Content uploaded by Yong Li
Author content
All content in this area was uploaded by Yong Li on Jul 21, 2020
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Content uploaded by Kayode S. Are
Author content
All content in this area was uploaded by Kayode S. Are on Jul 21, 2020
Content may be subject to copyright.
Contents lists available at ScienceDirect
Soil & Tillage Research
journal homepage: www.elsevier.com/locate/still
Farmland size increase significantly accelerates road surface rill erosion and
nutrient losses in southern subtropics of China
Yong Li*
,1
, Kayode Steven Are
1
, Zhaohua Qin
1
, Zhigang Huang, Toyin Peter Abegunrin,
Albert Assa Houssou, Hao Guo, Minghua Gu, Lanchao Wei
Key Laboratory of Agro-Environment and Agro-Product Safety, Guangxi University, 530004 Nanning, China
ARTICLE INFO
Keywords:
Farmland heterogeneity
Agricultural intensification
Road extension
Sediment production
Water pollution
ABSTRACT
Farmland size increase (FSI), which is an increase in operational size and consolidation of fragmented croplands,
is considered an effective scheme to improve agricultural productivity and reduce fertilizer use with environ-
mental protection benefits. But FSI involving road extension can enhance rill formation on road surfaces, and
increase road erosion and associated nutrient losses in a watershed. These effects have not been experimentally
quantified yet. Thus, we determined the changes in road surface rill erosion and associated nutrient losses in
response to FSI before (2017) and after (2018), in an intensive sugarcane production area within Nala wa-
tershed, in southern subtropics of China. We divided the watershed into three sub-watersheds (lower (S1), left
(S2) and right upper (S3) regions), and surveyed three categories of roads: (i) old roads (existing roads before
2017), (ii) new roads (constructed in 2018), and (iii) old-new roads (repaired and extended old roads in 2018).
The road surface rill erosion and associated nutrient losses were much higher in the lower region (S1) than the
upper regions (S1 and S2) of the watershed due to variations in field density (number of farms per unit area) and
traffic intensity. Field density reduced by 53 % while traffic intensity increased by 41 % after FSI. Following FSI,
annual road surface rill erosion, nitrogen, phosphorus and organic carbon losses increased by 88 %, 79 %, 67 %,
and 46 %, respectively. Road surface rill erosion had a significant inverse relationship with field density
(R² = 0.535, P< 0.05) but positively correlated with traffic intensity (R² = 0.728, P< 0.01). The accelerated N
and P losses offset the decrease in fertilizer use by 5–22 % due to FSI. Our results highlight the significance of
including soil erosion in FSI scheme for environmental protection. Accelerated road erosion following FSI should
be minimized by integrated soil and watershed management practices.
1. Introduction
Feeding over 7 billion people is a major challenge with global food
demand, and set to increase between 59 % and 98 % by 2050 (Valin
et al., 2014). To achieve food security, sustainable agricultural in-
tensification, involving farmland size increase (FSI), has been con-
sidered as an alternative to increase aggregate food production (Tilman
et al., 2011;Pretty and Bharucha, 2014). Farmland size increase, an
increase in operational size, reduction in field boundaries and con-
solidation of fragmented croplands, enables farmers with large farms to
capture the benefits of technological progress and increasing returns to
size (Sheng and Chancellor, 2019). Studies have shown that FSI im-
proves labour productivity, increases the efficiency of farm machinery,
and reduces excessive fertilization and environmental degradation
without compromising crop yield (Manjunatha et al., 2013;Ju et al.,
2016;Jin and Zhou, 2018;Wu et al., 2018;Khataza et al., 2019). For
instance, Wu et al. (2018) reported that a 1% FSI results in a 0.3 %
decrease in fertilizer application and would consequently reduce global
agrochemical use by 30–50 %. In most cases, the previous studies relied
on socio-economic data to assess the environmental protection benefit
of FSI in reducing fertilizer use without experimentally quantifying FSI
effect on nutrient release or soil erosion. However, FSI may increase
road erosion and associated nutrient losses due to increase in slope
length of the upslope (Li et al., 2020), and sediment linkage from roads
to streams via rilled pathways.
One of the important agricultural policies in China to FSI is the right
to transfer land to another household or firm as the new land operator
(MOA, 2014). By 2013, about 20 % of rented-out land was transferred
to farmers' professional cooperatives, more than 9 % to corporate or-
ganisations, and the rest to individual households including newly-
https://doi.org/10.1016/j.still.2020.104689
Received 16 November 2019; Received in revised form 26 April 2020; Accepted 7 May 2020
⁎
Corresponding author.
E-mail address: liyong@caas.cn (Y. Li).
1
These authors contributed equally to this work.
Soil & Tillage Research 204 (2020) 104689
0167-1987/ © 2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/).
T
named “Family Farms” (MOA, 2014). The land transferring system is
designed to increase the farmland size and accelerate agricultural
professionalism in China (Ju et al., 2016). The new land operators who
are large-scale, intensive and professional farmers, consolidate the ex-
isting fragmented lands by reducing the field density (number of farms
per unit area) for intensive agricultural mechanization (Wang et al.,
2019). An increase in subsidy for agricultural machinery purchases by
the Chinese government further accelerated the development of me-
chanization (Yang et al., 2013), and encouraged construction of new
roads (mostly unpaved roads) within agricultural fields, to facilitate
efficient delivery of farm inputs and agricultural produce (Zhang and Li,
2018;Fig. 1). Despite their significance in efficient delivery of farm
inputs and use of fertilizer for the crop production system, it is unclear
yet whether FSI can significantly accelerate road erosion and associated
nutrient losses within intensively managed agricultural areas.
Road erosion assessments have received considerable attention, and
various assessments compared their magnitudes to the adjacent un-
disturbed lands (Ziegler et al., 2000;Croke and Mockler, 2001;Dubé
et al., 2004;Ramos-Scharrón and MacDonald, 2007;Cao et al., 2014;
Ramos-Scharrón, 2018;Zhang et al., 2019). For instance, road erosion
rates recorded by Ramos-Scharrón and MacDonald (2007) from the US
Virgin Islands were up to four orders of magnitude higher than the
undisturbed hillslope areas; whereas, Croke and Mockler (2001) re-
ported erosion rates being six times higher in coastal southeast Aus-
tralia. In Puerto Rico, Ramos-Scharrón (2018) reported an annual road
erosion rate of 204 Mg ha
−1
yr
−1
, which was 200 times greater than
adjacent undisturbed plots (1.6 Mg ha
−1
yr
−1
). Recently, Zhang et al.
