1. Does hydrocarbon contamination induce water repellency and changes
in hydraulic properties in inherently wettable tropical sandy soils?
Ammishaddai Takawira, Willis Gwenzi ⁎, Phillip Nyamugafata
a
Department of Soil Science and Agricultural Engineering, University of Zimbabwe, P.O. Box MP167, Mt. Pleasant, Harare, Zimbabwe
a b s t r a c ta r t i c l e i n f o
Article history:
Received 4 March 2014
Received in revised form 24 July 2014
Accepted 27 July 2014
Available online xxxx
Keywords:
Pedotransfer functions
Petroleum hydrocarbons
Saturated hydraulic conductivity
Soil moisture retention curve
van Genuchten parameters
Water repellency
Hydrophobicity influences soil hydrological and ecological functions. Compared to naturally-occurring and fire-
induced hydrophobicity, limited information is available on the impacts of hydrocarbon contamination on water
repellency and hydraulic properties. Water repellency and hydraulic properties were measured on laboratory
simulated, and field contaminated soils, 1 and 5 years after an accidental petroleum hydrocarbon spill. The objec-
tives were; (1) to compare the water droplet penetration test (WDPT) to the molarity of ethanol droplet (MED)
test, (2) to investigate the effect of hydrocarbon contamination on water repellency and hydraulic properties, and
(3) to evaluate the performance of pedotransfer functions for hydraulic properties. The WDPT and MED tests
gave qualitatively similar water repellency results as evidenced by a significant positive correlation (p b 0.05,
r2
= 0.95) between the mean time for the two methods. Laboratory simulated hydrocarbon contamination in-
duced soil water repellency. Saturated hydraulic conductivity (Ks) increased linearly with level of contamination
(p b 0.05; r2
≈ 0.8), indicating that rapid flow of water attributed to a reduction of the dielectric constant,
and hence water–soil matrix interactions. No water repellency was observed in contaminated field soils
(WDPT b 3 s), but the residual signature of hydrocarbon contamination was evident in other soil properties par-
ticularly electrical conductivity. This indicates that natural soils were inherently wettable and that hydrocarbon-
induced hydrophobicity could be transient. This non-persistence was attributed to high decomposition rates
stimulated by tropical conditions and nutrients added to promote revegetation. Predictions of pedotransfer func-
tions were comparable to measured hydraulic data (p b 0.05, r2
N 0.8), confirming their general validity for water
and solute transport modeling even on contaminated soils. The study confirmed the hypothesis that hydrocarbon
contamination induces water repellency and reduces soil moisture retention at low suction (b100 kPa) for lab-
oratory contaminated soils, but effects were inconsistent for field samples. However, the increased saturated hy-
draulic conductivity associated with laboratory contaminated soils contradicted the original hypothesis. The
findings imply that storms falling on initially dry recently contaminated soils may trigger contaminant transport
and erosion via enhanced surface runoff, and rapid spreading of contaminants once they reach the groundwater
systems. These hydrological impacts are critical for remediation of contaminated sites. Future research could use
a contamination chronosequence/gradient to provide comprehensive information on the temporal evolution of
water repellency and hydraulic properties under field conditions.
© 2014 Elsevier B.V. All rights reserved.
1. Introduction
Hydrophobicity or water repellency is a well-recognized phenome-
non influencing soil hydrological behavior and agricultural productivity.
Water repellent soils resist wetting, and inhibit infiltration (Dekker and
Ritsema, 1994). Naturally-occurring and fire-induced water repellency
has been the subject of several studies conducted in Mediterranean en-
vironments in Australia, Spain, Portugal and Chile (Doerr et al., 2000,
2006) and sandy dunes in Netherlands (Dekker and Ritsema, 1994),
where water repellency appears more widely reported than in other
regions. Besides fire and antecedent soil moisture, the occurrence and
severity of water repellency are also influenced by soil type and proper-
ties (Badía et al., 2013; Doerr et al., 1996), vegetation type, soil manage-
ment and land use practices (Harper et al., 2000; Zavala et al., 2009). For
example, although water repellency has often been associated with
coarse-textured soils, several studies have shown that severe water re-
pellency also occurs in various soil types including those that are fine-
textured, aggregated, acidic and alkaline soils (Doerr et al., 2000;
Jordán et al., 2013; Mataix-Solera and Doerr, 2004).
In earlier studies, researchers investigated the origin and character-
istics of water repellency (Doerr et al., 2000; Jordán et al., 2013), evalu-
ation methods (e.g. Doerr, 1998; King, 1981; Letey, 1969; Letey et al.,
2000; Watson and Letey, 1970), impacts on hydrological behavior in-
cluding preferential flow (Dekker and Ritsema, 1994; Wallach, 2010;
Geoderma 235–236 (2014) 279–289
⁎ Corresponding author.
E-mail addresses: shaddytakaz@gmail.com (A. Takawira), wgwenzi@yahoo.co.uk,
wgwenzi@agric.uz.ac.zw (W. Gwenzi), pnyamugafata@gmail.com (P. Nyamugafata).
http://dx.doi.org/10.1016/j.geoderma.2014.07.023
0016-7061/© 2014 Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Geoderma
journal homepage: www.elsevier.com/locate/geoderma

2. Wallach and Jortzick, 2008) and amelioration and management prac-
tices (e.g. Dlapa et al., 2004). The causes of hydrophobicity include
plant derived waxes, humic and fulvic acids and organic compounds
from forest fires (Arcenegui et al., 2007; Doerr et al., 2000; Huffman
et al., 2001; Ritsema et al., 1993; Scott, 2000). Water repellency influ-
ences water redistribution via reduced infiltration, enhanced surface
runoff and erosion, and preferential flow or fingering (Doerr et al.,
2000). These changes in hydrological balance, may in turn impact on
soil–plant water relations, resulting in impeded seed germination,
stunted plant growth and reduced plant productivity (Mainwaring
et al., 2004). Other researchers have investigated the potential to ame-
liorate water repellency and the associated impacts through localized ir-
rigation, tillage, and application of clays and surfactants or wetting
agents (Buczko et al., 2006; Dlapa et al., 2004; Kostka, 2000). Several
methods exist for evaluating the occurrence and severity of soil water
repellency, the molarity of ethanol droplet (MED) and the water droplet
penetration test (WDPT) being the most prominent (Dekker and
Jungerius, 1990; Dekker and Ritsema, 1994; King, 1981). However,
comparative studies on their performance particularly on hydrocarbon
contaminated soil are limited. Therefore uncertainty exists about the
sensitivity and comparability of results between the two methods.
In comparison to other areas, little is known about the occurrence of
water repellency in the predominant tropical soils of southern Africa.
The reason for this lack of information is unclear, but could be indicative
of the general lack of hydrological research in the region. An exception is
a study by Scott (2000) documenting water repellency and reduced in-
filtration and enhanced runoff in an exotic eucalyptus timber and pine
plantation, and natural Acacia dominated miombo woodland in South
Africa. The miombo woodlands are the dominant native vegetation
type in southern Africa, covering over 3.6 million km2
across 11 coun-
tries (Timberlake and Chidumayo, 2011). The miombo woodlands con-
sist predominantly of deciduous broad-leaved leguminous trees with a
well-developed grass understory, giving rise to frequent and wide-
spread veld fires. Although documented cases of naturally-occurring
or fire-induced water repellency are scarce in the region, soil contami-
nation through anthropogenic activities could potentially cause water
repellency. In particular, wastewater irrigation, soil application of bio-
solids and hydrocarbon contamination may introduce hydrophobic
organic compounds into the soil system. However, compared to
naturally-occurring and fire induced hydrophobicity, little is known
about the impacts of contamination on water repellency and soil hy-
draulic properties.
Aislabie et al. (2004) noted that few studies exist on the impacts of
hydrocarbon contamination and associated additives on water repellen-
cy and moisture retention. In an arid region, wastewater irrigation has
been reported to cause water repellency (Wallach et al., 2005). A
study conducted in Canada on weathered oil-contaminated sites
showed that some long-chain and polycyclic aliphatic organic com-
pounds of petroleum origin were water repellent (Roy et al., 1999).
On Barrow islands in Australia, George et al. (2011) observed that
flowline additives associated with oilfield installation had no effect on
water repellency. In the Antarctic region, hydrocarbon-contaminated
soils were weakly hydrophobic, but impacts on moisture retention
were negligible (Aislabie et al., 2004). Elsewhere, hydrocarbon contam-
ination was also reported to alter soil field capacity, porosity, soil bulk
density and optimum water content even at low hydrocarbon contam-
ination levels (Adams and Cruz, 2008; Adams et al., 2008; Caravaca and
Rolda, 2003; Rahman et al., 2010). In other studies, soil contamination
by petroleum hydrocarbons was reported to increase the moisture re-
tention of soil at high suction values (Burckhard et al., 2004; Hyun
et al., 2008), while a decline in water retention was observed by Roy
and McGill (1998). These changes often result in reduced plant growth
and productivity (Adams and Cruz, 2008). In summary, the findings of
these earlier studies are inconsistent, and often contradicting. More-
over, the bulk of these studies were drawn from cool and humid tem-
perate and arctic conditions (Adams and Cruz, 2008; Balks et al., 2002;
Foght and Waterhouse, 2004; Quyum, 2000). By contrast, there is a pau-
city of information on the impacts of hydrocarbon contamination on
water repellency and hydraulic properties in tropical environments typ-
ical of southern Africa. Unlike temperate and Arctic environments, the
tropics experience distinct warm to hot and seasonally dry climatic con-
ditions, resulting in diverse soil types. These unique climatic and soil
characteristics constrain the extrapolation and generalization of find-
ings obtained in other environments.
