One of the keys to understanding climate change on tectonic timescales is the detailed knowledge of changes in oceanic and atmospheric circulation triggered by tectonic processes such as the opening and closing of oceanic gateways and the uplift history of great mountain belts. The upper Paleogene and Neogene sequences from Leg 202 (Sites 1236–1242) form a latitudinal transect off the coast of Central and South America, making these sequences particularly attractive for investigating the closure history of the CAS and the uplift history of the Andes (Mix, Tiedemann, Blum, et al., 2003). These tectonic events undoubtedly influenced the environment of the southeast Pacific. However, whether the paleoceanographic changes registered at Sites 1236–1242 reflect a response to the uplift of the Andes, closure of the CAS, paleodrift of the sites, or global climate change is difficult to separate, as their dynamics overlap in time (Fig. F6). Several geologic lines of evidence suggest that major uplift of the Andes was attained during the past 10 m.y. (Curry, Shackleton, Richter, et al., 1995; Gregory-Wodzicki, 2000), which temporally overlaps with the tectonic closure of the CAS from 13 to 2.7 Ma (Dengo, 1985; Duque-Caro, 1990; Collins et al., 1996). Even if the timing and identification of major changes in the sill depth and size of the CAS or in the uplift history of the Andes are known, the sensitivity level for profound changes in ocean circulation and climate may occur at any time during the tectonic process. Moreover, the site's paleodrift (e.g., from pelagic to hemipelagic environments or across oceanic fronts) has to be considered to distinguish the initial climatic-oceanographic response from drift-induced effects (Fig. F7).

The studies of Steph et al. (this volume, 2006) and Groeneveld et al. (this volume) at Site 1241 shed new light on closure-related variations in the thermal structure of the tropical east Pacific upper water column. These studies provide both high-resolution planktonic Mg/Ca temperature estimates (2- to 6-k.y. sampling distance) and oxygen isotope records (2- to 4-k.y. sampling distance) that span the time interval from 5.5 to 2.5 Ma. The combination of these proxies permits isolation of the 18O water signal, which is a measure of changes in ice volume and local salinity.

The comparison of reconstructed SSTs with those inferred from paleodrift of the sites provides an approach to assess whether the observed Pliocene temperature changes could be ascribed to plate tectonic movement. For example, the plate tectonic backtrack would have localized Site 1241 at 12 Ma close to the Equator several hundred miles southwest of its modern position (Fig. F7) (Mix, Tiedemann, Blum, et al., 2003). Assuming that the overall oceanic conditions remained constant during the Pliocene (i.e., reflecting modern oceanic conditions), we would expect a deepening of the thermocline and a rise of SSTs as the site moved out of the equatorial upwelling region into warmer waters of the North Equatorial Counter Current (NECC). Significant differences from these predicted trends would imply substantial changes in regional oceanographic conditions. Figure F8 provides a comparison between predicted (drift-induced, assuming no changes in geographic temperatures relative to modern oceanography) and reconstructed temperature changes for Site 1241 (Groeneveld et al., this volume). Planktonic Mg/Ca temperature reconstructions from the mixed-layer dweller G. sacculifer suggest regional warming of 1°–1.5°C in the tropical northeast Pacific from 4.8 to 3.7 Ma. This temperature increase would be consistent with the plate tectonic movement of Site 1241 toward a more northern position. However, the following cooling trend of 2°–3°C from 3.7 to 2.4 Ma contrasts the predicted drift-related trend and clearly points to regional long-term cooling. Assessment of salinity changes suggests that the cooling was also accompanied by mixed-layer freshening (Groeneveld et al., this volume). The timing and duration of these trends suggest a close link to the intensification of NHG, which started at ~3.6 Ma and ended at ~2.4 Ma with the buildup of large ice sheets in the arctic realm (Mudelsee and Raymo, 2005).

