You are on page 1of 12

Soil & Tillage Research 198 (2020) 104550

Contents lists available at ScienceDirect

Soil & Tillage Research


journal homepage: www.elsevier.com/locate/still

Consolidation effects on relationships among soil erosion properties and soil T


physical quality indicators
G.V. Wilsona,*, Tianyu Zhangb, R.R. Wellsc, Baoyuan Liud
a
USDA-ARS National Sedimentation Laboratory, 598 McElroy Dr., Oxford, MS, United States
b
Institute for Peat and Mire Research, Northeast Normal University, China
c
USDA-ARS National Sedimentation Laboratory, 598 McElroy Dr., Oxford, MS, United States
d
State Key Laboratory of Soil Erosion and Dryland Farming of on the Loess Plateau, Northwest A&F University, China

A R T I C LE I N FO A B S T R A C T

Keywords: Consolidation of soil by wetting and drying cycles following tillage rapidly changes the soil physical properties
Wetting/drying but little is known about how these changes impact soil erosion. The objective of this study was to develop
Shear stress equations to predict soil erodibility (Kd) and critical shear stress (τc) in response to consolidation based upon
Erodibility changes in soil bulk density, saturated hydraulic conductivity, water content, soil penetration strength and
surface shear strength following a series of wetting and drying cycles. Air-dried soil from four contrasting soil
series was loosely filled into 20 soil cylinders (80 total) to simulate freshly tilled conditions. Soil physical and
erosion properties were determined at six time periods (0, 1, 3, 5, 7, 10 days) following daily simulated rainfall
of 33 mm h−1 for 1 h and 24 h of drainage/drying. The Jet Test device was used to determine Kd, and τc. The bulk
density increased with time due to consolidation following the simulated precipitation events with the largest
increase (50–100 % of the total) occurring after the first wetting/drying cycle. The Onstad consolidation model
tended to over-predict this initial surface bulk density increase, whereas, an exponential model better re-
presented the surface and depth-averaged bulk density changes for the four soils tested. The shear strength and
soil penetration resistance increased dramatically after the first wetting/drying cycle then decreased to a fairly
stable value in response to subsequent wetting/drying cycles. The saturated hydraulic conductivity, Ks, de-
creased so rapidly with accumulated rainfall that an exponential decay model could not match the decrease in Ks
as the soil consolidated. The best indicators of erodibility response to consolidation were accumulated rain,
surface bulk density, saturated hydraulic conductivity and soil penetration resistance at the 1.3 cm depth. The
best indicators of critical shear stress were accumulated rainfall, soil penetration resistance at 1.3 cm and sa-
turated hydraulic conductivity. Because these erosion parameters had intercepts that were dependent upon the
soil series, prediction of changes in soil erosion with time following tillage will be soil-specific.

1. Introduction tillage. Wilson et al. (2019) identified indicators of soil physical quality
in a long-term no-till and showed that subsequent tillage so dramati-
Many studies have established the correlation between soil health cally changed the soil physical properties of the surface that they were
(soil quality) and individual soil properties (Andrews et al., 2004; no longer valid soil quality indicators. They recommended that the
Reynolds et al., 2009). Karlen et al. (2019) noted that awareness of and dynamic changes due to reconsolidation of the soil following tillage be
interest in soil health has increased exponentially in the last decade. included in future soil quality assessments.
They proposed that excessive tillage is a major factor in the connection Green et al. (2003) noted that tillage effects on soil hydraulic
between the decline in soil health and soil degradation, i.e. erosion. properties has been extensively studied and it is well established that
Dick (2018) stated that for soil health to be useful, it must be sensitive tillage initially decreases the bulk density and increases the porosity. As
to land management and temporally dynamic. Yet, one of the main noted by many others (Moret-Fernández et al., 2016; Moret and Arrúe,
complications of such analysis is the temporal changes in soil properties 2007; Veiga et al., 2008), these properties are far from static as the
due to climate effects, especially rainfall/runoff events following energy of raindrops, redistribution of particles by splash and rill


Corresponding author.
E-mail addresses: glenn.wilson@ars.usda.gov (G.V. Wilson), zhangty100@nenu.edu.cn (T. Zhang), robert.wells@ars.usda.gov (R.R. Wells),
baoyuan@bnu.edu.cn (B. Liu).

https://doi.org/10.1016/j.still.2019.104550

Available online 13 January 2020


0167-1987/ © 2019 Elsevier B.V. All rights reserved.
G.V. Wilson, et al. Soil & Tillage Research 198 (2020) 104550