Knowledge of soil hydraulic properties is crucial for understanding
the hydrology and remediation of contaminated sites (Gwenzi et al.,
2011). Soil hydraulic properties particularly saturated hydraulic
conductivity (Ks) and soil moisture retention (SMR) influence soil mois-
ture storage, deep drainage, runoff and infiltration, and provide key
inputs for water balance and solute transport models (Gwenzi,
2010; Gwenzi et al., 2011; Holländer et al., 2009). Most existing water
and solute transport models rely on hydraulic properties estimated
from pedotransfer functions derived for uncontaminated natural soils
(Bohnhoff et al., 2009; Holländer et al., 2009). Hydrocarbon contamina-
tion could potentially cause water repellency and associated changes in
hydraulic properties. Consequently, PTFs for Ks and SMR developed for
uncontaminated natural soils may fail to predict field measurements
on such contaminated soils. Therefore, there is need to evaluate the ca-
pacity of existing pedotransfer functions to predict saturated hydraulic
conductivity and soil moisture retention for hydrocarbon contaminated
soils. In the current study we investigated the hypothesis that hydrocar-
bon contamination induces water repellency and reduces moisture re-
tention and saturated hydraulic conductivity in inherently wettable
tropical sandy soils. The objectives of the study were; (1) to compare
the water droplet penetration test (WDPT) to the molarity of ethanol
droplet (MED) as water repellency tests, (2) to investigate whether
hydrocarbon contamination induces water repellency and changes in
soil hydraulic properties, and (3) to evaluate the performance of
pedotransfer functions for soil moisture retention curve and saturated
hydraulic conductivity.
2. Materials and methods
2.1. Description of study sites
The study was conducted on two field sites in Zimbabwe; Ruwa
(E 031° 13′ 04.0″, S 17°52′ 52.7″, altitude: 1521 m asl) and Goromonzi
(E 031° 24′ 10.9″, S 180° 07′ 54.0″, altitude: 1609 m asl). The sites
were located along the Mutare highway, approximately 10 (Ruwa)
and 30 km (Goromonzi) from Harare, the capital city of Zimbabwe.
The highway links Zimbabwe to the international seaport of Beira in
Mozambique, and is frequently used by oil tankers for the transport of
oil and other petroleum products. The two sites were approximately
10 km apart, and had similar soils, vegetation types and climatic condi-
tions. The climate of the area is tropical, characterized by distinct warm
wet summers (27 °C) and cool dry winters (17.5 °C). Average annual
rainfall is about 800 mm, occurring mainly in summer stretching from
November to February. Soils are predominantly in-situ sands derived
from granites. They are classified as Harare 6G.2 according to the
Zimbabwe soil classification system, corresponding to Udic Kandiustalf
(USDA, 1994) and Gleyic Luvisol (FAO, 1988) (Nyamapfene, 1991). Nat-
ural vegetation in the study area is miombo woodlands consisting of de-
ciduous trees and a grass understorey.
Two sampling sites representing hydrocarbon-contaminated and
uncontaminated soils (control) were selected within each study site.
Contaminated soils were selected from sites that experienced a
large spillage of petroleum hydrocarbons through road accidents in-
volving oil tankers. Petroleum hydrocarbon contamination occurred in
2007 at Ruwa and 2012 at Goromonzi. Sampling was conducted be-
tween January and April 2012, approximately 5 and 1 year later, respec-
tively. No clean-up or remediation was conducted at the Ruwa
site, while at Goromonzi, post-contamination remediation involved
280 A. Takawira et al. / Geoderma 235–236 (2014) 279–289

3. application of nutrient solution containing 9:1 phosphorus to nitrates at
a rate of 20 l ha−1
diluted in 200 l. The application of the nutrient solu-
tion was meant to promote vegetation growth. Field observations
showed that contaminated sites differed from the control sites in
terms of soil color and smell. Moreover, compared to the control sites,
sedges and grasses growing on the contaminated sites were sparse, yel-
lowish and stunted. Control sampling areas were uncontaminated sites
identified within natural undisturbed miombo woodlands approxi-
mately 1 km from the road and upstream of the contaminated area.
The identification of contaminated and control sites was conducted
with the assistance of officials from the Hazardous Substances Inspec-
torate of the Zimbabwe Environmental Management Agency (EMA).
EMA documents and maintains an up-to-date database of all freights
and accidents involving hazardous substances in Zimbabwe. The agency
also provides emergency response services including the clean-up and
remediation of contaminated sites. Therefore historical information in-
cluding the nature of the hydrocarbons, year of contamination and
post-contamination clean-up and rehabilitation activities were readily
available for the study sites. Additional information on the spatial extent
of the contamination, and whether there were any fire outbreaks were
obtained from local people who witnessed the accidents, and have lived
in the study sites since then.
To address the study objectives, soil samples for laboratory simula-
tion of hydrocarbon contamination were collected from the control
sites. Contaminated and control field soil samples were also collected
from both sites for general soil characterization, water repellency tests
and measurement of hydraulic properties. Details of the sampling pro-
tocol are described in Section 2.2.
2.2. Evaluation of water repellency
2.2.1. Laboratory hydrocarbon contaminated soils
Disturbed composite samples from the control within each site were
used for laboratory simulation of hydrocarbon contamination. Each
composite sample weighing approximately 50 kg consisted of five ran-
dom sub-samples collected from the top 10-cm depth using a spade.
The samples were air-dried under room conditions, thoroughly mixed
and then passed through a 2-mm sieve. Six levels of hydrocarbon con-
tamination replicated three times were tested; control (0.0), 30, 60,
120 and 150 mg of diesel per gram of soil. These contamination levels
are consistent with experimental values used in previous studies
(Adams and Cruz, 2008; Millioli et al., 2009) and concentrations mea-
sured in field contaminated soils (Aislabie et al., 2004). To facilitate ho-
mogeneous mixing with soil, the corresponding volume of diesel was
added to a 200-ml flask, and the volume made up to the mark with ac-
etone (Awada et al., 2004). The acetone was expected to evaporate
without altering the water repellency properties of the soil. Neverthe-
less, a control soil treated with 170 ml of acetone only was included to
confirm that acetone had no effect on water repellency. For each repli-
cate, one kilogram of soil was weighed into a container. The diesel–ace-
tone solution was then added and thoroughly mixed by hand. The soil
was then kept for 14 days under room conditions (25–26 °C). The soil
was mixed at two-day intervals during the aging process.
The contaminated soils were then oven-dried at 60 °C for 48 h, and
then allowed to equilibrate for two days under room conditions
(Contreas et al., 2008). A portion of soil from each replicate was then
put in a petri dish and the surface smoothened. Two methods were
compared for the evaluation of water repellency, namely, the water
drop penetration time method (WDPT) (Letey, 1969; Watson and
Letey, 1970) and the molarity of ethanol droplet test (MED) (King,
1981). For the WDPT, a syringe was used to place a droplet on the
smoothed surface and time required for the water droplet to penetrate
the soil surface recorded using a stopwatch. The WDPT values of each
sample were given by the average of seven drops of distilled water
placed onto the smoothened surface of 30–40 g of soil placed in a
petri dish (Contreas et al., 2008). Samples were classified as hydrophilic
or wettable (below 5 s), slightly water repellent (5–60 s), strongly
water repellent (60–600 s), severely water repellent (600–3600 s)
and extremely water repellent (more than 3600 s) (Dekker and
Jungerius, 1990; Dekker and Ritsema, 1994).
In the MED test water repellency was recorded as the molarity of
ethanol in a droplet of water needed to penetrate the soil within 10 s
(King, 1981). Solutions ranging from 0 to 6 M concentration at 0.2 M in-
tervals were used for the MED test (Roy et al., 1999). Results of the MED
tests were evaluated using the criteria developed by King (1981). The
classification system consists of four categories; non-water repellent
(MED = 0), low water repellency (0 b MED b 1), moderate water repel-
lency (MED = 1–2.2) and severe water repellency (MED N 2.2) (King,
1981).
2.2.2. Field hydrocarbon contaminated soils
To investigate the impacts of hydrocarbon contamination under
field conditions, contaminated and control soils were collected from
both sites. The soils were used for general characterization, water repel-
lency test and measurement of soil moisture retention and saturated
hydraulic conductivity. Metal cores (7 cm diameter and 5 cm height)
were used to collect undisturbed samples for the determination of
field water repellency, soil bulk density (Blake and Hartge, 1986) and
total porosity. As will be presented later, comparison of the MED and
WDPT data showed that the two methods gave similar results. There-
fore, the WDPT, which was quicker and cheaper, was used in subse-
quent water repellency tests conducted on field samples.
2.3. Determination of soil chemical properties
Soil pH, electrical conductivity (EC) and organic carbon (SOC) were
measured on triplicate samples from the contaminated and control
sites. Soil pH and electrical conductivity were measured in 1:5 soil:
water suspension using standard methods (Gwenzi et al., 2011;
Rayment and Higginson, 1992). Briefly, the soil–water suspension
were shaken for 1 h on a mechanical shaker and allowed to settle for
30 min (Gwenzi et al., 2011). Soil pH and electrical conductivity were
measured using EC and pH electrodes, respectively (model: Mettler To-
ledo). The modified Walkley–Black method was used for determining
SOC on samples passed through a 0.5-mm sieve (Okalebo et al., 1993).
