The discussion of the history of upwelling requires age models of sufficient resolution for comparison between sites and with records from other regions. On the whole, time resolution at the Leg 175 sites is excellent, mainly because of high sedimentation rates and the ubiquitous presence of nannofossils. The available information on biostratigraphic and magnetostratigraphic control is summarized in the Leg 175 Initial Reports volume (Giraudeau et al., 1998; individual site reports in Wefer, Berger, Richter, et al., 1998). Those interested in refining stratigraphic resolution will wish to consult these published data. For the purposes of this synthesis, ages are reassigned using one additional (arbitrary) criterion: avoidance of jumps in sedimentation rates. These assignments are then used to combine sediment properties of all kinds from several sites to detect trends and fluctuations. Events based on diatoms, radiolarians, and foraminifers were of secondary importance in age assignments, although these microfossils deliver important information concerning paleoceanographic history. Supplementary information useful in detailed age assignments within the late Quaternary, in cases, comes from the study of sediment cycles (magnetic susceptibility and color). The latter is only useful if independent age determinations are available, especially for sediments older than 1 Ma.
The assignments given here differ from the straight-line method of the Leg 175 Initial Reports volume (Wefer, Berger, Richter, et al., 1998) in that the sedimentation rate is taken to change gradually rather than abruptly at the point of stratigraphic control. They are similar because no additional data were used beyond those available in the Leg 175 Initial Reports volume (additional work on age models is provided in several papers in this volume, but these data could not be incorporated here). The modified age models are summarized in Table T1 ("Age models for Congo and Angola sites") and Table T2 ("Age models for Namibia and South Africa sites"). Table T1 refers to Sites 1075-1079, Table T2 to Sites 1081-1087. (Site 1080 had stratigraphic problems and is not considered.) The intent is to provide a convenient means of initial age assignment for any set of samples from the Leg 175 sites. Such samples are tagged by depth below seafloor, which is the reason that ages are assigned to a regular succession of depths (in meters below seafloor). Values are given to three digits after the decimal point, not to claim (unrealistically) high precision, but because sedimentation rates are very sensitive to differences in age assignments. Errors are routinely of magnitude 0.01 m.y. and greater.
The age models for the Congo sites show that the oldest sediment reached on the Congo Fan is somewhat greater than 2 m.y. old. Site 1076 does not reach that far back because of a doubled section, owing to slumping. A period of high sediment influx, presumably a debris flow, is apparent at Site 1079 off Angola. But for this disturbance, sedimentation rates would be similar to those of the other sites. An unusually high rate of accumulation is seen at Site 1078 throughout the section. (For reassessment of composite depths at Site 1077, see Jansen and Dupont, Chap. 20, this volume.)
The age models for the sites on Walvis Ridge and off Walvis Bay (Sites 1081-1084) show similar sedimentation rates, reaching an age of 2 Ma at depths between 130 and 200 mbsf, which is normal for continental slope deposition. Rates for Site 1084 are distinctly greater, with 2 Ma occurring at 340 mbsf. This site, off Lüderitz Bay, has exceptionally high contributions of plankton remains (including diatom debris) as a result of being in the sphere of influence of a strong upwelling cell (perhaps the strongest off Namibia). In keeping with this influence, Site 1084 also displayed unusual chemical properties, with exceptionally high levels of phosphate and ammonia in interstitial waters (Murray et al., 1998), as well as strong development of biogenic gases (Meyers et al., 1998).
It cannot be assumed that when using Tables T1 and T2 in assigning ages to a series of samples taken at high density from several adjacent cores in the same hole that a continuous stratigraphy is thereby obtained. Setting aside errors within each model, it is also necessary to make allowance for core expansion and loss of sediment between adjacent cores, the "core gaps." A first estimate of the size of gaps between cores is readily made by using the "growth" percentages when matching adjacent holes to determine "composite depth" (see individual site chapters in Wefer, Berger, Richter, et al., 1998). These percentages reflect core expansion. Such expansion makes it necessary to add depth increments beyond the driller's depths when correlating physical properties markers from one hole to another at the same site. Expansion percentages differ from one hole to another, and they change downhole. The values shown in Table T3 are estimates taken from the graphic summaries in the site chapters (except for Site 1087, which has a drafting mistake) and have a typical error of ~1%.
