A long-standing question in evolutionary biology is the extent to which adaptation to novel stressors can buffer populations from extinction. This question is arguably one of the most important questions for evolutionary biologists to answer in this age of rapid global change. Knowledge of how best to manage genetic variation in organisms faced with dramatic changes in their abiotic or biotic environment, such as those associated with anthropogenic climate change or emerging infectious disease, is critical for mitigating these threats. Genetic responses to rapid environmental changes are often characterized by selective sweeps, whereby a newly beneficial allele increases in frequency and becomes fixed in response to selection imposed by the novel stressor. The classic example of a selective sweep is referred to as a “hard sweep”, which is when a de novo mutation increases rapidly in frequency, resulting in a significant decrease in genetic diversity in and adjacent to the selected locus. A “soft sweep”, in contrast, is when a relatively more common neutral allele becomes beneficial in the new environment. Because such a previously neutral allele tends to be present in multiple haplotype blocks, genetic diversity is often maintained as the newly beneficial allele increases in frequency. Detection of hard sweeps is typically straightforward; however, soft sweeps can be challenging to detect due to the relative lack of gametic disequilibrium between the beneficial mutation and the nearby genomic background. In this issue of Molecular Ecology, Gignoux-Wolfsohn et al. (2021) test for subtle signatures of selection in bat populations recovering from mass-mortality events caused by white-nose syndrome (WNS). By combining long-term population monitoring, timely sampling, whole genome sequencing, and sensitive analytical approaches, the authors reveal evidence of population recovery driven by selection acting on standing genetic variation, characterized by soft sweeps and numerous loci of small effect. In doing so, the authors demonstrate an exemplary framework for uncovering adaptive responses to novel and dramatic stressors – knowledge that is essential to our efforts to preserve biodiversity in the face of rapid environmental change.

Emerging infectious diseases (EIDs) of wildlife are increasing and represent one of the leading threats to global biodiversity (Jones et al., 2008). White-nose syndrome (WNS), caused by the fungal pathogen Pseudogymnoascus destructans, is one of the most devastating examples of an EID threatening wildlife populations. Native to Europe and Asia, WNS was first detected in the United States in 2006 and has swept through North American bat populations, leading to dramatic declines, particularly in populations of the little brown bat Myotis lucifugus. However, variation in population trajectories following initial declines – ranging from extirpation to eventual positive population growth – imply the capacity of some M. lucifugus populations to adapt in response to selection by WNS. Identifying the targets of this selection has been challenging, and, in particular, an outstanding question has been whether these targets are consistent across different populations. By sampling multiple populations at different time points comprising pre- and post-WNS individuals, Gignoux-Wolfsohn et al. (2021) sought to uncover the extent to which selection in response to WNS has occurred in parallel among populations by examining changes in allele frequencies associated with WNS-induced mass mortality events.

Gignoux-Wolfsohn et al. (2021) focused on four WNS-affected M. lucifugus hibernacula in the northeastern United States: Hibernia Mine, Walter Williams Preserve, Aeolus Cave, and Greeley Mine (Figure 1). Pre-WNS samples comprised nonsurviving individuals collected immediately following the first outbreaks of WNS, whereas post-WNS samples were collected from living survivors seven years (one to two generations) later. They genotyped 132 individuals at over 30 million SNPs using a cost-effective probabilistic approach for accurately calculating population-level allele frequencies from low coverage whole genome sequences. Using these data, they assessed population structure and estimated site frequency spectra for each population to quantify genetic variation (θ) and Tajima's D. Then, Gignoux-Wolfsohn et al. (2021) identified loci under selection by focusing first on loci with large changes in allele frequency in the same direction in both focal populations with pre- and post-WNS data to maximize computational efficiency and reduce false positives. Specifically, they quantified the proportion of allele frequency change attributable to selection by producing a null model accounting for other potentially confounding causes of allele frequency change over time. To account for sampling error, the authors recalculated allele frequencies after bootstrapping across individuals. To account for genetic drift, the authors projected changes in pre-WNS population allele frequencies after two generations of genetic drift under a Wright-Fisher simulation model. Finally, the authors estimated selection coefficients for candidate loci using an approximate Bayesian computation model and accounted for the potential effect of immigration on allele frequency change by comparing observed immigration rates with those predicted by observed allele frequency changes. To identify loci with smaller allele frequency changes that may be indicative of small effects on disease resistance, the authors performed a Cochran-Mantel-Haenszel test for allele frequency changes associated with survivorship across populations.

