Photo from USDA Agricultural Research Service
They have also done DNA analysis of the millions or billions of bacteria found in soil samples taken from woodlands, fields, back yards, and dirt roads. Some students even collected slaughterhouse blood from MSU’s Department of Animal Science to study the bacterial changes caused when blood mixes with soil.
The goal of this research, supported by the National Institute of Justice, is to develop a “soil individualization technique” that would allow a forensic expert to produce objective, statistical data to conclusively show, for example, that the soil on a shovel possessed by a suspect matches the soil at the burial site where a victim was found. Using next-generation sequencing (NGS) of bacterial DNA, Foran has utilized supervised classification techniques to classify the bacterial makeup of a soil sample and give values that are, he said, “completely objective.”
The resulting statistical data from bacterial analysis meet the Daubert standard on the admissibility of scientific testimony and can thus be presented in court, he noted. The techniques also respond to the 2009 National Academy of Sciences report on forensic science that called for “the development and establishment of quantifiable measures of the reliability and accuracy of forensic analyses.”
“Soil comparisons have historically been based on class characteristics,” Foran said. The factors being compared — grain size, pH, color, and other physical traits of soil — can have varying degrees of objectivity; and, more importantly, an expert taking the stand in a criminal case who testifies that two soil samples match or have the same origin is primarily using subjective judgment.
“Our goal is to identify a soil sample’s origin and say it came from here, to the exclusion of any other place,” he said. “We’re trying to individualize it and identify its origin, in a completely objective manner.”
The bacterial populations in soil provide an immense supply of data that can be used for its “individualization,” which allows a forensic investigator to determine that a specific sample came from a specific place. “Soil is an extremely rich source of bacteria,” Foran said. “Soil contains all of the fertilizer that plants, for instance, use, and bacteria thrive on it. So, in a gram of soil, just a teaspoon of soil, literally millions or billions of bacteria exist, each one being an individual life form.”
Foran and his graduate students have been working on DNA profiling of soil bacteria since one student, who majored in microbiology as an undergraduate, came up with the idea several years ago. The work progressed incrementally, but the recent emergence of next-generation sequencing of DNA is making forensic analysis of soil more practical and effective. In a 2016 paper in the Journal of Forensic Sciences, Foran and co-authors noted that, “next-generation sequencing of [a bacterial marker] shows tremendous potential for forensic soil analysis.” They concluded that the DNA-based identification strategies they have developed “demonstrate the utility of next-generation sequencing in producing soil bacterial profiles, helping to link a suspect, victim, or evidentiary item to a crime scene.”
“The forensic applications of soil bacteria analyses come from agricultural uses developed by the wider microbiology community,” Foran said. “At Michigan State University and several other research institutions, there are a number of people who utilize bacterial profiling of soil — although they wouldn’t term it profiling — because they are interested, for instance, in farming techniques. When do you add fertilizer? Do you till the land, or do you not till the land? Do you plow crops under in the fall or do you leave them standing? They look at the soil and ask what is healthy and what is not. They’ve been doing it for years, so it’s a great foundation for us because we’re not going out and just off the cuff inventing techniques, we are repurposing established methodologies.”
The basic technique of analyzing bacterial DNA goes back several decades to when scientists made evolutionary comparisons of different bacteria to determine which were related to which. A ribosomal gene common to all bacteria was identified; this gene has conserved areas and hyper-variable regions. Researchers can pinpoint the conserved areas, which are the same in all bacteria, amplify them with primers, and then look at the variable regions between the primers. This information often identifies the species or genus of bacterium.
“So you collect a soil sample, get DNA from it, amplify all of the bacteria that are in there, sequence that variable region and from that one sequencing reaction you’ll get maybe 150,000 sequences,” Foran said. That is a good representation of all of the bacteria in that soil sample, and with NGS he can look at 96 or more soil samples at once. “We can have questioned soil samples in there, known soil samples, aged soil samples, and on and on, and you analyze the collection of bacterial sequences and ask which is most similar to which.”
He can also look at the changes in bacterial populations over time, which is important in forensic cases. For example, if a victim’s body is unearthed months after it was buried, will the bacterial profile from the site still match the soil on a shovel stored in a suspect’s shed? “When we look at temporal differences we find that the soil sample we collected from a burial site — and we’ve collected days, weeks, and months later — types back most closely to that sample on the shovel, as opposed to known samples from other areas,” Foran said.
“However, the environment in which the shovel is stored, whether hot or cold, moist or dry, also affects the bacterial population,” Foran said, “so it is most effective to collect a sample from a crime scene and age it under similar conditions.” “The changes are pretty predictable,” he said. “No matter what soil samples you use, the changes that happen are consistent. Certain species (of bacteria) increase really fast, and certain species drop off equally fast, depending on the environment.” The result is that the comparison between two locations is still statistically accurate.
Contamination of a soil sample by blood is also a factor in forensic analysis, but it can be accounted for. “Some bacteria grow really well in blood,” Foran said, “so if we know there is blood in with the soil, we’d expect certain bacteria to have increased rapidly, while other species may do poorly.”
“Once the bacterial profiles and comparisons are done,” Foran said, “they must be presented in court in scientific, yet understandable, ways.” “We’re doing our statistical analyses in two ways,” he said. “One is for the jury, so they can look at something and say, ‘Oh, I get it.’ The other is for a purely objective analysis that actually puts a number on it.”
Using abundance charts and multi-dimensional scaling, Foran can provide visual interpretations of the data that, while somewhat subjective, allows juries to see a graphic representation of the data, grouping bacterial populations that are more, and less, similar.
“Although that is useful for the jury,” Foran said, “it doesn’t provide an actual statistical value.” To obtain an objective number, he uses supervised classification techniques. There are several such techniques, but in the end they all provide confidence values. The result is an “objective, statistical number” about the likelihood of two samples having a common origin, “and you simply present those values for any site in question, such as a suspect’s backyard versus a burial site,” Foran said.
As the techniques for matching bacterial populations improve with the advent of next-generation sequencing of DNA, Foran hopes the use of bacterial profiling will gain wider acceptance in crime laboratories. For the moment, crime labs are not using next-generation sequencing, but as the technique becomes standard in non-forensic settings, Foran expects crime labs will incorporate it into their work for a host of forensic investigations.
“Once they have that technology in their labs,” he said, “my guess is that they’ll be much more accepting of using it for bacterial profiling.”
The research discussed in this article was supported by NIJ-funded grant 2013-R2-CX-K010 to Michigan State University.