Alarm calls of sagebrush converge when herbivory is high.

Herbivory is a major threat to virtually all plants, so adaptations to avoid herbivory will generally be selected. One potential adaptation is the ability to 'listen in' on the volatile cues emitted by plants that are experiencing herbivory and to then respond by ramping up defences. The nature of these volatile cues is poorly understood. Sagebrush (Artemisia tridentata) plants that were exposed to cues of experimentally damaged neighbours experienced less herbivory; this induction was most effective if emitter and receiver plants had similar volatile emission profiles, termed chemotypes. Previously, we observed that sagebrush populations that were in locations with high herbivory exhibited little diversity of volatiles compared to populations with low herbivory. Several hypotheses could produce this correlation. High risk of herbivory could have selected for individuals that converged on a common 'alarm cue' that all individuals would respond to. In this case, individuals of locally rare chemotypes that were less able to eavesdrop would experience more damage than common chemotypes when herbivores were abundant. Alternatively, low chemotypic diversity could allow higher levels of damage to plants. In this case, rare chemotypes would experience less damage than common chemotypes. We examined the chemotypes of sagebrush individuals from multiple sites and found that rare chemotypes experienced more damage than common chemotypes when herbivores were abundant. This pattern was seen among sites and among years with different densities of herbivores. This result is consistent with the hypothesis that herbivory selects for individuals that are effective communicators and shapes the communication system.

Risk of herbivory negatively correlates with the diversity of volatile emissions involved in plant communication.

Plant-to-plant volatile-mediated communication and subsequent induced resistance to insect herbivores is common. Less clear is the adaptive significance of these interactions; what selective mechanisms favour plant communication and what conditions allow individuals to benefit by both emitting and responding to cues? We explored the predictions of two non-exclusive hypotheses to explain why plants might emit cues, the kin selection hypothesis (KSH) and the mutual benefit hypothesis (MBH). We examined 15 populations of sagebrush that experience a range of naturally occurring herbivory along a 300 km latitudinal transect. As predicted by the KSH, we found several uncommon chemotypes with some chemotypes occurring only within a single population. Consistent with the MBH, chemotypic diversity was negatively correlated with herbivore pressure; sites with higher levels of herbivory were associated with a few common cues broadly recognized by most individuals. These cues varied among different populations. Our results are similar to those reported for anti-predator signalling in vertebrates.