Datums are constrained, on average, to ±1.6 m (Table 1). The age model employed is based on (1) the "Leg 154 time scale" (Curry, Shackleton, Richter, et al., 1995) and uses a combination of several tuned age models (Shackleton et al., 1990; Hilgen, 1991a, 1991b; Shackleton et al., 1995) and the Cande and Kent (1992) revision of the geopolarity time scale to convert ages from tables in Berggren et al. (1985), and (2) selected published ages for planktonic foraminifer datums in Berggren et al. (1995a, 1995b). Some datum ages, especially in the middle to late Miocene transition, may therefore differ from those in Berggren et al. (1995b) for reasons addressed in Schneider et al. (1997). The Ceara Rise ages cited in the results presented below are based on the astrochronological time scales of Bickert et al. (1997), Tiedemann and Franz (1997), and Shackleton and Crowhurst (1997). A comparison of the published ages of datums and those determined at Ceara Rise and at Site 999 is presented in Table 3. The ranges of all species that occur in more than two samples from Hole 999A are presented in Appendix A. Table 4 presents the records of those species that occur in two or fewer samples in Hole 999A. Table 5 is a list of datums (paleomagnetic and foraminifer) used to construct the age model and assign the ages to samples (Appendix B) that are used in the following section. Ceara Rise ages were used for datums in the Miocene below the FO of G. plesiotumida.
Berggren et al. (1995b) suggest an age of 8.1 Ma for FO of Candeina nitida based on results from Site 806 in the western equatorial Pacific, but at Ceara Rise this event was determined to be at 8.44 Ma. At Site 999 the FO of C. nitida is between Samples 165-999A-29X-5, 41-43 cm, and 29X-2, 41-43 cm (10.28 ± 0.12 Ma).
The age of the FO of Fohsella birnageae is 16.7 Ma according to Berggren et al. (1995b), but in Hole 999A the FO of this species is between Samples 165-999A-49X-2, 46-48 cm, and 48X-CC, 36-38 cm (17.24 ± 0.05 Ma). Chaisson and Leckie (1993) also found F. birnageae at Site 806 (western equatorial Pacific) in sediments dated older than the published age (Berggren et al., 1995b). In the lowest sample in its range at Site 999 it accounts for >3% of the assemblage. At Ceara Rise this species is found down to the base of Zone N5 (21.6 Ma; Pearson and Chaisson, 1997). On the other hand, F. birnageae does not extend up into the lower half of Zone N8 at Site 999, as is indicated by Kennett and Srinivasan (1983). Instead, its occurrence is confined to Zone N7 in Hole 999A.
The age of the LO of Fohsella peripheroronda is 14.6 Ma according to Berggren et al. (1995b). In Hole 999A the LO of this species is between Samples 165-999A-37X-2, 60-62 cm, and 36X-CC, 35-37 cm, for an estimated age of 14.16 ± 0.06 Ma. However, stratigraphic problems related to incomplete recovery in Core 165-999A-36X and the presence of several volcanic ash layers in Core 37X may have distorted the position of this datum.
The series of fohsellid FOs that mark the bases of Zones N10, N11, and N12 are all located in Hole 999A, although the interval may be somewhat condensed. The numbers of Fohsella praefohsi and Fohsella fohsi are low in all samples examined. The FO of Fohsella peripheroacuta (14.7 Ma; Pearson and Chaisson, 1997) is between Samples 165-999A-38X-2, 42-44 cm, and 37X-CC, 25-27 cm, and this species' abundance occasionally exceeds 3% of the assemblage. The FO of Fohsella praefohsi (14.0 Ma; Pearson and Chaisson, 1997) is between Samples 165-999A-36X-CC, 35-37 cm, and 36X-5, 35-37 cm. The LOs of both F. peripheroacuta and F. praefohsi are between Samples 165-999A-33X-6, 42-44 cm, and 33X-3, 42-44 cm (12.51 ± 0.41 Ma). These LOs are only one sample (3.44 m) below the LO of Fohsella fohsi (11.8 Ma) between Samples 165-999A-33X-3, 42-44 cm, and 32X-CC, 37-40 cm (11.79 ± 0.31). This is one indication of the condensation of this part of Hole 999A. Another indication is the generally poor preservation of fohsellid specimens. The absence and rarity, respectively, of F. fohsi lobata and F. fohsi robusta, which globally appear in the upper portion of Zone N12 (Kennett and Srinivasan, 1983), suggests that there is sediment missing from this section. In addition, the bottom of Core 165-999A-32X and the top of Core 33X were highly fractured as a result of the coring process (Sigurdsson, Leckie, Acton, et al., 1997). In addition to its rarity, the keel morphology of F. fohsi is not well developed at this site (see "Taxonomic Notes"), which makes firm placement of the base of Zone N12 difficult.
