Drug rotation (cycling), in which multiple drugs are administrated alternatively, has the potential for limiting resistance evolution in pathogens. The frequency of drug alternation could be a major factor to determine the effectiveness of drug rotation. Drug rotation practices often have low frequency of drug alternation, with an expectation of resistance reversion. Here we, based on evolutionary rescue and compensatory evolution theories, suggest that fast drug rotation can limit resistance evolution in the first place. This is because fast drug rotation would give little time for the evolutionarily rescued populations to recover in population size and genetic diversity, and thus decrease the chance of future evolutionary rescue under alternate environmental stresses. We experimentally tested this hypothesis using the bacterium Pseudomonas fluorescens and two antibiotics (chloramphenicol and rifampin). Increasing drug rotation frequency reduced the chance of evolutionary rescue, and most of the final surviving bacterial populations are resistant to both drugs. Drug resistance incurred significant fitness costs, which did not differ among the drug treatment histories. A link between population sizes during the early stages of drug treatment and the end-point fates of populations (extinction versus survival) suggested that population size recovery and compensatory evolution before drug shift increase the chance of population survival. Our results therefore advocate fast drug rotation as a promising approach to reduce bacterial resistance evolution, which in particular could be a substitute for drug mixing when the latter has safety risks.

This dataset was collected from the experiment we performed. See the article for details.