Soil & Tillage Research 175 (2018) 130–138

Contents lists available at ScienceDirect

Soil & Tillage Research

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

Effect of soil tillage and vegetal cover on soil water infiltration MARK
a,⁎ b c
Wilk Sampaio de Almeida , Elói Panachuki , Paulo Tarso Sanches de Oliveira ,
Roniedison da Silva Menezesb, Teodorico Alves Sobrinhoc, Daniel Fonseca de Carvalhod
a
Federal Institute of Education, Science and Technology of Rondônia, Campus Colorado do Oeste, BR 435, Km 63, Colorado do Oeste, RO, 76993-000, Brazil
b
State University of Mato Grosso do Sul, Agronomy Department, CEP 79200-000 Aquidauana, MS, Brazil
c
Federal University of Mato Grosso do Sul, CxP. 549, Campo Grande, MS, 79070-900, Brazil
d
Federal Rural University of Rio de Janeiro, Engineering Department, BR 465, Km 7, CEP 23897-000 Seropédica, RJ, Brazil

A R T I C L E I N F O A B S T R A C T

Keywords: Land cover and land use change have altered hydrologic processes. However, there are still some uncertainties
Soil tillage on the magnitudes of these effects in tropical regions. Here, we evaluate the effect of soil tillage and land cover
Infiltration models on soil water infiltration, through measurement of this parameter in areas under bare soil, soybeans (conven-
Stable infiltration rate tional tillage and no-tillage) and pasture. We use a portable rainfall simulator with a constant intensity rain
(60 ± 1.715 mm h−1) in plots of 0.7 m2 in size. In total, 96 rains were applied, 24 in each treatment, after soil
preparation and in five stages after soybean sowing (every 20 days). We use the observed data to fit three
infiltration models, Kostiakov-Lewis, Horton and Philip, whose the quality of adjustment, was verified by cor-
relation coefficient (R), root mean square error (RMSE) and the Nash-Sutcliffe efficiency (NSE). Our results show
that the stable infiltration rate (SIR) in the no-tillage soybean system is greater than in the other systems by
40 days. Furthermore, SIR values in bare soil (BS) and soybean under conventional tillage (SCT) systems do not
differ (p < 0.05), except at 80 days after soybean sowing (SIR BS = 14.28 mm h−1 and SIR
SCT = 24.00 mm h−1). We find that Horton’s model adjusts the best to the different systems, with an R = 0.88,
0.86, 0.47 and 0.63, RMSE = 4.46, 4.87, 5.76 and 8.12, and NSE = 0.78, 0.74, 0.23 and 0.45, for the BS, SCT,
SNT and PA systems, respectively. However, the Kostiakov-Lewis and Philip models adjust is very close, in-
dependent of the presence of vegetal cover and soil tillage. We conclude that soil water infiltration is more
influenced by vegetal cover, depending on the type of land use, than by the soil tillage system.

1. Introduction limited vegetation cover on soil surface, porosity and infiltration de-
crease, causing intensification of the soil erosion process (Carvalho
Increasing global demand for food, fiber, and energy has favored the et al., 2015). Didone et al. (2014) also report problems of erosion in
conversion of native vegetation areas into agricultural lands, mainly in many areas under no tillage cultivation in southern Brazil because of
Brazil for soybeans production (Gardi et al., 2014; Merten et al., 2015) the limited presence of crop residues, removal of structures for surface
and pasture (Gibbs et al., 2010). Areas under native vegetation in the runoff control (terraces), down-slope cultivation, and soil compaction.
Cerrado (Brazilian Savannah) that have been transformed for agri- Soil water infiltration is influenced by several factors, such as tillage
culture use are approximately 50% of the original extension of this and vegetal cover, surface roughness, soil porosity and density, amount
biome (Klink and Machado, 2005; Gibbs et al., 2015). Currently, large of organic carbon, size and stability level of the aggregates, and soil
areas under pasture have been replaced by mechanized soybean and water content. These factors that influence infiltration, consequently
sugarcane monocultures because pasture areas in the Cerrado can be interfere on runoff. According Carlesso et al. (2011) land use and land
easily converted to cultivated fields to accommodate soybean expansion cover greatly affects infiltration, and has an important influence on
(Barona et al., 2010; Merten and Minella, 2013). raindrop interception. Increasing the percentage of plant canopy, re-
In these areas, conventional soil tillage predominates to increase soil sidue cover, soil surface roughness, and the crop evapotranspiration,
porosity and water infiltration and as a traditional practice of many will increase infiltration rate at the beginning of the rainfall event, thus
farmers. However, surface sealing tends to occur in just a few days after reducing runoff.
tillage, due to the direct impact of raindrops on the soil. Because the Cover crop residues are important to increase water infiltration into

⁎
Corresponding author.
E-mail addresses: [email protected], [email protected] (W.S. de Almeida), [email protected] (E. Panachuki), [email protected] (P.T.S. de Oliveira),
[email protected] (R. da Silva Menezes), [email protected] (T.A. Sobrinho), [email protected] (D.F. de Carvalho).

http://dx.doi.org/10.1016/j.still.2017.07.009
Received 6 February 2017; Received in revised form 9 June 2017; Accepted 15 July 2017
0167-1987/ © 2017 Published by Elsevier B.V.
W.S. de Almeida et al. Soil & Tillage Research 175 (2018) 130–138

soil and reduce surface runoff and erosion, and serve as a primary form
of organic matter input that enhances soil biological activity, conserves
moisture and moderates soil temperature (Derpsch et al., 2014). Despite
these benefits, the high profitability of soybeans and favorable pro-
duction prospects over the last decade in Brazil have led farmers to
adopt a poorly diversified crop sequence (in which cover crops are
rarely included in the system) that does not produce enough residue to
allow for permanent cover throughout the cropping season (Merten
et al., 2015). As in southern Brazil (Didone et al., 2014), the size of
agricultural machinery has increased in the Cerrado region to reduce
operations time in the farms.
Besides that, in Brazil, areas with conservational soil systems, such
as no-tillage and adequate management pastures have increased
(Panachuki et al., 2010). However, water infiltration under different
land uses is still poorly studied (Oliveira et al., 2015). Furthermore, it is
still necessary to better understand the influence of vegetal cover on soil
water infiltration and the fitness of infiltration models from measure-
ments data (Carvalho et al., 2015).
Our hypothesis is that the vegetal cover on soil surface in associa-
tion with cultivation practices changes distinctly the soil water in-
filtration on agricultural systems. Therefore, the objective of this study
is to evaluate the effect of soil tillage and vegetal cover on the in-
filtration processes in areas under bare soil, soybean (conventional
tillage and no-tillage), and pasture. We also assess the performance of
three infiltration models (Kostiakov-Lewis, Horton and Philip) to fit the
observed data of 96 rain events in plots of 0.7 m2 in size.

