With the increase of IoT devices, P2P-based IoT platforms have been attracting attention because of their capabilities of building and maintaining their networks autonomously in a decentralized way. In particular, Skip Graph, which has a low network rebuilding cost and allows range search, is suitable for the platform. However, when data observed at geographically close points have similar values (i.e. when data have strong spatial autocorrelation), existing types of Skip Graph degrade their search performances. In this paper, we propose a query transfer method that enables efficient search even for spatially autocorrelated data by adaptively using two-types of Skip Graph depending on the key-distance to the target key. Simulation results demonstrate that the proposed method can reduce the query transfer distance compared to the existing method even for spatially autocorrelated data.