Satellites and AI used to track UK hedgehogs in bid to slow decline

Conservation technologists and environmental policymakers should take note. This project demonstrates that satellite AI tools built for precision agriculture or urban planning can be repurposed for biodiversity monitoring at a fraction of the cost of bespoke field surveys. The unanswered question is whether the UK government's nature recovery targets will fund this kind of infrastructure or leave it dependent on academic grants.

Britain's hedgehog numbers have collapsed. Researchers are turning to satellites for answers

The UK's hedgehog population has been in sharp retreat for decades. Rural numbers have fallen by up to 75% since 2000 according to a 2022 survey, and the species now sits on the Near Threatened list of the International Union for the Conservation of Nature. Researchers at the University of Cambridge have concluded that conventional monitoring methods cannot keep pace with the scale of the problem, and have turned to an AI tool called Tessera to build something more systematic.

Tessera works by processing detailed satellite images of the British landscape and identifying locations where hedgehog habitats exist or have recently been lost. The resolution is fine enough to capture individual hedgerows, which are a critical feature of hedgehog movement and feeding. Where cloud cover interrupts the satellite view, the AI fills in predictions based on surrounding terrain data, maintaining coverage across the map.

The project's wider aim goes beyond simply recording where hedgehogs currently live. Prof Silviu Petrovan of People's Trust for Endangered Species, who is involved in the work, describes the goal as understanding the specific barriers that prevent animals from finding food and mates and moving through the countryside without risk. Pinpointing those friction points is a prerequisite for meaningful intervention.

For conservation organisations, this matters because AI & Tech tools are increasingly being evaluated not just for commercial applications but for their potential to address environmental problems where human survey capacity is limited. The hedgehog mapping project sits within a broader push to apply machine learning to species monitoring worldwide, a trend that has attracted both enthusiasm and caution from researchers.

How the satellite maps work alongside GPS collar data

The Tessera-generated maps are designed to be used in combination with other data sources rather than in isolation. One of the most direct pairings is with miniature GPS trackers attached to individual hedgehogs, which log real-time movement data. Where the satellite system shows habitat at landscape scale, the GPS layer adds behavioural detail, showing exactly which routes animals use, how far they travel in a night, and where they encounter obstacles.

A similar tracker programme is already operating in Northern Ireland, where hedgehogs carrying small backpack-style devices have provided detailed movement data to inform local conservation planning. The Cambridge team's satellite approach could eventually complement that ground-level work by providing the broader landscape context that individual trackers cannot supply on their own.

Tracker data from such programmes has shown that hedgehogs in fragmented urban and suburban environments often cannot cross roads or navigate through enclosed garden boundaries, effectively trapping small populations in areas too limited to sustain breeding numbers. Identifying those chokepoints on a national scale, rather than street by street, is exactly the kind of problem that satellite-derived habitat maps are better placed to address than any ground survey programme could be.

The maps also have a monitoring function over time. When new housing developments are approved or agricultural land use changes, the satellite data can be re-run to assess what those changes mean for nearby hedgehog populations. That kind of longitudinal tracking has historically been difficult to sustain with volunteer-based survey methods, which tend to produce snapshots rather than continuous records.

This capacity to integrate satellite outputs with tracker data, planning records, and environmental assessments places the Tessera approach closer to the modelling work used in climate and land use research than to traditional wildlife surveys. The expansion of AI skills across the UK workforce may eventually make it easier to deploy and interpret these tools outside of specialist academic settings, broadening their practical reach.

The environmental cost question researchers cannot yet answer

Not everyone within the conservation research community greets AI tools with uncritical enthusiasm. Some researchers have flagged the energy consumption of the machine learning systems that underpin tools like Tessera as a concern, particularly as the volume of satellite data being processed continues to grow. The tension is real: a technology designed to protect ecosystems carries its own environmental footprint.

The counterargument, made implicitly by those running the hedgehog project, is that targeted conservation AI operates at a fundamentally different scale from the large commercial models used in content generation or financial services. A system processing satellite tiles across Britain is not drawing on the same infrastructure as a general-purpose language model serving millions of requests. Whether that distinction will satisfy critics as conservation AI scales up is an open question.

The broader debate over AI's power consumption is one that researchers working on environmental applications will need to engage with more directly. As seen with other AI deployments that have generated unintended consequences, the case for careful oversight applies even when the stated purpose is beneficial. Transparency about the energy and compute costs of conservation AI would strengthen, rather than weaken, the case for using it.

What this means for UK wildlife conservation funding

The Cambridge hedgehog project illustrates a pattern now visible across UK environmental research: academic teams are doing technically sophisticated work that sits well beyond the reach of the conservation charities that would most benefit from it. People's Trust for Endangered Species and similar organisations operate on tight budgets and depend on volunteer networks that produce inconsistent data. Satellite AI tools could transform that picture, but only if funding routes exist to move research from university labs into routine conservation practice. Britain's nature recovery targets are legally binding under domestic legislation introduced after Brexit, yet the gap between policy ambition and on-the-ground monitoring capability remains wide. Previous attempts to nationalise species monitoring, such as the national bird surveys run through the British Trust for Ornithology, took decades to establish at scale. The hedgehog may prove a useful test case for whether AI can compress that timeline significantly.