The evolutionary battle between farmers and the pests that attack their crops began during the earliest days of agriculture, when farmers selected the healthiest plants from one season as the source of seed for the next, making it more likely that each generation would produce more of the repellents and toxins needed to fend off pests. But the pests represented a moving target. As a result, pest populations evolved characteristics and behaviors that enabled them to resist such toxins and repellents. In recent years, this conflict has escalated sharply, as modern farmers not only rely on old methods of dealing with pests but also develop new ones, and the resulting pressure on pests to adapt thereby increases. For this reason, many of the shorter-term triumphs by farmers have carried within them the seeds of longer-term failure. Some years ago, for example, geneticists theorized that one way to control caterpillars would be to incorporate into crop plants genes that stimulated the production of caterpillar toxins derived from a species of bacteria, Bacillus thuringiensis, or Bt. Since this bacterium and the caterpillars had both been around for millions of years, geneticists assumed that if the caterpillars had the genetic potential to develop resistance to the toxins, they would have done so. What the geneticists overlooked was the fact that in natural environments, outbreaks of this bacterium were extremely rare, so the pressure on the caterpillar to adapt to the toxins was quite low in nature. The caterpillars' genetic potential may never have had occasion to prominently display itself. Indeed, this point was borne out when some caterpillar populations developed significant resistance to these bacterial toxins as a result of farmers' unusually heavy reliance on them for controlling caterpillars. Food deprivation appears to offer another effective approach to pest control. Farmers alternate planting of an insect's normal host plant with a plant that it cannot feed on, and the pest dies of starvation. But some insects have even begun to adapt to this normally quite effective strategy. During the 1980s, farmers in South Dakota reported experiencing trouble with the northern corn rootworm, despite the fact that they were routinely rotating corn and soybean crops. Scientists discovered that the rootworm had in fact adapted to this strategy. In large areas of the Midwest, where corn is grown every year, the corn rootworms produce eggs that remain in the soil for one winter and then hatch and feed on young corn roots in the following spring. In areas where fanners rotated corn with other crops that the rootworm could not eat, about 40 percent of the rootworm eggs were remaining in diapause, a rest state, for a second winter. This part of the rootworm population now exhibited a diapause that was synchronized to the rotation of the corn crop.