Ecosystem gross primary productivity during the COVID-19 lockdown

During the spring of 2020, many countries around the world imposed lockdown measures involving economic activity and movement restrictions to contain the outbreak of the novel coronavirus disease (COVID-19), thereby leading to changes in air pollutant concentrations (Venter et al., 2020). The unprecedented reductions in primary pollutant emissions created a unique opportunity to assess the response of photosynthetic activity of terrestrial ecosystems to atmospheric changes in air quality. Our hypothesis was that a concentration decrease in particulate matter (PM) and the resulting change in light scattering may have affected photosynthesis via changes in direct and diffuse radiation, while a reduction of ozone precursor emissions may have negatively impacted the formation of ozone and reduced its phytotoxic effects. Thus, we analysed turbulent fluxes from eddy covariance measurements and meteorological data collected at the Integrated Carbon Observation System (ICOS) ecosystem stations, and also air pollution data from a continental-scale chemistry transport model (LOTOS-EUROS). Using observations from 44 sites in Europe spanning eleven countries and nine vegetation types, we calculated a 4-month (March-April-May-June, hereafter ‘spring’) anomaly of gross primary productivity (GPP) as the cumulative difference of GPP between 2020 and the reference period from 2015 to 2019. For 34 out of 44 sites, we found that the means between 2020 and the reference GPP were different at the 5% significance level. We further classify these sites into four groups according to modelling and simulation analyses and related data.

• Group 1 included 16 sites where the GPP anomaly was predominantly driven by changes in meteorology. A 7-31% GPP reduction of eight sites in this group was attributed to several different factors such as reduced incoming shortwave radiation (SW_IN), increased vapour pressure deficit (VPD), late growing season and legacy effects. The remaining eight sites experienced an increase in GPP (5-20%) which coincided with increased SW_IN and reduced diffuse fraction (K_d).
• Group 2 consisted of five sites where the GPP anomaly was primarily linked to drought-related effects as indicated by an exceptional increase in the Bowen ratio (δß > 29%), declines in soil water content (SWC) and precipitation.
• Group 3 was represented by five sites where the GPP anomaly was presumably affected by both meteorology and pollutants. All sites in this group experienced an increase in GPP of 14-47% that coincided with enhanced SW_IN (2-13%), reduced atmospheric concentrations of NO₂ (28-47%), NO (33-57%), O₃ (2-3%), SO₂ (5-7%), PM10 (4-14%), PM2.5 (9-17%) and increased NH₃ (1-5%).
• There were eight grassland and savannah sites in Group 4 where the ecosystem management interacted with meteorology to mainly increase GPP by 10-41%.

We first conclude that meteorology and pollutant concentrations during the spring were different between 2020 and 2015-2019 period. Second, our analyses showed that the GPP anomaly in the spring of 2020 was explained by the balance between positive and negative impacts of biophysical drivers. GPP increased when the combined effects of enhanced SW_IN, increased air temperature and reduced pollutant concentrations overtook the negative impact of changes in VPD, SWC and K_d.

Acknowledgements. We would like to thank ICOS site investigators for sharing eddy covariance data.