Hydraulic redistribution in three Amazonian trees.

About half of the Amazon rainforest is subject to seasonal droughts of 3 months or more. Despite this drought, several studies have shown that these forests, under a strongly seasonal climate, do not exhibit significant water stress during the dry season. In addition to deep soil water uptake, another contributing explanation for the absence of plant water stress during drought is the process of hydraulic redistribution; the nocturnal transfer of water by roots from moist to dry regions of the soil profile. Here, we present data on patterns of soil moisture and sap flow in roots of three dimorphic-rooted species in the Tapajós Forest, Amazônia, which demonstrate both upward (hydraulic lift) and downward hydraulic redistribution. We measured sap flow in lateral and tap roots of our three study species over a 2-year period using the heat ratio method, a sap-flow technique that allows bi-directional measurement of water flow. On certain nights during the dry season, reverse or acropetal flow (i.e.,in the direction of the soil) in the lateral roots and positive or basipetal sap flow (toward the plant) in the tap roots of Coussarea racemosa (caferana), Manilkara huberi (maçaranduba) and Protium robustum (breu) were observed, a pattern consistent with upward hydraulic redistribution (hydraulic lift). With the onset of heavy rains, this pattern reversed, with continuous night-time acropetal sap flow in the tap root and basipetal sap flow in lateral roots, indicating water movement from wet top soil to dry deeper soils (downward hydraulic redistribution). Both patterns were present in trees within a rainfall exclusion plot (Seca Floresta) and to a more limited extent in the control plot. Although hydraulic redistribution has traditionally been associated with arid or strongly seasonal environments, our findings now suggest that it is important in ameliorating water stress and improving rain infiltration in Amazonian rainforests. This has broad implications for understanding and modeling ecosystem process and forest function in this important biome.