Contents lists available at ScienceDirect

Catena

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

Magnetic susceptibility characteristics of surface soils in the Xilingelegrassland and their implication for soil redistribution in wind-dominatedlandscapes: A preliminary study

Liang Liua,b, Zhuodong Zhangb, Keli Zhangb,⁎, Hongyuan Liub,c, Suhua Fud,e

a State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Institute of Soil and Water Conservation, Northwest A&F University, Yangling 712100,Shaanxi, Chinab State Key Laboratory of Earth Surface Processes and Resource Ecology, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, Chinac Key Laboratory of Wetland Ecology and Environment, Northeast Institute of Geography and Agroecology, Chinese Academy of Sciences, Changchun 130102, Chinad State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Chinese Academy of Sciences and Ministry of Water Resources, Yangling 712100, Chinae Faculty of Geographical Science, Beijing Normal University, Beijing 100875, China

A R T I C L E I N F O

Keywords:Magnetic susceptibility techniqueWind erosionGrazing intensitySoil grain sizeTemperate grassland

A B S T R A C T

Wind erosion processes in the typical temperate Xilingele grassland of North China result in significant regionalsurface soil fine particle and carbon loss. They increasingly restrict local grass industry sustainable productionand grassland ecosystem protection. It is challenging to link wind erosion and deposition at landscape scale usingclassical field monitoring or the expensive fallout environmental radionuclides tracing techniques. The low-costbut efficient magnetic susceptibility (MS) technique has been successfully demonstrated to have great potentialto trace soil water erosion processes and patterns at large spatial and temporal scales. However, so far soil winderosion research using MS technique has not been reported. This study had a trial to determine the variations ofsoil magnetic susceptibility on relative flat grassland by a grid soil sampling and to establish the relationshipbetween wind erosion parameters and variations of MS in surface soils. 160 grid sampling sites were spaced at aninterval of 400 m across a study transect with 12.8 km long and 1.6 km wide. 319 soil samples were collectedfrom the surface soils (0–1 cm and 1–6 cm layers). Grazing intensity of the sampling sites were investigated, andthe samples were measured for mass-specific low-frequency magnetic susceptibility (χlf), absolute frequency-dependent magnetic susceptibility (χfd), percentage frequency-dependent magnetic susceptibility (χfd%), soilgrain size and organic carbon concentrations. The results showed that the χlf, χfd and χfd% values in surface soilsranged from 30.0 to 97.8 × 10−8 m3 kg−1, 1.2 to 6.1 × 10−8 m3 kg−1 and from 3.2 to 8.0%, respectively. Thevariations of soil χlf values were closely related to grazing intensity, soil grain size and organic carbon con-centrations, suggesting that soil erosion processes were very sensitive to soil properties. Moreover, the MSparameters (χlf, χfd%) were positively correlated with the soil erosion rates and negatively correlated with thedust deposition rates, indicating that MS parameters could potentially identify the erosion and dust depositionstages of wind dominated erosion processes in semi-arid grassland, respectively. These preliminary experimentalresults implied that magnetic susceptibility signals in surface soils will hopefully serve as a useful tool in theaccuracy assessment of wind dominated erosion and deposition in the temperate grassland regions.

1. Introduction

Wind erosion is a dynamic soil degradation process that comprisesthe detachment, transport and deposition of soil particles (Skidmore,1986). Measurement and quantification of wind erosion are difficult atlarger scales because they involve complex and dynamic processes thatare random and without specific boundaries (Shao, 2000).

Conventional methods such as field monitoring (Musick and Gillette,1990; Hoffmann et al., 2011) and wind tunnel experiments (Burri et al.,2013; Wang et al., 2013) can provide data at the plot scale and for shortterm, and wind erosion models developed from these data also have thesame limitations (Van Pelt et al., 2007). Alternative technologies areneeded to determine wind erosion for larger spatial scales (i.e., land-scape scale to regional scale) and longer temporal scales (i.e., tens to

https://doi.org/10.1016/j.catena.2017.12.009Received 8 November 2016; Received in revised form 31 October 2017; Accepted 8 December 2017

⁎ Corresponding author at: State Key Laboratory of Earth Surface Processes and Resource Ecology, School of Geography, Faculty of Geographical Science, Beijing Normal University,Xinjiekouwai Str. 19, 100875 Beijing, China.

E-mail address: keli@bnu.edu.cn (K. Zhang).

Catena 163 (2018) 33–41

0341-8162/ © 2017 Published by Elsevier B.V.

T

hundreds of years) to identify the spatial pattern of soil redistributioninduced by wind, which is important for a better understanding of re-gional dust sources and sink patterns and implementing more rationalland management. The use of the tracing technique is an effectivemeans to meet these needs.

Fallout environmental radionuclides (FRNs) have been employed asreliable tracers for soil redistribution induced by water (Zapata et al.,2002; Ritchie and Ritchie, 2007). FRNs recent application in winderosion studies confirms its feasibility and implied potential in winderosion research (Yan and Shi, 2004; Van Pelt et al., 2007; Liu et al.,2008; Funk et al., 2012; Yang et al., 2013), although there are somelimitations in this field such as the challenge to find appropriate re-ference sites (Chappell, 1999). The FRN technique has been considereda cost-effective method (Ritchie and McHenry, 1990); however, it stillrequires high costs in terms of both equipment and measuring time (Liuet al., 2016). Another restriction of using FRNs in evaluating soil re-distribution is their fixed and limited time spans. For example, the useof 137Cs has been limited to evaluation of soil erosion and depositionsince the 1960s.

