INTENSIFICATION OF NHG

Between 4.8 and 3.4 Ma, the ¹⁸O_benthic record reveals no significant long-term trend, pointing to warm early Pliocene climate conditions without prominent changes in ice volume (Tiedemann et al., 1994). After 3.2 Ma, ¹⁸O_benthic increases from 2.5 to values of more than 3.5 for MIS 96, 98, and 100 at ~2.5 Ma. This increase, which can be globally observed, marks intensification of NHG (Ruddiman and Raymo, 1988; Raymo et al., 1989; Kwiek and Ravelo, 1999). The preceeding cooling trend (MIS M2; 3.4 to 3.3 Ma), as shown by the ¹⁸O_benthic and SST_Mg/Ca records and which was abruptly terminated by the so-called mid-Pliocene warmth period (Crowley, 1996), may indicate a failed attempt of the climate system to start NHG (Haug and Tiedemann, 1998).

The decrease of SST_Mg/Ca since 3.7 Ma starts several hundred thousand years earlier than the start of intensification of NHG at ~3.2 Ma as generally shown by global ¹⁸O_benthic records, which represent global ice volume (Lisiecki and Raymo, 2005). Average SST_Mg/Ca decreases by ~2°C between 3.7 and 2.4 Ma. ¹⁸O_G.sacculifer increases by 0.8, which agrees with the change expected from the SST_Mg/Ca decrease (2°C = ~0.5) and the ice volume effect of 0.4, which would give a total of 0.9.

Spectral analyses on the SST_Mg/Ca record also indicate a change in spectral character at ~3.7 Ma. The interval from 4.8 to 3.7 Ma is characterized by a concentration of variance at eccentricity- and precession-related frequencies (Fig. F5B) with limited power at the 41-k.y. cycle. The interval from 3.7 to 2.4 Ma, comprising the tropical cooling trend in SST_Mg/Ca, in contrast, provides a strong 41-k.y. cycle and negligible variability at the precession band. The pronounced dominance of the 41-k.y. cycle after 3.7 Ma indicates that high-latitude climate forcing has played an important role in driving SSTs in the tropical east Pacific because the presence of a strong obliquity related cycle typically reflects a high-latitude origin.

The similarity of the Mg/Ca and ¹⁸O records of G. sacculifer with the ¹⁸O_benthic record for the time period younger than 3.7 Ma is expressed by cross-spectral analysis between the proxies and the obliquity signal (Laskar, 1990) (Table T1). Coherency of all three proxies with obliquity exceeds 0.90 and coherency between SST_Mg/Ca and ¹⁸O_benthic is 0.91, whereas none of the proxies significantly lags or leads obliquity. This suggests that the 41-k.y. rhythm in tropical east Pacific ¹⁸O_benthic, SST_Mg/Ca, and ¹⁸O_G.sacculifer was mainly driven by variations of global importance, triggered by high-latitude processes that caused intensification of NHG (Figs. F3, F4).

Although the tropical cooling trend since 3.7 Ma is not reflected by an increase in ice volume before 3.4 Ma, evidence for early Pliocene high-latitude cooling is provided by sediment records from the Irminger and Norwegian Seas (Kleiven et al., 2002; St. John and Krissek, 2002). These records show significant pulses of ice-rafted debris back to 3.5 Ma, indicating the existence and development of glaciers large enough to produce icebergs well before intensification of NHG.

The warm Pliocene before intensification of NHG (Ravelo and Andreasen, 2000) is probably associated with atmospheric CO₂ concentrations that were higher than preindustrial concentrations by ~100 ppmv (Van der Burgh et al., 1993; Raymo et al., 1996). Although the mid-Pliocene warm period is generally assumed to have continued until 3.2 Ma, the initiation of an earlier decrease in atmospheric CO₂ cannot be excluded. Therefore, the lead of cooling tropical SSTs over the increase of global ice volume could indicate that a decrease in atmospheric CO₂ concentrations might have played an important role in intensification of NHG (Li et al., 1998). Decreasing atmospheric CO₂ concentrations would lead to global cooling, but global ice volume will not start to increase before a certain threshold is reached, thereby resulting in a lag of global ice volume to global temperatures (Raymo, 1998; DeConto and Pollard, 2003).

The expression of intensification of NHG is contradicted by Ravelo et al. (2004). The authors argued that the main phase of intensification of NHG is not linked to changes in the tropics, but rather to changes in the subtropics. The main changes in the tropics occurred between 4.5 and 4.0 Ma, simultaneously with the critical threshold in the shoaling of the Isthmus of Panama (Haug et al., 2001; Billups et al., 1999), and after 2.0 Ma, when the present-day SST gradient between the east and west Pacific became established (Cannariato and Ravelo, 1997; Chaisson and Ravelo, 2000). These changes also suggest that the main influence of the subtropics on the tropics, namely providing the source for cooler upwelling waters, did not develop until 2.0 Ma. If deepwater temperatures at high latitudes started to decrease during the middle Pliocene and then propagated toward the tropics, the thermocline would have to shallow in order to influence SSTs. However, the main phase of shallowing of the thermocline at Site 1241 ended around 4.0 Ma (Steph et al., this volume), showing no significant changes around the period when SST_Mg/Ca started to decrease at ~3.7 Ma (Fig. F2). From our data, we therefore conclude that intensification of NHG is paralleled by cooling in the tropical east Pacific, which even preceeded intensification of NHG by ~500 k.y.

The ¹⁸O_salinity record does not seem to show intensification of NHG. Instead of the expected trend toward higher salinity because of increasing global ice volume since 3.2 Ma, a freshening trend is present for this time period (Fig. F2E), which had already started at ~3.6 Ma when SST_Mg/Ca also started to decrease. Also, the global obliquity signal of 41 k.y., which dominates the ¹⁸O_benthic, ¹⁸O_G.sacculifer , and Mg/Ca records, is not significant in the ¹⁸O_salinity record (Fig. F4D). This suggests that the variations in local SSS in the Pliocene east Pacific are mainly controlled by low-latitude processes and can be explained by changes in evaporation/precipitation that are most likely related to latitudinal variations in the tropical rainbelt (ITCZ) and the establishment of modern conditions (Benway and Mix, 2004) as a consequence of ongoing shoaling of the Isthmus of Panama. The average values of ¹⁸O_salinity for the time period 3.6 to 2.4 Ma are on average 0.4 lower than for the time period between 4.8 and 3.6 Ma, indicating generally fresher water masses at Site 1241. The present-day distribution of mixed-layer salinity in the east Pacific shows that lower saline water masses are related to the position of the ITCZ/NECC and the depth of the thermocline. Therefore, we conclude that between 3.6 and 2.4 Ma, mixed-layer water mass properties at Site 1241 became not only cooler but also less saline. This indicates that with intensification of NHG, the ITCZ possibly moved into a more southerly position with weaker southeast trade winds, a relatively weaker SEC and NECC, a strengthened EUC, and a shallower thermocline in comparison with the period from 4.8 to 3.6 Ma.