Journal of Soils and Sediments

https://doi.org/10.1007/s11368-019-02488-z

SEDIMENTS, SEC 1 • SEDIMENT QUALITY AND IMPACT ASSESSMENT • RESEARCH ARTICLE

Spatial variability assessment of La and Nd concentrations in coastal

China soils following 1000 years of land reclamation
Lin Wang 1,2 & George Christakos 1,3 & Chunfa Wu 2 & Jiaping Wu 1

Received: 9 April 2019 / Accepted: 3 October 2019

# Springer-Verlag GmbH Germany, part of Springer Nature 2019

Abstract
Purpose Rare earth elements (REE) play a prominent role in the understanding of geochemical processes. Lanthanum (La) and
neodymium (Nd) are considered as two representative examples of REE. The objectives of this study were to investigate their
concentrations and spatial variability in coastal soils and to assess the effect of long-term reclamation. The coastal plain of Cixi
City (China) has undergone reclamation for over 1000 years, providing an ideal place to perform such research.
Materials and methods In total, 329 soils samples were collected by means of a nested sampling methodology, combined with
grid and transect techniques. The total contents, as well as their diethylenetriamine penta-acetic acid (DTPA) extractable con-
centrations, were measured by an inductively coupled plasma-mass spectrometry (ICP-MS). Statistics and geostatistics tools
were used to analyze their spatial variability. Combined variogram models were developed to quantitatively describe their spatial
structure. Variance components were estimated to assess the proportion of their total variance captured by reclamation history and
different sampling stages.
Results and discussion Both the La and Nd concentrations showed great variability in space. Reclamation activity accounted for
42–84% of their total variance. There were significant differences in their contents between reclaimed zones (P < 0.05). They
exhibited moderate spatial dependence, and the discontinuity in space generally occurred around sea dikes, as indicated by the
spherical components of the La and Nd concentration variogram models. A gradually increasing tendency of elemental contents
with reclamation time was observed, which was also reflected in the linear component of the variogram models. However, in the
areas covered by the 3rd to the 10th zones, intensive application of phosphorus fertilizer, land consolidation, or household
appliance factories led to large fluctuations of elemental contents being superimposed on the relatively homogeneous content
distributions, thus causing the observed periodic features of their variogram models. Additionally, a 100-m sampling distance was
found to be appropriate for future REE investigations.
Conclusions Geostatistical modeling showed a great potential to analyze quantitatively the spatial variability of La and Nd and to
assess human forces on soil genesis. The investigation of the spatial distribution of REE identified the prominent role played by
the geogenic component. By means of spatial analysis, a time-series of elemental contents on a millennium scale was also
reconstructed. Accumulation of their concentrations with time, as well as pollution, occurred in the surface soils. Reclamation
history was the most important factor controlling La and Nd spatiotemporal variability.

Keywords Coastal soils . Concentration accumulation . Land reclamation . Rare earth elements . Spatial distribution

Responsible editor: Marcel van der Perk

1 Introduction
* Jiaping Wu
[email protected] Rare earth elements (REE), from lanthanum (La) to lu-
1
tetium (Lu), constitute a coherent group in the periodic
Institute of Islands and Coastal Ecosystems, Zhejiang University,
Zhoushan 316000, China
table. Although their clarkes are small, they are of great
2
importance and have many technical applications. They
Jiangsu Key Laboratory of Agricultural Meteorology, Nanjing
University of Information Science & Technology, Nanjing 210044, are widely used to record subtle geochemical evolution
China in natural systems (Henderson 1984; Aide and Aide
3
Department of Geography, San Diego State University,
2012) and are routinely treated as tracers in geosciences
California, CA, USA or in environmental studies as indicators of geogenic or
 J Soils Sediments

