The role of age as a cause of social behaviour in male African savannah elephants (Loxodonta africana).

Social interactions can affect every aspect of an animal’s life. Male savannah elephants (Loxodonta africana) spend most of their time with other bulls, so their companion choices are likely to have a significant effect on their long-term fitness. Compared to females, male elephants have received limited social behaviour research, despite being more involved in negative human-elephant interactions. Furthermore, studies of bull social structure do not display strong agreement in their results, which may be due to genuine inter-population differences or the use of statistical methods since proven unreliable.

In this thesis, I reevaluate the social structure of male elephants and question whether males form groups to obtain social information from more experienced bulls or to gain predator protection. To do this, I first develop an adaptation to social association estimation methods, then use a combination of social network analysis, threatening playback experiments, and causal modelling to investigate the effect of age on male social behaviour. Contrary to my hypotheses, I found no evidence of males grouping for the specific purpose of gaining social learning opportunities, with no effect of male age on social position, or for males protecting one another when the group is in danger.

With no previous studies investigating male group threat responses, my experiments provide new evidence against predator protection as a reason for male-male association. However, the formation of weak connections with many social partners may still facilitate rapid information spread, even if social learning is not the primary purpose of group formation, because each individual could pass the information to many others. In the future, we may be able to use this to promote elephant avoidance of human-dominated areas. Since I also show that male social structures are similar across populations, conservation practices based on social learning may be similarly effective in different places.