Observing the group movements and group behaviour of animals in the wild is a question that has been long inspiring our team. We aimed to investigate the collective movements and social relationships of nearly 300 Przewalski's horses living in the reserve of the Hortobágy National Park (Hungary). This population is especially interesting, since the horses here live in a complex, multilevel herd, which is a rare social structure among animals. However, recording the movements of 300 individuals at the same time is a big challenge. For a decade, we have been working with different aerial filming techniques, drones and helium-filled blimps, and tried to find solutions to track movements of individuals on aerial videos recorded by these flying devices. Finally, our recent study yielded unexpected results. We found that by collecting high-resolution data, even a few minutes of collective movements can provide enough information to detect the social structure of a population and even draw conclusions about the past and future dynamics of the group. For a short summary watch our video abstract: 

Multilevel societies are rare social structures, where larger groups are composed of smaller sub-groups. They are mainly found in primates, but also occur in cetaceans, elephants and equids. Individuals in these societies form stable groups (e.g. one-male multi-female harems or a group of related females led by a matriarch), and then these groups form larger, looser communities. Yet, the majority of human social systems is structured in a multilevel way.

Przewalski’s horses, the last wild horses, are among the few species that may display multilevel social structure. They live in year-round stable harem groups, which in the reserve of Hortobágy National Park (Hungary), come together and form a large herd. Przewalski’s horses have been living in Hortobágy since 1997. Interestingly, in the first years after founding the population, harems lived in their own home ranges and rarely interacted with each other, but for a decade now, they form a massive herd. We aimed to study the behaviour of this complex herd made up of harems. Since the establishment of the reserve, the wild horses have been individually recognized by the park staff who regularly collected data on population changes. Thanks to this population monitoring, we know the parentage of the animals, also confirmed with genetic sampling, as well as their place in the social system, since it is regularly recorded which individual belongs to which harem.

In our study, we observed the collective movements of the Przewalski’s horse herd at Hortobágy, by taking aerial videos with drones while the herd was moving around the reserve. One drone recorded the movements of the herd, and another drone ensured individual identification. Based on the drone footage, we determined the movement routes of all individuals in the herd, 278 horses in total, with high spatial and temporal resolution. Individuals in a group coordinate their movements and align with each other. Thus, by detecting the fine interactions between the individuals during movements we can understand the group's social network. Additionally, we combined the short-term movement observations of a few minutes with the long-term population monitoring data of the national park going back two decades.

Based on the data of several minutes long movements, we built proximity networks of the herd, and we found that bonds in this network are related to kinship and familiarity between the animals. For example, female horses (mares) were closer to each other in the network if they have been harem mates for a longer time. On the other hand, kinship may play a significant role in the organization of harems into herds, since harems of sibling stallions (the single breeding male in a harem) were closer to each other in the network than harems of unrelated stallions. Between the closer harems, at the same time, the dispersal of mares was greater, which also contributes to the relations between harems through familiarity. An interesting feature of the detected network is that older and larger harems, which typically belonged to older and more experienced stallions, occupied more central locations. A possible explanation is that harem stallions form alliances to protect their harems more effectively against bachelor males.

One of the surprising outcomes of the study is that we could also infer future group dynamics by observing current movement. We showed that the mares that lived in different harems at the time of the aerial observations but became harem mates in the two years after the observations, were already moving in more similar routes than other mares. Thus, it is possible to identify which mares will leave their harem in the next two years and which harem they will transfer to.

The decades-long monitoring in the Hortobágy National Park’s reserve provided an exceptional opportunity to explore the social network of an entire population and its dynamics. Not only did we learn new, previously unknown details about the social life of Przewalski's horses, but we highlighted that drone observations, which can even be applied in wild populations, can provide very detailed information. Hopefully, our results can contribute to the conservation of this endangered species.