Weakling ants cheat by pruning the trees.

In the ruthless ant wars that rage through the branches of acacia trees, a perennial loser species holds its own by some nasty pruning, researchers claim.

In a novel defense, an African ant that loses 60 to 80 percent of its overt battles with other species chews off the shoots of the trees it lives in, preventing bridges for invaders, says Maureen L. Stanton of the University of California, Davis.

The trees have evolved adaptations to make them ant-friendly. They produce nectar and swollen thorns that the ants turn into onion-domed chambers. Yet the pruning ant, Crematogaster nigriceps, cheats on its host instead of repaying the hospitality, Stanton says. The ant's pruning kills flower buds, effectively sterilizing the tree, she and her colleagues report in the Oct. 7 NATURE.

"Cheating has evolved to let a competitively wimpy species survive," she says. "This isn't the first time we've seen plant-ants cheating," Stanton notes, but in most other cases the evolutionary pressures that turn ants to parasitism "are poorly understood."

Four ant species compete for occupancy of the 2-meter-high Acacia drepanolobium trees, which form "a very bizarre, Munchkin forest," Stanton says. After tying branches together to invite tree-to-tree invasion, Stanton's team found that the pruners lost most fights.

"It's total carnage," she recalls. The morning after a battle, dead ants under a tree form a half-inch-thick carpet.

The researchers documented that C. nigriceps shapes tree canopies into dense, highly branched masses that don't reach to neighboring trees. Acacia protected from ants for two full growing seasons grew less dense, with 25 percent fewer branches than did trees with unimpeded pruners.

What impresses one biologist about Stanton's study is the impact of the ants' pruning. "You don't think of tree architecture as a result of ant behavior," says Lee Dyer of Mesa State College in Grand Junction, Colo.

Daniel H. Janzen of the University of Pennsylvania in Philadelphia can think of only one other ant species that prunes its trees. That ant nibbles plants in Nicaragua into snarls dense enough to foil hungry birds.

Janzen wonders about the details of African acacia reproduction. For example, he asks, could an acacia patch actually be one multitrunked clone, well able to sacrifice some flower power? Or is the landscape so crowded that reducing the number of flowers doesn't matter much until a disaster clears a hole?

Ant taxonomist John T. Longino of Evergreen State College in Olympia, Wash., welcomes the new study as an antidote to a tendency to analyze ant-plant relationships as one-to-one interactions. "They always were communities," he says.

Stanton points out that one of the big questions in ecology is, "Why doesn't the best competitor drive everyone else into extinction?" Part of the answer, she contends, is that the weaklings have tricks.