DISSERTATION RESEARCH: Accounting for spatial autocorrelation in species distribution models using a Bayesian framework: consequences for predictions across space and time

论文研究：使用贝叶斯框架解释物种分布模型中的空间自相关：跨空间和时间预测的后果

• 批准号: 1404187
• 负责人: Jason Knouft
• 金额: $ 1.96万
• 依托单位: Saint Louis University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-08-01 至 2015-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1404187&HistoricalAwards=false
• 关键词: DISSERTATION; RESEARCH; Accounting; spatial; autocorrelation

Ecologists combine an understanding of current conditions with models to predict the effects of future environmental conditions on populations and communities. Many of the models used are incomplete, hindering these important predictions. This project improves predictions of where species will occur as environments change by including new data on the similarity, or degree of dependency, among observations that are closely associated in space. It will significantly advance fundamental research in ecology. Simultaneously, the research will result in a clearer general understanding of how species will respond to increasing habitat degradation, invasive species, disease, and climate change, thereby contributing directly to biodiversity conservation. The goal of this research is to investigate the relationship between stream flow variability, spatial relatedness, and fish species occurrence across sites within the Big River watershed in East Central Missouri while developing statistical techniques to account for the effects of spatial autocorrelation on species distribution predictions. Spatial autocorrelation is the positive association between the proximity of sample locations and the similarity of data at each location. It reduces the accuracy of current species distribution models. Stream fish assemblages at 50 sites across the watershed will be sampled and predicted based on spatial relatedness to other sampling sites and flow variability data estimated from high resolution in-stream depth gauges. The Big River watershed is a primary focus of collaborative conservation efforts by the Missouri Department of Conservation and The Nature Conservancy in Missouri. The data and results generated from this study will be of practical use by these agencies during their attempts to balance human activities and the conservation of biodiversity in this unique aquatic ecosystem. The project will significantly enhance ongoing dissertation research by providing field tests of model predictions, broadening and strengthening graduate student training.

生态学家将对当前条件的理解与模型相结合，以预测未来环境条件对人口和社区的影响。许多使用的模型是不完整的，阻碍了这些重要的预测。该项目通过纳入在空间中密切相关的观察中有关相似性或依赖程度的新数据，改进了对物种在环境变化时将出现在何处的预测。这将显著推进生态学的基础研究。同时，这项研究将使人们更清楚地了解物种如何应对日益严重的栖息地退化、入侵物种、疾病和气候变化，从而直接促进生物多样性的保护。本研究的目的是研究密苏里中东部大河流域各地点的流量变异性、空间相关性和鱼类发生之间的关系，同时开发统计技术来解释空间自相关对物种分布预测的影响。空间自相关是样本位置的接近性和每个位置数据的相似性之间的正相关。它降低了当前物种分布模型的准确性。根据与其他采样点的空间相关性和高分辨率流内深度计估计的流量变异性数据，将对流域内50个地点的溪流鱼类种群进行采样和预测。大河流域是密苏里州自然保护部和密苏里州自然保护协会合作保护工作的主要焦点。这项研究产生的数据和结果将在这些机构试图平衡人类活动和保护这一独特的水生生态系统中的生物多样性时具有实际用途。该项目将通过提供模型预测的实地测试，扩大和加强研究生培训，大大加强正在进行的论文研究。