Highly maneuverable Unmanned Aerial Vehicles (UAVs) are being increasingly deployed to replace humans in complex environments, and optimizing the task allocation among UAVs can enhance task efficiency and quality. This paper presents a distributed task allocation method for multi-UAV systems, where each task has a Minimum Task Requirement (MTR) regarding the number of UAVs needed, while the environment dynamics arise from changes in task quantity and requirement. Firstly, a dynamic task allocation model based on dynamic awareness is established to adapt to environmental variations. Subsequently, the task allocation model is structured as a Coalition Formation Game (CFG) model, and a Bilateral Altruism Preference (BAP) order is proposed. The integrated decision-making framework is called CFG-BAP, which addresses the dual challenges of member selection within coalitions and resource competition between coalitions by considering both group and individual benefits, allowing sub-coalitions whose MTR constraint is not fulfilled to loot members from other sub-coalitions. Finally, given the coalition utility function, preference order, and coalition formation principles, it is demonstrated that CFG-BAP has a Nash stable partition. The efficacy and superiority of the proposed scheme are validated via experimental simulations.