Properties of magnetic mineralogy of Alaskan loess: evidence for pedogenesis

Quaternary International 62 (1999) 93}102

Properties of magnetic mineralogy of Alaskan loess: evidence for pedogenesis Xiu Ming Liu!,*, Paul Hesse!, Tim Rolph", James E. BegeH t# !School of Earth Sciences, Macquarie University, NSW 2109, Sydney, Australia "Department of Geology, University of Newcastle, NSW 2308, Australia #Department of Geology and Geophysics, University of Alaska, Fairbanks, Alaska 99775-0760, USA

Abstract The in situ pedogenic enhancement of ferrimagnetic content provides the well-established patterns of magnetic susceptibility variation within mid- to low-latitude loess deposits such as those of China and Central Europe. However, this pattern of high magnetic susceptibility in palaeosols, and lower values in unweathered loess, is not replicated in the higher-latitude loess deposits of Alaska and Siberia. In these localities the relationship is inverted, with high values in loess, and low values in palaeosols. This inverse relationship has been explained by the idea that magnetic susceptibility is re#ecting the magnitude of an aeolian ferrimagnetic component of consistent mineralogy, the grain size of which is related to average wind velocity. However, the results of the magnetic study presented in this paper suggest that there are di!erences in magnetic properties between Alaskan loess and palaeosols, not only in magnetic grain size and concentration but also in magnetic mineralogy. This complicates the simple hypothesis of a &wind velocity' signal by introducing an additional factor into the climatic signal. In contrast to the enhancement of susceptibility observed in palaeosols of the Loess Plateau, China, we suggest that the low magnetic susceptibility values in the Alaskan palaeosol units are a re#ection, at least in part, of the alteration of the ferrimagnetic content by post-depositional processes associated with waterlogging (i.e. gleying) of the soils. ( 2000 Elsevier Science Ltd and INQUA. All rights reserved.

1. Introduction The magnetic susceptibility (s) of loess and palaeosol sequences in China has been recognised and widely used as a proxy palaeoclimatic indicator since Heller and Liu (1982) recognised its variation with loess and soils. It was later shown that the maximum s values occur within the most developed soil horizon (interglacial soil S5 using the Chinese nomenclature) while the s minima coincided with the most weakly weathered loess units (Liu et al., 1987, 1988). This positive relationship between the s value and the degree of pedogenesis has been reported over most parts of the Chinese Loess Plateau, so that palaeoclimatic proxy records of s for the last 2.5 Ma, from the Luochuan and Xifeng sections, have been widely used for correlations of Quaternary stratigraphy within continents and for linking palaeoclimatic signals between continents and oceans (Heller and Liu, 1984; Kukla, 1987;

* Corresponding author. Tel.: ##612 98507111. E-mail address: [email protected] (Y.M. Liu).

Kukla et al., 1988}1990; Hoven et al., 1989; Petit et al., 1990). It is now generally accepted that the s variations within the Chinese loess deposits arise from a combination of the in situ processes of compaction and leaching (Heller and Liu, 1984, 1986) and the inorganic and/or organic (biogenic) formation of ultra"ne ferrimagnetic minerals during pedogenesis (Liu et al., 1990, 1992; Zhou et al., 1990; Maher and Thompson, 1991, 1992; Zheng et al., 1991; Rolph et al., 1993; Evans and Heller, 1994; Heller and Evans, 1995; Hunt et al., 1995; Sun et al., 1995). An inverse relationship between magnetic susceptibility and soil development was reported for Alaskan loess by BegeH t and Hawkins (1989). They showed that the loess units displayed high s values while soil units displayed low values, the direct opposite of the Chinese relationship. A similar negative relationship has recently been reported from Siberian loess (Chlachula et al., 1997). As the susceptibility variations observed in Alaskan loess sections are very much like those observed in marine cores (low s in interglacials), and the latter are commonly interpreted as re#ecting the variation in aeolian transport of dust to marine sites, the authors therefore extended

1040-6182/00/$20.00 ( 2000 Elsevier Science Ltd and INQUA. All rights reserved. PII: S 1 0 4 0 - 6 1 8 2 ( 9 9 ) 0 0 0 2 7 - 0

94

X.M. Liu et al. / Quaternary International 62 (1999) 93}102

this rationale to the interpretation of the Alaskan loess (BegeH t and Hawkins, 1989). Subsequent experiments led to the suggestion (BegeH t et al., 1990) that susceptibility variations re#ected density sorting during aeolian transport. They suggested that similar processes operate on a large scale during natural windstorms, so that loess deposited near the source region or by strong winds will be characterised by higher s values, whereas distal loess will be relatively depleted in magnetite and show lower s values (BegeH t et al., 1990; BegeH t, 1996). At a given site, the alternation between strong winds in glacial periods, and reduced wind intensity during interglacials, will therefore introduce a #uctuating s signal. This hypothesis was subsequently used by Chlachula et al. (1997, 1998) to explain the similar relationship in Siberian loess. However, neither the Alaskan nor Siberian loess had been subjected to detailed investigations of magnetic mineralogy. This is signi"cant because the hypothesis of a simple wind-intensity relationship requires that the magnetic content of loess and soils varies only in terms of grain size and concentration, with a coarser and more abundant magnetic fraction in the loess. In this paper we present the results of a detailed magnetic investigation of samples of Alaskan loess. Our intention is to address the validity of this wind-intensity hypothesis.

