Soil moisture redistribution as a mechanism of facilitation in savanna tree–shrub clusters

Abstract

Plant-soil water relations were examined in the context of a selective removal study conducted in tree-shrub communities occupying different but contiguous soil types (small discrete clusters on shallow, duplex soils versus larger, extensive groves on deep, sandy soils) in a subtropical savanna parkland. We (1) tested for the occurrence of soil moisture redistribution by hydraulic lift (HL), (2) determined the influence of edaphic factors on HL, and (3) evaluated the significance of HL for overstory tree-understory shrub interactions. Diel cycling and nocturnal increases in soil water potential (Psisoil), characteristic signatures of HL, occurred intermittently throughout an annual growth cycle in both communities over a range of moisture levels (Psisoil=-0.5 to -6.0 MPa) but only when soils were distinctly stratified with depth (dry surface/wet deep soil layers). The magnitude of mean (+/-SE) diel fluctuations in Psisoil (0.19+/-0.01 MPa) did not differ on the two community types, though HL occurred more frequently in groves (deep soils) than clusters (shallow soils). Selective removal of either Prosopis glandulosa overstory or mixed-species shrub understory reduced the frequency of HL, indicating that Prosopis and at least one other woody species was conducting HL. For Zanthoxylum fagara, a shallow-rooted understory shrub, Prosopis removal from clusters decreased leaf water potential (Psileaf) and net CO2 exchange (A) during periods of HL. In contrast, overstory removal had neutral to positive effects on more deeply-rooted shrub species (Berberis trifoliolata and Condalia hookeri). Removal of the shrub understory in groves increased A in the overstory Prosopis. Results indicate the following: (a) HL is common but temporally dynamic in these savanna tree-shrub communities; (b) edaphic factors influencing the degree of overstory/understory development, rooting patterns and soil moisture distribution influence HL; (c) net interactions between overstory and understory elements in these woody patches can be positive, negative and neutral over an annual cycle, and (d) Prosopis-mediated HL is an important mechanism of faciliation for some, but not all, understory shrubs.