The size of a core gap may be estimated as follows. Let us say we are interested in the likely gap between two cores taken near 150 mbsf at Site 1075. First, we note that the growth rate is 9% at 100 mbsf and 11% at 150 mbsf. A similar change of 2% is obtained for the next deeper 50 m (Table T3). Thus, the composite depths are 109, 166.5, and 226 meters composite depth (mcd). The difference between the top and bottom of this section is 100 mbsf and 117 mcd, for an expansion of 17%. A core whose in situ length is 9.5 m (as is typically the case) consequently expanded by ~1.6 m, for a total length of 11.1 m. Given a core recovery of 10.2 m (see the "Site Summary," section in the individual site chapters in Wefer, Berger, Richter, et al., 1998), the gap is 0.9 m of expanded sediment, or ~0.75 m (8%) of the original record. For an instantaneous rate of sedimentation near 8 cm/k.y. (see below), the length of the missing time span is just under 10 k.y. When growth percentages do not change, the calculations are much simpler. For Site 1082, the percent growth values stay near 10%. Thus, the expansion of each individual core is near 10%. Recovery of 107.8% of length cored in Hole 1075C (see table 1 in Shipboard Scientific Party, 1998b) implies that gaps are minor (typically a few tenths of meter). For a sedimentation rate of 10 cm/k.y. (see below), the typical gap has but a few thousand years in Hole 1075C (recovery in Hole 1075A was much less satisfactory).
We recommend the following procedures in assigning age to any given sample:
The procedure results in initial age models for sequences of samples, which can then be refined according to additional information, for example, based on oxygen isotope stratigraphy (e.g., see Giraudeau et al., Chap. 7, this volume; Uliana et al., Chap. 11, this volume).
The first three sites of Leg 175 (Sites 1075, 1076, and 1077) were drilled off the Congo and contain sediments typical for a distal fan-type environment (Wefer, Berger, Richter, et al., 1998). Age control was excellent, especially for Site 1075, which has a well-behaved stratigraphy (Fig. F8A). For Site 1075, nannofossil events ("Nanno" in Fig. F8) and magnetic reversals ("Pmag" in Fig. F8) are available (Shipboard Scientific Party, 1998b), as well as a detailed study of cycles of magnetic susceptibility and color ("Cycles" in Fig. F8) (Berger et al., 1998b). In plotting the line representing the age model, here and in the following instances, it was assumed that nannofossil events will give approximate but absolute time control, magnetic reversals will yield refinement, as long as they can be assigned with confidence to a stratigraphic position, and cycles will yield further refinement once the proper time frame is agreed on. Tie points determined for different holes of the same site were treated the same; that is, depths given (in meters below seafloor) were taken as is, without adjustment for any differences in sequences between holes. Mostly, the error is within the bounds set by biostratigraphic resolution in the Leg 175 Initial Reports volume.
Sudden jumps in sedimentation rate were avoided as much as possible within the constraints set by the age tie points. Tie points have errors from several sources: uncertainties arising from the original age assignments to events, proper recognition of events, or unusual position of events within the regional framework of sedimentation (e.g., early regional extinction or dissolution or else unfavorable conditions delaying first occurrences or redeposition delaying last occurrences). Magnetic reversal stratigraphy has to deal with diagenetic interference (see Frost and Yamazaki, Chap. 8, this volume). Cycles may be improperly assigned to known periods of orbital forcing. This problem arises especially when the response is complex, as in color cycles in the case at hand. As it happens, the stratigraphy derived from Milankovitch-based cyclostratigraphy for Site 1075 (Berger et al., 1998b) agrees very well (within error limits) with the biostratigraphy and the magnetic stratigraphy (Fig. F8A). Obviously, if this were not so, the cyclostratigraphy would have to be rejected.
The available data for Site 1075 can be accommodated with a sedimentation rate stratigraphy ranging from just over 5 cm/k.y. near the bottom to 15 cm/k.y. at the top of the record (shaded line in Fig. F8); that is, there is a general increase in sedimentation rate throughout the Quaternary (which is distinctly greater than expected from increasing water content only). Also, there is evidence for a maximum in sedimentation rate centered between 1.2 and 1.1 Ma, a time of distinct change in climate and vegetation in the drainage basin of the Congo (L. Dupont, pers. comm., 2001; Uliana et al., Chap. 11, this volume). A change toward drier conditions would readily explain a pulse of increased erosion and sedimentation. However, without confirmation from other sites, effects from regional shifting of depocenters on the Congo Fan cannot be excluded.
