Magnetic Mineralogy of Soils Across the Russian Steppe: Climatic Dependence of Pedogenic Magnetite Formation

Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 www.elsevier.com/locate/palaeo

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 15January 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, and especially 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 exists regarding the role of climate vs. dust flux for the magnetic properties of modern loessic soils. Here, we examine the mineralogical 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 analyses identify in situ formation of ferrimagnets in these grassland soils; increased ferrimagnetic concentrations are associated with higher annual rainfall. XRD and electron microscopy show the soil-formed ferrimagnets are ultrafine- grained ( 6 V50 nm) and pure. Ferrimagnetic contributions to Mo«ssbauer spectra range from 17% in the parent loess to 42% for a subsoil sample from the highest rainfall area. Total iron content varies little but the systematic magnetic increases are accompanied by decreased Fe2þ content, reflecting increased silicate weathering. For this region, parent materials are loessial deposits, topography is rolling to flat and duration of soil formation effectively constant. The variations 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 anhysteretic remanence. This magnetic response correlates with that of the modern soils across the Chinese Loess Plateau. Such correlation suggests that the rainfall component of the climate system, not dust flux, is a key influence on soil magnetic properties in both these regions. ß 2003 Elsevier B.V. All rights reserved.

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

1. Introduction ing soils, is an increasingly important natural source of climatic and environmental information. The magnetic mineralogy of sediments, includ- The pedogenic (i.e. in situ, soil-formed) magnetic properties of well-drained, bu¡ered and unpolluted soils appear to be causally related to cli- * Corresponding author. Fax: +44-1524-947099. mate, and speci¢cally, rainfall (Maher et al., E-mail address: [email protected] (B.A. Maher). 1994; Han et al., 1996). Evaluation of the robust-

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart 322 B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 ness of this climate/magnetism couple is impor- mate variables. The magnetic susceptibility (the tant, as it may provide a quantitative transfer ‘magnetisability’) of the Chinese soils is domi- function, enabling calculation of palaeoprecipita- nantly contributed by trace amounts (V0.3%) of tion through the Quaternary period, via analysis magnetite of ultra¢ne grain size ( 6 V30 nm). of the magnetic properties of palaeosols inter- (Because of its ¢ne grain size, this magnetite is bedded within the loess sequences of central variably oxidised at its surface towards maghe- Asia (Heller et al., 1993; Maher et al., 1994). Ro- mite). To isolate the in situ, soil-formed magnetic bust and quantitative proxies of rainfall are nota- signal, the pedogenic susceptibility (Mped) was de- bly scarce within the palaeoclimate context, and ¢ned for each soil as the maximum susceptibility are especially signi¢cant for the densely-populated value (M) of the B horizon minus the M of the regions of monsoonal Asia. parent loess. Maher et al. (1994) found strong Sensitivity of certain soil iron compounds to positive correlation (R2 = 0.94) between the loga- climate has previously been documented in the rithm of pedogenic M and annual rainfall. Simi- mainstream soil science literature. Notably, larly, Han et al. (1996) examined a further 63 top- Schwertmann and Taylor and collaborators dem- soil samples across the Loess Plateau and also onstrated (mostly on the basis of bulk X-ray dif- identi¢ed a direct (albeit polynomial) relationship fraction (XRD) analyses) the response of haema- between rainfall and susceptibility. tite and goethite to di¡erent, climatically-driven Fig. 1 summarises the magnetic and rainfall pedogenic regimes. For example, the goethite^ data for modern soils across the Chinese region, haematite ratio in soils varies systematically along and also published magnetic data for soils across climatic, hydrological and topographic transects, the Northern Hemisphere temperate zone. Some as a result of variations in pH, soil temperature, of the scatter in the data may re£ect inclusion of water activity and organic matter (Schwertmann, ‘unsuitable’ soils, i.e. soils with conditions inimi- 1988). Conversely, the magnetic properties of soils cal to pedogenic formation of ferrimagnets. These are often dominated by formation of the strongly include: poorly-drained or excessively acidic soils, magnetic (ferrimagnetic) oxides, magnetite and eroded soils, or soils developing on slowly-weath- maghemite ^ usually in concentrations indetect- ering or iron-de¢cient substrates. Polluted soils able by XRD but easily measurable by routine and burnt soils, on the other hand, may be exces- magnetic methods of analysis. Hence, magnetic sively enriched in ferrimagnets (Maher, 1986; analyses of soils provide an additional, sensitive Maher and Thompson, 1999). However, another window on soil iron and its response to climate. possibility is that the scatter is real and that the Apparent coupling between magnetism of loes- magnetism/rainfall relationship is less signi¢cant sic soils and rainfall was ¢rst observed for modern than has been proposed. Kukla and co-workers soils developed on the near-horizontal, homoge- (Kukla et al., 1988; Porter et al., 2001) have pro- nous parent substrates of the famous Loess Pla- posed that the variations in magnetic susceptibil- teau region of north-central China (Maher et al., ity across and within the Chinese loess sequences 1994; Liu et al., 1995). At the present day, this are due to di¡ering rates of input of low-suscep- region experiences a strong gradient in rainfall, tibility dust. Indeed, Porter et al. (2001) suggest from values around 300 mm yr31 in the west, to that 84% of the susceptibility variance of the V550 mm yr31 in the central Plateau and V700 modern soils across the Loess Plateau is due to mm yr31 in the south. Most rainfall occurs in the the so-called ‘dust-dilution’ e¡ect. For the Chi- summer, due to monsoonal transport of warm, nese Loess Plateau, the confounding factor in pin- moisture-laden air from the Paci¢c across the Chi- pointing the respective roles of climate and dust nese mainland. The extent of westward penetra- £ux is that these two factors co-vary across the tion of the monsoonal rainbelt varies with the region. That is, annual rainfall increases and dust intensity of the summer monsoon system. Maher £ux decreases from the western to the southern et al. (1994) examined the relationships between and eastern areas of the Plateau. Hence, the pedogenic magnetic susceptibility and modern cli- ‘dust’ school of thought interprets the higher

