The use of landforms to predict the variability of soil

Geoderma 155 (2010) 55–66
Contents lists available at ScienceDirect
Geoderma
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / g e o d e r m a
The use of landforms to predict the variability of soil and orange attributes
D.S. Siqueira a,⁎, J. Marques Jr. b, G.T. Pereira c
a
b
c
Universidade Estadual Paulista(UNESP), Jaboticabal, São Paulo, Brazil
Department of Soils and Fertilizers, Universidade Estadual Paulista(UNESP), Jaboticabal, São Paulo, Brazil
Department of Exact Sciences, Universidade Estadual Paulista(UNESP), Jaboticabal, São Paulo, Brazil
a r t i c l e
i n f o
Article history:
Received 28 May 2009
Received in revised form 17 November 2009
Accepted 23 November 2009
Available online 24 December 2009
Keywords:
Geostatistics
Canonical correlation
Landscape
Citrus
a b s t r a c t
Relief may be considered an integrating factor that expresses the interaction of various soil and plant
attributes. This work aimed to analyze the potential use of landforms to predict the variability of soil and
orange attributes. The study area is located in the São Paulo state, Brazil. The following soil attributes were
analyzed: clay content, organic matter content, water content, aggregate stability, macropores, micropores,
total pore volume, saturated soil hydraulic conductivity, soil density, and soil resistance to penetration at
0.00–0.20-m depth. The orange attributes analyzed are total soluble solids, total titratable acidity, ratio,
production, concentrated juice yield, and fruit size, which were performed in three periods (July, August and
September). The soil and fruit data were submitted to descriptive statistical, geostatistical, and canonical
correlation (CCA) analyses. The mean soil and fruit attributes were significantly different for the landforms
by Tukey's test at 5% probability. The analysis of the geostatistical results showed that the spatial variability
of the soil and fruit attributes is influenced by landforms. We point out that the temporal variability of fruit
attributes is also influenced by landforms, resulting in different ripening gradients for each of the relief
compartments. The
fi rst canonical pair explained 77% of the attribute variance. The landforms were
shown to be efficient in mapping the variability of the soil and orange attributes and contributed to the
understanding of the soil–plant system.
© 2009 Elsevier B.V. All rights reserved.
1. Introduction
Orange culture adapts well to several types of soils, as long as they
have suitable physical attributes. Poor soil physical quality results in
poor water dynamics and storage in the profile, thus reducing the soil
volume exploited by plant roots (Santana et al., 2006). Paiva et al.
(1998) reported that soil water availability and root penetration are
more important for citric fruit culture than the supply of nutrients.
The other authors state the importance of soil physical attributes for
the production of citric fruit (Fidalski et al., 2007), the establishment
of sustainable culture systems (Reynolds et al., 2002), and the quality
of the soil as a whole (Doran and Parkin, 1996; Daily et al., 1997).
Among the several physical attributes used in the quantification of
the physical quality of soils (Topp and Zebchuk, 1979), soil density,
water penetration ratio in soil, hydraulic conductivity, and aggregate
stability stand out. Kemper and Derpsch (1981) and Roth et al. (1988)
used these attributes to evaluate the physical attributes related to
vegetable growth in tropical climate soils. Mapping these attributes
can help define soil use and management strategies better (Schaffrath
et al., 2008). However, information on the variability of soil properties
for agricultural purposes is obtained by random sampling (Sauer et al.,
⁎ Corresponding author. Tel.: +55 16 3609 2601; fax: +55 16 3609 2672.
E-mail address: [email protected] (D.S. Siqueira).
0016-7061/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.geoderma.2009.11.024
2006). The accuracy of the maps produced based on such information
is reduced, since the mapping unit is given by a single value, which
may not be representative of the mapped unit (Voltz and Webster,
1990).
To define the area limits and classes more precisely, investigators
have used geostatistics to study the spatial variability of the physical
attributes of soils (Viera et al., 1981; Voltz and Goulard, 1994; Bogaert
and D'Or, 2002). However, the comprehension and representation of
the variability of soil attributes are still limited, even with the use of
geostatistics. This is due either to the limited number of points used,
which interferes in the calculation of the semi-variance, or else to the
lack of a theoretical basis to determine the factors of variability of soil
attributes (Bogaert and D'Or, 2002).
The soil formation factors mentioned by Jenny (1941) are the main
conditioners of the initial variability of soil attributes (Trangmar et al.,
1985). Gessler et al. (2000) and Kravchenko and Bullock (2002)
reported that topography is an important factor in the spatial
distribution of soil attributes and that it affects vegetable development. The relief model proposed by Troeh (1965) is one of the major
landscape models used to explain the relation between soil attributes
and topography.
The influence of relief curvature on the variability of soil attributes
has been confirmed by several authors (Nizeyimana and Bicki, 1992;
Hammer et al., 1995; Pennock, 2003; Montgomery, 2003; McBratney
et al., 2003; Landi et al., 2004; Montanari et al., 2005; Tomer et al.,
56
D.S. Siqueira et al. / Geoderma 155 (2010) 55–66
2006; Camargo et al., 2008). Others have reported on the influence of
relief curve lines on crop yields (Thwaites and Slater, 2000; Manning
et al., 2001; Zebarth et al., 2002; Kravchenko and Bullock, 2002;
Norton et al., 2003; Si and Farrell, 2004; Razaei and Gilkes, 2005;
Martín et al., 2005; Zhang and Zhang, 2005; Terra et al., 2006; Rees
et al., 2007). In general, independent of the use of curvature models,
the studies that determine the soil quality in orange cultures present
several chemical, physical, and hydrologic attributes. The soil
attributes are described by univariate statistical analysis, which
compromises the interpretation of the results. The conclusions
drawn do not take variable interaction into consideration (Zhang
and Oxley, 1994; Fidalski et al., 2007). Kravchenko et al. (2005)
reported that crop yield results from a complex interaction between
several variables, including soil attributes and relief. An et al. (1997)
and Liu et al. (1996) call attention to the lack of studies on variable
interaction in the soil–plant system. Canonical correlation analysis
(CCA) is one of the methods used to investigate a large number of
variables. CCA has been used in the study of the intrarelation between
soil attributes (McBratney and Webster, 1981, Odeh et al., 1991) and
the soil–plant interrelation (Fu et al., 2004; Martín et al., 2005).
Theoretically, CCA has greater capacity to describe interdependent
phenomena related to soil attributes and crop yields.
Relief may be considered an integrating factor that expresses the
interaction of several soil and plant attributes. Furthermore, the time
and labor requirements for the acquisition of relief data are low in
comparison with the measurement of other soil attributes (Kravchenko and Bullock, 2002). Thus, relief may be used to map zones of
minimal soil and plant property variability, which is also called
specific management zone. Specific management zones are areas with
soil properties that have maximum homogeneity. The identification of
these zones ensures a more precise localization of the boundaries
between distinct areas, thus allowing the transfer of cultivation
techniques with greater ease and economy to similar environments
(Mallarino et al., 2001). This work aimed to analyze the potential use
of landforms to predict the variability of soil and orange attributes.
2. Materials and methods
2.1. Area characteristics
The study area is located in the municipality of Gavião Peixoto, in
the north of São Paulo State, Brazil (21° 75′ S and 48° 46′ W) (Fig. 1). It
is located in the geomorphological province of Planalto Ocidental
Paulista, in an altitude range of 525–580 m. The climate is classified as
temperate humid with dry winter and hot summer (CWa), according
to Köppen classification. The temperature ranges from 18-22 °C. The
study area is located in the northwestern of the São Paulo state. The
region's soil is predominantly formed by arenitic sediments from the
Adamantina Formation of the Bauru Group. Occupy approximately
9 million hectares (32%) of the São Paulo state (IPT, 1981).