erosion, and physiochemical processes of particle to particle bonding consolidation (Shainberg et al., 1996; Schiettecatte et al., 2005; Knapen
(aggregation) change the surface condition with time. Ahuja et al. et al., 2007b). This is generally considered a result of the soil erodibility
(1998) explained the reconsolidation processes involved in wetting as (Kd) being highest and the critical shear stress being lowest (essentially
(i) raindrop impact, (ii) effective stress approaches zero as soil wets up, zero) immediately after tillage (Knapen et al., 2007b). West et al.
and (iii) water movement condenses the matrix. As such, temporal (1992) was possibly the first to show that consolidation reduced the
changes due to natural reconsolidation by wetting and drying cycles erodibility (≈80 %) compared to freshly tilled soil but, in contrast to
can overwhelm the effects of different management treatments (Viega common thought, they found that the critical shear stress was also re-
et al., 2008; Green et al., 2003; Logsdon et al. 1993). Zhang et al. duced (≈40 %) by consolidation. King et al. (1995) found that critical
(2017) explained that the matrix becomes compacted due to its own shear was lowest immediately after tillage and increased (≈66 %) due
weight as the effective stress approaches zero in response to soil wetting to consolidation following one month of light rainfall. However, they
after tillage. Green et al. (2003) proposed that the soil properties also found that the erodibility did the opposite of what was expected
“gradually revert(s) back to the original state due to reconsolidation by and increased by 39 % in the aged (consolidated) conventional till plots
natural forces” but the temporal dynamics of this change have not been and by 35 % in the aged no-till plots. Reichert et al. (2001) compared
well quantified. Osunbitan et al. (2005) observed a gradual increase in recent conventionally tilled soil to consolidated (2 month) con-
bulk density with time for four different management systems that they ventionally tilled soil and found the expected relationship of decreased
reasoned to be due to rainfall induced wetting and drying cycles. Zhang (≈70 %) erodibility and increased (≈19 %) critical shear stress due to
et al. (2017) also observed a gradual decrease in porosity with time consolidation. These inconsistent results could be the result of treat-
after tillage to a relatively stable value after four wetting/drying cycles. ment differences such as differences in soils tested, tillage methods, and
In contrasts to the observations of a gradual shift in soil properties, rainfall (simulated or natural). The few studies that have quantified the
Mapa et al. (1986) found that the first wetting and drying cycle was erodibility and critical shear stress changes with time as the soil con-
responsible for almost all of the change in hydraulic properties. Their solidates have, therefore, been inconsistent and none have corre-
experiments involved wetting by drip irrigation with the energy im- sponded these changes in erosion properties to other soil physical
parted by raindrop impacts excluded. In contrast, Cassel (1983) mea- properties that could be soil health indicators.
sured physical and hydraulic properties of tilled soil exposed to natural Knapen et al. (2007a) assembled a database from the literature that
rainfall. Cassel found that essentially all of the bulk density increase due included 470 Kd and 522 τc values. These included three field runoff
to consolidation (settling) resulted from the first two rainfall events plot measurements, laboratory flume measurements and impinging
(two wetting/drying cycles). Even wheel traffic locations exhibited submerged Jet measurements. They noted that the most common ap-
further bulk density increases as a result of the first two rainfall events proach was using flow of clear water to detach particles and collecting
then stabilized. An extensive literature review by Strudley et al. (2008) the entrained particles at the outlet whether in a field plot or laboratory
concluded that the pronounced effects of tillage on physical and hy- flume. While the impinging Jet test was developed for measurements on
draulic properties disappear rapidly after just the first wetting/drying streambanks and is still used predominately for such applications
cycle. (Mahalder et al., 2018), it is now considered an acceptable method to
Onstad et al. (1984) used simulated rainfall to determine changes in measure soil erodibility and critical shear for other erosion scenarios
surface bulk density with time (ρt) based upon subsidence of the surface including agricultural conditions (Liu et al., 2017). The advantages of
to determine the following asymptotic relationship between the initial the Jet device used in the laboratory is the ability to control soil and
bulk density after tillage (ρi) and the cumulative precipitation with time hydrologic conditions. Liu et al. (2017) used the Jet Test to quantify the
(Pt): effects of antecedent soil moisture and water-potential gradients on soil
erosion properties for agricultural soil conditions (modest bulk den-
ρt = ρi + A[Pt/(1+ Pt)] (1)
sities). They found that critical shear stress can be predicted from
where ‘A’ is a regression coefficient. Green et al. (2003) used this re- knowledge of the silt content, void ratio and water potential gradient,
lationship to estimate changes in other hydraulic properties with time and soil erodibility can be predicted from knowledge of critical shear
due to consolidation. Sandin et al. (2018) also used simulated rainfall stress, clay content and void ratio. However, this method has yet to be
but on nine different soils packed into cylinders (10 cm high and 6.8 cm used to determine the effects of changes in physical soil properties due
diameter) to determine changes in soil structure. They found that to consolidation following tillage on erosion properties.
wetting and drying cycles decreased the total porosity in all soils except The objectives were to quantify the effects of rainfall-induced soil
the two clay soils and the greatest decrease was in the silt loam soils. consolidation on soil physical and erosion properties, and develop a
Additionally, for six of the soils the pore size distribution shifted to regression equation to calculate Kd and τc using one or more of the
smaller pore sizes and decreased pore network connectivity for four of following soil physical properties: soil bulk density, Ks, water content,
those six soils. These changes in pore structure with consolidation di- soil penetration strength and surface shear stress.
rectly impact the hydrologic and soil erosion response of soils to rainfall
events. 2. Materials and methods
Soil erosion properties have received considerably less quantifica-
tion with regards to changes with time following tillage. Schiettecatte 2.1. Field sites
et al. (2005) used simulated rainfall on loess derived silt loam soil to
find that successive rainfall significantly affected bulk density and Bulk disturbed soil samples were collected from three sites in
erodibility, with a linear increase in bulk density and a decrease in Mississippi (USA) and one site in Kansas (USA) as shown in Fig. 1:
erodibility. The most common expression relating soil erosion rate (Er, Goodwin Creek Experimental Watershed (GCEW), Little Topashaw
cm s−1) to physical properties is the excess shear stress Canal (LTC) watershed, the North Mississippi Experiment Station in
Holly Springs (HSES), and Cheney Lake Watershed (CLW) in Kansas,
Er = Kd(τ – τc)b (2) respectively. The four contrasting soils from these respective sites were:
Where Kd (cm3 N−1s−1) is the erodibility coefficient, τ (Pa) is the shear Calloway (Fine-silty, mixed, active, thermic Aquic Fraglossudalfs), Ar-
stress, τc (Pa) is the critical shear stress, ‘b’ is an exponent for non- kabutla (Fine-silty, mixed, active, acid, thermic Fluventic En-
linearity but is typically taken to be one (Knapen et al., 2007a). It is doaquepts), Providence (Fine-silty, mixed, active, thermic Oxyaquic
common knowledge that soil erosion potential is greatest after tillage Fragiudalfs), and Ost-Clark series combination (Fine-loamy, mixed,
and that soil detachment decreases with time after tillage as a result of superactive, mesic Udic Argiustolls, and Fine-loamy, mixed, super-
active, mesic Udic Calciustolls), respectively. The Calloway series

2
G.V. Wilson, et al. Soil & Tillage Research 198 (2020) 104550

Fig. 1. Experimental sites: Goodwin Creek Experimental Watershed (GCEW), Little Topashaw Canal (LTC), the North Mississippi Experiment Station in Holly Springs
(HSES), and Cheney Lake Watershed (CLW) and respective soils of Calloway, Arkabutla, Providence, and Ost-Clark.

