Distributed and collaborative beamforming (DCBF) has been attracting increased interest due to its desirable scalability and enhanced robustness, whereas slow convergence would hinder practical applications of the conventional DCBF algorithms for massive arrays. In this work, we consider the distributed reduced-rank beam coordination over an array network, equipped with a collection of potentially massive arrays interconnected via a certain topology. We develop a diffusion reduced-rank beam coordination algorithm along with an inherently adaptive combination scheme based on the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">combination matrix</i> for beam coordination, leveraging the joint iterative alternating optimization methodology. We propose two efficient adaptive implementations with parametric matrix combination schemes, which could further enhance the robustness of the distributed reduced-rank beam coordination algorithm against the spatial variations of the signal and noise powers. The proposed distributed reduced-rank beam coordination algorithm could not only remarkably speed the initial convergence in comparison with its full-rank counterparts under the small-sample conditions, but also achieve comparable steady-state performance with sufficiently large sample size. The tracking capability could also be enhanced with the proposed algorithm as a by-product. Illustrative simulations validate the efficiency of the proposed reduced-rank beam coordination algorithm.