A growing body of research documents the importance of plant genetic effects on arthropod community structure. However, the mechanisms underlying these effects are often unclear. Additionally, plant genetic effects have largely been quantified in common gardens, thus inflating the estimates of their importance by minimizing levels of natural variation. Using \*Valeriana edulis\*, a dioecious plant with genetically based sex determination, we conducted surveys and experiments on wild-grown individuals to document field patterns of arthropod association between the sexes and the mechanisms underlying these plant genetic effects. Three years of surveys revealed strong and consistent sex-biased arthropod association in wild-grown plants: female plants supported 4-fold, 1.5-fold, and 4-fold higher densities of aphids, aphid predators, and aphid-tending ants, respectively, compared to males. There was mixed evidence that the female bias for aphids was due to higher plant quality, while we found no difference between plant sexes in aphid preference or the top-down effects of predators and tending ants. Female bias for ants was due to both the greater attractiveness of female plants (direct effect mediated by floral nectar) and an independent, weaker effect of higher aphid abundance on females (density-mediated indirect effect). Conversely, the female bias for predators was driven solely by the greater attractiveness of female plants. We did not find interaction modification, i.e., ant–aphid and predator–aphid interactions were equivalent between plant sexes. Plant sex explained 0.24%, 2.28%, and 4.42% of the variance in aphids, predators, and ants, respectively, values comparable to but slightly weaker than those previously reported from common-garden studies. In contrast to the prediction of diminished plant genetic effects with increasing trophic level, we show how weak indirect effects on predators and parasitoids (via herbivores) can be complemented by strong direct effects via common plant traits (floral resources). In summary, we document direct and indirect effects of genetically based sex on a multi-trophic arthropod community that were expressed in wild-grown plants across multiple years.