Table 1 Great Bend Sand Plains. The Ost slopes range from 0 to 8 %, whereas
Standard soil property values for the four soils in this study. the Clark slopes range from 0 to 15 %. The soil samples were collected
Property Providence† Calloway†† Ost-Clark Arkabutla
adjacent to ephemeral gullies. Standard soil properties for the four soil
series are reported in Table 1.
Sand, % 15 2 48 23 The GCEW site (Calloway series) is typically managed as pasture or
Silt, % 69 88 24 66 forest in contrast to the alluvial flood plains where row crop agriculture
Clay, % 16 10 27 12
pH 6.4 5.5 7.7 6.5
is predominately conventionally tilled cotton (Gossypium hirsutum)
OC, % 0.52 0.97 0.68 1.05 and soybean (Glycine max), (Wilson et al., 2015). For the LTC site
EC, dS m−1 0.17 na 0.25 0.15 (Arkabutla series), the land is typically managed as either no-till corn
AS, % 23 na 10 16 (Zea mays), or no-till cotton (Gossypium hirsutum) followed by con-

ventional tillage (disking) and re-establishment of broad rows with
soil properties derived from Wilson et al. (2008a) and Wilson (2011).
†† deep furrows for sweet potato (Ipomoea batatus) production (Wilson
soil properties derived from National Cooperative Soil Characterization
Database for pedon ID: S1959MS107001.
et al., 2008b). Conventional tillage at the HSES involves disking twice
followed by harrowing and planting on flat rows without furrows
consists of deep, somewhat poorly drained silt loam soils that formed in (McGregor et al., 1998). The field site at CLW (Ost-Clark soil) was a
thick loess deposits on nearly level to gently sloping (0–3%) uplands conventional tillage wheat (Triticum aestivum) field.
and stream terraces in the Southern Mississippi Valley Silty Uplands.
They have a seasonally high water table perched over a thick fragipan. 2.2. Sample preparation and test procedures
The Arkabutla (Ar) series consists of deep, somewhat poorly drained,
moderately permeable soils that formed in silty alluvium. These, level Soil collected from the surface (0−15 cm) of the four sites was air-
to nearly level (0–2%), soils are on flood plains in the Southern Mis- dried and sieved to < 2 mm. Air-dried soil was filled into PVC cylinders
sissippi Valley Silty Uplands. The Providence series consists of moder- (15.24 cm inner diameter and 20 cm height, height may be adjusted) for
ately well drained soils with a fragipan. Permeability is moderately testing. Each PVC cylinder was designed with the top 5 cm composed of
slow. These silt loam soils formed in thin loess deposits above sandy a stack of 10 individual PVC rings (0.5 cm in height) secured together
and loamy sediments. They are nearly level to moderately steep by packing tape. This design allowed PVC rings to be removed as the
(0–15%) soils in uplands and terraces of the Southern Coastal Plain and soil consolidated to avoid a gap between the soil surface and the cy-
Southern Mississippi Valley Silty Uplands Major Land Resource Areas. linder’s top. The bottom of soil cylinders consisted of a porous mem-
Both the Ost and Clark series are very deep well drained loam soils that brane secured to the PVC to allow for free drainage. Each cylinder was
formed in alluvium and located on paleo-terraces in river valleys of the loosely filled with air-dried sieved soil to simulate conditions im-
mediately after tillage. A total of 20 soil cylinders were prepared for

3
G.V. Wilson, et al. Soil & Tillage Research 198 (2020) 104550

Fig. 2. The best-fit power functions for surface (0−1 cm) bulk density (A) and for depth-averaged bulk density (B) as a function of accumulated rainfall for the
Providence, Calloway, Ost-Clark and Arkabutla. Note that data points correspond to the six time periods tested (0, 1, 3, 5, 7 and 10 d) of daily wetting and drying
cycles.

each soil type (80 total) in which four replicated samples were prepared over the 10 wetting cycles and was applied for 1 h on daily basis. Fol-
for each of the five consolidation time periods. Within each set, two of lowing calibration to the desired rainfall application rate, the spatial
the four replicated samples were for soil property measurements (soil variability in rainfall was determined using a 10 × 10 grid of plastic
bulk density, saturated hydraulic conductivity, water content, soil pe- cups, placed at the same height as the soil columns. Samples were al-
netration strength and surface shear strength) and two were used for Jet lowed to freely drain for 24 h before testing and measurements were
tests to determine soil erosion properties. also made prior to rainfall as the starting condition following “tillage”.
The six time periods in which soil properties were measured received 0,
28, 87, 153, 220 and 324 mm accumulated rainfall, and corresponded
2.3. Soil property measurements
to the initial condition and subsequent wetting/drying cycles 1, 3, 5, 7,
10, respectively.
The effects of consolidation on soil physical properties by rainfall
After the daily rainfall was allowed to freely drain (wetting/drying
and the effects on soil erosion properties was determined at six time
cycle), the depth of subsidence due to consolidation was measured on
periods (0, 1, 3, 5, 7, 10 days) of daily simulated rainfall. A variable-
the surface along with ancillary soil properties. Bulk density was de-
swing nozzle type rainfall simulator was used as the rain source. The
termined at the surface (0−1 cm) with a 1 cm deep soil core (surface)
distance between the nozzle and soil surface was 3 m. The rainfall si-
and by measuring the oven-dry weight of the whole cylinder after
mulator (Meyer and Harmon, 1979), consists of single nozzle set at
correcting the volume for the depth of subsidence (depth averaged).
6.0 ± 0.2 MPa pressure. The rainfall intensity averaged 32.4 mm h−1

