Empirical Critical Levels of Ozone for U.S. Tree Species and Their Uncertainties with Machine Learning

Exposure to ambient ozone concentrations can impact vegetation through an array of cascading effects. In particular, research has shown that tree species experience biomass loss and increased mortality due to ozone exposure. Recent U.S. Environmental Protection Agency (EPA) analysis (U.S. EPA, 2020) introduced a new causality determination for ecological effects of ozone on tree survival. Robust exposure-response functions have been developed for reduced growth and yield in trees, but the relationship between ozone exposure and tree survival has not been sufficiently investigated. The ozone exposure level at which a harmful effect occurs is referred to as a "critical level." Previously, we used machine learning (ML) to characterize the uncertainty of potential negative impacts on 108 tree species from atmospheric deposition of N and S. ML modeling allows representation of ecological processes with greater accuracy than other approaches and provides more flexibility in the form of a modeled relationship between deposition and tree growth and survival. In this work, we apply our methodology to understand the impact of ozone exposure on tree growth and survival, and describe the critical levels of individual tree species for ozone. Further, we describe the uncertainty of the critical levels. With increasing interest in ozone impacts on trees, there is a need for additional evidence to understand the relationship (and its uncertainty) between ozone exposure and tree growth and survival. Our work provides a new set of relationships to understand these impacts.