Hattie Tanner v. Joan Yukins

RECOMMENDED FOR FULL-TEXT PUBLICATION Pursuant to Sixth Circuit I.O.P. 32.1(b) File Name: 17a0180p.06 UNITED STATES COURT OF APPEALS FOR THE SIXTH CIRCUIT HATTIE TANNER, ┐ Petitioner-Appellant, │ │ > No. 15-1691 v. │ │ │ JOAN YUKINS, │ Respondent-Appellee. │ │ ┘ Appeal from the United States District Court for the Eastern District of Michigan at Detroit. No. 2:04-cv-71155—Victoria A. Roberts, District Judge. Argued: May 4, 2017 Decided and Filed: August 15, 2017 Before: DAUGHTREY, MOORE, and KETHLEDGE, Circuit Judges. _________________ COUNSEL ARGUED: Matthew R. Cushing, JONES DAY, Columbus, Ohio, for Appellant. Raina I. Korbakis, OFFICE OF THE MICHIGAN ATTORNEY GENERAL, Lansing, Michigan, for Appellee. ON BRIEF: Matthew R. Cushing, Chad A. Readler, Danielle L. Scoliere, JONES DAY, Columbus, Ohio, for Appellant. Raina I. Korbakis, OFFICE OF THE MICHIGAN ATTORNEY GENERAL, Lansing, Michigan, for Appellee. _________________ OPINION _________________ KAREN NELSON MOORE, Circuit Judge. Hattie Mae Tanner, who was convicted of murder in 2000, argues that the district court erred by denying habeas relief on two grounds. No. 15-1691 Tanner v. Yukins Page 2 First, Tanner argues that the Michigan Supreme Court unreasonably applied Ake v. Oklahoma, 470 U.S. 68 (1985), when it held that the trial court properly denied Tanner’s trial counsel funding for a serology or DNA expert, and that the district court erred by upholding the Michigan Supreme Court’s application of Ake. Second, Tanner argues that the Michigan Supreme Court unreasonably applied Jackson v. Virginia, 443 U.S. 307 (1979), when it held that there was sufficient evidence to convict Tanner, and that the district court erred by upholding the Michigan Supreme Court’s application of Jackson. We agree with Tanner that she was convicted based on insufficient evidence and that the Michigan Supreme Court unreasonably applied Jackson. We REVERSE the district court’s judgment denying habeas relief on Tanner’s Jackson claim. Because we hold that Tanner is entitled to habeas relief on the ground that the Michigan Supreme Court unreasonably applied Jackson, we do not address whether the Michigan Supreme Court also unreasonably applied Ake. I. BACKGROUND A. The Crime Scene Sharon Watson, a bartender at Barney’s Bar and Grill, was stabbed to death in the basement of Barney’s sometime after 1:00 a.m. on March 22, 1995. People v. Tanner, 660 N.W.2d 746, 751 (Mich. Ct. App. 2003). Watson’s murder appears to have happened during the course of a robbery. Id. Watson’s boyfriend, Jerry Dockum, testified that at around 1:30 a.m. on March 22, Watson called him to tell him that she was closing the bar early. Id. at 752. When Watson was not home by 2:00 a.m., Dockum grew concerned and called the bar. Id. No one answered. Id. Dockum contacted Watson’s sister, Gloria Loring, who eventually went to Barney’s accompanied by Maria Coller, a former Barney’s employee who had keys, and Maria Coller’s husband, Ron Coller. Id. at 753. Loring and the Collers arrived at Barney’s around 5:30 a.m. Id. When they arrived, the lights were on, the television was blaring, the outside doors were locked, and Watson’s car was in the parking lot behind the bar. Id. A pack of Budweiser was on the floor near the side door with a napkin on top of it. Id. A note for a takeout order of beer was No. 15-1691 Tanner v. Yukins Page 3 on the cash register behind the bar. Watson’s coat was on the back of a chair, and Watson’s purse was on the back of the bar. Id. There was a knife behind the bar. Id. Shortly after arriving, the Collers called 911 and Barney’s owners. After Mr. Coller opened the door to the basement and observed loose cash at the bottom of the stairs, Mrs. Coller called 911 a second time. When they noticed that the door to the basement office was closed, Mrs. Coller called 911 a third time. Id. After one of Barney’s owners, Tom Bliler, arrived, they opened the door to the basement office. Id. They found Watson’s body in the basement. Id. Bliler estimated that $1,009 had been stolen from the safe, suggesting robbery. Id. Detective Michael VanStratton, who at the time was the crime lab supervisor of the Battle Creek Police Department, arrived at the scene. He testified that Watson had blood stains smeared across her body, an excessive amount of blood on her neck and chest, and stab wounds to her chest. Id. at 753–54. Because of the disarray in the office and the wounds to Watson’s arms, VanStratton concluded that there had been a struggle. Id. Crime-scene technicians found “diluted bloodstains on the stainless steel sink area directly behind the bar.” Id. In addition to the items the Collers had already noticed—namely, the six pack of beer with the napkin on it, the note about the take-out beer order, the knife, and Watson’s purse—technicians also found two drinking glasses and a cash register receipt. Id. In the basement, technicians found a bloodstain on the wall at the bottom of the stairs. Id. at 754. VanStratton had originally arrived at Barney’s around 7:00 a.m. without the necessary equipment to process the crime scene. Id. at 753. VanStratton testified that when he returned to Barney’s with the necessary equipment around 8:00 a.m., “‘some of the areas which [he] thought might be critical for investigation’”—including the area behind the bar—“‘had already been occupied by people that came in that morning.’” Id. There were about seven non-law- enforcement people in Barney’s, including Watson’s