16.2 News & Views 793 MH

Measurements of stream flow around the world have documented an increase in the amount of water that runs off the continents and returns to the ocean. This trend has been occurring since the beginning of the century, yet changes in precipitation over land do not sufficiently account for this increase. On page 835 of this issue, Gedney et al. identify an important contributor to increasing global runoff — decreased evaporation resulting from the influence of elevated atmospheric carbon dioxide on plant physiology (Fig. 1, overleaf). Carbon dioxide is the currency of plant photosynthesis: plants take up CO2 from the atmosphere and incorporate it into their tissues in the form of organic carbon compounds. This uptake of CO2 is accomplished through plant stomata — small openings in the surface of leaves (pictured above) that open and close to allow the exchange of CO2 and other gases with the atmosphere. During gas exchange, water is inevitably lost to the atmosphere, again through stomatal openings. This is the process of plant transpiration, which, on a global scale, mediates the transfer of water from the soil into plant tissues, and out through stomata to the atmosphere. On vegetated land, plant transpiration can make a substantial contribution to total surface evapotranspiration, which represents the sum of plant transpiration and other surface evaporation. Plants can regulate the opening and closing of stomata in response to changing environmental conditions; in a high-CO2 atmosphere they are more efficient in their use of soil moisture. The stomata do not open as much or for as long, and less water is lost from leaves to the atmosphere. As a consequence, plants acquire enough carbon through their stomata with less water uptake from the soil. The result is that continental evapotranspiration is reduced, more moisture is left in the soil, and this additional surface water can lead to increased continental runoff . Using a technique known as ‘optimal fingerprinting’ (also known as ‘detection and attribution’), Gedney et al. show that this direct effect of elevated CO2 on plant transpiration is the dominant contributor to observed increases in continental runoff. Optimal fingerprinting is simply a statistical regression in which a model simulation is compared with observations to isolate which processes in the model are consistent with the observed data. If a model simulation is consistent with observations, the process that drives the model trend is said to be ‘detected’ in the observations; if the observed trend is also inconsistent with other plausible explanations, then the trend can be ‘attributed’ to a specific cause. Gedney et al. investigated four plausible contributors to observed increases in runoff: climate change leading to changes in temperature and precipitation; land-use change and consequent changes in vegetation cover; socalled ‘solar dimming’, resulting from an increasingly hazy atmosphere; and the direct effect of CO2 on plant transpiration. The effects of each of these on surface runoff were simulated using a sophisticated land-surface and vegetation model, and the results of the model were compared with historical observations of continental runoff. The authors’ analysis shows that model-simulated runoff trends are consistent with the observed trend only when the direct effect of CO2 on transpiration is included in the simulation. So they attribute increases in continental runoff over the past century to the physiological effect of elevated atmospheric CO2. Detection and attribution has been widely EY E O F SC IE N C E/ SP L 16.2 News & Views 793 MH 10/2/06 5:21 PM Page 793