Introduction

The tectonic closure history of the Central American Seaway (CAS) is complex and a long lasting process that started during the latest Oligocene and early Miocene1. Recent tectonic evidence from land suggests that deep and intermediate connections between both ocean basins were already closed during the middle Miocene2. However, manifold evidence from ocean records indicate the existence of shallow (≤200 m) connections between both ocean basins until the Pliocene and that the closure of these were sufficient to affect global climate3,4,5,6,7,8,9,10,11. Most notably, paleoceanographic data from either side of the Panamanian land bridge indicate that the further restriction of the CAS reached a critical threshold between ~4.8–4 Ma with marked impacts on ocean circulation and global climate5,6,7,8. Climate model simulations of this closure indicate a significant increase of the AMOC leading to higher sea surface temperatures (SST) and sea surface salinities (SSS) in the Northern Atlantic7,9,12. In contrast, the Southern Hemisphere experienced cooling and freshening through the increased transport of warmth to the Northern Atlantic, a climatic consequence known as “heat piracy”. Following CAS closure, the restriction of the Indonesian Seaway between ~4 and 3 Ma played a prominent role in changing ocean currents and climate in the tropical eastern Indian Ocean and southwest Pacific Ocean13,14,15,16 such as the onset of aridity in northwestern Australia15. In particular, this tectonic re-organisation of the Indonesian region caused a change in throughflow from warm and salty South Pacific to fresher and cooler North Pacific subsurface water masses with most prominent changes during 3.5–2.95 Ma13,16. The constriction of the Indonesian Seaway has been suggested to precondition the onset of Northern Hemisphere Glaciation16,17. Other tectonic-induced closures and openings discussed here include the end of the Messinian Salinity Crisis during the latest Miocene (5.96–5.33 Ma) when the Mediterranean Sea was periodically blocked and connected again to the Atlantic Ocean with simulated effects on the AMOC18,19. Also, we discuss possible climatic effects due to the opening of the Bering Strait that is assumed of having taken place during the Pliocene and proposed to be important for North Atlantic SST and the AMOC strength20,21,22.

In this study, we focus on Pliocene climatic changes on the ocean surface deciphered from proxy records of the South Atlantic Deep Sea Drilling Project (DSDP) Site 516A (30°17′S; 35°17′W) and of North Atlantic DSDP Site 552A (56°02′N23°13′W; Fig. 1). Our strategy is based on data and model simulations showing that the sea surface at the selected core locations likely reacted sensitively on past changes in AMOC in general23,24,25,26,27,28 and in particular in response to the constriction/closure of the CAS7,9,12. Site 516A lies within the Subtropical Gyre at the edge of the Brazil Current, which constitutes the South Atlantic Current and is suitable to monitor SST and SSS signature responses in the Southern Hemisphere due to changes in AMOC strength. For instance, a weakening of the AMOC should have resulted in the strengthening of the warm Brazil Current leading to warmer temperatures and increasing salinities23,24. North Atlantic Site 552A is located within the influence of the warm and saline North Atlantic Current at the surface layer, which transports heat and salt towards the northern North Atlantic27 and is therefore well situated to monitor SST and SSS changes in this northern limb of the AMOC7,9,12,25,26,27. Accordingly, changes in the strength of the AMOC are likely reflected in the interhemispheric SST difference between sites 552A and 516A and hence, can be used for assessing the strength of the AMOC25,28. The underlying assumption is that changes in the volume flux of warm waters towards the North Atlantic are directly related to North and South Atlantic SST25,28. We followed this approach using our SST and SSS reconstructions in combination with benthic δ13C values from the South Atlantic Site 1264 (smoothed record; ref. 29), which are indicative of changes in North Atlantic Deep Water strength29,30. North Atlantic Deep Water represents the deep-water return route of the shallow warm water transport towards the North Atlantic. Hence, we here monitor changes in the entire AMOC – including the shallow and the deep circulation. Paired Mg/Ca and δ18O measurements of the planktic foraminifera Globigerinoides sacculifer and Globigerina bulloides were used to reconstruct SSTMg/Ca and changes in relative SSS expressed as δ18Oseawater values (Supplementary Information). We examine long-term trends, as well as interhemispheric and interbasinal gradients in SSTMg/Ca and δ18Oseawater (expressed as smoothed records) to explore long-term supra-regional oceanographic and climate variability responding to tectonic changes. For consistent age control, we established benthic δ18O stratigraphies for sites 552A and 516A (Supplementary Information).

Figure 1: Modern annual sea surface temperature distribution at 30 m water depth54.
figure 1

ODP/DSDP sites are indicated. Paleoceanographic proxy data were generated for South Atlantic DSDP Site 516A and North Atlantic Site 552A (red dots). Chart was created with Ocean Data View (http://odv.awi.de; version 4.5.1)55.

