You are on page 1of 7

J. Soil Sci. Soc. Sri Lanka, Vol.

23, 2011

ORIGINAL PAPER ACCEPTED FOR PUBLICATION IN JOURNAL OF SOIL SCIENCE SOCIETY OF SRI
LANKA
SPATIAL VARIABILITY OF SOIL TEXTURE, ORGANIC CARBON AND CATION EXCHANGE
CAPACITY OF A REDDISH BROWN LATASOLIC SOIL IN A SLOPY LANDSCAPE

V.G.D.Nayanaka, U.W.A. Vitharana* and R.B. Mapa

Department of Soil Science, Faculty of Agriculture, University of Peradeniya


* Corresponding Author: uvithara@pdn.ac.lk

ABSTRACT

Detailed and quantitative information on spatial variability of soil properties is a pre-requisite for decision making in site-
specific soil management and land use planning. Classical maps are published at a coarse scale thus they are not sufficient to
provide detailed soil spatial information requirement. Therefore, there is a need of exploring short-scale variability of soil
properties and produce detailed maps in different soilscapes. This study was conducted at the Dodangolla experimental station
of Faculty of Agriculture, University of Peradeniya to investigate the spatial variability of key soil properties in a selected slopy
landscape and to produce digital maps of soil variability. A seven hectare pasture land that consisting of Reddish Brown
Latasolic Great Soil Group (Ultisols) was selected for the study. Clay, sand, organic carbon and cation exchange capacity
(CEC) of top (0-30 cm) and subsoil (30-60 cm) were determined at 55 sampling locations identified using a combined grid and
random sampling scheme. The variogram analysis indicated high to moderate spatially structured variability of top and subsoil
properties. It was also evident that a large proportion of total variation (52% to 100%) was accounted by the spatially
structured component. Range of spatial dependencies of topsoil varied from 75 m to 95 m whereas that of subsoil properties
ranged from 68 m to 81 m. These variogram parameters could be used to develop optimum sampling schemes for detailed
characterization of these key soil properties in a similar soilscape. Digital map layers of each soil property further explained
the short-scale variability of the area. This short-scale variability can be used to implement sustainable land management
practices while minimizing the detrimental effects on environment. This study also revealed that short-scale variability of soil
properties is governed by the landscape processes.

Keywords: site-specific soil management, sampling scheme, Variograms, Geo-statistics

INTRODUCTION processes can greatly influence the reliability of results


(Burrough, 1983; Foussereau et al., 1993; Wilding et al.,
Spatial variability of soil properties has been recognized for 1994).
many years (Burrough, 1993) and this could be divided into
two main categories, namely systematic and random Traditionally, the information on spatial variation of soil
variability. Systematic variation is a gradual change of soil properties are inventoried as area-class maps which are
properties as a function of soil forming factors and/or soil published at coarser scales (1:250000 and 1:500000).
management. The random variation entails either spatial Therefore, these maps are not detailed enough to provide soil
difference of soil properties which cannot be explained in information for sustainable land management strategies.
terms of known soil-forming factors or resulted due to the Moreover, the soil spatial variability is not quantitatively
measurement errors at the scale of the study (Jenny, 1980; expressed by these maps (Heuvelink and Webster, 2001).
Quine and Zahng, 2002). Systematic variation leads to Besides, producing detailed soil maps by conventional soil
develop short-scale variability of soil properties which can be surveying methods are expensive and laborious as it requires
quantified using spatial analysis and mapping. Traditional soil large number of field observations. Therefore, there is a need
management practices do not account for the short scale to construct quantitative (continuous) maps of key soil
variability of soil properties and only considers that the soil properties by means of accurate and cost-effective methods.
properties are more homogenous within and among Classical statistical techniques are not sufficient to
agricultural fields (Vitharana, 2008). Moreover, many characterize the spatial variability of soil (Goovaerts, 1997).
researchers have revealed that the information on short scale In this context, geo-statistics is considered as an alternative
variability of soil properties can be used as a basis for approach in which spatial correlation (autocorrelation) of
sustainable land management strategies such as site specific variables those can be quantified through variogram analysis
soil management, process based land use planning and (Webster & Oliver, 2001) and subsequent accurate mapping
environmental modeling. Cahn et al. (1994) emphasized that of key soil properties. The objectives of this study were to
the spatial patterns of soil properties and nutrient quantify the spatial variability of key soil properties which
concentrations need to be characterized at detailed level to influence crop production through variogram analysis, use
develop site-specific soil management practices. Further, this information to produce accurate maps by means of
accuracy of soil spatial information as inputs for many ordinary kriging and to identify the relationships between
logical, empirical and physical models of soil-landscape variability of soil properties and topography.
Spatial Variability of Soil Properties of a Reddish Brown Latasolic Soil. Nayanaka et.al.

MATERIALS AND METHODS median, mean, standard deviation (SD), coefficient of


skewness and coefficient of variation (CV).
Study area
Study area (Fig. 1) was a seven hectare pasture land in the Spatial analysis
University experimental station, Dodangolla at Kundasale Spatial analysis involved the calculation of omnidirectional
(central coordinates: 717' 51" N, 80 42' 12" E) which has experimental variograms and subsequent fitting of theoretical
been previously used for annual crop cultivation. Reddish models. Variowin spatial analysis software (Pannatier, 1997)
Brown Latasolic soils (Ultisols: taxonomic equivalent) is the was used for spatial analysis. The strength of spatial
soil great group found in this area (Pushpananda, 1985). variability of each soil property was explained using the
variogram model parameters. Further, Relative nugget effect
(RNE) which is the ratio between nugget variance to sill
variance was calculated to understand the degree of spatially
structured variability. Ordinary point kriging interpolation
technique was used to create continuous digital maps of each
soil property using GSLIB software (Deutsch and Journel,
1998).

RESULTS AND DISCUSSION

Descriptive statistics of the soil properties are given in the


Table 1. The Kolmogrov-Smirnov normality test indicated
that top and subsoil organic matter content, cation exchange
capacity, sand content were normally distributed whereas top
and subsoil clay were not normally distributed at 5%
probability level. This was further explained by the
corresponding coefficient of skewness values (Table 1).

Topsoil clay content ranged from 3.5% to 30.9% whereas that


of the subsoil varied from 3.5% to 26.3% indicating a similar
variability of clay fractions in the two soil layers. This is also
reflected with almost similar CV values for the top and the
subsoil clay fractions (Table 1). Both top and subsoil sand
Fig. 1. Satellite image showing the Dodangolla University fractions had almost similar variability as reflected by CV
Farm (top) and the study area (bottom) (yellow line shows values. Three textural classes namely sandy loam, loamy sand
the boundary) and sandy clay loam were identified for the topsoil according
to the USDA soil textural triangle whereas four textural
Soil sampling and laboratory analysis classes ranging from sandy loam to sand were observed for
A coupled random and grid sampling scheme was used to
the subsoil. Topsoil organic matter content ranged from
obtain total of 55 soil samples (Fig. 2) from both top soil (0 -
0.99% to 2.41% whereas that of subsoil ranged from 0.20% to
30cm) and sub soil (30 - 60cm). First, half of the sampling
2.16% indicating a larger variability of subsoil in compared
points were located on each grid node of a 50 m regular grid
with the topsoil. According to the classification by Hillel
and the remaining sampling points were assigned as random
(1980) all the properties showed a medium variability
pairs within each grid cell. All sampling locations were geo-
(CV=10-100%).
referenced using a Garmin GPS (Garmin Inc. USA) receiver.
Three sub samples were taken from both top and subsoil Spatial variability of soil properties
within a one meter radius at each sampling location. Sub Experimental variograms of top and subsoil clay, sand,
samples were bulked to obtain a composite sample for top organic matter and CEC and theoretical models fitted are
and subsoil. Soil samples were air dried and sieved through given in the Fig. 3. Spherical model was found to be the best
2 mm sieve prior to laboratory analysis. Soil samples were fit for all experimental variograms of all properties. The
analyzed for soil texture (Gee and Or, 2002), CEC (Summer corresponding model parameters are listed in Table 2. The lag
and Miller, 1996) and organic carbon (Nelson and Sommer, distance at which the variogram reaches the sill variance
1996). represents the range of spatial correlation. It is considered that
the observations within the range are spatially correlated
Exploratory data analysis (auto-correlated) whereas those spaced greater than the range
Exploratory data analysis was performed for all variables.
are considered as spatially independent (Goovaerts, 1997).
This included calculation of statistical parameters such as
J. Soil Sci. Soc. Sri Lanka, Vol. 23,2011

Table 1. Descriptive statistics for the measured soil properties

Mean Median Min Max SD CV Skewness


Topsoil
Clay (%) 12.6 11.2 3.5 30.9 6.7 53.1 0.92
Sand (%) 74.3 75.2 50.5 87.3 7.9 10.7 -0.68
Organic matter (%) 1.54 1.56 0.99 2.41 0.31 19.94 0.22
CEC (cmolc/kg) 13.84 13.90 6.50 23.78 4.32 31.19 0.32
Subsoil
Clay (%) 10.8 10.6 3.5 26.3 5.5 50.8 0.53
Sand (%) 74.2 74.1 57.3 90.3 8.2 11.1 0.05
Organic matter (%) 1.26 1.35 0.20 2.16 0.56 43.96 -0.1
CEC(cmolc/kg) 14.13 13.40 7.0 24.40 4.23 29.92 0.34

Range of spatial dependencies for the topsoil properties property. Cambardella et al. (1994) proposed the use of
varied from 75 m to 95 m whereas that of the subsoil varied RNE% to quantify the degree of spatial dependency of a
from 68 m to 81 m. Webster and Oliver (2001) suggested that variable. Accordingly, top soil clay, subsoil clay and CEC
the range value of a variogram can be used to find out the showed strong spatial dependency (RNE <25%) whereas
maximum sampling interval. These results indicate that the other properties showed moderate level (RNE = 25% -75%)
samples should be taken within the sampling interval of less of spatial dependencies. Top and subsoil clay showed a
than 68 m for detailed characterization of these soil nugget variance of zero indicating a negligible level of
properties. Further, when spatial variability of individual soil random variation. According to Cambardella et al. (1994)
properties is concerned, corresponding ranges can be strong spatially dependent properties are controlled by the
considered when designing sampling schemes for each soil intrinsic factors such as soil forming processes.

Table 2. Model parameters of the fitted variogram models of top and subsoil properties

Property Model Variogram parameters


Nugget Sill Range(m) RNE*%
Topsoil
Clay Spherical 0.00 43.99 80.6 0.00
Sand Spherical 19.43 68.34 75.6 28.43
OM Spherical 0.03 0.09 95.4 33.33
CEC Spherical 2.09 18.24 94.5 11.45
Subsoil
Clay Spherical 0.00 30.00 67.5 0.00
Sand Spherical 18.75 67.66 72.0 27.71
OM Spherical 0.15 6.31 81.4 48.38
CEC Spherical 5.94 17.28 78.0 34.37
* Relative nugget effect

(a) (b)
Semivariance

Semivariance

Fig. 3. Omnidirectional experimental variograms (dots) fitted models (curves) for top soil (a) clay, (b) sand, (c) organic
matter, (d) CEC and subsoil (e) clay (f) sand (g) organic matter (h) CEC
Spatial Variability of Soil Properties of a Reddish Brown Latasolic Soil. Nayanaka et.al.

