PALAE0 ELSEVIER
Palaeogeography, Palaeoclimatology, Palaeoecology 131 (1997) 327-353
Mineral magnetic 'tracing' of aeolian dust in southwest Pacific sediments Paul P. Hesse
*
School of Earth Sciences, Macquarie University, N S W 2109, Australia
Received 1 May 1995; accepted 30 September 1995
Abstract
Reliable magnetic signatures were sought to trace aeolian dust in Tasman Sea sediments where the aeolian fraction had also been chemically isolated. Modern dust samples from eastern Australia were found to have a strong ferrimagnetic signal and a weaker antiferromagnetic (AFM) background. The ferrirnagnetic component could not be detected in the deep-sea sediments, where a finer biogenic magnetite component prevails over a background AFM and two other ?authigenic components. The modern dust is thought to be unrepresentative of prehistoric dust, probably as the result of contamination from local agricultural soils. The AFM component in the sediments, however, was found to be highly correlated with the independently determined aeolian concentration, except in intervals with a high magnetic susceptibility, high Zfd signal, possibly from an authigenic component. Because of the non-aeolian contribution to magnetic hardness (SIRM+IRM_3oo) and different relationships observed in different cores, the magnetic tracing of aeolian dust is currently only valid as a measure of relative changes in concentration in sediments with similar chemistry and low frequency dependence of susceptibility. © 1997 Elsevier Science B.V.
Keywords: dust; wind transport; magnetic susceptibility; deep-sea sedimentation; Tasman Sea; West Pacific; magnetic properties
1. Introduction
Aeolian dust as a detrital component in deepsea sediments is an important recorder of continental environments and atmospheric circulation and is a direct link between marine and terrestrial sediment records. Its analysis, however, is both tedious and prone to complication, particularly by contaminating volcanic and hemipelagic components (Olivarez et al., 1991). Mineral magnetics,
* Tel.: +61 02 9850 8384. Fax +61 02 9850 8428. E-mail
[email protected]. 0031-0182.,/97/$ l 7.00 © 1997 Elsevier Science B.V. All rights reserved, PII S0031-0182 ( 9 7 ) 0 0 0 1 0 - 2
or environmental magnetism, has been applied to recent sediments with the purpose of identifying the magnetic components and using them to trace the movement of sediment or determine the relative contributions of different components to a site ( T h o m p s o n and Oldfield, 1986). The ability to distinguish between different sources depends both on the existence of distinctive magnetic 'signatures' and the ability to distinguish between them by appropriate laboratory techniques. Aeolian dust should contain minerals representative of the surficial environment of the continents from which it is derived and therefore be distinguishable from marine authigenic or weathering components.
328
P.P. Hesse / Palaeogeography, Palaeoclimatology, Pahwoecology 131 (1997) 32 7-353
Available evidence suggests that magnetic tracing of aeolian dust is indeed viable. Modern dusts collected at Barbados show differences in the magnetic character of summer Saharan and winter South American dusts consistent with differences between the source areas (Oldfield et al., 1985). The red-brown summer dusts, most probably from the Sahara, have an antiferromagnetic (AFM; haematite/goethite) signal typical of arid region soils. Saharan dust collected in the western equatorial Atlantic (Carder et al., 1986) also showed high concentrations of Fe-rich minerals, suggesting that iron minerals may be a characteristic feature of Saharan dust either as independent grains or coatings on desert quartz. More recently, the concentration of the chemically isolated aeolian fraction of Arabian Sea sediments has been found to have a strong relationship with several magnetic parameters and, importantly, to be characterized by an A F M magnetic signal (de Menocal et al., 1991; Bloemendal et al., 1993). These few studies provide an empirical basis for application of the magnetic tracing of aeolian dust in deep-sea sediments. How general the AFM signal is for dusts from other continents, and how it is preserved in marine sediments, is still largely unknown. In this study the technique is applied to dust derived from Australia, possibly the largest source of dust in the Southern Hemisphere. In addition to exploring the potential for magnetic tracing of dust in this region, there are wider problems in the analysis of dust in marine sediments which require investigation. An as yet uninvestigated application of the magnetic tracing technique is in the differentiation of the aeolian component from major siliclastic contamination, most commonly from hemipelagic and volcanic sources (Olivarez et al., 1991 ).
1.1. Australian dust in Tasman Sea sediments
Dust raising is a common occurrence in arid and semi-arid Australia today (McTainsh et al., 1989) and is highly responsive to short-term fluctuations in rainfall (Yu et al., 1993). In the southeast, the zonal westerly winds are known to carry dust, entrained in inland Victoria, New South Wales and
South Australia, east over the Tasman Sea. Several instances of dust storms continuing out to sea and even depositing dust (or ~red rain') on the New Zealand Alps have been documented (Chapman and Grayson, 1903; Loewe, 1943; Glasby, 1971; Lourensz and Abe, 1983; Knight et al., 1995) and are summarized by McTainsh (1989). The aeolian contribution to Tasman Sea sediments has recently been investigated directly by isolation of the aeolian fraction following progressive removal of non-aeolian biogenic and authigenic components (Hesse, 1994b). In four cores from the northern Tasman Sea the relative concentration of the aeolian component is inversely related to carbonate content; however, mass accumulation rates ( M A R ) of the two fractions are strongly positively related (Hesse, 1994b). This reflects both the apparent lack of other major sediment components and the importance of marked changes in sedimentation rate in the calculation of aeolian MAR. Dust accumulation rates increase toward the south in the Tasman Sea (Fig. 1 ), with a marked increase between 3 3 S and 36")30'S in interglacial periods and between 30 33'S and 3 3 S in glacial periods (Hesse, 1994b). This latitudinal zone coincides with the present day summer position of the subtropical ridge (STR) of high pressure at around 35~S (Streten and Zillman, 1984) and therefore represents the average northward limit of westerly winds during summer. Cold fronts and their associated winds in the westerly flow are responsible for the transport of dust across the southeast Australian coast (Sprigg, 1982; McTainsh, 1989), although some events extend north of the average position of the STR (Knight et al., 1995). The approximately 33 (or 350 km) northward shift of the northern boundary of the east Australian dust plume in glacial intervals presumably records the migration of the STR northward in response to glacial conditions. This places the position of the dust plume and zonal westerly winds during the glacial periods further south than previous reconstructions (e.g. Thiede, 1979). Dust flux to the Tasman Sea increased markedly (up to 3 times) over interglacial levels during glacial maxima of the late Quaternary (Hesse,
P.P. Hesse/Palaeogeography, Palaeoclimatology, Palaeoecology 131 (1997)327-353
~
329
dust plume lin interglacial northern dust plume limit 2000m Isobath 0 500 L kilometres
SO-36-61i CI/8L " -
-~G---C~7S ~ , E39 ~"
E39.72e I Tasman eE26.3 Basin
-5561m eEZG.4
• NEW ZEALAND
Fig. 1. Location of core sites in the Tasman Sea and the east Australian dust plume (Hesse, 1994b). Modern dust samples were collected from the eastern highlands of Australia and Mildura.