2. Materials and methods

2.1. Characterization of the study area

Fig. 1. The portable rainfall simulator InfiAsper during a rainfall application in bare soil
The study was carried out during the wet season (November 2013 system: 1 – motor, 2 – water application, 3 – blocking device, 4 – upper frame, 5 – runoff
through May 2014) in the municipality of Aquidauana, MS, Brazil collector, 6 – water pump, 7 – tank, 8 – excess water collector and 9 – electric panel
(20°27′ S, 55°40′ W; altitude of 191 m). According to the Brazilian control.
System of Soil Classification (Santos et al., 2013), the soil is a Argissolo
Vermelho Distrófico típico (Red Dystrophic Ultisol, USA) with a soil surface), mean drop diameter of 2.0 mm and pressure of 32 kPa. Galva-
texture (sand, silt and clay) of: 770 g kg−1, 110 g kg−1 and 120 g kg−1 nized steel rectangular sheets (runoff collector) of 1.0 × 0.7 m (0.7 m2)
(0–24 cm) and 610 g kg1, 140 g kg−1 and 250 g kg−1 (50–65 cm) that were kept fixed in the field until the end of the study surrounded the
(Schiavo et al., 2010). According to the Köppen climate classification, experimental plots, which received precipitation. Alves Sobrinho et al.
the climate is Aw (Tropical wet-dry climate), with annual temperatures (2008) report a detailed information of design, construction and operation
and precipitation ranging from 24 to 26 °C, and 1300 to 1600 mm, of this rainfall simulator, which can produce kinetic energy > 90% the
respectively (Alvares et al., 2013). The slope steepness of the experi- kinetic energy of corresponding natural rain.
mental area is approximately 0.03 m m−1 (Santos et al., 2014). Before rains simulations, soil wetting was performed, with which
standardization of all the treatment soil humidity was guaranteed (Cogo
2.2. Treatments, experimental design and soil analysis et al., 1984). To obtain soil moisture before of rain simulations, we
sampled soil in different soil profile: 0–10, 10–20 and 20–40 cm.
We studied the following soil covers and managements: bare soil The depth of water infiltration (DWI) was estimated by the differ-
(BS) after conventional tillage (in the direction of slope); soybeans ence between the artificial rain and the surface runoff (SR). The SR was
(Glycine max L.) cultivated in conventional tillage (SCT) also in the calculated each minute through the relation between the volume of
direction of slope; soybeans (Glycine max L.) cultivated in no-tillage water and the experimental plot area. The infiltration rate was calcu-
(SNT); and pasture (Brachiaria ruziziensis) (PA). The experimental de- lated by the relation between the DWI and the considered sampling
sign was completely randomized, arranged in subdivided plots (in time. The stable infiltration rate (SIR) of water in the soil was obtained
time), with four repetitions for treatment. In each of the plots, the when the SR remained constant. Time for the surface runoff was con-
system of culture was evaluated (BS, SCT, SNT and PA) and in the sub- sidered as that passed from the start of the applied precipitation until
plots, the six levels of soil vegetal cover were measured after 0, 20, 40, the moment at which the superficial water runoff began.
60, 80 and 100 days after soybean sowing (DAS).
Characterization of soil bulk density, porosity (macroporosity and
microporosity) and aggregates̕ stability (mean geometric diameter – 2.3. Infiltration rate estimation models
MGD and pondered mean diameter – PMD) in each experimental plot
were performed in samples collected in the layers 0–10, 10–20 and The infiltration values observed in the field were adjusted to the
20–40 cm in depth (Donagema et al., 2011). Soil resistance to pene- models of Kostiakov-Lewis (Kostiakov, 1932; Lewis, 1937) (Eq. (1)),
tration was measured by three readings in un-deformed samples using a Horton (Horton, 1933, 1939) (Eq. (2)) and Philip (Philip, 1957, 1969)
digital penetrometer (Serafim et al., 2008). (Eq. (3)) as a function of the corresponding time, as described by
We use a portable rainfall simulator (Alves Sobrinho et al., 2008) Assouline (2013). In the equations, i is the estimated instant infiltration
(Fig. 1) calibrated with a constant rain intensity of 60 ± 1.715 mm h−1 rate (mm h−1), io and if are the observed initial and the stable in-
(considering the height of the sprinkler beak, 2.30 m in relation to the soil filtration rate (mm h−1) and t is the infiltration time (min).

131
W.S. de Almeida et al. Soil & Tillage Research 175 (2018) 130–138

Table 1 Table 2
Soil physical properties for different soil profile depth. Soil moisture by treatment before the field tests in each of the soil layers.