anthropogenic pollution sources (Dołęgowska and research on bioavailable REE is rarely mentioned in the liter-
Migaszewski 2013). ature. It is true that total REE content in soils can be regarded
Rare earth elements are not only of theoretical and soil- as a storage pool for REE but cannot reflect the potential
genetic importance, but they also have applied values availability for plant uptake (Hu et al. 2006; Loell et al.
(Vodyanitskii 2012). They are prerequisites of modern indus- 2011). Thus, the application of REE as soil genesis tracers is
try, and their use in agriculture has yielded positive effects in hindered. Accordingly, the need for investigating spatial and
terms of crop yield and body weight of poultry (Pang et al. temporal REE characteristics from a new perspective should
2001; Hu et al. 2006). In China, elevated REE concentrations be stressed.
have been observed in soils adjacent to REE mines and refin- In view of the above considerations, geostatistics provides
ing plants (Wei et al. 2001; Hao et al. 2015). They have been a set of mathematical modeling tools allowing the rigorous
applied in Chinese agriculture for over four decades (Lu et al. and physically meaningful characterization of spatial/
2006). Especially, the production and application of phospho- temporal data variability (Christakos 1992, 1998, 2017;
rus (P) fertilizers lead to REE emissions to the soils. The Saldaña et al. 1998; Olea 1999; Nickel et al. 2014). It can
experiment performed in Bulgaria during 1960–1990 showed quantify spatial or spatiotemporal relations and dependencies
that with an annual P-fertilizer application rate of 300 kg ha−1, between natural variables based on available observations at a
an amount of 30–170 g REE ha−1 is expected to enter the soils set of specified locations (Saldaña et al. 1998). Since 1980s,
(Hu et al. 2006). Continuous, intensive application can double geostatistics has been widely used in soil investigations
the REE content in soils during a time period of 159 years (Christakos 1985; Wilding 1985; Webster and Oliver 2007;
(Laveuf and Cornu 2009). River or groundwater irrigation and Lark 2012). But, so far, few studies have been carried out
waste spreading on soils are also responsible for REE inputs based on real-world data sets to determine the spatial variation
into soils (Aubert et al. 2002; Laveuf and Cornu 2009). of REE within and between seemingly homogeneous
Following anthropogenic inputs, REE accumulated in soils reclaimed areas.
and crops will ultimately enter the food chain. Excessive REE Accordingly, the objectives of the present study were three-
contents can cause serious consequence for ecosystems and fold: (i) to investigate the total and bioavailable levels of La
human health (Tyler 2004; Brioschi et al. 2013). The potential and Nd in the soils; (ii) to determine quantitatively their spatial
contamination by REE is expected to grow rapidly in the near variabilities; and (iii) to reveal the effects of long-term recla-
future (Miao et al. 2008). mation on them. Lanthanum (La) and neodymium (Nd) were
In most coastal areas, reclamation has become a convenient taken as two representative REE examples, since they have
approach of solving land shortage problems. Cixi City in been consistently among the most abundant REE in soils and
southeast China is a remarkable case that has undergone an could represent the elements with odd atomic number and the
over 1000-year-long reclamation process. Its sea dikes con- elements with even atomic number, respectively. Although
struction and land use types have been recorded in great detail, cerium (Ce) has high contents in soils, its geochemical behav-
which provides an ideal opportunity to assess temporal and ior is special, due to variable valences. Other REE were also
spatial soil changes after reclamation. Following such long- measured in this study but had low concentrations (around the
term reclamation, the landscape and ecosystem have been detection limit).
drastically modified (Tian et al. 2016; Xu et al. 2019). As a
matter of fact, it is necessary to assess such human activities
on soil genesis, since knowledge of soil dynamics and direc- 2 Materials and methods
tions of pedogenic evolution under land reclamation is useful
for the improvement and sustainable use of soil resources. 2.1 Site description
Since an efficient way to trace such changes is to study soil
chronosequences using REE, it has been considered critical to The study area is located along the coastal plain of the Cixi
use the REE signature to quantitatively trace reclamation pro- City, southeast China. This area was divided by the sea dikes
cesses (Chen et al. 2014). into 11 reclamation zones (Fig. 1). According to agricultural
At present, REE content distributions in different soil types chorography (Cixi Agricultural and Forestry Bureau 1991),
and in different regions, as well as their geochemical fractions since 1047 AD, nongovernmental earthy dikes have begun
and transferring in soil profiles, have been fully discussed in to emerge to defend against seawater intrusion. With the con-
the literature (Hu et al. 2006; Cheng et al. 2012; Mihajlovic tinual deposition of sediments, the coastline extended out-
et al. 2014; Pédrot et al. 2015). But, quantitative information wardly. Every few hundreds of years or even every few de-
about the spatial features of REE is currently not available, cades, new sea dikes were conducted to enclose more land
and, also, temporal variation of REE in soils is rarely paid resources. Most of the ancient dikes (in particular, the 1st to
attention to. Systematic research concerning the effect of the 7th ones) were destroyed due to lack of maintenance dur-
long-term reclamation on REE is rather limited. And relevant ing many years, but their location could still be interpreted
J Soils Sediments

Fig. 1 Sampling sites along the

coastal plain, Cixi City, China
(digital numbers, 1, 2, 3, and so
on, indicate the different
reclaimed zones)

from local chorography and remote sensing images. Having variability of La and Nd concentrations. The nested sampling
experienced long-term reclamation, initial micro-topography, distances considered were approximately 100, 60, 36, 21, 12,
soil types, and soil properties have been modified accordingly. 7, 4, 2, and 1 m, which were divided according to the Golden
In the first two reclaimed zones, the long reclamation and Ratio rule (Fig. 1). This design favored the investigation of
cultivation history caused the high soil fertility, so that rice had spatial dependency at different spatial scales (Saldaña et al.
been cultivated for more than 600 years. The soil types of these 1998; Webster and Oliver 2007). The transect where soil sam-
two zones belonged to the same soil group (Anthrosols), but fell ples were collected was perpendicular to the seashore in order
into different soil series (Wang et al. 2014a). Only soils in the to assess the reclamation history effect. And most of the sam-
first zone were issued from the alternating deposition of marine ples distributed along such transect, as the main objective was
and lacustrine materials. And parent materials of soils in the to analyze the spatial variability of elemental concentrations
other zones were marine sediments. influenced by reclamation history.
From the 3rd to the 7th zones, soils have been reclaimed for The samples were air-dried at room temperature, passed
a period of 60–300 years. The main land use type was upland, through a 2-mm polyethylene sieve, and were prepared for
including vegetable and nursery due to high salinity, where soil pH, texture, and extractable REE concentrations measure-
soils belonged to Cambosols. ment. Afterwards, some samples were ground into finer pow-
In the 8th to the 11th zones, the sea dikes were built in der using an agate mortar and passed through a 100-mesh
succession since 1952, and the soil types belonged to polyethylene sieve, in order to measure organic matter (OM)
Halosols. Among them, cotton plantation was the main land and total REE concentrations.
use type in the 8th and 9th zones. In the 10th zone, the main The pH values were measured in 1:2.5 soil-water suspen-
land uses included fishery and cotton plantation. Integrated sions using a pH meter (PB-10; Sartorius, Germany). The
industries were once widely spread in this zone, including amount of OM was determined by the Walkley-Black potas-
metallurgy and machinery manufacturing. Meanwhile, the sium biochromate volumetric method (Nelson and Sommers
11th zone was a newly reclaimed region, where uncultivated 1982). Soil texture was analyzed by the hydrometer method
land and fishery were distributed widely. Neither industry nor (Gee and Bauder 1986).
agriculture was present. According to measurements, pH values had a wide range
covering from acidic to strongly alkaline (5.28–10.1). The
2.2 Sampling and analyses OM values varied from 2.74 to 62.1 g kg−1. Clay content
varied from 50.4 to 342 g kg−1, with silt content from 204 to
Soil samples were collected at the coastal plain in Cixi during 694 g kg−1, and sand content from 95.6 to 746 g kg−1. Soil
October 2005. The total number of observations was 329. The texture belonged mainly into the silty clay loam category,
0–20-cm surface soils were collected by plastic spades and at according to the International Society of Soil Science (ISSS
this depth they were most severely influenced by human ac- 1929) classification.
tivity. Nested sampling, combined with transect and grid The diethylenetriamine penta-acetic acid (DTPA) extraction
methods, was used to quantify and model the spatial method was used to predict elemental bioavailability, which is a
 J Soils Sediments

