Â鶹ÒùÔº

A study recently in Communications Earth & Environment reveals how AI and network science can help authorities and conservation organizations combat the illegal wildlife trade by identifying trafficking hubs—even at previously unflagged airports and before incidents are reported.

Researchers from the University of Southern California School of Advanced Computing (a unit of the USC Viterbi School of Engineering) and the University of Maryland, College Park, analyzed the characteristics of almost 2,000 global airports.

Their model predicted 307 airports as potentially involved in illegal wildlife trading, despite no recorded seizures in the available data. Of those, 11 emerged as high-confidence "hidden hotspots"—including two U.S. airports, Dallas Fort Worth International and Denver International, that had not been previously flagged in global trafficking databases.

The predictive model used patterns in historical trafficking data and current insights about key airport features, such as an airport's centrality within flight networks, to identify locations likely involved in the illegal trade. The incidence of flora-related crimes at an airport, along with the strength of local counter-trafficking or law enforcement resilience measures, also emerged as significant predictors.

This approach offers conservation organizations, federal agencies and other decision makers novel insight into global illegal wildlife trade patterns via airports. In addition to the U.S. airports, other hotspots were identified in China, Indonesia, Italy, Mexico, and the Philippines.

The paper, titled "Encoding and Decoding Illegal Wildlife Trade Networks Reveals Key Airport Characteristics and Undetected Hotspots," was lead authored by Hannah Murray, a Ph.D. student in the Thomas Lord Department of Computer Science at USC, with co-authors USC Computer Science Associate Professor Bistra Dilkina, co-director of the USC Center for AI in Society, and Meredith Gore, a professor and research director in the Department of Geographical Sciences at the University of Maryland, College Park.

"These findings can empower decision-makers to make more proactive choices on how to prevent wildlife trafficking, rather than the current reactive approaches," said Murray, a student leader at the USC Center for AI in Society (CAIS).

"The most significant outcome of our model is the practical insights it offers to those invested in combating the illegal wildlife trade, such as how to allocate limited resources and prioritize where interventions are needed to make the most impact."

Revealing the unseen

The illegal taking, trading, and transporting of wild animals, plants, and their products is a major driver of biodiversity loss—but efforts to curb this activity remain underdeveloped, said Dilkina, a leader in the AI for biodiversity conservation space.

"Illegal wildlife trade is the second biggest threat to wildlife after habitat loss and fragmentation, and we urgently need to address it more effectively in order to preserve key biodiversity," she said, "Yet, the knowledge and data-driven tools available for the fight against wildlife trafficking are limited. It doesn't have to be this way."

Murray, who discovered her passion for wildlife conservation while working on her master's in data science at Georgia Tech, where Dilkina previously taught, said her work is inspired by the challenge of revealing what is "unseen" in the illegal wildlife trade.

"What we do know comes from documented seizures," she said. "But what about the incidents that go undetected or are never reported? Our model offers a tool for making those invisible patterns visible."

From insight to action

Armed with this new information, customs authorities could consider starting or increasing cargo and hand-luggage screening at the newly-flagged airports.

"Airlines operating at those locations might also require crew members to complete annual wildlife trafficking awareness training, such as those offered by the International Air Transport Association," said Gore. "In parallel, the conservation community could step up engagement with the airline and passenger transport sectors in Oceania to support awareness-building, monitoring, and improved data collection."

Understanding complex relationships

Dilkina's research on data-driven methods to combat wildlife trafficking includes a prior project focused on detecting and interdicting illicit supply chains, and an ongoing major international initiative called Operation Pangolin, which brings together cutting-edge sensor technology, big data, AI, and interdisciplinary conservation science to combat one of the world's most trafficked species.

Dilkina emphasized the potential of this latest work to inform our understanding of the complex relationships between wildlife trafficking and airport operations.

"Using machine learning models allows us to capture complex nonlinear relationships between the different factors that may play a role in the likelihood of illegal wildlife trade activity at an airport," said Dilkina.

"Importantly, we can also use the model to extrapolate beyond the training data to airports that may not have prior seizures reported and hence uncover possible hidden hotspots."

A foundation for the future

The research expands on earlier work by USC alumnus Aaron Ferber, now a postdoctoral researcher at Cornell University, who developed an AI model to predict flows of illegally traded wildlife.

The team hopes this latest study will lay a foundation for future research that gives decision-makers a stronger, data-driven edge in combating illegal wildlife trade. The approach could also be adapted to address other forms of illicit activity, such as drug and human trafficking.

"Authorities and conservation actors have been frustratingly stuck reacting to offenders who are constantly innovating," said Gore.

"This research leap-frogs over these challenges using the best available open data, exposing previously entombed information about the illegal wildlife trade in the global airline network and laying a soundtrack for a substantial advancement in computational approaches with regard to conservation criminology and team science."