The method is based on wet oxidation of soil organic carbon using po-
tassium dichromate with external heating at 145 °C.
2.4. Determination of soil hydraulic properties
Particle size distribution, soil bulk density, total porosity, soil mois-
ture retention and saturated hydraulic conductivity were measured to
evaluate the effects of hydrocarbon contamination on hydraulic proper-
ties. Particle size analysis was done using a combination of wet sieving
and sedimentation (Gee and Bauder, 1986). The core method (Blake
and Hartge, 1986) was used for the determination of bulk density
using metallic cores measuring 7 cm by 5 cm. The soil cores were
then oven dried at 105 °C for 24 h. Total porosity (n) was calculated as;
n = 1 − ρb/ρs, where ρb is the dry bulk density of the soil (kg m−3
),
and ρs is the particle density assumed to be 2650 kg m−3
(Hillel, 1998).
The pressure plate method was used to determine soil moisture re-
tention curves on core samples collected from the field (Klute and
Dirksen, 1986). Saturated hydraulic measurements were conducted on
repacked laboratory contaminated soils. Samples were subjected to suc-
tions of 2, 5, 10, 33, 100 and 200 kPa and allowed to equilibrate in the
pressure plate chamber. At equilibrium, samples were oven-dried at
105 °C and then weighed to determine soil moisture content at each
suction level. Single- and dual-porosity PTF models were then fitted to
the data using the RETC software (van Genuchten et al., 1991).
Section 2.5 presents the details of the PTFs investigated and the perfor-
mance evaluation criteria.
281A. Takawira et al. / Geoderma 235–236 (2014) 279–289

4. As water repellency tests showed that all field contaminated sam-
ples were non-water repellent, evaluation of the effect of hydrocarbon
contamination on Ks was limited to laboratory contaminated samples.
Ks was determined by the constant-head method (Reynolds and
Elrick, 2002) using a set-up developed by Verboom (1991). The setup
was originally designed and applied to investigate the effect of electro-
lytes on hydraulic conductivity of Zimbabwean soils (Verboom, 1991).
The set-up consisted of a plastic permeameter consisting of two double
rings; an outer ring with an internal diameter of 59 mm attached to
inner ring with internal diameter of 37 mm. Laboratory contaminated
bulk samples from the previous experiment were repacked in the plas-
tic permeameters to a bulk density of approximately 1500 kg m−3
. Wire
gauze was placed on the soil surface to reduce the impact of water drop-
lets. A burette was used to provide a constant supply of water by
allowing water to fall on the soil surface from a height of 25 cm to main-
tain a constant head of 2 cm above the soil column. The soil column had
a height of 5 cm and a diameter of 37 mm. To reduce edge effects, out-
flow data were collected from the internal ring and used to compute Ks.
Outflow data were measured using a balance to the nearest 0.00 g and
time monitored using a stopwatch. Cumulative outflow data were plot-
ted against time to obtain a straight line whose gradient represented av-
erage flow rate. Darcy's law was used to calculate Ks using a hydraulic
head of 7 cm and column dimensions of 5 cm height and diameter of
37 mm (Reynolds et al., 2000, 2002).
2.5. Evaluation of pedotransfer functions
To evaluate the capacity of existing pedotransfer functions (PTFs),
laboratory-measured saturated hydraulic conductivity (Ks) and soil
moisture retention curve (SMRC) data were compared to PTF model
predictions. Specifically, four prominent PTFs for estimating Ks were
evaluated (Sobieraj et al., 2001; Tietje and Hennings, 1996). These
PTFs have been investigated in previous studies conducted on natural
soils and artificial substrates including mine wastes (Gwenzi et al.,
2011; Sobieraj et al., 2001; Tietje and Hennings, 1996). The van
Genuchten (1980) model is the most widely used PTF for prediction of
soil moisture retention curves (Ippisch et al., 2006). Three forms of the
single-porosity van Genuchten (1980) model; VG1, VG2 and VG3
were used for SMRC prediction. The three forms differed in the value
of the van Genuchten parameter, m (Table 1). RETC software (version
6.02) (van Genuchten et al., 1991) was used for model fitting and esti-
mation of the air-entry value (α) and pore size distribution (n) param-
eters. Table 1 presents the PTF models and their input data on particle
size distribution and soil bulk density.
2.6. Data analysis
Shapiro–Wilk's and Barlett's tests were used to test data normality
and homogeneity of variances, respectively, at 5% level. Two-sample t-
test and analysis of variance (ANOVA) were conducted to test effects
of site, hydrocarbon contamination, and interactions on water repellen-
cy and hydraulic properties. Non-normal data were transformed to
achieve normality, and those that failed to meet assumptions for para-
metric tests were analyzed using non-parametric statistical tests. Re-
gression was used to test the relationship between the two methods
for determining water repellency, and relations between hydrocarbon
contamination levels, water repellency (WR) and hydraulic properties.
Statistical tests were done at p = 0.05 using Minitab version 16.0 statis-
tical package. Least significant differences (lsd) were used for the sepa-
ration of treatment means at probability level (p) of 0.05. Correlation
was used to evaluate the performance of PTF models for SMRC using
the coefficient of determination (r2
) to assess the goodness of fit.
Given that Ks varies considerably with method of measurement
(Gwenzi, 2010; Gwenzi et al., 2011; Johnston et al., 2009; Muñoz-
Carpena et al., 2002), comparison of measured Ks values to those pre-
dicted by PTF model focused on evaluating whether values were within
the same orders of magnitude. Therefore, box-and-whisker plots show-
ing median, minimum, maximum and 25th and 75th percentile values
of measured and predicted values were used for this purpose.
3. Results
3.1. General soil properties
The soils at the two study sites were classified as well-sorted pre-
dominantly fine sandy loam and fine loamy sand with approximately
80–88% sand (Fig. 1). At both sites, the differences in particle size distri-
bution between the control and contaminated soils were considered in-
sufficient to justify classifying into different textural classes according to
the USDA Soil Classification (Fig. 1). Therefore, the soils were considered
to have comparable particle size distribution.
The soils at both sites had mean bulk density values of
1350–1613 kg m−3
. Soil bulk density, porosity, SOC and EC varied
significantly (p b 0.05) between the control and contaminated sites,
but the trend was inconsistent (Table 2). The Ruwa control site had sig-
nificantly (p = 0.002) lower bulk density, and hence higher total poros-
ity than the contaminated site, while comparable values were observed
for the Goromonzi sites (Table 2). On the other hand, the Goromonzi
contaminated site had higher SOC than the control, while those for
the Ruwa soils were comparable. Soil EC was consistently higher for
the contaminated sites than the control. All soils were characterized
by similar slightly acidic soil pH. No remarkable pH differences were ob-
served between the contaminated soil and the control at the Goromonzi
site. The pH of the Ruwa contaminated soil was significantly higher than
that of the control.
3.2. Soil water repellency
Laboratory-contaminated soil samples were used to compare the
WDPT and the MED tests. The average time for the WDPT and the
MED increased exponentially with increasing concentration of hydro-
carbon contamination, with coefficients of determination (r2
) of 0.94
and 0.91, respectively (Fig. 2(a)). Accordingly, for any given concentra-
tion of hydrocarbon contamination, the mean time for WDPT and that of
MED were also positively correlated (p b 0.001, r2
= 0.97–0.99)
(Fig. 2(b)), demonstrating that the two methods yielded qualitatively
similar results.
The WDPT showed that hydrocarbon contamination exceeding
30 mg g− 1
caused soil water repellency at both sites. Significant
(p b 0.05) differences in WDPT were observed among all treatments ex-
cept the control and the 30 mg g−1
concentration, which were similar
Table 1
Pedotransfer functions used to predict soil moisture retention curve (SMRC) and saturated
hydraulic conductivity (Ks, mm h−1
). Input data for PTFs are moisture retention data for
SMRC and sand (Sa): 50–2000 μm, silt (Si): 2–50 μm and clay (C): ≤2 μm and dry soil
bulk density (ρb) for Ks.
Reference Pedotransfer function
Soil moisture retention curve (SMRC):
1. Van Genuchten (1980) (VG1)
θ hð Þ ¼
θr þ θs −θr
1þ αhj j
n
½ Š
m h≤0
θs hN0
(
2. Van Genuchten (1980)
(Mualem) (VG2)
VG1 with m = 1–1/n
3. Van Genuchten (1980)
(Burdine) (VG3)
VG1 with m = 1–2/n
Saturated hydraulic conductivity (mm h−1
):
1. Puckett et al. (1985) Ks = 156.96 exp[−1975C]
2. Dane and Puckett (1994) Ks = 303.84 exp[−0.144C]
3. Cosby et al. (1984) Ks = 25.4 × 10[−0.6 + 0.0126Sa − 0.0064C]
4. Jabro (1992) Ks = 10 × 10[9.56–0.81log(Si) − 1.09log(C) − 4.64ρ
b
]
In the van Genuchten model, θ is the volumetric water content, h is pressure head (cm); θr
and θs are the residual and saturated water contents, respectively. The parameter α is re-
lated to the inverse of the air-entry value, and n is a measure of the pore-size distribution
(van Genuchten, 1980).
