Water sourcing by trees in a mesic savanna: Responses to severing deep and shallow roots

Abstract

The co-dominance of trees and grasses is a much-debated topic in savanna ecology. One of the proposed mechanisms facilitating the co-dominance of these two different life-forms is root niche partitioning. Specifically, the rooting zone is hypothesized to have two layers: grasses are thought to be the superior competitors for water in the upper soil horizons, whereas trees avoid competitive exclusion by sourcing water from soil layers beyond the vertical reach of grasses. We examined this two-layer hypothesis through a field experiment in which we severed deep and shallow roots of Terminalia sericea saplings, 6 weeks before the onset of the new growing season, using a novel technique. During the course of the growing season, physiological responses to the root severing were measured. Severing lateral (shallow) roots resulted in rapid abscission of the old leaves, smaller leaves after the onset of the following growing season and lower xylem pressure potentials – indicating higher water stress – compared to the other treatment groups. Effects of severing deep-growing taproots were small, demonstrating the importance of lateral roots for maintaining the trees’ water balance, notably in drier times. We have shown that saplings of the common and locally dominant savanna tree T. sericea manage to coexist with grasses without avoiding competition through spatial root separation. The results of our study therefore do not support the two-layer hypothesis for explaining tree–grass coexistence in this mesic savanna.

Introduction

One of the world's most extensive biomes, the savanna, is generally defined as a discontinuous layer of trees overlying a continuous layer of grasses, and is characterized by a distinctly seasonal rainfall pattern (Frost et al., 1986). The mechanisms enabling the coexistence of trees and grasses in tropical and subtropical savannas have been comprehensively researched (for reviews see: Scholes and Archer, 1997, Sankaran et al., 2004, Midgley et al., 2010). The results of this research have led to a general agreement among savanna ecologists that resource availability, fire frequency and intensity, and herbivory influence savanna structure in complex ways. However, the factors preventing competitive exclusion of any of these two life-forms are not well understood and the relative importance of the different processes shaping the structure of savannas is under continuing debate (Higgins et al., 2000, Higgins et al., 2010, Van Langevelde et al., 2003, Sankaran et al., 2008, Coetsee et al., 2010, Midgley et al., 2010).

Rainfall is regarded as a key determinant of potential woody cover in savanna systems. In a continental scale analysis, Sankaran et al. (2005) showed that rainfall and maximum woody biomass are positively correlated for systems receiving between 100 and 650 mm of rainfall per year (Sankaran et al., 2005, Sankaran et al., 2008). Above this threshold, rainfall is presumed to be sufficient for woody biomass to increase to a state of complete canopy closure, but disturbances such as fire and herbivory prevent this from happening (Bond et al., 2003, Sankaran et al., 2005).

A number of studies have suggested that trees in savannas with their deeper roots have access to deep soil water that is unavailable to grasses (Walter, 1971). In this competition-based model, Walter (1971) assumed grasses to be superior competitors for water in the surface soil layers, with trees predicted to persist only where enough water percolates beyond the reach of grasses. This model, referred to as the two-layer hypothesis, has empirical support (Walker and Noy-Meir, 1982, Weltzin and McPherson, 1997), but several studies have shown a substantial overlap in root niches between trees and grasses (Le Roux et al., 1995, Seghieri et al., 1995, Mordelet et al., 1997, Ludwig et al., 2004, February and Higgins, 2010, Kulmatiski et al., 2010). Despite all this research, the extent to which trees and grasses partition resources is still unclear. Studies of soil resource competition in savannas have relied on inference from rooting patterns and/or from stable isotopes. Root distribution may not indicate actual water uptake, whereas stable isotope tracer studies merely provide a snapshot of water uptake patterns.

In this study, we further explore below ground resource partitioning between trees and grasses in mesic savannas. To examine the extent to which trees rely on water uptake from soil layers that are either within or beyond the vertical reach of grasses we conducted a manipulative experiment in which we simply restricted tree roots to the grass-roots layer. Root trenching is commonly used to study the effect of water competition between trees and understory crops on crop yield in agroforestry systems (Powell and Bork, 2006, Burner et al., 2009). In natural systems, Ludwig et al. (2004) found increased grass production after tree root trenching in a savanna, while Barberis and Tanner (2005) showed increased survival and growth in tree seedlings in response to trenching in a tropical forest. Although the effects of root severing on the water relations of mesquite trees have been examined (Ansley et al., 1990, Ansley et al., 1991) there has been no research testing the two-layer hypothesis using manipulation of tree water access in shallow and deep soil compartments through the severing of tree roots. In this study we manipulate the vertical and horizontal access of trees to the rooting zone of grasses, by severing the taproot and the lateral roots using a novel trenching technique which minimizes soil disturbance. We then monitored a range of water stress indicators, including water status and leaf-level responses, throughout the 2006–2007 growing season. If, as is predicted by the two layer hypothesis, trees are able to avoid competition with grasses for resources through deep rooting, then trees should respond more strongly to taproot severing than to lateral root severing.