Soil magnetic susceptibility (MS) can also serve as a tracing tool forsoil redistribution and it is a commonly used magnetic property that ismeasured easily and economically (Jordanova, 2017). MS does notrequire time spans that are as specific as those associated with FRNs,and it can be notably flexible in its interpretation. In general, naturalsoil magnetism originates from soil forming process, which is mainlyinfluenced by regional hydrothermal environment. On the basis of thesame soil parent material, the stronger the pedogenesis is, the strongerthe soil magnetism is. This is a long-term process lasting hundreds of orthousands of years at least (Jordanova, 2017). Additionally, enhance-ment of MS in natural surface soils worldwide is a common phenom-enon (Mullins, 1977; Thompson and Oldfield, 1986). By comparison,short-term accelerated erosion events can induce the redistribution ofMS profiles in original natural surface soils. This new features of MSprofiles are usually connected with soil redistribution process. There-fore, the MS variation between topsoil and subsoil can be used toquickly infer a link with water erosion and deposition on a hillslope(Sadiki et al., 2009; Jordanova et al., 2014; Liu et al., 2015). Con-siderable effort has been devoted in evaluating the performance of MSin tracing soil redistribution since its initial application. So far, MS hasbeen proven to be an effective and efficient tracing property in soils indifferent regions, such as Eastern Rif, Morocco (Sadiki et al., 2009),hilly regions in Iran (Mokhtari Karchegani et al., 2011; Ayoubi et al.,

2012a), Northeast Bulgaria (Jordanova et al., 2014) and NortheastChina (Liu et al., 2015). However, the MS technique has only beenemployed for soil redistribution induced by water and its application inassessing wind dominated erosion has not been reported. This process issimilar to the development of a process in which FRNs are initially usedto trace the soil redistribution associated with water erosion and then totrace the soil redistribution associated with wind erosion. To ourknowledge, as a natural property of soil, the use of MS should be fea-sible in tracing wind erosion as well as water erosion, and it needs to betested and verified.

Major wind erosion studies focus on croplands and desert regions. Inthe last two decades, there has been increased attention regardingtemperate grasslands, which have better vegetation cover than desertsand are located in the mid-latitude regions with continental climate(Archibold, 1995) because they are sensitive and vulnerable to winderosion induced by climate change and anthropogenic influences suchas overgrazing (Shinoda et al., 2011; Wang et al., 2015). The Xilingelegrassland is a typical temperate grassland that is severely impacted bywind dominated erosion. It has changed from a best-preserved grass-land to one of the seven most severe wind erosion districts in NorthChina (Han et al., 2008). The physical and chemical properties of soilhave been considerably influenced by soil erosion in this area (Kölblet al., 2011). Wind erosion at landscape scale is hindering sustainableecology and agricultural development in the Xilingele grassland.

The objectives of this study were preliminarily to (i) investigate thespatial distribution and variations of MS in the surface soils at thelandscape scale, (ii) understand the links between MS and soil redis-tribution dominated by wind erosion; and (iii) further assess the fea-sibility of MS technique to study wind erosion to complement ourpresent research methods.

2. Materials and methods

2.1. Description of the study site

The study area is a semi-arid grassland transect 12.8 km long and1.6 km wide (Fig. 1c), located in the southern part of the Xilin RiverCatchment, 70 km southeast of the district capital city, Xilinhot, InnerMongolia (Fig. 1a). Its terrain is typically undulating, with an elevationranging from 1150 to 1350 m above sea level (Fig. 1c). The area isdominated by a mid-latitude continental semi-arid climate that featuresdry and cold winters and warm summers, with an average annual

Fig. 1. Map of the study area (a) and sampling sites (b, c) in the study. (b) is a satellite image acquired on September 10, 2012.

L. Liu et al. Catena 163 (2018) 33–41

34

temperature of 3.0 °C and an average annual precipitation of 263.5 mmfor the period from 1981 to 2010 (National Meteorological InformationCenter in China). Approximately 75% of the rainfall occurs betweenJune and September. The annual average wind speed is 3.4 m s−1, andthe prevailing wind direction during the wind erosion season isnorthwest.

The dominating soil in the semi-arid region, formed on basalt pla-teaus, is Kastanozem, according to the FAO/UNESCO System of SoilClassification. The parent material is universally fine-sand loess, only afew detrital acid igneous rock distributed around volcano craters. >80% of soil particles in the silt and sand fractions indicate that thewind-blown dust plays an important part in soil pedogenesis (Funket al., 2012; Hoffmann et al., 2008b). Detailed soil texture informationof the study transect is provided in Table 1. Usually, the soils of thestudy area are vulnerable to wind erosion due to their texture anderosive climate in the winters and early springs.

The transect of the study area is a part of a typical steppe ecosystemdominated by the species Leymus chinensis in the undulating easternterrain and Stipa grandis in the flat western plain (Fig. 1b). Multiplefactors influencing wind erosion include climate, topography, soil type,vegetation, grazing intensity and land use in the selected transect. Landuse types include grazing grassland and farmland. Grazing intensityincludes different levels of grazing ranging from ungrazed sites withhigh dense vegetation to sites with intensive grazing.

2.2. Field survey and sampling

A total of 160 survey and sampling sites with a grid of400 m× 400 m were designed to investigate MS and other properties(clay, silt, sand contents and organic carbon concentrate) of surfacesoils in the transect (Fig. 2). Field investigation and soil sampling werecarried out in May 2014. The vegetation conditions including vegeta-tion height and coverage were investigated at each site, and the topo-graphic conditions including slope position, aspect and gradient werealso measured. Grazing intensity was determined based on the vege-tation conditions according to the classification standards of Hoffmannet al. (2008a). Four-tier grazing intensity was identified in the transect,with ungrazed areas (prohibited grassland since 1979 or 1999) fol-lowed by light, moderate and heavy grazing intensities. Particularly,there is no vegetation cover on the farmland in the study transect fromwinter to next spring seasons each year. The farmland usually is re-cognized as the areas suffering from serious wind erosion (Hoffmannet al., 2008a). Therefore, farmland was listed as the fifth grazing in-tensity, representing severe grazing intensity in the study.

