Human-elephant conflict situations continue to be a growing conservation issue. Conflict situations typically arise as the result of elephants traversing places with infrastructure where they become disruptive to human activities. Developing passive options for managing elephant movement and reducing human-elephant conflict situations is a conservation priority. In this paper, we report elephant responses to a synthetic honey bee alarm pheromone blend, and the relationship between strength and frequency of deterrence responses from elephants. We show that a blend of isoamyl acetate, 2-heptanol and 2-nonanol, volatile chemicals known to stimulate honey bee vigilance, defence and mass attack, produced stronger repellence than isoamyl acetate alone, or blended with 2-heptanol. Subadult elephants, irrespective of sex, were the demographic group most likely to show deterrence responses to the alarm blends. In some cases, entire herds of elephants were repelled from alarm pheromone treatments placed at waterholes, suggesting that the method has potential for manipulation of elephant movements on a landscape. Integration of synthetic honey bee alarm pheromones with other elephant behavioral management tools is discussed.

Field data collection, as described in the Wildlife Biology paper. 

Ranked deterrence responses of elephants to a synthetic honey bee alarm pheormone; demographic analysis (sex, age), and herd size.

Data analyzed using logistic regression analysis, generalized linear models.

Quoted from paper: we formulated isoamyl acetate, 2-heptanone, and 2-nonanol (20% V/V) with SPLAT^® (Specialized Pheromone and Lure Application Technology, ISCA Technologies, Riverside, California, USA) and deployed the three-component formulation at waterholes in the OWNR as previously described by Wright et al. (2018), to observe elephant reactions. Waterholes (1-3 per week depending on availability) with observation hides were selected for the work. They were large enough to permit deployment of at least a control and treatment at each location, and in some locations, two treatments simultaneously. The blends with the chemical cues were replaced at each exposure. Responses were video recorded by observers hidden within game-viewing hides, and analyzed to detect responses to the treatments. Elephants moving within 1m of the treatments were considered to be responders. In cases where other individuals were apparently alerted by the responder, the herd response was considered a single response, rather than the sum of individuals within the herd. Responses were categorized on an ordinal scale.