The most recurrent soil class is dark red dystrophic latosol with
moderate A horizon and texture range from mean to clayish. Table 1
gives the description of the four soil profiles of the study area. Pêra-rio
Fig. 1. Area location and digital elevation model (DEM) giving the relief form division (Troeh, 1965), location of the soil profiles, and the surface water flow simulation.
57
D.S. Siqueira et al. / Geoderma 155 (2010) 55–66
Table 1
Description of the area soil profiles.
Horizon
Depth
Color
Munsell scale
m
Sand
Silt
Clay
g kg− 1
pH
H2O
OM
Al+ 3
CEC
g dm− 3
mmolc
dm− 3
m
V
%
Profile 1 — dystrophic red latosol (Oxisol)
A
0.0–0.25
2.5YR 3/4
Bw2
1.14–1.50
2.5YR 4/6
702
528
68
172
230
300
5.1
4.6
18
9
0
0
47.8
21.7
0.0
0
54.0
26.3
Profile 2 — dystrophic red latosol (Oxisol)
A
0.0–0.13
5YR 4/6
Bw2
1.05–1.49
2.5YR 4/6
820
641
100
89
170
270
5.2
4.4
16
8
0
6
47.8
21.7
0.0
40.8
64.6
28.3
Profile 3 — dystrophic red latosol (Oxisol)
A
0.0–0.25
5YR 4/6
Bw2
1.19–1.37
2.5YR 4/6
675
581
95
69
230
350
5.0
4.4
20
8
0
3
63.3
34.0
0.0
20
51.0
35.3
Profile 4 — dystrophic red–yellow latosol (Oxisol)
A
0.0–0.30
5YR 4/4
Bw2
1.29–1.70
5YR 5/6
560
572
210
88
230
340
4.8
4.6
20
9
0
0
52.9
23.8
0.0
0
47.1
24.4
OM — organic matter, CEC — cation exchange capacity, m — saturation by aluminum, and V — saturation by bases.
orange cultivar has been grown in the area for 14 years. The plant
spacing used is 7 × 3.5 m. Table 2 shows the mean attributes of
oranges from the study area in 2002 and 2003. The area was divided
into two relief types according to Troeh's (1965) curvature model,
concave (48.41 ha) and linear (45.17 ha) (Fig. 1). The digital elevation
model and the surface water flow simulation were obtained with
software Surfer V8.00 (Golden Software, Inc., 1999).
2.2. Sampling and evaluation of soil attributes
The soil attributes macropore, micropore, soil density, total pore
volume, aggregate stability, clay content, and organic matter content
were analyzed in samples from 312 sites in an area of 93.58 ha.
Saturated soil hydraulic conductivity, resistance to penetration, and
water content were determined in 144 sampling sites. The soil
attributes were determined in samples collected at 0.00–0.20 m deep
under the orange tree canopies, outside the agricultural machine
traffic and agricultural operation area. According to Cintra et al.
(1999), Souza et al. (2004), and Santana et al. (2006), 60% of citrus
roots are 0.00–0.20 m deep. All sampling sites were georeferenced
with a GPS device with SAD 69 system (South American Datum, Zone
22 South).
The samples were harrowing, dried and the soil was sifted through
the sieve of 9.51 mm in diameter mesh for reviews of aggregation, and
the other part sifted through the sieve of 2.0 mm. The separation and
aggregate stabilities were determined by the method described by
Table 2
History of the attributes of the fruit in the area (Leão, 2004; Leão et al., 2006).
Attributes
Landforms
Concave
Linear
CV
Mean
Kemper and Chepil (1965). The aggregates were placed in contact
with water on a sieve of 4.76 mm for 15 min, the mass of material
retained on each sieve was placed in an oven at 105 °C. From the
results obtained percent aggregate with diameter larger than 2 mm,
percent aggregate with diameter range 1–2 mm, and mean geometric
diameter (MGD) in mm were calculated. The clay content (g kg− 1)
was determined by the pipette solution using 0.1 N NaOH as
dispersing chemical and mechanical stirring apparatus at low speed
for 16 h, following the methodology proposed by Embrapa (1997),
according to Stokes Law.
The content of organic matter (g dm− 3) was obtained by the
organic carbon content determined by wet oxidation according to the
method described in Embrapa (1997). Saturated soil hydraulic
conductivity (mm h− 1) was determined with a Guelph permeameter
using a hydraulic load of 5 cm and was measured at a constant rate of
infiltration in the field according to Reynolds and Elrick (1985). To
determine the soil porosity (total porosity − TP, macropores and
micropores), the undisturbed soil samples were saturated for 48 h in
the pan with water up to two thirds the height of the ring. After the
saturation period, the samples were drained in the potential equal to
−0.006 MPa using a tension table (Embrapa, 1997). Soil density
(kg dm− 3) was determined by the method known volume in the
volume rings (Embrapa, 1997), the samples were collected with a
sampler adapted cylinders with a mean size of 5.04 cm in diameter
and 4.01 cm high. The resistance to penetration (MPa) was determined at each grid point, using an impact penetrometer model IAA/
Planalsucar with cone angle of 30°, which was calculated as Stolf
(1991). Soil water content (%) was determined by gravimetric
method (Gardner, 1986) in samples collected at the time of the test
of resistance to penetration in the field.
2.3. Sampling and evaluation of fruit attributes
Mean
Range
Range
CV
Year 2002
SS (°Brix)
R (non-dimensional)
CY (kg sampling unit− 1)
JY (crates tons of
concentrated juice− 1)
12.3
15.0
1147
221
3.5
6.7
38
76
5.6
7.5
0.75
6.6
11.5
15.0
1104
248
4.8
5.4
106
97
6.0
7.1
2.4
6.7
Year 2003
SS (°Brix)
R (non-dimensional)
CY (kg sampling unit− 1)
JY (crates tons of
concentrated juice− 1)
12.0
18.4
1293
252
2.3
6.9
920
102
3.8
6.6
15.2
6.46
11.3
17.7
1321
279
3.3
6.0
1120
104
5.0
7.7
16.7
7.6
CV — coefficient of variance, SS — total soluble solids, R — ratio, CY — crop yield, and JY —
juice yield.
Orange samples were taken in 312 sampling units measuring
220.5 m2. Each sampling unit was formed by nine plants; only one
central plant was georeferenced. The fruit attributes were evaluated
in three different ripening stages in 2005 (period 1, July; period 2,
August, and period 3, September). In July, the fruits were green, but
had a satisfactory size; in August, they were in the initial ripening
stage, and in September, the fruits were ripe.
Fruit (50) collected from the canopies of trees within each
sampling unit was analyzed for total soluble solids (SS), total titration
acidity (TA), and ratio (R) according to Pozzan and Triboni (2005). SS
was determined by refractometry; the results are given in °Brix. TA
(grams of citric acid in 100 ml of juice) was determined by titration
with 0.3125 N NaOH. The citric acid represents 70 to 90% of the acids
58
D.S. Siqueira et al. / Geoderma 155 (2010) 55–66
contained in juice (Davis and Albrigo, 1994). R, a non-dimensional
value, was obtained by dividing SS by TA. These measurements were
performed as described by Horwitz (1960).
Brix is the measurement of the soluble solids quantity in a sucrose
solution, which is used by the food industry to measure the approximate
amount of sugars in juices. The ratio is the proportion between soluble
solids and titratable acidity (The Orange Book, 1998). The ratio attribute
indicates the fruit maturity as well as is used by the industry to determine
the harvest time. According to Figueiredo (1991) the Pêra Rio cultivar
presents an average of 11.8° Brix , 0.95% acidity and 12.5 ratio.
Crop yield (CY) (kg per sampling unit) was determined by
weighing the fruit collected in each sampling unit in big bags using
a load cell. Juice characterization and yield (JY) were evaluated by
calculating the number of orange crates necessary to produce 1 ton of
concentrated fruit juice (65 °Brix). Fruit size (FS) was indirectly
evaluated by determining the amount of fruit necessary to fill a
standard 40.8 kg orange crate.
