Strange bouquet

Julia Hinde reports on a new controversial classification of plants, based not on their appearance but on their genes

Step into Mark Chase's Kew Gardens laboratory and there is not a plant in sight. The whiteness and sterility - far removed from the rows of mounted specimen trays and pressed flowers one associates with taxonomists - is striking.

For Dr Chase - who this week publishes with 50 collaborators a new classification of flowering plants - studies not whole flowers or plants, but their tiniest components, their genes. He has spent the past ten years in front of a computer with sheets of DNA sequences up to 2,000 letters long.

What Chase and his collaborators are suggesting - a flowering plant classification based on DNA instead of visible physical attributes - is set to transform how botanists and chemists see relationships between plants, and their evolutionary line. No longer is the Asian lotus described as a close relative of the water lily, which it resembles. Instead, Chase suggests, its closest relative is the English plane tree, more commonly found in London squares than tropical jungles. His new classification may bring the 200-year-old question of which species of plants are related - a subject initiated by Swedish naturalist Karl von Linnaeus - to a 21st-century conclusion.

Chase, an American, has not always been interested in such questions. His original work concerned inch-tall, pale-green orchids from South America. Fascinated by their profusion (there are more than 20,000 naturally occurring species) and their sophisticated relationship with fungi, he went in search of orchids' origins and their closest relatives.

The literature, he says, was full of contradictions. Until recently, taxonomy has been based on morphological features and, less commonly, chemistry. One taxonomist would emphasise a particular physical feature of a plant, such as its complex flower or its seed leaves, while another would point to its acid production. Each would use that feature to find the plant's closest relatives. But Chase began investigating different techniques. By the late 1980s, he was one of the first taxonomists to suggest relatedness through plant DNA. After stints at US universities, Chase, now 47, crossed the Atlantic to Kew.

By now his interests had spread beyond orchids. He was studying genes from a number of species and from both sides of the traditional divide in plant taxonomy - monocytoledons, plants with one seed leaf, such as orchids, and dicots, those with two seed leaves, such as beans. With two colleagues, he started comparing the genes responsible for particular plant functions, looking how they varied between species. Starting with the gene that codes for rubisco, a protein which is important in photosynthesis, they called for anyone studying the gene to contact them. They heard from 40 researchers, who between them had sequenced 500 genes. In a field that has traditionally worked independently, the sheer scale of the collaboration was impressive.

Chase brought the data together in a 1993 paper, co-authored by all but six of the worldwide collaborators. He was looking for similarities in the gene sequence between each species. The closer the sequences, the more closely related the families, he suggested.

"The results from 1993 didn't look like any other classification," says Chase. "It didn't recognise the split between mono and dicotyledons, but instead a split between plants in terms of pollen character." It was something that could only be seen by looking through a microscope when taxonomists have historically favoured visible classifications.

Chase, his collaborators and critics were not certain whether sequencing a different gene, involved in translating DNA into protein rather than related to photosynthesis, would produce similar relationships between families. It had taken ten years to look at one gene, but sequencing techniques were now much faster.

According to Chase, the results and relationships for the second gene were so similar that people who had scoffed at the 1993 paper had to concede it might have some value.

A third gene - this time involved in the breakdown of energy in the plant - again showed the same relatedness between species, persuading Chase and his colleagues they were on to something. The new classification, published this week in the Annals of the Missouri Botanical Garden, is the result. "It is like saying bingo," says Chase. "It's like taking 500 species, throwing them in the air and letting them fall. Each ends up in exactly the same place three times in a row. I think when people really absorb this, they will have to admit it is very difficult to ignore."

So how does Chase explain the water lily and lotus's morphological similarities, when DNA evidence shows them to be less closely related than the morphologically opposite lotus and plane tree, which are genetically each other's closest relative?

"What we envisage is that the plane tree and the lotus shared a common ancestor," he says. The fossil record suggests they have both been around for 100 million years, putting their common ancestor around the time of the early dinosaurs. They may no longer display the characteristics of the common ancestor because one has taken a land and the other an aquatic form, just as the whale comes from mammals but has modified to its aquatic environment. Conversely, the water lily and the lotus have come from totally different origins, but have become similarly adapted because they live in similar aquatic environments - like fish and whales.

Chase stresses the classification is open to challenge now the data is published: it is "robust, repeatable and refutable", he says.

No genetic material has been sampled from 18 of the supposed families which contain either particularly rare or short-lived species, or those endemic to politically unstable areas. Within families there is also uncertainty, while according to some the interpretation of the gene sequences is not always straightforward. Changes may be necessary.

"It is amazing progress and profoundly exciting work," says Steve Blackmore, keeper of the department of botany at the Natural History Museum. "But there is no way molecular work will mean we don't need morphological understanding. We are moving towards integration of many data sets including DNA. We will need to add more and more data until we get no more differences."

Chase is convinced his research is reliable. "In the past, people's intuition has got in the way. They have emphasised one feature and ignored the rest that didn't add up."

He says classification is not just for the "esoteric botanist". It will be an essential tool as scientists genetically manipulate plants.

This is the equivalent of the Human Genome Project, says Chase. "That will enable us to deal with human diseases we've never understood before. It will completely change medicine. This is the same thing for the botanical world. It will enable us to get at problems we didn't even know existed."