The dating of cycles of magnetic susceptibility in terms of Milankovitch forcing (a method supported by recent results from throughout the South Atlantic) (von Dobeneck and Schmieder, 1999; Schmieder et al., 2000) is readily ported from Site 1075 (Berger et al., 1998b) to the other Congo sites (Fig. F8B, F8C), especially to Site 1077. Site 1076 seems disturbed (Giraudeau et al., 1998); the stratigraphy suggests large-scale coherent sliding, doubling the section in the central part of the site (Fig. F8B). Again, sedimentation rates increase within the Quaternary, from ~7 to ~20 cm/k.y. A distinct maximum is centered near the age of 220 ka. Before 0.5 Ma, rates hover around 6 to 10 cm/k.y., both above and below the slide contact. Site 1077 likewise shows the general increase within the Quaternary (from 5 to >15 cm/k.y.), with a distinct acceleration near 1 Ma. A maximum is centered near 0.75 Ma (following the mid-Pleistocene climate shift). The onset of 100-k.y. climate cycles (after 0.7 Ma) (Berger and Wefer, 1992) is accompanied by a decrease in sedimentation rate, which increases again after 0.3 Ma. (Increased water content contributes to the rate increase in the upper portion of the record.)
The Angola Sites 1078 and 1079 show very different patterns in sedimentation rate history. Site 1078 (Fig. F9A) shows decreasing rates from near 60 cm in the early part of the record to near 20 cm in the late part. Maximum penetration at 165 mbsf is estimated to have been reached at 0.35 Ma. Site 1079 (Fig. F9B) has three distinct sections with respect to sedimentation rate: one below 90 mbsf, with rates near 5 cm/k.y., one between 90 and 50 mbsf, with apparently redeposited material of an age near 0.22 Ma, and one above 50 mbsf, with rates slightly greater than 20 cm/k.y. The tie points from oxygen isotopes are based on analysis of Uvigerina auberiana and Globobulimina spp. at Scripps Institute of Oceanography (SIO) (Pérez et al., Chap. 19, this volume).
Sites 1081-1083 are referred to as the Walvis sites because they are either on Walvis Ridge (Site 1081) or on the slope of Walvis Basin (Sites 1082 and 1083). The closely related Lüderitz site is in the spheres of influence of both Walvis Bay and Lüderitz Bay, reflecting strong Namibian upwelling activity.
Site 1081 reaches back well into the late Miocene to ~9 Ma (Shipboard Scientific Party, 1998g). Judging from the constancy of the apparent sedimentation rate (Fig. F10A), the record is expected to be continuous. Stratigraphic control is excellent back to 2 Ma (Fig. F10B), with nannofossil events and magnetic reversal stratigraphy in good agreement. The addition of tie points from cycles of magnetic susceptibility adds nothing to the resolution but suggests that magnetic susceptibility is tied to glacial-interglacial cycles, as at the Congo sites.
There is one serious discrepancy between nannofossil stratigraphy and magnetic reversal stratigraphy at Site 1081: the last occurrence of Discoaster brouweri, dated at 1.95 Ma, occurs at 149 mbsf ± 2.2 m. According to the paleomagnetic data, this age level (1.95 Ma) is crossed above 120 mbsf (Fig. F10B; tie points at 1.95 Ma). An assumption that D. brouweri became regionally extinct at ~2.1 Ma (which would help resolve the discrepancy) is attractive, considering that at each of the Walvis sites (but not at the Lüderitz site) the last occurrence (LO) of D. brouweri is well below the corresponding reversal (bottom of the Olduvai Chron). Once this assumption is made, sedimentation rates for the three sites are relatively steady (4 cm/k.y. for Site 1081, rising to 6 cm/k.y. in the Quaternary; 8-2 cm/k.y. for Site 1082; and 6-8 cm/k.y. for Site 1083). The one exception is a pulse of high sedimentation rates at Site 1081 centered at 2.3 Ma (following cooling and ice buildup in the Northern Hemisphere) (Tiedemann et al., 1994; Mix et al., 1995; Bickert et al., 1997; Haug and Tiedemann, 1998; Maslin et al., 1998). The significance of this apparent pulse is not clear; it is not seen at the other Walvis sites, at least not when the stratigraphic data are interpreted as here proposed, which minimizes steps and pulses.