2.4. Statistical and geostatistical analyses
The data were submitted to descriptive statistical analysis by
calculation of mean, range, and coefficient of variation. The mean soil
attribute values were submitted to univariate ANOVA analysis with
software Minitab 14 (Minitab Release, 2000) to determine variations
in soil and fruit attributes according to the landforms. The spatial and
temporal variabilities of fruit attributes were evaluated by geostatistics (Matheron, 1963; Vieira et al., 1983; Isaaks and Srivastava, 1989).
After the mathematical model fitting, the data were interpolated by
Kriging Eq. (1)
K
for ∑ λi = 1
i=1
ð1Þ
where:
Z(h)
λi
Z(xi)
estimated Z value,
optimized weight of the neighboring points of position xi
based on the fitted semivariogram parameters,
absolute value of Z at position xi.
The spatial influence model was validated using the Jack Knifing
technique (Vieira and Lombardi Neto, 1995). Geostatistical analysis
was performed with software Surfer V8.00 (Golden Software, Inc.,
1999).
2.5. Canonical correlation analysis (CCA)
The relation between the physical–hydrologic attributes and fruit
attributes of the set of soil during the different ripening periods was
analyzed by CCA. In CCA, the interaction between the predicting and
the predicted variable sets is known as canonic pair Eq. (2).
S = a′X and P = b′Y
ð2Þ
where:
S
P
a′, b′
X, Y
canonical variable relative to the predicting attributes,
canonical variable relative to the predicted attributes,
vector coefficients obtained from the correlation matrices of
the soil and fruit attributes, respectively,
respective soil and fruit attribute measurement vectors.
The sampling points were paired and a network with 144 points
covering all the soil and fruit attributes in the different fruit ripening
periods was obtained. The canonical pairs were constructed based on
the soil and fruit attributes of 39 points (1 point = 0.4 ha− 1). The
predicting canonical variable (S) was formed by the following soil
attributes: MGD, clay content, organic matter content, hydraulic
conductivity, soil resistance to penetration, water content, TPV, and
soil density at 0.00–0.20-m depth. The predicted canonical variable
(P) was formed by the following fruit attributes: SS, TA, CY, and FS in
the three fruit ripening periods evaluated (July, period 1; August,
period 2, and September, period 3). The canonical pair that best
represented the linear combination of the two sets of variables was
chosen based on the highest values of the canonical correlation (λ),
the shared variance (λ), and the redundancy index, which is the
percent of the original variance of the fruit attributes explained by
canonical variance S. Additionally, Wilk's Lambda test was used to
verify the statistical significance (p) (Hair et al., 1995). After choosing
the canonical pair, the weights for the soil and fruit attributes,
coefficients a and b, respectively, were used to estimate the canonical
pair values for 105 points. We point out that from the 144 sampled
points for which all the soil and fruit attributes adopted were
evaluated, 39 points were used in the CCA analysis to construct the
variable interaction model. The remaining 105 points were used to
validate and to spatialize the model by geostatistics Eq. (3) (Mingoti,
2005).
S ′n = a1 A1 + … + an An ; P ′n = b1 B1 + … + bn Bn
ð3Þ
where:
index of the canonical variable estimated for point n, n
ranging from 1 to 105,
a [a1,..., an] coefficient of each of the eight soil attributes that were
studied,
A [A1,..., An] standardized values of the soil attributes at point n,
index of the canonical variable estimated for point n, n
Sn′
ranging from 1 to 105,
b [b1,..., bn] coefficient of each of the fruit attributes studied in the
three periods,
B [B1,..., Bn] standardized fruit attribute values.
Sn′
CCA was performed with software Statistica V7.00.
3. Results and discussion
3.1. Statistical and geostatistical analyses
The highest soil attribute range, except that of TPV, resistance to
penetration, and water content values, were found for the concave
form (Table 3). The range of the soil attributes influenced the
coefficient of variation (CV). The soil attributes that presented CV in
the same variation class in the two landforms were clay content, TPV,
and soil density (low class — CV b 12%), organic matter content,
resistance to penetration, and water content (mid class, CV range of
12–24%), and percent aggregates with 2–1 mm diameter, and
hydraulic conductivity (high class, CV N 24%) (Warrick and Nielsen,
1980). The CV values for the percentage of aggregates with diameter
larger than 2 mm, MGD, macropores, and micropores were the
highest for the concave relief form. These results corroborate those
reported by Montanari et al. (2005), who also found a larger variation
in soil attributes in the concave relief form.
Except for clay content and soil resistance to penetration, the mean
soil attribute values were statistically different by Tukey's test at 5%
(Table 3). Momtaz et al. (2009) also observed a significant difference
in the variation of soil attributes in different landforms. In the concave
form, they found higher mean values of percent aggregate in diameter
range of 2–1 mm, hydraulic conductivity, macropores, micropores,
TPV, and soil density. In the linear form, they found higher mean
59
D.S. Siqueira et al. / Geoderma 155 (2010) 55–66
Table 3
Descriptive statistics and mean test for the physical–hydraulic attributes at 0.00–0.20-m depth and fruit attributes in July (period 1), August (period 2), and September (period 3) in
year 2005.
Attributes
Landforms
Concave
Linear
Mean
Soil
Clay content (g kg− 1)
OM (g dm− 3)
Ø N 2 mm (%)
Ø between 2− 1 mm (%)
DMG (mm)
Hydraulic conductivity (mm h− 1)
Macropores (%)
Micropores (%)
TPV (%)
Soil density (kg dm− 3)
Soil resistance to penetration (MPa)
Water content (%)
Fruit
SS1 (°Brix)
SS2 (°Brix)
SS3 (°Brix)
TA1 (g of citric acid 100 ml of juice− 1)
TA2 (g of citric acid 100 ml of juice− 1)
TA3 (g of citric acid 100 ml of juice− 1)
R1 (non-dimensional)
R2 (non-dimensional)
R3 (non-dimensional)
CY1 (kg sampling unit− 1)
CY2 (kg sampling unit− 1)
CY3 (kg sampling unit− 1)
JY1 (crates tons of concentrated juice− 1)
JY2 (crates tons of concentrated juice− 1)
JY3 (crates tons of concentrated juice− 1)
FS1 (fruit crate− 1)
FS2 (fruit crate− 1)
FS3 (fruit crate− 1)
Range
CV
Mean
Range
CV
251
17
80.4
4.3
3.7
9
26.5
22.8
49.3
1.42
4.3
9.6
a
a
a
a
a
a
a
a
a
a
a
a
160
21
69.3
22.7
4.9
49
45.5
34.6
23.9
0.52
4.9
7.4
11.9
19.6
18.7
96.9
32.0
74.9
28.0
25.7
9.8
7.8
23.0
18.4
256
18
88.6
2.3
4.4
4
24.1
22.2
46.2
1.38
4.3
10.7
a
b
b
b
b
b
b
a
b
b
a
b
160
14
59.3
14.5
4.6
43
25.4
26.1
28.0
0.48
5.7
9.6
10.2
16.1
10.7
89.2
18.9
132.5
21.7
20.7
8.4
6.1
25.0
19.2
10.39
10.43
11.91
1.26
1.08
0.87
8.30
9.70
13.85
2,007
1,527
1,666
269
269
230
245
220
233
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
2.44
3.16
5.80
0.73
0.51
0.85
5.28
4.39
8.97
1,291
1,224
1,883
69
99
131
94
102
318
4.8
6.4
9.8
8.1
7.5
13.4
8.0
6.9
9.4
11.9
14.7
22.8
5.7
7.1
10.7
7.2
8.9
17.4
10.25
10.60
11.81
1.40
1.19
0.94
7.32
8.95
12.68
2,202
1,533
1,865
291
274
244
245
227
229
a
a
a
b
b
b
b
b
b
b
a
b
b
b
b
a
b
a
2.70
3.53
5.68
0.48
0.62
0.62
2.29
4.33
5.81
2,110
1,851
2,684
119
111
141
101
124
196
4.9
7.1
9.9
6.4
8.2
12.4
7.0
7.7
8.2
17.8
19.0
23.2
7.0
8.5
10.5
8.1
11.1
16.9
CV — coefficient of variation, OM — organic matter, MGD — mean geometric diameter, and TPV — total pore volume.