friends and Barney’s employees. Id. at 754. People were making coffee behind the bar and were in the area where the bloodstains, knife, and takeout beer were found. Id. “Detective VanStratton testified that because ‘there was some important evidence behind the bar,’ it was the first area that was isolated,” although people had gathered and made coffee in that area earlier in the morning, before law enforcement isolated it. Id. at 753–54. No. 15-1691 Tanner v. Yukins Page 4 B. Witness Accounts Detective David Walters of the Battle Creek Police Department focused the investigation on Tanner, Dion Paav, and Robert Cady. Id. at 751, 755. On May 24, 1995, Walters interrogated Tanner and made an audio recording of the interrogation. Id. at 755. According to Walters’s trial testimony, Walters showed Tanner a photograph of the knife recovered from the crime scene. Walters testified that Tanner said the knife was hers and that she recognized it by the alteration she had made to the blade for cleaning crack pipes. Id. On cross-examination, Walters acknowledged that the transcript of the audio recording shows Tanner saying that the knife was not hers. Id. at 757. Walters conceded that Tanner’s answers to many questions were inaudible in the recording. Id. Walters also testified that although the thirty-two page transcript included 261 instances where Tanner’s response is transcribed as “inaudible[]” Walters did not send the tape to the Michigan State Police Crime Laboratory to enhance the sound quality. Id. Walters testified that he interviewed Tanner about Watson’s murder again on June 7, 1995. Id. at 756. This interview took place in a police car, with Detective David Adams also present. Id. According to Walters, during this interview Tanner admitted to accompanying Cady to Barney’s around the time of Watson’s murder. Id. Walters testified that Tanner said she stayed in the car while Cady went inside, and that after she and Cady left Barney’s, they bought beer, cashed a check, and purchased crack. Id. Walters testified that he asked Tanner whether she was responsible for killing Watson, and she shook her head no. Id. Walters testified that he asked what circumstances might have led her to commit that sort of murder, Tanner responded that she might have done so “‘if that bitch had treated her bad.’” Id. There is no audio recording of this interview, and Detective Adams, who was also in the police car during the interview, did not testify. Tanner also testified at trial, and characterized her answers about the knife differently than Walters did. Id. at 760. Tanner testified that she told Walters that the knife in the photo looked like a knife she used to have but was not her knife. Id. She testified that she told Walters it could not be hers because it was a straight-bladed knife and her knife was a folding knife. Id. No. 15-1691 Tanner v. Yukins Page 5 Additionally, a friend of Watson, Catherine Huskins, testified that Watson had found a nonfolding knife before she was murdered. Id. Watson told Huskins’s husband that she was going to keep the nonfolding knife in her purse. Id. Huskins never saw the knife that Watson apparently carried in her purse; she only heard about it from Watson. Id. According to Cady’s trial testimony, on March 21, 1995 he got off work at approximately 10:55 p.m. Id. at 751. He had planned to meet Paav after work, but could not reach him. Id. After midnight on March 22, Cady called Tanner, drove to her house, and went with her to purchase crack cocaine. Cady and Tanner returned to Tanner’s house and smoked the crack. Id. Approximately a half hour later, Cady left, without Tanner, to cash a check at Barney’s. Id. He arrived at Barney’s around 1:00 a.m. Id. Barney’s appeared to be closed, but Cady entered through the open side door. Id. at 751–52. Watson was at the bar working, and there was a white male who Cady did not recognize in the bar. Id. Watson told Cady that she could not cash his check because she had already closed out her cash register. Id. at 752. Cady indicated that Watson would close out her cash register with customers in the bar only if they were trusted regular customers, and that Watson would close out her cash register with Cady in the bar. Id. at 752–53. Cady then went to Green’s Tavern, where he was able to cash the check. Id. at 752. He also had a beer while at Green’s Tavern. Id. At around 1:30 a.m., Cady called Tanner to tell her he was going to return to her house. Id. Cady went to buy more crack and then returned to Tanner’s house, arriving around 2:30 a.m. Id. He drove home around 2:45 a.m. Id. On his way home, he passed by Barney’s and noticed that the lights were on. Id. He found it unusual for Barney’s light to be on at that time, but he continued driving home without stopping. Id. Tanner’s trial testimony about the night of March 21 mirrored Cady’s. Id. at 760. She testified that Cady came to her house where they smoked crack together and then he left to go cash a check. Id. She testified that he returned to her house after cashing the check and then left again around 2:30 or 3:00 a.m. Id. She testified that she did not go anywhere with Cady except to purchase crack around 12:00 a.m. Id. She specifically said she did not go to Barney’s. Id. No. 15-1691 Tanner v. Yukins Page 6 Other witnesses also corroborated aspects of Cady’s and Tanner’s accounts of that night. Tanner’s mother testified that Cady was at her house when she woke up in the morning, and that neither Cady nor Tanner had any blood on