Forest fine-root production and nitrogen use under elevated CO₂: Contrasting responses in evergreen and deciduous trees explained by a common principle

Despite the importance of nitrogen (N) limitation of forest carbon (C) sequestration at rising atmospheric CO₂ concentration, the mechanisms responsible are not well understood. To elucidate the interactive effects of elevated CO₂ (eCO₂) and soil N availability on forest productivity and C allocation, we hypothesized that (1) trees maximize fitness by allocating N and C to maximize their net growth and (2) that N uptake is controlled by soil N availability and root exploration for soil N. We tested this model using data collected in Free-Air CO₂ Enrichment sites dominated by evergreen (Pinus taeda; Duke Forest) and deciduous [Liquidambar styraciflua; Oak Ridge National Laboratory (ORNL)] trees. The model explained 80-95% of variation in productivity and N-uptake data among eCO₂, N fertilization and control treatments over 6 years. The model explains why fine-root production increased, and why N uptake increased despite reduced soil N availability under eCO₂ at ORNL and Duke. In agreement with observations at other sites, the model predicts that soil N availability reduced below a critical level diminishes all eCO₂ responses. At Duke, a negative feedback between reduced soil N availability and N uptake prevented progressive reduction in soil N availability at eCO₂. At ORNL, soil N availability progressively decreased because it did not trigger reductions in N uptake; N uptake was maintained at ORNL through a large increase in the production of fast turnover fine roots. This implies that species with fast root turnover could be more prone to progressive N limitation of carbon sequestration in woody biomass than species with slow root turnover, such as evergreens. However, longer term data are necessary for a thorough evaluation of this hypothesis. The success of the model suggests that the principle of maximization of net growth to control growth and allocation could serve as a basis for simplification and generalization of larger scale forest and ecosystem models, for example by removing the need to specify parameters for relative foliage/stem/root allocation.