Vegetation Effects on Soil Moisture and Groundwater Recharge in Subtropical Coastal Sand Dunes

Abstract

Vegetation can potentially have a strong influence on the water budget and hydrological processes in vegetated ecosystems due to canopy rainfall interception and the partitioning of rainfall into throughfall and stemflow, as well as root water uptake from the vadose zone or groundwater table. This study aims to explore the potential effects of vegetation on soil moisture dynamics and groundwater recharge in a subtropical coastal area of eastern Australia. This area is characterized by highly permeable sands, intense summer rainfall events and three typical vegetation covers (exotic pine plantation, native woodland and grassland). First of all, the spatial variability of both throughfall and stemflow at the soil surface was investigated in a 12-year-old managed pine plantation over one year on Bribie Island using tipping-bucket rain gauges. Rainfall loss by canopy interception and subsequent evaporation from this pine plantation and a native banksia woodland were also quantified and compared using field measurements and two analytical models of rainfall interception. In addition, the potential hydrological impacts of changes in vegetation cover in this shallow sandy groundwater system (depth to water table l 2 m) was evaluated by estimating groundwater recharge and discharge by evapotranspiration (ETg) under the three contrasting vegetation covers over a 2-year period using the water table fluctuation method and the White method, respectively. To further monitor the actual water percolation processes in deep sand dune profiles (depth to water table g 10 m), spatial patterns and seasonal dynamics of root-zone soil moisture were quantified under three contrasting vegetation covers on North Stradbroke Island by combining two geophysical techniques: surface electric resistivity tomography (surface ERT) and spatial time domain reflectometry (spatial TDR). Based on the field investigations of rainfall and root distributions at the under-canopy and inter-canopy zones, spatial distributions of vadose zone soil moisture and deep drainage in this subtropical coastal forest overlying deep sand dunes were finally simulated using HYDRUS models. On the Bribie Island sites, the highest throughfall was found on the east side of the tree trunks (~85% of gross rainfall) and the lowest in the midway between tree rows (~68% of gross rainfall) in the pine plantation. These spatial patterns persist for around 84% of recorded rainfall events. This is explained by canopy interception of the inclined rainfall resulting from the prevailing easterly wind direction throughout the experiment. Annual rainfall interception loss in the banksia woodland was lower (~16% of gross rainfall) than that in the pine plantation (~23% of gross rainfall) due to the lower canopy storage capacity and higher aerodynamic resistance of the banksia woodland. The RGAM and WiMo models predicted the interception losses from these forest stands reasonably well. The average annual gross recharge was largest at the sparse grassland site, followed by the exotic pine plantation and then native banksia woodland. Lower recharge values at forested sites are most likely resulted from higher rainfall interception losses and shallower water table depths. The pine plantation extracted more groundwater through ETg than the banksia woodland, whereas sparse grassland was found not consuming groundwater. In the open pine forest on North Stradbroke Island, the joint use of surface ERT and spatial TDR methods allowed spatially monitoring of root-zone moisture dynamics of the forest soils and the detection of typical features of rainfall interception, root water uptake and preferential infiltration of stemflow. Both surface ERT measurements and HYDRUS modelling identified higher soil moisture and deep drainage at the inter-canopy area relative to those under the canopy due to lower rainfall interception loss and higher root water uptake. The HYDRUS modelling experiments indicated deep drainage was underestimated by 130 mm to 162 mm (9.7%n12.0% drop compared to baseline scenario) as a consequence of uniform representation of spatial root systems in one- or two-dimensional HYDRUS models.The results of this study confirmed the vegetation in these coastal systems has a significant impact on the spatial distribution of rainfall at the soil surface and root water uptake, changing water infiltration and evapotranspiration patterns. Recharge in these shallow sandy aquifers is governed by seasonal rainfall but restricted in the wet season by wet antecedent soil moisture when the water table is approaching the soil surface, i.e., potential recharge is rejected. Groundwater use by vegetation is largely driven by potential ET but also limited by the depth to water table. The establishment of commercial pine plantations in these areas of native vegetation may reduce deep drainage and ultimately groundwater recharge, especially during extensive dry seasons, due to higher interception losses and groundwater uptake. In recharge modelling, our HYDRUS simulations show that translating the hydrological effect of the two-dimensional tree structure (rainfall redistribution and heterogeneous roots) to a one-dimensional lumped vertical conceptualization needs to be undertaken with caution.