A total of 53 nannofossil events were recognized in the lower Oligocene to Pleistocene sediments studied (Tables T2, T3). This allows the subdivision of four upper Paleogene zones of Martini (1971) and his 21 Neogene to Quaternary zones or three late Paleogene zones of Okada and Bukry (1980) and their Neogene to Quaternary 15 zones (Fig. F3). A brief description of the result is given here with focus on biostratigraphic problems that occurred and solutions used during this study.
Upper Oligocene sediments were only recovered from Site 1148. The LO of Emiliania formosa marks the NP21/NP22 or CP16b/CP16c zonal boundaries. This event was also useful for defining the NP21/NP22 zonal boundary in adjacent areas of the SCS, such as in the Celebes Sea (Shyu and Müller, 1991) and in the West Philippine Sea (Shipboard Scientific Party, 2002). At Site 1148, only one specimen of E. formosa was observed at a depth of 671.89 mcd, above the FO of S. distentus. This specimen is clearly a reworked fossil. E. formosa is actually absent in the 851.4-m-long sequence at Site 1148, suggesting that Site 1148 does not penetrate through the NP21/NP22 zonal boundary and the sediments at the base of Site 1148 are still in the lowermost lower Oligocene Zone NP22 with an age of younger than 32.8 Ma.
In our materials, we did not observe typical R. umbilicus but found >14-µm-sized R. hillae. This allowed us to recognize the LO of R. umbilicus at 730.33 mcd at Site 1148. It is the earliest event at this site and serves as the marker for the NP22/NP23 boundary and assigns the interval below it to Zone NN22 or to Subzone CP16c (Table T3; Fig. F3).
In many open-ocean areas, for example in the Southern Ocean, the LO of Isthmolithus recurvus appears between the LO of R. umbilicus and the LO of E. formosa (Poore et al., 1982; Wei and Thierstein, 1991). This species is absent in our materials. It is also absent in the Eocene to the lower Oligocene sediments at Site 1201 in the West Philippine Sea (Shipboard Scientific Party, 2002). All of these imply that I. recurvus might be absent or rare in the low Pacific latitudes, supporting Wei (1992), who suggested that the FO and LO of I. recurvus are only applicable in the mid and high latitudes but not useful in low latitudes.
The FO of S. ciperoensis at Site 1148 was determined by separating typical S. ciperoensis from Sphenolithus aff. ciperoensis, using the method of Okada (1990). As a result, the FO of true S. ciperoensis was placed at depth 617.16 mcd and the FO of S. distentus was placed at 671.89 mcd (Table T2). In our materials studied, however, Sphenolithus aff. ciperoensis appears sporadically down to 655.52 mcd, much lower than the level of the FO of true S. ciperoensis, and differing from the case of ODP Leg 115, where S. ciperoensis and Sphenolithus aff. ciperoensis show the same stratigraphic range (Backman et al., 1990).
The next event, the LO of S. distentus, was determined at the depth of 477.34 mcd, marking the NP24/NP25 zonal boundary. Trace S. distentus was found occasionally; for example, it appears in a few samples of Core 184-1148A-48X, where redeposited carbonate mud clasts or layers were observed (Wang, Prell, Blum, et al., 2000). Therefore, we considered the trace occurrence of this species in that core as reworked. The near absence of S. distentus in Core 184-1148-49X, where several slumping layers are present (Wang, Prell, Blum, et al., 2000), indicates slumping occurred after the LO of S. distentus (27.5 Ma), or above Zone NP24. S. predistentus disappears with S. distentus together in the same sample at Site 1148 (Table T2).
The most difficult work at Site 1148 is to surely determine the events in Cores 184-1148A-47X and 48X, the upper Oligocene to lower Miocene interval. The problems were how to determine a number of late Oligocene LO events from sediments dominated by reworking, faulting, and slumping and to identify early Miocene events from poorly preserved fossils.
Sample 184-1148A-48X-6, 80–81 cm (462.32 mcd), contains common to few upper Oligocene Reticulofenestra bisectus, Zygrhablithus bijugatus, S. ciperoensis, and Chiasmolithus altus (Figs. F4, F5). In a normal sediment sequence, the sequence of the LOs of these species should appear as follows: R. bisectus (23.9 Ma), Z. bijugatus (24.5 Ma), S. ciperoensis (24.75 Ma), and C. altus (26.1 Ma). The appearance of these LOs in the same sample suggests that we did not see the real LOs of these species; the uppermost ranges of R. bisectus, Z. bijugatus, and S. ciperoensis, and, possibly, the uppermost range of C. altus, are truncated. In the same core, we observed abundant Sphenolithus capricornutus and Sphenolithus delphix in Sample 184-1148-48X-4, 81–82 cm, where rare R. bisectus, Z. bijugatus, and S. ciperoensis also appear (458.57 mcd) (Table T2; Figs. F4, F5).
