dollars in cash. But Herod countered with alibi testimony from his mother, stepfather, and brother. He was at his mother’s home on the night of the crime watching the college football national championship. The defense also emphasized inconsistencies in the testimony elicited by the prosecution,

8

No. 25-40247

including information the Gallaghers withheld in their original statements to police and discrepancies between Kelly’s and the Gallaghers’ testimony. 1 At bottom, the prosecution’s case turned, in large part, on the DNA evidence. As the district court’s opinion carefully outlines, the defense challenged nearly every other piece of prosecution evidence. See id. at *17– 21; see also United States v. Ausby, 916 F.3d 1089, 1095 (D.C. Cir. 2019) (concluding materiality was satisfied where admittedly false expert testimony was the “primary evidence that directly contradicted” the defense’s case theory and the defense “plausibly explained the remaining evidence”). This is not a situation where the evidence of guilt was so overwhelming as to undermine the import of the challenged DNA evidence. See Medellin v. Dretke, 371 F.3d 270, 281 (5th Cir. 2004) (concluding allegedly suppressed evidence was not material “in light of the overwhelming evidence establishing his guilt”); Wearry v. Cain, 577 U.S. 385, 392–93 (2016) (recognizing materiality of evidence where the state’s trial evidence resembled “a house of cards, built on the jury crediting [its witness’s] account rather than [defendant’s] alibi”). The allegedly suppressed scientific evidence and allegedly false DNA-related testimony infected

_____________________

1 Cell phone evidence also tied Herod to the crime. Two phones were found in Herod’s truck on the day of his arrest. Cell tower data reflected that one of the phones traveled between Santa Fe and Texas City on the evening in question, consistent with the timing of the crime as reported by the Gallaghers and Kelly. Someone using one of those phones had called Alissia at her reported time, confirmed by records from the cell provider. But there were significant oddities in the cell phone evidence—the phone that called Alissia was registered just 11 days after Alissia’s phone registration, to someone listing the same address (a California P.O. box), with the same account holder birthday, and the same recovery phone number. Some of this information was not presented to the jury—and therefore is part of the basis for Herod’s ineffective assistance of counsel claims. See Herod, 2025 WL 1001609, at *20 (“Although not presented by Herod’s counsel, evidence in the trial record showed that the x1537 phone had multiple unexplained similarities to Alissia Gallagher’s phone.”).

9

No. 25-40247

Herod’s “entire trial with error of constitutional dimensions.” See Prible, 43 F.4th at 514 (explaining petitioner’s burden to overcome procedural default). Herod has shown prejudice sufficient to overcome the procedural bar.

III.

Having overcome the procedural default, we next assess whether the non-retroactivity principle in Teague v. Lane, 489 U.S. 288 (1989), bars Herod’s Brady and Napue claims.

Teague prohibits federal habeas courts from applying “new constitutional rules of criminal procedure” to convictions and sentences finalized “before the new rules are announced.” 489 U.S. at 310. This non- retroactivity rule, much like the procedural default bar, is a “threshold question in every habeas case.” Matthew v. Johnson, 201 F.3d 353, 359 (5th Cir. 2000) (citation modified); Lambrix v. Singletary, 520 U.S. 518, 524 (1997) (recognizing that Teague-bar analysis may follow procedural-bar analysis but precedes the merits). Courts must assess “whether granting [petitioner] the relief he seeks would create a new rule of constitutional law.” Graham v. Collins, 506 U.S. 461, 466 (1993) (emphasis added) (citation modified); id. at 467 (“New rules will not be applied or announced in cases on collateral review unless they fall into one of two exceptions.” (emphasis added) (citation modified)).