universal method and particularly suitable for calcareous soils be the cornerstone of geostatistical analysis, which means that it
(Feng et al. 2005). The DTPA extractant was prepared accord- is vital to estimate, model, and interpret the variogram function
ing to Lindsay and Norvell (1978). A 6.67-g soil sample was correctly (Goovaerts 1998; Webster and Oliver 2007). As the
mixed with 20-ml DTPA solution. After shaking and filtration, empirical variogram values were found to fluctuate periodically,
the filtrate was acidified with three drops of concentrated nitric rather than increase monotonically, a combination of spherical,
acid (HNO3). Then the solution was diluted 1/4 and stored in linear, and periodic models were fitted to these empirical values.
plastic bottles for extractable measurements. According to Webster and Oliver (2007), when variogram
For total contents measurement, a 0.1-g soil sample was shapes appear complex, the best way to describe them is to
placed into a Teflon crucible and digested in the acid mixture combine two or more simple models, rather than to apply com-
of 12 ml concentrated hydrofluoric acid (HF), 2 ml perchloric plicated mathematical functions. The variogram model parame-
acid (HClO4), and 1 ml HNO3 on an electric hot plate at a ters can be used to estimate concentration values at unsampled
temperature of 100–200 °C. The remaining dry materials after places, if two-dimension data are obtained in the future.
digestion were dissolved by 1 ml HNO3 (1:1) solution to ob- The combined variogram model fitted to the empirical data
tain a completely clear solution. And then they were diluted to is as follows (Webster and Oliver 2007):
50 ml with 1% HNO3 solution.
Concentrations of La and Nd, both in DTPA-extractable and
total forms, were finally analyzed by inductively coupled
plasma-mass spectrometry (ICP-MS) (Agilent 7500a; Agilent,
Santa Clara, CA, USA). Iron (Fe) and manganese (Mn) con-
centrations were measured synchronously by ICP-MS to test
their relationship with La and Nd. Measurement quality was
controlled by certified standard reference materials (GBW
07401-GBW 07408, soil, China). In each experiment, replicate
samples and blanks analyses were performed and the recovery where h (lag interval) denotes the distance between a pair of
rates of La and Nd were between 95 and 103%. Relative stan- spatial locations, is a spherical model component, sh is a
dard deviation of each measurement was within 5%. linear model component, and is a Fourier
function (s, a, a1, b1, c, c0, and ω are empirical coefficients, the
2.3 Data processing physical meaning of which is explained next). The spherical
function is one of the most frequently used models in
Data normality was checked using the Kolmogorov–Smirnov geostatistics. It is characterized by three parameters: nugget ef-
(K–S) test. The total Nd content (Ndt) was approximated by a fect (c0), sill (c0 + c), and range (a). The nugget effect (c0) is
normal (Gaussian) distribution. But, the total La concentration determined as the positive intercept of the variogram curve with
(Lat), DTPA-extractable La (Lad), and Nd (Ndd) data were not the ordinate and represents short-scale variability finer than the
normally distributed, and they were log-transformed to stabilize sampling interval or measurement errors (Burgos et al. 2006).
their variances for geostatistical analysis purposes according to The sill (c0 + c) is the maximum value where variogram stabi-
Webster and Oliver (2007). The percentage of DTPA-extractable lized around (Webster and Oliver 2007). The nugget to sill ratio
concentrations to the total ones, i.e., the available ratio, was also (c0/(c0 + c)) indicates the degree of spatial dependence. The
calculated to determine their potential bioavailability. Analysis of range a is the distance at which a variogram reaches the sill,
variance (ANOVA) was performed to assess the difference in beyond which no spatial dependence exists. The coefficient s
elemental contents between different reclaimed zones. represents the gradient of lines. The coefficients a1 and b1 are
Pearson’s correlation coefficients were calculated between La used to describe amplitude of the waves, and the coefficient ω
and Nd concentrations, as well as soil properties. To assess the represents their wavelengths. Model selection and parameter es-
proportion of concentration variance captured by reclamation timation were based on the coefficient of determination (R2).
history and to determine the effects of different sampling stages,
the variance components were estimated by the restricted max-
imum likelihood (REML) method (constrained by the nonnega-
tive condition). This procedure was conducted using SPSS soft- 3 Results
ware (v13.0, SPSS Inc., Chicago, IL, USA).
Geostatistical methods were used to analyze the spatial vari- 3.1 Descriptive statistics
ability of elemental concentrations between observation points.
Experimental (or empirical) variogram values were calculated Descriptive statistics of total and DTPA-extractable concen-
using GS+ software (v3.1, Gamma Design Software, trations of La and Nd are presented in Table 1. The Lat con-
Plainville, MI, USA). The variogram function is considered to centrations ranged from 27.3 to 55.0 mg kg−1, with a mean
J Soils Sediments

Table 1 Descriptive statistics of

total and DTPA-extractable La Elements Mean Minimum Maximum Std. CV(%) Skewness Kurtosis
and Nd concentrations (in mg
kg−1) Lad 1.58 0.404 4.06 0.617 39.1 1.29 1.70
Ndd 2.07 0.588 6.14 1.15 55.6 1.43 1.16
Lat 38.6 27.3 55.0 3.54 9.17 0.532 1.40
Ndt 33.2 24.6 42.4 2.43 7.32 0.116 0.366