Based on semiquantitative observations (Fig. F4), we took the common occurrence of S. ciperoensis as its LO (24.75 Ma) in Sample 184-1148A-48X-6, 80–81 cm (462.32 mcd). Its sporadic and few presence in the samples above was considered as reworking. Abundant S. capricornutus and S. delphix in Sample 184-1148-48X-4, 81–82 cm, were determined as their LOs (23.7 and 23.8 Ma, respectively). In this case, the presence of few R. bisectus and Z. bijugatus above 462.32 mcd was also considered to be due to reworking. Consequently, the LOs of R. bisectus and Z. bijugatus were not seen and their uppermost ranges were truncated. As for C. altus, it appears very sporadically in upper Oligocene sediments; we were unable to determine if its LO in this interval is caused by reworking, and therefore we did not use it for further discussion.
For defining the NP25/NN1 and the CP19b/CN1a zonal boundaries, we still followed Berggren et al. (1995), who used the LO of R. bisectus (23.9 Ma). The determination of the LO of S. ciperoensis and the truncation of the uppermost range of R. bisectus at Site 1148 suggests that the uppermost range of Zone NP25 was truncated as well, but its upper part exists at Site 1248 (Fig. F5).
Rio et al. (1990) found a short acme interval of S. delphix with S. capricornutus slightly below the FO of Discoaster druggii in the subtropic and tropic Indian Ocean, and these two species are virtually restricted to the upper part of Zone NN1 (Subzone CN1c). An abundant S. delphix interval between the LO of S. ciperoensis and the FO of D. druggii is also present in ODP Hole 807A from the Ontong Java Plateau (Fornaciari et al., 1993). According to these studies, the common S. capricornutus and S. delphix in Sample 184-1148-48X-4, 81–82 cm, suggest that at least the upper part of Zone NN1 is preserved at Site 1148.
Above Core 184-1148-48X, the FO of D. druggii, a marker for the NN2/NN1 zonal boundary, was difficult to identify in samples between 405.7 and 458.7 mcd from Core 184-1148A-47X because of heavy overgrowth of Discoaster species in this interval. D. druggii and other discoasters are overgrown, showing thick and wide rays and making identification difficult. After a careful examination, a few moderately preserved D. druggii were observed at 454.35 mcd, assigning the lowest occurrence of this species (23.2 Ma) at this level and defining the NN2/NN1 zonal boundary.
The Oligocene/Miocene boundary as defined by calcareous nannofossils is placed at the top of Zone NP25 by some authors and within Zone NN1 by others. According to Berggren et al. (1995), the age of the Oligocene/Miocene boundary is 23.80 Ma, placed within Zone NN1. The LO of S. delphix, with an age of 23.80 Ma, determined for Site 1148 allows us conveniently to place the Oligocene/Miocene boundary at the level of the LO of S. delphix, within Zone NN1 and between the depths of 457.82 and 458.57 mcd at Site 1148.
Above the FO of D. druggii, the FO of Sphenolithus belemnos was only obtained at Site 1148, defining the NN2/NN3 and CN1/CN2 zonal boundaries (Fig. F3). The LO of Triquetrorhabdulus carinatus appears much lower than the FO of S. belemnos at Site 1148; thus, it is not suitable for marking the NN2/NN3 zonal boundary in the area studied (Table T2). Fornaciari et al. (1990) estimated an age of 19.8 Ma for this event; however, we were unable to estimate the age of the LO of this species due to the lack of detailed age control, for example magnetic events in this interval.
The first event found at both Sites 1148 and 1146 is the LO of S. belemnos (Fig. F3), which marks the NN3/NN4 and CN2/CN3 zonal boundaries (Berggren et al., 1995). S. belemnos was observed between Samples 184-1146A-63X-CC, 33–39 cm, and 35–42 cm (633.76–643.1 mcd), placing the bottom of Site 1146 in Zone NN3 (18.3–19.2 Ma).
Above the LO of S. belemnos, the events recognized from Site 1148 can be well correlated with the same events from Site 1146 (Fig. F3). All ages and depths of these events are given in detail in Tables T2 and T3.
The next event is the LO of Helicosphaera ampliaperta, found at both sites, marking the NN4/NN5 or CN3/CN4 zonal boundaries (Fig. F2). The LO of Sphenolithus heteromorphus is used to define the NN5/NN6 or CN4/CN5a zonal boundaries in this study (Fig. F2). Within the Zone NN6 interval, we recognized two other events (i.e., the LO of Cyclicargolithus floridanus and the FO of T. rugosus) in agreement with several previous records (Olafsson, 1989; Rio et al., 1990; Raffi and Flores, 1995).
