Ignoring our observation plane circling above the frozen Lake Superior wilderness, the eight gray wolves seemed as harmless as your beloved pooch cavorting with its pals in the yard. Trotting along Siskiwit Bay, they playfully nipped and pawed each other, pausing occasionally to roll in the snow.

But then the alpha male and female moved purposefully away from the shore. They passed through a clearing and plunged into thick woods, the others strung out behind.

They had eaten little for three days. Now they needed to hunt.

A mile northwest, a moose calf lumbered amid fragrant evergreen stands, nibbling sprigs of balsam fir. It was unaware that the pack, guided by a remarkably acute sense of smell, was closing in.

Overhead, John Vucetich watched intently, taking mental notes even though he knew what was coming: the violent climax of a drama that has long fascinated scientists conducting one of the world's longest studies of a predator and its primary prey.

Vucetich, of Michigan Tech University, is co-leader of a team closely monitoring Isle Royale's moose and wolves for five decades. Both species have had their ups and downs, but now may be facing their biggest threat.

Declines in pack and herd populations, coming as average temperatures have been rising, make the scientists wonder if global warming may be writing a new story line for the narrative that played out as the plane followed the hungry pack below.

Wolves' ruthless killing prowess is the stuff of legend. But moose can kick with lethal force. Researchers have recovered wolf skeletons with broken ribs. If a healthy adult moose stands its ground, wolves usually retreat.

A fleeing moose, however, is vulnerable. The wolves will try to bring it down quickly but may stalk it for days, wearing it down using hit-and-run tactics. Their preferred targets are the old, the sick and the calves, like the one that was coming into view.

The wolf pack suddenly attacked _ and soon the bawling calf could be seen heaving and flailing.

One wolf got a solid grip on the snout, another latched onto the hind quarters, and two advanced broadside. The others lagged behind, unneeded.

For several days, the wolves would feast on their kill in the blood-soaked snow. A wolf can gobble 20 pounds of meat in one sitting.

"They'll be fat and happy," Vucetich said.

Until they are hungry again.

Neither moose nor wolves are native to Isle Royale, which is actually a rugged archipelago _ one 45-mile-long island and 450 smaller ones, some little more than weathered rocks, bunched on the giant lake's northwestern side.

If Isle Royale were farther from the Ontario mainland, the two species probably wouldn't be here. If it were much closer, deer, bears and bobcats might have found their way over and complicated the predator-prey situation.

Around 1900, a few moose _ strong swimmers _ somehow managed the 15-mile crossing. They multiplied rapidly, and by mid-century had so overbrowsed the forest that mass starvation loomed. That's when an enterprising wolf couple journeyed to Isle Royale across a rarely solid ice bridge.

The two species soon formed a bond of interdependence that drew the attention of Purdue University conservation biologist Durwood Allen, who began the study in 1958. Ecologist Rolf Peterson, who arrived as a grad student a dozen years later, would eventually take over, aided by Vucetich and other proteges.

They have made Isle Royale a gold standard for documenting symbiotic relationships between predator and prey species and their natural surroundings.

"It's the most well-known wolf study in the world," said Douglas Smith, a project alumnus who now directs wolf research at Yellowstone National Park.

"Nature is so slow to change and evolve, and the short-term studies just give us snapshots of reality. To really begin to grasp the complexity of what's happening, you need decades."

Smith credits Peterson with debunking numerous myths about wolves, including their image as indiscriminate thrill killers. They're actually quite choosy, culling weaker members of herd species.

"There are such strong feelings about wolves, so much of it based on fear instead of facts," said Sharee Johnson, a director of the International Wolf Center in Ely, Minn. "One of the biggest fears is that wolves take too many prey animals. The Isle Royale study shows us definitively that that's not the case, that it balances out."

Peterson and Vucetich say the biggest long-term danger to the island's moose is not the wolf.

Climate change is the likely culprit behind a steady dropoff in moose numbers over the past decade. Isle Royale is on the southern edge of their range, and recent summers have been the hottest since the study began. The moose are showing signs of stress.

No one is predicting their demise for now. But if they decline much further, the wolves _ which rely almost entirely on moose for food _ could disappear.

Winter out here is not for wimps.

The national park, snowbound and bitterly cold, is closed. But the bare trees and white landscape provide the best opportunities of the year for bird's-eye views of animals on the ground. So this is where Peterson, Vucetich, pilot Don Glaser and a helper or two can be found from mid-January into March.

"You're a long way from anywhere if a problem comes up," Phyllis Green, the park superintendent, said while visiting the crew.