1994b). The elevated levels of dust accession in glacial intervals are in agreement with evidence for drier conditions in the continent (Chappell, 1991) and high levels of dune building (Wasson, 1987). Glacial peaks in dust flux were evident in all cores examined, except the northernmost (SO-36-61), which lies outside the main dust plume, for the last three or four glacial stages (Hesse, 1994b). Earlier glacial periods do not show elevated levels of dust accession, in agreement with continental records that show desiccation of interior southeastern Australia beginning around 350-500 ka (An
et al., 1986; Readhead, 1988). In the marine record these intervals of high dust accumulation rate are not consistently marked by high relative concentrations of dust. The accumulation rate of the only other major component of the deep-sea sediment, calcium carbonate, also increased in the glacial intervals, with the result that no clear pattern of concentration of either component is observed. Magnetic tracers of the aeolian component, therefore, would also only indicate the concentration of dust, from which accumulation rates could not be inferred directly.
330
P.P. Hesse / Palaeogeography, Palaeodimatolog3', Palaeoecology 131 (1997) 327 353
1.2. Magnetic' characterization of the aeolian ./?action in marine sediments Three methods have previously been used to trace the aeolian fraction in sediments magnetically: (1) Assumption of an antiferromagnetic (AFM; e.g. haematite/goethite) mineralogy for the magnetic component of dust, following Oldfield et al. (1985), and measurement of A F M signal strength as a proxy of dust concentration. (2) Using total susceptibility as a measure of aeolian concentration on the assumption that the aeolian component is the principal contributor to susceptibility. (3) Searching for an empirical relationship between the independently determined aeolian fraction and any of several magnetic measures, without prior assumption of an aeolian signal. In some ways the last path may be seen as a test of the first two methods. Robinson (1986) used H I R M (Hard Isothermal Remanent Magnetization) as an index of A F M concentration, and therefore relative aeolian contributions, to cores in the North Atlantic. Subsequently, Bloemendal et al. (1988) used the 'accumulation rate' of H I R M to estimate aeolian flux to equatorial Atlantic sites, determined as the product of the 'magnetic concentration' by the dry bulk density of the sample and the sedimentation rate determined for that sample interval. Magnetic 'accumulation rates' are only valid for parameters which are primarily responsive to absolute concentration (e.g. susceptibility or Bloemendal et al.'s H I R M ({IRM 3o0+SIRM }/2). Petit et al. (1990) used the 'accumulation rate' or 'flux' of susceptibility in their study of core RCI 1-120 in the southern Indian Ocean as an index of aeolian input, although they did not demonstrate a direct relationship between the susceptibility signal and the aeolian fraction, de Menocal et al. ( 1991 ) demonstrated a relationship between susceptibility and the chemically isolated aeolian fraction concentration in ODP cores 721 and 722 in the Arabian Sea. This relationship was subsequently found to be dependent on the diagenetic environment in the cores, with evidence for the removal of fine mag-
netic grains with depth, and to be equally strong for H I R M and A R M as for susceptibility (Bloemendal et al., 1993). In this paper two approaches are followed, equivalent to methods 1 and 3 above. The magnetic components present in modern dust samples from eastern Australia were identified and compared with magnetic components identified in deep-sea sediments. In the second, correlation was sought between the concentration of the chemically isolated aeolian fraction and several magnetic properties used in previous studies.
1.3. D([Jbrentiation o/mixed magnetic' components Magnetic susceptibility and remanence are known to be highly sensitive to the concentration of magnetic phases (Mullins, 1977; Robinson, 1990; Thompson and Oldfield, 1986) but also to mineralogy, grain shape and size (Mullins, 1977; Thompson and Oldfield, 1986). Rock magnetic studies have revealed the properties of magnetic minerals strictly defined by composition, size or shape but usually of uniform populations of ferrimagnetic grains (O'Reilly, 1984; Maher, 1988; Dunlop, 1990; Diaz Ricci and Kirschvink, 1992). In sediments there are quite often several magnetic components (Robinson, 1986) and many rock magnetic techniques are unhelpful in distinguishing between mixtures of them. In particular, emphasis on the role of magnetic grain size has dominated over consideration of mixed populations (Oldfield, 1991). Quantitative modelling of mixtures has recently progressed rapidly to handle a wider variety of mineralogies and domain states (Thompson, 1986; Bloemendal et al., 1988; Maher and Thompson, 1992). The models, however, do not yet approach the complexity of many natural sediments. The first section of this paper therefore deals with the development of empirical methods for the differentiation of magnetic components using isothermal and anhysteretic remanence measurements.
2. Sites and samples Two sets of samples were used in the mineral magnetic analyses: modern dust samples from east-
P.P. Hesse / Palaeogeography, Palaeoclimatology, Palaeoecology 131 (1997) 327 353
ern Australia, to characterize the aeolian signature, and deep-sea core samples from the Tasman Sea. Six cores were sampled and analysed (Fig. 1) for magnetic properties and carbonate content, of which three were also analyzed for aeolian content and four for oxygen isotope ratios of foraminiferal calcite. The cores range in length from under 2 m to 13 m and in maximum age from 200 ka to 1.05 Ma. The northernmost cores, SO-36-61 and C1/86 6GC3, come from the crest of the Lord Howe Rise, within the Antarctic Intermediate Water, near the oxygen minimum zone. Core E39.75 lies on the southern flank of the Lord Howe Rise in Circumpolar Deep Water (CDW). These three cores are topographically isolated from hemipelagic sediment sources and receive essentially only aeolian and biogenic input. The three southern cores (E39.72, E26.3 and E26.4) have poor carbonate preservation, owing to their greater depth near the carbonate compensation depth in CDW. The aeolian fraction was extracted from the three northern cores SO-36-61, C1/86 6GC3 and E39.75. Two rhyolitic tephras in SO-36-61 and E39.75 were identified by major element chemistry as New Zealand ashes and several other weathered, possibly ash, laminae were seen, particularly in E39.72 (Hesse, 1993). The volcanic content is small, however, and individual tephra layers are usually only several millimetres thick. Benthic foraminiferal oxygen isotope stratigraphies were determined for each of the three northern cores and E39.72. Carbonate preservation spikes in E26.3 and E26.4 are correlated with E39.72 to provide time control. Core chronologies are discussed in detail in Hesse (1994b). A suite of modern dust samples was collected from sites along the eastern highlands of Australia from 28 ° to 37°S. The dust was sampled from moss mats, or pollsters, on elevated rock outcrops. Moss mats on a variety of lithologies commonly hold a thick (up to 1 cm) accumulation of fine reddish brown material clearly not derived from weathering of the supporting rock. The dust trapping qualities of Israeli mosses have been noted by Danin and G a n o r (1991) and moss pollsters are commonly used in pollen analysis because they trap falling pollen grains. From a total of over 80 samples the 18 highest quality samples were
331
selected for analysis. The criteria used for selection were bedrock type (granites, acid volcanics and quartzites), height of outcrop and proximity of local dust sources. Outcrops over 80 cm above the soil were preferred because of the potential for rainsplash to contribute local soil material to low moss mats. Sites distant from ploughed fields, dirt roads and other local dust sources were preferred. An unknown factor is the possible removal of fines by rain-splash and wash. One sample was obtained from the roof cavity of an old house in Mildura, in northwestern Victoria. Over a period of several decades up to 5 cm of dust had accumulated from the frequent dust storms in the area (Yu et al., 1993). Some results of the magnetic analysis of the Tasman Sea cores have been published previously (Hesse, 1994a). Extraction of the magnetic fraction from samples in core E39.72 (Fig. 1 ) and transmission electron microscope observation revealed a bacterial magnetite component which is apparently responsible for ferrimagnetic remanence.