2. Samples and experiments Three Alaskan samples were collected in 1997; their stratigraphic positions are marked in Fig. 1. Sample AK1, with a low s value, is from a palaeosol equivalent to the peak of marine oxygen isotope (MOI) stage 3, while samples AK2 and AK3 are loess samples from the transition between MOI stages 5 and 6. For comparison, three samples, LC1, LC2 and LC3, from equivalent stratigraphic (MOI) locations at Luochuan, in the Chinese Loess Plateau, were also selected (Fig. 1) In this paper, the s and frequency-dependent magnetic susceptibility (s ) measurements were carried in Mac&$

Fig. 1. Magnetic susceptibility curves from Alaska (BegeH t, 1996) and Luochuan, China. The numbers indicate the stratigraphic location of the samples used in this study.

quarie University using a Bartington MS1. The thermomagnetic (Curie) curves and magnetic hysteresis were measured at the University of Liverpool, UK, and the high-temperature dependence of susceptibility was measured in the University of Western Australia using a Bartington MS2 susceptibility}temperature system. Low-temperature susceptibility was measured in the CSIRO Rock Magnetism laboratory in Sydney using a CSIRO-produced susceptibility bridge.

3. High-temperature properties The temperature dependence of magnetic properties provides information on magnetic grain size and chemistry. Thermomagnetic (Curie) curves re#ect the variation of saturation magnetisation (M or J ) with temperature S S (providing the applied "eld is of a su$cient magnitude). The curves are indicative of magnetic phase chemistry through both Curie temperatures and inversion temperatures. A common example of the latter is the inversion of maghaemite to haematite, the onset of which typically occurs above &3503C. Maghaemite is the end product of the low-temperature oxidation of magnetite. Partially oxidised &cation-de"cient' (CD) magnetite is also common, typically in the form of an oxidised surface on magnetite grains. The interpretation of the behaviour of s at high temperatures is less certain. It re#ects predominantly the in#uence of sample unblocking temperature spectra, which is both grain size and chemistry related, with large multidomain (MD) grains unblocking close to their Curie temperature and smaller single-domain (SD) and pseudo-single-domain (PSD) grains unblocking at temperatures below their Curie temperature. In addition, the s behaviour will also re#ect the inversion of any CD magnetite}maghaemite phases. As a generalisation, increases in s with heating either represent the unblocking of a magnetic phase or the production of a new magnetic phase. Conversely, reductions in s either represent the inversion of a magnetic phase to a less magnetic phase or the decrease in exchange energy with temperature up to the Curie point (in reality somewhat lower than the Curie point because the sample magnetisation is not at saturation). The observed peaks in s during heating, due to the unblocking of a magnetic phase, are referred to as Hopkinson peaks. Fig. 2 illustrates the high-temperature s variation of two Alaskan samples, a palaeosol (AK1) and a loess (AK3). Sample AK1 shows a gradual decrease in s until &3503C, at which point there is a minor increase before the earlier decrease is reinstated above &4103C. The susceptibility decreases to zero at &5703C, consistent with a magnetite Curie temperature. On cooling there is a major increase in s. This occurs rapidly between 570 and 4003C, indicating the production of a large magnetite phase during heating. Below 4003C the increase con-

X.M. Liu et al. / Quaternary International 62 (1999) 93}102

Fig. 2. High-temperature dependence of magnetic susceptibility (]10~5 SI) for Alaskan loess (AK3) and soil (AK1). The loess sample shows a decrease in s after the procedure while the soil sample shows an increase. This implies a di!erence in the magnetic content of the two samples.

