Unsaturated hydraulic properties of vegetated soil under single and mixed planting conditions Open Access

Vegetation is known to affect the hydrology and hence slope stability (Osman & Barakbah, 2011; Smethurst et al., 2015). Plant roots cause changes in soil matric suction (Simon & Collison, 2002; Veylon et al., 2015; Ng et al., 2016a, 2018; Ni et al., 2017) through evapotranspiration and soil hydraulic properties, including the soil water retention curve (SWRC) and the soil hydraulic conductivity function (SHCF). Some studies (Table 1) have shown an increase in water retention capability when plant roots are present in the soil (Scanlan & Hinz, 2010; Rahardjo et al., 2014; Leung et al., 2015; Ng et al., 2016a, 2016b; Jotisankasa & Sirirattanachat, 2017), probably because of the blockage of soil pore space by roots (Buczko et al., 2007). However, some studies have reported an opposite result (Ng et al., 2016a; Jotisankasa & Sirirattanachat, 2017), arguably because of the formation of soil cracks due to, for instance, repeated soil shrinkage/swelling and root decay/growth (Vergani & Graf, 2015; Ng et al., 2016a; Ni et al., 2017; Leung et al., 2018).

There is a dearth of test data about the effects of plant roots on the SHCF (Table 2). Jotisankasa & Sirirattanachat (2017) show that root effects on hydraulic conductivity were prominent only when matric suction of the soil was less than 10 kPa, whereas the hydraulic conductivity measured by Song et al. (2017) found that roots affect unsaturated hydraulic conductivity for the entire suction range considered (<100 kPa). Thus, the presence of plant roots does not necessarily always reduce or increase unsaturated hydraulic conductivity, depending both on the plant and soil types. Indeed, although Rahardjo et al. (2014) and Jotisankasa & Sirirattanachat (2017) tested the same grass type, the soil hydraulic properties of the vegetated soils measured were different, possibly because of the different soil types considered in these two studies. Moreover, there has been no study that investigates the effects of multiple plant functional groups (i.e. mixed planting of herbaceous and woody species) on both the SWRC and SHCF (Tables 1 and 2).

The aim of this study is to investigate the unsaturated hydraulic properties of soil with four different vegetation conditions (i.e. bare, grass-only, tree-only and mixed tree–grass planting). Replications of instrumented soil columns were subjected to a controlled drying/wetting cycle, the results of which were used to determine the root effects on the SWRC and SHCF by way of the instantaneous profile method. Any plant-induced changes in the two soil hydraulic properties were interpreted with plant root traits.

Completely decomposed granite (CDG; silty sand, SM) was used for testing. At a dry density of 1777 kg/m³ (the compaction level considered in this study), the saturated hydraulic conductivity, k_s, of the CDG was 1·4 × 10⁻⁶ m/s. The other index properties are summarised in Table 3.

A tree (Schefflera heptaphylla; ivy tree) and a grass (Cynodon dactylon; Bermuda grass) species were selected for testing. These species are ecologically suitable for slope rehabilitation and restoration in many parts of Asia (GEO, 2011). Before transplantation, tree individuals with shoot length of 800 ± 35 mm (mean ± standard error of mean) and root depth of 140 ± 15 mm were provided by Tung Kee Garden Horticulture Ltd in Hong Kong. Grass turf with shoot length of 50 ± 12 mm and root depth of 40 ± 14 mm was used for testing.

After transplantation, the plants were left to grow for 4 months in a plant room (relative humidity 60 ± 5%, air temperature 25 ± 1°C, radiant energy 120 (μmol/m²)/s) to facilitate plant growth (Ng et al., 2016a). During the growing period, all bare and planted columns were irrigated every 3 days so that the soil moisture content was close to the field capacity of the CDG (20–22% by mass).

Soil columns (400 mm high and 200 mm dia.; Fig. 1) were constructed for this study. The CDG was compacted to the column up to a depth of 350 mm at a dry density of 1777 kg/m³. Drainage holes were made at the bottom of each column for free drainage. In total, nine planted columns were constructed, three for the tree-only cover, three for the grass-only cover and three for the mixed tree–grass plantation. One bare column was used as control.

A vertical array of miniature-tip tensiometers (2100 F, Soil Moisture Equipment Cooperation) was installed in each column to measure negative pore water pressure or matric suction of the soil (Fig. 1). At the same instrument depths, an array of four calibrated soil moisture probes (SM 300, Delta-T Device Ltd) was installed to measure the soil volumetric water content (VWC).