Globigerina falconensis occurs through three intervals at Site 999; ~11-9, 6-4, and 2-0.5 Ma. It is almost always rare and its FO at Site 999 (at 12.4 Ma in Zone N12) is well above its published first appearance in Zone N7 (Kennett and Srinivasan, 1983). Globigerina bulloides is present more consistently, but its FO at Site 999 coincides with that of G. falconensis and is well above its published first appearance in Zone N9. It is usually rare at this site and exceeds 3% of the assemblage in only four samples (two in the late Miocene and the early Pliocene "pachyderma interval" and two in the late Pliocene).
The interpolated age of the LO of Globigerinoides obliquus above Sample 165-999A-5H-CC, 14-16 cm (1.31 ± 0.05 Ma), corresponds well to its LO at Ceara Rise (1.3 Ma). Gs. obliquus is more abundant than Globigerinoides ruber below Sample 165-999A-11H-5, 42-44 cm (3.11 ± 0.05 Ma), and is much more abundant than Gs. ruber below Sample 165-999A-13H-5, 42-44 cm (3.72 ± 0.08 Ma). It occurs quite consistently to the base of lower Miocene Zone N7, but only one specimen is found in the remaining five samples between Sample 165-999A-49X-2, 46-48 cm, and the bottom of the section examined. Sample 165-999A-31X-2, 42-44 cm (10.78 Ma), represents a transition; above this point Gs. obliquus dominates the assemblage compared to Globigerinoides sacculifer and below it is subordinate to Gs. sacculifer.
The LO of Globigerinoides extremus is between Samples 165-999A-10H-5, 42-44 cm, and 10H-3, 42-44 cm (2.74 ± 0.05 Ma), considerably lower than its published last appearance (1.77 Ma; Berggren et al., 1995a) or its LO at Ceara Rise (1.98 Ma). By contrast, this species seems to persist downsection at Site 999 to levels equivalent to ages much older than its published age (8.3 Ma; Berggren et al., 1995b) or its Ceara Rise age (8.58 Ma; Chaisson and Pearson, 1997). The FO of Gs. extremus is between Samples 165-999A-29X-CC, 37-39 cm, and 29X-5, 41-43 cm, which corresponds to an age of 10.31 ± 0.11 Ma. Candeina nitida, the published age of which is close to that of Gs. extremus, is also found lower than expected at this site, which suggests either a problem with the age model or that both of these species evolved in the western Caribbean Sea.
The LO of Globigerinoides altiapertura at Site 999 is between Samples 165-999A-44X-CC, 38-40 cm, and 43X-CC, 34-37 cm (16.20 ± 0.21 Ma), in the lower part of Zone N8/N9, which corresponds to its published range (Kennett and Srinivasan, 1983). This species is usually more abundant than the similar Gs. obliquus where their ranges overlap in Hole 999A.
The age of the FO of Globigerinoides conglobatus was found to be 6.20 Ma at Ceara Rise. It is consistently present in Hole 999A down to Sample 165-999A-15H-CC, 28-31 cm. Below this point only scattered specimens are encountered, but the lowest specimen found is in Sample 165-999A-19H-CC, 27-30 cm. Gs. conglobatus is therefore present consistently at Site 999 only after 4.80 Ma, but its FO at this site is at 6.27 ± 0.04 Ma, agreeing well with the Ceara Rise age (6.20 Ma).