282 A. Takawira et al. / Geoderma 235–236 (2014) 279–289

5. (Fig. 3). According to the classification system of Dekker and Ritsema
(1994), the soils from both sites were grouped into three water repel-
lency classes; non-water repellent (control and 30 mg g−1
), slightly
water repellent (60.0 mg g−1
) and strongly water repellent (120 and
150 mg g−1
). The control and contaminated field soils at both sites
were classified as wettable (Dekker and Ritsema, 1994), with very low
WDPT time (b1 s), which were approximately 5-times lower than 5-s
minimum for a slightly wettable soil. At both sites, the WDPT for
the control soils in a native miombo woodland were significantly higher
(p = 0.05) than that of the contaminated soils. The results of the MED
test for both Ruwa and Goromonzi soils were qualitatively similar to
that of the WDPT. Based on MED, the control and 30 mg g−1
were
classified as non-water repellent (MED = 0), while the 60, 120 and
150 mg g−1
hydrocarbon concentrations were classified as severely
water repellent with (MED: 5–8). Both WDPT and MED tests classified
the control soils treated with acetone as non-water repellent, demon-
strating that acetone used as a solvent for the petroleum hydrocarbons
had no effect on the observed water repellency. WDPT was strongly
and positively correlated to the concentration of hydrocarbon contami-
nation, with regression coefficients (r) of 0.90 and 0.95 for Ruwa and
Goromonzi, respectively. For Goromonzi, the times for WDPT and
MED were related to hydrocarbon concentration (mg g−1
) (HC)
by power functions; WDPT = 1.10e0.033 ∗ HC
, r2
= 0.94 and MED =
1.32e0.037 ∗ HC
, r2
= 0.91, respectively. A similar relationship was
obtained for Ruwa.
3.3. Impact of hydrocarbons on soil hydraulic properties
3.3.1. Soil moisture retention curve
Soil moisture content for the control and contaminated soils de-
clined gradually for suctions below 100 kPa for both sites, followed by
a rapid decline at 200 kPa (Fig. 4). The effect of hydrocarbon contamina-
tion was more pronounced for laboratory-contaminated soils than field
samples. Hydrocarbon contamination significantly (p b 0.01) reduced
soil moisture retention in laboratory contaminated soils (Fig. 4). At
low suction ranges (2–33 kPa), the 120 and 150 mg g−1
treatments
had consistently lower (p b 0.05) moisture content than the other treat-
ments (0–60 mg g−1
), which were comparable. On the other hand, the
effects of hydrocarbon contamination on soil moisture retention of field
samples were inconsistent between sites. At Ruwa, the control soil had
significantly (p b 0.05) higher moisture content at 2-kPa suction than
the contaminated soil, but no significant effects were observed at
other suction levels (Fig. 4). At Goromonzi, for any given suction, soil
moisture retention of the contaminated soils was similar to that of the
control (Fig. 4).
Evaluation of PTFs for SMRC only focussed on single-porosity models
of van Genuchten with various functions (e.g. Mualem, 1976) for esti-
mating the pore size distribution parameter (n). The soil moisture re-
tention curves for the laboratory contaminated soils was best
described by the single porosity van Genuchten (1980) model (VG1)
(r2
= 0.92–0.99, p ≤ 0.003) (Fig. 4(a)). Similar results were observed
for the control and contaminated soils at both sites (p b 0.001, r2
=
0.98) (Fig. 4(b) and (c)). No attempts were made to fit double or
dual-porosity models (e.g. Mualem, 1976), which are often used on
well-structured soils and those with a substantial proportion of rock
fragments and fractures, where both preferential and matrix flow are
dominant. This was motivated by the fact that particle size distribution
and soil moisture retention data showed no evidence of bimodality.
The effects of hydrocarbon contamination on the van Genuchten
parameters n, α and air-entry value 1
α
À Á
of field- and laboratory-
contaminated soils were also investigated. Under laboratory simulation
hydrocarbon contamination had a significant effect (p b 0.05) on n, α
and air-entry value particularly at concentrations above 30 mg g−1
(Table 3). In general, hydrocarbon contamination significantly in-
creased α, and consequently reduced the air-entry value. On the other
hand, the pore size distribution index n for laboratory contaminated
Fig. 1. Semi-logarithmic plot of particle size distribution for control (○) and hydrocarbon contaminated (●) field soils at Goromonzi (a) and Ruwa (b).
Table 2
Summary properties of field soil samples from Goromonzi and Ruwa used in the study. Values shown are mean ± standard error (SE) for sample size n = 3 except for soil bulk density
and total porosity (n = 5).
Characteristic Goromonzi Ruwa
Control Contaminated Control contaminated
Soil bulk density (kg m−3
) 1600 ± 44.1 1470 ± 30.9 1350 ± 51.2 1613 ± 30.9
Total porosity (%) 39.7 ± 1.7 44.4 ± 1.2 49.1 ± 1.9 39.7 ± 1.8
Soil organic carbon (%) 1.12 ± 0.03 1.43 ± 0.01 0.97 ± 0.03 0.91 ± 0.01
pH (1:5 soil:water) 6.2 ± 0.1 6.0 ± 0.0 6.0 ± 0.0 6.4 ± 0.0
EC (1:5 soil:water) (μS m−1
) 37 ± 0.9 61 ± 1.3 37 ± 2.5 80 ± 3.6
283A. Takawira et al. / Geoderma 235–236 (2014) 279–289

6. soils generally decreased with increasing hydrocarbon contamination.
Field contaminated soils had significantly lower α (p b 0.05) and
hence higher air-entry value than the uncontaminated control, but ef-
fects on n were not significant at both field sites.
3.3.2. Saturated hydraulic conductivity
Hydrocarbon contaminated laboratory soils caused a significant in-
crease in Ks at both sites (Fig. 5), but site effects were not significant.
At Goromonzi, all hydrocarbon contaminated soils had significantly
higher (p b 0.001) Ks than the control. A similar trend was observed at
Ruwa, but the 30 mg g−1
treatment had Ks similar to that of the control.
Although site had no significant effect on Ks, Ruwa had slightly higher
Ks (4.27 × 10−5
m s−1
) than Goromonzi (3.27 × 10−5
m s−1
).
A significant linear relationship (p b 0.05) was observed between
the level of hydrocarbon contamination (HC) and Ks at both Ruwa
(Ks = 5 × 10−7
∗ HC + 1 × 10−5
, r2
= 0.88) and Goromonzi (Ks =
3 × 10−7
∗ HC + 9 × 10−5
, r2
= 0.78).
Evaluation of five PTFs on Goromonzi soil revealed that predicted
median Ks values (1.0–4.0 × 10−5
m s−1
) were of similar orders of
magnitude to laboratory-measured ones (2 × 10−6
–2.4 × 10−5
m s−1
)
(Fig. 6(a)). The Puckett, and Dane and Puckett PTF models yielded the
closest estimate of Ks for the control and the 30 mg g−1
hydrocarbon con-
centration, but tended to under-estimate Ks for higher level of hydrocar-
bon contamination (60–150 mg g−1
). The opposite trend was observed
for the Jabro PTF model, which over-predicted Ks for low hydrocarbon
concentrations (control and 30 mg g−1
) by approximately an order of
magnitude, but gave estimates comparable to laboratory measured
values for higher hydrocarbon concentrations (60–150 mg g−1
)
(Fig. 6(a)). The performance of PTFs on Ruwa soil was generally
comparable to that of Goromonzi soil. The only exception was that the
Puckett, and Dane and Puckett PTF models performed equally well for
the 60 mg g−1
hydrocarbon contamination (Fig. 6(b)). The Jabro PTF
model also tended to under-predict the median Ks for all samples, but
its values were within the range for the 150 mg g−1
hydrocarbon con-
tamination. For both soils, laboratory-measured Ks showed higher vari-
ability than predicted values. However, it is noteworthy that the
evaluation of PTFs compared predicted values based on input data mea-
sured on control and contaminated field samples to those measured on
the control and laboratory-contaminated samples. This was motivated
by the findings of the current and previous research (e.g. Roy et al.,
1999) demonstrating that hydrocarbon contamination of laboratory
soils had negligible effects on the input data for the PTFs for Ks (PSD,
bulk density and soil organic carbon).
4. Discussion
Hydrocarbon contamination via accidental spills, leakages from
oilfield installations and disposal of used petroleum products potential-
ly alters soil hydrological and ecological functions. Understanding the
impacts of hydrocarbon contamination on soil and chemical properties
is critical for remediation of contaminated sites. The current study in-
vestigated water repellency and hydraulic properties of laboratory and
field contaminated sandy loam and loamy sands derived from granitic
parent material in tropical Zimbabwe. The soils were inherently wetta-
ble, and consisted predominantly of sand fraction (80–91%). Here, we
discuss the key findings on general soil properties, water repellency, hy-
draulic properties and evaluation of prominent PTFs for the prediction
of SMRC and Ks, and highlight the implications.
The time for WDPT and MED tests were positively correlated, indi-
cating that the two methods gave comparable results. Although previ-
ous research (e.g. Dekker and Ritsema, 1994) has reported that the
WDPT test tends to be less sensitive than the MED test, our results
showed that for soils similar to those studied here, either of the two
methods may be used for water repellency evaluation. This observation
Fig. 2. (a) Relationships between hydrocarbon contamination and time for WDPT ( ) and MED (○) water repellency tests, and (b) correlation between time for WDPT and MED for lab-
oratory contaminated Goromonzi soil. Similar results were obtained for the Ruwa soil (data not shown).
