Data from: Elephant damage, not fire or rainfall, explains mortality of overstorey trees in Serengeti

Generalizations about the drivers of tree demography in tropical savannahs continue to prove difficult because of the complex and dynamic interactions involved, and because multi-year datasets spanning meaningful gradients in potential drivers are lacking. Overstorey trees play disproportionate roles in the long-term dynamics and functioning of savannah ecosystems. Understanding demographic patterns in these trees is complicated by their resprouting ability after being top-killed and few studies have attempted to separate top-kill from true mortality events. We examined the interactive effects of fire frequency, mean annual precipitation and elephant herbivory on overstorey (>2m) tree mortality between 2009 and 2014 across 32 permanent vegetation plots in the Serengeti Ecosystem, Tanzania. Mean tree mortality over the study period was 0.28 ± 0.02 (0.07 yr−1). Among trees that were top-killed (19.1% of all individuals), 31.2% resprouted. Mortality was driven largely by elephant herbivory, though mortality rates varied considerably across space and tree species. Fire frequency and mean annual rainfall were weak predictors of tree mortality. Over the five year period, chronic elephant herbivory (i.e., repeated through time) was a stronger predictor of tree mortality than maximum elephant herbivory, suggesting that repeated, low-intensity damage from elephants was more important to mortality than acute, but infrequent, damage. Elephants disproportionately damaged certain tree species relative to their availability on the landscape, and the tolerance to damage differed by species as well. Acacia senegal was strongly selected and appeared to have low tolerance to damage when it occurred, resulting in a mortality rate of 0.736 ± 0.052 (0.234 yr−1) over the study period. Topkilling and resprouting rates varied across species, with Acacias generally having low resprout rates. Synthesis. Our study highlights the considerable role that chronic herbivory plays in structuring savannah tree populations, irrespective of prevailing fire and rainfall conditions, and provides important vegetative context to the dramatic recent declines and (in the case of Serengeti) increases in savannah elephant densities in sub-Saharan Africa.针对热带稀树草原（tropical savannahs）树木种群动态的驱动因子，学界迄今仍难以形成普适性认知，这一方面源于该系统内复杂且动态的相互作用，另一方面则因缺乏覆盖潜在驱动因子有效梯度的多年期数据集。 冠层乔木（overstorey trees）在稀树草原生态系统的长期动态与功能维持中，发挥着远超其丰度占比的关键作用。由于这类树木在遭受顶枯（top-kill，即地上组织完全死亡）后仍具备萌蘖能力，且鲜有研究尝试将顶枯事件与真正的植株死亡加以区分，因此解析其种群统计模式颇具挑战。 本研究以坦桑尼亚塞伦盖蒂生态系统内32个永久性植被样地为研究对象，分析了2009至2014年间火频度、年平均降水量与大象植食压力对高度>2米的冠层乔木死亡率的交互影响。研究期内乔木平均死亡率为0.28 ± 0.02（年均0.07 yr⁻¹）。在遭受顶枯的个体（占总个体数的19.1%）中，有31.2%成功萌蘖。 尽管乔木死亡率在空间分布与树种间存在显著差异，但整体上主要受大象植食压力驱动。火频度与年平均降水量对乔木死亡率的预测能力较弱。 在五年研究期内，持续性大象植食压力（即长期反复发生的取食干扰）对乔木死亡率的预测能力强于峰值植食压力，这表明大象反复施加的低强度损伤，相较于偶发的高强度急性损伤，对乔木死亡的影响更为显著。 大象对部分乔木树种的取食损伤强度与其在样地内的丰度不成正比，且不同树种对损伤的耐受能力也存在显著差异。阿拉伯胶金合欢（Acacia senegal）是大象优先取食的树种，且在遭受损伤后耐受能力较低，导致其研究期内死亡率达0.736 ± 0.052（年均0.234 yr⁻¹）。 不同树种的顶枯率与萌蘖率存在显著差异，金合欢属（Acacia）树种的萌蘖率普遍较低。 研究总结：本研究表明，无论区域火况与降水条件如何，持续性植食压力在塑造稀树草原乔木种群结构中发挥着重要调控作用；同时，该研究为撒哈拉以南非洲近期稀树草原大象种群的剧烈下降（塞伦盖蒂地区则为种群增长）现象，提供了重要的植被生态学背景解释。