(2019) reported that 45 % of the total length of unpaved roads in
China’s Loess Plateau was destroyed by road erosion. However, most of
these works identified rainfall intensity and duration, the magnitude of
runoff from the road segment, road surface materials and hydraulic
characteristics, road gradient, traffic, construction and maintenance,
and contributing road area as the influencing factors of road erosion. It
is obvious that sediment generated from the roads is a key environ-
mental concern due to its potential impact on aquatic ecology. It is
worth noting that the negative impacts of road erosion on the en-
vironment are often neglected once the roads are repaired. However,
water quality impairment by road erosion is dependent not only on the
source strength but also on the understanding of the connectivity of the
delivery path (Croke et al., 2005;Ramos-Scharrón and LaFevor, 2018).
In the context of this study, connectivity of sediment to the streams is
considered on the basis of numbers of extended roads that intercepted
the stream channels and increased rill volumes on the roads due to FSI.
According to Ramos-Scharrón and LaFevor (2018), road densities are
insufficient to explain the watershed impacts of roads on runoff and
sediment yields. But there is a need for understanding of how FSI in-
volving road extension can influence road-stream crossing, and sub-
sequent increase in soil and nutrient export from agricultural lands to
rivers and lakes. Road-stream crossings have been considered as an
important and direct cause of road-stream connectivity (Lane and
Sheridan, 2002;Thomaz et al., 2014, Ramos-Scharron and LaFevor,
2018). It is also essential to know that most farmers litter the roads with
agrochemicals (fertilizers, herbicides, pesticides, etc.,) before and
during the application, allowing the chemical residues to drain directly
into the river channels, as observed in our study site (Fig. A.
1
).
The aforementioned studies, and several others, assessed the
Fig. 1. Graphical illustration of cause–consequence relationship of road erosion and associated nutrient losses.
Y. Li, et al. Soil & Tillage Research 204 (2020) 104689
2
negative impacts of road erosion and provided insights into the factors
controlling road sediment generation, transport, and road-stream con-
nectivity. But a key knowledge gap remains on how FSI scheme influ-
ences road surface rill erosion in a hilly agricultural watershed. Thus,
the objective of this study was to quantify soil and nutrient losses by
road surface rill erosion following FSI in an intensively agricultural area
of southern subtropical China. Our hypothesizes are that: (i) FSI de-
creases landscape heterogeneity (dissimilar farm plots’ configuration
with nonuniform overland flow due to field boundaries), and influences
road surface rill erosion driven by hillslope overland flow; (ii) road
extension (RE) in FSI scheme enhances sediment and associated nu-
trient export from hillslope to the stream channel due to increase in
road-stream crossing (ii) traffic intensity causes spatial differences in
road sediment production and associated nutrient losses between sub-
watersheds.
2. Materials and methods
2.1. Study site
The study was conducted in Nala watershed (107° 39′ 29′′–107° 40′
17′′E, 22° 20′ 50′′–22° 20′ 36′′N) of Pearl River Basin, in Guangxi
Zhuang Autonomous Region of China (Fig. 2). The watershed, which
covers an area of 1.2 km
2
is part of the watersheds that drain to the
Kelan Reservoir. The watershed is located in a core intensive sugarcane
production area. This watershed belongs to typical areas under the FSI
scheme by the Chinese government. From the end of 2017 (exactly
December 2017) to March 2018, the small farmlands were converted to
large farmland sizes in Nala watershed. Many new roads were con-
structed and old roads were widened. The mean altitude of the wa-
tershed is about 1200 m a.s.l. It has a sloping and rolling topography
with an average gradient of about 35 %, while the elevation between
valley and hill top is between 142 and 182 m. The dominant lithology in
this area is limestone with a reddish lateritic soil, classified as Rendzic
Leptosols (topsoil texture: silty loam). The area belongs to a humid
subtropical climate with an annual mean temperature of 20.8 ℃. Pre-
cipitation and rainfall intensity data were collected in 2017 and 2018
using a tipping bucket-recording rain gauge installed at a monitoring
station within the watershed. Annual precipitations of 1532 mm and
1548 mm were recorded for 2017 and 2018, respectively. Analysis of
the monthly rainfall distributions and intensities for the two-year
period did not differ significantly (Li et al., 2020; Fig. A.
2
). Approxi-
mately 84 % of the annual precipitation falls between March and
September).
The study site was divided into three sub-watersheds (S1, S2 and S3)
based on topographic directions and drainage areas (Fig. 2). The sub-
watershed 1 (S1) occupied the lower region of the watershed and covers
0.21 km
2
land area. The sub-watershed 2 (S2) occupied the upper right
region of the watershed and covers 0.57 km
2
land area whilst sub-wa-
tershed 3 (S3) occupied the upper left region of the watershed and
covers 0.42 km
2
land area.
2.2. Watershed roadmaps and field survey
The roadmaps and field surveys were conducted during 2017 using
an aerial photo. The length of the road segments (a continuous road
surface that begins and ends at an intersection or change in slope
gradient) were measured. The location of each road (for its start and
end point) was determined using a global positioning system (GPS,
Trimble Juno 3E). The road lengths were measured using rolling meter
tape while their gradients were measured with a clinometer. In 2017,
farmland plots were based on relevant sub-plot data and maps of the
parcel-to-household policy in China. Data such as the distribution map
of the fields in numbers and size were obtained by exporting the ima-
geries into GIS environment using ArcGIS Map 10. In 2018, the farm-
land plots were delineated based on satellite imageries and field mea-
surements, with the results finally calculated using ArcGIS Map 10.
Based on 5-meter resolution of digital elevation model (DEM) generated
by 1:2000 scale topographic map, areas of the sub-watersheds and the
catchment area (runoff contributing area) were delineated using ArcGIS
Map 10. Field density, which is the number of farms per unit area, is
measured by dividing the number of farms/holdings in each year by the
watershed area.
Fig. 2. Study location and the road networks.
Y. Li, et al. Soil & Tillage Research 204 (2020) 104689
3
2.3. Measurement of road and rill characteristics, and traffic intensity
In this study, the road characteristics (i.e. length, width, area, gra-
dient and density) were measured in the watershed in 2017 and 2018.
The studied roads were classified as old, old-new and new based on the
time of construction and maintenance. Three road types were studied:
(i) old road – existing roads and last maintained before 2017; (ii)
old–new road – existing road that was repaired and extended in early
2018; and (iii) new road – newly built roads in 2018 (Fig. 2). According
to the farmers cultivating the watershed, road maintenance in the wa-
tershed is scheduled for every two years. The roads were constructed
simply by displacing surface soil of farmland with a crawler tractor as
described by Cao et al. (2014). During the study period, all categories of
roads in the watershed were unpaved. The roads were constructed
without any engineering standards (i.e. no culverts, inadequate and
poorly maintained side drains, poor compaction, and the road surfaces
are not crown shaped, Fig. 3).
