Among several Pleistocene zonations proposed by Martini (1971), Bukry (1973, 1975), Gartner (1977), Okada and Bukry (1980), and Martini and Muller (1986), Gartner's zonal scheme, which recognizes seven zones within the Pleistocene, is the most
popular and was used in this study as the framework of our Pleistocene biostratigraphic zonation. In recent years, nannofossil workers have correlated the Pleistocene nannofossil events with oxygen isotope and paleomagnetic data in order to establish a more precise and detailed Pleistocene nannofossil zonal scheme for Zone NN19 (e.g., Pujos, 1985; Takayama and Sato, 1987; Matsuoka and Okada, 1989, 1990; Sato and Takayama, 1992; Raffi et al., 1993; Wei, 1993). The Pleistocene nannofossil events used in this paper (Table 1) combine the opinions of these authors.
The delineation of the Pliocene/Pleistocene boundary has long been controversial. Most nannofossil workers rely on the Martini (1971), Okada and Bukry (1980), or Gartner (1977) zonal schemes and therefore use the last occurrence (LO) of Discoaster brouweri to mark the Pliocene/Pleistocene boundary. However, with the acceptance of the Vrica section (Calabria, southern Italy) as the stratotype for the boundary, the Pliocene/Pleistocene boundary issue appears to have been resolved with the use of the first occurrence (FO) of Gephyrocapsa oceanica s.l. (approximately 1.6 Ma, above the Olduvai subchron) as the marker (Rio et al., 1990b). In this study, we used the FO of Gephyrocapsa oceanica s.l. to recognize the Pliocene/Pleistocene boundary.
The Pliocene biostratigraphic zonation of Martini (1971) was followed in this study. Some modifications were necessary for the lower Pliocene: Zones NN14 and NN15 were always combined as previously suggested by Raffi and Rio (1979) and Rio et al. (1990b). The base of this interval was defined in this study by the FO of Pseudoemiliania lacunosa, instead of the FO of Discoaster asymmetricus as would be better estimated by quantitative analyses because D. asymmetricus appears to be absent from the lower part of its expected range. According to Lohman (1986) and Wei et al. (1988), the FO of Pseudoemiliania lacunosa approximates the Discoaster asymmetricus acme. For the uppermost Pliocene Zone NN19a, the definition of Zone MNN19a by Rio et al. (1990b) was followed.
The LO of Triquetrorhabdulus rugosus and the FO of Ceratolithus acutus are the best index events for the Miocene/Pliocene boundary (Cita and Gartner, 1973; Rio et al., 1990a). These marker species are not present in our samples, so the Miocene/Pliocene boundary is not recognized in the studied sites. The LO of Discoaster quinqueramus was used in this study to separate Miocene from Pliocene nannofossil assemblages. Above the LO of D. quinqueramus, the lowermost Pliocene biostratigraphic event in our material is represented by the FO of Pseudoemiliania lacunosa. A few samples between the FO of P. lacunosa and the LO of D. quinqueramus were placed with uncertainty in the lower Pliocene, but no biozone could be identified. The lowermost Pliocene is missing in the four holes where Pliocene sediments were sampled.