Globigerinoides ruber is present in Hole 999A from Sample 165-999A-15H-2, 42-44 cm, to the top of the section (the last ~4.5 m.y.). It is often a dominant species in the assemblage above Sample 165-999A-11H-2, 42-44 cm (the last ~3 m.y.). But for the first half of its range (Sample 165-999A-30X-2, 43-45 cm, to 15H-5, 42-44 cm; 10.5-4.7 Ma) this species is rare or absent from the record.
Kennett and Srinivasan (1983) tentatively regarded Globigerinoides subquadratus as the ancestor of Gs. ruber and their ranges are shown as contiguous. But Bolli and Saunders (1985), whose work was based in the Caribbean, did not recognize Gs. subquadratus and showed a discontinuous range for Gs. ruber that corresponds to the combined ranges of Gs. ruber and Gs. subquadratus in Kennett and Srinivasan (1983) but with a break in the upper Miocene part of the range. Indeed, the LO of Gs. subquadratus is between Samples 165-999A-31X-5, 42-44 cm, and 31X-2, 42-44 cm (10.84 ± 0.6 Ma, is just below (and ~0.4 m.y. before) the FO of Gs. ruber at Site 999. Gs. subquadratus is consistently present from its LO to the bottom of the section examined and is frequently the dominant species in the assemblage below Sample 165-999A-40X-CC, 28-30 cm (before 15.37 Ma).
Globigerinoides seigliei occurs in four samples in the upper Miocene section of Hole 999A. It was also reported by Keigwin (1978) at DSDP Site 502. The species does not seem to be cosmopolitan; we have never encountered it except at Site 999.
Globigerinoides sacculifer is present in every sample examined in Hole 999A. Saccate specimens are present in nearly all samples above Sample 165-999A-43X-2, 42-44 cm (the last ~16 m.y.), and the FO of saccate Gs. sacculifer is between Samples 165-999A-49X-2, 46-48 cm, and 48X-CC, 36-38 cm (17.24 ± 0.03 Ma). This species (mostly non-saccate specimens) dominates the assemblage in the lower to middle Miocene between Samples 165-999A-40X-CC, 28-30 cm, and 49X-2, 46-48 cm (~15.5-17.3 Ma). Saccate specimens are more abundant than non-saccate ones in several samples above Sample 165-999A-14H-CC, 30-35 cm (the last 4.5 m.y.), but not further down the section.
The LO of Globigerinoides fistulosus is found between Samples 165-999A-8H-5, 42-44 cm, and 8H-2, 42-44 cm (2.10 ± 0.05 Ma), at Site 999, and this event is marked by a single specimen. In samples examined for this study the species is extremely rare and not well developed morphologically. However, high-resolution work by R.D. Norris (pers. comm., 1998) at Site 999 has revealed a more complete record and also well-developed Gs. fistulosus specimens down to its FO.
The members of this normally temperate-zone genus are surprisingly well represented at western Caribbean Site 999. The FO of Globoconella inflata is between Samples 165-999A-8H-2, 42-44 cm, and 7H-CC, 42-44 cm (2.00 ± 0.06 Ma). This is closer to the 2.09 Ma age given by Berggren et al. (1995a) than to the FO observed at Ceara Rise (2.18 Ma). Gs. inflata at Site 999 are all the "normal" morphotype rather than the Gc. triangula ecophenotype that was observed in the lower part of the Gc. inflata range at Ceara Rise and in the eastern tropical Atlantic on Leg 108 (Weaver and Raymo, 1989) and Leg 159 (Norris, 1998). No Globoconella puncticulata were observed at Site 999, but at Ceara Rise both Gc. puncticulata and transitional specimens between Gc. puncticulata and Gc. triangula were found.
