Abstract. Biogenic volatile organic compounds (BVOC), such as isoprene and monoterpenes, play an important role in atmospheric processes. BVOC species are oxidized in the atmosphere and influence levels of ozone. The less volatile amongst the BVOC and their oxidation products are important for the formation and growth of secondary biogenic aerosol. In this way, the Earth's radiation balance is affected.

Geographic distribution and temporal changes in BVOC emissions are highly uncertain. Here we assessed changes in emission patterns across Europe since the Last Glacial Maximum (LGM) with a dynamic vegetation model. This model reproduces European tree species distribution and includes a process-based algorithm for terpenoid production. In a set of simulations the model was driven with paleoclimate anomalies and reconstructed CO₂ concentrations. We quantified three main driving factors for the changes in emissions of isoprene and monoterpenes since the LGM: (1) the changes in climate, with temperature changes as the most important factor affecting plant physiology and terpenoid production in all plant species, (2) a change in species distribution related to the changes in climate, causing local shifts in emission characteristics of the vegetation, and (3) a change in CO₂ concentration, causing opposing effects on the availability of different substrates for terpenoid production. The effect of atmospheric CO₂ concentration is particularly uncertain, but sensitivity simulations showed an increase in European BVOC emissions in all sensitivity experiments irrespective of the use of a direct inhibition of terpenoid production by CO₂. The effects of climate change on physiology and terpenoid production resulted in an overall relatively uniform increase of emissions in Europe over the simulation period, but regionally the effect of changes in species distribution and the related changes in emission capacities resulted in changes of emissions that can dominate over the physiology effects. This may have consequences for regional atmospheric chemistry simulations for the past, that have to rely on suitable geographic patterns of forest emissions.