In situ magnetic identification of giant, needle-shaped magnetofossils in Paleocene–Eocene Thermal Maximum sediments

Significance

Giant magnetofossils are the preserved remains of iron-biomineralizing organisms that have so far been identified only in sediments deposited during ancient greenhouse climates. Giant magnetofossils have no modern analog, but their association with abrupt global warming events links them to environmental disturbances. Thus, giant magnetofossils may encode information about nutrient availability and water stratification in ancient aquatic environments. Identification of giant magnetofossils has previously required destructive extraction techniques. We show that giant, needle-shaped magnetofossils have distinct magnetic signatures. Our results provide a nondestructive method for identifying giant magnetofossil assemblages in bulk sediments, which will help test their significance with respect to environmental change.

Abstract

Near-shore marine sediments deposited during the Paleocene–Eocene Thermal Maximum at Wilson Lake, NJ, contain abundant conventional and giant magnetofossils. We find that giant, needle-shaped magnetofossils from Wilson Lake produce distinct magnetic signatures in low-noise, high-resolution first-order reversal curve (FORC) measurements. These magnetic measurements on bulk sediment samples identify the presence of giant, needle-shaped magnetofossils. Our results are supported by micromagnetic simulations of giant needle morphologies measured from transmission electron micrographs of magnetic extracts from Wilson Lake sediments. These simulations underscore the single-domain characteristics and the large magnetic coercivity associated with the extreme crystal elongation of giant needles. Giant magnetofossils have so far only been identified in sediments deposited during global hyperthermal events and therefore may serve as magnetic biomarkers of environmental disturbances. Our results show that FORC measurements are a nondestructive method for identifying giant magnetofossil assemblages in bulk sediments, which will help test their ecology and significance with respect to environmental change.

The Paleocene–Eocene Thermal Maximum (PETM; ∼56 Ma) is a geologically rapid global warming event with many characteristics that make it an analog for the Anthropocene (1, 2). These characteristics include rapid warming of the sea surface and atmosphere, increased seasonality of precipitation and temperature, and biological extinctions (1). The PETM is identified globally by a −3‰ carbon isotope excursion (CIE) in bulk marine carbonate and is characterized by three stratigraphic intervals: 1) a preonset excursion, 2) the main CIE, and 3) a recovery toward baseline δ¹³C levels (1). The main CIE interval is further subdivided into the CIE onset and CIE core, which correspond to the first ∼6 to 10 kyr and 100 to 200 kyr of the PETM (1, 3, 4). The Wilson Lake A (WL-A) core from Wilson Lake, NJ, contains a continental shelf section of the PETM within the Marlboro Clay and, within the Marlboro Clay, an expanded and nearly complete record of the CIE onset and CIE core (1, 5). Several near-shore and offshore cores complement the WL-A record and enable a broader understanding of how Paleogene coastal ecosystems responded to the rapid onset of global hyperthermal conditions (5–8). The New Jersey continental shelf experienced an overall rapid influx of clay, mineralization of iron oxides, dinoflagellate blooms, and benthic foraminifera species turnover coincident with the CIE onset (5, 9, 10).

Abundant conventional and giant magnetofossils, the fossil remains of magnetotactic bacteria and other iron-biomineralizing microorganisms, were identified in several New Jersey cores. These magnetofossils are interpreted to be the predominant source of the PETM magnetic enhancement of these cores (7, 11–14), although alternative sources have been suggested (15–18). Giant magnetofossils have so far only been identified in sediments from the PETM and the Middle Eocene Climatic Optimum, leading to the interpretation that they are unique to hyperthermal events (6, 7, 11–14). For example, Chang et al. (6) suggest that giant magnetofossils are linked to oceanic deoxygenation during the PETM.

Previous studies, which used micromagnetic simulations, electron holography, or both, suggest that giant magnetofossils have distinct magnetic properties (6, 11, 14, 18, 19). These interpretations are limited, however, by assumptions regarding crystal arrangement, spacing, and magnetic domain structure, and they lack independent confirmation of defining characteristics. Additionally, some of these methods have been applied to only a few giant magnetofossil morphologies.