tinues but at a much reduced rate, suggesting the presence of a magnetic phase with Curie temperatures distributed between 4003C and room temperature. In fact s is still increasing at room temperature, suggesting that the distribution of Curie temperatures extends below room temperature. At room temperature, the s value has increased approximately "ve-fold compared to the preheating value. This major increase in v supports the interpretation of the peak in the heating curve at &4103C as due to the production of a new ferrimagnetic phase rather than a Hopkinson peak of an existing ferrimagnetic phase. Loess sample AK3 displays a completely di!erent thermomagnetic behaviour, most noticeably in a 60% reduction of s after heating. In detail, the heating curve shows an initial increase in s compatible with the unblocking of a ferrimagnetic phase with a maximum unblocking temperature of &2503C. This is followed by a decrease in s which accelerates as the Curie point of the sample is approached. For this sample the Curie temperature appears at about 5853C. On cooling, the increase in s is initiated at &5803C, compatible with a magnetite phase. The increase in s has essentially terminated by &3603C, after which no further increase is observed as temperature falls. This behaviour is compatible with the inversion, to haematite, of an oxidised ferrimagnetic phase with a Curie temperature that is consistent with a cation-de"cient magnetite. This behaviour is the same as that reported for Chinese loess (Sun et al., 1995; Hunt et al., 1995), where the decrease of v between 250 and 4503C was attributed to the inversion of maghaemite to haematite, and the subsequent sharp decrease from 500 to 5903C as due to magnetite. In our case the decrease

95

continues up to &6103C, suggesting, as noted earlier, the presence of a CD magnetite phase (partially oxidised surface layer?) which subsequently inverts to haematite. In Fig. 3 the thermomagnetic (Curie) curves of the three Alaskan samples are compared with the three equivalent samples from the Luochuan loess section in China. The behaviour of sample AK1 con"rms and enhances the interpretation of the s behaviour. The heating and cooling curves display a #at (heating) to convexdown (cooling) nature which is suggestive of paramagnetic behaviour or a component with Curie temperatures extending below room temperature. There is some suggestion on the heating curve of a convex-down segment centred at &4003C which may be attributed to the inversion of maghaemite (see discussion below of samples AK2 and AK3). There is a barely observable Curie temperature in both the heating and cooling curves which seems close to 6003C, slightly in excess of a typical magnetite value. Di!erences from the s behaviour are the absence of a hump in the heating curve and a much reduced increase in magnetisation on cooling (approximately two-fold compared to the near "ve-fold increase in s). This di!erence can be interpreted either as a lower level of alteration in the Curie balance or that the phase produced during heating makes a disproportionately large contribution to s. In the latter case, the production of an ultra"ne, superparamagnetic (SP) component could be inferred. Samples AK2 and AK3 show behaviour more typical of Chinese loess; a concave-down in#ection at &350}4503C associated with the inversion of maghaemite (Liu et al., 1992; Hunt et al., 1995; Sun et al., 1995), and a Curie temperature between &590 and 6103C, consistent with magnetite that has su!ered partial surface oxidation, with the oxidised component inverting to haematite during the completion of the heating to 7003C (OG zdemir et al., 1993; Dunlop and OG zdemir, 1997). This is supported by the more typical magnetite Curie temperature observed during cooling. The observation that sample AK1 shows an increase in both s and M after heating suggests a fundamental 4 di!erence between this soil sample and the two loess samples. The soil AK1 apparently contains certain iron-bearing minerals, which alter to a more magnetic phase during heating. At present it is not clear whether these minerals are absent from the loess or whether the soil contains organic matter which encourages reducing conditions within the sample during heating.

4. Room temperature properties Room temperature (RT) magnetic measurements for the Alaskan loess samples include susceptibility, s, frequency-dependent susceptibility, s , and hysteresis para&$ meters (M , M and H ). The magnetic hysteresis curves S RS C

96

X.M. Liu et al. / Quaternary International 62 (1999) 93}102

Fig. 3. Thermomagnetic (Curie) curves from Alaska (AK) and Luochuan (LC). All samples display a loss of magnetisation during heating, except for Alaskan soil AK1 which shows a signi"cant increase. The loss of magnetisation is attributed to the inversion of cation-de"cient ferrimagnetic phases, implying their absence from the Alaskan soil sample.

for the Alaska and Chinese samples are shown in Fig. 4. In all cases the loops are closed by 300 mT, consistent with a ferrimagnetic phase. The linear increase in magnetisation at higher "elds re#ects the paramagnetic component. Any hard (antiferromagnetic) phase is too small to be distinguished. The main di!erence between the Alaskan and Chinese samples is that for the soil samples, AK1 has a (relatively)

larger paramagnetic component than LC1 while for the loess samples, AK2 and AK3 have a (relatively) larger ferrimagnetic component than their Chinese equivalents. The hysteresis parameters for all samples are listed in Table 1, together with a number of other parameters determined from the other analyses. The s and s data from the Alaskan loess demonstrate &$ a di!erent type of relationship than the corresponding

X.M. Liu et al. / Quaternary International 62 (1999) 93}102

97

Fig. 4. Magnetic hysteresis loops for Alaskan (AK) and Luochuan (LC) samples. The ferrimagnetic content dominates the shape of the hysteresis loops for the Alaskan loess samples (AK2 and AK3) while for the Alaskan soil (AK1) and the Chinese samples (LC1}3) the loops show a proportionally greater in#uence of a paramagnetic (or hard antiferromagnetic) component.