Fig. 3. Effect of concentration of hydrocarbon contamination on water repellency mea-
sured by the water droplet penetration test (WDPT) on Ruwa (○) and Goromonzi (●)
laboratory-contaminated soils. Data are mean of three replicates. Errors bars show 1 stan-
dard error.
284 A. Takawira et al. / Geoderma 235–236 (2014) 279–289

7. is consistent with a few earlier studies showing agreement between
WDPT and MED results (e.g. Badía et al., 2013). In this regard, consider-
ing that no special reagents are required for the WDPT, this method is
ideal for rapid water repellency evaluation under field conditions, par-
ticularly where laboratory facilities are lacking.
Subsequent water repellency evaluation based on the WDPT and
MED tests both revealed that the uncontaminated soils were wettable.
Following hydrocarbon contamination, all samples became water repel-
lent except the 30 mg g−1
treated sample, evidently confirming our
original hypothesis that hydrocarbon contamination induces water re-
pellency in inherently wettable tropical sandy soils. In earlier studies,
water repellency has been documented in natural sandy soils (Dekker
and Ritsema, 1994; Mainwaring et al., 2004; Wahl, 2008), while limited
evidence exists for hydrocarbon contaminated soils. In these studies,
water repellency has been attributed to fire, type and vegetation spe-
cies, and hence soil organic carbon (Wahl, 2008). Here, we present the
first evidence suggesting that uncontaminated sandy soils associated
with the miombo woodlands of southern Africa are inherently wettable.
However, hydrocarbon contamination through spills, leakages and dis-
posal of petroleum hydrocarbons has the capacity to induce water re-
pellency. In this case, we attribute the induced water repellency to the
presence of hydrophobic long-chain aliphatic and aromatic compounds
in petroleum hydrocarbons.
Contrary to our expectation, Ks increased linearly (p b 0.05) as the
concentration of hydrocarbons increased between 30 and 150 mg g−1
.
Although elucidating the exact mechanisms was beyond the scope of
the current study, we hypothesize that the presence of hydrophobic pe-
troleum hydrocarbons caused a drop in the dielectric constant of water.
The resulting reduction of electrostatic interactions between the dipolar
water molecules and the charged soil matrix triggers rapid and exten-
sive outflow of pore water. Indeed, a recent study by Calla et al.
(2011) show that the dielectric constant of saline water dropped
when the diesel increased from 0 to 120% by weight. Further evidence
supporting this hypothesis is provided by the two laboratory
studies on the effects of liquid hydrocarbons as soil permeants on
Ks (Fernandez and Quigley, 1985, 1988). Fernandez and Quigley
(1985) observed that decreasing the dielectric constant from 80 to 2 in-
creased Ks of clay soils by 5 orders of magnitude from 5 × 10−9
to 1 × 10
−4
cm s−1
. As a result, using alcohol and liquid aromatics as permeants
at 30% of the pore volume increased Ks by 10 and 1000 fold, respectively
(Fernandez and Quigley, 1985). However, the magnitude of increase ob-
served on the predominantly sandy soils used in the current study
Fig. 4. Soil moisture retention curves for laboratory contaminated soils (a) and field contaminated soils (●) and control (○) at Ruwa (b) and Goromonzi (c) sites. RETC software
(version 6.02) was used to fit the single-porosity van Genuchten (VG) pedotransfer function to measured data (van Genuchten et al., 1991).
285A. Takawira et al. / Geoderma 235–236 (2014) 279–289

8. appears lower than values reported for clay (Fernandez and Quigley,
1985), probably reflecting the textural differences.
Laboratory simulated contamination above 30 mg g−1
significantly
reduced soil moisture retention at low suctions, suggesting enhanced
soil water loss. However, the impacts of hydrocarbon contamination
on SMRC of field contaminated soils were less pronounced and showed
no clear trend, further suggesting that under tropical conditions, the im-
pacts could be transient. Earlier research investigating impacts of hydro-
carbon contamination on SMRC have yielded inconsistent results (e.g.
Ellis and Adams, 1961; Roy and Mcgill, 1998). For example, Roy and
McGill (1998) observed that wettable soils had higher moisture reten-
tion than water repellent ones, while Ellis and Adams (1961) observed
the opposite. Although the mechanisms accounting for this inconsisten-
cy observed in earlier studies (e.g. Roy and McGill, 1998) are unclear,
this could reflect differences in soils, levels of contamination and age ef-
fects. For example, while recent hydrocarbon contamination may re-
duce water retention as observed under laboratory conditions, aging
of hydrocarbons may increase soil organic matter content and hence
moisture retention. Moreover, the effect of aging on hydrocarbon con-
centrations could also account for the inconsistent effects of field con-
tamination on SOC and SMRC observed between the two sites. In the
current study, the observed reduction in moisture retention for
Table 3
van Genuchten parameters α, air-entry value 1
α
À Á
and n for laboratory and field hydrocar-
bon contaminated soils. Data are means of three replicates ± standard errors. Means
followed by different letters for laboratory contaminated soils and each field site are signif-
icantly different at probability level p = 0.05.
Treatment α (kPa−1
) 1
α kPað Þ n
Laboratory-contaminated soils:
Hydrocarbon concentration (mg g−1
soil)
0 0.056 ± 0.007a
18.4 ± 2.0a
1.216 ± 0.135a
30 0.049 ± 0.000a
20.3 ± 0.1a
1.723 ± 0.115b
60 0.287 ± 0.028b
3.6 ± 0.4b
1.147 ± 0.071ac
120 0.382 ± 0.015b
2.6 ± 0.1b
1.072 ± 0.064ac
150 0.744 ± 0.116c
1.4 ± 0.2b
1.031 ± 0.026ac
p value b0.001 b0.001 0.002
Field contaminated soils:
Goromonzi:
Control 0.304 ± 0.012a
3.3 ± 0.1a
1.356 ± 0.086a
Contaminated soil 0.047 ± 0.021b
22.0 ± 0.3b
1.568 ± 0.092a
p value b0.001 0.02 0.374
Ruwa:
Control 0.505 ± 0.042a
2.0 ± 0.2a
1.612 ± 0.101a
Contaminated soil 0.301 ± 0.023b
3.3 ± 0.1b
1.631 ± 0.097a
p value 0.015 0.03 0.257
Fig. 5. Effects of hydrocarbon contamination levels on constant-head saturated
hydraulic conductivity (Ks) of laboratory contaminated sandy soils from Goromonzi
(a) and Ruwa (b) sites. Data are means ± standard error (SE) for three replicates.
Means with different letters are significantly different using least significant difference
(lsd) at p = 0.05.
Fig. 6. Box-and-whisker plot comparison of saturated hydraulic conductivity (Ks) values
measured by constant-head laboratory method and predicted by three pedotransfer func-
tion models (Puckett, Dane and Puckett and Jabro) for Goromonzi (a) and Ruwa (b) soils.
Data shown are median, minimum, maximum, and 25th and 75th percentile values.
286 A. Takawira et al. / Geoderma 235–236 (2014) 279–289

9. laboratory contaminated soils was consistent with increased soil hy-
draulic conductivity attributed to the reduced dielectric constant of
water in the presence of hydrocarbons. Water repellency tests on field
samples showed that soils from both control and contaminated sites
were wettable. Unlike other sandy soils in Australia (King, 1981) and
Netherlands (Dekker and Ritsema, 1994; Ritsema et al., 1993), results
showed that sandy soils derived from granites associated with miombo
natural woodlands are inherently wettable. Contrasting water repellen-
cy results on similar textures probably reflected the effects of vegetation
type and species on the amount and forms of soil organic carbon. How-
ever, t-test comparisons indicate that the WDPT for control soils were
higher than that of the contaminated soils. This observation implies
that the organic compounds causing water repellency were relatively
higher in natural woodlands than contaminated soils. Despite the
large pulse spills, our findings on field contaminated soils indicate no
water repellency. There is lack of information on the persistence of
hydrocarbon-induced water repellency in tropical soils. However, the
findings of few studies conducted in temperate and arctic regions sug-
gest that hydrocarbon-induced water repellency persists long after con-
tamination (Roy and McGill, 1998). Several reasons could account for
the putative lack of persistence in the current study. First, given that
the Ruwa site was contaminated in 2007, there exists a possibility that
breakdown of hydrocarbons could have occurred during the 5-year
post contamination period. Moreover, contrary to the low temperatures
that could retard microbial activity in the temperate and arctic regions
(e.g. Roy et al., 1999), relatively high temperatures associated with trop-
ical conditions in Zimbabwe could account for the lack of persistence.
Second, the application of nutrient solution to promote revegetation
could also have stimulated native microbes responsible for the decom-
position of hydrocarbons. The stimulation of microbial activity causing
hydrocarbon degradation through nutrient application has been report-
ed in previous studies (Garcia-Blanco, 2004; Garcia-Blanco et al., 2007).
However, determining the key mechanism accounting for the observed
lack of persistence will require conducting further experimentation
comparing similar sites with and without nutrient addition.
