1 INTRODUCTION

Much of the current literature on forest transformations driven by rising temperatures, increasing hydroclimate variability and changes disturbance regimes has focussed on areas that experience frequent drought or fires, such as the western United States (Allen et al., 2015; Hartmann et al., 2018; Romme et al., 2011; Westerling et al., 2011), or in high-latitude regions where species and ecosystem distributions are closely regulated by temperature (Payette, 2021). Palaeoecological evidence can provide a deeper time perspective of these dynamics by helping further our understanding of rapid population fluctuations and corresponding ecosystem transitions, including in areas that are today seemingly resilient. In the Great Lakes region, mesic tree taxa have experienced multiple abrupt population declines during the Holocene and provide a rich study system for understanding past rapid vegetation changes (Booth, Brewer, et al., 2012; Booth, Jackson, et al., 2012; Shuman, 2012; Shuman et al., 2009; Wang et al., 2016). For example, the range-wide population collapse of eastern hemlock (Tsuga canadensis) at 5.5 to 5.3 thousand years before present (ka BP) has been extensively studied and is often associated with one or more severe drought events suggesting high sensitivity to hydroclimatic variation (Booth, Brewer, et al., 2012; Oswald & Foster, 2012; Shuman, 2012; Shuman et al., 2009). However, other causes of the eastern hemlock collapse have been suggested, including regional temperature changes (Shuman et al., 2023) and outbreaks of hemlock looper or other pests or pathogens (Bhiry & Filion, 1996).

American beech (Fagus grandifolia, herein beech), another common mesic tree species in the eastern United States, has undergone rapid population fluctuations throughout the Holocene (Booth, Jackson, et al., 2012; Wang et al., 2016), but the patterns and causes are less well-studied. After the last glacial maximum, beech expanded northward from the southeastern United States, reaching the Great Lakes region between 8.0 and 6.0 ka (Bennett, 1985, 1988; Bernabo & Webb, 1977; Davis, 1981a; Webb et al., 1984; Williams, 1974). After establishment in the Great Lakes region, beech populations experienced repeated declines across sites in Ohio, Indiana (IN), Michigan (MI) and Wisconsin (Wang et al., 2016). At Spicer Lake, IN, beech populations rapidly expanded at 6.8 ka and subsequently ranged from near 30% to <5% pollen abundance (Wang et al., 2016). Beech populations at Spicer Lake experienced five abrupt and well-dated declines (5.3 ka, 4.3–4.0 ka, 3.2–2.0 ka, 1.2 ka, 1 ka), all followed by abrupt increases, except for the decline at 1 ka, from which beech has not recovered (Wang et al., 2016).

Better constraints on the timing of the beech declines across sites is essential to testing causal hypotheses. The declines at Spicer Lake appear to be asynchronous with declines at other sites in the Great Lakes region (Wang et al., 2016). However, this apparent asynchrony could be caused by poor dating constraints, as many records were collected decades ago and rely on relatively few bulk-sediment radiocarbon dates, which are prone to biasing due to hardwater effects (Grimm et al., 2009). Synchronous beech declines across the region would suggest macro-scale extrinsic drivers, such as temperature variations or pest outbreaks, as has been invoked for eastern hemlock (Shuman et al., 2009, 2023; Booth, Brewer, et al., 2012). Asynchronous beech variations would suggest localized interactions between extrinsic, intrinsic and disturbance processes, such as localized shifts in fire regime, local hydroclimate variability or species interactions leading to shifts in the dominant taxon. Cross-scale interactions are also possible, for example, local-scale feedbacks interacting with macro-scale extrinsic drivers, creating sub-regional temporal mosaics with clusters of synchronized declines (Williams et al., 2011). These intrinsic and extrinsic processes can further interact with disturbance regimes to amplify or mitigate rates of ecological change (Ratajczak et al., 2018) with transition zones between ecosystems being particularly susceptible to rapid changes in community composition (Hupy & Yansa, 2009; Nelson & Hu, 2008; Wiles, 2023; Williams et al., 2009).

