Exploring new strategies for ozone-risk assessment: A dynamic-threshold case study

Highlights: • Flux-based ozone impact response functions were implemented in a canopy model. • Both linear and nonlinear ozone response functions led to more realistic simulation. • Seasonal changes in ozone tolerance points to the need for seasonal thresholds. Tropospheric ozone is a dangerous atmospheric pollutant for forest ecosystems when it penetrates stomata. Thresholds for ozone-risk assessment are based on accumulated stomatal ozone fluxes such as the Phytotoxic Ozone Dose (POD). In order to identify the effect of ozone on a Holm oak forest in central Italy, four flux-based ozone impact response functions were implemented and tested in a multi-layer canopy model AIRTREE and evaluated against Gross Primary Productivity (GPP) obtained from observations of Eddy Covariance fluxes of CO₂. To evaluate if a clear phytotoxic threshold exists and if it changes during the year, six different detoxifying thresholds ranging between 0 and 5 nmol O₃ m⁻² s⁻¹ were tested. The use of species-specific rather than more general response functions based on plant functional types (PFT) increased model accuracy (RMSE reduced by up to 8.5%). In the case of linear response functions, a threshold of 1 nmol m⁻² s⁻² produced the best results for simulations of the whole year, although the tolerance to ozone changed seasonally, with higher tolerance (5 nmol m⁻² s⁻¹ or no ozone impact) for Winter and Spring and lower thresholds in Summer and Fall (0–1 nmol m⁻² s⁻¹). A "dynamic threshold" obtained by extracting the best daily threshold values from a range of different simulations helped reduce model overestimation of GPP by 213 g C m⁻² y⁻¹ and reduce RMSE up to 7.7%. Finally, a nonlinear ozone correction based on manipulative experiments produced the best results when no detoxifying threshold was applied (0 nmol O₃ m⁻² s⁻¹), suggesting that nonlinear functions fully account for ozone detoxification. The evidence of seasonal changes in ozone tolerance points to the need for seasonal thresholds to predict ozone damage and highlights the importance of performing more species-specific manipulative experiments to derive response functions for a broad range of plant species.