Subscript t denotes total concentration

Subscript d denotes DTPA-extractable concentration extracted by diethylenetriamine penta-acetic acid (DTPA)
Std. standard deviation, CV coefficient of variation

value of 38.6 ± 3.54 mg kg−1. The Lad concentrations ranged 3.3 Variogram analysis
between 0.404 and 4.06 mg kg−1, with a mean value of 1.58 ±
0.617 mg kg−1. And the concentration of DTPA-extractable The standardized empirical variogram values of La and Nd
La contributed to about 4.09% of the total La. concentrations (dots) and their fitted models (lines) are
The Ndt concentrations varied between 24.6 and 42.4 mg displayed in Fig. 3. Their model parameters are listed in
kg−1. Their mean value was 33.2 ± 2.43 mg kg−1, which was Table 2. For both of these elements, the spherical components
lower than that of Lat. The mean Ndd concentration was 2.07 ± of the models contributed most to the total variance (more than
1.15 mg kg−1. The Ndd content was higher than the corre- 75%), with a sill from 0.600 to 0.900. Concentrations of La
sponding level of Lad. Higher bioavailability of Nd than La and Nd (in both forms) showed considerable spatial variability
was observed, accounting for about 6.23% of its total content. (dependence). Their ranges varied from 3000 to 4500 m,
Contents of La and Nd showed high variability. The coef- which approximately corresponded to twice the average span
ficients of variation (CV) values of DTPA-extractable contents of each reclaimed zone. The physical interpretation of these
were larger than those of total contents. According to CV results was that La and Nd concentrations were correlated
classification (Wilding 1985), the former fell into the high within the above ranges, and sea dikes formed the boundary
variability category, whereas the latter displayed low variabil- of their continuity in space.
ity, which implied that the bioavailable contents might be Based on their nugget to sill ratio, La and Nd concentra-
more sensitive to human activities. tions in either form showed moderate spatial dependency, ac-
cording to the classification of Cambardella et al. (1994). By
3.2 Comparison between reclaimed zones comparison, La concentrations had a relatively smaller range
and a larger nugget effect than those of Nd concentrations (for
According to the ANOVA and multi-comparison analyses re- either form). Physically, this meant that the spatial pattern of
sults (Fig. 2), there were significant differences of the mean La concentrations was less continuous than that of Nd
values of La and Nd concentrations in both forms (total and concentrations.
DTPA-extractable) between different reclaimed zones at the The periodic components contributed to a small percentage
0.95 confidence level. The highest Lad mean values occurred of the total variance. Their amplitudes varied from 0.167 to
in the first two zones (about 2.25 mg kg−1), whereas the lowest 0.223. And their wavelengths were about 4000 m, which was
one was observed in the 11th zone (1.06 mg kg−1), i.e., the also approximately twice the average span of each reclamation
former were nearly double the latter. The Ndd concentrations zone. The periodic components of the variogram models im-
showed similar spatial patterns. The first two zones had the plied the existence of potential periodic features of the La and
highest mean Ndd values (about 3.74 mg kg−1), which were Nd contents in space. And the linear components, accounting
approximately three times the corresponding value in the 11th for the smallest percentage of the total variance, always had
zone (1.19 mg kg−1). Both of Lad and Ndd concentrations positive but very gentle gradients, indicating a long-range
displayed relative homogeneity from the 3rd to the 10th zone, physical trend in spatial variation (Webster and Oliver 2007).
especially obvious from the 3rd to 9th zone.
The Lat concentrations had their highest mean value in the 1st
zone (43.7 mg kg−1), and the lowest one in the 11th zone 3.4 Variance components
(34.2 mg kg−1). Similarly, the 1st zone had the highest mean
zonal Ndt value (36.2 mg kg−1), and the 11th zone had the lowest The variance components of La and Nd concentrations calcu-
one (30.5 mg kg−1). Both of Lat and Ndt concentrations tended to lated using the REML method were plotted as cumulative
vary rather little from the 2nd to the 10th zones. Likewise, the percentages in Fig. 4. In this figure, the first stage “Zone”
first two zones had the highest available ratio values of La and indicated variability caused by different reclamation zones.
Nd concentrations, while the 11th zone had the lowest ones. The second stage, an additional distance (500 m), entered
 J Soils Sediments

Fig. 2 Comparisons of La and Nd mean concentrations in different Scheffé test (P < 0.05). Digital numbers, 1, 2, 3, etc. denote different
reclaimed zones (histograms correspond to the mean concentration zones. Subscript t denotes total concentration. Subscript d denotes
values. Error bars are ± one standard deviation of the mean. Different extractable concentration extracted by diethylenetriamine penta-acetic
letters indicate significant differences between reclaimed zones by the acid (DTPA))

the calculations in order to search an appropriate sampling also showed a similar relationship. Furthermore, the total La
strategy for future investigations. and Nd concentrations showed a significant correlation with
The first stage “Zone” explained the largest part of their their corresponding DTPA-extractable contents.
total variance. For Ndd, about 84% of the total variance was Concentrations of La and Nd in both the total and DTPA-
obtained at this stage. For Lat, Ndt, and Lad, this stage extractable forms exhibited strong positive correlation with
accounted for about 42–49% of their total variance. Fed but a relatively weak relationship with Mnd. And they also
The first two stages, covering reclamation zones and 500-m showed a strong negative correlation with Fet and Mnt con-
stages, totally explained approximately 60% of the total vari- centrations (P < 0.01).
ance of Lat and Ndt. For Lad, however, the first three stages,
i.e., the sampling spans longer than 100 m, accounted for
about 60% of the total variance. Thus, the 100-m sampling
density could be sufficient to reveal the main spatial structures 4 Discussion
of the elements of interest.
4.1 Comparison with background values