Soil samples were collected from depths of 0 to 1 cm and 1 to 6 cmbelow the surface using a stainless-steel spatula and shovel at each site(Fig. 2). A total of 319 soil samples were collected at the 160 samplingsites.

2.3. Laboratory measurement

The samples were air-dried, ground and passed through a 2-mmsieve for measurements of MS and grain size distribution. Subsampleswere then ground and collected by totally passing the original samplesthrough a 0.149-mm sieve for the measurement of soil organic carbonconcentration (SOC).

The mass-specific MS at low frequency (χlf), absolute frequency-dependent MS (χfd) and the percentage frequency-dependent MS (χfd%)of the samples were determined using a Bartington MS2 magneticsusceptibility (Oxfordshire, UK) meter with an MS2B sensor at both low(470 Hz) and high (4700 Hz) frequencies (Dearing, 1994). For grain-size analysis, the samples were evaluated using the sieve-pipettemethod which is a combination of wet sieving of the fraction> 50 μm(in diameter) and the pipette sampling method based on Stokes' law for

Table 1Soil texture of the surface soils (0–1 cm and 1–6 cm layers) across the transect in the studyarea.

Grain size Particle composition (%)

0–1 cm soil layer (n = 160) 1–6 cm soil layer(n = 159)

Sand (2.0–0.05 mm) 56.9 ± 11.2a 62.0 ± 8.9Silt (0.05–0.002 mm) 26.9 ± 8.3 21.7 ± 6.2Clay (< 0.002 mm) 16.0 ± 3.7 15.6 ± 3.7

a The mean value ± standard deviation.

Fig. 2. Map of the magnetic susceptibility sampling sites across the transect and the selected 137Cs sampling sites in the study by Funk et al. (2012).

L. Liu et al. Catena 163 (2018) 33–41

35

the silt (2–50 μm) and clay (< 2 μm) fractions (Dane and Topp, 2002).The organic carbon concentration for each sample was measured by thedry combustion method using the Vario TOC cube analyzer (ElementarAnalysensysteme GmbH, Hanau, Germany).

2.4. Data analysis

Two group statistical results are derived from one dataset with totalMS data and one dataset with asterisk excluding several MS outliers(Table 2). According to our field investigation, surface soils near or involcano craters in the grid sampling sites are developed on acid igneousrocks. These soils possess much higher χlf values(> 100 × 10−8 m3 kg−1) and lower χfd% values, compared with theaverage MS values of the total soil samples. Then, MS data of the out-liers (n = 6 in the surface layers) are excluded from the whole dataset(n = 160 in the surface layers) in the all of follow-up soil analysis.

Descriptive statistical parameters were calculated, including thetotal number of samples, median, geometric mean, standard deviation,coefficient of variation, minimum, maximum, kurtosis and skewness.These parameters were used to determine the average values, disper-sion degree and skewness of the data. t-test was applied to identify thesignificant differences of MS values between 0–1 cm and 1–6 layers.Significant differences of MS, grain size content and organic carbonconcentration between various grazing intensities were analyzed usinga one-way ANOVA.

Paired MS values (χlf, χfd%) and the rates of soil erosion and de-position dominated by wind were acquired using spatial map matchingto evaluate the relationships between MS and soil redistribution.Specifically, one MS dataset for a depth of 0–6 cm was acquired byweighted calculation using the total MS data for soil samples collectedat depths of 0–1 cm and 1–6 cm. Maps of the spatial distribution of MSvalues (χlf, χfd%) within a depth of a 0–6 cm were then interpolatedusing the Inverse Distance Weighted method, IDW. Finally, the pairedMS values (χlf, χfd%) and soil erosion and deposition rates were de-termined from the maps of the spatial distribution of MS values (χlf,χfd%) within a depth of 0–6 cm using the pinpointed locations of theknown 137Cs sampling sites (Fig. 2) from the previous study by Funket al. (2012).

3. Results and discussion

3.1. Characteristics of soil magnetic susceptibility

The statistical features of the entire dataset of MS values for the0–1 cm and 1–6 cm soil layers across the transect of the study area are

shown in Table 2. The χlf values in the surface soils (0–1 cm and 1–6 cmsoil layers) ranged from 31.0 to 97.8 × 10−8 m3 kg−1 and from 30.0 to96.1 × 10−8 m3 kg−1, respectively. The mean χlf values of the twosurface soil layers were close and varied with the same magnitude(Table 2). The geometric mean of the χlf value in the 0–1 cm layer wasslightly higher than in the 1–6 cm layer. This increase in the χlf valueswithin the surface soil profile (Table 2, Fig. 3a and b) agreed with theprevailing magnetic increase phenomenon of undisturbed soil profilesobserved in most types of soils in the temperate zone (Mullins, 1977;Thompson and Oldfield, 1986).