SS — total soluble solids, TA — total titration acidity, R — ratio, CY — crop yield, JY — juice yield, and FS — fruit size.
1 — July, 2 — August, and 3 — September.
Means followed by the same letter in the same line are not significantly different by Tukey's test at 5% probability.
values for organic matter (18 g dm− 3), percentage of aggregates
larger than 2 mm, MGD, and water content.
The analysis of the hydraulic conductivity values of the saturated
soil revealed that the concave form presented values in the fast-tomoderate flow class. In contrast, the linear form presented values in
the slow flow class (4 mm h− 1) (USDA, 1993) due to the higher total
porosity of the concave relief form. Souza et al. (2006) confirmed the
relation between TPV and hydraulic conductivity. According to Ahuja
et al. (1984), the movement of water in soil depends more on the soil
structure than texture, as it is affected by the soil pore system.
Fruit attributes SS1, SS2, TA2, CY1, CY2, CY3, R1, R2, R3, FS1, and FS2
had the lowest range values in the concave form (Table 3). Among
these attributes, SS, TA, and JY reflect the fruit quality. Thus, the
concave form provides a more homogeneous environment in terms of
fruit quality. Except for CY1 in the linear form and TA3, CY2, CY3, and
FS3 in the concave and linear forms, which presented mean class CV, all
the other attributes of both landforms and the three periods belonged
to the low class (Warrick and Nielsen, 1980). However, it stands out
that the same attributes within the same variation class had a lower CV
in the concave form, except TA1, TA3, R3, R3, and FS3. All fruit
attributes were statistically different by Tukey's test at 5% significance,
except SS1, SS2, SS3, CY2, FS1, and FS3. Kravchenko and Bullock
(2002), working with soybean, observed variation in quality attributes
of grain, protein content and oil in different parts of the landscape.
Table 4 gives the geostatistical results of soil attributes. The
exponential and spherical model fitted the data spatial variance
structure. This result corroborates those of McBratney and Webster
(1986), confirming that the exponential and spherical models best fit
the soil attribute data. The geostatistical analysis showed that all soil
attributes analyzed were spatially dependent on soil depth 0.00–
0.20 m, as was also observed by the other authors (Boehm and
Anderson, 1997; Sobieraj et al., 2002) who investigated the spatial
variability of soil attributes as a function of landforms. The attributes
clay content, water content and soil resistance to penetration
presented strong spatial dependence (DSD, degree of spatial dependence [C0 / C0 + C1 ⁎ 100] lower than 25%) (Cambardella et al., 1994).
These attributes directly influenced the soil structure and the water
flow in the soil profile. Cambardella et al. (1994) found that the
variables that presented strong spatial dependence were more
influenced by the intrinsic soil attributes, such as soil formation
factors, including relief. Drees et al. (1994) and Noorbakhsh et al.
(2008) reported that the physical attributes are largely a consequence
of changes in the soil structure caused by water flow in the soil profile.
The values ranged from 195 to 600 mm, presenting a mean value
of 350 m. The values of all semivariogram models obtained by the Jack
Knifing technique were satisfactory, presenting a mean close to zero
and variance close to 1 (Vieira and Lombardi Neto, 1995).
The geostatistical results demonstrated spatial dependence structure for all fruit attributes in all periods (Table 4), which shows that
fruit attributes are influenced by culture location in the landscape. The
DSD of all fruit attributes ranged from strong to mild in the different
periods, expect SS1, SS2, and JY3 (Cambardella et al., 1994). Farias
60
D.S. Siqueira et al. / Geoderma 155 (2010) 55–66
Table 4
Fitted model and semivariogram parameters for the soil attributes at 0.00–0.20-m depth.
Attributes
Model
C0a
Soil
Ø N 2 mm
Ø between 2 and 1 mm
MGD
OM
Clay content
Hydraulic conductivity
Water content
Soil resistance to penetration
Macropores
Micropores
Soil density
TPV
Exponential
Exponential
Exponential
Exponential
Spherical
Spherical
Exponential
Exponential
Spherical
Exponential
Spherical
Spherical
80.00
7.20
0.68
5.20
150.00
21.00
0.00
0.00
27.00
10.00
0.07 10− 1
19.80
Fruit
SS1
SS2
SS3
TA1
TA2
TA3
R1
R2
R3
CY1
CY2
CY3
JY1
JY2
JY3
FS1
FS2
FS3
Spherical
Spherical
Exponential
Exponential
Spherical
Spherical
Exponential
Exponential
Spherical
Spherical
Exponential
Exponential
Spherical
Exponential
Spherical
Exponential
Exponential
Exponential
2.25 10− 1
0.35
0.90
0.28 10− 2
0.28 10− 2
0.06 10− 1
0.14
0.31
0.75
75.00 10− 3
43.00 10− 3
120.00 10− 3
110.00
250.00
500.00
220.00
300.00
1090.00
C0 + C1b
150.00
9.80
1.00
8.20
800.00
36.50
3.50
1.32
41.00
29.50
0.01
26.00
0.28
0.43
1.35
0.01
0.01
0.01
0.30
0.48
1.07
105.00 10− 3
68.00 10− 3
175.00 10− 3
290.00
450.00
650.00
360.00
515.00
1660.00
a (m)c
Jack Knifingd
DSDe
Mean
Variance
300
300
390
330
280
480
240
195
310
279
600
500
− 0.01
0.00
0.00
0.00
− 0.01
− 0.01
0.00
0.00
− 0.01
0.01
− 0.01
0.01
1.47
1.76
1.48
1.57
1.27
1.53
1.70
0.01
1.78
1.70
0.02
1.64
53.33
73.47
68.00
63.41
18.75
57.53
0.00
0.00
65.85
33.90
70.00
76.15
470
210
480
111
90
130
210
450
130
420
240
450
90
195
150
450
270
570
0.00
− 0.01
0.00
0.02
0.04
− 0.01
0.00
0.00
0.00
− 0.01
0.00
0.00
0.00
0.00
0.00
− 0.03
0.00
− 0.01
1.99
1.78
1.72
0.81
0.98
1.72
0.72
1.47
1.36
1.54
0.15
1.63
1.45
0.10
1.14
1.96
0.10
1.44
80.36
81.40
66.67
28.00
28.00
60.00
46.67
64.58
70.09
71.43
63.24
68.57
37.93
55.56
76.92
61.11
58.25
65.66
OM — organic matter, MGD — mean geometric diameter, and TPV — total pore volume.
SS — total soluble solids, TA — total titration acidity, R — ratio, CY — crop yield, JY — juice yield, and FS — fruit size.
1 — July, 2 — August, and 3 — September.
a
Nugget effect.
b
Sill.
c
Range (m).
d
Model validation test.
e
Degree of spatial dependence (%): C0 / C0 + C1 ⁎ 100.
et al. (2003) also observed spatial variability for fruit attributes within
the same sampling unit.
The other authors (Fu et al., 2004; Zhang and Zhang, 2005; Martín
et al., 2005; Rees et al., 2007) who studied the soil–plant interaction
confirmed a strong relationship between crop yield and landforms.
Stevenson et al. (2001) reported that the soil–relief interaction
explains 40% of the crop yield variability. The range values were from
90 to 570 m, presenting a mean value of 284 m. The similarity
between soil and fruit attribute range values stands out.
