Geochemistry, Geophysics, Geosystems

Diagenetic sensitivity of paleoenvironmental proxies: A rock magnetic study of Australian continental margin sediments

Authors


Abstract

A rock magnetic study of the upper 40 m of the Ocean Drilling Program Site 133-820A recovered at the outer edge of the northeastern Australian continental margin shows that downcore variations in magnetic parameters are diagenetically driven and correlate with the changes in global sea level. We identified intervals enriched in single-domain (SD) magnetite in the studied section. Unlike previous studies that postulated a detrital source, we show that the bulk of the SD fraction is biologically produced and is likely to be authigenic. The abundance of SD magnetite thus cannot be used as an indicator for provenance or sediment transport mechanisms. The biogenic magnetite was preserved during the high sedimentation rate periods, likely due to a short residence time in the corrosive zone of active iron reduction. The presence of the biogenic magnetite thus can be used as an indicator for a low degree of reductive dissolution. More advanced dissolution during the periods of slow sedimentation, coincident with sea level highstands, resulted in significant changes in the composition of the detrital assemblage, particularly in relative abundances of different mineral phases. The overall stability of the detrital magnetic minerals toward dissolution varies in the studied section as hematite > magnetite > goethite. Due to the higher stability of hematite, reductive diagenesis in general will lead to the lowering of the goethite/hematite (G/H) ratio, which can be mistaken for an increased aridity in the sedimentary source areas in the conventional interpretation for this proxy. The sensitivity of the G/H proxy to diagenesis should be taken into account in paleoenvironmental studies.

1. Introduction

Marine sediments with their good biostratigraphic age control and nearly continuous sedimentation are of considerable interest in environmental magnetism studies. The variations in composition and grain size distribution of detrital ferrimagnetic minerals may contain valuable information on climatically driven changes in provenance, sediment transport processes, or the changing climatic conditions in the source area of the sediments [e.g., Verosub and Roberts, 1995, Evans and Heller, 2003]. An example of such an environmental parameter that recently came into common use is the relative abundance of two pedogenic iron oxides goethite (G) and hematite (H). Goethite versus hematite content in the detrital fraction of marine sediments is routinely interpreted as to reflect the precipitation regime in the source area of the sediments; high G/H values are thought to indicate higher precipitation, whereas low values imply drier/warmer conditions [e.g., Harris and Mix, 1999, 2002; Clift, 2006; Zhang et al., 2007].

Diagenetic alteration due to oxidation [Henshaw and Merrill, 1980] and dissolution [Karlin and Levi, 1983] of the primary iron oxides or due to precipitation of new authigenic phases [Colley et al., 1984; Canfield and Berner, 1987; Jarvis and Higgs, 1987; Karlin et al., 1987] can compromise the environmental magnetic record by modifying the initial detrital assemblage of magnetic minerals. Reductive dissolution of magnetite in suboxic and anoxic environments, with preferential depletion of finer grains leading to an increase in the mean grain size, is a well-known process routinely reported in studies of marine sediments [Karlin and Levi, 1983, 1985; Canfield and Berner, 1987; Channell and Hawthorne, 1990; Karlin, 1990a, 1990b; Leslie et al., 1990a, 1990b; Bloemendal et al., 1992]. The effects of diagenesis on goethite and hematite are much less understood. Generally, goethite and hematite (undifferentiated) are considered to be more resistant to diagenesis; the onset of dissolution of the high-coercivity minerals was found to occur at greater depth than that of magnetite [e.g., Yamazaki et al., 2003; Emiroglu et al., 2004; Rey et al., 2005; Kawamura et al., 2007]. The relative stabilities of goethite and hematite toward reductive dissolution are especially poorly defined. Although synthetic goethite and hematite have similar thermodynamic stabilities, their dissolution rates relative to each other and to magnetite can vary significantly depending simply on whether the rates are calculated in terms of surface area (goethite < hematite < magnetite), or in terms of mass (hematite < magnetite < goethite) [e.g., Cornell and Schwertmann, 2003; Poulton et al., 2004]. It is unclear how applicable these stability schemes are to studies of natural sediments as magnetic properties of mineral grains depend on both their volume and shape.

An authigenic phase of the highest significance for paleomagnetism is arguably biogenic magnetite (magnetosomes) produced by magnetotactic bacteria. Magnetosomes are chemically pure crystals with sizes (35–120 nm) that fall within the stable single-domain size range [Bazylinski and Moskowitz, 1997; Cornell and Schwertmann, 2003]. The single-domain (SD) size is the most efficient for recording natural remanent magnetization [e.g., Dunlop and Özdemir, 1997]; in some sediments biogenic magnetite can be the dominant carrier of the stable natural remanence [e.g., Abrajevitch and Kodama, 2009].

The presence of fossil magnetosomes (magnetofossils) in the sediments could be a paleoecological indicator on its own. The abundance and morphology of fossil magnetosomes have been shown to correlate with climatically driven variations in organic carbon content and redox condition of sediments; a decrease in total magnetofossil abundance during the glacial and an increase during the interglacial periods have been reported by several studies [e.g., Hesse 1994a; Lean and McCave, 1998; Yamazaki and Kawahata, 1998; Dinarès-Turell et al., 2003].

