magnetic mineralogy of soils across the russian steppe

Magnetic mineralogy of soils across the Russian Steppe:climatic dependence of pedogenic magnetite formation

B.A. Maher a;�, A. Alekseev b, T. Alekseeva b

a Centre for Environmental Magnetism and Paleomagnetism, Lancaster Environment Centre, Department of Geography,Lancaster University, Lancaster LA1 4YB, UK

b Institute of Physicochemical and Biological Problems of Soil Science, Russian Academy of Sciences, Pushchino, Russia

Received 15 January 2003; received in revised form 23 July 2003; accepted 19 August 2003

Abstract

Formation of ferrimagnets in well-drained, buffered, unpolluted soils appears to be related to climate, andespecially rainfall. If robust, this magnetism/rainfall couple can be used to estimate past rainfall from buried soils,particularly the multiple soils of the Quaternary loess/soil sequences of Central Asia. However, dispute existsregarding the role of climate vs. dust flux for the magnetic properties of modern loessic soils. Here, we examine themineralogical basis of the magnetism/rainfall link for a climate transect across the loess-mantled Russian steppe,where, critically, dust accumulation is minimal at the present day. Magnetic and independent mineralogical analysesidentify in situ formation of ferrimagnets in these grassland soils ; increased ferrimagnetic concentrations areassociated with higher annual rainfall. XRD and electron microscopy show the soil-formed ferrimagnets are ultrafine-grained (6V50 nm) and pure. Ferrimagnetic contributions to Mo«ssbauer spectra range from 17% in the parent loessto 42% for a subsoil sample from the highest rainfall area. Total iron content varies little but the systematic magneticincreases are accompanied by decreased Fe2þ content, reflecting increased silicate weathering. For this region, parentmaterials are loessial deposits, topography is rolling to flat and duration of soil formation effectively constant. Thevariations in soil magnetic properties thus predominantly reflect climate (and its co-variant, organic activity) ^statistical analysis identifies strongest relationships between rainfall and magnetic susceptibility and anhystereticremanence. This magnetic response correlates with that of the modern soils across the Chinese Loess Plateau. Suchcorrelation suggests that the rainfall component of the climate system, not dust flux, is a key influence on soilmagnetic properties in both these regions.< 2003 Elsevier B.V. All rights reserved.

Keywords: soil magnetism; palaeoclimate; loessic soils; Russian steppe

1. Introduction

The magnetic mineralogy of sediments, includ-

ing soils, is an increasingly important naturalsource of climatic and environmental information.The pedogenic (i.e. in situ, soil-formed) magneticproperties of well-drained, bu¡ered and unpol-luted soils appear to be causally related to cli-mate, and speci¢cally, rainfall (Maher et al.,1994; Han et al., 1996). Evaluation of the robust-

0031-0182 / 03 / $ ^ see front matter < 2003 Elsevier B.V. All rights reserved.doi:10.1016/S0031-0182(03)00618-7

* Corresponding author. Fax: +44-1524-947099.E-mail address: [email protected] (B.A. Maher).

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart

Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341

www.elsevier.com/locate/palaeo

mailto:[email protected]

http://www.elsevier.com/locate/palaeo

ness of this climate/magnetism couple is impor-tant, as it may provide a quantitative transferfunction, enabling calculation of palaeoprecipita-tion through the Quaternary period, via analysisof the magnetic properties of palaeosols inter-bedded within the loess sequences of centralAsia (Heller et al., 1993; Maher et al., 1994). Ro-bust and quantitative proxies of rainfall are nota-bly scarce within the palaeoclimate context, andare especially signi¢cant for the densely-populatedregions of monsoonal Asia.

Sensitivity of certain soil iron compounds toclimate has previously been documented in themainstream soil science literature. Notably,Schwertmann and Taylor and collaborators dem-onstrated (mostly on the basis of bulk X-ray dif-fraction (XRD) analyses) the response of haema-tite and goethite to di¡erent, climatically-drivenpedogenic regimes. For example, the goethite^haematite ratio in soils varies systematically alongclimatic, hydrological and topographic transects,as a result of variations in pH, soil temperature,water activity and organic matter (Schwertmann,1988). Conversely, the magnetic properties of soilsare often dominated by formation of the stronglymagnetic (ferrimagnetic) oxides, magnetite andmaghemite ^ usually in concentrations indetect-able by XRD but easily measurable by routinemagnetic methods of analysis. Hence, magneticanalyses of soils provide an additional, sensitivewindow on soil iron and its response to climate.

Apparent coupling between magnetism of loes-sic soils and rainfall was ¢rst observed for modernsoils developed on the near-horizontal, homoge-nous parent substrates of the famous Loess Pla-teau region of north-central China (Maher et al.,1994; Liu et al., 1995). At the present day, thisregion experiences a strong gradient in rainfall,from values around 300 mm yr31 in the west, toV550 mm yr31 in the central Plateau and V700mm yr31 in the south. Most rainfall occurs in thesummer, due to monsoonal transport of warm,moisture-laden air from the Paci¢c across the Chi-nese mainland. The extent of westward penetra-tion of the monsoonal rainbelt varies with theintensity of the summer monsoon system. Maheret al. (1994) examined the relationships betweenpedogenic magnetic susceptibility and modern cli-

mate variables. The magnetic susceptibility (the‘magnetisability’) of the Chinese soils is domi-nantly contributed by trace amounts (V0.3%) ofmagnetite of ultra¢ne grain size (6V30 nm).(Because of its ¢ne grain size, this magnetite isvariably oxidised at its surface towards maghe-mite). To isolate the in situ, soil-formed magneticsignal, the pedogenic susceptibility (Mped) was de-¢ned for each soil as the maximum susceptibilityvalue (M) of the B horizon minus the M of theparent loess. Maher et al. (1994) found strongpositive correlation (R2 = 0.94) between the loga-rithm of pedogenic M and annual rainfall. Simi-larly, Han et al. (1996) examined a further 63 top-soil samples across the Loess Plateau and alsoidenti¢ed a direct (albeit polynomial) relationshipbetween rainfall and susceptibility.

Fig. 1 summarises the magnetic and rainfalldata for modern soils across the Chinese region,and also published magnetic data for soils acrossthe Northern Hemisphere temperate zone. Someof the scatter in the data may re£ect inclusion of‘unsuitable’ soils, i.e. soils with conditions inimi-cal to pedogenic formation of ferrimagnets. Theseinclude: poorly-drained or excessively acidic soils,eroded soils, or soils developing on slowly-weath-ering or iron-de¢cient substrates. Polluted soilsand burnt soils, on the other hand, may be exces-sively enriched in ferrimagnets (Maher, 1986;Maher and Thompson, 1999). However, anotherpossibility is that the scatter is real and that themagnetism/rainfall relationship is less signi¢cantthan has been proposed. Kukla and co-workers(Kukla et al., 1988; Porter et al., 2001) have pro-posed that the variations in magnetic susceptibil-ity across and within the Chinese loess sequencesare due to di¡ering rates of input of low-suscep-tibility dust. Indeed, Porter et al. (2001) suggestthat 84% of the susceptibility variance of themodern soils across the Loess Plateau is due tothe so-called ‘dust-dilution’ e¡ect. For the Chi-nese Loess Plateau, the confounding factor in pin-pointing the respective roles of climate and dust£ux is that these two factors co-vary across theregion. That is, annual rainfall increases and dust£ux decreases from the western to the southernand eastern areas of the Plateau. Hence, the‘dust’ school of thought interprets the higher


B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341322

soil magnetic values in the south and east of thePlateau as dominantly re£ecting reduced input oflow-susceptibility dust.