Several middle Miocene globoconellids ordinarily associated with temperate latitudes appear in Hole 999A samples. Globoconella miozea is found most consistently and its LO between Samples 165-999A-31X-2, 42-44 cm, and 30X-CC, 42-44 cm (10.76 ± 0.03 Ma), agrees with the LO shown by Kennett and Srinivasan (1983), but is much later than the 15.7 Ma age cited by Berggren et al. (1995b) based on the record at Site 747 on the Kerguelen Plateau. Berggren et al. (1995b) list the LO of Globoconella praescitula at 11.9 Ma (based on Site 747), but the LO of this species at Site 999 is between Samples 165-999A-31X-CC, 37-40 cm, and 31X-5, 42-44 cm (11.00 ± 0.05 Ma). Similarly, the LO of Globoconella panda is listed by Berggren et al. (1995b) at 11.8 Ma, but this event is between Samples 165-999A-31X-2, 42-44 cm, and 30X-CC, 42-44 cm (10.76 ± 0.03 Ma), the same position as Gc. miozea. However, Kennett and Srinivasan (1983) show the range of Gc. panda extending into the late Miocene Globorotalia mayeri Zone (= part of Zone N17). Gc. miozea, Gc. panda, and Gc. praescitula are most abundant and most consistently present during a brief interval between their LOs and the level between Samples 165-999A-33X-CC, 18-20 cm, and 33X-6, 42-44 cm (~13.0-10.8 Ma). Below this level they are either absent (Gc. panda) or present only sporadically.
Globorotalia tumida is present only sporadically in the Site 999 record. It is found most consistently with Zone N22 (the last 2 m.y.) and is essentially absent for all of the Pliocene, except the uppermost part. Gr. tumida appears in a single sample in the uppermost Miocene in Hole 999A (Sample 165-999A-18H-2, 42-44 cm; 5.63 Ma). The FO of the species marks the base of Zone N18, but this event could not be determined reliably because of the rarity of Gr. tumida in Hole 999A. Berggren et al. (1995a) list the age of this datum as 5.6 Ma. The Leg 154 time scale (Curry, Shackleton, Richter, et al., 1995; Chaisson and Pearson, 1997) includes an incorrect conversion of the Berggren et al. (1985) age (5.2 Ma) for this datum. The correct conversion (using Cande and Kent, 1992) should be 5.7 Ma. However, in Hole 925B at Ceara Rise the age of the FO of Gr. tumida was found to be 5.82 (Chaisson and Pearson, 1997) according to the astrochronological time scale of Shackleton and Crowhurst (1997).
The LO of Globorotalia plesiotumida at Site 999 is between Samples 165-999A-14H-2, 42-44 cm, and 13H-CC, 28-31 cm (3.84 ± 0.06 Ma). Berggren et al. (1995a) do not include an age for this datum, but Berggren et al. (1985) did include this datum age. In Curry, Shackleton, Richter, et al., (1995) and Chaisson and Pearson (1997) this datum age was incorrectly converted to 4.4 Ma. The correctly converted age is 4.3 Ma. At Site 925 on Ceara Rise the age of this datum was 3.77 Ma, fairly close to the age estimated at Site 999.
The FO of Gr. plesiotumida marks the base of Zone N17. Some confusion surrounds the age of this datum. Berggren et al. (1985) did not include an age for this datum. Chaisson and Leckie (1993) misplotted the depth of the datum and miscalculated the age. Berggren et al. (1995b) corrected the error of Chaisson and Leckie (1993) and presented a revised age of 8.3 Ma. At Site 925 on Ceara Rise the astrochronological age of this datum was determined to be 8.58 Ma (Chaisson and Pearson, 1997). At Site 999 the event is between Samples 165-999A-26X-CC, 17-20 cm, and 26X-5, 50-52 cm. Plotting the depth against the Ceara Rise age (Fig. 3) results in an unchanging sediment accumulation rate at Site 999 through the late Miocene, whereas using the Berggren et al. (1995b) age results in a faster rate above the LO of Gr. plesiotumida and a slower rate below this event.
The FO of Gr. lenguaensis is between Samples 165-999A-34X-CC, 33-37 cm, and 34X-5, 39-41 cm (13.57 ± 0.04 Ma). This age is older than the astrochronological age (12.85 Ma) estimated at Site 925 on Ceara Rise (Chaisson and Pearson, 1997). This species is not recorded in Hole 999A above Sample 165-999A-28X-2, 42-44 cm (9.15 ± 0.07 Ma). This range bears no resemblance to the published one (Kennett and Srinivasan, 1983) or to ages for the top of the range suggested by Berggren et al. (1995b) or the Leg 154 results (Chaisson and Pearson, 1997), which suggests that its range truncation is a response by this species to regional paleoceanographic conditions.
