The identification and biogeochemical interpretation of

Robert E. Kopp* and Joseph L. Kirschvink

Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA 91125, USA

*Corresponding author. Department of Geosciences, Princeton University, Princeton, NJ 08544, USA. E-mail: rkopp AT princeton.edu

Received 16 April 2007; accepted 6 August 2007. Available online 14 August 2007.

Abstract Magnetotactic bacteria, which most commonly live within the oxic-anoxic transition zone (OATZ) of aquatic environments, produce intracellular crystals of magnetic minerals, specifically or greigite. The crystals cause the bacteria to orient themselves passively with respect to the geomagnetic field and thereby facilitate the bacteria’s search for optimal conditions within the sharp chemical gradients of the OATZ. The bacteria may also gain energy from the redox cycling of their crystals. Because magnetotactic bacteria benefit from their magnetic moments, natural selection has promoted the development of traits that increase the efficiency with which the intracellular crystals impart magnetic moments to cells. These traits also allow crystals produced by magnetotactic bacteria (called magnetofossils when preserved in sediments) to be distinguished from abiogenic particles and particles produced as extracellular byproducts of bacterial metabolism. Magnetofossils are recognizable based on their narrow size and shape distributions, distinctive morphologies with blunt crystal edges, chain arrangement, chemical purity, and crystallographic perfection. This article presents a scheme for rating magnetofossil robustness based on these traits. The magnetofossil record extends robustly to the Cretaceous and with lesser certainty to the late Archean. Because magnetotactic bacteria predominantly live in the OATZ, the abundance and character of their can reflect environmental changes that alter the chemical stratification of sediments and the water column. The magnetofossil record therefore provides an underutilized archive of paleoenvironmental information. Several studies have demonstrated a relationship between magnetofossil abundance and glacial/interglacial cycles, likely mediated by changes in pore water oxygen levels. More speculatively, a better-developed magnetofossil record might provide constraints on the long-term evolution of marine redox stratification. More work in modern and ancient settings is necessary to explicate the mechanisms linking the abundance and character of magnetofossils to ancient biogeochemistry.

Keywords: bacteria, magnetite, greigite, , biogeochemistry

Published in Earth-Science Reviews (2008), 86: 42-61. doi: 10.1016/j.earscirev.2007.08.001 Copyright © 2007 Elsevier B. V. R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

1. Introduction conventional fossils, the presence of magnetofossils reflects environmental Traditionally, paleobiology has focused conditions, specifically conditions that on studying the products of biologically- facilitate magnetotactic bacteria growth and controlled mineralization (BCM), the process magnetofossil preservation. Magnetofossils of inducing minerals to precipitate following a are therefore a largely untapped proxy for template established by organic molecules ancient biogeochemistry. (Lowenstam, 1981). Such products are clear indications of the existence and nature of past 2. Ecology of magnetotactic bacteria life. A fossil shell, for instance, is an unequivocal biosignature that conveys 2.1. Magnetotaxis and Redox Zonation information about the paleoenvironment of its formation and hints to the evolutionary Magnetotactic bacteria produce complexity of the organism that created it. membrane-bound magnetite or greigite BCM is rare, however, in microbes. Thus, the crystals within vesicles called study of ancient microbes usually relies upon (Bazylinski and Frankel, 2004; Gorby et al., alternative techniques, each with distinctive 1988; Komeili et al., 2006; Komeili et al., 1 strengths and weaknesses: techniques such as 2004; Matsunaga and Okamura, 2003). The the study of stromatolites and carbonaceous crystals provide the bacteria with a net , the interpretation of organic magnetic moment, which they employ for biomarker compounds and isotopic signatures, magnetotaxis, movement directed by the local and the phylogenetic analysis of genomic data. magnetic field. Many magnetotactic bacteria Though many prokaryotes precipitate grow preferentially under specific, narrow carbonate, sulfide, or oxide minerals as redox conditions (Figure 1). Magnetite extracellular metabolic byproducts, producers are often microaerophiles or nitrate- magnetotactic bacteria are among the few reducers found in the suboxic conditions of the prokaryotes to engage in BCM (Blakemore, oxic-anoxic transition zone (OATZ) 1975). These bacteria are defined by the (Bazylinski and Moskowitz, 1997). Simmons ability to precipitate intracellular crystals of et al. (2004) found that, in Salt Pond, ferrimagnetic minerals, specifically magnetite Massachusetts, magnetite producers were and greigite, and have been found from several particularly concentrated at the top of the divisions of the Proteobacteria (DeLong et al., OATZ, where oxygen diffusing from above 1993; Maratea and Blakemore, 1981; Simmons and iron diffusing from below produce a peak et al., 2004) and from the Nitrospira (Spring et in particulate Fe(III) abundance. There are al., 1993; Spring and Schleifer, 1995). Like exceptions to this typical distribution, however. fossil shells, bacterial magnetite and greigite Some magnetite producers can grow under bear the signs of natural selection’s optimizing aerobic conditions (e.g. Schüler and influence. They can therefore be identified in Baeuerlein, 1998), although they do not sediments, where they are given the name produce magnetite at high oxygen levels. At magnetofossils. Like the presence of more least one magnetite producer, Desulfovibrio

1 Whether the magnetosomes are invaginations of the cytoplasmic membrane or true organelles is a matter of current debate (Kobayashi et al., 2006; Komeili et al., 2006). R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

Magnetotactic Bacteria Magnetite Growth Preservation Oxygen Microaerophilic Maghemitization Magnetite Producers Preservation of Anaerobic Unoxidized Magnetite Fe(II) Greigite Reductive Dissolution

Depth Producers

Sulfide

Concentration

Figure 1: Schematic representation of chemical gradients, typical optimal growth positions of different types of magnetotactic bacteria, and typical diagenetic fates of magnetite. Redox gradients can exist over scales from millimeters to meters. magneticus RS-1, is a strictly anaerobic sulfate magnetotactic bacterium found in reducing bacterium (Sakaguchi et al., 2002; Pettaquamscutt River Estuary, Rhode Island, Sakaguchi et al., 1993), although it is only produces both greigite and magnetite, with a weakly magnetotactic and may use its greater proportion of greigite particles magnetosomes for an alternate purpose (Pósfai produced under more reducing conditions et al., 2006). (Bazylinski et al., 1995). Greigite producers prefer more reduced In environments with sharp redox conditions and are likely strictly anaerobic gradients, magnetotaxis likely provides sulfate reducers (DeLong et al., 1993). magnetotactic bacteria with a selective Simmons et al. (2004) found that greigite- advantage by allowing them to search in one producing multicellular magnetotactic dimension instead of in three dimensions for prokaryotes (MMPs) grew in greatest optimal geochemical conditions (Kirschvink, abundance near the dissolved Fe(II) 1980). The cells’ magnetic moment causes concentration peak at the base of the OATZ, them to align passively with the local magnetic while other greigite producers grew in deeper, field. Redox gradients are often nearly more sulfidic waters. One sulfate-reducing vertical, and, except at the geomagnetic greigite bacterium, found in microbial mats equator, the geomagnetic field has a vertical associated with methane-seep carbonate component. For cells with moments > 10-15 concretions in the Black Sea, is a member of a Am2, equivalent to that produced by ~17 cubic syntrophic partnership engaged in anaerobic magnetite crystals with 50 nm edge lengths, oxidation of methane (Reitner et al., 2005). A >90% of the cell's velocity is directed along

3 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria magnetic field lines (Frankel and Blakemore, Magnetotaxis performs a function 1989). However, because other gradient analogous to that of the sheaths of Thioploca: organisms thrive in spite of performing three- namely, enabling the magnetotactic bacteria to dimensional biased random walks to find swim rapidly and directly between electron optimal conditions, additional functions for the donor rich and electron acceptor rich regions. magnetosomes have been proposed, including Moreover, as in the colorless sulfur bacteria, iron storage (Chang and Kirschvink, 1989) and sulfur globules have been found in a number of energy storage (Vali and Kirschvink, 1991). magnetotactic bacteria (Cox, 2002; Moench, The discovery of membrane-bound iron oxide 1988; Spring et al., 1993). As Spring et al. inclusions in the non-magnetotactic, iron- (1993) suggested, these globules may act as reducing bacterium Shewanella putrefaciens electron donor reserves. Some magnetotactic (Glasauer et al., 2002) lends credence to these bacteria also contain additional energy storage suggestions. compounds, such as polyphosphate granules and polyhydroxyalkanoates (Keim et al., 2005). 2.2. Magnetotaxis as a way to short-circuit Indeed, crystals diffusion themselves may act as intracellular storage For most organisms living at sharp batteries, becoming partially oxidized while redox gradients, metabolism is limited by the the bacterium is in oxidizing waters and diffusive fluxes of nutrients, but, as reviewed reduced back to stochiometric magnetite while by Schulz and Jørgensen (2001), several the bacterium is in reducing waters (Vali and species of large colorless sulfur bacteria have Kirschvink, 1991). Thermodynamic and found ways of bypassing diffusive limitations. growth rate calculations indicate that such a These microaerophilic or nitrate-reducing pathway is a feasible supplementary organisms employ two distinctive strategies: metabolism at the centimeter and sub- (1) overcoming diffusive limitation of electron centimeter length scales characteristic of donors and acceptors through various sedimentary redox zonation (Kopp, 2007). If approaches to motility, and (2) storage of correct, this magnetosome battery hypothesis intracellular electron donor and acceptor would explain why magnetotactic bacteria live reserves as a buffer against external variability. predominantly near sharp redox gradients and Thioploca, for instance, forms elevator-like why some bacteria (e.g., Spring et al., 1993) sheaths many centimeters long, within which produce many more magnetosomes than are filaments swim up and down (Jørgensen and necessary for magnetotaxis. It also suggests Gallardo, 1999). The sheaths provide it with that magnetotaxis might have been an direct paths between electron-donor rich and evolutionary exaptation, a metabolic pathway electron-acceptor rich environments. adapted for sensitivity to the geomagnetic field Thioploca can accumulate nitrate in storage after its initial evolution. vacuoles at concentrations as high as 0.5 M when at the more oxidized end of their shafts, 3. Traits and identification techniques for then use the nitrate to oxidize sulfide when at magnetotactic bacteria the more reduced end. Conversely, they can also partially oxidize sulfide to form elemental sulfur granules as an electron donor reserve.

4 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

Multidomain Multidomain

300 Metastable SD Elongate Irregular

) Cuboidal Single Domain 100 Stable SD Elongate Prismatic Length (nm

30 Superparamagnetic Superparamagnetic

0 0.2 0.4 0.6 0.8 1 Shape Factor (Width/Length)

Figure 2: Single domain stability field of magnetite as a function of shape factor (width/length ratio) and length. Diagrams on the right schematically represent the arrangement of magnetic moments in two domain, single domain, and superparamagnetic particles. The lower, SD/superparamagnetic boundary is determined for rectangular parallelpipeds with unblocking times of 100 s (dashed line) and 4.5 Gy (solid line) following Butler and Banerjee (1975) and Diaz-Ricci and Kirschvink (1992). Because the calculation ignores magnetocrystalline anisotropy, it overestimates the minimum SD length at shape factors close to 1. The dashed upper boundary of the stable SD field and the solid upper boundary of the metastable SD field are taken from the micromagnetic models of Witt et al. (2005) for characteristic magnetosome crystal shapes. Shaded regions mark size and shape of crystals from magnetotactic bacteria (Arató et al., 2005; Bazylinski et al., 1995; Devouard et al., 1998; Farina et al., 1994; Meldrum et al., 1993a; Meldrum et al., 1993b; Moench, 1988; Sakaguchi et al., 1993; Thornhill et al., 1994; Vali and Kirschvink, 1991).