them. Id. Todd Green testified that Cady did go to Green’s Tavern after midnight, where he cashed a check, drank a beer, and made a telephone call. Id. Two witnesses spotted a truck and unidentified individuals outside of Barney’s. Kevin Sage testified that he saw a light-colored truck with a wooden cap at Barney’s around 1:15 a.m. Sage said that the driver appeared to be a white man with a beard, and that there was a passenger who Sage did not get a good look at. Id. Nancy Chantrene testified that at 2:47 a.m. she passed Barney’s on her way to work and she saw a light-colored truck with a cap parked at the bar. Id. at 760–61. C. Charges The Battle Creek Police Department sought warrants for Tanner, Paav, and Cady. Id. at 751. In the fall of 1995, the prosecutor refused to issue the warrants because, in the prosecutor’s view, there was insufficient evidence. Id. at 751. A new prosecutor took office in 1997. Id. In May 2000, a warrant was issued for Tanner, but not for Paav or Cady. Id. The warrant charged Tanner with open murder, felony murder, and armed robbery. Id. As far as the record in this case reveals, neither Cady nor Paav was ever charged with any crime related to Watson’s murder. D. Forensic Experts Prior to trial, Tanner’s trial counsel requested funds for an expert to help him understand the prosecution’s DNA and blood evidence against Tanner. Specifically, the prosecution had evidence that the blood sample from the sink matched Tanner’s blood type. The Michigan Court of Appeals quoted the following colloquy that took place at the hearing on Tanner’s motion: The Court: Can’t you do that through your cross-examination of the DNA person that the People bring? Mr. Sullivan [defense counsel]: Imagine if the government had to—the only way they could prepare their case and be informed and develop their defense or their No. 15-1691 Tanner v. Yukins Page 7 government’s strategy is to wait for the defense witness to take the stand and cross them to learn. I’m not asking for a lot of money, Judge. I would think most of the DNA experts wouldn’t charge very much for an indigent defendant. The Court: Well, I beg to differ with you, Mr. Sullivan. That’s a fairly sophisticated area of expertise. Mr. Sullivan: I’m not asking that there be reanalysis, Judge. I’m just asking for consultation with them about what these results would mean to them. I’m not asking for them to analyze blood, any of that. I’m not—I’m not asking for money for lab work or anything like that. The Court: You only want to consult with someone and have them review what the People have? Mr. Sullivan: Yes. The Court: And be able to talk with them, is that it? Mr. Sullivan: Then—right, and then based upon the conversation that I would have, maybe call them as a witness for the defense. The Court: That opens a door to a lot more money, Mr. Sullivan. Mr. Kabot, what’s your position as it relates to this request? Mr. Kabot [assistant prosecutor]: My position is this, Judge—and Mr. Sullivan knows this full well—the blood stain that was on the sink upstairs in the bar, that was tested for serology. There is a breakdown of that—of that blood along with the blood samples that were submitted by a number of other people, because again they were tested serologically to see—they were tested for PGM enzymes and there is a breakdown. And as Mr. Sullivan knows[,] the report came back indicated that Ms. Hattie—Ms. Tanner’s blood was included within the specified group that could have been a donor of that blood. DNA-wise there really isn’t anything that links this defendant to anything that was found in that bar through DNA. The Court: Even in the basement? Mr. Kabot: Even in the basement and I think Mr. Sullivan understands that also. That what we’re talking about in the basement and the blood in the basement every report that we received back where DNA was performed excluded Hattie Tanner. The Court: So the point—I took it from Mr. Sullivan’s argument we were going in a different direction—at this point in time the only thing you have apparently that by inference would link Ms. Tanner to the crime scene is she has the same blood type as was found on the stain in the kitchen and there is nothing in the basement in terms of the DNA analysis that at all links Hattie Tanner to that scene, is that it, Mr. Kabot? Mr. Kabot: That’s my understanding, Judge. Now understand I just got this case the other day and I had it at home over the weekend and I was reading it. But No. 15-1691 Tanner v. Yukins Page 8 from each of the reports that we received back—forensic reports concerning DNA I saw on each of those that Ms. Tanner was excluded by way of DNA as being the donor for whatever blood sample they were testing. The Court: Would you even call this person as a prosecution witness then? Mr. Kabot: The only thing that I’m going to do, your Honor, as far as Megan Clement[] is concerned or anyone else as far as DNA is concerned is—there are some questions I have to ask them, but I mean as far as Mr. Sullivan’s cross- examination of them I’m not sure how much there’ll even be for the DNA people because again the bottom line is the results of the testing says right there. You know, Hattie Tanner’s excluded from being the donor of blood on the victim’s clothing and the blood stains that they recovered in the immediate area of the victim’s body which was in the basement. So how much cross-examination is there really going to be concerning that issue, only Mr. Sullivan knows. The Court: Mr. Sullivan. Mr. Sullivan: Can I offer a middle position here? The Court: Sure. Mr. Sullivan: How you [sic] about you approve enough money to me to consult the DNA expert, and then based on that consultation if I can persuade you that some money should be kicked in for him to testify then we can revisit that area, and that wouldn’t put off the trial. The Court: Well, Mr. Sullivan, the thing that occurs to me is that—and I’m not saying this would happen—but potentially you would confer with this expert and that expert would say I think the prosecution’s expert is all wrong, it really doesn’t link the defendant with the blood in the basement. Why do you need someone when you already have the prosecution witnesses saying it excludes your client? Id. at 765–66. Determining that “the prosecution’s expert would exonerate and exculpate the defendant from the evidence found at the scene,” the trial court denied defendant’s request for funds to hire a DNA or serology expert. Id. at 766. At trial, the prosecution presented two forensic experts. Nibedita Mahanti testified as an expert in DNA analysis. Id. at 759. She testified that the DNA on the knife and napkin (and a few other items) matched Watson. Id. She testified that the DNA in the blood on Watson’s shirt came from an unknown female. Id.; R. 16-3 (Trial Tr. at 230) (Page ID #427). The blood on Watson’s shirt did not match Watson, and it also did not match Tanner, Cady, or Paav. Tanner, 660 N.W.2d at 759. No. 15-1691 Tanner v. Yukins Page 9 Megan Clement testified as an expert in DNA and serology. Id. at 758. Like Mahanti, Clement testified that the blood stains on Watson’s shirt came from an unknown person. Id. Clement testified that the blood found on the sink behind the bar did not contain enough DNA to develop a profile. Id. Clement also testified generally about serological testing, although she did not testify about the results of the serological testing conducted on the evidence in this case. Clement explained that there are four common bloodtypes, A, B, AB, and O, and there are ten types of the enzyme phosphoglucomutase (PGM). Id. It is possible to analyze a sample of blood for ABO type and PGM subtype. Clement testified that “if you type for ABO and PGM, you look at the frequency of ABO times the frequency [of] the PGM to come up with what percentage of the population would have both of the characteristics in the sample detected.” Id.; R. 16-2 (Trial Tr. at 163) (Page ID #479). Regarding the frequency of blood types, Clement testified that, In the Caucasian population a type B blood is approximately ten percent and a PGM two plus one plus is approximately 20 percent, actually 21 percent. If you multiply the two for a person to have type B blood and be a two plus one plus it would be approximately two percent of the population of the Caucasian population. . . . In the African-American population a type B is much higher. It’s approximately 20 percent. And the PGM two plus one plus is about 19.8 percent, so approximately 20 percent as well. So a person who [has] blood type B and PGM two plus one plus in the African-American population would be approximately four percent so it would be twice as common in the African- American population as the Caucasian. Tanner, 660 N.W.2d at 758; R. 16-2 (Trial Tr. at 165) (Page ID #480). On cross-examination, Clement was asked how many people in the United States would have blood type B and PGM two plus one plus. In response, Clement testified that because African Americans comprise twenty-six percent of the United States population of 280 million, “26% of 280 million people times four percent of that” means that “possibly millions” of people were blood type B two plus one plus. Tanner, 660 N.W.2d at 759; R-16-2 (Trial Tr. at 174) (Page ID #482). The demographic information Clement provided is wrong. According to US Census data, in 1995, the year of Watson’s murder, the US population was approximately 263 million people No. 15-1691 Tanner v. Yukins Page 10 and African Americans comprised 12.6% of the total US population. See Resident Population Estimates of the United States by Sex, Race, and Hispanic Origin, CENSUS.GOV (Jan. 2, 2001), https://www.census.gov/population/estimates/nation/intfile3-1.txt. In 2000, the US population was approximately 276 million people and African Americans comprised 12.8% of the total US population. Id.; see also Tanner, 660 N.W.2d at 769 n.7. Additionally, Clement’s testimony implied that the blood must have come from an African-American person, but Clement had no way of knowing the contributor’s race. Clement was asked, “Just based upon your projected statistical information how many people have type B two plus one plus?” She responded by saying “I think it’s about 26 percent of the overall U.S[.] population is African-American so if you -- 26 percent of 280 million people times four percent of that.” After this response, defense counsel added “And same with Caucasian?” to which Clement responded, “That’s correct,” and “And then you throw some other millions in there for some of these mixed races all the other categories that we haven’t talked about here today?” to which Clement responded “Yes.” R. 16-2 (Trial Tr. at 174–75) (Page ID #482). As a result of these follow up questions, Clement’s testimony was not outright false. But Clement’s response was potentially misleading because it suggested that Clement had reason to believe that the blood came from an African-American person, and she had no basis to believe that. Similarly, Clement