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 323

Fig. 1. Pedogenic magnetic susceptibility (susc.A or B horizon3susc.C horizon) vs. annual rainfall, Chinese Loess Plateau, the Russian steppe 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 form and/or may be actively dissolved (Maher, 1998). Hence the rainfall/susceptibility climofunction will break down at this point. soil magnetic values in the south and east of the of Dokuchaev, who ¢rst identi¢ed regional scale Plateau as dominantly re£ecting reduced input of soil/climate links (Dokuchaev, 1883), but also low-susceptibility dust. other Russian soil scientists, including Babanin Given the scarcity of quantitative rainfall prox- (1973), Vadyunina and Smirnov (1976) and Vo- ies available to researchers for palaeoclimatic re- dyunitsky (1981), the ¢rst to make large-scale ¢eld construction, it is critical to test these opposing and laboratory analyses of soil magnetic proper- hypotheses ^ climate dependence vs. dust £ux de- ties. pendence. The soil magnetism/climate relationship was tested recently in a geographically independent, pollution-free and geomorphologically sta- 2. Sites and methods ble region, the Russian steppe (Maher et al., 2002). In contrast to the Chinese Loess Plateau The sampled transect exceeds V1000 km, region, minimal dust accumulation occurs across southwest to northeast across the loess-mantled, this region at the present day. Here, we examine exhumed marine plain from the northern £anks of in detail the mineralogical basis of the soil mag- the Caucasus, where glacial moraine is onlapped netism/climate link observed in this region, by an- by the loess, to the northwestern margins of the alysing the magnetic and mineralogical properties Caspian Sea (Fig. 2). The northeastern sector of of 22 modern steppe soils, from a transect span- the transect grades into calcareous clay loams, on ning s 1000 km across the loessic plain from the marine deposits of late Pliocene/Pleistocene age. northern Caucasus to the Caspian Sea. Climate This geomorphologically stable, near-horizontal data for the region are available for the last loessic and marine plain is contiguous to the V100 years, from stations across the area. This west with the eastern European loess belt. The work follows in the pioneering footsteps not only steppe in this region is notably free of any distur-

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart 324 B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341

Fig. 2. Location map with sample sites, sur¢cial geology and climate data for the sampled region (from Maher et al., 2002). bance to the continuous grassland cover; it is ex- tion in or below the B horizon (Table 1). The soils tremely sparsely populated and no pollution sour- are well bu¡ered, with pH values ranging from 7.2 ces exist for the sample sites. As shown in Fig. 2, to 8.2. Analysis of their clay mineralogy shows the climate of the region exhibits a marked gra- that smectites dominate in the soils in the north- dient in precipitation, from V500 mm/yr for the east, whilst mica predominates elsewhere across Stavropol region to V300 mm/yr around Volgo- the sampled region. grad. Precipitation is fairly evenly distributed Soil augers were used to drill soil cores to 2 m throughout the year. Summer temperatures reach depth, and volume magnetic susceptibility was V25‡C, winter temperatures vary from V35to measured for each core. Highest susceptibility val- 310‡C, with temperatures exceeding 10‡C on ues were observed for the top 40 cm and lowest s V170 days/yr (State Meteorological Organisa- values for the parent substrates. Soil samples were tion, 1966, 1968). Fig. 2 shows the locations of the transported to the laboratory in sealed polythene 22 transect sample points. Most of the soils are bags, where subsamples were taken at 10-cm in- light or dark variants of Kastanozem pro¢les tervals from the top 40 cm of each pro¢le, togeth- (FAO/UNESCO classi¢cation), i.e. well-drained er with a sample of parent material (typically 150^ soils with brown, humic topsoils (i.e. Ah horizons, 180 cm depth). After drying at 40‡C, they were with more than 50% of roots concentrated in the gently disaggregated and packed into 10-cc poly- upper 25cm of the soil) overlying a brown to styrene sample holders. The following magnetic cinnamon, argic (clay-enriched) or cambic measurements were made on each sample, using (slightly weathered) subsoil or B horizon, and the methods outlined by Maher et al. (1999): low- often with carbonate and/or gypsum accumula- and high-frequency magnetic susceptibility, an-

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 325

hysteretic remanence and incremental remanence acquisition and AF demagnetisation. For selected representative samples, additional hysteresis parameters were obtained using a vibrating sample Clay content magnetometer, and magnetic extraction proce- 4 dures (Hounslow and Maher, 1996) were used to

CaSO concentrate the magnetic carriers for independent

3 investigation by XRD and microscopy (optical and transmission electron microscopy). Before- and after-extraction magnetic measurements were made so that the extraction e⁄ciency could be assessed. Finally, Mo«ssbauer analysis was applied %%%% Humus CaCO to representative bulk and magnetic extract samples. O

2 The instrumentation and experimental proce- pH, H dures are detailed in Appendix A. b bulk from 0^20 cm Fe (rel. unit) 3. Results

a 3.1. Magnetic measurements þ 2

Fe pedogenic decrease Magnetic susceptibility, MARM and IRM data have been outlined elsewhere (Maher et al., 1

3 2002). Brie£y, all the soils display higher magnetic , 8 kg 3 3 susceptibility values in their A and B horizons ped 10 m M than their C horizons (Fig. 3). Susceptibility values range from consistent minima of V5^ 20U1038 m3kg31 for the C horizon samples (the Loessic 13Loessic 30Loessic 252.4 4556 2.8Loessic 42 7550 2.3 7.6 7.2 1.2 2.4 7.8 2.0 0.8 2.50.0 0.0 8.2 0.0 0.0 0.0 1.8 18 20 20 0.0 0.0 22 material lower values associated with the marine deposits to the northeast), to a maximum of V95U1038 This ratio enables exclusion of any di¡erences in parent material content, so that comparison can be made 3 31

þ m kg within the A horizon of Pro¢le A99-11. 2 PM Susceptibility maxima for each pro¢le occur with- )/Fe þ

2 A in the upper 30 cm of the individual soil pro¢les. 44.55E 44.14E 43.38E 44.17E

Fe Similarly, values of frequency-dependent suscepti- 3 ssbauer spectra). þ

« bility (normalised to the low-frequency value) 2 PM mm/yr % Rainfall Lat./Long. Parent range mostly between 0 and 4% in the parent (Fe

U substrates but from 5to 12% in the A and B horizon samples (Fig. 4). High percentages ( s 6%) A99-5380 48.02N, A99-11 490 45.03N, A99-6 310 47.18N, Pro¢le ID of frequency-dependent susceptibility (measured at 0.47 and 4.7 kHz) re£ect the presence of signi¢cant numbers of superparamagnetic (SP) ferrimagnetic grains, with grain diameters 6 V20 nm (e.g. Bean and Livingston, 1959; Dunlop, 1981;

pedogenic decrease = 100 Maher, 1988; Dearing et al., 1996). Fig. 4 also þ 2 shows the variation in anhysteretic remanence Fe bulk relative units (square of Mo Fe (ARM, divided by the DC ¢eld applied to become a b Kastanozem (Light Chest- nut soil) Kastanozem (Chestnut soil) A99-10 450Kastanozem (Dark Chest- nut soil) 45.42N, Kastanozem (Light Chest- nut soil) between the soils in terms of intensity of weathering during the period of soil formation. Table 1 Soil pro¢le information and analyticalSoil data type for FAO/UNESCO the soil(Russian types classi¢cation) represented across the sampled steppe transect an anhysteretic susceptibility, MARM), normalised