The LO of Globoquadrina venezuelana is between Samples 999A-11H-5, 42-44 cm, and 11H-2, 42-44 cm (3.00 ± 0.05 Ma), in the lower part of Zone N21/N20 at Site 999. This is slightly above the top of the range indicated by Kennett and Srinivasan (1983), but agrees well with the age estimated at Ceara Rise (3.08 Ma; Chaisson and Pearson, 1997).
The LO of Dentoglobigerina altispira is listed as identical to that of Menardella multicamerata (3.09 Ma) by Berggren et al. (1995a). At Site 999 both events are between Samples 165-999A-11H-5, 42-44 cm, and 11H-2, 42-44 cm (3.00 ± 0.05 Ma), which is close to the age determined on Ceara Rise (3.11 ± 0.02 Ma).
The LOs of Globoturborotalita species at Site 999 are earlier than have been recorded at other tropical sites. See Table 2 to compare the estimated ages of these events at Site 999 with those noted at other low-latitude sites. The LOs of Globoturborotalita woodi and Gt. apertura at Site 999 are in Sample 165-999A-9H-CC, 30-32 cm (2.54 ± 0.05 Ma). Gt. apertura first occurs in the Site 999 record in Sample 165-999A-29X-2, 41-43 cm (10.05 ± 0.16 Ma), in late Miocene Zone N16 in agreement with its published first appearance (Kennett and Srinivasan, 1983). By contrast, Gt. woodi is not found at Site 999 below Sample 165-999A-33X-CC, 18-20 cm (13.32 ± 0.10 Ma), in middle Miocene Zone N12, well above its published first appearance in the earliest Miocene Globorotalia kugleri Zone (= N4) (Kennett and Srinivasan, 1983). However, Gt. woodi is much more consistently present in the Miocene portion of Site 999 than is Gt. apertura.
The LO of Globoturborotalita nepenthes at Site 999 is in Sample 165-999A-15H-CC, 28-31 cm, corresponding to 4.69 ± 0.07 Ma and ~0.3 Ma earlier than in the western tropical Atlantic at Ceara Rise (see Table 2). This species occurs quite regularly in the Site 999 record and its FO in Sample 165-999A-32X-5, 42-44 cm (11.32 ± 0.08 Ma), is considered a reliable datum to mark the base of Zone N14.
The age of the FO of Hirsutella margaritae is listed as 6.4 Ma in Berggren et al. (1995a), somewhat younger than the age in Berggren et al. (1985) (7.0 Ma; when it is adjusted for changes in the geological time scale by astronomical tuning as per Shackleton et al., 1995). The FO of H. margaritae is between Samples 165-999A-20H-2, 42-44 cm, and 19H-CC, 27-30 cm (6.27 ± 0.04 Ma), agreeing well with the revised Berggren et al. (1985) age, but older than the age of this event at Ceara Rise (6.09 Ma). By contrast, this species' LO at Site 999 is between Samples 165-999A-12H-CC, 36-39 cm, and 12H-5, 42-44 cm (3.37 ± 0.06 Ma). This is younger than the age of this event (3.6 Ma) as listed by Berggren et al. (1985) and as determined at Ceara Rise (3.85 Ma; Chaisson and Pearson, 1997). It is also younger than the age (3.58 Ma) provided by Berggren et al. (1995a). However, Kennett and Srinivasan (1983) show the range of H. margaritae ending just below the base of Zone N21 (3.2 Ma). This event is difficult to constrain at this site because of numerous "transitional" forms that we have called "praehirsuta" (see "Taxonomic Notes").
Hirsutella hirsuta ostensibly descends from H. margaritae, but Kennett and Srinivasan (1983) show a stratigraphic gap between the base and top of their respective ranges. Berggren et al. (1995a) claim an age of 0.45 Ma for the FO of this species. This young date is refuted by evidence from Leg 172 sites (Keigwin, Rio, Acton, et al., 1998) and at Site 999, where the FO of H. hirsuta is between Samples 165-999A-12H-CC, 36-38 cm, and 12H-5, 42-44 cm (3.37 ± 0.06 Ma). This range nearly closes the gap between the FO of H. hirsuta at Site 999 and the LO of H. margaritae at Site 999, giving some credence to the ancestor-descendant relationship. H. hirsuta is a temperate latitude species and occurs only sporadically at Site 999.