noted that type B two plus one plus blood is twice as common in the African-American population as the Caucasian population, but she failed to note that because the United States’ Caucasian population is more than twice the size of the United States’ African- American population, there are more Caucasian people than African-American people who are type B two plus one plus. In other words, Clement failed to note that there were more than twice as many Caucasian people as African-American people who could have left blood near the sink at Barney’s. In sum, there were millions of people who could have contributed the blood found on the sink and most are not African American. Clement’s possible suggestion that the contributor was more likely to be African American is relevant because Tanner is African American. Tanner, 660 N.W.2d at 759. Marie Bard-Curtis testified as a serology expert. Id. Other than the blood on the sink, the samples she tested were insufficient to produce results. Id. The blood on the sink was sufficient No. 15-1691 Tanner v. Yukins Page 11 to produce results; it was ABO type B and PMG subtype two plus one plus. Id. According to Bard-Curtis, “The ABO type determined on the blood on the bar sink was type B. Hattie Mae Tanner was blood type B. The PGM subtyping detected on the sample from the bar sink was two plus one plus and Hattie Mae Tanner was also two plus one plus.” Id. at 759; R. 16-2 (Trial Tr. at 240–41) (Page ID #429–30). Bard-Curtis was not asked about the frequency of various blood types in the population, and she did not mention that millions of people would have the same ABO type and PMG subtype as the blood on the sink. Throughout his cross-examination of both Bard-Curtis and Clement, Tanner’s counsel indicated that he was confused about the subject matter. At one point he asked Bard-Curtis whether “there was a homogenous sample” and then, when Bard-Curtis responded that she did not know, he says “I apologize, Judge. I’m not positive I know what that word means.” R. 16-3 (Trial Tr. at 244–45) (Page ID #430–31). Before starting re-cross, in which Tanner’s counsel tried to ask the expert about whether the blood on the sink could have come from more than one person, counsel noted, “This might be beyond my expertise.” Id. at 250 (Page ID #432). Similarly, during his cross-examination of Clement, Tanner’s counsel said, “I want to go back to serology for a minute. When blood is tested for serology for the—I’m just talking about something I know nothing about” before proceeding to ask her general questions about how labs conduct serological testing. Id. at 172–73 (Page ID #481–82). Tanner’s trial counsel brought up the blood found on Watson’s shirt in his cross- examination of Dr. Mahanti. He asked, “With regard to shirt item No. H . . . excuse me . . . [y]our number 140.98G, you said that came from an unknown contributor, correct? . . . Specifically your conclusion is that the donor of that blood on that submitted piece of evidence came from a female contributor, correct? . . . But neither Ms. Watson nor Ms. Tanner?” R. 16-3 (Trial Tr. at 230) (Page ID #427). Dr. Mahanti responded to these questions in the affirmative. Id. In his closing argument, trial counsel mentioned in passing that the blood on Watson’s shirt belonged to an unidentified woman. He said, The results as to the shirt listed on this evidence is that there was blood on that shirt that was not the victim’s. It was not Dion Paav’s, Rob Cady’s, or Hattie Tanner’s, but it was blood from somebody else. A female contributor was the testimony of the doctor on the shirt. How does that fit into their theory? And is No. 15-1691 Tanner v. Yukins Page 12 that blood? Do we know it -- did they check? Is B, two plus one plus, all that kind of stuff? Interesting that shirt though. R. 18-2 (Trial Tr. at 109–10) (Page ID #540). Trial counsel did not emphasize, either in his cross-examination of Dr. Mahanti or in his closing argument, that the blood of an unknown person was on the shirt the victim was wearing while she was stabbed multiple times during a struggle. That is, he does not emphasize for the jury the likelihood that the blood came from Watson’s murderer. Trial counsel’s closing argument mentioned the truck spotted outside of Barney’s— “What about pickup trucks? How’s that fit into what the government’s told you?”—but he failed to connect the fact that there were unidentified people spotted in a truck outside of Barney’s and the blood of an unidentified person found on the victim’s shirt. Id. at 110 (Page ID #540). Instead, he said simply “whether you can ever find those trucks or not, that’s not my client’s issue.” Id. E. Verdict After both sides presented their evidence, the trial judge denied Tanner’s motion for a directed verdict, relying in part on the serology evidence. The trial judge said, “if the jury believes Mr. Walters . . . a rational trier of fact could find that based on [Tanner’s] statement as read by Mr. Walters; to wit, she was there, that it was her knife, and you couple that with the fact that her blood type was found in a stain at the bar that would form the basis where a rational trier of fact could conclude that it was, in fact, Hattie Tanner who committed the crime.” Tanner, 660 N.W.2d at 772. The jury convicted Tanner of first-degree felony murder, second-degree murder, and armed robbery. Tanner, 660 N.W.2d at 751. She was sentenced to life without parole for the first-degree felony murder conviction. She was initially sentenced to