However, the small size of biogenic magnetite crystals makes them particularly vulnerable to dissolution; exposed to the reducing environment biogenic grains dissolve quickly because of their large surface to volume ratio. In a steady state system of coupled sediment accumulation and reductive diagenesis, fossil magnetosomes are not likely to be preserved [e.g., Vali and Kirschvink, 1989; Snowball, 1994; Tarduno and Wilkison, 1996; Robinson et al., 2000]. Consequently, the intervals of enhanced bacterial magnetite concentrations found in sedimentary sequences may record conditions favorable to magnetofossils' preservation rather than conditions favorable to the bacterial growth.

To better understand what controls the distribution of magnetofossils and the effects of diagenesis on the high-coercivity minerals we conducted a rock magnetic study of a carbonate-siliciclastic sedimentary sequence recovered in Leg 133, Hole 820. Previous rock magnetic and electron microscopy studies found evidence of bacterial magnetite in the Hole 820 sediments [Barton et al., 1993a], but the importance of the biogenic contribution to the natural magnetization and its environmental significance have not been adequately addressed. Downcore variations in the content of the high-coercivity fraction were also detected, but the relative contributions of hematite and goethite have not been studied.

2. Description

Leg 133 Site 820 is one of the set of closely spaced holes drilled at the outer edge of the northeastern Australian continental margin in September 1990 [Shipboard Scientific Party, 1991]. It is located on the upper slope of the margin, approximately 10 km from the outer edge of the Great Barrier Reef and 50 km from the Australian shoreline (Figure 1). Hole 820A was drilled in 278.0 m of water and had a total recovery of 145.8 m. A hydraulic piston corer was used for first 15 cores and penetrated from 0 to 140.2 mbsf (102.5% recovery).

Figure 1.

Location map showing the main bathymetric features of the northeastern Australian continental margin and the location of ODP Site 820.

The recovered sequence is characterized by alternation between terrigenous and carbonate-rich lithologies with a well-developed cyclicity within the sediments [Davies et al., 1991] (Figures 2a, 2b, and 2c). The variations in sedimentation pattern were attributed to rapid changes in global sea level driven by climatic oscillations [McKenzie et al., 1993]. The top 40 m of the core contains an extremely high resolution section of carbonate slope sediments and represents the time span of reasonably well-understood oscillations in sea level during the past 140,000 years [Peerdeman and Davies, 1993].

Figure 2.

Downhole variations in (a) oxygen isotopes, (b) grain size, and (c) terrigenous fraction content of the sediments are thought to be controlled by glacioeustatic sea level fluctuations [Peerdeman and Davies, 1993; Peerdeman et al., 1993]. Shipboard measurements of rock magnetic parameters: (d) susceptibility and (e) natural remanent magnetization (NRM) [from Barton et al., 1993a]. (f) Variations in the saturation isothermal remanent magnetization (total SIRM) in our study closely follow that of the shipboard NRM measurements. As both of these parameters are controlled by quantity and mineralogy of magnetic minerals, the good correlation between them suggests that the ferrimagnetic assemblage was not altered during core storage. The variations in the susceptibility and NRM relate to fluctuation of sea level. Two prominent peaks at about 7 (peak 1) and 32 (peak 2) meters below surface (mbsf) correspond to marine transgressions [Barton et al., 1993a, 1993b]; the sharp susceptibility maxima occur at the start of the transgressions. The susceptibility peaks are superimposed on a weakly magnetic background representative of the sea level highstand sedimentation. Surface zone sediments (0–3 mbsf) accumulated during the Holocene highstand [Peerdeman and Davies, 1993] shows an NRM peak in the upper 2 m followed by a drop to “background” values. The high NRM values of the surface zone are comparable to that of peaks 1 and 2. However, unlike in peaks 1 and 2, the high NRM values do not have their counterpart in susceptibility.

Although pore water geochemical data, especially strong gradients in sulfate near the top of the core suggest strong reducing conditions, total sulfur concentrations in the sediments were below detection limits and no sulfides were found in the upper 40 m of the sedimentary column [Shipboard Scientific Party, 1991]. Therefore, diagenesis probably did not reach the stage of authigenic iron sulfide production in this depth interval. The relatively low degree of diagenetic alteration (absence of authigenic sulfides) is probably due to low (<0.4%) total organic carbon content of the sediments.

Paleomagnetic analyses of sediment samples from Site 820 showed significant downcore variations in magnetic parameters [Barton et al., 1993a]. In the upper 40 m of the core, two prominent susceptibility peaks at about 7 mbsf (peak 1) and 32 mbsf (peak 2) are superimposed on a background of smaller oscillations (Figure 2d). The NRM intensity record (Figure 2e), similar to the susceptibility record, is characterized by relatively high NRM values at about 7 and 32 mbsf superimposed on a relatively uniform, weakly magnetic background. In addition, there is a high NRM intensity zone in the uppermost 2 m (surface zone) that does not have a susceptibility counterpart.

The anhysteretic remanent magnetization measurements showed high values, indicating a high content of a single-domain fraction, in the uppermost 1.5 m and in both high-susceptibility zones. Transmission electron microscopy (TEM) examinations of magnetic extracts from four samples (0.77, 5.07, 6,17 and 10.76 mbsf) found euhedral grains of SD size magnetite arranged in chains, characteristic of bacterial magnetosomes [see Barton et al., 1993a, Figure 20]. High abundance of bacterial magnetite was found in samples from the surface and peak 1 high-magnetization zones.

