One of the most perplexing questions in climate is why global climate over the past 40 m.y. changed from very warm conditions (the "Greenhouse World") to conditions of unipolar (Southern Hemisphere) and later bipolar glaciation (the "Icehouse World"). Partial answers include plate tectonic processes, elevation and erosion of vast plateaus, opening and closing of oceanic gateways, and changing concentrations of atmospheric greenhouse gases. The southeast Pacific is a key location for examining the Neogene and Paleogene response of climate, oceanography, and biogeochemical systems to events such as the closure of the Isthmus of Panama and the uplift of the Andes.

Andean Uplift, Eolian Dust, Volcanism, Climate Change, and Biogeochemistry

The Neogene uplift of the Andes Mountains is expected to have caused extensive changes in South American climate, wind-driven oceanic surface circulation, and, hence, productivity by reorganizing the pattern of atmospheric circulation and the hydrological cycle. But the timing and consequences for climate and ocean circulation, possibly responding to critical thresholds in the uplift history, are poorly constrained.

Continued uplift raised the Andes to an altitude of >4000 m by the end of the Neogene, establishing a barrier for southeast trade winds in the subtropics and for westerly winds in the midlatitude regions of South America (Lenters et al., 1995). Today, the blocking of the westerly winds results in enhanced precipitation on the western side of the mountain range (Chilean and Patagonian glaciers) and causes a strong rain shadow on the eastern side (Patagonian desert). A reversed rainfall pattern marks the low-latitude regions, where the barrier leads to enhanced precipitation from the trade winds on the eastern side of the mountain chain. Heavy rainfalls form a major source for the Amazon River, which drains its sediment load into the Atlantic. The western side suffers from a rainshadow effect and drier climate conditions, as expressed in the Atacama Desert. Moreover, the interplay between the uplift of the Andes and the semipermanent subtropical high-pressure cell over the southeast Pacific may have significantly amplified the aridification along the western margin of South America, as suggested by theoretical studies (Hay, 1996). The Andes constrict the counterclockwise flow on the east side of the high-pressure cell and force low-level high-velocity winds to follow the coast line. This results in enhanced coastal upwelling, lower sea-surface temperatures, reduced evaporation, and increased onshore aridity.

Marked changes in the erosion history of the Andes were detected during Leg 154 (Ceara Rise), which drilled the Atlantic side of South America. The increase in the Amazon River's supply of terrigenous sediments and its change in clay mineralogy indicate major uplift phases from 12 to 8 Ma and since ~4.6 Ma (Curry et al., 1995), as well as substantial regional climate changes (Harris and Mix, 2002). Uplift over the past ~10 m.y. is consistent with paleobotanical reconstructions from the central and Colombian Andes (Gregory-Wodzicki, 2000) and from ~4.6 Ma (Hooghiemstra and Ran, 1994; van der Hammen et al., 1973). The early Pliocene phase is paralleled by the subduction of Cocos Ridge at ~5 Ma, which formed during the passage of the Cocos plate over the Galapagos hotspot and dramatically elevated the Central American volcanic arc and led to the final phase of the closure of the Isthmus of Panama (Dengo, 1985). Unfortunately the land record can not be used to assess the details of Neogene water mass changes that respond to mountain building, as the land record of uplifted marine sediments contains major hiatuses, for example in Chile between 10 and 3.5 Ma (Martinez-Pardo, 1990).

During Leg 202, we recovered a latitudinal transect of sediment records off the Pacific coast of South America that spans the time interval of the last 32 m.y. and offers an opportunity to assess the impact of Andean uplift on long-term changes in volcanic activity, continental aridity, upwelling/productivity, trade winds, and dust transport.