concurrent terms of forty to sixty years and twenty-five to fifty years for the second-degree murder and armed robbery convictions, but the trial judge sua sponte vacated those sentences. Id. No. 15-1691 Tanner v. Yukins Page 13 F. Procedural History After being found guilty by a jury in the Circuit Court for Calhoun County, Michigan, Tanner appealed to the Michigan Court of Appeals. The Michigan Court of Appeals reversed her convictions and her sentence for the first-degree felony murder conviction and remanded for a new trial. Tanner, 660 N.W.2d at 751. It ruled that the trial court erred by denying Tanner’s counsel’s motion for expert funds. Id. at 769–70. It also ruled that the trial court erred by denying Tanner’s motion for a directed verdict on the charge of first-degree felony murder on the theory that Tanner aided and abetted Cady. Id. at 774. The Michigan Court of Appeals determined that the prosecution did not introduce sufficient evidence to prove that Cady murdered Watson, meaning that it did not produce sufficient evidence to prove that Tanner aided and abetted Cady in murdering Watson. It also determined that the prosecution introduced sufficient evidence to prove that Tanner committed felony murder as the principal, as opposed to by aiding and abetting Cady, but noted that the members of the jury were not asked to specify whether they found Tanner guilty of felony murder as the principal or as an aider and abettor. Id. at 775 & n.9. Judge Danhof dissented in part, arguing that the trial court was correct to deny Tanner’s counsel request for funds to pay for a serologist. Id. at 775 (Danhof, J., dissenting). In a short per curiam opinion, the Michigan Supreme Court reversed the judgment of the Court of Appeals and reinstated Tanner’s felony-murder conviction and sentence. People v. Tanner, 671 N.W.2d 728, 731 (Mich. 2003). The Michigan Supreme Court addressed the sufficiency of the evidence supporting Tanner’s murder conviction only in a short footnote to the opinion. Justice Kelly dissented from the Michigan Supreme Court’s per curiam opinion. Id. at 731 (Kelly, J., dissenting). After exhausting her state remedies, Tanner filed a pro se federal habeas petition, which the district court denied in 2005. See Tanner v. Yukins, 776 F.3d 434, 435–36 (6th Cir. 2015). Tanner was unable to file a timely notice of appeal because prison guards refused to let her access the law library to pick up legal papers that she needed to send to the court before the deadline. Id. at 436. The district court granted Tanner’s certificate of appealability, apparently not realizing that the notice of appeal was filed one day past the deadline. Id. at 436–37. The district court permitted Tanner to appeal on two grounds: “that the state trial court violated her No. 15-1691 Tanner v. Yukins Page 14 right to due process when it denied her request for DNA and serological experts, and that there was insufficient evidence to convict her of felony-murder.” Id. at 437. By the time the appeal was docketed with this court, it was too late for Tanner to request an extension of time to file the notice of appeal. Id. This court’s show-cause order was the first notice to Tanner that her notice of appeal was untimely. Id. Tanner explained the situation in an affidavit filed in response to the show-cause order, but this court dismissed Tanner’s appeal for lack of jurisdiction. Id. Tanner then filed a § 1983 case against the guards who prevented her from timely filing her notice of appeal. Id. A jury found in her favor and awarded compensatory and punitive damages. Id. After the verdict finding that prison guards prevented Tanner from filing a timely notice of appeal, Tanner filed in the district court a motion for relief from judgment under Federal Rule of Civil Procedure 60(b)(6), seeking a procedural avenue to appeal the denial of her habeas petition. Id. The district court denied the motion for lack of jurisdiction. Id. This court granted a certificate of appealability on the issue of whether the district court erred by denying the 60(b)(6) motion. Id. This court then reversed the judgment of the district court and directed the district court to “vacate the judgment, entered on November 8, 2005, that denied Tanner’s petition for a writ of habeas corpus and to re-enter the judgment, thereby starting anew the 30–day period under Rule 4(a)(1)(A) in which to file a notice of appeal.” Id. at 444. The district court reinstated its order on May 6, 2015. R. 54 (05/06/2015 Order) (Page ID #1001). The second time around, Tanner timely filed a notice of appeal. R. 59 (06/03/2015 Notice of Appeal) (Page ID #1007– 08). This case is now before us to address the same issues on which the district court originally granted a certificate of appealability in 2005: first, that the Michigan Supreme Court unreasonably applied Ake when it held that the trial court properly denied Tanner’s trial counsel funding for a serology or DNA expert, and, second, that the Michigan Supreme Court unreasonably applied Jackson v. Virginia when it held that there was sufficient evidence to convict Tanner. No. 15-1691 Tanner v. Yukins Page 15 II. DISCUSSION A. Legal Standards “In reviewing a district court order granting an application for a writ of habeas corpus relief, we review legal conclusions de novo.” Newman v. Metrish, 543 F.3d 793, 795 (6th Cir. 2008), cert. denied, 558 U.S. 1158 (2010). “Although we generally review the district court’s findings of fact for clear error, we review de novo when the