Calcareous nannofossils are generally well represented in the studied material. Abundance, preservation, and assemblage composition vary with water depth, latitude, and stratigraphic interval. Miocene and Pleistocene assemblages are better preserved and more diversified than the Pliocene ones.
The distribution patterns of 30 middle Miocene to Pleistocene calcareous nannofossil index species, established by different counting methods, are shown in Fig. 4, Fig. 5, Fig. 6, Fig. 7, Fig. 8, Fig. 9, Fig. 10, Fig. 11, Fig. 12, Fig. 13, Fig. 14, Fig. 15, Fig. 16, Fig. 17, Fig. 18, Fig. 19, Fig. 20, Fig. 21. Together these species define 50 biohorizons, which are listed and discussed in Table 2, Table 3, Table 4, Table 5. Their stratigraphic position at each site is reported in Appendix B. Following the strategy outlined above, on the basis of the established distribution patterns and available magnetostratigraphy (Fig. 22, Fig. 23, Fig. 24, Fig. 25), the reliability of the 50 biohorizons is evaluated below and the results are summarized in Table 2, Table 3, Table 4, Table 5. Then, on the basis of the regionally reliable biohorizons, a biostratigraphic scheme integrated with the GPTS and GCS is proposed (Fig. 2), upon which the age assignments at single sites are based. In the following section the data obtained for the Pleistocene, Pliocene, and Miocene time intervals are discussed.
The Pleistocene Series was formally defined in 1983 by the Global Stratotype Section and Point (GSSP) of the Vrica section in southern Italy (Aguirre and Pasini, 1985), just below the top of the Olduvai Subchron (C2n Subchron in the terminology of Cande and Kent, 1995), with an age of 1.806 Ma (Lourens et al., 1996). In this work, the base of the Pleistocene is approximated with magnetostratigraphy when available, or by means of the first occurrence (FO) of Gephyrocapsa oceanica s.l., which occurs some 80 k.y. above the Vrica GSSP (Raffi et al., 1993; Rio, Raffi, et al., 1997). A subdivision of the Pleistocene Series has not been attempted because it is controversial (Van Eysinga, 1975; Ruggeri and Sprovieri, 1977, 1979; Rio, 1982; Berggren and Van Couvering, 1974; Haq et al., 1987; Ruggeri et al., 1984; Haq and Van Eysinga, 1986; Rio et al., 1991; Berggren et al., 1985; Berggren, Hilgen, et al., 1995; and Van Couvering, 1997), and the subdivisions proposed up to now cannot be recognized by means of the calcareous nannofossils.
Sediments belonging to the Pleistocene Series are well represented at all sites of Leg 167 (Fig. 3), and in some cases (i.e., Sites 1017, 1018, 1019, and 1020) the Pleistocene sequence is expanded.
The classification of Pleistocene sediments by means of calcareous nannofossils in the light microscope is based on the eight biohorizons reported in Table 2. Two additional biohorizons in the late Pleistocene, the FO and the acme beginning (AB) of Emiliania huxleyi, are not discussed in this paper because they cannot be confidently recognized with the light microscope technique used in this study.
The gephyrocapsids are a major component of the Pleistocene calcareous nannofossil assemblage, both in the California margin area and worldwide. They provide at least five globally useful biohorizons (P2, P3, P5, P6, and P8 in Table 2). Taxonomy and nomenclature of gephyrocapsids are complex and contradictory. For the present study, the biometric subdivision proposed by Rio (1982) and Raffi et al. (1993) has been followed.
The gephyrocapsids appeared in the late Miocene (Bonci and Pirini Radrizzani, 1992; Pujos, 1987) with small forms (<4 µm; "small Gephyrocapsa spp."). They remained a secondary component of the calcareous nannofossil assemblages into the late Pliocene, when their abundance increased and they underwent a gradual size increase from ~4 to ~7 µm (Rio, 1982). Based on this progressive size increase, it is possible to define at least two different biohorizons: the FO of Gephyrocapsa oceanica s.l. (defined as gephyrocapsids >4 µm, with an open central area) and the FO of large Gephyrocapsa spp. (gephyrocapsids >5.5 µm; Rio, 1982; Raffi et al., 1993). These two bioevents have been referred to differently in the literature. The FO of G. oceanica s.l. corresponds to the FOs of medium-sized Gephyrocapsa spp. of Raffi et al. (1993) and Gephyrocapsa spp. A-B of Wei (1993), whereas the FO of large Gephyrocapsa of Rio (1982) and Raffi et al. (1993) corresponds to the FO of Gephyrocapsa spp. A-B >5.5 µm of Wei (1993).
Following the early Pleistocene rapid size increase, gephyrocapsids underwent a major evolutionary change with the virtual disappearance of medium- and large-sized forms and the onset of an interval characterized by the dominance of small Gephyrocapsa spp. (Gartner, 1977; Rio, 1982; Raffi et al., 1993). This major paleontologic change is referred to in the literature as the LO of large Gephyrocapsa spp. by Rio, Raffi, et al. (1990) and Raffi et al. (1993) and as the LO of Gephyrocapsa spp. A-B of Wei (1993)
The end of the interval of dominance of the small Gephyrocapsa spp. was referred to as the acme end (AE) of small Gephyrocapsa spp. by Gartner (1977) and thereafter used as a zonal boundary definition in the early Pleistocene. This bioevent is marked by the reentrance (RE) of normal-sized gephyrocapsids (size between 4.0 and 5.5 µm), among which there is a form characterized by having the bridge in the central area parallel to the short axis of the placolith. This form is missing in the underlying interval. Following Rio (1982), such specimens have been labelled Gephyrocapsa sp. 3. Gartner's AE of small Gephyrocapsa spp. has been referred to as Gephyrocapsa spp. C-D by Wei (1993) and as the RE of Gephyrocapsa spp. by Raffi et al. (1993).
In the California margin, all these biohorizons based on gephyrocapsids can be recognized across the entire Leg 167 latitudinal transect, as discussed below.