3.1. The fingerprint of natural selection magnetic moments are strong enough that the In the modern world, Fe is often a magnetic aligning effect dominates thermal scarce nutrient, its availability limited by its agitation, selection should favor traits that insolubility under oxic conditions. Because maximize the moment produced per atom of natural selection has not eliminated Fe used. magnetosome production, magnetotactic The most fundamental characteristic of bacteria must gain some adaptive advantage biologically controlled mineralization is that it from the sequestration of this precious resource occurs under biological control. Magnetosome in magnetosomes. Natural selection should crystals are produced within membrane-bound therefore favor traits that maximize the vesicles in a process orchestrated by a efficiency with which the bacteria employ Fe. mechanism that genetic studies are just starting Since magnetotaxis requires that the cells’ to reveal (Bazylinski and Frankel, 2004;

5 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

Arató et al., 2005; Devouard et al., 1998; Kirschvink and Lowenstam, 1979). Almost all magnetotactic bacteria produce crystals that, at Skewed Gaussian least in the arrangement that naturally occurs in the cell, act as single domain (SD) particles (Figure 2). While the moments of smaller, superparamagnetic (SP) particles are buffeted N by thermal noise and those of larger, Gaussian multidomain (MD) particles are reduced by the formation of domains with moments aligned in different directions, in single domain particles Log!Normal the entire crystal contributes to producing a stable net magnetization. Natural selection 0 100 200 would predict such an outcome in organisms Length (nm) that are capable of controlling the Figure 3: Schematic representation of particle size microenvironment in which the crystals form distributions. Most magnetotactic bacteria produce skewed Gaussian distributions, although some produce and that gain a selective advantage from their Gaussian distributions. Open system chemical growth magnetic moment. processes give rise to log-normal distributions. Whereas open system chemical growth processes lead to log-normal size distributions, Komeili et al., 2006; Komeili et al., 2004; magnetotactic bacteria produce distributions Matsunaga and Okamura, 2003; Scheffel et al., with sharper cutoffs at larger sizes. Cultured 2006). Biochemical regulation allows bacteria generally produce negatively skewed magnetotactic bacteria to produce crystals with crystal size distributions, although a few three broad categories of adaptive traits: (1) bacteria produce Gaussian size distributions narrow size and shape distributions, (2) chain (Arató et al., 2005; Devouard et al., 1998; arrangement, and (3) chemical purity and Pósfai et al., 2001) (Figure 3). crystallographic pefection. Several aspects of Common magnetosome magnetite these traits were previously discussed by morphologies include equidimensional cubo- Thomas-Keprta et al. (2000). Some of these octahedra, elongate hexaoctahedral prisms, and traits can be partially assessed at a bulk level irregular and elongate tooth, bullet (Thornhill through the techniques of rock magnetism and et al., 1994), and arrowhead (Bazylinski et al., ferromagnetic resonance (FMR) spectroscopy 1995) shapes (Figure 4), while common (e.g., Kopp et al., 2006b), while others traits magnetosome greigite morphologies include require detailed electron microscopy to equidimensonal cubo-octahdra and elongate identify. rectangular prisms (Bazylinski et al., 1994). (Cubo-octahedra are frequently labeled as “cuboidal” and elongate hexaoctahedra are 3.2. Size and Shape Distributions frequently labeled as “prismatic,” a convention we adopt for linguistic simplicity. In the Magnetosome crystals typically exhibit literature, where crystals have not been species-specific, narrow distributions of size examined closely, cubo-octahedra have and shape factor (length/width ratio) (e.g. sometimes been identified as cubic or

6 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

Figure 4: TEM images of magnetite-producing magnetotactic bacteria exhibiting different magnetosome crystal morphologies. (a) Cubooctahedral magnetite in laboratory-grown Magnetospirillum magnetotacticum MS-1, (b) hexaoctahedral prismatic magnetite from bacteria living in the sediments of Lake Ammersee, Gernany, (c-d) irregular elongate magnetite from bacteria living in the sediments of Lake Chiemsee, Germany. Scale bar is 250 nm in all images. Images courtesy A. Kobayashi (a) and H. Vali (b-d). octahedral.) The characteristic magnetosome Ricci and Kirschvink, 1992; Kopp et al., crystal shapes have less sharp edges than do 2006a). In equidimensional magnetite and equidimensional octahedral or rectangular greigite particles, magnetic anisotropy and thus parallelepipeds. This ‘rounding’ reduces the coercivity are controlled by the inherent outward warping of magnetization at the magnetocrystalline anisotropy. In crystal ends, which produces magnetic stochiometric magnetite, magnetocrytalline ‘flower’ structures in straight-edged particles anisotropy leads to room-temperature bulk (Kirschvink, 2001; Witt et al., 2005). It thereby coercivities of ~15 mT. Increasing the extends the size to which single-domain anisotropy of a particle increases coercivity crystals can grow (Witt et al., 2005). Because and permits larger particles to remain within the bending of magnetization in flower the single domain field. Shape anisotropy structures reduces net crystal magnetization, increasingly governs magnetic anisotropy for the elimination of sharp edges also directly crystals with width-to-length ratios less than increases the magnetic moment per iron atom ~0.9. A 50 nm long magnetite particle with a (Kirschvink, 2001). width-to-length ratio of 0.5, for instance, has a Magnetic anisotropy energy causes the room-temperature bulk coercivity of ~50 mT. magnetization of a domain to align At 0 K, an infinitely long magnetite rod would preferentially in certain crystallographic have a bulk coercivity of ~150 mT, the directions. In SD particles, it therefore maximum possible for magnetite. controls the particle’s coercivity, the field In magnetotactic bacteria with elongate required to remagnetize the particle (Diaz- magnetite particles, the crystals are typically

7 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria elongated along a [111] axis (but for 1995; Heywood et al., 1990; Pósfai et al., exceptions see Mann et al., 1987; Taylor and 1998), fueling speculation that [100] axes are Barry, 2004; Taylor et al., 2001; Vali and the greigite magnetocrystalline easy axes Kirschvink, 1991). Because [111] axes are the (Bazylinski and Moskowitz, 1997). magnetocrystalline easy axes, this choice of The most direct method of assessing axis causes the magnetocrystalline anisotropy size and shape factor distributions, but also the to enhance the shape anisotropy produced by most labor-intensive and subject to sampling elongation. Greigite crystals are often biases, involves direct measurement of elongated along a [100] axis (Bazylinski et al., particles in magnetic extracts under

AMB-1 mnm18 AMB-1 mnm18 (equidimensional, isolated)

AMB-1 AMB-1 (equidimensional, in chains)

MV-1 MV-1 (elongate, in chains) Derivative of IRM Derivative of Absorption

Wilson Creek Wilson Creek (detrital)

10 100 1000 0 200 400 600 Field (mT) Field (mT)

Figure 5: Example coercivity spectra (left) and FMR spectra (right). From top to bottom, spectra are shown for AMB-1 mutant mnm18 (which predominantly produces isolated particles of cuboidal magnetite), wildtype AMB-1 (which produces chains of cuboidal magnetite), MV-1 (which produces chains of elongate prismatic magnetite), and detrital magnetic particles from the Wilson Creek Formation of Mono Basin, California. Comparison of the wildtype AMB-1 and MV-1 coercivity spectra demonstrates the effect of elongation on particle coercivity, while comparison of the bacterial and detrital coercivity spectra illustrates the narrower coercivity distribution of biologically controlled magnetite. Whereas magnetocrystalline anisotropy causes the FMR spectrum of mnm18 to be slightly asymmetric in the high field direction, the stronger anisotropy produced by particle elongation and chain arrangement causes the spectrum of MV-1 to be much broader and strongly low-field extended. Wildtype AMB-1, with anisotropy dominated by chain arrangement, is more mildly low-field extended. The detrital magnetic particles of the Wilson Creek sediments, being more heterogeneous in size, shape, and arrangement, produce an extremely broad but fairly symmetric spectrum.

8 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria transmission electron microscopy (TEM). A bacteria and having few branch points. If the variety of magnetic techniques allow the particle size distribution of the magnetic construction of coercivity spectra of samples, crystals is known, sampling statistics can which are functions of both size and shape. assess the probability that the crystals within a The characteristically narrow distributions of chain are not a random sub-sample of the biologically controlled magnetic minerals can distribution. Chains with these traits are highly sometimes be observed in the dispersion of the suggestive of a biological origin. Techniques coercivity (Egli, 2004) (Figure 5). Anisotropy for identifying these chains are therefore key to fields play a major role in controlling the width developing the fossil record of magnetotactic and asymmetry of FMR spectra, which can be bacteria. used to identify elongate magnetic particles as TEM imaging of magnetic extracts is well as the narrow distributions characteristic currently the only technique for producing of biological control (Kopp et al., 2006a; Kopp high-resolution images of chains. Magnetic et al., 2006b) (Figure 5). extraction, however, involves disrupting the matrix containing the magnetic particles, and the extraction process provides a good 3.3. Chain Arrangement mechanism for physically forming strings of In most magnetotactic bacteria, the particles. Thus, short chains found in extracts magnetosomes are maintained aligned in one are ambiguous as to their origins; only long or more chains by a cytoskeletal structure chains with characteristic magnetofossil traits (Komeili et al., 2006), an organic sheath are indicative of biological origin. Scanning (Kobayashi et al., 2006), and anchoring electron microscopy (SEM) permits imaging proteins (Scheffel et al., 2006). The chain axis particles in situ (Friedmann et al., 2001; Maher is typically aligned with the easy axes of the et al., 1999) but lacks the resolution needed to individual particles (Dunin-Borkowski et al., characterize particle shape and so is only 2001). The chain serves the same basic useful for magnetofossil identification when physical function as particle elongation: it combined with TEM of extracts (e.g., Figure 6, increases the stability of the state in which c-d). particle moments are aligned along the chain For unoxidized magnetite, the axis by enhancing magnetic anisotropy. Moskowitz test, which compares the thermal Whereas mutant AMB-1 producing isolated demagnetization behavior of low temperature particles of nearly equidimensional magnetite saturation remanence magnetizations acquired has a room-temperature bulk coercivity of ~13 after cooling in zero field and that acquired mT, AMB-1 producing short chains of after cooling in a strong field, can also indicate equidimensional particles has a room- the presence of chains (Moskowitz et al., 1993; temperature coercivity of ~25 mT (Kopp et al., Weiss et al., 2004b). At the Verwey transition, 2006a). which occurs at 125 K in stochiometric Linear strings of magnetic particles can magnetite, magnetite shifts from having cubic also be produced by physical processes (Kopp symmetry at higher temperatures to having et al., 2006b). Biologically produced chains uniaxial symmetry at lower temperatures. are distinguished by being composed of Magnetite chains exhibit relatively greater particles from a size and shape factor demagnetization of the field-cooled remanence distribution characteristic of magnetotactic upon warming through the Verwey transition

9 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria than do other arrangements of magnetite. The purity can be assessed most accurately by effect likely results from the influence of the analytical techniques coupled to TEM, such as chain structure on the selection of an energy dispersive X-ray spectroscopy (EDS), elongation axis during cooling below the but can also be assessed in magnetite at a bulk Verwey temperature (Moskowitz et al., 1993). level from shifts in the Verwey transition However, the physics underlying this temperature and Néel temperature. However, observation (Carter-Stiglitz et al., 2002; Carter- these transition temperatures are also affected Stiglitz et al., 2004) is only partially by diagenetic oxidation. understood, and the test is subject to false With the exception of twinning along negatives generated by limited particle the [111] easy axis, crystallographic defects oxidation. also reduce the magnetic moment of magnetite Ferromagnetic resonance spectroscopy particles. Thus, crystallographic defects are is sensitive to the magnetic anisotropy rare in magnetosome magnetite (Devouard et produced by particle chains as well as the al., 1998). The absence of such defects can be homogeneity that distinguishes biological assessed only by high-resolution TEM. chains from physical strings (Kopp et al., Greigite magnetosome crystals are 2006a; Kopp et al., 2006b) (Figure 5). It is often less strictly controlled than magnetite rapid and insensitive to particle oxidation, and c r y s t a l s , b o t h c h e m i c a l l y a n d thus is currently the best bulk technique crystallographically. Some greigite-producers capable of screening samples for the presence can incorporate up to ~10 atomic percent Cu of likely magnetofossil chains. into their magnetosome crystals (Bazylinski et al., 1993; Pósfai et al., 1998). Greigite magnetosome crystals also commonly exhibit 3.4. Chemical Purity and Crystallographic planar defects along (222)-type planes, Perfection believed to be associated with the conversion of mackinawite into greigite (Pósfai et al., In general, magnetite produced by 1998). These differences suggest that greigite magnetotactic bacteria is nearly pure iron precipitation by magnetotactic bacteria may be oxide, with concentrations of trace elements less regulated than magnetite precipitation and like Ti, Al, and Cr significantly lower than in could hinder identification of greigite most abiotic magnetite (Thomas-Keprta et al., magnetofossils. 2000). Such purity is expected based both on selection for efficiency in the use of Fe, as trace elements reduce the magnetic moment of 3.5. Scoring magnetofossil identifications magnetite particles, and on the extensive use of pure iron as a specific metal cofactor in We suggest the following scheme for numerous enzymatic systems. Some bacteria rating possible magnetofossils: do produce magnetite that is slightly oxidized; Context and Robustness: Magnetofossil the Verwey transition can be reduced from the identifications are more reliable if the samples 125 K of stochiometric magnetite to under consideration come from an understood temperatures as low as ~100 K, which stratigraphic, geochemical, and paleomagnetic indicates up to 0.4% cation depletion (Kopp et context or were collected as part of a broader al., 2006a; Moskowitz et al., 1993). Chemical study aimed at understanding this context.