χfd is used as an indicator of the concentration of nano-sized(d < 0.03 μm) superparamagnetic (SP) minerals in soils and the re-lative contribution of the SP fraction in soils is expressed as χfd%(Dearing, 1994). Environmental samples can be divided into fourclasses based on the semi-quantitative relationships between χfd%value and SP mineral content in soils (Dearing, 1994): samples withχfd%< 2% and SP concentrations< 10%; samples with χfd% between2 and 10% in which there is a mixture of SP and coarser non-SP grains;samples with χfd% ranging between 10 and 14% and SP concentra-tion> 75%. χfd values for the 0–1 cm and 1–6 cm soil layer variedfrom 1.2 to 6.0 × 10−8 m3 kg−1 (Fig. 3c) and from 1.4 to6.1 × 10−8 m3 kg−1 (Fig. 3d), respectively; χfd% values for the 0–1 cmand 1–6 cm soil layer ranged from 3.3 to 8.0% (Fig. 3e) and from 3.2 to7.4% (Fig. 3f), respectively (Table 2). These suggested that there weremixtures of nano-sized SP grains and coarser non-SP grains in the soilsof the study area. In general, natural soils have higher χfd% values withstronger pedogenesis. χfd% value varies with the concentrate of nano-sized SP grains, which is sensitive to long-term wind erosion processes.Therefore, average χfd% value in the 0–1 cm soil layer was significantlysmaller than that in the 1–6 cm soil layer (Table 2, Fig. 3e and f), in-dicating that wind erosion processes prevailed in the study area.

3.2. Wind erosion factors and magnetic susceptibility in surface soils

Vegetation cover is one of the most important factors influencingwind erosion. Grazing intensity has direct impacts on the vegetationcoverage and wind erosion in the Xilingele grassland (Hoffmann et al.,2008a). Thus, it is important to examine the links between soil MS andgrazing intensity to evaluate their relationships to wind erosion pro-cesses.

The differences of χfd values (△χfd) between 0–1 cm layer and1–6 cm layer under each grazing intensity classification were analyzed(Table 3). Table 3 shows that △χfd values were positive for ungrazedareas (△χfd > 0) and were mostly negative (△χfd ≤ 0) in areas withother grazing intensities. These results indicated that soil redistribution

Table 2Statistics for magnetic susceptibility data for the 0–1 cm and 1–6 cm soil layers across the transect of the study area.

Magnetic parameter Soil depth n Mean Geometricmean

Standarddeviation

Coefficient ofvariation

Minimum Median Maximum Range Kurtosis Skewness Pb

(cm)

χlf (10−8 m3 kg−1) 0–1 160 68.6 66.0 24.6 0.36 31.0 65.2 293.1 262.0 44.6 5.5 0.0021–6 159 65.4 63.4 18.1 0.28 30.0 63.7 172.4 142.5 10.6 2.3

χfd (10−8 m3 kg−1) 0–1 160 3.3 3.2 1.0 0.30 1.2 3.3 7.5 6.3 2.4 0.9 0.1271–6 159 3.4 3.2 1.1 0.33 1.4 3.3 6.8 5.4 −0.1 0.5

χfd% (%) 0–1 160 4.9 4.8 0.8 0.17 2.4 4.9 8.0 5.6 1.5 0.3 0.0001–6 159 5.1 5.0 0.9 0.18 3.1 5.1 7.4 4.3 −0.5 0.1

χlfa (10−8 m3 kg−1) 0–1 154 65.2 64.0 12.0 0.18 31.0 64.8 97.8 66.7 0.4 −0.3 0.000

1–6 154 63.1 61.8 12.3 0.19 30.0 63.6 96.1 66.1 0.2 −0.1χfd

a (10−8 m3 kg−1) 0–1 154 3.2 3.1 0.9 0.27 1.2 3.2 6.0 4.8 −0.1 0.1 0.0551–6 154 3.3 3.1 1.1 0.32 1.4 3.2 6.1 4.7 −0.3 0.4

χfd%a (%) 0–1 154 4.9 4.8 0.8 0.16 3.3 4.9 8.0 4.7 1.5 0.5 0.0001–6 154 5.1 5.1 0.9 0.17 3.2 5.2 7.4 4.1 −0.5 0.1

n, the number of samples.a The datasets have excluded the outliers (n = 6) that represent the poor soils formed on acidic igneous rocks.b P < 0.05 indicates significant differences of magnetic susceptibility values between 0–1 cm and 1–6 cm layers using the pair-sample t-test.

L. Liu et al. Catena 163 (2018) 33–41

36

dominated the grassland surface soils in the study region and probablyvaried with the spatial variations of grazing intensity. They also impliedthat dust deposition existed in other areas, varying with the grazingintensity. These findings were in accord with the results of previousstudies (Funk et al., 2012; Hoffmann et al., 2011) showing that thestudy area were suffering wind erosion.

As shown in Fig. 4a, the χlf values in the surface soils decreased oneach grazing intensity except for ungrazed areas. As the grazing in-tensity increases, the loss of magnetic minerals in surface soils alsoincreased due to the lack of sufficient vegetation coverage for pro-tecting topsoil, which had been confirmed by the relationship betweensoil loss induced by wind and vegetation coverage observed in this area

(Yan et al., 2013). However, Fig. 4b depicts there was no distinct trendas a whole between χfd% values and grazing intensity, indicating veryfine soil particles (SP fraction) were not well matched with grazingintensity. This was probably attributed that the χfd% values in surfacesoils include the environmental information of both wind erosion anddust deposition processes. The dust deposition processes are complexdynamic eolian processes. Thus, in view of the present preliminaryfinding, further research will be necessary to build a direct link betweenthe χfd% values in natural surface soils and grazing intensity. At last,the fact that the χlf and χfd% values for ungrazed areas are less thanthose for other grazing intensities (Fig. 4) should be ascribed to the biasof representation of the small sample size (n = 4).

Fig. 3. Spatial distribution of soil magnetic susceptibility in the 0–1 cm and 1–6 cm soil layers across the transect.