3. Methods
All core samples used for magnetic analysis were dried at 35°C for up to 1 week in plastic cuvettes, weighed and packed with cotton wool to prevent movement of the sample within the cubes. Samples were not crushed after drying since the sediment was later to be sieved for microfossils. All moss mat samples were soaked in distilled water overnight and then sieved at 1 mm while agitated in an ultrasonic bath for 30 min. The moss was discarded and the fines then passed through a 63 gm sieve and the >63 lam fraction also discarded. The sample was concentrated by centrifuging and then allowed to settle and dry in a warming oven at 35°C for up to 1 week. The dried samples were then prepared for magnetic analysis as outlined above. Several drops of hydrochloric acid were added to two samples, MF32 and MF68a to aid settling. The results show that this has probably altered the mineralogy significantly, justifying the decision not to chemically remove organic matter from the samples.
©8
:D
1-..~ 0
0
0
0
~D
-
.
°
.
0
o
o
°
o
I
o °
~2,
I
0
o
0
I
o O~
~,;%0
o °~ o
o
(10-8 m3kg-l)
o? 0~o
)~
@~oU o~
S ~oOO°~-0;°°
i
(10-8 m3kg-I) I
0
t~
0
(..3
0
0
0
0
-
0
o,
,
0
I
I
0 I
0
0
o ° 0%
~D
X ( 10-8 m3kg-I
J
0
o°
I
o o
6~ °
)~ (10 -8 m3kg -1)
0
~
0
I
(:~
('~
CD
0
E 0
0
~8
tao
Q
o~
I
0
0
~
o_
0
O°
Oo¢
%0
\
o2:oO o,:oo
o
0
(10-8 m3kg-l)
I
0
( 10-8m3kg - 1)
380
I
oooo o
Q
i,
e oo
0
)~
0
I
0
0
:°
~
CD
t~
J
O
&
..
b-)
P.P. Hesse/Palaeogeography, Palaeoclirnatology,Palaeoecology131 (1997) 327-353 Susceptibility measurements were made using a Bartington susceptibility meter (M.S.2) at two frequencies: 4 . 7 k H z and 0.47kHz (low and high frequency, respectively). Mass specific susceptibility (low frequency) was calculated and is used here with SI units (m 3 kg-~). Anhysteretic magnetization was applied to all samples after susceptibility measurement with a peak AF field of 100 m T and DC field of 0.04 mT. Anhysteretic remanent magnetization ( A R M ) was measured on a Molspin magnetometer and is expressed either in S! units of A m 2 kg-1 or as the susceptibility of A R M , XARM,in units of m 3 kg 1 (Maher, 1988; Oldfield, 1991). Isothermal remanent magnetization ( I R M ) was generated in a pulse magnetizer. IRM84 o m T is also referred to as saturation isothermal remanent magnetization, SIRM, as it is the largest remanence which can be produced on the pulse magnetizer and is more than adequate for saturation of ferrimagnetic grains (Thompson and Oldfield, 1986). In addition two reverse field I R M s were generated, after SIRM, at - 1 0 0 m T and - 3 0 0 m T ( I R M - l o o and I R M 3o0). Remanent magnetizations are expressed in units of A m 2 k g - 1.
4. Tracing the modern dust magnetic signature 4.1. Magnetic analysis of deep-sea sediments 4.1.1. Calcium carbonate 'dilution' of the magnetic signal In natural sediments magnetic susceptibility and the intensity of remanence measurements are strongly dependent on the concentration of the magnetic minerals. This is particularly true of mass-specific magnetic susceptibility, g (Mullins, 1;977; T h o m p s o n and Oldfield, 1986; Robinson, 1990). A major cause of variation of susceptibility in marine sediments is the contribution of non-
333
magnetic biogenic material, especially calcium carbonate. The Tasman Sea cores show an increasing dependence of susceptibility on carbonate content with depth, as carbonate dissolution becomes more pronounced (Fig. 2). By no means all of the variability of susceptibility in those cores is due to carbonate dilution, especially where more than one population of samples is present. Each group is seen as a band of points with negative gradient, possibly three in the case of E39.72 and two each in E39.75 and C1/86 6GC3 (Fig. 2 b - d ) . These populations may possibly be the result of varying (1) grain size, (2) mineralogy or (3) concentration of grains within the non-carbonate fraction. Further magnetic analysis is required to distinguish between these possibilities.
4.2. Characterization of magnetic mineralogy Ferrimagnetic (e.g. magnetite and titanomagnetite) and A F M minerals are the two dominant and most readily distinguishable magnetic minerals in sediments. The several different methods used in the literature to characterize the relative contributions of antiferromagnetic and ferrimagnetic components to the total sediment magnetic signal are based on their contrasting 'hardness', or coercivity (Fig. 3a). To this end, rapid estimates of, or surrogates for, the coercivity of remanence have been sought. Oldfield suggests three measures of hardness (Oldfield, 1991, tables 1 and 2). These are similar in principle to the measures used by Robinson (1986); I R M loo/SIRM ( = S ratio) and S I R M - I R M - 3 o o ( = H I R M ) , and also by Bloemendal et al. (1988); - I R M 3oo/SIRM ( = S ratio) and ( I R M 3oo+SIRM)/2 ( = H I R M ) . The HIRM (Hard Isothermal Remanent Magnetization) is ideally dependent only on the concentration of the A F M components within the
Fig. 2. Magnetic susceptibility as a function of calcium carbonate content in the six Tasman Sea cores. Cores are arranged from (a) to (f) from north to south, which is also the order of increasing water depth, except for the deepest two cores. The modern lysocline lies close to 3600 m and cores below that depth show a greater range of carbonate values, in part reflecting variable carbonate preservation. The low range of carbonate content in the two shallowest cores may be because of variable input of aeolian dust and only minimal carbonate dissolution. In E39.75 and E39.72 at least two populations of samples are clearly seen, due to different dominant magnetic mineralogies.
334
Haematite
Size Coarse SD Fine SD
Magnetite
IO
1000
100
Reverse
Size (pm)
Field (mT)
(b)
I
3 _ z ‘z ~------~-------.......________:__~_-______________________ z 2
1 Haematite
:
0 c’ocmr SD
0 ? Jk
-I pd
SD
Magnctitc I
I 0
-I IRM.