Despite the lack of water repellency, residual signature of hydrocar-
bon contamination was evident at both sites. For instance, EC was con-
sistently higher in contaminated soils than the control soils, signifying a
residual salt load. The high EC could be associated with fuel additives
and/or soluble products of hydrocarbon decomposition. At Goromonzi,
the recently (b1 year) contaminated soil had higher SOC than the con-
trol. The enhanced soil structural stability arising from the relatively
higher SOC could account for the significantly lower density and more
porous soils observed on the contaminated soil than the control. At
Ruwa, the 5-year old contaminated site had significantly higher pH,
bulk density and lower porosity than the control. An increase in soil
pH on soils contaminated with flowline additives for oil pipelines has
also been reported under tropical conditions in the Barrow Island in
Australia (George et al., 2011). The high bulk density, and low total po-
rosity were attributed possibly to compaction associated with its close
proximity to the road compared to the control. The control and contam-
inated sites had comparable SOC, suggesting that over the 5-year period,
carbon derived from hydrocarbon could have undergone rapid decom-
position to background levels.
Evaluation of PTFs showed that the single-porosity van Genuchten
function developed initially for uncontaminated natural soils provided
the best fit for soil moisture retention. The good fit for both control,
and laboratory and field contaminated soils supports earlier observa-
tions indicating that the impact of hydrocarbon contamination on
SMRC was minimal. However, laboratory simulated hydrocarbon con-
tamination increased α and reduced the air-entry value, an observation
consistent with the reduced moisture retention in Fig. 4(a). The oppo-
site effect was observed for field contaminated soil suggesting improved
moisture retention on contaminated soils evident in soil moisture re-
tention curves in Fig. 4(b) and (c). Although the mechanisms are un-
clear, this could reflect changes in the nature and amount of soil
organic carbon especially for the Goromonzi site. However, in general,
the values of n and α were comparable to those reported for natural
sandy soils (2.0 to 3.2 kPa) (e.g. Wang et al., 2009; Zhu and Mohanty,
2002). Exceptions were laboratory (0 and 30 mg g−1
) and field contam-
inated soils from Goromonzi which had comparatively higher air entry
values than the other treatments, probably due to fitting errors associat-
ed with the RETC software. Overall, the van Genuchten parameters
appeared more sensitive to hydrocarbon contamination than the
laboratory-measured soil moisture retention data. Therefore, we infer
that water and solute transport models based on the van Genuchten
function may be applied to the studied soils with minimum bias in
model outputs.
Evaluation of PTFs for Ks indicate that the Puckett, and Dane and
Puckett models provide Ks estimates closer to the measured values
than the Jabro model, which tended to over-estimate. The opposite
trend was observed for the 60–150 mg g−1
hydrocarbon contamina-
tion, where the Jabro model provided a better estimate than the other
two. The discrepancy between measured and observed Ks for the
60–150 mg g−1
was associated with the observed unexpected increase
in Ks with hydrocarbon contamination. These PTFs, originally developed
for natural uncontaminated soils, do not account for the phenomenon
associated with hydrocarbon contamination. It is also noteworthy that
the PTFs for Ks used in the current study all involve clay as an input. Con-
sidering that the soils studied here contained 80–88% sand, it is also
likely that the variation of predicted Ks among PTFs possibly reflects dif-
ferences in input parameters. Overall, both measured and predicted Ks
values were within the range reported in literature for such textural
classes (Sobieraj et al., 2001; Tietje and Hennings, 1996). Although the
capacity of PTFs to predict measured hydraulic properties of uncontam-
inated soils has been reported in earlier studies (Dikinya, 2005; Kool
and Parker, 1988), our results further confirm the validity of such
models on hydrocarbon contaminated soils investigated in the current
study.
The induced water repellency and increase in saturated hydraulic
conductivity have implications on hydrological and ecological functions
of contaminated soils. Water repellency alters the water balance
through reduced infiltration, enhanced surface runoff and erosion, and
reduced soil moisture. In a seasonally dry environment, reduced soil
moisture may have adverse impacts on soil-plant-water relations, caus-
ing shifts in plant species through invasion by alien species and vegeta-
tion die-off. Indeed, field observations showed that contaminated sites
had sparse, yellowish and stunted vegetation consisting predominantly
of sedges and grass species, indicative of reduced vegetation growth and
vigor and possibly mortality. Moreover, enhanced surface runoff may
transport hydrocarbons and other associated contaminants into surface
water bodies, posing significant public and environmental risks. The ob-
served increase in Ks on contaminated soils suggests that once saturat-
ed, rapid water and contaminant transport including preferential flow
may occur in the sub-surface into groundwater systems. The generally
high residence times of groundwater in aquifers, coupled with anaero-
bic conditions and low microbial activity in the saturated zone, imply
that hydrocarbons and other contaminants tend to persist in groundwa-
ter systems. This will exert a strong influence on groundwater remedi-
ation and clean-up, and pose significant public and environmental
risks particularly in southern Africa, where approximately 90% of rural
households depend on groundwater sources for domestic supply.
5. Summary, conclusions and outlook
Hydrocarbon contamination of soils occurs frequently yet limited re-
search has investigated water repellency and changes in soil hydraulic
properties induced by such contamination. In addition, comparison of
water repellency methods and evaluation of existing PTFs for estimating
hydraulic properties on such soils have been lacking. The current study
measured water repellency and hydraulic properties, and evaluated
PTFs on laboratory-and field-contaminated tropical sandy soils. The
287A. Takawira et al. / Geoderma 235–236 (2014) 279–289

10. WDPT and MED methods for water repellency evaluation gave compa-
rable results. Laboratory simulated hydrocarbon contamination of soils
induced water repellency (WR) and a significant increase in saturated
hydraulic conductivity (Ks) in an inherently wettable sandy soil. The in-
crease in Ks was attributed to the reduction of the dielectric constant of
water caused by the presence of hydrophobic hydrocarbons. Contrary to
studies conducted in temperate and arctic environments where WR
persists for over 20 to 50 years after hydrocarbon contamination, the
current study showed that hydrocarbon contaminated field soils were
wettable, as evidenced by WDPT far below the minimum threshold for
water repellent soils. This indicates that under tropical conditions,
such induced water repellency could be transient or non-persistent.
The induced water repellency could potentially offset the hydrological
balance and adversely affect soil–plant–water relations and soil ecology.
On recently contaminated sites, the associated increase in Ks may trigger
rapid movement of water and contaminants into the groundwater sys-
tems, and influence remediation or clean-up process. Non-persistence
of water repellency in the tropics could be attributed to high biodegra-
dation rates stimulated by warm conditions. At the Goromonzi site, nat-
ural breakdown could have been further stimulated by nutrients added
to promote revegetation of the contaminated sites. The fact that both
relatively old (5 years) and recent (b1 year) sites had similar water re-
pellency further suggests that this breakdown could be rapid under
tropical conditions. However, the residual signature of hydrocarbon
contamination on field samples was evident through stunted vegetation
and changes in electrical conductivity, pH, soil organic carbon, bulk den-
sity and total porosity. Despite these changes in soil properties, the re-
sults suggest that the existing PTFs provide reliable estimations of
SMRC and Ks as evidenced by values within the same order of magni-
tude. By inference, water and solute balance models relying on these
PTFs to estimate storages and fluxes may yield outputs with minimum
bias. Coupled with additional field data, these data are sufficiently accu-
rate for large-scale simulation of water and solute transport on contam-
inated soils using existing models based on the PTFS evaluated here.
In general, the findings support the hypothesis that hydrocarbon
contamination induces water repellency and reduces moisture reten-
tion on inherently wettable tropical sandy soils. However, the positive
linear relationship between hydrocarbon contamination and Ks
contradicted the original hypothesis, an observation attributed to a de-
crease in the dielectric constant of water. Future research should focus
on elucidating the key mechanisms responsible for the increased Ks in
hydrocarbon contamination. Further field investigations based on a
contamination chronosequence or gradient are also required to confirm
our laboratory findings. Such studies will also provide comprehensive
information on the temporal evolution of water repellency and hydrau-
lic properties, and the fate and degradation mechanisms of hydrocar-
bons under tropical conditions.
Acknowledgments
We are grateful to staff from the Hazardous Substance Inspectorate
of the Zimbabwe Environmental Management Agency (EMA) for de-
tailed site information, and their time and effort during site identifica-
tion and sampling. We also thank staff from the Department of Soil
and Agricultural Engineering at the University of Zimbabwe for assis-
tance during laboratory analyses. Laboratory reagents were partly pro-
vided by WG's biochar research project funded by the Swedish
International Foundation for Science (IFS) Grant Number C/5266-1.
The authors are solely responsible for the experimental design, data col-
lection and interpretation, manuscript compilation and conclusions
drawn thereof. We are also grateful to Professor Jirka Simunek of
the Department of Environmental Sciences, University of California Riv-
erside for help with interpretation of RETC results. The manuscript also
benefited from valuable comments provided by two anonymous re-
viewers, to whom we are sincerely thankful.
References
Adams, R.H.,Cruz, J.Z., 2008. Water repellency in oil contaminated sandy and clayey soils.
Int. J. Environ. Sci. Technol. 5 (4), 445–454.
Adams, R.H.,Osorio, F.J.G.,Cruz, J.Z., 2008. Water repellency in oil contaminated sandy and
clayey soils. Int. J. Environ. Sci. Technol. 5 (4), 445–454.
Aislabie, J.M.,Balks, M.,Foght, J.M.,Waterhouse, J., 2004. Hydrocarbon spills on antarctic
soils: effects and management. Environ. Sci. Technol. 38 (5), 1265–1274.