Results and Discussion

North Atlantic DSDP Site 552A shows an inverse SSTMg/Ca development compared to South Atlantic Site 516A and Southwest Pacific Site 590B14 during the entire time period studied, suggesting that a long-term interhemispheric seesaw might have existed in the Atlantic Ocean at least since the latest Miocene (Fig. 2a,b). This tight anti-correlation between records, indeed, encourages us to use the Pliocene SSTMg/Ca gradient between North Atlantic Site 552A and South Atlantic Site 516A as a reflection of AMOC variability25,28. Robust independent support for our approach25,28 stems from a recent benthic δ13C record from Site 1264 in the Southeast Atlantic (Fig. 2c; ref. 29). This site location is bathed in the North Atlantic Deep Water and the δ13C record has been shown to be very sensitive to changes in North Atlantic Deep Water export into the South Atlantic30. The close similarity between the interhemispheric SSTMg/Ca gradient between sites 552A and 516A and the benthic δ13C record from the Southeast Atlantic Site 1264 (ref. 29; lower δ13C values mean less influence from North Atlantic Deep Water29,30) impressively support the tight connection of North Atlantic Deep Water formation and the Atlantic interhemispheric temperature gradient25,28 (Fig. 2c). This tight relationship is also reflected in our calculated δ18Oseawater gradient pointing to an enhanced interhemispheric SSS gradient with relative freshening at North Atlantic Site 552A and more saline conditions at South Atlantic Site 516 at times of AMOC weakening (Fig. 3a, b,c; refs 9, 12 and 23). Together, both our SSTMg/Ca and δ18Oseawater gradients reacted sensitively due to changes in AMOC.

Figure 2: Pliocene paleoceanographic changes.
figure 2

(a) G. sacculifer SSTMg/Ca records from Southern Hemisphere sites 516A (red line) and 590B14 (blue line). (b) G. bulloides SSTMg/Ca record from Site 552A (blue line), and alkenone derived SST from Site 607 (ref. 42) from the North Atlantic (brown line). (c) Interhemispheric SSTMg/Ca gradient between sites 552A and 516A (red; interpreted as deviation from 0) and benthic δ13C record from Site 1264 (ref. 29; smoothed black line). (d) G. tumida subsurface Mg/Ca derived temperatures from Site 1241 (ref. 8; green line), and other sites 848 (blue line), 849 (red line), and 853 (black line) from ref. 32. (e) Sand percentages at Site 999 (ref. 5; smoothed line). Shaded areas indicate distinct changes in proxy records due to the tectonic constrictions of the CAS and the Mediterranean Seaway IS = Indonesian Seaway. Thick lines represent smoothed lines based on a Stineman function with ±10% data range (performed with Kaleidagraph 4.1).

Figure 3: Surface salinity reconstructions during the Pliocene.
figure 3

(a) δ18Oseawater record from South Atlantic Site 516A (red line). (b) North Atlantic Site 552A δ18Oseawater record (blue line). (c) δ18Oseawater gradient between sites 552A and 516A as indication of relative changes of ancient salinities (red; interpreted as deviation from 0) and benthic δ13C record from Site 1264 (ref. 29; smoothed black line). Thick lines represent smoothed lines based on a Stineman function with ±10% data range (performed with Kaleidagraph 4.1).

We note an increasing SSTMg/Ca gradient between the North and South Atlantic of ~3 °C during ~5.6–5 Ma with most distinct cooling of the North Atlantic Site 552A (~1.5 °C) and warming of the South Atlantic Site 516A (~2 °C; Fig. 2a,b,c) at ~5.3 Ma. This event appears synchronous to the Messinian Salinity Crisis. During this time the Mediterranean Sea was periodically blocked from and connected again with the North Atlantic18 with effects on the salinity of Mediterranean Outflow Water19. A recent modelling study19 suggested that a related change from more to less saline Mediterranean Outflow Water conditions would have caused a significant weakening of the AMOC in line with a typical bipolar temperature and salinity asymmetry. Our observed pattern of an enlarged bipolar SSTMg/Ca gradient in the Atlantic Ocean during 5.6–5 Ma is synchronous to the evolving interhemispheric SSS gradient with a mean amplitude of ~0.5‰ (Figs 2c and 3c). Jointly, with the significant decrease of ~0.3‰ of the benthic δ13C record from the Southeast Atlantic Site 1264 at ~5.3 Ma29, these changes in the proxy records indicate a substantial weakening of the AMOC (Figs 2c and 3c).