I QO/SIRM
Fig. 3. (a) Gcnerdhzed coercivity of remancnce for magnetite and haematite of varmu sizes. Magnetite (after Thompson and Oldfield. 1986) shows softer remanence, with normalized IRM approaching - 1 at fields less than 100 mT. Haematite (after Bancrjee. 1971: Oldfield and Robinson, 1986) shows harder remanrncc, and higher coercivity. with increasing sire in the SD range. Vertical dashed SIRM values. (b) Bivariate plot of IRM _ ,,,,.!SIRM and IRM ,,,,,:SIRM values for lines indicate the IRM ,,,,,.!SIRM and IRM _10o..1 the magnetite and haematite size fractions shown in (a).
bulk sediment and aims to measure the concentration of antiferromagnetics by the intensity of remanence remaining at IRM _30,,. AFM concentration determined by HIRM is considered later. Both of the S-ratio values attempt to show the
hardness of the magnetic signal (i.e. the relative strength of AFM to ferrimagnetic components) and are therefore not dependent on or responsive to the concentration of non-magnetic minerals (clays or carbonate). These concentration-independent
P.P. Hesse/Palaeogeography, Palaeoclimatology, Palaeoecology 131 (1997)327-353
parameters are therefore valuable in determining the dominant magnetic mineralogy in a sample. Plotting IRM_loo/SIRM against IRM-3oo/ SIRM (Fig. 3b) effectively reduces the coercivity curve over this range of field strengths to a single point. Magnetite of all sizes plots in a narrow field whereas SD haematite size classes are dispersed (Fig. 3b); however, any mixing between the two mineralogies will give points plotting along a narrow band, irrespective of size of either component (Fig. 3b). For systems in which only magnetite and haematite are present, either parameter would appear to be an ideal measure of relative mixing independent of size effects. Underestimation of true SIRM (grown in a field of 840 mT, less than the true saturation magnetization) results in slightly higher negative values of both parameters for haematite. Tasman Sea samples plotted in this manner show a distribution resembling an acute 'V', considered to represent mixing between three end-member components (Fig. 4). Two of these components are consistent with the predicted fields for magnetite and haematite (and/or goethite). Mixing of these two components is seen where points fall along the line linking the end-member 'pure' values. The prominent third end-member is most likely a third mineralogy and can be highlighted by calculating the distance from the magnetite-AFM mixing line (rescaled so that for points on the line IRM_ loo/SIRM = IRM 3oo/SIRM). This 'D' (difference) component value is defined as - ( I R M _ 3 o o / S I R M - IRM loo/SIRM), where both parameters are rescaled using end-member values determined from the graph (Fig. 4a). The principal advantage of selecting the end-member values is in the ability to compare different cores, even though the scale and zero value are somewhat arbitrary. Because of the much stronger magnetite remanence, IRM 3oo/SIRM values do not reflect linear mixing of magnetite and A F M components. At SD magnetite concentrations greater than approximately 10% IRM 3oo/SIRM shows little change (Bloemendal et al., 1993) and therefore antiferromagnetic minerals may be volumetrically more abundant even in samples where high negative IRM 3oo/S1RM indicates magnetite dominance of
335
remanence. Coarser multi-domain magnetite is less strongly magnetic and therefore IRM-3oo/SIRM is sensitive to a wider range of mixtures of MD magnetite with A F M components (Bloemendal et al., 1993). In this sense the 'concentrationindependent' parameters are, in fact, sensitive to small changes in the relative magnetite concentration when concentration is low but are rather insensitive to antiferromagnetic abundance. The Tasman Sea samples, therefore, are dominated magnetically by magnetite and the ferrimagnetic 'D' component and only relatively few samples have low magnetite concentration (Fig. 4a). Antiferromagnetic minerals may comprise a background which is only detected in 'windows' of low magnetite concentration. While magnetite-AFM and magnetite-'D' component mixtures are observed, 'D' c o m p o n e n t - A F M mixtures are almost totally absent. It can be seen that IRM loo/SIRM cannot differentiate between the 'D' component and magnetite A F M mixtures and that IRM 3oo/SIRM, while largely distinguishing between the 'D' component and AFM, gives similar values for magnetite and the 'D' component. It can be seen from Figs. 3 and 4 that IRM 3oo/SIRM is the most convenient measure of A F M versus ferrimagnetic ('D' component and 'magnetite') contribution. In discussion of the cores, 1RM_3oo/SIRM and the 'D' component values are used to describe the mineralogical composition of the samples.
4.3. Characterization of magnetic grain size The theoretical and observed dependence of many magnetic properties on grain (crystal) size has been used by many workers to characterize the size of the magnetic components in sediments or rocks. Of the many used (King et al., 1982; Robinson, 1986; Thompson and Oldfield, 1986; Maher, 1988; Oldfield, 1991; Bloemendal et al., 1993) most are based on studies of sized samples of a single mineral, usually stoichiometric magnetite. Non-magnetite and mixed mineralogies therefore complicate simple interpretations of sizesensitive parameters. Measures of grain size are commonly derived by comparing two parameters, for example, A R M / z (King et al., 1982) or SIRM/z
336
P.P. Hesse ,' Palaeogeography, Pulaeoclimatology, Palaeoecology 131 (1997) 327 353
(a) All Cores - 6 7 2 points -0,7
Antiferromagnetiq /~ C)
~
Magnetite 'D' Component
=.0,-~a" ~o
/ ]
° "'" .~'~ =~ g .~! o
-0.8 -
~-0.9
-I .0
i
-1.0
-0.8
-0.6
-0.4
-0.2
IRM_ 1oo/SIRM
(b) 1
IRM
~'. .o \'..',
'SIRM IRM 300/SIRM
"~
"1 ',"., ",,
O'
i 'L L" '~"
I I
I I
I
I
I
I
Antiferromagnetics magnetite-AFM mixing lines Fine SD Magnetite 'D' Component
.
10
.
.
.
.
.
.
.
I
, - - i -"-'," -,-..
100
-,,
1000
Reverse Field (mT) Fig. 4. (a) Bivariate plot of IRM_mo/SIRM and IRM 3oo/SIRM values for all 672 deep-sea samples used in this study (squares). The positions of three inferred end-member components are shown by the large open symbols. Mixing between these components is inferred to give rise to the points scattered along the dashed lines between the end-members. (b) Inferred coercivity curves of the components identified in (a) (thick lines) and mixing between the magnetite and AFM components (thin dotted lines). The 'D' component curve intersects these mixing lines as it is rapidly reversed in the interval between 100 and 300 mT.
(Thompson and Oldfield, 1986). Such inter-parametric ratios have the benefit of being concentration independent but those which use Z give similar values for sizes both coarser and finer than the SD/MD boundary. Maher (1988) has shown that ZARM/SIRM is diagnostic for SP and SD grains
but is also sensitive to the packing density of the grains in the sediment because interacting SP grains behave like SD grains (Fig. 7). As measures of fine and coarse grain size (Oldfield, 1991) in magnetite, ZARM/SIRM and 1RM2o/ARM, respectively, are internally consistent. In this study a
P.P. Hesse/Palaeogeography, Palaeoclimatology, Palaeoecology 131 (1997) 327-353
logarithmic distribution was found, such that either parameter presented on a logarithmic scale is in fact equally sensitive at both high and low values and hence only the more commonly used ZARM/S1RM will be discussed hereafter. Samples in this study could be interpreted as representing an array of magnetite particle sizes stretching between two end-points of different particle size (e.g. samples from SO-36-61, Fig. 5b,c). However, samples of all mineralogies described above have values of ZARMand SIRM and therefore must also plot on the same fields. Comparison of the 'size' parameter ZARM/SIRM with 'mineralogy' indicators, IRM 3oo/SIRM and IRM_loo/SIRM,
337
reveals an apparent dependence (e.g. Fig. 5). Mixing between, or gradation of, magnetite grain sizes alone cannot account for samples with low negative IRM_3oo/SIRM values (compare Figs. 5a and c) which are diagnostic for A F M mineralogies. We can propose two scenarios to account for the distributions seen in Fig. 5: (1) two magnetite populations of different sizes, or a continuum of sizes, are mixed with an A F M component; (2) a single population of fine-grained magnetite is mixed with an A F M component. Both scenarios temporarily neglect the 'D' component. The first scenario is difficult to sustain since samples approaching the coarse magnetite field are absent; -0.70
(a)
-0.75
o oo° o
@AFM
component
~
fine SD Magnetite ( ~ coarse SD Magnetite O 'D' Component
-o.8o
o~ oo
~ -0.85~, ~,:~ -0.90-
;~
o
o
o% o 08
o
o ~
Du
0
-0.95 -1.00 -1.0 I
I
10
I
L
10
~
-0.6
-0.4
-0.2
, IR,M-lqO/SIRM i
(b)
E
-0.8
,
(c)
>
K ~e3
o
& o
100-
¢%~,
0
©
o oOo o
< i
i
oO o o
o
g" o
o
tm
>
o g
~°¢¢3E3 o
o oo |
o~8O % 0 o
"100
o
o
©
i
o-
~ |
!