Another discrepancy between nannofossil stratigraphy and magnetic reversal stratigraphy arises at Site 1082 below 400 mbsf in sediments older than 3.5 Ma. According to the "Site 1082" chapter (Shipboard Scientific Party, 1998h), the magnetostratigraphy is ambivalent between 3 and 5 Ma and two differing models are proposed as viable. The straight-line biostratigraphic interpolation (Fig. F10C) favors "1082A-Model 1" (Shipboard Scientific Party, 1998h, p. 294, fig. 12). This discrepancy must not be confused with the one discovered for Site 1086, whereby paleomagnetic tie points are systematically younger than assigned nannofossil ages between 3.5 and 5 Ma. That situation is the reverse of the one described here and is owing to the use of different timescales (Berggren et al., 1995, vs. Lourens et al., 1996) (see Shipboard Scientific Party, 1998m). Site 1083 shows excellent agreement between nannofossil tie points and magnetic stratigraphy (Fig. F11A). The sedimentation rate is near 7.5 cm/k.y. throughout the sequence.
Site 1084 has the highest sedimentation rates of any of the sites between Walvis Ridge and the cape; rates range from ~8 cm/k.y. in the Pliocene portion of the record to well above 15 cm/k.y. in the late Quaternary. Periods of increase are centered near 2.8 Ma (a Northern Hemisphere cooling step) (Whitman and Berger, 1992; Tiedemann et al., 1994; Mix et al., 1995; Bickert et al., 1997; Haug and Tiedemann, 1998; Maslin et al., 1998) and near 1.1 Ma (a time of climate change in central Africa) (L. Dupont, pers. comm., 2001; Uliana et al., Chap. 11, this volume). The sedimentation rate maximum between 2.5 and 2.2 Ma coincides with a maximum in opal content (the Matuyama Diatom Maximum) (Lange et al., 1999; Pérez et al., Chap. 4, this volume). The maximum near 1 Ma marks a secondary diatom maximum associated with a carbonate minimum. The sediments of Site 1084, as mentioned, bear a strong imprint of hyperproductivity; they also show the greatest expansion of any of the Leg 175 sections (owing to high gas pressure). Thus, it is reasonable to assign much of the fluctuations in sedimentation rate (and the overall increase since the early Pliocene) to changes in supply of biogenous matter. Terrigenous contributions may vary similarly; only detailed analysis can resolve the question of the relative importance of the two components.
The Cape Basin sites (1085, 1086, and 1087) have records consisting of calcareous ooze, with carbonate values typically well over 60 wt% (Shipboard Scientific Party, 1998m, 1998k; Meyers and Robinson, Chap. 2, this volume, for Site 1087). Sedimentation rates at Site 1085 are higher than in typical pelagic carbonate oozes, varying from ~2 cm/k.y. in the middle Miocene to >5 cm/k.y. in the Pliocene (Fig. F12A, F12B). Maximum values appear just before 10 Ma (at the very time of a distinct carbonate minimum and a total organic carbon [TOC] maximum) (Shipboard Scientific Party, 1998k) and between 4 and 4.5 Ma (a secondary TOC maximum). A generalized increase is centered on 6.5 Ma, at the time of early glaciation pulses in the Northern Hemisphere (Larsen et al., 1994) when the Mediterranean dried up (Hsü et al., 1973a, 1973b) and the Himalayas first experienced considerable uplift (Raymo, 1994). It is reasonable to speculate that increases in sedimentation rates resulted from increases in export production and biogenic supply resulting from wind-forced upwelling and mixing.
The record of Site 1086 (Fig. F12B) suggests one way how sedimentation rates are decreased during times of increased upwelling. Site 1086 shows effects of winnowing in the uppermost portion of the record and a hiatus comprising most of the Brunhes Chron. Overall, sedimentation rate is low (<2 cm/k.y.), with the exception of a maximum centered near 5.7 Ma. The data for nannofossil stratigraphy and magnetostratigraphy provide for alternative age models, as mentioned. This is not because of an inherent discrepancy but because of the use of different timescales by physicists and geologists during Leg 175 (Berggren et al., 1995, for reversals; Lourens et al., 1996, for nannofossil events). The difference is as expected, given the two timescales. Using exclusively the timescale of Lourens et al. and making the corresponding correction to the depth-age line for Site 1085 (Fig. F12A) results in removal of the sedimentation-rate peak near 4.3 Ma and appearance of another peak near 5.3 Ma, that is, a position close to the peak at Site 1086. This illustrates the hazards of elaborating on changes in sedimentation rate (or using mass-flux rates in plotting sedimentation patterns) when age models are subject to substantial revision.
