The evolutionary consequences of seasonality: assessing demography and balancing selection in real time. Alan O. Bergland¹, Emily Behrman², Katherine O'Brien², Paul Schmidt², Dmitri A. Petrov¹. 1) Dept. of Biol., Stanford Univ, Stanford, CA; 2) Dept. of Biol., Univ. of Penn., Philadelphia, PA.

   Assessing the change in genetic composition of a population through time provides a means to understand the evolutionary forces acting on that population. For D. melanogaster living in temperate environments, seasonal climatic fluctuations cause severe demographic events and act as strong selective agents. To identify the evolutionary impact of seasonality in flies, we performed whole genome resequencing of large samples of flies collected in a temperate, Pennsylvania orchard during the spring and fall over three consecutive years and contrast changes in allele frequency through time with changes in allele frequency along a North American latitudinal cline. We demonstrate that populations collected through time are as differentiated as populations separated by 5-10 latitude. Using forward simulations, we show that such levels of differentiation are more consistent with models that incorporate temporal variation in selection coefficients and population size than purely demographic models. Accordingly we performed a genome-wide scan for SNPs that vary in frequency in a cyclic fashion and conservatively identify ~1000 such sites that strongly vary in frequency by ~20% between the seasons and infer that per-generation selection coefficient acting on these SNPs is on the order of 10%. We show that a large fraction of the seasonal SNPs can evolve independently under simple models of natural selection and that patterns of polymorphism surrounding seasonal SNPs are consistent with adaptive evolution. We also show that frequencies of seasonal SNPs following the first frost of the winter respond in a predictable way. Finally, we contrast seasonal SNPs with clinal SNPs and find that while seasonal SNPs are weakly clinal, there is surprisingly little overlap. Taken together, our results highlight the abundance of balancing selection driven by seasonal variation in selection coefficients and corroborate models put forward by T. Dobzhansky over 50 years ago.