The bimodal distribution of population 1 may indicate that the group actually consists of two populations. Figure F10B would allow definition of second population in this group. However, for the following discussion use populations 1-3 as defined in the data population section.
To check the significance of the populations for interpretation in terms of depositional processes, we plotted the data groups with their mean grain size against depth and age (Fig. F12). Blank areas within the first column indicate areas of no sample coverage. The mean of the grain-size populations correlates in general with the depositional energy of a system or depositional process. An arbitrarily chosen definition of the populations led to a random downhole distribution. Instead, the populations show distinct groups of spatial and temporal occurrence and absence. Population 1 is most common in four distinct time/depth intervals (intervals 1, 2, 4, and 5). Populations 2 and 3 are more continuous, except that population 3 is less common in younger/shallower deposits.
Our tentative correlation with depositional environments has been drawn from a comparison to a sequence stratigraphic model developed on board the ship (see "Lithostratigraphy" in Shipboard Scientific Party, 1999b). Population 3 is clearly related to bioturbated upper sequence boundaries that may be influenced by contourite currents. Population 1 is certainly glacial, with its maximum values correlating to ice-rafting events. However, most of population 1 may be indirectly ice-derived shelf material redeposited by turbidites. Especially noteworthy is the low number of population 1 events around 3-6 Ma, a global warm period (e.g., Messinian salinity crisis).
The depositional significance of population 2 is less certain; we assume an interglacial turbiditic origin. The density curves in Figure F12 represent counts of events at a certain depth and age ranges normalized to the total number of samples analyzed within this interval. The density curve of the glacial population 1 may therefore potentially be linked to parameters such as ice volume, shelf ice extent, or periods of advance and retreat of ice sheets.