district court’s decision in a habeas case is based on a transcript from the petitioner’s state court trial, and the district court thus makes no credibility determination or other apparent finding of fact.” Id. at 795–96 (internal quotation marks omitted); see also Taylor v. Withrow, 288 F.3d 846, 850 (6th Cir.), cert. denied, 537 U.S. 1007 (2002). The de novo standard of review does not constrain our review of the district court’s judgment, but the applicable law tightly constrains our review of the state appellate court and the jury. On a sufficiency-of-the-evidence claim, “the relevant question is whether, after viewing the evidence in the light most favorable to the prosecution, any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt.” Jackson, 443 U.S. at 319. The Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA) allows a petitioner in state custody to present a federal court with a claim if the claim was adjudicated on the merits in state court, but the federal court may grant a writ of habeas only if the state court’s decision was either “contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court,” 28 U.S.C. § 2254(d)(1), or “was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding,” 28 U.S.C. § 2254(d)(2); see also Taylor v. Withrow, 288 F.3d at 851 (quoting Williams v. Taylor, 529 U.S. 362, 405 (2000)). Thus, two layers of deference apply, one to the jury verdict, and one to the state appellate court. See Brown v. Konteh, 567 F.3d 191, 204–05 (6th Cir. 2009), cert. denied, 558 U.S. 1114 (2010) (“In an appeal from a denial of habeas relief, in which a petitioner challenges the constitutional sufficiency of the evidence used to convict him, we are thus bound by two layers of deference to groups who might view facts differently than we would. First, as in all sufficiency-of-the-evidence challenges, we must determine whether, viewing the trial testimony and exhibits in the light most favorable to the prosecution, any rational trier of fact No. 15-1691 Tanner v. Yukins Page 16 could have found the essential elements of the crime beyond a reasonable doubt. In doing so, we do not reweigh the evidence, re-evaluate the credibility of witnesses, or substitute our judgment for that of the jury. Thus, even though we might have not voted to convict a defendant had we participated in jury deliberations, we must uphold the jury verdict if any rational trier of fact could have found the defendant guilty after resolving all disputes in favor of the prosecution. Second, even were we to conclude that a rational trier of fact could not have found a petitioner guilty beyond a reasonable doubt, on habeas review, we must still defer to the state appellate court’s sufficiency determination as long as it is not unreasonable.” (internal citations omitted)). B. Sufficiency of the Evidence Acknowledging the double deference standard that applies in this case, the evidence still leaves us no choice but to conclude that the Michigan Supreme Court unreasonably applied Jackson, in violation of 28 U.S.C. § 2254(d)(1). See Newman, 543 F.3d at 797; see also Brown v. Palmer, 441 F.3d 347, 352 (6th Cir. 2006). The inculpatory evidence in this case establishes, at best, “reasonable speculation” that Tanner was in the Barney’s parking lot around the time of the murder and that she was last in possession of the murder weapon approximately a month before the murder. Newman, 543 F.3d at 797. There are three pieces of inculpatory evidence. First, Walters testified that Tanner told him that she was in the Barney’s parking lot around the time of the murder. Second, Walters testified that, during a separate interview, he showed Tanner a photograph of the knife recovered from the crime scene, and Tanner said the knife was hers. Third, there was blood near the upstairs sink that matched Tanner’s blood type and PGM subtype. There are several gaps in this evidence. First, Tanner did not tell Walters that she went into Barney’s on the night of the murder. Walters testified that Tanner told him that she was in the parking lot outside Barney’s on her way to buy crack with Cady, and that she waited in the car while Cady went into Barney’s to cash a check. Tanner, 660 N.W.2d at 756. Tanner did not say that she entered Barney’s. Moreover, even if the evidence placed Tanner in Barney’s— which it does not—there is no evidence showing that Tanner murdered Watson while in Barney’s. And even if the prosecution had proved that Cady murdered Watson—although the No. 15-1691 Tanner v. Yukins Page 17 prosecution did not indict, let alone convict, Cady—there is no evidence showing that Tanner helped him, either from within Barney’s or from the parking lot. Walters even testified that when he asked Tanner whether she was responsible for Watson’s murder, she answered in the negative. Id. Second, even accepting that Tanner told Walters that the murder weapon was her knife does not advance the prosecution’s case very far. We note that there are several problems with Walters’s testimony about the knife. One problem is that the transcript of Walters’s conversation with Tanner flatly contradicts Walters’s testimony. In the transcript, Tanner’s answer to the question whether the knife was hers was “no.” Id. at 757. A related problem is that, like the transcript, Tanner’s trial testimony also contradicts Walters’s. Tanner testified that she actually told Walters that the pictured knife looked like a knife she used to own, but was not hers; her knife was a folding knife, not a straight-bladed