A biogenic origin for anomalous fine-grained magnetic material at the Paleocene-Eocene boundary at Wilson Lake, New Jersey



[1] The Paleocene-Eocene Thermal Maximum, which occurred ∼55.5 Ma, was caused by a massive release of carbon, as indicated by an ∼3‰ negative carbon isotope excursion recorded in the marine, atmospheric, and terrestrial reservoirs. One suggested source for the carbon, a cometary impactor, is based on the sudden appearance and high concentration of single-domain (SD) magnetite in Paleocene-Eocene (P-E) boundary cores from the North Atlantic continental margin. We evaluate the potential sources of SD magnetite at the P-E boundary by presenting new magnetic hysteresis, low-temperature magnetic remanence, and transmission electron microscopy data from the North Atlantic coastal ocean. Our results show a similar increase in SD material but demonstrate that the magnetic material has a biogenic origin. These findings indicate that the high concentrations of SD magnetite immediately above the P-E boundary are the result of unusual accumulations and/or preservation of magnetotactic bacteria. Such bacteria typically occupy the oxic-anoxic transition zone near the sediment-water interface or in the water column. The high abundances of SD magnetite in sediments from across the shelf may be an artifact of nonsteady state redox conditions and exceptional preservation of SD magnetite. It may also indicate that the oxic-anoxic redox boundary shifted into the water column. The latter explanation implies transient eutrophy of the coastal ocean in this region, most likely due to seasonally enhanced runoff, and increased stratification and nutrient loading.

1. Introduction

[2] The Paleocene-Eocene Thermal Maximum (PETM), which occurred approximately 55.5 Ma immediately following the Paleocene-Eocene (P-E) boundary, is one of the most abrupt and dramatic climate events of the Cenozoic. The event is characterized by abrupt warming in deep-sea and surface waters [Kennett and Stott, 1991; Zachos et al., 1993, 2006], an ∼3‰ negative carbon isotope excursion (CIE) recorded in the marine, atmospheric, and terrestrial reservoirs [Kennett and Stott, 1991; Koch et al., 1992], a carbonate dissolution horizon in deep-sea sediments [Zachos et al., 2005], enhanced continental weathering and runoff [Kelly et al., 2005; Ravizza et al., 2001; Schmitz and Pujalte, 2007; Villasante-Marcos et al., 2007], and extinctions and radiations of marine and terrestrial flora and fauna [Maas et al., 1995; Thomas and Shackleton, 1996]. The CIE and carbonate dissolution horizon are consistent with the release of a large mass of 12C-enriched carbon.

[3] Several carbon sources for the light carbon have been suggested, including volcanism [Bralower et al., 1997; Schmitz et al., 2004]. Another suspected and favored source is the dissociation of 1,400 to 2,800 Gt of gas clathrates and subsequent release of methane from the seafloor [Dickens et al., 1995]. Possible dissociation triggers include a thermal threshold [Dickens et al., 1995], orbital forcing [Lourens et al., 2005], changes in ocean circulation [Katz et al., 2001], and volcanism [Svensen et al., 2004]. One hypothesis posits a cometary impact not only supplied some of the light carbon, but also triggered the CIE and PETM [Kent et al., 2003].

