RESULTS

A modified form of the ternary classification scheme used by Comas, Zahn, Klaus, et al. (1996, fig. 5) is used to classify sediments in this study. The modification was necessitated by the sieve sizes employed. In Figure 8, this modification has the effect of "moving" data points towards the clay and fine silt (top) apex, parallel to the tie line between clay and fine silt and medium and coarse silt. Figure 8 confirms the visual description made by the shipboard scientists and, given the above, that the dominant textural lithology in lithostratigraphic Unit I throughout the Alboran Basin is clay and silty clay. Downhole distribution of sand (>63 µm), medium and coarse silt (63-32 µm + 32-20 µm), and fine silt and clay (<20 µm) is shown in Figure 9. The average and standard deviation for sand content is 4.1% ± 1.5% (Hole 976B), 4.0% ± 2.5% (Hole 977A), 4.4% ± 1.7% (Hole 978A), and 3.2% ± 2.4% (Hole 979A). For the medium and coarse silt fractions, these measures are 6.0% ± 1.6% (Hole 976B), 4.3% ± 2.% (Hole 977A), 8.0% ± 2.5% (Hole 978A), and 8.7% ± 2.7% (Hole 979A), and for clay and fine silt, they are 89.9% ± 2.8% (Hole 976B), 91.7% ± 3.9% (Hole 977A), 87.7% ± 3.1% (Hole 978A), and 88.1% ± 3.7% (Hole 979A). These data show a consistent downhole distribution of the textural components in Unit I in all four holes. No between-hole correlatable trends could be discerned.

XRD data were collected for each unsorted sample and for each of the four grain-size splits for each sample giving a total of 800 analyses. In addition, two repeats were run for unsorted samples (Samples 161-976B-1H-2, 64-68 cm, and 161-977A-1H-3, 53-57 cm) to check duplication of results, and blanks were run on the glass slides and plastic sample holders used during data collection. Normally, SIROQUANT quantification was performed twice on each sample, to check duplication on the unsorted, >63-µm, and <20-µm fractions, and in each case, the value derived from the second SIROQUANT analysis is reported, where these values fell within the error range reported for the first run. Otherwise, the analysis was rerun until two succeeding runs fulfilled this criterion. Because of time constraints, the 63- to 32-µm and 32- to 20-µm fractions were not analyzed using SIROQUANT. Again, this did not compromise the study, which was principally aimed at examining detrital mineral distribution in the sand and clay fractions.

Percentages for each of the minerals in the suite described above in the unsorted, >63-µm, and <20-µm fractions are given in Table 2, together with ² values for each analysis.

Downhole trends for quartz, "clay" (chlorite + illite + kaolinite + muscovite + talc), and total carbonate (dolomite + calcite) are shown in Figure 9. These values, with the addition of total feldspar (albite + bytownite + orthoclase), have been normalized after removal of authigenic pyrite, halite, and gypsum. For comparison, the original shipboard estimates of quartz + feldspar + rock fragments + mica (QFRM) have also been plotted on Figure 9 (column 2) for each hole. As the latter values were calculated from different samples to those used in the study reported here, they give a reasonable indication of downhole trends that is independent of sample location, and, therefore, provides an independent check for the data. Comparison of the original shipboard data with the quartz XRD data for Hole 976B shows that the trends seen in the former broadly correlate with the XRD-derived data in the upper half of the hole (peaks in both data sets at around 10, 50, and broadly from 130 to 170 mbsf), but cannot be matched in lithostratigraphic Unit I below 220 mbsf. No similar correlations exist for Holes 977A, 978A, or 979A, most probably reflecting large errors associated with visual estimation.

The "clay" group is considered to include both authigenic material (precipitated, and directly from the breakdown of feldspar and rock fragments) as well as detrital material, principally muscovite. These phyllosilicates have not been considered separately because SIROQUANT had difficulty in modeling their variable compositions using pure muscovite and talc hkl files. Similarly, the feldspar group has not been included because an unknown proportion of original detrital material has broken down into clay. For these reasons, only quartz has been used as an indication of detrital input.

