Temporally variable environments are the norm rather than the exception in nature. Yet, the ecological consequences of this variability and the evolutionary responses it invokes remain poorly understood. In this thesis, a previously proposed theory of competitive coexistence was further developed that yielded a nonconventional prediction: fluctuating environments can support stable coexistence of competitors even in the absence of negative frequency-dependent selection. It was confirmed by laboratory competition using bacteria. After generalization of the theory by simulation, an alternative to the genetic drift model emerges that explains the rich polymorphism observed in nature. Next, a dynamic theory of bet-hedging was developed and tested by experiments, which also proved stochastic phenotypic switching as a highly adaptive bet-hedging strategy. Besides, this new theory predicts that the standard theory of bet-hedging should fail under certain conditions of direct biological relevance. With the ecological and evolutionary models tested, a forward evolutionary experiment was carried out to study adaptation to fluctuating environments. Hypothesis free, this effort captured unexpected strategies to cope with cyclic environments, revealing the generative effect of trade-off in the presence of two-dimensional selection. Together, these three projects offer a multi-perspective picture of the complex process of adaptation in fluctuating environments, with an emphasis on mechanisms—ecological or physiological—that underlie the emergence of varied evolutionary responses.