Here we show independent, physical evidence of the magnetic signature of giant magnetofossils in situ (i.e., not in extracts) using low-noise, high-resolution first-order reversal curves (FORCs). FORC routines measure the response of all magnetic particles, including giant magnetofossils, within a bulk sediment sample. We also present micromagnetic simulations of giant needle-shaped crystals whose morphologies were characterized with transmission electron microscopy (TEM) of magnetic extracts. Our results show that giant, needle-shaped magnetofossils produce a high-coercivity component distinct from conventional magnetofossils that is identified using a specific FORC measurement protocol. We argue that these are definitive magnetic signatures of giant magnetofossils, which further supports the interpretation that the magnetic enhancement of the WL-A core has a biogenic origin. The link between giant magnetofossils, hyperthermal events, and oceanic deoxygenation makes the magnetic signature of giant needles a powerful tool for identifying giant magnetofossil assemblages and, by extension, testing their ecological significance in the context of global change events in the geologic record.

Results

Low-Noise, High-Resolution FORCs.

Previous studies suggest that the magnetic enhancement of the WL-A core is due to abundant magnetofossils (12, 13) found in magnetic extracts. Other sources of single-domain magnetite particles, such as debris from a comet or pyrogenic magnetite produced during wildfires, have also been proposed (15, 16, 18, 20). One of the reasons for the ambiguity in interpreting the origin of magnetic nanoparticles in the Marlboro Clay is the selective extraction of magnetofossils with large magnetic moments, in contrast to abiotic iron oxides; this selective extraction might explain their dominance in TEM observations (18, 20). Here we provide a detailed characterization of the magnetofossil signature in a bulk sediment sample by comparing low-noise, high-resolution FORC measurements on specimen WL35950b, from the CIE onset interval at WL-A, to that of BAL13, a sediment sample from Lake Baldeggersee in Switzerland (Fig. 1), where the magnetic signature of conventional magnetofossils was isolated for the first time (21). The magnetic signature of conventional magnetofossils consists of two narrow coercivity components associated with single-domain particles, referred to as biogenic soft and biogenic hard (21, 22). These magnetic characteristics have been confirmed by analysis of several freshwater and marine magnetofossil-rich sediments (23–26) and are therefore considered representative of conventional magnetofossils. The single-domain character and narrowness of these biogenic coercivity components reflect the tightly genetically controlled biomineralization process in producing highly uniform magnetic structures.

Fig. 1.

Low-noise, high-resolution FORC measurements for WL35950b and BAL13. (A) Three distinct coercivity distributions identified in WL35950b. Arrows indicate the coercivity components BS (biogenic soft), BH (biogenic hard), BN (biogenic needles), and HC (high coercivity). (B) Same as A for BAL13. (C) FORC diagram for WL35950b. Arrows point to the following features: doublet of positive and negative amplitudes produced by reversible magnetic moment rotation in uniaxial SD (single-domain) particles (SDR); high-coercivity SDR termination (SDRT); high-coercivity termination of flux closure annihilation (FCAT) from particles and/or aggregates of particles (clumps, collapsed chains) with vortex-like magnetization states; and FCAT for biogenic needles (FCAT [BN]). (D) FORC diagram for BAL13, plotted to scale with C. (E) Isolated central ridge contribution to the FORC diagram of WL35950b, with 10× vertical exaggeration highlighting the vertical offset with respect to B_u = 0. The dashed line is the expected B_u value for each vertical profile of the central ridge. (F) Vertical central ridge offset for WL35950b (dashed line) and BAL13 (solid line) B_c, with 1σ confidence intervals (shaded). These offsets are related inversely to the thermal activation volume of SD particles. Arrows point to coercivity ranges dominated by viscous particles (VISC), biogenic particles (BS, BH, and BN), and high-coercivity particles (HC). The SDRT marks the transition from biogenic to high-coercivity contributions.