In this study, a method similar to the ones presented in Anderson
and MacDonald (1998) and Zhang et al. (2019) was adopted in the
assessment of road surface rill erosion as the main form of road-related
soil losses. The rill length, width and depth on the road segments were
measured with a metric tape and a steel ruler. The ruts created on the
road surfaces by vehicular traffic were also considered in our mea-
surement. The rill depth and width were measured at the cross-section
intervals of 2–5 m based on the road dimensions, which were fitted into
trapezoidal shape. The rills were measured from the starting point to
Fig. 3. Road features in the watershed showing (a) rills on old unpaved road; (b) initiation of ruts on the road by traffic wheel; (c) rills developed from ruts; (d)
expanded rill directly linked to stream channel.
Y. Li, et al. Soil & Tillage Research 204 (2020) 104689
4
the point it discharged to the side drains, stream channels or where we
had sediment deposition, as a road erosion segment, such that there
were always one or more rills on a road segment (Fig. 3). For rill that
drains to another rill, the length was measured from its starting point to
the confluence. A cylindrical core of 5 cm in height and diameter was
used to take soil samples (4 core samples per road segment) for road
surface bulk density (ρ
b
) following Grossman and Reinsch (2002) pro-
cedure. The road segment area for each road type was calculated fol-
lowing equation (Eq. 1):
A
x
=l
x
*w
x
(1)
where A
x
is road segment area (m
2
), l
x
is the road segment length
(m), and w
x
is the road width (m) of road type x. In 2017 and 2018, the
road density in a sub-watershed was calculated as the total road length
divided by the area of the sub-watershed.
Traffic intensity was measured based on evaluation of traffic levels
(cart, light-vehicle and truck traffic) using farmers’ operational records
and the Darwish and Abu Bakar (2015) method. The traffic intensity
index (number of passages yr
−1
,T
tf
) with respect to the weight of trucks
(which often visit during harvest time) and other vehicular types on the
roads was computed according to (Eq. 2):
= + +
=
T np[ ( ) ]
tf
i
n
iL WL
1 12
(2)
where,
npi
is the number of passage of trucks per month =
Y
T
c
w
,
=np 0
i
in a month when there is no truck movement. Y
c
is the total weight of
harvested sugarcane per watershed (tons) and T
w
is the capacity of the
vehicle (tons).
L
is the impact scoring factor of loaded truck, and
WL
is
the impact scoring factor of truck without load. γ is the number of
passages of other types of vehicles, and i is monthly value (i.e. 1–12th
month). The scoring factors
L
and are defined as:
=
+
=
+
D
D D
D
D D
and
L
L
L WL
WL
WL
L WL
D
L
is the distance travelled with load (km) and D
WL
is the distance
travelled without load (km)
2.4. Road surface rill erosion rates, nutrient losses and data analysis
Soil samples were collected from the rill wall of each category of
road and analyzed for road erosion and nutrient losses. The soil and
associated nutrients (total nitrogen (TN), total phosphorus (TP) and soil
organic carbon (SOC)) were analysed following standard procedures.
An air-dried soil sample was gently ground and allowed to pass 0.5 mm
sieve, and thereafter analysed for organic carbon, total nitrogen and
total phosphorus (TP). The SOC content was determined by titration
(Nelson and Sommers, 1982) while TN was analyzed using the Kjeldahl
method (Bremner, 1996). Phosphorus (P) was analyzed colorimetrically
using a hot perchloric acid digestion (Sommers and Nelson, 1972).
Road surface rill erosion (R
e
, kg m
−2
) by each road segment was
calculated annually based on the rill volumes as follows (Eq. 3):
=
=
d w l
A
Re *
b
i
n
i i i
xi
1
(3)
where ρ
b
is the road surface bulk density, iis the road type and d
i
,w
i
and l
i
are respectively, the depth, width and length of the rill on the
road segment. A
xi
is the cross-sectional area of the road. Annual road
erosion after FSI (R
e (after FSI)
) was determined based on the changes in
rill volumes pre- and post-FSI as (Eq. 4):
=R R R
e after FSI e cum after FSI ebefore FSI( ) ( ) ( )
(4)
where R
e (cum after FSI),
is the cumulative road surface rill erosion post-
FSI in 2018, and R
e (before FSI)
is the road surface rill erosion pre-FSI in
2017.
Total soil loss (SL
w
) rate from road surface rills per annum in a
watershed (A
w
) was calculated as (Eq. 4):
=
=
SL Sd A
A
*
wi
n
r
xi
w
1
(4)
where Sd
r
is the average annual sediment yield of each road (t ha
−1
yr
−1
).
Nutrient loss (NL) by road erosion was calculated as (Eq. 5):
=
=
NL SL C*
i
n
wi ti
1
(5)
where SL
w
is the soil loss by road surface rill erosion (t ha
−1
yr
−1
), C
t
is
the concentration (g kg
−1
) of the sediment associated nutrient from the
road and i is the road number.
Fertilizer application rates were obtained from farmers’ records in
both 2017 and 2018. The environmental protection benefits of FSI in
reducing fertilizer use (details in Appendix A.1) was calculated from
data obtained from the nutrient offsets by road surface rill erosion (Eq.
6):
=
N P N P
Fert Fert
N or P offset (%) ( / / )
( ) * 100%
i
after IFS before IFS
after IFS before IFS
(6)
where N/P
before IFS
and N/P
after IFS
are the nutrient (N or P) losses
before and after IFS, respectively, while Fert
before IFS
and Fert
after IFS
are
the applied fertilizers (N or P) before and after IFS, respectively.
Soil sampling from the rill walls was conducted with three re-
plications on every 5-m rill, from which mean and standard error were
calculated. Road surface rill erosion and other variables including total
nitrogen, total phosphorus and organic carbon were analysed using
one-way analysis of variance (ANOVA) at α = 0.05 probability to test
the significance levels. When treatment effects were significant (p≤
0.05), the means were separated using Least Significant Difference
(LSD). Pearson correlation analyses were used to determine the re-
lationships between road surface rill erosion and influencing field
density, runoff contributing area of the roads, traffic intensity, surface
bulk density and road gradients.
3. Results
3.1. Road characteristics
Table 1 summarizes road characteristics before and after FSI in 2017
and 2018. There was significant increase in road numbers, density and
surface area following construction of more roads in 2018 across the
entire watershed. The road numbers, density and surface area in the
entire watershed increased by 163 %, 121 % and 252 %, respectively, in
2018 as compared to 2017. However, there were significant differences
in road numbers, densities and surface areas between the lower (S1)
and upper (S2 and S3) reach of the watershed. The increase in road
numbers and density was significantly higher in S3 (350 %, 178 %)
than S2 (167 %, 132 %) and S1 (80 %, 63 %), respectively, as compared
to their respective values in 2017. However, the road surface area in S2
increased by 298 %, which was significantly higher than the increase
found in S1 (208 %) and S3 (182 %). The number of road-stream
crossings increased from 2 in 2017 to 6 in 2018 (Fig. 2), indicating an
increase from 2 out of 16 (12.5 %) to 6 out of 42 (14.3 %) in the entire
watershed.