Given the scarcity of quantitative rainfall prox-ies available to researchers for palaeoclimatic re-construction, it is critical to test these opposinghypotheses ^ climate dependence vs. dust £ux de-pendence. The soil magnetism/climate relationshipwas tested recently in a geographically indepen-dent, pollution-free and geomorphologically sta-ble region, the Russian steppe (Maher et al.,2002). In contrast to the Chinese Loess Plateauregion, minimal dust accumulation occurs acrossthis region at the present day. Here, we examinein detail the mineralogical basis of the soil mag-netism/climate link observed in this region, by an-alysing the magnetic and mineralogical propertiesof 22 modern steppe soils, from a transect span-ning s 1000 km across the loessic plain from thenorthern Caucasus to the Caspian Sea. Climatedata for the region are available for the lastV100 years, from stations across the area. Thiswork follows in the pioneering footsteps not only

of Dokuchaev, who ¢rst identi¢ed regional scalesoil/climate links (Dokuchaev, 1883), but alsoother Russian soil scientists, including Babanin(1973), Vadyunina and Smirnov (1976) and Vo-dyunitsky (1981), the ¢rst to make large-scale ¢eldand laboratory analyses of soil magnetic proper-ties.

2. Sites and methods

The sampled transect exceeds V1000 km,southwest to northeast across the loess-mantled,exhumed marine plain from the northern £anks ofthe Caucasus, where glacial moraine is onlappedby the loess, to the northwestern margins of theCaspian Sea (Fig. 2). The northeastern sector ofthe transect grades into calcareous clay loams, onmarine deposits of late Pliocene/Pleistocene age.This geomorphologically stable, near-horizontalloessic and marine plain is contiguous to thewest with the eastern European loess belt. Thesteppe in this region is notably free of any distur-

Fig. 1. Pedogenic magnetic susceptibility (susc.A or B horizon3susc.C horizon) vs. annual rainfall, Chinese Loess Plateau, the Russiansteppe and additional sites in the Northern Hemisphere temperate zone (Maher and Thompson, 1999). At high rainfall totals (i.e.beyond the range shown on this graph), soils may become decalci¢ed, poorly bu¡ered and pedogenic magnetite may not formand/or may be actively dissolved (Maher, 1998). Hence the rainfall/susceptibility climofunction will break down at this point.


B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 323

bance to the continuous grassland cover; it is ex-tremely sparsely populated and no pollution sour-ces exist for the sample sites. As shown in Fig. 2,the climate of the region exhibits a marked gra-dient in precipitation, from V500 mm/yr for theStavropol region to V300 mm/yr around Volgo-grad. Precipitation is fairly evenly distributedthroughout the year. Summer temperatures reachV25‡C, winter temperatures vary from V35 to310‡C, with temperatures exceeding 10‡C onsV170 days/yr (State Meteorological Organisa-tion, 1966, 1968). Fig. 2 shows the locations of the22 transect sample points. Most of the soils arelight or dark variants of Kastanozem pro¢les(FAO/UNESCO classi¢cation), i.e. well-drainedsoils with brown, humic topsoils (i.e. Ah horizons,with more than 50% of roots concentrated in theupper 25 cm of the soil) overlying a brown tocinnamon, argic (clay-enriched) or cambic(slightly weathered) subsoil or B horizon, andoften with carbonate and/or gypsum accumula-

tion in or below the B horizon (Table 1). The soilsare well bu¡ered, with pH values ranging from 7.2to 8.2. Analysis of their clay mineralogy showsthat smectites dominate in the soils in the north-east, whilst mica predominates elsewhere acrossthe sampled region.

Soil augers were used to drill soil cores to 2 mdepth, and volume magnetic susceptibility wasmeasured for each core. Highest susceptibility val-ues were observed for the top 40 cm and lowestvalues for the parent substrates. Soil samples weretransported to the laboratory in sealed polythenebags, where subsamples were taken at 10-cm in-tervals from the top 40 cm of each pro¢le, togeth-er with a sample of parent material (typically 150^180 cm depth). After drying at 40‡C, they weregently disaggregated and packed into 10-cc poly-styrene sample holders. The following magneticmeasurements were made on each sample, usingthe methods outlined by Maher et al. (1999): low-and high-frequency magnetic susceptibility, an-

Fig. 2. Location map with sample sites, sur¢cial geology and climate data for the sampled region (from Maher et al., 2002).



hysteretic remanence and incremental remanenceacquisition and AF demagnetisation. For selectedrepresentative samples, additional hysteresis pa-rameters were obtained using a vibrating samplemagnetometer, and magnetic extraction proce-dures (Hounslow and Maher, 1996) were used toconcentrate the magnetic carriers for independentinvestigation by XRD and microscopy (opticaland transmission electron microscopy). Before-and after-extraction magnetic measurements weremade so that the extraction e⁄ciency could beassessed. Finally, Mo«ssbauer analysis was appliedto representative bulk and magnetic extract sam-ples.

The instrumentation and experimental proce-dures are detailed in Appendix A.

3. Results

3.1. Magnetic measurements

Magnetic susceptibility, MARM and IRM datahave been outlined elsewhere (Maher et al.,2002). Brie£y, all the soils display higher magneticsusceptibility values in their A and B horizonsthan their C horizons (Fig. 3). Susceptibility val-ues range from consistent minima of V5^20U1038 m3kg31 for the C horizon samples (thelower values associated with the marine depositsto the northeast), to a maximum of V95U1038

m3kg31 within the A horizon of Pro¢le A99-11.Susceptibility maxima for each pro¢le occur with-in the upper 30 cm of the individual soil pro¢les.Similarly, values of frequency-dependent suscepti-bility (normalised to the low-frequency value)range mostly between 0 and 4% in the parentsubstrates but from 5 to 12% in the A and B ho-rizon samples (Fig. 4). High percentages (s 6%)of frequency-dependent susceptibility (measuredat 0.47 and 4.7 kHz) re£ect the presence of sig-ni¢cant numbers of superparamagnetic (SP) ferri-magnetic grains, with grain diameters 6V20 nm(e.g. Bean and Livingston, 1959; Dunlop, 1981;Maher, 1988; Dearing et al., 1996). Fig. 4 alsoshows the variation in anhysteretic remanence(ARM, divided by the DC ¢eld applied to becomean anhysteretic susceptibility, MARM), normalisedT

able

1So

ilpro¢

leinform

ationan

dan

alytical

data

forthesoiltypesrepresentedacross

thesampled

steppe

tran

sect

Soiltype

FAO/U

NESC

O(R

ussian

classi¢cation)

Pro¢le

IDRainfall

Lat./L

ong.