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart 326 B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341

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

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

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 327

Fig. 5. IRM acquisition for representative parent material and upper (A and B) soil horizons (annual rainfall in brackets). to the SIRM attained after application of a 1 Tesla ent material samples vary little, around 2.2U1033 ¢eld. The parent substrates display a narrow Am2kg31 and their patterns of remanence acqui- range of low MARM/SIRM values (Fig. 4), while sition are similar (Fig. 5). Most remanence is ac- the B and A horizon samples are characterised quired at ¢elds of 10^100 mT (V50%) but there is by higher and more variable values. MARM has also signi¢cant remanence acquisition (V40%) been shown to be a sensitive indicator of magnetic at higher ¢elds (300^1000 mT). The A and B ho- grain size and magnetic interactions in natural rizon samples acquire slightly more low-¢eld rem- samples. Highly interacting, single domain anence (up to 10% at ¢elds less than 10 mT) and (V30^50 nm) ferrimagnets, such as the magnetic also more remanence at ¢elds up to 100 mT chains made by magnetotactic bacteria, give rise (V70%). Their SIRM values are 2^3Uhigher to high MARM values. Equally ¢ne-grained but than those of their parent samples. The magnetic non-concatenated grains found in magnetically- hardness of the high ¢eld IRM (HIRM, i.e. ac- enhanced soils (Oº zdemir and Banerjee, 1982; quired beyond 100 mT) was also examined, by Maher, 1988; Maher et al., 1999) produce mod- ¢rst applying a 1 T ¢eld to representative samples, erate MARMs, while larger, multidomain magnetic to produce an SIRM, which was then af demag- grains give rise to low MARM values (Fig. 4; netised in a ¢eld of 100 mT. The remanence Dankers, 1978). The o¡set in MARM values be- remaining following this demagnetisation (the tween the sub-micrometre synthetic magnetites HIRM100 mT af ) re£ects the concentration of and the soil samples most likely re£ects the pres- demonstrably stable, high-coercivity haematite ence in the soils of haematite, which carries only a (rather than goethite, which acquires most rema- very low MARM. nence at ¢elds greater than the 1 T applied here). ‘Saturation’ IRM (SIRM1T) values for the par- The HIRM values for the parent materials are

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart 328 B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341

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

generally lower than those of the soil samples, 3.2. Mineralogical analysis of magnetic extracts with the exception of those soils formed in the highest rainfall areas (Fig. 6). This suggests that To obtain independent mineralogical data on soil formation in the less humid areas results in the magnetic carriers in the parent substrates the formation of haematite as well as low concen- and the magnetically-enriched A and B horizons, trations of ferrimagnetic minerals. However, as magnetic extractions were performed on samples seen in Fig. 6, there is an indirect relationship from six representative pro¢les spanning the cli- between annual rainfall and HIRM100 mT af , the mate transect. Prior to magnetic extraction, car- amount of haematite diminishing with increased bonate was dissolved from the samples to ensure rainfall. Finally, for the bulk samples, hysteresis e¡ective sample dispersion and extraction. The loops were measured for a number of representa- magnetic properties of the treated samples were tive samples. Fig. 7 shows data for the parent measured to check the carbonate leach had not material and A1 samples from pro¢le A99-10. In altered them signi¢cantly. The samples were comparison with their parent substrate, the top- then sieved into s 38- and 6 38-Wm size frac- soil samples display steeper, thinner loops. Their tions. As shown in Fig. 8, all of the samples dis- ferrimagnetic contributions to susceptibility range play much higher values of M and ARM in their from 85to 95%,compared to 70^79% in the 6 38-Wm fractions than their s 38-Wm fractions. parent loess; paramagnetic contributions to mag- This magnetic di¡erentiation between fractions netic susceptibility are 5^15% in the soils and 21^ appears increasingly marked for the more mag- 30% in the parent substrates. The topsoils also netic soils, located in areas with higher annual display lower coercive force values. Modelling of rainfall values. the remanence components using the ‘Hystear’ A magnetic probe (MP) procedure, which ex- program of von Dobeneck (1996) indicates their poses the circulating sample slurry to a relatively magnetic assemblage is signi¢cantly softer overall low but high-gradient magnetic ¢eld, was applied (Fig. 7). to the 6 38-Wm fractions, to extract the ¢ne mag-

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 329

association with clay minerals). For the weakly magnetic C horizons, more of the susceptibility is 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 procedure provided su⁄cient extract for analysis by XRD. Their extract mineralogy comprises haematite, with trace quantities of magnetite/maghemite, and abundant quartz, clay minerals (Alekseev et al., in press) and feldspars (Fig. 9). Silicate minerals appear in the magnetic extracts due to their paramagnetic nature, and/or the presence within them of magnetic inclusions (Hounslow and Maher, 1996; Fig. 10), and/or because of the strong association between clay minerals and the magnetic iron oxides. Examination of the MP extracts from the parent materials by transmission electron microscopy (TEM) reveals the presence of three distinct components (Fig. 11): lath-like particles (probably goethite), larger geometric Fe-rich particles, with occasional substitution with Ti and Cr, of s V1 Wm diameter (detrital magnetite), and, rarely, some ultra¢ne Fe-rich particles ( 6 100 nm). Analysis of the Fig. 7. Hysteresis loops for the A horizon and parent materi- sparse ME extracts from the parent materials, al, pro¢le A99, together with their modeled distribution of by optical microscopy, identi¢es the presence of remanence components, derived from von Dobeneck’s (1996) ‘Hystear’ program. opaque detrital grains of haematite, and magnetite and ilmenite with oxidised, haematitic rims (martite). Additional mineral components include netic grains. Subsequently, a magnetic edge (ME) quartz, pyroxenes, hornblende, rutile and biotite procedure was applied to the s 38-Wm fraction, (Fig. 9). to concentrate the larger magnetic carriers. Both The XRD spectra for the MP extracts from the sets of methods follow those described by von B horizon samples display some di¡erences from Dobeneck (1985) and Hounslow and Maher those for the parent C horizon samples (Fig. 9). (1996, 1999). To assess the e¡ectiveness of the The magnetite/maghemite peaks are relatively extractions, measurements of M and ARM were broadened in the B horizons, indicating poorer made before and after the procedures. The extrac- crystallinity. The di¡erence in crystallinity of the tion e⁄ciencies for M are 18^58%, with an average topsoil and C horizon ferrimagnets can be ac- of V40%; for ARM, V20^70% of the signal was counted for from the microscopy observations. removed (Table 2). The variable extraction e⁄- Microscopy of the parent material extracts iden- ciency for the susceptibility carriers matches that ti¢es the presence of relatively large (0.5^2 Wm), reported by Maher (1998) for a range of modern geometric Fe-rich particles, with common substi- soils. For the A and B horizon samples, the ¢nest- tution by Ti and Cr, of inherited (detrital) origin. grained (SP) ferrimagnets appear di⁄cult to re- TEM examination of the topsoil magnetic ex- move even by the MP procedure (indeed, they tracts (Fig. 11) reveals the presence of: goethite- may even be concentrated in the non-extracted like laths (V500 nm in length), larger, geometric residue due, probably, to stable aggregation and Fe-rich particles (V200^500 nm diameter), and,