4
G.V. Wilson, et al. Soil & Tillage Research 198 (2020) 104550

Soil shear strength of the immediate surface was measured with a power functions for the Providence, Calloway, Ost-Clark and Arkabutla
pocket vane shear tester (GEOTEST, E-285) on the cylinder surface. Soil soils respectively: ρb = 1.324 Pt 0.010, r2 = 0.840; ρb = 1.280 Pt 0.007,
shear strength of the top 5 cm was measured with a shear vane tester r2 = 0.851; ρb = 1.271 Pt 0.005, r2 = 0.534; ρb = 1.228 Pt 0.009,
(GEONOR, H-60). Soil penetrometer readings were made with a cone r2 = 0.588 and the respective functions for the depth-averaged bulk
penetrometer (Field Scout, SC900) in 2.5 cm depth increments from the density were ρb = 1.238 Pt 0.007, r2 = 0.954; ρb = 1.150 Pt 0.011,
surface to 12.5 cm. Water content was measured by weight prior to r2 = 0.916; ρb = 1.142 Pt 0.006, r2 = 0.937; ρb = 1.045 Pt 0.007,
saturated hydraulic conductivity measurement (Ks). The Ks was de- r2 = 0.860.
termined by saturating soil cylinders from the bottom up with tap water Onstad et al. (1984) proposed that Eq 1 could represent the
then measuring steady outflow under a constant head. The two soil asymptotic increase in bulk density with water applications. Fitting Eq.
cylinders dedicated at each wetting/drying cycle for these soil prop- 1 to the surface bulk density data produced coefficients (A) ranging
erties were subsequently oven-dried to determine the dry weight of the from 0.065 to 0.121 (Table 2). Based upon these coefficients, the On-
whole cylinder for water content and bulk density calculations. stad model for all four soils predicted a 97 % increase in bulk density
The submerged Jet Erosion Test (Hanson and Cook, 2004) was used occurring following the first rainfall. This was somewhat representative
to measure the soil erodibility, Kd, and critical shear, τc on two separate of the Calloway and Ost-Clark soil response to the first event (Fig. 3) but
soil cylinders for each wetting/drying cycle dedicated to soil erosion even for these cases it over-predicted how sharply the soil bulk density
properties. The Jet test was identical to the device used by Liu et al. increased. The Onstad model greatly over-predicted the response for
(2017) with procedures similar to Hanson and Cook (2004). The pres- the Providence and Arkabutla series suggesting that the Onstad model
sure head at the jet orifice was 324 mm water and the distance from the gave too much emphasis on the initial wetting/drying cycle. The Ost-
orifice to the initial soil surface was 58 mm. The depth of scour was Clark and Arkabutla series exhibited a 100 % increase by the end of the
measured directly below the point-gauge every 5 s. Scour depth mea- fifth wetting/drying cycle (87 mm accumulated rain) then exhibited
surements stopped when the scour depth stabilized, i.e. maximum was dramatic decreases in bulk density. In contrast, the Providence and
reached. Because the Jet test is destructive, a new soil sample that had Calloway series exhibited more gradual increases to a maximum surface
been through the previous wetting and drying cycles was used each bulk density. A better model for such responses to cumulative pre-
time. cipitation, Pt, with time is the exponential model
The SAS Proc Reg procedure with SAS statement options Stop = 5,
ρt = ρi + (ρm-ρi) [1-exp (-Pt/ Pm)] (3)
Best = 20 used to determine the combinations of soil physical proper-
ties that best describe soil erodibility and critical shear stress. The Where Pm is the cumulative precipitation at which bulk density reaches
statement options listed above restrict SAS from including more than its final or maximum value, ρm. The exponential model provided more
five independent variables in multiple regression analysis and instructs gradual increases in surface bulk density that better represented the
SAS to report the 20 best equations (coefficients for the various para- observed responses for the four soils (Fig. 3).
meter combinations) for each step. Once the best soil quality indicators The depth averaged bulk density of the soil core based upon the
were selected for predicting either erodibility or critical shear stress, depth of subsidence and total core dry weight gave much different re-
SAS Proc GLM was performed to determine the level of significance of sults (Fig. 2B). Not only were the bulk densities lower, as would be
each variable and their interaction terms. The non-significant interac- expected from consolidation being integrated over the depth of the soil
tion terms were eliminated one at a time (least significant eliminated) core as opposed to at the surface where the consolidation is more im-
until only significant interaction terms remained, then non-significant pacted, but the patterns were more consistent with those predicted by
main variables were eliminated one at a time unless they occurred in an Onstad et al. (1984). Eq. 1 was fit to the depth-averaged bulk density
interaction term. data and the coefficients (A) ranged from 0.066 to 0.110 (Table 2). The
improved fit is reflected in the depth-averaged bulk density exhibiting
the dominate increase after the first wetting/drying cycle then reaching
3. Results and discussion
a stable value after five cycles (Fig. 2B) as opposed to decreasing after
the seventh cycle as observed for the surface bulk density (Fig. 2A). As a
3.1. Bulk density
result of reaching a stable depth-averaged bulk density, the power
function exhibited excellent fits with coefficients of determination >
For all four soils, the bulk density of the surface (0−1 cm) increased
0.86 for all four soils. While the Onstad model predicts 97 % of the final
in response to rainfall (Fig. 2A). The majority of the increase was fol-
increase will occur after the first wetting/drying cycle, the depth-
lowing the first rainfall wetting and drying cycle with increases of 6.0,
averaged bulk density only exhibited > 47 % of its full increase after
7.0, 5.1, and 3.5 % for the Providence, Calloway, Ost-Clark, and Ar-
the first cycle, > 70 % after three cycles, and by the fifth cycle (193 mm
kabutla, respectively. This is equivalent to 50, 89, 102, and 48 %, re-
rainfall) it was essentially 100 % consolidated for all four soils. Thus,
spectively, of the overall increase after 10 rainfall wetting/drying cycles
while the fit of the Onstad model to the depth average bulk density was
(Table 2). Note in Fig. 2A that three of the four soils exhibited a de-
superior to the surface bulk density, the proposed exponential model,
crease in bulk density after the seventh wetting and drying cycle
Eq. 3, still provided a better representation of the changes with wet-
(220 mm accumulated rainfall). The increase in surface (0−1 cm) bulk
ting/drying cycles.
density (ρb) with accumulated rainfall (Pt) was best fit by the following

Table 2
Surface bulk density and depth averaged (core) bulk density changes with time due to consolidation during the ten precipitation events (days 1–10) and their Eq. 1 A
parameters.
Surface Bulk Density Core Bulk Density

st
Initial Bulk Density Total increase 1 Cycle increase Eq 1 A Initial Bulk Density Total increase 1st Cycle increase Eq 1 A

−3 −3
Soil Series g cm % % g cm % %
Providence 1.27 12.0 50.1 0.121 1.20 7.8 55.3 0.082
Calloway 1.24 7.8 89.3 0.087 1.10 12.2 46.9 0.110
Ost-Clark 1.24 4.9 102.5 0.065 1.11 5.8 104.8 0.066
Arkabutla 1.18 7.4 47.8 0.101 1.01 9.1 67.3 0.074

5
G.V. Wilson, et al. Soil & Tillage Research 198 (2020) 104550

Fig. 3. The Onstad et al. (1984) model (Eq. 1) and exponential model (Eq. 3) for predicting the surface bulk density changes with accumulated rainfall for the (A)
Providence, (B) Calloway, (C) Ost-Clark, and (D) Arkabutla soil series. Note that data points correspond to the six time periods tested (0, 1, 3, 5, 7 and 10 d) of daily
wetting and drying cycles.

