Vous êtes sur la page 1sur 9

Catena 165 (2018) 251–259

Contents lists available at ScienceDirect

Catena
journal homepage: www.elsevier.com/locate/catena

Spatial variability of soil properties and delineation of soil management T


zones of oil palm plantations grown in a hot and humid tropical region of
southern India

Sanjib K. Beheraa, , Ravi K. Mathura, Arvind K. Shuklab, K. Suresha, Chandra Prakashb
a
ICAR-Indian Institute of Oil Palm Research, Pedavegi, West Godavari, Andhra Pradesh 534450, India
b
ICAR- Indian Institute of Soil Science, Nabibagh, Berasia Road, Bhopal, Madhya Pradesh 462038, India

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

Keywords: Sustainable soil nutrient management with proper understanding of spatial variability of soil properties helps in
Site-specific nutrient management enhancement of crop productivity and avoiding soil degradation. It has more importance in oil palm plantations,
Precision agriculture since the area under oil palm cultivation is on rise globally. Moreover, the crop is a nutrient-requiring one and
Spatial distribution effective nutrient management contributes about 50% of fresh fruit bunch (FFB) production. Therefore, the
Principal component analysis
present study was carried out to assess spatial distribution of soil properties and to delineate soil management
Variable application rate
zones (MZs) in oil palm plantations of a hot and humid tropical region of India for efficient soil nutrient
management. A total of 180 geo-referenced representative soil samples (from 0 to 0.20 m depth) were collected
from oil palm plantations of Pedavegi and Denduluru mandals of west Godavari district of Andhra Pradesh, India.
Collected soil samples were processed and analysed for soil properties like pH, electrical conductivity (EC), soil
organic carbon (SOC), available phosphorus (P), available potassium (K), exchangeable calcium (Ca), ex-
changeable magnesium (Mg), available sulphur (S) and available boron (B). The values of soil properties varied
widely with low (7.7%) to moderate (29.0 to 77.4%) coefficient of variations. Semivariogram analysis and
ordinary kriging revealed varied spatial distribution pattern with moderate to strong spatial dependence for
most of the soil properties. Development of the MZs was carried out by principal component (PC) analysis and
fuzzy c-means clustering. Three PCs with eigen values > 1 and accounting 60.31% of total variance were used
for further analysis. On the basis of fuzzy performance index and normalized classification entropy, three MZs
were identified. The MZs differed significantly with respect to studied soil properties. Thus, the study empha-
sized that the methodology for delineating MZs could be effectively used for site-specific soil nutrient man-
agement in oil palm plantations and other crops for maximizing crop production in the study area.

1. Introduction spatial distribution of soil properties can be carried out by geostatistical


methods (Mueller et al., 2003). According to Saito et al. (2005), pre-
Soil is a heterogeneous body and key to ecosystem processes which diction of values at un-sampled locations can be done by geostatistical
govern nutrient cycling (Fitter et al., 2005). Sustainable soil manage- estimation by considering spatial correlation between estimated and
ment with proper understanding of soil properties helps in maintaining sampled points and reducing estimation error and involved costs. Ad-
or improving the level of soil fertility and avoiding soil degradation, dressing soil heterogeneity by delineation of management zone (MZ) of
which is a global problem of importance (Thapa and Yila, 2012; Zhao soil is a technique, in which an area is divided into different zones
et al., 2013). Soil properties vary spatially due to combined impact of having homogenous characters (Ortega and Santibanez, 2007; Xin-
physical, chemical and biological processes operating in soil along with Zhang et al., 2009; Peralta et al., 2015). Several researchers have de-
human/animal activities (Goovaerts, 1998). Proper understanding of lineated soil MZs in different agro-ecosystems including different crops
the spatial distribution of soil properties and their mapping is the key to for site-specific soil management using geostatistics, principal compo-
site-specific soil management for sustainable crop production by vari- nent analysis (PCA) and fuzzy c-means classification (Davatgar et al.,
able-rate application of nutrients (Behera and Shukla, 2015; Brevik 2012; Tripathi et al., 2015; Shukla et al., 2017; Nawar et al., 2017).
et al., 2016; Shukla et al., 2017; Bogunovic et al., 2017a). Assessment of Oil palm is the highest oil producing crop of the world, having


Corresponding author.
E-mail addresses: sanjibkumrbehera123@gmail.com, Sanjib.Behera@icar.gov.in (S.K. Behera).

https://doi.org/10.1016/j.catena.2018.02.008
Received 2 June 2017; Received in revised form 6 February 2018; Accepted 7 February 2018
0341-8162/ © 2018 Published by Elsevier B.V.
S.K. Behera et al. Catena 165 (2018) 251–259

