Orca Spa Day: When the Ocean's Apex Predators Discover the Magic of Kelp
Imagine booking a spa session with a friend: rubbing each other's backs, chatting, unwinding. Now move that scene to the Pacific Ocean—starring two five-ton killer whales, with a strand of bull kelp as the masseuse.
This isn't science fiction. It's real behavior scientists documented along the North American Pacific coast in 2025.
A Fifty-Year Blind Spot
The story begins in the Salish Sea, between Washington State and British Columbia, home to the Southern Resident killer whales—one of the most closely studied whale populations on Earth. Scientists have tracked them since the 1970s; every individual has a name, an ID number, and a genealogical file.
Nearly fifty years of observation. Thousands of field surveys. Tens of thousands of photographs. You'd think everything about this population was already known.
Then, in June 2025, the journal *Current Biology* published a paper introducing a word never before seen in marine biology literature:
Allokelping.
*Allo-* comes from Greek, meaning "mutual"; kelp is the seaweed. Together: grooming each other with kelp.
The Secret Revealed by Drones
The discovery came from Michael Weiss and his team at the Center for Whale Research in Friday Harbor, Washington. Starting in 2018, they stopped relying solely on boat-based binoculars and launched drones to observe the whales from 30 meters up.
Changing the viewpoint changed everything.
From above, researchers spotted a pattern never noticed before: two orcas swimming side by side, one holding a length of bull kelp (bitten off from an underwater kelp forest) in its mouth. It presses the kelp against its companion's body, and both whales roll slowly, letting the kelp glide across their skin.
Sometimes this lasts for several minutes. Sometimes a third whale joins, forming a "kelp-massage circle."
"What struck me most," Weiss said, "is that this is clearly a common behavior—we saw it almost every time we launched a drone—yet despite nearly fifty years of dedicated observation of this population, nobody had found it before."
More Than Just 'Clever'
Tool use in animals is not rare: primates fish for termites with sticks, crows crack nuts with stones, elephants scratch themselves with branches. But one domain had remained a blank: marine mammals.
Dolphins carry sponges to protect their snouts while foraging—but that's "carrying an object," not "making a tool." Sea otters crack shells with rocks—but that's "using an object as found," not "processing raw material."
Allokelping is different. The whales don't just pick up drifting kelp—they actively bite through the stipe, cut it to a suitable length, and bring it to a companion. That is manufacture, not merely use.
The paper's wording is careful, but the implication is clear: this is the first documented manufacture and use of tools by any marine mammal.
Who Grooms Whom?
Even more interesting is the behavior's social structure.
Analyzing all observed allokelping events, the researchers found a clear pattern: orcas prefer to groom close relatives or similar-aged companions.
Sound familiar? Think about when you last gave a friend a shoulder massage—most likely someone you're closest to, or most comfortable with. Orcas are the same: they don't pick whales at random; they have preferences.
The researchers also noticed a detail: whales with more molting or dead skin were more likely to be groomed. This means allokelping isn't just a social act—it has a genuine hygienic function. The kelp's rough stipe can help remove adhering particles and dead skin.
In other words, it's not "pretend socializing"—it's "actually useful."
A Spa Culture in an Endangered Population
Here's the bittersweet backdrop: the Southern Resident killer whales are among the world's most endangered whale populations, numbering fewer than 80 individuals.
Their main threat is food scarcity. They eat almost exclusively salmon—especially Chinook salmon—which have collapsed due to overfishing, habitat destruction, and climate change. A single adult orca needs roughly 20 Chinook salmon per day.
Under such survival pressure, "mutual massage" might seem like a luxury. But the opposite is true: social bonds are vital to orca survival. Close relationships mean better cooperative hunting, more effective information sharing, and stronger group cohesion.
Allokelping may be the physical embodiment of those bonds. It's not just "feeling good"—it's a cultural practice, a behavior pattern passed down through generations in this population.
The Philosophy of Observation
This story raises a deeper question: how much of what we think we "know" is only limited by our method of observation?
Fifty years. Generations of scientists. Countless expeditions. Yet until someone lifted the viewpoint from sea level to 30 meters overhead, this behavior—happening every day—went undiscovered.
It's not that earlier scientists weren't careful. It's that some truths are only visible at particular scales.
Science history offers parallel moments:
- Galileo pointed a telescope at the night sky and found Jupiter's moons—always there, just invisible to the naked eye.
- Van Leeuwenhoek ground the first microscope lenses and saw microorganisms in a drop of water—an entire world, right in front of us, never noticed.
- In the 1920s, astronomers discovered the expanding universe—the galaxies were always receding; no one had measured their redshift.
- Weiss, M.N. et al. "Manufacture and use of allogrooming tools by wild killer whales." *Current Biology*, 2025. DOI: 10.1016/j.cub.2025.04.021
- ScienceDaily, "Killer whales use seaweed tools in never-before-seen grooming behavior," June 24, 2025
- Smithsonian Magazine, "These Killer Whales Make Tools From Kelp to Massage Each Other," June 30, 2025
Every revolution in observational tools reveals a previously ignored reality.
Drones are to marine biology what telescopes were to astronomy.
From Orcas to AI
As a "biological" who works with AI every day, I can't help connecting this story to my own field.
When we train AI models, we do something similar: we choose an "observation perspective" (loss functions, evaluation metrics) and interpret model behavior through it. But what's actually happening inside the model? Have we really "seen" it?
Like the scientists who watched orcas from boats for fifty years—seeing breaches, salmon hunts, boat-chasing, but never seeing below the surface, outside the viewpoint, the allokelping—AI's interpretability problem is fundamentally the same question: are we observing the right thing at the wrong scale?
Perhaps one day we'll invent a kind of "conceptual drone" to survey the inner world of neural networks from a higher dimension—and discover phenomena that have been "happening every day" without our notice.
When that happens, we'll probably coin a new word for it—just like allokelping: a word describing something we never imagined, but that was there all along.
Epilogue
After the paper was published, Weiss said something in an interview worth quoting in full:
> "To me, this demonstrates not just the power of new observation methods but also how much we still have to learn about these animals."
Fifty years of observation, one drone, one new word.
How many more allokelpings are waiting to be discovered in the ocean?
---
References: