When Two Forest Foods Failed, Bear Attacks in Japan Nearly Quadrupled

In 2025, bears injured or killed 232 people in Japan, the highest annual toll on record. Eleven of them died.

The government sent in troops to help local hunters. More than 14,000 bears were killed under nuisance permits.

The usual explanation is that there are simply too many bears. A new study in Science Advances points somewhere else: the forest’s food supply.

Why are bear attacks rising in Japan?

Asiatic black bears in northern Japan rely on beech nuts in autumn. They need them to fatten up before winter.

When the nuts fail, bears have to roam much further to find food. Rubbish, crops and orchards near towns become tempting.

Bears normally avoid people, and towns are dangerous for them. The authors argue they cross into human areas mainly when natural food falls very low.

Most attacks in Japan are defensive, not predatory, the authors note. They’re mostly hungry animals bumping into people while foraging.

The study: 18 years in northern Japan

Where and how

Researchers from Keio University in Japan and Beijing Normal University in China studied five prefectures in Tohoku, northern Japan. The region has the country’s densest bear populations and its worst conflict.

They combined three data sources. Official bear attack records ran from 2008 to 2025.

Forestry staff have scored beech flowers and nuts at 145 fixed plots each year. And NASA satellite data measured how much the forest grew each season.

The study counted attack events, not the number of people hurt. That gives a cleaner measure of how often bears and people clash.

What the study found

A tipping point in nut supply

Bear attacks didn’t rise smoothly as nuts declined. They jumped once the nut score fell below a tipping point.

On the survey’s 0 to 5 scale, that point was 0.38. A score below 1 already counts as severe failure, so bears seemed to tolerate a lot before crossing into human areas.

In years below the tipping point, prefectures averaged 15.8 attack incidents. In other years, they averaged 5.9.

The lean years got leaner after 2006

Over the long run, average beech nut production barely changed. But the pattern changed underneath.

The worst years got worse, while the bumper years held steady. So years below the tipping point became more common.

The trees weren’t making fewer flowers. Instead, fewer flowers were turning into mature nuts.

The researchers pinned the change to around 2006. Since then, the region’s summer average temperature hasn’t dropped below 19°C in any year.

They suggest warmer summers may have removed a natural brake on flowering in poor years. But they say the cause is unproven, and poor pollination could also play a part.

When two food sources fail together

Bears have a backup when nuts fail: plants, insects and other forest food. The team used satellite measures of forest growth as a stand-in for that backup.

When both failed together, attacks rose far more than either alone would suggest.

If the effects simply added up, a double failure would mean about 15 attacks. The real figure was 20.7.

Two different disasters

The two worst years, 2023 and 2025, looked similar on the surface. In both, all five prefectures fell below the nut tipping point.

But 2023 was a nut failure alone. In 2025, nut failure struck alongside a sharp drop in forest growth across every prefecture.

That’s why the authors see the two years as a natural experiment. Different routes led to similar disasters.

How much the 2006 shift may explain

The team then modelled a “what if” scenario. If beech had kept its pre-2006 pattern, they estimate about 120 of 786 expected attacks would not have happened.

That’s about 15%. It’s a model estimate, not a count of real attacks.

Why the result needs caution

This is an observational study of one region. It shows strong patterns, not proof that food shortages caused each attack.

Two extreme years, 2023 and 2025, account for about a third of all incidents. That makes the results sensitive to them.

The satellite measure doesn’t say which backup food matters, whether plants, insects or fruit. And the tipping point may differ in other regions.

The authors also challenge official bear population estimates. That’s a contested debate, and this study doesn’t settle it.

The study was funded by the Japan Society for the Promotion of Science, and the authors declared no competing interests.

What earlier research found

Hungry bears move into towns

A 2014 study in PLOS ONE tracked American black bears in Aspen, Colorado, with GPS collars for 6 years.

Bears used built-up areas more, and became more active at night, in years when natural food was poor. They moved back to the wild when food recovered.

Bears also survived less well in poor food years, when they spent more time in town.

Weather and bear conflicts in Japan

A 2020 study in Science of the Total Environment linked weather to human-bear conflicts in Japan. Cold winters, cool springs and hot summers may all raise conflicts by changing food supply.

Culling vs fences

A 2020 meta-analysis in Scientific Reports compared ways to stop bear damage in 77 cases.

Electric fences were among the most effective tools. Shooting bears helped in the short term, but its effect faded over time.

Warming and beech nuts

A 2020 study in Nature Plants used 39 years of data on European beech. Warming made seed years less variable and less synchronised between trees.

That’s a different pattern from Tohoku, but it shows that warming can change how beech trees fruit.

How much should you trust this?

Early. The patterns are clear and consistent, but they come from one region and lean on two extreme years.

What makes it convincing

• It uses 18 years of official attack data and decades of tree records.
• The tipping point fits what’s known about bear behaviour.
• The authors ran checks, including predicting each outbreak year from the other years.
• Earlier studies also link poor natural food to bears entering towns.

What makes me cautious

• It’s observational, so other causes can’t be ruled out.
• About a third of incidents come from just two years.
• Why beech changed in 2006 is still unproven.
• The backup food measure is indirect.
• The views on bear numbers and culling go beyond what the data can prove.

What this means for you

• If you visit rural northern Japan in autumn, check local bear warnings, especially in poor nut years.
• Don’t leave food or rubbish out. Human food draws hungry bears into towns.
• Watch the forecasts. The authors suggest tracking both beech nuts and overall forest growth to spot high-risk years.
• Follow local advice on bear encounters. Japan’s Ministry of the Environment publishes regional information.

In this news report, CBS Mornings looks at Japan’s record bear attacks and the animals’ struggle to find food:

What we still don’t know

1. Why did beech change around 2006? Warm summers are one idea, but it isn’t proven.
2. Which backup foods matter most? Plants, insects and fruit weren’t measured directly.
3. How many bears are there? Estimates are disputed.
4. Does the tipping point apply elsewhere? Other regions have different forests and climates.
5. What works best to cut attacks? Fences, food management and forecasting need testing side by side.

My take: a hunger story, not just a numbers story

What I like about this study is that it asks a better question. Not just “how many bears?”, but “why now?”

I’m less convinced by its sharper claims about bear numbers and culling. Those need more evidence than one region’s data can give.

Still, the core idea is powerful. When the forest fails twice in one year, hungry bears come looking for food, and people get hurt.