FORC measurements M(B_r, B) are represented as two-dimensional density functions in FORC diagrams, which are defined as the second mixed derivative of magnetization, M, and plotted in coordinates of magnetic coercivity, B_c, and magnetic offset or interaction, B_u (27, 28) (see SI Appendix, SI Text, for a glossary of magnetic terminology). Three coercivity distribution types, characterized by specific magnetization responses to magnetic particles with different domain states (29), were obtained from our FORC measurements (Fig. 1 A and B). The first is f_dcd, the first derivative of direct current demagnetization curves M(B_r, 0). The median of f_dcd defines the coercivity of remanence B_cr. The second distribution is f_irr, the first derivative of the irreversible component of the ascending hysteresis branch. In contrast to other coercivity distributions, f_irr is also defined for negative fields, where transitions between magnetic states that occur without reversing the field direction are recorded. Accordingly, f_irr(B < 0) = 0 for uniaxial, noninteracting single-domain particles. Nonzero values of f_irr(B < 0) in Fig. 1 A and B correspond to flux closure nucleation in FORCs beginning at positive reversal fields. The third coercivity distribution is f_cr, the central ridge coercivity distribution, and is defined as the integral of the central ridge over B_u. The central ridge is a sharp signature of FORC diagrams located along or close to B_u = 0 (Fig. 1 C and D). It is produced by isolated particles or groups of particles, such as magnetosome chains, possessing magnetic states capable of single-domain remanence (29–32). FORC signatures typical of magnetofossil-rich sediments include the following three characteristics: a central ridge; a doublet of positive and negative amplitudes, nearly antisymmetric with respect to the B_u = −B_c diagonal; and positive amplitudes above the central ridge with a similar distribution along B_c (31, 33) (SI Appendix, Text S1.1).

The isolated central ridge of the WL35950b FORC diagram is characterized by a small, B_c-dependent vertical offset δB_u (dashed line in Fig. 1E) caused by the effect of thermal activations on the switching field (B_sw) of single-domain particles (34). Larger offsets are caused by smaller energy barriers which, in magnetite, correspond to smaller blocking volumes. This offset is compared with that of BAL13 in Fig. 1F. Between 0 and 10 mT, where magnetically viscous particles contribute to the central ridge, δB_u decreases sharply from 0.7 to ∼0.4 mT. It then remains nearly constant over the coercivity range of biogenic soft magnetofossils (10 to 50 mT) and increases slightly to ∼0.45 mT over the coercivity range of biogenic hard magnetofossils (50 to 100 mT). The corresponding offset observed in the FORC distribution of sample BAL13 cannot be calculated beyond 110 mT because the central ridge disappears above this B_c value. In contrast, δB_u continues to increase almost linearly in WL35950b up to B_c ∼ 140 mT. Between 140 and 210 mT, a broad peak with a maximum δB_u of ∼0.7 mT is observed. Above ∼210 mT, δB_u returns to the ideal continuation of the linear trend observed over the coercivity range for biogenic soft and biogenic hard magnetofossils (Fig. 1). As seen with lower-resolution measurements up to B_c = 800 mT, the coercivity distribution associated with the central ridge reaches a minimum at ∼300 mT and then increases again at higher fields (SI Appendix, Fig. S1). This minimum represents the transition from SD magnetite contributions, whose theoretical upper coercivity limit is 300 mT, to high-coercivity minerals (e.g., goethite) (35).

Micromagnetic Modeling.

We performed micromagnetic simulations of the magnetic hysteresis produced by giant, needle-shaped magnetofossils identified within the CIE onset interval at WL-A (Fig. 4, Table S1). These simulations predict the magnetic behavior and the associated FORC diagram signature of the giant needles. The key parameters extracted from the simulated hysteresis loops include the field at which the loop magnetization reverses sign (B_c), the switching field (B_sw, the field at which the magnetization state reverses), and squareness (M_rs/M_s, the saturation remanence M_rs normalized to saturation magnetization M_s). Simulated hysteresis loops indicate that B_c and B_sw for needle-shaped particles with lengths of 1,000 nm and widths up to 100 nm reach maximum values (∼310 mT) when field orientations are parallel and perpendicular, respectively, to particle elongation (Fig. 2 and SI Appendix, Fig. S2). The average B_c is ∼92 mT, and the average B_sw is ∼148 mT. Notably, however, the averages for the two thinnest needles (21 and 25 nm, both 1,000 nm long) are B_c ∼110 mT and B_sw ∼165 mT. The ∼165 mT switching field value matches the peak central ridge offset at 170 mT (Fig. 1F). Generally, B_c and B_sw decrease with increasing needle width due to the emergence of nonhomogeneous switching modes (e.g., curling). B_sw follows this trend at all field angles (ϕ), while B_c values are width-independent for ϕ > 60°. The angular dependences of B_c and B_sw generally track those for the model of coherent rotation of atomic spin moments (36) for the three thinnest needle morphologies and the curling model (37) for the 100-nm-wide needle (Fig. 2B). At large field angles the spin moments continue rotating, yielding negative magnetization values before the switching event occurs, so that B_sw will be larger than B_c. At small field angles, B_sw coincides with B_c. For the 21- to 50-nm-wide needles the switching process can include a buckling or fanning component (38), whereby moments at needle extremities tend to rotate at a different rate than moments in the center (e.g., SI Appendix, Figs. S3 F, G, J, and K and S4 F, G, J, and K and Movies S1–S6). For the 100-nm-wide particle, maximum curling occurs just prior to the switching event (SI Appendix, Figs. S3C and S4C and Movies S7 and S8). The angular dependence of M_rs/M_s follows the Stoner–Wohlfarth model for the magnetization of single-domain particles (Fig. 2A) (36) because all needles are almost homogeneously magnetized in a null field. We also modeled a 150-nm-wide needle (SI Appendix, Fig. S5) that switches nonhomogeneously via vortex nucleation and annihilation; this magnetic behavior will not contribute to the central ridge signal at high fields.