The Teague analysis, largely unaddressed in the parties’ briefing, 2 proceeds in three steps: First, identifying the date on which Herod’s

_____________________

2 The state raised Teague for the first time on appeal. Herod asks the court to deem any Teague arguments waived. True, the state “may waive the Teague bar implicitly by failing to raise it.” Blakenship v. Johnson, 118 F.3d 312, 316 (5th Cir. 1997) (recognizing that Teague is not an unwaivable jurisdictional doctrine). But “absent compelling reasons to the contrary,” the court must “apply Teague even when it has been implicitly waived by the State” by raising it anew on appeal. Jackson v. Johnson, 217 F.3d 360, 361 (5th Cir. 2000); Jones v. Davis, 890 F.3d 559, 564 (5th Cir. 2018). Herod has provided no compelling

10

No. 25-40247

conviction and sentence were finalized. Matthew, 201 F.3d at 359. Second, “surveying the legal landscape as it then existed” and assessing whether the “rule he seeks was required by the Constitution” at the time. Id. (citation modified). And, third, if a new rule is needed, whether it falls within either of two narrow exceptions to Teague’s non-retroactivity bar. Id.

A.

A “conviction and sentence become final for purposes of retroactivity analysis when the availability of direct appeal to the state courts has been exhausted and the time for filing a petition for a writ of certiorari has elapsed or a timely filed petition has been finally denied.” Caspari v. Bohlen, 510 U.S. 383, 390 (1994). Herod’s conviction and sentence were finalized in June 2014. Herod, 2025 WL 1001609, at *1 (CCA denied Herod’s petition for discretionary review on March 12, 2014); Sup. Ct. R. 13 (requiring petitions for certiorari in criminal cases to be filed within 90 days of state court order denying discretionary review).

B.

Herod’s claims require new rules if, in June 2014, “it would have been anything but clear to reasonable jurists” that Herod’s trial violated the Constitution. Graham, 506 U.S. at 468. New rules “break[] new ground, impose[] a new obligation on the states or the federal government, or [were] not dictated by precedent existing at the time the defendant’s conviction became final.” Hughes v. Dretke, 412 F.3d 582, 591 (5th Cir. 2005). “Dictated by precedent means that no other interpretation was reasonable.” United States v. London, 937 F.3d 502, 507 (5th Cir. 2019) (citation modified); Lambrix, 520 U.S. at 538. “Unless reasonable jurists hearing the defendant’s

_____________________

reasons not to assess Teague and asserted at oral argument only that his claims break no new constitutional ground.

11

No. 25-40247

claim at the time his conviction became final would have felt compelled by existing precedent to rule in his favor, a federal habeas court is barred from doing so on collateral review.” Hughes, 412 F.3d at 591. Note, however, that “when all we do is apply a general standard to the kind of factual circumstances it was meant to address, we will rarely state a new rule for Teague purposes.” Chaidez v. United States, 568 U.S. 342, 348 (2013).

Consider, first, Herod’s arguments that do not ask the court to craft new rules: his Brady argument based on the nondisclosure of scientific debate and his Napue arguments. See infra Part IV.A.2, B. Those arguments ask the court to apply rules of “general application,” “designed for the specific purpose of evaluating a myriad of factual contexts.” Chaidez, 568 U.S. at 348 (citation modified). As such, ruling in Herod’s favor on those claims would not be “so novel that it forges a new rule.” Id. (internal quotation marks and citation omitted). Legal principles from the cases Herod cites “speak directly” to the issues he raises. See Saffle v. Parks, 494 U.S. 484, 490 (1990) (addressing situation where the existing cases did not “speak directly, if at all, to the issue presented”).