In comparison, recently reported SST reconstructions from Pacific Sites 846 (Lawrence et al., 2006) and 847 (Wara et al., 2005), which are positioned farther south of Site 1241 in the eastern equatorial upwelling region, also reveal significant Pliocene long-term cooling (Fig. F8). However, the timing for the onset of the cooling deviates significantly, even though both sites are located in the equatorial cold tongue within the influence of the South Equatorial Current (SEC). The SST record from Site 847 (estimated from Mg/Ca measurements in G. sacculifer) suggests that equatorial cooling began at ~2.5 Ma, whereas the alkenone-based SST record from Site 846 suggests that the cooling started at ~4.3 Ma, well before the intensification of NHG. This difference in timing between sites cannot be ascribed to paleodrift, as both sites drifted eastward toward the center of the cold tongue (if at all, this probably only predated the true regional cooling at the two sites). The different timing of the Pliocene cooling in the equatorial east Pacific (EEP) most likely resulted from a combination of global long-term cooling, regional changes in surface water conditions, and/or from differences in the depth habitat between species responsible for the SST proxy measured (as the alkenone unsaturation index of the organic remains from coccolithophorids should reflect shallower water temperatures than those derived from the Mg/Ca ratio of the mixed-layer dweller G. sacculifer). Hence, the present evidence makes it difficult to define a simple mechanism for Pliocene EEP cooling and to evaluate to which degree the cooling contributed to or resulted from NHG. However, the gradual and long-term character of Pliocene climate change points to slow tectonic forcing such as the closing of gateways or mountain uplift. Cane and Molnar (2001) proposed that the restriction of the Indonesian Gateway between 4 and 3 Ma probably reduced the atmospheric heat transport from the tropics to the higher northern latitudes, thereby cooling the arctic regions. The shoaling of the CAS may have increased the moisture supply to high northern latitudes via an intensification of the Atlantic meriodional overturning circulation (Driscoll and Haug, 1998). Enhanced freshwater delivery to the Arctic Ocean via Siberian rivers could facilitate sea ice formation, thereby increasing the albedo and reducing the heat transfer from the ocean into the atmosphere. Mountain uplift (Tibetan Plateau and Andes) may have led to weathering-induced CO2 removal and to changes in southeast Pacific atmospheric circulation during the Miocene and Pliocene (Raymo and Ruddiman, 1992; Hay 1996). Although the precise timing of when these tectonic processes became climatologically efficient is not well constrained, they have to be considered as an indirect cause of NHG and tropical cooling in the east Pacific.

Although the link between tropical gateway closures and NHG is still a matter of debate (Berger and Wefer, 1996; Haug and Tiedemann, 1998; Cane and Molnar, 2001; Ravelo et al., 2004; Haug et al., 2005), paleoceanographic studies suggest a close link between the formation of the Panama Isthmus and major oceanographic changes during the early Pliocene between 4.7 and 4.2 Ma, when shoaling of the CAS reached a critical threshold for upper ocean water mass exchange (Keigwin, 1982; Haug et al., 2001; Steph et al., 2006, this volume; Groeneveld et al., this volume). Restricted exchange of Pacific-Caribbean surface water masses led to the establishment of the modern Atlantic/Pacific salinity contrast that may be linked to the atmospheric net freshwater transport from the tropical Atlantic and Caribbean into the EEP (e.g., Jousaumme et al., 1986; Benway and Mix, 2004). In addition, results from general circulation models as well as paleoceanographic studies suggest reorganization of equatorial Pacific surface circulation and an increased volume transport of heat and salt into the North Atlantic via an intensified Gulf Stream, favoring North Atlantic Deep Water (NADW) formation and Atlantic carbonate preservation (Maier-Reimer et al., 1990; Farrell et al., 1995; Tiedemann and Franz, 1997; Cannariato and Ravelo, 1997; Mikolajewicz and Crowley, 1997; Haug and Tiedemann, 1998; Billups et al., 1999; Haug et al., 2001; Prange and Schulz, 2004).

Site 1241 was drilled in an ideal position to monitor oceanographic changes that may have resulted from the closure of the CAS, as it is close to the gateway region and comprises a complete late Neogene sequence of carbonate-rich and well-preserved foraminifers (except during the late Miocene carbonate crash noted by Lyle et al., 1995) from relatively shallow water depth (2000 m). Steph et al. (this volume) assessed changes in upper ocean stratification by comparing Mg/Ca temperature and 18O records from shallow- and deep-dwelling planktonic foraminifers (G. sacculifer and G. tumida) that span the time interval from 5.5 to 2.5 Ma at Site 1241. Their study indicates an early Pliocene shoaling of the tropical east Pacific thermocline that was marked by a 6°C temperature decrease (5.3–4.0 Ma) at the bottom of the photic zone, associated with two major steps at 5.3 and 4.5 Ma (Fig. F8). This decrease cannot be explained by the site's tectonic drift (Fig. F8). Instead, the early Pliocene shoaling seems to be a regional phenomenon of the tropical northeast Pacific, expanding to Sites 847 and 851, which are located ~1000 and 2000 nmi southwest of Site 1241, respectively. At Sites 847 and 851, planktonic 18O records from G. tumida document an increase in 18O after ~4.5 Ma (Cannariato and Ravelo, 1997; Chaisson and Ravelo, 2000; Wara et al., 2005; Ravelo et al., 2006), which has been interpreted as a temperature decrease at the bottom of the photic zone by assuming that the 18O records primarily represent a temperature signal.

