Phenotypic heterogeneity is adaptive for microbial populations under starvation

To persist in variable environments populations of microorganisms have to survive periods of starvation and be able to restart cell division in nutrient-rich conditions. Typically, starvation signals initiate a transition to a quiescent state in a fraction of individual cells, while the rest of the cells remain non-quiescent. It is widely believed that, while quiescent cells (Q) help the population to survive long starvation, the non-quiescent cells (NQ) are a side effect of imperfect transition. We analysed regrowth of starved monocultures of Q and NQ cells compared to mixed, heterogeneous cultures in simple and complex starvation environments. Our experiments, as well as mathematical modelling, demonstrate that Q monocultures benefit from better survival during long starvation, and from a shorter lag phase after resupply of rich medium. However, when the starvation period is very short, the NQ monocultures outperform Q and mixed cultures, due to their short lag phase. In addition, only NQ monocultures benefit from complex starvation environments, where nutrient recycling is possible. Our study suggests that phenotypic heterogeneity in starved populations could be a form of bet hedging, which is adaptive when environmental determinants, such as the length of the starvation period, the length of the regrowth phase, and the complexity of the starvation environment vary over time. ImportanceNon-genetic cell heterogeneity is present in glucose starved yeast populations in the form of quiescent (Q) and nonquiescent (NQ) phenotypes. There is evidence that Q cells help the population to survive long starvation. However, the role of the NQ cell type is not known, and it has been speculated that the NQ phenotype is just a side effect of imperfect transition to the Q phenotype. Here we show that, in contrast, there are ecological scenarios in which NQ cells perform better than monocultures of Q cells or naturally occuring mixed populations containing both Q and NQ. NQ cells benefit when the starvation period is very short and environmental conditions allow nutrient recycling during starvation. Our experimental and mathematical modeling results suggest a novel hypothesis: the presence of both Q and NQ phenotypes within starved yeast populations may reflect a form of bet hedging, where different phenotypes provide fitness advantages depending on environmental conditions.