RESULTS

Three log facies are recognized in the FMS images. Warping and contortion of bedding in Hole 1118A (280-285 and 320-330 mbsf) indicate slumping, but this deformation is so minor that we do not assign these deposits to an additional facies. The criteria used for facies recognition are the grain sizes and bed thicknesses of both sand-silt turbidites and intercalated muds. The grain sizes reported below are taken from handwritten shipboard core descriptions in the same intervals as the FMS images that characterize each facies. Because the images have a lower resolution than visual descriptions of cores, sedimentary structures could not be used to define log facies.

Facies A dominates the succession at Sites 1118, 1109, and 1115 (Fig. F4). It consists of graded beds of silt and very fine grained to medium-grained sand alternating with beds of silty to clayey mud (Fig. F5). The proportion of mud varies from ~35% to 80%; it is commonly moderately to intensely burrowed (Fig. F6C). In cores (Fig. F6) and FMS images, the sand and silt beds show sharp bases and gradational tops; a few beds have irregular scoured bases. Sand beds range in thickness from <5 to 50 cm, whereas silt beds are 10 to 100 cm thick.

Facies B consists mostly of thick alternating intervals of silt and clayey mud; 5- to 20-cm-thick beds of sand are rare (~5%). The silt beds are 200 to 800 cm thick, and the clayey muds are 200 to 800 cm thick. Silt beds appear to be largely ungraded. Cores that sample the muds and some of the silts show bioturbation (Fig. F7). The silts are locally laminated.

Facies C is restricted to Site 1118, closest to the axis of the rift basin. It consists of interbedded very fine grained to medium-grained sands (~40%), silts (~30%), and clays (~30%). The greater sand content and coarser overall grain size of this facies are shown by the DPORO profile in Figure F3 from ~700 to 850 mbsf. The normally graded sands are mostly 5 to 80 cm thick, with rare graded beds 150 to 230 cm thick. These beds have sharp bases, gradational tops, local wavy to parallel laminations, and scattered mud clasts (Figs. F8, F9).

The graded silt and sand beds that characterize Facies A and C are interpreted as turbidity current deposits, consistent with the conclusions of shipboard scientists (Robertson et al., in press). The thick silt beds of Facies B are not believed to be turbidites. The widespread burrowing suggests that they are hemipelagic deposits that formed largely by rain-out from suspension. A similar depositional mechanism is inferred for the clayey muds of Facies B.

Bed-thickness values for sand and silt turbidites were transformed to logarithms, binned into ~20 classes, and plotted as bar graphs (Fig. F10). These plots suggest that the bed-thickness populations are not lognormal—although there is a central mode, the populations are positively skewed with a long tail of particularly thick beds (Crowther, 2000). The tail of thicker beds is better accommodated by exponential and power-law models. At all sites, there is an approximately exponential decrease in frequency with increasing bed thickness, with a long tail of beds much thicker than the mode (Fig. F11).

Plots in log-log space of the number of beds thicker than T, vs. T, consist of segmented curves with one or more dominant linear trend (Fig. F12). These linear segments suggest power-law relationships, with values (Equation 2) ranging from 1.4 to 5.6, but mostly <3.0. These values are higher than for other turbidite successions (Hiscott et al., 1992, 1993; Pirmez et al., 1997) because the range in bed thickness is small (most beds <100 cm thick), even though the number of beds is large. In all plots, data points for T <~10 cm deviate from a straight-line trend.

Plots of bed number vs. interpolated age for each site (Fig. F13) show some similarities but also striking differences across the asymmetric rift basin. Bathyal conditions (water >150 m deep) occurred first at Sites 1109 and 1115. The number of sand and silt turbidite beds that accumulated per million years rapidly reached ~930 at Site 1109 and ~440 at Site 1115 (recurrence intervals of 1075 and 2270 yr, respectively). These high turbidity-current frequencies persisted until at least 3.45 Ma at Site 1109 and 3.1 Ma at Site 1115. Because of a gap in the FMS data at Hole 1109D (Fig. F4), it is possible that both these sites continued to receive turbidites at high frequency until ~3.1 Ma. After ~3.1 Ma, the frequencies of turbidity current events were reduced at Sites 1109 and 1115 (respectively ~170 and ~70/m.y.; recurrence intervals = ~5,900 and ~14,300 yr).

Site 1118 began to subside strongly by ~3.6 Ma, and until 3.45 Ma (or perhaps as late as ~3.2 Ma; see above) was receiving deposits of turbidity currents at the same rate as Site 1109 (Fig. F13). After that time, even though the frequency of depositing turbidity currents reaching Sites 1115 and 1109 decreased markedly, the frequency at Site 1118 remained high (~1500 beds/m.y.; recurrence interval = 670 yr). This continued until 2.6 Ma.