3.2. Road surface rill characteristics and erosion
The road surface rill characteristics varied significantly between
2017 and 2018 in the entire watershed (Table 2). The average number,
length and volume of the rills on road surfaces in 2018 increased by 150
%, 27.7 % and 18.2 %, respectively, as compared to their respective
values in 2017. More road surface rills and the corresponding decrease
in road bulk density were recorded in 2018 on constructed new and old-
Y. Li, et al. Soil & Tillage Research 204 (2020) 104689
5
new roads than the existing old roads (Tables 1 and 2). Compared to
2017, increase in the number of road surface rills in 2018 was highest in
S3 (380 %) followed by S2 (180 %) and the least in S1 (64 %). In
contrast, the increase in volume of rills on road surfaces was highest in
S1 (222 %) followed by S2 (219 %) and the least in S3 (176 %).
The annual road surface rill erosion and its relative change between
2017 and 2018 in the entire watershed and sub-watersheds are sum-
marized in Fig. 4. Road erosion increased significantly in the entire
watershed, but the increase varied between lower (S1) and upper (S2
and S3) regions of the watershed. In entire watershed, road erosion
increased from 4.49 t ha
−1
yr
−1
in 2017 to 8.43 t ha
−1
yr
−1
with an
increase of 88 % in 2018 as compared to 2017. For sub-watersheds, the
magnitudes of the road surface rill erosion were in the order of S1 > S2
> S3 in both 2017 and 2018. The annual increase of 176 %, 42 %, and
85 % was obtained for road surface rill erosion in S1, S2 and S3, re-
spectively, in 2018 as compared to 2017 (Fig. 4). Comparing the road
types, rill erosion on new and old-new roads increased by 33–410 %
than old roads (Fig. 5).
3.3. Nutrient losses associated with road surface rill erosion
Nutrient losses from the road surfaces followed trends similar to
surface rill erosion in 2017 and 2018 across the watershed (Fig. 6). The
annual N, P and SOC losses by road surface rill erosion increased from
6.19, 2.03 and 83.34 kg ha
−1
yr
−1
in 2017 to 11.07, 3.40 and
121.48 kg ha
−1
yr
−1
in 2018, respectively. Thus, the annual increase in
nutrient losses was 46–79 % across the entire watershed (Fig. 6). For
sub-watersheds, there were much greater losses of N, P and SOC by road
Table 1
Road characteristics in the watershed before (2017) and after (2018) farmland size increase.
Year Watershed Road type Road counts
b
Road gradient (
o
) Bulk density (g cm
−1
) Road Length(km) Road area (m
2
) Road width (m)
c
Road density (km/km
2
)
2017
a
S1 Old 5 3 ± 1.41 2.05 1.23 3277 2.7 5.9
S2 Old 9 12 ± 2.83 1.67 3.02 8803 2.9 5.3
S3 Old 2 1.5 ± 0 1.66 0.74 3713 5.0 1.8
Entire 16 5.5 1.79 5.00 15,793 3.2 4.2
2018 S1 Old 1 3 ± 1 2.05 0.27 1365 5.0 1.3
Old-New 4 8.8 ± 3.18 1.88 0.96 4780 5.0 4.6
New 4 9 ± 2.83 1.59 0.79 3962 5.0 3.8
Entire S1 9 6.9 1.84 2.02 10,107 5.0 9.6
S2 Old 2 12 ± 2.83 1.67 0.92 4591 5.0 1.6
Old-New 7 7.9 ± 2.53 1.66 2.11 10,530 5.0 3.7
New 15 9.9± 97 1.63 3.98 19,903 5.0 7.0
Entire S2 24 9.9 1.65 7.01 35,023 5.0 12.3
S3 Old 2 1.5 ± 0 1.74 0.74 3713 5.0 1.8
New 7 10.3 ± 93 1.66 1.35 6760 5.0 3.2
Entire S3 9 5.9 1.70 2.09 10,473 5.0 5.0
Entire 42 7.8 1.71 11.12 55,603 5.0 9.3
a
S1, S2 and S3 represent lower watershed 1, left upper watershed 2 and right upper watershed 3, respectively.
b
The road gradient value ± followed by its standard deviation.
c
Road density was calculated as the road length divided by the watershed area (The areas of S1, S2, S3 and the entire watershed are 0.21, 0.42, 0.57 and 1.20 km
2
,
respectively).
Table 2
Rill characteristics of different roads in the watershed before (2017) and after (2018) farmland size increase.
Year Watershed Road type Rill count Rill length (m) Rill volumes (m
3
)
Entire Maximum mean Entire maximum mean
2017 *S1 Old 146 1041 504 7.1b 96 42 0.66b
S2 Old 171 2161 868 12.6b 125 54 0.73b
S3 Old 35 1005 633 28.7a 68 42 1.94a
Entire watershed 352 4207 12.0 289 0.82
2018 S1 Old 45 504 504 11.2a 30 30 0.67b
Old-New 101 1343 672 13.3a 165 91 1.63a
New 94 1335 696 14.2a 114 51 1.21a
Entire S1 240 3182 13.3a 309 1.29
S2 Old 31 1186 868 38.3a 82 54 2.65a
Old-New 140 2436 854 17.4b 110 25 0.79b
New 308 4414 802 14.3b 210 67 0.68b
Entire S2 479 8036 16.8 401 0.84
S3 Old 35 1005 633 28.7a 68 42 1.94a
New 133 1871 522 14.1b 120 29 0.90b
Entire S3 168 2876 17.1 188 1.12
Entire watershed 887 14,094 15.9 898 1.01
a
S1, S2 and S3 represent lower watershed 1, left upper watershed 2 and right upper watershed 3, respectively. Total rill length is the sum of the rill lengths, maximum
rill length is the longest rill length on each road and the mean rill length is the total rill length divided by the rill counts. Total rill volume is the sum of the rill
volumes, maximum rill volume is the largest rill volume obtained on the road of a particular road type in a particular year. The mean rill volume is the total rill
volume divided by the rill counts. Within a column in the same year among the road types in a sub-watershed, means followed by the same letter are not significantly
different.
Y. Li, et al. Soil & Tillage Research 204 (2020) 104689
6
surface rill erosion in the lower (S1) than upper (S2 and S3) sub-wa-
tersheds in 2018. The new and old-new roads contributed significantly
more to N, P and SOC losses than the old roads in each of the sub-
watersheds (Fig. 7). The magnitudes of N, P and SOC losses on new and
old-new roads were significantly higher than on old roads by 49–539 %,
53–249 % and 17–1211 %, respectively.