Depth (cm) BS SCT SNT PA Overall mean Depth Humidity previous to the rain (cm3 cm−3)
Soil physical properties
BS SCT SNT PA Overall mean
Macroporosity (%)
0–10 17.15 b A 18.47 a A 14.16 b A 21.67 a A 17.86 A 0 DAS
10–20 10.93 a B 11.13 a B 13.96 a A 9.92 a B 11.48 B 0–10 24.62 a A 20.72 a A 17.30 b A 22.51 a A 21.29 A
20–40 8.55 a C 8.45 a B 9.03 a B 8.15 a B 8.54 C 10–20 22.56 a A 23.17 a A 17.99 b A 21.88 a A 21.39 A
20–40 21.25 a A 20.30 a A 17.77 a A 24.50 a A 20.95 A
Microporosity (%)
0–10 18.73 a B 16.85 a B 18.74 a A 17.77 a A 18.02 B 20 DAS
10–20 20.50 a A 19.14 a A 17.82 a A 19.26 a A 19.18 A 0–10 12.11 a A 12.44 a A 15.75 a A 12.89 a A 13.29 A
20–40 19.87 a A 19.86 a A 19.25 a A 17.96 a A 19.23 A 10–20 14.96 a A 15.74 a B 16.59 a A 19.03 a B 16.58 B
20–40 15.92 a A 17.78 a B 17.36 a A 17.15 a B 17.05 B
Total porosity (%)
0–10 35.88 b A 35.33 b A 32.90 b A 39.43 a A 35.89 A 40 DAS
10–20 31.43 a B 30.27 a B 31.78 a A 29.17 a B 30.66 B 0–10 11.82 a A 6.77 b A 11.63 a A 12.01 a A 10.57 A
20–40 28.41 a B 28.30 a B 28.28 a B 26.10 a C 27.77 C 10–20 13.27 a A 9.64 a B 12.51 a A 15.24 a A 12.66 B
20–40 15.16 a A 12.43 a B 15.25 a A 16.06 a A 14.72 C
Organic C (g kg−1)
0–10 42.25 a A 31.33 b A 29.80 b A 36.48 a A 34.96 A 60 DAS
10–20 19.57 b B 31.44 a A 19.46 b B 18.78 b B 22.31 B 0–10 8.54 a A 4.61 a A 6.20 a A 8.81 a A 7.04 A
20–40 17.53 a B 18.10 a B 16.51 a B 14.81 a B 16.74 C 10–20 10.31 a A 4.56 b A 6.47 b A 8.70 a A 7.51 A
20–40 13.01 b A 4.51 a A 7.02 a A 6.90 a A 7.86 A
Bulk density (Mg m−3)
0–10 1.43 b B 1.39 b B 1.54 a B 1.30 c C 1.42C 80 DAS
10–20 1.59 a A 1.55 a A 1.57 a B 1.60 a B 1.58 B 0–10 14.60 a A 8.09 b A 13.81 a A 11.84 a A 12.08 A
20–40 1.66 a A 1.61 a A 1.65 a A 1.69 a A 1.65 A 10–20 15.66 a A 10.53 b A 15.77 a A 16.25 a B 14.55 B
20–40 15.12 a A 10.95 b A 16.90 a A 17.73 a B 15.17 B
MGD (mm)
0–10 1.14 b A 1.15 b A 1.29 b A 2.55 a A 1.54 A 100 DAS
10–20 1.14 a A 1.50 a A 0.90 a A 1.30 a B 1.21 B 0–10 13.23 a A 13.08 a A 22.56 b A 16.30 a A 16.29 A
20–40 0.84 a A 0.78 a A 0.79 a A 0.95 a B 0.84 C 10–20 15.58 a A 12.63 a A 22.85 b A 20.29 b A 16.86 A
20–40 15.87 a A 13.97 a A 19.43 b A 18.17 b A 17.83 A
PMD (mm)
0–10 1.91 b A 2.07 b A 2.26 b A 3.44 a A 2.42 A
BS = bare soil; SCT = soybeans cultivated in conventional tillage; SNT = soybeans cul-
10–20 1.92 a A 2.48 a A 1.46 a B 2.07 a B 1.98 B
tivated in no-tillage; and PA = pasture. Mean values followed by letters (lowercase in the
20–40 1.15 a A 1.30 a B 1.07 a B 1.47 a B 1.25C
file and capital in the column) indicate differences according to the Scott Knott test
PR (MPa) (p < 0.05). DAS: days after soybean sowing.
0–10 0.32 b C 0.44 b B 0.99 a A 0.42 b B 0.54 B
10–20 0.91 a B 1.07 a A 0.77 a A 1.35 a A 1.03 A n
20–40 1.43 a A 1.00 a A 1.03 a A 1.37 a A 1.21 A ⎡ ∑ (Yobs − Yest )2 ⎤
⎢ ⎥
BS = bare soil; SCT = soybeans cultivated in conventional tillage; SNT = soybeans cul- NSE=1 − ⎢ i n= 1 ⎥
⎢ 2⎥
tivated in no-tillage; and PA = pasture. MGD = mean geometric diameter, ⎢∑ (Yobs − Yobs ) ⎥
PMD = pondered mean diameter and PR = penetration resistance obtained through a ⎣ i=1 ⎦ (5)
bench penetrometer. Mean values followed by letters (lowercase in the file and capital in
the column) indicate differences according to the Scott Knott test (p < 0.05). RMSE values of 0 indicate a perfect fit (Moriasi et al., 2007). The
NSE ranges between −∞ and 1.0 (1 inclusive), with NSE = 1 being the
Additionally, α, β, b and k are statistical parameters of the models and optimal value. Values between 0.0 and 1.0 are generally viewed as
are estimated by nonlinear regression using the Gauss-Newton method acceptable levels of performance, whereas values < 0.0 indicates that
using the software DataFit version 9.1.32 (DataFit, 2017), with which the mean observed value is a better predictor than the simulated value,
the standard deviation of the squares̕ sum, associated to the infiltration which indicates unacceptable performance (Nash and Sutcliffe, 1970;
rate values observed in the field, is minimized. Moriasi et al., 2007).
i = if + α k t α −1 (1)
2.4. Statistical analysis
i = if + (io − if ) e−β t (2)
All data analysis and plotting were performed using the statistical and
1 −0,5 computing environment R version 3.4.0 (R Core Team, 2017). The variables
i=b+ kt
2 (3)
stable infiltration rate and physical soil properties were submitted to a
To verify the quality of the models’ adjustments a correlation ana- hierarchical clustering algorithm as an exploratory analysis. We used the
lysis along with a residue analysis of the models was performed using Scott-Knott package (SNK) to perform it (available in https://cran.r-project.
the DataFit version 9.1.32. Therefore, we used the correlation coeffi- org/web/packages/ScottKnott/index.html). The Scott-Knott package is a
cient (R), root mean square error (RMSE) (Moriasi et al., 2007) (Eq. (4)) code wrote by Jelihovschi et al. (2014) to run a Scott-Knott algorithm (Scott
and Nash-Sutcliffe efficiency (NSE) (Nash and Sutcliffe, 1970). (Eq. and Knott, 1974) in the environment R. The SNK has two main functions:
(5)). In these equations, Yobs represents the observed data, Yest the es- SK and SK.nest. The SK function is used for clustering the treatment means
timated data through the models, n the number of observations and of a main factor. The SK.nest function, in turn, is used for clustering
Yobs the arithmetic mean of the observations, as showed: treatment means relative to interactions among factors, that is whenever the
treatment means belong to a factor nested in others. The Scott-Knott algo-
n
⎡ ⎤ rithm divides the treatments means in groups without overlapping, its result
⎢ ∑ (Yobs − Yest )2 ⎥
also uses a probabilistic approach aimed at finding the groups: the SK al-
RMSE=⎢ i=1 ⎥
gorithm takes the maximum between group sum of squares, which is used
⎢ n ⎥
⎢ ⎥ in a likelihood ratio test with an asymptotic χ2 distribution (Jelihovschi
⎢
⎣ ⎥
⎦ (4) et al., 2014).