data from Chinese loess. In the former case the susceptibility is greater in the loess samples (AK2; AK3) while s is greater in the soil (AK1); in the latter case both &$ values are greater in the soil (LC1) than the loess samples (LC2; LC3; see Table 1). Additionally, the s value for &$ the Alaskan soil (AK1) is signi"cantly less even than the Chinese loess samples while the Alaskan loess shows almost no frequency dependence of susceptibility. A sim-

ilar phenomenon was reported for the Siberian loess (Chlachula et al., 1998). In Chinese loess, s is typically &$ higher than 5% (except in the more arid NW sections), while in the Siberian loess only four small peaks of s '1% were found, and these corresponded to &$ palaeosols formed during oxygen isotope stage 1, 3, 5 and 7, respectively. Frequency-dependent susceptibility is considered to be a sensitive indicator of the presence of

98

X.M. Liu et al. / Quaternary International 62 (1999) 93}102

Table 1 Magnetic parameters for the two sets of sample from Alaska and Luochuan! Sample

M S (]10~2 Am2 kg~1)

M /M RS S

H C (mT)

s (]10~7 m3 kg~1)

s &$ (%)

*M S (]10~2 A m2 kg~1)

*M S (%)

AK1 AK2 AK3 LC1 LC2 LC3

2.78 15.40 20.60 8.02 4.60 3.01

0.19 0.11 0.11 0.15 0.16 0.15

12.52 10.79 10.25 6.47 6.96 8.37

3.63 11.56 17.75 14.69 7.55 5.27

3.40 0.83 0.30 10.02 8.03 6.15

!1.29 6.09 6.50 1.39 0.81 0.53

!77 37 47 24 26 30

!M : saturation magnetisation; M /M : ratio of saturation remanent magnetisation/M ; H : saturation coercivity; s: susceptibility; s : frequency S RS S S C &$ dependence of susceptibility; *M : magnetisation di!erence at 203C between heating and cooling. S

ultra"ne ((30 nm) ferrimagnetic material formed in situ during pedogenesis. Therefore, although the inverse correspondence between palaeosols and susceptibility dominates the record of Siberian and Alaskan loess, consistent with the &wind intensity' model of susceptibility enhancement in loess, there is clearly also a pedogenic record of susceptibility enhancement in soils.

5. Low-temperature susceptibility properties Low-temperature magnetic susceptibility (LTs) provides information on the grain size and chemistry of the magnetic minerals and the paramagnetic content. The curves of LTs for the three Alaskan samples are shown in Fig. 5. All three samples have in common a gradual increase in s as temperature decreases, but the size of the increase varies between the three samples, being largest for sample AK1 and smallest for sample AK3. This behaviour is consistent with the presence of a paramagnetic component (1/¹ relationship). In addition samples AK2 and AK3 show a peak in s centred at !1703C, which is absent from the AK1 curve. Magnetite shows a Verwey transition within the temperature interval !153 to !1433C, however partial oxidation of the magnetite can smear the transition to lower temperatures and eventually suppress it completely (OG zdemir et al., 1993). In view of the other evidence outlined above for the presence of partially oxidised magnetite and maghaemite in the Alaskan samples it seems reasonable that this lower Verwey transition temperature should be observed in these samples. The absence of this transition in sample AK1 implies that this sample has a smaller magnetic grain size, with a greater surface : volume ratio, such that surface oxidation has led to complete suppression of the Verwey transition (OG zdemir et al., 1993). Certainly the frequency-dependent s data indicate that sample AK1 has a "ner magnetic component, but we should also consider the alternative interpretation which is that the apparent suppression of the Verwey transition may simply arise from the fact that the paramagnetic susceptibil-

Fig. 5. Low-temperature susceptibility curves for the three Alaskan samples. The curve for the soil sample (AK1) is dominated by a &1/¹ paramagnetic relationship while loess samples AK2 and AK3 show a combination of paramagnetic and MD magnetite behaviour.

ity dominates the ferrimagnetic susceptibility (see Table 1) resulting in a Verwey transition that is masked by the paramagnetic response. The Curie temperature curve (Fig. 3) is inconclusive as far as these two alternatives are concerned. Although the Curie temperature is slightly in excess of the magnetite value, this is attributed to the quick heating rate (about 20 min for each measurement). Moreover, although there is some evidence for a convexdown trough centred on &4003C, rather than evidence for maghaemite inversion, this may just re#ect the production of a new ferrimagnetic phase that is obvious in the high-temperature s data (Fig. 2). A further complication is that even if the ferrimagnetic grains are partially (or fully) oxidised, their "ner grain size may increase their inversion temperature (Bando et al., 1965) and produce a distributed range of inversion temperatures re#ecting the grain size distribution. Finally, the Curie curves are dominated by a paramagnetic-like response which tends

X.M. Liu et al. / Quaternary International 62 (1999) 93}102

to mask the more subtle response of the small ferrimagnetic component.

