Predicting the risk of non-target damage to a close relative of a target weed using sequential no-choice tests, paired-choice tests and olfactory discrimination experiments

We investigated host-plant utilisation by the candidate biocontrol agent Paradibolia coerulea (Coleoptera: Chrysomelidae) on the target plant Spathodea campanulata Beauv. (Bignoniaceae) and a closely related non-target plant, Kigelia africana (Lam.) Benth. (Bignoniaceae). Paired-choice and sequential no-choice experiments were performed and coupled with olfactory discrimination experiments to test the insects' responses to volatiles from both plant species as well as to cues from conspecific beetles. Although K. africana was utilised by P. coerulea, S. campanulata was preferred for both adult feeding and oviposition. Interestingly, whereas females were attracted to olfactory cues emitted by S. campanulata, males demonstrated no such olfactory discrimination. Females were also attracted to cues deposited by males, and males were deterred by cues from other males, but neither sex responded to female olfactory cues. Very few eggs were recorded on K. africana and none of the larvae that hatched on K. africana survived the first instar. Both S. campanulata and K. africana are suitable for adult feeding, but persistent utilisation of K. africana in the field is unlikely because larval development is only possible on S. campanulata and because the adult females are strongly attracted to volatiles emitted by the target plant. Nevertheless, if P. coerulea is released as a biocontrol agent, spill-over adult feeding could potentially occur on K. africana growing sympatrically with S. campanulata. Because P. coerulea cannot complete its development on K. africana, non-target damage will only occur where the target plant is present, with an intensity dependent on densities of adult beetles locally.