These findings for surface and depth averaged values of bulk density Where θf and Pf are the final measured water content at the final cu-
were consistent with Kool et al. (2019) who found that the bulk density mulative precipitation. This asymptotic model (Fig. 4B) over-predicted
was lowest immediately after tillage. They proposed that by disrupting the increase in water content at the first wetting/drying cycle. How-
the soil structure, tillage results in an unstable soil condition. ever, it provided better overall fits than standard statistical models,
They found that the largest change (i.e. increase) in bulk density such as the power function (Fig. 4A) due to the water content stabi-
occurred the first few weeks after tillage. This study corroborated the lizing after the third cycle at a field capacity value below the maximum
results of others (Cassel, 1983′ Onstad et al., 2984; Mapa et al., 2986; (saturated) value represented by θs.
Strudley et al., 2008) that found the majority of the change occurring
after the first wetting/drying cycle.
3.3. Shear strength

3.2. Water content The soil shear strength (SS) exhibited consistent patterns among
soils and between the surface (Fig. 5A) and 5 cm depth (Fig. 5B) but to
The soil prior to the first wetting/drying cycle was air-dry to si- different magnitudes of strength. Despite the sharp increase in water
mulate a well cultivated field. Thus, the water content increased rapidly content from the initial condition to the first wetting/drying cycle, the
over the first three rainfall events and remained stable following the SS increased dramatically. Further increases in water content led to
third wetting/drying cycle (Fig. 4). Given that water had ceased subsequent decreases in soil SS to a fairly stable value after the third
draining in all cases prior to the next rainfall simulation, this stable wetting/drying event. The main differences among soils were that the
value represents field capacity. Zhang et al. (2017) observed that initial increases were much less (> 300 %) for the Providence than for
wetting/drying cycles following tillage incrementally increased the the other three soils (> 1150 %). This pattern of increase then decrease
water content at all depths. They observed a shift towards higher water to a stable value held true for the surface and 5 cm depth with the
content, θ, at a specific water pressure which resulted in the soil water difference with depth being that the SS was about an order of magni-
holding capacity improving with wetting/drying cycles. The best fit tude lower for the 5 cm depth than the surface. The best fit equations
equations in Fig. 4A representing the 10 days of wetting/drying were for soil shear strength (SS) as a function of accumulated rainfall for the
the following power functions for the Providence, Calloway, Ost-Clark Providence, Calloway, Ost-Clark and Arkabutla, respectively for the
and Arkabutla, respectively: surface were: SS = 37.70 P t0.141, r2 = 0.949; SS = 20.67 P t0.244,
θ = 0.115 P t0.245, r2 = 0.989; θ = 0.110 P t0.257, r2 = 0.991; r2 = 0.812; SS = 27.30 P t0.264, r2 = 0.901; SS = 29.35 P t0.242,
θ = 0.110 P t0.250, r2 = 0.990; θ = 0.101 P t0.253, r2 = 0.987. However, r2 = 0.820 and the respective functions for the 5 cm depth were:
such functions predict that the water content continues to increase with SS = 1.70 P t0.081, r2 = 0.365; SS = 2.39 P t0.126, r2 = 0.314; SS = 1.94
accumulated rainfall which fails to account for a maximum water P t0.236, r2 = 0.684; SS = 2.68 P t0.166, r2 = 0.598.
content, i.e. saturation being attainable.
To better model the volumetric water content, θt, with time, the
following asymptotic relationship was used 3.4. Soil penetration resistance

θt = θsPt/(θs/2 + Pt) (4) The soil penetration resistance (SPR) followed the same patterns as
the soil shear strength with regards to changes with accumulated
Where θs is a maximum (saturated) water content calculated by
rainfall (Fig. 6A). The best fit functions for the Providence, Calloway,
θs = θf Pf/(Pf - θf/2) (5) Ost-Clark and Arkabutla, respectively were: SPR = 18.0E-04P t2 –

6
G.V. Wilson, et al. Soil & Tillage Research 198 (2020) 104550

Fig. 4. The power function model (A) and the asymptotic model (B) in Eq. 4 of water content as a function of accumulated rainfall for the Providence, Calloway, Ost-
Clark and Arkabutla. Note that data points correspond to the six time periods tested (0, 1, 3, 5, 7 and 10 d) of daily wetting and drying cycles.

0.825P t + 110.3, r2 = 0.206; SPR = 0.8E-04P t2 – 0.494P t + 69.3, best-fit power functions with the intercept fitted (Fig. 7A) for the Pro-
r2 = 0.195; SPR = -0.7E-04P t2 – 0.058P t + 32.2, r2 = 0.099; SPR = vidence, Calloway, Ost-Clark and Arkabutla were: Ks = 2E-06 P t−0.004,
-0.4E-04P t2 - 0.003P t + 6.7, r2 = 0.061. For each soil series, the SPR r2 = 0.674; Ks = 2E-06 P t-0.005, r2 = 0.597; Ks = 1E-06 P t-0.007,
initially increased dramatically following the first wetting/drying cycle r2 = 0.720; Ks = 1E-06 P t-0.006, r2 = 0.334)
then dramatically decreased following the subsequent wetting/drying The best fit equations tended to under-predict Ks for the initial and
cycles down to a fairly stable value. Unlike the SS, the SPR tended to first wetting/drying cycle value and over-predict Ks for all but the final
increase with depth or, as seen for the Providence series in Fig. 6B, the value of the remaining wetting/drying cycles. Setting the intercept to
SPR increased from the surface (1.25 cm) to 3.75 cm depth and then the initial Ks value prior to rainfall made the model fits significantly
stabilized or slightly decreased. This SPR-depth relationship was seen worse (r2 of 0.18 to 0.71; Fig. 7B). The respective functions with the
following each wetting/drying cycle but with differing magnitudes re- intercept set (Fig. 7B) to the initial Ks value measured prior to rainfall
flective of the relationships observed in Fig. 6A. were: Ks = 3E-06 P t−0.005, r2 = 0.499; Ks = 3E-06 P t-0.008, r2 = 0.356;
Ks = 2E-06 P t-0.008, r2 = 0.706; Ks = 2E-06 P t-0.009, r2 = 0.183). While
the fits to the initial and first wetting/drying cycle were improved, the
3.5. Hydraulic conductivity
model still over-predicted the Ks for the later (after the initial) rainfall
events.
The saturated hydraulic conductivity, Ks, decreased rapidly over the
These results of decreasing Ks with wetting/drying cycles is con-
first 3 wetting/drying cycles and reached a stable low value by the fifth
sistent with results of Kool et al. (2019) who observed a decrease in Ks
cycle (Fig. 7). The Ks exhibited exponential decay with accumulated
with time after tillage although they did not address the relation to
rainfall (Fig. 7A) with r2 values of 0.33 to 0.72. The decrease in Ks with
wetting or drying cycles. However, they observed a steep reduction in
rainfall was more rapid than the exponential function suggested. The

7
G.V. Wilson, et al. Soil & Tillage Research 198 (2020) 104550

Fig. 5. The best-fit power functions for soil shear strength as a function of accumulated rainfall at the surface (A) and the respective functions for the 5 cm depth (B)
for the Providence, Calloway, Ost-Clark and Arkabutla. Note that data points correspond to the six time periods tested (0, 1, 3, 5, 7 and 10 d) of daily wetting and
drying cycles.