(c) (d)
Semivariance

Semivariance
(e) (f)
Semivariance

Semivariance
(g) (h)
Semivariance

Semivariance

Fig. 3. Continued

Digital maps of topsoil properties draped over the Digital High topsoil organic matter content (1.7% to 2.7%) was
elevation model (DEM) of the study area are presented in the observed in a part of the southern corner of the field where
Fig. 4a to d to explain the spatial variability of soil properties enrichment of clay was also evidenced (Fig. 4c). Many
and their relationships with the landscape. researchers have documented that the low lying landscape
positions are likely to accumulate organic materials due to
Topsoil clay map (Fig. 4a) showed a large variability across
slow decomposition rates caused by anaerobic conditions and
the hilly landscape. Large part of the area had moderate clay
deposition of organic material from upper landscape
content (10% to 20%) whereas smallest clay contents (<5%)
positions. Furthermore, organic matter content can increase
were observed in the north-western corner of the field which
with clay content as organic matter form organo-mineral
is the highest position of the field. All most opposite spatial
complexes which are resistant to decomposition (Brady and
distribution was observed for the topsoil sand content (Fig.
Weil, 1998). However, the spatial distribution of topsoil
5b). Moor et al. (1993) and Gessler et al. (2000) have
organic matter did not significantly coincide with the clay
observed distributions of large sand content and smaller clay
contents of the other parts of the field. Poor correlation
contents in the upper positions (crest and shoulder positions)
between topsoil organic matter and clay (r = 0.1) also
of the landscape. Water erosion occurring in the hilly
confirmed the absence of a strong linear relationship.
landscape is the reason for such spatial distributions of sand
According to the crop history, this land has been used for
and clay fractions in the upper positions of the field. Finer
annual crop cultivation and later it has been converted to a
particles such as clay and silt are more susceptible to erosion
pasture land. Therefore, addition of organic manure may have
by water compared to the sand particles. Therefore, such finer
masked the distribution of soil organic carbon under natural
particles can be removed from upper landscape positions and
conditions. Spatial distribution of CEC was similar to that of
deposited in low-lying areas in the hilly landscape. This study
the organic carbon (Fig. 4d). Larger CEC (15 cmolc/Kg to 24
clearly revealed this by showing a comparatively large clay
cmolc/Kg of soil) was observed in the same low lying
contents (20% to 30%) in the center and the bottom part of
landscape positions where organic matter was also high (Fig.
the field (Fig. 4a) and lower clay contents at the upslope
4c). Comparatively low CEC values were distributed in
areas.
northwestern corner (hill top) and the ridges in the center
J. Soil Sci. Soc. Sri Lanka, Vol. 23,2011

positions of the landscape. Larger sand content (>75%) was to that of topsoil clay (Fig. 5a and b). Further, compared to
also observed in those areas in field (Fig. 4b). it is a well topsoil, areas with low clay content (<5%) appeared
known fact that soil texture is one of the important factors prominently in the subsoil indicating a minimum disturbance
contributing to the cation exchange capacity in soil. Sand of previous management practices to subsoil clay. Fig. 5c
fraction of the texture usually consists of primary minerals shows that the low-lying area in the southern corner of the
such as Quartz and Feldspars which have low CEC compared field and concave slope positions (north-east direction) had
with the clay minerals (Brady and Weil, 1998). Therefore, high subsoil organic matter content (1.5% to 2.1%) whereas a
both organic matter and sandy texture can be the possible small area with low organic matter (<0.5%) was noticed in
reason for spatial distribution of CEC in the field. Correlation the highest position of the landscape at the north western
coefficients of 0.5 and -0.4 between CEC and organic matter corner.
and CEC and sand also supported this evidence. Subsoil clay
and sand contents showed small difference in spatial patterns

Fig. 4. Digital map layers of topsoil (a) clay%, (b) sand%, (c) organic matter%, (d) and CEC cmol(+)/Kg (g) draped on the
digital elevation model

Many researchers including Moor et al. (1993) and Gessler et concave area in north-eastern corner and the area along the
al. (2000) documented high organic matter associated with western corner were identified with large CEC values (Fig.
footslope and concave areas in landscape. However, it is 5d). Comparatively high organic matter and clay content in
important that such soil-landscape relationships need to be low lying area of northern corner and concave area of north
quantified in order to use such relationships to map soil eastern corner (Fig. 5a and c) had attributed to high CEC
properties accurately. Low-lying area in the southern corner, values in those areas.
Spatial Variability of Soil Properties of a Reddish Brown Latasolic Soil. Nayanaka et.al.