Parent

material

Mped,

1038

m3kg

31

Fe2

þ

pedo

genic

decrease

a

Fe b

ulkb

(rel.un

it)

pH,

H2O

Hum

usCaC

O3

CaS

O4

Claycontent

mm/yr

%%

%%

%from

0^20

cm

Kastano

zem

(Light

Chest-

nutsoil)

A99-6

310

47.18N

,44.55E

Loessic

1325

2.4

7.6

1.2

0.8

0.0

20

Kastano

zem

(Light

Chest-

nutsoil)

A99-5

380

48.02N

,44.14E

Loessic

3042

2.4

8.2

1.8

0.0

0.0

22

Kastano

zem

(Chestnu

tsoil)

A99-10

450

45.42N

,43.38E

Loessic

4556

2.8

7.2

2.0

0.0

0.0

20

Kastano

zem

(DarkChest-

nutsoil)

A99-11

490

45.03N

,44.17E

Loessic

7550

2.3

7.8

2.5

0.0

0.0

18

aFe2

þpedo

genicdecrease=100U

(Fe2

þ PM3Fe2

þ A)/Fe2

þ PM

Thisratioenab

lesexclusionof

anydi¡erencesin

parent

materialcontent,so

that

compa

risoncanbe

mad

ebetw

eenthesoils

interm

sof

intensityof

weatheringdu

ring

theperiod

ofsoilform

ation.

bFebu

lkrelative

units(squ

areof

Mo «ssbau

erspectra).



Fig. 4. MARM (normalised with respect to the SIRM) vs. Mfd (as a % of low frequency M), for sized pure magnetite powders andfor A, B and C horizons of the Russian steppe soils.

Fig. 3. Magnetic susceptibility with soil depth for representative soil pro¢les, A99-n, across the sampled transect, with annualrainfall (mm) given in brackets (from Maher et al., 2002).



to the SIRM attained after application of a 1 Tesla¢eld. The parent substrates display a narrowrange of low MARM/SIRM values (Fig. 4), whilethe B and A horizon samples are characterisedby higher and more variable values. MARM hasbeen shown to be a sensitive indicator of magneticgrain size and magnetic interactions in naturalsamples. Highly interacting, single domain(V30^50 nm) ferrimagnets, such as the magneticchains made by magnetotactic bacteria, give riseto high MARM values. Equally ¢ne-grained butnon-concatenated grains found in magnetically-enhanced soils (Oº zdemir and Banerjee, 1982;Maher, 1988; Maher et al., 1999) produce mod-erate MARMs, while larger, multidomain magneticgrains give rise to low MARM values (Fig. 4;Dankers, 1978). The o¡set in MARM values be-tween the sub-micrometre synthetic magnetitesand the soil samples most likely re£ects the pres-ence in the soils of haematite, which carries only avery low MARM.

‘Saturation’ IRM (SIRM1T) values for the par-

ent material samples vary little, around 2.2U1033

Am2kg31 and their patterns of remanence acqui-sition are similar (Fig. 5). Most remanence is ac-quired at ¢elds of 10^100 mT (V50%) but there isalso signi¢cant remanence acquisition (V40%)at higher ¢elds (300^1000 mT). The A and B ho-rizon samples acquire slightly more low-¢eld rem-anence (up to 10% at ¢elds less than 10 mT) andalso more remanence at ¢elds up to 100 mT(V70%). Their SIRM values are 2^3Uhigherthan those of their parent samples. The magnetichardness of the high ¢eld IRM (HIRM, i.e. ac-quired beyond 100 mT) was also examined, by¢rst applying a 1 T ¢eld to representative samples,to produce an SIRM, which was then af demag-netised in a ¢eld of 100 mT. The remanenceremaining following this demagnetisation (theHIRM100 mT af ) re£ects the concentration ofdemonstrably stable, high-coercivity haematite(rather than goethite, which acquires most rema-nence at ¢elds greater than the 1 T applied here).The HIRM values for the parent materials are

Fig. 5. IRM acquisition for representative parent material and upper (A and B) soil horizons (annual rainfall in brackets).



generally lower than those of the soil samples,with the exception of those soils formed in thehighest rainfall areas (Fig. 6). This suggests thatsoil formation in the less humid areas results inthe formation of haematite as well as low concen-trations of ferrimagnetic minerals. However, asseen in Fig. 6, there is an indirect relationshipbetween annual rainfall and HIRM100 mT af , theamount of haematite diminishing with increasedrainfall. Finally, for the bulk samples, hysteresisloops were measured for a number of representa-tive samples. Fig. 7 shows data for the parentmaterial and A1 samples from pro¢le A99-10. Incomparison with their parent substrate, the top-soil samples display steeper, thinner loops. Theirferrimagnetic contributions to susceptibility rangefrom 85 to 95%, compared to 70^79% in theparent loess; paramagnetic contributions to mag-netic susceptibility are 5^15% in the soils and 21^30% in the parent substrates. The topsoils alsodisplay lower coercive force values. Modelling ofthe remanence components using the ‘Hystear’program of von Dobeneck (1996) indicates theirmagnetic assemblage is signi¢cantly softer overall(Fig. 7).

3.2. Mineralogical analysis of magnetic extracts

To obtain independent mineralogical data onthe magnetic carriers in the parent substratesand the magnetically-enriched A and B horizons,magnetic extractions were performed on samplesfrom six representative pro¢les spanning the cli-mate transect. Prior to magnetic extraction, car-bonate was dissolved from the samples to ensuree¡ective sample dispersion and extraction. Themagnetic properties of the treated samples weremeasured to check the carbonate leach had notaltered them signi¢cantly. The samples werethen sieved into s 38- and 6 38-Wm size frac-tions. As shown in Fig. 8, all of the samples dis-play much higher values of M and ARM in their6 38-Wm fractions than their s 38-Wm fractions.This magnetic di¡erentiation between fractionsappears increasingly marked for the more mag-netic soils, located in areas with higher annualrainfall values.

A magnetic probe (MP) procedure, which ex-poses the circulating sample slurry to a relativelylow but high-gradient magnetic ¢eld, was appliedto the 6 38-Wm fractions, to extract the ¢ne mag-

Fig. 6. SIRM and HIRM100 m Taf for parent materials and representative soil samples vs. annual rainfall.



netic grains. Subsequently, a magnetic edge (ME)procedure was applied to the s 38-Wm fraction,to concentrate the larger magnetic carriers. Bothsets of methods follow those described by vonDobeneck (1985) and Hounslow and Maher(1996, 1999). To assess the e¡ectiveness of theextractions, measurements of M and ARM weremade before and after the procedures. The extrac-tion e⁄ciencies for M are 18^58%, with an averageof V40%; for ARM, V20^70% of the signal wasremoved (Table 2). The variable extraction e⁄-ciency for the susceptibility carriers matches thatreported by Maher (1998) for a range of modernsoils. For the A and B horizon samples, the ¢nest-grained (SP) ferrimagnets appear di⁄cult to re-move even by the MP procedure (indeed, theymay even be concentrated in the non-extractedresidue due, probably, to stable aggregation and

association with clay minerals). For the weaklymagnetic C horizons, more of the susceptibilityis carried by paramagnets, such as clay minerals,which are also ine⁄ciently extracted by the low-¢eld, high-gradient MP method.