10 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

Magnetofossil reports from localities lacking Table 1: Proposed magnetofossil robustness criteria robust paleomagnetic data should be viewed Context Criteria cautiously, as they are more likely to have Environment analogous to S ! 3; or undergone diagenetic or metamorphic younger magnetofossil- S ! 2 and C ! 3; or processes that altered primary magnetic bearing environments; S ! 2 and C ! 2 and ChP carriers, including any possible magnetofossils. Paleomagnetic data robust Our suggested criteria for considering a Environment analogous to S ! 3 and ChP; or magnetofossil identification to be robust are younger magnetofossil- S ! 3 and C ! 3; or given in Table 1. Table 2 lists reports of pre- bearing environments; S ! 3 and C ! 2 and ChP Paleomagnetic data not Quaternary magnetofossils from samples robust collected with contextual information, while Environment analogous S = 4 and ChP; or Table 3 lists reports of pre-Quaternary to younger S ! 3 and C !3 and ChP; magnetofossils from grab samples lacking such magnetofossil-bearing or information. For samples where paleomagnetic environment; Sediments S !3 and C ! 2 and ChP have undergone burial and CrP information is available, we also report a metamorphism or paleomagnetic quality (PQ) score follow the paleomagnetic data criterion of van der Voo (1990). PQ ranges remagnetized from zero to seven, where samples with PQ ! 4 Unique environment S = 4 and C ! 3 and ChP are considered robust. For demonstrably and CrP remagnetized units, PQ is listed as “*”. Single domain (criterion SD): As a indicates the absence of any evidence for basic requirement, all claims of magnetofossils chains, while one indicates that either SEM or should be supported by magnetic or electron low-temperature thermal demagnetization microscopy evidence indicating the presence indicates the presence of chains. Two indicates of a significant amount of single domain either (a) that FMR indicates the presence of magnetite, maghemite, or greigite. chains or (b) that short chains of ambiguous Size and shape (score S): Beyond the origin were imaged in the TEM of magnetic basic single domain criterion, we score the size extracts. Three indicates either (a) the TEM and shape of particles in a sample based on (1) identification of short chains and FMR data coercivity or FMR spectra indicating narrow indicating that they are a significant, in situ distributions of size and shape, (2) TEM component of the sample, or (b) the TEM evidence for SD particles with truncated edges identification of long chains in magnetic (cubo-octahedral or hexa-octahedral extracts. Four indicates the TEM identification morphologies, for example), (3) TEM evidence of long chains in magnetic extracts combined for elongate SD particles (such as with SEM or FMR evidence confirming that hexaoctahedral and irregular elongate the chains occur in situ. particles), and (4) statistical TEM evidence for Chemical perfection (criterion ChP): SD populations with narrow size and shape Criterion ChP reflects whether the particles distributions. Each of these four lines of relatively pure and, in particular, Ti-free. evidence contributes one point to score S, Among other techniques, EDS and low- which ranges from zero to four. temperature magnetometry can assess purity. Chains (score C): We grade the quality Crystallographic perfection (criterion of chain identification from zero to four. Zero CrP): Criterion CrP reflects whether high

11 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

Figure 6: TEM images of magnetofossils. (a) Quaternary magnetofossils in a magnetic extract from nannofossil ooze from Ocean Drilling Project Hole 1006D, Santaren Channel, west of the Great Bahama Bank. Cuboidal, prismatic, and bullet morphologies are visible (scale bar = 200 nm). (b) Prismatic Miocene magnetofossils from clay-rich sediments, DSDP Leg 73, Angola Basin, South Atlantic Ocean (scale bar = 200 nm). (c-d) Cretaceous magnetofossils from chalk beds at Culver Cliff, United Kingdom. (c) shows prismatic and irregular elongate magnetosome morphologies in a magnetic extract from Culver Cliff chalk, with some intact chains of prismatic crystal (scale bar = 100 nm), while (d) shows intact magnetite chains imaged in situ with SEM (scale bar = 500 nm). (e) Putative 1.9 Ga magnetofossils from the Gunflint Formation, Ontario, Canada (scale bar = 100 nm). (Images courtesy M. Hunslow (a, c-d) and H. Vali (b). (e) reproduced from Chang (1988).) resolution TEM indicates the absence of and Maher, 1996; Maher et al., 1999; crystallographic defects other than twinning Montgomery et al., 1998). These around the magnetic easy axis. magnetofossils were identified in the course of a broad paleomagnetic study that yielded robust results, with a paleomagnetic quality 3.6. Example magnetofossil scores index of 5. Rock magnetic experiments revealed that the magnetic carrier in these As an example application of this samples is a low coercivity phase, such as scoring system, consider the magnetofossils magnetite, and TEM confirmed that the from the Cretaceous chalk deposits of southern dominant carrier is SD magnetite. Thus, the England (Table 2 and Figure 6, c-d) (Hounslow samples pass the SD test.

12 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

The TEM images show predominantly 4. Fossil record of magnetotactic bacteria hexaoctahedral prisms and lesser quantities of cubo-octahedral and bullet-shaped crystals. 4.1. Phanerozoic Magnetofossils However, no coercivity or FMR spectra are available, nor have size and shape distributions Magnetofossils are a common been compiled from TEM images. Thus, the contributor to sedimentary magnetism in a samples earn points for particles with truncated variety of Quaternary environments (e.g., edges and for elongate SD particles, yielding a Figure 6a). Magnetite magnetofossils, often S score of 2. TEM images of magnetic extracts partially oxidized to maghemite, have been show particles in chains, while SEM images found as a major magnetic component of confirm that these chains occur in situ, yielding lacustrine (Dearing et al., 1998; Kim et al., a C score of 4. No analyses have been 2005; Oldfield et al., 2003; Pan et al., 2005; performed to test the chemical purity or Peck and King, ; Snowball, 1994), microbial crystallographic perfection of the magnetite mat (Stolz et al., 1989), hemipelagic (Dinares- particles. Turell et al., 2003; Housen and Moskowitz, Because the Cretaceous sediments yield 2006; Stolz et al., 1986), pelagic (Hesse, 1994; robust paleomagnetic data, we would test the Hilgenfeldt, 2000; Lean and McCave, 1998; magnetofossils against the first set of criteria in Petersen et al., 1986; Yamazaki and Ioka, Table 1. With a S score of 2 and a C score of 4, 1997; Yamazaki et al., 1991) and carbonate we would therefore judge these magnetofossils platform (Maloof et al., 2007; McNeill, 1990; to be robust. McNeill et al., 1988; Sakai and Jige, 2006) As another example, consider the sediments. Greigite magnetofossils have not putative magnetofossils from the yet been identified in Quaternary sediments. Paleoproterozoic Gunflint Formation of The pre-Quaternary magnetofossil Canada (Table 3 and Figure 6e). These samples record, in contrast, is sparse; more are not placed within stratigraphic context at a magnetofossil-bearing localities have been resolution finer than the formation level. Rock identified in the Quaternary than in all the rest magnetic data indicate the presence of of Earth history. The most robust pre- significant SD magnetite, an observation Quaternary identifications come from confirmed by TEM. The samples thus pass the Mesozoic and Cenozoic sediments, although SD test. The SD particles present are crudely possible magnetofossils have also been found cubo-octahedral in shape, although they appear in rocks as old as the late Archean (Tables 2, to be corroded; one might generously grant a S 3). To date, pre-Quaternary magnetofossils score of 1. TEM of magnetic extracts reveal have been identified in carbonate platform short chains of ambiguous origin, yielding a C sediments, basinal sediments, and continental score of 2. No analyses have tested the shelf deposits. chemical purity or crystallographic perfection of the crystals. Regardless of the leniency with Carbonate platform and atoll sediments. In which one picks a set of robustness criteria weakly magnetic Quaternary carbonate from Table 1, these magnetofossils are not sediments, magnetofossils are frequently the robust. primary carrier of syndepositional remanence magnetization, and the limited data available

13 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria suggest the same is true of ancient carbonate coercivity data indicate that biogenic SD sediments. magnetite is the main remanent magnetization Magnetofossils have been robustly carrier (Belkaaloul and Aissaoui, 1997). identified in Pliocene platform and atoll Lower Cambrian (Tommotian) sediments. In Pliocene to Recent limestones limestones of the Pestrotsvet Formation, and dolstones from San Salvador Island, Siberia Platform, similarly contain plausible Bahamas, TEM investigation of magnetic magnetofossils that marginally fail to meet our extracts reveals that the dominant magnetic robustness criteria. The sediments are well particles are cuboidal and bullet-shaped SD preserved, having experienced no magnetite crystals, sometimes arranged in metamorphism, and provide robust short chains (McNeill et al., 1988). Similarly, paleomagnetic data (Kirschvink and Rozanov, in a core of Pliocene limestones and dolostones 1984). Their magnetic properties reflect a of the Mururora atoll, Aissaoui et al. (1990) dominant role for SD magnetite (Chang et al., found predominantly cuboidal and prismatic 1987), and TEM investigation reveals that the SD magnetite particles, mixed with fragments magnetite particles occur in cuboidal and of lithogenic, multidomain titanomagnetite. prismatic shapes. At least in extract, the Coercivity spectra suggested that the single magnetite crystals occasionally form short domain particles dominate the samples' chains. magnetic properties. Plausible Jurassic (Bathonian- Basinal sediments. Most robust reports of Sinemurien) magnetofossils that marginally Phanerozoic magnetofossils come from basinal fail to meet our robustness criteria have been sediments, predominantly unlithified sediments found in carbonate platform sediments of the from marine drill cores but also including Paris and Jura Basins (Belkaaloul and lithified sediments exposed on land. Aissaoui, 1997; Vali et al., 1987). In both In Pliocene to Recent clays from the basins, single domain magnetite, which North Pacific, deposited below the lysocline at constitutes a minor component (~15% in the depths in excess of 5000 m, Yamazaki and Ioka Paris Basin; "low concentrations" in the Jura) (1997) found that the magnetic mineralogy was of the bulk magnetic mineralogy, is present in a mixture of SD, SP, and PSD magnetite, as cuboidal and prismatic shapes. Magnetic well as maghemite and hematite. TEM images extracts from the Jura Basin carbonates contain revealed that the SD fraction consisted of short chains of SD particles (Vali et al., 1987), cuboidal and prismatic crystals, sometimes while the Paris Basin extracts contain pairs and aligned in chains, while the larger grains were clusters of particles (Belkaaloul and Aissaoui, irregularly shaped. Yamazaki and Ioka 1997). The Jura samples were collected identified the SD particles as magnetofossils without paleomagnetic data, and Vali et al. and proposed an aeolian origin for the larger (1987) do not report any stratigraphic grains. information that would allow their samples to Yamazaki et al. (1991) examined early be tied to other samples from the well-studied Miocene siliceous sediments from similar Northern Calcareous Alps. The Paris Basin depths in the central equatorial Pacific Ocean. carbonates carry robust paleomagnetic data, Rock magnetic measurements indicated that and despite the dominance by volume of SD magnetite was the dominant stable detrital and authigenic MD magnetite, magnetic component of the sediments, and

