Paleomagnetic, rock magnetic and micro-textural analyses from a Middle Pleistocene lacustrine sequence in the southern Apennines (Italy) indicate the presence of greigite and magnetite as the main magnetic minerals at different stratigraphic levels. In all cases a normal polarity characteristic remanent magnetization (ChRM) was observed, in agreement with an 40Ar–39Ar age of 0.712 ± 0.018 Ma from volcaniclastic sanidine. After correction for bedding tilt, ChRM directions carried by greigite do not coincide with the expected geocentric axial dipole field direction at the site latitude, whereas the magnetite ChRMs directions do. The data indicate that the greigite magnetization was acquired after tilting and after lock-in of the magnetite ChRM. The estimated delay for the remanence carried by greigite with respect to deposition is ∼300 ka. Scanning electron microscope analyses indicate that alteration of detrital volcanic minerals has occurred and that authigenic greigite is generally present in agglomerates and around volcanic grains. This observation is consistent with a late diagenetic origin of greigite due to anoxic conditions and availability of dissolved sulfide associated with decomposition of organic matter in the paleolake. Documentation of a late diagenetic magnetization confirms that care should be taken when using greigite-bearing sediments for magnetostratigraphic and tectonic studies.
 Greigite (Fe3S4) is an authigenic magnetic mineral that forms in sulfate-reducing environments and which can be important in paleomagnetic and environmental magnetic studies. It has been widely identified in marine and lake sediments, where it is often the main magnetic carrier [Roberts, 1995]. The reliability of paleomagnetic data from greigite-bearing rocks is strongly debated because the timing of NRM acquisition by greigite is often not well constrained. On one hand, magnetosomal greigite can survive over geological time scales and can record a primary NRM [Hüsing et al., 2007; Vasiliev et al., 2008]. Conversely, in many cases, greigite-bearing sediments have inconsistent polarity records [Florindo and Sagnotti, 1995; Roberts and Weaver, 2005; Sagnotti et al., 2005] and even contradictory polarities within the same stratigraphic horizon [Jiang et al., 2001]. This indicates a late diagenetic origin of greigite. Discrepancies in the timing of remanence acquisition by greigite can have adverse effects for high-resolution magnetostratigraphy [Jiang et al., 2001] or tectonic studies [Rowan and Roberts, 2008], i.e., rock magnetic and textural analyses are needed to help constrain the origin and timing of greigite formation.
 In this paper, we present paleomagnetic results from a Quaternary continental sequence deposited in the Piano del Gaudo basin (Latitude 40°44′35″N, Longitude 15°06′55″E), an intramontane basin in the Picentini Mountains (southern Apennines, Italy). Paleomagnetic data have been integrated with detailed rock magnetic analyses. In particular, we investigated the magnetic properties of greigite-bearing sediments, which are present at different stratigraphic levels within the sedimentary section. Greigite and other iron sulfide minerals were identified using a scanning electron microscope (SEM).
2. Geological Setting
 Lacustrine sediments crop out at ∼1050 m asl, in the Picentini Mountains (southern Italy). The sediments consist of 15 m of clays, marls, and silts which, in the upper part of the section, are interbedded with 4-m-thick layer of volcaniclastic sediments (Figure 1a). The volcaniclastic sediments are characterized by submillimeter-sized sanidine, leucite, biotite and clinopyroxene crystals. The abundance of these minerals and scarce evidence of reworking suggest that they were rapidly deposited into the lake. The entire sequence is gently tilted 15° toward the west and is slightly deformed by small scale NW–SE oriented normal faults. The deformation is likely related to the extensional tectonic phase responsible of the formation of the nearby (2 km away) NNW–SSE Acerno basin, dated between 400 and 260 ka on the basis of tephrochronology and pollen analysis [Munno et al., 2001]. Sanidine crystals from the volcaniclastic layers were sampled for 40Ar–39Ar analysis. Seventeen out of 19 total fusion analyses on multigrain splits (see Table S1 of the auxiliary material) gave ages from 0.66 ± 0.06 Ma to 0.95 ± 0.04 Ma. The youngest thirteen analyses gave ages that overlap within analytical errors, with an error-weighted mean of 0.712 ± 0.018 Ma (±2σ, Table S1). On the basis of petrographic and sedimentological analyses, we interpret the 0.712 ± 0.18 Ma age as a maximum age of deposition as the sediments are produced by erosion of pre-existing volcanic products.