Fig. 2.

Summary of results from micromagnetic simulations on the 21-, 25-, 50-, and 100-nm-wide needles observed in magnetic extracts from WL35900 and WL35800 (50 and 100 nm) and inferred from our low-noise, high-resolution FORC measurements (21 and 25 nm). Simulations were performed as a function of the angle (ϕ) between the applied field and the particle elongation direction, which was oriented in the z direction. The angular dependence was compared to Stoner–Wohlfarth (36) calculations for coherent rotation in an infinite cylinder (dashed and solid black lines, S-W). (A) Magnetic squareness (M_rs/M_s) as a function of ϕ. The angular dependence follows the S-W model, M_rs/M_s(ϕ) = cosϕ. The dashed line is for M_rs/M_s(0°) = 1, which is the case for the 21- to 50-nm-wide needles (see also SI Appendix, Fig. S2). (B) Coercivity (B_c) and switching field (B_sw) as a function of ϕ. For ϕ < 60°, the symbols for B_c and B_sw overlap. The gray dashed lines are calculations of the nucleation field for cases exhibiting fanning (S ≤.1.63) and curling (S > 1.63) of moments. S is a parameter that depends on particle width, exchange constant, and M_s (see SI Appendix, Text S2, for details of the theoretical calculations).

TEM.

Magnetofossil dimensional measurements (Table S1) are used to predict the domain state of the nine magnetofossil categories preserved within the CIE onset interval at WL-A. We identified nine magnetofossil categories: immature, cuboctahedra, small bullets, medium bullets, elongate prisms, large bullets, giant bullets, spindles, and needles; the latter three are giant magnetofossil morphologies (11, 14). Giant magnetofossils are only identified within magnetic extracts from the CIE onset interval, so we use only measurements from extracts WL35900 and WL35800 for our analysis. Conventional magnetofossil morphologies (e.g., cuboctahedra, small bullets, medium bullets, and large bullets) are within or close to the stability range of isolated single-domain particles (Fig. 3) (39, 40). If chain geometries are considered, then the single-domain stability range is expanded by the stabilizing effect of positive magnetostatic interactions (40, 41); in this context, all particles identified as conventional magnetofossils are within the single-domain stability range. This is expected if magnetofossils are optimized for magnetic navigation, in which case they must provide the maximum magnetic moment with the minimum amount of iron (42). In contrast, giant bullets, spindles, and elongate prisms cluster around the upper limit of the single-domain stability range. Specific micromagnetic simulations of these morphologies indicate that they can have single-domain and non–single-domain magnetic structures depending on their size, shape, and arrangement in chain structures (11). Together, these observations suggest that magnetic navigation might not have been the only purpose of giant bullets, spindles, and elongate prisms. Alternatively, abundant bioavailable iron, from enhanced continental weathering during the PETM (1), may have decreased the pressure to biomineralize morphologies with the most efficient magnetic moments. Our micromagnetic simulations indicate that giant needles possess a stable single-domain–like remanence without the need for chain arrangements if their widths do not exceed ∼150 nm for particles ≥500 nm in length. This is close to the upper limit of the few available TEM observations of these crystals (e.g., Fig. 4); however, the real size distribution is likely wider and would therefore exceed the single-domain stability range of isolated crystals. We note that our few TEM observations are limited by the fact that giant needles are less abundant and they may be more difficult to dislodge from bulk sediment than conventional magnetofossils.

