Abstract
Antibiotic and pesticide resistance of pathogens are major and pressing worldwide issues. Resistance evolution is often considered in simplified ecological contexts: treated versus nontreated environments. In contrast, antibiotic usually present important dose gradients: from ecosystems to hospitals to polluted soils, in treated patients across tissues. However, we do not know whether adaptation to low or high doses involves different phenotypic traits, and whether these traits trade-off with each other. In this study, we investigated the occurrence of such fitness trade-offs along a dose gradient by evolving experimentally resistant lines of Escherichia coli at different antibiotic concentrations for similar to 400 generations. Our results reveal fast evolution toward specialization following the first mutational step toward resistance, along with pervasive trade-offs among different evolution doses. We found clear and regular fitness patterns of specialization, which converged rapidly from different initial starting points. These findings are consistent with a simple fitness peak shift model as described by the classical evolutionary ecology theory of adaptation across environmental gradients. We also found that the fitness costs of resistance tend to be compensated through time at low doses whereas they increase through time at higher doses. This cost evolution follows a linear trend with the log-dose of antibiotic along the gradient. These results suggest a general explanation for the variability of the fitness costs of resistance and their evolution. Overall, these findings call for more realistic models of resistance management incorporating dose-specialization.