EAGER: Scaling Up Plant Demographic Rates with Imagery from Unoccupied Aerial Systems

该奖项全部由《2021年美国救援计划法案》（公法117-2）资助。随着气候变化和入侵物种导致更大、更频繁的火灾，野火对美国西部的生态系统构成了越来越大的威胁。了解原生植物种群从野火中恢复的速度有多快（如果有的话），以及哪些地方没有出现先前优势物种的补充，将有助于恢复退化的生态系统。研究植物种群恢复最常用的方法是在田间小区或沿着50至100米的样带标记和测量单株植物。后勤方面的考虑限制了这些类型的野外测量的空间覆盖范围，导致多公里尺度的野火和地块级响应之间的不匹配。这里提出的研究将使用无人机来开发新的方法来测量大面积的植物种群恢复。这些方法将包括在航空图像中检测和区分单个植物的计算机算法，以及量化无人机检测到的植物生长、存活和繁殖的统计模型。这一定量框架将使研究小组能够研究一种生态上重要的优势植物物种如何在山间西部对大面积火灾和大面积燃烧区域内的大规模环境变化做出反应。该研究将用于预测具有重要生态意义的大艾属植物种群的恢复。研究结果将通过推进无人机技术来监测火灾后的恢复，从而帮助这一广阔地区的土地管理。来自无人机系统（UAS）的新型遥感数据可以通过提供足够高分辨率的图像来检测大空间范围内的单个植物，从而帮助植物人口统计学的空间模型。该项目将开发一个框架，通过将计算机视觉技术与分层贝叶斯模型相结合，从无人机图像推断植物的人口统计率。本研究将重点研究美国西部具有较高保护价值的大艾属植物——三叉戟蒿（Artemisia tridentata）的空间种群动态。由于野火摧毁了山艾树的栖息地，景观恢复取决于大山艾树能否重新定居受干扰的地区。然而，像大多数低体型植物一样，我们对大山艾树人口统计的理解几乎完全来自于米尺度的田间地块，限制了对大空间尺度的外推。该研究将开发一个可重复的工作流程来识别UAS图像中的单个植物，量化植物的人口统计率，同时考虑到航空图像中不完美的检测，并确定空间协变量对山艾树人口统计的影响。该研究将推进空间植物人口统计学的定量建模方法，并立即应用于濒危生态系统。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This award is funded in whole under the American Rescue Plan Act of 2021 (Public Law 117-2).Wildfires pose a growing threat to ecosystems in the American West, as climate change and invasive species promote larger and more frequent fires. Understanding how fast, if ever, native plant populations can recover from wildfire, and where recruitment of prior dominants is not occurring will aid efforts to restore degraded ecosystems. Plant population recovery is most often studied by marking and measuring individual plants in field plots or along 50 to 100 m transects. Logistical considerations limit the spatial coverage of these types of field measurements, resulting in a mismatch between multi-kilometer-scale wildfires and plot level responses. The research proposed here will use drones to develop new methods to measure plant population recovery over large areas. These methods will include computer algorithms to detect and distinguish individual plants in aerial imagery, and statistical models to quantify drone-detected plants' growth, survival, and reproduction. This quantitative framework will enable the research team to study how an ecologically important dominant plant species across the inter-mountain west is responding to wide spread fire and to large-scale environmental variation within the vast burned areas. The research will be applied to forecast population recovery of big sagebrush plants, a species with critical ecological importance. The results will aid land management across this vast region by advancing drone technology to monitor post-fire recovery. Novel remote sensing data from unoccupied aerial systems (UAS) could aid spatial models for plant demography by providing imagery with fine enough resolution to detect individual plants across large spatial extents. This project will develop a framework to infer plant demographic rates from UAS imagery by coupling computer vision techniques with hierarchical Bayesian models. The research will focus on spatial population dynamics of big sagebrush, Artemisia tridentata, a species with high conservation value in the American West. As wildfires decimate sagebrush habitat, landscape recovery depends on whether big sagebrush can recolonize disturbed areas. However, like most low-statured plants, our understanding of big sagebrush demography is almost entirely derived from meter-scale field plots, limiting extrapolation to large spatial scales. The research will develop a reproducible workflow to identify individual plants in UAS imagery, quantify demographic rates of plants while accounting for imperfect detection in aerial imagery, and determine the impacts of spatial covariates on sagebrush demography. This research will advance quantitative modeling approaches for spatial plant demography with immediate application to an imperiled ecosystem.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.