Previous work on terrigenous material in deep-sea sediment indicates that given similar extraction procedures and analytical methods, measurements of grain size are reproducible (e.g., Rea, 1994). For closely spaced lower Paleogene sediment samples at Site 215, the median grain size determined for the extracted terrigenous component is similar to that reported by Hovan and Rea (1992). We therefore conclude that our approach for isolating and examining the terrigenous component will render robust grain-size information.
The grain-size distributions of the terrigenous component extracted from upper Paleocene–lower Eocene sediment at Site 1267 are similar to those observed for the terrigenous component extracted from contemporaneous sediment at nearby Site 527 (Rea and Hovan, 1995). In both cases, the isolated component has a flat pattern characterized by low weight percents of material spread across a wide grain-size range (Fig. F2). This pattern suggests that hemipelagic material dominates the terrigenous component (Rea and Hovan, 1995; Boven and Rea, 1998).
Grain-size analyses of the terrigenous component in time-coincident samples at Site 1263 clearly vary from those at Site 1267 (Fig. F2). Isolated samples from Site 1263, the shallower location, contain a pronounced grain-size mode centered between 5 and 7 µm. Such a grain-size distribution is consistent with a large fraction of eolian material in the terrigenous component (Boven and Rea, 1998).
Even in open-ocean settings, the terrigenous component of deep-sea sediment may comprise a mixture of eolian and hemipelagic material (Rea, 1994; Rea and Hovan, 1995). Using extensive data sets of terrigenous grain-size distributions from Pacific Ocean sediments, Boven and Rea (1998) developed a two-component mixing model to determine the relative proportion of each component. This approach is based on the observation that hemipelagic grain-size distributions are broad with peak weight percents typically low near 1%, whereas eolian grain-size distributions depict a dominant mode with greater peak weight percents near 5% (Boven and Rea, 1998). The model compares the peak weight percent of a sample's grain-size distribution to end-member eolian and hemipelagic distributions. According to this model, our results suggest that the terrigenous component of sediment deposited near the P/E boundary contains ~6%–11% eolian material at the relatively deep Site 1267 but as much as 20%–40% eolian material at the relatively shallow Site 1263. This indicates that the shallow site received a higher concentration of eolian material than did the deep sites. However, sites 1263 and 1267 have likely received a similar flux of eolian material through time because they are proximal. This indicates that the deep flank of Walvis Ridge received a higher flux of hemipelagic sediment than the shallow crest. This finding conforms to general views of hemipelagic deposition, considering the paleodepths of Sites 1267 (~3200 mbsl at ~55 Ma) and 1263 (~1500 mbsl at ~55 Ma). Hemipelagic material should have settled and moved at >2000 m water depth, such that greater amounts of hemipelagic material would have accumulated at Site 1267 (and Site 527).
An eolian grain-size record can probably be gleaned from the lower Paleogene sequence recovered at Site 1263 on Walvis Ridge. However, there is an issue regarding measurement because, even at this shallow location, the terrigenous component contains a mixture of hemipelagic and eolian material. Large amounts of hemipelagic material broaden the grain-size distribution so that the median and modal grain size of the terrigenous component differs significantly (Fig. F2). Whereas previous studies of eolian grain size have generally used the median for paleoenvironmental reconstructions, we suggest that the mode more accurately measures the grain size of the eolian fraction in mixed hemipelagic-eolian systems (where an eolian signal is detectable). This is particularly evident in samples with lower eolian percents, such as the Site 1263 sample illustrated in Figure F2A. Figure F2A shows the lowest Site 1263 eolian percent and the highest mode-median offset of 1.1 µm, whereas samples shown in Figure F2B–F2E have eolian percents ranging from 32% to 38% and a mode-median difference of only 0.1–0.2 µm. Further, samples collected from Site 215 that were shown to be of eolian origin (Hovan and Rea, 1992) record a mode-median average difference of 0.1 µm.