After 4 months of growing, the surface of all planted and bare columns were ponded with water until basal percolation was observed and suction at all instrumented depths became zero. Then, all columns were left in the plant room for evapotranspiration for 6 days (referred to as the drying test). Subsequently, the ten columns were ponded again, but with a controlled constant water head of 20 mm for 2 h using a Mariotte's bottle (referred to as the wetting test). During both the drying and wetting tests, the bottom holes of each column remained open for free drainage. Responses of suction, VWC and any basal percolation were recorded continuously.

After testing, root traits including root volume and root depth were measured from each planted column, following the procedures described by Reubens (2010). The root volume ratio, R_v, was obtained by normalising the measured root volume by the soil volume of that depth range.

The SWRC of each column was obtained by relating the measured suction and VWC at the same instrument depth. The VWC of each SWRC was divided by the soil porosity to obtain the degree of saturation, assuming that there is no soil volume change upon drying and wetting processes. Indeed, element tests performed by both Chiu & Ng (2012) and Leung & Ng (2016) show that CDG compacted to a similarly high dry density to that of the present study has negligible volume change when suction is less than 100 kPa. Moreover, there was no observed collapse during the first wetting. Each SWRC was fitted by the equation proposed by van Genuchten (1980) 

where S_r is the soil's degree of saturation; s is matric suction; a is related to the air-entry value (AEV); n and m control the shape of an SWRC.

The SHCF of each column was determined by the instantaneous profile method (Watson, 1966; Ng & Leung, 2012; Leung et al., 2016). The measured SHCF was then compared with the equation proposed by van Genuchten (1980) 

where k_r is the relative soil hydraulic conductivity, which is the ratio between soil hydraulic conductivity k and saturated hydraulic conductivity k_s.

The k_s value of each vegetated case was determined by back-analysing the suction data obtained during the wetting phase of each test using the numerical model developed by Shao et al. (2017). The k_s values are summarised in Table 3.

Statistical analysis was performed using Microsoft Excel. Significant differences were assessed with one-way ANOVA (analysis of variance), followed by post hoc Fisher's least-significant-difference test. Results were considered statistically significant when p-value ≤0·05. Different letters (i.e. a, b, c and d) were used to indicate statistical significance of differences among groups when the p-value is ≤0·05. This means that when any two groups (e.g. suction in bare and grass-only soil) have the same letter, they have no statistical difference. On the contrary, when they have different letters, the groups are significantly different statistically.

Figure 2 shows the R_v distributions with depth for grass-only, tree-only and mixed grass–tree cases, respectively. The R_v value of grass roots distributed almost linearly along the depth, peaked at the soil surface. The trees have a parabolic distribution of R_v, with the maximum R_v located approximately at the mid-depth of their root zone. The peak R_v of trees was almost 70% larger than that of grass in both single and mixed planting conditions. In the top 85 mm, the R_v of trees is statistically significantly higher than that of grasses (p-value <0·01). Whether the trees and grasses were planted individually or together (i.e. mixed plantation) has minimal effects on the R_v (Fig. 2).

When grown in relatively coarse soil (e.g. the silty sand tested in this study), plant roots tend to grow laterally to explore a greater soil volume for resources such as water and nutrients (Hamer et al., 2016). On the contrary, due to the relatively poor aeration and low hydraulic conductivity in fine-grained soil, root growth would be more restricted and localised (Travlos & Karamanos, 2006).

Figure 3(a) shows the measured and fitted drying SWRCs of the bare, grass-only and tree-only soils. The SWRCs of grass-only and tree-only soils are similar to each other (Table 4), and the amount of VWC retained for a given suction in these vegetated cases is statistically higher than that of the bare soil (p-value <0·001). Although the parameter n which describes the desorption rate of SWRC is similar between the bare and vegetated soils, the parameter a (which controls AEV) of both vegetated soils is noticeably lower than that of the bare case. This is consistent with the models proposed by Scanlan & Hinz (2010) and Ng et al. (2016b), who hypothesise that root occupancy in the pore space of coarse-grained soil would reduce the soil pore diameter, causing an increase in matric suction according to the capillary law. Indeed, the root diameter range, for both grasses and trees, is 0·15–2 mm. Recalling the capillary law and for a given surface tension, this diameter range affects the soil pore space that corresponds to a low range of matric suction (no more than 2 kPa). However, for fine-grained soil with clay content >12%, there are many factors possibly affecting the soil hydraulic properties, such as the release of organic matter as root exudates in the rhizosphere (Helliwell et al., 2014), soil aggregation due to plant–bacteria interaction in soil (Horn & Smucker, 2005) and/or the formation of microcracks/fissures associated with the continual drying–wetting process due to root water uptake (Daly et al., 2015).