-0.70 -0.75 -0.80 -0.85 -0.90 -0.95 -1.00 IRM_300/SIRM
-1.0
-0.8
°o
o
i
|
-0.6 -0.4 IRM_ 100/SIRM
-0.2
Fig. 5. Bivariate plots of the three parameters used to differentiate magnetic components, ZARM/SIRM, I R M Io0/SIRM and IRM_3oo/SIRM, for core SO-36-61. The three plots are arranged as though collapsed from a box in which (c) forms the horizontal plane. (a) The 1RM_loo/SIRM and I R M 3oo/SIRM field and components are the same as in Fig. 4a. Samples to 65 cm depth are shown by symbols with a thicker outline and indicate probable 'D' component contribution. These shallow samples in SO-36-61 fall on a slightly different distribution to the overall pattern (Fig. 4a), and this is seen clearly in (b), In both (b) and (c) the field between the coarse and fine SD magnetite end-members and between coarse magnetite and A F M end-members are empty, except for a single tephra sample.
P.P. Hesse / Palaeogeography. Palaeoclimatology, Palaeoecology 131 (1997) 327 353
338
netite grain size are also affected by the presence of other magnetic minerals. The commonly used 'size' parameter, ZARM/SIRM,is ambiguous at low values and cannot distinguish between coarse magnetite and AFM minerals. Antiferromagnetics have low 7~ARM/SIRMand low negative IRM 3oo/ SIRM and therefore, in mixed mineralogies values of 7~ARM/SIRM alone, may respond principally to mineralogy rather than magnetite grain size. Because no single parameter could reliably be interpreted in terms of variations in grain size only, bivariate plots of the three parameters ZARM/SIRM, IRM_loo/ARM and IRM_3oo/SIRM (e.g. Fig. 5) are necessary to clarify mineralogy and grain size effects. The frequency dependence of susceptibility, 7,rd, is widely used to identify grain sizes in the superparamagnetic (SP) range (Maher, 1988). Maher showed that, for sized samples of equant magnetite grains dispersed in a non-magnetic matrix, both Zfd and /~ARM/SIRMincrease with decreasing size to the SD/SP boundary (Fig. 7). A single finer sample, of mean crystal size 0.012/am, gave negligible Zrd and moderate ZARM/SIRM, which Maher
that is, it would imply that, although fine magnetite occurs without the AFM component, coarser magnetite always does (i.e. we never see low ZARM/SIRM, high negative IRM_3oo/SIRM samples). In the second scenario all the patterns are explained simply by mixing between the proposed fine magnetite population (high ZAR~/SIRM, high negative IRM_3oo/SIRM) and the proposed AFM population (low ~RM/SIRM, low negative IRM 3oo/SIRM). This situation is complicated only slightly when data from all six cores are compiled (Fig. 6). What appeared as a narrow band of points in SO-36-61 appears as a much broader scatter of points with hard remanence. This scatter of AFM-dominated samples may be due to either a range of AFM particle sizes or mineralogy (e.g. haematite and goethite); however, the field near the expected coarse magnetite end-member is empty. Particle size variation may account for the small scatter of values of ZARM/SIRM seen in the ferrimagnetic (fine SD) cluster (Fig. 6). The samples considered here demonstrate that magnetic parameters commonly used to infer mag10
4:
I0
°o
100
oo~
<
,----ce
o~
o
:"
:Q°
-o.o
-&
-o s
-o;o
mM_3odSmM
Oo
100
o
g
°°°~°
:o°° J 5
,.oo
-,.o
°
°-
-L
mM_loo/Sn~M
Fig. 6. Plots of all deep-sea sediment samples, as in Fig. 5b,c. End-member mineralogies (large open shapes) as for Figs. 4 and 5. Bars attached to the A F M component represent possible range of values of ZA~M/SIRM for the end-member A F M component(s) to account for the observed scatter. In both plots the field near the coarse SD magnetite end-member is empty, except for a single altered tephra sample from E26.4 which displays erratic behaviour.
P.P. Hesse / Palaeogeography, Palaeoclimatology, Palaeoecology 13l (1997) 32 7-353
339
12.0
12.0
(a) Deep-Sea Sediments
00.022 10.0 -
O0.016
10.0 -
8.08.0-
0 0.023
k~ ,.A
°~~
6.0-
Zfd(%) 6.0-
~ee~ ~:c:~ °
~
4.0-
4.0GO 00..O.O34 2.0 -
~
C3 3
O 0.032
....... 10
©0.069
o Q
Doc
o
o
-2.0
100 ZARM/SIRM (10-5 A-lm)
D
0.0-
O0.0q 2
I
~oo
z
c
D
0.0
2.0-
.
10
.
.
.
.
.
.
|
XARM/SIR M 100
12.0 (b) M o d e m D u s t S a m p l e s J []~
Fig. 7. Data from Maher (1988) showing the behaviour of frequency dependence of susceptibility (Zfd) and ZARM/SIRM for sized synthetic magnetite crystals. Diamonds represent points in the 'new MT' series with lower concentrations of magnetite than the 'MT' series (squares). Numbers indicate mean diameter of magnetite crystals in the 'new MT series' in microns.
10.0 []
8.0-
6.0LJ
suggested may have been due to the high frequency field (the same as used in this study) being too low to block in the SP fraction. The frequency dependence of the samples was also found to be highly dependent on the concentration, or clustering, of magnetite grains (Maher, 1988) (Fig. 7). In Tasman Sea samples with high or moderate values of Z, the frequency dependence of Z (Zfd%) was calculated as a percentage of the LF volume susceptibility; Zfd% = ((ZLV- - Z H V ) / Z L V ) X 100. Where the raw values of volume susceptibility are below 10- ~ m 3, Zra% cannot be accurately determined and such samples have been excluded from the following discussion. The samples cover a range from a cluster with high ZARM/SIRM and high Zrd to a tail with equally high ZAgM/SIRM but low Zfd (Fig. 8a). The low Zfd, high ZARM/SIRM samples seen, almost entirely from E39.72, are not predicted by the experimental results (Maher, 1988), except perhaps for clustered SP magnetite grains (Fig. 7). These samples are not enriched in 'D' component and obviously not the low XARM/SIRM AFM-influenced group (Fig. 6). Measurement of extracted bacterial magnetosomes (Hesse, 1994a) from E39.72 revealed
4.0w
2.0-
0.0-
-2.0
.