6. Discussion To summarise, the magnetic measurements reported in this paper have revealed that the palaeosol sample from Alaska (AK1) has magnetic characteristics which are dominated by paramagnetic and/or hard magnetic materials, and that it contains minerals which are converted to ferrimagnetic minerals during heating to temperatures above &4003C. Conversely, the Alaskan loess samples (AK2 and AK3) have magnetic characteristics dominated by a coarser-grained ferrimagnetic component which is composed of magnetite that has been partially (surface) or fully oxidised to cation-de"cient magnetite and maghaemite. Consequently, there are data which suggest that the di!erence in magnetic properties between the loess and palaeosol samples from Alaska may not only re#ect a di!erence in magnetic grain size and concentration, as suggested by BegeH t et al. (1990) and BegeH t (1996), but also in the type and origin of the magnetic components. This would complicate the suggestion that the Alaskan susceptibility}climatic signal re#ects only the in#uence of wind intensity. If the thermally unstable component of magnetisation (represented by the loss in magnetisation, measured at 203C, between heating and cooling; Fig. 3; *M , Table 1) S is taken as a measure of the contribution by maghaemite (including both fully and partially oxidised magnetite), it is found that the Alaskan loess generally has a greater content of maghaemite than Chinese loess/soils in both absolute and relative terms (palaeosol AK1 is an exception due to the increase of magnetisation during heating). Thermally unstable maghaemite in Chinese loess was interpreted as both of primary aeolian origin (coarser grain size) and the result of pedogenic enhancement (Liu, 1997; Liu et al., 1999). The Alaskan loess samples show the same characteristic thermal behaviour as the samples of Chinese loess, namely that a higher percentage of maghaemite is associated with loess (Fig. 3 and Table 1). The more obvious transition around 3503C, and the greater size of the thermally unstable component in both absolute and relative terms, imply that the Alaskan loess contains a coarser aeolian maghaemite and was closer to its source area than the Chinese Loess Plateau samples were to their source. The Alaskan soil AK1 shows di!erent magnetic behaviour which may indicate that some aspect of climate (humidity, precipitation?) during the interglacial periods resulted in the gradual conversion of ferrimagnetic minerals into hard and/or paramagnetic minerals, resulting in the low measured s today (see following discussion). Although sites such as Luochuan and Xifeng are often considered to be typical of the Chinese Loess Plateau, in

99

fact there is a considerable regional variation in the relationship between loess, soils and magnetic susceptibility. Although typically this variation is in terms of the absolute values of s, and the relative enhancement in soils as a re#ection of regional variations in pedogenic intensity, in the southern part of the Plateau there is evidence that the usual relationship of increasing s with increasing pedogenic intensity has broken down. At Baoji, adjacent to the Wei River, it was found that the highest s value was not in the most well-developed soil, S5, but in a younger soil, S3 (Fig. 6). Of relevance to this observation is the fact that well-developed soils (eg. S5, S6, S7 and S8) in this southern region show abundant, black Fe}Mn mottling and coatings on grains, which is interpreted as evidence of gleying that results from the seasonal waterlogging of the upper part of soil pro"les (Ding, 1988). Pedogenesis, and associated enhancement of s, show a roughly positive proportional relationship with humidity throughout most of the Plateau, with a number of studies indicating a positive relationship between precipitation and s values (Liu et al., 1992, 1995; Maher and Thompson, 1995). However, this evidence from the southern Plateau suggests that once humidity has exceeded a certain critical value, the positive relation between humidity/pedogenesis and s breaks down (Liu, 1989). Waterlogging in the upper part of a soil pro"le can produce reducing conditions that gradually transform part of the magnetite and maghaemite content into hydroxides such as goethite which are more suited to the soil environment but only make a small contribution to the s. The s curve for the Baoji section is shown in Fig. 6, together with the s curve for Xifeng. Baoji, which lies

Fig. 6. Comparison of magnetic susceptibility records s (]10~5 SI) from Xifeng, towards the centre of the Chinese Loess Plateau, and from Baoji in the southern part of the Plateau. The peak v value of S5 is the highest of the youngest "ve soils at Xifeng, but at Baoji it becomes the lowest value. At Baoji it is suggested that high precipitation levels have altered the ferrimagnetic minerals in S5. See text for a detailed discussion.