[4] The idea for a bolide impact at the P-E boundary is based primarily on an abrupt and distinct change in magnetic grain size (from multidomain (MD) to single domain (SD) magnetite) across the CIE and initial PETM in outer shelf sediments from the New Jersey continental margin. High concentrations of SD magnetite in shelf sediments are rare and warrant explanation. Kent et al. [2003] argue that the fine-grained magnetic material condensed from an impact ejecta plume, an idea they bolster with Ir, Os, and 3He data from other P-E boundary sections. However, this explanation is inconsistent with recent geochemical studies on the most complete P-E boundary sections, which suggest that Ir anomalies are restricted to northern European sites and are coincident with an explosive magmatic flare-up in the North Atlantic [Schmitz et al., 2004; Storey et al., 2007]. Moreover, a detailed investigation of 3He accumulation in deep-sea sediments indicates no cometary helium source and instead suggests a rapid onset to, short duration of (<120 ka), and rapid recovery from the PETM, as well as increased sedimentation rates during the event [Farley and Eltgroth, 2003]. Other evidence for a cometary impact, such as shocked quartz and glass spherules, is lacking from known PETM sections [Schmitz et al., 2004]. Also, comparison of rock magnetic properties from the New Jersey sections with coeval marine sections from the South Atlantic, California, and New Zealand suggest that the SD magnetite is a local rather than a global phenomenon [Lippert et al., 2004; Villasante-Marcos et al., 2007]. Thus the abundance of SD magnetite in the New Jersey P-E boundary cores may have a biogenic, not extraterrestrial, origin, as dismissed by Kent et al. [2003] but argued for by Dickens and Francis [2003].

[5] Here we evaluate potential sources of SD magnetite in the New Jersey sections by presenting new magnetic hysteresis data from a core near the study areas examined by Kent et al. [2003]. Our new, high-resolution data show a similar increase in SD material, but indicate that the magnetic material has a biogenic origin. We also present results which provide additional insight to the domain state and magnetic mineralogy of the Wilson Lake samples and thus paleoenvironmental conditions.

2. Materials and Methods

[6] The Paleocene-Eocene boundary was recovered from ∼92 to 112 m in the U.S. Geological Survey (USGS) Wilson Lake (New Jersey (Figure 1)) core. The Wilson Lake core is one of several continental cores drilled into the Paleocene-Eocene neritic inner to middle shelf environment on the North American continental margin [Miller, 1997]. Other proximal cores include the shallow water Clayton core ∼5 km to the west, as well as the deeper Ancora (45–70 m depth) and Bass River (∼105–140 m depth) cores from Ocean Drilling Program Leg 174AX farther to the east [Miller et al., 1998; Van Sickel et al., 2004]. P-E boundary sediments from Wilson Lake consist of deltaic-neritic inner shelf clays and glauconitic sandstones deposited during a sea level transgression [Gibson et al., 1993]. Previous paleogeographic reconstructions suggest Wilson Lake was positioned in the north temperate zone (e.g., http://www.scotese.org), but recent reconstructions by Kopp et al. [2007] suggest a more southerly position in the subtropical evaporative belt (∼28°N latitude).

Figure 1.

Map showing the location of Wilson Lake (∼39°39′N, 75°2′W) and other USGS and Ocean Drilling Program cores [from Miller, 1997].

[7] The P-E boundary is well marked by nannofossil biostratigraphy [Gibbs et al., 2006] and here is defined as the onset of the CIE in bulk and single-species foraminifera carbon isotope records, between stratigraphic levels 109.23 and 109.85 m [Zachos et al., 2006] (Figure 2). The carbon isotope stratigraphy suggests that the Paleocene-Eocene transition is complete, although there are two unconformities in the lower Eocene, during the final recovery stages of the CIE [Gibbs et al., 2006]. The CIE persists in bulk and benthic foraminifera records for at least 13 m (from ∼109.85 to 96.94 m). Although the duration of the CIE is not well constrained in the New Jersey cores, 3He and other age models from deep-sea sections suggests a rapid onset (few thousand years) and short duration (<180 ka) [Farley and Eltgroth, 2003; Westerhold et al., 2007].

Figure 2.

