Leaf-cutting ants as road engineers: the width of trails at branching points in Atta cephalotes

We used a simple engineering principle, which suggests that the width of a road needed for a smooth traffic flow is proportional to the peak traffic volume ("engineering hypothesis"), to analyze the adaptive significance of trail width at branching points in the leaf-cutting ant Atta cephalotes. Since the flow of outgoing ants splits at trail bifurcations and merges when ants return to the nest through the same paths, the sum of branch widths should equal the width of the trail section upstream of the bifurcation. We measured the width of branches and their preceding trail section and also performed field measurements and manipulations to analyze ant flow, number of collisions, and ant speed in different trail sectors. Contrary to the prediction of the "engineering hypothesis", the sum of branch widths was larger than the width of the trail immediately before the bifurcation. Our data contradict the "trail addition hypothesis" and support the "border effect hypothesis" to explain this pattern. First, the width of the widest branch was smaller than the width of the trail upstream of the bifurcation, an unexpected result if one branch is merely the continuation of the basal trail. Second, ants collided with obstacles more often in the margin than in the central portion of the trail, relocated ants from central to margin trail sectors reduced their speed, and ant flow was higher in the central sections of the trail. Since the delaying effect of trail margins increases as the trail width decreases, ants may build branches wider than expected to reduce the border effect. The delaying effect of trail margins should be included in the analysis of costs and benefits to fully understand the adaptive value of the design of ant trail networks.