14 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

, 1996; ce amazaki and Ioka, amazaki et al., ali and Kirschvink, Sour (Y 1997) (McNeill et al., 1988) (Aissaoui et al., 1990) (Chang and Kirschvink, 1985) (Butler et al., 1999) (Y 1991) (Kirschvink and Chang, 1984; Petersen et al., 1986; V 1989) (Kent et al., 2003; Kopp et al., 2007; Lippert and Zachos, 2007) (Hounslow and Maher Montgomery et al., 1998) (Belkaaloul and Aissaoui, 1997) (Chang et al., 1987) 1-33 Sed Rate (m/My) 0.4-2.7 1 12-66 ~40 ~30 2-6 2-9 ~25-60 30-40 ~30 5-50 Lithified N Y Y Y Y N N Y Y Y Y

Depth > 5000 m supratidal to euphotic euphotic ~600-900 m supratidal to euphotic 5200-5600 m 3800-4600 m euphotic sub-euphotic shelf euphotic euphotic , marl, and calcareous ooze Lithology clay limestone and dolostone limestone and dolostone claystone limestone and evaporite and clay clay claystone limestone (chalk) limestone limestone Setting pelagic basinal carbonate platform carbonate atoll hemipelagic basinal hemipelagic shelf pelagic basinal pelagic basinal hemipelagic shelf pelagic shelf carbonate platform carbonate platform Leg , Crete , , Siberia Atoll, incentown Fm., North Pacific San Salvador Bahamas Mururoa French Polynesia Potamida Clay Calcare di Base, Sicily Central Equatorial Pacific Angola Basin, South Atlantic (DSDP 73) V Atlantic Coastal Plain, New Jersey Southern England Paris Basin, France Pestrotsvet Formation, Labaia Lena River Locality Age Pliocene Pliocene Pliocene Miocene Miocene Miocene Eocene to Quaternary Eocene (PETM) Cretaceous Jurassic Cambrian CrP nd nd nd nd nd nd + nd nd nd nd ChP nd nd nd nd nd nd + + nd + nd " C 2 2 0 0 0 0 4 3 4 0 2 4 2 2 2 S 2 3 3 3 " 4 4 + + + + + SD + + + + + + able 2: Magnetofossil reports from pre-Pleistocene localities with stratigraphic context 5 5 5 4 5/* nd 5 T PQ 4 5 5 5

15 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

TEM examination revealed crystals with work on the Potamida clays found that the cuboidal, prismatic, and bullet shapes intensity of remanent magnetization decreases characteristic of magnetofossils. in sediments deposited during magnetic Several studies have examined reversals (Valet and Laj, 1981). This led magnetofossils in sediments of the Angola Kirschvink (1982) to speculate that Basin, South Atlantic Ocean, collected during magnetofossils carry the magnetization in Deep Sea Drilling Project (DSDP) Leg 73 at these sediments and that the decreased Sites 519, 521, 522, and 523 (Figure 6b) magnetization might result from conditions (Chang and Kirschvink, 1984; Kopp et al., during a reversal that disfavor large 2006b; Petersen et al., 1986; Vali and populations of magnetotactic bacteria. Chang Kirschvink, 1989). The sediments range in age and Kirschvink (1985) examined magnetic from Eocene to Quaternary and in composition extracts from the clay under TEM and found from nearly pure carbonate ooze to carbonate- cuboidal and prismatic SD magnetite, which poor clays. Magnetic properties indicate that they interpreted as magnetofossils, as well some sediments have a significant amount of octahedral magnetite that they interpreted as MD titanomagnetite, whereas others are byproducts of bacterial metabolism. dominated by SD magnetite (Petersen et al., Miocene marls from #$ka, Poland, 1986), often partially oxidized (Kirschvink and deposited under brackish conditions in the Chang, 1984). Magnetic mineralogy does not foredeep of the Western Carpathians, contain correlate with lithology (Petersen et al., 1986). greigite particles with a Gaussian size TEM images indicate that SD magnetite occurs distribution resembling that of crystals produce in cuboidal, prismatic, and bullet shapes and is by multicellular magnetotactic prokaryotes, arranged in small and long chains, clumps, and which led Pósfai et al.(2001) to speculate that meshes (Petersen et al., 1986; Vali and these particles were produced by magnetotactic Kirschvink, 1989). Ferromagnetic resonance bacteria. Though inconclusive, this is the only spectroscopy of Oligocene-Miocene sediments report of possible greigite magnetofossils. bearing SD magnetite reveals magnetic properties that are dominated by elongate Continental shelf sediments. Continental shelf magnetite particles or magnetite aligned in situ deposits have been less frequently investigated in chains (Kopp et al., 2006b). While most of than carbonate platform or basinal deposits but these sediments carry stable remanent are the source of the two most robust magnetizations (Tauxe et al., 1984), some clay- identifications of magnetofossils in lithified rich Miocene sediments do not, an observation sediments. Vali and Kirschvink (1989) linked to TEM Kent et al. (2003) examined cores observations of aggregation and partial through Atlantic Coastal Plain sediments of dissolution of magnetofossils. New Jersey that record the Paleocene-Eocene The first reported identifications of Thermal Maximum (PETM), the ~220 ky magnetofossils in consolidated sediments came initial Eocene global warming event (Rohl et from the Miocene (Tortonian-Messinian) al., 2000). Sediments that precede and follow Potamida clays of Crete, which were deposited the PETM in the cores are glauconitic silts and in the Kastelli sub-basin at depths of ~600-900 have hysteresis properties suggesting a m (Meulenkamp, 1979; van Hinsbergen and dominant magnetic contribution from Meulenkamp, 2006). Initial paleomagnetic lithogenic input. In contrast, PETM sediments

16 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria are kaolinite-rich clays, ~3-10 times more The oldest robust magnetofossils yet magnetic than underlying and overlying found come from Cretaceous (Coniacian- sediments, and have hysteresis properties Campanian) chalk deposits of southern suggesting that their magnetic mineralogy is England (Hounslow and Maher, 1996; Maher dominated by single domain magnetite. The et al., 1999; Montgomery et al., 1998). These anomalous magnetic properties do not appear sediments have been the subject of to be lithologically controlled, because they paleomagnetic, rock magnetic, SEM, and TEM persist for the duration of the PETM despite studies. Rock magnetic properties indicate that kaolinite concentrations that range in the most sediment magnetization is dominated by SD proximal core between ~8% and ~50%. Kent magnetite, occasionally with some hematite. et al.’s TEM examination of clay suspensions The most abundant magnetite particles are found only isolated, equidimensional particles prismatic, but cuboidal and bullet shapes, as of single domain magnetite. They therefore well as other elongate irregular morphologies, concluded that magnetotactic bacteria had not are also present. The crystals often occur in produced the particles and instead proposed long chains, which have been imaged both in that the particles might have been produced as magnetic extracts examined under TEM and in condensates from an impact ejecta plume. situ adhering to clay particles under SEM Kopp et al. (2007) re-examined one of (Figure 6, c-d). the cores through PETM sediments with FMR Though not a robust magnetofossil to check Kent et al.’s observation that the identification, Butler et al. (1999) found that m a g n e t i t e o c c u r r e d a s i s o l a t e d , the magnetization in the microbial limestones equidimensional particles. Whereas FMR and evaporites of the Miocene (Messinian) properties of silty clays deposited above and Calcare di Base, Sicily, was carried by cubic below the PETM clay indicate a predominantly sub-micron magnetite crystals. They detrital magnetic mineralogy, the properties of interpreted these particles as magnetosome the PETM clay indicate elongate and/or chain crystals but did not provide detailed magnetite of biogenic origin. Contrary to Kent descriptions or images. et al.’s TEM imaging of clay suspensions, TEM imaging of magnetic extracts reveals the presence of abundant cuboidal, prismatic, and 4.2. Putative Precambrian Magnetofossils bullet-shaped SD magnetite particles, occasionally in chains. An independent rock A l l p u t a t i v e P r e c a m b r i a n magnetic and TEM study of another core by magnetofossils come from carbonate platform Lippert and Zachos (2007) confirms this result. sediments, mostly stromatolitic limestones and The findings undermine the evidence for an cherts. This bias is a product of sample initial Eocene cometary impact but indicate selection and does not necessarily reflect the that the Atlantic margin PETM sediments are a distribution of magnetofossils. The samples rich magnetofossil deposit. The Kopp et al. studied are all hand samples from outcrops, result is the first FMR-driven discovery of and the findings are reported with limited magnetofossils and confirms the utility of contextual information. No Precambrian FMR as a technique for detecting magnetofossils meet our robustness criteria. magnetofossils in ancient sediments. Chang and Kirschvink (1989) found that limestones from the Ediacaran Nama

17 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

Table 3: Magnetofossil reports from pre-Pleistocene localities with limited or no stratigraphic context

SD S C ChP CrP Age Locality Setting Lithology Source Northern Calcareous Alps, carbonate (Vali et al., + 2 2 nd nd Jurassic Austria platform limestone 1987) (Chang and Nama Group, carbonate Kirschvink, + 1 2 nd nd Ediacaran Namibia platform limestone 1989) Bitter Springs Fm., South carbonate (Chang et al., + 2 2 nd nd Neoproterozoic Australia platform stromatolitic chert 1989) Vempalle Fm., carbonate (Chang et al., + 2 0 nd nd Mesoproterozoic India platform stromatolitic chert 1989) Gunflint Fm., carbonate (Chang et al., + 1 2 nd nd Paleoproeterozoic Canada platform stromatolitic chert 1989)

Tumbiana Fm., carbonate stromatolitic (Akai et al., nd 1 nd nd nd Late Archean Western Australia platform limestone 1997)

Noachian? Mars? carbonate globules many authors + 2.5 1 + + Holocene? ALH84001 Antarctica? in basalt (see text)

Group, Namibia, had coercivity spectra Paleoproterozoic Gunflint Formation, Canada, dominated by SD magnetite. TEM images of contain a mixture of SD and MD magnetite, magnetic extracts show cuboidal magnetite which appears in somewhat corroded cuboidal that is sometimes arranged in short chains. As shapes, sometimes in short chains (Figure 6e). with all rocks bearing putative Precambrian Akai et al. (1997) conducted no rock magnetofossils, no paleomagnetic data was magnetic investigations but, based on TEM collected from the Nama Group samples, but images alone, identified ~30-50 nm long, Meert et al. (1997) found that some parts of the elongated, irregular magnetite particles in the Nama Group had been largely remagnetized in acid-insoluble residue of recrystalized and the Paleozoic. partially silicified stromatolites from the late Chang et al. (1989) searched for Archean Meentheena Member, Tumbiana magnetofossils in a number of silificied Formation, Fortescue Group, Western Proterozoic stromatolites. The magnetic Australia. They suggested that these particles properties of samples from the Neoproterozoic were produced by magnetotactic bacteria, Bitter Springs Formation, Adelaide Basin, although they did not identify chains, examine South Australia, indicate a mixture of SD and crystal morphology closely, or report MD magnetite. TEM investigation shows that population statistics. Irregular elongate crystals SD magnetite occurred in octahedral and might alternatively be produced by cuboidal shapes; no chains were found. metamorphic alteration of siderite, as has been Samples from the Mesoproterozoic Vempalle observed in drill core from lower greenschist Formation, Cuddapah Basin, India, contain grade late Archean siderite of the Ghaap predominantly SD magnetite, which occurred Group, Transvaal Supergroup, South Africa in elongate prismatic shape, but again no (Tikoo et al., in prep.). chains were found. Samples from the