The difference in isothermal remanent magnetizations (IRM) acquired at 300 mT and 500 mT, was used to estimate the volume ratio of hematite to magnetite in the sediments [Barton et al., 1993a]. Volumetrically significant amounts of the high-coercivity phase were found throughout most of the sequence; however, the surface zone and peaks 1 and 2 each showed predominance of magnetite.

The downhole changes in magnetic properties were interpreted in the context of glacioeustatic sea level fluctuations [i.e., Barton et al., 1993a, 1993b; Peerdeman and Davies, 1993]. The sharp susceptibility maxima correlate with the start of marine transgressions, following lowstands in sea level (high δ18O, glacial maxima). The early transgression sediments are characterized by a stable SD remanence, with a significant contribution from ultrafine, superparamagnetic grains. During the later marine transgression, the susceptibility gradually returns to low values and the remanence is carried by single-domain magnetite. The highstand “background” sediments are characterized by multidomain (MD) magnetic assemblage.

Fluctuations in susceptibility and remanence within the “background zone” were found to be controlled predominantly by variations in the concentration, rather than in the composition of ferrimagnetics [Barton et al., 1993a]. However, the difference in the NRM intensity between the background and high-magnetization zones could not have been explained by carbonate dilution only; a change in the terrigenous fraction assemblage thus was proposed. As the prevailing southeasterly wind pattern at the location of Site 820 precludes a large influx of wind-borne material from the continental interior [Thiede, 1979; Hesse, 1994b], the following complicated scenario for variations in the type of magnetic particles that reached the slope was proposed by Barton et al. [1993b].

At low sea level, fine-grained terrigenous sediments were trapped in river estuaries and deltas on the exposed shelf. At the same time bacterial magnetite accumulated in inter-reefal lagoons and swamps. At the beginning of the early transgression, these reservoirs were tapped. The initial flush of superfine products accounts for the sharp susceptibility peaks with high SD and SP contents. As transgression proceeded, more distal fluviodeltaic reservoirs also were accessed, contributing predominantly SD grains. As interglacial highstand condition were established, the reefal carbonate communities flourished and carbonate contents of the sediments rose, diluting the terrigenous influx and lowering susceptibility. Average particle size increased as a consequence of more vigorous offshore transport processes (shelf currents, storms intensified boundary currents). The establishment of the MD background properties was probably enhanced by gradual dissolution of very fine magnetic grains.

The authors admitted that the evidence to sustain this hypothesis was limited and certain counterarguments can be made. The hypothesis, for example, does not account for markedly different magnetic properties of the sediments from the surface zone that accumulated during the Holocene highstand and the background sediments, also accumulated during the sea level highstands. In order to reevaluate the covariation of the biogenic and high-coercivity magnetic fractions with the postulated sea level changes, we restudied the upper 40 m of the Hole 820A, for which the high-resolution sedimentological and paleomagnetic data are available, using recently developed rock magnetic techniques.

3. Methods

Samples for this study were collected from the working half of the ODP section 133-820A. The upper 7 m of the section was sampled in ∼20 cm intervals; the interval from 7 to 40 mbsf was sampled in 50 cm intervals. A sample in the present study consists of ∼1.5 cm3 volume of sediment. Rock magnetic measurements were performed in the Paleomagnetism Laboratory at the Center for Advanced Marine Core Research, Kochi University (Japan).

Hysteresis and remanence cycles at room temperature up to maximum field of 1.0 T were measured with a Princeton Alternating Gradient Magnetometer (AGM) MicroMag2900 model (noise level < 10−9Am2). Saturation magnetization (Ms), saturation remanence (Mr), and coercivity (Bc) were obtained after subtracting the paramagnetic contribution. Remanence coercivity (Bcr) was obtained by demagnetizing the SIRM in a stepwise increasing back field.

First-order reversal curves (FORC) were measured with a VSM and processed using an algorithm of Harrison and Feinberg [2008]. The FORCs were measured with a field spacing of 0.38 mT, averaging time of 300 ms, and a 1T saturating field.

Thermomagnetic curves were acquired with a Curie balance (Natsuhara Giken); operational field 0.5 T. The samples were progressively heated in air to 700°C at a rate of 10°C per minute, and then cooled to room temperature at the same rate.

Isothermal remanent magnetization (IRM) were imparted stepwise up to a maximum peak field of ∼7 T with a MMPM-10 pulse magnetizer (Magnetic Measurements LTD) and measured with a 2G SQUID magnetometer with a noise level of less than 10−7 Am2. Effectively, thirty to thirty five field steps covering interval 2.8∼7000 mT were used for the statistical analysis of the IRM acquisition curves limited to symmetric distributions in log-space [Kruiver et al., 2001; Heslop et al., 2002]. The magnetic components are characterized by the saturation isothermal remanent magnetization (SIRM), a parameter that is proportional to the content of the mineral in a sample, the peak field (B1/2) at which half of the SIRM is reached, and the dispersion (DP) of its corresponding cumulative lognormal distribution [Kruiver et al., 2001].

The combined frequency (1, 10, 100 and 1000 Hz in fields of 0.3 mT) and low-temperature dependence of susceptibility between 300 and 10 K, as well as zero field cooled (ZFC) and field cooled (FC) remanence were measured for representative samples with a Quantum Design SQUID magnetometer (MPMS, sensitivity of ∼10−7 Am2). ZFC remanence was obtained by initial cooling from 300 K to 20 K in zero field prior to the application of a 2.5 T field at 20 K. FC remanence was obtained by cooling from 300 K to 20 K in a high magnetic field of 2.5 T.

