Atta cephalotes foragers transfer leaf fragments to "carrier" ants at junctions of new branch trails and the established trail. A more pronounced specialization into "harvesters" and "carriers" is exhibited by A. sexdens rubropilosa in which smaller ants harvesting in the tops of tall trees drop material to the ground where larger workers collect it. COLUMNSOF LEAF-CUTTER ANTS (Myrmeciinae: Attini), carrying leaf and flower fragments along trails to the nest, are a common sight in the neotropics. Workers cut fragments from a variety of trees, shrubs, and crop plants, and culture a fungus on them, the specialized hyphae of which serve as their sole food source (Weber 1966, Hubbell and Rockwood 1980). As in many other eusocial insects, including bees, termites, and especially ants, a trail pheromone helps to guide the workers between nest and food source (Wilson 1971, Moser and Blum 1963). W e describe here a novel two-stage relay process in foraging leaf-cutter ants, involving a transfer of material from one ant to another. This relay process, serves to increase the speed of new food-source utilization, and to increase the efficiency of exploitation of established food sources. Foraging efficiency is a significant problem for leaf-cutters which must often travel long distances, sometimes over 100 meters, to harvest leaves of tree species best able to support fungus growth (Cherrett 1968, Hubbell and Rockwood 1980). RELAY COOPERATION DURING NEW TRAIL FORMATION We studied the foraging of Atta cephalotes L. in a tropical dry forest in Santa Rosa National Park, Guanacaste Province, Costa Rica (lat. 10' 50", long. 85' 37"). Trails from the nest lead to particular trees or shrubs or to small, oval, ground-foraging areas where ants forage solitarily. The ants may later establish a trail leading to a particularly acceptable plant within the oval. The discoverer ant lays a pheromone trail connecting the new source to the main trail, and within an hour a branch is established. Laden A. cephalotes normally carry their loads all the way into the nest: The first ants to visit a new source, however, carry their loads only as far as the main trail, where they drop them, or antennate other ants and transfer their load to them. These "carriers" then carry the fragments to the nest. To learn more about branch trail formation and about the transfer of information regarding the locations of new, high-quality food sources to other workers, we performed experiments in October and November 1975. In these experiments, bread crumbs or leaves of Bursera ~imaruba, a preferred tree species, were placed in an oval ground-foraging area 1 meter from an established trail. Figure 1 shows that in one such experiment a trail to the bread crumbs (black circle) developed as a branch off the main trail (heavy line), and figure 1A traces the tortuous paths of the first two crumb-bearing ants back to the main trail. Both touched their gasters repeatedly to the ground, a behavior which indicates pheromone deposition (Moser and Blum 1963). Typically such individual trails gradually consolidate as more ants visit the food and lay pheromone Figure 1B shows the paths of three ants returning to the main trail 40 minutes after discwering the bread. The trail has become narrow and points directly to the nest (not shown in the figure). A decline in the percentage of ants dropping or transferring their loads upon reaching the main trail was always found to accompany trail consolidation. Table 1 shows how, in another experiment, this petcentage declined over a 50-minute period subsequent to the discovery of bread. In this experiment, 12 percent of the drops and transfers occurred along the way between the bread and the main trail, 83 percent occurred along the first meter of the main trail, 5 percent occurred along the second meter of the main trail, and none in the last two meters to the nest. This typical localization of dropping or transferFIGURE 1. A. Paths back to the main trail (heavy line) of the first two ants to discover the new food source (black circle). The x's mark the sites where the ants dropped or transferred their loads. B. Paths of three recruits 40 minutes after discovery of the food. These ants continued straight to the nest (to the right) with their loads. ring the bread led us to hypothesize that some abruptly encountered feature of the main trail, such as familiar landmarks, trail pheromone, or heavy ant traffic, stimulates these behaviors. As the new branch trail becomes more like the main trail, dropping and transferring is less often stimulated. Although main trails sometimes appear as distinct grooves worn in the ground, this characteristic was not a distinguishing feature in the experimental site, so we began exploring the other possibilities. W e determined in another experiment in M:iy 1976 that sudden encountering of trail pheromone by itself could produce dropping or transferring behavior. Two 46 x 61 cm cardboard rectangles were nailed to the ground for 30 hours, the "control cardboard" at a site where ants were not foraging, and the "trail cardboard" on an active trail, the long axis parallel to the trail. Ants marked the trail cardboard. The next evening each cardboard was laid between a pile of bread crumbs in an oval ground-foraging area and a main trail one meter away. Care was taken that the two sites where the cardboards were laid were equivForaging in Leaf-Cutter Ants 211 alent in terms of ground cover, slope, and traffic of the main trail; at neither site had the ants previously been exposed to bread. Each cardboard was oriented wlth its long axis parallel to the main trail, so that the pheromone trail on the trail cardboard lay perpendicular to the most direct path from crumbs to main trail. TABLE 1. Percentage o f ants dropping or transferring loads i n each o f five 10-minlcte periods after discovery o f the food.