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart 330 B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341

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 across the 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. additionally, an abundance of ultra¢ne-grained, the concatenation is likely to have occurred post- geometric Fe-rich particles (V10^50 nm) with oc- dispersion and -extraction. casional substitution with Mn. Whereas large Thus, XRD shows that B horizon samples from numbers of the ultra¢ne grains appear as tight these steppe soils have an additional, poorly crys- clusters of particles, some are assembled in chains, talline magnetite/maghemite component com- resembling the chains of ferrimagnets produced pared with their parent substrates. From electron by magnetotactic bacteria (Fig. 11C). None, how- microscopy, much of this additional, strongly ever, display the unique, elongate ‘bullet’- or magnetic material appears to be ultra¢ne-grained ‘boot’-shaped crystals that unequivocally de¢ne ( 6 V50 nm) and either free from foreign cation an intracellular, biogenic origin for such grains substitution or with occasional substitution with (e.g. Petersen et al., 1986; Vali et al., 1987) and Mn.

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 331

Table 2 Magnetic and Mo«ssbauer data for magnetic extracts from B horizons across the climate gradient and for a parent material sample 38 3 31 35 3 1 Sample ID Annual rainfall M 10 m kg MARM 10 m kg Sample mass Haematite, Magnetite+Maghemite, Bhf = 51.9 T 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 42 A99-10 B1 450 71 (49) 423 (53) 47 26 40 451 B1 380 62 (38) 228 (36) 4536 28 A99-6 B1 310 27 (47) 152 (52) 18 38 29 Parent material 451 C 22 (45) 102 (62) 25 40 17

3.3. Mo«ssbauer analysis (Table 1). However, room temperature spectra (Fig. 12) display increasing contributions by mag- Mo«ssbauer analysis was applied to the MP netite and maghemite in the topsoils from areas magnetic extracts from the B horizons of four with higher rainfall values. Ferrimagnetic contri- soils spanning the climate gradient, and from a butions to the observed spectra range from a min- parent material sample (Table 2). The sampled imum of 17% for a parent material sample to a transect displays little variation in terms of total maximum of 42% for the B horizon sample from iron content, with (relative) values from 2.3 to 2.8 the area with the highest annual rainfall (pro¢le

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

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart 332 B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341

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

A99-11, 490 mm rain p.a.). The haematite/goe- transect (Fig. 13). It is at a minimum for the thite content shows a relative decrease compared semi-arid zone close to the Caspian Sea (rain with the magnetite content, falling from 40% in V300 mm/yr) and rises to a maximum for the the parent material to 21% in the most magnetic more humid zone close to the northern Caucasus B horizon. Finally, the Fe2þ content of the parent region (V500 mm/yr). Table 3 provides a corre- material can be compared with that of the upper lation matrix examining the relationships between horizons, in order to identify the intensity of sil- the soil transect magnetic properties and the ma- icate weathering during soil formation (Alekseev jor climate variables. As shown by this matrix, the et al., 1996). Based on this parameter, the weath- strongest statistical relationships exist between an- ering intensity also shows a generally increasing nual rainfall and MLF and MARM (correlation coef- trend with increased annual rainfall totals (Table ¢cients of 0.93), and between summer rainfall and 1). MLF and MARM (correlation coe⁄cients of 0.85and 0.84, respectively). A negative correlation is evi- dent between annual rainfall and HIRM (30.68). 4. Discussion As established by Dokuchaev (1883) and Jenny (1941), soils, or any particular soil property, re- Magnetic and mineralogical analysis of this £ect the interplay of the ¢ve soil-forming factors: range of modern, mainly Kastanozem-type soils parent material, climate, organisms, topography shows that they contain varying amounts of ultra- and time. For this region of the Russian steppe, ¢ne-grained ferrimagnets. The parent substrates parent materials have uniformly low magnetic of the soils are consistently weakly magnetic but concentrations and variability, topography is roll- their A and B horizons contain signi¢cant addi- ing to £at (only inter£uve sites have been tional concentrations of magnetite and maghe- sampled) and duration of soil formation appar- mite. The pedogenic magnetic content of the soils ently constant (there has been minimal accumula- (e.g. magnetic susceptibility A; B horizon3magnetic tion of loess since the last glacial stage). Thus, in susceptibility C horizon) varies across the sampled terms of soil magnetism, the soil-forming equa-

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 333

Fig. 11. Transmission electron micrographs for the MP extracts: (A) and (B) from the C horizon samples, (C) and (D) from the B horizon samples. (A) goethite laths (V500 nm in length). (B) s 1-Wm geometric Fe-rich particles of detrital magnetite with occasional ultra¢ne-grained, geometric, Fe-rich particles ( 6 V100 nm). (C) Larger, geometric Fe-rich particles (V200^500 nm diameter) 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. tion can be reduced in this region to a climofunc- is annual rainfall. For pedogenic magnetic suscep- tion (since vegetation will co-vary with climate) ^ tibility (MB3MC), the climofunction for these Rus- that is, the soil magnetic properties vary mostly as sian steppe soils takes the form: a function of climate. Further, the correlation ma- Annual rainfall ¼ 86:4LnðM 3M Þþ90:1 trix (Table 3) identi¢es that the major climate var- B C iable which in£uences the soil magnetic properties The new steppe magnetic data can be incorpo-

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart 334 B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341