The timing of the thermocline rise suggests a causal relationship to the shoaling of the CAS, as suggested by modeling experiments with the UVic Earth System climate model (Steph, 2005; Schneider and Schmittner, 2006). Model simulations for different sill depths suggest subsurface cooling of ~3°C and thermocline shoaling at Site 1241 when the sill depth rose from 700 to 130 m, as well as freshening in the tropical Pacific mixed layer when the sill depth rose to <130 m. Accordingly, thermocline shoaling may have occurred prior to freshening. The study of Groeneveld et al. (this volume) suggests significant freshening at Site 1241 after 3.7 Ma. The general consent between model results and proxy data corroborates a link to the progressive closure of the CAS rather than linking the observed changes to a response that may have resulted from the Pliocene constriction of the Indonesian Gateway (Cane and Molnar, 2001) or from the uplift of the Andes.

The Neogene uplift of the Andes Mountains is likely to have caused extensive changes in South American climate, wind-driven oceanic surface circulation, and biogenic productivity by reorganizing the pattern of atmospheric circulation and the hydrological cycle. Sedimentological and paleobotanical evidence from the Atlantic and the Pacific sides of the Andes as well as the frequent occurrence of ash layers in Leg 202 sediment records point to major uplift phases since ~12 Ma that were associated with intense volcanism and aridification of the Atacama Desert (for details see Mix, Tiedemann, Blum, et al., 2003). Paleobotanical evidence as summarized by Gregory-Wodzicki (2000) suggests a different uplift history for the central and northern Andes. These data suggest that the central Andes had attained no more than half of modern elevation by ~10 Ma and imply surface uplift of ~2000–3500 m since the late Miocene. Major uplift of the Colombian Andes has been suggested to occur at a later stage, between 2 and 5 Ma, reaching no more than 40% of modern elevation by ~4 Ma and modern height by ~2.7 Ma. Based on oxygen isotopes in lacustrine carbonates, Garzione et al. (2006) infer rapid rise of the Bolivian Altiplano to its current elevation between ~10.3 and ~6.8 Ma (with a data gap from 10.3 to 7.6 Ma), coincident with an abrupt increase in physically weathered chlorite and oxide mineral evidence for increasing terrigenous aridity from 8 to 7 Ma recorded in marine sediments off the Amazon (Harris and Mix, 2002) and with sedimentological data from Atacama Desert that indicate that desertification commenced at ~8 Ma (Hartley and Chong, 2002).

Uplift-induced paleoceanographic changes in the eastern South Pacific are expected to derive mainly from reorganizations in atmospheric circulation. Progressive uplift of the Andes formed a barrier for the trade winds, which forced the low-level high-velocity winds on the western side to follow the coastline (Hay, 1996). The associated increase in trade wind strength should have initiated a chain reaction of environmental changes, including enhanced coastal upwelling, lower SSTs, reduced evaporation, and increased onshore aridity (desertification of the Atacama). South Pacific Sites 1236 and 1237 are best suited to prove such hypothesized climatic and oceanic changes as well as possible thresholds and feedbacks that might emerge from the uplift of the Andes.

At both Sites 1236 and 1237, eolian iron oxides are accompanied by a significant eolian siliciclastic fraction since the late Miocene (Fig. F9). After ~8 Ma, increases are evident in both hematite content at Site 1236 (especially after ~6 Ma) and total dust flux at Site 1237, indicating enhanced eolian deposition likely sourced from the Atacama Desert. Moreover, the increases in eolian deposition and aridification are paralleled by a pronounced increase in productivity at Site 1237 (Fig. F9).