yielding potential of 18.8 t oil ha−1 year−1 (Hoffmann et al., 2014). To 2.2. Collection of soil samples, processing and analysis
fulfil the demand of vegetable oil requirement, it is being cultivated in
28 developing countries of the world having significant level of palm oil Altogether 180 surface (0 to 0.20 m depth) soil samples were col-
production (Anonymous, 2017). In India, it presently occupies 0.3- lected from oil palm plantations situated at an interval of 1 to 2 km
million-hectare area in 14 states of the country as small holders' plan- within the study area. Soil samples were collected using hand auger
tation crop. However, an area of 19.3 million ha covering 19 states is from 3 m radius weeded palm basins. Five sub-samples were collected
having potential for oil palm cultivation in India (Rethinam et al., and mixed to obtain representative sample from each sampling loca-
2012). The Government of India is making efforts to enhance the area tion. Using hand-held Global Positioning System (GPS), the geo-
under oil palm cultivation in the country by planting this crop in fallow graphical coordinates viz. latitude and longitude of each sampling point
land and/or replacing less remunerative crops. Out of the 0.3 million ha were noted. The collected soil samples were air-dried. Stones and debris
of cultivated area, the west Godavari district of Andhra Pradesh lying in present in samples were removed. Grinding of soil samples was carried
hot and humid tropical climate is having 70, 000 ha area under oil palm out by using wooden pestle and mortar and then passed through a 2 mm
cultivation. sieve. The processed soil samples were stored in polyethene bottles for
Among the several factors influencing oil palm yield, nutrients play estimation of different properties. Soil pH and electrical conductivity
a key role as it contributes about 50% towards fresh fruit bunch (FFB) (EC) was estimated in 1:2.5 (w/v) soil and water suspensions (Jackson,
yield of oil palm (Woittiez et al., 2017). The nutrient requirement of oil 1973). Soil organic carbon content is an important and manageable soil
palm is relatively higher and it requires a large amounts of nitrogen (N), property and it influences many other soil properties. Available N status
phosphorus (P), potassium (K), magnesium (Mg) and boron (B). Ac- in soil can be predicted from SOC content value (Murphy, 2015; Palmer
cording to Mengel and Kirkby (1987), oil palm requires 162 kg of N, et al., 2017). The content of SOC was estimated by Walkley and Black
30 kg of P, 217 kg of K, 38 kg of Mg and 36 kg of calcium (Ca) to pro- method (Walkley and Black, 1934). Available P (Olsen et al., 1954),
duce 2.5 t oil ha-1 year-1 (equivalent to 10 t FFB ha-1 year-1, considering available K (Hanway and Heidel, 1952), exchangeable Ca (Jones Jr.,
oil to bunch ratio of 1:4). The nutrient-limited yield is site-specific and 1998), exchangeable Mg (Jones Jr., 1998), available S (Williams and
it is predominately owing to the influence of soil properties on avail- Steinbergs, 1969) and available B (Gupta, 1967) concentration in soil
ability of nutrients. Nutrient disorders like N/K imbalance, K, Mg and B samples were estimated by respective methods.
deficiencies in oil palm plantations adversely affect oil palm production
in India (Rao et al., 2014). Out of several soil properties, soil pH and 2.3. Statistical, geostatistical, principal component and fuzzy cluster
soil organic carbon (SOC) content mostly affect the nutrient availability algorithm analysis
by influencing chemical environment of soil (Tisdale et al., 1985). Soil
pH affects nutrient availability, whereas SOC modifies physical and Using SAS 9.2 software pack (SAS, 2011), the descriptive statistics
biological environment and chemical reactions in soil. Fertilizer re- revealing minimum, maximum, mean, standard deviation (SD) and
quirement in oil palm is effectively assessed by carrying out soil and coefficient of variation (CV) of studied soil properties was obtained. The
leaf analysis (Goh et al., 2003) as soil is the source of nutrients for oil test of normality for soil properties was also performed using the Kol-
palm and the leaf nutrient concentration and palm productivity is di- mogorov-Smirnov test (K-S test) and all the soil properties passed the
rectly correlated. It is therefore essential to assess soil properties of oil test. Pearson's correlation coefficient values demonstrating relationship
palm plantations for nutrient management in a sustainable manner. among the soil properties were obtained. For assessment of spatial
The spatial variability of soil properties in oil palm plantations of distribution pattern of soil properties, ArcGIS 10.4.1 software was used
India is expected to be high mainly owing to adoption of different land and semivariogram for each soil property was calculated using Eq. (1)
management practices and varied amount of fertilizer application as given below.
(Prasad et al., 2013). The information regarding spatial variability of
N(h)
soil properties and soil management zones in oil palm plantations of 1
India is limited. Keeping above facts in view, the present study was
γ(h) =
2N(h)
∑ [Z(X α + h)]2
α=1 (1)
carried out (i) to assess spatial distribution of the soil properties by
using the geostatistics and (ii) to delineate MZs in the oil palm plan- where γ(h), N (h), z(xα) and z(xα + h) represent semivariance for the
tations of a hot and humid tropical region of west Godavari district of lag distance h, number of sample pairs separated by the lag distance h,
Andhra Pradesh, India. measured value at αth sample location and measured value at point
α + hth sample location, respectively. Semivariogram models like cir-
cular, spherical, K-Bessel, exponential, stable and Gaussian were eval-
2. Materials and methods uated to determine the best-fit one for each soil property. The semi-
variogram models were chosen by using the cross validation technique,
2.1. Study area i.e. by comparing the actual values with the values estimated by kriging
using the semivariogram model. Prediction accuracy of semivariogram
The study was carried out in Pedavegi and Denduluru mandals of models was evaluated by mean square error (MSE).
west Godavari district of Andhra Pradesh, India situated at 16°07′N and n
∑i = 1 [z (x i , yi ) − z∗ (x i , yi )]2
81°01′E (Fig. 1). The area experiences hot and humid tropical climate MSE =
n (2)
receiving average rainfall of 950 mm. A significant portion of rainfall is
received during June to September. The mean highest (39 °C) and the where, n is the number of observation for each case, z(xi, yi) is the
mean lowest (23 °C) temperature prevails in the month of May and observed soil parameter, z ∗ (xi, yi) is the estimated soil property, and
December respectively. The mean highest (90%) and the mean lowest (xi, yi) are sampling coordinates.
(60%) relative humidity is obtained in the month of July–August and Interpolation mapping to estimate the values of soil properties at
February respectively. The area is having igneous, metamorphic and un-sampled locations was done using ordinary kriging (OK) procedure
sedimentary geological formations. Soils of the study area belong to (Goovaerts, 1998). Ordinary kriging was used because it is more reli-
orders Entisols, Alfisols and Vertisols (Soil Survey Staff, 2014) with able than other methods based on MSE (Meul and Van Meirvenne,
sandy clay loam to loamy sand in texture. Oil palm was grown as irri- 2003). Furthermore, OK is the best unbiased predictor for the random
gated crop and planted at 9 m × 9 m × 9 m distance viz. equilateral process as specific unsampled locations in case of sparse and randomly
triangular planting. Fertilizers were applied in 3 m radius weeded palm selected soil sampling. It has also an additional benefit of reducing
basins for four times a year. influence of outliers (Triantafilis et al., 2001). Goodness-of-prediction

252
S.K. Behera et al. Catena 165 (2018) 251–259

Fig. 1. Location of the study area (Pedavegi and Denduluru mandals of West Godavari district of Andhra Pradesh, India) with sampling points.