132
W.S. de Almeida et al. Soil & Tillage Research 175 (2018) 130–138

The analysis of variance (Anova) (F test, p < 0.05) was performed Table 3
considering the completely randomized design. The normality and Time necessary for runoff to begin in the systems at six evaluation stages.
homogeneity of residuals were verified, respectively by a Shapiro-Wilk
Management Time necessary for runoff to begin (min) in the systems (BS, SNC,
test (Shapiro and Wilk, 1965) (p < 0.05; p-value = 0.41) and a Bar- systems SNT and PA) at six evaluation stages
tlett test (Bartlett, 1937) (p < 0.05; p-value = 9.87 e−07). For the
significant effects of the treatments, the means were compared con- 0 DAS 20 DAS 40 DAS 60 DAS 80 DAS 100 DAS
sidering the significance level of 5% (p < 0.05), without any data
BS 9.32 b 1.55 b 0.79 b 1.34 c 1.23 b 0.46 b
transformations. SCT 6.67 b 1.34 b 1.46 b 1.39 c 1.89 b 0.54 b
SNT 9.39 b 4.78 b 4.92 b 7.36 b 4.77 b 4.00 b
3. Results and discussion PA 13.90 a 16.45 a 21.01 a 38.07 a 40.20 a 15.00 a

Time necessary for surface runoff to begin (min) in the systems BS = bare soil;
3.1. Soil properties
SCT = soybeans cultivated in conventional tillage; SNT = soybeans cultivated in no-til-
lage; and PA = pasture, during the six evaluation stages (0, 20, 40, 60, 80 and 100 days
The greatest macroporosity is verified for the superficial layer after soybean sowing. Different letters in the column indicate difference according to
(0–10 cm) in the pasture (PA), probably associated with the level of Scott Knott (p < 0.05).
organic matter derived from plant roots as has been reported by Prado
et al. (2014). We also find greater values of mean geometric diameter
(MGD) and pondered mean diameter (PMD) of aggregates compared to Table 4
the other systems, with PA being the most resistant to disaggregation, Stable infiltration rate (mm h−1) in the systems (BS, SNC, SNT and PA) at six evaluation
stages.
favoring aggregate stability. We note no difference between the BS, SCT
and SNT systems in the PMD (Table 1). However, the effect of aggregate Management Stable infiltration rate (mm h−1) in the systems (BS, SNC, SNT and
stability on soil structure was confirmed, mainly in the systems con- systems PA) at six evaluation stages
sidered as conservationists (SNT and PA). In these systems, the organic
matter provides greater aggregate stability, as has been observed in 0 DAS 20 DAS 40 DAS 60 DAS 80 DAS 100 DAS

previous studies (Panachuki et al., 2006; Prado et al., 2014). BS 26.71 b 16.14 b 14.29 b 22.93 b 14.29 c 5.71 c
The greatest macroporosity and smallest density values of the soil SCT 23.43 b 15.29 b 20.14 b 19.07 b 24.00 b 13.42 c
were detected in the superficial layer. The aggregate stability (MGD and SNT 46.39 a 33.36 a 39.36 a 52.79 a 54.21 a 47.82 a
PMD) tends to decrease with increasing depth, although differences PA 53.71 a 34.79 a 33.79 a 30.33 b 31.86 b 29.52 b

were observed for the MGD in the PA and for the PMD in the SNT and
Stable infiltration rate (mm h−1) in the systems BS = bare soil; SCT = soybeans culti-
PA. These differences among layers may be due to the lower amount of vated in conventional tillage; SNT = soybeans cultivated in no-tillage; and PA = pasture,
organic matter at greater depths (Prado et al., 2014), associated with during the six evaluation stages (0, 20, 40, 60, 80 and 100 days after soybean sowing.
great density and clay content in the sub-superficial layers (10–20 cm). (Scott Knott, p < 0.05).
The MGD (35.71, 47.44, 63.30 and 168.42%) and PMD (66.10,
59.23, 111.21 and 134.0%) values of the systems BS, SCT, SNT, and PA,
respectively, observed in the deepest soil layer (20–40 cm) were lower differences in this variable registered between the superficial and the
compared to those of the superficial layer. Our results indicate that sub-superficial layers, which may be associated with the soil water re-
aggregates stability decreases with soil depth, as verified by Souza et al. tention capacity.
(2005). These results are also related to the presence of vegetal cover on
the soil surface (Shukla et al., 2003), mainly in the SNT and PA systems. 3.2. Stable infiltration rate of soil water
The concentration of roots in the superficial layer contributes to ag-
gregates stability (Prado et al., 2014) through the incorporation of or- We find the highest surface runoff values in the pasture areas (PA),
ganic matter into the soil. independent of the stage of evaluation, probably due to the high
Bulk density values greater than or near to 1.45 Mg m−3 are con- amount of biomass (Guzha, 2004; Almeida et al., 2016). In the other
sidered limiting for most culture development (Corsini and Ferraudo, systems, only at 60 DAS, the surface runoff in the no-tillage treatment is
1999; Baquero et al., 2012). According to this fact, the superficial layer higher than in the BS and SCT. Similarly, increases in the required time
density in the BS and SNT systems is not restrictive to root growth. for runoff to begin are verified in the SCT and SNT 40 DAS, in asso-
However, in the other layers, values were greater than the critical limit, ciation with soybean development (Table 3). By intercepting raindrops,
and therefore, the soil may be considered compacted. Soil layers with vegetal cover diminishes the runoff velocity, which affects infiltration
high-density levels and low macroporosity intensify soil erosion pro- and time necessary to start the surface runoff (Panachuki et al., 2011).
cesses and limit the radicular growth of the plants (Corsini and The stable infiltration rate (SIR) in the SNT and PA systems did not
Ferraudo, 1999). differ among the stages 0, 20 and 40 DAS. In these two treatments, the
In the no-tillage system, high soil penetration resistance (PR) is SIR is greater than in the other treatments. From 60 DAS on, the no-
verified in comparison to the others, probably due to the soil being tillage SIR is significantly greater (p < 0.05) than in the other treat-
undisturbed. In general, this variable increased with soil depth, except ments. On the other hand, between the two conventional tillage sys-
in the bare soil, where no difference is found between the 0–20 and tems, with disturbed soil, (BS and SCT), the SIR differed at 80 DAS
20–40 cm layers. The latter might be the result of soil preparation, when the soybean proportionated greater vegetal cover in the SCT,
which promotes the breakdown of pore continuity, aggregates pulver- therefore, favoring water infiltration in the soil (Table 4). At this stage,
ization and superficial sealing (Santos et al., 2014). the highest percentage of soybean cover in the SCT tend to promote a
Wetting of the soil by dripping was made unnecessary by natural high soil water infiltration in this system, such as reported by Almeida
precipitation events that occurred before the phases that corresponded et al. (2016).
to the 20, 40, 80 and 100 DAS. The soil moisture values that preceded In agricultural systems under conventional tillage, water infiltration
the tests are similar among the layers inside each treatment (Table 2). is influenced by land preparation and its interaction with the physical
Only in the SCT (20 and 40 DAS) and the PA (20 and 80 DAS), properties of the soil and surface vegetal cover. In the initial and final

133
W.S. de Almeida et al. Soil & Tillage Research 175 (2018) 130–138

Table 5
Parameters and statistical indexes of the evaluated infiltration models tested for the bare soil system.