3.5 Correlation analysis China has the largest REE reserves (Ramos et al. 2016). Its
background value surveys were initiated in the 1980s, in
The correlation analysis revealed a significant relationship which 863 surface soil samples were collected from 41 soil
between elemental concentrations and soil properties groups and the contents of 15 different REE elements were
(Table 3). Concentrations of La and Nd in either form showed reported (Wei et al. 1991). The mean Lat concentration in the
a significant negative correlation with pH value and sand con- present study was close to the background average values in
tent, but had a significant positive correlation with OM, clay, China (39.7 mg kg−1) (Wei et al. 1991). About 37% of the
and silt contents (P < 0.05). total number of the samples had Lat values above this thresh-
The total concentrations of La and Nd showed significant old, mainly occurring in the first three zones. The mean Ndt
correlations with each other, and the DTPA-extractable forms concentration was higher than the background level in China
J Soils Sediments

Fig. 3 Empirical (raw) variograms (dots) with fitted models (solid lines) for the total and DTPA-extractable La and Nd concentrations across the sea
reclaimed zones

(26.4 mg kg−1) (Wei et al. 1991). About 99% of the samples 4.2 Reclamation effect
had higher Ndt values than this threshold. And their mean total
concentrations were also slightly beyond the corresponding The concentrations of La and Nd showed significant varia-
values for an average soil in the world (Kabata-Pendias tions across space in the Cixi City region. They exhibited a
2000). These results suggest that La and Nd concentrations spatial dependence structure, instead of a completely random
in the reclaimed regions had been increased, to some extent, distribution, according to the geostatistical analysis. This
mainly due to human activities. Further, the DTPA-extractable means that their distribution was controlled by certain system-
contents and potentially bioavailable fraction fell into the atic factors. Combined variogram models (with three compo-
ranges reported by other studies (Hu et al. 2006; Loell et al. nents) were applied in this study. Each component of these
2011). models systematically revealed different features of the

Table 2 Parameter values of

models fitted to the variograms of Component Parameters Lad Ndd Lat Ndt
La and Nd in the soils
Spherical Nugget, c0 0.400 0.200 0.400 0.400
Sill, c0 + c 0.900 0.600 0.700 0.800
Range (m), a 3000 4000 3500 4500
nugget to sill ratio (c0/(c0 + c)) 44.4% 33.3% 57.1% 50.0%
Linear Gradient, s 1.50e-5 5.53e-5 3.12e-5 2.85e-5
Periodic a1 − 0.0908 − 0.159 − 0.0854 − 0.104
b1 0.148 0.157 0.144 0.143
Wavelength (m), ω 4032 4004 4210 4035
Amplitude, W 0.174 0.223 0.167 0.177
 J Soils Sediments

fields and were in a prolonged reducing biochemical environ-

ment. Paddy cultivation had been found to favor the accumula-
tion of all REE in soils (Wei et al. 1991; Laveuf et al. 2012; Chen
et al. 2014). Hence, La and Nd concentrations, especially in the
total form, were significantly higher than those in the other zones.
From the 3rd to the 10th zone, similar upland cultivation
types caused the homogeneity of La and Nd concentrations. In
particular, soils in the 3rd to the 7th zone had been cultivated
for more than 60 years, where La and Nd concentrations
showed little variation. It seemed that after long-term upland
cultivation and cropping, soil bodies converged to increasing
homogenization as a function of age within 300 years, which
was also supported by Saldaña et al. (1998).
Nevertheless, fluctuations of La and Nd concentrations
Fig. 4 Accumulated variance percentages of the nested sampling existed in these zones, which were also mainly caused by land
distances for total and DTPA-extractable La and Nd concentrations use history. Around the densely populated residential areas,
their concentrations were elevated through fertilizer applica-
tion or irrigation water. Local farmers were inclined to apply
elemental spatial distribution, related with the reclamation considerable amounts of phosphorus fertilizers into soils, ap-
history and land use types. And the results of AVOVA and proximately fourfold the original values used during the 20-
variance component analyses further revealed that the key year period (1981–2003), leading to high available phospho-
factor was the reclamation history. According to Tyler rus contents in soils (Lu et al. 2006). Meanwhile, some other
(2004) and Laveuf and Cornu (2009), long-term intensive areas experienced land consolidation, i.e., the sublayer and
anthropogenic disturbance, like reclamation history and land surface soils were cycled during cultivation, and decreased
use changes, could bring about elemental redistribution in concentrations in the surface appeared. The above facts led
soils and soil evolution direction changes. to the periodic fluctuation of their variograms. Meanwhile,
the La and Nd amounts were relatively high in the 10th zone
among the upland farmed zones, although with relatively short
4.2.1 Land use effect reclamation time. Perhaps, this could also partly explain the
periodic characteristics of concentrations. Cixi City was fa-
Different reclaimed zones had quite different land use types. In mous for its factories producing household appliances, which
the first two reclaimed zones, soils were cultivated as paddy could partly cause the elemental accumulation within the up-
per soil layers (Zhang et al. 2000; Li et al. 2013). Besides,
Table 3 Correlation coefficients between La and Nd concentrations and rivulets and ponds spread over this zone, and water is consid-
soil properties ered as a carrier and an accumulator of pollutants dispersed in
the environment, which further promotes the elevation of REE
Lad Ndd Lat Ndt
(Laveuf and Cornu 2009; Moiseenko et al. 2013).
Ndd 0.786** The 11th zone was free from industry and agricultural pol-
Lat 0.484** 0.631** lution, having the lowest La and Nd levels, compared with
Ndt 0.511** 0.614** 0.909** other reclaimed zone. And the concentrations of heavy metal
Fed 0.521** 0.755** 0.530** 0.485** elements concentrations were also the lowest level (Wang
Mnd 0.150** − 0.135** 0.0716 − 0.0111 et al. 2014b). Soils in the study region mainly evolved from
Fet − 0.178** − 0.486** − 0.308** − 0.147** such sediments; this zone may be a suitable area to serve as the
Mnt − 0.431** − 0.758** − 0.396** − 0.348** reference for REE researches in the reclamation areas, for the
pH − 0.519** − 0.864** − 0.610** − 0.569** purpose of clarifying agriculture or industry impacts on REE
OM 0.603** 0.822** 0.517** 0.514** mobilization.
Clay 0.160** 0.158** 0.269** 0.390**
Silt 0.179** 0.129* 0.264** 0.204** 4.2.2 Reclamation history effect
Sand − 0.218** − 0.181** − 0.340** − 0.367**
The increasing trend of La and Nd concentrations with time
*Correlation significant at the 0.05 level was discovered by the gentle gradient in the linear variogram
**Correlation significant at the 0.01 level components. The first reclaimed zone had the highest elemen-
OM organic matter content tal contents with longest reclamation history, while the 11th
J Soils Sediments