The long-term history of eolian deposition in the subtropical southeastern Pacific is best recorded at Sites 1236 and 1237 (Fig. F33). Today, both sites underlie the path of eolian transport from the Atacama Desert in Chile. The southeast trade winds are the major dust carriers as indicated by the pattern of quartz distribution in southeast Pacific surface sediments (Molina-Cruz, 1977). The presence of terrigenous hematite and goethite, as well as other mainly clay-sized siliciclastics at Sites 1236 and 1237 (Fig. F33), is indicative of a far-field eolian component (Pye, 1987). The combined records of these tracers suggest that eolian dust has accumulated in the subtropical southeast Pacific since at least the late Oligocene. Shipboard data from Site 1236 provide no information about eolian deposition for the sediments older than middle Miocene because the sediment deposition was dominated by supply from a nearby carbonate platform. Conversely, at Site 1237, the iron oxide signal is lost, probably diagenetically, in sediments younger than ~8 Ma.

From ~31 to ~15 Ma, before the major uplift of the Andes, Site 1237 was located nearly parallel to its modern latitudinal position farther westward (a result of plate tectonic movement), ~2200-1500 km away from the South American coast, and thus received less dust (Fig. F33). Siliciclastic accumulation rates were lower than today by an order of magnitude and indicate no significant trend over this interval. The hematite record, however, indicates a change in the source region of dust, which was significantly enriched in hematite during the Oligocene and depleted during the Miocene. This gradual change between 26 and 22 Ma may indicate a change in weathering conditions that was possibly associated with a major change in global climate, when late Oligocene warming (~25 Ma) reduced the extent of Antarctic ice and led to warmer conditions that prevailed until the middle Miocene (Zachos et al., 2001). The timing of this shift in eolian sedimentation is also roughly coincident with the opening of the Drake Passage from 39 to 22 Ma (Barker and Burrell, 1977), which established a deepwater connection between oceans and the Antarctic Circumpolar Current. The presence of dust far away from the South American coast suggests that arid or semiarid conditions existed in subtropical South America prior to the uplift of the Andes. Aridity in this region could have been the result of loss of humidity by the southeast trade winds along their continental path across South America and the presence of the adjacent cool Peru-Chile Current (Frakes, 1979). Before the major uplift of the Andes, the trade winds probably had a more zonal distribution that allowed deposition of eolian hematite at Site 1237.

For the period between ~15 and ~8 Ma, a hematite record exists at both sites. Higher hematite content at Site 1237 during this interval is consistent with its closer proximity to a potential South American dust source area. A first slight increase in siliciclastic accumulation rates occurred at ~14 Ma. This may reflect the onset of aridity in the Atacama Desert, as suggested by Alpers and Brimhall (1988) to occur sometime between 15 and ~8 Ma, possibly in response to the uplift of the Andes. However, the increase in siliciclastic accumulation rates since ~14 Ma may also result from strengthened trade winds in response to enhanced global cooling during the Miocene from ~14 to 10 Ma, associated with the reestablishment of a major ice sheet on Antarctica (Zachos et al., 2001), or from the eastward migration of Nazca plate, which moves Site 1237 closer to the dust source (Fig. F33).

The abrupt loss of the iron oxide signal at Site 1237 (~8 Ma) appears to result from diagenetic reduction processes due to an increase in productivity and enhanced rain of organic matter to the seafloor rather than to reflect a change in the dust source area. This is corroborated by the fact that the reddish color signal of the iron oxides persists into the Pleistocene sequence at Site 1236, located ~500 km to the southwest of Site 1237, and thus had an even larger distance to the dust source.