18 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

4.3. Putative extraterrestrial magnetofossils (2003) have raised concerns that, even if magnetofossils are present in association with The oldest and most controversial the carbonate globules, they could have been putative magnetofossils come from what is deposited during the period the meteorite was almost certainly the most intensively studied exposed to surface conditions in Antarctica. hand sample in the history of the geological However, magnetotactic bacteria have not yet sciences, the ~4.5 Ga Martian meteorite been identified in endolithic or psychrophilic ALH84001 (Buseck et al., 2001; Friedmann et communities, either in Antarctica or on Mars. al., 2001; Golden et al., 2004; McKay et al., According to the robustness criteria 1996; Thomas-Keprta et al., 2000; Thomas- proposed in this article, the ALH84001 Keprta et al., 2001; Thomas-Keprta et al., magnetofossils are not robust, and it is unlikely 2002; Weiss et al., 2004b and many others). A that they will ever be accepted as robust by complete treatment of the debate over the most researchers. The challenges posed by biogenicity of magnetite particles in attempting to identify magnetofossils in a ALH84001 requires its own, full-length review sample that has been stripped of geological and is therefore beyond the scope of this context and is not analogous to any known article. Briefly, the magnetic minerals magnetofossil-bearing sample are great. To associated with carbonate globules in the overcome them would require unambiguous basaltic meteorite include stochiometric SP and identification of populations of single domain SD magnetite, as well as pyrrhotite. There are particles with distinctive bacterial several different morphological populations of morphologies and narrow size and shape magnetite crystals. ~27% of the magnetite distributions, the identification with TEM of crystals appear chemically pure and assume long chains of particles in magnetic extracts elongate prismatic shape, ~7% are whiskers with biogenic morphologies, and an extensive with typical inorganic trace impurities, and the search for potentially contaminating remainder take a variety of other shapes. The magnetotactic bacteria populations in arguments about biogenicity focus on the Antarctica. The controversy will most likely be prismatic crystals, which Thomas-Keprta et al. resolved only with study of fresh samples (2000) compare to crystals from marine collected in context on Mars. magnetotactic vibrio strain MV-1. Golden et al. (2004) argue that the elongate prismatic 5. Magnetofossil Taphonomy crystals from ALH84001 are distinct from those produced by MV-1 and more closely resemble those produced by thermal decomposition of Fe-rich carbonate, a claim 5.1. Taphonomy of Magnetite Magnetofossils disputed by Thomas-Keprta et al. (2004). After a magnetotactic bacterium dies, Friedmann et al. (2001) attempted to identify magnetite particles it produced can suffer three in situ magnetite chains using SEM with fates during early diagenesis that affect their limited success, while Weiss et al. (2004b) utility as fossils: oxidation, chain breakup, and used low-temperature magnetometry and FMR reductive dissolution (Figure 1). Due to spectroscopy to constrain the abundance of any changes in redox conditions during diagenesis, such chains to <~10% of the volume of particles can also experience a combination of magnetite. In addition, Kopp and Humayun these processes. Partial oxidation to

19 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria maghemite, driven by the oxidation and reducing bacteria will dissolve fine magnetite +2 subsequent diffusive loss of Fe from the particles, including magnetofossils. The long- crystal structure (Xu et al., 1997), is the least term preservation of magnetofossils thus destructive fate, as it does not affect the depends on the race between lithification, general morphology of the crystals. While it which should protect magnetofossils, and the suppresses the Verwey transition (Özdemir et metabolic activities of bacteria living in pore- al., 1993), thereby rendering the Moskowitz waters. Although little work has been done chain test (Moskowitz et al., 1993) ineffective, investigating the mechanisms and rates of it does not greatly alter ferromagnetic processes that protect magnetofossils from resonance spectra (Weiss et al., 2004a). dissolution, increased supply of organic carbon Although complete oxidation to hematite or does shift the balance toward the iron-reducing goethite destroys magnetofossils, the process is bacteria and against preservation. thermodynamically disfavored or kinetically inhibited in many common depositional and 5.2. Preservation Potential of Greigite early diagenetic settings. Greigite is a metastable mineral and The breakdown of the proteins and will transform into pyrrhotite or pyrite (Berner, lipids that hold the magnetosome crystals 1967; Morse et al., 1987). Nonetheless, within a cell can affect the arrangement of the greigite can persist for geologically relevant particles in sediments and potentially destroy periods of time in sulfide-limited the characteristic chain structure. Chains of environments. Greigite was first discovered in equidimensional particles, however, will tend Miocene lacustrine sediments of the Tropico to collapse into closed loops that leave the Group, Kramers-Four Corner area, California general head-to-tail (%%) dipole arrangement (Skinner et al., 1964), while the oldest greigite intact (Kobayashi et al., 2006; Kopp et al., described in a published report was formed 2006b). Elongate particles are more during early diagenesis in late Cretaceous susceptible to the loss of linearity and tend to sediments of the North Slope Basin, Alaska collapse into side-by-side dipole arrangement (Reynolds et al., 1994). In a meeting abstract, (⇄) (Kopp et al., 2006a), but can still be Niitsuma et al. (2004) described recognized by their distinctive morphologies. thermomagnetic evidence for greigite from the At high concentrations, both equidimensional 2.77 Ga Archean Mt. Roe shale, recovered in and elongate particles collapse into clumps. an Archean Biosphere Drilling Project core Reductive dissolution is a common fate from Western Australia, but Niitsuma et al. for fine magnetite particles (e.g., Hilgenfeldt, (2005) found no unambiguous evidence of 2000; Housen and Moskowitz, 2006; Maloof et greigite under TEM. al., 2007; Tarduno, 1994; Vali and Kirschvink, 1989). Fe(III) is a lower potential electron acceptor than nitrate but a higher potential 6. Magnetofossils as an archive of electron acceptor than sulfate, so once oxygen environmental change and nitrate are consumed, organisms turn to Fe 6.1. Glacial-Interglacial Variations in (III), including Fe(III) in magnetite (Dong et Magnetofossil Abundance al., 2000; Kostka and Nealson, 1995), as an electron acceptor. If sufficient organic matter is available to fuel microbial metabolism, iron-

20 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

Coincident with the discovery of the growth and that the bacteria producing first sedimentary magnetofossils, Chang and elongate crystals favored lower oxygen levels Kirschvink (1984) proposed that than did bacteria producing equidimensional magnetofossils might serve as a paleooxygen particles. indicator. At the time, all known Based on TEM imaging of deep-sea magnetotactic bacteria were microaerophiles, surface sediments and sediment trap so magnetosome magnetite was interpreted as measurements of organic carbon flux at eight a tracer of the presence of at least small sites in the Pacific Ocean, Yamazaki and quantities of oxygen. The discovery of the Kawahata (1998) similarly found that the ratio strictly anaerobic, sulfate-reducing, magnetite- of equidimensional to elongate magnetofossils producing strain RS-1 (Sakaguchi et al., 1993) was higher in areas with lower organic carbon provided a counter-example, but subsequent flux. They observed a parallel relationship work has nonetheless demonstrated a linkage between organic carbon and magnetofossil between magnetofossil abundance and morphology in a core through hemipelagic paleoenvironmental conditions. mud in the West Caroline Basin that extended Several studies have found evidence for to the Last Glacial Maximum. In the core, the a climatic influence on magnetofossil proportion of magnetofossils with populations. In sediment cores from ~4500 m equidimensional shapes increased from ~10% depth in the Tasman Sea, spanning the last to ~50% during the onset of the present ~780 ky, Hesse (1994) found that the interglacial stage, coincident with a decrease in abundance of SD magnetite decreased during weight percent organic carbon from ~1.4% to discrete events coincident with glacial stages. ~0.6%. TEM images indicated the magnetite was of In a sediment core from 3556 m depth bacterial origin. The change in abundance was at Chatham Rise, southwest Pacific Ocean, not controlled by dilution with carbonate but extending from Oxygen Isotope Stage 6 to the rather reflected a true change in magnetic present, Lean and McCave (1998) again mineralogy. During the reduced magnetization identified a decrease in total magnetofossil events, the ratio of equidimensional abundance correlated with increased organic magnetofossils to elongate magnetofossils also carbon levels during glacial intervals but, in decreased. Because elongate particles have contrast to Hesse (1994) and Yamazaki and larger surface area-to-volume ratios than Kawahata (1998), observed an increase in the equidimensional particles and therefore should proportion of equidimensional magnetofossils be more subject to reductive dissolution, Hesse during glacial stages. In a core from 488 m speculated that the morphological change depth through hemipelagic sediments of the reflected differences in growth rather than in Sicily Strait, spanning the last ~1 My, Dinares- preservation. A co-occurring reduction in Turell et al. (2003) likewise found that the total degree of bioturbation and dulling of sediment contribution of magnetofossils to sedimentary color indicated that sediment pore waters were magnetism was higher in interglacial stages more reduced during glacial intervals. Hesse and lower in glacial stages. therefore inferred that pore water oxygen Work on Quaternary sediments thus depletion, driven either by a drop in bottom demonstrates an empirical linkage between water oxygen levels or an increase in organic redox conditions and magnetofossils, with carbon flux, disfavored magnetotactic bacterial limited evidence suggesting a greater degree of

21 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria control by ecological factors than by three phases: an anoxic phase in the Archean taphonomic factors. Because most magnetite and earliest Proterozoic, an euxinic phase in producers live in the upper OATZ, redox the middle Proterozoic, and an oxic phase in changes that expand the OATZ should lead to the late Proterozoic and the Phanerozoic. larger populations of magnetotactic bacteria. Magnetofossil populations and preservation In sediments, the size and depth of the OATZ would likely differ among the three regimes. is controlled by a balance between the Studies on Quaternary sediments diffusive supply of oxygen from bottom waters suggest that, in a predominantly oxic ocean, and the consumption of oxygen for organic magnetofossil abundance is inversely carbon oxidation (van der Loeff, 1990). correlated with organic carbon burial (Hesse, Assuming the organic carbon supply remains 1994; Lean and McCave, 1998; Yamazaki and sufficient to deplete oxygen, both increased Kawahata, 1998). Such studies have been oxygen supply and decreased organic carbon confined to sediments with ! ~ 0.4 wt% supply to sediments lead to OATZ expansion . organic carbon; under conditions so organic Increased sedimentation rate can also dilute a poor that oxygen does not become depleted in constant organic carbon supply and similarly the sediments, no significant magnetotactic expand the OATZ. The relationship between bacteria population should grow. Greigite magnetofossil populations and organic carbon magnetofossils might be expected to be more content of sediments observed by Hesse (1994) abundant in higher productivity sediments but and Lean and McCave (1998) is consistent are in need of study. with this model. Taphonomic processes could Modern stratified water bodies like the produce a similar correlation, because although Pettaquamscutt River Estuary, Rhode Island some magnetite producers live in the lower (Bazylinski et al., 1995), Salt Pond, OATZ and under sulfidic conditions, such Massachusetts (Simmons et al., 2004), and conditions promote the reductive dissolution of Lake Ely, Pennsylvania (Kim et al., 2005), fine magnetite particles. Comparative provide analog environments for a possible microbiological, geochemical, and magnetic ancient euxinic ocean. Whereas in an oxic studies of marine sediments with different water body, the OATZ and thus the highest bottom water oxygen levels and organic carbon density of magnetotactic bacteria occur in the supplies would further elucidate and help sediments, in a euxinic water body, the OATZ quantify the magnetofossil paleoredox proxy. occurs in the water column. Kim et al. (2005) studied the magnetic properties of sediments from Lake Ely and found that magnetite 6.2. Magnetofossils as a tracer of global magnetofossils play a dominant role, with oxygen levels concentrations higher in organic-rich sediments than in organic-poor sediments. The linkage between paleo-redox However, a steady decline in saturation conditions and magnetofossil patterns keeps remanent magnetization from sediments alive Chang and Kirschvink (1984)’s hope that deposited in ~1700 CE to sediments deposited a rich Precambrian magnetofossil record might in ~1300 CE suggests that most of the serve as as a tracer of global changes in magnetite may dissolve during diagenesis. Kim oxygen level. Canfield (1998) proposed that et al. did not find signatures of greigite global ocean redox chemistry went through magnetofossils, but cautioned that sulfide

