DESCRIPTION AND RESULTS

Green clay layers are mainly concentrated in the uppermost Pliocene-Pleistocene section and, for Site 1148, in the lower Miocene section (Wang, Prell, Blum, et al., 2000). These centimeter-scale layers, consisting of generally coarser and stiffer material than background sediments, which are composed of nannofossil ooze and detrital clays, can be as thick as 3 cm and occur in high numbers, up to hundreds per core (i.e., Cores 184-1143B-12H and 184-1143C-4H) (Wang, Prell, Blum, et al., 2000). Discrete layers as well as discontinuous and bioturbated green intervals were observed (Fig. F2), and in many occurrences a clear association with iron sulfide minerals and pyrite nodules was found (Wang, Prell, Blum, et al., 2000). Smear slide analysis performed during visual inspection of the cores indicates that these layers are relatively poor in carbonate. A green mineral was recognized on smear slides and interpreted as being glauconite, but XRD analysis performed on board did not validate this observation. After the plastic bags containing the sampled sediments were opened, about two months after sampling, the green color characteristic of GCLs had disappeared and the color difference between the green clay and the host sediment samples was less distinct.

The studied sediments, both the GCL and host material, are fine grained, with material <8 µm averaging 82%. Samples are composed of phyllosilicates, which average 17%, and quartz, which represents ~12% of the GCL and 13% of the host sediments. Other minor constituents are K-feldspar, plagioclase, and calcite, which together represent not more than 13% of bulk mineral composition. On average, ~50% of the bulk mineralogical composition of each sample is represented by the "unquantified" mineral phase (see "Methods") (Table T3). Small amounts of pyrite were detected in both GCLs and normal sediments (Table T3). Micas (s.l.) and chlorite constitute the most abundant clay minerals detected in the carbonate-free <2-µm and 2- to 16-µm fractions. Micas average 52% and 49% of the <2-µm and 2- to 16-µm fractions, respectively, whereas chlorite represents 40% and 23% of the same fractions. Kaolinite is present in lower percentages, averaging 6% and 9% of the <2-µm and 2- to 16-µm fractions, respectively. Smectite, detected in the <2-µm carbonate-free fraction treated with ethylene glycol, makes up ~18% of the fraction. Generally, micas are of detrital origin, as it is shown by their sharp peaks (Fig. F3).

Background sediments bear ~25% more calcite than GCLs, whereas no consequential difference in pyrite percentages was observed. GCLs are generally coarser (average = 19.3%; >8 µm) than background sediments (average = 15.1%; >8 µm). Both GCLs and background sediments contain the same amount of K-feldspar, plagioclase, and phyllosilicates, mainly chlorite and smectite (Table T3).

As stated by several authors (see Moore and Reynolds, 1997), the term glauconite (used for a Fe-rich mica) is "ambiguous" and the distinction between glauconite and illite is not straightforward from both a chemical and a mineralogical point of view. Because the main difference between glauconite and illite is the weak or even absent 002 reflection in the glauconite XRD profile, due to a higher Fe content in the glauconite octahedral layers (Moore and Reynolds, 1997), it is rather difficult to detect glauconite by XRD. Although there is no direct evidence of the presence of glauconite in our samples, we cannot exclude it. In fact, a detailed analysis of clay minerals shows that mixed-layered clays are present in the GCL (Fig. F3), although in small amounts. These mixed-layered clays consist mainly of illite-smectite (IS) and chlorite-smectite (CS). We are able to estimate the content of Fe-rich mica at ~80%-85% and of chlorite at 75%-80% in GS and CS, respectively (Moore and Reynolds, 1997).

Concentration of the different phosphorus sedimentary phases and ferric iron are reported in Table T3. Fe-bound P and authigenic P are the most abundant phases, and their concentrations average 0.11 and 0.3 mg/g, respectively. Detrital P and organic-bound P are generally low, averaging 0.043 and 0.07 mg/g, respectively (Table T3). Ferric iron is present in relatively high quantities, averaging 2 mg/g. XRF results are shown in Table T4, and only few analyzed elements display important variations between normal sediments and GCLs. On the whole, GCLs bear higher concentrations of iron, nickel, and potassium compared to host sediments, whereas the latter are characterized by slightly higher amounts of authigenic P, CaO, barium, and strontium (Tables T3, T4). Rock-Eval data indicate that GCLs contain lesser amounts of organic carbon, which averages 0.26 wt% in GCLs and 0.34 wt% in host sediments.

We have interpolated the age for each GCL at meters composite depth using the shipboard splices and age models, and thus have a complete record of GCLs for each of the studied sites. The GCL's time distributions are represented in Figure F4. On average, eight GCLs per 20 k.y. are recorded in the studied sites (corresponding to 2.5 occurences per thousand years), and peaks of more than 10 occurrences per 20 k.y. are observed at Site 1143. Generally, GCLs are more recurrent in the upper part of the sequence. Sites 1143 and 1148 present the most continuous records of GCLs (Fig. F4).

All records were tested for randomness, using ² statistics, in order to assess if patterns or trends in the temporal distribution of GCLs exist. For all sites except Site 1143, the results indicate that GCLs are not distributed randomly in time (Table T5). A trend in the length of the intervals between GCL events is observed for Sites 1145 and 1146, with intervals becoming shorter toward the upper part of the sequence (Fig. F5). At Site 1148 it is possible to detect a cyclicity. We executed autocorrelation analysis on the detrended records of intervals between events, and positive peaks on the autocorrelogram were statistically significant only for Site 1148 (at 80% and 90% level of confidence) (Fig. F6). These peaks are located at 127, 140, 204, 270, and 477 k.y. Spectral analysis of GCL records from the other sites did not highlight any statistically significant cyclicity.