Weather, pollution and biotic factors drive net forest – atmosphere exchange of CO2 at different temporal scales in a temperate-zone mixed forest

By Joanna A. Horemans, et al.


• Daytime C sink capacity of the Brasschaat forest increased between 1999 and 2014.

• Daytime NEE was most strongly correlated to atmospheric CO2 levels and soil solution pH.

• Long term trend in night respiration was strongly correlated to ground water level.

• Random variation in day/night NEE were linked to cloudiness/air temperature.

• Time series decomposition plus random forest analysis useful to study NEE drivers.


Understanding the drivers of net ecosystem exchange of carbon (NEE) between forests and the atmosphere is crucial for the prediction of future global carbon dynamics. We therefore analyzed the long-term (1999–2014) ecosystem carbon fluxes of a mixed coniferous/deciduous forest (Brasschaat forest) in the Campine ecoregion of Belgium. The carbon uptake of this forest increased over the 16-year study period. By consecutively performing time series decomposition and the statistical technique of random forests, the correlative strength between multiple meteorological drivers, tropospheric pollutants and biotic indices with NEE was quantified at different time scales: i.e., long-term, seasonal and weekly, and separately for day- and nighttime NEE fluxes. The drivers that were most correlated with the trend in carbon sink capacity were the increasing atmospheric CO2 level and soil recovery from acidification. The radiation-saturated carbon uptake increased remarkably and explained much of the long-term variability of daytime NEE. When the long-term and seasonal variation were extracted the remaining weekly variation in daytime NEE was most strongly correlated with variation in the incoming radiation and cloudiness, and to a lesser extent by variation in vapor pressure deficit. In contrast to daytime NEE, nighttime NEE did not show a steady trend over time, but fluctuated, peaking in 1999 and in 2011. The long-term variability in nighttime NEE was most strongly correlated with the groundwater table depth. Air temperature was highly correlated to the seasonal as well as to the remaining weekly variation, i.e., after removal of the long-term and seasonal variability, in nighttime NEE. Biotic drivers (e.g., quantum yield and radiation-saturated carbon uptake) explained less of the variation in NEE on a seasonal and short-term scale, but were more important at the long term.

This (paywalled) article appeared on the Agricultural and Forest Meteorology website at