Research news

A groundbreaking new study by researchers the University of Sussex published in People and Nature reveals how AI-automated searches of online platforms can be used to identify where bats are hunted, traded, and consumed across Indonesia, one of the world’s hotspots for bat diversity.

The study presents a new method called “human–nature interface mapping”, which uses modelling techniques typically used in ecology to predict where and how people are likely to interact with wildlife.  

The researchers used AI to screen online news and social media posts from Google, Bing, Facebook, and Instagram in English and Indonesian for reports of bat hunting, sale or consumption. By relating these records to environmental and socio-economic factors such as human population density, or proximity to protected habitats, the team was able to predict where bat hunting, trade, and consumption was most likely to occur in areas where no data were available. The models’ accuracy was tested by assessing how well these predictions aligned with where bats are already known to be hunted, traded or consumed.  

Some of the key findings included:

• Online records as a new conservation tool: Data collected at low cost from publicly available online news and social media platforms can successfully identify hotspots of wildlife exploitation.
• Hotspot regions: Java, Sulawesi, and Sumatra emerged as areas with the highest predicted risk of bat exploitation, meaning that conservation action can now be focused in these locations.
• Hunting vs. trade: Hunting was predicted to occur more widely, and often in different locations than trade or consumption. This reflect the facts that animals may be transported long distances to their ultimate points of use
• Language matters: Data collected in Indonesian provided wider geographic coverage and produced differing predictions from models using only English-language searches, highlighting the importance of multilingual monitoring.
• Drivers of exploitation: Human population density and accessibility were the strongest predictors of bat hunting and trade, reinforcing how human proximity and infrastructure shape threats to wildlife populations.

A methodological first:

The research demonstrates how conservation culturomics — the study of human–wildlife interactions using digital data — can move beyond documenting trends to producing maps of threats to wild species. By integrating online data with environmental and socioeconomic variables, the approach provides an affordable, scalable way to target conservation interventions in data-scarce regions.

“This study presents a new way of using data gleaned from online searches to make predictions about where threats to wildlife populations are taking place” said Dr. Bronwen Hunter, lead author of the study. “It shows that we can harness our vast digital footprint to guide bat conservation, even where traditional field data are limited.”

Why it matters:

Bats are crucial for ecosystems, pollinating plants, dispersing seeds, and controlling insects. Yet in Indonesia, demand for bat meat and traditional medicine has driven population declines. Identifying exploitation hotspots is vital to protect these threatened species before local extinctions occur.

“By combining digital records with ecological modelling, we can pinpoint where bats are most at risk and prioritise where conservation action is needed,” added Professor Fiona Mathews.

Back to news list