Fig. 3.

Dimensional analysis of magnetofossils identified in WL35900 and WL35800 with predicted domain states at room temperature. The theoretical single-domain states for individual grains and magnetofossil chains are highlighted: 1) lower limit for a six-crystal chain with intercrystal gaps of 0 (lower curve) and 0.6 times the length of constituent crystals (41), 2) lower limit for isolated single-domain particles (39), 3) upper limits for isolated single-domain particles with long axes parallel to the 〈100〉 and 〈111〉 crystallographic axes (40), and 4) critical sizes for isolated crystals within a chain of three crystals (40). Dimensions of modern magnetosomal magnetite of various shapes from well-characterized magnetotactic bacteria and conventional magnetofossils are shown in color (for a complete list of data sources, see SI Appendix, Text S3).

Fig. 4.

TEM images of giant needles from magnetic extract WL35900 highlighted with red arrows. Two needles are shown in the Left panel, whereas the Right panel shows a single needle with several conventional magnetofossils in the upper right. Needles have a cylindrical-like morphology (Inset in Right) and some taper toward one end of the crystal.

Discussion

Magnetofossil Components Identified from the Central Ridge Coercivity Distribution (f_cr).

There are two principal differences between the three coercivity distributions obtained from our FORC measurements for WL359050b and BAL13 (Fig. 1 A and B). First, f_irr and f_cr are determined by in-field (i.e., induced) magnetization states, whereas f_dcd probes remanent magnetization states. Second, only a selected subset of micromagnetic transitions contribute to f_cr, namely, those to the positive saturation state is a sequence of transitions that started from negative saturation (29). In single-domain particles and structures with single-domain behavior, such as magnetosome chains, there is only one transition during FORC measurements, from a negative to a positive single-domain state. It is this transition that contributes to the central ridge, which explains its dominance in magnetofossil signatures. Slightly larger particles possess a few more magnetic states (43), perhaps related to the nucleation of magnetic vortices (32), but transitions to the positive single-domain state still contribute significantly to the central ridge. The number of magnetic states grows rapidly with particle size, and the relative contribution of the central ridge to the total magnetic signal decreases accordingly, until it disappears fully. The contribution of magnetic structures with non–single-domain behavior, such as collapsed chains, some giant magnetofossils, lithogenic magnetite, and titanomagnetite, explain the smaller amplitude of f_cr. The strong single-domain selectivity of the central ridge is similar to that of anhysteretic remanent magnetization (ARM) (44), so that f_cr is closely related to the coercivity distribution obtained from the alternating field demagnetization of ARM (44).

Since f_cr depends on fewer magnetic processes, it contains coercivity components with smaller dispersion parameters. This effect helps distinguish individual coercivity components and is clearly visible in BAL13 (Fig. 1B): the biogenic soft and biogenic hard coercivity components, originally identified with ARM demagnetization curves (21), are clearly recognizable in f_cr but are less evident in f_dc and f_irr due to their strong overlap in those coercivity measurements. The maximum coercivity range of the two conventional magnetofossil components reaches ∼120 mT, which is also the upper limit for single-domain coherent rotation signatures in the lower half of the FORC diagram (Fig. 1D). In contrast, the coercivity range of contributions in the upper half of the FORC diagram, which are caused by flux closure annihilation, is limited mostly to 80 mT (Fig. 1D).