4. Results

4.1. Temperature Dependence of Magnetization

Examples of thermomagnetic measurements in air for representative samples are shown in Figures 3a–3c. The thermomagnetic runs are irreversible for all samples; the cooling branches differ from the heating ones indicating that magnetic phases were modified by heating.

Figure 3.

(a-c) Examples of thermomagnetic runs for representative samples. An inflection or a small peak increase in magnetization between 100°C and 200°C is likely produced by dehydration of clays or poorly crystalline iron (oxyhydr)oxides. Samples from the upper part of the surface zone show a single stepwise decrease in saturation magnetization at ∼560°C–580°C, characteristic of (titano)magnetite or maghemite (Figure 3a). The majority of the samples show a peak increase in magnetization between 400°C and 600°C indicating that mineralogical changes are taking place in this temperature interval (Figures 3b and 3c). The content of the phases undergoing alteration can be estimated based on the size of the alteration peak, approximated by a triangle with the base of 2 × (Tmax−To) and the height of (SMmax−SMo) (Figure 3c). Downcore variations in (d) the estimated peak's area correlate better with (e) the total content of terrigenous fraction of the sediments rather than with (f) the ferromagnetic content-sensitive parameters like magnetic susceptibility.

An inflection or a small peak increase in magnetization between 100°C and 200°C is present in the majority of the samples (Figures 3a–3c); it is likely produced by dehydration of clays or poorly crystalline iron (oxyhydr)oxides, such as ferrihydrite or goethite [Cornell and Schwertmann, 2003]. The surface layer samples (depths up to 1.5 mbsf) show a single stepwise decrease in saturation magnetization at ∼560°C–580°C superimposed on a strong paramagnetic background (Figure 3a); this temperature range is characteristic of (titano)magnetite or maghemite.

The majority of the samples from the deeper levels also show a peak at 400°C–600°C on the heating branch (Figures 3b and 3c) indicating that some mineralogical changes are taking place in this temperature interval. A peak increase in magnetization can sometimes be used to identify ferrimagnetic phases involved in the reaction. The transformations of iron sulfides to magnetite [Roberts, 1995] and siderite to maghemite [Chadima et al., 2006] occur at temperatures above 300°C. These reactions should produce phases that are more magnetic than their parent phase, which normally leads to an increase in saturation magnetization at room temperature. In our samples, the saturation magnetization values at room temperature before and after the runs do not differ considerably, suggesting that the new magnetic phases formed at 500°C have been subsequently oxidized at higher temperatures. The thermal instability is particularly characteristic of maghemite (the product of siderite transformation) which converts to weakly magnetic hematite on heating.

The size of the high-temperature alteration peak controlled by the amount of a reagent available for chemical reactions seems to vary downcore. The shape of the peak can be roughly approximated by a triangle with an estimated area of (Tmax − To) × (SMmax − SMo), where Tmax and SMmax are the temperature and saturation magnetization at the peak's maximum, and To and SMo are the temperature and saturation magnetization at the initial rise (Figure 3c). The variations in the estimated peak's area show a much better correlation with the total content of the noncarbonate fraction, rather than with the parameters sensitive to the ferrimagnetic fraction content, such as susceptibility or NRM (Figures 3d–3f). Such relationships suggest that the high-temperature peak likely reflects alteration of terrigenous input-related minerals, either aluminosilicates, or diagenetic (Fe,Mn) carbonates, and thus could not be used for the identification of the NRM carriers.

4.2. IRM Acquisition

Typical examples of IRM acquisition plots for the studied samples are shown as gradient curves in Figure 4. Five distinctive magnetic components with variable contributions to the total IRM were identified by a statistical analysis [Kruiver et al., 2001; Heslop et al., 2002]. The low-coercivity component, with mean coercivity B1/2 of ∼12–25 mT and a wide dispersion DP of 0.36∼0.46 is present in all samples. An intermediate coercivity component (B1/2∼250–500 mT and DP ∼0.27–0.35) and a high-coercivity component (B1/2∼1.6–2.5 T; DP∼0.3–0.35) are ubiquitous as well. Based on their characteristic coercivities, the intermediate- and high-coercivity components can be identified as hematite and goethite, respectively [e.g., Heller, 1978; Rochette and Fillion, 1989; France and Oldfield, 2000; Kruiver et al., 2001; Abrajevitch et al., 2009]. The component with the lowest coercivity likely represents a mixture of titanomagnetite and maghemite. The wide dispersion of this component, which indicates large variations in grain size and stoichiometry of the ferrimagnetic phase, is characteristic of a detrital source [e.g., Egli, 2004], and/or of pedogenic ultrafine magnetite/magnehemite grains that precipitate inorganically [Maher, 1988] or by the action of dissimilatory iron reducing bacteria [Moskowitz et al., 1989]. Although the possibility of some contribution from ultrafine authigenic magnetite/maghemite grains cannot be excluded, for simplicity, this component will be referred to as “detrital magnetite” in the following discussion. The term “detrital” is used here in a general sense of transported and redeposited.

Figure 4.