We compared the percentage reduction in fertilizer use and nutrient
losses through road surface rill erosion between 2017 and 2018. The
annual applied N and P fertilizers reduced by 7% of N and 54 % of P in
2018 as compared to 2017 (Table 3). The reduction in fertilizer use was
much lower than N (79 %) and P (67 %) losses by the road surface rill
erosion. The amounts of N and P offset by the losses were 22 % and 5%,
respectively, due to FSI in 2018.
3.4. Field density and traffic intensity
Effects of FSI and RE on field density and traffic intensity are pre-
sented in Table 3. Following FSI in 2018, field density in the entire
watershed decreased by 53 % as compared to 2017. However, the de-
crease in field density varied across the sub-watersheds with a 71 %, 48
%, and 34 % decrease occurring in the lower (S1), left upper (S2) and
right upper (S3) regions, respectively (Table 3). The decrease in field
density reduced farmland fragmentations with loss of many field
boundaries in 2018 than 2017.
Following RE, traffic intensity increased significantly in the wa-
tershed (Table 3). Traffic intensity increased in the entire watershed by
41 % in 2018 as compared to 2017. However, the increase in traffic
intensity varied significantly among the different sub-watersheds,
where traffic intensity increased significantly in lower (S1) region than
upper (S2 and S3) regions of the watershed by 110–190 %.
Fig. 4. Annual road surface rill erosion and its changes before and after farmland size increase. S1, S2 and S3 represent sub-watersheds 1–3 at lower, left upper and
right upper watershed, respectively. Error bars are the standard error, and different letters indicate significant differences within a watershed at p≤ 0.05.
Fig. 5. Effects of road types on road
surface rill erosion after farmland size
increase (FSI) in 2018. S1, S2 and S3
represent sub-watersheds 1–3 at lower,
left upper and right upper watershed,
respectively. Error bars are the stan-
dard error, and different letters in-
dicate significant differences between
road types within a watershed at p≤
0.05.
Y. Li, et al. Soil & Tillage Research 204 (2020) 104689
7
Fig. 6. Annual (A) nitrogen (N), (B) phosphorus (P), and (C) soil organic carbon (SOC) losses before and after farmland size increase (FSI). S1, S2 and S3 represent
sub-watersheds 1–3 at lower, left upper and right upper watershed, respectively. Error bars are the standard error, and different letters indicate significant differences
within a watershed at p≤ 0.05.
Y. Li, et al. Soil & Tillage Research 204 (2020) 104689
8
3.5. The factors influencing road surface rill erosion
We analysed many influencing factors of road surface rill erosion but
found four key factors: field density, road density, traffic intensity index
and runoff contributing area, that were significantly correlated with road
surface rill erosion following FSI scheme. We found a significant inverse
relationship between road surface rill erosion and field density (P<
0.05, R
2
= 0.535). A decrease in field density following FSI resulted in
increase in road surface rill erosion and associated nutrient losses
(Fig. 8a). Following FSI and RE, road surface rill erosion had positive and
significant correlations with road density (R² = 0.609, P < 0.05, Fig. 8b)
and runoff contributing area (R² = 0.728, P< 0.01, Fig. 8c). A non-
linear regression analysis between road surface rill erosion and traffic
intensity index also showed a positive and significant relationship
(R
2
= 0.631, P< 0.01, Fig. 8d), suggesting that traffic intensity con-
tributed to spatial differences in road surface rill erosion between sub-
watersheds. However, road surface bulk density was slightly correlated
(R² = 0.219, P< 0.05) with road surface rill erosion but the correlation
between road gradients and road surface rill erosion was not significant
(R² = 0.209, P> 0.05; Fig. A.
3
). The impacts of surface bulk density and
road gradient on road erosion were probably concealed by variation in
traffic intensity and runoff contributing area in the watershed.
4. Discussion
This study highlights the negative environmental impact of FSI on
accelerating road surface rill erosion and soil nutrient losses of the in-
tensive agricultural areas in the southern subtropics of China. Following
FSI, the annual road surface rill erosion and associated nutrient losses
increased by 46–88 % compared to 7–54 % reduction in fertilizer use.
The accelerated N and P losses by FSI scheme offset the environmental
benefit of a decrease in fertilizer use by 5–22 % after FSI. The road
erosion and associated nutrient losses have not been included in the
environmental benefit of the FSI scheme (Chen et al., 2014;Ju et al.,
2016;Jin and Zhou, 2018;Wu et al., 2018). For instance, Wu et al.
(2018) made a global estimation of a 30–50 % decrease in agrochem-
icals (fertilizer, pesticides, etc.) and a 50 % decrease in environmental
impact of those agrochemicals following FSI. Comparing the estima-
tions with the nutrient losses by road surface rill erosion in our wa-
tershed study, the environmental benefit cited to support FSI in redu-
cing fertilizer use may have been overestimated by at least 3–11 % if
omitting road surface rill erosion alone (Appendix A.1).
The accelerated road surface rill erosion and associated nutrient
losses have a close relationship with FSI. Farmland size increase in-
volves the consolidation of fragmented lands by land operators, which
Fig. 7. Effect of road types on nutrient losses after farmland size increase (FSI) in 2018. Different letters indicate significant differences for a particular nutrient type
between road types within a watershed at p≤ 0.05.
Table 3
Annual field numbers, field density, farm size and traffic intensity and fertilizer application rates in the study watershed following farmland size increase.
Year Watershed Number of fields Field density (ha
−1
) Average farm size (ha) Traffic intensity (passage yr
−1
) Fertilizer use (kg ha
−1
)
N P
2017
a
S1 323 16a (0.065 ± 0.07)b 411.9a 308.4 47.8
S2 864 15a (0.066 ± 0.08)b 248.7b 308.4 47.8
S3 429 10b (0.098 ± 0.14)a 181.7b 308.4 47.8
Entire 1644 14 0.073 ± 0.018 842.3 308.4 47.8
2018 S1 109 5b (0.192± 0.29)a 649.8a 286.0 21.8
S2 380 7a (0.129 ± 0.22)b 313.2b 286.0 21.8
S3 326 7a (0.150 ± 0.25)b 227.7c 286.0 21.8
Entire 764 6 0.157 ± 0.026 1190.7 286.0 21.8
a
S1, S2 and S3 represent lower watershed 1, left upper watershed 2 and right upper watershed 3, respectively. Within a column in the same year, means followed
by the same letter are not significantly different.
