Elevated CO2, vegetation and the hydrologic cycle – the role of observations and models

Background/Question/Methods

Observations made at different scales often reveal different patterns within the same entity. The change in observed pattern with the change in scale of observation is often abrupt ('break in scale'). Linking patterns with different scales of observation has become a major challenge in all sciences; however, it appears that the better our ability to observe patterns becomes, the more 'breaks in scales' we find. This talk relates breaks in scales to hierarchical levels of organisation. It is argued that organisation is linked with optimality and that optimality makes a system predictable to a certain degree.

The effect of elevated atmospheric CO₂ concentrations (eCO₂) on vegetation water use is an example for different response patterns at different scales. At the leaf scale, eCO₂ can induce stomatal closure and reduced transpiration. This affect was incorporated in global models and held responsible for an observed increase in global river runoff during the past century. However, stomatal closure is unlikely the only means by which vegetation responds to eCO₂. If, for example, stomatal closure is offset by an increase in vegetation cover in the long term, the effect of eCO₂ on global transpiration could, in fact, be reversed.

Long-term effects of eCO₂ are difficult to capture in experiments or to deduce from past observations, because experiments are relatively short and long-term observations at the CO₂ levels expected in 20 years time do not exist. Alternatively, predictions of such long-term effects could be derived from models that do not simply extrapolate observed responses into the future.

The Vegetation Optimality Model (VOM) allows separation of different spatial and temporal scales of adaptation, without the need for parameterisation with observed responses. This frees up observational and experimental data for model testing and allows hypothetical predictions about vegetation response at time scales beyond those of free air CO₂ enrichment experiments.

Results/Conclusions

At an intensively monitored savanna site, the simulated short-term response (years) to eCO₂ turned out to result in a reduction of evapo-transpiration (ET) that is consistent with expectations. However, the simulated long-term response (decades) resulted in a large increase in evapo-transpiration. In most other ecosystems, the model also predicted a stronger reduction in ET in the short-term than in the long-term. These results confirm that adaptation of natural vegetation to environmental change may lead to different responses at different time scales, resulting in a potential bias if observed short-term responses are projected into the future.