The increase in the DSD characterizes a decrease in the spatial
dependence. Table 4 shows that the fruit attributes were homogeneous in all periods, except SS3, CY2, and FS2, which show that relief
influences the temporal variability of orange attributes, affording a
landform-dependent ripening gradient.
Fig. 2 shows the soil attribute distribution maps. The spatial distributions of clay content (Fig. 2a) and organic matter (Fig. 2b) were
quite similar for the two landforms. Thus, we can conclude that the
variability of the mineralogical attributes of the clay fraction, such as
hematite, goethite, kaolinite, and gibbsite, influences the soil aggregation of the different landform soils. Some authors (Deshpande et al.,
1968; Muggler et al., 1996; Lima et al., 2000) confirmed the influence of
mineralogical attributes on the microaggregation and the aggregation of
tropical climate soils. Studies on mineral variability and its relationship
with soil physical attributes in tropical climate soils in different
landforms, such as those by Camargo et al. (2008), are needed.
The results for saturated hydraulic conductivity (Fig. 2f) showed
that the concave landform values were the highest, which was due to
its higher TPV (Fig. 2i). Souza et al. (2006) found that, besides TPV,
pore morphology also contributes to a larger movement of water
along the soil profile, as is also demonstrated by the results given in
Table 3.
The analysis of the maps in Fig. 2 and the water flow map in Fig. 1
reveals that the water movement determined by the landforms
influences the spatial variability of soil attributes. The soil attribute
isolines are longer in the direction of the water flow in both the
concave and the linear landforms. Young and Hammer (2000)
mentioned that the interaction between relief and water flow leads
to the formation of pedoenvironments, which are the main factors
responsible for the variability of the soil attributes.
Fig. 3 displays the spatial distribution of the fruit attributes. The
concave landforms presented the highest SS values (Fig. 3a, b, and c).
The low SS contents found in low lands may be associated with the
greater dilution of fruit sugars as a result of water accumulation in this
part of the landscape. This corroborates previous results (Table 3,
Fig. 2k). The spatial distribution maps of SS (Fig. 3a, b, and c) and TA
(Fig. 3d, e, and f) presented an opposite concentration gradient,
probably because the sugar content increases and the citric acid
content decreases as the fruits ripen (The orange book, 1998).
The fruit attribute isolines (Fig. 3) also describe a behavior similar
to that of the water flow (Fig. 1). The variation in the CY maps (Fig. 3j,
D.S. Siqueira et al. / Geoderma 155 (2010) 55–66
61
Fig. 2. Spatial distribution maps of soil attributes at 0.00–0.20-m depth. (a) Clay content — g kg− 1, (b) organic matter content — g dm− 3, (c) aggregates with diameter larger than 2 mm —
%, (d) aggregates in diameter range 2–1 mm — %, (e) mean geometric diameter — mm, (f) saturated soil hydraulic conductivity — mm h− 1, (g) macropore — %, (h) micropore — %, (i) total
pore volume — %, (j) soil density — kg dm− 3, (k) soil resistance to penetration — MPa, and (l) water content — % (—) division of the landforms (Troeh, 1965).
l, and k) ranged from 1000 to 3800 kg sample− 1. The CY values for the
concave form were the lowest, in contrast to the FS values (Fig. 3p, q,
and r). The JY values (Fig. 3m, n, and o) ranged from 190 to
355 crate tons of juice− 1 (65 °Brix). The darker areas, mostly located
in the concave form, produced larger fruit, consequently requiring a
smaller number of orange crates to produce 1 ton of 65 °Brix juice. The
lighter areas, mostly located in the linear form, required up to 142%
more orange crates to produce 1 ton of 65 °Brix juice (Table 3),
reflecting the high fruit quality in the concave landforms and the
predominance of fruit with higher R (Fig. 3g, h, and i). These results
confirm those in Table 3 and those reported by Leão (2004) and Leão
et al. (2006) (Table 2) for the same area.
In Brazil, the purchase of orange raw material is made under some
quantitative characteristics as production amount ( number of boxes
containing 40.8 kg). This contributes to discourage investments by
orange producers.
The Brazilian Citrus Industry is promoting changes in the Trade Protocols, including the fruit quality as an essential criterion. The fruit quality is already used in Florida, USA, in order to price the fruit. In the
United States, the oranges have to be harvested when they reach a
ratio of 13 then processed to produce juice in a ratio of 15 to 18,which
is the consumer´s preference. In Brazil, even though the frozen juice
consumption is not large (5% of the total amount produced), the ratio
above 14 is more acceptable. Nevertheless, the processing is made
in a ratio of 12 to 13. The mapping of those ratio’s values areas,using
the landforms and matematical modeling help with evaluation of environments with more or less ratio potential, contributing to the harvest
planning and management.
62
D.S. Siqueira et al. / Geoderma 155 (2010) 55–66
63
D.S. Siqueira et al. / Geoderma 155 (2010) 55–66
Table 5
Data of the canonical pairs obtained by the linear combination of the physical–hydraulic
attributes at 0.00–0.20-m dept with other fruit attributes for the three periods (July,
August, and September) in year 2005.
Canonic
pair
Canonic
correlation (λ)
Shared
variance (λ2)
Variance of the fruit
attributes explained by S
pa
S1,P1
S2,P2
S3,P3
S4,P4
S5,P5
S6,P6
S7,P7
S8,P8
0.88
0.75
0.68
0.63
0.54
0.50
0.34
0.17
0.77
0.56
0.47
0.39
0.29
0.25
0.11
0.03
0.15
0.03
0.02
0.02
0.01
0.02
0.01
0.00
0.01
0.06
0.15
0.28
0.46
0.65
0.86
0.97
S — canonic variable for the soil attributes and P — canonic variable for the fruit
attributes.
a
p b 0.01. Significant at 1% probability in Wilk's Lambda test.
Table 6
Correlation of soil attributes at 0.00–0.20-m depth and fruit attributes in three periods
(July, August, and September) with the first pair of canonic variables (S1 and P1).
Soil attributes
S1
Fruit attributes
P1
Hydraulic conductivity
Soil resistance to penetration
Water content
Soil density
TPV
OM
Clay content
MGD
0.53
0.46
0.41
0.25
0.12
0.07
− 0.17
− 0.63
SS1
FS1
FS3
CY3
SS2
CY2
SS3
FS2
TA3
CY1
0.63
0.10
0.02
− 0.06
− 0.13
− 0.18
− 0.27
− 0.35
− 0.51
− 0.65
TPV — total pore volume, OM, organic matter, and MGD — mean geometric diameter.
S′1 — estimated canonic variable for the soil attributes and P′1 — estimated canonic
variable for the fruit attributes.
This way, the ratio values (Fig. 3g, 3h, 3i) present the concave area
as more favorable environment to grow oranges with high commercial
standard. Likewise, when the fruits are harvested at an optimal ripeness point, which is when they reach the maximum total soluble solids
content standardized in 65o
° Brix, the citrus industries will be able to decrease their energy expense during the concentrated juice manufacturing.
3.2. Influence of relief on the soil–fruit attribute interrelation
CCA produced eight canonical pairs. The correlation between the
original variables and the canonical variables was significant only for
the first canonical pairs (p b 0.01, Wilk's Lambda) (Table 5). The
correlation of the first canonical pair was 0.88. Martín et al. (2005)
found a correlation of 0.75 for the first canonical pair in the analysis of
the correlation of soil and plant attributes. Canonical variables S1 and
P1 shared 0.77 of the data variance, which indicates that the first
canonical pair expresses the linear combination of the soil and fruit
attributes the best. The relationship between the original variables
and the first canonical pair indicates that 15% of the variance of the
fruit attributes in the three periods is explained by S1. Martín et al.
Table 7
Fitted model and semivariogram parameters for the estimated canonic variables at 105
points.