22 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria levels in Lake Ely peak at ~60 µM. In settings arose first as an iron sulfide-based energy like Salt Pond (Simmons et al., 2004), where storage mechanism that was later exapted for microbiological studies have found greigite magnetotaxis and still later adapted for the use bacteria living in the water column, peak of magnetite. Even more speculatively, the sulfide concentrations exceed 350 µM. It primitive, iron sulfide-based magnetosome therefore remains unclear whether marine could could be a relict of a protobiotic iron- sediments from a euxinic environment would sulfur world (Wachtershauser, 1990). Testing have magnetofossil populations significantly these hypotheses requires the construction of a different from oxic environments, although rich magnetofossil record for all of Earth poor preservation might be expected due to history, as well as a fuller understanding of the higher organic carbon concentrations and diversity of magnetic microbes alive today and consequent diagenetic dissolution. the genetic mechanisms underlying biological Understanding the ecology of controlled magnetic mineralization. magnetotactic bacteria in an anoxic ocean where sulfide concentrations are low and iron cycling drives the biosphere (Walker, 1987) 7. Summary requires a better understanding of the evolution of magnetotaxis than currently exists. In such The existence of magnetotactic bacteria an ocean, the narrow, suboxic redox is testimony to the power of evolution, as the environment present in the OATZ today would intracellular magnetic particles they produce fill most of the ocean. The navigational value reflect the optimization processes of natural of magnetotaxis for bacteria like the magnetite selection. Magnetofossils, the remains of these producers who dwell preferentially at the top particles, are therefore distinguishable in of the OATZ would therefore be limited, sediments from magnetic particles produced by though the niche occupied by some greigite other means and leave a record that extends bacteria at the bottom of the OATZ would still firmly into the Mesozoic and more be present where high-sulfur water masses ambiguously into the Precambrian. came in contact with high-iron water masses. Most magnetotactic bacteria favor Thus, one might expect that the ability particular redox conditions, and their to produce magnetite crystals evolved after the preservation as fossils also depends upon redox advent of oxygenic photosynthesis and that parameters. Thus, the magnetofossil record magnetite magnetofossils trace locally oxic can serve as a proxy of climate-driven changes environments. RS-1 appears to provide a in organic productivity and bottom water counterargument; whether it is a valid oxygen and might have utility deep in Earth counterargument depends upon whether RS-1 history as a tracer of ancient oxygen levels. independently evolved the ability to produce The pre-Quaternary magnetofossil record is, magnetite after it evolved into a strictly however, currently sparse. Fully utilizing the anaerobic sulfate reducer. Alternatively, the magnetofossil proxy will require studies of the ability could have evolved in an ancestor with m e c h a n i s m s a n d e v o l u t i o n o f a different metabolism or have been laterally biomineralization in magnetotactic bacteria, acquired. investigations of the relationship between The magnetosome battery hypothesis magnetotactic bacteria growth, magnetofossil suggests the possibility that the magnetosome population, oxygen supply, and biological

23 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria productivity, and the development of a richer Between Microbes And Minerals. Reviews In Mineralogy, pp. 181-223. magnetofossil record. Belkaaloul, N.K. and Aissaoui, D.M., 1997. Nature and origin of magnetic minerals within the Middle Jurassic shallow-water carbonate rocks of the Acknowledgements Paris Basin, France: implications for magnetostratigraphic dating. Geophysical This work was funded by grants from the Agouron Journal International, 130: 411-421. Institute and the NASA Astrobiology Instrument Berner, R.A., 1967. Thermodynamic stability of Development program (to JLK) and from the Moore sedimentary iron sulfides. American Journal of Foundation and NSF Graduate Research Fellowships (to Science, 265: 773-785. REK). Mark Hounslow, Hojatollah Vali, and Atsuko Blakemore, R., 1975. Magnetotactic Bacteria. Science, Kobayashi graciously provided TEM images. We thank 190: 377-379. Jared Leadbetter, Adam Maloof, Cody Nash, Dianne Buseck, P.R., Dunin-Borkowski, R.E., Devouard, B., Newman, and Lael Vetter for helpful discussion. Frankel, R.B., McCartney, M.R., Midgley, P.A., Posfai, M. and Weyland, M., 2001. Magnetite References morphology and life on Mars. Proceedings Of The National Academy Of Sciences Of The Aissaoui, D.M., McNeill, D.F. and Kirschvink, J.L., United States Of America, 98: 13490-13495. 1990. Magnetostratigraphic dating of shallow- Butler, R.F. and Banerjee, S.K., 1975. Theoretical water carbonates from Mururoa Atoll, French single-domain grain size range in magnetite Polynesia - implications for global eustasy. and titanomagnetite. Journal of Geophysical Earth and Planetary Science Letters, 97: Research, 80: 4049-4058. 102-112. Butler, R.W.H., McClelland, E. and Jones, R.E., 1999. Akai, J., Iida, A., Akai, K. and Chiba, A., 1997. Mn and Calibrating the duration and timing of the Fe minerals of possible biogenic origin from Messinian salinity crisis in the Mediterranean: two Precambrian stromatolites in western Linked tectonoclimatic signals in thrust-top Australia. Journal of the Geological Society of basins of Sicily. Journal of the Geological Japan, 103: 484–488. Society, 156: 827-835. Arató, B., Szányi, Z., Flies, C., Schüler, D., Frankel, Canfield, D.E., 1998. A new model for Proterozoic R.B., Buseck, P.R. and Pósfai, M., 2005. ocean chemistry. Nature, 396: 450-453. Crystal-size and shape distributions of Carter-Stiglitz, B., Jackson, M. and Moskowitz, B., magnetite from uncultured magnetotactic 2002. Low-temperature remanence in stable bacteria as a potential biomarker. American single domain magnetite. Geophysical Mineralogist, 90: 1233-1241. Research Letters, 29: 1129. Bazylinski, D.A. and Frankel, R.B., 2004. Magnetosome Carter-Stiglitz, B., Moskowitz, B. and Jackson, M., formation in prokaryotes. Nature Reviews 2004. More on the low-temperature magnetism Microbiology, 2: 217-230. of stable single domain magnetite: Reversibility Bazylinski, D.A., Frankel, R.B., Heywood, B.R., Mann, and non-stoichiometry. Geophysical Research S., King, J.W., Donaghay, P.L. and Hanson, Letters, 31: L06606. A.K., 1995. Controlled Biomineralization of Chang, S.B.R., 1988. Bacterial magnetite in sedimentary deposits and its geophysical and palelogical Magnetite (Fe3O4) and Greigite (Fe3S4) in a Magnetotactic Bacterium. Applied and implication. Ph.D. Thesis Thesis, California Environmental Microbiology, 61: 3232-3239. Institute of Technology, Pasadena, CA. Bazylinski, D.A., Garratt-Reed, A.J., Abedi, A. and Chang, S.B.R. and Kirschvink, J.L., 1984. Bacterial Frankel, R.B., 1993. Copper association with magnetofossils as probes of Precambrian iron sulfide magnetosomes in a magnetotactic ecological and biochemical evolutionary bacterium. Archives of Microbiology, 160: events, 97th annual meeting, Geological 35-42. Society of America. Bazylinski, D.A., Garratt-Reed, A.J. and Frankel, R.B., Chang, S.B.R. and Kirschvink, J.L., 1985. Possible 1994. Electron-Microscopic Studies of Biogenic Magnetite Fossils from the Miocene Magnetosomes in Magnetotactic Bacteria. Marine Clay of Crete. In: J.L. Kirschvink, D.S. Microscopy Research and Technique, 27: Jones and B. McFadden (Editors), Magnetite 389-401. Biomineralization and Magnetoreception in Bazylinski, D.A. and Moskowitz, B.M., 1997. Microbial Organisms: A New Biomagnetism. Plenum biomineralization of magnetic iron minerals: Press, New York, NY, pp. 647-669. Microbiology, magnetism and environmental Chang, S.B.R. and Kirschvink, J.L., 1989. significance, Geomicrobiology: Interactions Magnetofossils, the magnetization of sediments, and the evolution of magnetite

24 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

biomineralization. Annual Review of Earth and Atomic-Force Microscopy. Journal of Planetary Sciences, 17: 169-195. Microscopy-Oxford, 173: 1-8. Chang, S.B.R., Kirschvink, J.L. and Stolz, J.F., 1987. Frankel, R.B. and Blakemore, R.P., 1989. Magnetite and Biogenic Magnetite As A Primary Remanence m a g n e t o t a x i s i n m i c r o o rg a n i s m s . Carrier In Limestone Deposits. Physics Of The Bioelectromagnetics, 10: 223-237. Earth And Planetary Interiors, 46: 289-303. Friedmann, E.I., Wierzchos, J., Ascaso, C. and Chang, S.B.R., Stolz, J.F., Kirschvink, J.L. and Winklhofer, M., 2001. Chains of magnetite Awramik, S.M., 1989. Biogenic magnetite in crystals in the meteorite ALH84001: Evidence stromatolites. 2. Occurrence In ancient of biological origin. Proceedings Of The sedimentary environments. Precambrian National Academy Of Sciences Of The United Research, 43: 305-315. States Of America, 98: 2176-2181. Cox, B., 2002. Organization and Elemental Analysis of Glasauer, S., Langley, S. and Beveridge, T.J., 2002. P-, S-, and Fe-rich Inclusions in a Population of Intracellular Iron Minerals in a Dissimilatory Freshwater Magnetococci. Geomicrobiology Iron-Reducing Bacterium. Science, 295: Journal, 19: 387-406. 117-119. Dearing, J.A., Boyle, J.F., Appleby, P.G., Mackay, A.W. Golden, D.C., Ming, D.W., Morris, R.V., Brearley, A., and Flower, R.J., 1998. Magnetic properties of Lauer, H.V., Treiman, A.H., Zolensky, M.E., recent sediments in Lake Baikal, Siberia. Schwandt, C.S., Lofgren, G.E. and McKay, Journal Of Paleolimnology, 20: 163-173. G.A., 2004. Evidence for exclusively inorganic DeLong, E.F., Frankel, R.B. and Bazylinski, D.A., 1993. formation of magnetite in Martian meteorite Multiple evolutionary origins of magnetotaxis ALH84001. American Mineralogist, 89: in bacteria. Science, 259: 803-806. 681-695. Devouard, B., Posfai, M., Hua, X., Bazylinski, D.A., Gorby, Y.A., Beveridge, T.J. and Blakemore, R.P., 1988. Frankel, R.B. and Buseck, P.R., 1998. Characterization of the Bacterial Magnetosome Magnetite from magnetotactic bacteria: Size Membrane. Journal of Bacteriology, 170: distributions and twinning. American 834-841. Mineralogist, 83: 1387-1398. Hesse, P.P., 1994. Evidence for Bacterial Diaz-Ricci, J.C. and Kirschvink, J.L., 1992. Magnetic Paleoecological Origin of Mineral Magnetic Domain State And Coercivity Predictions For Cycles in Oxic and Sub-Oxic Tasman Sea Biogenic Greigite (Fe3s4) - A Comparison Of Sediments. Marine , 117: 1-17. Theory With Magnetosome Observations. Heywood, B.R., Bazylinski, D.A., Garrattreed, A., Journal Of Geophysical Research-Solid Earth, Mann, S. and Frankel, R.B., 1990. Controlled 97: 17309-17315. Biosynthesis Of Greigite (Fe3S4) In Dinares-Turell, J., Hoogakker, B.A.A., Roberts, A.P., Magnetotactic Bacteria. Naturwissenschaften, Rohling, E.J. and Sagnotti, L., 2003. 77: 536-538. Quaternary climatic control of biogenic Hilgenfeldt, K., 2000. Diagenetic dissolution of biogenic magnetite production and eolian dust input in magnetite in surface sediments of the Benguela cores from the Mediterranean Sea. upwelling system. International Journal Of Palaeogeography Palaeoclimatology Earth Sciences, 88: 630-640. Palaeoecology, 190: 195-209. Hounslow, M.W. and Maher, B.A., 1996. Quantitative Dong, H., Fredrickson, J.K., Kennedy, D.W., Zachara, extraction and analysis of carriers of J.M., Kukkadapu, R.K. and Onstott, T.C., 2000. magnetization in sediments. Geophysical Mineral transformation associated with the Journal International, 124: 57-74. microbial reduction of magnetite. Chem. Geol, Housen, B.A. and Moskowitz, B.M., 2006. Depth 169: 299-318. distribution of magnetofossils in near-surface Dunin-Borkowski, R.E., McCartney, M.R., Posfai, M., sediments from the Blake/Bahama Outer Frankel, R.B., Bazylinski, D.A. and Buseck, Ridge, western North Atlantic Ocean, P.R., 2001. Off-axis electron holography of determined by low-temperature magnetism. m a g n e t o t a c t i c b a c t e r i a : m a g n e t i c Journal of Geophysical Research, 111: G01005. microstructure of strains MV-1 and MS-1. Jørgensen, B.B. and Gallardo, V.A., 1999. Thioploca European Journal of Mineralogy, 13: 671-684. spp.: filamentous sulfur bacteria with nitrate Egli, R., 2004. Characterization of individual rock vacuoles. FEMS Microbiology Ecology, 28: magnetic components by analysis of remanence 301-313. curves, 1. Unmixing natural sediments. Studia Keim, C.N., Solórzano, G., Farina, M. and Lins, U., Geophysica Et Geodaetica, 48: 391-446. 2005. Intracellular inclusions of uncultured Farina, M., Kachar, B., Lins, U., Broderick, R. and magnetotactic bacteria. International Debarros, H.L., 1994. The Observation of Microbiology, 8: 111-117. Large Magnetite (Fe3o4) Crystals from Kent, D.V., Cramer, B.S., Lanci, L., Wang, D., Wright, Magnetotactic Bacteria by Electron and J.D. and Van der Voo, R., 2003. A case for a