criterium (G) values were also estimated for checking the accuracies of Table 1
interpolated maps (Agterberg, 1984; Tesfahunegn et al., 2011). Positive Descriptive statistical parameters of soil properties of the study area.
G values indicate that the map obtained by interpolating data from the
Soil properties Minimum Maximum Mean SD CV (%)
samples is more accurate than the area average (Parfitt et al., 2009).
Negative and close to zero G values indicate that the area average pH 5.44 8.26 7.31 0.57 7.72
predicts the values at unsampled locations as accurately as or even EC (dS m-1) 0.07 0.48 0.23 0.09 40.6
SOC (%) 0.27 1.43 0.82 0.25 30.46
better than the sampling estimates.
P (kg ha-1) 9.24 105.0 53.4 22.1 41.3
Principal component analysis (PCA) is a multivariate analysis that K (kg ha-1) 12.9 443 160 109 68.3
produces new orthogonal variables called principal components (PCs) Ca (mg kg-1) 439 1609 914 265 29.0
from data set using orthogonal transformation. The correlation analysis Mg (mg kg-1) 48.0 432 231 89.6 38.7
values were used as input for PCA. Out of many PCs, PCs having more S (mg kg-1) 9.55 80.8 40.3 14.3 35.4
B (mg kg-1) 0.15 11.2 3.24 2.51 77.4
than one eigen values were considered to delineate MZ. A bi-plot was
used to visualize the grouping of soil properties. The commonly used EC: electrical conductivity; SOC: soil organic carbon; P: available phosphorous; K:
fuzzy c-means classification method was adopted to partition the da- available potassium; Ca: exchangeable calcium; Mg: exchangeable magnesium; S: avail-
tasets into different clusters having common trait (Berget et al., 2008). able sulphur; B: available boron; SD: standard deviation; CV: coefficient of variation.
The field was divided into two to seven clusters by using FuzME soft-
ware (with settings of maximum iteration = 300, minimum zones = 2, available S and B varied from 439 to 1609 mg kg-1 (with mean value of
maximum zones = 7, stopping criterion = 0.0001, and fuzziness ex- 914 ± 265 mg kg-1), 48.0 to 432 mg kg-1 (with mean value of
ponent = 1.5) (Minasny and McBratney, 2006). Maximum of seven 231 ± 89.6 mg kg-1), 9.55 to 80.8 mg kg-1 (with mean value of
clusters was considered for practical use as MZs. Fuzzy performance 40.3 ± 14.3 mg kg-1) and 0.15 to 11.2 mg kg-1 (with mean value of
index (FPI) and normalized classification entropy (NCE) estimated as 3.24 ± 2.51 mg kg-1) respectively. This is in line with the findings of
given below were used to get optimum number of clusters. Behera et al. (2016) who reported mean values of 5.35, 0.13 dS m-1,
c n 19.8 g kg-1, 270 mg kg-1, 24.7 mg kg-1, 914 mg kg-1, 203 mg kg-1,
c ⎡ ∑ ∑ (μik )2 ⎤ 23.2 mg kg-1 and 0.70 mg kg-1 for soil pH, EC, OC, NH4OAc-K, Bray's-P,
FPI = 1 − s ⎢1 − i=1 k=1 ⎥
c − 1⎣ n ⎦ (3) exchangeable Ca and Mg, CaCl2-S and hot water soluble B respectively,
in surface soils of oil palm plantations of west coastal area of India.
n c
n ⎡ ∑k = 1 ∑i = 1 μik log a (μik ) ⎤ Behera et al. (2015) also recorded mean values of 6.94, 0.53 dS m-1,
NCE = − 11.6 g kg-1, 179 mg kg-1, 92.9 mg kg-1, 820 mg kg-1, 159 mg kg-1,
n − c⎢
⎣ n ⎥
⎦ (4)
21.8 mg kg-1 and 5.81 mg kg-1 for soil pH, EC, OC, NH4OAc-K, Olsen-P,
where, c, n, μik and loga indicate number of cluster, number of ob- exchangeable Ca and Mg, CaCl2-S and hot water soluble B respectively
servations, fuzzy membership and natural logarithm, respectively. FPI in surface soils of oil palm plantations of Karnataka state of India.
estimates degree of fuzziness and NCE measures amount of dis- Alvares et al. (2011) reported pH, soil organic matter (SOM), ex-
organization created by specific classes. Optimum cluster number was changeable K, Ca and Mg values of 3.4 to 5.0, 1.0 to 56.0 g kg-1, 0.1 to
obtained with lowest FPI and NCE values. Differences among the MZs 15.3 mmolc kg-1, 0.5 to 49.0 mmolc kg-1and 0.3 to 25.0 mmolc kg-1 re-
were tested by variance analysis. spectively in Eucalyptus grown soils of Capao Bonito of Brazil. The CV of
soil properties varied from low (< 10%) to moderate (10 to 100%)
3. Results and discussion according to the criteria given by Nielsen and Bouma (1985). The CV
value was the lowest (7.72%) for pH and the highest (77.36%) for
3.1. Variability of soil properties available B. Our findings are in agreement with the observations made
by Tesfahunegn et al. (2011) who reported CV values of 8.6, 62.7, 14.2,
The values of studied soil properties varied widely (Table 1). The 30.4 and 23.9% for soil pH, OC, exchangeable K, Ca and Mg respec-
mean values were 7.31 ± 0.57 (varied from 5.44 (acidic) to 8.26 (al- tively in soils of northern Ethiopia. The CV values of 43.0, 80.0 and
kaline)), 0.23 ± 0.09 dS m-1 (varied from 0.07 to 0.48 dS m-1), 27.0 for organic matter content and available P and K respectively were
0.82 ± 0.25% (varied from 0.27 to 1.43%), 53.4 ± 22.1 kg ha-1 recorded by Zhang et al. (2007) in black soils of northeast China. Wide
(varied from 9.24 to 105 kg ha-1) and 160 ± 109 kg ha-1 (varied from variability in soil properties was also reported in soils of Lesotho
12.9 to 443 kg ha-1) for soil pH, EC, SOC and available P and K re- (Nikheloane et al., 2012) and Turkey (Kilic et al., 2012). This variability
spectively. Whereas, the values for exchangeable Ca and Mg and in soil properties is attributed to combined effect of soil types, climatic

253
S.K. Behera et al. Catena 165 (2018) 251–259

Table 2
Pearson's correlation coefficient values showing relationship among soil properties of the study area.

Variable PH EC SOC P K Ca Mg S B

PH 1.000
EC -0.139⁎ 1.000
SOC -0.084 0.338⁎⁎ 1.000
P 0.059 0.139⁎ -0.032 1.000
K -0.058 0.534⁎⁎ 0.286⁎⁎ 0.199⁎⁎ 1.000
Ca 0.091 0.154⁎ 0.308⁎⁎ 0.161⁎ 0.367⁎⁎ 1.000
Mg 0.206⁎⁎ -0.040 0.088 0.037 0.194⁎⁎ 0.372⁎⁎ 1.000
S -0.018 0.207⁎⁎ 0.232⁎⁎ 0.359⁎⁎ 0.325⁎⁎ 0.233⁎⁎ 0.186⁎⁎ 1.000
B -0.015 0.617⁎⁎ 0.254⁎⁎ 0.114 0.639⁎⁎ 0.117 0.135⁎ 0.227⁎⁎ 1.000

EC: electrical conductivity; SOC: soil organic carbon; P: available phosphorous; K: available potassium;
Ca: exchangeable calcium; Mg: exchangeable magnesium; S: available sulphur; B: available boron.