The thin bark of beech makes it vulnerable to damage by fire, while a shallow root system makes it also susceptible to changes in soil moisture (Tubbs & Houston, 1990). During the Medieval Climate Anomaly (1.05–0.6 ka BP), beech declines were associated with increased fire frequency at some sites in northern Michigan (Booth, Jackson, et al., 2012), but at Spicer Lake, there was largely no consistent relationship between beech abundance and fire (Wang et al., 2016). At some sites, fire regimes can stabilize vegetation composition. For example, in Minnesota's Big Woods, the late-Holocene shift from oak woodland to mesic forest only occurred at the sites where fire was absent due to the presence of natural firebreaks (Calder, 2016; Grimm, 1984). Drought could also cause beech declines, but the Spicer Lake record lacks robust paleohydrology proxies (Wang et al., 2016). More independent paleohydrology proxies are needed, such as past lake-level variations (Digerfeldt et al., 1992; Pribyl & Shuman, 2014) or other indicators of the regional moisture balance to further evaluate the relationship between beech declines and hydroclimatic history (Booth, 2008; Adams et al., 2015). Variations in beech abundances may also be affected by changes in temperature, which appear to have affected the region during the Holocene (Puleo et al., 2020; Shuman et al., 2023) and pathogen outbreaks, which are known from modern forests. For example, some studies attribute late-Holocene declines in beech in New England and the Great Lakes region to Little Ice Age cooling (Fuller et al., 1998; Gajewski, 1987), whereas the accidental introduction of beech bark disease to the United States in the 1890s CE has devastated many beech forests, with mortality rates of 50% and infestation rates of 80 to 95% (Beckman et al., 2021; Stephanson & Ribarik Coe, 2017).

Even with independent proxies of hydroclimate variation, the potential for different sensitivities and response times between lacustrine and terrestrial ecosystems remains a source of uncertainty. A relatively new approach, which integrates the balance between carbon and water fluxes and thus shows promise for inferring shifts in the physiological sensitivity and response time of vegetation to climate forcing, is stable carbon isotopic (δ¹³C) analyses of fossil pollen grains of C₃ plants (Griener et al., 2013; Jahren, 2004; Loader & Hemming, 2004; Nelson, 2012). The δ¹³C values provide a seasonally integrated signal of intrinsic water use efficiency (iWUE), which is the ratio of photosynthesis and stomatal conductance of water (Farquhar et al., 1989). In terms of water-use strategies, plants that prioritize carbon gain relative to water loss have low iWUE and δ¹³C values, whereas those that prioritize water conservation relative to carbon gain have high iWUE and δ¹³C values (Bacon, 2009; Farquhar & Sharkey, 1982) and dry conditions tend to favour the latter (Sperry et al., 2017). Atmospheric CO₂, atmospheric pollution and climate (temperature and precipitation) have been shown to affect plant iWUE during the historical record via their influences on photosynthesis and/or stomatal conductance (Mathias et al., 2023), though the former two factors are unlikely to have been significant controls of iWUE during the Holocene prior to the industrial revolution. Pollen δ¹³C values (δ¹³C_pollen) represent landscape- to population-level shifts in iWUE because multiple pollen grains must be combined to produce enough carbon for δ¹³C analysis and each sediment sample comprises pollen grains from many individual plants within pollen source radii on the order of 10s of km (Prentice, 1988). The δ¹³C data from specific taxa can be coupled with pollen assemblage data to assess the influence of variations in iWUE on abundance changes for individual taxa and changes in community composition (Griener et al., 2013).

Here, we seek to better understand the drivers of rapid changes in beech abundances in the southern Great Lakes region, through a new well-dated, multi-proxy record for Story Lake, Indiana. This record includes proxies for past vegetation composition (pollen), iWUE of beech (δ¹³C_beech) and fire regime (charcoal) from the same core. The Story records are compared with a recent lake-level reconstruction from nearby Lake Lavine, MI (Ray-Cozzens, 2022) as an independent indicator of past hydroclimate variations. We also compare the history of beech variations at Story with those at two nearby sites within 15 km of Story Lake (Appleman and Pretty Lakes) and one more distal lake located 120 km to the west (Spicer Lake). These comparisons allow us to test hypotheses about the synchrony of vegetation changes at landscape to regional scales and thereby assess the relative importance of local-scale vegetation-fire-climate feedbacks and regional-scale climatic drivers on rapid changes in mesic tree populations in the southern Great Lakes region.