10
.
.
.
.
.
.
I
100
XARM/SIRM (10-5 A- 1m) Fig. 8. (a) Frequency dependence of susceptibility related to ZARM/SIRM for the deep-sea sediment samples (with reliable Zfd, see text) showing a band with high ZARM/SIRMbut a broad range of frequency dependence of susceptibility. Samples plotted are those shown in figure 9-III so the distribution neglects most samples with low ZARM/SIRM. (b) Modern dust samples, showing a continuum of sizes from coarse to fine SD magnetite, following the results presented in Fig. 7.
sizes consistent with those expected for the fine magnetite component detected by magnetic analysis. The magnetosomes have a broad range of geometries and fill most of the SD field but do not extend to the small sizes (0.01 gm) suggested by the Zfd values. There is also no indication of down-
P,P. Hesse/Palaeogeography, Palaeoclimatology, Palaeoecology 131 (1997)327 353
340
a SO-36-61
b C1/86 6GC3
[RM_300/S1RM IO
-({9
c E39.75 PC
IRM-300/SIRM
{~
-017
.........
o
(}
i
.
8
i
,
i
d E39.72
IRM_300/SIRM ....
,
i ........
1
e E26.3
[RM.~oo/SIRM
g
......
o
0 .....
8
f E26.4
IRM-300/SIRM . . . . . . . . . . . .
IRM_300/S1RM
8
07
.....
~9
{~N
(]7
o Stage 60 2(N}
4(~1 "
40(I
6(~}"
(,{~)
*'~'-
*'~'
,
>
. . . . . . .
2{~) -
il, )
4(x)'
S u b sta~e I i
I F e r r i m a g n e t i c and Antiferromagnetic Components
8(~) Brunhcs blatsuyama
12{~1"
'D' C o m p o n e n t
011 o
~2{}0
oil "
'D' C o m p o n e n t
0i3 "
0i5 "
0i {) i
01 ,
i
'D' C o m p o n e n t
03 .
200
i
1} ~ .
t
0l o i
oil ,
'D' c o m p o n e n t
~}i} ,
2{R)
0i5 •
0i
o~
0~
o
Ul o •
2(}~)
4(~1
'D' Component
'D' C o m p o n e n t
i} 5
Ol '
i
0.3 ,
r
05 ,
010 o
20{1
200"
(H
4IR)"
0 I0 ,
i
0.3o 05o .
I
~
.
i
tephru (o 921
II 'D' Component x{~}
x(x} -
I{~K)
Zfd (%LF)
Zid (% LF)
Xfd (% LF)
Xfd (~LF)
)
. ~ , ~ • 7
0
200 -
2(~)"
4(X)"
~)-
OX)"
III S u p e r p a r a m a g n e t i c Component
2 12(~)"
P.P. Hesse/Palaeogeography, Palaeoclimatology, Palaeoecology 131 (1997) 327-353
core fining of the magnetite fraction (c.f. Fig. 9Illd), for example as the result of dissolution (Bloemendal et al., 1993). SP maghaemite, produced by low-temperature oxidation of the SD magnetite fraction, was previously suggested as a separate component accounting for the high Xrd samples in E39.72 (Hesse, 1994a). The sudden, simultaneous increase in 7~faand sediment colour at 5 m depth in E39.72 (Fig. 9-IIId) was explained as marking a stepwise increase in bottom water oxygen concentration at around 350 ka. A similar pattern is seen in E39.75 (Fig. 9-IIic). This explanation may not be necessary, or adequate, to account for the distribution seen in Fig. 8a, if Zfd values are truly indicative of grain size. The high ZARM/SIRM, high Zfa samples seen in all cores are consistent with predicted values for fine SD magnetite (Maher, 1988) and therefore it is the group of low Zfa samples from the bottom of E39.72 which may represent an altered or alternative component. The sudden increase in Zfa could be explained as a diagenetic front and the frequency dependence record to be responding to some diagenetic change in the sediment not reflected in the other magnetic parameters; however, this cannot explain the intervening apparently non-altered (or non-oxic) intervals above the 'front'. The significance of the )~fd data is therefore not clear, and open to interpretation.
4.4. Downcore distribution of magnetic components The downcore distributions of the components, and their distribution in the sequence of cores from a range of marine environments, gives some indication of the origin of the components and particularly their diagenesis. In Fig. 9 IRM 3oo/SIRM, 'D' Component and Xrd are shown for the six cores, arranged in order from north to south (left to right), which is also close to the order of increasing depth, from 1350 m to
34l
4810m. The two shallowest sites lie within the oxygen minimum zone, slightly below core Antarctic Intermediate Water (AAIW). The remaining sites are under the influence of Circumpolar Deep Water ( C D W ) but with increased mixing of Antarctic Bottom Water ( A A B W ) with depth. Consequently, oxygen concentration of bottom waters increases with depth in the sequence, or from left to right in Fig. 9. Overall, there is a decrease in AFM-dominated sediments, indicated by IRM-3oo/SIRM close to - 0 . 7 , with increasing water depth (Fig. 9-1). This is in contrast to the trend toward higher aeolian content in the deeper southern cores (Hesse, 1994b). In the two shallowest cores the A F M component dominates while in E39.75 and E39.72 it is restricted to discrete, well spaced intervals down the cores. In the two deepest cores there is no detectable A F M influence on IRM_ 30o/S1RM. This trend is somewhat surprising in that the most likely A F M minerals, goethite and haematite, both containing solely ferric iron, are most common at sites with the lowest environmental oxygen today. The pattern is most likely due to preference of the magnetite-producing bacteria for high oxygen conditions (Hesse, 1994a) and the dominance of the magnetic signal by the biogenic SD magnetite. In E39.72 the A F M intervals were observed to coincide with glacial isotope stages (Hesse, 1994a) however the A F M intervals in E39.75, although similar in appearance, are not correlatable and the lower two such intervals extend over both glacial and interglacial stages. The A F M intervals are most probably the result of low oxygen conditions, either as the result of lower water mass oxygen concentration or greater consumption of oxygen by breakdown of organics in the uppermost sediments, and hence were periods unsuitable for the aerophilic magnetotactic bacteria. The A F M signal appears unchanged with depth and sediment chemistry is, therefore, probably sub-oxic or oxic to a
Fig. 9. Three diagnostic magnetic measures, I R M _ 3oo/SIRM, 'D' component and Zfd, for the six cores arranged in order from north to south (left to right). Dotted stratigraphic time-lines represent Termination II (stage 6.0; ~ 130 ka), sub-stage 11.3 ( ~ 410 ka) and the B r u n h e s / M a t s u y a m a palaeomagnetic reversal. The isotopic levels were determined in each of the four northern-most cores and correlated to E26.3 and E26.4 from E39.72 using the carbonate preservation stratigraphy. Weak volume susceptibility gave unreliable Xfd values for SO-36-61, C1/86 6GC3 and intervals of E39.75 and E39.72, not shown.