25 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

comet impact trigger for the Paleocene/Eocene Kopp, R.E., Nash, C.Z., Kobayashi, A., Weiss, B.P., thermal maximum and carbon isotope Bazylinski, D.A. and Kirschvink, J.L., 2006a. excursion. Earth and Planetary Science Letters, Ferromagnetic resonance spectroscopy for 211: 13-26. assessment of magnetic anisotropy and Kim, B.Y., Kodama, K.P. and Moeller, R.E., 2005. magnetostatic interactions: A case study of Bacterial magnetite produced in water column mutant magnetotactic bacteria. J. Geophys. dominates lake sediment mineral magnetism: Res., 111: B12S25. Lake Ely, USA. Geophysical Journal Kopp, R.E., Raub, T.D., Schumann, D., Vali, H., International, 163: 26-37. Smirnov, A.V. and Kirschvink, J.L., 2007. Kirschvink, J.L., 1980. South-seeking magnetic bacteria. Magnetofossil Spike at the Paleocene-Eocene Journal of Experimental Biology, 86: 345-347. Thermal Maximum: Ferromagnetic resonance, Kirschvink, J.L., 1982. Paleomagnetic Evidence For rock magnetic, and electron microscopy Fossil Biogenic Magnetite In Western Crete. evidence from Ancora, New Jersey, USA. Earth And Planetary Science Letters, 59: Paleoceanography, 22, PA4103, doi: 388-392. 10.1029/2007PA001473. Kirschvink, J.L., 2001. A seventh criterion for the Kopp, R.E., Weiss, B.P., Maloof, A.C., Vali, H., Nash, identification of bacterial magnetofossils. Eos C.Z. and Kirschvink, J.L., 2006b. Chains, Trans. AGU, 82: S131. clumps, and strings: Magnetofossil taphonomy Kirschvink, J.L. and Chang, S.B.R., 1984. Ultrafine- with ferromagnetic resonance spectroscopy. Grained Magnetite in Deep-Sea Sediments - Earth and Planetary Science Letters, 247: Possible Bacterial Magnetofossils. Geology, 10-25. 12: 559-562. Kostka, J.E. and Nealson, K.H., 1995. Dissolution and Kirschvink, J.L. and Lowenstam, H.A., 1979. Reduction of Magnetite by Bacteria. Mineralization and magnetization of chiton Environmental Science & Technology, 29: teeth: paleomagnetic, sedimentologic, and 2535-2540. biologic implications of organic magnetite. Lean, C.M.B. and McCave, I.N., 1998. Glacial to Earth and Planetary Science Letters, 44: interglacial mineral magnetic and 193-204. palaeoceanographic changes at Chatham Rise, Kirschvink, J.L. and Rozanov, A.Y., 1984. SW Pacific Ocean. Earth and Planetary Science Magnetostratigraphy of lower Cambrian strata Letters, 163: 247-260. from the Siberian Platform: a paleomagnetic Lippert, P.C. and Zachos, J.C., 2007. A biogenic origin pole and a preliminary polarity time-scale. for anomalous fine-grained magnetic material Geol. Mag., 121: 189-203 . at the Paleocene-Eocene boundary at Wilson Kobayashi, A., Kirschvink, J.L., Nash, C.Z., Kopp, R.E., Lake, New Jersey. Paleoceanography, 22, Sauer, D.A., Bertani, L.E., Voorhout, W.F. and PA4104, doi:10.1029/2007PA001471. Taguchi, T., 2006. Experimental observation of Lowenstam, H.A., 1981. Minerals formed by organisms. magnetosome chain collapse in magnetotactic Science, 211: 1126. bacteria: sedimentological, paleomagnetic, and Maher, B.A., Thompson, R. and Hounslow, M.W., 1999. evolutionary implications. Earth And Planetary Introduction. In: B.A. Maher and R. Thompson Science Letters, 245: 538-550. (Editors), Quaternary Climates, Environments, Komeili, A., Li, Z., Newman, D.K. and Jensen, G.J., and Magnetism. Cambridge University Press, 2006. Magnetosomes are cell membrane Cambridge, pp. 1-48. invaginations organized by the actin-like Maloof, A.C., Kopp, R.E., Grotzinger, J.P., Fike, D.A., protein MamK. Science, 311: 242-245. Bosak, T., Vali, H., Poussart, P.M., Weiss, B.P. Komeili, A., Vali, H., Beveridge, T.J. and Newman, and Kirschvink, J.L., 2007. Sedimentary Iron D.K., 2004. Magnetosome vesicles are present Cycling and the Origin and Preservation of before magnetite formation, and MamA is Magnetization in Platform Carbonate Muds, required for their activation. Proceedings of the Andros Island, Bahamas. Earth and Planetary National Academy of Sciences, 101: Science Letters, 259: 581-598. 3839-3844. Mann, S., Sparks, N.H.C. and Blakemore, R., 1987. Kopp, R.E., 2007. The identification and interpretation Structure, morphology, and crystal growth of of microbial biogeomagnetism. Ph.D. Thesis, anisotropic magnetite crystals in magnetotactic California Institute of Technology, Pasadena, bacteria. Proc. R. Soc. Lond. B, 231: 477-487. California, 190 pp. Maratea, D. and Blakemore, R.P., 1981. Aquaspirillum Kopp, R.E. and Humayun, M., 2003. Kinetic model of magnetotacticum sp. nov., a magnetic spirillum. carbonate dissolution in Martian meteorite International Journal of Systematic ALH84001. Geochimica et Cosmochimica Bacteriology, 31: 452-455. Acta, 67: 3247-3256. Matsunaga, T. and Okamura, Y., 2003. Genes and proteins involved in bacterial magnetite particle

26 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

formation. Trends in Microbiology, 11: Niitsuma, S., Nedachi, M., Hoashi, M., Suganuma, Y., 536-541. Sugita, T., Ohno, K., Hisamitsu, T., Kodama, McKay, D.S., Gibson, E.K., ThomasKeprta, K.L., Vali, K. and Niitsuma, N., 2004. Rock magnetism of H., Romanek, C.S., Clemett, S.J., Chillier, the black shale in the Mt. Roe Basalt, Pilbara X.D.F., Maechling, C.R. and Zare, R.N., 1996. Craton, Western Australia, Japan Earth and Search for past life on Mars: Possible relic Planetary Science Joint Meeting. Japan biogenic activity in Martian meteorite Geoscience Union, Makuhari, Japan. ALH84001. Science, 273: 924-930. Oldfield, F., Wake, R., Boyle, J., Jones, R., Nolan, S., McNeill, D.F., 1990. Biogenic magnetite from surface Gibbs, Z., Appleby, P., Fisher, E. and Wolff, G., Holocene carbonate sediments, Great Bahama 2003. The late-Holocene history of Gormire Bank. Journal Of Geophysical Research-Solid Lake (NE England) and its catchment: a Earth And Planets, 95: 4363-4371. multiproxy reconstruction of past human McNeill, D.F., Ginsburg, R.N., Chang, S.B.R. and impact. Holocene, 13: 677-690. Kirschvink, J.L., 1988. Magnetostratigraphic Özdemir, Ö., Dunlop, D.J. and Moskowitz, B.M., 1993. Dating Of Shallow-Water Carbonates From The effect of oxidation on the Verwey San-Salvador, Bahamas. Geology, 16: 8-12. transition in magnetite. Geophysical Research Meert, J.G., Eide, E.A. and Torsvik, T.H., 1997. The Letters, 20: 1671-1674. Nama Group revisited. Geophys. J. Int., 129: Pan, Y.X., Petersen, N., Davila, A.F., Zhang, L.M., 637-650. Winklhofer, M., Liu, Q.S., Hanzlik, M. and Meldrum, F.C., Mann, S., Heywood, B.R., Frankel, R.B. Zhu, R.X., 2005. The detection of bacterial and Bazylinski, D.A., 1993a. Electron- magnetite in recent sediments of Lake Microscopy Study of Magnetosomes in 2 Chiemsee (southern Germany). Earth And Cultured Vibrioid Magnetotactic Bacteria. Planetary Science Letters, 232: 109-123. Proceedings of the Royal Society of London Peck, J.A. and King, J.W., 1996. Magnetofossils in the Series B-Biological Sciences, 251: 237-242. sediment of Lake Baikal, Siberia. Earth and Meldrum, F.C., Mann, S., Heywood, B.R., Frankel, R.B. Planetary Science Letters, 140: 159-172. and Bazylinski, D.A., 1993b. Electron- Petersen, N., Vondobeneck, T. and Vali, H., 1986. Fossil Microscopy Study of Magnetosomes in a Bacterial Magnetite In Deep-Sea Sediments Cultured Coccoid Magnetotactic Bacterium. From The South-Atlantic Ocean. Nature, 320: Proceedings of the Royal Society of London 611-615. Series B-Biological Sciences, 251: 231-236. Pósfai, M., Buseck, P.R., Bazylinski, D.A. and Frankel, Meulenkamp, J.E., 1979. Lithostratigraphy and relative R.B., 1998. Iron sulfides from magnetotactic chronostratigraphic position of the sections bacteria: Structure, composition, and phase Apostoli and Potamidha 1 and 2. Utrecht transitions. American Mineralogist, 83: Micropaleontol. Bull., 21: 9-21. 1469-1481. Moench, T.T., 1988. Bilophococcus magnetotacticus Pósfai, M., Cziner, K., Marton, E., Marton, P., Buseck, gen. nov. sp. nov., a motile, magnetic coccus. P.R., Frankel, R.B. and Bazylinski, D.A., 2001. Antonie van Leeuwenhoek, International Crystal-size distributions and possible biogenic Journal of General and Molecular origin of Fe sulfides. European Journal Of Microbiology, 54: 483-496. Mineralogy, 13: 691-703. Montgomery, P., Hailwood, E.A., Gale, A.S. and Pósfai, M., Moskowitz, B.M., Arato, B., Schüler, D., Burnett, J.A., 1998. The magnetostratigraphy of Flies, C., Bazylinski, D.A. and Frankel, R.B., Coniacian late Campanian chalk sequences in 2006. Properties of intracellular magnetite southern England. Earth And Planetary Science crystals produced by Desulfovibrio magneticus Letters, 156: 209-224. strain RS-1. Earth and Planetary Science Morse, J.W., Millero, F.J., Cornwell, J.C. and Rickard, Letters, 249: 444-455. D., 1987. The chemistry of the hydrogen Reitner, J., Peckmann, J., Blumenberg, M., Michaelis, sulfide and iron sulfide systems in natural W., Reimer, A. and Thiel, V., 2005. waters. Earth-Science Reviews, 24: 1-42. Concretionary methane-seep carbonates and Moskowitz, B.M., Frankel, R.B. and Bazylinski, D.A., associated microbial communities in Black Sea 1993. Rock magnetic criteria for the detection s e d i m e n t s . P a l a e o g e o g r a p h y , of biogenic magnetite. Earth And Planetary Palaeoclimatology, Palaeoecology, 227: 18-30. Science Letters, 120: 283-300. Reynolds, R.L., Tuttle, M.L., Rice, C.A., Fishman, N.S., Niitsuma, S., Kakegawa, T., Nagasu, T. and Nedachi, Karachewski, J.A. and Sherman, D.M., 1994. M., 2005. Discovery of greigite from Archean Magnetization and geochemistry of greigite- rock?, Japan Earth and Planetary Science Joint bearing Cretaceous strata, North Slope basin, Meeting. Japan Geoscience Union, Makuhari, Alaska. Am J Sci, 294: 485-528. Japan. Rohl, U., Bralower, T.J., Norris, R.D. and Wefer, G., 2000. New chronology for the late Paleocene

