Xu Xing's dinosaur finds range from a pint-sized creature with four wings to the feathered ancestor of Tyrannosaurus rex. Between them, they have cemented the evolutionary link between dinosaurs and birds. He talks to Phil McKenna about his work

How did you become interested in dinosaurs?

I grew up in a remote region in Xinjiang province. I didn't even know what a dinosaur was when I was young, not even when I was in high school. I was interested in physics, but when I got into Beijing University I was assigned to study palaeontology. I was later admitted as a graduate student to the Chinese Academy of Sciences to continue studying palaeontology. I was only interested because it meant I could stay in Beijing and didn't have to go back to Xinjiang, but early in my career I got to study some very interesting fossils. Now it's hard to imagine how I could live without dinosaurs.

Before your discoveries of feathered dinosaurs, what was the prevailing thinking on the relationship between dinosaurs and birds?

It was generally accepted that birds were descended from dinosaurs. People had found many dinosaurs that shared striking similarities with early birds, yet a few things didn't quite fit. The time sequence didn't seem to be correct, for instance. Most of these bird-like dinosaurs were from the Cretaceous, from 145 million to 65 million years ago, but the earliest known bird, Archaeopteryx, was much older - from the Jurassic, 200 million to 145 million years ago. Also, if birds were descended from dinosaurs, you would predict that their dinosaur ancestors should have feathers or feather-like structures. But at that time there was no fossil evidence for this.

What have you found that changed this?

In 1996 a feathered dinosaur, Sinosauropteryx, was discovered in north-eastern China by other Chinese researchers. So in 1997 we organised an expedition in that area, and that year we discovered a second feathered dinosaur, Beipiaosaurus. That specimen supported the conclusion earlier researchers had made that primitive feathers are widely distributed among bird-like dinosaurs. Then in 1998 we found another feathered genus, Sinornithosaurus. Collectively, these fossils were the first solid evidence of primitive feathers and they were found in dinosaurs. This provided very good evidence supporting the idea that birds evolved from dinosaurs; before this, feathers were only known in birds.

Many of the early bird-like feathered dinosaurs you describe could not fly. What did they use their feathers for?

Some believe the initial purpose of feathers was for flight, some say it was for insulation. Based on my observations, I don't think it was either. Primitive feathers were not good for flight: they were weak, hair-like structures. They were also very rigid, not an ideal structure for insulation, and they were less densely populated on the body than you would expect for insulation. It's still not entirely clear, but they appear to initially have had a display function.

The transition from dinosaurs to birds involved some unexpected turns. Do any of your other findings tell us about this process?

In 2003 we made another, even bigger discovery - a new species called Microraptor gui. This species is amazing because it has feathers attached to its legs as well as its forelimbs, so we called it a four-winged dinosaur.

This was one of the most important discoveries in understanding the transition from dinosaurs to birds because it revealed an entirely new stage of morphology during the transition. A hundred years ago, the American palaeontologist William Beebe predicted that there would be a phase in evolution involving a hypothetical four-winged "tetrapteryx". He suggested bird ancestors used not only arm feathers but also big feathers attached to their legs. But at the time there was no fossil evidence to support his theory. Our discovery of M. gui showed that there was a four-winged stage in the evolution of birds. Later we found others: Pedopenna from Inner Mongolia, and in 2009 we published on another species, Anchiornis huxleyi, or "near bird". They are all from different lineages but all have four wings.

What was the response to finding M. gui?

Its existence argues so strongly against the mainstream understanding of the dinosaur-to-bird transition that some people didn't believe it was a real specimen, or even that it was the true morphology for the species. Later, our group and others found more specimens - more than 1000 in all - of four-winged dinosaurs across a number of different lineages, which showed that it is definitely a common and important condition for the origin of birds.

What about the problematic issue of the time sequence?

Anchiornis belongs to one of the most bird-like groups of dinosaurs and lived about 160 million years ago, roughly 10 million years before the first known bird. Since 2001 we have discovered a lot of specimens from Xinjiang, where there is a huge exposure of Jurassic rocks, including the feathered dinosaur Guanlong, the earliest known ancestral form of T. rex, as well as many others that haven't yet been published. And just last month in Science (DOI: 10.1126/science.1182143), we described Haplocheirus sollers from the Alvarezsauridae, a bizarre group of bird-like dinosaurs with a large claw on one hand and very short, powerful arms. This creature has helped us solve the puzzle of how dinosaurs came to look like birds and takes the evolution of another group of bird-like dinosaurs back a further 63 million years. Until this find we had no direct evidence that dinosaurs like this lived 160 million years ago in the late Jurassic. So in terms of establishing the time sequence, this is a huge discovery.

You bought some fossils from farmers and dealers. Is this a common practice - and do you have any concerns about it?

It is true that we purchased many of the specimens from Liaoning province - including some of the first feathered dinosaurs we found - from local farmers or dealers. The density of fossil-bearing rocks is very rich there and the farmers are also very poor. It's common in China to buy fossils, though it is not encouraged. In all other areas - Inner Mongolia, and Xinjiang and Shandong provinces - we have dug all the fossils ourselves. I don't like to buy fossils, and I will say I've bought fewer and fewer in recent years. My colleagues have the same idea. It's not good to buy fossils as it encourages people to dig illegally, and it's not good for science because fossils can be damaged in the process. But in China, so many people are digging and it's not just the museums that buy them. The most difficult decision is when you see a very scientifically important fossil: do you let it go or keep it for science? If you don't buy it, it will go somewhere else and will be a loss for science, but if you do buy it, it will encourage farmers to continue digging.

What are you working on now?

I have found four new dinosaur species that will be published soon. One species is the first known ceratopsid, or "horned dinosaur", from outside of North America. Many of my new finds are from a quarry that recently opened in Shandong province and is the largest known dinosaur graveyard in the world. We have already explored thousands of dinosaur bones and found several new species there. I'm also filming a documentary for National Geographic, on the evolution of feathers.