Ks as bulk density increased for ρb > 1.06 g cm−3. Given the well es- reached its stable (maximum) bulk density by the third cycle. Given
tablish relationship of bulk density to wetting/drying cycles after til- these relationships, a polynomial relationship to accumulated rainfall
lage, it is not surprising that Kool et al. (2019) observed a sharp de- best represented the changes with wetting/drying cycles. However, it
crease in Ks following tillage. should be noted that the coefficient for the quadratic term for the
Providence series was essentially zero such that a linear model would
provide equally as good a fit for the decrease in erodibility with rainfall.
3.6. Erosion properties
The polynomial relations for soil erodibility (Kd) as a function of ac-
cumulated rainfall for the Providence, Calloway, Ost-Clark and Arka-
Given the bulk density increases with accumulated rainfall, con-
butla, respectively were: Kd = 6E-04P t2 – 0.802P t + 286.1,
solidation of the surface resulted in the erodibility decreasing (10–39%)
r2 = 0.884; Kd = 17E-04P t2 – 0.375P t + 242.6, r2 = 0.343; Kd = 21E-
for all but the Arkabutla series which exhibited an increase (37 %) in
04P t2 – 0.329P t + 204.3, r2 = 0.677; Kd = 11E-04P t2 - 0.836P t +
erodibility after the first wetting/drying cycle (Fig. 8). The erodibility
504.6, r2 = 0.343.
did tend to decrease with subsequent wetting/drying cycles for the
The critical shear stress initially decreased (91 %) for the
Arkabutla as well as the Providence series, however, it tended to sub-
Providence but decreased by 29 and 52 % for the Calloway and
sequently increase for the Ost-Clark and Calloway series. The sub-
Arkabutla soils, respectively, and showed no change after the first
sequent increase after the first wetting/drying cycle is likely in response
wetting/drying cycle for the Ost-Clark soil. However, the critical shear
to the decrease in SS and SPR given that the bulk density had essentially

8
G.V. Wilson, et al. Soil & Tillage Research 198 (2020) 104550

Fig. 6. The soil penetration resistance (A) as a function of accumulated rainfall for the Providence, Calloway, Ost-Clark and Arkabutla for the Providence, Calloway,
Ost-Clark and Arkabutla and the SPR for the Providence soil as a function of soil depth prior to rainfall (0.0 mm) and for four of the five wetting/drying cycles tested
(27.6, 87.1, 152.5 and 220.3 mm). Note that data points correspond to the six time periods tested (0, 1, 3, 5, 7 and 10 d) of daily wetting and drying cycles.

stress tended to decrease with continued wetting/drying cycles for all The erosion properties generally followed the expected pattern of
soil series. The best fit relations for critical shear stress (τc) as a function erodibility decrease and critical shear increase with consolidation in
of accumulated rainfall for the Providence, Calloway, Ost-Clark and response to wetting and drying cycles. However, the initial response to
Arkabutla, respectively were: τc = 0.007Pt + 0.494, r2 = 0.744; the first event was not consistent with this pattern for all soils. The
τc = 0.0017Pt + 0.466, r2 = 0.318; τc = 0.0045Pt + 0.141, erodibility and critical shear stress were highly dynamic in response to
r2 = 0.880; τc = -0.0002 Pt + 1.596, r2 = 0.001. The critical shear soil physical properties, such as water content, bulk density, Ks, shear
stress exhibited a strong linear increase with accumulated rainfall for strength and penetration resistance, which changed with each wetting
the Providence and Ost-Clark soils (r2 of 0.74 and 0.88, respectively) and drying cycle. These dynamic changes could explain the incon-
and a weak linear relationship for the Calloway (r2 = 0.32) but no re- sistencies observed in measurements at a set point in time in previous
lationship for the Arkabutla (r2 = 0.00). Thus, shear stress would need studies (West et al., 1992; King et al., 1995; Reichert et al., 2001).
to increase with wetting and drying cycles for erosion to be initiated.
Given the increase with bulk density and decrease in Ks, runoff would
be expected to increase as the soil consolidates with each wetting/ 3.7. Overall combined effects
drying cycle and, thus, the shear stress involved would likely exceed the
increases in critical shear stress. The relationships of soil erosion properties (erodibility and critical
shear stress) to accumulated rainfall (Figs. 8 and 9) do not account for

9
G.V. Wilson, et al. Soil & Tillage Research 198 (2020) 104550

Fig. 7. The saturated hydraulic conductivity, Ks, as an exponential function of accumulated rainfall with the intercept fitted (A) and set (B) to the initial Ks value
measured prior to rainfall. Note that data points correspond to the six time periods tested (0, 1, 3, 5, 7 and 10 d) of daily wetting and drying cycles.