L. Liu et al. Catena 163 (2018) 33–41

37

However, the features of MS values in the surface soil profiles (the0–1 cm and 1–6 cm layers) were obviously different in ungrazedgrassland and areas with other grazing intensities based on the resultsin the Fig. 4. For the ungrazed surface soils (Fig. 4a, b), there was atrend of synchronous variation between the contents of nano-sized SPminerals and total ferrimagnetic minerals with increases in soil depth.Compared with the ungrazed sites, opposite variations of χfd% valueswith increases in soil depth existed in other grazing intensities Thisindicated very fine surface soil particles, including nano-sized SPgrains, in the surface soils (0–1 cm layer) could be easily eroded bywind without sufficient vegetation protection (Colazo and Buschiazzo,2015). The largest amounts of magnetic mineral loss in soil on arableland and heavily grazed land were probably due to the disturbance ofthe mixture effect of plowing and the high-intensity livestock grazingon surface soils.

The relationships between soil grain size and χlf values for differentgrazing intensities are shown in Fig. 5. Positive trends existed in therelationships between the clay fractions and the χlf values as well as therelationship between the silt fractions and the χlf values. By contrast,negative trends existed between the sand fractions and the χlf values.This indicated that there was a strong link between the ferrimagneticmineral content in surface soils and finer particle fractions, which wasconsistent with the results of investigations of the magnetic character-istics of most natural soils (Maher and Taylor, 1988; Mullins, 1977).Additionally, there was notable variation between various grazing in-tensities and soil grain size fractions, as shown in Fig. 6. A highergrazing intensity caused a lower concentration of finer particles (clay

and silt) in the surface soils in the 0–6 cm layer, which were directlysubject to wind erosion based on the results of the quantitative effects ofwind erosion on the soil texture under various vegetation coverageconditions (Colazo and Buschiazzo, 2015; Yan et al., 2013). In general,the concentration of fine ferrimagnetic particles in surface soils of thegrassland in the study area were reduced with the increase in grazingintensity, which confirms the previous findings that overgrazing by li-vestock was accelerating the wind erosion impact on surface soils in theXilingele grassland (Han et al., 2008; Yan et al., 2013).

SOC is a very important component of soils (Ayoubi et al., 2012b).The amount of SOC usually represents the degree of soil maturity. Inaddition, SOC can promote the formation of secondary ferrimagneticminerals, the major contributors to soil MS value, in natural soils(Maher and Taylor, 1988). Fig. 7 depicts a higher grazing intensitymatched with a lower amount of SOC, indicating that the reduction ofvegetation coverage caused the depletion of SOC (Yan et al., 2013).Because of pedogenic significance of the MS parameters, MS could beused as a tool connecting ferrimagnetic minerals in soils with grazingintensity.

3.3. Implications for wind erosion based on magnetic susceptibility in soils

The MS values and the erosion rates determined by Funk et al.(2012) based on 137Cs inventories were compared to evaluate the linkbetween MS and wind erosion (Fig. 8, Table 4). There was a distinctincrease in the χlf values with the decrease of the soil erosion rate,while there was reduction in the χlf values with the increase of the dustdeposition rate (Fig. 8a). The similar variations of χfd% values with soilerosion or dust deposition rate are depicted in Fig. 8b.

The application of magnetic susceptibility techniques in the eva-luation of wind erosion is seldom reported. Previous MS-related re-search in semi-arid and arid regions of China mainly focused onQuaternary climates and environments at geologic time scales (Maherand Thompson, 1999; Maher et al., 2009). In comparison, wind erosionprocesses involve the rapid change of grain size distribution in surfacesoils at time scales related to human activities, which typically rangefrom a few years to decades. Fine soil particles, including ferrimagneticminerals, can be easily eroded because of human disturbance andtransported farther away due to wind sorting (Chepil, 1957; Chien andWan, 1983). The composition of soil particle in surface soils can thenbecome coarser over time (Yan et al., 2013). In general, the MS para-meters (χlf, χfd%) are indicators of the concentrations and various grainsizes of ferrimagnetic minerals in soils (Dearing, 1994; Liu et al., 2012);therefore, it is possible that the MS parameters could be consideredpotential indicators of soil redistribution by wind.

Table 3Statistics for the difference between 0–1 cm and 1–6 cm soil layer regarding the fre-quency-dependent magnetic susceptibility (△χfd) under different grazing intensities.

Grazing intensity n △χfd > 0 △χfd = 0 △χfd < 0

n Ratioa

(%)n Ratioa

(%)n Ratioa

(%)

Ungrazed 4 4 100 0 0 0 0Light 13 3 23.1 0 0 10 76.9Moderate 80 27 33.8 3 3.8 50 62.5Heavy 41 21 51.2 6 14.6 14 34.1Severe (Farmland) 16 6 37.5 4 25 6 37.5

n, the number of samples.a Assuming that △χfd equals the χfd value for the 0–1 cm layer minus the χfd value for

the 1–6 cm layer. Ratio (%) equals the number of sampling sites with △χfd > 0,△χfd = 0, and △χfd < 0 in the percentage of the total number of sampling sites foreach grazing intensity.

Fig. 4. Variation of soil magnetic susceptibility (χlf, χfd%) for five grazing intensities in the 0–1 cm and 1–6 cm soil layers. The grazing intensity numbers 1, 2, 3, 4 and 5 denote ungrazed,light, moderate, heavy, and severe grazing intensities (equivalent to arable land), respectively. Error bar represents the standard deviation.

