Except for the dead bodies, the landscape is idyllic. There is a forest of shortleaf pines, with boxelder and white ash trees nearby. There is restored prairie land, bottomland hardwood, and an area called Palmetto Flat. The overall ecology is not only inviting but important, as the soil conditions, temperature, rainfall and heat all affect how the human bodies scattered across the land decompose.
The cadavers are purposefully placed there to help scientists, medical examiners, and criminal investigators. The donated bodies at the Southeast Texas Applied Forensic Science Facility (STAFS), near Sam Houston State University in Huntsville, answer questions about the biology after death, that time when a human body goes from being the home of a living person to, as researchers put it, a “rich nutrient source” for insects and microbes.
Scientists have long known about the important progression of insect populations during the first two weeks of decomposition, and how that can serve as a clock to determine how long a body has been dead. But in recent years, several researchers, supported by National Institute of Justice grants, have tried to extend the clock to a month or more by focusing on the microbes that consume a body after death.
“Microbes are everywhere and everyone has them,” said Jessica Metcalf, an evolutionary biologist at Colorado State University, who is one of the leading scientists in developing the microbiome as postmortem interval clock. “Unlike insects, which may not be present in the winter or in other circumstances, there are always microbes,” she said. “We have them in our bodies and they are in the environment.”
Although individual humans live in mostly symbiotic relationships with the trillions of bacteria, fungi, and viruses in and on the body, at the moment of death the immune system stops and the relationship with the microbes changes dramatically. Waves of microbes, first those inside the body and then those from the surrounding environment, start breaking down organs and other tissue. Knowing which microbes appear and when, as decomposition progresses, is the basis for the postmortem microbial clock. Microbial clocks may cover longer timescales and also be a more precise predictor of the postmortem interval than insects.
Using microbe progression as a clock to determine the postmortem interval has been made possible in the last decade due to advances in DNA sequencing and increasingly sophisticated data analysis, scientists note. Those advances have allowed Metcalf and other researchers to understand the progression of the microbes and to realize that, like the stages of insect development but on a much grander scale, the stages of microbial progression on a cadaver can be reliably tracked.
In a 2018 paper in Forensic Science International: Genetics[1], based on several of her NIJ grants, Metcalf described the current knowledge gaps in using microbes as a postmortem interval indicator and noted that “developing and transitioning new forensic science technologies into the justice system requires overcoming scientific, investigative, and legal hurdles.” To do that, scientists must gather mountains of data on the progression of microbes in cadavers that are decomposing in all seasons in an array of environments.
The STAFS facility and six other “body farms” in the United States provide the fundamental data on the microbial progression of decomposition, but analyzing and categorizing the trillions of microbes presents a problem of scale. Metcalf and her colleagues are gathering data from STAFS and two of the other facilities for their current postmortem interval research.
“Because mammalian cadavers are an important, concentrated source of nutrients in an ecosystem, it is likely that communities of microbes have evolved over hundreds of millions of years to efficiently recycle these nutrients,” Metcalf wrote in the genetics article. “Therefore, mammalian decomposition may include a predictable succession of microbes that reflects the different stages of metabolic function and pathways in decomposition.”
Although the abundance of microbes in and on humans makes them invaluable to researchers trying to determine postmortem intervals, the diversity and complexity of microbial communities make them much more difficult to track than insects. And such communities, Metcalf said, are “shaped by many environmental variables and the challenge is trying to get a hold on whether those variables are having so much influence [on the microbes] that we can’t recover temporal information.”
Sibyl Bucheli, an entomologist at Sam Houston State University, collaborates with Metcalf on determining postmortem interval through the microbiome, and is expanding the research to move beyond bodies decomposing in nature. “Most people don’t die in the woods naked,” she said, and wonders what would happen to the microbes in and around a body left in a house, or put in a steel drum, or wrapped in a carpet, or placed in water?
“The insects are going to be excluded, but the bacteria are always going to be there,” Bucheli said. “Plus, even in situations where we have our insects, how are they interacting with the bacteria? The whole thing is just this big ecosystem, so I started thinking in these different ways. If the insects are excluded from the decomposition ecosystem, can we use microbes the same way we would use insects?” And if the insects aren’t excluded, she asked, how would they influence the microbial ecosystem?
Another complicating factor could be drugs in a cadaver. At STAFS and the other decomposition facilities, researchers are not provided any information about the medical histories of the bodies they receive so the role of drugs in decomposition is difficult to determine. Does a body filled with strong chemotherapy drugs as a result of cancer treatments decompose differently than a body without those drugs? Does a heroin or fentanyl user’s body decompose at a faster or slower rate than a drug-free body? Do the drugs affect how the insects interact with the body? Is the microbiome progression different due to the drugs?
