Variation in foraging behavior may indicate differences in food availability and allow assessment of restoration actions. Ants are prominent bioindicators used in assessing ecological responses to disturbance. However, behavioral data have been poorly incorporated as an index. The foraging performance of red harvester ants was quantified in order to evaluate the success of a restoration ecology experiment in the tropical dry forest of Sierra de Huautla, Morelos, in central Mexico. Foraging performance by granivorous, Pogonomyrmex barbatus, ants was diminished after 6 and 8 years of cattle grazing and wood harvest were excluded as part of a restoration experiment in a highly degraded biome. Despite investing more time in foraging, ant colonies in exclusion plots showed lower foraging success and acquired less seed biomass than colonies in control plots. In line with the predictions of optimal foraging theory, in restored plots where ant foraging performance was poor, ants harvested a higher diversity of seeds. Reduced foraging success and increased harvest of non-preferred foods in exclusion plots were likely due to the growth of herbaceous vegetation, which impedes travel by foragers. Moreover, by 8 years of exclusion, 37% of nests in exclusion plots had disappeared compared to 0% of nests in control plots. Ants' foraging success and behavior were sensitive to changes in habitat quality due to the plant successional process triggered by a restoration intervention. This study spotlights on the utility of animal foraging behavior in the evaluation of habitat restoration programs.