1 INTRODUCTION

Large herbivorous mammals (hereafter, ‘megafauna’) have strong interactions with vegetation. Megafauna consume vegetation in large quantities, thereby altering plant biomass, community composition and structure (Pringle et al., 2023). Megafauna also cause physical disturbance to ecosystems by trampling, debarking and rooting (Pringle et al., 2023). Due to these processes, megafauna play an important role in creating and maintaining open areas. This is the case in the temperate forest biome, where closed woody vegetation would otherwise dominate under mild climatic conditions (Svenning, 2002). Today, megafauna assemblages are a fraction of their past size (Smith et al., 2018). Larger species existed in much greater diversity and abundance before the arrival of modern humans ~50,000 years ago (Bergman et al., 2023; Smith et al., 2018). The extinction of so many megafauna species likely had cascading effects on the abundance, structure and composition of vegetation communities (Bakker et al., 2016; Svenning et al., 2024). Extant smaller species such as Capreolus capreolus (roe deer), Cervus elaphus (red deer) and Sus scrofa (wild boar) are unlikely to have produced functionally equivalent vegetation interactions (Pringle et al., 2023). In the European temperate forest biome, pollen-based reconstructions have revealed high percentages of open and light-woodland vegetation before modern humans (Pearce et al., 2023). Climate and other environmental processes, such as soil moisture availability (le Roux et al., 2013) and degree of continentality (Giesecke et al., 2008), may have influenced vegetation patterns at this time. However, it is likely that megafauna assemblages, and their associated disturbance regimes, had a major influence on vegetation structure, composition and species interactions (Davoli et al., 2023; Svenning et al., 2024).

Many trees and shrubs depend on disturbance for reproduction, to reduce competition from other taxa and to provide open, sunlit conditions (Vera, 2000). However, different disturbance mechanisms can have diverging effects. For example, Pinus (pine), Betula (birch) and Salix (willow) are positively related to fire events (Molinari et al., 2020), whereas Quercus (oak) and Corylus (hazel) are fire-tolerant but equally benefit from canopy openings driven by fire or large grazers (Molinari et al., 2020; Vera, 2000). Meanwhile, Taxus (yew) is sensitive to fire but can thrive in grazed systems (Busing et al., 1995; Omarova & Asadulaev, 2016). Evidence suggests that intense fire regimes may not have been widespread in Europe before the arrival of modern humans (Milner et al., 2016; Sandom, Faurby, et al., 2014). Conversely, the megafauna-rich conditions of past European landscapes may have promoted taxa such as Quercus, Corylus and Taxus (Pearce et al., 2023; Svenning & Magård, 1999).

Deciduous Quercus (deciduous oaks, with Quercus robur and Q. petraea being the most widespread and the only species in northern parts of Europe; hereafter ‘Quercus’) is a light-demanding taxon that thrives in disturbed, transitional habitats (Bobiec et al., 2018). Quercus seed dispersal and recruitment occur mostly in dynamic, mosaic landscapes; the taxon fails to regenerate under dense canopies in modern European forests (Bobiec et al., 2018). Abundant Corylus (hazel; largely Corylus avellana: common hazel; hereafter ‘Corylus’) is usually indicative of scrub woodland subject to ongoing disturbance (Vera, 2000). Corylus may thrive under the canopy of lightly shading trees such as Quercus and Fraxinus excelsior (common ash) (Coppins & Coppins, 2003); however, Corylus seedlings require high-light conditions and Corylus pollen production is enhanced in open and sunny sites (Bégeot, 1998; Vera, 2000). As a result, it is considered a ready coloniser of open landscapes (Kollmann & Schill, 1996; Vera, 2000) and fails to regenerate under a dense canopy (Vera, 2000). Taxus baccata (common yew; hereafter ‘Taxus’) is usually considered a shade-tolerant taxon, but also thrives in full sun (Linares, 2013). However, in modern forests, Taxus depends on canopy openings for reproduction, regeneration, growth and survival and declines under a dense canopy of shading trees (Linares, 2013; Svenning & Magård, 1999). Taxus requires substantial light for optimal growth, experiencing significant reductions in growth rates, biomass allocation and reproductive success under shaded conditions (Iszkuło et al., 2012; Perrin & Mitchell, 2013). Limited light availability in dense canopies, dominated by faster- and taller-growing species such as Abies alba (silver fir) and Fagus sylvatica (beech), hampers Taxus growth, survival and regeneration (Iszkuło et al., 2012; Perrin & Mitchell, 2013). Consequently, Taxus populations decline in dense forests where light is scarce (Iszkuło et al., 2012; Perrin & Mitchell, 2013). Additionally, Taxus is highly toxic to many large herbivores, except for deer and possibly certain other ruminants (Cortinovis & Caloni, 2015; Thomas & Polwart, 2003) and therefore may outcompete species more susceptible to browsing (Dhar et al., 2007; García & Ramón Obeso, 2003). On the other hand, Taxus is vulnerable to fire due to its thin bark and difficulty regenerating, particularly in dry environments (Mola et al., 2014; Thomas & Polwart, 2003) and its slow growth and low seedling recruitment further delay population recovery post-fire (Thomas & Polwart, 2003). Therefore, Taxus would likely have been favoured in past ecosystems with high levels of herbivory provided by diverse megafauna (Davoli et al., 2023) but would have been disfavoured by frequent fires.