the contributions of soil physical properties and their interactions. The series, Pt, ρt, LogSPR1.3) with the interactions of soil series with Pt and
best indicators of soil erosion were determined by multiple regression LogSPR1.3. Thus, the regression equation was dependent upon the soil
analysis to all soil properties (bulk density of the surface, saturated series with the overall and the best indicators of erodibility changes
hydraulic conductivity, shear strength at 0 and 2.5 cm depth, and soil with wetting and drying cycles (r2 = 0.97) were accumulated rain, and
penetration resistance at 0, 1.3, 3.8, 6.3, 8.8, and 11.3 cm depths) in- the associated changes in bulk density, and soil penetration resistance
cluding their log transformed values and the accumulated precipitation immediately below the surface (1.3 cm depth) (Table 3). The final
(Pt) for each soil series. The model was restricted to producing the 20 model for critical shear was τc= f (soil series, Pt, LogSPR1.3, LogKs) with
best equations based upon all possible combinations of five or fewer the interactions of soil series with LogKs and LogSPR1.3. Thus, the re-
independent variables. The multiple regression analyses suggested that, gression equation was again dependent upon the soil series with the
of the 21 variables included, the most common physical properties for overall best indicators for changes in critical shear stress (r2 = 0.93)
soil erodibility across the four soil series were: accumulated precipita- with time due to consolidation by wetting and drying cycles being the
tion (Pt), bulk density (ρt), log-transformed soil penetration resistance accumulated precipitation and the associated changes in soil penetra-
at 1.3 cm depth (LogSPR1.3) and log-transformed saturated hydraulic tion resistance at 1.3 cm and saturated hydraulic conductivity but not
conductivity (LogKs). The most common properties for critical shear bulk density (Table 4).
across the four soils were the same four variables as for erodibility. Unlike the findings of West et al. (1992) in which erodibility and
The generalized linear model (SAS Proc GLM) was used to de- critical shear could not be explained by any soil properties, both erosion
termine which of these four variables and their interaction terms were properties were found to relate to the hydraulic conductivity and soil
significant. After eliminating the non-significant interaction terms one penetration resistance and erodibility was also related to bulk density.
at a time and the non-significant main variables that were not included The findings that accumulated precipitation was correlated to both
as an interaction term, the final model for erodibility was Kd= f(soil erodibility and critical shear are in contrast to Morrison et al. (1994)

10
G.V. Wilson, et al. Soil & Tillage Research 198 (2020) 104550

Fig. 8. The soil erodibility (Kd) as a function of accumulated rainfall with the best fit regression equation. Note that data points correspond to the six time periods
tested (0, 1, 3, 5, 7 and 10 d) of daily wetting and drying cycles.

who concluded that cumulative rainfall was not consistently correlated Table 3
to rill erosion parameters. However, the findings of a correlation to soil The intercepts (Int) and coefficients for each variable and interaction terms for
penetration but not soil strength is consistent with Morrison et al. predicting erodibility (Kd).
(1994) who found very little correlation between critical shear and soil Type INT PT ρT LogSPR1.3
strength parameters.
Providence 1173.6 −0.2599 −718.372 −15.67
Calloway 1145.7 0.2819 −718.37 −47.53
4. Conclusions Ost-Clark 1107.9 0.4095 −718.37 −52.06
Arkabutla 1371.7 −0.5011 −718.37 30.47
The bulk density generally increased with time due to consolidation
following ten simulated precipitation events. The largest increase
(50–100 % of the total) occurred following the first wetting and drying density responses for the four soils tested.
Soil water content increased rapidly over the first three wetting/
cycle. The model proposed by Onstad et al. (1984) tended to over-
predict this initial increase in surface bulk density but did an adequate drying cycles then remained stable at field capacity following the
subsequent cycles. These changes in water content impacted the soil
job of representing the depth average bulk density changes. An ex-
ponential model that exhibited a more gradual increase in bulk density physical properties (shear strength, soil penetration resistance, and
saturated hydraulic conductivity) in contrasting, yet interactive ways.
was found to better represent the surface and depth-averaged bulk

Fig. 9. The critical shear stress (τc) as a function of accumulated rainfall with the best fit regression equation. Note that data points correspond to the six time periods
tested (0, 1, 3, 5, 7 and 10 d) of daily wetting and drying cycles.

11
G.V. Wilson, et al. Soil & Tillage Research 198 (2020) 104550

Table 4 Buzzword. CSA News, pp. 12–17 November.


