The eco-hydrology of drought: What are the survival mechanisms of trees? Exploring the hypothesis that trees will create deep fine root mass to prevent carbon loss under severe drought.

Abstract

Trees play an important role in the biosphere on earth, both in the ecosystems they are part of and through the role they play in the biosphere-atmosphere interaction as a source of water to the atmosphere by transpiration. Global climate change will increase the magnitude and frequency of droughts in many parts of the world among which Southern Europe, therefore increasing our knowledge of the drought resistance and vulnerability of tree species is most pressing. Evaporation observations of forests in the Netherlands during periods of regular drought show that trees are able to evaporate almost at full potential, while surrounding crops and grasses show an evaporation reduction. This reduction in transpiration indicates a reduction in the carbon uptake of vegetation, because these processes are linked. There are doubts whether the underlying process that causes this phenomenon is taken into account correctly in current simulation models. The hypothetical mechanism that will be investigated during this research project is that trees can relocate fine root mass under drought to utilize water stored deeper in the soil and retain transpiration under limiting soil moisture conditions. To correctly simulate the hydrological consequences of climate change, under the foreseen increase in droughts, it is important to investigate this adaptation mechanism and potentially make it part of eco-hydrological models. The key assumption is that a tree will maximize its energy (i.e. carbon) gain by the optimal choice of the following trade-off strategies: (1) keeping photosynthesis going by investing in replacement of fine root mass and extracting water from deeper down the profile or (2) decreasing maintenance respiration by decreasing leave area.

This hypothesis is investigated with a Vegetation Optimality Model that simulates vertical carbon and water fluxes. The underlying assumption of the optimality approach is that vegetation maximizes its ‘Net Carbon Profit’ (NCP) to be most fit from an evolutionary perspective. The hypothetical root optimization strategy proposed in this research project was modelled as to achieve maximum NCP for the tree. The assumption that NCP is maximized by vegetation, allows formulating a single objective function accounting for both productivity and ‘‘water stress’’. The model was run using half-hourly Eddy covariance measurement data made available by the CarboEurope Integrated Project Ecosystem Component Database. The model was used for the Hainich forest site (DE-Hai) in Central Germany, where it was calibrated for the tree species Fagus sylvatica L. (european beech), and for Puechabon (FR-Pue) in the Herault region in France, calibrated for Quercus ilex L (holm oak). Root distribution and the maximal electron transport capacity (Jmax) are allowed to adapt dynamically during a model run, while ce and me (unitless empirical parameters that define the slope of the curve between photosynthesis and evapotranspiration), cRl (leaf respiration coefficient), Jmaxtop (electron transport capacity at the top of the canopy) and phenology were optimized off-line for the year 2005.

The model’s simulations were compared to the flux measurements of carbon (Net Ecosystem Exchange) and water (latent heat) and validated for the summer of 2005. Phenology was simulated using a degree-day method based solely on temperature. The difference between measurements and model outcome was minimized by an unconstrained nonlinear optimization function, taking into account the difference in measurement errors by weighing the carbon and water components. Off-line stochastic optimization of the 4 parameters (ce, me, cRl, Jmaxtop) was done with the DiffeRential Evolution Adaptive Metropolis (DREAM) algorithm. The results for the half hourly simulations in the summer (JJA) at the Hainich site showed very good agreement with the observed data both in magnitude and dynamics. In most of the model runs the root water uptake flux (Qr) and consequently the internal water storage of the tree (Mq) fluctuate strongly, from unrealistically high to low values with each time step, possibly as a result of numerical instability. Yearly results from the optimized model show some discrepancy in both carbon and water fluxes, as the simulated fluxes are generally overestimated in spring and for a short period during summer. Overall, the yearly dynamics of soil moisture for the year 2003-2007 give realistic results and water and carbon fluxes show good agreement with the observations. The difference between observed and simulated fluxes results from the late onset and early decline of leaves as prescribed by the phenology, since fluxes are overestimated during the period of full vegetative cover to balance the underestimation without the presence of leaves.

To test the hypothesis and examine the feasibility of investing in the relocation of fine root mass, the simulated cumulative NCP was evaluated after 1 to 7 days. The results for the summer of 2003, during the extreme drought, show that the root mechanism does not take place when evaluation of NCP is done after 1 or 2 days at all and only starts to make a difference after an evaluation period of 4 days or more. The simulations of the water flux increase after roots have been repositioned to deeper soil layers, allowing for higher water uptake and consequently carbon assimilation. Thus, it is beneficial to relocate fine root mass since the NCP is higher with the mechanism in place. The hypothetical root mechanism allows trees to continue transpiration after soil moisture has been depleted from the top soil layers. Currently, most hydrological models will simulate a decrease of evapotranspiration for forests during moderate droughts, whereas it should only occur if droughts are severe enough. The main conclusions, with the hypothetical root mechanism implemented in the model, are that there is indeed an increase in NCP, but there is no proof that this gives a better fit to the data, because the phenology, onset and decline of Leaf Area Index (LAI), is not simulated correctly. The model that was used during this project would benefit from (1) a better representation of the root water uptake flux. To further investigate the hypothetical root mechanism it is advised to (2) change the cost of creating new fine root mass and/or the vascular system respiration. Whereas nearly all eco-hydrological model simulations could be improved from (3) building a better phenology model.