For the C horizon samples, the MP procedureprovided su⁄cient extract for analysis by XRD.Their extract mineralogy comprises haematite,with trace quantities of magnetite/maghemite,and abundant quartz, clay minerals (Alekseev etal., in press) and feldspars (Fig. 9). Silicate min-erals appear in the magnetic extracts due to theirparamagnetic nature, and/or the presence withinthem of magnetic inclusions (Hounslow andMaher, 1996; Fig. 10), and/or because of thestrong association between clay minerals andthe magnetic iron oxides. Examination of theMP extracts from the parent materials by trans-mission electron microscopy (TEM) reveals thepresence of three distinct components (Fig. 11):lath-like particles (probably goethite), larger geo-metric Fe-rich particles, with occasional sub-stitution with Ti and Cr, of sV1 Wm diameter(detrital magnetite), and, rarely, some ultra¢neFe-rich particles (6 100 nm). Analysis of thesparse ME extracts from the parent materials,by optical microscopy, identi¢es the presence ofopaque detrital grains of haematite, and magne-tite and ilmenite with oxidised, haematitic rims(martite). Additional mineral components includequartz, pyroxenes, hornblende, rutile and biotite(Fig. 9).

The XRD spectra for the MP extracts from theB horizon samples display some di¡erences fromthose for the parent C horizon samples (Fig. 9).The magnetite/maghemite peaks are relativelybroadened in the B horizons, indicating poorercrystallinity. The di¡erence in crystallinity of thetopsoil and C horizon ferrimagnets can be ac-counted for from the microscopy observations.Microscopy of the parent material extracts iden-ti¢es the presence of relatively large (0.5^2 Wm),geometric Fe-rich particles, with common substi-tution by Ti and Cr, of inherited (detrital) origin.TEM examination of the topsoil magnetic ex-tracts (Fig. 11) reveals the presence of: goethite-like laths (V500 nm in length), larger, geometricFe-rich particles (V200^500 nm diameter), and,

Fig. 7. Hysteresis loops for the A horizon and parent materi-al, pro¢le A99, together with their modeled distribution ofremanence components, derived from von Dobeneck’s (1996)‘Hystear’ program.



additionally, an abundance of ultra¢ne-grained,geometric Fe-rich particles (V10^50 nm) with oc-casional substitution with Mn. Whereas largenumbers of the ultra¢ne grains appear as tightclusters of particles, some are assembled in chains,resembling the chains of ferrimagnets producedby magnetotactic bacteria (Fig. 11C). None, how-ever, display the unique, elongate ‘bullet’- or‘boot’-shaped crystals that unequivocally de¢nean intracellular, biogenic origin for such grains(e.g. Petersen et al., 1986; Vali et al., 1987) and

the concatenation is likely to have occurred post-dispersion and -extraction.

Thus, XRD shows that B horizon samples fromthese steppe soils have an additional, poorly crys-talline magnetite/maghemite component com-pared with their parent substrates. From electronmicroscopy, much of this additional, stronglymagnetic material appears to be ultra¢ne-grained(6V50 nm) and either free from foreign cationsubstitution or with occasional substitution withMn.

Fig. 8. The % of the 6 38- and s 38-Wm particle size fractions for a range of representative A, B and C horizon samples acrossthe sampled transect, and the contribution of each fraction to the measured magnetic susceptibility (upper diagram) and MARM

(lower diagram). Annual rainfall (mm/yr) also given for each sample site.



3.3. Mo«ssbauer analysis

Mo«ssbauer analysis was applied to the MPmagnetic extracts from the B horizons of foursoils spanning the climate gradient, and from aparent material sample (Table 2). The sampledtransect displays little variation in terms of totaliron content, with (relative) values from 2.3 to 2.8

(Table 1). However, room temperature spectra(Fig. 12) display increasing contributions by mag-netite and maghemite in the topsoils from areaswith higher rainfall values. Ferrimagnetic contri-butions to the observed spectra range from a min-imum of 17% for a parent material sample to amaximum of 42% for the B horizon sample fromthe area with the highest annual rainfall (pro¢le

Table 2Magnetic and Mo«ssbauer data for magnetic extracts from B horizons across the climate gradient and for a parent material sam-ple

Sample ID Annual rainfall M 1038m3kg31MARM 1035m3kg1 Sample mass Haematite,

Bhf = 51.9 TMagnetite+Maghemite,Bhf = 45.8/48.8 T,Bhf = 50.5 T

(mm) (ext. e¡.%) (ext. e¡.%) (mg) (spectr.%) (spectr.%)

A99-11 B1 490 96 (31) 612 (48) 37 21 42A99-10 B1 450 71 (49) 423 (53) 47 26 40451 B1 380 62 (38) 228 (36) 45 36 28A99-6 B1 310 27 (47) 152 (52) 18 38 29Parent material 451 C 22 (45) 102 (62) 25 40 17

Fig. 9. XRD spectra for MP extracts from the parent material and B horizon samples. Abbreviations: H, haematite; Mt, magne-tite; Mh, maghemite; Qu, quartz; Fsp, feldspars; Sm, smectite; Mi, mica; Chl, chlorite; K, kaolinite.



A99-11, 490 mm rain p.a.). The haematite/goe-thite content shows a relative decrease comparedwith the magnetite content, falling from 40% inthe parent material to 21% in the most magneticB horizon. Finally, the Fe2þ content of the parentmaterial can be compared with that of the upperhorizons, in order to identify the intensity of sil-icate weathering during soil formation (Alekseevet al., 1996). Based on this parameter, the weath-ering intensity also shows a generally increasingtrend with increased annual rainfall totals (Table1).

4. Discussion

Magnetic and mineralogical analysis of thisrange of modern, mainly Kastanozem-type soilsshows that they contain varying amounts of ultra-¢ne-grained ferrimagnets. The parent substratesof the soils are consistently weakly magnetic buttheir A and B horizons contain signi¢cant addi-tional concentrations of magnetite and maghe-mite. The pedogenic magnetic content of the soils(e.g. magnetic susceptibility A; B horizon3magneticsusceptibility C horizon) varies across the sampled

transect (Fig. 13). It is at a minimum for thesemi-arid zone close to the Caspian Sea (rainV300 mm/yr) and rises to a maximum for themore humid zone close to the northern Caucasusregion (V500 mm/yr). Table 3 provides a corre-lation matrix examining the relationships betweenthe soil transect magnetic properties and the ma-jor climate variables. As shown by this matrix, thestrongest statistical relationships exist between an-nual rainfall and MLF and MARM (correlation coef-¢cients of 0.93), and between summer rainfall andMLF and MARM (correlation coe⁄cients of 0.85 and0.84, respectively). A negative correlation is evi-dent between annual rainfall and HIRM (30.68).