The thoughtful study design of Gignoux-Wolfsohn et al. (2021) allowed them to identify genetic changes that were consistent with a soft selective sweep. They found that genome-wide diversity was largely maintained despite dramatic population declines of 80%–98% and an increase in Tajima's D, supporting a population bottleneck in one of the two populations with pre- and post-WNS data. After accounting for sampling error, genetic drift, and immigration, they found evidence for strong selection at 63 candidate loci with large allele frequency changes occurring in parallel among populations following WNS-induced mass mortality. These loci showed characteristics consistent with a soft sweep, having nonzero allele frequencies pre-WNS and with surrounding regions showing no evidence of changes in nucleotide diversity or Tajima's D that would indicate a hard sweep. Genes containing candidate loci were largely associated with hibernation and metabolism instead of immunity. In addition, the authors identified almost ten thousand loci of putative small effect that were significantly associated with survival. Together, these results suggest that the capacity of populations to recover following WNS-induced mass mortality events is dependent on standing genetic variation.

Critical to this study was the ongoing monitoring of bat populations. This monitoring provided several benefits (Figure 2); first, it provided long-term census data, including population trajectories following WNS-induced mass mortality, that could be used to contextualize observed genomic responses to WNS. This combination of field data and genomics greatly strengthened the inferential power of this study beyond what could be achieved using either approach alone. Second, pre-WNS monitoring enabled timely detection of WNS arrival so that sampling could be conducted at appropriate time points both during and following WNS-induced mass mortality events. This serial sampling facilitated comparisons of population allele frequencies pre- and post-disease that were critical to achieving the power necessary to detect subtle signatures of selection.

Despite these strengths, this study contained some limitations to the sampling design that are instructive for future work. For example, pre-WNS samples only included nonsurviving individuals collected during mass mortality events (no survivors were sampled), meaning that pre-WNS allele frequencies may have been somewhat biased. Although the authors correctly point out that survivors represented a small proportion of these populations and would have had minimal impact on pre-WNS allele frequency estimates, this bias in estimating allele frequencies could be substantially higher if a similar sampling approach were used for populations with lower mortality rates. In addition, estimates of pre-WNS population sizes were 10 years old for the Williams hibernaculum, were taken concurrently with WNS arrival for Hibernia, and were nonexistent for the others, which limited quantification of WNS-induced population declines. Both of these limitations could have been avoided through enhanced population monitoring and sampling predisease arrival. Nonetheless, together with a small number of other recent examples (Epstein et al., 2016; Gurgel et al., 2020; Poff et al., 2018), this study underlines the importance of long-term monitoring of wildlife populations to establish baseline demographic, environmental, and genetic conditions against which novel environmental stressors and their associated population impacts and genomic responses can be accurately measured and contextualized.

Disease surveillance has long been recognised as being critical to our ability to detect and respond to emerging threats to wildlife, domestic animal, and human health (Lipkin, 2013). We concur with this view but propose that this surveillance be coupled with genetic sampling so that genomic changes associated with disease-driven selection events can be accurately quantified (Figure 2). This recommendation equally applies to biodiversity threats other than EIDs, such as climate change. Extreme weather events such as the unprecedented heatwaves during the recent northern summer represent major selection events (Schiermeier, 2021). Such events are becoming increasingly frequent and severe, but our capacity to understand their effects is limited by a lack of baseline data against which we can measure genetic change. Ambitious national and state genomic monitoring programmes are necessary to fill this void but are currently neglected in biodiversity conservation plans (Hoban et al., 2021). Nonetheless, recent efforts to develop plans for monitoring genomic diversity in wildlife (e.g., Posledovich et al., 2021) are an important step forward in global efforts to preserve biodiversity in an era of rapid environmental change.

