Most of us are familiar with sloths, the bear-like animals that hang from trees, live life in the slow lane, take a month to digest a meal and poop just once a week. Their closest living relatives are anteaters and armadillos, and if that seems like an odd pairing, there’s a reason why. Today, there are only two sloth species, but historically, there were dozens of them, including one with a bottle-nosed snout that ate ants and another that likely resembled the ancestors of modern armadillos.

Most of these extinct sloths also didn’t live in trees, because they were too big. The largest sloths, in the genus Megatherium, were about the size of Asian bull elephants and weighed roughly 8,000 pounds.

“They looked like grizzly bears but five times larger,” said Rachel Narducci, collection manager of vertebrate paleontology at the Florida Museum of Natural History.

Narducci is co-author of a new study published in the journal Science in which scientists analyzed ancient DNA and compared more than 400 fossils from 17 natural history museums to figure out how and why extinct sloths got so big.

Ground sloths varied widely in size, from the truly massive Megatherium—which could rip foliage off the tops of trees with its prehensile tongue and acted as a sort of ecological stand in for giraffes—to the modestly chunky Shasta ground sloth that terrorized cacti in the desert southwest of North America.

The same cannot be said for sloths that developed an affinity for tree climbing. Those that lived entirely in the canopy were and are uniformly small, with an average weight of 14 pounds, while those that spent part of their time on the ground averaged about 174 pounds.

You don’t have to be a scientist to puzzle out why trees enforce a strict weight limit. It’s the same reason why modern tree sloths have a strange elastic quality to them: Branches break when put under too much strain, and sloths are not generally known for their ability to swiftly avert sudden disaster.

Tree sloths have reportedly survived falls of up to 100 feet. However, given that falls from even moderate heights can cause severe damage and some trees in the Amazon Rainforest top out at just under 300 feet, it makes evolutionary sense to be as small as possible when going out on a limb.

What’s less clear is why some ground sloths grew to such excessive sizes while others seemed content with being merely large. There may have been several reasons, which is why it’s been so hard for scientists to answer the question with confidence.

Larger sizes might have been advantageous for finding food or avoiding predators, for example. Ground sloths had a special fondness for caves, and their size undoubtedly played a role in their ability to find and make shelters. The moderately sized Shasta ground sloth favored small, natural caves bored by wind and water into the cliffsides of the Grand Canyon, like the alveoli of a gigantic, geologic lung. These also doubled as convenient latrines; in 1936, paleontologists discovered a mound of fossilized sloth poop, bat guano and packrat middens more than 20 feet thick in Rampart Cave, near Lake Mead.

Larger sloths weren’t restricted to pre-existing caves. Using claws that are among the largest of any known mammal, living or extinct, they could carve their own from bare earth and rock. Many of the caves they left behind are still around with claw-mark décor along the interior walls, evidence of their ancient nesting excavations.

Other factors that may have contributed to their size discrepancy include climate, the degree of relatedness among sloth species and metabolic rates. The ability to accurately discriminate between these several possibilities required a substantial amount and various types of data.

The authors combined information about the shape of fossils with DNA from living and extinct species to create a sloth tree of life that traced the sloth lineage all the way back to their origin more than 35 million years ago. With this scaffold in place, they added results gleaned from decades of research about where sloths lived, what they ate and whether they were climbers or walkers.

Because the authors were specifically interested in the evolution of size, they collected data for the final analytical ingredient by measuring hundreds of museum fossils, which they used to estimate sloth weight.

This is where the Florida Museum played a special role. “We have the largest collection of North American and Caribbean-island sloths in the world,” Narducci said. She carefully took several measurements of 117 limb bones and shared the numbers with her colleagues.

The authors mixed all this information together, computationally stirred it and got back a fully baked answer.

The result: Size differences among sloths has been primarily influenced by the types of habitats they lived in and, by extension, climate change.

“Including all of these factors and running them through evolutionary models with multiple different scenarios was a major undertaking that had not been done before,” Narducci said.

The sloth dynasty coincided with significant, life-altering changes in Earth’s climate. The oldest thing that scientists can reasonably consider to be a sloth is called Pseudoglyptodon, which lived 37 million years ago in Argentina. Analyses from the study indicate the earliest sloths would have likely been small ground dwellers, about the size of a great Dane.