Correlation is significant at 0.05 level.
⁎⁎
Correlation is significant at 0.01 level.

conditions and crop-soil management practices (Mallarino et al., 1999; Table 3 and Fig. 2. The best-fit model (model with low value of error)
Foroughifar et al., 2013). was found to be circular, spherical and K-Bessel for soil pH, EC and SOC
content respectively. Available P and K and exchangeable Ca and Mg
had exponential best-fit model whereas, available S and B were having
3.2. Correlation among soil properties
stable and Gaussian best-fit model respectively. This is in line with the
findings of Ferreira et al. (2015) who recorded exponential best-fit
The values of Person's correlation coefficient demonstrating re-
model for soil pH, EC, N and organic matter in olive tree orchards of
lationship among the studied soil properties are presented in Table 2.
Portugal. Similarly, Behera et al. (2016) recorded exponential, stable,
Soil properties showing positive and negative correlation are presumed
Guassian, K-Bessel and spherical best-fit model for soil properties of oil
to exhibit similar and opposite spatial distribution pattern respectively.
palm plantations of west coastal area of India. The nugget (which re-
Soil pH was negatively and positively correlated with EC and ex-
veals the micro-variability) values were small for soil pH, EC, SOC and
changeable Mg respectively. Soil EC was positively correlated with
available B (varied from 0 to 0.78) and large for available P, K and S
SOC, available P, K, exchangeable Ca, available S and B. This is because
and exchangeable Ca and Mg (varied from 67.4 to 64,136.1). Large
soil EC is an indirect measurement of many soil properties influencing
nugget values indicated that available P, K and S and exchangeable Ca
soil fertility and crop yield (Corwin and Lesch, 2005). Soil EC is used as
and Mg were affected to a greater extent by ecological processes over a
measure to differentiate soil types for site-specific management. The
small scale (Zhang et al., 2007) and selected sampling distance could
content of SOC was positively correlated with available K, exchange-
not capture the spatial dependence well. Whereas, sill values indicating
able Ca, available S and B. Soil organic carbon is an important part of
the variance of the sampled population at large separation distance if
SOM which influences soil physical, chemical and biological properties
the data have no trend, were also higher for available P, K and S and
affecting soil nutrient availability to crops. Available P in soil was po-
exchangeable Ca and Mg. The variation in nugget and sill values of the
sitively correlated with available K and S and exchangeable Ca. Avail-
studied soil properties derive support from the observations recorded
able K was positively correlated with exchangeable Ca and Mg and
by Tesfahunegn et al. (2011) and Behera et al. (2016).
available S and B. Positive correlation was also recorded between ex-
According to Cambardella et al. (1994), nugget to sill ratio values
changeable Ca vs exchangeable Mg and available S, exchangeable Mg
of < 0.25, 0.25–0.75 and > 0.75 reveal strong (attributed to intrinsic
vs available S and B and available S vs available B. In line with the
factors), moderate (attributed to both intrinsic and extrinsic factors)
findings of the present study, Behera and Shukla (2015) reported po-
and weak (attributed to extrinsic factors) spatial dependence respec-
sitive correlation of soil pH and SOC with exchangeable K, Ca and Mg
tively. Based on nugget to sill ratio, the studied soil properties had
and among exchangeable K, Ca and Mg in some cropped acid soils of
different spatial dependences. Soil EC, SOC, available K and B had
India. Visualizing the correlations among the studied soil properties,
strong spatial dependence whereas available P, S and exchangeable Mg
PCA was carried out to derive principal sources of data variability.
had moderate spatial dependence class. Soil pH and exchangeable Ca
were having weak spatial dependence class. The strong spatial depen-
3.3. Spatial distribution of soil properties dence of the soil properties like EC, SOC, available K and B is controlled
by inherent soil characteristics like soil mineralogy and texture whereas
Parameters of sermivariograms of soil properties are given in

Table 3
Semivariogram parameters of the soil properties of the study area.

Soil properties Model Nugget Partial sill Sill Nugget/sill Range (m) SDC MSE G (%)

pH Circular 0.290 0.020 0.310 0.940 12,600 Weak 0.50 51


EC (dS m‐1) Spherical 0.003 0.010 0.013 0.230 9481 Strong 0.00 47
SOC (%) K-Bessel 0.000 0.060 0.060 0.000 5513 Strong 1.11 50
P (kg ha‐1) Exponential 348 160 507 0.690 11,129 Moderate 18.5 45
K (kg ha‐1) Exponential 1965 11,684 13,648 0.140 10,727 Strong 16.9 39
Ca (mg kg‐1) Exponential 64,136 6851 70,987 0.900 14,400 Weak 110.1 25
Mg (mg kg‐1) Exponential 2891 4948 7839 0.370 4446 Moderate 67.9 58
S (mg kg‐1) Stable 67.40 159 227 0.300 4858 Moderate 0.03 41
B (mg kg‐1) Gaussian 0.780 6.99 7.77 0.100 7830 Strong 0.01 59

EC: electrical conductivity; SOC: soil organic carbon; P: available phosphorous; K: available potassium; Ca: exchangeable calcium; Mg: exchangeable magnesium; S: available sulphur; B:
available boron; SDC: spatial dependency class; MSE: mean square error; G: goodness-of-prediction criterium.

254
S.K. Behera et al. Catena 165 (2018) 251–259

(a) pH (c) SOC


(b) EC

(d) P (e) K (f) Ca

(h) S (i) B
(g) Mg

Fig. 2. Experimental semivariograms and fitted models for (a) pH, (b) EC, (c) SOC, (d) P, (e) K, (f) Ca, (g) Mg, (h) S, and (i) B of the study area.