But Herod’s Brady argument that the state unconstitutionally suppressed his exclusion is barred by Teague. Herod does not identify a single person or document, from before or during trial, that suggests the state knew or believed Herod was excluded as a contributor. “Brady claims involve the discovery, after trial, of information which had been known to the prosecution but unknown to the defense.” West v. Johnson, 92 F.3d 1385, 1399 (5th Cir. 1996) (emphasis added) (citation modified); see also Caitlin Plummer & Imran Syed, “Shifted Science” and Post-Conviction Relief, 8 Stan. J. C.R. & C.L. 259, 285 (2012) (recognizing that Brady claim necessarily fails in “the case of true shifted science” because “a prosecutor cannot disclose something that she does not know and cannot be expected to know”). Brady does not require the state to perform different forensic tests to create

12

No. 25-40247

impeachment or exculpatory evidence. See United States v. Alverio-Melendez, 640 F.3d 412, 424 (1st Cir. 2011) (“The failure to create exculpatory evidence does not constitute a Brady violation.”); United States v. Garvin, 270 F. App’x 141, 144 (3d Cir. 2008) (“Although it hardly bears mention, an implicit prerequisite of any Brady claim is that favorable, material evidence actually exists.”); United States v. Nguyen, 98 F. App’x 608, 609 (9th Cir. 2004) (“Brady does not require the government to interview witnesses or otherwise create exculpatory evidence not then in existence.”). Herod cites no authority suggesting that the prosecution suppresses a fact absent evidence the prosecution even knew about it. Transforming this record into suppressed Brady material goes far beyond what precedent demanded at the time of trial—and demands even now. See West, 92 F.3d at 1399. Therefore, relief is “barred by Teague.” Id.

C.

Two exceptions permit the court to apply new rules to old convictions, but neither aids Herod here. The first applies to rules placing certain conduct “beyond the power of the criminal law-making authority to proscribe.” Gilmore v. Taylor, 508 U.S. 333, 345 (1993) (citation modified). The second encompasses a “small core” of “watershed rules of criminal procedure.” Id. (citation modified). The paradigmatic example is the guarantee of representation in criminal trials from Gideon v. Wainwright, 372 U.S. 335 (1963). Saffle, 494 U.S. at 495. Even if valid, the new Brady rule that Herod asks the court to employ has “none of the primacy and centrality of the rule adopted in Gideon.” Id.; Edwards v. Vannoy, 593 U.S. 255, 264 (2021) (noting that the Supreme Court has “never found that any new procedural rule” satisfies the second exception). Neither Teague exception applies.

13

No. 25-40247

IV.

Initial hurdles cleared, we turn to the merits. Recall that § 2254(d)’s deferential review standards do not apply where the claims were procedurally defaulted in state court, as here. Nelson, 952 F.3d at 658. But § 2254(e)(1) still applies, even where § 2254(d) does not. See Austin v. Davis, 876 F.3d 757, 778 & n.196 (5th Cir. 2017). So, this court must presume correct any state court factual findings. 28 U.S.C. § 2254(e)(1). An applicant can overcome this presumption only by clear and convincing evidence. Id. The relevant facts here are the state habeas trial court’s findings of fact. See Murphy v. Davis, 901 F.3d 578, 595 (5th Cir. 2018) (recognizing that Texas trial court issued findings of fact that were not “directly inconsistent” with CCA’s procedural dismissal and were entitled to the presumption of correctness).

We begin with Herod’s Brady claim.

A.

Three elements make up a Brady violation: “(1) the evidence at issue was favorable to the accused, either because it was exculpatory or impeaching; (2) the evidence was suppressed by the prosecution; and (3) the evidence was material.” Grace, 123 F.4th at 805 (quoting United States v. Glenn, 935 F.3d 313, 319 (5th Cir. 2019)).

1.

Start with a threshold matter. The district court’s Brady analysis assesses only materiality because it concluded the state conceded Brady’s favorability and suppression elements. The state failed to brief the Brady claim’s merits in the district court, relying only on its procedural default arguments. Here, the state argues it neither conceded, nor intended to

14

No. 25-40247

concede, suppression on the merits of the Brady claim. The state maintains that its concession was limited to cause.

The district court deemed Brady’s first two elements conceded because of this paragraph of the state’s brief:

As Herod points out, cause can parallel the first two Brady requirements: 1) that evidence favorable to the accused was 2) suppressed by the State, either willfully or inadvertently. Like the defendant in St[r]ickler [v. Greene, 527 U.S. 263 (1999)], where the Supreme Court determined that the State suppressed documents favorable to the accused for impeachment purposes, it appears that Herod can meet the first two Brady requirements and can therefore show cause why he did not raise the claim when he filed his first state applications.