Between 2.6 and ~2.0 Ma, all sites recorded the arrival of turbidity currents at a sharply reduced rate as compared with rates that prevailed at 3.5 Ma. There was an increase in the frequency of turbidite deposition from 2.0 to 1.8 Ma at Sites 1109 and 1118. FMS data are not of sufficient quality in Hole 1115C to reveal the same increase, but cores deposited from 2.0 to 1.8 Ma at this site (~90-110 mbsf) are uniformly muddy, suggesting that it did not receive the same increased influx of turbidites.

In plots designed to reveal power-law tendencies in bed-thickness populations, the deviations in straight-line behavior for T < 10 cm (Fig. F12) may be the result of poor resolution of the very thinnest sand and silt beds in the FMS images, so that beds only a centimeter or so thick may have been missed. Alternatively, it is possible that the thinnest turbidites are mainly mud grade rather than consisting of silt or sand, so that they were not recognized or counted. Facies descriptions by the shipboard party (Taylor, Huchon, Klaus, et al., 1999) indicate that ~1-cm-thick beds are common, favoring the first alternative.

The segmented nature of these graphs for T > 10 (Fig. F12) can be explained using suggestions published by Malinverno (1997). He shows that natural bed-thickness data sets can be expected to plot as segmented linear trends with different slopes (i.e., different values, Equation 2) if there is a relationship between bed length and bed thickness that depends on the bed volume. For example, smaller flows and their deposits might be confined by topographic features in the basin (e.g., channels or subdued depositional topography), whereas larger flows might be free to spread over wider areas (Rothman and Grotzinger, 1995; Malinverno, 1997). In this case, the value would be greatest for the thicker beds (larger flows), just as in Figure F12. Multichannel seismic profiles between Sites 1118 and 1109 show a broad north-northwest-trending trough (paleochannel) at the level of a widespread erosional surface situated below the Pliocene turbidite succession (Goodliffe et al., 1999). Site 1109 was drilled in this trough and Site 1118 along its flank. Confinement of turbidity currents by such underlying topography might have prevented the free spreading of flows across the floor of the rift basin, thus accounting for the segmented nature of bed-thickness plots. Channels with smaller dimensions (<10 m depth) might also have been present but are unresolved in the multichannel seismic data.

The Woodlark Basin is a site of active rifting and profound subsidence, where earthquakes are potentially responsible for triggering turbidity currents and other sediment gravity flows (e.g., silt-laden liquefied flows or debris flows) like those that generated the Pliocene basinal succession. Earthquake magnitudes typically follow power-law distributions (Turcotte, 1989); hence, earthquakes have been proposed as a triggering mechanism for turbidity currents whose deposits likewise conform to power-law distributions (Hiscott et al., 1993; Beattie and Dade, 1996). According to Kuribayashi and Tatsuoka (1977) and Keefer (1984), earthquakes with magnitudes of >5.0 are required to cause significant liquefaction for the initiation of sediment gravity flows. As the distance from the epicenter increases, so does the minimum magnitude required for liquefaction, so that a magnitude 7.0 shock can liquefy sediment as far as ~100 km from the epicenter. The bed-thickness data sets from Holes 1118A, 1109D, and 1115C are broadly consistent with power-law behavior, so that seismic triggering is a possible explanation for the genesis of sediment gravity flows during the late early Pliocene to late Pliocene. Turbidity-current recurrence intervals of ~1000 yr to several thousands of years (Fig. F13) are similar to those determined for thin-to medium-bedded turbidites from the Izu-Bonin forearc basin by Hiscott et al. (1992, 1993), another area with strong seismic activity. Hiscott et al. (1992, 1993) found these recurrence intervals for turbidites to be one to two orders of magnitude greater than the recurrence intervals for large-magnitude earthquakes in the subduction zones of the western Pacific. A Web-based search (wwwneic.cr.usgs.gov/neis/epic/epic_rect.html) of the number of earthquakes of magnitude 4.0 since 1973 in two areas, one south of Japan (33°-35°N, 135°-139°E) and the other in the source area for the Woodlark Basin (8°-10°S, 150°-153°E), identified 389 and 78 events, respectively. Hence, seismic events are less frequent in the rift setting than above subduction zones by a factor of ~5, and depositing turbidity currents are estimated to occur ~0.5-1.5 orders of magnitude less frequently than potential seismic triggers. One must imagine that the availability of seismic triggers is not alone sufficient to generate turbidity currents; in addition, failures large enough to initiate far-traveled turbidity currents must only occur after an extended period of deposition on basin slopes to accumulate the unstable mass that eventually fails (Hiscott et al., 1993).