The effect of ground cover upon the communities of beneficial arthropods established in the canopy of lemon trees was investigated, by comparing three ground-cover management treatments applied: RV, resident vegetation; S, sowed selected species; and BS, bare soil by controlling weeds with herbicide. Over two consecutive years, arthropod communities in the tree canopy were sampled periodically by beating and suction techniques. Significantly higher numbers of beneficial arthropods were found in the RV and S treatments in comparison with bare soil. Spiders and parasitoid wasps were the two most common groups, representing, respectively, 70% and 19% of all catches in beating samples and 33% and 53% in suction samples. For the RV and S treatments, significant seasonal deviations from the bare soil treatment were observed using principal response curves. Similar seasonal patterns were observed over the two years. The RV and S treatments showed significant positive deviations from the BS treatment in late spring and summer, accounted for the higher numbers of parasitoid wasps, coccinelids and lacewings present. By contrast, the seasonal deviations observed for the spider community differed from those of the remaining arthropods. During late winter and early spring, the RV and S treatments presented a higher abundance of spiders in the tree canopy, in comparison with bare soil, whereas in the summer significantly more spiders were found in the bare soil treatment. Spider movements between tree canopy and ground vegetation layers may justify this result.