zone had the lowest values, among all the reclaimed zones. In Cornu 2009). Thus, their correlation was rather complex, as
general, increased REE contents affected by anthropogenic was suggested by Yan et al. (1999), Laveuf and Cornu (2009),
alterations are the result of the application of agricultural fer- and Yusoff et al. (2013).
tilizer, and youthful soils usually had little weathering of REE- The strong correlation between La and Nd concentrations
bearing minerals and little REE mobility (Aide and Aide, observed in the present work is also supported by previous
2012; Ramos et al. 2016). Further, the first zone had higher studies (Tyler 2004; Aide and Aide 2012). These two elements
contents than the second one, although under the same culti- have analogous atomic mass and ionic radius, share similar
vation types, which was mainly contributed to the result of its chemical and physical properties, and tend to exist together
longer reclamation history, besides different parent materials. naturally (Henderson 1984; Hu et al. 2006). For each element,
Under long-term reclamation conditions, soil physico- the close relationship between total and extractable contents
chemical properties of the plain also showed an obvious evo- supported the fact that total element contents contributed to
lution with time, including increasing clay and OM contents, the available proportion of each element (Loell et al. 2011). It
and decreasing heavy metal contents, as wells as calcium car- seems that long-term reclamation caused the accumulation of
bonate leaching (Wang et al. 2014a,b; Ma et al. 2015). La and Nd contents in surface soils but did not alter their
Previous literature on a millennium-scale soil chronosequence coexistence relationship.
has also reported similar temporal trends for soil properties
and elemental concentrations under agricultural practices
(Sun et al. 2011; Chen et al. 2014; Wang et al. 2014b).
5 Conclusions
4.2.3 Soil properties influence
The coastal plain of Cixi City is a region that has been
The total content and bioavailability of La and Nd concentra- reclaimed since 1047 AD. As a result of reclamation, the nat-
tions were strongly affected by the physico-chemical proper- ural environment has been drastically modified. This work
ties of soils. Firstly, the particle size of soils strongly con- investigated quantitatively the levels and spatial features of
trolled the distribution of La and Nd. Soils abundant in clay La and Nd concentrations in the coastal soils.
fraction were usually enriched in REE, whereas samples with In this context, geostatistical tools provided a rigorous
a high portion of sand were depleted in REE (Miao et al. 2008; quantitative description of their spatial structure. By means
Aid and Aid 2012; Perelomov et al. 2012). Moreover, clay of spatial analysis, a time-series of elemental contents on a
minerals played an important role in increasing the mobility millennium scale was reconstructed, and a gradually accumu-
and bioavailability of REE in soils (Liang et al. 2005; Laveuf lating trend as a function of reclamation time was observed. It
and Cornu 2009; Li et al. 2013). was found that La and Nd pollution caused by the long-term
Secondly, the negative relationship between DTPA- reclamation history has occurred on the plain. The fitted
extractable contents and pH has also been supported by pre- variogram models consisted of three parts: the spherical, the
vious studies (Tyler 2004; Loell et al. 2011). Usually, low soil periodic, and the linear components. The physical interpreta-
pH favored conversion of elements from precipitate fractions tion of the model parameters linked them directly to the dom-
into soluble ones. Rare earth elements release increased dras- inant distribution features, like spatial correlation range, nug-
tically in acid soil conditions and reduced in alkaline condi- get effect, gradual trend, and local periodicity. All these fea-
tions (Cao et al. 2001; Hu et al. 2006; Li et al. 2013). tures were, in turn, associated with reclamation activities, in
Thirdly, OM had a high capacity to adsorb or chelate pos- which case the reclamation history was proved to be the key
itively charged REE (Tyler 2004; Laveuf and Cornu 2009). Its factor controlling spatiotemporal characteristics of La and Nd.
fraction representing humic and fulvic acids preferentially ac- Moreover, parent materials and soil properties also made
cumulated REE (Aide and Aide 2012). Hence, strongly posi- certain contribution to elemental distribution in the region of
tive correlation between OM and elemental contents occurred interest. Critical parameters were determined quantitatively,
on this coastal plain. like the optimal 100-m sampling interval, based on the nested
Finally, Fe and Mn, as well as their oxides, were also im- sampling method and variance components analysis, which
portant in the distribution and contents of La and Nd. They can be particularly valuable in future REE investigations. In
could scavenge REE by several processes, including co-pre- sum, the present study could significantly improve our under-
cipitation, adsorption, and ion exchange, during soil ageing standing of REE potential pollution and soil genesis under
and crystallization processes (Rankin and Childs 1976; Cao long-lasting land reclamation.
et al. 2001; Yusoff et al. 2013). Oxidative conditions allow the
Funding information This work was supported partially by the
precipitation of Fe and Mn as oxides with REE, while reduc-
Department of Science and Technology of Zhejiang (2016C04004) and
ing conditions lead to the dissolution of these Fe- and Mn- by the National Natural Science Foundation of China (Grant NSFC
oxides, which release all REE (Cao et al. 2001; Laveuf and 41671399).
 J Soils Sediments