To characterize late Neogene variations in the thermal structure of the upper water column at Site 1237, Abe et al. (this volume) and Wara and Ravelo (this volume) reconstructed alkenone- and Mg/Ca-derived SST records that span the time interval of the last 6 m.y. (Fig. F10). Assuming no changes in regional oceanography, tectonic drift of the site would predict long-term cooling of surface waters and shoaling of the thermocline (Fig. F10) over the past 6 m.y. as the site approached the coastal upwelling area off Chile. In contrast, the alkenone-derived SST record (Abe et al., this volume) suggests no marked trend and relatively warm temperatures between 5 and 3 Ma. However, temperatures significantly declined by more than 5°C since 3 Ma, a trend that was possibly amplified by the site's paleodrift. This cooling trend is consistent with that observed in other subtropical upwelling areas (e.g., Marlow et al., 2000) (Fig. F10), and the range of late Pleistocene temperature variability agrees with alkenone temperatures reconstructed at a nearby piston core (Calvo et al., 2001; Prahl et al., 2006). The alkenone-derived Pliocene–Pleistocene cooling trend, however, conflicts with Mg/Ca temperature reconstructions of Wara and Ravelo (this volume), which suggest no significant long-term temperature changes over the past ~6 m.y at Site 1237 (the conversion of Mg/Ca ratios determined at tests from G. sacculifer was carried out by applying the equation of Anand et al., 2003) (Fig. F10). Another peculiarity is that the Mg/Ca–based mixed-layer temperatures are at times warmer than the alkenone-based surface temperatures, which is not very likely. Whether the absolute temperature estimates are biased by factors other than in situ temperature, which affected the alkenone unsaturation index (Uk´37), or the amount of Mg2+ incorporated into foraminiferal tests during calcification, which include, for example, diagenetic processes, selection of paleotemperature equations, and redistribution of sediment material, is difficult to assess without further data. The application of species-specific Mg/Ca temperature equations for G. sacculifer (Nürnberg et al., 2000; Dekens et al., 2002) instead of the multispecies equation of Anand et al. (2003) would have resulted in even warmer (~1°–2°C) mixed-layer temperatures (e.g., Steph et al., this volume). Effects such as preferential removal of Mg2+ in response to enhanced carbonate dissolution would result in lower Mg/Ca temperature estimates (e.g., Dekens et al., 2002). Another question is to what extent the Uk´37 signal is biased by seasonal cycles in coccolithophorid production, as the Uk´37-SST calibration is based on annual mean SSTs (Müller et al., 1998). If the Mg/Ca signal of G. sacculifer and the sedimentary Uk´37 signal at Site 1237 both primarily reflect that of the warm season, the alkenone-derived temperatures (Fig. F10) might be too low. Assuming that the long-term trends in both temperature records are reliable, at least, provides an interesting perspective. The habitat-related temperature gradient between both temperature records may point to enhanced stratification prior to ~2.5 Ma and to an enhancement in upwelling over the past 2.5 m.y., when subsurface and surface temperatures were more similar. This interpretation would be consistent with the observed increase in trade wind strength, continental aridification (Atacama), and productivity as indicated by a strengthened increase in dust, organic carbon, and biogenic opal accumulation rates (Fig. F9). The timing of these wind-induced changes between 3.0 and 2.5 Ma suggests amplification in atmospheric circulation that may have been driven as a threshold response to a steeper pole–equator temperature gradient, owing to the amplification of polar glaciation rather than to the uplift of the Andes.

This hypothesis is further constrained by paleoceanographic evidence from the Benguela upwelling system off the coast of southwest Africa. The Benguela upwelling system is analog to the eastern boundary current and upwelling system off western South America, but along a mountain range that experienced no significant uplift during the late Neogene. Marlow et al. (2000) presented an alkenone-derived SST record (0–7 Ma) from the Benguela upwelling system that is consistent with the general evolution of SSTs at Site 1237 (Abe et al., this volume). SSTs of the Benguela upwelling system remained warm and relatively uniform during the late Miocene and the early Pliocene and significantly declined by ~10°C since 3.2 Ma (Fig. F10). The decrease in SST in the Benguela system has been interpreted to reflect enhanced wind-driven upwelling, consistent with increased trade wind strength and aridification of Africa throughout the cooling transition (amplification of NHG), as evidenced by records of eolian dust flux to marine sediments (Tiedemann et al., 1994; Ruddiman and Janecek, 1989; Hovan and Rea, 1991). The similarity of temporal changes in SST, upwelling, and dust flux between the Benguela and Chile upwelling systems suggests that the uplift of the Andes was probably of secondary importance for generating the observed oceanic changes at Site 1237 during the last 6 m.y., as the Benguela upwelling system was not affected by mountain uplift. Hence, the expected atmosphere-oceanic response of the southeast Pacific to the uplift of the Andes, if any, likely occurred prior to 6 Ma during the time of major uplift. A full test of this hypothesis must await additional reconstructions of temperature and other oceanographic properties in the interval that spans the likely time of Altiplano uplift near 8 Ma.