In the 1980s, a large lake — Lago Guri — was created as part of a hydroelectric project in Venezuela. Islands in the lake have enabled ecologists to test a fundamental hypothesis in their discipline.
Why is the world green? Why have grazing animals with their insatiable appetites not consumed all vegetation and reduced the land to dust? There have been hypotheses, of course, but as with many large-scale ecological problems, it has not proved easy to test any proposal with controlled experiments. One suggestion is that the intensity of grazing is held in check by predation of carnivores on the herbivores, and this hypothesis has at last proved testable. Writing in Journal of Ecology, John Terborgh and his colleagues1 describe a large-scale experiment in which the degree of predation upon grazers varies and the consequences for vegetation can be measured. They show that, without top predators, the world would be less likely to remain a green and pleasant land.
Animal life is supported by the primary production of green plants, and current knowledge2 suggests that for every species of terrestrial plant there are about five species of animal. Undoubtedly, many more species of animal (especially insects) await description than do plants. Not all of these animals feed directly on living plants; some consume dead plant litter, and others prey on the plant consumers. But in the light of such energetic dependency of animals on a plant food-base, it is remarkable that vegetation survives at all — and not only survives, but dominates the biomass of most land ecosystems. The most widely accepted explanation for this, first put forward by Hairston et al.3, is that herbivore numbers are controlled by ranks of predators that keep their populations in check and inadvertently ensure that green plant production continues.
An opportunity to test the hypothesis on a meaningful scale arose when a valley in Venezuela was flooded to develop a hydroelectric scheme, and a lake — Lago Guri (Fig. 1) — was created. The lake is 4,300 km2 in area, and contains many islands of different sizes. Before the valley was flooded, commercial logging of the valley floor was carried out, but the elevated regions were left untouched and survive as forested islands. Terborgh et al. have recorded the ecological consequences of fragmentation of the forest into these isolated units over many years4, and have described the relationship between island size and species richness, which follows the model described by the theory of island biogeography5. Species losses, predictably, have been greater on the small islands.
Islands of less than 2 hectares (20,000 m2, or about 5 acres) lost many of their vertebrate species within a few years of isolation, and these smaller islands also began to display higher densities of herbivores6 — especially invertebrates, including leaf-cutter ants, but also some vertebrates such as iguana, howler monkey, agouti and tortoise. Land masses of more than 75 ha retained greater numbers of vertebrate grazers, including deer, peccary and a full range of primates, but they also supported predators of these vertebrates, including raptors (such as harpy eagle), snakes, ocelot, puma and jaguar. The Hairston ‘green world’ hypothesis would predict that the very small islands that lacked predators and developed high densities of herbivores should experience a decline in vegetation. Medium-sized islands (less than 15 ha) with some vertebrate predators, such as the armadillo that preys on leaf-cutter ants, would be less severely affected, and large islands with a full complement of predators would remain unchanged.
Terborgh's team periodically surveyed the vegetation of all three types of island. The small islands typically contained about 300 individual trees, so all of these were tagged and their sizes and condition noted. Sample areas (usually about 0.6 ha in extent) with similar tree densities were selected on the medium-sized and larger islands, and the individual trees were recorded in the same way. Changes rapidly became evident on the small islands, which by 1997 had densities of small saplings only 37% of those on the large islands; recruitment and mortality of trees and shrubs had evidently been strongly affected by the increased herbivory under conditions of low predation. By 2002, the density figure for small islands had fallen to 25% of that of the large islands. Tree and shrub mortality over a five-year period was quite high on all islands, but was greatest on the small ones, which experienced 46% mortality compared with 32% on the large islands.
The researchers consider other causes, but conclude that the loss of animals that preyed upon vertebrate grazers and leaf-cutter ants on the small islands set in motion a trophic cascade that destabilized the food web. Such cascades, where the removal of one trophic level (in this case, top predators) causes knock-on effects through other trophic levels, are well documented from aquatic communities7. They have proved difficult to demonstrate in terrestrial ecosystems, although (for example) the loss of wolves from most of the national parks of the United States has led to increases in vertebrate grazers and overgrazing.
Terborgh et al.1, however, have quantified these effects with great precision and have demonstrated both the extent and pace of the trophic cascade. It remains to be seen whether overgrazing will lead to the total destruction of vegetation on the small islands, and whether that would then lead to herbivore extinction followed by plant reinvasion and the establishment of a new order.