Those sentences appear under a heading labeled “Cause/State’s suppression of favorable evidence under Brady.” And the paragraph compares Herod to a defendant in another case, where the prosecution unquestionably suppressed impeachment evidence. Strickler, 527 U.S. at 282 (“[T]here is no dispute about the fact that [five documents] were known to the Commonwealth but not disclosed to trial counsel.”). The district court concluded that the state had not argued “that the DNA evidence was not suppressed or not favorable.” Herod, 2025 WL 1001609, at *22 n. 13.

But the very next paragraph of the state’s district court brief indicates the scope of its concession. It reads:

In his subsequent state applications, Herod presented specific allegations and evidence that the relevant scientific evidence is currently available which was not available at the time of his trial because this evidence was unascertainable through the exercise of reasonable diligence. He has also shown that the DNA evidence was not available until February 9, 2017, two years after he filed his first state habeas applications. Thus, Herod has

15

No. 25-40247

established cause for his failure to bring the claim in state court earlier.

To prove cause, a petitioner must prove an objective external factor by either “active governmental interference” or “the reasonable unavailability of the factual or legal basis for the claim.” Hughes, 530 F.3d at 341. And the state expressly conceded that the scientific evidence—specifically the 2017 DNA mixture report—was not available at the time of Herod’s trial through “the exercise of reasonable diligence.” But the state argued at a district court hearing that Herod cannot establish cause for his Brady claim regarding the DNA interpretation methods, analogizing to a Fourth Circuit case addressing suppression under similar circumstances. See United States v. Higgs, 663 F.3d 726, 738 (4th Cir. 2011) (concluding no Brady claim existed because reports about concerns with a forensic testing method were available to the defense, and therefore the state did not violate its Brady obligations). These arguments indicate that the state never intended to concede that it suppressed the criticisms of its DNA methodology.

Granting habeas relief “imposes significant costs on state criminal justice systems.” Shinn v. Ramirez, 596 U.S. 366, 377 (2022). As such, it is designed to “guard only against extreme malfunctions.” Id. (citation modified). To be sure, the state neglected its Brady arguments before the district court. But context reveals its concession was limited to cause. Given the magnitude of granting habeas relief and the context of the state’s concession, we will review the state’s suppression arguments.

2.

Having rejected Herod’s first Brady argument under Teague, we reject Herod’s remaining Brady argument—that the state suppressed the existence of scientific debate about its DNA mixture interpretation method—on its merits. Herod asserts that the prosecution “failed to turn over that there

16

No. 25-40247

were indeed difficulties with interpreting mixture DNA evidence at the time of Herod’s trial.” In Herod’s view, “the favorable evidence suppressed . . . is that DPS employees were aware at the time of trial that there was great disagreement in the forensic DNA community about the validity of the mixture interpretation techniques used by DPS.” The state argues that this evidence could not be suppressed because it was in the public domain at the time.

Information is not suppressed if the defense could have discovered it through reasonable diligence. See, e.g., Pondexter v. Quarterman, 537 F.3d 511, 526 (5th Cir. 2008); United States v. Runyan, 290 F.3d 223, 246 (5th Cir. 2002) (“Evidence is not suppressed if the defendant knows or should know of the essential facts that would enable him to take advantage of it.” (citation modified)). The scientific debate surrounding the CPI method was available online and in various publications well before Herod’s trial. All signs point to this being a publicly available scientific debate, and therefore not something that the prosecution could “hide” from the defendant and require him to “seek.” See Banks, 540 U.S. at 696. With “some industry,” Herod could have cross-examined Browder about the debate surrounding the CPI method or put on evidence about the changing industry guidelines and other scientists’ concerns. 3 Runyan, 290 F.3d at 246.