Since the late Miocene and Pliocene, eolian iron oxides are complemented by a significant eolian siliciclastic fraction. Since ~8 Ma, increases are evident in both hematite content at Site 1236 and total dust flux at Site 1237, indicating enhanced eolian deposition (Fig. F33). In accordance, recent sedimentological data from middle Miocene to upper Pliocene successions in the modern Atacama Desert indicate that desertification commenced at ~8 Ma and was punctuated by a phase of increased aridity at ~6 Ma (Hartley and Chong, 2002). Moreover, the increase in eolian deposition and aridification is paralleled by a pronounced increase in productivity (reaching a maximum at ~6 Ma) along the equatorial upwelling belt (Sites 1239 and 1241 and Leg 138 sites) and in the southeast Pacific, in a region outside the major coastal upwelling zone but within the influence of the Peru-Chile Current (Sites 1236 and 1237) (Fig. F34). A change in atmospheric circulation could explain these features. Intensified trade winds would enhance equatorial upwelling/productivity and the northward advection of nutrient-rich waters transported by the Peru-Chile Current. Modern upwelling along the Peru continental margin occurs over a broader region than just a very narrow coastal strip (Fig. F34). This may be because the wind stress curl associated with the topographically steered winds (the result of an effective mountain barrier) is positive, creating a broader upwelling zone than a strictly uniform wind with an equatorward component (Pickard and Emery, 1990). Thus, the late Miocene paleoposition of Site 1237 was possibly influenced by upwelling features. Enhanced coastal upwelling, in turn, would have led to increased aridification of the Atacama Desert.

The inferred enhancement in wind speed at ~8 Ma is consistent with a coarsening in eolian sediments at south equatorial Sites 848 and 849 (Hovan, 1995), which were recovered during Leg 138. At about the same time, the first occurrence of discrete ash layers in sediments of the South Pacific record the onset of intense volcanism in the central Andes and possibly a phase of major uplift that led to the deposition of 55 ash layers at Site 1237 over the last 9 m.y. (Fig. F35). Taken together, this chain of evidence may indicate a critical threshold in the uplift history at ~8-9 Ma by forming an effective topographic barrier that enhanced the steering of trade winds along the coast, resulting in stronger meridional flow, enhanced eolian transport, coastal upwelling, and biogenic productivity. This event is also recorded in the Atlantic Ocean as a rapid increase in the chlorite:kaolinite ratio off the Amazon, suggesting an abrupt increase in physical weathering following uplift (Harris and Mix, 2002).

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 the modern elevation by ~10 Ma and imply surface uplift on the order 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 its modern elevation by ~4 Ma and a modern height by ~2.7 Ma. The temporal distribution of ash layers along the latitudinal transect of the Leg 202 sediment records might be a useful proxy for tracing the different evolution of the mountain range from Central to South America, if major uplift phases were accompanied by intense volcanism. The sediment records of Leg 202 comprise >200 volcaniclastic horizons that were deposited during the last ~9 m.y. (Fig. F35). Maxima in ash layer frequency occurred at ~8-6 Ma off the coast of southern Peru (~16S). Farther to the north, between 2S and 8N, significant amounts of ash layers did not occur before ~5 Ma. The Miocene occurrence of ash layers at north equatorial Site 1241 is interpreted to reflect the volcanic activity of the Galapagos hotspot based on the presence of black lapilli-sized scoria. The deposition of such ash layers ceased at ~6 Ma, probably because of the northeastward movement of Site 1241 away from the Galapagos hotspot. The interval of the last 5 m.y. is marked by ash layers enriched in clear glasses, possibly originating in Central America.

Another maximum in ash layer frequency marks the interval of the last 3 m.y. at all sites along the transect between 16S and 8N. The late appearance of ash layers in the northern sediment records during the Pliocene might be indicative of the major uplift phase in the northern Andes between 5 and 2 Ma.

However, it is difficult to conclusively prove a primary teleconnection between the uplift of the Andes and the observed changes in volcanism and atmospheric and oceanic circulation because the inferred uplift rates have limited confidence in absolute paleoelevation and age control. Even if the timing and identification of major changes in the uplift history are known, thresholds for profound changes in climate and ocean circulation are unknown. Furthermore, it is difficult to distinguish the initial climate response in equatorial and southeast Pacific sediment records from effects that may result from the plate tectonic drift of site locations toward the continent. Nevertheless, the sediment records recovered during Leg 202 have an excellent potential to provide detailed insights into the complex tectonics-climate connection.