342
P.P. Hesse/ Palaeogeography, Palueoclimatology, Palueoeeolog), 131 (1997) 327 353
depth of at least 10m in E39.75 and 13m in E39.72 ( Fig. 9-I ). The 'D' component is largely restricted to the core tops of all cores except E26.4 (Fig. 9-II ). The depth of the 'D' component zone varies and has either a sharp lower boundary (SO-36-61, C1/86 6GC3, E39.72) or a gradual decline from lower absolute values (E39.75, E26.3). It is possible that this component is restricted to the uppermost oxic layer of the sediment and is destroyed under suboxic conditions. Sediment chroma is strong in the top of E39.72, indicating oxic conditions (Lyle, 1983), but overall does not match the 'D' component distribution (Hesse, 1994a). Likewise, in C1/86 6GC3 and SO-36-61 there is no relationship between 'D' component distribution and sediment colour (Hesse, 1993). Bottom water oxygen concentration near E26.4 is high because of strong AABW influence (Mulhearn, 1985) and may explain the strong presence of the 'D' component throughout the core. As observed above, samples show almost no mixing between the 'D' component and A F M end-members (Fig. 4) but the 'D' component is strongly linked to magnetite in the core tops and in E26.4, suggesting a genetic or environmental link. The apparently rapid diagenesis seen in some cores may be evidence of very small grain volume, possibly poorly crystalline or micro-crystalline iron minerals. 4.4. I. Magnetic analysis o f modern dust samples All the modern dust samples are dominated by a ferrimagnetic component, probably magnetite, and have a much weaker antiferromagnetic background (Fig. 10). The values of XARM/SIRM and Zfa (Fig. 8) indicate a range of magnetite grain sizes from coarse SD to fine SD, although values of Zfa are higher at given values of 7~ARM/SIRM than those found by Maher and Taylor (1988) for sized pure magnetite grains, possibly as the result of A F M contributions. The samples reveal a weak, but definite, contribution of haematite to the overall magnetic signal (Fig. 10a), with the majority of samples displaced away from the magnetite endmember of the magnetite~AFM mixing line. There may also be a small 'D' component influence on several samples, displaced to the right of the magnetite-AFM mixing line. Antiferromagnetic
concentration, estimated by SIRM +IRM-3oo, is generally high but variable (Fig. 10d), possibly due to the presence of non-magnetic organics in the samples, and has no correlation with IRM 3oo/SIRM as expected if AFM contribution to IRM 3oo/SIRM is drowned out by the strong magnetite signal. In Australia the iron-oxide minerals, goethite, haematite, maghaemite and magnetite, are both common and widespread products of pedogenic processes (Taylor et al., 1983 ). Goethite and haematite are the most abundant soil iron oxides ( Taylor et al., 1983), reflecting the strongly oxidizing soil environment common throughout most of Australia, especially in the arid and tropical zones. Taylor et al. (1983) describe maghaemite as commonest in very weathered soils of the tropics and subtropics but it is also common in more temperate regions where it may be formed by high temperature inversion of haematite during bushfires (Anand and Gilkes, 1987). Maghaemite may also form authigenically in soils (Verosub et al., 1993). Magnetite may occur either as a primary mineral, a biological (Fassbinder et al., 1990) or authigenic product (Taylor et al., 1987; Maher and Taylor, 1988; Zhou et al., 1990; Maher and Thompson, 1991 ), however the range of soil conditions under which biogenic and authigenic magnetites are formed is almost completely unknown. Haematite and goethite are present in desert soils and dust samples (Walker and Costin, 1971; Mokma et al., 1972) and are indicated by the prominent red colour of the moss mat samples used here. While the prominent magnetite contribution to the modern dust samples is, therefore, not necessarily expected for arid source dusts, the AFM signal is expected for dusts derived from desert soils. 4.4.2. Tracing the modern aeolian magnetic signal in Tasman Sea sediments The modern dust samples, although apparently forming a spectrum of compositions, have a distinctive magnetic character, notably coarse to fine SD magnetite mixed with an AFM component. The low XARM/SIRM and soft remanence consistent with coarser SD magnetite (Fig. 10) is not observed at all in the deep-sea cores (Figs. 4a and 6). An A F M component is prominent in the
343
P.P. Hesse / Palaeogeography, Palaeoclimatology, Palaeoecology 131 (1997) 327 353
-0.7
O
AFM component
~
O
'D' Component
O
fine SD Magnetite
&
coarse SD Magnetite
(a)
-0.8
i
~
-0.9
COo -1.0 -1.0 10
I
I
I
-0.8
-0.6
-0.4
-0.2
10
(b)
E
O
(c)
>
O D
©
0
o
tP
tz m o t~a q o
loo
0
O
&
-0.7
I
!
-0.8
-0.9
-1.0
IRM_3oo/SIRM 1.0
<
o
o
e~ <
O4
o o
¢o ~e
I00
O
m
-1.0
i
I
I
I
-0.8
-0.6
-0.4
-0.2
IRM_ 100]SIRM
(cl) oo
0.8-
m
0.6o~
+
N r~
o Q
0.4-
D Q
m o
0.2m
0.0 - 1.00
I
I
I
-0.95
-0.90
-0.85
-0.80
IRM_300/SIRM Fig. 10. Magnetic components in modern dust samples. Bivariate plots presented as for Fig. 5. The two samples closest to the AFM end-member in (a), (b) and (c) were treated with a small quantity of HCI to aid settling during preparation and their magnetic mineralogy has probably been affected as a result. (d) AFM concentration (SIRM + IRM 3oo)plotted against IRM _ 3oo/SIRM, which shows relative AFM contribution to the SIRM signal (increasing to right).
344
P.P. Hesse / Palaeogeography, Palaeoclimatology, Palaeoecology 131 (1997) 327 353
(a) SO-36-61 Aeolian (%) 0
5 . |
oI
10 I
(b) C1/86 6GC3 Aeolian (%)
15 I
20 I
0
5 I
o /
IO I
(c) E39.75 PC Aeolian % (adj)
15 I
20 I
10 I
o
20 I
30 I
40 I
50 I
60 I
Depth (cm) • ~....
oo0
o.oo
0 . 0 5 o.1o SIRM+IRM-300 (10-3 Am2kg-l)
0.05
,::~?-:__-~ -
o.Lo
" SIRM+IRM-300--. (10-3 Am2kgd) 5o0
I
__
Aeolian
........