The LO of Reticulofenestra asanoi (a >6.5-µm circular reticulofenestrid) has been evidenced as a useful event by Takayama and Sato (1987). Wei (1993) proved this event globally synchronous at an age of ~0.88 Ma. In the California margin, R. asanoi has been analyzed quantitatively only at Site 1014 (Fig. 10, Fig. 11), where the species is well represented and its extinction is a clear event occurring in the middle part of Chron 1r (Fig. 11). The age-depth plot suggests an age of 0.88 Ma (Fig. 23; Table 2), in agreement with previous evaluations (Takayama and Sato [1987], 0.83 Ma; Matsuoka and Okada [1989], 0.81 Ma; Wei [1993], 0.88 Ma; Bassinot et al. [1994], 0.88 Ma). The LO of R. asanoi is a useful and reliable biohorizon within the mid-Pleistocene in this area.
Another reliable biohorizon observed in the California margin is the LO of P. lacunosa. The species is well represented, and its extinction is easily determined at all sites studied with detailed quantitative counting (Fig. 5, Fig. 11, Fig. 17, Fig. 19). However, it must be noted that at all sites, except Site 1014, the event appears to be abrupt, most probably because of the low-resolution sampling. In fact, at Site 1014, where the time spacing between samples is ~6 k.y., P. lacunosa shows strong abundance fluctuations before its extinction (Fig. 10, Fig. 11), in agreement with the findings by Thierstein et al. (1977).
The LO of H. sellii has been proposed by Gartner (1977) as a useful early Pleistocene event. However, it has been proved to be time transgressive over various water masses by Backman and Shackleton (1983), Wei (1993), and Raffi et al. (1993). In the studied area helicoliths occur discontinuously (e.g., Site 1010, Fig. 5), making it difficult to pinpoint the LO of H. sellii. However, at Sites 1011, 1012, 1016, 1017, and 1018 the event is associated with or just below the LO of large Gephyrocapsa spp. (Fig. 7, Fig. 8, Fig. 12, Fig. 13, Fig. 14). Therefore, it is found in a stratigraphic position comparable to that observed in the Mediterranean (Rio, Raffi, et al., 1990) and middle North Atlantic Ocean successions (Raffi et al., 1993). The LO of H. sellii in the California margin has been considered as an unreliable event because the species is too rare in the early Pleistocene.
The LO of C. macintyrei has been proposed as an early Pleistocene biohorizon by Gartner (1977), but it has been proved to be time transgressive (Raffi et al., 1993; Wei, 1993; Table 2). Specifically, the event normally occurs slightly above the FO of G. oceanica s.l. (Rio, Raffi, et al., 1990; Raffi et al., 1993; Wei, 1993; Gartner, 1977), but it has been associated with the MIS 57/58 transition (1.67 Ma) in some areas, and with the MIS 55 (1.6 Ma) in others (Raffi et al., 1993; Wei, 1993; Table 2). Calcidiscus macintyrei is generally well represented in the California margin. However, it occurs discontinuously and is rare in its final range (see Site 1014 in Fig. 11, Site 1020 in Fig. 17, and Site 1021 in Fig. 18, Fig. 19). With reference to those sites where the sampling resolution was sufficiently high, the LO of C. macintyrei has been observed in its "right" position only at low-latitude Site 1014 (Fig. 10, Fig. 11,). At mid-latitude Site 1020, the event is definitively below the FO of G. oceanica s.l., and at Site 1021, studied in low resolution, the event seems to be associated with the Olduvai (C2n) Subchron, with an estimated age of 1.915 ± 0.015 Ma (Fig. 25; Table 2). These data seem to indicate that the LO of C. macintyrei in the California margin is more time transgressive than previously thought, and hence a poorly reliable event.
The biostratigraphic resolution provided by the standard zonations of Okada and Bukry (1980) for the Pleistocene is limited (Fig. 2), and normally workers refer to additional biohorizons for a finer subdivision. The set of biohorizons discussed above makes it possible to propose a better resolved and more reliable zonation for the Pleistocene that can be applied to the California margin area and in most of the low- to high-latitude areas. Therefore, the following emendation of the early to mid-Pleistocene zonation of Bukry (1973) as coded by Okada and Bukry (1980) is proposed:
Bukry (1973) and Okada and Bukry (1980) considered these biostratigraphic intervals as subzones. Because they are widely recognizable over various water masses as demonstrated by the extensive literature available, it has been proposed to rank them at zone level. The definitions and occurrences of these four new biostratigraphic units are given in Appendix C in addition to any related remarks.
The Pliocene/Miocene boundary has not been formally defined yet, but following a common practice, in this paper it is equated with the reestablishment of the open marine conditions in the Mediterranean, namely at the base of the Mediterranean Zanclean Stage (Rio et al., 1991; Hilgen and Langereis, 1988). The recognition of the Miocene/Pliocene boundary is generally considered difficult at a global scale (e.g., Benson et al., 1991; Benson and Hoddell, 1994; Benson and Rakic-El Bied, 1996; and Suc et al., 1997). Recently, however, it has been shown that the base of the Zanclean is associated with the upper part of Chron C3r, some 80 k.y. below the Thevra (C3n.4n) Subchron (Zijderverld et al., 1986; Channell et al., 1988; Hilgen and Langereis, 1988), at an age of 5.33 Ma (Lourens et al., 1996). This chronological information allows us to approximate the Miocene/Pliocene boundary by the LO of Discoaster quinqueramus (5.54 Ma according to Backman and Raffi, 1997) and the FO of Ceratolithus acutus (5.37 Ma according to Backman and Raffi, 1997). It must be noted that the practice of recognizing the Miocene/Pliocene boundary by means of these biohorizons is well established among nannofossil paleontologists (Bukry, 1973; Rio, Fornaciari, et al., 1990; Raffi and Flores, 1995). Because of the unreliability of the FO of C. acutus in the California margin area (see below), magnetostratigraphy and/or the LO of D. quinqueramus are used for approximating the Miocene/Pliocene boundary.
