Predicting habitat affinities of herbaceous dicots to soil wetness based on physiological traits of drought tolerance.

Background and Aims: Soil water availability is an important mechanism filtering plant species but the functional basis of this filtering in herbaceous dicots is poorly studied. The authors address three questions: Which physiological traits best predict different levels of drought tolerance or avoidance in herbaceous dicots? To what degree can species' habitat preferences along the gradient of soil moisture availability be predicted by their physiological responses to drought? What are the direct and indirect relationships between the physiological traits and how do they interact to determine the species' habitat preferences? Methods: Twenty-five species of herbaceous dicots whose field distributions span a gradient of soil moisture from continually moist to dry were chosen. Under controlled conditions, watering was stopped in a treatment group, the plants were monitored until death of the above-ground tissues and compared with a control group watered at field capacity. Fourteen traits related to plant water economy were measured, including stomatal conductance, net photosynthesis and the visual wilting of leaves. Traits were then analysed using a cumulative link model and path analysis. Key Results: Five physiological traits (stomatal conductance and net photosynthesis measured at soil field capacity, water use efficiency, stomatal conductance and soil water potential measured when leaves begin to wilt) related to the ability to acquire resources (when water is not limiting) or conserve water (when it is limiting) best predicted different levels of drought tolerance or avoidance in herbaceous dicots. Conclusions: Species' habitat preferences can be fairly predicted by their physiological responses to drought ( R 2 = 0·48). Strong direct and indirect relationships between the five identified traits (plus net photosynthesis at wilting and the time until death) led to synergistic and antagonistic relationship in a path analysis model. To allow better prediction of species distributions along a wetness gradient, the next step would be to link these physiological traits to more accessible functional traits.