Harvest may be too heavy to last.

What are now rich forest areas for harvesting Brazil nuts might wane into an impoverished old age unless harvesters change their ways, warns a large international group of scientists.

Brazil nuts alone among internationally traded seed crops come entirely from wild collections in the forest, rather than from farms. Conservationists have praised nut collection as a model for generating income from a tropical forest without destroying it.

That happy view may need rethinking, according to Carlos A. Peres of the University of East Anglia in Norwich, England, and his 16 colleagues. An analysis of tree ages in 23 spots in lowland Amazonia suggests that moderate and intense gathering claims so many seeds that not enough remain to replace old trees, the researchers report in the Dec. 19 Science.

Jason Clay of the World Wildlife Fund--U.S. in Washington, D.C., comments that if this threat should materialize, shrinking income might drive harvesters out of business and away from the trees. "Having people in the forest who have a vested interest in it is one of the things that can help keep the forest as it is," he says.

Brazil nut trees (Bertholletica excelsa) grow globes enclosing up to 25 angular, brown Brazil nuts. An individual tree can yield nuts for 150 years. People gather the nuts, as do native rabbit-size rodents called agoutis.

To judge the effects of human Brazil nut harvests, Peres and his colleagues examined nut-collection sites in Brazil, Peru, and Bolivia and categorized each site as lightly, moderately, or heavily harvested. The researchers also surveyed the Brazil nut trees for size, a standard way of estimating tree ages. Some sites had many slim, youthful trees in line to replace oldsters as they die, but other sites had hardly any members of the next generation.

The researchers then performed statistical tests to see whether various environmental factors might explain the differences in tree sizes. The strongest, most consistent effect was that sites where people conduct moderate or heavy nut collection contain few new trees.

The researchers also ran a computer model predicting the size of trees in a forest where people picked all the nuts. Those results matched the tree-size data that the researchers had observed for heavily harvested sites.

Tropical-forest specialist Lourens Poorter of Wageningen University in the Netherlands cautions that it's hard to prove that a lack of new trees indicates overharvesting. "They showed it can be the case, but it's not 10o percent sure," he says.

Study coauthor Pieter A. Zuidema of Utrecht University in the Netherlands responds that the new study can distinguish causes for sapling deficits on the basis of Brazil nut biology and data from various harvest regimens.