The Pliocene Series has been subdivided into three intervals (early, mid-, and late Pliocene) following a recent formal decision of the International Commission on Stratigraphy (Rio et al., 1998; Castradori et al., 1998). The early Pliocene corresponds to the Zanclean Stage, the mid-Pliocene to the Piacenzian Stage, and the late Pliocene to the Gelasian Stage (Rio et al., 1998; Castradori et al., 1998). This three-fold subdivision of the Pliocene Series is easily recognizable by means of calcareous nannofossil biostratigraphy. The early Pliocene (Zanclean)/mid-Pliocene (Piacenzian) boundary, defined in Sicily by a lithologic level corresponding to the Gilbert/Gauss boundary, at an age of 3.66 Ma, is approximated by the LO of Sphenolithus spp. (3.7 Ma according to Lourens et al., 1996) and the LO of R. pseudoumbilicus (3.82 Ma according to Shackleton et al., 1995). In the California margin area, the LO of Sphenolithus spp. is not reliable. The early/mid-Pliocene boundary has been placed by using magnetostratigraphy, when available, or the paracme end (PE) of Discoaster pentaradiatus, which is closely associated with the Gilbert/Gauss boundary in this area (see discussion below). The mid-Pliocene (Piacenzian)/late Pliocene (Gelasian) boundary is defined in Sicily by a lithic level close to the Gauss/Matuyama boundary, at an age of 2.55 Ma (Rio et al., 1998). In the California margin area it is well approximated by the closely spaced successive extinctions of Discoaster surculus (~2.53 Ma) and D. pentaradiatus (~2.52 Ma).
The Pliocene Series is well represented at all Leg 167 sites (Fig. 3) with a thickness of as much as 270 m at Site 1014 (Fig. 10).
The Pliocene calcareous nannofossil assemblage is generally poorly diversified and badly preserved, especially in the lower Pliocene sections where barren intervals are present (Fig. 3). The 16 biohorizons that have been proposed as useful for the correlation and classification of the ~3.52-m.y.-long Pliocene Series have been reported in Table 3. In the following section the reliability of all these biohorizons in the California margin is discussed.
Discoasterids provide at least eight biohorizons for subdividing the Pliocene Series (Pl5, Pl7, Pl10, and Pl12-Pl16 in Table 3). They are known to be missing in sediments underlying high latitudes and high-nutrient water masses (Chepstow-Lusty et al., 1989, 1991, 1992; Chepstow-Lusty and Chapman, 1995). The vastly different latitudinal and ecological conditions represented in Leg 167 sites make it difficult to evaluate the reliability of the discoasterid biohorizons in the entire area.
The simultaneous LOs of Discoaster brouweri and Discoaster triradiatus (the last representative of the discoasterids; Ericson et al., 1963), generally used in all low- and middle-latitude zonations (Table 3; Fig. 2), have been detected at Sites 1010 (Fig. 4, Fig. 5), 1011 (Fig. 7), 1012 (Fig. 8), 1013 (Fig. 9), 1014 (Fig. 10, Fig. 11), 1016 (Fig. 12), 1018 (Fig. 14), 1020 (Fig. 16, Fig. 17), 1021 (Fig. 18, Fig. 19), and 1022 (Fig. 21). At the low-latitude Sites 1010 and 1012 and at mid-latitude Site 1021 the D. brouweri and D. triradiatus LOs occur slightly above the base of the Olduvai Subchron (Fig. 4, Fig. 8, and Fig. 18), in agreement with the previous evaluations of Raffi et al (1993), Wei (1993), and Channell et al. (1990). However, at mid-latitude Sites 1020 and 1022 (Fig. 16, Fig. 17, and Fig. 21), located near the core of the Northern California Current (Lyle, Koizumi, Richter, et al., 1997), the LOs of D. brouweri and D. triradiatus are associated with the LOs of Discoaster pentaradiatus and Discoaster surculus, at a lower stratigraphic level with respect to previous sites. Wise (1973), who studied Site 173 located 16 km from Site 1022, observed as well an early LO of D. brouweri in this area of cold and probably high-productivity water masses. The LOs of D. brouweri and D. triradiatus are diachronous events in the California margin, useless for long-distance correlation in the area. Bukry (1973) used the LO of D. brouweri as a zonal boundary. Because the LO of D. brouweri is of limited geographic applicability, it would be better utilized as a subzonal boundary definition.
Due to the high abundance of D. triradiatus in the upper part of the range of D. brouweri (Takayama, 1970), Backman and Shackleton (1983) defined the AB of D. triradiatus as a late Pliocene biohorizon. In the investigated area, D. triradiatus is well represented only at the low-latitudes Sites 1010 (Fig. 4, Fig. 5), 1012 (Fig. 8), and 1014 (Fig. 10, Fig. 11), whereas it is rare or absent at other mid-latitude sites. At Site 1010, the AB of D. triradiatus occurs in the basal part of the Matuyama (C2r) Chron (Fig. 4) at an interpolated age of 2.29 ± 0.1 Ma (Fig. 22; Table 3). This age estimate is within the range of previous evaluations (from 2.15 to 2.25 Ma according to Berggren, Hilgen, et al., 1995). However, the AB of D. triradiatus seems to be of limited utility in the California margin, because the species is too rare.
Close to the Gauss/Matuyama boundary, concomitantly with the enhancement of Northern Hemisphere glaciation, the productivity of discoasterids decreased drastically, and all species except D. brouweri and D. triradiatus became extinct (Backman and Pestiaux, 1987). This major turnover in the calcareous nannofossil assemblage occurred in ~300-400 k.y. and the successive shortly spaced LOs of D. tamalis, D. surculus, and D. pentaradiatus are generally used for biostratigraphic purposes (Fig. 2; Table 3). These three biohorizons have been recognized at all sites (Site 1010, Fig. 4, Fig. 5; Site 1011, Fig. 7; Site 1012, Fig. 8; Site 1013, Fig. 9; Site 1014, Fig. 10, Fig. 11; Site 1016, Fig. 12; Site 1018, Fig. 14; Site 1020, Fig. 16, Fig. 17; Site 1021, Fig. 18, Fig. 19; and Site 1022, Fig. 21). Quantitative and semiquantitative distribution patterns around extinctions are provided for Sites 1010 (Fig. 5), 1014 (Fig. 11), 1020 (Fig. 17), and 1021 (Fig. 19).
