SEARCH

SEARCH BY CITATION

This paper reports significant effects of a co-occurring plant species (Lythrum salicaria, Lythraceae) on the reproductive success of the perennial herb Filipendula ulmaria (Rosaceae). We studied 15 Filipendula populations in the Skeppsvik Archipelago; seven of which were monospecific and eight mixed with Lythrum. All the Filipendula populations studied harbored the chrysomelid beetle Galerucella tenella, and in 2005 seed set was strongly negatively correlated with the percentage leaf area consumed. Moreover, data from 2004 showed that 25–100% of the G. tenella larvae were parasitized by the hymenopteran parasitoid Asecodes mento, and we found a strong cascading top-down effect of parasitism in 2004 on Filipendula seed set in 2005. In 2004, parasitism (at the population level) was negatively correlated with percentage leaf area consumed and positively correlated with seed set in 2005. The parasitoid Asecodes also parasitized G. calmariensis, which is monophagous on Lythrum. Mixed populations of Filipendula and Lythrum supported higher densities of their shared ‘bodyguard’Asecodes. Further, Y-tube bioassays showed that floriferous Filipendula attracted more than twice as many gravid Asecodes females as floriferous Lythrum. Taken together, these findings suggest that coexistence of the two plants results in ‘associational resistance’ for Filipendula and ‘associational susceptibility’ for Lythrum. This scenario was supported for Filipendula since, for this species, we found lower leaf consumption followed by higher seed production in mixed than in monospecific populations. Considered together, our results show that bodyguards may increase the reproductive fitness of a perennial herb, and that the strength of the cascading ‘bodyguard’ effect can be strongly influenced by co-occurring plants through ‘apparent competition’. This is the first paper to demonstrate that, in the wild, plant species may use odors to compete for ‘bodyguards’, thereby causing asymmetrical ‘apparent competition’ between the herbivores involved. Our data emphasize the need to consider community factors in studies of trophic interactions.