Fig. 12. Room temperature Mo«ssbauer spectra for magnetic extracts from topsoil samples spanning the climate gradient. Note the variation in spectral intensity (y-axis), which increases by V50% from spectrum (C) to (A). rated into Maher and Thompson’s (1995) North- mate system in in£uencing the pedogenic mag- ern Hemisphere temperate zone dataset. As can netic properties of these modern soils. be seen from Fig. 13, the magnetic data from As described previously, soils can take slightly the Chinese Loess Plateau and the Russian steppe di¡erent magnetic enhancement pathways, likely are highly correlated both with rainfall and with related to other climate variables, such as season- each other. The steppe data thus independently ality (Maher and Thompson, 1999). These Rus- substantiate, and indeed can be used to re¢ne, sian soils take a magnetically slightly harder en- the previously observed magnetism/rainfall rela- hancement path compared, for instance, with tionship. This high degree of correlation, in two some of the most enhanced palaeosols from the geographically independent regions, identi¢es the Chinese Plateau. dominance of the rainfall component of the cli- Whereas the magnetic susceptibility, MARM and

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 335

SIRM of these steppe soils are dominated by goethite in the wetter, more magnetic soil pro- rather small concentrations of the ferrimagnets, ¢les. magnetite and maghemite, the high-¢eld rema- In laboratory experiments, any of the iron ox- nence (HIRM100 mT af ) is contributed by the ide and hydroxide species can be formed in con- weakly magnetic oxide, haematite (the ‘saturating’ ditions realistic in terms of soil environments (i.e. ¢eld used here was 1 T, which is insu⁄cient at room temperature and pressure, and near-neu- to induce signi¢cant remanence from goethite). tral pH), via oxidation of Fe2þ/Fe3þ suspensions. HIRM100 mT af values are mostly higher in the Fig. 14 (adapted from Schwertmann and Taylor, upper soil horizons (compared with the parent 1987) identi¢es possible pathways of oxide forma- loess). HIRM values are also higher in the less tion, and the environmental factors which favour magnetic soils, at the more arid end of the formation of any particular oxide. Higher oxida- sampled transect. These increases in HIRM tion rates, higher organic matter and lower pH indicate formation of haematite, as well as ferri- (V4^6) favour formation of goethite (Taylor et magnets, during soil development, especially in al., 1987), whilst haematite is favoured by higher the more arid soils. The HIRM values indicate temperatures, decreased water activity and higher haematite concentrations of between V0.08 and pH (V7^8) (Schwertmann and Taylor, 1987; 4% (compared with magnetite/maghemite con- Schwertmann, 1988). In soils, magnetite forma- centrations of 0.04^0.2%). These magnetic data tion requires the initial presence of some Fe2þ match well with the Mo«ssbauer analysis, which cations. Even in generally well-drained and oxic also indicate decreasing amounts of haematite/ soils, 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 the Northern Hemisphere dataset. Whilst correlation exists between rainfall and several di¡erent magnetic parameters, a multi-parameter proxy approach performs no better statistically than that based on magnetic susceptibility alone.

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart 336 B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341

formed in soil micro-zones, made temporarily an- 0.31 0.19 0.29 0.47 0.30 oxic during periods of soil wetness, via the activity 3 3 3 3 3 of iron-reducing bacteria (Starkey and Halvorson, 1927; Munch and Ottow, 1980; Lovley et al., 0.46 0.36 0.53 0.67 0.12 0.40 0.47 1.00 1987; Rossello-Mora et al., 1995). In this model 3 3 3 3 3 of pedogenic magnetite formation, intermittent 0.23 0.03 0.55 0.36 0.19 wetting and drying of soils will thus favour for- 3 3 3 mation of magnetite (with bacterial mediation), in the presence of organic matter and a weathering 0.02 0.20 0.23 0.17 0.81 0.06 0.61 0.68 0.02 prcep temp sum_rain win_rain 3 3 3 3 3 source of iron. This relationship between soil wet- / ting and drying and formation of magnetite may 0.29 0.52 0.09 0.09 0.49 0.31 0.35 0.35 0.19 0.37 0.15 ARM fd M 3 3 3 3 3 account for the strong, and therefore possibly MM / causal, correlation between pedogenic magnetite 0.18 0.61 0.92 0.12 0.68 0.23 0.56 0.17 0.07 0.13 0.49 0.89 0.02 0.59 1.00 0.23 0.65 0.37 0.12 0.09 0.31 ARM M 3 3 3 3 3 3 3 3 and rainfall. Because the action of each iron-reducing bacterium can mediate the formation of 0.12 0.28 0.14 0.140.17 0.63 1.00 0.91 0.47 0.10 0.72 0.17 0.21 0.02 0.71 0.93 0.150.75 0.15 0.17 0.00 0.69 0.83 0.13 0.66 0.15 0.14 0.17 0.23 0.20 0.07 0.23 1.00 0.10 0.15 0.46 0.23 0.72 0.650.10 1.00 HIRM hundreds of ferrimagnetic grains, this pathway M 3 3 3 3 3 3 3 3 3 3 can account for the magnetic concentrations ob- / served in the steppe and loess soils. Further bio- 0.14 1.00 0.86 0.04 0.36 0.54 0.76 0.67 0.38 ARM SIRM M 3 3 0.86 0.91 3 3 logical/climatic coupling may arise from vegetation/iron mineral interactions. Correlation has been observed between soil magnetic concentra- 0.64 0.81 0.51 0.40 0.670.09 1.00 0.14 0.69 0.85 0.350.63 0.07 0.47 1.00 0.590.07 0.72 0.650.70 0.82 0.68 0.40 0.72 0.30 0.17tions 0.36 (as measured by susceptibility or saturation 3 3 3 3 HIRM100 3 3 3 3 3 3 3 3 3 remanence) and organic carbon (Maher, 1998). Plants can transport iron from deeper soil layers 0.67 0.50 0.54 0.12 0.73 0.35 0.74 0.68 0.61 0.19 0.02 0.10 0.67 to the surface via leaf litter fall; it is also possible 3 3 %IRM, 0.3^1 T 3 3 3 3 3 3 3 3 3 that magnetite, as the inorganic core of plant phy- toferritin, may constitute a more direct (but so far 0.13 0.14 0.76 0.38 0.86 0.42 0.60 0.68 1.00 0.73 0.03 0.82 0.73 1.00 unquanti¢ed) source of soil ferrimagnets, of ultra- 20 mT 3 3 3 0.89 3 3 3 ¢ne (1^50 nm) grain size (McClean et al., 2001). Evans and Heller (1994) have noted that the mag- 0.17 0.59 0.74 0.65 SIRM %IRM, 0^ 3 3 0.83 3 3 netic grain size distributions of the palaeosols spanning the Chinese Loess Plateau appear very similar (as indicated by their magnetic coercivity 0.21 0.73 0.73 0.69 ARM 3 M 3 0.93 3 3 spectra). On this basis, they suggest that magnetotactic bacteria are responsible for pedogenic mag-