D. pentaradiatus shows variable abundance close to its extinction, but its LO datum has been detected at all sites. At both Sites 1010 (low-latitude) and 1021 (mid-latitude; Fig. 4 and Fig. 18) the LO of D. pentaradiatus is associated with the lowermost part of the Matuyama Chron (C2r.2r). The age-depth plots of Sites 1010 (Fig. 22) and 1021 (Fig. 25) suggest an age of 2.44 ± 0.05 Ma and an age of up to 2.51 Ma (Table 3), respectively. There are contradictions in the literature concerning the synchroneity of the LO of D. pentaradiatus (compare Wei, 1993 and Berggren, Hilgen, et al., 1995). Previous age evaluations range from 2.52 Ma (Tiedeman et al., 1994) and 2.51 Ma (Lourens et al., 1996), to 2.36/2.51 Ma (Wei, 1993). These ages compare well with the evaluation at Sites 1010 and 1021, and suggest that the LO of D. pentaradiatus can be considered as a fairly good event even if the species is sometimes rare and scattered.
The LO of Discoaster surculus has been recognized at all sites and occurs either together or slightly below the LO of D. pentaradiatus as observed by all previous authors (Müller, 1978; Ellis, 1979; Raffi and Rio, 1979; Rio, Raffi, et al., 1990). Quantitative or semiquantitative distribution patterns established at Sites 1010 (Fig. 5), 1014 (Fig. 11), 1020 (Fig. 17), and 1021 (Fig. 19) indicate that the species is common to abundant around its extinction, which appears to be abrupt. At Sites 1010 (low latitude) and 1021 (mid-latitude) the LO of D. surculus is associated with the lowermost part of the Matuyama Chron (C2r.2r), with an interpolated age of up to 2.55 Ma (Fig. 22, Fig. 25; Table 3) in fairly good agreement with previously reported evaluations (Lourens et al., 1996; Wei., 1993; Berggren, Hilgen, et al., 1995; Table 3). The LO of D. surculus appears to be a reliable event in the California margin, and, probably, the easiest detected Pliocene discoasterid event in the area.
The final range of Discoaster tamalis has been established in detail at Sites 1014 (Fig. 11) and 1020 (Fig. 17). The species shows an abrupt decline followed by a few spikes of occurrence that are probably related to a tail of low productivity of the species also known from other areas (Rio, Raffi, et al., 1990). The LO of D. tamalis is defined as the drop of abundance that is easily detected at all sites. At Sites 1010 (low latitude) and 1021 (mid-latitude), the LO of D. tamalis is associated with the late Gauss Chron (lower part of Chron C2An.1n; Fig. 4, Fig. 18), with an interpolated age of ~2.97 Ma (Fig. 22, Fig. 25; Table 3). This age evaluation is not in strong contrast with previous estimates (2.83 Ma; Table 3), considering that it is based on an interpolation from sediment accumulation rates. Therefore the LO of D. tamalis is a fairly useful event in the California margin area.
In the early Pliocene the only biohorizon used from the zonal scheme of Okada and Bukry (1980) is the first common and continuous occurrence (FCO) of Discoaster asymmetricus (Fig. 2; Table 3). This event has been detected at Sites 1011 (Fig. 7), 1014 (Fig. 10, Fig. 11), 1016 (Fig. 12), and 1021 (Fig. 18, Fig. 19) and it can easily be recognized in the California margin, where D. asymmetricus is fairly well represented (Fig. 11, Fig. 19). At Site 1021 (sampled in low resolution; Table 1), the FCO of D. asymmetricus is associated with Chron 2Ar at an age of 3.925 ± 0.065 Ma (Fig. 25). Previously, this event has been associated with the top of the Cochiti (C3n.1n) Subchron with an estimated age of 4.2 Ma (Berggren, Hilgen, et al., 1995; Table 3) and with Subchron C2Ar with an estimated age of 4.12 Ma (Lourens et al., 1996). Despite the possible diachroneity with respect to other areas, the FCO of D. asymmetricus has been considered a regionally useful biohorizon.
Within the early Pliocene discoasterid assemblages of the California margin, a peculiar distribution pattern of Discoaster pentaradiatus has been detected: this species is virtually missing in a short interval close to the Gilbert/Gauss boundary at Site 1021 (Fig. 18, Fig. 19) and in a similar stratigraphic position at Site 1014 (Fig. 10, Fig. 11). An analogous absence interval (paracme) of D. pentaradiatus had been observed by Driever (1981, 1988) and Rio, Raffi, et al. (1990) in the Mediterranean area. These authors proposed two biohorizons, the paracme beginning (PB) and PE of D. pentaradiatus, that are very useful in the Mediterranean area, where they are associated with the Gilbert/Gauss transition as well. Specifically, Lourens et al. (1996) estimated ages of 3.93 and 3.61 Ma for the PB and PE of D. pentaradiatus respectively. Note that the PB of D. pentaradiatus has been recognized at Sites 1011 (Fig. 7) and 1016 (Fig. 12), and the PE of D. pentaradiatus has been observed at Sites 1010 (Fig. 4, Fig. 5), 1011 (Fig. 7), 1012 (Fig. 8), 1016 (Fig. 12), and 1020 (Fig. 16, Fig. 17). The PB and PE of D. pentaradiatus are considered reliable in the California margin area and basically correlatable with the events observed in the Mediterranean.
