Wildlife monitoring provides the observational basis for studying animal behaviour, habitat use and ecosystem change. However, existing methods are often difficult to scale, depend on high volumes of species-specific data, or express subpar adaptive capabilities. Aerial robots offer high-resolution, scalable observations, though a key challenge remains ensuring data collection does not unduly disturb the animals. This thesis presents a disturbance-aware reinforcement learning framework for autonomous wildlife monitoring with aerial robots. The framework combines synthetic animal movement models, empirical animal trajectories, spatial resource maps, a geometry-based disturbance model, aerial sensing agents, and reward shaping that explicitly balances monitoring quality against induced animal disturbance. Animal disturbance is modelled as a continuous field that drives behavioural states including calm, avoidance, and fleeing. Policies were trained using Deep Q-Network, Proximal Policy Optimization, and Soft Actor-Critic across four synthetic movement models and evaluated against empirical trajectories from Pigeons, lapwings, and Jackals. Effective policy operating zones are highlighted by the knee of the Pareto frontier. Learned policies consistently outperformed tuned rule-based controllers. More importantly, policy performance obtained when training on synthetic movements is transferred to evaluation against unseen empirical GPS trajectories. While Soft Actor-Critic excelled in single-agent scenarios, Proximal Policy Optimization exhibited more stable learning dynamics and was selected for sensitivity and robustness experimentation. Small multi-agent teams enhanced performance, whereas larger teams introduced coordination challenges without clear synergistic benefits. The learned policies demonstrated robustness to mild wind but degraded in strong, gusty conditions. The proposed framework offers a way to gain insights into control strategies before field deployment, where extensive experimentation may be costly, difficult, or disruptive to animals.