Better tired than lost: turtle ant trail networks favor coherence over short edges

Abstract

1To create and maintain a backbone routing network is a basic challenge in many engineered and 2 biological systems [1][2][3], from wireless sensor networks and robot swarms to neural circuits 3 and blood circulation. Optimal routing in such networks often seeks to minimize transport 4 delay [4][5][6][7][8], but routing decisions may be influenced by variability in the terrain [9][10][11][12]. Here 5 we find that turtle ants build trail networks that emphasize coherence, keeping the ants together 6 on the trail in a heterogeneous environment, rather than minimizing the distance travelled. The 7 routing backbone of turtle ants (Cephalotes goniodontus), an arboreal species that forages in the 8 1 tree canopy of tropical forests, connects many nests [13][14][15], using trail pheromone that the ants 9 put down continuously, not just on the way back from a food source. Unlike species that forage 10 on the ground, arboreal ants are constrained to travel within the vegetation network of branches 11 and vines. We compared observed turtle ant trails with random, hypothetical trails in the same 12 surrounding vegetation. Strikingly, the trails do not minimize distance travelled, but instead 13 minimize the total number of nodes in the backbone, and favor nodes with 3d configurations 14 that are easily reinforced with pheromone. Thus, rather than forming the shortest paths, turtle 15 ants take advantage of natural variation in the environment to build coherent trails. This ensures 16 that the nests and food sources stay connected, at the expense of longer travel time. This design 17 principle may be beneficial in applications where distributed agents, such as swarms of robots, 18 must coordinate using a communication backbone in complex environments to collectively 19 solve a task, such as building a structure, searching for resources, or surveying terrain [16]. 20 Introduction 21The goal of a backbone routing network is to ensure that there is a path for any two 22 devices on the network to communicate. In real-world systems, however, physical variation in the 23 environment can affect the accuracy and rate of communication [9][10][11][12][17][18][19]. Understanding the 24 physical structure of the environment may improve the design of routing algorithms [16,[20][21][22][23][24], for 25 example, by reducing the search space of possible routing paths and steering network construction 26 away from parts of the terrain that are difficult to reach. 27The 14,000 species of ants have evolved diverse distributed routing and search algorithms to 28 search for, obtain, and distribute resources [25] in the diverse environments that they inhabit [26-29 29]. For example, species such as Formica and Argentine ants, which forage and build trails in 30 a 2d-plane, have minimal constraints on trail geometry [30][31][32], and can minimize the distance 31 travelled by forming trails with branch points that approximate Steiner trees [27, 33, 34], an NP-32 complete generalization of the minimal spanning tree concept. By contrast, many specie...

Goals and Limitations of Modeling Collective Behavior in Biological Systems

“…This shows that this model is plausible because it is consistent with field observations. Next, we asked what objectives are optimized by the algorithm that the ants use [45]. Unlike species that forage on the ground and can go anywhere in a 2D plane, arboreal ants never leave the tree canopy, and so the configuration of their trail networks is constrained by the vegetation.…”

Arboreal Ants Suggest Surprising Algorithms for the Shortest Path Problem and its Variants

“…Chandrasekhar, Marshall, Austin, Navlakha, and Gordon [Cha+19] noticed in field observations that turtle ants favour paths which minimize the probability of getting lost. Our result in Theorem 1 suggests a plausible way in which ants converge to such paths.…”

“…More recently, Chandrasekhar, Marshall, Austin, Navlakha, and Gordon [Cha+19] used data from field observations of the trail networks of turtle ants to test what objective functions these trails may be optimizing. They identified each node or junction in the vegetation where an ant had a choice among more than one edge in the direction it was traveling.…”