Effects of spatial variations in soil evaporation caused by tree shading on water flux partitioning in a semi-arid pine forest

Abstract

In dry environments, water availability is a major limitation to forest productivity while losses to soil evaporation (E) are a significant component in ecosystem hydrology. We report on a 3-year study (2004–2007) in a semi-arid pine forest in Southern Israel (40-year-old Pinus halepensis; leaf area index = 1.5; mean precipitation 285 mm year⁻¹) that estimated soil E, assessed its spatial variability and identified the factors influencing it. We used a modified and specially calibrated soil respiration chamber to directly measure E on a weekly basis at 14 permanently installed soil collars across the range of soil surface conditions. Results showed that spatial variability in E was large, with SD of ±47% between measurement sites. E fluxes measured in sun-exposed areas were on average double those in shaded areas (0.11 mm h⁻¹ vs. 0.06 mm h⁻¹). The spatial variability in E correlated with radiation (measured in the photosynthetically active range), which was up to 92% higher in exposed compared to shaded sites, and with soil water content, which was higher in exposed areas during the wetting season but higher in shaded areas during the drying season. The fraction of shaded forest floor area was described as a function of canopy geometry (mean tree height, crown width and stand density) and the daily variation in solar altitude. Simple simulations based on the relationship between E and the shaded fraction indicated that E/P (precipitation) for the Yatir forest could decrease from 0.53 (undeveloped canopy of 10% cover) to 0.30 (full canopy closure). However, according to our analysis, increasing canopy cover will also increase intercepted precipitation and transpiration such that current precipitation inputs will not be able to support forest growth above a canopy cover of 65%. Combining direct measurements of environmental conditions and canopy characteristics with such simulations can provide a simple predictive and management tool to optimize tree water use in dry environments.

Introduction

The hydrological balance between ecosystem water components can be simply represented as

$$P=R+S+\text{ET}\text{and}\text{ET}=T+E+I_{\text{p}}$$

where P is precipitation inputs, R is runoff, S is deep soil recharge beyond the rooting zone and ET is evapotranspiration. The latter component is the vapor flux made up of T, which is plant transpiration, E, which is soil evaporation, and I_p, which is evaporation of water intercepted by plant foliage. ET is often the largest water flux in dryland ecosystems, usually accounting for more than 90% of annual precipitation (Zhang et al., 2001). The proportion of T in these warm and water-limited systems is influenced by the partitioning between the other components of the hydrological budget (i.e. losses to R, S, E and I_p). Among these losses, E is expected to be the largest flux in water-limited environments with sparse vegetation cover, motivating research to improve its measurement and to quantify its interactions with environmental conditions.

T and E are both affected by the balance between atmospheric demand, soil water content and hydraulic conductivity, reflecting our difficulty in partitioning ET into its components and measuring each flux separately. The ability to distinguish between specific fluxes is critical for understanding ecosystem processes such as plant water availability, water-use efficiency, vegetation patterns, ecosystem productivity and soil respiration, as well as for improving large-scale models (Hutjes et al., 1998, Law et al., 2002, Ciais et al., 2005, Huxman et al., 2005, Lawrence et al., 2007).

Forests in dry environments are characterized by low tree density, presumably to allow sufficient water for all trees, even during dry years. In such ecosystems, the fraction of bare soil surface is relatively large, resulting in high heterogeneity in surface conditions (mainly related to various shading situations) and consequently in soil moisture content, runoff production, redistribution and infiltration. Below-canopy radiation in sun-exposed areas can be high (Diawara et al., 1991, Morecroft et al., 1998) and the spatial variability in radiation load between sun-exposed and tree-shaded areas seems to be maximized for a canopy cover of 20–40% (Martens et al., 2000). Canopy structure and leaf area index (LAI) are known to affect below-canopy radiation characteristics (Kuuluvainen and Pukkala, 1989), which can potentially influence E and increase its spatial variability (Breshears et al., 1998). Despite the significant contribution of E to ecosystem ET and the expected large spatial variability, little quantitative information is available on this topic. Such information could help develop predictive and management tools to improve water use and water-use efficiency in forest ecosystems of dry environments.

