1. The abiotic environment drives species abundances and distributions both directly and indirectly through effects on multi-trophic species interactions. However, few studies have documented the individual and combined consequences of these direct and indirect effects. 2. We studied an ant-tended aphid along an elevational gradient, where lower elevations were more arid. Hypotheses of stronger species interactions at lower elevations and a greater sensitivity of higher trophic levels to climate led us to predict increased top-down control of aphids by natural enemies (third trophic level) but even stronger protection from mutualist ants (fourth trophic level) with increasing aridity. As a result, we predicted that mutualism strength and aphid abundance would increase with aridity. 3. We documented patterns of aphid abundance and tested for both the direct and multi-trophic indirect effects of aridity on aphid performance. To do so, we used both observational and manipulative methods across two years in replicate high and low elevation valleys, where summer temperatures decreased by 3.7°C and precipitation increased by 27 mm/mo from low to high elevations. 4. Aphid colonies were 75% larger in the most (vs. least) arid sites, and this was best explained by changes in interactions with predators and ants. Aphids were unaffected by the direct effects of the abiotic environment or its indirect effects via host plant quality. In contrast, natural enemy effects increased with aridity; under ant exclusion, natural enemies had no effect on aphids in the least arid sites but depressed colony growth by 252% in the most arid sites. Ant activity also increased with aridity, with ants discovering more aphid colonies and experimental baits and allocating more foragers per aphid, although there was no effect of aridity on ant abundance or community composition. Correspondingly, the mutualist services provided by ants increased with aridity; ants provided no benefits to aphids in the least arid sites but doubled colony growth in the most arid sites. 5. In summary, an elevational cline in herbivore abundance was driven by a monotonic increase in trophic level sensitivity to aridity. These findings illustrate that predicting species responses to climate change will require a multi-trophic perspective.