Lithology and nannofossil biostratigraphic zonations (far left) for Wilson Lake plotted versus subsurface depth (m) (biostratigraphic scheme follows the nanoplankton scheme of Martini [1971]). Stable isotope, weight percent Corg and CaCO3 [Zachos et al., 2006], saturation magnetization (Ms), squareness (Mrs/Ms), and coercivity (Hc) are plotted versus subsurface depth. The stable isotope data are from analyses of Morozovella velascoensis (acuta), Acarinina soldadoensis, Subbotina spp., and Cibicidoides spp. The wavy lines at 94.79 and 96.32 m represent unconformities. The lower unconformity truncates the upper portion of the carbon isotope excursion layer. The gray bar in the carbonate panel indicates the Apectodinium dinoflagellate acme. Stars indicate stratigraphic levels at which LTD and TEM measurements were made.

[8] We measured standard magnetic hysteresis parameters (remanent magnetization, saturation remanent magnetization, coercivity, and coercivity of remanence (Mr, Mrs, Hc, Hcr, respectively)) to characterize the average magnetic grain size in each sample [Day et al., 1977; Dunlop, 2002a]. Bulk sediment samples (∼20 mg each) were collected every 10–30 cm over a 20 m interval and measured on a Princeton Measurements Corporation alternating gradient magnetometer at the University of California, Santa Cruz. We measured the low-temperature demagnetization (LTD) of zero field-cooled (ZFC) and field-cooled (FC) saturation isothermal remanent magnetizations (ZFC and FC SIRM, respectively) of a subset of eight samples to better understand the domain state and magnetic mineralogy of the Wilson Lake sediments. Low-temperature measurements were made on a Quantum Designs MPMS-2 at the Institute of Rock Magnetism at the University of Minnesota following methods described by Moskowitz et al. [1993]. Magnetic separates of samples from the same stratigraphic intervals as LTD samples, as well as additional samples at the P-E boundary, were prepared for transmission electron microscopy (TEM) following methods adapted from work by Tarduno and Wilkison [1996]. Samples were viewed with a JEOL 1200EX TEM operating at 80 kV and images were captured on Gatan Model 792 Bioscan digital camera at the University of California, Santa Cruz.

3. Results

[9] Magnetic hysteresis properties of Wilson Lake sediments show patterns similar to those found in nearby P-E boundary cores from outer shelf environments (Figure 1) [Kent et al., 2003; Lippert et al., 2004], as well as some pelagic and hemipelagic sections [Karlin, 1990b; Tarduno and Wilkison, 1996]. Saturation magnetization (Ms) and squareness (Mrs/Ms) remain relatively low throughout the latest Paleocene and then rise abruptly with the onset of the CIE at 109.85 m (Figure 2). Squareness values throughout the CIE interval exceed 0.4, and corresponding coercivity ratios (Hcr/Hc) are <2.0. Squareness and coercivity values remain high throughout the CIE and return to preexcursion values in the later stages of the CIE recovery (above 96 m). CIE samples have properties characteristic of a mixture of 30–50% cubic single domain (SD) magnetite and multidomain (MD) magnetite [Dunlop, 2002a]. Pre- and post-CIE samples have typical MD to superparamagnetic (SP) hysteresis behavior. These patterns are similar to results presented by Kent et al. [2003], especially those from the Clayton and Ancora sections. The higher resolution of our data set, however, reveals a previously unrecognized peak in magnetic coercivity (Hc) between 110 and 109 m, coincident with the transition from glauconitic sands to the clays, the onset of the CIE, and a peak in organic carbon levels.

[10] Low-temperature demagnetization experiments provided additional constraints on the magnetic mineralogy and origin of the SD material. For pre- and post-CIE samples, thermal demagnetization of ZFC and FC SIRM are nearly identical before the superparamagnetic (SP) signal at temperatures <40 K is removed, but diverge slightly after accounting for this signal, with the FC measurements yielding higher remanence (Figure 3b, inset). In contrast, CIE samples display divergent demagnetization curves with significantly greater FC remanence and a weak Verwey transition (Tv) depressed ∼20 K (Figure 3b, inset). We also calculated the ratio of remanence lost in warming through Tv in both ZFC and FC measurements (δZFC and δFC, Figure 3b), following Moskowitz et al. [1993]. CIE samples have delta ratios that range from 1.22 to 1.77.