Quantifying vegetation communities before and after the late-Quaternary megafauna extinctions is challenging. The extinctions occurred with the arrival of Homo sapiens ~50,000 BP, during the last glacial period (Bergman et al., 2023; Svenning et al., 2024), when forests and their tree species had strongly reduced distributions and abundances due to the cold, dry climate (Svenning et al., 2008; Willis & van Andel, 2004). Modern conservation practices are increasingly focused on restoring ecological function (Perino et al., 2019) and the role of long-term data (e.g. on functional baselines) has been recognised for almost 20 years (Willis et al., 2005, 2010). Pre-degradation baselines can provide unique ecological insights into systems before human-induced degradation, supplying modern restoration efforts with important context. However, we note that reconciling these insights with modern restoration efforts requires acknowledging contemporary ecological complexities, such as changing climate regimes and other anthropogenic influences, such as wildfire initiation, which may necessitate additional considerations beyond pre-degradation baselines alone. To establish a pre-degradation baseline for current and future systems, warm interglacial conditions are most relevant. The Last Interglacial in Europe (Eemian; ~129,000–116,000 BP) is the most recent interglacial period prior to the Holocene. It occurred before the late-Quaternary megafauna extinctions and had a faunal structure broadly comparable to the preceding 10 million years or more (Blanco et al., 2021; Croitor & Brugal, 2010). This period is geologically recent enough to limit evolutionary changes, as most extent species were already present (Van Kolfschoten, 2000). Finally, the Last Interglacial is climatically comparable to the warming Holocene, particularly in its central mesocratic phases, and has been described as a testbed for assessing environmental responses and climate feedbacks under conditions warmer than the pre-industrial benchmark (Kalis et al., 2003; Kühl et al., 2007; Salonen et al., 2018). Therefore, the Last Interglacial presents a long-term representative baseline for vegetation structure and dynamics under warm conditions prior to the late-Quaternary faunal downgrading (Smith et al., 2018).

Pollen-based vegetation reconstruction using the REVEALS model (Sugita, 2007) has shown that Quercus and Corylus were present in high abundance during the Last Interglacial in Europe (Pearce et al., 2023). Studies using raw pollen data from the Last Interglacial have shown a Corylus peak of up to 50% (Rychel et al., 2014; Suchora et al., 2022). Past abundances of Quercus and Corylus are better known for the Holocene, where REVEALS modelling shows 1%–15% Quercus cover and 10%–15% mean Corylus cover at 8200–5700 BP (Githumbi et al., 2022; Serge et al., 2023). Taxus was present during the Last Interglacial to a varying degree (3%–15% in Central Europe; Malkiewicz, 2018; Schläfli et al., 2021). In the Holocene, Taxus pollen is infrequent and cover is considered low (2%–6%; Deforce & Bastiaens, 2007; Pérez Díaz et al., 2013). However, in both the Holocene and the Last Interglacial, Taxus cover has only been examined using raw pollen percentages, which are less robust than REVEALS modelling for reconstructing past cover (Anderson et al., 2006; Sugita, 2007).