Parameters and indexes 0 DAS 20 DAS 40 DAS 60 DAS 80 DAS 100 DAS

Kostiakov-Lewis’s Model
α 0.3123 0.3145 0.0019 0.0016 0.0674 0.0018
k 120.4535 152.6062 22,724.6639 22,151.5614 725.5983 29,341.0492
R 0.84 0.84 0.83 0.84 0.86 0.96
RMSE 4.82 5.85 5.62 4.30 5.32 2.89
NSE 0.70 0.71 0.69 0.70 0.74 0.92

Horton’s Model
io 59.14 58.28 60.00 60.85 60.85 61.71
if 26.71 16.14 14.29 22.93 14.29 5.71
β 0.1027 0.1120 0.3007 0.3565 0.2130 0.3192
R 0.91 0.88 0.89 0.81 0.86 0.93
RMSE 3.62 5.19 4.31 4.56 5.32 3.76
NSE 0.83 0.77 0.79 0.66 0.74 0.87

Philip’s Model
b 20.7816 9.4501 5.2077 16.6720 5.7056 −4.7966
k 90.4996 111.0329 94.5660 70.3416 105.7299 111.7107
R 0.88 0.87 0.85 0.77 0.86 0.92
RMSE 4.17 5.32 4.95 5.02 5.32 4.00
NSE 0.78 0.76 0.73 0.59 0.74 0.85

α, k, β and b are statistical parameters of the models, obtained in software DataFit; io and if are the observed initial and the stable infiltration rate (mm h−1); R = correlation coefficient;
RMSE = root mean square error; NSE = Nash-Sutcliffe efficiency and DAS = days after soybean sowing.

Table 6
Parameters and statistical indexes of the evaluated infiltration models tested for the soil conventional tillage soybean system.

Parameters and indexes 0 DAS 20 DAS 40 DAS 60 DAS 80 DAS 100 DAS

Kostiakov-Lewis’s Model
α 0.2819 0.1067 0.2974 0.0021 0.2813 0.0064
k 144.9174 421.5940 144.3197 19,154.5971 150.6677 7945.9512
R 0.74 0.92 0.79 0.86 0.85 0.93
RMSE 7.27 3.62 6.24 4.37 5.00 3.79
NSE 0.54 0.84 0.62 0.74 0.73 0.86

Horton’s Model
io 59.14 58.28 60.00 60.85 60.85 61.71
if 23.43 15.29 20.14 19.07 24.00 13.42
β 0.1198 0.2080 0.1380 0.3227 0.1196 0.2339
R 0.78 0.91 0.77 0.84 0.91 0.94
RMSE 6.66 3.83 6.49 4.66 4.00 3.58
NSE 0.61 0.82 0.59 0.71 0.83 0.88

Philip’s Model
b 17.0280 7.9324 14.8839 12.1195 17.3293 3.9802
k 95.8270 96.4689 94.1463 80.6283 99.5429 109.1377
R 0.77 0.91 0.80 0.80 0.89 0.92
RMSE 6.90 3.68 6.10 5.16 4.38 3.99
NSE 0.59 0.83 0.64 0.64 0.79 0.85

α, k, β and b are statistical parameters of the models, obtained in software DataFit; io and if are the observed initial and the stable infiltration rate (mm h−1); R = correlation coefficient;
RMSE = root mean square error; NSE = Nash-Sutcliffe efficiency and DAS = days after soybean sowing.

stages of culture development, soil susceptibility to the erosive action of puddle formation, which simultaneously favors infiltration. These re-
rain is high. In the present study, this effect was minimized by the sults confirm the interference of the combination of stage and culture
presence of soybean vegetal cover from 40 DAS on when the stable system on variations in soil water infiltration. Soil tillage and vegetal
infiltration rate increased. When water reaches the soil at lower velo- cover, mainly at the beginning and end, respectively, of the culture
city, it tends to have a lower runoff and even stay for higher periods of cycle, may influence these variations the most, as observed by Carvalho
time in soil surface depressions, which favor infiltration (Cogo et al., et al. (2015).
1984).
Up to 20 DAS, the pasture mean SIR was 15.85% greater than that 3.3. Performance of infiltration models
verified in the no-tillage soybean system. However, from 40 DAS on, the
pasture SIR decreased. This decrease could be due to the higher soybean In general, our findings indicate that the soil water infiltration es-
vegetal cover, as well as to the greater water content stored in the timated values by Kostiakov-Lewis, Horton and Philip are acceptable,
pasture and cultural residuals of corn in the SNT. This phenomenon according R, RMSE and NSE (Tables 5–8). We also note that the Horton
might have functioned as a physical barrier to water runoff through model had the best adjustment for the conditions at which the assays