extrinsic factors like soil management including tillage and fertilizer respectively. Almost all portion of Pedavegi mandal and southern por-
application influence moderate and weak spatial dependence of soil tion of Dendulru mandal was having soil EC value of < 0.25 dS m−1.
properties. The range value of semivariogrm is the maximum distance Mosaic distribution pattern of SOC was recorded in the study area.
within which autocorrelation or spatial dependence exists. The range About 2, 30 and 68% of the study area were having low (< 0.50%),
values of studied soil properties varied from 4446 (exchangeable Mg) to medium (0.50 to 0.75%) and high (> 0.75%) value of SOC. Medium
14,400 m (for exchangeable Ca) (Table 3). Beyond this value, auto- level of SOC was recorded in western and central portion of the study
correlation does not exist. Large range value reveals that measured soil area. Varied SOC content in the study area is attributed to micro-cli-
properties (pH, available P, K and exchangeable Ca) are affected by mate, land use management and various biotic processes, such as pro-
natural and anthropogenic factors over greater distance than soil duction of biomass, litter deposition and their decomposition (Saha
properties (EC, SOC, exchangeable Mg, available S and B) having et al., 2014; Nadal-Romero et al., 2016; Trigalet et al., 2016). Lower
smaller ranges (Lopez-Granados et al., 2002). According to Kerry and concentration of available P and K were recorded in north-western and
Oliver (2004), soil sampling interval should be less than half the northern and southern portion of the study area respectively. Lower
semivariogram range value. Therefore, the range values obtained for concentration of exchangeable Mg was observed in northern portion of
the studied soil properties could be used to guide future soil sampling the study area. Available S exhibited varied distribution pattern in the
planning in oil palm plantations and in other crops of similar areas. study area. About 3% of the study area was having low
Hence, soil sampling interval should be less than the range values of soil (< 0.5 mg kg−1) in available B concentration. The varied distribution
properties recorded in the present study. pattern of soil properties in the study area is ascribed to different soil
The most accurate predictions for soil properties with the lowest types and adoption of various land management practices including
MSE values was carried out using cross-validation technique (Table 3). fertilizer application in the study area (Sharma et al., 2011; Bogunovic
Lowest MSE values indicate that kriging predictions of soil properties et al., 2017b). The spatial distribution maps demonstrate variability of
are closer to measured values. The accuracy of kriged interpolation soil properties in the study area and these maps could be utilized for
maps of soil properties was also measured by the G values (Table 3) site-specific soil nutrient management in the area.
which varied from 25 to 59%. This is in line with the observations made
by Mueller et al. (2003) and Tesfahunegn et al. (2011). The positive G 3.4. Soil management zones
values for the soil properties revealed more accurate prediction com-
pared to average values of the area. The procedure of ordinary kriging Principal component analysis was performed to aggregate and
was adopted for generation of distribution maps of soil properties summarize the variability in the 9 variables, retaining PCs having eigen
(Fig. 3). The soil properties exhibited different distribution in the study values > 1 and a cumulative loading of > 60%. Using this criterion,
area. Perusal of distribution map revealed that about 78 and 22% of the only first 3 PCs were considered for the final analysis, accounting for
study area were having soil pH value of 6.5 to 7.5 and 7.5 to 8.5 re- 60.31% of the total variability (Table 4, Fig. 4). Principal component 1
spectively. The study area was non-saline in nature. About 74 and 26% (PC1) explained 31.5% of the total variance, which was dominated by
area were having soil EC value of < 0.25 and 0.25 to 0.5 dS m−1 EC, SOC, available K, S and B and exchangeable Ca. Soil pH and

255
S.K. Behera et al. Catena 165 (2018) 251–259

Fig. 3. Ordinary kriged distribution maps of soil properties of the study area.

exchangeable Mg had most significant effect on the second principal (Fig. 5). Exchangeable Ca and Mg and available P and S constituted one
component (PC2), and it explained 16.3% of total variability. Further, group and soil EC, SOC and available K and B formed the second group.
third principal component (PC3) explained an additional 12.5% of the The first three PCs were chosen to MZs. Optimum numbers of MZs were
variance and was predominated by available P. In summary, PCA ag- created by using scores of three selected PCs through fuzzy c-means
gregated 9 studied properties into 3 PCs describing majority of spatial classification in FuzMe software. Two functions namely FPI and NCE
variability. This is in line with the findings of Tripathi et al. (2015) and were used to obtain optimum number of MZs (Table 5). Three MZs
Khaledian et al. (2017) who reported three and four PCs, respectively selected as the values of FPI and NCE were minimum at this number.
from PCA by aggregating and summarizing the variability of soil The resultant maps exhibiting three MZs are presented in Fig. 6. The t-
properties in eastern India and northeast Iran respectively. A bi-plot test was performed to assess the effectiveness of using combination of
analysis (PC1 vs PC2) exhibited two major groupings of soil properties PCA and fuzzy c-means clustering for characterizing the spatial

Table 4
Principal component analysis of soil properties and loading coefficient for the first three principal components.

Principal component Eigen values Component loading (%) Cumulative loading (%)

PC1 2.835 31.497 31.497


PC2 1.465 16.281 47.778
PC3 1.128 12.532 60.310
PC4 0.982 10.915 71.225
PC5 0.754 8.381 79.606
PC6 0.668 7.417 87.023
PC7 0.494 5.491 92.513
PC8 0.394 4.374 96.887
PC9 0.280 3.113 100

Principal component loading for each variable

pH EC SOC P K Ca Mg S B
PC1 ‐0.045 0.716 0.533 0.340 0.825 0.524 0.317 0.552 0.748
PC2 0.589 ‐0.447 ‐0.086 0.245 ‐0.095 0.501 0.671 0.236 ‐0.292
PC3 ‐0.060 ‐0.018 ‐0.392 0.791 ‐0.029 ‐0.240 ‐0.303 0.437 ‐0.060

256
S.K. Behera et al. Catena 165 (2018) 251–259

Fig. 4. Kriged maps of three principal components.

4 Mg
PH
3 Ca

2
P S
PC2 (16.29 %)

0
OC K
-1
B
-2
EC
-3

-4

-5
-5 -4 -3 -2 -1 0 1 2 3 4 5
PC1 (31.51 %)

Fig. 5. Principal component analysis bi-plot (PC1 vs PC2) of soil properties of the study Fig. 6. Soil management zones of the study area.
area.

MZ 1 to MZ 2 had low K availability in different proportion, where


Table 5 additional K supply would improve the oil palm productivity of the
FPI and NCE calculated for identifying the optimum clusters for the study area.
zones. Likewise, efforts must be done to improve SOC content by dif-
Class FPI NCE ferent agricultural practices in MZ 2 which is having lowest SOC con-
tent among the MZs. Hence, based on the values of soil properties in
1 0.213 0.237 different MZs, different management options can be devised for dif-
2 0.167 0.181
ferent MZs. Thus, the utility of MZs is enhanced if the farm managers
3 0.110 0.118
4 0.131 0.142 have adequate knowledge about the soil properties and devise agri-
5 0.139 0.156 cultural management accordingly. Moreover, adaptation of site-specific
6 0.146 0.163 management by the farmers depends upon its simplicity, functionality
7 0.170 0.194 and economic feasibility. Hence, cluster analysis creates an opportunity
for identification MZs in an area and adoption of site-specific soil
management strategies for enhanced oil palm production.
variability of soil properties for delineating MZs. The analysis of var-
This approach of delineating soil management zones could be
iance revealed that the three MZs created were distinctly different from
adopted in other crops for efficient soil nutrient management by con-
each other. This is in line with findings of other researchers (Xin-Zhang
sidering stable and predictable spatial information related to crop yield.
et al., 2009; Davatgar et al., 2012; Tripathi et al., 2015; Shukla et al.,
The cluster analysis, which helps in reducing within zone variability
2017). The mean values of soil properties of three MZs are given in
and identification of soil management zones for site-specific soil man-
Table 6. The geographical area of MZ 1, MZ 2 and MZ 3 were 30.1, 47.8
agement to increase the crop production, also reduces the number of
and 22.1% respectively. Soil properties varied significantly among the
soil samples to be analysed for preparation of MZ maps for a particular
MZs. The highest value of soil pH was recorded in MZ 2 whereas the
crop cultivation practices.
content of SOC followed the order MZ 3 > MZ 1 > MZ 2. The con-
centration available P, K and S and exchangeable Ca was the lowest in
MZ 1 and the highest in MZ 3. Among the MZs, MZ 2 had lowest 4. Conclusions
available B concentration and the highest exchangeable Mg con-
centration. The heterogeneity of soil properties in MZs is attributed to In the present study, spatial variations of some selected soil prop-
soil types, soil and nutrient management practices. It highlights the erties of oil palm plantations grown in a hot and humid tropical region
importance of for site-specific soil management for sustaining soil and of southern India were assessed using geostatistical tools and clustered
oil palm productivity as well as avoiding land degradation in the study into three soil management zones by PCA and fuzzy c-means classifi-
area. The average values of soil properties could be used as reference cation. The study exhibited wide variation in the values of the soil
for site-specific management of soil nutrient management via different properties in oil palm plantations of the study area. Soil pH was ne-
application rates (Khosla et al., 2008; Nawar et al., 2017). For example, gatively correlated with EC and positively correlated with exchangeable

