Summary1. Arthropod communities are structured by complex interactions between bottom-up (resource-based) and top-down (natural enemy-based) forces. Their relative importance in shaping arthropod communities, however, continues to be under debate. Bottom-up and top-down forces can be affected by intraspecific plant variation, for example by differences in concentrations of secondary metabolites that affect herbivore abundance through plant quality (bottomup) or attract natural enemies of these herbivores (top-down). 2. Our objective was to investigate whether herbivore abundance is more strongly affected by plant-mediated bottom-up or top-down forces. 3. We used a model system of four cultivars of Brassica oleracea that show a high degree of variation in several plant traits, resistance to herbivores and attraction of natural enemies. During two field seasons, we recorded the abundance of several herbivorous and carnivorous insect species. To assess the relative importance of bottom-up and top-down forces, we quantified chemical and morphological traits of the cultivars (bottom-up) and assessed parasitization of herbivores and predator oviposition on plants inoculated with a controlled number of herbivores (top-down). 4. We show that intraspecific variation in plant chemistry and morphology consistently affects the abundance of insect herbivores and their natural enemies, resulting in cascading effects on tritrophic interactions in the associated insect community. Foliar profiles of glucosinolates and leaf toughness appeared most important for these effects. Brassica oleracea cultivars that harboured the largest numbers of herbivores also harboured the largest numbers of natural enemies. Differences in the fraction of herbivores parasitized and in predator oviposition on plants inoculated with a controlled number of herbivores could not explain the differences in natural abundance of herbivores. 5. Although abundance of herbivores is most likely influenced by a combination of bottom-up and top-down forces, it appears that in the tritrophic system investigated, bottom-up forces (plant chemistry and morphology) were more important for herbivore abundance than plantmediated top-down forces (attraction and arrestment of natural enemies).

Euseius gallicus is a new phytoseiid species recently described from southern France that has shown potential as a biocontrol agent for thrips and whitefly in rose, if pollen is supplied as an additional food source. To investigate if the use of E. gallicus provides improved thrips and whitefly control, we conducted laboratory experiments examining the biology of E. gallicus and a semi-field experiment comparing the biological control efficiency of E. gallicus with that of Amblydromalus limonicus and Amblyseius swirskii, two phytoseiid species commonly used for biological control of thrips and whitefly in roses. Euseius gallicus had high oviposition rates on Typha latifolia pollen (3.9 eggs/day) and on young whitefly eggs (3.6 eggs/day). Oviposition on first instar Frankliniella occidentalis larvae was lower (1.5 eggs/day). Euseius gallicus predated 2.6 first instar thrips larvae per day; however, predation of thrips larvae was nearly zero when T. latifolia pollen was offered as a supplementary food source. Euseius gallicus females did not enter diapause under short-day conditions, and juvenile development was completed at 13 . When released in combination with T. latifolia pollen on roses in the semi-field trial, E. gallicus developed the largest population of the predatory mites tested; however, despite reducing the whitefly population, it had no control effect on the thrips population.