Figure 3.

(a) Day plot of magnetic hysteresis data from Wilson Lake showing the magnetic grain size of bulk samples, generalized fields for pelagic and lacustrine sediments [see Dunlop, 2002b], and magnetic grain size fields (thin solid lines; PSD, pseudosingle-domain; MD, multidomain). Thick solid line is a theoretical mixing curve of MD and uniaxial single-domain (SD) magnetite; dashed line is a mixing curve of MD and cubic SD magnetite. Points indicate the percentage of SD grains in the mixture [Dunlop, 2002a]. Samples from the carbon isotope excursion (CIE) are shown by open diamonds. (b) Moskowitz plot [Moskowitz et al., 1993] of low-temperature demagnetization (LTD) results from Wilson Lake samples compared to common reference materials (GS-15, Geobacter metallireducens, a dissimilatory iron-reducing bacterium). Sample symbols are the same as in Figure 3a. Insets show representative LTD curves; solid squares: field cooled (FC); open circles: zero-field cooled (ZFC).

[11] We observed three general types of material in TEM images of magnetic extracts. Large (>200 nm), irregular grains (Figure 4a), probably of detrital origin, occur in all extracts. Ultrafine (<20 nm) opaque particles are present throughout the section in small abundance, except below the P-E boundary, where they dominate (Figure 4b). The third type of material is characterized by bullet-shaped, cubo-octahedral, or cubic grains 50 to 70 nm in length and with axial ratios >0.7 (Figures 4a, 4c, and 4d). These latter grains are dispersed throughout the sampled section, but occur in distinct chains between 109.85 and 108 m and constitute a significant component of the extracts at these stratigraphic levels. Dimensional analysis of 325 equant and bullet-shaped grains (Figure 4d) indicates that these grains are within the theoretical limits for single domain magnetite [Butler and Banerjee, 1975], consistent with the hysteresis measurements. We note that our results (abundances of grains, and diversity of grain sizes, shapes, and organization), are in sharp contrast to those presented by Kent et al. [2003] for the Clayton core just 5 km west of the Wilson Lake core but are similar to those presented by Kopp et al. [2007]. We attribute these differences and similarities mostly to extraction techniques and sample-handling procedures.

Figure 4.

Transmission electron microscope images of magnetic extracts from Wilson Lake. Scale bar is shown in each photomicrograph. (a) Typical detrital grains with dispersed 50–70 nm subeuhedral grains (from 109.12 m). (b) Typical ultrafine material found below the CIE (from 109.85 m) (c) Typical equant grains arranged in chains (from 109.58 m). (d) Plot of axial ratio versus particle length of 325 representative euhedral grains measured from TEM images. Also shown are theoretical magnetic grain size fields from Butler and Banerjee [1975]. Gray boxes indicate grain size ranges of magnetosomes in representatives of a few extant magnetotactic bacteria [Kirschvink, 1983].

4. Discussion

[12] We interpret the abrupt change in hysteresis properties coincident with the onset of the CIE at Wilson Lake as a change in the dominant grain size of preserved magnetic material. A Day plot of hysteresis properties (Figure 3a) suggests that CIE level sediments are characterized by a uniform population of as much as 50% cubic SD magnetite. This conclusion is supported by TEM of magnetic extracts. A strong cubic SD magnetite signal is not characteristic of pelagic, estuarine, or laucustrine sediments [Dunlop, 2002b], and therefore its source is significant. Cubic SD magnetite is common only to magnetotactic bacteria and possibly young oceanic basalts [Gee and Kent, 1995]. The former are a diverse group of prokaryotes that biomineralize intracellular, membrane-bound magnetite or greigite crystals called magnetosomes. Because crystal growth is biologically controlled, magnetosomes have a narrow size distribution (SD magnetite, 40–120 nm), are morphologically distinct (typically cubes, prisms, and bullets) and are often arranged in chains [Blakemore, 1982; Frankel, 1984; Sparks et al., 1990]. Magnetotactic bacteria are typically found at or below the oxic/anoxyic transition in modern laucustrine and marine water columns and sedimentary environments [Bazylinski et al., 1995; Kim et al., 2005; Peterman and Bleil, 1993; Simmons et al., 2004; Stoltz, 1992]. At Wilson Lake, there is no obvious detrital source for young oceanic basalt, which makes a biogenic origin for the fine-grained magnetic particles attractive.