100

X.M. Liu et al. / Quaternary International 62 (1999) 93}102

150 km south of Xifeng, has 100 mm more annual precipitation and 250 mm less evaporation. At Xifeng, soil S3 has the lowest s value of the last "ve interglacial soils, implying either the lowest degree of pedogenesis or a lower level of precipitation at the site during this interglacial. Conversely, at Baoji, S3 has the largest s value and the preceding soils, S4 and S5, are considerably lower. The interpretation of the records of these two sites is that soils S4 and S5 were formed during interglacials where precipitation levels were such that at Baoji, waterlogging and gleying occurred, while at Xifeng, with its lower precipitation levels, this waterlogging did not occur. Subsequently, the level of precipitation in the following interglacial was considerably lower such that gleying did not occur (or was much reduced) at Baoji while at Xifeng the reduced precipitation led to a reduction in s enhancement (compared to S4 and S5). This is clearly a simplistic interpretation and the precise relationship between environmental conditions and s enhancement in soils is more complex (Maher, 1998). However, the dominant role of precipitation in s enhancement is well established and this interpretation is a natural development of that argument. We have attempted to illustrate this relationship in Fig. 7. The relationship is not quanti"ed and so we have not put numbers on the axes. Our contention is that precipitation levels in the Chinese Loess Plateau can at the extreme lead to a reversal of the trend of susceptibility enhancement (e.g. S5 at Baoji) while in Alaska the level of waterlogging is such that during pedogenesis there is always a signi"cant reduction in ferrimagnetic content compared to the parent loess. The Alaskan samples come from Halfway House, near Fairbanks, which is located about 183 further north than the Chinese Loess Plateau. There are signi"cant di!erences between the two locations in terms of physical geography, solar radiation, and hydrology, resulting in quite di!erent monthly temperature and precipitation levels (Fig. 8). Based on the data of 33 meteorological stations, the Chinese Loess Plateau has an average annual precipitation (including snowfall) of about 526 mm and annual evaporation of 1620 mm. This represents almost a three-fold imbalance in precipitation}evaporation. Fairbanks has an annual rainfall of about 300 mm but has in addition snowfall of 1780 mm (Alaska Climate Research Center). Fairbanks also has a cooler and longer winter and currently only about 90 d in summer are frost-free (Anderson, 1969). This leads to the development of permanently frozen subsoil which prevents adequate sub-surface drainage and maintains a high moisture content in the upper part of soil pro"les. Under such conditions it is plausible that waterlogging, similar to that inferred in the Baoji section, may have occurred in Alaska in previous interglacials. Recently, three important studies of Alaskan loess have provided great insight into the mechanism of the

Fig. 7. An idealised relationship between precipitation and modi"cation of the ferrimagnetic content of loess during pedogenesis. It is suggested that in Alaska the predominant behaviour is that of reduction in magnetic susceptibility due to waterlogging of the soils, while in China the susceptibility generally follows a positive relation with rainfall, probably due to less waterlogging and greater evaporation.

susceptibility}climatic signal. One study (Vlag, 1998) produced detailed measurements for two sites near Fairbanks; Halfway House and Gold Hill Steps. At these sites both the uppermost and basal soils demonstrated high s and frequency-dependent s, similar to the Chinese Loess. This suggests that during at least two interglacial intervals the environmental conditions in Alaska were more typical of those experienced on the Chinese Loess Plateau, and therefore we should extend the &Alaska' zone in Fig. 7 further to the left into the region of susceptibility enhancement. Whether these particular soils formed under conditions of lower precipitation, or warmer conditions (with better soil drainage due to lower permafrost) can at present only be a matter of conjecture. The second study (Rosenbaum et al., 1997) reported geochemical data which showed a maximum in the concentration of chemically immobile titanium within palaeosol horizons. Recently BegeH t (1999) has shown that scanning electron microscope (SEM) imagery of ferro-magnetic grains separated from palaeosols in Alaskan loess show pitting and other chemical dissolution features consistent with pedogenic gleying. All these provide strong support for the suggestion that the lower susceptibility associated with soil units in Alaska (and possibly Siberia) is due to the post-depositional destruction of magnetic minerals. Siberia has a similar physical environment to Fairbanks, Alaska, and we therefore suggest that our discussion of the v behaviour and origin in the Alaskan loess can be applied equally well to the Siberian loess.

X.M. Liu et al. / Quaternary International 62 (1999) 93}102

101

study areas and especially when they are in signi"cantly di!erent environments. Acknowledgements We thank Dr. S.L. Yang for kind help in obtaining VSM and Curie data; Dr. Zheng H.B. for high-temperature K(t) measurement; the Alaska Climate Research Centre for Fairbanks present-day climatic data; Drs. P.W. Schmidt and D. Clark for helpful discussion. This research is supported by Australian Research Council and Macquarie University Research Grant. References

Fig. 8. Comparison of monthly averaged temperature (3C) and monthly averaged precipitation (mm) from Fairbanks, Alaska (Hare and Hay, 1974) and Luochuan, China.