Y. Li, et al. Soil & Tillage Research 204 (2020) 104689
9
reduced field heterogeneity and boundaries in the study site. A major
distinction between small and large farmland size is the number of field
boundaries, which control the speed of overland flow. Our study evi-
denced that a 1% decrease in field density increased road surface rill
erosion by 1.6 % and nutrient losses by 0.9–1.5 % in the entire wa-
tershed. In the studied watershed, Li et al. (2020) recently reported 64
% increase in slope length and disappearance of about 39 % of field
boundaries following FSI. The disappearance of many field boundaries
in the watershed reduced the heterogeneity of the hillslopes and con-
sequently widened the runoff contributing area that intercepted the
road networks. This validates our hypothesis that FSI will increase
hillslope runoff by a decrease in hillslope heterogeneity and conse-
quently increase road surface erosion. So far, only Van Oost et al.
(2000) considered a relationship between soil erosion and impact of
field boundaries in slowing down runoff contribution from hillslopes.
They reported a 12–29 % increase in hillslope erosion following
200–300 % FSI on Belgian loam belt. Their values on hillslope erosion
were significantly lower than accelerated road erosion (87.6 %) esti-
mated in our study following FSI (53.3 %). However, these two studies
agreed that accelerated soil erosion by FSI for different erosion types
has an inverse relationship with field boundaries. Although, rainfall has
been considered as a primary influencing factor of road erosion (Ziegler
et al., 2001;Lane and Sheridan, 2002;Ramos-Scharrón and MacDonald,
2007;Ramos-Scharrón and LaFevor, 2018;Zhang et al., 2019) analysis
of rainfall data showed no significant difference in monthly
distributions and intensities during the study years in the watershed (Li
et al., 2020; Fig. A.
2
).
Another important factor accelerating road erosion is RE in the FSI
scheme. Road extension involves the construction of new roads and an
increase in road density to facilitate farm mechanization and efficient
delivery of agricultural products. The farmers make new roads by using
tractors to displace the surface soil of farmland without engineering
standards. The new roads are less compacted and have loose surface
materials with rich nutrients, thus creating more rills on new roads than
old roads following heavy rainfall. Both the rill numbers and volumes
on new roads increased more than old roads by 76–894 %, resulting in a
significant increase in road surface rill erosion and associated nutrient
losses following RE. More than a twofold increase in road density was
observed following FSI, and this perhaps increased the number of road-
stream crossings from 13 % to 14 % of the road networks in 2017 and
2018, respectively (Fig. 2). The increase in road-stream crossing and
rilled pathways may have contributed to the sediment connectivity
from agricultural lands to the stream channels during the study (Li
et al., 2020). Compared with isolated road segments, Ziegler et al.
(2004) reported that road sediment transports may be greatly influ-
enced by upslope runoff intercepted by road networks at the watershed
scale. Many studies also reported that road and hillslope erosion have
positive relationship with surface rill density (Ziegler et al., 2000;Li
et al., 2003;Dubé et al., 2004;Ramos-Scharrón and MacDonald, 2007;
Fu et al., 2010;Rizalihadi et al., 2013;Cao et al., 2014;Ramos-
Fig. 8. Relationships between road surface rill erosion and (a) field density, (b) road density, (c) traffic intensity index, and (d) runoff contributing area in 2017 and
2018.
Y. Li, et al. Soil & Tillage Research 204 (2020) 104689
10
Scharrón, 2018;Zhang et al., 2019).
Agricultural traffic following RE in the study watershed significantly
influenced road surface rill erosion and associated nutrient losses. Road
extension across the watershed allowed the farmer to broaden farm
activities at a near and far distance, especially the initially non-roaded
areas (Fig. 1). This resulted in extra traffic, especially truck on the
roads. The truck wheel, due to crushing and upward forcing of fine-
grained sediment from road bed, created ruts on road surfaces, with
subsequent increase in surface rill volumes and sediment production
(Sheridan et al., 2006;van Meerveld et al., 2014;Sosa-Pérez and
MacDonald, 2017). Before RE in the studied watershed, heavy-duty
trucks stopped at a meeting point, and relied on tractors and carts to
transport agricultural inputs and produce to and from non-roaded area
(Fig. A.
4
). However, RE following FSI increased traffic intensity index
by 41 %, with truck passages accounting for more than 70 % of the total
road traffic. This perhaps caused significant spatial variation in the ruts
created by vehicular wheels and road surface rill erosion between roads
with high and low traffic. For example, the road networks in our study
site were built such that the vehicular movement to and from the other
two sub-watersheds (S2 and S3) must pass through the sub-watershed
S1. This perhaps caused higher traffic intensity estimated in S1 than S2
and S3 of the watershed, and consequently accelerated road surface rill
erosion and associated nutrient losses in S1. Other studies (Reid and
Dunne, 1984;Ziegler et al., 2001;Sheridan et al., 2006;van Meerveld
et al., 2014;Sosa-Pérez and MacDonald, 2017) reported 2–25 times
increase in road erosion following an increase in traffic intensity. Our
observed results are little lower than the range of other studies fol-
lowing an increase in traffic intensity since we only accounted for road
erosion from surface rills. The differences in road erosion following an
increase in traffic intensity are dependent on the types of traffic and soil
surface characteristics.
Thus, the current environmental benefit assessment of FSI in redu-
cing fertilizer overuse may be only one part of the story. If we only
consider the positive effect of fertilizer reduction but omit soil erosion
accelerated by FSI, we may overestimate the environmental benefit of
the scheme. We highlighted only the environmental pollution risk of
road surface rill erosion and associated nutrient losses following FSI in
our study, but this risk could be substantial if other delivery pathways
are included. Previous studies by Oshunsanya et al. (2019) and Li et al.
(2020) show that soil erosion, rather fertilization, is a major driving
force in agricultural non-point source pollution. It is evident from our
study that both accelerated soil erosion and fertilizer overuse should be
considered as the major driving forces for agriculture-derived water
pollution. From scientific view, the study provides new insights into the
mechanisms and roles of FSI in increasing sediment and nutrient export
from agricultural lands to the river channels. Scientific efforts must be
geared towards identifying other hydrological pathways, apart from
road surface rills, for sediment and nutrient losses within agricultural
watersheds following FSI scheme. Our results also suggest that the
policy makers should rethink the withdrawal of fertilizer subsidy in
“Zero Growth of Chemical Fertilizer Use Program” (Ju et al., 2016;Jin
and Zhou, 2018), and include soil and water conservation practices for
greater benefits of the FSI scheme.