knife like the one pictured. Id. at 760. Yet another problem with Walters’s testimony is that Catherine Huskins’s testimony casts doubt on it. Huskins testified that Watson found a nonfolding knife and said she was going to keep it in her purse. Id. at 760. Huskins’s testimony suggests that there is at least a possibility that Watson’s own knife was the murder weapon. Despite these problems, a jury was entitled to credit Walters’s testimony, and this court is not entitled to question the jury’s credibility finding. See Brown v. Konteh 567 F.3d at 205. As a result, we must accept that Tanner told Walters that the photo of the murder weapon was a photo of her knife. But accepting that Tanner told Walters that the murder weapon was her knife still does not advance the prosecution’s case very far. According to Walters’s testimony, Tanner told him that she had last handled the knife three or four weeks before Watson’s murder. Tanner, 660 N.W.2d at 756. Tanner also made clear that other people, including Paav, had access to the knife. Id. Even if the knife used to murder Watson was in Tanner’s possession at one point, there is no indication that it was in her possession close to the time of the murder. That leaves the third piece of evidence, the blood. The blood found near Barney’s sink matched Tanner’s blood type and PMG subtype. Millions of people share Tanner’s blood type and PGM subtype. Any one of those people could have contributed the blood. The experts who testified at trial about serology did not provide precise (or even accurate) information about the No. 15-1691 Tanner v. Yukins Page 18 frequency of type B two plus one plus blood in the population, but they did at least acknowledge that millions of people have type B two plus one plus blood. Tanner, 660 N.W.2d at 758; R. 16- 2 (Trial Tr. at 165, 174) (Page ID #480, 482). Further undermining the importance of the blood near the sink, there is no way of knowing whether the blood belonged to the perpetrator or to one of the people who gathered at Barney’s the morning after Watson’s murder. When VanStratton arrived at Barney’s with the equipment necessary to process the crime scene, Barney’s employees and friends of Watson were already gathered in the area behind the bar where officers eventually found the blood. Tanner, 660 N.W.2d at 758. One of them could have bled near the sink that morning. Not only are there gaps in the prosecution’s inculpatory evidence, there is also exculpatory evidence. Most crucially, an unidentified woman’s blood was on the victim’s shirt. The blood did not come from Tanner or from any of her hypothesized accomplices (Cady or Paav). The blood on Watson’s shirt is particularly relevant because VanStratton concluded that Watson was killed during a struggle. Id. at 754. Someone who may have been the contributor of this blood was outside of Barney’s around the time of the murder. Two different witnesses, Sage and Chantrene, spotted a truck outside Barney’s early in the morning of March 22. Id. at 760–61. Sage saw two people in the truck. According to Sage, the driver was a man, but the passenger could have been a woman. Id. at 760. The State of Michigan has repeatedly responded to the gaps in the prosecution’s case and the existence of potentially exculpatory evidence by pointing out that a jury convicted Tanner. We are mindful that, as a federal habeas court, we are not to substitute our judgment for the jury’s judgment. See Brown v. Konteh, 567 F.3d at 205. We are also mindful that “a properly instructed jury may occasionally convict even when it can be said that no rational trier of fact could find guilt beyond a reasonable doubt . . . .” Jackson, 443 U.S. at 317. “[W]hen such a conviction occurs . . . it cannot constitutionally stand.” Id. at 318. In this case, the blood on Watson’s shirt would have created reasonable doubt for a rational jury. None of the evidence that implicates Tanner is sufficient to overcome the No. 15-1691 Tanner v. Yukins Page 19 reasonable doubt created by the presence of a unknown woman’s blood on the victim’s shirt. This is particularly true given that witnesses observed unknown individuals outside of Barney’s around the time of the murder. It would be difficult to overestimate the importance of blood on the shirt of a victim who was stabbed in the chest during a struggle. It is hard to imagine any scenario in which the contributor of this blood would not be a key suspect. And it is impossible to see how a rational jury could have found the defendant guilty beyond a reasonable doubt without an explanation for the unknown person’s blood on the victim’s shirt. Therefore, based on the strength of the potentially exculpatory evidence and the comparative weakness of the incriminating evidence, there is no way to read the record here to support the Michigan Supreme Court’s conclusion that a rational trier of fact could have found Tanner guilty of Watson’s murder beyond a reasonable doubt. In finding that there was sufficient evidence to convict Tanner, the Michigan Supreme Court unreasonably applied Jackson. Because we determine that the Michigan Supreme Court unreasonably applied Jackson, we do not consider Tanner’s claim that the Michigan Supreme Court unreasonably applied Ake. III. CONCLUSION For the foregoing reasons, we REVERSE the judgment of the district court and remand the case to the district court with directions to grant the writ, setting aside Tanner’s conviction and freeing her unconditionally from state supervision.