进化生物学中一个长期存在的问题是对新压力源的适应可以在多大程度上缓冲种群免于灭绝。在这个全球快速变化的时代，这个问题可以说是进化生物学家要回答的最重要的问题之一。了解如何最好地管理面临非生物或生物环境发生巨大变化的生物体的遗传变异，例如与人为气候变化或新发传染病相关的生物体，对于减轻这些威胁至关重要。对快速环境变化的遗传反应通常以选择性扫描为特征，由此新的有益等位基因频率增加，并响应新压力源施加的选择而变得固定。选择性扫描的经典示例称为“硬扫描”，这是当从头突变的频率迅速增加时，导致所选基因座及其附近的遗传多样性显着降低。相反，“软扫描”是指相对更常见的中性等位基因在新环境中变得有益。因为这种以前的中性等位基因往往存在于多个单倍型块中，所以随着新的有益等位基因频率的增加，遗传多样性通常得以保持。硬扫描的检测通常很简单。然而，由于有益突变和附近基因组背景之间相对缺乏配子不平衡，软扫描可能难以检测。在本期 相反，“软扫描”是指相对更常见的中性等位基因在新环境中变得有益。因为这种以前的中性等位基因往往存在于多个单倍型块中，所以随着新的有益等位基因频率的增加，遗传多样性通常得以保持。硬扫描的检测通常很简单。然而，由于有益突变和附近基因组背景之间相对缺乏配子不平衡，软扫描可能难以检测。在本期 相反，“软扫描”是指相对更常见的中性等位基因在新环境中变得有益。因为这种以前的中性等位基因往往存在于多个单倍型块中，所以随着新的有益等位基因频率的增加，遗传多样性通常得以保持。硬扫描的检测通常很简单。然而，由于有益突变和附近基因组背景之间相对缺乏配子不平衡，软扫描可能难以检测。在本期 硬扫描的检测通常很简单。然而，由于有益突变和附近基因组背景之间相对缺乏配子不平衡，软扫描可能难以检测。在本期 硬扫描的检测通常很简单。然而，由于有益突变和附近基因组背景之间相对缺乏配子不平衡，软扫描可能难以检测。在本期分子生态学，Gignoux-Wolfsohn 等人。( 2021 ) 测试从白鼻综合征 (WNS) 引起的大规模死亡事件中恢复的蝙蝠种群选择的微妙特征。通过结合长期种群监测、及时采样、全基因组测序和敏感的分析方法，作者揭示了由作用于现有遗传变异的选择驱动的种群恢复的证据，其特征是软扫描和许多小效应位点。在这样做的过程中，作者展示了一个示范性框架，用于揭示对新奇和戏剧性压力源的适应性反应——这些知识对于我们在面对快速的环境变化时保护生物多样性的努力至关重要。

野生动物的新发传染病 (EID) 正在增加，是对全球生物多样性的主要威胁之一（Jones 等人，2008 年）。由真菌病原体Pseudogymnoascus destructans引起的白鼻综合征 (WNS)是 EID 威胁野生动物种群的最具破坏性的例子之一。WNS 原产于欧洲和亚洲，于 2006 年在美国首次被发现，并席卷了北美蝙蝠种群，导致数量急剧下降，特别是在小棕蝙蝠Myotis lucifugus的种群中。然而，随着最初的下降，人口轨迹的变化——从灭绝到最终的人口正增长——意味着一些M的能力。 萤火虫种群适应 WNS 的选择。确定这种选择的目标一直具有挑战性，特别是一个悬而未决的问题是这些目标在不同人群中是否一致。Gignoux-Wolfsohn 等人通过在不同时间点对包括 WNS 前后个体的多个群体进行采样。（2021 年）通过检查与 WNS 引起的大规模死亡事件相关的等位基因频率的变化，试图揭示针对 WNS 的选择在人群中平行发生的程度。

Gignoux-Wolfsohn 等人。( 2021 ) 专注于四个受 WNS 影响的M。 萤火虫美国东北部的冬眠：Hibernia Mine、Walter Williams Preserve、Aeolus Cave 和 Greeley Mine（图 1）。WNS 前样本包括在 WNS 第一次爆发后立即收集的非幸存者，而 WNS 后样本是在七年后（一到两代）从活着的幸存者那里收集的。他们使用具有成本效益的概率方法对超过 3000 万个 SNP 的 132 个人进行了基因分型，以准确计算来自低覆盖全基因组序列的群体水平等位基因频率。利用这些数据，他们评估了每个种群的种群结构和估计的位点频谱，以量化遗传变异 (θ) 和 Tajima's D。然后，Gignoux-Wolfsohn 等人。( 2021) 通过首先关注具有 WNS 前后数据的两个焦点群体中等位基因频率在同一方向上发生较大变化的基因座来确定选择中的基因座，以最大限度地提高计算效率并减少误报。具体来说，他们通过产生一个解释等位基因频率随时间变化的其他潜在混杂原因的空模型来量化归因于选择的等位基因频率变化的比例。为了解决抽样误差，作者在跨个体自举后重新计算了等位基因频率。为了解释遗传漂变，作者在 Wright-Fisher 模拟模型下预测了两代遗传漂变后 WNS 前种群等位基因频率的变化。最后，作者使用近似贝叶斯计算模型估计了候选基因座的选择系数，并通过比较观察到的迁移率与观察到的等位基因频率变化预测的迁移率来解释迁移对等位基因频率变化的潜在影响。为了识别具有较小等位基因频率变化的基因座，这些变化可能表明对疾病抵抗力的影响很小，作者对与人群存活相关的等位基因频率变化进行了 Cochran-Mantel-Haenszel 检验。

