How Forensic Genealogy Led to an Arrest in the Phoenix ‘Canal Killer’ Case
The man accused in Phoenix’s “Canal Killer” case may have his ancestors to blame for his 2015 arrest.
Records show forensic genealogy was key in leading police to Bryan Patrick Miller, a man now facing a death-penalty trial in the early 1990s slayings of Angela Brosso and Melanie Bernas.
Using a method with little precedent in the world of criminal justice, a California genealogist named Colleen Fitzpatrick handed police what would amount to a case-busting lead: the suspect’s last name.
“The name Miller came up in my analysis,” Fitzpatrick wrote in an email that she later forwarded to investigators.
Miller, a now-44-year-old father, was arrested within weeks of Fitzpatrick’s emails with police. His trial is scheduled to begin April 28.
Naming a suspect through genealogy
Police had long held samples of crime-scene DNA they believed belonged to the killer, and Miller had remained in Phoenix for much of his adult life. But as investigators often bemoan, a DNA sample is useful only if you have a suspect’s sample against which to compare it.
The first crime occurred in November 1992, when 22-year-old Angela Brosso failed to return home from a bike ride. Police soon would find her headless body near 25th Avenue and Cactus Road, and her head in the Arizona Canal several days later.
Ten months later, 17-year-old Melanie Bernas would face a chillingly similar fate. In September 1993, the high school student also had been out on a bike ride before her apparent abduction. Her body, which was discovered the next day, also was found in the canal.
After months of speculation, police confirmed what the community had feared. The two murders were connected by a single suspect’s DNA.
Over the years, it is likely officials had run the suspect’s DNA through databases such as CODIS, a national repository for the profiles of known criminal offenders. But because Miller had never been to prison, his DNA profile would not have been included in this database.
After Miller’s January 2015 arrest, police confirmed the link was made through surreptitiously collecting Miller’s DNA and matching it with that of the unknown person’s. (Collection methods can include picking up a suspect’s used cigarette or water bottle, but police have not released details on this specific operation.)
What remained unclear was how an investigation that had fielded thousands of tips, dozens of leads and crossed state lines had winnowed to a pool of suspects small enough for detectives to track down one man’s DNA.
Police emails provided to The Arizona Republic through a public-records request provide the missing link in the investigation’s timeline.
That link began in October 2014, when genealogist Fitzpatrick attended a human-identification conference in Phoenix.
Investigators at the conference also met Fitzpatrick, who has carved a niche in identifying people, including helping to name an unknown child on the Titanic and the remains of a serviceman who died in a plane crash in 1948, according to her website.
By late December 2014, Fitzpatrick was providing email updates to Sgt. Troy Hillman of the homicide department’s cold-case unit.
Phoenix police and Fitzpatrick declined to comment for this story, with Phoenix police citing the upcoming criminal trial. Fitzpatrick said she would defer to Phoenix police on when she could comment. The Republic consulted with genealogy experts to help decipher some of the more technical portions of the emails.
It appears Phoenix police began by providing Fitzpatrick with a profile of the suspected killer’s DNA.
While much of the documents released to The Arizona Republic were redacted, Fitzpatrick indicates in them that she tapped at least one publicly available DNA database.
In a Dec. 27, 2014 email, Fitzpatrick mentions the private commercial companies Family Tree DNA and Ancestry.com — which both have their own database of DNA profiles — and a commercial kit called Y-filer. Fitzpatrick notes that only Family Tree DNA has a certain marker test from a certain area in the profile.
Fitzpatrick already has flagged the surname by this email.
“I checked Ancestry.com briefly to see if there were any Millers living in the apartment complex near the Canal,” she wrote. “Of course there were quite a few. I haven’t had a chance to look through them to see if there is a benefit for making the map — it may be too crowded.”
In an email from Fitzpatrick to a conference employee written shortly after Miller’s arrest in January 2015, Fitzpatrick said Miller had been on Phoenix police’s “extensive list of candidates.”
The fact that the name “Miller” emerged in her analysis, she said, “caus(ed) the Phoenix PD to take a closer look at him.”
It is unclear from the emails exactly how Fitzpatrick was directed to the surname.
Bennett Greenspan, the founder and CEO of commercial genetic testing company Family Tree DNA, was briefed on this story and posed a possible scenario.
Greenspan said there are several companies, including Family Tree DNA, Ancestry.com and 23andMe, that will compare a DNA signature to those already in their database. The service can be used to locate distant relatives or to help adopted individuals locate their biological families.
The databases for each company are populated by others who have submitted their DNA profile and opted in to allow others to find them, Greenspan said.
He said Family Tree DNA has the ability to test for Y-STR, which comes from the Y, or paternal chromosome. Records indicate Fitzpatrick analyzed the Y-STR for the Canal Killer case.
Y-STR stands for short tandem repeats, which Greenspan described as a genetic “stutter.” The stutters, he said, are passed on “from a father to his son, literally unaltered for generations.”
Greenspan said it’s likely that one or more of Miller’s relatives — even distant — had opted to submit their DNA to one of these databases. The database would then generate the names of those with the same Y-STR. And because male lineage is more likely to carry a surname through generations, it was likely that the database generated a relative or relatives with the last name of “Miller.”