The same biogenic components are identified in the central ridge distribution for WL35950b. However, in this case an additional, higher-coercivity contribution dominates the 120 to 210 mT range (Fig. 1A). We attribute this higher-coercivity contribution to the needles observed with TEM (Fig. 4) rather than other giant magnetofossil morphologies, like bullets and spindles, which likely exhibit vorticity and fanning/curling (thus lower coercivities) (11) because of their large widths. The high-coercivity contribution to the central ridge is not unique to needles: it also has been observed in some igneous rocks (45) and in sediments with silicate minerals that contain inclusions of fine-grained single-domain magnetite (46). We note that TEM observations and the sedimentology of WL-A sediments (5) indicate that these alternative sources do not contribute to the magnetic signal in our specimens. Micromagnetic simulations of conventional magnetosome chains show that some chain configurations (i.e., >10 magnetosomes or with interparticle spacings ∼1 nm) may have coercivity tails that reach ∼180 mT (19, 47), but these contributions are smaller than those of the needles described here. The uniaxial single-domain nature of this needle component is also identified in the FORC diagram by the larger field range of coherent single-domain rotation contributions (lower half plane) and by a small contribution (upper half plane) caused by flux closure annihilation (Fig. 1C). A minor fraction of the needles must therefore be sufficiently large to nucleate magnetic vortices during FORC measurements, such as the 150-nm-wide needle of the micromagnetic simulation shown in SI Appendix, Fig. S5.

Central Ridge Offset as a Distinct Signature of Biogenic Needles.

The central ridge offset provides insight into the physical nature of the magnetic components biogenic soft, biogenic hard, and biogenic needle (Fig. 1F). The small vertical offset of the central ridge is related to the measurement timing asymmetry of the FORC protocol: decreasing the applied field from positive saturation to a negative reversal field B_r switches all single-domain particles with switching fields ≤ −B_r, and these particles are switched back to positive saturation near B = −B_r. Thermally activated switching occurs during the pause at B_r and during measurement at B ∼ −B_r. More time is effectively spent at B_r than near −B_r, which requires a slightly larger field B = −B_r + ∆B_fluc to switch all particles back to positive saturation. ∆B_fluc is the difference between fluctuation fields for thermally assisted switching at ±B_r (34, 48).

The central ridge offset corresponds exactly to ∆B_fluc and depends on the amplitude of the energy barrier of single-domain particles in proximity of the switching field, with ∆B_fluc → 0 as the energy barrier grows to infinity. For single-domain particles switching by coherent rotation, this energy barrier is proportional to particle volume (49). The energy barrier continues to increase with volume in single vortex particles (50), an effect which can also contribute to the central ridge signal. The central ridge offset at B_c is thus a measure of the thermal activation volume of a single particle, or groups of particles with collective behavior (such as magnetosome chains), switching at B_c.

A small but significant central ridge offset (0.4 to 0.7 mT) is visible in both samples after careful isolation of the central ridge from the remaining FORC contributions using VARIFORC (VARIable FORC) processing (51) (Fig. 1E; see also SI Appendix, Text S1.2–S1.5). The initial offset decrease originates from viscous single-domain particles via the direct relationship between switching field and particle volume. Viscous particles, such as isolated immature magnetofossils, contribute to f_cr at the B_c → 0 limit, before the onset of the biogenic soft component at ∼10 mT (21). The relatively constant offset over the coercivity range of conventional magnetofossils is also observed in the FORC distribution for BAL13. This is consistent with a common origin of the biogenic soft and biogenic hard components in the two samples. TEM results (SI Appendix, Table S1) further suggest that conventional magnetofossils observed in magnetic extracts from WL-A can be categorized into biogenic soft or biogenic hard based on relative shape anisotropy and probable chain structure. The maximum central ridge offset occurs in the coercivity range of the biogenic needle component, which is attributed to single-domain particles with a smaller activation volume than that of conventional magnetofossil chains. Identification of these central ridge offset signatures depends largely on the tight field control used for FORC measurements and enhanced signal-to-noise ratio obtained by stacking multiple sets of curves (see details in Materials and Methods).

Implications for Studying Hyperthermal Events.

