Context: Unmanned aerial vehicles or drones are powerful tools for wildlife research. Identifying the impacts of these systems on target species during operations is essential to reduce risks of disturbance to wildlife, to minimise bias in behavioural data, and to establish better practices for their use.

Aims: We evaluated the responses of captive Antillean manatees to the overhead flight of a small aerial drone.

Methods: We used aerial and ground videos to compare manatee activity budgets and respiration rates in three 15-min sampling periods: ‘before’, ‘during’ and ‘after’ flights with a DJI Phantom 3 Advanced. The drone was hovered stationary for 3 min at five altitudes (100 m, 40 m, 20 m, 10 m, 5 m) to determine whether manatees display behavioural responses compared with the control period, and whether they respond more at lower altitudes. Only one flight was performed per manatee group to avoid bias owing to habituation to the drone.

Key results: Manatees responded to drone flights by (1) increasing their activity levels during and after flights, therefore signalling after effects; (2) decreasing their respiration rate during flights; and (3) displaying behavioural reactions including grouping, tail-kicking, fleeing from their original position and moving under submerged structures. From the 11 individuals displaying behavioral reactions, 9 reacted in the first ~2 min of flight, preventing assessments of altitude effects and suggesting manatees responded to the drone sound at take-off.

Conclusions: Behavioural changes of responding manatees were similar to previous reports of disturbance responses to boats and drones in this species. Our use of a control period showed shifts in respiration rates and activity budgets that persisted after flights. Several manatees reacted to the drone from the time of take-off and first minutes of flight, indicating that the sound of the electric rotors could be a strong negative stimulus to manatee and highlighting the importance of establishing safe distances for take-off.

Implications: Future studies should consider that drones could elicit conspicuous and inconspicuous responses in manatees. Our results emphasise the need for control data on animal behaviour to better assess the impact of drones on wildlife and to design non-invasive protocols.

背景：无人驾驶飞行器或无人机是野生动物研究的有力工具。确定这些系统在运营期间对目标物种的影响对于减少对野生动物的干扰风险、最大程度地减少行为数据中的偏差以及建立更好的使用实践至关重要。

目标：我们评估了圈养的安的列斯海牛对小型空中无人机的头顶飞行的反应。

方法：我们使用空中和地面视频来比较三个 15 分钟采样周期内的海牛活动预算和呼吸率：使用 DJI Phantom 3 Advanced 的“飞行前”、“飞行中”和“飞行后”。无人机在五个高度（100 m、40 m、20 m、10 m、5 m）静止悬停3分钟，以确定海牛与控制期相比是否表现出行为反应，以及它们是否在较低的高度反应更多。每个海牛群只进行一次飞行，以避免由于对无人机的习惯而产生偏差。

主要结果：海牛对无人机飞行的反应是 (1) 在飞行期间和飞行后增加其活动水平，从而发出后效信号；(2) 在飞行过程中降低呼吸频率；(3) 表现出行为反应，包括分组、踢尾、逃离原位和在水下结构下移动。从表现出行为反应的 11 个人中，有 9 个人在飞行的前 2 分钟内做出了反应，从而阻止了对高度影响的评估，并表明海牛在起飞时对无人机的声音做出了反应。

结论：响应海牛的行为变化与之前关于该物种对船只和无人机的干扰反应的报告相似。我们对控制期的使用显示了飞行后呼吸率和活动预算的变化。几只海牛从起飞和飞行的第一分钟就对无人机做出了反应，表明电动旋翼的声音可能对海牛产生强烈的负面刺激，凸显了为起飞建立安全距离的重要性。

启示：未来的研究应该考虑到无人机可能会在海牛身上引起显眼和不显眼的反应。我们的结果强调需要控制动物行为的数据，以更好地评估无人机对野生动物的影响并设计非侵入性协议。