The FO of Hirsutella cibaoensis is either between Samples 165-999A-26X-2, 40-42 cm, and 25X-CC, 20-23 cm (8.05 ± 0.03 Ma), or between Samples 165-999A-31X-2, 42-44 cm, and 30X-CC, 38-40 cm (10.76 ± 0.03 Ma). This taxon is not present in any of the intervening samples. However, similarly "deep" specimens were found in several samples at Ceara Rise. The younger Site 999 age is close to the one provided by Berggren et al. (1995b); their age is 7.8 Ma for this datum based on (corrected) data from Site 806 (western equatorial Pacific; Chaisson and Leckie, 1993). H. cibaoensis occurs too sporadically at the top of its range at Site 999 to determine an age for its LO.
The FO of Hirsutella juanai is between Samples 165-999A-28X-5, 42-44 cm, and 28X-2, 42-44 cm (9.37 ± 0.16 Ma). Berggren et al. (1995b) supply an age of 8.1 Ma for this datum based on replotted data from Site 806 (as above), but the age of this datum was determined to be 9.76 Ma at Ceara Rise (Chaisson and Pearson, 1997).
Only sinistrally coiled Menardella menardii are present at Site 999 in upper Pliocene and Pleistocene samples (with the exception of lower Pleistocene Sample 165-999A-6H-5, 42-44 cm; 1.56 Ma). M. menardii is present only sporadically between Samples 165-999A-17H-2, 42-44 cm, and 8H-2, 42-44 cm (~5.3-2.0 Ma), but when it is present the specimens are dextrally coiled. The record is dominated by sinistrally coiled specimens below Sample 165-999A-17H-2, 42-44 cm, to the FO of M. menardii between Samples 165-999A-33X-6, 42-44 cm, and 33X-3, 42-44 cm (~12.5-5.3 Ma).
The FO of Menardella limbata is between Samples 165-999A-30X-5, 42-44 cm, and 30X-2, 42-44 cm (10.58 ± 0.06 Ma), if a single suspect specimen in Sample 165-999A-31X-CC, 37-40 cm, is not considered. The FO at Site 999 is quite close to the FO determined at Ceara Rise (10.57 Ma; Chaisson and Pearson, 1997). The LO of this species at Site 999 is between Samples 165-999A-10H-5, 42-44 cm, and 10H-3, 42-44 cm (2.74 ± 0.05 Ma). This is older than the age that was established for the LO of M. limbata at Ceara Rise (2.38 Ma; Chaisson and Pearson, 1997). The Site 999 record is dominated by dextrally coiled M. limbata from between Samples 165-999A-18H-2, 42-44 cm, and 17H-CC, 21-23 cm (5.60 ± 0.06 Ma) to its LO. Below the level between Samples 165-999A-21H-CC, 41-44 cm, and 21H-5, 42-44 cm (6.85 ± 0.06 Ma), nearly all M. limbata specimens are sinistrally coiled.
The LO of Menardella multicamerata is between Samples 165-999A-11H-5, 42-44 cm, and 11H-2, 42-44 cm (3.00 ± 0.05 Ma), which corresponds well with the age determined at Ceara Rise (3.11 Ma; Chaisson and Pearson, 1997). The FO of M. multicamerata at Site 999 is between Samples 999A-18H-2, 42-44 cm, and 17H-CC, 21-23 cm (5.60 ± 0.06 Ma), slightly younger than at Ceara Rise (~6.2 Ma; Chaisson and Pearson, 1997).
