Toward detection of CO2 fertilization of tree growth and biomass accumulation in Amazon forests

Synthesis studies of old-growth tropical forest plot networks indicate a pantropical net carbon sink of more than 1 Pg C/yr. However a number of confounding factors limit our ability to attribute these changes to direct CO2 fertilization of tree growth and forest productivity. Of primary importance is determining if the plots adequately sample natural disturbance and recovery gradients, and the larger landscape successional mosaic. In addition, forest biomass dynamics which include tree growth, recruitment and mortality can interact in complex ways with changes in forest productivity and biomass accumulation. This study represents a novel approach to determine the sensitivity of different sampling strategies for detecting tropical forest CO2 fertilization while accounting for these confounding factors. Our approach, developed for Amazon forests in Brazil and Peru, combines extensive field plot data on biomass dynamics, remote sensing analyses to generate disturbance probability distribution functions, and individual-based simulation modeling for placing plot-level results into a landscape context. Results indicate that forest plots significantly larger than 10 ha are required to maximize the signal-to-noise ratio for detecting CO2 fertilization. We also present a field sampling strategy for quantifying site-to-site differences in forest biomass accumulation rates, which is useful for detecting regional differences in tropical forest sensitivity to rising atmospheric CO2 concentration. Overall, this approach is useful in developing field campaigns that explicitly account for landscape heterogeneity in testing key predictions of Earth system models.