The early Pliocene zonal scheme of Okada and Bukry (1980) is largely based on biohorizons defined by ceratoliths (FO of Ceratolithus acutus, FO of Ceratolithus rugosus, and LOs of Amaurolithus primus and Amaurolithus tricorniculatus; Fig. 2). In addition, Bukry (1973) suggested as a secondary criterion for defining the bottom of Zone CN10b the LO of Triquetrorhabdulus rugosus, a species evolutionarily related to the ceratoliths (Raffi et al., 1998). T. rugosus, A. tricorniculatus, and members of the genus Ceratolithus are virtually missing in the early Pliocene sediments of the California margin and, hence, the biostratigraphic intervals based on them are not recognizable in the area. A. primus is rare, whereas Amaurolithus delicatus is common and continuously present. The distribution pattern of A. primus has been established at low-latitude Site 1014 (Fig. 11) and mid-latitude Site 1021 (Fig. 19). The species occurs discontinuously, particularly at Site 1021. At this latter site A. primus apparently becomes extinct within the Cochiti (C3n.1n) Subchron (Fig. 19), whereas previously it had been associated with the top of the Sidufjal (C3n.3n) Subchron in oceanic areas and with the Nunivak (C3n.2n) Subchron in the Mediterranean (Berggren, Hilgen, et al., 1995; Table 3). Because of these results, the LO of A. primus is considered a poorly reliable event in the California margin. The distribution pattern of A. delicatus has been established at Sites 1014 and 1021 (Fig. 11, Fig. 19), but the species is well represented also at the other sites. Its extinction is easily detected and occurs above the FCO of D. asymmetricus and close to the LO of R. pseudoumbilicus, in a stratigraphic position that is comparable to that observed in the Mediterranean area by Rio, Raffi, et al. (1990). At Site 1021, the LO of A. delicatus is associated with Chron C2Ar (upper Gilbert) with an interpolated age of 3.962 ± 0.065 Ma (Fig. 25; Table 3). The inferred age is only indicative because of the low-resolution sampling available at this site (Table 1). The LO of A. delicatus seems to be consistently correlatable, at least in the California margin area, and is retained as a regionally reliable biohorizon.
The early/mid-Pliocene boundary (corresponding to the Gilbert/Gauss transition) is characterized by a major turnover in the calcareous nannofossil assemblage. In particular, the final extinction of two major elements of the Neogene calcareous nannofossil assemblage occurred very close to the boundary. In fact, the last representatives of the genus Sphenolithus and of large reticulofenestrids (Reticulofenestra pseudoumbilicus) became extinct at 3.66 and at 3.82 Ma respectively (Shackleton et al., 1995). As already noted above, approximately in this time interval, the last member of the genus Amaurolithus (A. delicatus) became extinct, and the distribution pattern of D. pentaradiatus was affected by an absence interval detected both in the Mediterranean and in the California margin. In addition, in the latest part of the early Pliocene, the placolith Pseudoemiliania lacunosa made its first occurrence.
However, as already noted, in the California margin the genus Sphenolithus is often missing in the early Pliocene successions, except at low-latitude Site 1010 (Fig. 5) and mid-latitude Site 1021 (Fig. 19). At Site 1021 the LO of Sphenolithus spp. occurs much earlier than indicated in the literature, below the FCO of D. asymmetricus at the base of Chron C3n.1n (Fig. 18). At Site 1010 (Fig. 4, Fig. 5) a barren interval does not allow the observation of this extinction. In the California margin, the LO of Sphenolithus spp. is clearly a useless biohorizon.
Reticulofenestra pseudoumbilicus is common to abundant at all sites, but close to its final exit it may be missing for a short interval (Site 1021, Fig. 18, Fig. 19). The LO of R. pseudoumbilicus has been detected at Sites 1011 (Fig. 7), 1012 (Fig. 8), 1014 (Fig. 10, Fig. 11,), 1016 (Fig. 12), 1020 (Fig. 16, Fig. 17), and 1021 (Fig. 18, Fig. 19). At all of these sites it occurs together or just above the LO of A. delicatus. The age-depth plot at Site 1021 suggests an age of 3.83 ± 0.03 Ma (Fig. 25; Table 3), in good agreement with previous evaluations (Shackleton et al., 1995; Table 3). The LO of R. pseudoumbilicus seems to be a reliable and useful biohorizon in the California margin.
The FO of Pseudoemiliania lacunosa was proposed as a useful biohorizon in the late early Pliocene of the Mediterranean region (Raffi and Rio, 1979; Rio, Raffi, et al., 1990) and in the northern part of the California margin (Wise, 1973; Fig. 2). In the investigated area, P. lacunosa in its initial range is rare, small, and in poorly preserved material can be confused with small reticulofenestrids. The FCO of P. lacunosa is defined as the point at which its abundance reaches 1% in a counting of 50-100 reticulofenestrids. The distribution pattern of P. lacunosa has been established at Site 1014, where the FCO datum of the species occurs between the FCO of D. asymmetricus and the LO of R. pseudoumbilicus (Fig. 10) at a similar stratigraphic position as in the Mediterranean area (see Rio, Raffi, et al., 1990). However, this event has been considered moderately reliable because P. lacunosa is rare and difficult to recognize in its initial range.
The FO of Helicosphaera sellii was proposed as a useful event in the Mediterranean area by Rio, Raffi, et al. (1990). As already stated, helicoliths are rare in the California margin. The FO of H. sellii has been detected only at Site 1021 (Fig. 18, Fig. 19) where it occurs within Chron C2Ar, at an estimated age of 4.15 ± 0.03 Ma (Fig. 25; Table 3), in a position comparable to that observed in the Mediterranean region (Table 3). However, the species is too rare and it occurs discontinuously for its FO datum to be considered as useful in the California margin.
The previous discussion indicates that the middle and late Pliocene zones and subzones proposed by Okada and Bukry (1980) can all be recognized in the California margin area, except for Subzone CN12d at high-latitude sites (i.e., Sites 1020 and 1022) because of the early extinction of D. brouweri and D. triradiatus. The short Subzones CN12b and CN12c, defined by the extinction of D. surculus and D. pentaradiatus respectively (Fig. 2), have been recognized only at those sites sampled at high resolution (Site 1014; Fig. 10) or characterized by high accumulation rates (Sites 1011, 1012, 1018, and 1021; Fig. 7, Fig. 8, Fig. 14, and Fig. 18, respectively).
