高维空间数据的不完全检测联合种群分布模型
Joint species distribution models with imperfect detection for high-dimensional spatial data
摘要：
确定物种和群落的空间分布是生态学和保护的关键目标。联合物种分布模型使用多物种检测-非检测数据来估计物种和群落分布。这些数据的分析是复杂的物种之间的残留相关性，不完善的检测和空间自相关。虽然存在方法来适应这些复杂性中的每一个，但文献中很少有同时解决和探索所有三个复杂性的例子。在这里，我们开发了一个空间因子多物种占用模型，明确占物种的相关性，不完善的检测和空间自相关。所提出的模型使用空间因子降维方法和最近邻高斯过程来确保具有大量物种（例如，> 100）和空间位置（例如，100，000）。我们将所提出的模型性能与五种候选模型进行比较，每种模型都解决了三种复杂性的一个子集。我们在spOccupational软件中实现了所提出的模型和竞争模型，旨在通过一个可访问的，文档齐全的开源R包来促进应用程序。通过模拟，我们发现忽略这三个复杂性会导致模型预测性能低下。通过对美国大陆98种鸟类的案例研究，空间因子多物种占用模型在候选模型中具有最高的预测性能。我们提出的框架，连同其在spOccupational的实现，作为一个用户友好的工具，以了解物种分布和生物多样性指标的空间变化，同时解决常见的复杂性，在多物种检测-非检测数据。
英文摘要：
Determining spatial distributions of species and communities are key objectives of ecology and conservation. Joint species distribution models use multi-species detection-nondetection data to estimate species and community distributions. The analysis of such data is complicated by residual correlations between species, imperfect detection, and spatial autocorrelation. While methods exist to accommodate each of these complexities, there are few examples in the literature that address and explore all three complexities simultaneously. Here we developed a spatial factor multi-species occupancy model to explicitly account for species correlations, imperfect detection, and spatial autocorrelation. The proposed model uses a spatial factor dimension reduction approach and Nearest Neighbor Gaussian Processes to ensure computational efficiency for data sets with both a large number of species (e.g., > 100) and spatial locations (e.g., 100,000). We compare the proposed model performance to five candidate models, each addressing a subset of the three complexities. We implemented the proposed and competing models in the spOccupancy software, designed to facilitate application via an accessible, well-documented, and open-source R package. Using simulations, we found ignoring the three complexities when present leads to inferior model predictive performance. Using a case study on 98 bird species across the continental US, the spatial factor multi-species occupancy model had the highest predictive performance among the candidate models. Our proposed framework, together with its implementation in spOccupancy, serves as a user-friendly tool to understand spatial variation in species distributions and biodiversity metrics while addressing common complexities in multi-species detection-nondetection data.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
