Beyond the Bear

Roughly 37,000 years ago, a brown bear wandered a cooler, forested Vancouver Island, before the last great ice sheets reached their maximum extent. When the animal died, one of its bones came to rest deep inside a limestone cave, where the sheltered environment preserved it while countless others were lost. Thousands of years later, cavers found it, and scientists at the Hakai Institute sent it for radiocarbon dating. The result startled everyone: roughly 37,000 years old, the oldest animal bone yet recovered from the region.

Ask paleoecologist Chris Hebda what the world looked like when that bear was alive, and the honest answer is that no one really knows. The period is, as he puts it, “a bit of an unknown lost world”—old enough that the glaciers which came afterward scraped away most of the evidence. “We actually don't know very much” about what lived on northern Vancouver Island then, he says. A bone this old is a rare witness from a time that is mostly blank.

Which is why, at Hakai, the bear became important less for what it revealed about the past than for what it taught scientists about recovering the past at all.

Ancient DNA is changing which questions scientists can realistically answer about the distant past. Fragments of genetic material preserved in bones and sediments can trace how species are related, reconstruct vanished ecosystems, reach back tens of thousands of years—but only when they survive, and they rarely survive well. Time shatters the DNA; chemistry alters it; modern contamination floods it. When the team extracts DNA from an ancient bone, paleogenomics technician Mac Barrera notes, more than 99 percent of it usually belongs to something else—bacteria, fungi, the microbes that colonized the specimen over millennia. The DNA they actually want can be little more than a whisper. “You're working with the worst possible starting point,” says molecular DNA technologist Danielle Grant.

Recovering something useful from that is not a matter of dropping a sample into a sequencing machine. It is a matter of the hundreds of small decisions made before sequencing begins. Every improvement the team has made serves one modest, stubborn goal: to recover a little more from samples that might otherwise be written off as failures. 

That ambition shaped the lab from the start. When Hakai set out to build it several years ago, the point was never to buy equipment; it was to build capability. The early days, Hebda recalls, were “a bit of a leap of faith”—an attempt to make a dream meet reality. What began as little more than a “garden shed on Quadra” was designed, unusually, clean from the ground up rather than retrofitted from an older building, and staffed by people drawn from different disciplines—archaeology, molecular biology, genomics, bioinformatics, environmental DNA—each arriving with a different set of tools.

ancient DNA scientist Tyler Murchie has spent his energy less on chasing discoveries than on building the methods that make future discoveries possible. Every specimen, he points out, has its own problem: lake sediment behaves unlike cave sediment, a bone unlike permafrost, and even two bones from the same cave can hold their DNA in completely different ways. So instead of running every sample down one identical line, the team asked what would happen if the process adapted to the sample.

The clearest expression of that idea is a system Danielle Grant has helped build. Rather than treat the work as a single pipeline, the lab stops repeatedly to ask what each sample needs next. Is there enough DNA? Are naturally occurring compounds interfering? Would another method do better? Press on, or go back? She pictures it not as an assembly line but as a flower: every sample starts with the same stem—whatever preservation has left behind—and from there each takes its own path, some needing extra coaxing, others moving quickly, until a whole project becomes a bouquet, flowers from one garden, each a little different.

Even what counts as normal is still being worked out. “Ancient DNA normal is not like modern DNA normal,” she says. Much of the craft is learning to tell the difference—and, just as often, refusing to quit. “It takes a lot for us to give up on a sample.”

It is worth remembering what all this careful method is for. When the glaciers advanced across coastal British Columbia, Hebda says, they scoured away most of the physical record of the ecosystems that came before—the sediments, the bones, the ordinary evidence a paleoecologist would normally read. Ancient DNA offers another way back to what the ice erased. And unlike a single fossil, it can begin to return a whole living world: not one animal, but the plants and animals and microbes that shared a landscape, and how they fit together. The bear is one bone. The methods being built around it are aimed at recovering the rest of the picture.

This is where the bear earns its keep. Laboratories usually test new methods on synthetic DNA—molecules built to order. The bear offers something far more valuable: authentic ancient DNA, complete with all the damage, contamination, and unpredictability the team ultimately has to overcome. Because the bone is large enough to sample repeatedly, every improvement can be measured against the same specimen. That consistency has made it an ideal benchmark for refining the lab's methods.

But the bear isn't simply a test specimen. Each time the team sequences another portion of its genome, they also learn a little more about the animal itself. The same work that is improving the laboratory is gradually revealing the bear's own evolutionary history. Every advance serves two purposes at once.

Barrera's work meets the 99-percent problem head-on. Rather than sequence everything and hope, he is helping develop ways to use baits—probes that fish out just the fragments the team is after and leave most of the rest behind—recovering, from the same sample, on the order of 1,000 times more useful information. The bait panels, designed by Murchie, can already target plants and animals across vast spans of geography and time—but that reach has a price: the commercial baits run thousands of dollars apiece, are good for a single use, and only cost more as the targets multiply. So rather than keep buying them, the team is learning to make many of the tools themselves. What has been a fixed and costly supply becomes something they can adjust, expand, and eventually pass to other labs. Owning the synthesis opens a further door, too: with the process in their hands, they can start refining the wet-lab steps around it. The instinct, here as everywhere, is to turn a one-time solution into a reusable one.

That all of it actually works is easiest to see in an unlikely place: fossilized ground-squirrel dung. For years, attempts to pull DNA from those coprolites came back empty, defeated by too little sequenceable DNA and too many interfering compounds. Once the lab's methods were brought to bear, the samples gave up not only the squirrel but its entire diet—plants and animals no one expected to find preserved in feces, including DNA from a mammoth. The team published it with a methods supplement running to some 150 pages: an unusually complete showing of the work, in a field where the effort behind a discovery is usually condensed to a paragraph or two.

For all that, no one on the team talks like they have mastered anything. What comes through instead is a kind of restless humility. Even now, Murchie estimates, most of the signal is lost at every stage—perhaps 99 percent in the lab, and another 90 percent in the computation that follows. “This cookie that we're not getting,” he says, “is actually most of the cookie.” And yet they have been amazed, more than once, by a crumb. From a handful of dirt, or a single old bone, the lab can now reconstruct the genomes of several species at once—and that, Murchie points out, is what's left after almost everything has been lost. Grant describes her favorite part of the work as the constant questioning—not accepting that a result means what it appears to, but asking, again, whether it really does. These are not the words of people claiming certainty. They are the words of people building better questions.

The pieces reinforce one another in the hands of the people using them: the checkpoints decide what is worth pursuing, the bait enrichment recovers more of it, in-house software designs the tools and wrangles the data, and because Hakai runs its own sequencing rather than shipping samples to an outside facility and waiting on a black box, even that last step is something the team can question and improve. That kind of end-to-end control is rare, and it is beginning to draw notice. Grant recently presented the lab's quality-control work at the first International Conference on Palaeogenomics in Stockholm, where other scientists wanted to know how Hakai coaxes results from samples that ought to be hopeless; several of its methods are being readied for publication so other labs can use them too. Discoveries matter—but methods that let everyone make better discoveries may matter more.

Hebda has been chasing one question for as long as he has done this work: what actually lived here, in the deep past, on a coast whose record the ice mostly destroyed. For years it stayed out of reach. “We're finally reaching the point,” he says, “where I think we're going to be able to answer those questions.” The tools that will answer them are being sharpened now, quietly, on a bear.

The bear remains a story in its own right. But its greatest contribution may prove to be the quietest of all—not what it revealed about one ancient animal, but the growing ability to recover the thousands of stories that, until now, science simply had no way to hear.