IRM (gradient) acquisition curves for representative samples of (a) the upper part of the surface zone, (b) the lower part of the surface zone, (c) peak 1, (e) peak 2, and (d, f) the background zone. Five distinctive magnetic components with variable contributions to the total IRM were identified by a statistical analysis [Kruiver et al., 2001; Heslop et al., 2002]. The low-coercivity (log B1/2∼1.3) component with a wide distribution is interpreted as corresponding to detrital magnetite/maghemite. An intermediate coercivity component (log B1/2∼2.3) and a high-coercivity component (log B1/2∼3.2) were identified as hematite and goethite, respectively. Distinctive narrow peaks (gray highlights) with log B1/2∼1.6 and ∼1.9 are interpreted as corresponding to biogenic magnetite.

An additional component in the low-coercivity range with logB1/2 of 1.55∼1.65 (B1/2 ∼32–44 mT) and a relatively narrow dispersion of 0.2∼0.24 is present in the samples from all high NRM zones (Figures 4a, 4c, and 4e). A distinctive component with logB1/2 of 1.9–1.98 (B1/2 of 79–97 mT) and a very narrow distribution (DP∼0.14) is mostly confined to the surface high-magnetization zone (Figure 4a). The narrow shapes of these two peaks indicate that the corresponding populations of magnetic grains have a restricted coercivity range, which is a sign of uniform grain sizes and/or composition. Such uniform assemblages are usually attributed to the biogenic production of magnetic grains (magnetosomes) by magnetotactic bacteria [e.g., Kruiver and Passier, 2001; Egli, 2004; Kawamura et al., 2007]. Different bacterial strains produce magnetosomes of different morphology (equant, prismatic, elongated, bullet-shaped, isolated, in chains, etc.,) [Moskowitz et al., 1988, 1993; Penninga et al., 1995; Hanzlik et al., 2002]. Since several bacterial strains occur simultaneously in natural sediments, variations in magnetosomes are known to result in coexistence of several distinct “biogenic” components [Egli, 2004].

Apart from the presence of the biogenic components in the high-magnetization zones, other IRM acquisition parameters show systematic variation with depth (Figure 5). The concentrations of the low-coercivity detrital fraction (Figure 5a), hematite (Figure 5b) and goethite (Figure 5c) (approximated by the SIRM values obtained by the IRM composition analysis) are low in the background zones and significantly higher in peaks 1 and 2. There is generally a proportional relationship between the content of the low-coercivity detrital fraction and the high-coercivity hematite and goethite components (Figures 6a and 6b), suggesting that these three components are likely coexisting in the original detrital assemblage. The slope of the correlation line, however, seems to differ between the background and high-magnetization zones (Figure 6b). The detrital low-coercivity component shows systematic compositional variations with depth as well; the mean coercivity and dispersion of this component are lower in the samples from the high-magnetization zones compared to the background intervals samples (Figures 5f and 5g).

Figure 5.

Downcore variations in the content (SIRM) of (a) detrital magnetite, (b) hematite, (c) goethite, and (d) total content of ferrimagnetic minerals as obtained by the IRM component analysis. The contribution of the detrital fraction (detrital magnetite +hematite + goethite) into the total SIRM is shown by gray fill; the contribution of biogenic magnetite is shown by black fill. Downcore variations in relative abundances of (e) goethite and hematite, (f) mean coercivity, and (g) dispersion parameter of the detrital magnetite component indicate systematic compositional differences between the high-magnetization zones (the surface zone and peaks 1 and 2) and background sediments.

Figure 6.

Covariations between the detrital magnetite fraction and other ferrimagnetic fractions identified by the coercivity analysis of the IRM acquisition curves: (a) versus goethite, (b) versus hematite, (c) versus biogenic magnetite, and (d) versus the G/H ratio. The goethite and hematite contents of the samples generally show a good correlation (R2 > 0.7) with the detrital magnetite content, suggesting that these components are likely to be the parts of the detrital package. However, the high-magnetization zones and the background sediments show different relationships between the components (Figure 6b). The content of the biogenic magnetite in the samples does not show any correlation with the detrital ferrimagnetic fraction (Figure 6c). The relationship between the detrital magnetite, goethite, and hematite of the sediments (Figure 6d) indicate that although the surface zone sediments have generally low average content of the detrital ferrimagnetic fraction, comparable to that of the background sediments, the composition of the surface zone detrital assemblage is close to that of the high-magnetization zones, which is particularly evident in the high G/H values.

The relative contributions of goethite and hematite into the total SIRM (shown as G/H ratio in Figure 5e) also vary. The ratio is higher in the high-magnetization zones, and low in the background interval. The surface zone of sediments has a relatively low content of the detrital low-coercivity component (consistent with the generally lower terrigenous fraction content) but have the high G/H ratios, comparable to those of peaks 1 and 2 (Figure 6d).

