There is a demand for mechanistic studies to explore underlying drivers behind observed patterns of biodiversity in urban areas. We describe a two-year field experiment in which we manipulated bottom-up (resource availability) and top-down (bird predation) forces on arthropod communities associated with a native plant, Encelia farinosa, across three land-use types—urban, desert remnant, and outlying natural desert—in the Phoenix metropolitan area, Arizona, USA. We monitored the trophic structure, richness, and similarity of the arthropod communities on these manipulated plants over a two-year period. We predicted that (1) increased water resources increase plant productivity, (2) increased productivity increases arthropod abundances, and (3) in the urban habitat, top-down forces are greater than in other habitats and limit arthropod abundances. We also predicted that urban remnant habitats are more similar to urban habitats in terms of arthropod richness and composition. Strong interannual differences due to an unusual cold and dry winter in the first year suppressed plant growth in all but urban habitats, and arthropod abundances in all habitats were severely reduced. In the following year, arthropod abundances in desert and remnant habitats were higher than in urban habitats. Water had positive effects on plant growth and arthropod abundance, but these water effects emerged through complex interactions with habitat type and the presence/absence of cages used to reduce bird predation. Plants grew larger in urban habitats, and phenology also differed between urban and desert habitats. The results from caging suggest that bird predation may not be as important in cities as previously thought, and that arthropods may retard plant growth. As expected, desert communities are strongly bottom-up regulated, but contrary to predictions, we did not find evidence for strong top-down control in the city. Remnant habitats were intermediate between desert and urban habitats in terms of diversity, richness, evenness, arthropod composition and phenology, with urban habitats generally lowest in terms of diversity, richness, and evenness. Our study shows that control of biodiversity is strongly altered in urban areas, influenced by subtle shifts in top-down and bottom-up controls that are often superseded by climatic variations and habitat type.