With validation by micromagnetic modeling and TEM imaging of magnetic extracts, we present unequivocal identification of giant, needle-shaped magnetofossils derived from central ridge characteristics of FORC diagrams. This confirms that the CIE magnetic enhancement is caused predominately by conventional and giant magnetofossils. Our FORC measurement protocol is optimized for rigorous quantification of the vertical offset of the central ridge associated with the needle component. The needle component, and potentially other magnetofossil signatures, is also identified, albeit more ambiguously, using lower-resolution FORC diagrams. Together, these data indicate that the magnetic signatures described here are those of in situ magnetic particles, not a subset of particles removed from their taphonomic context via artificial sorting and reorganization by magnetic extraction. In situ identification of central ridge components with coercivity peaks >20 mT and small fluctuation fields ≪B_c, indicates abundant single-domain particles, or chains of such particles, with strong uniaxial anisotropies compatible with those of conventional magnetofossils and giant needles. The central ridge contributes to ∼40% of the saturation remanent magnetization (M_rs). This is also the lower limit for the total magnetofossil contribution, which is significantly higher than the ∼10% estimated by Wang et al. (18). The magnetization carried by magnetofossils is much larger if the FORC contributions above and below the central ridge, associated with transitions from single-domain–like to flux-closure magnetic states and vice versa, are considered. As shown with selective chemical extraction of a magnetofossil-rich sediment, the central ridge contributes to ∼65% of the M_rs carried by <0.3 µm magnetite particles (24, 52). Using this estimate, the magnetofossil contribution to the M_rs of WL35950b increases to ∼66%. The presence of nonbiogenic, isolated single-domain equidimensional or irregular magnetite particles of the type attributed to an impact (15) cannot be excluded. However, their contribution, if present, must be minor, given the absence of a coercivity peak at B_c = 0 and the viscous signatures that unavoidably occur with single-domain magnetite particles lacking uniaxial anisotropy and biological size control (52–54).

Giant needles have so far only been found in sediments deposited during hyperthermal events, and they are often accompanied by other giant magnetofossil morphologies (e.g., spearheads, spindles, and bullets) (6, 7, 11, 14). Although the physiology and ecology of the organisms that biomineralize giant needles is unknown, the association of these magnetofossils with the onset of rapid and large changes in the temperature and chemistry of marine environments suggests they could be used to efficiently identify environmental perturbations. The capability of detecting giant, needle-shaped magnetofossils using robust, nondestructive magnetic measurements on bulk sediment samples, in contrast to magnetic separates, is promising for scientists studying global change events. The ability to rapidly find giant magnetofossil assemblages in the geologic record will help identify the origin of these unusual magnetofossils and the specific ecology of the organisms that biomineralized them. The needle component may help identify intervals of substantial environmental change that are more subtly expressed by other microorganisms and geochemical proxies. Moreover, giant magnetofossils appear to be linked to oceanic deoxygenation stimulated by rapid planetary change, particularly warming events. By studying the occurrence of giant magnetofossils we can better understand how sensitive marine ecosystems responded to past climate change events.

Materials and Methods

Sample Preparation for FORC Measurements.

We prepared a dry, powdered sediment chip (∼0.116 g) from the WL-A core (WL35950b). This specimen is from the CIE onset interval of the PETM. Although our measurements were performed on disaggregated sediment, we consider the measurements in situ: we did not apply any magnetic extraction, and we have not disrupted the submicron arrangement of particles at a scale relevant to magnetostatic interactions. The vertical offset of the central ridge, which we conclude is a characteristic of biogenic needles, is also visible in FORC diagrams of intact, unpowdered bulk sediment chips (SI Appendix, Fig. S1), despite the fact that this measurement has a lower signal-to-noise ratio than those completed on the powdered sample. Future experiments will help identify the optimal measurement routine to resolve the vertical offset with confidence while also minimizing user and instrument time. A freeze-dried sediment sample from Lake Baldeggersee (BAL13) (22) is included for comparison. This sample corresponds to a depth interval of 13 to 14 cm in the gravity core described by Egli (22) and is characterized by approximately equal contributions of biogenic soft, biogenic hard, and a low-coercivity detrital component. Both specimens were secured inside gelatin capsules and then affixed to a vibrating sample magnetometer (VSM) probe for high-resolution FORC measurements.

High-Resolution FORCs.