Accessible only by boat or floatplane, Isle Royale draws about 17,000 visitors a year. All but 1 percent of the park is wilderness: no roads, no cars, sparse facilities. Most users are hardy backpackers, eager for solitude.

From May to October, the scientists handle tasks such as trapping wolves and fitting them with radio collars, collecting bones, seeking wolf dens and monitoring the vegetation that moose eat.

During winter, they bunk in a staff lodge at the Windigo information center, near the southwestern end of the island.

A wood stove in the living room is the only source of heat. The indoor plumbing is turned off, so they drill through foot-thick ice on Washington Harbor for drinking water and convert the unheated bathroom to a refrigerator.

The outhouse is a short stroll from the side door. But be sure to carry the toilet seat with you _ the one propped near the stove, decorated with painted holly leaves. Bit more comfy than the outdoor metal seat on a below-zero night.

As Vucetich took the final observation flight one overcast afternoon, Peterson baked a hamburger casserole in the propane-fired oven.

"We usually skip lunch when we're out in the field, except maybe a candy bar," Peterson explained.

Soon, Glaser burst through the front door, stamping snow from his boots. "Honey, I'm home," he boomed to no one in particular.

A bush pilot in Alaska most of the year, Glaser has flown for the Isle Royale study since the late 1960s. White-haired, lightly bearded, with ruddy features and a mischievous grin, he's a prankster, delighting in testing the professors' vocabulary with five-dollar words.

But he's all business in the cockpit of the single-engine Piper Super Cub. The plane is so tiny there's barely room for two people sitting back to back. Still, it lands quickly on frozen lakes and has the maneuverabilty to track animals.

Also spending a couple of weeks on the island was Leah Vucetich, John's wife, who earned her doctorate at Michigan Tech studying Isle Royale deer mice. For hours at a time, she was bundled up in an unheated garage near the bunkhouse, processing bits of wolf and moose scat (feces) to extract DNA samples. The goal: insight into everything from gender balance to pregnancy rates and what the animal was eating.

"If you do any kind of wildlife biology," she said, "you have to be OK with poop."

On Isle Royale in winter, you have to be OK with lots of things.

Close quarters, for example. When not working outside, everyone gravitates toward the toasty living room, which doubles as an office.

A diesel generator runs a few hours daily, enabling the scientists to charge up the batteries that power their computers, digital cameras and other electronic gadgets. Wires are draped across tables and an old piano used as a desktop. Snowshoes and boots line the walls near the front door. Wood is piled on one side of the stove; on the other is a clothes rack draped with woolen socks and gloves.

No one complains. In summer, the Vucetiches sleep in a nearby yurt _ a circular, portable tent _ and Peterson lives with his wife, Candy, in a log cabin toward the other end of the island.

"We've got pretty much all the conveniences we need," Peterson said at dinner, finishing off the casserole as Glaser retrieved a gallon of vanilla ice cream from their makeshift freezer _ the back steps.

John Vucetich, sipping Merlot from a recycled jelly jar, agreed: "We've just gotten used to things that other people might think of as hardships."

A 36-year-old specialist in population biology, he quickly became captivated by the mission of tracing two interdependent species in a closed setting. Peterson assigned him to work with Leah, a fellow grad student and kindred spirit who in junior high school had a pet boa constrictor. "People thought that was just weird," she recalled.

But for a budding wildlife biologist, it made sense. Just as spending the seven coldest weeks of the year on Isle Royale makes perfect sense to Vucetich and Peterson.

Relaxing by the fire after dinner, they talked until bedtime about wolves: the intricacies of forming packs and staking out territory, the struggle to survive in a forbidding environment where most die within four years. The scientists' attitudes were respectful, yet clear-eyed.

"When people made wolves the symbol of evil, that was wrong," Vucetich said. "But it's just as wrong to make them a symbol of all that's good, some mysterious icon of the wilderness. We need to learn about them on their own terms."

The next morning, Peterson hiked into the backcountry to collect the skeletal remains of a moose killed by wolves.

Bones and teeth, like scat, are storehouses of information about the animals and their environment. Increasingly, they are showing elevated levels of the carbon isotope linked with global warming.

Peterson, a soft-spoken and unassuming Minneapolis native, retired a couple of years ago from his Michigan Tech professorship _ which enabled him to spend even more time on Isle Royale. Even after nearly four decades, the project has never grown stale to him.

"Somebody introduced me at a talk and said, 'He's going to die out here,'" he said wryly. Not exactly a pleasant thought, but he didn't dispute it.