Canonic
variables
Model
S′1
P′1
Exponential
Exponentia
C0a
1.40
0.05
C0 + C1b
1.98
0.27
a (m)
510
450
c
Jack Knifingd
Mean
Variance
− 0.01
0.01
1.57
0.03
DSDe
70.71
18.52
estimated canonic
S′1 — estimated canonic variable for the soil attributes and P′—
1
variable for the fruit attributes.
a
Nugget effect.
b
Sill.
c
Range (m).
d
Test for the validation of the model.
e
Degree of spatial dependence (%).
(2005) found that 29 and 24% of the variance of the plant attributes in
2000 and 2001, respectively, were explained by S1.
The correlation between the original variables and the canonical
pairs is given in Table 6. The soil attribute that presented the highest
positive correlation with canonical variable S1 was hydraulic
conductivity (r = 0.53), while SS1 presented the highest positive
correlation with canonical variable P1 (r = 0.63). The most significant
negative correlation between soil attributes and S1 was that of MGD
(r = −0.63). The fruit attribute with the most significant negative
correlation with P1 was CY1 (r = − 0.65). These results confirm those
presented in Table 3 and Figs. 2 and 3. The landscape areas with the
highest hydraulic conductivity values (Fig. 2f), TPV (Fig. 2i), and MGD
(Fig. 2e) gave the highest SS1 values (Fig. 3a, b, and c) and the lowest
CY1 values (Fig. 3j, l, and k), which indicate that the areas with the
highest crop yields are not necessarily the ones with the highest
concentrated juice yields. The linear landforms presented the best
physical conditions in terms of MGD, TPV, and hydraulic conductivity
(Table 3, Fig. 2). However, they also presented low chemical quality
(Table 2). Canonic variable S1′ presented moderate spatial dependence
(25% b DSD b 75%) and canonic variable P1′ presented strong spatial
dependence (DSD b 25%) (Cambardella et al., 1994) (Table 7). The
range of S1′ is close to the ranges of hydraulic conductivity, soil density,
and TPV (Table 4). The fruit attributes that presented ranges similar to
that of P1′ were SS1, SS3, R2, CY1, CY3, and FS1 (Table 4). According to
Wu et al. (2002), when there is an interaction between variable
groups, they present the same spatial structure. Thus, the distributions of these values in the landscape are similar. The values obtained
by the Jack Knifing technique with reduced mean close to zero and
variance close to 1 validate the fitted semivariogram models (Vieira
and Lombardi Neto, 1995). The well-defined level (C0 + C1) close to
the variance line, as well as the number of pairs, equal to or greater
than 50, involved in the semi-variance calculation of the first lags,
reinforces the validity of the models (Fig. 4a and c) (Wollenhaupt
et al., 1997; Burrough and Mcdonnel, 2000).
The spatial distribution maps (Fig. 4b and d) demonstrate the
aptitude of the concave and linear landforms for soil and fruit
attributes. Fig. 4b shows that the best values for the interaction of
physical–hydrologic attributes are those for the concave landforms.
The plant response to this environment is given in Fig. 4d, which also
gives the highest values for the concave landforms. Each attribute
correlates with these values either positively or negatively (Table 6).
Fu et al. (2004) investigated the relationship between soil attributes,
topography, and vegetable diversity by CCA and also found the best
environments to be the top part of the relief.
Fig. 3. Spatial distribution maps of the fruit attributes 1st column — July (period 1), 2nd column — August (period 2), 3rd column — September (period 3), (a, b, and c) total soluble
solids — °Brix, (d, e, and f) total titration acidity — grams of citric acid 100 ml juice− 1, (g, h, and i) ratio — non-dimensional, (j, l, and k) production — kg sample− 1, (m, n, and o) juice
yield — crates tons of concentrated juice− 1, (p, q, and r) fruit size — fruit crate− 1, (—) division of landforms (Troeh, 1965).
64
D.S. Siqueira et al. / Geoderma 155 (2010) 55–66
Fig. 4. Fitted semivariograms for estimated canonic variables S′1 (a) and P′1 (c), spatial distribution maps for estimated canonic variables S1′ (b) and P1′ (d), (—) division of landforms
(Troeh, 1965).
4. Conclusions
The division of the landforms helped in the understanding of the
interrelations and intrarelations between physical–hydrologic soil
attributes and orange attributes in different periods of the year. The
associated use of geostatistics and CCA allowed the efficient representation of these relations and an understanding of the influence of
relief on the variability and interaction of these attributes. Thus,
besides the analysis of crop response, it also allows for investigating
the response causes and for the mapping of this response.
The concave landforms also presented a higher aptitude for water
movement in the soil profile, while the linear form had a greater
aptitude for the formation of large aggregates. The concave landforms
favored fruit attributes, leading to higher SS, while the linear form
resulted in higher CY in period 1 (July). Therefore, CCA, geostatistics,
and landforms must be considered in similar studies for a better
understanding of the soil–plant system, for specific management zone
mapping, and the analysis of zone crop aptitude.
Acknowledgements
The authors are grateful to Fundação de Amparo a Pesquisa do
Estado de São Paulo (FAPESP) for the scientific initiation scholarship
and research grant.
References
Ahuja, L.R., Naney, J.W., Green, R.E., Nielsen, D.R., 1984. Macroporosity to characterize
spatial variability of hydraulic conductivity and effects of land management. Soil
Science Society of America Journal 48, 699–702.
An, S.Q., Wang, Z.F., Liu, Z.L., Hong, B.G., Zhao, R.L., 1997. Effects of soil factors on species
diversity in secondary forest communities. Acta Ecologica Sinica 17 (1), 45–50. in
Chinese with English abstract.
Boehm, M.M., Anderson, D.W., 1997. A landscape-scale study of soil quality in three
prairie farming systems. Soil Science Society of America Journal 61, 1147–1159.
Bogaert, P., D'Or, D., 2002. Estimating Soil Properties from Thematic Soil Maps.
Burrough, P.A., Mcdonnel, R.A., 2000. Principles of Geographical Information Systems.
Oxford University Press, UK.
Camargo, L.A., Marques Júnior, J., Pereira, G.T., Horvat, R.A., 2008. Variabilidade espacial
de atributos mineralógicos de um latossolo sob diferentes formas do relevo. II —
correlação espacial entre mineralogia e agregados. Revista Brasileira de Ciência do
Solo 32, 2279–2288.
Cambardella, C.A., Moorman, T.B., Novak, J.M., Parkin, T.B., Karlen, D.L., Turco, R.F.,
Konopka, A.E., 1994. Field scale variability of soil properties in Central Iowa soils.
Soil Science Society of America Journal 58, 1501–1511.
Cintra, F.L.D., Libardi, P.L., de Jorge, L.A.C., 1999. Distribuição do sistema radicular de
porta-enxertos de citros em ecossistema de tabuleiro costeiro. Revista Brasileira de
Fruticultura 21, 313–317.
Davis, F.S., Albrigo, L.G., 1994. Citrus. CAB internacional, Wallingford. 254 pp.
Daily, G.C., Matson, P.A., Vitousek, P.M., 1997. Ecosystem services supplied by soil. In:
Daily, G.C. (Ed.), Nature's Services: Societal Dependence on Natural Ecosystems.
Island Press, Washington, DC, pp. 113–132.
Deshpande, T.L., Greenland, D.J., Quirk, J.P., 1968. Changes in soil properties associated
with the remove of iron and aluminium oxides. Journal of Soil Science 1, 18–22.
Doran, J.W., Parkin, T.B., 1996. Quantitative indicators of soil quality: A minimum data
set. In: Doran, J.W., Jones, A.J. (Eds.), Methods for Assessing Soil Quality. SSSA Spec.
Publ., vol. No. 49. SSSA, Madison, WI, pp. 25–37.
D.S. Siqueira et al. / Geoderma 155 (2010) 55–66
Drees, L.R., Karathanasis, A.D., Wilding, L.P., Blevins, R.L., 1994. Micromorphological
characteristics of long-term no-till and conventionally tilled soils. Soil Science
Society of America Journal 58, 508–517.
