EPSCOR Research Fellows: NSF: Advancing National Ecological Observatory Network-Enabled Science and Workforce Development at the University of Maine with Artificial Intelligence

Biodiversity is essential to human wellbeing and the functioning of the Earth’s ecosystems. The vision of the proposed fellowship is to leverage artificial intelligence (AI) to extract information related to life on Earth from imagery generated by the NSF-funded National Ecological Observatory Network (NEON). This extraction is intended to fuel scientific discovery related to biodiversity and nature’s contributions to people. This project will focus on automating extraction of information on beetles from images. Beetles occur in most habitats and are important components of ecosystems as they play key roles as pollinators of plants, predators of pests on plants and crops, and decomposers in soils. This project will provide information that will be helpful in better understanding why certain species of beetles occur in different areas and how they might respond to environmental change. Beyond these scientific impacts, the work will further improve the research capacity of the University of Maine System through curricula PI Record will develop as part of this effort and through PI Record’s mentoring of graduate and undergraduate students on research related to AI for biodiversity science. The proposed work will also provide enhanced infrastructure for research by funding the construction of a specialized imaging system, which if successful in processing samples could be deployed across NEON domains to aid in imaging beetle specimens at scale.

Intraspecific variation (ITV) of traits provides a currency for assessing the roles of abiotic and biotic processes in community assembly, as it reflects the mechanisms driving species occurrence and responses to change. However, a lack of information on multiple traits for many individuals across communities at a biogeographic extent limits the ability to fully understand what traits matter to community assembly, especially for animals. This project will fill this knowledge gap by providing information on ITV from National Ecological Observatory Network beetle specimens at a continental scale. This project will also contribute to larger global efforts to document insects given recent conservation concerns about wide-scale insect declines because the data products produced by this project will contribute to enhanced automated monitoring of insects because images taken and traits measured can also be used to identify specimens from images. The specific research objectives are: 1) to develop machine learning algorithms to extract morphological trait measurements from NEON beetle specimens, 2) to test the ability of these algorithms to automate measurements of traits from different types of images, and 3) to better understand which of the measured traits are most closely linked to beetle diversity across NEON. The host site (the Translational Data Analytics Institute at The Ohio State University which houses two NSF-funded centers, Imageomics and AI for Biodiversity Change) will build collaborations with PI Record and provide training for a graduate student from her research group. PI Record will develop curricula to share the knowledge gained from this fellowship with the next generation of workers in Maine, which will be sustained through her continued teaching.

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.

Award Number: 2429418
Principal Investigator: Sydne Record
Funds Obligated: $230,552
State: ME