E is often calculated as a residual when ET is measured with eddy covariance or energy balance techniques and T is measured with sap flux techniques (Baldocchi and Vogel, 1996, Rana and Katerji, 2000, Oishi et al., 2008). Microlysimeters and time-domain reflectometers (TDRs) measure soil water loss or gain directly, but in vegetated environments they estimate total ET, without distinguishing specific paths of water loss (Baker and Spaans, 1994, Wythers et al., 1999). ET can be partitioned based on an approach that combines measurements of the stable oxygen and hydrogen isotopic compositions of water and water vapor with conventional flux measurements (Wang and Yakir, 2000, Yakir and Stenberg, 2000, Yepez et al., 2003). This approach, however, is laborious and is usually conducted for short durations, although the introduction of new laser-based isotopic measurements may change this (Wen et al., 2008). Another approach to ET partitioning uses chambers, which measure gas flow within a confined volume of air inside a chamber positioned on the soil surface. A possible advantage of this methodology is that the measured segment can be specified and relatively high spatial resolution can be obtained. Accordingly, the chamber method has been used to measure E (Iritz et al., 1997, Stannard and Weltz, 2006, Ward and Micin, 2006, Daikoku et al., 2008), whole-tree T (Greenwood and Beresford, 1979, Denmead et al., 1993), crop water use (Reicosky et al., 1983, Dunin and Greenwood, 1986, Pickering et al., 1993, Grau, 1995) and below-canopy ET (Dugas et al., 1997, Steduto et al., 2002, Arnone and Obrist, 2003, McLeod et al., 2004). The disadvantages of this approach may include calibration difficulties, up-scaling complications, low portability and perhaps most importantly, alterations in ambient conditions.

The objective of this study was to directly measure E in a low LAI semi-arid pine forest, define its spatial variability and identify factors that influence its contribution to total ET in order to assess how changes in these factors will affect the hydrological balance of the forest.

Gas-exchange chambers, which are most widely used for soil respiration measurements, are divided into two groups: flow-through steady-state chambers and closed, non-steady-state chambers (Hutchinson and Livingston, 2001). Positioning a soil chamber on the ground immediately changes the ambient conditions within the soil chamber, mainly with respect to radiation, temperature, pressure and turbulence. Problems of reduced radiation inside the chamber are conventionally addressed by using transparent walls, which enable penetration of most of the radiation (Greenwood and Beresford, 1979, Stannard and Weltz, 2006) but may sometimes produce artifacts of increased radiation (Steduto et al., 2002). For non-steady-state chambers, the foremost distortion is caused by build-up of gas concentration and pressure inside the chamber, decreasing the vapor flux and constraining the effective period of flux measurement. These effects have led to a preference for short measurement durations (usually up to 5 min, depending on chamber size), which are implemented before significant distortion develops. Other chamber effects are caused by pressure differences between the headspace and open atmosphere, resulting in suction of gas or depression of flux from the soil (Davidson et al., 2002), or radial diffusion below the chamber walls (Healy et al., 1996). The direction of these effects is mostly parallel, causing an underestimation of fluxes.

Due to the rising interest in forest floor soil CO₂ efflux, recent technical improvements have been made to these soil chambers, such as optimizing chamber dimensions to minimize concentration changes while maintaining high sensitivity; pre-installation of collars in the soil to increase sealing and prevent both radial diffusion and gas bursts during chamber installation; use of chamber fans at a speed sufficient to eliminate stratification but avoid pressure effects; and use of vents to equalize between chamber and ambient pressure. Davidson et al. (2002) summarized the results of several modeling, laboratory and field studies for CO₂ fluxes and concluded that non-steady-state chambers produce artifacts ranging from negligible to 15% underestimation, depending on chamber design, soil texture and water content.

Measurement of water-vapor fluxes with soil chambers is less common, but calibration is easier based on weight loss due to water E. In addition to this absolute calibration, less direct energy balance (Bowen ratio) and soil water budget (TDR or microlysimeter) are also employed. While most of the abovementioned chamber designs have generally been found to produce results that are in good agreement with other methods, in some reports, chamber measurements have been found to overestimate vapor fluxes (Grau, 1995, Dugas et al., 1997), attributed mainly to the effects of fan speed. In some cases, correction factors have been applied, mainly due to vapor condensation on the chamber walls (McLeod et al., 2004, Stannard and Weltz, 2006).

These studies imply that while technical difficulties and calibration issues need to be carefully considered, soil chambers still provide one of the few means to directly measure vapor fluxes from the soil under vegetated field conditions at high spatial and temporal resolutions.