Embrapa, 1997. Manual de métodos de análise de solo, 2.ed. Ministério da Agricultura e
do Abastecimento, Rio de Janeiro, p. 212.
Farias, P.R.S., Nociti, L.A.S., Barbosa, J.C., 2003. Agricultura de precisão: mapeamento da
produtividade em pomares cítricos usando geoestatística. Revista Brasileira de
Fruticultura 25, 235–241.
Figueiredo, J.O., 1991. Variedades copa de valor comercial. In: Rodriguez, O., Viégas, F.C.P.,
Pompeu Jr., J., Amaro, A.A. (Eds.), Citricultura Brasileira, pp. 228–264. Campinas.
Fidalski, J., Tormena, C.A., Silva, A.P., 2007. Qualidade física do solo em pomar de
laranjeira no noroeste do paraná com manejo da cobertura permanente na
entrelinha. Revista Brasileira de Ciência do Solo 31, 423–433.
Fu, B.J., Liu, S.L., Ma, K.M., Zhu, Y.G., 2004. Relationships between soil characteristics,
topography and plant diversity in a heterogeneous deciduous broad-leaved forest
near Beijing, China. Plant and Soil 261, 47–54.
Gardner, W.H., 1986. Water content. In: Klute, A. (Ed.), Methods of Soil Analyses.
Agronomy Monography, vol. 9. ASA, Madison, pp. 493–541.
Gessler, P.E., Chadwick, O.A., Chamran, F., Althouse, L., Holms, K., 2000. Modeling soillandscape and ecosystem properties using terrain attributes. Soil Science Society of
America Journal 64, 2046–2056.
Golden Software, Inc., 1999. Surfer Version 8.00. Surface Mapping System, Golden, CO.
Hair, J.F., Anderson, R.E., Tatham, R.L., Black, W.C., 1995. Multivariate Data Analysis with
Readings, 4. ed. Macmillan Publishing International, New York.
Hammer, R.D., Young, N.C., Wolenhaupt, T.L., Barney, T.L, Haithcoate, T.W., 1995. Slope
class maps form soil survey and digital elevation models. Soil Science Society of
America Journal 59, 509–519.
Horwitz, C.O., 1960. Official Methods of Analysis of the Association of Official Analytical
Chenusts, 9. ed. Board, Whashington, p. 1015.
IPT, 1981. Instituto de pesquisas tecnológicas, Mapa geomorfológico do Estado de São
Paulo. São Paulo. Escala — 1:1.000.000.
Isaaks, E.H., Srivastava, R.M., 1989. An Introduction to Applied Geoestatistics. Oxford
University Press, New York.
Jenny, H., 1941. Factors of Soil Formation. McGraw-Hill Book Co., New York.
Kemper, B., Derpsch, R., 1981. Soil compaction and root growth in Parana. In: Russel, R.S.,
Igue, K., Mehta, Y.R. (Eds.), The Soil/Root System in Relation to Brazilian Agriculture.
IAPAR, Londrina, PR, pp. 62–81.
Kemper, W.D., Chepil, W.S., 1965. Size distribution of aggregates. In: Black, C.A. (Ed.),
Methods of Soil Analysis: American Society Agronomy, pp. 499–510.
Kravchenko, A.N., Bullock, D.G., 2002. Correlation of corn and soybean yield with
topography and soil properties. Agronomy Journal 75, 75–83.
Kravchenko, A.N., Robertson, G.P., Thelen, K.D., Harwood, R.R., 2005. Management,
topographical, and weather effects on spatial variability of crop grain yields.
American Journal 97, 514–523.
Landi, A., Mermut, A.R., Anderson, D.W., 2004. Carbon distribution in a hummocky
landscape from Saskatchewan, Canada. Soil Science Society of America Journal 68,
175–184.
Leão, M.G.A., Siqueira, D.S., Marques Junior, J., Pereira, G.T., Barbieri, D.M., 2006.
Influência das formas da paisagem no rendimento e na qualidade da fruta cítrica.
Anais do Congresso Brasileiro de Agricultura de Precisão CONBAP-Congresso
brasileiro de Agricultura de Precisão São Pedro-SP.
Leão, M.G.A., 2004. Relação entre variabilidade dos atributos de um latossolo e
qualidade da fruta cítrica. 119 f. Dissertação (Mestrado em Ciência do Solo) —
Faculdade de Ciências Agrárias e Veterinárias, Universidade Estadual Paulista,
Jaboticabal.
Lima, J.M., Sharon, J., Anderson, J., Curi, N., 2000. Phosphate-indiced clay dispersion as
related to aggregate size and composition in Hapludoxs. Soil Science Society of
America Journal, Madison 64, 892–897.
Liu, C.M., Li, C.Z., Shi, M.H., Liang, H.Y., 1996. Multivariate statistical analysis techniques
applicated in differentiation of soil fertility. Acta Ecologica Sinica 16, 444–447 (in
Chinese with English abstract).
Mallarino, A.P., mazhar, U.H., Wittry, D., Bermudez, M., 2001. Variation in soybean
response to early season foliar fertilization among and within fields. Agronomy
Journal 93, 1220–1226.
Manning, G., Fuller, L.G., Eilers, R.G., Florinsky, I., 2001. Topographic influence on the
variability of soil properties within an undulating Manitoba landscape. Canadian
Journal of Soil Science 81, 439–447.
Martín, N.F., Bollero, G.A., Bullock, D.G., 2005. Associations between field characteristics
and soybean plant performance using canonical correlation analysis. Plant and Soil
55, 273-39.
Matheron, G., 1963. Principles of geostatistics. Economic Geology 58, 1246–1266.
McBratney, A.B., Webster, R., 1981. Spatial dependence and classification of soil along a
transect in north-east Scotland. Geoderma 26, 63–82.
Mcbratney, A.B., Webster, R., 1986. Choosing functions for semi-variograms of soil
properties and fitting them to sampling estimates. Soil Science, Baltimore 37, 617–639.
Mcbratney, A.B., Santos, M.L.M., Minasny, B., 2003. On digital soil mapping. Geoderma 117,
3–52.
Mingoti, S.A., 2005. Análise de dados através de métodos de estatística multivariada:
uma abordagem aplicada. Belo Horizonte, Minas Gerais.
Minitab Release, 2000. Making Data Analysis Easler: Version 13.1.
Momtaz, H.R., Jafarzadeh, A.A., Torabi, H., Oustan, Sh., Samadi, A., Davatgar, N., Gilkes, R.J.,
2009. An assessment of the variation in soil properties within and between landform
in the Amol region. Iran Geoderma 149, 10–18.
Montanari, R., Marques Júnior, J., Pereira, G.T., Souza, Z.M., 2005. Forma da paisagem
como critério para otimização amostral de latossolos sob cultivo de cana-de-açúcar.
Pesquisa Agropecuária Brasileira 40, 69–77.
65
Montgomery, D.R., 2003. Predicting landscapescale erosion rates using digital elevation
models. Comptes Rendus Geoscience 335, 1121–1130.
Muggler, C.C., Curi, N., Silva, M.L.N., Lima, J.M., 1996. Características pedológicas de
ambientes agrícolas nos Chapadões do Rio Corrente, sudoeste da Bahia. Pesquisa
Agropecuária Brasileira 31, 221–232.
Nizeyimana, E., Bicki, T.J., 1992. Soil and soil landscape relationships in the North
Central region of Rwanda, East-Central Africa. Soil Science 153, 225–236.
Noorbakhsh, S., Schoenau, J., SI, B., Zeleke, T., Qian, P., 2008. Soil properties, yield, and
landscape relationships in South-Central Saskatchewan Canada. Journal of Plant
Nutrition 31, 539–556.
