Ecological risk assessment and food webs: identifying ecosystem tipping points under multistress

预测生态系统对环境变化的反应受到其内在复杂性和缺乏长期高分辨率数据的阻碍。这一挑战尤其严峻的一个领域是生态风险评估，在这一领域，迫切需要从我们对化学品对个体生物体影响的了解扩展到对社区和生态系统进程的影响。预测模型可能有助于弥合这一知识差距。然而，由于系统的多样性和不同生态尺度上的生态系统功能的阵列，为绝大多数生态系统建立预测模型是极具挑战性的。即使在有可能建立这种模型的情况下，目前的努力仍然高度针对具体系统。因此，在ERA的背景下，模型的可实现的价值在于他们的潜力，以确定generalisable phenomenon.在生态时间序列分析领域，其中一个领域已经看到了一个快速发展，在过去的十年是预警信号（EWS）领域。这些方法通过观察到的系统时空动态变化来推断系统即将发生的状态变化，这可以作为弹性下降的症状（从扰动中快速恢复的能力），然后达到不可逆转的转变。这种方法是现象学的，旨在确定一般的，广泛适用的信号，而不考虑系统的基本动态或变化的外部驱动因素（如化学应激）。事实上，EWS已被确定在基于丰度的数据，例如，在人口丰度增加的自相关性;基于性状的数据，例如，下降的平均个体大小;和空间数据，例如，在连接的补丁的丰度增加的相关性。早期预警信号的普遍性给人们带来了希望，即它们可以作为一种确定优先次序的工具，以确定使系统特别容易受到化学应激源影响的特征（例如社区结构或特征组成），并确定风险评估应关注的触发点。此外，我们知道相对较少的EWS表现在生态群落和生态系统过程中，面对多重压力。对生物多样性社区中压力驱动的预警信号进行预测，需要一个多物种的建模框架，允许多种压力对生理、行为、生存和生殖产生影响。生物能量食物网模型（BEFW）就是这样一个模型。这个博士将使用食物网模型（特别是BEFW）来发展EWS理论，用于社区和生态系统水平的过程，其中多种化学压力影响多个物种。具体来说，学生将：* 评估使用这些信号评估即将发生的弹性损失的潜力。确定这些信号对压力源影响和社区结构的不同假设的鲁棒性。在丰度、性状和空间指标中确定恢复力损失的最大信息信号集。通过使用商业上可用的食物链模型（例如美国环保局的AQUATOX）模拟污染物影响，评估最佳信号集的一般性

Predicting the response of ecological systems to environmental change is hampered by their inherent complexity and a lack of long-term, high-resolution data. One area where this challenge is particularly acute is ecological risk assessment (ERA), where there is a pressing need to scale from our knowledge of chemical impacts on individual organisms to implications for community and ecosystem processes. Predictive models may serve to bridge this knowledge gap. However, building predictive models for the vast majority of ecological systems is extremely challenging because of system diversity and the array of ecosystem functions at different ecological scales. Even where such modelling is possible, current efforts remain highly system-specific. Thus, the realisable value of models in the context of ERA lies in their potential to identify generalisable phenomena.In the domain of ecological time series analysis, one area which has seen a rapid advancement in the last decade is the field of early warning signals (EWSs). These methods infer the approaching state-change of a system through observed changes in its spatio-temporal dynamics, which can act as symptoms of declining resilience-the capacity to recover quickly from perturbations-before reaching an irreversible transition. Such approaches are phenomenological, aiming to identify general, broadly applicable signals without considering the underlying dynamics of the system or the external drivers of change (e.g. chemical stressors). Indeed, EWSs have been identified in abundance-based data, e.g. increasing autocorrelation in the population abundance; trait-based data, e.g. declining mean body size of individuals; and spatial data, e.g. increased correlation of abundances in connected patches. The generalisability of early warning signals gives hope that they can act as a prioritisation tool to identify the characteristics that make systems particularly vulnerable to chemical stressors (e.g. community structure or trait composition) and identify trigger points that the risk assessment should focus on. Although this approach could be a valuable risk-assessment tool, it is not yet clear how to operationalise the use of such signals. Moreover, we know relatively little about how EWSs manifest in ecological communities and ecosystem processes in the face of multiple stressors. Generating predictions about stressor-driven early warning signals in biodiverse communities requires a modelling framework that is multi-species and allows multiple stressor impacts on physiology, behaviour, survival and reproduction. The bioenergetic food web model (BEFW) is one such model.This PhD will use the food web models (notably BEFW) to develop EWS theory for the community- and ecosystem-level processes where multiple chemical stressors are affecting multiple species. Specifically, the student will:*Evaluate the potential for using such signals for assessment of impending resilience loss.*Determine the robustness of these signals to different assumptions about stressor impacts and community structure.*Identify the set of maximally informative signals of resilience loss among abundance, trait and spatial metrics.*Assess the generality of the optimal set of signals by simulating pollutant impacts using commercially available food chain models (e.g. US EPA AQUATOX)