中文翻译：

---

火灾和水文气候对五大湖南部地区橡树和山毛榉群落流行率的相互作用

1 简介

当前许多关于由气温上升、水文气候变异性增加和干扰制度变化驱动的森林转型的文献都集中在经常经历干旱或火灾的地区，例如美国西部（Allen 等，  2015；Hartmann 等，  2018；Romme 等，  2011；Westerling 等，  2011），或在物种和生态系统分布受到温度密切调节的高纬度地区（Payette，  2021）。古生态学证据可以帮助我们进一步了解人口的快速波动和相应的生态系统转变，包括今天看似具有恢复力的地区，从而为这些动态提供更深入的时间视角。在五大湖地区，中生树类群在全新世期间经历了多次种群数量突然下降，并为了解过去快速植被变化提供了丰富的研究系统（Booth, Brewer, et al.,  2012；Booth, Jackson, et al.,  2012 ） ；舒曼，  2012；舒曼等，  2009；王等，  2016）。例如，东部铁杉（ Tsuga canadensis）在距今 5.5 至 5300 年（ka BP）的大范围种群崩溃已得到广泛研究，并且通常与一次或多次严重干旱事件有关，表明对水文气候变化高度敏感。 Booth、Brewer 等，  2012；Oswald 和 Foster，  2012 ； Shuman 等，  2009）。然而，东部铁杉崩溃的其他原因也被提出，包括区域温度变化（Shuman等人，  2023年）和铁杉尺蠖或其他害虫或病原体的爆发（Bhiry＆Filion，  1996年）。

美国山毛榉（Fagus grandifolia，本文为山毛榉）是美国东部另一种常见的中生树种，在整个全新世经历了快速的种群波动（Booth, Jackson, et al.,  2012；Wang et al.,  2016），但模式和原因研究较少。在末次盛冰期之后，山毛榉从美国东南部向北扩张，在 8.0 至 6.0ka 之间到达五大湖地区（Bennett，  1985，1988；Bernabo & Webb，  1977；Davis，  1981a；Webb 等，  1984；Williams） ，  1974）。在五大湖地区定居后，俄亥俄州、印第安纳州 (IN)、密歇根州 (MI) 和威斯康星州的山毛榉种群数量经历了反复下降（Wang 等，  2016）。在印第安纳州斯派塞湖，山毛榉种群在 6.8 ka 时迅速扩张，随后花粉丰度范围从近 30% 到 <5%（Wang 等人，  2016）。斯派塞湖的山毛榉种群经历了五次突然且日期明确的下降（5.3 ka、4.3-4.0 ka、3.2-2.0 ka、1.2 ka、1 ka），随后均突然增加，除了 1 ka 时的下降外，山毛榉尚未恢复（Wang et al.,  2016）。

更好地限制不同地点山毛榉数量减少的时间对于检验因果假设至关重要。斯派塞湖的下降似乎与五大湖地区其他地点的下降不同步（Wang 等人，  2016 年）。然而，这种明显的异步性可能是由不良的测年限制造成的，因为许多记录是几十年前收集的，并且依赖于相对较少的大量沉积物放射性碳测年，而由于硬水效应，这些记录很容易出现偏差（Grimm等，  2009）。该地区山毛榉的同步衰退表明存在宏观外在驱动因素，例如温度变化或害虫爆发，正如东部铁杉所引用的那样（S human 等人，  2009 年、2023 年；Booth、Brewer 等人，  2012 年）。异步山毛榉变化表明外在、内在和干扰过程之间存在局部相互作用，例如火灾状况的局部变化、局部水文气候变化或导致主要分类单元变化的物种相互作用。跨尺度相互作用也是可能的，例如，局部尺度反馈与宏观尺度外在驱动因素相互作用，创建具有同步下降集群的次区域时间马赛克（Williams et al.,  2011）。这些内在和外在过程可以进一步与扰动机制相互作用，以放大或减轻生态变化率（Ratajczak等，  2018），生态系统之间的过渡区特别容易受到群落组成快速变化的影响（Hupy＆Yansa，2009；Nelson＆胡，  2008；怀尔斯，  2023；威廉姆斯等人，  2009）。