7. Conclusions (1) From the analysis of this small number of samples from Alaska it is possible to make preliminary conclusions regarding the rock magnetism responsible for the observed pattern of magnetic susceptibility. (2) There are signi"cant di!erences between the magnetic characteristics of soil and loess samples from the Alaskan section. These di!erences exist not only in magnetic grain size, but also in magnetic mineralogy. These di!erences cannot be fully explained simply by a model of wind intensity modulation. (3) Soil saturation, or waterlogging, could be a signi"cant factor in the low magnetic susceptibility of soils at the Alaskan site through the post-depositional alteration of ferrimagnetic minerals to weakly magnetic a-ferric oxyhydroxides. (4) The applicability of the results to other sites in Alaska has not been investigated. However they suggest a need for further important research into climatic}magnetic relationships. In view of these interpretations it is suggested that great care should be taken when applying existing s-pedogenic palaeoclimatic interpretations, particularly in new

Anderson, J.R., 1969. Alaska. In: World Atlas of Agriculture. The Committee for the World Atlas of Agriculture, v3, pp. 35}40. Bando, Y., Kiyama, M., Takada, T., Kachi, S., 1965. The e!ect of particle size of c-Fe O on the transformation from c-form to 2 3 a-form. Japanese Journal of Applied Physics 4, 240}241. BegeH t, J., 1996. Tephrochronolgy and paleoclimatology of the last interglacial cycle recorded in Alaska loess deposits. Quaternary International 34}36, 121}126. BegeH t J., 1999. Continous late Quaternary proxy climate records from loess in Beringia, Quaternary Science Reviews, in press. BegeH t, J., Hawkins, D., 1989. In#uence of orbital parameters on Pleistocene loess deposition in central Alaska. Nature 337, 151}153. BegeH t, J., Stone, D., Hawkins, D., 1990. Paleoclimate forcing of magnetic susceptibility variations in Alaskan loess. Geology 18, 40}43. Chlachula, J., Evans, M.E., Rutter, N.W., 1998. A magnetic investigation of a late Quaternary loess/palaeosol record in Siberia. Geophysical Journal International 132, 128}132. Chlachula, J., Rutter, N.W., Evans, M.E., 1997. A late Quaternary loess-palaeosol record at Kurtak, southern Siberia. Canadian Journal of Earth Science 34, 679}686. Ding, Z.L., 1988. Pedostratigraphy and paleoclimates of the loess deposites in China. Ph.D. Dissertation, Institute of Geology, Chinese Academy of Sciences, 202pp. (in Chinese with English abstract). Dunlop, D.J., OG zdemir, OG ., 1997. Rock Magnetism: Fundamentals and Frontiers. Cambridge University, Cambridge, 573pp. Evans, M.E., Heller, F., 1994. Magnetic enhancement and palaeoclimate: study of a loess/palaeosol couplet across the Loess Plateau of China. Geophysical Journal International 117, 257}264. Hare, F.K., Hay, J.E., 1974. The climate of Canada and Alaska. In: Bryson, R.A., Hare, F.K. (Eds.), Climate of North America. Elsevier, Amsterdam, pp. 49}192. Heller, F., Evans, M.E., 1995. Loess magnetism. Reviews of Geophysics 33 (2), 211}240. Heller, F., Liu, T.S., 1982. Magnetostratigraphical dating of loess deposits in China. Nature 300, 431}433. Heller, F., Liu, T.S., 1984. Magnetism of Chinese loess deposits. Geophysical Journal of the Royal Astronomical Society 77, 125}141. Heller, F., Liu, T.S., 1986. Palaeoclimatic and sedimentary history from magnetic susceptibility of loess in China. Geophysical Research Letters 13, 1169}1172. Hovan, S.A., Rea, D.K., Pisias, N.G., Shackleton, N.J., 1989. A direct link between the loess and marine d18O records: aeolian #ux to the north Paci"c. Nature 340, 296}298. Hunt, C.P., Banerjee, S.K., Han, J.M., Solheid, P.A., Oches, E., Sun, W.W., Liu, T.S., 1995. Rock-magnetic proxies of climate change in the loess}palaeosol sequences of the western loess plateau of China. Geophysical Journal International 123, 232}244.