257
S.K. Behera et al. Catena 165 (2018) 251–259

Table 6
Mean values of soil properties in different management zones.

Management zone No. of points pH EC SOC P K Ca Mg S B % area

1 69 7.12b 0.20b 0.81b 35.0c 106.48c 807.16c 194.53c 30.83c 2.42b 30.1
2 53 7.73a 0.16b 0.68c 63.67b 114.81b 946.12b 260.18a 44.65b 1.81c 47.8
3 58 7.17b 0.32a 0.94a 66.05a 266.05a 1013.12a 249.49b 47.73a 5.52a 22.1

Different letters within each column indicate significant difference between the management zones at 0.05 level.
EC: electrical conductivity; SOC: soil organic carbon; P: available phosphorous; K: available potassium; Ca: exchangeable calcium; Mg: exchangeable magnesium; S: available sulphur; B:
available boron.

Mg. The content of SOC was positively correlated with available K, S Cambardella, C.A., Moorman, T.B., Novak, J.M., Parkin, T.B., Karlen, D.L., Turco, R.F.,
and B and exchangeable Ca. Geostatistical analysis revealed circular, Konopka, A.E., 1994. Field scale variability of soil properties in central Iowa soils.
Soil Sci. Soc. Am. J. 58, 1501–1511. http://dx.doi.org/10.2136/sssaj1994.
spherical, K-Bessel, exponential, stable and Gaussian best-fit semivar- 03615995005800050033x.
iogram models for studied soil properties. Soil properties revealed Corwin, D.L., Lesch, S.M., 2005. Apparent soil electrical conductivity measurements in
spatial heterogeneity with weak to strong spatial dependence. Three agriculture. Comput. Electron. Agric. 46, 11–43.
Davatgar, N., Neishabouri, M.R., Sepaskhah, A.R., 2012. Delineation of site specific nu-
MZs of the study area generated using PCA and fuzzy c-means differed trient management zones for a paddy cultivated area based on soil fertility using
significantly for soil properties. The mean values of soil properties in fuzzy clustering. Geoderma 173–174, 111–118. http://dx.doi.org/10.1016/j.
each zone could be utilized as a reference for variable-rate fertilizer geoderma.2011.12.005.
Ferreira, V., Panagopoulos, T., Andrade, R., Guerrero, C., Loures, L., 2015. Spatial
application. The information of such MZs may help in optimizing future
variability of soil properties and soil erodibility in the Alqueva reservoir watershed.
soil sampling approaches. The results of the present study along with Solid Earth 6, 383–392. http://dx.doi.org/10.5194/se-6-383-2015.
developed MZ maps could be utilized for site-specific soil nutrient Fitter, A.H., Gilligan, C.A., Hollingworth, K., Kleczkowski, A., Twyman, R.M., Pitchford,
J.W., 2005. Biodiversity and ecosystem function in soil. Funct. Ecol. 19, 369–377.
management and devising future soil sampling strategies in oil palm
http://dx.doi.org/10.1111/j.1365-2435.2005.00969.x.
plantations and also in other crops of the study area for maximizing Foroughifar, H., Jafarzadeh, A.A., Torabi, H., Pakpour, A., Miransari, M., 2013. Using
crop production. geostatistics and geographical information system techniques to characterize spatial
variability of soil properties including micronutrients. Commun. Soil Sci. Plant Anal.
44 (8), 1273–1281. http://dx.doi.org/10.1080/00103624.2012.758279.
Acknowledgements Goh, K.J., Hardter, R., Fairhust, T.H., 2003. Fertilizer for maximum return. In: Fairhust,
T.H., Hardter, R. (Eds.), Oil Palm: Management for High and Sustainable Yields.
The authors acknowledge the financial support extended by Indian International Potash Institute, Singapore, pp. 279–306.
Goovaerts, P., 1998. Geostatistical tools for characterizing the spatial variability of mi-
Council of Agricultural Research (ICAR), New Delhi, India for carrying crobiological and physico-chemical soil properties. Biol. Fertil. Soils 27, 315–334.
out the study. The authors thank the editor and the anonymous re- Gupta, U.C., 1967. A simplified method for determining hot-water soluble boron in
viewers for their suggestions for improving the quality of the manu- podzol soils. Soil Sci. 103, 424–428.
Hanway, J.J., Heidel, H., 1952. Soil Analyses Methods as Used in Iowa State College Soil
script. Testing Laboratory. 57. Iowa Agric., pp. 1–31.
Hoffmann, M.P., Vera, A.C., van Wijk, M.T., Gilder, K.E., Oberthur, T., Donough, C.,
Appendix A. Supplementary data Whitbred, A.M., 2014. Simulating potential growth and yield of oil palm (Elaeis
guineensis) with PLMSIM: model description, evaluation and application. Agric. Syst.
131, 1–10.
Supplementary data associated with this article can be found in the Jackson, M.L., 1973. Soil Chemical Analysis. Indian ed. Prentice Hall of India, New
online version, at https://doi.org/10.1016/j.catena.2018.02.008. Delhi, India.
Jones Jr., J.B., 1998. Soil test methods: past, present, and future. Commun. Soil Sci. Plant
