Predicting invasive species SPREAD by integrating big data on the environment and species-occurrence using machine-learning modelingproject period: 4 25 2025 through 4 24 2027PURPOSE: knowing the…

This Project Grant award from the U.S. Geological Survey's Assistance to State Water Resources Research Institutes program (CFDA 15.805) provides $261,500 to Montana State University (MSU) for a 4/25/2025 - 4/24/2027 project period. The project aims to develop and test machine learning models that can accurately predict locations at high risk of aquatic invasive species (AIS) invasion in the Upper Mississippi River System (UMRS). Key activities include integrating environmental and species-occurrence data, improving web tool outputs to visualize invasion risk, collecting environmental DNA samples to quantify AIS and biodiversity, and collaborating with water resource managers to continuously improve the tools. The goal is to provide user-friendly modeling capabilities that can enhance AIS monitoring, prevention, and management efforts by identifying high-risk locations. Sub-awards may be issued to conduct eDNA sampling and metabarcoding analysis from 20 additional waterbodies.

Predicting invasive species SPREAD by integrating big data on the environment and species-occurrence using machine-learning modelingproject period: 4 25 2025 through 4 24 2027PURPOSE: knowing the highest-risk locations of invasion for multiple AIS will improve management in many ways including monitoring for early detection, establishing boat wash check stations, and posting signage to educate the public. The project will improve managers ability to protect waters from AIS through the development of user-friendly modeling tools that accurately predict high-risk locations in the umrs and beyond. This web tool will visualize invasion hotspots on easy-to-read maps, to improve understanding of invasion risk by managers, the public, policymakers, and students. This web tool will report the physical, chemical, and biological drivers of AIS occurrence. Activities: the project will (1) develop and test code that incorporates finer-spatial scale environmental data (remote-sensing data) to facilitate combining environmental data and species-occurrence data, and assess accuracy and uncertainty in machine-learning models predicting AIS occurrence, (2) write code to improve web tool outputs including heat maps and visualizations of model uncertainty and covariate contributions to model performance, (3) develop video tutorials explaining how to install and run software, (4) zoom-meet managers monthly (and at two workshops) to continuously integrate feedback to improve tools and teach managers to use modeling, (5) collect samples for edna-metabarcoding from 20 new water bodies (of AIS concern) to teach managers edna sampling and add to the 100 water bodies meta-barcoded by USGS collaborators in the umrs, (6) conduct metabarcoding for the 20 waterbodies to quantify presence absence of all plants (native and invasive) and estimate species-diversity (BIODIVERSITY) for modeling, (7) quantify reproducibility and effects of metabarcoding sequencing data-filtering choices on downstream inferences (BIODIVERSITY, hotspot maps) to provide best practices for researchers, (8) optimize habitat suitability models with versus without edna data, including novel metabarcoding-based model covariates of BIODIVERSITY to improve understanding of biological, physical, chemical drivers of AIS SPREAD, (9) validate web tool accuracy and uncertainty, and evaluate usefulness by end-user managers, (10) transfer technology and publish manuscript to make the deliverables from this project widely available. - deliverables and outcomes: models and heatmaps produced for and by managers to inform the development of plans for early detection and rapid response and to guide the allocation of limited resources for these activities. A web tool with scripts to empower managers to easily repeat model construction (e.g., semi-annually) using new data or improved predictor variables. Advancement of knowledge of the relationship between the physical, chemical, and biological drivers of occurrence (SPREAD) for seven of the most problematic AIS in the umrs: eurasian milfoil, curly leaf pondweed, starry stonewort, reed canary grass, zebra quagga mussels, and faucet snail. A improving AIS management by understanding drivers of SPREAD. 3 standardized and validated metabarcoding bioinformatic pipeline with best practices for (and quantified effects of) data filtering choices on inference about species richness (BIODIVERSITY), habitat suitability modeling, and relative contributions of physical, chemical, and biological predictor variables of AIS occurrence. Publications and new improved AIS management plans. - beneficiaries: AIS managers from multiple states, tribes, and counties.