An Ant Colony Optimization (ACO) approach for a dynamic Traveling Salesperson Problem (TSP) is studied in this paper. In the dynamic version of the TSP cities can be deleted or inserted over time. Specifically, we consider replacing a certain number of cities with new ones at different frequencies. The aim of the ACO algorithm is to provide a good solution quality averaged over time, i.e. the average taken of the best solution in each iteration is optimized. Several strategies for pheromone modification in reaction to changes of the problem instance are investigated. The strategies differ in their degree of locality with respect to the position of the inserted/deleted cities and whether they keep a modified elitist ant or not.