Norton, J.B., Sandor, J.A., White, C.S., 2003. Hillslope soils and organic matter dynamics
within a Native American agroecosystem on the Colorado Plateau. Soil Science
Society of America Journal 67, 225–234.
Odeh, I.O.A., Chittleborough, D.J., McBratney, A.B., 1991. Elucidation of soil–landform
interrelationships by canonical oridination analysis. Geoderma 49, 1–32.
Paiva, A.Q., Souza, L.S., Ribeiro, A.C., Costa, L.M., 1998. Disponibilidade de água em uma
toposseqüência de solos de tabuleiro do Estado da Bahia e sua relação com indicadores
do crescimento da laranjeira. Revista Brasileira Ciência do Solo 22, 367–377.
Pennock, D.J., 2003. Terrain attributes, landform segmentation, and soil redistribution.
Soil and Tillage Research 69, 15–26.
Pozzan, M., Triboni, H.R., 2005. Citros. Centro APTA Citros Sylvio Moreira, p. 803. c. 26.
Razaei, S.A., Gilkes, R.J., 2005. The effects of landscape attributes and plant community
on soil physical properties in rangelands. Geoderma 125, 145–154.
Rees, H.W., Chow, T.L., Gregorich, E.G., 2007. Spatial and temporal trends in soil
properties and crop yield at a site under intensive up- and down-slope potato
production in northwestern New. Canadian Journal of Soil Science 87, 383–398.
Reynolds, W.D., Elrick, D.E., 1985. In situ measurement of field-saturated hydraulic
conductivity, sorptivity, and the α-parameter using the Guelph permeameter. Soil
Science 140, 292–302.
Reynolds, W.D., Bowman, B.T., Drury, C.F., Tan, C.S., Lu, X., 2002. Indicators of good soil
physical quality: density and storage parameters. Geoderma 110, 131–146.
Roth, C.H., Meyer, B., Frede, H.G., Derpsch, R., 1988. Effect of mulch rates and tillage
systems on infiltrability and other soil physical properties of a Oxisol in ParanaÂ,
Brazil. Soil Tillage Research 11, 81–91.
Santana, M.B., da Souza, L.S., Souza, L.D., Fontes, L.E.F., 2006. Atributos físicos do solo e
distribuição do sistema radicular de citros como indicadores de horizontes coesos
em dois solos de tabuleiros costeiros do Estado da Bahia. Revista Brasileira de
Ciência do Solo 30, 1–12.
Sauer, T.J, Cambardella, C.A., Meek, D.W., 2006. Spatial variation of soil properties
relating to vegetation changes. Plant and Soil 280, 1–5.
Schaffrath, V.R., Tormena, C.A., Fidalski, J., Gonçalves, A.C.A., 2008. Variabilidade e
correlação espacial de propriedades físicas de solo sob plantio direto e preparo
convencional. Revista Brasileira de Ciência do Solo 32, 1369–1377.
Si, B.C., Farrell, R.E., 2004. Scale-dependent relationship between wheat yield and
topographic indices: a wavelet approach. Soil Science Society of America Journal
68, 577–587.
Sobieraj, J.A., Elsenbeer, H., Coelho, R.M., Newton, B., 2002. Spatial variability of soil
hydraulic conductivity along a tropical rainforest catena. Geoderma 108, 79–90.
Souza, L.D., da Cunha Sobrinho, A.P., da Ribeiro, L.S., da Souza, L.S., da Ledo, C.A.S., 2004.
Avaliação de plantas cítricas, em diferentes profundidades de plantio, em Latossolo
Amarelo dos Tabuleiros Costeiros. Revista Brasileira de Fruticultura 26, 241–244.
Souza, Z.M., Marques Jr., J., Cooper, M., Pereira, G.T., 2006. Micromorfologia do solo e sua
relação com atributos físicos e hídricos. Pesquisa Agropecuaria Brasileira 41, 487–492.
Stevenson, F.C., Knight, J.D., Wendroth, O., Kessel, C.Van, Nielsen, D.R., 2001. A
comparison of two methods to predict the landscape-scale variation of crop yield.
Soil and Tillage Research 58, 163–181.
Stolf, R., 1991. Teoria e teste experimental de fórmulas de transformação dos dados de
penetrômetro de impacto em resistência do solo. Revista Brasileira de Ciência do
Solo 15, 229–235.
Terra, J.A., Shaw, J.N., Van Santen, E., 2006. Soil management and landscape variability
affects field-scale cotton productivity. Soil Science Society of America Journal 70,
98–107.
The Orange Book. Lund, 1998. Tetra Pak.
Thwaites, R.N., Slater, B.K., 2000. Soil-landscape resource assessment for plantations—a
conceptual framework towards an explicit multi-scale approach. Forest Ecology
and Management 138, 123–138.
Tomer, M.D., Cambardella, C.A., James, D.E., Moorman, T.B., 2006. Surface-soil
properties and water contents across two watersheds with contrasting tillage
histories. Soil Science Society of America Journal 70, 620–630.
Topp, G.C., Zebchuk, W., 1979. The determination of soil–water desorption curves for
soil cores. Canadian Journal of Soil Science 59, 19–26.
Trangmar, B.B., Yost, R.S., Uehara, G., 1985. Application of geostatistics to spatial studies
of soil properties. Advances in Agronomy 38, 54–94.
Troeh, F.R., 1965. Landform equations fitted to contour maps. American Journal of
Science 263, 616–627.
USDA, 1993. Soil Survey Division Staff. Soil survey manual. Washington, (Agriculture
handbook, 18).
Viera, S.R., Nielson, D.R., Biggar, J.W., 1981. Spatial variability of field-measured
infiltration rate. Soil Science Society of America Journal 45, 1040–1048.
Vieira, S.R., Hatfield, J.L., Nielsen, D.R., Biggar, J.W., 1983. Geoestatiscal theory and
application to variability of some agronomical properties. Hilgardia 51, 1–75.
Vieira, S.R., Lombardi Neto, F., 1995. Variabilidade espacial do potencial de erosão das
chuvas do Estado de São Paulo. Bragantia 54, 405–412.
Voltz, M., Webster, R., 1990. A comparison of kriging, cubic splines and classification for
predicting soil properties from sample information. Journal of Soil Science 41,
473–490.
66
D.S. Siqueira et al. / Geoderma 155 (2010) 55–66
Voltz, M., Goulard, M., 1994. Spatial interpolation of soil moisture retention curves.
Geoderma 62, 109–123.
Warrick, A.W., Nielsen, D.R., 1980. Spatial Variability of Soil Physical Properties in the
Field. In: Hillel, D. (Ed.), Applications of Soil Physics. Academic Press, New York, pp.
319–344.
Wollenhaupt, N.C., Mulla, D.J., Crawford, G., 1997. Soil sampling and interpolation
techniques for mapping spatial variability of soil properties. In: Pierce, F.J., Sadler, E.J.
(Eds.), The State of Site-specific Management for Agriculture, pp. 19–53. Madison.
Wu, J., Norvell, W.A., Hopkins, D.G., Welch, R.M., 2002. Spatial variability of grain
cadmium and soil characteristics in a durum wheat field. Soil Science Society of
America Journal 66, 268–275.
Young, F.J., Hammer, R.D., 2000. Defining geographic soil bodies by landscape position,
soil 112 taxonomy, and cluster analysis. Soil Science Society of America Journal 64,
989–998.
Zebarth, B.J., Rees, H., Walsh, J., Chow, L., Pennock, D.J., 2002. Soil variation within a
hummocky podzolic landscape under intensive potato production. Geoderma 110,
19–23.
Zhang, J., Oxley, R.R.B., 1994. A comparison of three methods of multivariate analysis of
upland grasslands in NorthWales. Journal of Vegetation Science 5, 71–76.
Zhang, H., Zhang, G.L., 2005. Landscape-scale soil quality change under different
farming systems of a tropical farm in Hainan, China. Soil Use and Management 21,
58–64.