RESEARCH PRODUCT

Using previously collected information on weed risk and ecology of non-native plants in Hawaiʻi (Weed Risk Assessment Database), this project developed a screening system focused on wildfire risk by asking land managers to assign the relative fire risk for weeds naturalized in Hawai‘i. This particular expertise was used to train a machine learning model to rank hundreds of other species based on plant traits and other information from the scientific literature. The Weed Fire Risk Assessment Database assesses 360+ plant species for their fire-promoting traits and provides a computer model-generated fire risk score for each plant species, information that may also be useful to predict post-fire responses, since available fire regeneration characteristics of each plant species are also included. The Database is presented in several formats for convenience.