A significant part of the world cargo transportation is carried out by water transport. Therefore, the optimization of fleet management is one of the key means of reducing environmental pollution. In this article we consider an algorithm to optimize feeder fleet operations in order to minimize the sum of tour lengths of feeders and, therefore, to reduce water and air pollution. Based on the specificity and intensity of maritime operations within Mediterranean Sea and Basin, we consider the problem of optimization for the maritime cargo delivery as a multiple traveling salesman model and use an adapted ant colony algorithm to solve it. The algorithm, along with minimizing the route length, allows to minimize the level of harmful emissions into water and atmosphere. To increase the ant colony algorithm performance, we analyze values of its input parameters, whereupon they are adjusted so that the algorithm efficiently produces solutions very close to the optimal solution. Besides, an early stop parameter implementation is considered in order to increase the computation efficiency of the algorithm. In addition, a comparison of the algorithm with the early stop parameter and the algorithm with pre-defined number of iterations is performed. The gain may be up to a few percent, by which the route length is shortened, but it is a substantial reduction of expenses for maritime cargo delivery and harmful emissions.