27 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

thermal maximum and its environmental Tarduno, J.A., 1994. Temporal Trends of Magnetic implications. Geology, 28: 927-930. Dissolution in the Pelagic Realm - Gauging Sakaguchi, T., Arakaki, A. and Matsunaga, T., 2002. Paleoproductivity. Earth and Planetary Science Desulfovibrio magneticus sp. nov., a novel Letters, 123: 39-48. sulfate-reducing bacterium that produces Tauxe, L., Tucker, P., Petersen, N.P. and LaBrecque, intracellular single-domain-sized magnetite J.L., 1984. Magnetostratigraphy of Leg 73 particles. International Journal of Systematic Sediments. In: K.J. Hsü and J.L. LaBrecque and Evolutionary Microbiology, 52: 215-221. (Editors), Initial Reports of the Deep Sea Sakaguchi, T., Burgess, J.G. and Matsunaga, T., 1993. Drilling Project. U.S. Govt. Printing Office, Magnetite formation by a sulphate-reducing Washington, pp. 609-622. bacterium. Nature, 365: 47-49. Taylor, A.P. and Barry, J.C., 2004. Magnetosomal Sakai, S. and Jige, M., 2006. Characterization of matrix: ultrafine structure may template magnetic particles and magnetostratigraphic biomineralization of magnetosomes. Journal of dating of shallow-water carbonates in the Microscopy, 213: 180-197. Ryukyu Islands, northwestern Pacific. The Taylor, A.P., Barry, J.C. and Webb, R.I., 2001. Structural Island Arc, 15: 468-475. and morphological anomalies in Scheffel, A., Gruska, M., Faivre, D., Linaroudis, A., magnetosomes: possible biogenic origin for Plitzko, J.M. and Schuler, D., 2006. An acidic magnetite in ALH84001. Journal of protein aligns magnetosomes along a Microscopy, 201: 84-106. filamentous structure in magnetotactic bacteria. Thomas-Keprta, K.L., Bazylinski, D.A., Kirschvink, Nature, 440: 110-114. J.L., Clemett, S.J., McKay, D.S., Wentworth, Schüler, D. and Baeuerlein, E., 1998. Dynamics of iron S.J., Vali, H., Gibson, E.K. and Romanek, C.S., uptake and Fe3O4 biomineralization during 2000. Elongated prismatic magnetite crystals in aerobic and microaerobic growth of ALH84001 carbonate globules: Potential Magnetospirillum gryphiswaldense. Journal of Martian magnetofossils. Geochimica Et Bacteriology, 180: 159-162. Cosmochimica Acta, 64: 4049-4081. Schulz, H.N. and Jørgensen, B.B., 2001. Big bacteria. Thomas-Keprta, K.L., Clemett, S.J., Bazylinski, D.A., Annual Review of Microbiology, 55: 105-137. Kirschvink, J.L., McKay, D.S., Wentworth, Simmons, S.L., Sievert, S.M., Frankel, R.B., Bazylinski, S.J., Vali, H., Gibson, E.K., McKay, M.F. and D.A. and Edwards, K.J., 2004. Spatiotemporal Romanek, C.S., 2001. Truncated hexa- distribution of marine magnetotactic bacteria in octahedral magnetite crystals in ALH84001: a seasonally stratified coastal salt pond. Presumptive biosignatures. Proceedings Of The Applied and Environmental Microbiology, 70: National Academy Of Sciences Of The United 6230-6239. States Of America, 98: 2164-2169. Skinner, B.J., Grimaldi, F.S. and Erd, R.C., 1964. Thomas-Keprta, K.L., Clemett, S.J., Bazylinski, D.A., Greigite Thio-Spinel of Iron - New Mineral. Kirschvink, J.L., McKay, D.S., Wentworth, American Mineralogist, 49: 543-&. S.J., Vali, H., Gibson, E.K. and Romanek, C.S., Snowball, I.F., 1994. Bacterial magnetite and the 2002. Magnetofossils from ancient Mars: a magnetic properties of sediments in a Swedish robust biosignature in the Martian meteorite lake. Earth And Planetary Science Letters, 126: ALH84001. Applied And Environmental 129-142. Microbiology, 68: 3663-3672. Spring, S., Amann, R., Ludwig, W., Schleifer, K.H., Van Thomas-Keprta, K.L., Clemett, S.J., Schwartz, C., Gemerden, H. and Petersen, N., 1993. Morphew, M., McIntosh, J.R., Bazylinski, Dominating role of an unusual magnetotactic D.A., Kirschvink, J.L., Wentworth, S.J., bacterium in the microaerobic zone of a McKay, D.S., Vali, H., Gibson, E.K. and freshwater sediment. Applied and Romanek, C.S., 2004. Determination of the Environmental Microbiology, 59: 2397-2403. three-dimensional morphology of ALH84001 Spring, S. and Schleifer, K.H., 1995. Diversity of and biogenic MV-1 magnetite: comparsion of magnetotactic bacteria. Systematic and Applied results from electron tomography and classical Microbiology, 18: 147-153. transmission electron microscopy (abstract), Stolz, J.F., Chang, S.B.R. and Kirschvink, J.L., 1986. Lunar and Planetary Science XXXV, pp. 2030. Magnetotactic Bacteria And Single-Domain Thornhill, R.H., Burgess, J.G., Sakaguchi, T. and Magnetite In Hemipelagic Sediments. Nature, Matsunaga, T., 1994. A morphological 321: 849-851. classification of bacteria containing bullet- Stolz, J.F., Chang, S.B.R. and Kirschvink, J.L., 1989. shaped magnetic particles. FEMS Microbiology Biogenic Magnetite In Stromatolites .1. Letters, 115: 169-176. Occurrence In Modern Sedimentary Tikoo, S.M., Kopp, R. E., Vali, H., and Kirschvink, J.L., Environments. Precambrian Research, 43: in prep. Rock magnetic characterization of the 295-304. late Archean Ghaap Group, Transvaal

28 R. E. Kopp and J. L. Kirschvink / Fossil magnetotactic bacteria

Supergroup, from Agouron drill core GKP01: Walker, J.C.G., 1987. Was the Archaean biosphere No signs of life, but a magnetic upside down? Nature, 329: 710-712. paleothermometer for siderite metamorphism. Weiss, B.P., Kim, S.S., Kirschvink, J.L., Kopp, R.E., Precambrian Research. Sankaran, M., Kobayashi, A. and Komeili, A., Valet, J.-P. and Laj, C., 1981. Paleomagnetic record of 2004a. Ferromagnetic resonance and low two successive Miocene geomagnetic reversals temperature magnetic tests for biogenic in western Crete. Earth and Planetary Science magnetite. Earth and Planetary Science Letters, Letters, 54: 53-63. 224: 73-89. Vali, H., Forster, O., Amarantidis, G. and Petersen, N., Weiss, B.P., Kim, S.S., Kirschvink, J.L., Kopp, R.E., 1987. Magnetotactic Bacteria And Their Sankaran, M., Kobayashi, A. and Komeili, A., Magnetofossils In Sediments. Earth And 2004b. Magnetic tests for magnetosome chains Planetary Science Letters, 86: 389-400. in Martian meteorite ALH84001. Proceedings Vali, H. and Kirschvink, J.L., 1989. Magnetofossil Of The National Academy Of Sciences Of The dissolution in a paleomagnetically unstable United States Of America, 101: 8281-8284. deep-sea sediment. Nature, 339: 203-206. Witt, A., Fabian, K. and Bleil, U., 2005. Three- Vali, H. and Kirschvink, J.L., 1991. Observations of dimensional micromagnetic calculations for Magnetosome Organization, Surface Structure, naturally shaped magnetite: Octahedra and and Iron Biomineralization of Undescribed magnetosomes. Earth And Planetary Science Magnetic Bacteria: Evolutionary Speculations. Letters, 233: 311-324. In: R.B. Frankel and R.P. Blakemore (Editors), Xu, W.X., Peacor, D.R., Dollase, W.A., Van Der Voo, R. Iron biominerals. Plenum Press, New York, pp. and Beaubouef, R., 1997. Transformation of 97-116. titanomagnetite to titanomaghemite: A slow, van der Loeff, M.M.R., 1990. Oxygen in Pore Waters of two-step, oxidation-ordering process in MORB. Deep-Sea Sediments. Philosophical American Mineralogist, 82: 1101-1110. Transactions of the Royal Society of London Yamazaki, T. and Ioka, N., 1997. Environmental rock- Series A, 331: 69-84. magnetism of pelagic clay: Implications for van der Voo, R., 1990. The reliability of paleomagnetic Asian eolian input to the North Pacific since data. Tectonophysics, 184: 1-9. the Pliocene. Paleoceanography, 12: 111-124. van Hinsbergen, D.J.J. and Meulenkamp, J.E., 2006. Yamazaki, T., Katsura, I. and Marumo, K., 1991. Origin Neogene supradatchment basin development on Of Stable Remanent Magnetization Of Crete (Greece) during exhumation of the South Siliceous Sediments In The Central Equatorial Aegean core complex. Basin Research, 18: Pacific. Earth And Planetary Science Letters, 103-124. 105: 81-93. Wachtershauser, G., 1990. The Case for the Yamazaki, T. and Kawahata, H., 1998. Organic carbon Chemoautotrophic Origin of Life in an Iron- flux controls the morphology of magnetofossils Sulfur World. Origins of Life and Evolution of in marine sediments. Geology, 26: 1064-1066. the Biosphere, 20: 173-176.

29