“Drugs are important,” said Bucheli. “Chemotherapy, illicit drugs, prescribed drugs, all of these things are going to matter. But we don’t know, so we just have to do our best and collect as much data as we can.” At the STAFS facility, two bodies were placed at the same time about 10 feet apart, she said. A few days later one was covered with flies, and the other wasn’t. Was it differences of drugs in the bodies, or some other biological difference? The researchers don’t yet know.
Bucheli’s NIJ-supported research on the effect of indoor versus outdoor microbial decomposition has been delayed due to COVID restrictions. Bodies could not be accepted if the person died of a COVID-related illness, Bucheli said. But early in the research, before the pandemic hit, several bodies were placed on the STAFS grounds and data were collected. “We have five weeks straight of microbe collection, so we had completed our first year of gathering data,” she said.
The goal of all of the research, both Bucheli and Metcalf noted, is to create comprehensive datasets of microbes that allow for a more accurate determination of the postmortem interval. There are now two datasets that are useful to investigators — physiological changes to the body in the hours after death and insect progression in the week or two after death. “Our hope is that we can give people three datasets, with the third one being microbial data,” Bucheli said.
To create a microbial dataset that is useful to criminal investigators and solid enough to meet scientific standards in court, scientists are using machine learning, often with an algorithm called a random forest classifier. Machine learning is particularly good at recognizing patterns and relationships in vast amounts of data, and random forest is a method that can classify a new sample based on prior data.
“We’re using hundreds to thousands of species of microbes to try to predict postmortem interval,” Metcalf said. “And to do that you need some sort of computational approach that takes advantage and leverages all that information. So, the machine learning approach we’re using is the same type of thing that [streaming services] use to figure out what it thinks you’re going to want to watch.”
She has found in her early computer models that the machine learning approach remains accurate even with bodies left in vastly different soil types, such as a forest versus a desert. Forest soil contributes significantly different microbes to the decomposition process than desert soil, but the method still works. Although the environments and some of the microbes can be fairly distinct, Metcalf said, machine learning “is not necessarily using all of the data. It uses the microbes that end up being important for the model.” Despite different decomposer communities emerging in different environments, “there are still enough generalizable decomposer microbes in common across all of these environments that we can build a model that predicts postmortem interval pretty well.”
Bucheli noted that “with the machine learning we’re training these computer programs to learn these microbial datasets. Then they can tell us what stage of decomposition we’re in.” If a body is in the “wet” stages of decomposition in the first couple of weeks after death, insects can be used to determine the postmortem interval within two or three days. But in cases where exposure to insects is limited or a body has entered a later “dry” stage of decomposition, microbes may be the best postmortem interval clock.
“I think microbial data are going to be really useful in two ways,” Bucheli said. “The first is when the insects have been excluded (the body is indoors or otherwise isolated from insects), and the second is the line between when insects are active and when the bacteria are active. In my ideal world, all three datasets will be helpful, but if any one of them is excluded, then we can rely on the other ones. The idea is the more you know, the more data you have, the better.”
Her current NIJ-supported research to determine the difference between indoor and outdoor decomposition is intended to develop a new dataset, and her goal is to eventually add more datasets that could make microbial postmortem interval determination a part of the criminal justice system. “Next we might look at a dataset on aquatic decomposition,” she said, “and the next dataset might be burned remains, or possibly hanged remains. Every time we can add one of these datasets then our window of the postmortem interval is going to get smaller, and our estimation will get more precise.”
Metcalf said the microbial models can currently predict the postmortem interval in the first three weeks of decomposition at plus or minus three days. “But we’re still in research and development because we’re still trying to understand in which cases we have to separate databases,” she said. Can a dataset based on a person dying naturally in the woods be useful in determining the postmortem interval of a body found indoors wrapped in a carpet? Are there consistent, predictable patterns despite the many variables that come with decomposing bodies?
“It’s still in its infancy,” Bucheli said of the microbial postmortem interval research. Beyond the data itself, protocols must be developed for properly gathering evidence, packaging it, and processing it in the lab. “But first,” she said, “We have to build this library for law enforcement to even have something to compare [the evidence] to.”
About This Article
The research described in this article was funded in part by three NIJ grants to Colorado State University; 2019-DU-BX-0010 (Estimating the postmortem interval of human skeletal remains using rapid, inexpensive microbiome tools), 2015-DN-BX-K016 (Microbial clocks for estimating the postmortem interval of human remains at three anthropological research facilities), and 2016-DN-BX-4194 (Improving estimates of the postmortem interval with metagenomics and metabolomics). The research was also funded by NIJ grant 2019-DU-BX-0025, awarded to Sam Houston State University.