In this paper, an efficient distributed approach for implementing the approximate message passing (AMP) algorithm, named distributed AMP (DAMP), is developed for compressed sensing (CS) recovery in sensor networks with the sparsity K unknown. In the proposed DAMP, distributed sensors do not have to use or know the entire global sensing matrix, and the burden of computation and storage for each sensor is reduced. To reduce communications among the sensors, a new data query algorithm, called global computation for AMP (GCAMP), is proposed. The proposed GCAMP based DAMP approach has exactly the same recovery solution as the centralized AMP algorithm, which is proved theoretically in the paper. The performance of the DAMP approach is evaluated in terms of the communication cost saved by using GCAMP. For comparison purpose, thresholding algorithm (TA), a well known distributed Top-K algorithm, is modified so that it also leads to the same recovery solution as the centralized AMP. Numerical results demonstrate that the GCAMP based DAMP outperforms the Modified TA based DAMP, and reduces the communication cost significantly.