On the contrary, the recognition of most of the early Pliocene zones and subzones of Okada and Bukry (1980) has been difficult because of the scarcity of ceratoliths. Hence, for the biostratigraphic classification of the early Pleistocene sediments of the California margin, Zone CN10 and Subzone CN11a have been combined (Fig. 2).
The late Miocene corresponds, by general agreement, to the Tortonian and Messinian Mediterranean stages. However, the position in time of the base of the historical Tortonian stratotype in the Rio Mazzapiedi-Castellania section (Piedmont Tertiary Basin, Northern Italy) is controversial (Rio, Cita, et al., 1997). No formal definition of the Tortonian stage (middle Miocene/late Miocene boundary) has been proposed to date. Therefore, this paper conforms to the most widespread usage by equating the base of the Tortonian to the first appearance of Neogloboquadrina acostaensis, occurring within Zone CN6 in the Mediterranean (Fornaciari et al., 1996) and in many oceanic areas (Berggren et al., 1995b). The bottom of Zone CN6 has been convincingly associated with the lower part of Chron C5n (Raffi et al., 1995; Schneider et al., 1997). Hence, the late Miocene is bounded by approximately the bottom of Chron C5n and the base of the Zanclean Stage at 5.33 Ma (Fig. 2). This usage is basically in agreement with Berggren et al. (1985), Berggren, Hilgen, et al. (1995), and Haq et al. (1987).
The late Miocene is entirely represented only at Site 1010, where it is 70 m thick, and at Site 1021, where it is ~100 m thick (Fig. 3). At Sites 1011 and 1014, only the upper part of the late Miocene is present (Fig. 3).
Late Miocene calcareous nannofossil assemblages are very characteristic with respect to those of the underlying middle Miocene. To remark this Bukry (1975b) introduced the "Sorolian Coccolith Stage," largely based on the dominance of five-rayed discoasterids. Considering both standard zonations and the most recent literature, at least 20 biohorizons have been proposed for the biostratigraphic classification and correlation of the ~5.5-m.y.-long late Miocene time interval (Table 4; Fig. 2). The presence and distribution patterns of all the species defining the biohorizons reported in Table 4 have been checked. Some of the late Miocene biohorizons (the FOs of Discoaster hamatus [base of CN7], Catinaster calyculus, Catinaster coalitus [base of CN6], Discoaster bellus, and the LCO of Coccolithus miopelagicus) fall in barren intervals at both Sites 1010 and 1021, and, hence, their reliability cannot be evaluated. Also, some biohorizons (FOs of Discoaster loeblichii and Discoaster neorectus, defining the base of Subzone CN8b) are based on species that are missing in the area. The remaining biohorizons are discussed below.
Discoasterids are generally rare to scarce in the late Miocene sediments of the California margin, even at low-latitude Site 1010, where, nevertheless, they are better represented (compare Sites 1010 [Fig. 5, Fig. 6] and 1021 [Fig. 19, Fig. 20]). This finding is not unexpected because the area was affected by high-productivity conditions that are unfavorable for discoasterids that thrive better in oligotrophic water masses (Chepstow-Lusty et al., 1989, 1991, 1992; Chepstow-Lusty and Chapman, 1995).
As already mentioned, some of the species used by Okada and Bukry (1980) in their zonation are missing or very rare. In particular, the standard zonation marker species D. hamatus (Fig. 4, Fig. 5), D. berggrenii, and Discoster quinqueramus (Fig. 4, Fig. 5, Fig. 11, Fig. 18, Fig. 20) are too rare for allowing the recognition of the LO of D. hamatus (Zone CN8 base) and the FO of D. berggrenii (Zone CN9 base). However, it is worth mentioning that, when determined, their position is often comparable to that known in other oceanic areas (Fig. 26). In particular, note how the FO of D. berggrenii at Sites 1010 (Fig. 4, Fig. 5) and 1021 (Fig. 20, Fig. 21) is associated with the interval of absence of R. pseudoumbilicus as is seen in the equatorial Pacific Ocean (Raffi et al., 1995). Likewise, the LO of D. hamatus at low-latitude Site 1010 (Fig. 4, Fig. 5) occurs close to the FO of M. convallis as suggested by Rio, Fornaciari, et al. (1990) for the tropical Indian Ocean.
The only discoasterid biohorizon used in the zonation of Okada and Bukry (1980) that could be retained as moderately useful in the studied area is the LO of D. quinqueramus (Zone CN10 base). This event has been detected at Sites 1010 (Fig. 4, Fig. 5), 1011 (Fig. 7), 1014 (Fig. 10, Fig. 11), and 1021 (Fig. 18, Fig. 20). At low-latitude Site 1010, this biohorizon is associated with the middle part of Chron 3r, at an interpolated age of 5.55 ± 0.05 Ma (Fig. 22; Table 4), which compares well with previous evaluations (Gartner et al., 1984; Monechi, 1985; Bleil, 1985; Muza et al., 1987; Berggren, Kent, et al., 1995; Raffi et al., 1995; Backman and Raffi, 1997; Table 4;). However, if the reliability of the LO of D. quinqueramus at low latitudes is confirmed by direct comparison with magnetostratigraphy, not very much can be said about the reliability of this biohorizon at middle latitudes, where the species is very rare. It can be noted, however, that it occurs in the same relative biostratigraphic position (ranking) observed in low-latitude areas (Fig. 26).
The distribution patterns of the following discoasterids, sometimes used for defining supplementary biohorizons or useful for recognizing biostratigraphic intervals, have been monitored: Discoaster surculus (Fig. 6, Fig. 20), Discoaster pentaradiatus (Fig. 6, Fig. 20), Discoaster bollii (Fig. 6), and Discoaster neohamatus (Fig. 6).
The FO of Discoaster surculus was used as a secondary marker for the base of Zone CN9 by Bukry (1975a) and the FCO of the species was considered a useful event within the lower part of Zone CN9 by Raffi et al. (1995). In the California margin area, Discoaster surculus is consistently represented only at low-latitude Site 1010, where its FCO datum occurs in the same stratigraphic position proposed by Raffi et al. (1995; Fig. 26).