At various points throughout their evolutionary history, sloths adopted a semi-arboreal lifestyle. Not all of them stayed in the trees, however. The largest sloths, including Megatherium and Mylodon, likely evolved from a tree-adapted sloth that ultimately decided to stay firmly planted on the ground.

Against this background of indecisive climbers and walkers, the size of sloths hardly changed at all for about 20 million years, irrespective of their preferred method of locomotion. Then something earth-shattering occurred.

A giant wound opened up between modern-day Washington state and Idaho down through parts of Oregon and Nevada, and magma boiled out of it. This left a nearly 600,000 cubic mile scab over the Pacific Northwest. It’s still visible in some places along the Columbia River, where millions of years of running water have cut through and polished a colonnade of basalt.

These rock pillars have a distinct hexagonal shape caused by the way in which the magma hardened and cracked as it cooled. The volcanic event that made them was a slow burn that lasted roughly 750,000 years and aligned with a period of global warming called the Mid-Miocene Climatic Optimum. The greenhouse gases emitted by the volcanic eruption are currently considered the likeliest cause of the warming.

Sloths responded by getting smaller. This may be because warmer temperatures brought increased precipitation, which allowed forests to expand, thereby creating more habitat for smaller sloths. Size reduction is also a common way for animals to deal with heat stress and has been documented in the fossil record on several different occasions.

The world remained warm for about a million years after the volcano fell silent. Then, the planet resumed a longstanding pattern of cooling that has continued in fits and starts to the present. Sloths reversed course too. The more temperatures fell, the bulkier they became.

Arboreal and semi-arboreal sloths had the obvious limitation of having to live near trees, but ground sloths lived just about anywhere their feet would take them. They climbed the Andes Mountains, fanned out through open savannas, migrated into the deserts and deciduous forests of North America and made a home for themselves in the boreal forests of Canada and Alaska. There were even sloths adapted to marine environments. Thalassocnus lived in the arid strip of land between the Andes and the Pacific. They survived in this harsh region by foraging for food in the ocean.

“They developed adaptations similar to those of manatees,” Narducci said. “They had dense ribs to help with buoyancy and longer snouts for eating seagrass.”

These varied environments presented unique challenges that ground sloths met, in part, by beefing up. “This would’ve allowed them to conserve energy and water and travel more efficiently across habitats with limited resources,” Narducci said.

“And if you’re in an open grassland, you need protection, and being bigger provides some of that. Some ground sloths also had little pebble-like osteoderms embedded in their skin,” Narducci said, referencing the bony plating that sloths had in common with their armadillo relatives, a trait that was also recently discovered in spiny mice.

Equally as important, larger bodies helped sloths contend with cooling climates. They reached their greatest stature during the Pleistocene ice ages, shortly before they disappeared.

“About 15,000 years ago is when you really start to see the drop-off,” Narducci said.

There’s still debate about what happened to sloths, but given that humans arrived in North America at about the same time sloths went extinct in droves, it’s not hard to speculate. Paradoxically, the large size that kept them safe from most predators and insulated from the cold became a liability. Neither fast nor well-defended, ground and semi-arboreal sloths were easy pickings for early humans.

Arboreal sloths watched the carnage unfold below them from the safety of the treetops, but even there, they didn’t escape without losses. Long after their ground-dwelling relatives had gone extinct everywhere else, two species of tree sloth in the Caribbean held out until 4,500 years ago. Humans arrived in the Caribbean about the same time that Egyptians were building the pyramids. Caribbean tree sloths went extinct not long after.

Alberto Boscaini, Néstor Toledo François Pujos, Eduardo Soto, Sergio Vizcaíno and Ignacio Soto of the Consejo Nacional de Investigaciones Científicas y Técnicas, Daniel Casali, Susana Bargo of the Universidad Nacional de La Plata, Max Langer of the Universidade de São Paulo, Juan L. Cantalapiedra of the Universidad de Alcalá, Gerardo De Iuliis of the University of Toronto and Timothy Gaudin of the University of Tennessee at Chattanooga are also co-authors of the study.