134
W.S. de Almeida et al. Soil & Tillage Research 175 (2018) 130–138

Table 7 of the evaluated models, an adjustment pattern similar to the observed

Parameters and statistical indexes of the evaluated infiltration models in the no-tillage data in the field was verified (Tables 5–8).
soybean system.
The smallest difference between minimum and maximum RMSE was
Parameters and 0 DAS 20 DAS 40 DAS 60 DAS 80 DAS 100 DAS obtained for the Horton model excepting for the bare soil, when the Philip
indexes model had a major sensibility to estimated soil water infiltration, and
pasture systems, which to Kostiakov-Lewis presents minor difference. In the
Kostiakov-Lewis’s Model
other hand, the greatest difference for RMSE values occurred in the pasture
α 0.2261 0.4162 0.5560 0.2930 0.4610 0.5895
k 63.8093 72.8466 43.2740 28.2076 15.8721 25.3329 system, for Horton, Kostiakov-Lewis and Philip, by stage are obtained for
R 0.41 0.56 0.51 0.27 0.52 0.43 the Phillip and Kostiakov-Lewis models, primarily in the systems with dis-
RMSE 6.16 8.66 7.37 5.30 2.26 5.53 turbed soils. These low RMSE values demonstrate the reliability of these
NSE 0.17 0.32 0.26 0.07 0.27 0.18 models for predicting the water infiltration rate in the soil, mostly under
Horton’s Model conventional tillage. On the other hand, the large RMSE values, in SNT and
io 59.14 58.28 60.00 60.85 60.85 61.71 PA, demonstrate the negative effect of vegetation cover on estimation of soil
if 46.39 33.36 39.36 52.79 54.21 47.82
water infiltration (Shao and Baumgartl, 2016).
β 0.1359 0.0751 0.0486 0.1509 0.0727 0.0459
R 0.44 0.62 0.55 0.25 0.52 0.41 Through analysis of the NSE, we verify that for soil disturbed con-
RMSE 6.07 8.20 7.16 5.33 2.25 5.57 dition, the three models present major values compared to undisturbed
NSE 0.19 0.39 0.30 0.06 0.27 0.17 soil systems, and however the NSE values are similar between the
Philip’s Model models in all systems evaluated. The same comportment occurs with
b 43.8537 30.1597 39.6903 51.8903 53.8023 48.6588 the R and RMSE. These results corroborate the influence of vegetation
k 33.9720 71.2559 50.7497 17.4783 16.0490 29.8813 on soil water infiltration. Therefore, vegetation affect infiltration by
R 0.43 0.58 0.51 0.27 0.52 0.42
intercepting and storing rainfall, protecting the soil surface against the
RMSE 6.10 8.52 7.40 5.30 2.25 5.55
NSE 0.19 0.34 0.26 0.07 0.27 0.18 impact of raindrops, and changing the soil properties (Panachuki et al.,
2006; Santos et al., 2014; Carvalho et al., 2015; Shao and Baumgartl,
α, k, β and b are statistical parameters of the models, obtained in software DataFit; io and 2016) and thus the infiltration characteristic (Puigdefábregas, 2005;
if are the observed initial and the stable infiltration rate (mm h−1); R = correlation Shao and Baumgartl, 2016). Consequently, the simulation accuracy
coefficient; RMSE = root mean square error; NSE = Nash-Sutcliffe efficiency and
could be affected by the presence of vegetation probably because of the
DAS = days after soybean sowing.
fact that these infiltration models were primarily developed for bare
soils (Shao and Baumgartl, 2016).
Table 8 The estimation of statistical parameters made it possible to compare
Parameters and statistical indexes of the evaluated infiltration models in the pasture. the models. The Horton model presents a trend to fit better to observed
infiltration data, but the adjustment of three models was acceptable
Parameters and 0 DAS 20 DAS 40 DAS 60 DAS 80 DAS 100 DAS
indexes according a correlation and residual analysis, as reported by R, RMSE
and NSE. The adjustment of the Kostiakov-Lewis and Philip models are
Kostiakov-Lewis’s Model similarly throughout the culture cycle in the different treatments.
α 0.3741 0.5662 0.4927 0.5667 0.5519 0.4928
Therefore, for each culture system, the plant cover and soil tillage ef-
k 16.8834 56.5057 70.9909 72.3657 66.2238 85.3436
R 0.23 0.41 0.46 0.67 0.77 0.68 fects on soil water infiltration are considered, as well as estimation
RMSE 5.23 13.63 13.23 8.74 5.71 8.92 through the Kostiakov-Lewis, Horton and Philip models. Shao and
NSE 0.05 0.17 0.22 0.45 0.60 0.46 Baumgartl (2016) verified the capacity of three modified models
Horton’s Model (Holtan, Green-Ampt and Horton) on small field plots using sequences
io 59.14 58.28 60.00 60.85 60.85 61.71 of rainfall events produced by a field rainfall simulator. They concluded
if 53.71 34.79 33.79 30.33 31.86 29.52 that in all validation data sets, R2 values of the three models had sig-
β 0.0885 0.0409 0.0540 0.0411 0.0453 0.0547
nificant (p < 0.05) and negative correlations with vegetation cover
R 0.26 0.49 0.56 0.80 0.90 0.80
RMSE 5.20 13.04 12.43 6.97 3.85 7.25 and root content, differently of our results. In our study, R2 (data not
NSE 0.07 0.24 0.31 0.65 0.82 0.64 shown) and RMSE had a positive correlation with vegetation.
Philip’s Model
Fig. 2 shows an example of the fitted models to the observed soil
b 52.9028 34.9776 31.3528 30.5582 31.7391 26.7781 water infiltration in 80 DAS. The models present similar adjustment for
k 14.9841 69.9992 82.6211 89.8772 79.9954 98.3335 the different systems, but the Horton model shows best adjustment for
R 0.24 0.40 0.47 0.66 0.76 0.69 pasture (Fig. 2D). Several authors also have found best performance of
RMSE 5.22 13.68 13.16 8.86 5.79 8.78
the Horton model under this condition (Carvalho et al., 2015;
NSE 0.06 0.16 0.22 0.43 0.58 0.48
Panachuki et al., 2006; Tomasini et al., 2010; Santos et al., 2014).
α, k, β and b are statistical parameters of the models, obtained in software DataFit; io and
if are the observed initial and the stable infiltration rate (mm h−1); R = correlation 4. Conclusions
coefficient; RMSE = root mean square error; NSE = Nash-Sutcliffe efficiency and
DAS = days after soybean sowing.
In this study, we evaluate the effect of soil tillage and land cover on
water soil infiltration through measurement of this parameter in areas
under bare soil, soybeans (conventional tillage and no-tillage) and
were conducted, with R values of 0.88, 0.86, 0.47 and 0.63 for the BS, pasture. We perform 96 field tests using a rainfall simulator with a
SCT, SNT and PA systems, respectively. The RMSE values for Horton’s constant intensity rain (60 ± 1.715 mm h−1) in plots of 0.7 m2 in size.
model are 4.46, 4.87, 5.76 and 8.12; for NSE the values are 0.78, 0.74, We also use the observed data to fit three infiltration models
0.23 and 0.45. Some of the tested models underestimated the infiltra- (Kostiakov-Lewis, Horton and Philip).
tion rates at the beginning of rain application (minors times), whereas Our results indicate that water infiltration into the soil is more in-
others overestimated it at the final test (the approximately 50 min fi- fluenced by the vegetal cover, depending on land use type, than by the
nals), especially the Kostiakov-Lewis and Philip models. However, in all soil tillage system. We find that initially, soil tillage causes greater

135
W.S. de Almeida et al. Soil & Tillage Research 175 (2018) 130–138

Fig. 2. Soil water infiltration observed and estimated (mm h−1) in time (min) by Kostiakov-Lewis, Horton and Phillip models in the systems bare soil (A), soybean cultivated in
conventional tillage (B), soybean cultivated in no-tillage (C) and pasture (D), at stage 80 DAS (days after soybean sowing).