4.3. Hysteresis Parameters

All of the hysteresis loops have “normal” shapes, with no evidence of wasp-waisted or pot-bellied contours [Tauxe et al., 1996] (Figures 7a–7f). There are systematic variations in hysteresis parameters with depth (Figures 7g–7j). Saturation magnetization Ms, a concentration-dependent parameter, is high in peaks 1 and 2, but close to the background values in the surface high-magnetization zone (Figure 7g). Coercivity is higher in the upper parts of all high-magnetization zones (Bc > 20mT, Bcr > 40 mT), but gradually decrease toward the base (Figures 7h and 7i). Zones of high coercivity correspond to the zones of high values of Mr/Ms (Figure 7j). Prominent minima in Bcr mark the bases of peaks 1 and 2 (Figure 7h). As Bcr is sensitive only to remanence and thus insensitive to the superparamagnetic (SP) grains, the drop in Bcr indicates the coarsening of the median grain size. The change in overall shape of hysteresis loops toward the base of high-magnetization zones, the narrowing of the loops with lowering of Bc and Mr/Ms values (Figures 7c–7f), suggest some addition of SP fraction as well.

Figure 7.

Representative examples of the hysteresis curves (corrected for paramagnetic contributions) for (a) surface zone, (b) background, and (c-f) high-magnetization interval samples. Downcore variations in (g) saturation magnetization Ms, (h) coercivity of remanence Bcr, (i) coercivity Bc, and (j) Mr/Ms. High values of the saturation magnetization in peak 1 and 2 zones indicate high concentration of the ferrimagnetic minerals in these zones. High coercivity values Bcr and Bc and Mr/Ms ratio in the upper part of these zones indicate the high content of single-domain size (SD) magnetite. Systematic drop in values of these parameters toward the base of the high-magnetization zone indicate coarsening of the median grain size and increasing content of very fine, superparamagnetic (SP) fraction. The increase in the SP contribution is evident in the “narrowing” of the hysteresis loops toward the base of the peak (c = > d, e = > f).

Compared to the theoretical hysteresis trends for SD-SP and SD-MD mixtures [Dunlop, 2002a, 2002b] on a Day plot [Day et al., 1977] the samples plot generally parallel to the SD-MD mixing model curve, but displaced toward the SP-SD mixing curve (Figure 8). Multidomain grains dominate in the background interval (>60% of MD), while SD gains are abundant in the high-magnetization zones (>60% of SD grains). The SD contribution in the high-magnetization zone estimated by the comparison with the theoretical trends agrees well with the relative contribution of the “biogenic” components into the total SIRM (60–70%) indicating that the major part of the SD magnetite is biologically produced.

Figure 8.

Hysteresis parameters, Mr/Ms and Bcr/Bc plotted on a Day et al. [1977] plot with the theoretical mixing curves expected for an admixture of MD, SD, and SP magnetite [Dunlop 2002a, 2002b]. The curves 1 and 2 are given for different median sizes of the MD grains. The percentages along the model curves refer to the volume fraction of the SP or MD component. Background interval samples are dominated by MD grains (>60% of MD), while SD gains are abundant in the high-magnetization zones (>60% of SD grains).

4.4. FORC Diagrams

High-resolution FORC diagrams for representative samples from the high-magnetization zones show a narrow ridge with a wide maximum between 15 and 60 mT along the Bc axis and a very narrow distribution along the Bu axis (Figure 9). The narrow ridge is a unique signature of noninteracting single-domain particles [Egli et al., 2010]; such a feature is characteristic of magnetosomes [Egli et al., 2010; Li et al., 2010]. The central ridge's maximum of ∼35 mT is comparable with values for biogenic magnetite in modern lake sediments [e.g., Egli, 2004, Egli et al., 2010]. The ridge is absent in the FORC diagrams of the background interval samples (not pictured).

Figure 9.

High-resolution FORC diagrams for representative samples of (a) the high-magnetization surface zone, (b) peak 1, and (c) peak 2 showing a narrow peak along the Bu axis, a feature typical of magnetosomes [Egli et al., 2010]. FORC diagrams are plotted using an algorithm of Harrison and Feinberg [2008] with a smoothing factor of 7. Measurements are not normalized by mass.

4.5. Low-Temperature Remanence and Susceptibility

Examples of zero field cooled (ZFC) and field cooled (FC) warming curves are shown in Figures 10a–10c. No distinctive remanence transitions were observed although many samples show a slight kink at ∼100–120 K, possibly an indication of a suppressed Verwey transition. The ZFC and FC curves of the samples from the background zone are almost identical and show the continuous loss of remanence on warming, typical of poorly ordered materials (Figure 10a). In-phase susceptibility of the background samples is essentially independent of frequency (Figure 10d) demonstrating the absence of SP grains.

Figure 10.

(a, b, c) Low-temperature saturation remanence curves after samples were cooled in a high field (FC) and a zero field cooled (ZFC) and (d, e, f) frequency dependence of in-phase susceptibility for representative samples of the background (Figures 10a and 10d) and high-magnetization (Figures 10b, 10c, 10e, and 10f) zones. In the background interval samples, ZFC and FC curves are almost identical and show the continuous loss of remanence on warming (Figure 10a), a feature typical of poorly ordered materials. In-phase susceptibility of the background samples is essentially independent of frequency (Figure 10d) indicating the absence of SP grains. In the high-magnetization zones, ZFC and FC curves diverge at ∼250–300 K (Figures 10b and 10c) indicating the presence of a ferrimagnetic fraction with the range of blocking temperatures that are below room temperature. The frequency dependence of susceptibility at temperatures below 300 K (Figures 10e and 10f) indicates the presence of SP grains. Samples from the base of the high-magnetization zone (Figure 10f) show a significantly higher frequency dependence (higher SP content) compared to the samples from the upper part of the zone (Figure 10e).

The samples from the high-magnetization zones (Figures 10b and 10c) show bifurcated ZFC-FC curves that merge at about 250–300 K indicating the presence of a ferrimagnetic fraction with the range of blocking temperatures that are below room temperature. The frequency dependence of susceptibility at temperatures below 300 K (Figures 10d–10f) indicates the presence of SP fraction. There is a general correlation between the magnitude of separation between the ZFC and FC curves and the frequency dependence (estimated as (χ1 Hzχ997Hz)/χ1Hz)). The highest frequency dependence up to 25% is found at the base of the zones 1 and 2 (Figure 10f); in the samples from the surface zone and the upper parts of the zones 1 and 2, the frequency dependence does not exceed 10% (Figure 10e).

5. Discussion

5.1. Biogenic Magnetite

A combination of rock magnetic parameters and TEM identification of magnetosomes [Barton et al., 1993a] strongly suggests that biogenic magnetite is present in all high-magnetization zones. Hysteresis parameters indicate a high concentration of SD particles in the surface high-magnetization zone and peaks 1 and 2, which is consistent with, although not diagnostic of magnetofossils's presence. The IRM components with the narrow distributions indicative of uniform grain size and stoichiometry of ferrimagnetic particles [Egli, 2004] and the noninteracting character of the particles indicated by the high-resolution FORC diagrams [Egli et al., 2010] are diagnostic of magnetosomes. The absence of the Verwey transition in the low-temperature experiments, and thus the failure of Moskowitz et al. [1993] test, is likely due to partial oxidation of magnetosomes that suppressed the Verwey transition [Housen and Moskowitz, 2006].

The relative contribution of the biogenic magnetite-carried remanence into the total NRM of the sediments can be roughly estimated using IRM as a proxy for NRM [e.g., Dunlop and Özdemir, 1997]. The estimated contributions of biogenic components into the total SIRM exceed 70% in the surface high-magnetization zone, and range from 40 to 60% in peaks 1 and 2 (Figure 5d). Biogenic magnetite is likely to be mainly responsible for a usable paleomagnetic signal in the studied section; the stable primary magnetization has been isolated only in the intervals containing biogenic magnetite whereas a viscous magnetization dominated in the background sediments [Barton et al., 1993a].

Previous studies made note of the substantial contribution of SD particles to the NRM. To explain the down core variations in the SD content, Barton et al. [1993b] proposed a complex hypothesis based on the idea that SD magnetite was transported from detrital sources on the continent and deposited downslope (see section 2 for a brief summary). Our study, however, shows that the content of SD (biogenic) magnetite does not correlate with any other detrital fraction (Figure 6c). More likely, biogenic magnetite is produced in situ.

5.2. The Difference Between the Background and High-Magnetization Zone Sediments

There seems to be distinct compositional differences between the background zone sediments and the high-magnetization zones. In the background zone, there is a strong correlation between the amount of detrital magnetite and hematite, and a significantly lower G/H ratio compared to the high-magnetization zones (Figures 5 and 6). In the high-magnetization zones, there is strong correlation between magnetite and hematite as well, but with a shallower slope (Figure 6b).

In the previous studies [Barton et al., 1993a, 1993b], the downcore variations in the relative contribution from the high-coercivity minerals were interpreted as to indicate the different sources of the detrital sediments. A higher supply of SD magnetite was postulated during the transgressions following the lowstands of the sea level, which differed markedly from the times of the highstand and regression when the detrital fraction was enriched in MD magnetite and hematite.

However, the surface zone of sediments that was accumulated during the Holocene highstand has a magnetic mineral assemblage similar to that of the high-magnetization zones, especially apparent in its high goethite content (Figure 6d), rather than to the highstand background zone. If the magnetic assemblage of the surface magnetization zone is taken to represent the detrital (high sea level) sediment source, its marked difference from the background zones can be well explained by a diagenetic modification of the primary assemblage.

Overall, the surface zone of sediments follows the classical scenario for a steady state system of coupled sediment accumulation and reductive diagenesis [e.g., Snowball, 1994; Tarduno and Wilkison, 1996; Robinson et al., 2000]. Biogenic magnetite is produced in a narrow, well defined subsurface layer where suboxic redox conditions are optimal to the living bacteria [Petermann and Bleil, 1993]. As oxidation front migrates with the accumulation of the sediments, the zone of bacterial magnetite proliferation and the zone of active iron reduction migrate accordingly. A long-term survival potential of a ferrimagnetic grain depends on its residence time in the corrosive zone. Magnetofossils (single-domain magnetite) left behind in the iron reduction zone dissolve quickly because of their large surface to volume ratio. Reductive dissolution also affects detrital minerals; with the loss of biogenic magnetite indicating the advancing dissolution, there is an increase in the mean coercivity of the detrital magnetite component (log B1/2) from 1.27 to 1.35 and dispersion parameter from 0.36 to 0.45 (Figures 5f and 5g). Such trends are consistent with progressive dissolution and preferential removal of small grains resulting in overall coarsening of the assemblage. This coarsening trend is likely to continue with longer time spent in the corrosive zone, and might eventually reach the “background zone” characteristics with even higher log B1/2 of 1.38.