5. Conclusion
This study highlights the negative environmental impact of FSI in
increasing road erosion and associated nutrient losses in an intensive
agricultural watershed. Road surface rill erosion and associated nu-
trient losses, as shown in our results, increased significantly following
FSI. The accelerated road surface rill erosion could be linked to a re-
duction in field density with disappearance of many field boundaries,
which could have acted as barriers to slow down the water flow to the
road networks. On the other hand, road extension, following FSI, in-
creased the road-stream crossings, and this perhaps influenced the se-
diment and hydrological connectivity between hillslopes and stream
channels. The observed spatial differences in road surface rill erosion
between sub-watersheds were linked to traffic intensity with a sig-
nificant correlation between road surface rill erosion and traffic in-
tensity. Although, FSI scheme reduced fertilizer use across the wa-
tershed, but the N and P losses offset the observed decrease in fertilizer
use. Thus, omission of soil erosion in FSI scheme may have over-
estimated the environmental benefit cited to support FSI in reducing
fertilizer overuse. From scientific view, long-term monitoring and ob-
servation studies are required to verify all the pathways connecting
sediment and associated nutrient losses from hillslopes and river
channels in FSI assessment. Policy makers are also encouraged to in-
clude effective soil conservation and watershed management program
in IFS scheme to minimize soil erosion and agricultural diffuse pollu-
tion.
Declaration of Competing Interest
The authors declared that they have no conflicts of interest to this
work.
Acknowledgements
This work was supported by Science and Technology Major Project
of Guangxi (Guike AA17204078) and Base Scientific Technology and
Talents Programme (Guike AD17195098).
Appendix A. Supplementary data
Supplementary material related to this article can be found, in the
online version, at doi:https://doi.org/10.1016/j.still.2020.104689.
References
Anderson, D.M., Macdonald, L.H., 1998. Modelling road surface sediment production
using a vector geographic information system. Earth Surf. Process. Landforms 23,
95–107.
Bremner, J.M., 1996. Nitrogen-total. In: Sparks, D.L., Page, A.L., Helmke, P.A., Loeppert,
R.H., Soltanpour, P.N., Tabatabai, M.A., Johnston, C.T., Sumner, M.E. (Eds.),
Methods of Soil Analysis, Part 3. Soil Science Society of America, Madison, pp.
1085–1121.
Cao, L., Zhang, K., Liang, Y., 2014. Factors affecting rill erosion of unpaved loess roads in
China. Earth Surf. Process. Landforms 39, 1812–1821. https://doi.org/10.1002/esp.
3569.
Chen, X., Cui, Z., Fan, M., Vitousek, P., Zhao, M., Ma, W., Wang, Z., Zhang, W., Yan, X.,
Yang, J., Deng, X., Gao, Q., Zhang, Q., Guo, S., Ren, J., Li, S., Ye, Y., Wang, Z., Huang,
J., Tang, Q., Sun, Y., Peng, X., Zhang, J., He, M., Zhu, Y., Xue, J., Wang, G., Wu, L.,
An, N., Wu, L., Ma, L., Zhang, W., Zhang, F., 2014. Producing more grain with lower
environmental costs. Nature 514, 486–489. https://doi.org/10.1038/nature13609.
Croke, J., Mockler, S., 2001. Gully initiation and road-to-stream linkage in a forested
catchment, southeastern Australia. Earth Surf. Process. Landforms 26, 205–217.
https://doi.org/10.1002/1096-9837(200102)26:2<205::AID-ESP168>3.0.CO;2-G.
Croke, J., Mockler, S., Fogarty, P., Takken, I., 2005. Sediment concentration changes in
runoff pathways from a forest road network and the resultant spatial pattern of
catchment connectivity. Geomorphology 68, 257–268. https://doi.org/10.1016/j.
geomorph.2004.11.020.
Darwish, T., Abu Bakar, K., 2015. Traffic density estimation in vehicular ad hoc networks:
a review. Ad Hoc Netw. 24, 337–351. https://doi.org/10.1016/j.adhoc.2014.09.007.
Dubé, K., Megahan, W., Mccalmon, M., 2004. Washington Road Surface Erosion Model
(WARSEM) Manual. Department of Natural Resources, 428 State of Washington, US.
Fu, B., Newham, L.T.H., Ramos-Scharrón, C.E., 2010. A review of surface erosion and
sediment delivery models for unsealed roads. Environ. Model. Softw. 25, 1–14.
Grossman, R.B., Reinsch, T.G., 2002. Bulk density and linear extensibility: core method.
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. 208–228.
Jin, S., Zhou, F., 2018. Zero growth of chemical fertilizer and pesticide use: china’s ob-
jectives, progress and challenges. J. Res. Ecol. 9 (1), 50–58. https://doi.org/10.5814/
j.issn.1674-1764x. 2018.01.006.
Ju, X., Gu, B., Wu, Y., Galloway, J.N., 2016. Reducing China’s fertilizer use by increasing
farm size. Global Environ. Change 41, 26–32. https://doi.org/10.1016/j.gloenvcha.
2016.08.005.
Khataza, R.R.B., Hailu, A., Doole, G.J., Kragt, M.E., Alene, A.D., 2019. Examining the
relationship between farm size and productive efficiency: a Bayesian directional
distance function approach. Agric. Econ. 50, 237–246. https://doi.org/10.1111/
agec.12480.
Lane, P.N.J., Sheridan, G.J., 2002. Impact of an unsealed forest road stream crossing:
Y. Li, et al. Soil & Tillage Research 204 (2020) 104689
11
water quality and sediment sources. Hydrol. Process. 16, 2599–2612. https://doi.
org/10.1002/hyp.1050.
Li, Y., Poesen, J., Yang, J.C., Fu, B., Zhang, J.H., 2003. Evaluating gully erosion using
137Cs and
210
Pb/
137
Cs ratio in a reservoir catchment. Soil Tillage Res. 69, 107–115.
https://doi.org/10.1016/S0167-1987(02)00132-0.
Li, Y., Tang, C., Huang, Z., Hussain, Z., Are, K.S., Abegunrin, T.P., Qin, Z., Guo, H., 2020.
Increase in farm size significantly accelerated stream channel erosion and associated
nutrient losses from an intensive agricultural watershed. Agric. Ecosyst. Environ. 295,
106900.
Manjunatha, A.V., Anik, A.R., Speelman, S., Nuppenau, E.A., 2013. Impact of land frag-
mentation, farm size, land ownership and crop diversity on profit and efficiency of
irrigated farms in India. Land Use Policy 31, 397–405. https://doi.org/10.1016/j.
landusepol.2012.08.005.
MOA (Ministry of Agriculture of China), 2014. The status of transferring rural household
responsibility land in 2013. Rural Business and Management 5, 42.
Nelson, D.W., Sommers, L.E., 1982. Total carbon, organic carbon and organic matter. In:
Page, A.L., Miller, R.H., Keeney, D.R. (Eds.), Methods of Soil Analysis. Part 2
Chemical and Microbiological Properties, pp. 539–579.
Oshunsanya, S.O., Li, Y., Yu, H., 2019. Vetiver grass hedgerows significantly reduce ni-
trogen and phosphorus losses from fertilized sloping lands. Sci. Total Environ. 661,
86–94. https://doi.org/10.1016/j.scitotenv.2019.01.129.
