Despite the potential benefits of a maritime Internet of Things, a number of issues hinder its realization, including the need for wide area coverage and cost-effectiveness. Toward these needs, a model is first presented for uplink maritime radio communications that hierarchically employ Wi-Fi and cellular links for data transmission from marine user equipments (MUEs) on a ship to a cellular base station with distributed antennas (DAs) along a coastline. Then, the performances are evaluated given in terms of average data rate and outage probability in the hierarchical maritime radio networks. When evaluating the performances, heterogeneous channel characteristics of the hierarchical networks are taken into account in which Rayleigh fading is considered for the MUE-ship links, whereas shadowed-Rician fading is considered for the ship-DA links. Moreover, the effects due to various transmission errors that can be occurred at different communication layers are reflected in determining the quality of service (QoS) requirements of a ship. With the help of authors' analysis, an antenna selection algorithm is proposed under which the minimum size of a service cloud can be found to support the QoS requirements of MUEs with a ship. The numerical evaluations demonstrate the validity of the proposed algorithm.