The FO of D. pentaradiatus was associated with Zone CN8 by Bukry (1973) and Raffi et al. (1995), although rare occurrences of the species were observed within Zone CN7 in the tropical Indian Ocean (Raffi et al., 1995). In the California margin D. pentaradiatus is rare and scattered in its initial range, and therefore its FO is difficult to pinpoint. At Site 1010 (low latitude; Fig. 6), this event is recorded in the lower part of Zone CN8, which is the same location as in the equatorial Pacific Ocean. However, at Site 1021 (middle latitude; Fig. 20) the first D. pentaradiatus has been detected only within the lower part of Subzone CN9a.
The LO of D. bollii was considered as occurring close to the LO of D. hamatus (CN7/CN8 boundary; Gartner, 1992; Raffi et al., 1995). In the California margin, D. bollii has been observed only at low-latitude Site 1010 (Fig. 6), where it becomes extinct in the same stratigraphic position indicated by previous authors.
The FO of Discoaster neohamatus was recorded within Zone CN7 (Bukry, 1973; Gartner, 1992) and was considered to be diachronous within this zonal interval (Raffi et al., 1995). In the California margin in agreement with Bukry (1981), the presence of D. neohamatus has been detected only at low-latitude Site 1010 (Fig. 6), where its FO datum occurs apparently in the same stratigraphic position (terminal part of Zone CN7) as observed in the equatorial Pacific Ocean by Raffi et al. (1995).
The species Catinaster calyculus and Catinaster coalitus are stratigraphically important in the lower part of the late Miocene (Fig. 2). Specifically, the FO of C. coalitus is the original definition of the bottom of Zone CN6. The FO of C. calyculus was used to define the bottom of Subzone CN7b (Fig. 2). In addition, Bukry (1973) suggested that the LOs of C. calyculus and C. coalitus occur above the LO of D. hamatus (Zone CN8 base). However, there are contrasting opinions among various authors about the distribution of Catinaster spp. probably because the species of the group seem to be affected by ecological factors that are not fully understood yet. In fact, Rio, Fornaciari, et al. (1990) found the FO of C. calyculus within Zone CN6. Raffi et al. (1995) observed that in both the equatorial Indian Ocean and the Pacific Ocean the LOs of C. calyculus and C. coalitus are in the lower and upper part of Zone CN7, respectively. In the California margin, the genus Catinaster is well represented both at low and middle latitudes (Fig. 6, Fig. 18, Fig. 20), but whereas C. calyculus is common, C. coalitus is few to rare.
As already mentioned, the FOs of C. coalitus and C. calyculus have not been detected because they occur in a barren interval at both Sites 1010 and 1021 (Fig. 6, Fig. 20). The LO of C. calyculus occurs close to the FO of M. convallis, whereas the LO of C. coalitus occurs in the middle part of Zone CN7 (Site 1010, Fig. 4, Fig. 6; Site 1021, Fig. 18, Fig. 20). These results are in agreement with the data of Raffi et al. (1995) and suggest that both events may be considered useful and provide good correlations between the low-latitude Site 1010 and mid-latitude Site 1021 (Fig. 26).
The appearance and rapid evolution of horseshoe-shaped nannoliths (ceratoliths) in the upper part of the late Miocene provide biohorizons that allow us to finely subdivide this time interval. The zonation of Okada and Bukry (1980) utilized the FO of Amaurolithus primus as the boundary definition of the base of Subzone CN9b. Recently, Raffi and Flores (1995) subdivided Subzone CN9b of Okada and Bukry (1980) into three intervals on the base of the total range of Nicklithus amplificus. Ceratoliths are well represented in the area (Fig. 5, Fig. 6, Fig. 10, Fig. 11, Fig. 18, Fig. 19, Fig. 20), but only the species A. primus and A. delicatus have been recognized. Specifically, N. amplificus is missing; therefore the subdivision of Subzone CN9b proposed by Raffi and Flores (1995) cannot be applied in the area.
A. delicatus is thought to appear slightly above the FO of A. primus (Gartner and Bukry, 1975; Rio, Fornaciari, et al., 1990; Hodell et al., 1994; Raffi and Flores, 1995; Raffi et al., 1995; Benson and Rakic-El Bied, 1996), and a high-resolution sampling together with a high sedimentation rate is needed to recognize this spacing. The simultaneous FOs of both species have been recognized at Sites 1010 (Fig. 6) and 1021 (Fig. 20). At Site 1010, the FOs of A. primus and A. delicatus are associated with Chron 3Bn/3Br at an interpolated age of 7.23 ± 0.11 Ma (Fig. 22). This result is in fairly good agreement with previous evaluations by Raffi et al. (1995) and Berggren, Kent, et al. (1995) who recorded the A. primus FO in Chron 3Br. Backman and Raffi (1997) estimated an age of 7.39 Ma on the base of astrocyclostratigraphy. The FO of Amaurolithus spp. in the California margin area has been considered a useful event.