[13] The increase in Mrs/Ms values at the P-E boundary, which reflects a transition from poor to strong magnetic remanence and therefore effective magnetic grain size, is accompanied by a peak in coercivity values (Hc). Tarduno and Wilkison [1996] noted similar patterns in hysteresis properties of young pelagic sediments from the South Pacific, and TEM studies indicate that the coercivity peak records an accumulation of magnetotactic bacteria. Similarly, we suggest the changes in magnetic grain size in the Wilson Lake core record an unusual abundance of magnetotactic bacteria preserved in PETM sediments.

[14] Our low-temperature demagnetization experiments support this conclusion. The absence of a Verwey transition in pre- and post-CIE samples indicates that magnetic mineralogy is dominated by either strongly oxidized magnetite or other weakly magnetic minerals. The presence of glauconite below the P-E boundary favors the latter interpretation. Abrupt decreases in magnetic remanence from 10 to 40 K may indicate pyrrhotite [Rochette et al., 1990] or significant amounts of superparamagnetic material in these samples. Transmission electron microscopy suggests that pre-CIE samples are dominated by ultrafine magnetic material well below the theoretical grain size threshold between superparamagnetic and single-domain material (Figure 4b). In contrast, CIE samples exhibit a weak Verwey transition (Tv) depressed ∼20 K, which suggests the presence of nonstoichiometric SD magnetite. The low-temperature remanence behavior of CIE samples is typical of a mixture of detrital and biogenic magnetite [Moskowitz et al., 1993; Smirnov and Tarduno, 2000; Weiss et al., 2004], an interpretation supported by our TEM results (Figures 4a and 4c). Moreover, Moskowitz et al. [1993] suggest that δZFC and δFC can indicate the presence of biogenic magnetic and discriminate it from detrital or other authigenic forms in bulk sediment samples [see also Weiss et al., 2004]. Pre- and post-CIE samples have delta values comparable to superparamagnetic and multidomain grains, while CIE samples are strikingly similar to values from extracted and intact chains of magnetosomes (Figure 3b).

[15] We interpret our TEM results as clearly indicating the presence of magnetotactic bacteria preserved in CIE and PETM sediments. Chains of equant, ∼70 nm grains are indistinguishable from TEM images of extant magnetotactic bacteria, leading us to conclude that these grains are indeed fossil magnetosomes. Numerous dispersed grains similar in size and shape to grains constituting intact chains are probably isolated magnetosomes and chains disaggregated by bioturbation and sample processing. Magnetosomes are absent below the onset of the CIE, where much finer-grained magnetic material is present, and decrease in abundance throughout the CIE recovery. Thus the anomalous magnetic material found in Wilson Lake and proximal P-E boundary sections does not have a bolide origin as suggested by Kent et al. [2003], but rather a biogenic origin as suggested by Dickens and Francis [2003]. Similar findings using ferromagnetic resonance and TEM have been reported for a nearby midshelf sequence at Ancora, New Jersey [Kopp et al., 2007]. Insufficient magnetic concentration and extraction techniques may have prevented Kent et al. [2003] from observing similarly abundant and distinct magnetic grain assemblages in their studied sections.

