Meeting Abstract

P1.111  Tuesday, Jan. 4  An evolutionary simulation of predator-prey interactions in early fish-like vertebrates DAHAL, Bidur K*; PHAM, Tong T; LIN, Khine; LIEW, Chun W; ROOT, Robert G; LONG JR, John H; Lafayette College; Lafayette College; Lafayette College; Lafayette College; Lafayette College; Vassar College liew@cs.lafayette.edu

Based on evidence from living fish species, we predict that predation and avoidance contributed significantly to the evolution of physical traits in early vertebrates. Within this ecological context, we test the hypothesis that the axial skeleton of fish evolved as an adaptation for enhanced locomotion, prey capture, and predator avoidance. To test this hypothesis, we combine a physics-based digital simulation of self-propelled and autonomous fish with a genetic algorithm. The stiffness and length of the tail is coded as a suite of evolvable characters in hypothetical vertebrates based on the Tunicate Tadpole larvae. A simulation starts with a single pair of predator-prey, swimming independently in two dimensions. Each fish has sensors capable of detecting the other’s presence. Upon detection, the predator starts swimming toward the prey, which takes flight. This pattern continues until the predator catches, or loses sight of, the prey. We carry out the experiments in phases of alternating optimizations: each phase has either the predator or the prey evolve while the other remains unchanged. The fitness function is different for the predator and prey, reflecting different abilities for predation and avoidance. This work was supported by NSF DBI-0442269.