Inspection of Figure 9 shows no consistent downhole trends for any of the mineral components in unsorted samples or in any of the grain-size splits, except for an inverse relationship between total carbonate, most commonly the dominant component, and either quartz and/or total "clay." One example of this inconsistent variability is seen in the total carbonate plots for the unsorted samples (Fig. 9, column 2). In this case values decrease downhole in Holes 976B, 977A, and 978A, but are more or less constant downhole in 979A. Similar inconsistent trends occur for all other components. Statistics have not been calculated for the individual mineral components because the data are not considered absolute as discussed earlier.

The only obvious variation in the bulk mineralogy of lithostratigraphic Unit I between holes is the direct relationship between water depth and >63-µm-fraction carbonate content. The highest sand-sized carbonate content (and correspondingly lowest quartz content) occurs in Holes 977A and 978A, both of which lie beneath almost 2000 m of water (Fig. 9, column 3), whereas the overall carbonate content is much lower in Holes 976B and 979A, both of which lie in just over 1000 m of water.

More interesting findings result from an examination of various mineralogical parameters and sedimentation rates between holes. These correlations are given in Figures 10 through 13. All between-hole correlations were undertaken using a downhole time scale in preference to depth to facilitate direct comparison and highlight temporal synchroneity. As this time scale is a derivative of depth (see above), it is subject to precision and accuracy errors in the resolution of the age determinations, and so exact time correlations cannot be expected. The correlations indicated on Figure 10, Figure 11, and Figure 12 were undertaken by visual matching of peaks in the smoothed downhole data only where these are defined by more than one point.

Figure 10, Figure 11, and Figure 12 show cross-hole correlations for unsorted, >63-µm, and <20-µm fractions for the components quartz, carbonate (dolomite + calcite), and quartz normalized against "clay" and feldspar, after the removal of dolomite, calcite, pyrite, gypsum, and halite (hereafter referred to as normalized quartz). The latter parameter was calculated to remove the dominating effect of pelagic carbonate to see if any underlying trends were apparent in the remaining components, principally the quartz. Curves so derived show that there are no clear basin-wide correlatable trends in the distribution of sand-sized quartz—apparent correlation in unsorted and >63-µm fractions is believed to be a partial artifact of much stronger carbonate correlations, which are described below. The good correlation of cycles of quartz distribution in the <20-µm fraction seen in the upper graph of Figure 12 is almost certainly attributable to the good correlation in the dominant carbonate fraction (Fig. 12, <20-µm carbonate), as all but the coincidence at ~0.5 Ma disappear after carbonate is removed (Fig. 12, <20-µm quartz).

In contrast, there is a reasonably strong basin-wide correlation in carbonate distribution between the sites, particularly for the <20-µm fraction. Correlation is least clear in the unsorted fraction (Fig. 10), probably because more obvious temporal correlations in the >63-µm (Fig. 11) and <20-µm (Fig. 12) fractions are not synchronous with each other. The frequency of cyclicity in the >63-µm fraction is ~0.75 m.y., whereas strongly correlated cycles in the <20-µm samples have regularly repeated every 0.5 m.y. since ~2.5 Ma.

Sedimentation rates across the basin are broadly correlated as indicated on Figure 13, although this is less convincing than the carbonate data. Between 1.0 and 2.0 Ma, peaks in the <20-µm carbonate cycles seem to correspond with sedimentation-rate highs, except at Site 978, but no such coincidence is observed in sediments younger and older than this. There is a moderately well-defined correlatable quartz peak in both the >63-µm and <20-µm data at around 2.6-2.7 Ma that coincides with the poorly defined sedimentation-rate high at that time. However, the sedimentation-rate high from 100 to 300 ka corresponds with upward decreasing carbonate, particularly in the <20-µm fraction, and highly variable quartz content.

SEM examination of selected samples confirmed shipboard determinations that the principal carbonate component of the sand (>63-µm) fraction is foraminifer tests, whereas the dominant carbonate component of the <20-µm fraction is calcareous nannofossils, but with fragments of foraminifer tests also present (Fig. 14). In addition, rounded to subrounded quartz is present in both the >63-µm and <20-µm fractions, with rounded to subhedral feldspar additionally present in the <20-µm fraction. This is interpreted to indicate the presence of detrital quartz in both the >63-µm and <20-µm fractions.