Artist reconstruction of _Miyumba chrisdickmani_. Neville GardenWhen you think of carnivorous marsupials, you probably picture the Tasmanian devil or perhaps a spotted-tailed quoll. But these famous predators are only the largest members of a remarkable family of marsupials called dasyurids.Today the dasyurid family contains almost 80 living species. Some live in trees, others in deserts. Some eat insects and weigh only a few grams, while others can weigh up to 14 kilograms and feast on wombats and wallabies. Together they occupy habitats stretching from New Guinea’s tropical rainforests to Australia’s deserts and alpine woodlands.Despite this extraordinary diversity, one of the biggest questions about their evolution has remained unanswered. Where did they come from?Our new research, published today in Australian Zoologist, describes what is now the earliest known member of the dasyurid family – a mouse-sized predator we have named Miyumba chrisdickmani. It pushes the fossil record of dasyurids back by more than five million years and provides the first clear picture of what the earliest members of this iconic Australian group looked like. A long-running puzzleFor decades, scientists have puzzled over the origins of dasyurids.Early researchers assumed they must be an ancient group because their molar teeth exhibit a remarkably primitive design, little changed from the earliest Australian marsupial ancestors.But DNA studies later suggested something very different. Modern dasyurids are actually a comparatively young branch of the marsupial family tree, diverging from their closest relatives around 18 million years ago.The problem was that the fossil record didn’t seem to agree.Almost every fossil dasyurid known was younger than five million years old. One supposed dasyurid, Barinya wangala, from the Miocene Epoch, appeared to have filled this evolutionary gap. But recent work showed it actually belongs to an entirely different family of bizarre snail-eating marsupials called the malleodectids. Suddenly, the fossil record contained no confirmed dasyurids at all before the start of the Pliocene roughly 5.3 million years ago. This left an apparent evolutionary ghost lineage stretching back millions of years.Our new fossil helps solve that mystery. The Riversleigh World Heritage Area on Waanyi country in northwestern Queensland is one of the world’s richest fossil deposits. Timothy Churchill Tiny teeth hold important clueMiyumba chrisdickmani was about the size of a modern antechinus or large dunnart, weighing only around 30 grams. It is known from six fossil lower jaws recovered from the Riversleigh World Heritage Area on Waanyi country in northwestern Queensland. This is one of the world’s richest fossil deposits. It preserves Australian ecosystems that span more than 25 million years. Although tiny, the animal’s teeth preserve an important evolutionary clue.For decades, palaeontologists struggled to identify fossil dasyurids because there was no unique dental feature that clearly separated them from their extinct relatives. Miyumba reveals one at last.In most primitive marsupial predators, the third lower premolar is the largest of the three premolars. In modern dasyurids, however, that tooth is actually smaller than both the first and second premolars. Miyumba possesses exactly this arrangement. This makes it the earliest known fossil to display what appears to be a defining feature of the dasyurid family. That seemingly subtle difference finally gives palaeontologists a reliable way to recognise early members of this important group.The earliest known dasyuridTo test where Miyumba belonged, we compared its anatomy with living and extinct marsupials using two different evolutionary analyses. One examined hundreds of anatomical features preserved in fossils. The other combined anatomical evidence with DNA sequence data from living species while also incorporating the ages of fossil species to estimate when major groups evolved.Both approaches reached the same conclusion. Miyumba chrisdickmani represents the earliest and most primitive known dasyurid. The analyses also identified another small Riversleigh fossil, Mayigriphus orbus, as one of the earliest members of the lineage that eventually gave rise to today’s dunnarts, planigales and other small arid-adapted dasyurids.Together these discoveries push the origin of the dasyurid family back to around 23 million years ago, roughly five million years earlier than previous estimates. Read more: We checked 2000 museum specimens and discovered a tiny new ‘ferocious’ Australian mammal An early life in rainforestsPerhaps the biggest surprise is what these fossils reveal about where modern dasyurids came from.Today, many of Australia’s smallest dasyurids thrive in dry grasslands and deserts. Yet Miyumba lived in the lush rainforest ecosystems that covered much of northern Australia during the Early Miocene. Its anatomy also suggests it was most closely related to modern rainforest species found only in New Guinea.This hints that the earliest dasyurids evolved in rainforests before some lineages adapted to the increasingly dry environments that spread across Australia over the past 15 million years. While many ancient rainforest species disappeared from mainland Australia, some of their descendants may have survived in New Guinea’s tropical forests, where similar environments persist today.A second mystery might be solvedThe timing may also explain another long-standing mystery. Early dasyurids appear just as another group of Australian marsupial predators, the keeunamorphians, disappeared from the fossil record. It’s possible that as these older predators declined, early dasyurids expanded into the ecological niches they left behind before eventually diversifying into the remarkable array of carnivorous marsupials we know today.A handful of tiny fossil jaws may not seem like much. But they reveal the missing opening chapter in the evolutionary history of Australia’s most successful marsupial predators and show that sometimes the smallest fossils answer the biggest questions.Timothy Churchill receives funding from the Australian Research Council, the Riversleigh Society and the Phil Creaser CREATE Fund at UNSW. He is a council member of the Royal Zoological Society of New South Wales.