The SWRCs of tree-only soils reported by Ng et al. (2016a) are superimposed in Fig. 3(b). They tested the same tree species and soil type as the present study and obtained the SWRC from soil that was planted with multiple trees with different spacings (60 and 180 mm; namely, test S60 and S180). When the tree spacing was wide, the SWRC was similar to that of single tree-only soil in the present study (Table 4). This is because the tree spacing is wide enough that the growth and water uptake action from each tree individual were not affected by the neighbouring trees (Ng et al., 2016a). For closer tree spacing, the water retention capability reduced as compared to the bare soil. The SWRC of this close tree spacing case is similar to the one obtained under the mixed planting condition in this study. In both instances, root decay is observed due to interspecies (tree–grass) competition and intra-species (tree–tree) competition. This may have created soil macro-pores (Ghestem et al., 2011), causing not only an increase in saturated hydraulic conductivity but also a reduction of water-holding capacity.

Figure 4(a) compares the relative drying SHCFs, k_r (i.e. normalised by k_s of the respective case). Each SHCF is obtained at 50 mm depth within the root zone, so any root effects can be investigated. Both Fig. 4(a) and Table 5 show that the reduction rate of k_r with respect to an increase in suction (characterised by the parameter n), decreased in grass- and tree-only cases, but increased in the mixed-species cases (p-value <0·001). This means that the presence of plant roots, depending on the plant types and planting method (i.e. single as opposed to mixed), does not only affect the AEV, but also plays a prominent role in affecting the ease of water flow in unsaturated soil (see both the parameters a and n in Table 5; p-value <0·001).

In Fig. 4(b), the best-fitted SHCFs of the four cases are compared with the predicted ones based on the best-fitted SWRC and k_s using the van Genuchten (1980) equation. Not surprisingly, the best-fitted and predicted k_r values for the bare soil are only slightly different. However, evidently, for tree- and grass-only cases, the predicted reduction rate of k_r is greater than the best-fitted one. On the contrary, for mixed tree–grass soil, the predicted reduction rate of k_r is less than the best-fitted case. Comparison of the results in Tables 4 and 5 reveals that, for a given vegetated condition, the fitted parameters for the SWRC are not always the same as those for the SHCF. This implies that the presence of plant roots changed the soil pore size and its distribution, which are the fundamental properties that govern soil water retention and hydraulic conductivity (Scholl et al., 2014; Ng et al., 2016b). Indeed, most existing predictive equations of the SHCF, including that suggested by van Genuchten (1980; equations (1) and (2)), do not take into account the root effects on the changes of soil pore size distribution and hence soil hydraulic properties. Based on the comparison in Figs 4(a) and 4(b), it may be important to link both the parameters a and n in the van Genuchten (1980) equation, or an equivalent parameter that describes the reduction rate of k_r in other prediction equations, with root trait(s).

This study has used the instantaneous profile method to quantify the effects of plant roots on unsaturated hydraulic properties of vegetated silty sand, under single- and mixed-species planting conditions. The water retention ability of both the tree-only and grass-only soils was greater than that of the bare soil. Although there was no discernible difference in terms of the rate of water desorption, the air-entry value of the silty sand increased substantially due to the presence of roots. However, under mixed-species planting where root decay was found, vegetated soil showed an evident reduction of the air-entry value. Compared with the bare soil, soils planted with single species showed reduced saturated hydraulic conductivity, whereas soils with mixed-species planting showed an increase due to preferential flow through soil macro-pores associated with root decay. Prediction of SHCF based on the known SWRC using an existing equation works well for bare soil, but there are discrepancies with measurements for all vegetated soil cases, either planted with single or mixed species. The rate of reduction of hydraulic conductivity is substantially overestimated for the tree- and grass-only cases, but is underestimated for the mixed planting case.

The authors acknowledge research grant 51778166 awarded by the National Natural Science Foundation of China, research grant 2012CB719805 from the National Basic Research Program (973 Program) administered by the Ministry of Science and Technology of the People's Republic of China and research grant HKUST6/CRF/12R awarded by the Research Grants Council of the Government of the Hong Kong SAR. The second author acknowledges the EU Marie Curie Career Integration Grant under the project ‘BioEPIC slope’.

a

fitting parameter in van Genuchten's equation (van Genuchten, 1980)

k

soil hydraulic conductivity

k_r

relative soil hydraulic conductivity

k_s

saturated hydraulic conductivity

m

fitting parameter in van Genuchten's equation (van Genuchten, 1980)

n

fitting parameter in van Genuchten's equation (van Genuchten, 1980)

R_v

root volume ratio

s

matric suction

S_r

degree of saturation of soil