Declines in plants and herbivorous insects due to land use abandonment and intensification have been studied in agricultural areas worldwide. We tested four hypotheses, which were complementary rather than mutually exclusive, to understand the mechanisms driving biodiversity declines due to abandonment and intensification. These predict that biodiversity decline is caused by a decline in resource diversity, changes in disturbance regime, surrounding landscape conversion, and a decrease in biomass production. We compared plant richness and butterfly and orthopteran richness and diversity among three land use types in seminatural grasslands: abandoned, traditional, and intensified terraces. Then, we examined effects of changes in resource (plant) richness, frequency of disturbance (mowing), and surrounding landscapes on butterfly and orthopteran diversity to understand the mechanisms driving decline after land abandonment and intensification. Plant and herbivore richness and diversity were significantly lower in abandoned and intensified grasslands than in traditional grasslands. This trend was consistent throughout the seasons in both years of the study. Changes in mowing frequency and surrounding landscape explained plant richness declines as a consequence of land abandonment and intensification. Declines in herbivorous insects were explained by plant richness declines and changes in mowing frequency, but not by landscape changes. Plant and herbivore richness were maximized at an intermediate mowing frequency (approximately twice per year), which is typical practice on traditional terraces. This is the first report demonstrating that the intermediate disturbance hypothesis explained well the biodiversity declines in agricultural ecosystems. The richness and diversity responses of herbivore functional groups to plant richness, mowing frequency, and surrounding landscapes were generally inconsistent with predictions. We found significant trends in which butterfly and orthopteran species with low abundance in traditional terraces were lost in abandoned and/or intensive terraces. This may suggest that the number of individuals of most herbivorous species decreased randomly with respect to life-history traits following a decline in plant richness after changes in disturbance frequency. This study demonstrates that declines in herbivorous insects can be explained by multiple factors, and provides a unified explanation for biodiversity declines in both abandoned and intensified use of agricultural lands, which have often been studied separately.