As established by Dokuchaev (1883) and Jenny(1941), soils, or any particular soil property, re-£ect the interplay of the ¢ve soil-forming factors:parent material, climate, organisms, topographyand time. For this region of the Russian steppe,parent materials have uniformly low magneticconcentrations and variability, topography is roll-ing to £at (only inter£uve sites have beensampled) and duration of soil formation appar-ently constant (there has been minimal accumula-tion of loess since the last glacial stage). Thus, interms of soil magnetism, the soil-forming equa-

Fig. 10. Optical micrographs for the ME extracts from the B1 horizon of soil pro¢le D451, in transmitted light: (A) view of theextract (magni¢cationU100); (B) and (C) opaque inclusions in transparent grains and rock fragments (magni¢cationU200).



tion can be reduced in this region to a climofunc-tion (since vegetation will co-vary with climate) ^that is, the soil magnetic properties vary mostly asa function of climate. Further, the correlation ma-trix (Table 3) identi¢es that the major climate var-iable which in£uences the soil magnetic properties

is annual rainfall. For pedogenic magnetic suscep-tibility (MB3MC), the climofunction for these Rus-sian steppe soils takes the form:

Annual rainfall ¼ 86:4 LnðMB3MCÞ þ 90:1

The new steppe magnetic data can be incorpo-

Fig. 11. Transmission electron micrographs for the MP extracts: (A) and (B) from the C horizon samples, (C) and (D) from theB horizon samples. (A) goethite laths (V500 nm in length). (B) s 1-Wm geometric Fe-rich particles of detrital magnetite with oc-casional ultra¢ne-grained, geometric, Fe-rich particles (6V100 nm). (C) Larger, geometric Fe-rich particles (V200^500 nm di-ameter) of detrital magnetite, with additional ultra¢ne fraction. (D) Ultra¢ne-grained, geometric, Fe-rich particles (V10^50 nm)with some (post-extraction?) arrangement in chains.



rated into Maher and Thompson’s (1995) North-ern Hemisphere temperate zone dataset. As canbe seen from Fig. 13, the magnetic data fromthe Chinese Loess Plateau and the Russian steppeare highly correlated both with rainfall and witheach other. The steppe data thus independentlysubstantiate, and indeed can be used to re¢ne,the previously observed magnetism/rainfall rela-tionship. This high degree of correlation, in twogeographically independent regions, identi¢es thedominance of the rainfall component of the cli-

mate system in in£uencing the pedogenic mag-netic properties of these modern soils.

As described previously, soils can take slightlydi¡erent magnetic enhancement pathways, likelyrelated to other climate variables, such as season-ality (Maher and Thompson, 1999). These Rus-sian soils take a magnetically slightly harder en-hancement path compared, for instance, withsome of the most enhanced palaeosols from theChinese Plateau.

Whereas the magnetic susceptibility, MARM and

Fig. 12. Room temperature Mo«ssbauer spectra for magnetic extracts from topsoil samples spanning the climate gradient. Notethe variation in spectral intensity (y-axis), which increases by V50% from spectrum (C) to (A).



SIRM of these steppe soils are dominated byrather small concentrations of the ferrimagnets,magnetite and maghemite, the high-¢eld rema-nence (HIRM100 mT af ) is contributed by theweakly magnetic oxide, haematite (the ‘saturating’¢eld used here was 1 T, which is insu⁄cientto induce signi¢cant remanence from goethite).HIRM100 mT af values are mostly higher in theupper soil horizons (compared with the parentloess). HIRM values are also higher in the lessmagnetic soils, at the more arid end of thesampled transect. These increases in HIRMindicate formation of haematite, as well as ferri-magnets, during soil development, especially inthe more arid soils. The HIRM values indicatehaematite concentrations of between V0.08 and4% (compared with magnetite/maghemite con-centrations of 0.04^0.2%). These magnetic datamatch well with the Mo«ssbauer analysis, whichalso indicate decreasing amounts of haematite/

goethite in the wetter, more magnetic soil pro-¢les.

In laboratory experiments, any of the iron ox-ide and hydroxide species can be formed in con-ditions realistic in terms of soil environments (i.e.at room temperature and pressure, and near-neu-tral pH), via oxidation of Fe2þ/Fe3þ suspensions.Fig. 14 (adapted from Schwertmann and Taylor,1987) identi¢es possible pathways of oxide forma-tion, and the environmental factors which favourformation of any particular oxide. Higher oxida-tion rates, higher organic matter and lower pH(V4^6) favour formation of goethite (Taylor etal., 1987), whilst haematite is favoured by highertemperatures, decreased water activity and higherpH (V7^8) (Schwertmann and Taylor, 1987;Schwertmann, 1988). In soils, magnetite forma-tion requires the initial presence of some Fe2þ

cations. Even in generally well-drained and oxicsoils, like the loessic soils here, Fe2þ can be

Fig. 13. Pedogenic magnetic susceptibility vs. annual rainfall across the sampled steppe transect (inset) and included within theNorthern Hemisphere dataset. Whilst correlation exists between rainfall and several di¡erent magnetic parameters, a multi-param-eter proxy approach performs no better statistically than that based on magnetic susceptibility alone.



formed in soil micro-zones, made temporarily an-oxic during periods of soil wetness, via the activityof iron-reducing bacteria (Starkey and Halvorson,1927; Munch and Ottow, 1980; Lovley et al.,1987; Rossello-Mora et al., 1995). In this modelof pedogenic magnetite formation, intermittentwetting and drying of soils will thus favour for-mation of magnetite (with bacterial mediation), inthe presence of organic matter and a weatheringsource of iron. This relationship between soil wet-ting and drying and formation of magnetite mayaccount for the strong, and therefore possiblycausal, correlation between pedogenic magnetiteand rainfall. Because the action of each iron-re-ducing bacterium can mediate the formation ofhundreds of ferrimagnetic grains, this pathwaycan account for the magnetic concentrations ob-served in the steppe and loess soils. Further bio-logical/climatic coupling may arise from vegeta-tion/iron mineral interactions. Correlation hasbeen observed between soil magnetic concentra-tions (as measured by susceptibility or saturationremanence) and organic carbon (Maher, 1998).Plants can transport iron from deeper soil layersto the surface via leaf litter fall ; it is also possiblethat magnetite, as the inorganic core of plant phy-toferritin, may constitute a more direct (but so farunquanti¢ed) source of soil ferrimagnets, of ultra-¢ne (1^50 nm) grain size (McClean et al., 2001).Evans and Heller (1994) have noted that the mag-netic grain size distributions of the palaeosolsspanning the Chinese Loess Plateau appear verysimilar (as indicated by their magnetic coercivityspectra). On this basis, they suggest that magneto-tactic bacteria are responsible for pedogenic mag-netite formation in these soils. However, thesebacteria produce ferrimagnetic crystals within bio-logically-constrained membranes within their cellsand so the ferrimagnets tend to have a narrowgrain size spread and thus a narrow coercivityspectrum. Such magnetic behaviour is in contrastwith the observation that coercivity spectra tendto broaden, rather than constrict, with increasingdegree of soil development (e.g. Fig. 7). While thegrain size of the bacterial magnetite could subse-quently be altered during soil formation, due toprocesses of dissolution, it also seems unlikelythat such dissolution would operate to similar de-T

able

3Correlation

matrixforsoilmagneticvariab

lesan

dclim

atevariab

les

MLF

(10-

8SI)