The stratigraphy of Site 1087 is well constrained by nannofossil events; magnetic data are useful back to 3.5 Ma (where the discrepancy noted at Site 1086 first becomes important). A continuous record apparently reaches deep into the upper Miocene; middle Miocene and pre-Neogene sediments also are present below disconformities. Sedimentation rates typically vary between 5 cm/k.y. in the late Miocene to ~4 cm/k.y. in the Pliocene and Pleistocene, where the variability may be greater. (Higher stratigraphic resolution also can result in greater variability of apparent sedimentation rates.) A pronounced maximum of sedimentation rate is centered near 3.2 Ma (the first strong cooling event of a series of cooling steps leading into Northern Hemisphere glaciation) (see Tiedemann et al., 1994; Mix et al., 1995; Bickert et al., 1997; Haug and Tiedemann, 1998; Maslin et al., 1998). The period is characterized by a strong TOC maximum centered near 140 mbsf (Meyers and Robinson, Chap. 2, this volume).
Given the age models here derived from the shipboard data of Leg 175, by allowing for uncertainties in position of events and by minimizing abrupt changes in sedimentation rates, "most likely" ages can now be assigned to the events used to produce the depth-age models and also to the events not used. The latter include all determinations based on radiolarians, diatoms, and foraminifers. Our concern was that the lesser abundance of guide fossils in these groups and their greater sensitivity to problems arising from poor preservation introduces additional uncertainties into the age models, which might as well be avoided.
Separate tabulations were made for the position of biostratigraphic events at the Congo and Angola sites (1075-1079), the Walvis sites (1081-1084), and the Cape Basin sites (1085-1087) (Tables T4, T5, T6). The three regimes were then compared to check for possible asynchrony (Table T7).
In Table T4 (as in the companion tables), there are three entries for age estimates for each stratigraphic event. The first is the age used in making the age models during Leg 175 (see Giraudeau et al., 1998). These are based on various earlier studies tying fossil events to age scales of varying reliability. The last serious change to the Quaternary age scale was in 1990 (Shackleton et al., 1990), and the scale for the last 3 m.y. or so now seems stable (e.g., Tiedemann et al., 1994; Mix et al., 1995; Bickert et al., 1997; Haug and Tiedemann, 1998; Maslin et al., 1998). Uncertainties in the early Pliocene were discussed above in connection with the age model of Site 1086. (Uncertainties for pre-Pliocene age determinations may still be substantial in places; T. Bickert, pers. comm., 2001.) The "effective ages" listed are valid within the framework of the stratigraphy of Leg 175, with emphasis on the nannofossil record and the minimizing of abrupt changes in sedimentation rate. The first of these "effective ages" is weighted by the time range within which the event occurs, given the depth range that constrains it. If the depth range is small (and the sedimentation rate high), the weight is correspondingly great. (The weight is taken as a function of 1/range.) The second "effective age" listing is based on the same data but without weighting. Differences resulting from the two methods are minor.
On the whole, the "effective ages" closely follow the "defined ages" in all three regions, as expected. Deviations are typically <50 k.y., corresponding to a depth difference of <5 m in deposits with a sedimentation rate of 10 cm/k.y. Five meters is also the typical range of uncertainty in assigning a depth position to an event when studying core catcher samples. If datums from different holes are used as though they were from the same hole (as we did), the uncertainty roughly doubles for any one sample. For several samples, there should be a convergence on the "defined age."
A number of discrepancies between "defined" and "effective" ages stand out and may be significant in terms of asynchronous first appearance or last occurrence relative to the biostratigraphic standard used.
For example (Table T7), the LO of Axoprunum angelicum (0.46 Ma) may be diachronous, occurring later than expected at the Congo and Walvis sites but not at the Cape Basin sites. Redeposition can move LO events upward. Alternatively, the LO valid for the open ocean is not valid for the coastal ocean.