In this study, we compare the abundances of Quercus, Corylus and Taxus to assess how their populations changed before and after the late-Quaternary megafauna extinctions in the temperate forest biome. We quantify percentage cover for each taxon in the Last Interglacial and early–mid-Holocene using the latest REVEALS reconstructions (Pearce et al., 2023; Serge et al., 2023). Whilst direct testing of megafauna impact on vegetation was unfeasible in this study, we assess the extent to which observed abundance differences relate to climatic variations and estimate the residual variance. We were specifically interested in answering the following research questions: (1) How does the percentage cover of Quercus, Corylus and Taxus differ between the Last Interglacial and the early–mid-Holocene (before the onset of agriculture)? (2) How far can these differences be explained by climatic differences between the two periods? Overall, we expected that Quercus, Corylus and Taxus would be less abundant in the early–mid-Holocene than in the Last Interglacial as they depend on disturbed, transitional habitats for regeneration (Bobiec et al., 2018; Coppins & Coppins, 2003; Svenning & Magård, 1999), although human-enhanced fire regimes (Milner et al., 2016) may have compensated for the megafauna decline in the case of Quercus and Corylus. Furthermore, if changes in the abundance of these taxa were caused by human-linked changes in megafauna herbivory (Svenning et al., 2024), we expect that climate variables will have poor predictive ability. Together, our research questions will shed light on the pre-degradation abundances of important European tree species and their potential drivers, with implications for restoration initiatives and forest management (Lindbladh et al., 2007; Palli et al., 2023).

中文翻译：

---

在大型动物群落第四纪晚期灭绝之前，欧洲温带林地中受干扰的乔木和灌木丰度较高

 1 引言

大型食草哺乳动物（以下简称“巨型动物”）与植被有很强的互动。巨型动物大量消耗植被，从而改变植物生物量、群落组成和结构（Pringle 等人，2023 年）。巨型动物还通过践踏、剥皮和生根对生态系统造成物理干扰（Pringle 等人，2023 年）。由于这些过程，大型动物在创建和维护开放区域方面发挥着重要作用。温带森林生物群落就是这种情况，在温和的气候条件下，封闭的木本植被将占主导地位（Svenning，2002 年）。今天，巨型动物群落的规模只是过去的一小部分（Smith et al.， 2018）。在 ~50,000 年前现代人类到来之前，较大的物种以更大的多样性和丰度存在（Bergman 等人，2023 年;Smith et al.， 2018）。如此多的巨型动物物种的灭绝可能对植被群落的丰度、结构和组成产生了级联效应（Bakker 等人，2016 年;Svenning等人，2024 年）。现存的较小物种，如 Capreolus capreolus（狍子）、Cervus elaphus（马鹿）和 Sus scrofa（野猪）不太可能产生功能等效的植被相互作用（Pringle et al.， 2023）。在欧洲温带森林生物群落中，基于花粉的重建揭示了现代人类之前开阔和浅色林地植被的比例很高（Pearce 等人，2023 年）。气候和其他环境过程，例如土壤水分可用性（le Roux 等人，2013 年）和大陆性程度（Giesecke 等人）。， 2008） 可能影响了当时的植被模式。然而，大型动物群落及其相关的干扰机制很可能对植被结构、组成和物种相互作用产生了重大影响（Davoli 等人，2023 年;Svenning等人，2024 年）。

