Engineering neo-biomimetics, i.e. imitation models based on body structures and behavior of living organisms, relied upon to solve complex problems, have been studied in various fields. In distributed networks, such as ad-hoc networks and wireless sensor networks, the behavior of a variety of true slime molds which are capable of constructing multipath flow networks based on the amount of body, has been studied. Ad hoc networks only consist of mobile terminals (nodes) that can relay packets along an established route. However, link relations and the available bandwidth of the nodes change dynamically due to the mobility of nodes. In addition, the speed of communication between nodes also varies due to node positions and their communication-related quality. Thus, practical use of ad-hoc networks still remains an issue, because it is difficult to establish stable routes under such environments. This study aims to propose an adaptive load balancing routing technique that adaptively diversifies the transmission paths based on the available bandwidth, residual battery life, and the data transmission volume, by applying a mathematical model of slime mold routing, known as the physarum solver. We confirm the effectiveness of its adaptive behavior in dynamic environments using computer simulations.