Arcenegui, V.,Mataix-Solera, J.,Guerrero, R.,Zornoza, R.,Mayoral, A.M.,Morales, J., 2007.
Factors controlling the water repellency induced by fire in calcareous Mediterranean
forest soils. Eur. J. Soil Sci. 58, 1254–1259.
Awada, A.S.,Anai, K.K.,Ukushima, M.F., 2004. Preparation of artificially spiked soil with
polycyclic aromatic hydrocarbons for soil pollution analysis. Anal. Sci. 20, 239–241.
Badía, D., Aguirre, J.A., Martí, C., Márquez, M.A., 2013. Sieving effect on the intensity
and persistence of water repellency at different soil depths and soil types from NE-
Spain. Catena 108, 44–49.
Balks, M.R.,Paetzold, R.F.,Kimble, J.M.,Aislabie, J.,Campbell, I.B., 2002. Effects of hydrocar-
bon spills on the temperature and moisture regimes of Cryosols in the Ross Sea
region. Antarct. Sci. 14 (4), 319–326.
Blake, G.R.,Hartge, K.H., 1986. Bulk density, In: Klute, A. (Ed.), Methods of Soil Analysis
Part 1 ASA Monograph No. 9, 2nd ed. American Society of Agronomy, Madison, Wis-
consin, pp. 363–376.
Bohnhoff, G.L., Ogorzalek, A.S., Benson, C.H., Shackelford, C.D., Apiwantragoon, P., 2009.
Field data and water-balance predictions for a monolithic cover in a semiarid climate.
J. Geotech. Geoenviron. 135 (3), 333–348.
Buczko, U., Bens, O., Huttl, R.E., 2006. Tillage effects on hydraulic properties and
macroporosity in silty and sandy soils. Soil Sci. Soc. Am. J. 70, 1998–2007.
Burckhard, S.R.,Pirkl, D.,Schaefer, V.R.,Kulakow, P.,Levenet, B., 2004. A study of soil water-
holding properties as affected by TPH. Proceedings of the 2000 Conference of
Hazardous Waste.
Calla, O.P.N.,Hasan, S.,Almadian, N., 2011. Variability of dielectric constant of saline water
in combination with diesel in Cj band (5.3 GHz) and Ku band (13.4 GHz). Indian J.
Radio Space Phys. 40, 153–158.
Caravaca, F.,Rolda, A., 2003. Assessing changes in physical and biological properties in a
soil contaminated by oil sludges under semiarid Mediterranean conditions.
Geoderma 117, 53–61.
Contreas, S.,Cantón, Y.,Solé-Benet, A., 2008. Sieving crusts and macrofaunal activity con-
trol soil water repellency in semiarid environments: evidences from SE Spain.
Geoderma 145, 252–258.
Cosby, B.J.,Hornberger, G.M.,Clapp, R.B.,Ginn, T.R., 1984. A statistical exploration of soil
moisture characteristics to the physical properties of soils. Water Resour. Res. 20,
682–690.
Dane, J.H.,Puckett, W., 1994. Field soil hydraulic properties based on physical and miner-
alogical information. In: van Genuchten, M.Th., et al. (Eds.), Proceedings of the Inter-
national Workshop on Indirect Methods for Estimating the Hydraulic Properties of
Unsaturated Soils. University of California, Riverside, pp. 389–403.
Dekker, L.W.,Jungerius, P.D., 1990. Water repellency in the dunes with special reference
to the Netherlands. Catena Suppl. 18, 173–183.
Dekker, L.W.,Ritsema, C.J., 1994. How water moves in a water repellent sandy soil: I. Po-
tential and actual water repellency. Water Resour. Res. 30, 2507–2517.
Dikinya, O., 2005. Comparison of the instantaneous profile method and inverse modeling
for prediction of effective soil hydraulic properties. Soil Res. 43, 599–606.
Dlapa, P., Doerr, S.H., Lichner, L., Sir, M., Tesar, M., 2004. Effect of kaolinite and Ca-
montmorillonite on the alleviation of soil water repellency. Plant Soil Environ. 50
(2), 358–363.
Doerr, S.H., 1998. On standardizing the ‘water drop penetration time’ and the ‘molarity of
an ethanol droplet' techniques to classify soil hydrophobicity: a case study using me-
dium textured soils. Earth Surf. Process. Landf. 23, 663–668.
Doerr, S.H., Shakesby, R.A., Walsh, R.P.D., 1996. Soil hydrophobicity variations with
depth and particle size fraction in burned and unburned Eucalyptus globulus and
Pinus pinaster forest terrain in the Águeda Basin, Portugal. Catena 27, 25–47.
Doerr, S.H.,Shakesby, R.A.,Walsh, R.P.D., 2000. Soil water repellency: its causes, character-
istics and hydro-geomorphological significance. Earth-Sci. Rev. 51, 63–65.
Doerr, S.H.,Shakesby, R.A.,Dekker, L.W.,Ritsema, C.J., 2006. Occurrence, prediction and hy-
drological effects of water repellency amongst major soil and land-use types in a
humid temperate climate. Eur. J. Soil Sci. 57, 741–754.
Ellis Jr., R.,Adam Jr., R.S., 1961. Contamination of soils by petroleum hydrocarbons. Adv.
Agron. 13, 197–216.
FAO (Food and Agriculture Organization), 1988. UNESCO soil map of the world, revised
legend. World Resources Report, 60. FAO-UNESCO, Rome.
Fernandez, F.,Quigley, R.M., 1985. Hydraulic conductivity of natural clays permeated with
simple liquid hydrocarbons. Can. Geotech. J. 22 (2), 205–214.
Fernandez, F.,Quigley, R.M., 1988. Viscosity and dielectric constant control on the hydrau-
lic conductivity of clay soils permeated with water-soluble organics. Can. Geotech. J.
25, 582–589.
Foght, J.M.,Waterhouse, E.J., 2004. Critical review hydrocarbon spills on Antarctic soils: ef-
fects and management. Environ. Sci. Technol. 38 (5), 1265–1274.
Garcia-Blanco, S., 2004. Testing the Resource-ratio Theory as a Framework for
Supporting a Bioremediation Strategy for Clean-up of Crude Oil-contaminated Envi-
ronments (Ph.D. Dissertation) University of Cincinnati, Cincinnati, OH.
Garcia-Blanco, S.,Venosa, A.D.,Suidan, M.T.,Lee, K.,Cobanli, S.,Haines, J.R., 2007. Biostimu-
lation for the treatment of an oil-contaminated coastal salt marsh. Biodegradation 18
(1), 1–15.
Gee, G.W., Bauder, J.W., 1986. Particle size analysis, In: Klute, A. (Ed.), Methods of Soil
Analysis, Part 1. Physical and Mineralogical Methods, 2nd edn ASA SSSA, Madison,
WI, pp. 383–411.
288 A. Takawira et al. / Geoderma 235–236 (2014) 279–289

11. George, S.J.,Sherbone, J.,Hinz, C.,Tibbett, M., 2011. Terrestrial exposure of oilfield flowline
additives diminish soil structural stability and remediative microbial function. Envi-
ron. Pollut. 159, 2740–2749.
Gwenzi, W., 2010. Vegetation and Soil Controls on Water Redistribution on Recently Con-
structed Ecosystems in Water-limited Environments(PhD thesis) The University of
Western Australia, Perth.
Gwenzi, W.,Hinz, C.,Holmes, K.,Philips, I.R.,Mullins, I.J., 2011. Field-scale spatial variability
of saturated hydraulic conductivity on a recently constructed artificial ecosystem.
Geoderma 166 (1), 43–56.
Harper, R.J.,McKissock, I.,Gilkes, R.J.,Carter, D.J.,Blackwell, P.S., 2000. A multivariate frame-
work for interpreting the effects of soil properties, soil management and landuse on
water repellency. J. Hydrol. 231–232, 371–383.
Hillel, D., 1998. Environmental Soil Physics. Academic Press, San Diego, CA.
Holländer, H.M., Blume, T., Bormann, H., Buytaert, W., Chirico, G.B., Exbrayat, J.-F.,
Gustafsson, D.,Hölzel, H.,Kraft, P.,Stamm, C.,Stoll, S.,Blöschl, G.,Flühler, H., 2009. Com-
parative predictions of discharge from an artificial catchment (Chicken Creek) using
sparse data. Hydrol. Earth Syst. Sci. 13, 2069–2094.
Huffman, L.,Forest, A.N.,Collins, F., 2001. Strength of hydrophobicity and persistence front
of fire-induced and lodge pole soil under Colorado ponderosa. Hydrol. Process. 15,
2877–2892.
Hyun, S., Mi-Youn, A., Zimmerman, A.R., Minhee, K., Kim, G.-J., 2008. Implication of
hydraulic properties of bio-remediated diesel-contaminated soil. Chemosphere 71,
1646–1653.
Ippisch, O.,Vogel, H.J.,Bastian, P., 2006. Validity limits for the van Genuchten–Mualem
model and implications for parameter estimation and numerical simulation. Adv.
Water Resour. 29 (12), 1780–1789.
Jabro, J.D., 1992. Estimation of saturated hydraulic conductivity of soils from particle size
distribution and bulk density data. Trans. ASAE 35 (2), 557–560.