SIRM/Mlf

Mfd,%

MARM

SIRM

%IR

M,0^

20mT

%IR

M,

0.3^

1T

HIR

M10

0MARM/

SIRM

MHIR

MMARM/M

MARM/

Mfd

prcep

temp

sum_rain

win_rain

MLF(10-8S

I)1.00

30.80

0.56

0.98

0.95

0.93

30.67

30.64

0.81

30.12

30.18

0.61

0.92

0.12

0.68

0.23

SIRM/M

30.80

1.00

30.65

30.73

30.59

30.86

0.42

0.60

30.86

0.04

0.36

30.31

30.81

0.06

30.53

30.29

Mfd,%

0.56

30.65

1.00

0.46

0.42

0.65

30.50

30.51

0.40

30.28

30.56

30.29

0.52

0.09

0.09

0.49

MARM

0.98

30.73

0.46

1.00

0.96

0.91

30.73

30.69

0.85

30.21

0.02

0.71

0.93

0.15

0.75

0.15

SIRM

0.95

30.59

0.42

0.96

1.00

0.87

30.74

30.65

0.70

30.17

0.00

0.69

0.83

0.13

0.66

0.15

%IR

M0^

20mT

0.93

30.86

0.65

0.91

0.87

1.00

30.68

330.82

0.86

30.14

30.13

0.49

0.89

30.03

0.55

0.36

%IR

M0.3^

1T30.67

0.42

30.50

30.73

30.74

30.68

1.00

0.73

30.54

0.76

30.12

30.35

30.61

30.19

30.67

0.12

HIR

M10

030.64

0.60

30.51

30.69

30.65

30.82

0.73

1.00

30.67

0.38

30.09

30.35

30.68

0.02

30.40

30.30

MARM/SIR

M0.81

30.86

0.40

0.85

0.70

0.86

30.54

30.67

1.00

30.14

0.14

0.63

0.91

0.10

0.72

0.17

MHIR

M30.12

0.04

30.28

30.21

30.17

30.14

0.76

0.38

30.14

1.00

30.17

0.07

30.17

30.23

30.46

0.36

MARM/M

30.18

0.36

30.56

0.02

0.00

30.13

30.12

30.09

0.14

30.17

1.00

0.47

30.02

0.20

0.23

30.31

MARM/M

fd0.61

30.31

30.29

0.71

0.69

0.49

30.35

30.35

0.63

0.07

0.47

1.00

0.59

0.07

0.72

30.19

prcep

0.92

30.81

0.52

0.93

0.83

0.89

30.61

330.68

0.91

30.17

30.02

0.59

1.00

0.23

0.65

0.37

temp

0.12

0.06

0.09

0.15

0.13

30.03

30.19

0.02

0.10

30.23

0.20

0.07

0.23

1.00

0.10

0.15

sum_rain

0.68

30.53

0.09

0.75

0.66

0.55

30.67

30.40

0.72

30.46

0.23

0.72

0.65

0.10

1.00

30.47

win_rain

0.23

30.29

0.49

0.15

0.15

0.36

0.12

30.30

0.17

0.36

30.31

30.19

0.37

0.15

30.47

1.00



grees at sites across the Loess Plateau to producethe observed similarity of grain size distribution.An alternative explanation relates to the action ofthe Fe-reducing bacteria, which may provide theinitial source of Fe2þ cations required for magne-tite formation (Fig. 14). Laboratory syntheses ofmagnetite (Taylor et al., 1987) show that the sizeto which synthetic crystals grow is controlled par-ticularly by oxidation rate, pH, and Fe concentra-tion. To produce a ‘constant’ grain size distribu-

tion (as in the soils across the Plateau), magnetiteformation must thus be initiated under ‘constant’environmental conditions. Such constancy in thenatural environment can be explained if the bac-teria which produce the required Fe2þ (e.g. She-wanella sp. ; Geobacter sp.) only operate within,and/or create via their metabolism (Bell et al.,1987), a certain set of pH, Eh and Fe conditions(Fig. 15). In the laboratory experiments by Tayloret al. (1987) the magnetite grain size distribution

Fig. 14. Pathways of iron oxide formation in pedogenic environments (adapted from Schwertmann and Taylor, 1987). Iron-reduc-ing bacteria are likely to be involved at many of the reduction/dissolution steps (also see Fig. 15).



typical of the Chinese soils is produced with a pHof 7.5, at 26‡C and an oxidation rate of 4 ml air/min.

In contrast to these optimal magnetite-formingconditions, longer periods of dryness, with in-creased oxidation rates and reduced water activ-ity, would favour formation of the more oxic ironcompounds, haematite and goethite (Maher,1998; Ji et al., 2001). In arid environments, there-fore, little formation of pedogenic ferrimagnetswould be predicted, and thus pedogenic suscepti-bility would show little sensitivity to climate(Thompson and Maher, 1995). Similarly, in areasof excessively high rainfall (sV2000 mm/yr),with resultant decalci¢cation and thus decreasedbu¡ering capacity, pedogenic magnetite againmay not form or may be subjected to dissolution,and the rainfall/susceptibility climofunction wouldbreak down (Maher, 1998; Guo et al., 2001).

Thus, within a certain rainfall range (e.g. Fig.1), a climatically determined equilibrium magneticvalue may be reached due to feedback betweenformation of ferrimagnets and their subsequentloss by oxidation or dissolution (Maher, 1998;Maher et al., 2002).

5. Conclusions

(1) The concentration of SP- and SD-sized fer-rimagnets is greater in the A and B horizons ofthe Kastanozem-type soils spanning the Russianloessic steppe, compared with their rather homo-genous and weakly magnetic parent substrates.Magnetic susceptibility, frequency dependent sus-ceptibility and MARM/SIRM ratios of the topsoilsall increase systematically from the arid marginsof the Caspian Sea to the more humid fringes ofthe Caucasus Mountains. The magnetic measure-ments indicate that concentrations of magnetite/maghemite vary from V0.04% in the parent ma-terials to a maximum of V0.2% in the wettestpart of the transect, in the A and B horizons.

(2) Mo«ssbauer analysis of magnetic extractsfrom representative A, B and C horizon samplescon¢rm independently the increases in magnetite/maghemite concentrations in the upper soil hori-zons. From the Mo«ssbauer spectra, analysis of theFe2þ contents of the parent material and the Aand B horizons indicates that pedogenic forma-tion of ferrimagnets accompanies the weatheringof detrital Fe-silicates. Electron microscopy iden-ti¢es that much of the new, pedogenic magneticmaterial occurs as reasonably crystalline, pure(Fe-rich), ultra¢ne grains (6V50 nm).

(3) The soils in the semi-arid parts of the tran-sect also form haematite during their develop-ment, as indicated by their higher HIRM100 mT af

values, compared with the parent loess. However,the amount of haematite formed decreases withincreasing annual rainfall.

(4) Given that parent material, time and topog-raphy are e¡ectively constant across the sampledtransect, the observed spatial variations in soilmagnetic properties dominantly re£ect iron oxidetransformations in£uenced by climate (and its co-variant, organic activity). From statistical analy-

Fig. 15. Eh^pH stability ¢elds for iron compounds, togetherwith preferred redox/pH ranges for the major groups ofiron-oxidising (eg. Thiobacillus f., Leptothrix, Gallionella) andiron-reducing bacteria (Shewanella, Geobacter). After Zavarzi-na, 2001.



sis, the most signi¢cant correlation is between an-nual rainfall and both magnetic susceptibility andMARM (r=0.93).

