Web-based Decision Analytics For Mapping Host Species Distributions and Forecasting the Spread of Forest Pests and Pathogens

Pests and pathogens are a growing threat to farmland and forests around the globe. These biotic agents negatively affect biodiversity, productivity, and ecosystem services. Land managers, tasked to mitigate pest and pathogen impact, are turning to spatial decision support systems to forecast future spread and explore management scenarios. The Pest or Pathogen Spread Model (PoPS) is web-based interface that provides a flexible, generalized framework for modeling the spread of pests or pathogens across a landscape to support land-managers. One important factor underpinning accurate spread projections in this framework is the host species maps depicting the current distribution of plants capable of hosting a pest or pathogen. However, these spread models have often relied on generalized species data whose uncertainty is unknown, leading to reduced confidence in model results. Additionally, analysis is often needed to run over multiple spatial resolutions and extents. Therefore, there is a need to establish a data service that provides accurate and timely species distribution to the PoPS for reliable model results. We present a host species mapping web-interface that provides users with the ability to create, revise, and update host species maps as new data become available. This cloud-based system provides a graphical user interface for generating species data at a user-defined scale. The host species distribution maps are generated using repeat remote sensing observations in a Bayesian framework to provide estimates of certainty in species presence. These certainty values are thresholded using ROC analysis to produce binary maps of species presence and absence. The resulting maps are then validated using citizen science data from iNaturalist and Early Detection and Distribution Maps observations. Our web-interface provides a novel approach to generating species data across multiple scales, allowing users to easily create, examine, and use the data for modeling spread of pests and pathogens in forest ecosystems.