山毛榉的薄树皮使其容易受到火灾的损害，而浅根系使其也容易受到土壤湿度变化的影响（Tubbs＆Houston，  1990）。在中世纪气候异常期间（1.05–0.6 ka BP），密歇根州北部一些地点的山毛榉数量减少与火灾频率增加有关（Booth, Jackson, et al.,  2012），但在斯派塞湖，基本上没有一致的关系山毛榉丰度与火灾之间的关系（Wang et al.,  2016）。在某些地点，火势可以稳定植被组成。例如，在明尼苏达州的大森林中，全新世晚期从橡树林地到湿地森林的转变仅发生在由于存在天然防火带而没有火灾的地方（Calder，  2016；Grimm，  1984）。干旱也可能导致山毛榉数量减少，但斯派塞湖记录缺乏可靠的古水文学指标（Wang et al.,  2016）。需要更独立的古水文学指标，例如过去的湖泊水位变化（Digerfeldt et al.,  1992；Pribyl & S human,  2014）或区域水分平衡的其他指标，以进一步评估山毛榉衰退与水文气候历史之间的关系（Booth,  2008；亚当斯等人，  2015）。山毛榉丰度的变化也可能受到温度变化的影响，温度变化似乎在全新世期间影响了该地区（Puleo 等人，  2020 年；S human 等人，  2023 年）和现代森林中已知的病原体爆发。例如，一些研究将全新世晚期新英格兰和五大湖地区山毛榉的减少归因于小冰期变冷（Fuller 等，  1998；Gajewski，  1987），而山毛榉树皮病意外传入美国1890 年代，山毛榉毁坏了许多山毛榉森林，死亡率达 50%，侵染率达 80% 至 95%（Beckman 等人，  2021 年；Stephanson 和 Ribarik Coe，  2017 年）。

即使有水文气候变化的独立代理，湖泊和陆地生态系统之间潜在的不同敏感性和响应时间仍然是不确定性的来源。一种相对较新的方法是对 C 化石花粉粒进行稳定碳同位素 (δ ¹³ C) 分析，它整合了碳和水通量之间的平衡，从而有望推断植被对气候强迫的生理敏感性和响应时间的变化。_{3 个}工厂（Griener 等人，  2013 年；Jahren，  2004 年；Loader & Hemming，  2004 年；Nelson，  2012 年）。 δ ¹³ C 值提供了内在水分利用效率（iWUE）的季节性综合信号，即水的光合作用和气孔导度的比率（Farquhar 等，  1989）。在用水策略方面，相对于水损失优先考虑碳增益的植物具有较低的 iWUE 和 δ ¹³ C 值，而相对于碳增益优先考虑节水的植物具有较高的 iWUE 和 δ ¹³ C 值（Bacon，  2009；Farquhar） & Sharkey,  1982）和干燥条件往往有利于后者（Sperry et al.,  2017）。大气 CO ₂、大气污染和气候（温度和降水）已被证明在历史记录中通过对光合作用和/或气孔导度的影响来影响植物 iWUE（Mathias 等，  2023），尽管前两个因素不太可能在工业革命之前的全新世期间，iWUE 受到了重大控制。花粉 δ ¹³ C 值（δ ¹³ C_花粉）代表 iWUE 中从景观到种群水平的变化，因为必须组合多个花粉粒才能产生足够的碳用于 δ ¹³ C 分析，并且每个沉积物样品包含来自花粉内许多单个植物的花粉粒源半径约为数十公里（Prentice，  1988）。来自特定类群的δ ¹³ C数据可以与花粉组合数据相结合，以评估iWUE的变化对单个类群丰度变化和群落组成变化的影响（Griener等，  2013）。

在这里，我们寻求通过印第安纳州故事湖的新的、日期明确的多代理记录，更好地了解五大湖南部地区山毛榉丰度快速变化的驱动因素。该记录包括同一核心的过去植被成分（花粉）、山毛榉 iWUE（δ ¹³ C_山毛榉）和火势（木炭）的代理。将 Story 记录与密歇根州附近拉文湖最近的湖面重建结果进行比较（Ray-Cozzens，2022），作为过去水文气候变化的独立指标。我们还将 Story 的山毛榉变异历史与 Story Lake 15 公里范围内的两个附近地点（Appleman 和 Pretty Lakes）以及位于以西 120 公里处的一个更远的湖泊（Spicer Lake）进行了比较。这些比较使我们能够检验关于景观与区域尺度植被变化同步性的假设，从而评估局部尺度植被-火灾-气候反馈和区域尺度气候驱动因素对南部中生树种群快速变化的相对重要性五大湖地区。