Menardella exilis seems to enter the record slightly earlier at Site 999 than it does at Site 925 in the western tropical Atlantic, while Menardella pertenuis and Menardella miocenica seem to enter later. At Ceara Rise the FO of M. exilis is 4.45 Ma, but at Site 999 its FO is between Samples 165-999A-15H-CC, 28-31 cm, and 15H-5, 42-44 cm (4.69 ± 0.06 Ma). At Ceara Rise the FOs of M. pertenuis and M. miocenica are 3.52 and 3.77 Ma, but the FO of M. pertenuis is between Samples 165-999A-12H-5, 42-44 cm, and 12H-2, 42-44 cm (3.27 ± 0.06 Ma), and the FO of M. miocenica (which marks the base of Zone N20/N21) is between Samples 165-999A-11H-CC, 22-25 cm, and 11H-5, 42-44 cm (3.11 ± 0.06 Ma), at Site 999. The LO of M. exilis is between Samples 165-999A-9H-2, 42-44 cm, and 8H-CC, 24-26 cm (2.31 ± 0.05 Ma), significantly older than the 2.09 Ma age determined at Ceara Rise (Chaisson and Pearson, 1997). The published age for the LO of M. pertenuis is 2.60 Ma (Berggren et al., 1995a), but at Ceara Rise this event was determined to be at 2.33 Ma. At Site 999 it is between Samples 165-999A-9H-CC, 30-32 cm, and 9H-5, 30-32 cm (2.54 ± 0.05 Ma), (i.e. an age closer to the Ceara Rise age). M. pertenuis is always quite rare at Site 999 and is absent in several samples within its range. Menardella miocenica never exceeds 15% of the assemblage at Site 999 (as it did at Ceara Rise), but it is the most abundant Atlantic-endemic menardellid when it is present. In addition to not being as abundant as it was in the western equatorial Atlantic, M. miocenica is not as morphologically well developed at Site 999. Its LO is between Samples 165-999A-9H-5, 42-44 cm, and 9H-2, 42-44 cm (2.42 ± 0.05 Ma), older than published last appearance (2.3 Ma; Berggren et al., 1995a), but quite close to its LO at Ceara Rise (2.38 Ma; Chaisson and Pearson, 1997).
Keigwin (1978, 1982a) recorded the occurrence of sinistrally coiled Neogloboquadrina pachyderma at Site 502, which is ~200 km south-southwest of Site 999. Keigwin (1982a) suggested that this normally polar to subpolar species might have been associated with seasonal upwelling events in the western Caribbean. At Site 999 the sinistrally coiled specimens are common (15%-20% of the assemblage) in Sample 165-999A-23X-CC, 33-35 cm (7.34 Ma), and constitute 3%-14% of the assemblage in several samples between Samples 165-999A-20H-CC, 41-44 cm, and 18H-2, 42-44 cm (~6.6-5.6 Ma). The LO of sinistrally coiled specimens is between Samples 165-999A-15H-5, 42-44 cm, and 15H-2, 42-44 cm (4.51 ± 0.06 Ma). Dextrally coiled specimens also disappear from the Site 999 record at this level, but only temporarily. They are present in most samples examined between Sample 165-999A-12H-CC, 36-38 cm (3.42 Ma), and 2H-2, 42-44 cm (0.27 Ma).
Reconstructions of the paleogeography of the Central American Isthmus (Coates and Obando, 1996) show that during late Miocene (7-6 Ma) an archipelago existed west of Site 999. Subsequent uplift of the Cocos Ridge created a more continuous landmass in what is now Costa Rica and western Panama (Coates and Obando, 1996). The boundary between the North Equatorial Current (NEC) and the North Equatorial Countercurrent (NECC) in the Pacific is a zone of upwelling. In the modern ocean this boundary is at ~10°N, but during the late Miocene it may have been further north (Hovan, 1995). It is therefore possible that although there was enough surface flow through the seaway to allow the passage of the NEC and NECC, this upwelling band may have extended into the western Caribbean in the late Miocene until regional uplift in Central America halted the movement of significant amounts of surface water across the emerging isthmus. The timing of greatest sinistrally coiled N. pachyderma abundance (~6.5-5.6 Ma) corresponds to the late stage in the existence of the "Central American archipelago."