L. Liu et al. Catena 163 (2018) 33–41

38

Funk et al. (2012) showed that the concentration of 137Cs increasedwith the increase of eolian 137Cs deposits in surface soils in the Xilingelegrassland. On the contrary, the MS (χlf, χfd%) values increased with thedecrease of the dust deposition rate, especially for χfd% values, in thisstudy (Fig. 8a, b). This contradiction might be due to adsorption dif-ferences between soil magnetic materials and 137Cs on different sizes offine grains. The SP mineral concentrations and 137Cs radioactivity levelare both closely related to the clay particles in soils. The χlf values areindicators of the magnetic minerals in soils with various grain sizesranging from sand to clay. The χfd% values mainly indicate the amountof nano-sized SP grains (< 0.03 μm) in undisturbed soils (Dearing,1994), which is a primary contributor to the natural soil MS. However,as an artificial environmental radionuclide, 137Cs in natural soils isusually imported from external sources. 137Cs is mainly absorbed intothe fine-grain silt and clay fraction (< 36 μm) in soils (He and Walling,

1996; Bihari and Dezső, 2008). There is a particle size gap between0.03 μm and 36 μm due to different responses between the MS and137Cs to soil particle size. The dust grains fraction d < 0.03 μm is easilysuspended in the air for several years (Chien and Wan, 1983; Pye, 1987;Hoffmann et al., 2008b). By contrast, the soil grains (> 0.03 μm) candeposit in a shorter period of time. Thus, the wind-blown particles(> 0.03 μm) could significantly change the particle size composition ofsurface soils which was supported by the analysis of the soil grain sizedata in this study (Fig. 6). For example, compared with the ungrazedgrassland area, the average clay content of soils in the 0–1 cm layer inthe farmland area decreased by 31.9%, the average silt content de-creased by 34.5%, and the average sand content increased by 32.0%.Moreover, the eolian dust from the Mongolian Gobi desert, which is oneof the major dust sources for the Xilingele grassland (Hoffmann et al.,2008c), is mainly composed of quartz with negligible MS values (Maher

Fig. 5. Relationship between soil grain size (clay, silt or sand fraction) and magnetic susceptibility (χlf) in soils in the 0–1 cm and 1–6 cm soil layers under five different grazingintensities, including ungrazed, light, moderate, heavy, and severe grazing intensities (equivalent to arable land).

Fig. 6. Variation in soil grain size (clay, silt or sand fraction) for five grazing intensities in the 0–1 cm and 1–6 cm soil layers. The grazing intensity numbers 1, 2, 3, 4 and 5 denoteungrazed, light, moderate, heavy, and severe grazing intensities (equivalent to arable land), respectively. Error bar represents the standard deviation.

L. Liu et al. Catena 163 (2018) 33–41

39

et al., 2009). Consequently, the total amount of 137Cs increased with theaccumulation of the eolian deposits (> 0.03 μm), and the concentra-tion of native minerals with high magnetism in surface soils was dilutedby exotic materials (like fine quartz) with very low magnetism.Therefore, the above relationships between MS-soil erosion rate andMS-dust deposition rate (Fig. 8) indicated that MS could have the po-tential for use in distinguishing the stages of soil erosion and dust de-position associated with wind erosion processes if a reference site canbe introduced that has soil MS in a topsoil profile with no erosion ordeposition caused by wind.

The MS technique not only has technical advantages (Dearing,1994), but also provides a new perspective to quantitatively investigatesoil redistribution processes caused by water (de Jong et al., 1998;Dearing et al., 1986; Liu et al., 2015). This study further confirmed thatthe MS technique could hopefully be used to determine the spatialdistribution of soil erosion and dust deposition dominated by wind inthe semi-arid grassland. However, further studies are still needed toexamine the relationships between MS and the soil redistribution rateusing directly measured data, such as acquired MS data for various soilgrain sizes and MS data for soil erosion rates and deposition rates de-termined directly by field dust-trapping devices. The difference be-tween 137Cs and MS in their response mechanisms for fine-grain rangeof soils may be the next stage of research priorities for exploring the

relationship between these two tracing techniques in soil erosion re-search. Additionally, it is highly desirable to explore the backgroundinformation of soil MS involving the variations of MS values with dif-ferent land uses and in different soil profiles in this region, which canprovide more information about pedogenesis and historical wind ero-sion events during land degradation periods or at longer temporalscales.

4. Conclusions

Magnetic susceptibility characteristics of surface soils at landscapescale and related wind erosion parameters were investigated in theXilingele grassland of China. The results showed that (1) the χlf valuesin surface soils ranged from 30.0 to 97.8 × 10−8 m3 kg−1, χfd valuesfrom 1.2 to 6.1 × 10−8 m3 kg−1, and χfd% values from 3.2 to 8.0%;(2) the variations of χlf values in the 0–6 cm surface soil layer wererelated to grazing intensity, grain size and organic carbon concentra-tion, which suggested that response of the surface soil magnetism towind erosion was sensitive, and (3) the MS parameters (χlf, χfd%) werepositively correlated to soil erosion rates and negatively correlated withdust deposition rates, indicating that MS parameters could probably beused to identify the soil erosion and deposition stages of wind domi-nated erosion processes. Additionally, to support these preliminary re-sults, future studies under artificial control condition and natural con-dition will be conducted to explore the quantitative relationshipsbetween MS and the soil redistribution rate using directly measureddata. In general, inexpensive but efficient MS measurements have thepotential for accurate assessment of the spatial distribution of soil winderosion. The MS technique can provide a new perspective for exploringthe spatial distribution and formation mechanisms of soil redistributiondominated by wind. It also hopefully serve for soil degradation mon-itoring and assessment in temperate grassland regions.

Acknowledgments

This study was supported by the National Natural ScienceFoundation of China (Grant No. 41301282 and No. 41730748), theChinese Academy of Sciences (CAS) “Light of West China” Program andChina Postdoctoral Science Foundation (Grant No. 2016 M602884).Special appreciation was given to Qianqian Qiu and Chuanlong Sun fortheir assistance in soil analysis and Jing Wang for her field assistance.The authors would also like to thank the reviewers for the constructivecomments on this paper.

Fig. 7. Variation of soil organic carbon (SOC) for five grazing intensities in the 0–1 cmand 1–6 cm soil layers. The grazing intensity number 1, 2, 3, 4 and 5 denote ungrazed,light, moderate, heavy, and severe grazing intensities (equivalent to arable land), re-spectively. Error bar represents the standard deviation.