Discoaster kugleri is rare but continuously present in a short interval in the middle Miocene sequences at Sites 1146 and 1148, allowing the recognition of its FO and LO and the definition of the NN6/NN7 zonal boundary and the CN5a/CN5b subzonal boundary (Figs. F2, F3). Catinaster coalitus appears commonly to abundantly in our samples studied, and its FO serves as a good marker for the NN7/NN8 or CN5/CN6 zonal boundaries (Fig. F3). The LO of D. kugleri was found at a middle stratigraphic level in Zone NN7 at Sites 1146 and 1148.
Discoaster hamatus appears first with a notable abundance in the lower part of the upper Miocene sediments at Sites 1148 and 1146, assigning the NN8/NN9 and CN6/CN7 zonal boundaries for these two sites (Fig. F3). Zone NN8 or CN6 is a very short interval of ~10 m at Site 1146 and 4 m at Site 1148, in agreement with other studies (Raffi and Flores, 1995). The LO of D. hamatus marks the NN9/NN10 and CN7/CN8 zonal boundaries (Fig. F3).
Okada and Bukry (1980) used the FO of Catinaster calyculus to subdivide the CN7 Zone into Subzones CN7a and CN7b. However, its FO appears slightly below the FO of D. hamatus, about 2 m lower, for example, at Site 1148 and 3 m at Site 1146 (Tables T2, T3). Similar observations were reported from the Indian Ocean and the South Atlantic, respectively (Thierstein, 1974; Rio et al., 1990). Berggren et al. (1995) pointed out that the relationship between the FO of C. calyculus and the FO of D. hamatus is still poorly understood. Therefore, we did not subdivide Zone CN7. In the interval of Zone NN9, C. coalitus disappears first and then C. calyculus also becomes extinct below the LO of D. hamatus.
The top of Zone NN10 was defined by the FO of Discoaster quinqueramus (Martini, 1971), whereas that of Zone CN7 was defined by the FO of Discoaster berggrenii or the FO of Discoaster surculus (Okada and Bukry, 1980). Recently, D. quinqueramus and D. berggrenii were considered as synonyms, or D. berggrenii as a variety of D. quinqueramus (Rio et al., 1990; Raffi and Flores, 1995; Berggren et al., 1995). In our study, D. quinqueramus and D. berggrenii were still identified as two species and their FOs were found in the same sample at a site—for example, at 401.1 mcd at Site 1146 and at 240.92 mcd at Site 1148 (Tables T2, T3). Therefore, the NN10/NN11 and CN8/CN9 zonal boundaries can surely be defined (Fig. F2). Bukry (1973) further subdivided Zone CN8 into Subzones CN8a and CN8b by the FO of Discoaster neorectus and/or by the FO of Discoaster loeblichii. These two species are easily identifiable without confusion (Perch-Nielsen, 1985). However, they appear down to Zone NN9 in our northern Sites 1146 and 1148 and in the southern Site 1143 of the SCS (Wang, Prell, Blum, et al., 2000). According to Xu (1996), the same case appears in sediments from the offshore basin of the SCS. The much earlier appearance of these two species made the subdivision of Zone CN8 impossible.
In our samples, D. berggrenii disappears slightly later than D. quinqueramus, suggesting a slight difference in their stratigraphic ranges. In this paper, we used the LO of D. quinqueramus to define the NN11/NN12 and CN9/CN10 zonal boundaries (Fig. F3). The FO of Amaurolithus primus was used to subdivide Zone NN11 into Subzones CN9a and CN9b, in line with the definition of Okada and Bukry (1980). The FO and LO of Amaurolithus amplificus appear within Zone NN11.
The FO of Ceratolithus acutus and the LO of T. rugosus were found at the same level within Zone NN12. We differentiated Subzones CN10a and CN10b, following Bukry (1973).
According to Berggren et al. (1995), the age of the Oligocene/Miocene boundary is 23.80 Ma (Fig. F2). Recent studies suggest an age of 5.34 Ma for the LO of C. acutus, which is located within Zone NN12 and marks the top of Subzone CN10a (Young et al., 1994; Berggren et al., 1995). The next event above it is the FO of Ceratolithus rugosus with an age of 5.07 Ma, defining the tops of Zone NN12 and Subzone CN10b.
Following Berggren et al. (1995), we placed the Miocene/Pliocene boundary between the LO of C. acutus and the FO of C. rugosus in the upper part of Zone NN12 or between Subzones CN10a and CN10b.
Above the FO of C. rugosus, determination of the FO of D. asymmetricus in the sequences was somewhat difficult; because five rayed discoasters in that interval are poorly preserved, most arms were broken or lost. The FO of Discoaster asymmetricus in this study was determined based on finding relatively complete specimens. C. acutus becomes extinct somewhat earlier than the FO of D. asymmetricus. The LO of Amaurolithus tricorniculatus or Amaurolithus spp. marks the tops of Zones NN14 and CN10 (Fig. F2).
