Autonomous microgrids (MGs) are being installed in large remote areas to supply power where access to the utility grid is unavailable or infeasible. The power generation of such standalone MGs is largely dominated by renewable based energy sources where overloading or power deficiencies can be common due to the high intermittency and uncertainty in both load and power generation. Load-shedding is the most common mechanism to alleviate these problems to prevent system instability. To minimize load-shedding, most MGs are equipped with local battery energy storage (BES) systems to provide additional support. Furthermore, in the event of severe overloading or when BES capacity is insufficient to alleviate the overload, neighboring MGs can be provisionally coupled to provide mutual support to each other which is a more effective, economic and reliable approach. Such a coupling is preferred to be via power electronic converters to enhance the autonomy of the MGs. This paper proposes a two-stage, coordinated power sharing strategy among BESs and coupled MGs for overload management in autonomous MGs, through dynamic frequency control. Both local BES and the neighboring MGs can work in conjunction or individually to supply the required overload power demand. For this, BES' state of charge should be above a minimum level and extra power generation capacity needs to be available in the neighboring MGs. A predefined framework with appropriate constraints and conditions, under which the power exchange will take place, are defined and formulated. The proposed mechanism is a decentralized approach, operating based on local frequency and state of charge measurements, and without any data communication amongst the MGs. The dynamic performance of such a network, is evaluated through extensive simulation studies in PSIMR and verifies that the proposed strategy can successfully alleviate the overloading situation in the MGs through proper frequency regulation.