3. Paleomagnetic and Rock Magnetic Analyses
 We collected 120 oriented cores from eleven levels (sites) distributed over a total stratigraphic thickness of about 15 m (Figure 1a). Rock magnetic measurements include thermal demagnetization of a composite isothermal remanent magnetization (IRM), and thermomagnetic analysis of magnetic susceptibility (κ) measured in air.
 Blocking temperature spectra were determined by thermally demagnetizing a composite IRM produced by sequential application of 1.7, 0.6, and 0.12 T fields along the three orthogonal specimen axes [Lowrie, 1990]. For most of the samples, the magnetic mineralogy is dominated by low coercivity magnetic minerals. In some cases, the low coercivity component has a maximum unblocking temperature of 500–580°C, which indicates that low-Ti titanomagnetite and magnetite are the main magnetic carriers (Figures 1b and S1a in the auxiliary material). In some other sites, the maximum unblocking temperature is 360–390°C, which instead suggests the presence of an iron sulfide mineral (e.g., greigite) as the main magnetic carrier (Figures 1c and S1b in the auxiliary material). Some other sites contain both minerals with different maximum unblocking temperatures of 300–360°C and 580°C (Figure 1d). Two sites (TI20 and TI21) are dominated by high coercivity magnetic minerals and have a maximum unblocking temperature of ∼630°C, suggesting the presence of hematite. These sites were successively not involved in further paleomagnetic investigations.
 In the rest of sites, two types of behaviour were observed in thermomagnetic analysis of the bulk κ susceptibility (Figures S1c and S1d in the auxiliary material). In one case, thermomagnetic curves are characterized by a single drop at 550–580°C, which suggests the presence of magnetite as the main magnetic carrier (Figure S1c in the auxiliary material). In some other cases, the curves are marked by a decrease between 200° and 350°C, with a strong increase of κ below 350°C (Figure S1d in the auxiliary material). This behavior is typical of sediments where greigite is the dominant magnetic mineral [Roberts, 1995; Sagnotti and Winkler, 1999]. The combination of unblocking temperature range and chemical alteration at relatively low temperature in one set of samples suggests that magnetite (Fe3O4) and greigite (Fe3S4) are the dominant ferrimagnetic phases in the Piano del Gaudo basin.
 Paleomagnetic measurements of the natural remanent magnetization (NRM) were performed at the paleomagnetic laboratories in the University of Roma Tre (Italy) and ETH Zurich (Switzerland). In some cases, both thermal and alternating field (AF) demagnetization were carried out. Most specimens (116 out of 120 measured specimens) yielded reliable paleomagnetic results. Different paleomagnetic behavior is observed according to the main magnetic carriers present in each specimen.
 In magnetite-bearing specimens, a well-defined single normal polarity magnetization is isolated up to 480–500°C (Figure 1b). In greigite-bearing specimens, one or two components can be observed. When only one component occurs, it is characterized by a stable downward-directed magnetization up to 290–360°C (Figure 1c), which, before tectonic bedding correction, is oriented close to the expected geomagnetic field direction. Where two components are present, we recognize a component carried by greigite with normal polarity, which is stable up to 290–360°C, and a high-temperature component (530–580°C), due to magnetite, that also has normal polarity (Figure 1d). In some specimens, two magnetic components are evident in samples where combined thermal and AF demagnetization were used. Equal area stereographic projections of characteristic remanent magnetization directions for each magnetic carrier, before and after tilt correction, are shown in Figure 2. We obtain well-defined and statistically distinct mean paleomagnetic directions for both greigite- and magnetite-bearing specimens, where their 95% confidence ellipses are not overlapping.
 The mean paleomagnetic direction of the low-temperature component carried by greigite is D = 355.5°; I = 52.9° (α95 = 2.2°) before bedding correction. This direction is almost coincident with that of the expected geomagnetic axial dipole (GAD) field for the study area (triangle in Figure 2). When the tectonic tilt correction is applied, the mean direction is D = 336.0°; I = 50.5°, which is statistically distinct from the expected GAD reference direction. The mean paleomagnetic direction of the high-temperature components carried by magnetite before tectonic correction is D = 15.9°, I = 45.1° (α95 = 3.4°). This direction is statistically different from that of the component carried by greigite and from the present day GAD reference field direction. When the bedding correction is applied the mean direction is D = 0.2°, I = 48.5° (α95 = 3.4°). In this case, the mean paleomagnetic direction is close to the expected GAD direction even if its confidence interval does not intersect with the GAD direction (4° in between).