The ferrimagnetic assemblage of the high-magnetization zones 1 and 2 is similar to that of the subsurface sediments in their high biogenic magnetite content, the IRM acquisition parameters of the detrital magnetite fraction and the high G/H ratio (Figure 5), indicating that sediments of these zones have not reached an advanced stage of reductive dissolution. Increased sedimentation rates during the early transgressions [Peerdeman and Davies, 1993; Dunbar and Dickens, 2003] likely shorten the time spent by ferrimagnetic grains in the zone of active iron reduction. A partial dissolution of SD biogenic magnetite likely resulted in the production of the SP fraction. The further dissolution of the grains was likely arrested with the passage of the active reduction front, thus preserving the SP grains in the base of the peaks 1 and 2. During the late transgressions, when the sedimentation rates were the highest [Peerdeman and Davies, 1993; Dunbar and Dickens, 2003], the time spent in the active reduction zone was the shortest, leading to the preservation of the early assemblage, including the biogenic magnetite.

The long-term preservation of the biogenic magnetite in the studied sediments thus is enhanced during the high sedimentation rate periods. Although the change in sedimentation rate itself is driven by glacioeustatic sea level variations, the presence of magnetite is unlikely to have any climatic significance; rather it indicates a rapidly changing redox regime within the subsurface sediments.

5.3. Stability of Detrital Minerals Toward Reductive Dissolution

The overall stability of ferrimagnetic minerals toward reductive dissolution can be roughly estimated by comparing the least and the most affected sediments with similar initial ferrimagnetic assemblages. The grain size distribution and the average content of terrigenous fractions in 0–7 mbsf and 8.5–16.5 mbsf intervals are comparable (Figures 2b, 2c, and 11a). As a change in provenance of the terrigenous sediments is not likely at the sampling site [e.g., Thiede, 1979; Dunbar and Dickens, 2003], these two depth intervals likely have had generally similar initial detrital assemblages. The diagenetic dissolution of the detrital grains in the upper interval is minimal as evidenced by the preservation of the fine-grained biogenic magnetite. The sediments of the lower interval were subjected to more advanced dissolution. A comparison of the average contents of the detrital minerals in these two intervals indicates that the diagenetic processes resulted in the loss of ∼80% of detrital magnetite/maghemite, ∼65% of hematite and >90% of goethite (Figures 11b–11d). The preservation potential of these phases thus varies as hematite > magnetite > goethite.

Figure 11.

Variations in (a) the average content of the total terrigenous fraction, (b) detrital magnetite, (c) hematite, and (d) goethite between the sediments that were the least affected (0–7 mbsf interval, gray fill) and strongly affected by reductive dissolution (8.5–16.5 mbsf, black fill). The size of the column represents the mean value, while the error bars correspond to ±1 standard deviation of the corresponding parameter.

Due to the higher stability of hematite, reductive diagenesis in general will lead to the lowering of the G/H ratio. A diagenetically induced decrease in the G/H ratio can be mistaken for an increased aridity in the sedimentary source areas in the conventional interpretation of this proxy. The commonly held assumption that the G/H ratio in marine sediments is resistant to diagenetic changes is incorrect; the sensitivity of the G/H to diagenesis should be taken into account in the studies of marine sediments.

6. Conclusions

Our rock magnetic study of the upper 40 m of the ODP Site 133-820A recovered at the outer edge of the northeastern Australian continental margin showed downcore variations in magnetic parameters that are generally similar to the findings of Barton et al. [1993a]. We detected high concentrations of single-domain magnetite in the upper 1.5 m of the sedimentary column and in two distinct zones at ∼7 and 32 mbsf. The uniform size/composition of the SD particles indicated by the narrow dispersion of the corresponding IRM acquisition components, their noninteracting nature shown by the FORC analyses and TEM images of bacterial magnetosomes [Barton et al., 1993a] strongly suggest that SD magnetite is of biological origin. Contrary to Barton et al. [1993a, 1993b] who postulated a detrital source for SD magnetite, we found no correlation between the amount of the biogenic magnetite and the content of other detrital source phases. More likely, the biogenic magnetite was produced in situ during early diagenesis, and consequently, cannot be used as a proxy for provenance or sediment transport mechanisms.

The intervals enriched in the biogenic magnetite coincide with the high sedimentation rate periods. Although the change in sedimentation rate itself is climatically driven, the presence of the biogenic magnetite likely indicates the enhanced preservation due to a short residence time in the corrosive zone of active iron reduction, rather than bacterial proliferation.

During the sea level highstands characterized by a low sedimentation rate and thus longer time in the corrosive zone, the advanced dissolution led to a coarsening of the detrital magnetite/maghemite grains, an increase in the relative abundance of hematite, and a decrease in relative abundance of goethite. The overall resistance of the detrital magnetic minerals toward dissolution thus varies as hematite > magnetite > goethite.

Due to the higher stability of hematite, reductive diagenesis in general will lead to the lowering of the G/H ratio. The sensitivity of the G/H proxy to diagenesis should be taken into account in paleoenvironmental studies.

Acknowledgments

This research used samples and/or data provided by the Ocean Drilling Program (ODP). The ODP is sponsored by the U.S. National Science Foundation (NSF) and participating countries under management of Joint Oceanographic Instititions (JOI), Inc. This paper was improved as a result of constructive review comments from Ramon Egli, Christoph Geiss, and an anonymous reviewer. This is the contribution PM0111 of the Paleo/Rock Magnetism Laboratory, Kochi Core Center.