SIRM+IRM_300
750.
i..................... :-~
|000 0.00
-0.20 I
0
(era) 1
0.00 I
J
,
0.20 I
~
2001
0.10 0
'
0.00 I
0.10 I
0.20 ~
I - ~
~
I
0.30
030 1
0.25
000
0.25
0.50
0.75
I
I
I
I
0
•
200-] ~
"
I
0.20
SIRM+IRM_300 excess SIRM+IRM_300
excess SIRM+IRM_300
excess SIRM+IRM_300
I
0.10
2oo-
-
400"
600 -
II (10-3 Am2kg-1) 800"
1000
Z (CO3 free) 0 uh i
Depth
20 I
(cm) t ~
40 I
60 I
80 I
X (CO3 free) 100 t
0 u. /
.... ] ~
20 I
40 I
60 I
X (CO3 free) 80 I
I O0 I
0
0 I
,
50 l
-
-t 400
6o0
S~
~
III (10-8 m3kgd)
< 800" ~
) 1(~)0
,
%
1O0 I
,
150 I
P.P. Hesse / Palaeogeography, Palaeoclimatology, Palaeoecology 131 (1997) 327-353
marine cores (Fig. 9), preserved to at least 13 m depth, and is possibly of aeolian origin, although an authigenic or diagenetic origin cannot be ruled out. If the A F M component is in fact aeolian then it must be independent of the aeolian magnetite component. We can only speculate on the reasons for the absence of the aeolian magnetite signal in the marine sediments. One possibility is that the magnetite fraction may have been destroyed or altered within the marine sediment column. However, the lack of a detectable coarse SD magnetite component in even the core tops, as well as the apparent stability of the biogenic magnetite and A F M components in the marine sediments (Fig. 9), argues against diagenetic alteration. Alternatively, it is possible that, under modern land degradation conditions in the humid and semi-arid fringe of the Australian continent, the moss mat dust samples may be retaining dust which has no equivalent in the past, for example from subsoils or the cropping zone immediately to the west of the highlands.
5. Correlating down-core aeolian and magnetic records 5.1. H I R M
as a measure of aeolian content
The version of H I R M used here ( S I R M + IRM_3oo), essentially the same as Robinson's (1986) HIRM, measures the (mass specific) magnetic moment unreversed by application of a backfield of 300 m T to a sample previously saturated in a forward field. Whereas magnetite has essentially saturated at - 3 0 0 mT, A F M grains are capable of further significant magnetization. Ideally, S I R M + I R M _ 3 o o therefore measures the concentration of A F M components, even in samples which may be dominated by ferrimagnetic components. Concentration-dependent SIRM + IRM_ 3oo was calculated for cores SO-36-61, C1/86 6GC3 and
345
E39.75, for which the aeolian fraction was also isolated. Overall, S I R M + I R M _ 3 o o appears to be a good proxy for the aeolian content (Fig. 1 l-I). There is very close correlation between even small peaks in aeolian content and the S I R M + I R M _ 3 o o record in SO-36-61 and CI/86 6GC3, below approximately 50 cm (Fig. 11-Ia,b), and in E39.75 below about 2 m depth (Fig. 11-Ic). Above 2 m depth in E39.75, and near 3.5 m there is an unclear relationship, even while the values of SIRM + IRM_ 300 agree with the overall increase in aeolian content above 5 m depth. In a bivariate plot of the two parameters (Fig. 12c) this is revealed clearly as a broad scatter of points in the range above 20% aeolian content but a linear relationship for points with low aeolian concentration. The distributions in the other cores (Fig. 12a,b) is also clearly linear, in contrast to the log-linear distribution found by Bloemendal et al. (1993). The distributions for the two shallow cores coincide (dashed line in Fig. 12a,b) but differ from the distribution observed in E39.75 (dashed line extrapolated from low concentration values). This may be analogous to the different distributions found between the three diagenetic zones in the Indian Ocean cores (Bloemendal et al., 1993). SIRM + I R M - 3 o o underestimates aeolian content in the tephra layer at 260cm in SO-36-61. However, since the 'aeolian' fraction includes the volcanic component, this underestimation is welcome. The deviations from the linear aeolianSIRM + IRM-3oo relationship are not random and have their own clear distribution within the cores. 'Excess' SIRM + I R M _ 3 o o was calculated for each sample as the difference between aeolian content and SIRM+IRM_3oo, where values of each parameter were rescaled so that each of the dashed lines in Fig. 12a-c represented equivalence. Because SO-36-61 and C1/86 6GC3 have different distributions to E39.75, the relative 'excess' values are not quantitatively comparable. Excess SIRM +IRM_3o o is concentrated in the core tops
Fig. 11. (I) Aeolian concentration and SIRM + IRM 300 for the three northern cores. Values of aeolian content in E39.75 adjusted to correct for pretreatment effects ( Hesse, 1994b). ( I I ) Excess SIRM + IRM _ 30o (see text for description of method ). (III ) Magnetic susceptibility corrected for carbonate 'dilution' according to the formula {x/( 1 --carbonate content)}. Dashed time-lines as for Fig. 9.
P.P. Hesse / Palaeogeography, Palaeoelimatology, Palaeoeeology 131 (1997) 327 353
346
(a) SO-36-61
(b) C1/86 6GC3
0.15
(c) E39.75 PC 0.30
0.15 o
.f
0.25
,/.. '""
~O.lO ~." o
0,15,d a
+
0.05"
o
0.05 • ./-
'
=5
0,10
o
c/1
0,05
/*
0.00
0.00
' ' ' ' 1 ' ' ' ' | ' ' ' ' 1 ' ' ' '
0
5
10
15
20
....
I .... 5
0
Aeolian (%)
I .... 10
80-
0
o
1O0 oo
40
'
i
'
40
I
50
'
60
oo )7~ oG°° o~oo
50
o
2O
o~
o
I
I
I
I
I
I
-0.20
0.00
0.20
-0.20
0.00
0.20
excess SIRM+IRM_300
I
30
o o o°
o
¢,q
o o
'
%
o
S
I
20
(f) E39.75 PC
60
ch Oo
'
150
80
60-
I
10
Aeolian % (adj)
100
o
wZ 4 0 -
20
(e) C 1/86 6GC3
(d) SO-36-61
o ~ o o
0.00
I .... 15
Aeolian (%)
100
°"
0.20'
ev
o
m..'
o n
c
~ O.lO.
.,""
+
20
°
O
excess SIRM+IRM_300
0
I
|
- 0 . 5 0 - 0 . 2 5 0.00
I
0.25
i
0.5(} 0.75
excess SIRM+IRM_300
Fig. 12. (a) (c) SIRM + IRM 300 (10 3 Amakg - 1) as a function of aeolian concentration for the three northern cores. Dashed lines drawn by eye. ( d ) - ( f ) Carbonate-free susceptibility (10 8 m3kg ~) against excess S I R M + I R M 30o for the same cores. Excess SIRM + IRM-300 is calculated as the difference between SIRM + I R M 300 and aeolian concentration using the dashed lines in (a) (c) as ideal distributions, where excess SIRM + IRM 300=0 (see text)•
of SO-36-61 and C1/86 6GC3, as noted above, while there is a negative excursion in the tephra band at 260cm (Fig. ll-IIa,b). In E39.75 there are several intervals of excess SIRM + I R M _ 3 o o, decreasing in magnitude with depth, in the top four metres (Fig. 11-lie). This distribution is similar to that of frequency-dependent susceptibility (Fig. 9-llIc) and 'carbonate-free' susceptibility (Figs. 11-lIIc and 12f) (in effect the susceptibility not accounted for by the variable concentration of calcium carbonate in the core). It is possible that a distinct magnetic component accounts for these observations (i.e. a high susceptibility, highly frequency dependent component with harder IRM also contributing to SIRM +IRM-3oo). Even samples with IRM 3oo/SIRM approaching - 1 . 0 (Fig. 4a) retain a small amount of unreversed
remanence (
