Sedimentation at the Blake Ridge sites (Sites 994, 995, and 997) was uninterrupted through the late Pliocene and Quaternary, resulting in an undeformed and continuous sequence of nannofossil-rich clays extending below 100 mbsf at all three sites. Interstitial methane and sulfate profiles indicate limited seawater diffusion and no evidence of significant methane transport into the sulfate reduction zone, in contrast with Site 996. As a consequence, the Blake Ridge sites show the normal pattern of magnetic diagenesis in reduced marine sediments, which is illustrated by the MST susceptibility record for the upper part of Site 995 (Fig. 1). Magnetic diagenesis in reduced sediments involves a series of iron reduction steps within the first few meters below the seafloor, beginning with authigenesis of magnetite, through generation of incompletely reduced iron sulfides, including the ferrimagnetic phase greigite, to complete iron reduction to pyrite (Canfield and Berner, 1987; Karlin, 1990; Leslie et al., 1990). Susceptibility in the three Blake Ridge sites increases from the seafloor to about 2 mbsf. The increase in susceptibility presumably results from authigenesis of magnetite, adding to the depositional load of detrital magnetite. A decrease in susceptibility immediately below this peak reflects a first stage of further reduction of magnetite to sulfides; a second peak at about 3 mbsf may represent growth of greigite, before susceptibility precipitously declines at about 3.75 mbsf as iron sulfides are fully reduced to pyrite, which is paramagnetic.
On a Day plot, samples from the A995 set from Site 995 plot in the PSD field, or just to the right of the PSD field, indicating an admixture of MD or SPM grains with PSD grains (Fig. 2A). Samples migrate across the field in a systematic way with increasing depth, following a path that suggests an increase in the proportion of SPM grains from 0.13 to 0.6 mbsf, a decrease in the proportion of SPM grains from 0.6 to 3.83 mbsf, and an increase in the proportion of MD grains from 3.83 to 28.60 mbsf. Figure 3 shows the downhole behavior of DJH for the A995 set, which summarizes the trends seen in the Day plot.
Because the A995 set of anoxically stored samples is small, we chose to examine the rock magnetism of the larger B995 set, to see whether this pattern of behavior could be confirmed. However, the B995 set plots over a much wider region on a Day plot than does the A995 set (Fig. 2B), and the B995 samples do not show a systematic trend. The origin of the differing behavior of the two sets is suggested by comparison of the downhole distribution of the rock-magnetic parameters Jrs, Js, Hcr, and Hc of the two sample sets (Fig. 3). Below the susceptibility-decrease horizon at 3.75 mbsf, most B995 samples have Jrs (and less distinctly Js) values that are lower than the A995 samples. The difference between the two sets can be explained if greigite makes up a significant proponent of the remanence-carrying (i.e., larger than SPM) grains below 3.75 mbsf. Oxidation of greigite in the B995 samples would have produced a decrease in saturation remanence and magnetization. Oxidation also appears to have reduced and scattered coercivity of remanence, and scattered coercivity to both higher and lower values; such apparent changes to Hcr and Hc may merely reflect the very low saturation remanences of many of the B995 samples, which makes accurate determination of coercivity parameters very difficult. Scatter in both Hc and Hcr results in scattered, nonsystematic behavior in DJH for the B995 samples.
Despite its small size, the A995 set appears to be a more faithful record of the in situ magnetic diagenesis in the reference site than the B995 set. Surprisingly, the increase in susceptibility over the first 2 mbsf at Site 995 is accompanied by an initial decrease in Hc, Hcr, and DJH, indicating a decrease in the proportion of more stable, single-domain-sized ferrimagnet grains in the total magnetic population (Fig. 3). At the seafloor, DJH has a value >0.1, representing a position in the middle of the PSD field. At about 0.6 mbsf, DJH reaches a minimum of about 0.02, which represents an average ferrimagnet domain state just within the PSD field, very close to the boundary with the SPM/MD fields. This occurs despite an increase in saturation magnetization, and so may represent addition of SPM or near-SPM grains, occurring at the same time as SD magnetite is being lost. Magnetotactic bacteria have been reported to produce magnetite and greigite grains in the SD size range (e.g., Heywood et al., 1990), whereas individual ferrimagnet grains produced by dissimilatory bacteria are commonly smaller, in or near the superparamagnetic range (Lovley et al., 1987; Lovley and Phillips, 1988). Reduction in anaerobic sediments from the Chile margin at ODP Site 863 (Musgrave et al., 1995) also appears to have involved production of near-SPM greigite grains.
In the set of anaerobically stored samples, coercivity and DJH recover slightly over the interval from 1 to 4 mbsf. Both saturation magnetization and saturation remanence initially increase to 2 mbsf, and then decrease sharply. The change in DJH represents a small shift away from the SPM/MD field, and the trends in both DJH and saturation magnetizations below 2 mbsf could be explained by further reduction of ultra-fine-grained (near-superparamagnetic) magnetite and/or greigite to pyrite.
The subsidiary peak in susceptibility at 3-4 mbsf may represent an intermediate step in which greigite is generated, before further reduction to pyrite. At least some of this greigite may be SD in size as indicated by the trend in DJH over this interval and by the evidence that remanence-carrying greigite has been destroyed by oxidation below this depth. Below 4 mbsf, and extending down to the bottom of Site 995 at 700 mbsf, DJH mostly remains below 0.05, near the boundary between the PSD and SPM/MD fields. The trend on the Day plot suggests further conversion of SD or PSD greigite and/or magnetite grains to pyrite, leaving a residuum of larger, MD grains. Saturation remanence, saturation magnetization, pARM intensity, and susceptibility all decrease sharply below 4 mbsf, to substantially lower than the near-seafloor values (although scattered exceptional samples show higher values deeper in the sequence). Little of the original detrital load of magnetite survives below 4 mbsf, and the finer fraction of any ferrimagnetic minerals present—both any PSD or SD grains that may have been present in the detrital magnetite, and SPM or SD magnetite and greigite produced in the earlier stages of reduction—has been mostly reduced through to pyrite.
Samples from less than 1 mbsf have a single coercivity population, shown by histograms of pARM and dARM (Fig. 4). The peak in these histograms occurs at about 25-35 mT, typical of MD to PSD magnetite, consistent with a detrital origin. Below 4 mbsf pARM and dARM histograms fall into two classes: they are either strongly left-skewed, with a single peak at about 15 mT, or they are bimodal, with peaks at 15 and 35-55 mT. The left-skewed samples may be explained by preferential loss of finer grained, SD and PSD magnetite, relative to the coarser MD fraction. The samples with two peaks suggest two populations of magnetic phases. Thermal demagnetization of multicomponent isothermal remanence (Shipboard Scientific Party, 1996d) indicates the presence in these samples of a thermally metastable phase (decomposing at 250º-300ºC) with coercivity higher than magnetite, tentatively identified as greigite. The presence of greigite was positively confirmed by XRD in Site 995 samples. It is likely that the higher coercivity population present in these samples includes SD greigite.