许多树木和灌木依靠干扰进行繁殖，以减少来自其他分类群的竞争，并提供开阔、阳光充足的条件（Vera， 2000）。然而，不同的干扰机制可能会产生不同的影响。例如，松树（松树）、桦树（桦树）和柳树（柳树）与火灾事件呈正相关（Molinari et al.， 2020），而栎树（橡树）和 Corylus（榛树）具有耐火性，但同样受益于火或大型食草动物驱动的树冠开口（Molinari et al.， 2020;Vera，2000 年）。同时，红豆杉 （yew） 对火很敏感，但可以在放牧系统中茁壮成长（Busing et al.， 1995;Omarova & Asadulaev，2016 年）。有证据表明，在现代人类到来之前，强烈的火灾状态可能在欧洲并不普遍（Milner 等人，2016 年;Sandom、Faurby 等人，2014 年）。相反，过去欧洲景观中大型动物丰富的条件可能促进了 Quercus、Corylus 和 Taxus 等分类群的出现（Pearce 等人，2023 年;Svenning & Magård，1999 年）。

落叶栎（落叶橡树，Quercus robur 和 Q. petraea 是欧洲北部分布最广且唯一的物种;以下简称“栎”）是一种要求光照的分类单元，在受干扰的过渡栖息地中茁壮成长（Bobiec 等人，2018 年）。栎树种子的传播和补充主要发生在动态的马赛克景观中;该分类群在现代欧洲森林的茂密树冠下无法再生（Bobiec et al.， 2018）。丰富的 Corylus（榛子;主要是 Corylus avellana：普通榛子;以下简称“Corylus”）通常表示灌木林地受到持续干扰（Vera，2000 年）。 Corylus可能在轻度遮荫的树木的树冠下茁壮成长，如Quercus和Fraxinus excelsior（普通灰烬（Coppins & Coppins，2003年）;然而，Corylus 幼苗需要强光条件，并且在开阔和阳光充足的地方，Corylus 花粉的产生得到增强（Bégeot， 1998;Vera，2000 年）。因此，它被认为是开放景观的现成殖民者（Kollmann & Schill， 1996;Vera， 2000），并且无法在茂密的树冠下再生（Vera， 2000）。红豆杉（普通红豆杉;以下简称“红豆杉”）通常被认为是一种耐阴的分类群，但也在充足的阳光下茁壮成长（Linares，2013 年）。 然而，在现代森林中，红豆杉依靠树冠开口进行繁殖、再生、生长和生存，并在茂密的遮荫树冠下下降（Linares，2013 年; Svenning & Magård，1999 年）。 红豆杉需要大量的光线才能实现最佳生长，在阴凉条件下，生长速率、生物量分配和繁殖成功率会显著降低（Iszkuło等人，2012 年;Perrin & Mitchell，2013 年）。在茂密的树冠中，以生长更快和更高的物种为主，如 Abies alba（银冷杉）和 Fagus sylvatica（山毛榉），光线可用性有限，阻碍了红豆杉的生长、生存和再生（Iszkuło 等人，2012 年;Perrin & Mitchell，2013 年）。因此，在光线稀缺的茂密森林中，红豆杉种群数量下降（Iszkuło et al.， 2012;Perrin & Mitchell，2013 年）。此外，红豆杉对许多大型食草动物具有剧毒，除了鹿和可能的某些其他反刍动物（Cortinovis & Caloni，2015;Thomas & Polwart， 2003），因此可能胜过更易被浏览的物种（Dhar et al.， 2007;García & Ramón Obeso， 2003）。另一方面，红豆杉由于其薄的树皮和难以再生而容易受到火的影响，尤其是在干燥的环境中（Mola et al.， 2014;Thomas & Polwart， 2003）及其缓慢的生长和低幼苗的补充进一步延迟了火灾后的种群恢复（Thomas & Polwart， 2003）。因此，红豆杉在过去由各种大型动物提供高水平食草性的生态系统中可能会受到青睐（Davoli 等人，2023 年），但会因频繁的火灾而受到青睐。