Johnston, S.G.,Hirst, P.,Slavich, P.G.,Bush, R.T.,Aaso, T., 2009. Saturated hydraulic conduc-
tivity of sulphuric horizons in coastal floodplain acid sulphate soils: variability and
implications. Geoderma 151, 387–394.
Jordán, A.,Zavala, L.M.,Mataix-Solera, J.,Doerr, S.H., 2013. Soil water repellency: origin, as-
sessment and geomorphological consequences. Catena 108, 1–5.
King, P.M., 1981. Comparison of methods for measuring severity of water repellence of
sandy soils and assessment of some factors that affect its measurements. Aust. J.
Soil Res. 19, 275–285.
Klute, A.,Dirksen, C., 1986. Hydraulic conductivity and diffusivity: laboratory methods, In:
Klute, A. (Ed.), Methods of Soil Analysis, Part 1. Physical and Mineralogical Methods,
2nd ed. ASA SSSA, Madison, WI, pp. 687–734.
Kool, J.B.,Parker, J.C., 1988. Analysis of the inverse problem for transient unsaturated flow.
Water Resour. Res. 24 (6), 817–830. http://dx.doi.org/10.1029/WR024i006p00817.
Kostka, S.J., 2000. Amelioration of water repellency in highly managed soils and the en-
hancement of turfgrass performance through the systematic application of surfac-
tants. J. Hydrol. 231 (232), 359–368.
Letey, J., 1969. Measurement of contact angle, water drop penetration time, and critical
surface tension. Proceedings of the Symposium on Water-repellent Soils, 6–10 May
1968. University of California, Riverside, pp. 43–47.
Letey, J.,Carrillo, M.L.K.,Pang, X.P., 2000. Approaches to characterize the degree of water
repellency. J. Hydrol. 231 (232), 61–65.
Mainwaring, K.A., Morley, C.P., Doerr, S.H., Douglas, P., Llewellyn, C.T., Llewellyn, G.,
Matthews, I.,Stein, B.K., 2004. Role of heavy polar organic compounds for water repel-
lency of sandy soils. Environ. Chem. Lett. 2, 35–39.
Mataix-Solera, J.,Doerr, S.H., 2004. Hydrophobicity and aggregate stability in calcareous
topsoil from fire affected pine forests in southeastern Spain. Geoderma 118, 77–88.
Millioli, V.S.,Servulo, E.L.-C.,Sobral, L.G.S.,De Carvalho, D.D., 2009. Bioremediation of crude
oil-bearing soil: evaluating the effect of rhamnolipid addition to soil toxicity and to
crude oil biodegradation efficiency. Global NEST J. 11 (2), 181–188.
Mualem, Y., 1976. A new model for predicting the hydraulic conductivity of unsaturated
porous media. Water Resour. Res. 12, 513–522.
Muñoz-Carpena, R.,Regalado, C.M.,Álvarez-Benedí, J.,Bartoli, F., 2002. Field evaluation of
the new Philip–Dunne permeameter for measuring saturated hydraulic conductivity.
Soil Sci. 167 (1), 9–24.
Nyamapfene, K., 1991. Soils of Zimbabwe. Nehanda Publishers, Harare.
Okalebo, J.R., Gathua, K.W., Wormer, P.L., 1993. Laboratory methods of soil and plant
analysis: a working manual. Soil Science Society of East Africa Technical Publication
No. 1Marvel EPZ (Kenya) LTD, Nairobi, Kenya.
Puckett, W.E.,Dane, J.H.,Hajek, B.H., 1985. Physical and mineralogical data to determine
soil hydraulic properties. Soil Sci. Soc. Am. J. 49, 831–836.
Quyum, A., 2000. Water Migration Through Hydrophobic Soils(M.Sc. thesis) Department
of Civil Engineering. The University of Calgary, Alberta, p. 157.
Rahman, A.Z.,Hamzah, U.,Mohd, P.,Taha, R.,Ithnain, N.,Ahmad, N.S., 2010. Influence of oil
contamination on geotechnical properties of basaltic residual soil. Am. J. Appl. Sci. 7
(7), 954–961.
Rayment, G.E., Higginson, F.R., 1992. Australian laboratory handbook of soil and water
chemical methods. Australian Soil and Land Survey HandbookInkata Press, Mel-
bourne, Sydney.
Reynolds, W.D.,Elrick, D.E., 2002. Constant head well permeameter (vadose zone). In:
Dane, J.H., Topp, G.C. (Eds.), Methods of Soil Analysis, Part 4. Physical Methods. Soil
Science Society of America, Inc., Madison, WI, pp. 844–858.
Reynolds, W.D.,Bowman, B.T.,Brunke, R.R.,Drury, C.F.,Tan, C.S., 2000. Comparison of ten-
sion infiltrometer, pressure infiltrometer, and soil core estimates of saturated hydrau-
lic conductivity. Soil Sci. Soc. Am. J. 64, 478–484.
Reynolds, W.D.,Elrick, D.E.,Young, E.G.,Amoozegar, A.,Booltink, H.W.G.,Bouma, J., 2002. Sat-
urated and field-saturated water flow parameters. In: Dane, J.H., Topp, G.C. (Eds.),
Methods of Soil Analysis, Part 4. Physical Methods. Soil Science Society of America,
Inc., Madison, WI, pp. 797–878.
Ritsema, C.,Dekker, L.W.,Hendrickx, J.M.H.,Hamminga, W., 1993. Preferential flow mech-
anism in a water repellent sandy soil. Water Resour. Res. 29, 2183–2193.
Roy, J.L.,Mcgill, W.B., 1998. Characterization of disaggregated nonwettable surface soils
found at old crude oil spill sites. Can. J. Sci. 78, 331–344.
Roy, J.L., McGill, W.B., Rawluk, M.D., 1999. Petroleum residues as water-repellent
substances in weathered nonwettable oil-contaminated soils. Can. J. Soil Sci. 79,
367–380.
Scott, D.F., 2000. Soil wettability in forested catchments in South Africa; as measured
by different methods and as affected by vegetation cover and soil characteristics. J.
Hydrol. 232, 87–104.
Sobieraj, J.A.,Elsenbeer, H.,Vertessy, R.A., 2001. Pedotransfer functions for estimating sat-
urated hydraulic conductivity: implications for modeling storm flow generation. J.
Hydrol. 251, 202–220.
Tietje, O.,Hennings, V., 1996. Accuracy of the saturated hydraulic conductivity prediction
by pedotransfer functions compared to the variability within FAO textural classes.
Geoderma 69, 71–84.
Timberlake, J.,Chidumayo, E., 2011. Miombo ecoregion vision report. Occasional Publica-
tions in Biodiversity No. 20Biodiversity Foundation for Africa/WWF SARPRO, p. 80.
USDA (United States Department of Agriculture), 1994. Keys to Soil Taxonomy, 6th ed.
USDA Soil Conservation Service, Washington DC.
van Genuchten, M.Th., 1980. A closed-form equation for predicting the hydraulic conduc-
tivity of unsaturated soils. Soil Sci. Soc. Am. J. 44, 892–898.
van Genuchten, M.Th.,Leij, F.J.,Yates, S.R., 1991. The RETC code for quantifying the hydrau-
lic functions of unsaturated soils. Version 6.02 EPA Report 600/2-91/065. U.S. Salinity
Laboratory, USDA, ARS, Riverside, California (Available on-line http://www.pc-prog-
ress.com/en/Default.aspx?RETC.).
Verboom, W.C., 1991. The Influence of Soil Surface Settling and Sealing on the Water Dy-
namics of Three Zimbabwean Topsoils(PhD thesis) Department of Soil Science and
Agricultural Engineering, Faculty of Agriculture, University of Zimbabwe, Harare.
Wahl, N.A., 2008. Variability of water repellency in sandy forest soils under broadleaves
and conifers in north-western Jutland/Denmark. Soil and Water Research 3 (1),
155–164.
Wallach, R., 2010. Effect of soil water repellency on moisture distribution from a subsur-
face point source. Water Resour. Res. 46, W08521. http://dx.doi.org/10.1029/
2009WR007774.
Wallach, R., Jortzick, C., 2008. Unstable finger-like flow in water-repellent soils during
wetting and redistribution — the case of a point water source. J. Hydrol. 351 (1–2),
26–41.
Wallach, R.,Ben-Arie, O.,Graber, E.R., 2005. Soil water repellency induced by long-term ir-
rigation with treated sewage effluent. J. Environ. Qual. 34 (5), 1910–1920.
Wang, T.,Zlotnik, V.,Šimunek, J.,Schaap, M., 2009. Using pedotransfer functions in vadose
zone models for estimating groundwater recharge in semiarid regions. Water Resour.
Res. 45, W04412. http://dx.doi.org/10.1029/2008WR006903.
Watson, C.L.,Letey, J., 1970. Indices for characterizing soil water-repellency based upon
contact angle-surface tension relationships. Proc. Soil Sci. Soc. Am. 34, 841–844.
Zavala, L.M.,González, F.A.,Jordán, A., 2009. Intensity and persistence of water repellency
in relation to vegetation types and soil parameters in Mediterranean SW Spain.
Geoderma 152, 361–374.
Zhu, J.,Mohanty, B.P., 2002. Spatial averaging of van Genuchten hydraulic parameters for
steady-state flow in heterogeneous soils: a numerical study. Vadose Zone J. 1,
261–272.
289A. Takawira et al. / Geoderma 235–236 (2014) 279–289