_____________________

3 To the extent Herod argues the state suppressed the suspect-driven bias in Browder’s analysis, that argument fails too. “Suspect-driven bias” is an “approach that uses the known reference profile(s) to drive the interpretation of evidentiary data.” And, reviewing Browder’s 2011 report during the state habeas proceedings, expert Bruce Budowle explained that Browder “provided three different CPI calculations,” which is an indicator of suspect-driven bias. By listening to Browder’s testimony and looking at the 2011 expert report, the defense knew or should have known “the essential facts permitting [Herod] to take advantage” of the suspect-driven bias to impeach Browder’s expert testimony. Rector v. Johnson, 120 F.3d 551, 560 (5th Cir. 1997) (internal quotation marks and citation omitted).

17

No. 25-40247

To the extent the information suppressed was unidentified DPS employees’ knowledge of the scientific debate—information that was (arguably) not publicly available—this is not material. The “criticisms of [the DNA protocol] were already present in the public domain at the time of [Herod’s] trial.” Higgs, 663 F.3d at 738. So, evidence about the state’s knowledge of those criticisms is unlikely to cast more doubt on the verdict than publicly available scientific criticisms would have. See id. at 737–38 (concluding undisclosed FBI study was not improperly suppressed because the information in the study was “already present in the public domain”).

* * *

Herod’s Brady claim fails because the state did not suppress any evidence. And even if the state suppressed DPS employees’ knowledge of scientific debate, that evidence alone is not material. These scientific disputes were better suited to cross-examination or rebuttal evidence, not collateral review.

B.

All that remains is Herod’s Napue claim, alleging that Browder falsely testified that Herod “could not be excluded,” there was “no difficulty with the statistics,” and that was the “strongest language” she could use. Napue violations have three elements: “1) the testimony was actually false, 2) the state knew [or should have known] it was false, and 3) the testimony was material.” 4 Canales v. Stephens, 765 F.3d 551, 573 (5th Cir. 2014) (citation modified); United States v. Agurs, 427 U.S. 97, 103 (1976) (describing

_____________________

4 The parties dispute whether a Napue claim lies where the prosecution “should have known” that the testimony was false. The foregoing analysis proceeds even assuming that “should have known” suffices, see Agurs, 427 U.S. at 103, although Herod identifies no Fifth Circuit cases granting Napue relief purely based on information the prosecution “should have known.”

18

No. 25-40247

knowing use of perjured testimony using “knew[] or should have known” standard); Chambers v. Johnson, 218 F.3d 360, 364 (5th Cir. 2000).

“[W]hen it comes to forensic science, the concept of ‘false,’ as required by Napue, is not so easy to pin down.” Andy Hyman, Note, The Problem with Johnson and Lemond: Irreconcilable Science and the Brady-Napue No-Man’s Land, 50 Sw. L. Rev. 129, 135 (2020). Generally, evidence is “false” if “it is specific misleading evidence important to the prosecution’s case in chief.” Nobles v. Johnson, 127 F.3d 409, 415 (5th Cir. 1997) (internal quotation marks and citation omitted). The Supreme Court recently clarified that whether the witness believes his statement to be false is “beside the point” for a Napue claim. Glossip v. Oklahoma, 604 U.S. 226, 252 (2025). “What matters is that [the witness’s] testimony was false and a prosecutor knowingly let it stand nonetheless.” Id.

Consider each alleged false statement, in turn.

1.