These data include the Google map of the most important areas de-
Anal. 29, 1543–1552.
scribed in this article. Kerry, R., Oliver, M.A., 2004. Average variograms to guide soil sampling. Int. J. Appl.
Earth Obs. Geoinf. 5, 307–325.
References Khaledian, Y., Kiani, F., Ebrahimi, S., Brevik, E.C., Aitkenhead-Peterson, J., 2017.
Assessment and monitoring of soil degradation during land use change using multi-
variate analysis. Land Degrad. Dev. 28 (1), 128–141. http://dx.doi.org/10.1002/ldr.
Agterberg, F.P., 1984. Trend surface analysis. In: Gaile, G.L., Willmott, C.J. (Eds.), Spatial 2541.
Statistics and Models. Reidel, Dordrecht, The Netherlands, pp. 174-171. Khosla, R., Inman, D., Westfall, D.G., Reich, R.M., Frasier, M., Mzuku, M., Koch, B.,
Alvares, C.A., Goncalves, J.L.M., Vieira, S.R., da Silva, C.R., Franciscatte, W., 2011. Hornung, A., 2008. A synthesis of multi-disciplinary research in precision agriculture:
Spatial variability of physical and chemical attributes of some forest soils in south- site-specific management zones in the semi-arid western Great Plains of the USA.
eastern of Brazil. Sci. Agric. 68 (6), 697–705 (Piracicaba, Braz.). Precis. Agric. 9, 85–100.
Anonymous, 2017. World Palm Oil Production 2016/2017. Available at: https://www. Kilic, K., Kilic, S., Kocyigit, R., 2012. Assessment of spatial variability of soil properties in
worldpalmoilproduction.com/, Accessed date: 20 May 2017. areas under different land use. Bulgarian J. Agr. Sci. 18, 722–732.
Behera, S.K., Shukla, A.K., 2015. Spatial distribution of surface soil acidity, electrical Lopez-Granados, F., Jurado-Exposito, M., Atenciano, S., GarciaFerrer, A., De la Orden,
conductivity, soil organic carbon content and exchangeable potassium, calcium and M.S., Garcia-Torres, L., 2002. Spatial variability of agricultural soil parameters in
magnesium in some cropped acid soils of India. Land Degrad. Dev. 26, 71–79. http:// southern Spain. Plant Soil 246, 97–105.
dx.doi.org/10.1002/ldr.2306. Mallarino, A.P., Oyarzaba, E.S., Hinz, P.N., 1999. Interpreting within-field relationships
Behera, S.K., Rao, B.N., Suresh, K., Manorama, K., Ramachandrudu, K., Manoja, K., 2015. between crop yields and soil and plant variables using factor analysis. Precis. Agric. 1,
Distribution variability of soil properties of oil palm (Elaeis guineensis Jacq.) planta- 15–26.
tions in southern plateau of India. Indian J. Agric. Sci. 85 (9), 1170–1174. Mengel, K., Kirkby, E.A., 1987. Principles of Plant Nutrition. International Potash
Behera, S.K., Suresh, K., Rao, B.N., Mathur, R.K., Shukla, A.K., Manorama, K., Institute, Basel, Switzerland.
Ramachandrudu, K., Harinarayana, P., Prakash, C., 2016. Spatial variability of some Meul, M., Van Meirvenne, M., 2003. Kriging soil texture under different types of non-
soil properties varies in oil palm (Elaeis guineensis Jacq.) plantations of west coastal stationarity. Geoderma 112, 217–233.
area of India. Solid Earth 7, 979–993. http://dx.doi.org/10.5194/se-7-979-2016. Minasny, B., McBratney, A.B., 2006. A conditioned Latin hypercube method for sampling
Berget, I., Mevik, B., Nis, T., 2008. New modifications and applications on fuzzy c-means in the presence of ancillary information. Comput. Geosci. 32 (9), 1378–1388. http://
methodology. Comput. Stat. Data Anal. 52 (5), 2403–2418. dx.doi.org/10.1016/j.cageo.2005.12.009.
Bogunovic, I., Pereira, P., Brevik, E.C., 2017a. Spatial distribution of soil chemical Mueller, T.G., Hartsock, N.J., Stombaugh, T.S., Shearer, S.A., Cornelius, P.L., Barnhise,
properties in an organic farm in Croatia. Sci. Total Environ. 584-585, 535–545. R.I., 2003. Soil electrical conductivity map variability in limestone soil overlain by
http://dx.doi.org/10.1016/j.scitotenv.2017.01.062. loess. Agron. J. 95, 496–507.
Bogunovic, I., Trevisani, S., Seput, M., Juzbasic, D., Durdevic, B., 2017b. Short-range and Murphy, B.W., 2015. Impact of soil organic matter on soil properties — a review with
regional spatial variability of soil chemical properties in an agro-ecosystem in eastern emphasis on Australian soils. Soil Res. 53, 605–635. http://dx.doi.org/10.1071/
Croatia. Catena 154, 50–62. http://dx.doi.org/10.1016/j.catena.2017.02.018. SR14246.
Brevik, E.C., Calzolari, C., Miller, B.A., Pereira, P., Kabala, C., Baumgarten, A., Jordan, A., Nadal-Romero, E., Cammeraat, E., Perez-Cardiel, E., Lasanta, T., 2016. How do soil or-
2016. Soil mapping, classification, and pedologic modelling: history and future di- ganic carbon stock change after crop land abandonment in Mediterranean humid
rections. Geoderma 264 (Part B), 256–274. mountain areas? Sci. Total Environ. 566–567, 741–752. http://dx.doi.org/10.1016/j.