Fig. 8. Relationship between the mean erosion or deposition rate and magnetic susceptibility (χlf, χfd%) in surface soils.

L. Liu et al. Catena 163 (2018) 33–41

40

References

Archibold, O., 1995. Ecology of World Vegetation. Chapman & Hall, London.Ayoubi, S., Ahmadi, M., Abdi, M.R., Afshar, F.A., 2012a. Relationships of Cs-137 in-

ventory with magnetic measures of calcareous soils of hilly region in Iran. J. Environ.Radioact. 112, 45–51.

Ayoubi, S., Mokhtari Karchegani, P., Mosaddeghi, M.R., Honarjoo, N., 2012b. Soil ag-gregation and organic carbon as affected by topography and land use change inwestern Iran. Soil Tillage Res. 121, 18–26.

Bihari, Á., Dezső, Z., 2008. Examination of the effect of particle size on the radionuclidecontent of soils. J. Environ. Radioact. 99, 1083–1089.

Burri, K., Gromke, C., Graf, F., 2013. Mycorrhizal fungi protect the soil from wind erosion:a wind tunnel study. Land Degrad. Dev. 24, 385–392.

Chappell, A., 1999. The limitations of using Cs-137 for estimating soil redistribution insemi-arid environments. Geomorphology 29, 135–152.

Chepil, W.S., 1957. Sedimentary characteristics of dust storms. I. Sorting of wind-erodedsoil material. Am. J. Sci. 255, 12–22.

Chien, N., Wan, Z., 1983. Mechanics of Sediment Transport. Science Press, Beijing (inChinese).

Colazo, J.C., Buschiazzo, D., 2015. The impact of agriculture on soil texture due to winderosion. Land Degrad. Dev. 26, 62–70.

Dane, J.H., Topp, G.C., 2002. Methods of Soil Analysis, Part 4—Physical Methods. SoilScience Society of America, Madison.

Dearing, J.A., 1994. Environmental Magnetic Susceptibility, Using the Bartington MS2System. Chi Publishers, Kenilworth.

Dearing, J.A., Morton, R.I., Price, T.W., Foster, I.D.L., 1986. Tracing movements of topsoilby magnetic measurements - two case studies. Phys. Earth Planet. Inter. 42, 93–104.

Funk, R., Li, Y., Hoffmann, C., Reiche, M., Zhang, Z., Li, J., Sommer, M., 2012. Using Cs-137 to estimate wind erosion and dust deposition on grassland in Inner Mongolia-selection of a reference site and description of the temporal variability. Plant Soil351, 293–307.

Han, X., Yang, L., Zhou, H., Jiang, Y., 2008. The analysis on spatial distribution andcharacteristics of strong aeolian erosion districts in north China. J. Arid Land Res.Environ. 22, 113–116 (in Chinese with English abstract).

He, Q., Walling, D.E., 1996. Interpreting particle size effects in the adsorption of Cs-137and unsupported Pb-210 by mineral soils and sediments. J. Environ. Radioact. 30,117–137.

Hoffmann, C., Funk, R., Li, Y., Sommer, M., 2008a. Effect of grazing on wind drivencarbon and nitrogen ratios in the grasslands of Inner Mongolia. Catena 75, 182–190.

Hoffmann, C., Funk, R., Sommer, M., Li, Y., 2008b. Temporal variations in PM10 andparticle size distribution during Asian dust storms in Inner Mongolia. Atmos. Environ.42, 8422–8431.

Hoffmann, C., Funk, R., Wieland, R., Li, Y., Sommer, M., 2008c. Effects of grazing andtopography on dust flux and deposition in the Xilingele grassland, Inner Mongolia. J.Arid Environ. 72, 792–807.

Hoffmann, C., Funk, R., Reiche, M., Li, Y., 2011. Assessment of extreme wind erosion andits impacts in Inner Mongolia, China. Aeolian Res. 3, 343–351.

de Jong, E., Nestor, P.A., Pennock, D.J., 1998. The use of magnetic susceptibility tomeasure long-term soil redistribution. Catena 32, 23–35.

Jordanova, N., 2017. Soil Magnetism: Applications in Pedology, Environmental Scienceand Agriculture. Academic Press, London.

Jordanova, D., Jordanova, N., Petrov, P., 2014. Pattern of cumulative soil erosion andredistribution pinpointed through magnetic signature of Chernozem soils. Catena120, 46–56.

Kölbl, A., Steffens, M., Wiesmeier, M., Hoffmann, C., Funk, R., Krümmelbein, J.,Reszkowska, A., Zhao, Y., Peth, S., Horn, R., Giese, M., Kögel-Knabner, 2011. Grazingchanges topography-controlled topsoil properties and their interaction on differentspatial scales in a semi-arid grassland of Inner Mongolia, P.R. China. Plant Soil 340,35–58.

Liu, J., Qi, Y., Shi, H., Zhuang, D., Hu, Y., 2008. Estimation of wind erosion rates by using

Cs-137 tracing technique: a case study in Tariat-Xilin Gol transect, Mongolian. Chin.Sci. Bull. 53, 751–758.

Liu, Q., Roberts, A.P., Larrasoaña, J.C., Banerjee, S.K., Guyodo, Y., Tauxe, L., Oldfield, F.,2012. Environmental magnetism: principles and applications. Rev. Geophys. 50,1–50.

Liu, L., Zhang, K., Zhang, Z., Qiu, Q., 2015. Identifying soil redistribution patterns bymagnetic susceptibility on the black soil farmland in Northeast China. Catena 129,103–111.

Liu, L., Zhang, K., Zhang, Z., 2016. An improved core sampling technique for soil mag-netic susceptibility determination. Geoderma 277, 35–40.

