Forecasting the Onset of Cupressus Flowering in the Mediterranean Region

Cypress species and other members of the Cupressaceae family are widespread evergreen trees and shrubs, commonly used as ornamental plants. Some species, such as Mediterranean cypress (Cupressus sempervirens), Arizona cypress (Cupressus arizonica), and prickly juniper (Juniperus oxycedrus), widespread in Mediterranean, cause significant allergic reactions. This study aimed to develop a phenological model for Southern and Central Europe to predict the timing of cypress pollen release, enabling integration into atmospheric models for pollen dispersion forecasts. 

A key challenge is the microscopic similarity of all Cupressaceae pollen grains, which prevents species-level identification. Consequently, pollen observations report total Cupressaceae counts, complicating phenological modeling of allergenic species. 

For early-flowering species, thermal time models, such as growing degree days (GDD) or growing degree hours (GDH), are suitable. These models require defining a heat accumulation start date, a temperature threshold, and the cumulative heat required for flowering. Geographic variability and ornamental planting further influence flowering patterns, even between neighbouring locations. 

Pollen data were obtained from the European Aeroallergen Network (EAN), and temperature data from the ERA5 reanalysis dataset. Testing three start dates of the accumulation revealed that the autumn equinox was too early, while January 1st was too late, as J. oxycedrus and C. arizonica may flower before the new year. November 30th was optimal for detailed analysis. GDD/GDH was calculated using thresholds of 0°C, 1°C, 2°C, and 5°C, with normalized GDH (nGDH) yielding the most accurate results. 

When flowering onset was defined as 5% of the seasonal pollen index (SPI), the median GDH requirements ranged from 0.06–0.15 nGDH0 (SD 0.01–0.05) to 0.01–0.06 nGDH5 (SD 0–0.02). A 5°C threshold was too high leading to insufficient heat accumulation sensitivity, while 0°C was too low due to higher variability between years. Thresholds of 1°C and 2°C provided optimal accuracy with moderate inter-annual variability, making them suitable for forecasting the flowering onset.