Compliance with ethical standards Henderson P (1984) General geochemical properties and abundances of
the rare earth elements. In: Henderson P (ed) Rare Earth Element
Geochemistry. Elsevier, Amsterdam, pp 1–32
Conflict of interest The authors declare that they have no conflict of
Hu Z, Haneklaus S, Sparovek G, Schnug E (2006) Rare earth elements in
interest.
soils. Commun Soil Sci Plant Anal 37(9-10):1381–1420
International Society of Soil Science (ISSS) (1929) Minutes of the First
Commission Meetings, International Congress of Soil Science 1927.
Proceedings of the Congress of the International Society of Soil
References Science, Washington, pp 215–220
Kabata-Pendias A (2000) Trace elements in soils and plants, 3rd edn.
Aide MT, Aide C (2012) Rare earth elements: their importance in under- CRC Press, New York
standing soil genesis. ISRN Soil Sci 2012:1–11 Lark RM (2012) Towards soil geostatistics. Spat Stat 1:92–99
Laveuf C, Cornu S (2009) A review on the potentiality of rare earth
Aubert D, Stille P, Probst A, Gauthier-Lafaye F, Pourcelot L, Del Nero M
elements to trace pedogenetic processes. Geoderma 154(1-2):1–12
(2002) Characterization and migration of atmospheric REE in soils
Laveuf C, Cornu S, Guilherme LRG, Guerin A, Juillot F (2012) The
and surface waters. Geochim Cosmochim Acta 66(19):3339–3350
impact of redox conditions on the rare earth element signature of
Brioschi L, Steinmann M, Lucot E, Pierret MC, Stille P, Prunier J, Badot
redoximorphic features in a soil sequence developed from lime-
PM (2013) Transfer of rare earth elements (REE) from natural soil to
stone. Geoderma 170:25–38
plant systems: implications for the environmental availability of
Li X, Chen Z, Chen Z, Zhang Y (2013) A human health risk assessment
anthropogenic REE. Plant Soil 366(1-2):143–163
of rare earth elements in soil and vegetables from a mining area in
Burgos P, Madejón E, Pérez-de-Mora A, Cabrera F (2006) Spatial vari- Fujian Province, Southeast China. Chemosphere 93:1240–1246
ability of the chemical characteristics of a trace-element- Liang T, Zhang S, Wang L, Kung H, Wang Y, Hu A, Ding S (2005)
contaminated soil before and after remediation. Geoderma 130(1- Environmental biogeochemical behaviors of rare earth elements in
2):157–175 soil–plant systems. Environ Geochem Health 27(4):301–311
Cao X, Chen Y, Wang X, Deng X (2001) Effect of redox potential and pH Lindsay WL, Norvell WA (1978) Development of a DTPA soil test for
value on the release of rare earth elements from soil. Chemosphere zinc, iron, manganese, and copper. Soil Sci Soc Am J 42(3):421–428
44(4):655–661 Loell M, Albrecht C, Felix-Henningsen P (2011) Rare earth elements and
Cambardella CA, Moorman TB, Parkin TB, Karlen DL, Novak JM, relation between their potential bioavailability and soil properties,
Turco RF, Konopka AE (1994) Field-scale variability of soil prop- Nidda catchment (Central Germany). Plant Soil 349(1-2):303–317
erties in central Iowa soils. Soil Sci Soc Am J 58(5):1501–1511 Lu H, Zhao X, Luo H, Li R, Zhu Y (2006) Soil fertility changes of culti-
Chen L, Zhang G, Jin Z (2014) Rare earth elements of a 1000-year paddy vated soil in Cixi, Zhejiang. Chin Soil Sci 3:426–428 (in Chinese)
soil chronosequence: implications for sediment provenances, parent Ma C, Zheng R, Zhao J, Han X, Wang L, Gao X, Zhang C (2015)
material uniformity and pedological changes. Geoderma 230-231: Relationships between heavy metal concentrations in soils and recla-
274–279 mation history in the reclaimed coastal area of Chongming Dongtan of
Cheng H, Hao F, Ouyang W, Liu S, Lin C, Yang W (2012) Vertical the Yangtze River Estuary, China. J Soils Sediments 15:139–152
distribution of rare earth elements in a wetland soil core from the Miao L, Xu R, Ma Y, Zhu Z, Wang J, Cai R, Chen Y (2008)
Sanjiang Plain in China. J Rare Earths 30(7):731–738 Geochemistry and biogeochemistry of rare earth elements in a sur-
Christakos G (1985) Modern statistical analysis and optimal estimation of face environment (soil and plant) in South China. Environ Geol
geotechnical data. Eng Geol 22(2):175–200 56(2):225–235
Christakos G (1992) Random Field Models in Earth Sciences. San Diego, Mihajlovic J, Stärk H, Rinklebe J (2014) Geochemical fractions of rare
CA, pp 459 earth elements in two floodplain soil profiles at the Wupper River,
Christakos G (1998) Spatiotemporal information systems in soil and en- Germany. Geoderma 228-229:160–172
vironmental sciences. Geoderma 85(2-3):141–179 Moiseenko TI, Skjelkvåle BL, Gashkina NA, Shalabodov AD,
Christakos G (2017) Spatiotemporal random fields: theory and applica- Khoroshavin VY (2013) Water chemistry in small lakes along a
tions. Elsevier, Amsterdam transect from boreal to arid ecoregions in European Russia: Effects
of air pollution and climate change. Appl Geochem 28:69–79
Cixi Agricultural and Forestry Bureau, Zhejiang (1991) Agricultural cho-
Nelson DW, Sommers LE (1982) Total carbon, organic carbon, and organic
rography of Cixi. Shanghai Science and Technology Press,
matter. In: Page AL (ed) Methods of soil analysis. Part 2. Chemical
Shanghai, pp 11–32 (in Chinese)
and Microbial Properties. ASA and SSSA, Madison, pp 539–579
Dołęgowska S, Migaszewski ZM (2013) Anomalous concentrations of Nickel S, Hertel A, Pesch R, Schröder W, Steinnes E, Uggerud HT (2014)
rare earth elements in the moss-soil system from south-central Modelling and mapping spatio-temporal trends of heavy metal ac-
Poland. Environ Pollut 178:33–40 cumulation in moss and natural surface soil monitored 1990-2010
Feng M, Shan X, Zhang S, Wen B (2005) A comparison of the throughout Norway by multivariate generalized linear models and
rhizosphere-based method with DTPA, EDTA, CaCl2, and NaNO3 geostatistics. Atmos Environ 99:85–93
extraction methods for prediction of bioavailability of metals in soil Olea RA (1999) Geostatistics for engineers and earth scientists. Kluwer
to barley. Environ Pollut 137(2):231–240 Academic Publishers, Boston
Gee GW, Bauder JW (1986) Particle-size analysis. In: Klute A (ed) Pang X, Li D, Peng A (2001) Application of rare-earth elements in the
Methods of soil analysis. Part 1. Physical and mineralogical agriculture of China and its environmental behavior in soil. J Soils
methods. ASA and SSSA, Madison, pp 383–411 Sediments 1:124–129
Goovaerts P (1998) Geostatistical tools for characterizing the spatial var- Pédrot M, Dia A, Davranche M, Gruau G (2015) Upper soil horizons
iability of microbiological and physico-chemical soil properties. control the rare earth element patterns in shallow ground water.
Biol Fertil Soils 27:315–334 Geoderma 239-240:84–96
Hao Z, Li Y, Li H, Wei B, Liao X, Liang T, Yu J (2015) Levels of rare Perelomov LV, Asainova ZS, Yoshida S, Ivanov IV (2012)
earth elements, heavy metals and uranium in a population living in Concentrations of rare-earth elements in soils of the Prioksko-
Baiyun Obo, Inner Mongolia, China: A pilot study. Chemosphere Terrasnyi State Biospheric Reserve. Eurasian Soil Sci 45(10):983–
128:161–170 994
J Soils Sediments