Additional insights are thought to result from a comparison of sediment records from the southeast Pacific and southeast Atlantic off the coast of Namibia (Leg 175 and proposed for Leg 208). These regions share strong similarities. Both coasts are bordered by highlands that force the meridional component of the southeast trade winds in connection with the subtropical high-pressure cell to cause enhanced coastal upwelling and desertification (Namib and Atacama Deserts). However, the highlands along the coast of Namibia rise to an altitude of no more than 2000 m and thus may serve as a modern analog to the late Miocene elevation of the Andes and its impact on atmospheric and oceanic circulation. We mention these thoughts here in the hope of encouraging further research into the tectonic linkages, especially experiments with general circulation models, to test the sensitivity of atmospheric circulation patterns to progressive mountain uplift in the midlatitudes and their effect on oceanic heat transfer via the eastern boundary currents.

Closure of the Isthmus of Panama

The Neogene tectonic closure of the Central American isthmus from 13.0 to 2.7 Ma (Duque-Caro, 1990; Collins et al., 1996) resulted from the subduction of the Pacific Cocos and Nazca plates beneath the North and South American plates and later the Caribbean plate. The early phase of the closure restricted the exchange of deep and intermediate water between the Pacific and the Atlantic and is considered as a potential cause for the carbonate crash near the middle/late Miocene boundary (Lyle et al., 1995). This interval is characterized by periods of severe carbonate dissolution and has been documented in sediment records from the eastern equatorial Pacific (~9-11 Ma) and the Caribbean (~10-12 Ma). The dissolution events are more recently interpreted to reflect an intensified influx of corrosive southern source water from the Atlantic into the Caribbean and into the equatorial east Pacific, both in response to a strengthened global thermohaline circulation associated with enhanced production of North Atlantic Deep Water (Roth et al., 2000). However, the major phase of the carbonate crash occurred ~1 m.y. earlier in the Caribbean than in the Pacific, leaving open several questions concerning the timing and mechanisms.

In addition, the closure has always been an attractive candidate for the ultimate cause of the Pliocene intensification of the Northern Hemisphere glaciation since ~3.1 Ma (Mikolajewicz and Crowley, 1997). Closure-induced changes in global thermohaline circulation have been invoked to be the cause either for the onset (Berggren and Hollister, 1974) or for the delay (Berger and Wefer, 1996) or for setting the preconditions of the Northern Hemisphere glaciation (Haug and Tiedemann, 1998; Driscoll and Haug, 1998).

Although the link between the isthmus closure and the Northern Hemisphere glaciation is still a matter of debate, recent studies clearly identify a close link between the formation of the Isthmus of Panama and major oceanographic changes that occurred between 4.6 and 4.2 Ma, when the Panamanian sill shoaled to a water depth of <100 m (Haug and Tiedemann, 1998). The chain of evidence suggests the development of the modern Pacific-Atlantic salinity contrast of ~1 (Haug et al., 2001), a reorganization of the equatorial Pacific surface current system, the intensification of Upper North Atlantic Deep Water formation, and the development of the modern chemical Atlantic-Pacific asymmetry, which is reflected in a strong increase in Caribbean/Atlantic carbonate preservation and a remaining strong carbonate dissolution in the Pacific (Farrell et al., 1995b; Haug and Tiedemann, 1998; Haug et al., 2001). This major salinity contrast between ocean basins, driven in part by net freshwater transport as vapor across the Isthmus of Panama, is likely responsible for maintaining the global thermohaline "conveyor belt" circulation, which is dominated by North Atlantic Deep Water (Gordon, 1986; Broecker, 1991).

In contrast to Leg 138, the equatorial sediment records from Leg 202 provide only limited evidence for the middle-late Miocene carbonate crash, probably due to the shallower site locations. All information about the Pacific carbonate crash is based on records deeper than 3000 m water depth, a depth level that is highly sensitive to fluctuations in the carbonate compensation depth (CCD) in the eastern equatorial basins (Lyle et al., 1995). However, Site 1241 on Coccos Ridge was affected by enhanced carbonate dissolution between ~9 and 12 Ma (Fig. F36), although positioned in a water depth of ~2200 m, well above the modern CCD. The tectonic backtrack moves Site 1241 southwestward, closer to its origin at the Galapagos hotspot into the equatorial high-productivity belt, and into a water depth that was probably several hundred meters shallower during the middle to late Miocene than it is today.