136
W.S. de Almeida et al. Soil & Tillage Research 175 (2018) 130–138

Fig. 2. (continued)

changes on water infiltration than the effect of the vegetal cover, as was estimating the rate of soil water infiltration, mainly in conventional
verified under bare soil and soybeans with the conventional tillage management systems, as those of no-tillage and established pasture.
system. Furthermore, water infiltration tends to be smaller in areas However, the Kostiakov-Lewis and Philip model are acceptable to de-
under bare soil than in those under soil conservation systems. scribe the soil water infiltration and are similar to the Horton model.
The Horton infiltration model tends to be the most adequate for

137
W.S. de Almeida et al. Soil & Tillage Research 175 (2018) 130–138

Acknowledgements Trans. Am. Geophys. Union 20, 693–711.

Jelihovschi, E.G., Faria, J.C., Allaman, I.B., 2014. ScottKnott: a package for performing
the Scott-Knott clustering algorithm in R. Tend. Mat. Apl. Comput. 15, 3–17.
The authors thank the Brazilian foundation, Conselho Nacional de Klink, C.A., Machado, R.B., 2005. Conservation of the Brazilian Cerrado. Conserv. Biol.
Desenvolvimento Científico e Tecnológico – CNPq whose financial 19, 707–713.
Kostiakov, A.N., 1932. On the dynamics of the coefficient of water percolation in soils and
support made the development of the present study possible and the on the necessity of studying it from a dynamic point of view for purposes of ameli-
first author's scholarship (grant number 133053/2013-0); to Rosalba O. oration. Transactions of the Sixth Commission of the International Society of Soil
Fors, for English grammar editing; and we also are grateful to the Science, Part A (Moscow) 17–21.
Lewis, M.R., 1937. The rate of infiltration of water in irrigation practice. Trans. Am.
anonymous reviewers and the editorial committee for all suggestions Geophys. Union 18, 361–368.
that improved the quality of this paper. Merten, G.H., Minella, J.P.G., 2013. The expansion of Brazilian agriculture: soil erosion
scenarios. Int. Soil Water Conserv. Res. 1, 37–48.
Merten, G.H., Araújo, A.G., Biscaia, R.C.M., Barbosa, G.M.C., Conte, O., 2015. No-till
References
surface runoff and soil losses in southern Brazil. Soil Tillage Res. 152, 85–93.
Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Bingner, R.L., Harmel, R.D., Veith, T.L.,
Almeida, W.S., Carvalho, D.F., Panachuki, E., Valim, W.C., Rodrigues, S.A., Varella, 2007. Model evaluation guidelines for systematic quantification of accuracy in wa-
C.A.A., 2016. Erosão hídrica em diferentes sistemas de cultivo e níveis de cobertura tershed simulations. Trans. ASABE 50, 885–900.
do solo. Pesq. Agrop. Brasileira 51, 1110–1119. Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models: part 1.
Alvares, C.A., Stape, J.L., Sentelhas, P.C., Gonçalves, J.L.M., Sparovek, G., 2013. Köppen’s A discussion of principles. J. Hydrol. 10, 282–290.
climate classification map for Brazil. Meteor. Zeitsch. 22, 711–728. Oliveira, P.T.S., Wendland, E., Nearing, M.A., Scott, R.L., Rosolem, R., Rocha, H.R., 2015.
Alves Sobrinho, T., Macpherson, H.G., Gómez, J.A., 2008. A portable integrated rainfall The water balance components of undisturbed tropical woodlands in the Brazilian
and overland flow simulator. Soil Use Man 24, 163–170. Cerrado. Hydrol. Earth Syst. Sci. 19, 2899–2910.
Assouline, S., 2013. Infiltration into soils: conceptual approaches and solutions. Water Panachuki, E., Alves Sobrinho, T., Vitorino, A.C.T., Carvalho, D.F., Urchei, M.A., 2006.
Resour. Res. 49, 1–18. Avaliação da infiltração de água no solo, em sistema de integração agricultura-
Baquero, J.E., Ralisch, R., Medina, C.C., Tavares Filho, J., Guimarães, M.F., 2012. pecuária, com uso de infiltrômetro de aspersão portátil. Act. Scient. Agron. 28,
Propriedades físicas do solo e crescimento radicular de cana-de-açúcar em um 129–137.
Latossolo Vermelho. Rev. Bras. Ciênc. Solo 36, 63–70. Panachuki, E., Bertol, I., Alves Sobrinho, T., Vitorino, A.C.T., Souza, C.M.A., Urchei, M.A.,
Barona, E., Ramankutty, N., Hyman, G., Coomes, O.T., 2010. The role of pasture and 2010. Rugosidade da superfície do solo sob diferentes sistemas de manejo e influ-
soybean in deforestation of the Brazilian Amazon. Environ. Res. Lett. 5, 1–9. enciada por chuva artificial. Rev. Bras. Ciênc. Solo 34, 443–451.
Bartlett, M.S., 1937. Properties of sufficiency and statistical tests. Proc. R. Soc. 160, Panachuki, E., Bertol, I., Alves Sobrinho, T., Oliveira, P.T.S., Rodrigues, D.B.B., 2011.
268–282. Perdas de solo e de água e infiltração de água em Latossolo Vermelho sob sistemas de
Carlesso, R., Spohr, R.B., Eltz, F.L.F., Flores, C.H., 2011. Runoff estimation in southern manejo. Rev. Bras. Ciênc. Solo 35, 1777–1785.
Brazil based on Smith’s modified model and the Curve Number method. Agric. Water Philip, J.R., 1957. The theory of infiltration Sorptivity and algebraic infiltration equa-
Manag. 98, 1020–1026. tions. Soil Sci. 84, 257–264.
Carvalho, D.F., Eduardo, E.N., Almeida, W.S., Santos, L.A.F., Alves Sobrinho, T., 2015. Philip, J.R., 1969. Theory of infiltration. Adv. Hydrosci. 5, 215–296.