The intercepts (Int) and coefficients for each variable and interaction terms for Green, T.R., Ahuja, L.R., Benjamin, J.G., 2003. Advances and challenges in predicting
predicting critical shear stress (τc). agricultural management effects on soil hydraulic properties. Geoderma 116, 3–27.
Hanson, G.J., Cook, K., 2004. Apparatus, test procedures, and analytical methods to
Type INT PT LOGKs LogSPR1.3 measure soil erodibility in situ. Appl. Eng. Agric. 20, 455–462.
Karlen, D.L., Veum, K.S., Sudduth, K.A., Obrycki, J.F., Nunes, M.R., 2019. Soil health
Providence 8.583 0.0025 1.189 0.328 assessment: past accomplishments, current activities, and future opportunities. Soil
Calloway −0.236 0.0025 −0.098 0.071 Tillage Res. 195, 104365.
King, K.W., Flanagan, D.C., Norton, L.D., Laflen, J.M., 1995. Rill erodibility parameters
Ost-Clark −1.822 0.0025 −0.318 −0.025
influenced by long-term management practices. Trans. ASAE 38, 159–164.
Arkabutla −5.152 0.0025 −1.421 −1.175
Knapen, A., Poesen, J., Govers, G., Gyssels, G., Nachtergaele, J., 2007a. Resistance of soils
to concentrated flow erosion: a review. Earth. Rev. 80, 75–109.
Knapen, A., Poesen, J., De Baets, S., 2007b. Seasonal variations in soil erosion resistance
In response to the changes in soil properties as the soil consolidated due during concentrated flow for a loess-derived soil under two contrasting tillage
practices. Soil Tillage Res. 94, 425–440.
to cycles of wetting and drying, the soil erosion properties were highly
Kool, D., Tong, B., Tian, Z., Heitman, J.L., Sauer, T.J., Horton, R., 2019. Soil water re-
dynamic. Erosion property changes with accumulated rainfall were not tention and hydraulic conductivity dynamics following tillage. Soil Tillage Res. 193,
consistent among soil series. The erodibility, Kd, tended to decrease 95–100.
after the initial wetting/drying event for all soils except the Arkabutla. Liu, Q.J., Wells, R.R., Dabney, S.M., He, J.J., 2017. Effect of water potential and void ratio
on erodibility for agricultural soils. Soil Sci. Soc. Am. J. 81, 622–632.
For the remaining wetting/drying cycles the Kd continued to decrease Mahalder, B., Schwartz, J.S., Palomino, A.M., Zirkle, J., 2018. Estimating erodibility
for two of the soils but increase for the other two. In contrast, the cri- parameters for streambanks with cohesive soils using the mini jet test device: a
tical shear stress increased linearly with accumulated rainfall for all comparison of field and computational methods. Water 10, 304. https://doi.org/10.
3390/w10030304.
soils except the Arkabutla soil which exhibited no relationship with Mapa, R.B., Green, R.E., Santo, L., 1986. Temporal variability of soil hydraulic properties
accumulated rainfall. with wetting and drying subsequent to tillage. Soil Sci. Soc. Am. J. 50, 1333–1338.
The overall relationship of these erosion parameters with accumu- McGregor, K.C., Cullum, R.F., Mutchler, C.K., 1998. Long-term management effects on
runoff, erosion, and crop production. Trans. ASAE 42, 99–105.
lated rainfall and all the soil properties combined with their log- Meyer, L.D., Harmon, W.C., 1979. Multiple-intensity rainfall simulator for erosion re-
transformed values was analyzed to determine the best indicators of soil search on row sideslopes. Trans. ASAE 22, 100–103.
erosion changes with consolidation. The best indicators of erodibility Moret, D., Arrúe, J.L., 2007. Dynamics of soil hydraulic properties during fallow as af-
fected by tillage. Soil Tillage Res. 96, 103–113.
changes due to consolidation by wetting/drying cycles were accumu- Moret-Fernández, D., Peña-Sancho, C., López, M.V., 2016. Influence of the wetting pro-
lated rain, surface bulk density, saturated hydraulic conductivity, and cess on estimation of the water retention curve of tilled soils. Soil Res. 54, 840–846.
soil penetration resistance at the 1.3 cm depth. The best indicators of Morrison, J.E., Richardson, C.W., Laflen, J.M., Elliot, W.J., 1994. Rill erosion of a vertisol
with extended time since tillage. Trans. ASAE 37, 1187–1196.
critical shear stress changes due to consolidation were accumulated
Osunbitan, J.A., Oyedele, D.J., Adekalu, K.O., 2005. Tillage effects on bulk density, hy-
rainfall, soil penetration resistance at 1.3 cm and saturated hydraulic draulic conductivity and strength of a loamy sand soil in southwestern Nigeria. Soil
conductivity. Both of these erosion parameters had intercepts that were Tillage Res. 82, 57–64.
dependent upon the soil series. Thus, characterization of soil properties Reichert, J.M., Schafer, M.J., Cassol, E.A., Norton, L.D., 2001. Interrill and rill erosion on
a tropical sandy loam soil affected by tillage and consolidation. In: Stott, D.E.,
to predict changes in soil erosion with time following tillage is soil- Mohtar, R.H., Steinhardt, G.C. (Eds.), 2001 Sustaining the Global Farm Meetings, pp.
specific. 592–596 May 24-29.1999.
Reynolds, W.D., Drury, C.F., Tan, C.S., Fox, A., Yang, X.M., 2009. Use of indicators and
pore volume-function characteristics to quantify soil physical quality. Geoderma 152,
Declaration of Competing Interest 252–263.
Sandin, M., Jarvis, N., Larsbo, M., 2018. Consolidation and surface sealing of nine har-
None. rowed Swedish soils. Soil Tillage Res. 181, 82–92.
Schiettecatte, W., Jin, K., Yao, Y., Cornelis, W.M., Lu, J., Wu, H., Verbist, K., Cai, D.,
Gabriels, D., Hartmann, R., 2005. Influence of simulated rainfall on physical prop-
Acknowledgements erties of a conventionally tilled loess soil. Catena 64, 209–221.
Shainberg, I., Goldstein, D., Levy, G.J., 1996. Rill erosion dependence on soil water
content, aging, and temperature. Soil Sci. Soc. Am. J. 60, 916–922.
This work was supported by the U. S. Department of Agriculture and Veiga, M., Reinert, D.J., Reichert, J.M., Kaiser, D.R., 2008. Short and long-term effects of
the Chinese Academy of Sciences, China Scholarship Council funds. The tillage systems and nutrient sources on soil physical properties of a southern Brazilian
authors wish to express their appreciation to Allen Gregory, Tony Smith Hapludox. Rev. Bras. Cienc. Solo 32, 1437–1446.
West, L.T., Miller, W.P., Bruce, R.R., Langdale, G.W., Laflen, J.M., Thomas, A.W., 1992.
and the late Alan Hudspeth for their technical support.
Cropping system and consolidation effects on rill erosion in the Georgia Piedmont.
Soil Sci. Soc. Am. J. 56, 1238–1243.
Appendix A. Supplementary data Wilson, G.V., 2011. Understanding soil-pipeflow and its role in ephemeral gully erosion.
Hydrol. Processes 25 (15), 2354–2364.
Wilson, G.V., Cullum, R.F., Römkens, M.J.M., 2008a. Ephemeral gully erosion by pre-
Supplementary material related to this article can be found, in the ferential flow through a discontinuous soil-pipe. Catena 73 (1), 98–106.
online version, at doi:https://doi.org/10.1016/j.still.2019.104550. Wilson, G.W., Shields Jr, F.D., Bingner, R.L., Reid-Rhoades, P., DiCarlo, D.A., Dabney,
S.M., 2008b. Conservation practices and gully erosion contributions in the Topashaw
Canal Watershed. J. Soil Water Cons. 63, 420–429.
References Wilson, G.V., Rigby, J.R., Dabney, S.M., 2015. Soil pipe collapses in a loess pasture of
Goodwin Creek Watershed, Mississippi: role of soil properties and past land use. Earth
Ahuja, L.R., Fiedler, F., Dunn, G.H., Benjamin, J.G., Garrison, A., 1998. Changes in soil Surf. Process. Landf. 40, 1448–1463.
water retention curves due to tillage and natural reconsolidation. Soil Sci. Soc. Am. J. Wilson, G.V., Wells, R.R., Dabney, S.M., Zhang, T., 2019. Filling an ephemeral gully
62, 1228–1233. channel: impacts on physical soil quality. Catena 174, 164–173.
Andrews, S.S., Karlen, D.L., Cambardella, C.A., 2004. The soil management assessment Zhang, Meng, Lu, Yili, Heitman, Joshua, Horton, Robert, Tusheng, Ren, 2017. Temporal
framework: a quantitative soil quality evaluation method. Soil Sci. Soc. Am. J. 68, changes of soil water retention behavior as affected by wetting and drying following
1945–1962. tillage. Soil Sci. Soc. Am. J. 81, 1288–1295.
Dick, R., 2018. Soil Health: The Theory of Everything (terrestrial) or Just Another

12

You might also like