Vegetation-atmosphere interactions are a persistent source of uncertainty in projections of the global carbon cycle and climate. Current land-surface models are built on a framework describing plants and ecosystems as physical objects exchanging materials and energy with the atmosphere and hydrosphere. This approach requires large number of biologically controlled parameters, and their responses to environmental factors, to be specified. In the absence of an accepted theory, standard practice has been to assign generic, literature-derived parameter values and environmental responses to each of a small set of plant functional types (PFTs) standing for the dominant plants in each biome. This practice combines overcomplexity (as the parameter list quickly becomes very long) with oversimplification (because parameter values in the real world typically vary more within than between PFTs). On the other hand, despite the huge accumulation of plant functional trait data in recent years, analysis and interpretation of these data are usually limited to statistical relationships – providing an inadequate basis to infer general principles that could be used to develop more robust and reliable models.
The missing element in both current land-surface modelling and plant functional ecology is the universal principle of Darwinian selection, and its power to eliminate uncompetitive combinations of plant traits – thereby generating patterns of trait variation along environmental gradients that can be predicted quantitatively, as analytical solutions to well-defined optimality problems. This approach has provided novel, parameter-sparse, testable and well-tested representations of key leaf- and plant-level processes including photosynthesis, respiration of leaves and stems, stomatal behaviour and leaf economics, and ecosystem-level outcomes including gross primary production, transpiration and vegetation cover. Such representations do not require a prior classification of vegetation into PFTs. They do, however, predict previously unexplained (or sometimes wrongly explained) spatial and temporal patterns of variation in plant traits. Examples include: opposing trends of leaf thickness with latitude in deciduous and evergreen plants; differential responses of leaf and steam respiration to growth temperature, and their relative magnitudes; and declining leaf nitrogen concentrations over recent decades. These successes build confidence in a new set of defining equations for the plant and ecosystem properties that are most important both for plant and ecosystem responses to environmental changes, and for their feedbacks to the climate system.