Biodiversity helps keep ecosystems healthy.

For years, prominent biologists and conservationists have campaigned for the preservation of biological diversity, despite little proof of their assertion that reducing the number of plant and animal species upsets nature's balance. Now, two experimental studies illustrate the detrimental effects of species loss.

More than 10 years ago, ecologist David Tilman of the University of Minnesota in St. Paul and his colleagues began investigating how as many as 250 kinds of plants could thrive in midwestern grasslands, even though the flora competed for limited resources. They did not expect to address the question of the value of biodiversity, Tilman says.

For this experiment, the researchers created 207 4-meter-square plots distributed among one native prairie and three abandoned fields of different ages.

Each season, they clipped a different 0.33-square-meter section in each plot and analyzed its species composition and biomass--the weight of leaves, stems, and flowers combined. They left some plots alone and added specific amounts of nitrogen fertilizer or other nutrients to others.

In 1987, less than 300 millimeters of rain fell, down from an average of 450 mm. The next growing season brought just 200 mm. The more diverse the plant community, the less its productivity declined during these dry years and the faster it rebounded. These changes "provided data that have not been available before," Tilman says.

During the dry spell, plots with nine to 23 species in the clipped sections produced half as much as normal. With fewer than nine species, the plot's productivity declined precipitously. Those with just one or two types of plants dropped to one-eighth normal, note Tilman and John A. Downing of the University of Montreal.

The plots with many species regained their productivity a year later; those with five or fewer species took more than four years to recover, the researchers report in the Jan. 27 NATURE.

"[This work] demonstrates clearly that species diversity can make a difference in ecosystem4evel characteristics," comments Peter M. Vitousek, an ecologist at Stanford University.

Under the right conditions, a field with just one species can produce as much biomass as one with many plants. This observation had led some to suggest that the existence of lots of kinds of plants doesn't add much to the health of an ecosystem because those species are "redundant." This study shows otherwise.

"Biodiversity really is an insurance policy against catastrophe," Tilman says. "Areas with more species are more stable."

The presence of enough different plants ensures that some can withstand-whatever stress nature provides, be it fire, flood, drought, disease, or insect pest, he holds. Diverse fields are more likely to include plants that can tolerate the stress. Those plants use nutrients freed up by the loss of less fit species and help maintain the overall productivity of the ecosystem.

Earlier, Tilman had examined the drought's effect on species diversity, restricting his analysis to the unaltered plots. In this subset, the average number of species present dropped from 13 before the drought to eight in 1988. Only in 1993 did they see those species numbers rise again, Tilman told ScIENcE NEws.

The second study evaluated species diversity in normal conditions. For nine months, John H. Lawton maintained 16 1-square-meter plots in special chambers at the Natural Environment Research Council Center for Population Biology in Ascot, England. He subjected plots with two to 16 plant species and increasing numbers of soil critters and herbivores to controlled conditions, then monitored changes in ecosystem function.

Unpublished data from the study indicate that the simpler the ecosystem, the less able it was to take up carbon dioxide and the faster decomposition occurred. Species numbers also affect the cycling of water and nutrients, but in varying ways, Lawton told SCIENCE NEws.

Previously, he had argued that species number may not matter much to ecosystem function. However, "diversity makes a hell of a difference," he says now.

Vitousek and Lawton note that these two studies deal with just a few dozen species. It is unclear whether the correlation between biodiversity and ecosystem health holds in situations where many species coexist. Lawton suspects that those species may be redundant in normal conditions, such as those he studied, but may prove vital in catastrophes.

Nevertheless, this information applies to all ecosystems, even managed ones. "I think it will probably affect how people manage resources," Vitousek says. "I suspect they will hedge their bets more than they did."