Gignoux-Wolfsohn 等人的周到研究设计。（2021 年）使他们能够识别与软选择性扫描一致的基因变化。他们发现，尽管种群数量急剧下降了 80%–98% 并且 Tajima 的D，支持具有 WNS 前后数据的两个种群之一的种群瓶颈。在考虑了抽样误差、遗传漂变和移民后，他们发现了在 63 个候选基因座上进行强选择的证据，这些基因座在 WNS 引起的大规模死亡率后在人群中平行发生了较大的等位基因频率变化。这些基因座显示出与软扫描一致的特征，在 WNS 之前具有非零等位基因频率，并且周围区域没有显示核苷酸多样性或 Tajima's D变化的证据这将表明艰难的扫荡。包含候选基因座的基因主要与冬眠和新陈代谢有关，而不是与免疫有关。此外，作者确定了近一万个与生存显着相关的假定小效应位点。总之，这些结果表明，在 WNS 引起的大规模死亡事件后，种群恢复的能力取决于现有的遗传变异。

这项研究的关键是对蝙蝠种群的持续监测。这种监测提供了几个好处（图 2）；首先，它提供了长期人口普查数据，包括 WNS 引起的大规模死亡后的人口轨迹，可用于将观察到的基因组对 WNS 的反应背景化。这种现场数据和基因组学的结合极大地增强了这项研究的推理能力，超出了单独使用任何一种方法所能达到的程度。其次，WNS 前监测能够及时检测 WNS 到达，以便在 WNS 引起的大规模死亡事件期间和之后的适当时间点进行采样。这种连续采样促进了疾病前后人口等位基因频率的比较，这对于实现检测细微选择特征所需的能力至关重要。

尽管有这些优势，但本研究对抽样设计存在一些限制，这些限制对未来的工作具有指导意义。例如，WNS 之前的样本仅包括在大规模死亡事件期间收集的非幸存者（没有幸存者被抽样），这意味着 WNS 之前的等位基因频率可能有些偏差。尽管作者正确地指出，幸存者代表了这些人群的一小部分，并且对 WNS 前等位基因频率估计的影响很小，但如果将类似的抽样方法用于具有较低死亡率。此外，WNS 冬眠前的 WNS 种群规模估计为 10 岁，与 WNS 到达 Hibernia 同时进行，而其他人则不存在，这限制了 WNS 引起的人口下降的量化。这两个限制都可以通过加强人口监测和采样疾病前到达来避免。尽管如此，连同少数其他最近的例子（爱泼斯坦等人，2016 年；Gurgel 等人，2020 年；Poff 等人，2018 年），这项研究强调了对野生动物种群进行长期监测以建立基线人口、环境和遗传条件的重要性，据此可以准确测量和情境化新的环境压力源及其相关的种群影响和基因组反应.

长期以来，疾病监测一直被认为对我们检测和应对对野生动物、家畜和人类健康的新威胁的能力至关重要（Lipkin，2013 年）。我们同意这一观点，但建议将此监测与基因采样相结合，以便可以准确量化与疾病驱动的选择事件相关的基因组变化（图 2）。该建议同样适用于 EID 以外的生物多样性威胁，例如气候变化。极端天气事件，例如最近北方夏季前所未有的热浪是主要的选择事件（Schiermeier，2021）。此类事件变得越来越频繁和严重，但我们了解其影响的能力受到缺乏可以衡量遗传变化的基线数据的限制。雄心勃勃的国家和州基因组监测计划对于填补这一空白是必要的，但目前在生物多样性保护计划中被忽视（Hoban 等人，2021 年）。尽管如此，最近制定监测野生动物基因组多样性计划的努力（例如，Posledovich 等人，2021 年）是在环境快速变化的时代保护生物多样性的全球努力向前迈出的重要一步。