Water use by woody plants on contrasting soils in a savanna parkland: Assessment with δ2H and δ18O

Abstract

In savanna parklands of southern Texas, patches of grassland and ‘discrete clusters’ of small trees and shrubs occur on sandy loam surface soils underlain by an argillic horizon (claypan) at 40 cm. Large trees and shrubs in ‘groves’ occur on deep (2 m) sandy loam soils without an argillic horizon. δ2H and δ18O of rainfall, groundwater, and soil and plant water were measured to: (1) determine if coexistence in woody patches occurs via vertical stratification of soil water uptake; (2) document differences in plant water acquisition on contrasting soil types; and (3) evaluate recharge and evaporative losses of soil moisture from grassland vs. wooded landscape elements. Groundwater was isotopically similar to weighted rainfall, suggesting local recharge at this site. Linear regressions of soil water δ2H on δ18O yielded slopes less than the meteoric water line, indicating significant evaporative losses of soil moisture in all landscape elements. Interspecific differences in root density distribution were significant; some woody species had roots well below 1.6 m, while others had few roots below 0.8 m. δ2H and δ18O values of stem water from all plants in groves were lower than those of soil water in the upper 1.5 m of the profile, suggesting all species obtained their water from depths >1.5 m. Deep roots of trees and shrubs at this savanna parkland site thus appeared to have a functional significance that was not revealed by biomass or density determinations. Root densities of species in discrete clusters (claypan present) were typically greater than those of the same species in groves (claypan absent), especially in the upper 80 cm of the soil profile. Consistent with rooting profiles, δ2H and δ18O values of plant water indicated that trees and shrubs in discrete clusters with fine- textured subsoils obtained most of their water at depths <1.5 m. As with groves, there was no indication of water resource partitioning between species. In summary, we saw no isotopic evidence that co- occurring woody plants at this savanna parkland site were partitioning soil moisture vertically during late summer/early fall, despite marked differences in their root density distributions. This supports other lines of evidence which indicate that species interactions in tree/shrub clumps are competitive, and that species composition is therefore unstable in those landscape elements.