258
S.K. Behera et al. Catena 165 (2018) 251–259

scitotenv.2016.05.031. org/10.1097/SS.0b013e31821c0dab.
Nawar, S., Corstanje, R., Halcro, G., Mulla, D., Mouazen, A.M., 2017. Delineation of soil Shukla, A.K., Sinha, N.K., Tiwari, P.K., Prakash, C., Behera, S.K., Lenka, N.K., Singh, V.K.,
management zones for variable-rate fertilization: A review. Adv. Agron. 143, Dwivedi, B.S., Majumdar, K., Kumar, A., Srivastava, P.C., Pachauri, S.P., Meena,
175–245. http://dx.doi.org/10.1016/bs.agron.2017.01.003. M.C., Lakaria, B.L., Siddiqui, S., 2017. Spatial distribution and management zones for
Nielsen, D.R., Bouma, J., 1985. Soil spatial variability. In: Proceedings of a Workshop of sulfur and micronutrients in Shiwalik Himalayan region of India. Land Degrad. Dev.
the ISSS and the SSSA, Las Vegas, USA. 30th November to 1st December 1984. 243 28, 959–969. http://dx.doi.org/10.1002/ldr.2673.
Pudoc: Wageningen, The Netherlands. Soil Survey Staff, 2014. Keys to Soil Taxonomy, 12th edn. USDA-Natural Resources
Nikheloane, T., Olaleye, A.O., Mating, R., 2012. Spatial heterogeneity of soil physic- Conservation Service, Washington, DC.
chemical properties in contrasting wetland soils in two agro-ecological zones of Tesfahunegn, G.B., Tamene, L., Vlek, P.L.G., 2011. Catchment-scale spatial variability of
Lesotho. Soil Res. 50, 579–587. soil properties and implications on site-specific soil management in northern
Olsen, S.R., Cole, C.V., Watanable, F.S., Dean, L.A., 1954. Estimation of available phos- Ethiopia. Soil Tillage Res. 117, 124–139. http://dx.doi.org/10.1016/j.still.2011.09.
phorous in soils by extraction with sodium bicarbonate. In: Circular of United States 005.
Department of Agriculture No. 939. Thapa, G.B., Yila, O.M., 2012. Farmers' land management practices and status of agri-
Ortega, R.A., Santibanez, O.A., 2007. Determination of management zones in corn (Zea cultural land in the Jos Plateau, Nigeria. Land Degrad. Dev. 23, 263–277. http://dx.
mays L.) based on soil fertility. Comput. Electron. Agric. 58 (1), 49–59. http://dx.doi. doi.org/10.1002/ldr.1079.
org/10.1016/j.compag.2006.12.011. Tisdale, S.L., Nelson, W.L., Beaton, J.D., 1985. Soil Fertility and Fertilizers, 4th edn.
Palmer, J., Thorburn, P.J., Biggs, J.S., Dominati, E.J., Probert, M.E., Meier, E.A., Huth, McMilan, New York.
N.I., Dodd, M., Snow, V., Larsen, J.R., Parton, W.J., 2017. Nitrogen cycling from Triantafilis, J., Odeh, I.O.A., McBratney, A.B., 2001. Five geostatisical models to predict
increased soil organic carbon contributes both positively and negatively to ecosystem soil salinity from electromagnetic induction data across irrigated cotton. Soil Sci. Soc.
services in wheat agro-ecosystems. Front. Plant Sci. 8, 731. http://dx.doi.org/10. Am. J. 65, 869–878.
3389/fpls.2017.00731. Trigalet, S., Gabarron-Galeote, M.A., Oost, K.V., van Wesemael, B., 2016. Changes in soil
Parfitt, J.M.B., Timm, L.C., Pauletto, E.A., Sousa, R.O., Castilhos, D.D., de Avila, C.L., organic carbon pools along chronosequence of land abandonment in southern Spain.
Reckziegel, N.L., 2009. Spatial variability of the chemical, physical and biological Geoderma 268, 14–21. http://dx.doi.org/10.1016/j.geoderma.2016.01.014.
properties in lowland cultivated with irrigated rice. Rev. Bras. Ciênc. Solo 33, Tripathi, R., Nayak, A.K., Shahid, M., Lal, B., Gautam, P., Raja, R., Mohanty, S., Kumar,
819–830 (in Portuguese). A., Panda, B.B., Sahoo, R.N., 2015. Delineation of soil management zones for a rice
Peralta, N.R., Costa, J.L., Balzarini, M., Franco, M.C., Cordoba, M., Bullock, D., 2015. cultivated area in eastern India using fuzzy clustering. Catena 133, 128–136. http://
Delineation of management zones to improve nitrogen management of wheat. dx.doi.org/10.1016/j.catena.2015.05.009.
Comput. Electron. Agric. 110, 103–113. http://dx.doi.org/10.1016/j.compag.2014. Walkley, A.J., Black, I.A., 1934. An examination of the Degtjareff method for determining
10.017. soil organic matter and a proposed modification of the chromic acid titration method.
Prasad, M.V., Sarkar, A., Jameema, J., 2013. Performance of oil palm production tech- Soil Sci. 37, 29–38.
nologies. Indian Res. J. Ext. Educ. 10 (10–15), 2013. Williams, C.H., Steinbergs, A., 1969. Soil sulphur fractions as chemical indices of avail-
Rao, B.N., Suresh, K., Behera, S.K., Ramachandrudu, K., Manorama, K., 2014. Nutrient able sulphur in some Australian soils. Aust. J. Agric. Res. 10, 340–352.
Management in Oil Palm. Technical Bulletin. DOPR, Pedavegi, AP, India, pp. 1–24. Woittiez, L.S., van Wijk, M.T., Slingerland, M., van Noordwijk, M., Giller, K.E., 2017.
Rethinam, P., Arulraj, S., Rao, B.N., 2012. Assessment of Additional Potential Areas for Yield gaps in oil palm: A quantitative review of contributing factors. Eur. J. Agron.
Oil Palm Cultivation in India. Report submitted to Department of Agriculture and 83, 57–77. http://dx.doi.org/10.1016/j.eja.2016.11.002.
Cooperation, Ministry of Agriculture, Govt. of India. Directorate of Oil Palm Xin-Zhang, W., Guo-Shun, L., Hong-Chao, H., Zhen-Hai, W., Qing-Hua, L., Xu-Feng, L.,
Research, Pedavegi, Andhra Pradesh, pp. 1–84. Wei-Hong, H., Yan-Tao, L., 2009. Determination of management zones for a tobacco
Saha, D., Kukal, S.S., Bawa, S.S., 2014. Soil organic carbon stock and fractions in relation field based on soil fertility. Comput. Electron. Agric. 65, 168–175. http://dx.doi.org/
to land use and soil depth in the degraded Shiwaliks hills of lower Himalayas. Land 10.1016/j.compag.2008.08.008.
Degrad. Dev. 25, 407–416. http://dx.doi.org/10.1002/ldr.2151. Zhang, X.Y., Sui, Y.Y., Zhang, X.D., Meng, K., Herbert, S.J., 2007. Spatial variability of
Saito, H., McKenna, A., Zimmerman, D.A., Coburn, T.C., 2005. Geostatistical interpola- nutrient properties in black soil of northeast China. Pedosphere 17 (1), 19–29. http://
tion of object counts collected from multiple strip transects: ordinary kriging versus dx.doi.org/10.1016/S1002-0160(07)60003-4.
finite domain kriging. Stoch. Env. Res. Risk A. 19, 71–85. Zhao, G., Mu, X., Wen, Z., Wang, F., Gao, P., 2013. Soil erosion, conservation, and eco-
SAS Institute, 2011. The SAS system for Windows. In: Release 9.2. SAS Inst. Cary, NC. environment changes in the loess plateau of China. Land Degrad. Dev. 24, 499–510.
Sharma, P., Shukla, M.K., Mexal, J.G., 2011. Spatial variability of soil properties in http://dx.doi.org/10.1002/ldr.2246.
agricultural fields of Southern New Mexico. Soil Sci. 176 (6), 288–302. http://dx.doi.

259

Vous aimerez peut-être aussi