 Our results strongly suggest that magnetite and greigite formed during different times. The high-temperature component carried by magnetite is probably a primary post-depositional remanent magnetization (pDRM), acquired during the early part of the Brunhes epoch, in agreement with the radioisotopic age of 0.712 ± 0.018 Ma. Conversely, the low-temperature component carried by greigite is probably a chemical remanent magnetization (CRM) that was acquired after tectonic tilting of the stratigraphic section.
4. SEM–EDS Analysis
 Microtextures of representative samples from the Piano del Gaudo basin were investigated using back-scattered electron (BSE) imaging of polished sections from volcanic and non-volcanic greigite-bearing intervals. A Philips XL30 SEM at the Interdepartmental Laboratory of Electron Microscopy at Roma Tre University, operated at 25 kV, was used for these observations. Elemental analyses were obtained, using an X-Ray energy dispersive spectrometer (EDS), from point analyses (∼2 μm beam diameter) of individual mineral grains, or of clusters of smaller grains. We distinguished greigite and pyrite from examination of the iron to sulfur ratio: Fe/S = 0.5 for pure pyrite, and Fe/S = 0.75 for greigite [Roberts and Weaver, 2005]. We recognize distinct microtextures in samples from clay and volcanic-rich layers. In all samples, we found patches of iron sulfide aggregates and/or nodules with dimensions of 10s to 100s of μm across (Figures 3a–3c). The aggregates usually contain spherical (circular in cross-section) framboids made up of large (∼1 μm) octahedral pyrite grains, as well as less regularly shaped aggregates of pyrite and much finer-grained greigite crystals (∼0.5 μm, Figure 3b). Large titanomagnetite grains with variable Ti contents have also been found in all volcanic samples, which coexist with iron sulfide grains (Figure 3c).
 In particular, we observe that greigite occurs in two different settings. The most common case is when it occurs within iron sulfide aggregates, in which rhombohedral-shaped pyrite is documented (Figures 3a and 3b). These aggregates occur in elongated patches (Figure 3a), or in irregular distributions (Figure 3d). In the other case, greigite occurs as a secondary product of altered volcanic grains (e.g., phyllosilicates, Figure 3e). Here, EDS spectra of aggregated grains indicate that the Fe/S ratio tends to decrease from the centre to the edge of the larger aggregate. EDS analyses often yield intermediate Fe/S ratios between those expected for pyrite and greigite (Fe/S = 43/57 at. %) because the diameter of the electron beam is wider than individual grains and therefore pyrite and greigite grains were analysed simultaneously. No sample oxidation was observed from elemental spectra (Figure 3e). Based on the paleomagnetic evidence, these different types of greigite growth appear to have occurred long after deposition and are responsible for remagnetizations within the Piano del Gaudo basin.
 Among the several mechanisms proposed to explain the formation of iron sulfides, the most common process concerns chemical reactions driven by bacterial degradation of organic matter [Roberts and Weaver, 2005]. Iron reduction results in increase of in dissolved iron concentrations in pore waters, whereas sulfate reduction results in increased dissolved sulfide (H2S). The dissolved species react to eventually form pyrite (Fe2S), after the precursor greigite [Roberts and Weaver, 2005]. In other cases, greigite is closely associated with volcanic ash layers [Krs et al., 1990; Florindo and Sagnotti, 1995]. Volcanic input could provide a source of sulfide within marine/lacustrine basins, which can contribute to growth of late diagenetic iron sulfides.
 In the Piano del Gaudo basin, we recognize two distinct mechanisms for late diagenetic greigite growth. The first is common along the stratigraphic section and can be attributed to prevalent anoxic conditions within the small basin, which provided the necessary dissolved iron and sulfide by means of organic matter decomposition [Raiswell, 1982]. This mechanism is responsible for the formation of polyframboidal aggregates of iron sulfides, which probably remineralized organic matter (Figures 3a–3c).
