Suppose a tree falls in the forest, and takes crashing down with it one of the tenets of ecology. Suppose more specifically this particular tree happens to fall inside a test plot on an island in the Panama Canal, and a team of ecologists are there not to hear it nor to see it, but to see what happens next. (The old conundrum about whether there is a sound when a tree falls and no one — no human, that is — is there to hear it is speciesism at its worst. Almost certainly some creature, backboned or not, perceived the acoustic waves produced by the crashing tree.)

One of the wonders of tropical forests is the sheer diversity of trees, with many times more species per acre than are found in temperate climes. (The same is true, of course, for birds and butterflies, ants and frogs.) A favorite explanation has been that lurking in the understory are seedlings and seeds of fast-growing species that explode toward the light when a treefall opens up the canopy so that suddenly, in one place, light reaches the ground, perhaps for the first time in decades or centuries. As time goes by, a succession of different tree species (with different light-gathering abilities) fill the opening in stages. The understanding was that an intermediate rate of gap formation would produce the most diversity, because there would always be gaps in various stages of filling in.

Well, apparently it's not that simple. Stephen Hubbell, a Princeton ecologist, has been working on this very theory for nearly 20 years on Barro Colorado Island in the Panama Canal, the central research site of the Smithsonian Tropical Research Institute (STRI). Hubbell and Robin Foster, who is now at STRI and Chicago's Field Museum, and Richard Condit (STRI), working with an international team of students from both Princeton and STRI, have painstakingly taken repeated censuses of a 50-hectare (125-acre) plot. In 1982 and 1985, and again in 1990 and 1995, they tagged, measured, mapped and identified all woody plants with a diameter at chest height of one centimeter or more. It was an enormous undertaking, involving more than 300,000 trees of more than 300 species each time. Every year from 1983 through 1996, they measured canopy height and gaps on a grid of 20,301 sample points five meters apart. In particular, they followed the fortunes of 1,284 gaps from 1985 through 1995.

(There is now at STRI the Center for Tropical Forest Science, which coordinates similar studies at 16 other tropical sites in 13 countries. Scientists, all using the same methods so that the results are comparable, are following more than 2 1/2 million trees of more than 5,500 species.)

All that sweaty, totally unglamorous work has paid off. In a paper in Science, Hubbell and seven coauthors reported that species diversity in gaps is no greater than it is in the surrounding forest. Until now, most tropical biologists believed that disturbances — such as treefall gapsóinitiated a sequence where successional tree species replaced one another in the regrowth of the site. In forests that have many disturbances, most of the area is dominated by early pioneer species. In localities with infrequent disturbances, most of the forest is populated by late successional canopy species. With both these rates of disturbance, the diversity of species is low. But in a forest with a moderate amount of gaps, all at different stages of recovery and thus with a full range of successional species, diversity would be maximized. This explanation is known as the intermediate disturbance hypothesis.

It's time for rethinking. As the group wrote in Science, "Spatial and temporal variation in the gap disturbance regime did not explain variation in species richness." Hubbell and his colleagues lean toward a different explanation, that of recruitment limitation. Some trees do not disperse well: not many birds or animals carry their seeds far from the tree, nor does the wind. Some seeds simply do not germinate. Thus, when a gap forms, they may not be present in the immediately surrounding forest, and so there is no chance they will seize the opportunity. They're just not on the dock when their ship comes in.

In a commentary in the same issue of Science, David Tilman, a professor of ecology at the University of Minnesota, wrote: "Like a team that fails to appear at a sporting event, a species that is locally absent has forfeited any chance of competitive victory at the site. This can allow inferior competitors to win by default. If there is recruitment limitation, the winners of local competition are not necessarily the best competitors that exist in the region, but the best competitors that happened to colonize a particular site. This can lead to essentially unlimited diversity."

Hubbell et al. have covered the recruitment-limitation base with ten years of data from seed traps and censuses of seedlings. More than a million seeds of the 260 main census species were collected and identified over those years, most by Osvaldo Calderón of STRI. In a census of 2,000 quadrats, each a square meter, the most commonly encountered species was found in only 15 percent of them. Three-fifths of the main species in the 50-hectare plot were completely missing. The seedlings were just as poor in diversity. Three-quarters of the main species counted in the seedling census were present in less than 1 percent of the quadrants. Poorer competitors end up winning places in gaps because more often than not the better competitors never show up. As a result, species richness measured in gaps is no greater than species richness in control sites.

If we used time-lapse photography to look at a large swath of forest, we would see light hitting the forest floor through treefall gaps as flashes, with the randomness of tempo and location like cameras flashing in a stadium. Gaps are real and they are important for most species of trees to reach maturity, but they are not the primary explanation for the riches of tropical forest diversity.

The finding has immediate application. According to John Terborgh, codirector of Duke University's Center for Tropical Conservation, in North Carolina, it has been suggested that logging could be allowed in forests if it followed the natural pattern of light gaps. Done this way, logging would actually promote diversity, the argument ran. Now, Terborgh says, we know the proposal was based on a misunderstanding.

For Hubbell, finding out that something we "knew" was not so came as no great surprise. "We're still in the Middle Ages in biodiversity research," he has said. "We're still cutting bodies open to see what organs are inside."

By John P. Wiley, Jr.