Water erosion and soil water infiltration in different stages of corn development and Prado, E.A.F., Vitorino, A.C.T., Oliveira, W.H., Espindola, D.L.P., Arantes, H.P., 2014.
tillage systems. Rev. Bras. de Eng. Agric. e Ambiental 19, 1076–1082. Índice de dispersão de agregados de um Latossolo Vermelho distroférrico cultivado
Cogo, N.P., Moldenhauer, W.C., Foster, G.R., 1984. Soil loss reductions from conservation com cana sob aplicação de vinhaça. Sem.: Cên. Agr. 35, 2347–2356.
tillage practices. Soil Sci. Soc. Am. J. 48, 368–373. Puigdefábregas, J., 2005. The role of vegetation patterns in structuring runoff and sedi-
Corsini, P.C., Ferraudo, A.S., 1999. Efeitos de sistemas de cultivo na densidade e mac- ment fluxes in drylands. Earth Surf. Processes Landforms 30, 133–147.
roporosidade do solo e no desenvolvimento radicular do milho em Latossolo Roxo. R Core Team, 2017. R: A Language and Environment for Statistical Computing. R
Pesq. Agrop. Brasileira 34, 289–298. Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/.
Derpsch, R., Franzluebbers, A.J., Duiker, S.W., Reicosky, D.C., Koeller, K., Friedrich, T., Santos, H.G., Jacomine, P.K.T., Anjos, L.H.C., Oliveira, V.A., Lumbreras, J.F., Coelho,
Sturny, W.G., Sá, J.C.M., Weiss, K., 2014. Why do we need to standardize no-tillage M.R., Almeida, J.A., Cunha, T.J.F., Oliveira, J.B. (Eds.), 2013. Brazilian System of Soil
research? Soil Tillage Res. 137, 16–22. Classification (In portuguese: Sistema Brasileiro de Classificação de Solos), 3rd ed.
Didone, E.J., Minella, J.P.G., Reichert, J.M., Merten, G.H., Dalbianco, L., De Barros, Embrapa Solos, Brasília, Brazil pp. 353.
C.P.P., Ramon, R., 2014. Impact of no-tillage agricultural systems on sediment yield Santos, M.A.N., Panachuki, E., Alves Sobrinho, T., Oliveira, P.T.S., Rodrigues, D.B.B.,
in two large catchment in Southern Brazil. J. Soil Sediment 14, 1287–1297. http:// 2014. Water infiltration in an Ultisol after cultivation of common bean. Rev. Bras.
dx.doi.org/10.1007/S11368-013-0844-6. Ciênc. Solo 38, 1143–1152.
Donagema, G.K., Campos, D.V.B., Calderano, S.B., Teixeira, W.G., Viana, J.H.M., 2011. Schiavo, J.A., Pereira, M.G., Miranda, L.P.M., Neto, A.H.D., Fontana, A., 2010.
Manual de Métodos de Análise de Solos. Empresa Brasileira de Pesquisa Caracterização e classificação de solos desenvolvidos de arenitos da formação
Agropecuária. Centro Nacional de Pesquisa em Solos, Rio de Janeiro. Embrapa Solos. Aquidauana-MS. Rev. Bras. Ciênc. Solo 4, 881–889.
230 p. Disponível em https://www.embrapa.br/busca-de-publicacoes/-/publicacao/ Scott, A.J., Knott, M., 1974. A cluster analysis method for grouping means in the analysis
990374/manual-de-metodos-de-analise-de-solo. of variance. Biometrics 30, 507–512.
Gardi, C., Angelini, M., Barceló, S., Comerma, J., Cruz Gaistardo, C., Encina Rojas, A., Serafim, M.E., Vitorino, A.C.T., Peixoto, P.P.P., Souza, C.M.A., Carvalho, D.F., 2008.
Jones, A., Krasilnikov, P., Mendonça S.B, M.L., Montanarella, L., Muñiz Urgate, O., Intervalo hídrico ótimo em um Latossolo Vermelho distroférrico sob diferentes sis-
Schad, P., Vara Rodríguez, M.I., Vargas, R. (Eds.), 2014. Atlas de Suelos de América temas de produção. Eng. Agric. 28, 654–665.
Latina y el Caribe. Comisión Europea – Oficina de Publicaciones de La Unión Shao, Q., Baumgartl, T., 2016. Field evaluation of three modified infiltration models for
Europea, L-2995 Luxembourg pp. 176. the simulation of rainfall sequences. Soil Sci. 181, 45–56.
Gibbs, H.K., Ruesch, A.S., Achard, F., Clayton, M.K., Holmgren, P., Ramankutty, N., Foley, Shapiro, S.S., Wilk, M.B., 1965. An analysis of variance test for normality (complete
J.A., 2010. Tropical forests were the primary sources of new agricultural land in the samples). Biometrika 52, 591–611.
1980 and 1990. Proc. Natl. Acad. Sci. U. S. A. 107, 16732–16737. Shukla, M.K., Lal, R., Owens, L.B., Unkefer, P., 2003. Land use management impacts on
Gibbs, H.K., Rausch, L., Munger, J., Schelly, I., Morton, D.C., Noojipady, P., Soares-Filho, structure and infiltration characteristics of soils in the north Appalachian region of
B., Barreto, P., Micol, L., Walker, N.F., 2015. Brazil’s Soy Moratorium: supply-chain Ohio. Soil Sci. 168, 167–177.
governance is needed to avoid deforestation. Science 347. Souza, Z.M., Prado, R.M., Paixão, A.C.S., Cesarin, L.G., 2005. Sistemas de colheita e
Guzha, A.C., 2004. Effects of tillage on soil microrelief, surface depression storage and soil manejo da palhada de cana-de-açúcar. Pesq. Agrop. Brasileira 40, 271–278.
water storage. Soil Tillage Res. 76, 105–114. Tomasini, B.A., Vitorino, A.C.T., Garbiate, M.V., Souza, C.M.A., Alves Sobrinho, T., 2010.
Horton, R.E., 1933. The role of infiltration in the hydrological cycle. Trans. Am. Geophys. Infiltração de água no solo em áreas cultivadas com cana-de-açúcar sob diferentes
Union 14, 446–460. sistemas de colheita e modelos de ajustes de equações de infiltração. Eng. Agric. 30,
Horton, R.E., 1939. Analysis of runoff plat experiments with varying infiltration capacity. 1060–1070.

138