Temporally fluctuating environmental conditions are a ubiquitous feature of natural habitats. Yet, how finely natural populations adaptively track fluctuating selection pressures via shifts in standing genetic variation is unknown^1,2. Here we generated genome-wide allele frequency data every 1–2 generations from a genetically diverse population of Drosophila melanogaster in extensively replicated field mesocosms from late June to mid-December (a period of approximately 12 total generations). Adaptation throughout the fundamental ecological phases of population expansion, peak density and collapse was underpinned by extremely rapid, parallel changes in genomic variation across replicates. Yet, the dominant direction of selection fluctuated repeatedly, even within each of these ecological phases. Comparing patterns of change in allele frequency to an independent dataset procured from the same experimental system demonstrated that the targets of selection are predictable across years. In concert, our results reveal a fitness relevance of standing variation that is likely to be masked by inference approaches based on static population sampling or insufficiently resolved time-series data. We propose that such fine-scaled, temporally fluctuating selection may be an important force contributing to the maintenance of functional genetic variation in natural populations and an important stochastic force impacting genome-wide patterns of diversity at linked neutral sites, akin to genetic draft.

随时间波动的环境条件是自然栖息地的普遍特征。然而，自然种群如何通过固定遗传变异的变化来适应性地跟踪波动的选择压力尚不清楚^1,2 。在这里，我们从 6 月下旬到 12 月中旬（总共大约 12 代）在广泛复制的田间中生态系统中，每 1-2 代从遗传多样化的黑腹果蝇群体中生成全基因组等位基因频率数据。种群扩张、峰值密度和崩溃等基本生态阶段的适应是由重复中基因组变异极其快速、平行的变化所支撑的。然而，选择的主导方向反复波动，即使在每个生态阶段也是如此。将等位基因频率的变化模式与从同一实验系统获得的独立数据集进行比较表明，选择的目标多年来是可预测的。总之，我们的结果揭示了常备变异的适应性相关性，这种相关性很可能被基于静态总体抽样或未充分解析的时间序列数据的推理方法所掩盖。我们认为，这种精细的、随时间波动的选择可能是有助于维持自然群体中功能​​遗传变异的重要力量，也是影响相连中性位点全基因组多样性模式的重要随机力量，类似于遗传草案。