Back to Bigger Picture

BIGGER PICTURE Nature & food

The map that sends rangers beyond their usual path

A snare can wait silently for weeks. PAWS helps patrol teams decide which pieces of a vast protected area deserve their limited time.

AI-generated illustration: Conceptual ranger team planning a patrol beyond familiar forest routes.
Ranger team planning a patrol beyond familiar forest routes.AI-generated illustration of the idea, not a photograph of the actual people, equipment or event.
Field result
Over five times the usual snares found in a 2018 test
Planning unit
Typically 1 km × 1 km cells
Users
Rangers retain patrol decisions

How the idea works

  1. Combine patrol records and terrain

    Past finds are related to access routes and geography.

  2. Identify uncertain hotspots

    The model estimates risk while accounting for uneven observation.

  3. Search and update

    Rangers remove traps and add new field evidence.

The patrol cannot visit every square

A protected forest is too large for rangers to inspect every trail each day. PAWS uses their own patrol records to suggest where illegal activity may be missed. In Cambodia, the information became concrete assignments: four ranger teams incorporated 16 suggested hotspots into monthly patrols. A prediction mattered only if someone could actually reach and search the ground.

Sources: [1]

No recorded snare does not mean no snare

The model relates past observations to terrain, roads, rivers and other geographic conditions. Its researchers tackled a crucial bias: records are richest where rangers already go. An empty patch on the database might mean safety, an unsuccessful search, or no search at all. Modelling uncertainty helps avoid simply sending patrols back to the same familiar places.

Sources: [1][2]

Success was measured in traps removed

Harvard reported that a December 2018 Cambodian test found and removed more than five times as many snares as usual. A separate USC account of the following December–January period describes over 1,000 snares recovered by 24 rangers. These are tangible field discoveries; they are not interchangeable estimates of a percentage decline in poaching.

Sources: [1][3]

The forest knowledge still belongs to the rangers

Patrol routes must remain feasible and adapt to local conditions. More discoveries can reflect better targeting, more effort, or both; they do not by themselves prove animal populations recovered. PAWS makes accumulated field knowledge easier to use, while the people who know the terrain still carry out the difficult conservation work and feed new observations back.

Sources: [2][3]

The original sources

  1. Harvard John A. Paulson School of Engineering and Applied SciencesPreventing poaching
  2. Xu et al., arXivStay Ahead of Poachers: Illegal Wildlife Poaching Prediction and Patrol Planning Under Uncertainty with Field Test Evaluations
  3. USC ViterbiAI is for Animals

What to explore next

The kitchen bin can explain tomorrow’s breakfastThe bank sees a scam before the money leaves
Explore all 99 applications