Maher, B.A., Taylor, R.M., 1988. Formation of ultrafine-grained magnetite in soils. Nature336, 368–370.

Maher, B.A., Thompson, R., 1999. Quaternary Climates, Environments and Magnetism.Cambridge University Press, Cambridge.

Maher, B.A., Mutch, T.J., Cunningham, D., 2009. Magnetic and geochemical character-istics of Gobi desert surface sediments: implications for provenance of the Chineseloess plateau. Geology 37, 279–282.

Mokhtari Karchegani, P., Ayoubi, S., Lu, S.G., Honarju, N., 2011. Use of magnetic mea-sures to assess soil redistribution following deforestation in hilly region. J. Appl.Geophys. 75, 227–236.

Mullins, C., 1977. Magnetic susceptibility of the soil and its significance in soil science–areview. J. Soil Sci. 28, 223–246.

Musick, H.B., Gillette, D.A., 1990. Field evaluation of relationships between a vegetationstructural parameter and sheltering against wind erosion. Land Degrad. Dev. 2,87–94.

Pye, K., 1987. Aeolian dust and dust deposits. Academic Press, London.Ritchie, J.C., McHenry, J.R., 1990. Application of radioactive fallout Cs-137 for mea-

suring soil erosion and sediment accumulation rates and patterns: a review. J.Environ. Qual. 19, 215–233.

Ritchie, J.C., Ritchie, C.A., 2007. Bibliography of publications of Cs-137 studies related toerosion and sediment deposition. http://www.ars.usda.gov/Main/docs.htm?docid=15237, Accessed date: 21 July 2016.

Sadiki, A., Faleh, A., Navas, A., Bouhlassa, S., 2009. Using magnetic susceptibility toassess soil degradation in the eastern Rif, Morocco. Earth Surf. Process. Landf. 34,2057–2069.

Shao, Y., 2000. Physics and modelling of wind erosion. In: Atmospheric andOceanographic Sciences Library. Vol. 23 Kluwer Academic Publishers, Dordrecht.

Shinoda, M., Gillies, J.A., Mikami, M., Shao, Y., 2011. Temperate grasslands as a dustsource: knowledge, uncertainties, and challenges. Aeolian Res. 3, 271–293.

Skidmore, E.L., 1986. Soil Erosion by Wind: An Overview. In: El-baz, F., Hassan, M.H.A.(Eds.), Physics of Desertification, Dordrecht, pp. 263–271.

Thompson, R., Oldfield, F., 1986. Environmental magnetism. Allen & Unwin, London.Van Pelt, R.S., Zobeck, T.M., Ritchie, J.C., Gill, T.E., 2007. Validating the use of Cs-137

measurements to estimate rates of soil redistribution by wind. Catena 70, 455–464.Wang, X., Wang, G., Lang, L., Hua, T., Wang, H., 2013. Aeolian transport and sandy

desertification in semiarid china: a wind tunnel approach. Land Degrad. Dev. 24,605–612.

Wang, T., Xue, X., Zhou, L., Guo, J., 2015. Combating aeolian desertification in northernChina. Land Degrad. Dev. 26, 118–132.

Yan, P., Shi, P., 2004. Using the Cs-137 technique to estimate wind erosion in GongheBasin, Qinghai Province, China. Soil Sci. 169, 295–305.

Yan, Y., Xin, X., Xu, X., Wang, X., Yang, G., Yan, R., Chen, B., 2013. Quantitative effects ofwind erosion on the soil texture and soil nutrients under different vegetation cov-erage in a semiarid steppe of northern China. Plant Soil 369, 585–598.

Yang, M.Y., Walling, D.E., Sun, X.J., Zhang, F.B., Zhang, B., 2013. A wind tunnel ex-periment to explore the feasibility of using beryllium-7 measurements to estimate soilloss by wind erosion. Geochim. Cosmochim. Acta 114, 81–93.

Zapata, F., Garcia-Agundo, E., Ritchie, J.C., Appleby, P.G., 2002. Introduction. In: Zapata,F. (Ed.), Handbook for the Assessment of Soil Erosion and Sedimentation UsingEnvironmental Radionuclides, IAEA, pp. 1–13.

Table 4Magnetic susceptibility in surface soils in the selected sampling sites and the paired soil erosion or deposition rate based on 137Cs inventories.

Site 137Csa Magnetic susceptibility

Codeb Grazing intensity 137Cs inventory mean erosion (−) or deposition (+) rate Soil depth n Mean χlf Mean χfd%

(Bq·m−2) (t·km−2·a−1) (cm) (10−8 m3 kg−1) (%)

Stipa gr. steppe ST1 heavily grazed 1330 −131 0–6 1 64.2 4.5ST2 Moderately (1) 1667 −55 0–6 1 70.6 4.9ST3 Moderately (2) 1185 −170 0–6 1 56.3 5.0ST4 lightly 2524 95 0–6 1 61.0 4.5ST5 ungrazed since 1979 2787 140 0–6 1 58.5 4.2

Leymus ch. steppe RF moderately 1967 0 0–6 1 65.5 5.5LE1 Moderately (1) 1652 −58 0–6 1 71.1 5.8LE2 Moderately (2) 2158 33 0–6 3 61.6 5.6LE3 ungrazed since 1999 2339 64 0–6 3 61.1 5.2LE4 ungrazed since 1979 2332 62 0–6 3 62.2 5.1

Arable farmland AF 1500 −91 0–6 7 64.6 5.4

n, the number of samples.a The 137Cs inventories and the average annual rate of erosion or deposition data are originally from Funk et al. (2012). N is number of sampling sites.b The Codes of the sampling sites are showed in Fig. 2.

L. Liu et al. Catena 163 (2018) 33–41

41