In this study, R. pseudoumbilicus is identified for specimens having a maximum coccolith length >7 µm, in accord with Gartner (1967). The LO of Sphenolithus abies/neoabies appears shortly after the LO of R. pseudoumbilicus; the interval between these two events is ~1 m at Site 1146 and 3 m at Site 1148 (Tables T2, T3)
The LO of D. surculus is easily recognizable in the sections studied, and it was used to mark the NN16/NN17 and CN12/CN13 zonal boundaries (Fig. F2). The LO of D. tamalis was found in the relatively long interval of Zone NN16 in the Pliocene, subdividing this interval into Subzones CN12a and CN12b.
The tops of Zone NN17 and Subzone CN12c were defined by the LO of Discoaster pentaradiatus (Fig. F2). Zone NN17 is very thin at these two sites, ~10 m, in agreement with the record of this zone in other areas.
Reworking of discoasters is very common in the upper sections at Sites 1146 and 1148, causing difficulty in determining the LO of D. brouweri. The event was determined by its appearance with reasonable abundance.
The Pliocene/Pleistocene boundary as defined by calcareous nannofossils is usually set above the LO of D. brouweri. The appearance of larger species of Gephyrocapsa gives an indication of the younger age of the sediment (Perch-Nielsen, 1985). In our study, the Pleistocene/Pliocene boundary is constrained by the FO of Gephyrocapsa (medium) spp. and the LO of D. brouweri and is located between 1.69 and 1.96 Ma.
The lowermost Pleistocene Zone NN19 was defined from the LO of D. brouweri to the LO of Pseudoemiliania lacunosa. The LO of Calcidiscus macintyrei and the FO and LO of R. asanoi were recognized within this interval. H. sellii is rarely and sporadically present in the Pliocene and lower Pleistocene sequences at Sites 1148 and 1146 in the northern SCS and also at Site 1143 in the southern SCS (Wang, Prell, Blum, et al., 2000); thus, its LO is not determinable in this study.
Okada and Bukry (1980) used the FO of Gephyrocapsa carribeanica and the FO of Gephyrocapsa oceanica to subdivide the interval of Zone NN19 into Subzones CN13a, CN13b, and CN14a. As discussed in the taxonomic notes in this paper, there are morphological intergradients between different Gephyrocapsa species and a detailed study is needed to find reliable and applicable criteria to separate them. Thus, in the present study, we did not make a further subdivision for this interval.
The interval characterized by absences of P. lacunosa and E. huxleyi but by abundant Gephyrocapsa oceanica was defined as Zone NN20 and Subzone CN14b (Fig. F2). The FO of E. huxleyi was determined both by means of light microscope and scanning electron microscope and marks the bases of Zones NN21 and CN15. Gartner (1977) suggested this to be the acme of E. huxleyi. The determination of the acme of E. huxleyi was based upon a semiquantitative method.
The ages of these events were plotted against their depths to show variation in average linear sedimentation rates (Fig. F6). Sedimentation rates at Site 1146 are higher than those at Site 1148, most likely a result of having more terrestrial supplies due to its location closer to shore than Site 1148. Both sites show the same trend of downcore variation in sedimentation rates.
A very high sedimentation rate (35.6 cm/k.y.) was seen in the Oligocene section at Site 1148 and is probably a result of seafloor spreading (Wang, Prell, Blum, et al., 2000).
A significant unconformity was seen in the transition from upper Oligocene to lower Miocene. The nannofossil record indicated the absence of the uppermost part of Oligocene Zone NP25 and the lower part of Miocene Zone NN1. The time span of the unconformity was estimated to be ~1 m.y. based on the LO of S. ciperoensis (24.75 Ma) in the upper part of Zone NP25 and the LOs of S. capricornutus and S. delphix (23.7 and 23.8 Ma) in the upper part of Zone NN1. The sedimentation rate is only ~3 cm/k.y. in this interval. The sedimentation rate in the interval between the LOs of S. delphix and S. distentus (27.5 Ma) is also very low—only 5 cm/k.y.
Sedimentation rates increase since the early Miocene and reach the highest values in Pleistocene—for example, from 14.5 to 22 cm/k.y. and to 64.6 cm/k.y. at Site 1148, although it reaches to 105 cm/k.y. in the Pleistocene at Site 1146. This agrees with Wang, Prell, Blum, et al. (2000), who found Miocene sediments rich in carbonate in conjunction with a general increase of noncarbonate sediment accumulation after 2–3 m.y. and a more significant increase in the latter part of the last million years.