Multiple-input multiple-output (MIMO) communications based on orthogonal frequency division multiplexing (OFDM) are considered for improving the performance and bandwidth utilization of underwater acoustic systems in which cooperation is possible between distributed transmitters. The major challenge in such a framework - and the principal difference from the traditional case where multiple transmitters are co-located - is the fact that distributed transmitter-receiver pairs may experience significantly different Doppler distortion (e.g. two vehicles moving in different directions with respect to the receiver). The conventional approach of front-end resampling that corrects for a common Doppler scaling will then fail, rendering a post-FFT signal that is contaminated by transmitter-specific inter-carrier interference. To counteract this problem, we propose a front-end receiver structure that utilizes multiple resampling branches, each followed by FFT demodulation. As a result, a set of sufficient statistics are acquired, which are subsequently processed using custom-designed, linear or nonlinear detection schemes. Numerical results illustrate significant performance improvements as compared to the conventional, single-resampling schemes.