Our understanding of the causes of molecular evolution is not as certain as it was a decade ago when Kimura's neutral theory appeared to explain major features of DNA conservation and change. The last ten years have seen the development of empirical approaches and statistical tests for detecting selection in DNA and a proliferation of data that challenge our current understanding of the molecular evolutionary process. We begin this review with a discussion of protein polymorphism and divergence: two major areas of research where the strictly neutral model cannot explain general patterns in the data. We then present a survey of statistical methods for detecting positive selection, which includes tests for balancing selection, for sequence convergence, and for unusually high rates of evolution that cannot be accounted for by neutral models. Finally, we present findings of a number of groups working on within- and between-species variation in Drosophila: These highlight the importance of adaptive evolution, purifying selection, and recombination in understanding levels and patterns of nucleotide variation.