Herod first argues the prosecution should have known Browder falsely testified that Herod “could not be excluded” from DNA mixture on the white t-shirt. Some scholars suggest that a state “should have known” of the falsity only in situations “as in Giglio, where positive information already in the possession of the prosecution established that the testimony was false.” Wayne R. LaFave et al., The Duty to Correct False Evidence, 6 Crim. Proc. § 24.3(d) (5th ed. 2025). Here, Herod identifies no “positive information” from the time of trial establishing that Browder’s conclusion was false. While there is evidence that DPS should have known about problems with the CPI method generally and that Browder used suspect-driven bias (using the suspect’s sample to drive interpretation), there is no evidence that the state

19

No. 25-40247

should have known Browder’s results were inaccurate. 5 Browder’s technique was grist for the cross-examination mill. Too many inferences are required to reach the conclusion that the state “should have known” Herod was excluded at the time of trial simply because Browder used suspect-driven bias. 6 See Hafdahl v. Johnson, 251 F.3d 528, 533 (5th Cir. 2001) (“[T]he proper place to challenge [an expert’s] methods and the strength of [her] conclusions is cross-examination—not on collateral review.”); cf. also Glossip, 604 U.S. at 246–47 (concluding prosecution knowingly failed to correct false testimony that witness had never seen a psychiatrist because sheriff’s department records reflected that someone had diagnosed witness with a mental disorder and prescribed lithium).

2.

Herod next alleges that Browder falsely testified that there was “no difficulty with the statistics” she calculated from the DNA mixture on the

_____________________

5 We assume without deciding that DPS’s general knowledge of scientific debate and changing practices in DNA mixture interpretation is imputed to the prosecution team for Napue purposes. See Coulson v. Johnson, 273 F.3d 393, 2001 WL 1013186, at *7 (5th Cir. 2001) (per curiam) (recognizing “knowledge may be imputed from the police to the prosecution”); United States v. Antone, 603 F.2d 566, 569 (5th Cir. 1979) (recognizing in Napue analysis that we “decline[] to draw a distinction between different agencies under the same government, focusing instead upon the ‘prosecution team’ which includes both investigative and prosecutorial personnel”). The parties dispute how far this principle extends, and whether high-level DPS review of DNA interpretation practices constitutes prosecutorial knowledge in every case involving DPS.

6 The district court also concluded that “the experts also [now] agree that, even based on [Browder’s] biased assumption[s],” Herod “should have been excluded. Herod, 2025 WL 1001609, at *24. But there is no evidence that the state knew this at the time of trial. We hesitate to conclude the state should have known its forensic expert falsely testified based merely on two experts disagreeing with her methodology and conclusions after the fact. See Gimenez v. Ochoa, 821 F.3d 1136, 1142 (9th Cir. 2016) (“To the extent that this new testimony contradicts the prosecution’s expert testimony, it’s simply a difference in opinion—not false testimony.”).

20

No. 25-40247

white t-shirt. Browder repeatedly expressed confidence in her expert opinion. The district court concluded that DPS’s knowledge of “potential inaccuracies” in the CPI method means the state “knew or should have known at Herod’s trial that Browder’s expressed certainty in her statistics was false.” Herod, 2025 WL 1001609, at *26 (emphasis added).

That conclusion does not follow. At the time of trial, the CPI method was “not scientifically unacceptable.” 7 The method was flawed, at least in part, because of the risk of allele drop out and suspect-driven bias. But those flaws were not “appreciate[d]” by “[m]ost laboratories” using the CPI method. In fact, Browder’s flawed application of the CPI method was a “common historical misunderstanding” “not only in Texas but nationwide.” It stretches the bounds of Napue to suggest that the state should have known at the time of trial that there was “difficulty with the statistics” because of broad-based scientific debate about a not yet scientifically unacceptable interpretation method, and where Browder’s misapplication of that method was based on the exact same assumptions made by laboratories around the country.

3.

Herod lastly argues Browder testified falsely that “could not be excluded” is the “strongest language you could use in [evaluating] a

_____________________

7 The state habeas court found based on Budowle’s affidavit that the CPI method was used by “most laboratories in the United States.” It likewise found that the change in interpretation methods was based on “forensic scientists” becoming “more aware of the limitations” of the CPI method. Expert Robert Collins testified that Budowle was too “diplomatic” and “not harsh enough” about the continued use of CPI at the time of trial, and by 2013 Texas was one of a few labs resisting the changes. But Herod has not provided clear and convincing evidence to overcome the presumption of correctness we must apply to the state court’s finding that the CPI method was not “scientifically unacceptable” at the time of trial in 2012.

