Parameters for global ecosystem models

Abstract

Tianet al.¹ have used their process-based ecosystem model to estimate the net CO₂exchanges, also called net ecosystem productivity, for the years 1980-94. They deduced a large interannual variability ranging between −0.2 (from land to atmosphere) and +0.7 petagrams of carbon (Pg C) per year, the variability being mostly a function of soil moisture, which in turn is largely regulated by precipitation and temperature. These values were derived by including the modelled effects of increasing atmospheric levels of CO₂. The above numbers are the differences between net primary productivity and heterotrophic respiration. Over the given time period for the CO₂feedback case, these values were 5.0 (±0.3) and 4.8 (±0.1) Pg C per year, respectively. The calculated net ecosystem productivity was thus a small fraction, between −4% and +14%, of the net primary productivity, with an average over the 15-year period of +4%.

Main

We wish to point out that another carbon-release process, involving the emissions of isoprene (C₅H₈) and other volatile organic compounds (VOCs) from the Amazon vegetation to the atmosphere, must be taken into account. Guenther et al.² have estimated a global VOC emission of 1.15 Pg C per year, corresponding to an average global fraction of 2.4% of the net primary productivity. With the relatively larger VOC contributions from deciduous forests, this fraction will be of the order of 3% for tropical forests. The oxidation of isoprene to CO₂does not occur promptly, but involves many reactions in which compounds such as CO and oxidized hydrocarbons are formed; these compounds have lifetimes of the order of days to a month (in the case of CO). The neglect of VOC emissions can lead to a substantial underestimate of the ultimate release of carbon from the tropical ecosystems to the atmosphere.

In our research in the forested regions of Surinam during March 1998 (refs 3,4), we used a proton-transfer-reaction mass spectrometer on an aircraft to detect VOCs. We measured not only high concentrations of short-lived (hours) isoprene averaging 2.8 nmol mol⁻¹, but also high concentrations of up to 13.5 nmol mol⁻¹ of various VOC compounds with a positive ion mass of 43 atomic mass units such as propan-2-ol, as well as formic acid and acetic acid with average molar ratios approaching 10 nmol mol⁻¹.

In agreement with Guenther et al.², we conclude that a proper forest-ecosystem carbon budget analysis must take into account the emissions of VOCs as well as of CO₂. Not only are direct emissions of VOCs from intact vegetation important, but leaf wounding is a significant additional source of VOCs⁵; herbivores consume (and thus wound) about 40% of tropical forest above-ground net foliar production⁶. Research is greatly needed to make better estimates of these emissions and the processes affecting them in order to expand the model of Tian et al.¹. This is also important for an understanding of the air chemistry above the tropical forests and its influence on global chemistry⁷,⁸.

The occurrence of fires is an additional factor that must be taken into account. ‘Dry’ years, during which emissions of CO₂are highest owing to low soil humidity¹, are also years of more extensive fires in the tropics, further increasing the interannual variability of carbon release.

See also Schulman et al. Reply - Tian et al.