(5) The magnetic mineralogy of these steppesoils appears to re£ect present-day rainfall varia-tions across this geographic and climatic transect.These soils are subject to neither loess accumula-tion nor industrial pollution at the present day.

(6) The pedogenic magnetic response of thesewell-drained, near-neutral, Russian steppe soilsappears strongly correlated with that of the sim-ilarly well-drained and bu¡ered modern soilsacross the Chinese Loess Plateau (and across thewider Northern Hemisphere temperate zone).Such correlation suggests that the rainfall compo-nent of the climate system is a key in£uence onsoil magnetic properties in both these regions.This direct coupling of the soil magnetism ofmodern soils with present-day climate substanti-ates the use of magnetic climofunctions to makequantitative estimates of past rainfall variationsfrom the magnetic properties of buried palaeosolsfor both the Russian steppe and the ChineseLoess Plateau. However, as noted previously,the climofunction will be insensitive to climatein such arid environments (6V100 mm/yr rain)that pedogenic ferrimagnets do not form, or inhighly humid environments (sV2000 mm/yrrain) where gleying and/or increased soil aciditymay cause dissolution of pedogenic magnetite.

Acknowledgements

We are very grateful for the ¢nancial supportfrom the NATO Science Programme and the Rus-sian Foundation for Basic Research, which en-abled this project to be carried out.

Appendix A. Methods and instruments

A.1. Magnetic measurements

Each sample was dried and packed into 10-ccplastic cylinders. Magnetic susceptibility was mea-

sured at low (0.46 kHz) and high (4.6 kHz) fre-quencies, using a Bartington Instruments MS2susceptibility meter. ARMs were imparted in analternating ¢eld of 80 mT with a biasing DC ¢eldof 0.08 mT (Molspin AF demagnetiser, with dcattachment). The susceptibility of ARM (MARM) iscalculated by normalising the ARM by the inten-sity of the applied bias ¢eld. IRMs were imparted,after demagnetisation of ARMs, in pulsed ¢eldsof 10, 20, 50, 100 and 300 mT (Molspin pulsemagnetiser) and a DC ¢eld of 1000 mT (Newport4Q Electromagnet). HIRMs were AF demagne-tised at 100 mT. All magnetic remanences weremeasured using a £uxgate magnetometer (Mol-spin Ltd., sensitivity V1037 A m2). Magnetic hys-teresis was measured using a Molspin VSM Nuvo.

A.2. Magnetic extractions

Prior to magnetic extraction, the samples weredecalci¢ed using bu¡ered acetic acid and werethen particle-sized into 6 38-Wm and s 38-Wmfractions. Mineral grains were extracted usingthe magnetised probe method and the magneticedge method (Hounslow and Maher, 1999). Theamount of magnetic material extracted at eachstage was quanti¢ed by before- and after-ex-traction magnetic measurements (susceptibility,ARM).

A.3. XRD

The major and minor mineral phases in themagnetic separates were identi¢ed by XRD, usinga Philips PW1710 X-ray di¡ractometer withmonochromatic Cu and Co-radiation, automaticdivergence slit and scan speed 0.005‡ 2a s31. Theestimates of the mineral abundances were basedon subsequent peak intensities.

A.4. Mo«ssbauer analysis

Mo«ssbauer spectra were obtained with aMS1101E spectrometer with a constant accelera-tion drive system (57Co/Cr source with an activity



of about 64 mCi). The velocity scale was cali-brated relative to Fe and sodium nitroprusside.The relative content of total and divalent iron,as well the proportions of magnetite (maghemite)and goethite, in the magnetic extracts were estab-lished from numerical analyses. The room-tem-perature spectra for the magnetic extracts were¢tted with two magnetite sextets, one maghemitesextet and one haematite sextet, and doublets ofFe 3þ and Fe2þ.

A.5. Microscopy

Optical microscopy of polished sections madefrom s 38-Wm magnetic particles was used toprovide information on the grain size, their mor-phology and composition. It also enabled identi-¢cation of the presence and signi¢cance of ferri-magnetic inclusions within various silicateminerals. For examination of the ultra¢ne mag-netic separates, transmission electron microscopy(JEOL JEM-2000EX), with energy dispersiveX-ray analysis (Link Systems Ltd.) was used.

References

Alekseev, A.O., Alekseeva, T.V., Maher, B.A., in press. Mag-netic properties and mineralogy of iron compounds insteppe soils. Pochvovedenie N1.

Alekseev, A.O., Alekseeva, T.V., Morgun, E.G., Samoylova,E.M., 1996. Geochemical regularities of iron state in soils ofconjugate landscapes of Central Precaucasus. Litologiya ipoleznue iskopaemue N 1, 12^22.

Babanin, V.F., 1973. The use of magnetic susceptibility inidentifying forms of iron in soils. Sov. Soil Sci. 5, 487^493.

Bean, C.P., Livingston, J.D., 1959. Superparamagnetism.J. Appl. Phys. 30, 120S^129S.

Bell, P.E., Mills, A.L., Herman, J.S., 1987. Biogeochemicalconditions favouring magnetite formation during anaerobiciron reduction. Appl. Env. Microbiol. 53, 2610^2616.

Dankers, P.H.M., 1978. Magnetic Properties of DispersedNatural Iron Oxides of Known Grain Size. Ph.D. Thesis,University of Utrecht.

Dearing, J.A., Dann, R.J.L., Lees, J.A., Loveland, P.J.,Maher, B.A., O’Grady, K., 1996. Frequency dependent sus-ceptibility measurements of environmental materials. Geo-phys. J. Int. 124, 228^240.

Dokuchaev, V.V., 1883. Russian Chernozem. St. Petersburg.Dunlop, D.J., 1981. The rock magnetism of ¢ne particles.Phys. Earth Planet. Int. 26, 1^26.

Evans, M.E., Heller, F., 1994. Magnetic enhancement and pa-leoclimate: Study of a loess/palaeosol couplet across theLoess Plateau. Geophys. J. Int. 117, 257^264.

Guo, B., Zhu, R.X., Roberts, A.P., Florindo, F., 2001. Lackof correlation between paleoprecipitation and magnetic sus-ceptibility of Chinese loess/paleosol sequences. Geophys.Res. Lett. 28, 4259^4262.

Han, J., Lu, H., Wu, N., Guo, Z., 1996. Magnetic susceptibil-ity of modern soils in China and climate conditions. Stud.Geophys. Geodet. 40, 262^275.

Heller, F., Shen, C.D., Beer, J., Liu, X.M., Liu, T.S., Bronger,A., Suter, M., Bonani, G., 1993. Quantitative estimates andpaleoclimatic implications of pedogenic ferromagnetic min-eral formation in Chinese loess. Earth Planet. Sci. Lett. 114,385^390.

Hounslow, M.W., Maher, B.A., 1996. Quantitative extractionand analysis of carriers of magnetisation in sediments. Geo-phys. J. Int. 124, 57^74.