102

X.M. Liu et al. / Quaternary International 62 (1999) 93}102

Kukla, J.G., An, Z.S., 1989. Loess stratigraphy in central China. Palaeogeography, Palaeoclimatalogy, Palaeoecology 72, 203}225. Kukla, G., An, Z.S., Melice, J.L., Gavin, G., Xiao, J.L., 1990. Magnetic susceptibility record of Chinese loess. Transactions of the Royal Society of Edinburgh: Earth Science 81, 263}288. Kukla, G., Heller, F., Liu, X.M., Xu, T.C., Liu, T.S., An, Z.S., 1988. Pleistocene climates in China dated by magnetic susceptibility. Geology 16, 811}814. Liu, X.M., 1989, Magnetic study on Chinese loess. Ph.D. Dissertation, Institute of Geology, Chinese Academy of Sciences, 135pp. (in Chinese with English abstract). Liu, X.M., 1997. Thermomagnetic study on Chinese loess. In: Liu, J.Q., Yuan, B.Y. (Eds.), Quaternary Geology and Environment in China. Ocean Press, Beijing, pp. 313}322 (in Chinese). Liu, X.M., Hesse, P., Rolph, T., 1999. Origin of maghaemite in Chinese loess deposits: aeolian or pedogenic? Physics of Earth and Planatary Interiors 112, 191}201. Liu, X.M., Rolph, T., Bloemendal, J., Shaw, J., Liu T.S., 1995. Quantitative estimates of palaeoprecipitation the loess plateau of China. Palaeogeography, palaeoclimatology, palaeoecology 113, 243}248. Liu, X.M., Liu, T.S., Xu, T.C., Liu, C., Chen, M.Y., 1987. A study on magnetostratigraphy of a loess pro"le in Xifeng area. In: Liu, T.S. (Ed.), Aspects of Loess Research. China Ocean Press, pp. 164}174. Liu, X.M., Liu, T.S., Xu, T.C., Liu, C., Chen, M.Y., 1988. The Chinese loess in Xifeng, I. The primary study on magnetostratigraphy of a loess pro"le in Xifeng area, Gansu province. Geophysical Journal 92, 345}348. Liu, X.M., Shaw, J., Liu, T.S., Heller, F., Yuan, B.Y., 1992. Magnetic mineralogy of Chinese loess and its signi"cance. Geophysical Journal International 108, 301}308. LuK , H.Y., Han, J.M., Wu, N.Q., Guo, Z.T., 1994. Magnetic susceptibility analysis of modern soils in China and its palaeoclimatic signi"cance. Science in China-B 24, 1290}1297 (in Chinese). Maher, B.A., 1998. Magnetic properties of modern soils and Quaternary loessic paleosols: paleoclimatic implications. Palaeogeography, Palaeoclimatology, Palaeoecology 137, 25}54.

Maher, B.A., Thompson, R., 1991. Mineral magnetic record of the Chinese loess and palaeosols. Geology 19, 3}6. Maher, B.A., Thompson, R., 1992. Paleoclimatic signi"cance of the mineral magnetic record of the Chinese loess and paleosols. Quaternary Research 37, 155}170. Maher, B.A., Thompson R., 1995. Paleorainfall reconstructions from pedogenic magnetic susceptibility variations in the Chinese loess and paleosols. Quaternary Research 44, 383}391. OG zdemir, OG ., Dunlop, D.J., Moskowitz, B.M., 1993. The e!ect of oxidation on the Verwey transition in the magnetite. Geophysical Resarch Letters 20, 1671}1674. Petit, J.R., Mounier, L., Jouzel, J., Korotkevich, Y.S., Kotlyakov, V.I., Lorius, C., 1990. Palaeoclimatological and chronological implications of the Vostok core dust record. Nature 343, 56}58. Rolph, T.C., Shaw, J., Derbyshire, E., Wang, J.T., 1993. The magnetic mineralogy of a loess section near Lanzhou, China. In: Pye (Ed.), The Dynamics and Environmental Context of Aeolian Sedimentary Systems. Geol. Soc. Special Publication No 72. pp. 311}323. Rosenbaum, J.G., Reynolds, R.L., Muhs, D.R., Harlan, S.S., BegeH t, J.E., 1997. Geochemical constraint on the interpretation of magnetic property variations in loess/paleosol sequences from central Alaska, EOS, Transactions. AGU Fall Meeting 78, pp. F170-171. Sun, W.W., Banerjee, S.K., Hunt, C.P., 1995. The role of maghemite in the enhancement of magnetic signal in the chinese loess}paleosol sequence * an extensive rock magnetic study combined with citrate}bicarbonate}dithionite treatment. 1995. Earth and Planetary Science Letters 133, 493}505. Vlag, P., 1998. The magnetic signal in loess-sequences in Alaska. The Institute for Rock Magnetism Quarterly 8, 1}7. Zheng, H., Old"eld, F., Yu, L., Shaw, J., An, Z.S., 1991. The magnetic properties of particle-sized samples from the Luo Chuan loess section: evidence for pedogenesis. Physics of the Earth and Planetary Interiors 68, 250}258. Zhou, L.P., Old"eld, F., Wintle, A.G., Robinson, S.G., Wang, J.T., 1990. Partly pedogenic origin of magnetic variations in Chinese loess. Nature 346, 737}739.