Ramos SJ, Dinali GS, Oliveira C, Martins GC, Moreira CG, Siqueira JO, Wei FS, Liu TL, Teng EJ, Rui KS (1991) Soil background value of rare
Guilherme LRG (2016) Rare earth elements in the soil environment. earth elements of China. Environ Sci 12(5):78–82 (in Chinese)
Curr Pollut Rep 2(1):28–50 Wei Z, Yin M, Zhang X, Hong F, Li B, Tao Y, Zhao G, Yan C (2001) Rare
Rankin PC, Childs CW (1976) Rare-earth elements in iron-manganese earth elements in naturally grown fern Dicranopteris linearis in
concretions from some New Zealand soils. Chem Geol 18(1):55–64 relation to their variation in soils in South-Jiangxi region
Saldaña A, Stein A, Zinck JA (1998) Spatial variability of soil properties (Southern China). Environ Pollut 114(3):345–355
at different scales within three terraces of the Henares River (Spain). Wilding LP (1985) Spatial variability: its documentation, accommodation
Catena 33(3-4):139–153 and implication to soil surveys. In: Nielsen DR, Bouma J (eds) Soil
Sun Y, Li X, Mander Ü, He Y, Jia Y, Ma Z, Guo W, Xin Z (2011) Effect of Spatial Variability. Pudoc, Wageningen
reclamation time and land use on soil properties in Changjiang River Xu C, Pu L, Li J, Zhu M (2019) Effects of reclamation on C, N, and P
Estuary, China. Chin Geogr Sci 21(4):403–416 stoichiometry in soil and soil aggregates of a coastal wetland in
Tian B, Wu W, Yang Z, Zhou Y (2016) Drivers, trends, and potential eastern China. J Soils Sediments 19:1215–1225
impacts of long-term coastal reclamation in China from 1985 to Yan X, Kerrich R, Hendry MJ (1999) Sequential leachates of multiple
2010. Estuar Coast Shelf Sci 170:83–90 grain size fractions from a clay-rich till, Saskatchewan, Canada:
Tyler G (2004) Rare earth elements in soil and plant systems-a review. implications for controls on the rare earth element geochemistry of
Plant Soil 267(1-2):191–206 porewaters in an aquitard. Chem Geol 158(1-2):53–79
Vodyanitskii YN (2012) Geochemical fractionation of lanthanides in soils Yusoff ZM, Ngwenya BT, Parsons I (2013) Mobility and fractionation of
and rocks: a review of publications. Eurasian Soil Sci 45(1):56–67 REEs during deep weathering of geochemically contrasting granites
Wang L, Coles N, Wu C, Wu J (2014a) Effect of long-term reclamation in a tropical setting, Malaysia. Chem Geol 349-350:71–86
on soil properties on a coastal plain, southeast China. J Coast Res Zhang H, Feng J, Zhu WF, Liu CQ, Wu DS, Yang WJ, Gu JG (2000) Rare
30(4):661–669 earth element distribution characteristics of biological chains in rare
Wang L, Coles N, Wu C, Wu J (2014b) Spatial variability of heavy metals earth element high background regions and their implications. Biol
in the coastal soils under long-term reclamation. Estuar Coast Shelf Trace Elem Res 73(1):19–28
Sci 151:310–317
Webster R, Oliver MA (2007) Geostatistics for environmental scientists. Publisher’s note Springer Nature remains neutral with regard to
Chichester, UK, pp 77–138 jurisdictional claims in published maps and institutional affiliations.