 A different process is likely for greigite growth within the volcanic-rich horizon. In this case, iron-sulfide growth is enhanced by the high concentration of volcanic grains, which have high Fe-abundances. In this case, the process that produced the iron sulfide is related to dissolution of sheet silicates, which progressively react to form iron sulfides (Figure 3e). Replacement of silicates explains why greigite, which is present throughout the stratigraphic section, is much more abundant within the volcaniclastic-rich layer.
 Given the post-depositional origin of greigite, the main important question is whether it is possible to define the precise timing of greigite growth and the dating of remanence acquisition. Although some work suggests that early [Vasiliev et al., 2008] greigite growth can occur (only a few years in the case of Reynolds et al. ), compelling evidence suggests that in many cases iron sulfides can also have a late diagenetic origin. Liu et al. , Roberts and Weaver  and Rowan et al.  demonstrated that the formation of greigite and, therefore, remanence acquisition may be delayed by ∼10s to 100s of ka with respect to depositional age. Florindo and Sagnotti  suggested that growth of late diagenetic greigite occurred at least 150 ka after deposition of a volcaniclastic horizon. Larrasoana et al.  suggested that formation of greigite and pyrrhotite in marine sediments could occur over a time span ranging from a few thousand years to a few million years. In the Piano del Gaudo section, paleomagnetic data indicate a different temporal relationship between the remanence acquisition for magnetite and greigite, respectively, with respect to the time of tilting of the strata. The low-temperature magnetic component was acquired after tilting and is related to late diagenetic growth of authigenic greigite (Figure 2a). The high-temperature magnetic component is related to the presence of detrital magnetite and was acquired before tilting (Figure 2b). On this basis, we can establish that the growth of greigite occurred after the tectonic phase that was responsible for normal faulting and tilting of the stratigraphic sequence. This tectonic phase can be reasonably related to the formation of the nearby (2 km away) intramontane lacustrine Acerno basin, which has been dated between 400 and 260 ka [Munno et al., 2001]. Taking into account that the tectonic phase responsible of the onset of Acerno lacustrine sedimentation was probably active at ∼400 ka and that the maximum age of the volcaniclastic horizon in the Piano del Gaudo basin is 0.712 ± 0.018 Ma, then greigite growth could post-date deposition by as much ∼300 ka. This hypothesis is consistent with the results of Florindo and Sagnotti , and with the observed replacement of iron sulfide grains around relict volcanic minerals and the slow kinetics of sheet silicate dissolution. It has been observed that dissolution of silicate is a relatively slow process, and that the reaction half-life of sheet silicates with iron sulfides is of the order of hundreds of ka [Canfield et al., 1992]. Jiang et al.  and Roberts and Weaver  argued that the slow kinetics of silicate replacement by greigite provides one mechanism for remagnetization of sediments.
 This work shows that also greigite associated with pyrite was formed after tectonic tilting. Taking into account that sulphate is generally depleted below 12 m [Larrasoana et al., 2007], an additional sulphide source is required to form greigite in the lower part of the section. We propose two possible sources. One possibility could be related to oxidation of pyrite that can provide additional source of sulfide [Roberts and Weaver, 2005]. The occurrence of oxidation into the basin is suggested by the presence of hematite. A second possibility is based to the presence of several sulphate-rich water springs along the nearby (5 km away) Sele valley, which are alimented by the hydrogeological structure of eastern sector of Picentini Mountains [Duchi et al., 1995], i.e., the same sector where Piano del Gaudo lake was active in the Middle Pleistocene.
 In the Piano del Gaudo section, a syn-depositional magnetic signature has been modified by post-depositional greigite formation, which has partially or totally overprinted the primary magnetization carried by magnetite. Anoxic conditions within the small lacustrine basin and the availability of sulfur from volcanic minerals provided the conditions favorable for greigite growth. Greigite-bearing sediments acquired a chemical remanent magnetization after tectonic tilting of the sequence. Based on the inferred ages of the normal faulting that produced the tilting and on the timescales for reaction of sheet silicates with iron sulfides, late diagenetic greigite growth appears to have occurred up to ∼300 ka after deposition of the sedimentary succession.
 We are indebted to S. Lo Mastro for assistance with SEM analysis and F. Cifelle for assistance with sampling and field work. A. P. Roberts and an anonymous reviewer are gratefully acknowledged for their constructive reviews.