Reply to Miglietta et al.: Maximal transpiration controlled by plants
Proceedings of the National Academy of Sciences of the United States of America
We thank Miglietta et al. (1) for their interest in our study (2). Their first and main point arises from the idea that plant transpiration (T) is driven by atmospheric demand, giving plants limited control over the water they lose. Miglietta et al. (1) add that stomatal density (D) will not change in response to increasing atmospheric CO 2 concentration ([CO 2 ]), so future changes in T are unlikely. This idea neglects plant physiology and is not supported by recent observations (3). Maximal T
... ions (3). Maximal T is limited by maximal stomatal conductance (g smax ) with stomata fully open. When atmospheric demand exceeds a plant's water transport capacity, stomata close in minutes. To long-term gradual [CO 2 ] fluctuations, many perennial C3 plants adjust D and the size of fully opened stomata (a max ) to regulate g smax . Concomitant D and a max adjustments may therefore constrain T also at (multi)decadal time scales. By using the measured responses of solely long-lived C3 plants to historical [CO 2 ] increments (3) , we developed a model for the optimization of g smax under the constraint of a cost of water loss (2). This approach acknowledges that water loss is intrinsically linked to carbon uptake by diffusion through stomata. We stress that simulated changes in T are interpreted at canopy scales and not, as Miglietta et al. (1) imply, at regional scales. Our model was validated with measurements of D and a max on 519 leaves from five angiosperm and three conifer species (3). All show a consistent decrease in g smax (34% average) coeval with an approximately 100 ppm [CO 2 ] increase. Crucial is that plants reduced g smax by species-specific adaptations of both D and a max . Changes in D alone cannot be related to g smax without accounting for a max . The second point Miglietta et al. (1) raise is that, if plants respond to [CO 2 ], a decrease in T results in limited change in evapotranspiration (ET) because additional evaporation (E) answers the demand from potential ET. Although regional feedbacks were not our focus, we calculate here that a decrease in T from 120 to 60 W·m −2 results in a 30 W·m −2 decrease in ET, assuming T accounts for 50% in ET. This compares with other results for densely forested (sub)tropical regions (4). To demonstrate the limited stomatal control on ET, Miglietta et al. (1) refer to a model that simulates little change in long-term runoff in a temperate forest under elevated [CO 2 ] (5). As the referred model assumes that plants are not actively controlling T, we do not see how this proves their point. Moreover, the measurements on which this model was calibrated (6) show a 14% decrease in T under elevated [CO 2 ], resulting in 10% less ET. Changes in T occur in response to elevated [CO 2 ] and are related to plant physiology aimed to optimize carbon gain with minimal water loss. Crucial and interesting at a regional scale are the feedbacks between atmosphere and land that may dampen or accelerate hydrological cycling in future climates. We therefore acknowledge the final statement of Miglietta et al. (1) : "that projecting effects of increasing [CO 2 ] on the hydrologic cycle must account for soil and canopy processes as well as atmospheric feedback."