The age of the LO of Neogloboquadrina acostaensis was determined to be 5.1 Ma based on data from Site 806 in the western equatorial Pacific (Chaisson and Leckie, 1993), but at Ceara Rise aberrant neogloboquadrinids were misidentified as N. acostaensis and the LO datum of the species was not accurately placed. Similar small neogloboquadrinids at Site 999 were identified as dextrally coiled N. pachyderma or the "aco-pac" morphotype of Loubere (1988), and the top of the N. acostaenis range in Hole 999A was found between Samples 165-999A-12H-5, 42-44 cm, and 12H-2, 42-44 cm (3.27 ± 0.06 Ma). This age is younger than both the one determined in the western Pacific (Site 806; Chaisson and Leckie, 1993) and another age determined for the LO of N. acostaensis in the eastern Pacific at Site 847 (~3.6 Ma) (Chaisson, 1996). The FO of N. acostaensis is between Samples 165-999A-29X-5, 41-43 cm, and 29X-6, 40-42 cm (10.30 ± 0.01 Ma).
The FO of Praeorbulina sicana marks the base of Zone N8. In Hole 999A this has been drawn between Samples 165-999A-45X-5, 38-40 cm, and 45X-2, 42-44 cm (16.4 Ma), instead of below Sample 165-999A-46X-CC, 27-32 cm, although a single specimen of P. sicana was found there. It was assumed that this specimen was an aberrant Gs. sacculifer as no other P. sicana specimens were found in spite of extensive searching.
The first appearance of Orbulina spp. delimits the base of Zone N9 (15.1 Ma), but at Site 999 this species enters the record between Samples 165-999A-37X-CC, 25-27 cm, and 37X-5, 40-42 cm (14.56 ± 0.09 Ma). This species was extremely rare in the lowest part of its range at Ceara Rise (Pearson and Chaisson, 1997) and poor to moderate preservation in Hole 999A may have dissolved any rare Orbulina specimens just above the base of its range.
The LO of Sphaeroidinellopsis seminulina is between Samples 165-999A-13H-2, 42-44 cm, and 12H-CC, 36-8 cm (3.45 ± 0.06 Ma). This corresponds to a level older than either its published age (Berggren et al., 1995a) of 3.12 Ma or the Ceara Rise age of 3.11 Ma (Chaisson and Pearson, 1997).
At Ceara Rise the age of the LO of Sphaeroidinellopsis kochi was determined to be 4.45 Ma, but at Site 999 the event is between Samples 165-999A-16H-4, 42-44 cm, and 16H-2, 42-44 cm, corresponding to a significantly older age (4.91 ± 0.06 Ma).
The development from Sphaeroidinellopsis paenedehiscens to Sphaeroidinella dehiscens involves the appearance and gradual increase in size of a dorsal secondary aperture. Malmgren et al. (1996) document a 3-8 fold increase in the size of the secondary aperture after 3.6 Ma (their published age adjusted to accord with the Leg 154 time scale). Specimens with dorsal secondary apertures are entirely absent or quite rare at this site below Sample 165-999A-12H-2, 42-44 cm (before 3.77 Ma). At Site 999 rare sphaeroidinellid specimens with minute dorsal apertures were found down to Sample 165-999A-17H-CC, 21-23 cm (5.60 ± 0.04 Ma).
At Site 999 dextrally coiled Truncorotalia truncatulinoides are more consistently present than sinistrally coiled specimens, and the FO of this species between Samples 165-999A-8H-5, 42-44 cm, and 8H-2, 42-44 cm (2.1 ± 0.05 Ma), is marked by dextrally coiled specimens. This is slightly older than the age determined at Ceara Rise (1.92 Ma). Sinistrally coiled specimens are more numerous than dextrally coiled specimens in only two samples (Samples 165-999A-5H-CC, 14-16 cm, and 4H-CC, 16-20 cm. The FO of T. truncatulinoides (s) is between Samples 165-999A-8H-2, 42-44 cm, and 7H-CC, 42-44 cm (2.00 ± 0.04 Ma).
The FO of Truncorotalia crassaformis is between Samples 165-999A-15H-CC, 28-31 cm, and 15H-5, 42-44 cm. This position suggests an age close to 4.7 Ma as per Berggren et al. (1985; adjusted with the Leg 154 time scale; Curry, Shackleton, Richter, et al., 1995), and therefore this datum is used as a tie-point for constructing the biostratigraphic age model. However, Berggren et al. (1995a) suggests an age of 4.5 Ma for this datum and the FO of T. crassaformis was determined to be 4.31 Ma at Ceara Rise (Chaisson and Pearson, 1997).