,% netite formation in these soils. However, these 0.56 0.02 0.00 0.28 0.29 0.71 0.69 0.49 0.65 0.50 0.51 fd M 3 3 3 3 3 3 bacteria produce ferrimagnetic crystals within bio-

lf logically-constrained membranes within their cells M and so the ferrimagnets tend to have a narrow 0.80 0.56 0.98 0.95 0.93 0.651.00 0.46 0.42 0.65 0.86 0.40 0.850.70 0.86 0.73 0.46 1.00 0.960.31 0.91 0.59 0.42 0.96 1.00 0.87 0.81 0.52 0.86 0.650.91 0.87 1.00 0.53 0.09 0.75 0.66 0.55 0.29 0.49 0.150.150.36 0.12 3 3 3 SIRM/ 3 3 3 3 3 3 3 grain size spread and thus a narrow coercivity spectrum. Such magnetic behaviour is in contrast with the observation that coercivity spectra tend 0.18 0.36 0.12 0.04 0.80 1.00 0.67 0.42 0.64 0.60 LF (10- 3 M 8SI) 0.98 3 0.61 3 0.92 3 3 to broaden, rather than constrict, with increasing degree of soil development (e.g. Fig. 7). While the grain size of the bacterial magnetite could subse-

fd quently be altered during soil formation, due to M M M /SIRM 0.81 / / processes of dissolution, it also seems unlikely , % 0.56 LF(10-8SI) 1.00 fd ARM ARM HIRM ARM ARM M Table 3 Correlation matrix for soil magnetic variables and climate variables M SIRM 0.95 SIRM/ M M prcep %IRM 0^20mT 0.93 M %IRM 0.3^1T M temp 0.12 0.06 0.09 0.150.13 HIRM100 sum_rain 0.68 win_rain 0.23 M that such dissolution would operate to similar de-

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 337

Fig. 14. Pathways of iron oxide formation in pedogenic environments (adapted from Schwertmann and Taylor, 1987). Iron-reducing bacteria are likely to be involved at many of the reduction/dissolution steps (also see Fig. 15). grees at sites across the Loess Plateau to produce tion (as in the soils across the Plateau), magnetite the observed similarity of grain size distribution. formation must thus be initiated under ‘constant’ An alternative explanation relates to the action of environmental conditions. Such constancy in the the Fe-reducing bacteria, which may provide the natural environment can be explained if the bac- initial source of Fe2þ cations required for magne- teria which produce the required Fe2þ (e.g. She- tite formation (Fig. 14). Laboratory syntheses of wanella sp.; Geobacter sp.) only operate within, magnetite (Taylor et al., 1987) show that the size and/or create via their metabolism (Bell et al., to which synthetic crystals grow is controlled par- 1987), a certain set of pH, Eh and Fe conditions ticularly by oxidation rate, pH, and Fe concentra- (Fig. 15). In the laboratory experiments by Taylor tion. To produce a ‘constant’ grain size distribu- et al. (1987) the magnetite grain size distribution

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart 338 B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341

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

5. Conclusions

(1) The concentration of SP- and SD-sized ferrimagnets is greater in the A and B horizons of the Kastanozem-type soils spanning the Russian loessic steppe, compared with their rather homo- genous and weakly magnetic parent substrates. Magnetic susceptibility, frequency dependent susceptibility and MARM/SIRM ratios of the topsoils all increase systematically from the arid margins of the Caspian Sea to the more humid fringes of the Caucasus Mountains. The magnetic measurements indicate that concentrations of magnetite/ maghemite vary from V0.04% in the parent materials to a maximum of V0.2% in the wettest part of the transect, in the A and B horizons. Fig. 15. Eh^pH stability ¢elds for iron compounds, together (2) Mo«ssbauer analysis of magnetic extracts with preferred redox/pH ranges for the major groups of iron-oxidising (eg. Thiobacillus f., Leptothrix, Gallionella) and from representative A, B and C horizon samples iron-reducing bacteria (Shewanella, Geobacter). After Zavarzi- con¢rm independently the increases in magnetite/ na, 2001. maghemite concentrations in the upper soil horizons. From the Mo«ssbauer spectra, analysis of the typical of the Chinese soils is produced with a pH Fe2þ contents of the parent material and the A of 7.5, at 26‡C and an oxidation rate of 4 ml air/ and B horizons indicates that pedogenic forma- min. tion of ferrimagnets accompanies the weathering In contrast to these optimal magnetite-forming of detrital Fe-silicates. Electron microscopy iden- conditions, longer periods of dryness, with in- ti¢es that much of the new, pedogenic magnetic creased oxidation rates and reduced water activ- material occurs as reasonably crystalline, pure ity, would favour formation of the more oxic iron (Fe-rich), ultra¢ne grains ( 6 V50nm). compounds, haematite and goethite (Maher, (3) The soils in the semi-arid parts of the tran- 1998; Ji et al., 2001). In arid environments, there- sect also form haematite during their develop- fore, little formation of pedogenic ferrimagnets ment, as indicated by their higher HIRM100 mT af would be predicted, and thus pedogenic suscepti- values, compared with the parent loess. However, bility would show little sensitivity to climate the amount of haematite formed decreases with (Thompson and Maher, 1995). Similarly, in areas increasing annual rainfall. of excessively high rainfall ( s V2000 mm/yr), (4) Given that parent material, time and topog- with resultant decalci¢cation and thus decreased raphy are e¡ectively constant across the sampled bu¡ering capacity, pedogenic magnetite again transect, the observed spatial variations in soil may not form or may be subjected to dissolution, magnetic properties dominantly re£ect iron oxide and the rainfall/susceptibility climofunction would transformations in£uenced by climate (and its co- break down (Maher, 1998; Guo et al., 2001). variant, organic activity). From statistical analy-