The Miocene interval from ~9 to 12 Ma is marked by low carbonate accumulation rates (Fig. F34) and low carbonate concentrations averaging ~40 wt% (Fig. F36) but relatively high mean sedimentation rates of ~50 m/m.y. Biogenic opal, organic carbon, and siliciclastics became significant contributors to the sediment composition. The preservation of calcareous nannofossils, planktonic foraminifers, and even benthic foraminifers was affected strongly by carbonate dissolution (Fig. F36). The overall increase in biogenic opal (including laminated diatom oozes), organic carbon (up to 1.5 wt%), and sedimentation rates is indicative of high surface productivity and enhanced organic carbon rain. Fluctuations in the lysocline depth associated with a change in carbonate vs. biogenic opal productivity (>Corg/CaCO3-flux) and supralysoclinal carbonate dissolution in response to organic matter degradation are considered as major contributing factors to the carbonate dissolution events. These findings may revitalize the discussion about the mechanisms that contributed to the carbonate crash. At this stage of results, however, it is difficult to postulate any link to the closure of the Isthmus of Panama.

The most prominent feature of long-term changes in the Leg 202 sediment records is a pronounced late Miocene to early Pliocene maximum in biogenic accumulation rates, suggesting an interval of enhanced oceanic surface productivity between ~8 and ~4 Ma with a maximum at ~6 Ma (Fig. F34). A similar but shorter interval of rapid biogenic accumulation (6.7-4.5 Ma) was found at equatorial east Pacific sites during Leg 138 and is often referred to as the late Miocene to early Pliocene biogenic bloom (Farrell et al., 1995b). Compelling evidence that the biogenic bloom occurred throughout the entire tropical Indo-Pacific is summarized in Farrell et al. (1995b). The ultimate cause of the rise and fall of this bloom is unknown but has been linked to a variety of previous hypotheses including sea level variations, continental weathering, deepwater circulation, trade wind fluctuations, and the closure of the Isthmus of Panama. Sites 1236 and 1237 trace the productivity maximum for the first time farther south and suggest an early onset of this event in the eastern boundary current of the South Pacific. The onset and continuation of the productivity event was probably associated with an increase in trade wind circulation since ~8 Ma as discussed in the context of Andean uplift in the previous section. Whether the wind-driven northward advection of cool and nutrient-rich Southern Ocean waters via the eastern boundary current and its injection into the South Equatorial Current is a possible source for this biogenic bloom will be studied postcruise.

The fall of this event was possibly associated with a decrease in the strength of the southeast trade winds as indicated by grain-size studies on eolian sediments from south equatorial Sites 848 and 849 (Hovan, 1995). The grain-size records, spanning the interval of the last 10 m.y., reveal a pronounced early Pliocene minimum, inferring a decrease in wind speed. Site 1237 underlies the path of eolian transport from the Atacama Desert and is ideally located to verify this hypothesis. In addition to changes in atmospheric circulation, the emergence of the Isthmus of Panama may have played a critical role because the end of the bloom, at ~4.5 Ma, coincides with a critical threshold in the closure history that led to a decrease in equatorial east Pacific sea-surface salinities (Haug et al., 2001) and an eastward shift of maximum biogenic opal productivity (Farrell et al., 1995b). The preliminary results from Leg 202 provide no clear timing for the Pliocene fall of the productivity event (Fig. F34) because the carbonate accumulation rates continued to decrease into the Pleistocene at Sites 1236, 1237, and 1241. At Carnegie Ridge, off the coast of Ecuador, the picture is even more complicated because here the biogenic bloom persisted until ~3 Ma and Site 1238 and perhaps until ~2 Ma at Site 1239. However, equatorial Sites 1238, 1239, and 1241 are ideally located to monitor changes in biogeochemical cycles and equatorial oceanography that may result from the closure history of the Isthmus of Panama.