In both the equatorial Indian Ocean and the Pacific Ocean many authors recorded an interval of almost total absence of R. pseudoumbilicus (specimens >7 µm) (Rio, Fornaciari, et al., 1990; Gartner, 1992; Takayama, 1993; Raffi and Flores, 1995; Backman and Raffi, 1997). In the California margin a similar interval of absence (paracme) has been detected in the same stratigraphic position. The end of this interval of absence (PE) is associated with the lower part of Chron 3Ar (Raffi et al., 1995) at an age of 7.1 Ma (Backman and Raffi, 1997; Table 4), whereas its beginning (PB) is calibrated within the upper part of Chron 4An (Raffi et al., 1995) at an age of 8.78 Ma (Backman and Raffi, 1997; Table 4). In the California margin these two biohorizons have been detected at Sites 1010 (Fig. 4, Fig. 6) and 1011 (Fig. 7). At Sites 1014 and 1021 only the PE of R. pseudoumbilicus is present (Fig. 10, Fig. 11, Fig. 18, Fig. 20). At low latitudes (Site 1010) the PE of R. pseudoumbilicus occurs within Chron 3Ar at an estimated age of 7.11 ± 0.01 Ma (Fig. 22; Table 4), in agreement with previous evaluations. No magnetostratigraphic data were available to calibrate the PB of R. pseudoumbilicus, which seems to occur close to a suspected hiatus at Site 1010 (Fig. 4, Fig. 6) and close to barren intervals at Site 1021 (Fig. 18, Fig. 20). This event is close to the Minylitha convallis LO and is below or associated with the D. quinqueramus/berggrenii FO (Site 1010, Fig. 4, Fig. 6; Site 1021, Fig. 18, Fig. 20). At Site 1011 (Fig. 7) the PB of R. pseudoumbilicus is just above the D. quinqueramus/berggrenii FO, a discrepancy that can be explained by both the low sampling resolution (Table 1) and the discontinuous occurrence of D. quinqueramus/berggrenii. The PE is apparently reliable in the studied area, whereas it is not possible to make firm inference about the PB of this species. The PE and PB of R. pseudoumbilicus have been used to define the base of Subzone CN9bB and the base of Zone *CN8b (emended) respectively. In fact, despite the problems linked to the environmental characteristics of the study area, this event is more recognizable and correlatable even outside this region than the biohorizons utilized in the original definition of Okada and Bukry (1980; Fig. 2).
Bukry (1973) noted the appearance of M. convallis within Subzone CN8a. In the studied area, M. convallis is common to abundant (Sites 1010, Fig. 6; 1021, Fig. 20) and in a few samples is dominant. The LO of M. convallis is associated with Chron 4n.2n in the equatorial Pacific Ocean (Raffi et al., 1995) at an age of 7.8 Ma (Berggren, Kent, et al., 1995), whereas at mid-latitudes Gartner (1992) detected this event in the upper part of Chron 4n. The FO of M. convallis occurs in Chron 4Ar.2r in the equatorial Pacific Ocean (Raffi et al., 1995) at an age of 9.3/9.5 Ma (Berggren, Kent, et al., 1995) close to the LOs of Discoaster hamatus and Catinaster calyculus. Gartner (1992) observed the FO of M. convallis in Chron C5n at mid-latitude Site 608. During Leg 167, it was not possible to calibrate these biohorizons with the magnetostratigraphy. The LO of M. convallis has been recorded close to the D. quinqueramus/berggrenii FO (Sites 1010, Fig. 4; 1011, Fig. 7; and 1021, Fig. 18) in a stratigraphic position older than previous evaluations. The FO of M. convallis occurs above the C. calyculus LO (Site 1010, Fig. 4; Site 1011, Fig. 7; and Site 1021, Fig. 18), just above the LO of D. hamatus (Site 1010, Fig. 4). Regionally, these two events seem to be useful and fairly reliable.
The previous discussion shows how Subzone CN9b is the only biostratigraphic interval of the scheme of Okada and Bukry (1980) that is recognized with a moderate confidence in the California margin. All other biozones and subzones have been difficult to detect, even at the low-latitude Site 1010. The recognition of the recently proposed three-fold subdivision of Subzone CN9b by Raffi and Flores (1995) was also not possible because of the ecological exclusion of N. amplificus from the area. The correlation and, hence, the biostratigraphic classification of late Miocene sediments in this area are better achieved by utilizing the following biohorizons recently proposed in the literature:
Because of these results, the original definitions of the biostratigraphic intervals of Okada and Bukry (1980) have been "adjusted" (as marked by an asterisk [*]) and are outlined below.
The scheme proposed above is not intended to have any general value and must be considered as very preliminary for the California margin itself, because it is based only on a limited number of successions.
The oldest sediments recovered at Sites 1010 and 1021 during Leg 167 are advanced middle Miocene in age. Specifically, the base of the Site 1010 succession belongs to the middle Miocene Zone CN4 (Langhian; Fig. 2), and the base of the succession recovered at Site 1021 belongs to Zone CN5 (Serravallian; Fig. 2). Both sites have been studied at low resolution and are affected by barren intervals (Fig. 3). In the fossiliferous intervals, the assemblages are in a moderate to good state of preservation. However, no magnetostratigraphy is available, and, hence, the evaluation of the biostratigraphic reliability of the relevant biohorizons is very tentative.
At low-latitude Site 1010 the scheme of Okada and Bukry (1980) has been easily recognized because the marker species (S. heteromorphus and D. kugleri) are well represented (Fig. 4, Fig. 6). It should be noted that D. kugleri is present as common and continuous in a short interval and its FO and LO probably correspond to the FCO and LCO observed by Raffi et al. (1995) in the eastern equatorial Pacific Ocean.
Because both Zones CN4 and CN5 have a fairly long duration (Fig. 2), attempts have been made in the past to define additional biohorizons, most of which are listed in Table 5 (Bukry, 1973; Ellis, 1982; Gartner and Chow, 1985; Theodoridis, 1984, Olafsson, 1989, 1991; Fornaciari et al., 1990, 1993; Gartner, 1992, Fornaciari et al., 1996). These additional events include the LO of Discoaster kugleri, LO of Discoaster exilis, FO of Discoaster bollii, FO of Calcidiscus macintyrei, LCO of Calcidiscus premacintyrei, and LCO of Cyclicargolithus floridanus.
The species defining these biohorizons are all well represented at low-latitude Site 1010 and all the biohorizons they define have been recognized in the same relative position (ranking) observed in previous studies (Table 5) except for the LO of D. exilis. The latter occurs apparently well within Zone CN7, whereas previous authors recorded it either in the upper part of Zone CN6 (Bukry, 1973; Rio, Fornaciari, et al., 1990; Raffi et al., 1995) or at the base of Zone CN7 (Raffi et al., 1995). It is to be noted that at mid-latitude Site 1021, D. exilis is less abundant and discontinuously distributed (Fig. 20). It is, therefore, clear that it is not possible to make firm inferences about the reliability of these biohorizons on the basis of the available single succession.