This icy morning, dropped off by Glaser at the edge of Feldtmann Lake, Peterson swung a backpack loaded with gear across his shoulders and trudged off, pausing occasionally to examine tracks of deer mice and red foxes in the powdery snow. All was quiet, but for the rustle of windblown treetops and the scrunch, scrunch of snowshoes.

In tiptop shape at 58, the wiry biologist squirmed through thick stands of alder, cedar and spruce, homing in on the spot designated by his hand-held GPS unit. The terrain was rugged, swampy and not without danger: A few days earlier, walking alone, he'd plunged through a weak spot in the ice. "I just fell forward as fast as I could and got myself out," he said nonchalantly.

Peterson reached the site after a vigorous 90-minute tramp.

Bloodstains and paw prints marked the spot of the kill. Following drag marks down a dry creekbed, he found a thick mat of brownish-gray fur and scattered bones, which he stowed in plastic bags. He did likewise with a couple dozen scat piles and made notes, fingers stiff with cold.

Then back to camp.

Meanwhile, in a clearing near the Windigo lodge, a pungent odor arose from a 35-gallon garbage can of water, heated by a hissing propane burner. Vucetich dropped in a moose femur, or leg bone. It would cook about 10 hours to remove hair and muscle.

The Michigan Tech team has collected bones of more than 4,300 moose over the project's 50-year history. They offer hints about why the animals died: Low fat levels in marrow indicate malnourishment; arthritic malformations suggest they might have been unable to outrun wolves.

Such details are a treasure trove for wildlife biologists with an interest in predator and prey species.

And they're valuable clues for Peterson and Vucetich, who have a mystery to solve. The moose population has nose-dived in recent years. They need to know why.

For most of the 50-year study period, scientists have observed a pattern. When wolf numbers were low, moose flourished and the herd aged. That boosted kill opportunities for wolves, whose numbers rose as the moose declined, giving the island's vegetation time to recover from overbrowsing. In time, a shortage of moose would push wolf numbers down again. And so on.

Both have survived catastrophes. A parvovirus outbreak two decades ago nearly killed off the wolves. The moose population then exploded, only to tumble from 2,400 to 500 during the vicious winter of 1996, when deep snow made food hard to find.

They began to recover, but in recent years have fallen again _ harder than ever. Peterson and Vucetich estimated the herd at 385 in 2007, the lowest since the study began. The wolf count fell from 30 the previous year to 21.

And the animals began acting strangely.

Moose urgently need to fatten up during the short summers, but instead have been goofing off _ taking dips in the water, lounging in the shade. Wolves, which instinctively shun people, have wandered into campsites in broad daylight.

The biologists figure it must be the heat.

Not only does it make moose lazy. It's worsening an onslaught of ticks, which infest the big beasts, making them waste feeding time biting their fur and rubbing against trees.

Their dropoff has made the wolves more desperate for food and less shy around people. They've stepped up raids on each other's territory. An alpha female was caught trespassing on a rival pack's turf last year and paid with her life.

Will there be any wolves or moose to study in another 50 years?

No one knows. But if this year's winter study is any indication, it's far too early to count them out.

Peterson and Vucetich headed to camp in January fully expecting to find both species down further. Instead, there were 23 wolves _ two more than during the previous winter. A new pack had formed, bringing the total to four.

Later came another twist: The moose census produced an estimate of 650, up 40 percent in just one year. They may have undercounted the moose in 2007, when lack of snow made it harder to see them from the air.

What's certain is that the herd remains well below its historical average and doesn't figure to grow much soon. The biologists still expect wolf numbers to fall but acknowledge they've been mistaken before.

The sky was dark gray the next morning. Arctic gusts rolled in from Canada, making it too dangerous to fly.

Sometimes bad weather grounds the scientists for days. But there's always work to be done. Dust covered the board games and decks of cards stacked on a bookshelf.

"I think we played a game of hearts sometime this year," Peterson said.

Instead, John and Leah Vucetich entered data into a computer spreadsheet. Peterson tinkered with a video camera they'd begun concealing at kill sites to catch images of scavenging wolves.

Digital photography is among technological advances making their study more sophisticated, Peterson said. Another is fecal DNA analysis, the basis of a paper they plan to publish in a couple of years, on the effects of inbreeding.

Always more to learn.

Before long, the weather cleared and they were back in the air _ spotting kill sites, watching the newest pack hound a wounded moose for five days.

And they saw wolves breaking away from their packs to search for mates. New litters of pups would be born. Spring would come.

On the Net:

Isle Royale study homepage: http://www.isleroyalewolf.org

Copyright 2008 The Associated Press. All rights reserved. This material may not be published, broadcast, rewritten or redistributed.