Greenspan said he knew of Fitzpatrick, but didn’t know if it was his database that she used for the case. He said it was certainly possible that she submitted the profile anonymously.
“It’s a community-built database that allows people to compare, match and share data, share family stories — that’s what it’s designed for,” he said. “It may be used for forensic purposes, but they’re doing it around the system.”
The tests are typically used by private customers, he said. It could be two men with the last name “O’Brien” want to see if they’re related. Or someone who is adopted could use the test to find relatives and a likely biological family name.
“We sell 50,000 of these things per year. It is a repository — a treasure trove of potential information,” he said. “The truth is, if you don’t have anything to hide, it doesn’t matter.”
Familial testing and privacy concerns
The surname search used to point police to Miller is similar to that of familial searches that have gained notoriety in recent years.
With familial searches, police are able to test an unknown DNA profile with those of known offenders in criminal databases. If there are genetic similarities with a known criminal, police can use the information to build a family tree; eliminating or highlighting certain relatives based on their age or residence at the time of the crime.
One of the most celebrated examples of this method came in 2010, with an arrest in Los Angeles’ notorious “Grim Sleeper” case.
According to the Los Angeles Times, a familial search of prisons pointed to a convict whose DNA was similar to that of a serial killer. After police created and analyzed the family tree, the prisoner’s father, Lonnie David Franklin Jr., emerged as a suspect.
Earlier this year, Franklin was convicted of killing nine women and one teenage girl. In August, he was sentenced to death.
Familial testing can provide police a vital lead when other searches have come up empty. But critics have argued that it can place innocent people under police scrutiny solely based on their relation to a suspect.
Take the case of Michael Usry, a New Orleans filmmaker who recently became ensnared in a 1996 murder investigation of Idaho Falls teenager Angie Dodge.
According to the Associated Press, Usry’s father had submitted his DNA to a non-profit organization years earlier, and that data had later been acquired by Ancestry.com. In 2014, Idaho Falls police submitted the unknown person’s DNA to Ancestry.com, which produced a close match.
This signaled that the killer may have been a close relative to the donor, and police acquired a search warrant for Ancestry.com to turn over the name of the donor.
Usry was the right age to be the killer, and in turn was interrogated for hours. AP reports that Usry at last submitted a blood DNA sample, and remained under suspicion until he was ruled out.
Critics of the process say familial testing could have sweeping implications for privacy rights.
“When law enforcement is searching private databases of DNA and there’s no regulation and no judicial supervision, that does raise profound concerns about what is permissible activity by law enforcement,” said Steve Mercer, chief attorney for the Forensics Division of the Maryland Office of the Public Defender and a vocal opponent of familial testing.
Mercer noted that there already are several forms of law-enforcement activity that could be useful but are not tolerated by society. Police aren’t allowed to enter homes without a warrant, take someone into custody without probable cause or practice “stop and frisk” tactics.
“There’s so much that we take for granted that the police aren’t allowed to do in our everyday lives,” Mercer said. “But of course, if efficiency was the only limiting factor, then the police could do all of those things, because they’re going to catch some bad people doing that.”
David Kaye, an associate dean for research at Penn State University who studies genetics and its application in law, said privacy concerns will be a “significant question” if this field expands.
He said it would be important to see what types of disclosures the commercial companies make to their users.
“If you sign up, do they warn you that by signing up, you’re making information about your ancestry available to anyone? That it can be used by … authorities, including those investigating crimes?” he asked.
Generally speaking though, Kaye said, it could be argued that the commercial DNA databases are public domain, and therefore fair game for police.
“The question is, is this different than looking at someone walking down a public street?” he said. “Traditionally there’s been no expectation of privacy where the general public can find information.”
Kaye said potential harm depends on what police do with the surname once they have it. Though a separate DNA test is taken to show a direct match, there is a possibility that information could be leaked and publicly implicate an innocent person.
Still, Kaye argued, the same types of gaffes can come as a result of other investigative leads, like faulty eyewitnesses.
“There are issues present,” he said. “But I guess my own view would be that if police are discreet about this, it’s simply another clue.”
Could the search lead to more arrests?
Kaye said the prospect a surname search had been tossed around in literature about a decade ago, but he had never seen it come to fruition.
Records indicate that the method comes with limitations.
After the success of Miller’s arrest, Phoenix police apparently sent at least six other DNA profiles to Fitzpatrick for analysis. Records released to The Republic indicate their initial luck hasn’t yet repeated itself.
In various emails, Fitzpatrick explains having trouble with partial DNA profiles, and notes that the databases are “skewed in favor of Caucasian Europeans.”
“I have not checked for a name on that last case because I saw it was Native American,” Fitzpatrick wrote in an early 2016 email to Hillman.
Hillman and Fitzpatrick then discuss checking a DNA profile’s ethnicity before moving on to the more expensive surname search.
“This is pioneering work, so that it pays to be flexible as we find out more about how to use of the databases,” Fitzpatrick said during this conversation. “There are certainly other cases that can be solved. It’s a question of stacking the deck in our favor.”