[16] Assuming a biological control, the magnetic grain sizes and properties may be useful proxies for reconstructing changes in bacterial paleoecology, paleoproductivity, and/or changes in redox conditions at the sediment-water interface or in the water column [Hesse, 1994; Kim et al., 2005; Simmons et al., 2004; Stoltz, 1992; Tarduno and Wilkison, 1996]. Strong seasonality in the subtropics during the CIE would periodically increase runoff, water column stratification, and sedimentation rate, while delivering more nutrients to the coastal oceans [Gibbs et al., 2006; Schmitz and Pujalte, 2007]. Thick kaolinite deposits at Wilson Lake and other North Atlantic continental margin sites [Gibson et al., 2000; Schmitz et al., 2001] indicate high seasonal physical weathering and runoff rates during the initial CIE. Moreover, enhanced productivity inferred from nannoplankton assemblages [Crouch et al., 2001; Kaiho et al., 1996; Speijer et al., 1996] at the onset of the PETM has been documented in several nearshore environments, including Wilson Lake [Gibbs et al., 2006]. Increases in dinoflagellate abundances coeval with the onset of the CIE at Wilson Lake (Figure 2) [Zachos et al., 2006] are also consistent with increased nutrient supply and coastal eutrophication [Bujak and Brinkhuis, 1998]. The low absolute amount of organic carbon preserved at Wilson Lake (Figure 2) may indicate a seasonally extreme regional climate characterized by a prolong dry season punctuated by a brief, intense wet season during which erosion and runoff were especially high [Schmitz and Pujalte, 2007; Schmitz et al., 2001; Zachos et al., 2006]. It is important to note too that the amount of organic carbon remains approximately the same across the P-E boundary while sedimentation rates increase significantly, indicating that more organic carbon was delivered to the North Atlantic coastal oceans during the PETM.

[17] What is the link between bacterial ecology, nutrients, and paleoproductivity? The peak abundances and morphologic diversity of magnetosomes, inferred from the magnetic coercivity peak and revealed in TEM images, is coincident with peak organic carbon levels, suggesting a potential link between nutrient supply and the bacterial community [Kim et al., 2005; Lippert et al., 2004; Peterman and Bleil, 1993; Tarduno and Wilkison, 1996]. Extant magnetotactic bacteria not only live in anoxic sediments, but also flourish below the oxic-anoxic transition zone in the water column of seasonally stratified, nutrient-loaded estuaries [Simmons et al., 2004] and freshwater lakes [Kim et al., 2005]. High sedimentation rates of clays and organic carbon during the CIE, followed by lower rates after the PETM would produce nonsteady state pore water redox conditions that would promote the preservation of magnetosomes in P-E boundary sediments [Dickens and Francis, 2003; Finney et al., 1988; Karlin, 1990a]. Therefore we suggest that seasonally enhanced precipitation induced by the PETM increased continental runoff and fostered eutropic-like conditions that may have extended well out onto the North Atlantic subtropical shelf. Strong seasonal stratification, coupled with increased fertilization of coastal oceans and the concomitant increase in paleoproductivity of nannoplankton, lead to eutrophic conditions on the shelf, thereby raising the oxic-anoxic boundary into the water column where magnetotactic bacteria flourished. High sedimentation rates of clays and organic carbon during the CIE promoted the excellent preservation of these bacterial blooms.


[18] P.L. thanks Xixi Zhao for low-temperature remanence measurements and the IRM at the University of Minnesota for the use of their facilities. Funds for the IRM are provided by the National Science Foundation (NSF) Instruments and Facilities Program and by the University of Minnesota. Funds for TEM work were provided by an award to P.L. from the Janice A. Nowell Memorial fund, and we thank Jon Krupp for TEM technical support. We thank Laurel Bybell for access to the Wilson Lake core. We also thank Rob Coe and John Tarduno for helpful discussions and Gerry Dickens, Birger Schmitz, Jim Channell, and an anonymous reviewer for constructive and helpful reviews. This research was partially supported by National Science Foundation grant EAR-0120727.