21

No. 25-40247

mixture.” 8 He suggests on appeal that that this statement “misled” the jury because “the strongest language she could have used would have been that Herod was excluded from all DNA samples recovered from the scene.” But this statement is also not false, as it reflected a situation matching her expert opinion that Herod could not be excluded. Just because Browder could have reached a different, defense-favorable conclusion does not mean that “is excluded” is stronger language than “cannot be excluded.” This “strongest language” statement is not “specific misleading evidence.” See Nobles, 127 F.3d at 415 (citation modified).

* * *

Herod’s efforts to turn broad-based debate about scientific processes and after-the-fact expert opinions into evidence that the state should have known Browder falsely testified fails. If we ruled in Herod’s favor on these facts, prosecutors could be on the hook for presenting false testimony in virtually every case relying on developing or changing forensic science. Like his Brady claim, Herod’s Napue claim should have been rejected by the district court. 9

_____________________

8 The district court does not appear to have ruled that this statement was false.

9 Herod suggests on appeal that Browder was testifying falsely when she stated, “The mixture is separated to determine possible allele combinations before it’s ever compared to a profile. The separation of the mixture is not based on a comparison to the known samples. It would be the same regardless.” He suggests this was false testimony that Browder was not using suspect-driven bias. The district court did not address this as a false statement in its Napue analysis. And Herod did not identify this in his petition when analyzing the Napue claim, although he did mention it as false testimony earlier in his petition. But suspect-driven bias is using a known profile to drive data interpretation and select allele loci for CPI analysis, which led to Browder’s assumption that there were four contributors, not three. Browder appeared to be testifying here about separating the mixture first to determine allele combinations, not to select loci to calculate the CPI statistics. It is in no way clear from the record that separating the mixture, as described by

22

No. 25-40247

V.

Although Herod maintains that his claims were fully developed factually in the state court record, he asks the court to remand for further factual development if it concludes he has failed to establish either a Brady or Napue violation. Assuming Herod is not limited by 28 U.S.C. § 2254(e)(2), he is still not entitled to an evidentiary hearing on these claims. “[A] federal court must consider whether [an evidentiary] hearing could enable an applicant to prove the petition’s factual allegations, which, if true, would entitle the applicant to federal habeas relief.” Schriro v. Landrigan, 550 U.S. 465, 474 (2007). “[I]f the record refutes the applicant’s factual allegations or otherwise precludes habeas relief, a district court is not required to hold an evidentiary hearing.” Id.

The analysis above rejects Herod’s theories of relief based on the way they were alleged in his petition. So, an evidentiary hearing to further develop the allegations and legal theories in his complaint related to his Brady and Napue claims would not “entitle the applicant to federal habeas relief.” Id. Herod does not identify any factual allegations in his federal habeas petition that were not borne out by the state record that he would seek to prove at an evidentiary hearing. As such, we need not remand with instructions for the district court to conduct one. Id. at 475. It remains for the district court whether Herod is entitled to an evidentiary hearing related to any of his remaining claims.

VII.

The district court correctly concluded Herod established prejudice sufficient to overcome the procedural default of his Brady and Napue claims

_____________________

Browder, conflicts with Browder’s other statements or falsely denies her suspect-driven bias.

23

No. 25-40247

in state court. But it erroneously granted habeas relief despite significant barriers under Teague’s non-retroactivity principle and the merits. Both parties now apparently agree that Herod was excluded as a contributor to the DNA mixture on the white t-shirt. But we cannot shoehorn this conclusion into constitutional error infecting his trial under Brady or Napue. We REVERSE the district court’s grant of 28 U.S.C. § 2254 habeas relief premised on Herod’s Brady and Napue claims and REMAND for proceedings consistent with this opinion.