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 339 sis, the most signi¢cant correlation is between an- sured at low (0.46 kHz) and high (4.6 kHz) fre- nual rainfall and both magnetic susceptibility and quencies, using a Bartington Instruments MS2 MARM (r = 0.93). susceptibility meter. ARMs were imparted in an (5) The magnetic mineralogy of these steppe alternating ¢eld of 80 mT with a biasing DC ¢eld soils appears to re£ect present-day rainfall varia- of 0.08 mT (Molspin AF demagnetiser, with dc tions across this geographic and climatic transect. attachment). The susceptibility of ARM (MARM)is These soils are subject to neither loess accumula- calculated by normalising the ARM by the inten- tion nor industrial pollution at the present day. sity of the applied bias ¢eld. IRMs were imparted, (6) The pedogenic magnetic response of these after demagnetisation of ARMs, in pulsed ¢elds well-drained, near-neutral, Russian steppe soils of 10, 20, 50, 100 and 300 mT (Molspin pulse appears strongly correlated with that of the sim- magnetiser) and a DC ¢eld of 1000 mT (Newport ilarly well-drained and bu¡ered modern soils 4Q Electromagnet). HIRMs were AF demagne- across the Chinese Loess Plateau (and across the tised at 100 mT. All magnetic remanences were wider Northern Hemisphere temperate zone). measured using a £uxgate magnetometer (Mol- Such correlation suggests that the rainfall compo- spin Ltd., sensitivity V1037 Am2). Magnetic hys- nent of the climate system is a key in£uence on teresis was measured using a Molspin VSM Nuvo. soil magnetic properties in both these regions. This direct coupling of the soil magnetism of modern soils with present-day climate substanti- A.2. Magnetic extractions ates the use of magnetic climofunctions to make quantitative estimates of past rainfall variations Prior to magnetic extraction, the samples were from the magnetic properties of buried palaeosols decalci¢ed using bu¡ered acetic acid and were for both the Russian steppe and the Chinese then particle-sized into 6 38-Wm and s 38-Wm Loess Plateau. However, as noted previously, fractions. Mineral grains were extracted using the climofunction will be insensitive to climate the magnetised probe method and the magnetic in such arid environments ( 6 V100 mm/yr rain) edge method (Hounslow and Maher, 1999). The that pedogenic ferrimagnets do not form, or in amount of magnetic material extracted at each highly humid environments ( s V2000 mm/yr stage was quanti¢ed by before- and after-ex- rain) where gleying and/or increased soil acidity traction magnetic measurements (susceptibility, may cause dissolution of pedogenic magnetite. ARM).

Acknowledgements A.3. XRD

We are very grateful for the ¢nancial support The major and minor mineral phases in the from the NATO Science Programme and the Rus- magnetic separates were identi¢ed by XRD, using sian Foundation for Basic Research, which en- a Philips PW1710 X-ray di¡ractometer with abled this project to be carried out. monochromatic Cu and Co-radiation, automatic divergence slit and scan speed 0.005‡ 2a s31. The estimates of the mineral abundances were based Appendix A. Methods and instruments on subsequent peak intensities.

A.4. Mo«ssbauer analysis A.1. Magnetic measurements Mo«ssbauer spectra were obtained with a Each sample was dried and packed into 10-cc MS1101E spectrometer with a constant accelera- plastic cylinders. Magnetic susceptibility was mea- tion drive system (57Co/Cr source with an activity

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart 340 B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 of about 64 mCi). The velocity scale was cali- Evans, M.E., Heller, F., 1994. Magnetic enhancement and pa- brated relative to Fe and sodium nitroprusside. leoclimate: Study of a loess/palaeosol couplet across the Loess Plateau. Geophys. J. Int. 117, 257^264. The relative content of total and divalent iron, Guo, B., Zhu, R.X., Roberts, A.P., Florindo, F., 2001. Lack as well the proportions of magnetite (maghemite) of correlation between paleoprecipitation and magnetic sus- and goethite, in the magnetic extracts were estab- ceptibility of Chinese loess/paleosol sequences. Geophys. lished from numerical analyses. The room-tem- Res. Lett. 28, 4259^4262. perature spectra for the magnetic extracts were Han, J., Lu, H., Wu, N., Guo, Z., 1996. Magnetic susceptibility of modern soils in China and climate conditions. Stud. ¢tted with two magnetite sextets, one maghemite Geophys. Geodet. 40, 262^275. sextet and one haematite sextet, and doublets of Heller, F., Shen, C.D., Beer, J., Liu, X.M., Liu, T.S., Bronger, Fe 3þ and Fe2þ. A., Suter, M., Bonani, G., 1993. Quantitative estimates and paleoclimatic implications of pedogenic ferromagnetic mineral formation in Chinese loess. Earth Planet. Sci. Lett. 114, 385^390. A.5. Microscopy Hounslow, M.W., Maher, B.A., 1996. Quantitative extraction and analysis of carriers of magnetisation in sediments. Geo- Optical microscopy of polished sections made phys. J. Int. 124, 57^74. from s 38-Wm magnetic particles was used to Hounslow, M.W., Maher, B.A., 1999. Laboratory procedures provide information on the grain size, their mor- for quantitative extraction and analysis of magnetic minerals from sediments. In: Walden, J. Old¢eld, F., Smith, J.P. phology and composition. It also enabled identi- (Eds), Environmental Magnetism: A Practical Guide. Qua- ¢cation of the presence and signi¢cance of ferri- ternary Research Association, Cambridge, pp. 139^184. magnetic inclusions within various silicate Jenny, H., 1941. Factors of Soil Formation. McGraw-Hill. minerals. For examination of the ultra¢ne mag- Ji, J., Balsam, W., Chen, J., 2001. Mineralogic and climatic netic separates, transmission electron microscopy interpretations of the Luochuan loess section (China) based on di¡use re£ectance spectrophotometry. Quat. Res. 56, 23^ (JEOL JEM-2000EX), with energy dispersive 30. X-ray analysis (Link Systems Ltd.) was used. Kukla, G., Heller, F., Liu, X.M., Xu, T.C., Liu, T.S., An, Z.A., 1988. Pleistocene climates in China dated by magnetic susceptibility. Geol. 16, 811^814. References Liu, X.M., Rolph, T., Bloemendal, J., Shaw, J., Liu, T.S., 1995. Quantitative estimates of paleoprecipitation at Xifeng, Alekseev, A.O., Alekseeva, T.V., Maher, B.A., in press. Mag- in the Loess Plateau of China. Palaeogeogr. Palaeoclimatol. netic properties and mineralogy of iron compounds in Palaeoecol. 113, 243^248. steppe soils. Pochvovedenie N1. Lovley, D.R., Stolz, J.F., Nord, G.L., Phillips, E.J.P., 1987. Alekseev, A.O., Alekseeva, T.V., Morgun, E.G., Samoylova, Anaerobic production of magnetite by a dissimilatory iron- E.M., 1996. Geochemical regularities of iron state in soils of reducing microorganism. Nature 330, 252^254. conjugate landscapes of Central Precaucasus. Litologiya i McClean, R.G., Scho¢eld, M.A., Kean, W.F., Sommer, C.V., poleznue iskopaemue N 1, 12^22. Robertson, D.P., Toth, D., Gajdardziska-Josifovska, M., Babanin, V.F., 1973. The use of magnetic susceptibility in 2001. Botanical iron minerals: Correlation between nano- identifying forms of iron in soils. Sov. Soil Sci. 5, 487^493. crystal structure and modes of biological self-assembly. Bean, C.P., Livingston, J.D., 1959. Superparamagnetism. Eur. J. Mineral. 13, 1235^1242. J. Appl. Phys. 30, 120S^129S. Maher, B.A., Alekseev, A., Alekseeva, T., 2002. Variation of Bell, P.E., Mills, A.L., Herman, J.S., 1987. Biogeochemical soil magnetism across the Russian steppe: Its signi¢cance conditions favouring magnetite formation during anaerobic for use of soil magnetism as a paleorainfall proxy. Quat. iron reduction. Appl. Env. Microbiol. 53, 2610^2616. Sci. Rev. 21, 1571^1576. Dankers, P.H.M., 1978. Magnetic Properties of Dispersed Maher, B.A., Thompson, R., 1999. Paleomonsoons, I. The Natural Iron Oxides of Known Grain Size. Ph.D. Thesis, paleoclimatic record of the Chinese loess and palaeosols. University of Utrecht. In: Maher, B.A., Thompson, R. (Eds.), Quaternary Cli- Dearing, J.A., Dann, R.J.L., Lees, J.A., Loveland, P.J., mates, Environments and Magnetism. Cambridge University Maher, B.A., O’Grady, K., 1996. Frequency dependent sus- Press, pp. 81^125. ceptibility measurements of environmental materials. Geo- Maher, B.A., Thompson, R., Hounslow, M.W., 1999. Intro- phys. J. Int. 124, 228^240. duction to Quaternary climates, environments and magne- Dokuchaev, V.V., 1883. Russian Chernozem. St. Petersburg. tism. In: Maher, B.A., Thompson, R. (Eds.), Quaternary Dunlop, D.J., 1981. The rock magnetism of ¢ne particles. Climates, Environments and Magnetism. Cambridge Uni- Phys. Earth Planet. Int. 26, 1^26. versity Press, pp. 1^48.