量化第四纪晚期巨型动物灭绝前后的植被群落具有挑战性。灭绝发生在末次冰河期 ~50,000 BP 的智人到来（Bergman 等人，2023 年;Svenning et al.， 2024），当时由于寒冷干燥的气候，森林及其树种的分布和丰度大大减少（Svenning et al.， 2008;Willis & van Andel，2004年）。现代保护实践越来越注重恢复生态功能（Perino et al.， 2019），长期数据的作用（例如功能基线）已经得到认可近 20 年（Willis et al.， 2005， 2010）。退化前基线可以在人为退化之前为系统提供独特的生态见解，为现代恢复工作提供重要的背景。然而，我们注意到，将这些见解与现代恢复工作相协调需要承认当代生态复杂性，例如不断变化的气候状况和其他人为影响，例如野火的爆发，这可能需要在单独的退化前基线之外进行额外的考虑。为了为当前和未来的系统建立退化前基线，温暖的间冰期条件是最相关的。欧洲末次间冰期（艾米亚时期;~129,000–116,000 BP）是全新世之前最近的间冰期。它发生在第四纪晚期巨型动物灭绝之前，其动物群结构与之前的 1000 万年或更长时间大致相当（Blanco 等人，2021 年;Croitor & Brugal， 2010）。 这个时期在地质学上足够近，足以限制进化变化，因为大多数物种已经存在（Van Kolfschoten， 2000）。最后，末次间冰期在气候上与变暖的全新世相当，特别是在其中央中层阶段，并被描述为在比工业化前基准更温暖的条件下评估环境响应和气候反馈的试验台（Kalis 等人，2003 年;Kühl 等人，2007 年;Salonen et al.， 2018）。因此，末次间冰期为第四纪晚期动物群降级之前温暖条件下的植被结构和动态提供了长期代表性基线（Smith et al.， 2018）。

使用 REVEALS 模型（Sugita，2007 年）进行的基于花粉的植被重建表明，在欧洲末次间冰期期间，栎属和 Corylus 大量存在（Pearce 等人，2023 年）。使用末次间冰期原始花粉数据的研究显示，Corylus 峰值高达 50%（Rychel 等人，2014 年;Suchora et al.， 2022）。栎属和栎属的过去丰度以全新世而闻名，其中 REVEAL 模型显示，在 8200-5700 BP 处，1%-15% 的栎属覆盖和 10%-15% 的平均栎属覆盖（Githumbi 等人，2022 年;Serge et al.， 2023）。红豆杉在末次间冰期存在不同程度（中欧为 3%-15%;Malkiewicz，2018 年;Schläfli等人，2021 年）。在全新世，红豆杉花粉很少见，覆盖率被认为很低（2%-6%;Deforce & Bastiaens， 2007;Pérez Díaz et al.， 2013）。然而，在全新世和末次间冰期，红豆杉覆盖度都只使用原始花粉百分比进行检查，这不如 REVEAL 模型重建过去的覆盖物那么可靠（Anderson 等人，2006 年;Sugita，2007 年）。

在这项研究中，我们比较了 Quercus、Corylus 和 Taxus 的丰度，以评估它们在温带森林生物群落中第四纪晚期巨型动物灭绝前后的种群变化。我们使用最新的 REVEAL 重建量化了末次间冰期和全新世早期中期每个分类群的覆盖率（Pearce等人，2023 年;Serge et al.， 2023）。虽然在本研究中直接测试大型动物对植被的影响是不可行的，但我们评估了观察到的丰度差异与气候变化的关系程度，并估计残余方差。我们对回答以下研究问题特别感兴趣：（1） 栎属、科里鲁斯和红豆杉的百分比覆盖率在末次间冰期和全新世早期中期（农业开始之前）之间有何不同？（2） 这两个时期之间的气候差异可以在多大程度上解释这些差异？总体而言，我们预计栎属、Corylus 和 Taxus 在全新世早期中期的数量将低于末次间冰期，因为它们依赖于受干扰的过渡栖息地进行再生（Bobiec 等人，2018 年;Coppins & Coppins， 2003;Svenning & Magård， 1999），尽管人类增强的火灾机制（Milner等人，2016）可能已经弥补了Quercus和Corylus等大型动物的减少。此外，如果这些分类群丰度的变化是由人类相关的大型动物食草性变化引起的（Svenning et al.， 2024），我们预计气候变量的预测能力会很差。 我们的研究问题将共同阐明欧洲重要树种的退化前丰度及其潜在驱动因素，对恢复计划和森林管理产生影响（Lindbladh 等人，2007 年;Palli等人，2023 年）。