Pretty, J., Bharucha, Z.P., 2014. Sustainable intensification in agricultural systems – in-
vited Review. Ann. Bot. 114, 1571–1596. https://doi.org/10.1093/aob/mcu205.
Ramos-Scharrón, C.E., 2018. Land disturbance effects of roads in runoff and sediment
production on dry-tropical settings. Geoderma 310, 107–119. https://doi.org/10.
1016/j.geoderma.2017.08.035.
Ramos-Scharrón, C.E., LaFevor, M.C., 2018. Effects of forest roads on runoff initiation in
low-order ephemeral streams. Can. Water Resour. J. 54, 8613–8631. https://doi.org/
10.1029/2018WR023442.
Ramos-Scharrón, C.E., MacDonald, L.H., 2007. Runoff and suspended sediment yield from
an unpaved road segment. St. John, US Virgin Islands. Hydrol. Process. 21, 35–50.
https://doi.org/10.1002/hyp.6175.
Reid, L.M., Dunne, T., 1984. Sediment production from forest road surfaces. Can. Water
Resour. J. 20 (11), 1753–1761. https://doi.org/10.1029/wr020i011p01753.
Rizalihadi, A.M., Fatimah, B.E., Nazia, C.L., 2013. The influence of rill density on soil
erosion against USLE-soil erosion method. Proceedings of the 7th International
Conference on Asian and Pacific Coasts (APAC 2013) Bali, Indonesia, September 24-
26, 2013.
Sheng, Y., Chancellor, W., 2019. Exploring the relationship between farm size and pro-
ductivity: evidence from the Australian grains industry. Food Policy 84, 196–204.
https://doi.org/10.1016/j.foodpol.2018.03.012.
Sheridan, G.J., Noske, P.J., Whipp, R.K., Wijesinghe, N., 2006. The effect of truck traffic
and road water content on sediment delivery from unpaved forest roads. Hydrol.
Process. 20 (8), 1683–1699. https://doi.org/10.1002/hyp.5966.
Sommers, L.E., Nelson, D.W., 1972. Determination of total phosphorus in soils: A rapid
perchloric acid digestion procedure. Soil Sci. Soc. Amer. Proc. 36, 902–904. https://
doi.org/10.2136/sssaj1972.03615995003600060020.
Sosa-Pérez, G., MacDonald, L.H., 2017. Effects of closed roads, traffic, and road
decommissioning on infiltration and sediment production: a comparative study using
rainfall simulations. Catena 159, 93–105. https://doi.org/10.1016/j.catena.2017.08.
004.
Thomaz, E.L., Vestena, L.R., Ramos-Scharrón, C.E., 2014. The effects of unpaved roads on
suspended sediment concentration at varying spatial scales—A case study from
southern Brazil. Water Environ. J. 28, 547–555. https://doi.org/10.1111/wej.12070.
Tilman, D., Balzer, C., Hill, J., Befort, B.L., 2011. Global food demand and the sustainable
intensification of agriculture. Proc. Natl. Acad. Sci. U.S.A. 108 (50), 20260–20264.
https://doi.org/10.1073/pnas.1116437108.
Valin, H., Sands, R.D., van der Mensbrugghe, D., Nelson, G.C., Ahammad, H., Blanc, E.,
Bodirsky, B., Fujimori, S., Hasegawa, T., Havlik, P., Heyhoe, E., Kyle, P., Mason-
D’Croz, D., Paltsev, S., Rolinski, S., Tabeau, A., van Meijl, H., von Lampe, M.,
Willenbocke, D., 2014. The future of food demand: understanding differences in
global economic models. Agric. Econ. 45, 51–67. https://doi.org/10.1111/agec.
12089.
van Meerveld, H.J., Baird, E.J., Floyd, W.C., 2014. Controls on sediment production from
an unpaved resource road in a Pacific maritime watershed. Can. Water Resour. J. 50
(6), 4803–4820. https://doi.org/10.1002/2013WR014605.
Van Oost, K., Govers, G., Desmet, P., 2000. Evaluating the effects of changes in landscape
structure on soil erosion by water and tillage. Landscape Ecol. 15, 577–589. https://
doi.org/10.1023/a:1008198215674.
Wang, X., Yamauchi, F., Huang, J., Rozelle, S., 2019. What constrains mechanization in
Chinese agriculture? Role of farm size and fragmentation. China Econ. Review.
https://doi.org/10.1016/j.chieco.2018.09.002. (In press).
Wu, Y., Xi, X., Tang, X., Luo, D., Gu, B., Lam, S.K., Vitousek, P.M., Chen, D., 2018. Policy
distortions, farm size, and the overuse of agricultural chemicals in China. Proc. Natl.
Acad. Sci. U.S.A. 115 (27), 7010–7015. https://doi.org/10.1073/pnas.1806645115.
Yang, J., Huang, Z., Zhang, X., Reardon, T., 2013. The rapid rise of cross-regional agri-
cultural mechanization services in China. Am. J. Agric. Eco. 95, 1245–1251. https://
doi.org/10.1093/ajae/aat027.
Zhang, X., Li, Z., 2018. China’s Rural Development Road: Research Series on the Chinese
Dream and China’s Development Path. Springer Nature Singapore Pte Ltd. and Social
Sciences Academic Presshttps://doi.org/10.1007/978-981-10-5646-8_3.
Zhang, Y., Zhao, Y., Liu, B., Wang, Z., Zhang, S., 2019. Rill and gully erosion on unpaved
roads under heavy rainfall in agricultural watersheds on China’s Loess Plateau. Agric.
Ecosyst. Environ. 284, 106580. https://doi.org/10.1016/j.agee.2019.106580.
Ziegler, A.D., Sutherland, R.A., Giambelluca, T.W., 2000. Runoff generation and sediment
transport on unpaved roads, paths, and agricultural land surfaces in northern
Thailand. Earth Surf. Process. Landforms 25 (5), 519–534. https://doi.org/10.1002/
(SICI)1096-9837(200005) 25:5<519::AID-ESP80>3.0.CO;2-T.
Ziegler, A.D., Sutherland, R.A., Giambelluca, T.W., 2001. Interstorm surface preparation
and sediment detachment by vehicle traffic on unpaved mountain roads. Earth Surf.
Process. Landforms 26 (3), 235–250. https://doi.org/10.1002/1096-9837(200103)
26:3<235::AID -ESP171>3.0.CO;2-T.
Ziegler, A.D., Giambelluca, T.W., Sutherland, R.A., Nullet, M.A., Yarnasarn, S., Pinthong,
J., Preechapanya, P., Jaiaree, S., 2004. Toward understanding the cumulative im-
pacts of roads in upland agricultural watersheds of northern Thailand. Agric. Ecosyst.
Environ. 104, 145–158. https://doi.org/10.1016/j.agee.2004.01.012.
Y. Li, et al. Soil & Tillage Research 204 (2020) 104689
12