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart B.A. Maher et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 201 (2003) 321^341 341

Maher, B.A., 1998. Magnetic properties of modern soils and Schwertmann, U., Taylor, R.M., 1987. Iron oxides. In: Dixon, loessic palaeosols: Implications for palaeoclimate. Palaeo- J.B., Weed, S.B. (Eds.), Minerals in Soil Environments, 2nd geogr. Palaeoclimatol. Palaeoecol. 137, 25^54. ed. Soil Sci. Soc. Am., Madison. Maher, B.A., Thompson, R., 1995. Paleorainfall reconstruc- State Meteorological Organisation, 1966. Handbook on the tions from pedogenic magnetic susceptibility variations in Climate of USSR 13, part 2. Hydrometeorological Publish- the Chinese loess and paleosols. Quat. Res. 44, 383^391. ing House, Leningrad. Maher, B.A., Thompson, R., Zhou, L.P., 1994. Spatial and State Meteorological Organisation, 1968. Handbook on the temporal reconstructions of changes in the Asian paleomon- Climate of USSR 13, part 4. Hydrometeorological Publish- soon: A new mineral magnetic approach. Earth Planet. Sci. ing House, Leningrad. Lett. 125, 461^471. Starkey, H.L., Halvorson, H.O., 1927. Studies on the trans- Maher, B.A., 1988. Magnetic properties of some synthetic sub- formation of iron in nature, II. Concerning the importance micron magnetites. Geophys. J. 94, 83^96. of micro-organisms in the solution and precipitation of iron. Maher, B.A., 1986. Characterization of soils by mineral mag- Soil Sci. 14, 381^402. netic measurements. Phys. Earth Planet. Int. 42, 76^92. Taylor, R.M., Maher, B.A., Self, P.G., 1987. Magnetite in Munch, J.C., Ottow, J.C.G., 1980. Preferential reduction of soils, I. The synthesis of single-domain and superparamag- amorphous to crystalline iron oxides by bacterial activity. netic magnetite. Clay Miner. 22, 411^422. Soil Sci. 129, 15. Thompson, R., Maher, B.A., 1995. Age models, sediment Oº zdemir, Oº ., Banerjee, S.K., 1982. A preliminary magnetic £uxes and palaeoclimatic reconstructions for the Chinese study of soil samples from west-central Minnesota. Earth loess and palaeosol sequences. Geophys. J. Int. 123, 611^ Planet. Sci. Lett. 59, 393^403. 622. Petersen, N., von Dobeneck, T., Vali, H., 1986. Fossil bacterial Vadyunina, A.F., Smirnov, Yu.A., 1976. Natural remanent magnetite in deep-sea sediments from the South Atlantic magnetisation of some soils. Sov. Soil Sci. 8, 471. Ocean. Nature 320, 611^615. Vali, H., Fo«rster, O., Amarantidis, G., Petersen, N., 1987. Porter, S.C., Hallet, B., Wu, X, An, Z., 2001. Dependence of Magnetotactic bacteria and their magnetofossils in sedi- near-surface magnetic susceptibility on dust accumulation ments. Earth Planet. Sci. Lett. 86, 389^400. rate and precipitation on the Chinese Loess Plateau. Quat. Vodyunitsky, Y.N., 1981. Formation of ferromagnetics in sod- Res. 55, 271^283. podsolic soil. Pochvovedeniye 5, 114^122. Rossello-Mora, R.A., Caccavo, F., Osterlegner, K., Springer, von Dobeneck, T., 1985. Gesteinsmagnetische Untersuchungen N., Spring, S., Schuler, D., Ludwig, W., Amann, R., Vann- an Tiefseesedimenten des Sudatlantiks. Diplomarbeit Inst. canneyt, M., Schleifer, K.H., 1995. Isolation and taxonomic Allg. Angew. Geophys., Ludw.-Max. Univ., Mu«nchen. characterization of a halotolerant, facultatively iron-reduc- von Dobeneck, T., 1996. A systematic study of natural mag- ing bacterium. Syst. Appl. Microbiol. 17, 569^573. netic mineral assemblages based on modelling hysteresis Schwertmann, U., 1988. Occurrence and formation of iron loops with coercivity-related hyperbolic basis functions. oxides in various pedoenvironments. In: Stucki, J.W., Geophys. J. Int., 124, 675^694 Goodman, B.A., Schwertmann, U. (Eds.), Iron in Soils Zavarzina, D.G., 2001. Biogeochemical Factors of Iron Com- and Clay Minerals. NATO ASI Series C217, Reidel, Dor- pounds Transformation in Redox Conditions. Ph.D. Thesis, drecht, pp. 267^308. Moscow State University.

PALAEO 3203 14-11-03 Cyaan Magenta Geel Zwart