使用互相映射的现实与虚拟场景对生态位模型适用性的比较研究

Ecological niche models (ENMs) have become the primary method for quantifying species-environment relationships, and are now applied broadly in the context of species distribution patterns, invasive species dispersal, patterns of disease transmission, and global climate change. With more widespread use, methodological challenges are increasingly apparent. For example, a recent paper in SCIENCE demonstrated that blind selection of ENM algorithm and, uninformed decisions of parameters values lead to compounded errors in results. This illustrates the importance of applying a modeling framework that reflects the data and hypotheses being tested. Conversely, often the modeling framework is chosen based on ease of application or popular consensus of what works "best". In the end, poor model preparation will provide poor results. The treatment of ENM as a black box is thus a significant issue that has slowed the progress of development and application of this technique. Utilization of multiple ENM algorithms and different parameter settings can help illuminate this issue, however, few researchers make these comparisons. This may be due to the confounding effects of differences in, and interactions among, the statistical methods, species traits, data characteristics, and accuracy metrics considered. .This proposal employs simulation data to evaluate model conceptualization and implementation in a virtual world where the “truth” is known. We will develop three platforms to improve ease of ENM comparison and evaluation: one that generates a virtual world for simulation studies, one scientific-workflow-based tool that evaluates model performance, and one that calculates the similarity between virtual species and real datasets. The virtual world includes an environmental space, that mirrors the spatial distribution of environments on Earth, and virtual species each characterized by an eigen matrix and eigen vector, and expressed as an ellipsoid in environmental space. Changing the shape and placement of ellipsoids in environmental space will generate multiple virtual species with variable traits such as niche breadth and sampling density and bias. As an example, each virtual species will be modeled with 12 different ENM algorithms with varying parameters. Nine thresholding techniques will then be applied to all model outputs and model performance evaluated under the BAM framework, using measures of sensitivity, specificity, and the true skill statistic..Our simulation study will systematically test the effects of ENM algorithm choice, model parameters, and post-processing procedures on analysis results. This work provides an important baseline and guide for users that will improve the fidelity of ENM application. Using these platforms to compare model results between simulated and real data, in combination with traditional evaluation indices, will enable researchers to choose the most appropriate model and parameters for the specific data and research question.

物种的生态位是生态学的基础问题，生态位模型是利用数学工具估计生态位的重要方法，也是生物地理学重要的研究方向，广泛用于保护生物学、进化生物学等多个研究领域。但生态位模型选择的盲从性导致很多基于真实数据的研究工作存在理论缺陷，从而阻碍了该领域的发展。因此，模型的评估和最优化选择方法成为本领域亟待解决的问题之一。.本申请在生物地理学相关概念的基础上，通过对真实研究情景的分析、归纳和总结后，采用虚拟场景技术重现这些情景，并将该虚拟场景应用于生态位模型，在具有真实基础生态位的条件下，根据不同的应用场景准确的分析和比较模型结果，评估模型在不同场景下的性能，探讨模型的适用范围。.通过对具有不同特征的虚拟场景的评估，及与真实应用的比较和分析，本项研究结果将为研究者提供生态位模型研究的重要基础性数据，并由此为每个应用场景给出最佳的备选模型、参数设置及评价指标，以引导研究者正确的使用模型分析相关生物学问题。

物种的生态位指的是物种生存所需的环境条件的组合，生态学的基础问题，而生态位模型是利用数学工具建立物种分布区与所需环境条件的数学关系，是估计生态位的重要方法，也是生物地理学重要的研究方向，广泛用于保护生物学、进化生物学等多个研究领域。生态位模型选择的盲目性导致很多基于真实数据的研究工作存在理论缺陷，从而阻碍了该领域的发展。因此，模型的评估和最优化选择方法成为本领域亟待解决的问题之一。.本研究在生物地理学相关概念的基础上，首先通过对真实研究情景的分析、归纳和总结后，利用虚拟场景技术模拟重现了这些情景，并将该虚拟场景应用于已知的生态位模型，在具有真实基础生态位的条件下，根据不同的应用场景准确的分析和比较模型结果，评估了模型在不同场景下的性能，探讨了模型的适用范围等问题。.在此理论基础上，本项目利用虚拟场景模拟了物种分布在动态变化的环境下的变化情况，讨论了物种的形成与灭绝与环境变化的关系，物种的环境生态位宽度的纬度梯度格局的形成原因，物种多样性的纬度梯度格局形成原因以及物种多样性的海拔梯度格局的形成原因。在将模拟的结果与真实数据相比较的过程中，我们发现尽管大部分地区，模拟结果和真实数据匹配度较高，但在少部分地区模拟结果无法与真实数据相匹配。在排除模型问题后，我们发现这些不匹配是由于真实数据采样偏差造成的，因此，我们又对常用的生物多样性数据库，如GBIF，OBIS，IUCN分布图，eBird数据库等进行逐一分析，找出其数据的缺陷和采样偏差，并讨论这些真实数据的缺陷可能导致的生物多性评估方面的问题。.通过对具有不同特征的虚拟场景的评估，及与真实应用的比较和分析，本项研究取得了一系列的成果，在生态学主流期刊Nature Ecology & Evolution, Nature GeoScience, Global Ecology and Biogeography, Ecography等杂志上发表系列文章，这些结果结果为研究者提供生态位模型研究的重要基础性数据和方法指导，也为更好的进行生物多样性评价提供理论支持。