Hounslow, M.W., Maher, B.A., 1999. Laboratory proceduresfor quantitative extraction and analysis of magnetic mineralsfrom sediments. In: Walden, J. Old¢eld, F., Smith, J.P.(Eds), Environmental Magnetism: A Practical Guide. Qua-ternary Research Association, Cambridge, pp. 139^184.

Jenny, H., 1941. Factors of Soil Formation. McGraw-Hill.Ji, J., Balsam, W., Chen, J., 2001. Mineralogic and climaticinterpretations of the Luochuan loess section (China) basedon di¡use re£ectance spectrophotometry. Quat. Res. 56, 23^30.

Kukla, G., Heller, F., Liu, X.M., Xu, T.C., Liu, T.S., An,Z.A., 1988. Pleistocene climates in China dated by magneticsusceptibility. Geol. 16, 811^814.

Liu, X.M., Rolph, T., Bloemendal, J., Shaw, J., Liu, T.S.,1995. Quantitative estimates of paleoprecipitation at Xifeng,in the Loess Plateau of China. Palaeogeogr. Palaeoclimatol.Palaeoecol. 113, 243^248.

Lovley, D.R., Stolz, J.F., Nord, G.L., Phillips, E.J.P., 1987.Anaerobic production of magnetite by a dissimilatory iron-reducing microorganism. Nature 330, 252^254.

McClean, R.G., Scho¢eld, M.A., Kean, W.F., Sommer, C.V.,Robertson, D.P., Toth, D., Gajdardziska-Josifovska, M.,2001. Botanical iron minerals: Correlation between nano-crystal structure and modes of biological self-assembly.Eur. J. Mineral. 13, 1235^1242.

Maher, B.A., Alekseev, A., Alekseeva, T., 2002. Variation ofsoil magnetism across the Russian steppe: Its signi¢cancefor use of soil magnetism as a paleorainfall proxy. Quat.Sci. Rev. 21, 1571^1576.

Maher, B.A., Thompson, R., 1999. Paleomonsoons, I. Thepaleoclimatic record of the Chinese loess and palaeosols.In: Maher, B.A., Thompson, R. (Eds.), Quaternary Cli-mates, Environments and Magnetism. Cambridge UniversityPress, pp. 81^125.

Maher, B.A., Thompson, R., Hounslow, M.W., 1999. Intro-duction to Quaternary climates, environments and magne-tism. In: Maher, B.A., Thompson, R. (Eds.), QuaternaryClimates, Environments and Magnetism. Cambridge Uni-versity Press, pp. 1^48.



Maher, B.A., 1998. Magnetic properties of modern soils andloessic palaeosols: Implications for palaeoclimate. Palaeo-geogr. Palaeoclimatol. Palaeoecol. 137, 25^54.

Maher, B.A., Thompson, R., 1995. Paleorainfall reconstruc-tions from pedogenic magnetic susceptibility variations inthe Chinese loess and paleosols. Quat. Res. 44, 383^391.

Maher, B.A., Thompson, R., Zhou, L.P., 1994. Spatial andtemporal reconstructions of changes in the Asian paleomon-soon: A new mineral magnetic approach. Earth Planet. Sci.Lett. 125, 461^471.

Maher, B.A., 1988. Magnetic properties of some synthetic sub-micron magnetites. Geophys. J. 94, 83^96.

Maher, B.A., 1986. Characterization of soils by mineral mag-netic measurements. Phys. Earth Planet. Int. 42, 76^92.

Munch, J.C., Ottow, J.C.G., 1980. Preferential reduction ofamorphous to crystalline iron oxides by bacterial activity.Soil Sci. 129, 15.

Oº zdemir, Oº ., Banerjee, S.K., 1982. A preliminary magneticstudy of soil samples from west-central Minnesota. EarthPlanet. Sci. Lett. 59, 393^403.

Petersen, N., von Dobeneck, T., Vali, H., 1986. Fossil bacterialmagnetite in deep-sea sediments from the South AtlanticOcean. Nature 320, 611^615.

Porter, S.C., Hallet, B., Wu, X, An, Z., 2001. Dependence ofnear-surface magnetic susceptibility on dust accumulationrate and precipitation on the Chinese Loess Plateau. Quat.Res. 55, 271^283.

Rossello-Mora, R.A., Caccavo, F., Osterlegner, K., Springer,N., Spring, S., Schuler, D., Ludwig, W., Amann, R., Vann-canneyt, M., Schleifer, K.H., 1995. Isolation and taxonomiccharacterization of a halotolerant, facultatively iron-reduc-ing bacterium. Syst. Appl. Microbiol. 17, 569^573.

Schwertmann, U., 1988. Occurrence and formation of ironoxides in various pedoenvironments. In: Stucki, J.W.,Goodman, B.A., Schwertmann, U. (Eds.), Iron in Soilsand Clay Minerals. NATO ASI Series C217, Reidel, Dor-drecht, pp. 267^308.

Schwertmann, U., Taylor, R.M., 1987. Iron oxides. In: Dixon,J.B., Weed, S.B. (Eds.), Minerals in Soil Environments, 2nded. Soil Sci. Soc. Am., Madison.

State Meteorological Organisation, 1966. Handbook on theClimate of USSR 13, part 2. Hydrometeorological Publish-ing House, Leningrad.

State Meteorological Organisation, 1968. Handbook on theClimate of USSR 13, part 4. Hydrometeorological Publish-ing House, Leningrad.

Starkey, H.L., Halvorson, H.O., 1927. Studies on the trans-formation of iron in nature, II. Concerning the importanceof micro-organisms in the solution and precipitation of iron.Soil Sci. 14, 381^402.

Taylor, R.M., Maher, B.A., Self, P.G., 1987. Magnetite insoils, I. The synthesis of single-domain and superparamag-netic magnetite. Clay Miner. 22, 411^422.

Thompson, R., Maher, B.A., 1995. Age models, sediment£uxes and palaeoclimatic reconstructions for the Chineseloess and palaeosol sequences. Geophys. J. Int. 123, 611^622.

Vadyunina, A.F., Smirnov, Yu.A., 1976. Natural remanentmagnetisation of some soils. Sov. Soil Sci. 8, 471.

Vali, H., Fo«rster, O., Amarantidis, G., Petersen, N., 1987.Magnetotactic bacteria and their magnetofossils in sedi-ments. Earth Planet. Sci. Lett. 86, 389^400.

Vodyunitsky, Y.N., 1981. Formation of ferromagnetics in sod-podsolic soil. Pochvovedeniye 5, 114^122.

von Dobeneck, T., 1985. Gesteinsmagnetische Untersuchungenan Tiefseesedimenten des Sudatlantiks. Diplomarbeit Inst.Allg. Angew. Geophys., Ludw.-Max. Univ., Mu«nchen.

von Dobeneck, T., 1996. A systematic study of natural mag-netic mineral assemblages based on modelling hysteresisloops with coercivity-related hyperbolic basis functions.Geophys. J. Int., 124, 675^694

Zavarzina, D.G., 2001. Biogeochemical Factors of Iron Com-pounds Transformation in Redox Conditions. Ph.D. Thesis,Moscow State University.



magnetic mineralogy of soils across the russian steppe

Documents