BALTIMORE, Md. -- Studies of butterfly and ant life on Venezuelan islands newly created by the flooding of the world's second largest reservoir show how dramatically sudden isolation can change even tiny components of a once-integrated ecosystem.
By painstakingly investigating these small invertebrate species for three years, two researchers from Duke University's Nicholas School of the Environment found that inundation has served to exclude some common but highly specialized fruit-eating butterflies from the smallest surviving pieces of the former forest, while at the same time causing leaf cutter ant populations to explode there.
Their studies offer insights into the complex and often unexpected effects of ecological disturbances.
Ghazala Shahabuddin and Madhu Rao prepared reports on their work for presentation at the Ecological Society of America's 1998 Annual Meeting Aug. 3-6.
Their observations and experiments are a part of a larger forest fragmentation study designed by John Terborgh, a professor at the Nicholas School and the department of zoology at Duke. Terborgh, past recipient of a prestigious MacArthur fellowship and an authority on tropical ecosystems, proposed to use the mammoth Lake Guri dam project in east-central Venezuela to simulate what happens to forest habitats that get broken into "islands" by such land-bound activities as agricultural conversion into fields and pastures.
"Generally, it's been found that when chunks of land get separated, dramatic ecological changes take place in the fragments, generally leading to a lot of species extinction at some point," said Shahabuddin, who is one of Terborgh's doctoral students. "I decided to look at butterflies, because very few people look at such invertebrates in forest fragments," she added in an interview.
In 1995, Shahabuddin began to evaluate the ways in which approximately 40 different butterfly species that dine only on rotting fruit have adapted to the flooding, completed in 1986. Using a small motorboat, she and her assistants visited four tiny islands weekly to make butterfly inventories. The now-isolated islands were located between 0.1 and 3 kilometers (a kilometer is about 6/10th of a mile) from "colonizing sources," places where the butterflies were known to have established populations.
Comparing those butterfly counts with others made at "control" sites on larger island and mainland locations, she came to a quick conclusion. "The big thing I found was that there were hardly any butterflies on these small islands, whereas if you went to a larger island or a continuous forest the densities are much higher," she recalled.
She found this puzzling, because her studies also showed many of these butterfly species were perfectly capable of flying to islands even farther away from the mainland. "I was very confused about how I was going to sort out all these effects and study them," she added. "So I decided to concentrate on one butterfly species."
Shahabuddin chose Hamadryas februa, one of a group commonly called "checkerspot" or "calico" butterflies because of their wing patterns. As a group, checkerspot butterflies are widely distributed in the American tropics, ranging north as far as Texas, she said.
However, the species Hamadryas februa is only known to lay its eggs on an inconspicous vine called Dalechampia scandens. This specificity made it easier for her to study why these butterflies might be found on one island but not on another. A helpful clue would be the presence or absence of egg-laying sites.
In 1996 and 1997, she expanded her study area to include 11 small islands and nine different control sites. She also did inventories of Dalachampia, the butterflies' "host" plants. "The most striking thing was that there were big differences from island to island in the numbers of host plants," she said. "One would think that a small woody plant like that would be all over the place. But it just wasn't there on some islands."
Surveying the insects' food sources as well, Shahabuddin learned that "to find a fruiting tree in the forest is a pretty hard job," she said. "Last summer, I found that there wasn't more than one tree per hectare (about 2.4 acres) that was fruiting, and that was during the peak fruiting season."
To further refine her study, she decided to trace the butterflies' movements and breeding behavior with trapping-and-release experiments. She used rotting banana-baited traps to snare hungry butterflies in netting. Then she marked the trapped insect's wings, released them, and retrapped the marked butterflies later on at other locations.
These experiments showed this butterfly species "just doesn't like those small islands," Shahabuddin said. "I found that very little reproduction was taking place on small islands. The butterflies were not laying eggs there. They were laying lots more eggs on the larger islands, and on islands with a lot of host plants.
Apparently small islands were shunned because they lacked enough resources. "It's a very complicated system," she said. "If there is a high host plant density, butterflies tend to stick around longer. Consequently, they tend to lay more eggs. People assume these islands are pretty much similar in terms of resource distribution. That may be true for some species, like rodents, iguana or howler monkeys. But that's not true for some specialized butterflies."
Shahabuddin noted "many of the same processes we are studying have been found to occur in national parks, which resemble forest islands in the midst of a man-modified landscape. Such studies help to pinpoint the ecological changes taking place in our increasingly fragmented landscapes, and thus help to design conservation strategies."
Rao, another of Terborgh's students about to receive her Ph.D., joined Shahabuddin on her island-hopping expeditions to study a different question. Her interest was the abundance of ant species that surgically remove leaves from young trees and carry them away to their colonies to cultivate a fungus they feed on.
Rao found these leaf cutter ants seem to thrive on the very small islands that some butterflies avoid. In fact, her systematic census showed an average of 5.6 leaf cutter ant colonies per hectare on the smallest islands, compared to 2.3 on medium sized islands and only 0.72 on the largest.
The reason seems to be linked to the presence or absence of a major leaf cutter ant predator: armadillos. "They're totally absent from the small islands," she said. "Each armadillo requires three or four hectares in order to survive, and the average size of the smaller islands is much less."
She bolstered this observation with experiments using cages to protect ant colonies from armadillo predation, and found this significantly increased their survival rates. "What my research showed me is that the absence of armadillos has contributed largely to the hyper-abundance of ants on the small islands," Rao said.
Rao went on to investigate the environmental consequences of this "hyper-abundance." After determining which of 43 common species growing there were "preferred" food sources for the ants, she determined the "relative abundance" of preferred species was being reduced in the vicinity of the ant colonies she studied there.
"I found that preferred species were underrepresented in the understory of the smaller islands, indicating that hyper-abundance could eventually affect the tree species composition on small islands," she said.