Fig. 5. Digital map layers of subsoil (a) clay%, (b) sand%, (c) organic matter%, (d) and CEC cmol (+)/Kg (g) draped on the
digital elevation model
Burrough, P.A., 1993. Fractals and Geostatistical methods in
CONCLUSIONS landscape studies. p. 87-112. In: N. Lam and L. de Cola
(ed.) Fractals in geography. Prentice Hall, Englewood
This study was an attempt to quantify the spatially structured Clifts, NJ.
variability of key soil properties for crop production by means Burrough, P.A., 1983. Multi-scale sources of spatial variation in
of geo-statistical analysis and use of information to produce soil: The application of fractal concepts to nested levels
detailed and accurate maps. The variogram analysis showed of soil variation. Journal of Soil Science 34: 577597.
the presence of structured spatial variability of soil properties. Cahn, M.D., J.W. Hummel and B.H. Brouer. 1994. Spatial
The spatial dependencies of topsoil ranged from 75 m to 95m Analysis of Soil Fertility for Site Specific Crop
whereas that in subsoil properties exhibited a range of 68 m to Management. Soil.Sci. Soc. Am.J. 58:1240-1248.
Cambardella, C.A., T.B. Moorman, J.M. Novak, T.B. Parkin,
81 m. Digital maps exhibited a clear spatial variability of soil
D.L. Karlen, R.F. Turco, and A.E. Konopka. 1994.
properties further strengthening the information revealed by
Field-scale variability of soil properties in central Iowa
variograms. Further, spatial distribution of all soil properties
soils. Soil Sci. Soc. Am. J. 58:1501-1511.
except for soil organic matter largely resembled the general
Deutsch, C.V. and A.G. Journel. 1998. GSLIB, Geostatistical
topography of the slopy landscape. This indicated the
software library and user guide, 2ed. Oxford University
presence of satisfactory soil-landscape relationships which press. NY.
can be used for cost-effective characterization of the Foussereau, X., A.G. Hornsby, and R.B. Brown. 1993.
investigated soil properties. This study clearly showed that Accounting for variability within map units when
the within field variability of soil properties is significant and linking a pesticide fate model to soil survey. Geoderma
should be considered in site-specific soil management and 60: 257276.
land use planning. Gee, G.W., and J.W. Bauder. 2002. Particle size analysis, p. 383-
411, In A. Clute, (ed.) Methods of soil analysis, part 1.
REFERENCES Physical and mineralogical methods. Agronomy
monograph 9. American Society of Agronomy,
Brady, N. C. and R. R. Weil. 1998. The nature and properties of Medison, WI.
soils: R.R. Weil (ed.). A Simon and Schuster company, Gessler, P. E., O. A. Chadwick, F. Chamran, L. Althouse and K.
Prentice Hall, New Jersey. Holmes. 2000. Modeling SoilLandscape and
J. Soil Sci. Soc. Sri Lanka, Vol. 23,2011

Ecosystem Properties Using Terrain Attributes, Soil Sci.


Soc. Am. J. 64:20462056.
Goovaerts, P. 1997. Geostatistics for natural resources
evaluation. Oxford University Press, NY.
Heuvelink, R., and R. Webster. 2001. Modelling soil variation:
past, present, and future. Geoderma 100: 269-301.
Hillel. D. 1980. Application of soil physics, Academic press.
New York.
Jenny, H., 1980. The soil resource: origin and behaviour.
Ecological studies. Springer-Verlag, New York. 37:377.
Moore, I.D., P.E. Gessler, G.A. Nielsen, and G.A. Peterson.
1993. Soil attribute prediction using terrain analysis.
Soil Sci. Soc. Am. J. 57:443-452.
Nelson, D. W. and L. E. Sommer. (1996). Total carbon, organic
carbon and organic matter. P.961-1010. In A. L. Page
(ed.) Method of soil science analysis, part 3, chemical
methods: American society of agronomy, soil science
society of America, Madison, Wisconsin, USA.
Pannatier, Y. 1997. VARIOWIN: software for spatial data
analysis in 2D. Springer, NY.
Pushpananda, K. D. 1985. Semi-detailed soil survey of
university farm at Dodangolla. (II) Soil physical
characteristics. Unpublished B.Sc. thesis. University of
Peradeniya.
Quine, T.A. and Y. Zhang. 2002. An investigation of spatial
variation in soil erosion, soil properties and crop
production within an agricultural field in Devon, U.K.
Journal of soil and water conservation. 57: 50-60.
Summer, M. E. and W. P. Miller. (1996). Cation exchange
capacity and exchange coefficient. p. 1201-1229. In D.
L. Sparks (ed.) Method of soil science analysis, part 3,
chemical methods: American society of agronomy, soil
science society of America, Madison, Wisconsin, USA.
Vitharana, W.A.U. 2008. Spatial inventory techniques in support
of site-specific soil management. Ph.D. Thesis. Ghent
University, Ghent.
Webster, R., and M.A. Oliver. 2001. Geostatistics for
environmental scientists. John Wiley and Sons,
Chichester.
Wilding, L.P., J. Bouma, D. Goss. 1994. Impact of spatial
variability on modeling. P.61-75. In Bryant, R., and
R.W. Arnold. (ed.) Quantitative Modeling of Soil
Forming Processes.

You might also like