Specimens WL35950b and BAL13 were measured using a specialized, high-resolution, low-noise FORC measurement protocol with improved field precision to extract the thermal activation signature of single-domain particles contributing to the central ridge. The Lake Shore 8600 VSM used for these measurements provides excellent field control characteristics and the possibility of zeroing the Hall probe and monitoring the applied field at a 100-Hz rate. The typical reversal field overshoot <0.2 mT of the regular FORC protocol can be further decreased by a stepwise approach to B_r in which the field is ramped from saturation to B_r + k∆B, where ∆B is the field step used for FORC measurements, and k ≥ 1 is the number of approach steps. In this manner, field overshooting problems are eliminated. The following protocol was used for the FORC measurements of WL35950b: B_sat = 500 mT, B_c,max = 250 mT, B_u,min = −10.2 mT, B_u,max = +80.2 mT, ∆B = 0.4 mT, k = 3, pause at B_r = 3 s. Measurements were performed in point-by-point mode (the field was stabilized for 0.1 s before a measurement is taken) with averaging time = 0.15 s. BAL13 was measured with the same protocol, except for limiting B_c,max to 160 mT due to the lack of high-coercivity components. The maximum Hall probe bias during measurements at constant laboratory temperature was 10 µT. Signal-to-noise ratios well above the standard of regular FORC measurements were attained by stacking 32 sets of measurements for WL35950b and 13 sets of measurements for BAL13. These data were processed using optimal measurement averaging (ImportFORC), smoothing (CalculateFORC), and central ridge processing (IsolateCR) through VARIFORC (51). For full details on the measurement protocol, see SI Appendix, Text S1.2–1.5 and Figs. S6–S8. The signal-to-noise ratio of processed data for WL35950b in correspondence of the central ridge is 100 to 150 over the coercivity range covered by conventional magnetofossils; it is 60 to 100 over the coercivity range of giant needles, which ensures sufficiently precise quantification of the vertical central ridge offset. This is a FORC protocol designed specifically to characterize the distinct FORC signatures associated with the needle component. Although the resulting data are nosier, this component is also identified in lower-resolution FORC datasets that only require ∼10 h per measurement (SI Appendix, Text S4 and Fig. S1).

Micromagnetic Modeling.

Micromagnetic simulations for giant needle-shaped magnetofossils were performed using MERRILL (Micromagnetic Earth Related Robust Interpreted Language Laboratory), a micromagnetics package optimized for rock magnetism (55). We used the default settings for magnetite at room temperature, which include magnetocrystalline anisotropy with the easy axis along the 〈111〉 crystallographic direction. Particle elongation was set along the 〈001〉 direction (z in our coordinate system). Shape estimates and constraints were determined from TEM images of giant needles and from our high-resolution FORC measurements. We performed simulations for the following needle widths: 21, 25, 50, 100, and 150 nm. Needle length was 1,000 nm, except for the 150-nm-wide particle, which had a length of 500 nm. Volumetric meshes with unit cell sizes of 5 nm were generated using MEshRRILL (https://bitbucket.org/poconbhui/meshrrill/). The three widest particle sizes were used to model the giant needle-shaped magnetofossils observed in extracts from WL35900 and WL35800. The two narrower particle sizes were used to model giant needle-shaped magnetofossils inferred from the high-resolution FORC measurements of specimen WL35950b. Hysteresis loops were simulated in 10° increments between the z direction (parallel to elongation) and the x direction (perpendicular to elongation) for the 21-, 25-, and 50-nm needles. Loops were simulated in the same orientations for the 100-nm needle but in 15° increments. Only one loop was simulated for the 150-nm needle, with the field along the z direction. Saturating fields were chosen between 100 and 500 mT, depending on grain size and orientation. The upper hysteresis loop branch was obtained at 5-mT field steps, and the lower branch was calculated by inverting the upper branch.

TEM.

We used average magnetofossil lengths and widths of particles imaged using TEM of magnetic extracts (Table S1) for dimensional analyses. TEM was performed on the TECNAI TF30 scanning/TEM at John Hopkins University, in the Materials Characterization and Processing Facility, using an accelerating voltage of 300 keV. Particles were measured using Gatan's GMS 3 Digital Micrograph Software. Dimensional analyses predict the magnetic domain state of the magnetofossils identified from magnetic extracts from the CIE onset interval at WL-A, WL35900, and WL35800. We plotted magnetofossil length against the width/length of magnetofossils with the theoretical grain size limits for individual particles of magnetite and chains of interacting magnetofossils from Butler and Banerjee (39), Muxworthy and Williams (40), and Newell (41).

Data Availability.

AVI, FRC, and PDF data have been deposited in Figshare (https://doi.org/10.6084/m9.figshare.13182848). All other study data are included in the article and supporting information.