Understanding animal behaviour is essential for predicting and mitigating the impacts of natural and human-induced environmental changes on animal populations and the ecosystems they inhabit.Unmanned Aerial Vehicles (UAV) or drone-based aerial monitoring of wildlife has gained traction over the past decade, however, limited training data of wild animals in ecologically relevant scenarios has hindered the development of automated computer vision solutions for long-term tracking of animal movement and behavior. Here, we introduce the first large-scale UAV dataset to tackle the problem of multi-object tracking (MOT) and re-identification (Re-ID) in wild animals. The data is derived from an ongoing study on the mating behaviour (lekking) of antelopes (blackbuck) conducted with three simultaneously flying UAVs. Our dataset includes over 1.2 million annotations of 680 tracks across 12 video clips of 5.4K resolution, with videos averaging 66 seconds in length and featuring 30 to 130 individuals per video. Additionally, we introduce a novel task of animal re-identification (730 individuals) using videos from two UAVs. Our dataset aims to motivate the development of scalable methods to track the movement of wild animal groups over extended periods and large areas by integrating data from multiple sensors. We provide the baseline performance of two detectors and benchmark several state-of-the-art tracking methods on the dataset. Collected in collaboration with biologists, our dataset accurately reflects the challenges of tracking wild animals in socially, ecologically, and evolutionarily relevant contexts, emphasizing the importance of solving the MOT and Re-ID problems for research, conservation, and wildlife management.