From ~4.8–4 Ma, a strong SSS gradient developed between Caribbean Site 999 and tropical eastern Pacific Ocean Site 851 Ma6,31, while the eastern Pacific subsurface temperatures cooled and accordingly, the thermocline shoaled7,8,32,33,34 (Fig. 2d). At the same time, a continuous amplification of the AMOC was proposed5, witnessed by increasing sand percentages at Caribbean Site 999 (Fig. 2e; ref. 5). These were interpreted in terms of increasing carbonate preservation and hence, the presence of less corrosive, well-oxygenated North Atlantic Deep Water5. The benthic δ13C gradient between Caribbean sites 1000 and 925 also suggests the strengthening of the AMOC with better-ventilated Upper North Atlantic Deep Water7. All these developments are evident for the continuous constriction of the CAS, the climatically relevant effects of which include the strengthened transport of warmer, saltier North Atlantic Current waters towards the northern North Atlantic5,6,7,8,9,12. According to a multi model simulation9, the constriction of the CAS during the early Pliocene should have led to both: The shoaling of the thermocline in the tropical eastern Pacific and an interhemispheric “seesaw pattern” in SST and SSS. In consequence, the North Atlantic should have warmed with more saline conditions and the South Atlantic including the Southern Ocean should have freshened and cooled9.

However, a recent study based on benthic δ13C records from the deep North and South Atlantic30 suggested that during the early Pliocene only the production of upper North Atlantic Deep Water was increased due to the constriction of the CAS with minuscule climatic effects30.

Our proxy data are largely consistent with the model results and data studies suggesting significant climatic changes due to the constriction of the CAS5,6,7,8,9,12,22 (Fig. 4a). During ~4.8–3.8 Ma, synchronous sea surface cooling of ~2 °C at Southern Hemisphere sites 516 A and 590B14 and sea surface warming by ~2 °C in the North Atlantic (Site 552A) generate a diminishing interhemispheric SST gradient (~3.5 °C) between North and South Atlantic sites 552A and 516A (Fig. 2a,b,c). In accordance with the slightly increasing δ13C values at South Atlantic Site 1264 (ref. 29), this decreasing temperature gradient supports the amplification of the AMOC at ~4.8–3.8 Ma5,7,9,25 (Fig. 2c). Our δ18Oseawater values further imply the notion of significant climatic changes and a stronger AMOC during this time9 (Fig. 3c): During ~4.8–3.8 Ma, the δ18Oseawater gradient decreased between the North and the South Atlantic of ~0.5‰ suggesting more saline conditions at North Atlantic Site 552A.

Figure 4: Summary charts showing AMOC-related climatic changes during the constriction of the CAS.
figure 4

(a) and global cooling (b) respectively. Oceanographic changes as changes in AMOC variability, changes in temperatures and ocean currents are schematically indicated and based on the results of this study and other studies5,6,7,8,9,12,13,14,15,16,17,29,32,43,44,45,46,50,51,52. IS = Indonesian Seaway, BS = Bering Strait. Charts were created with Ocean Data View (http://odv.awi.de; version 4.5.1)55.

We note, however, in accordance to Bell et al. (ref. 30) that during ~4.8–3.8 Ma AMOC strengthening was better developed in the upper AMOC branch than in the deep AMOC as seen at Southeast Atlantic Site 1264 (ref. 29). Stronger circulation in the upper AMOC branch is supported by the distinctly increased sand percentages at Caribbean Site 999 (ref. 5; Fig. 2e) and the enlarged benthic δ13C gradient between Caribbean sites 1000 and 925 (ref. 7).

Our proxy data from both hemispheres support the hypothesis that the CAS constriction and the closely related strengthening of the AMOC even affected the tropical East Pacific by cooling/shoaling of the thermocline during ~4.8–4 Ma7,9 (Fig. 2d). Cessi et al. (ref. 35) proposed that the amplification of the AMOC, which is part of the globally spanning ocean circulation conveyor enhanced the heat transport from the tropical Pacific towards the North Atlantic. In consequence, the North Atlantic warmed, while the tropical Pacific thermocline cooled and shoaled35. The latter process was even fostered by the cooling of the Southern Ocean as evidenced by our sites 516A and 590B14 SSTMg/Ca records. The related intensified formation, northward spread, and equatorial upwelling of southern-sourced mode and intermediate waters due to strengthened wind circulation fostered global cooling through ocean-atmosphere processes36 and should have considerably contributed to the shoaling and cooling of the tropical eastern Pacific thermocline7,32,37. This process apparently started regionally as early as ~4.4 Ma, marked by the increasing SSTMg/Ca gradient between the equatorial West Pacific Site 806 and the East Pacific Site 84638,39.

Synchronously with the subsequent global cooling trend40,41, we observe during ~3.8–3 Ma a cooling of ~2–3 °C at the North Atlantic sites 552A and 607 (ref. 42) and a warming at South Atlantic Site 516A of the same magnitude leading to an increased interhemispheric temperature gradient between the North and South Atlantic of ~4 °C (less pronounced at Southwest Pacific Site 590B14) (Fig. 2a,b,c). Also, the δ18Oseawater gradient between the North and South Atlantic increased by ~0.5‰ with a relative freshening of North Atlantic Site 552A (Fig. 3c). Synchronously, benthic δ13C values at Site 1264 (ref. 29) decreased by ~0.25‰ indicative of a weaker North Atlantic Deep Water export into the South Atlantic (Figs 2c and 3c). All these proxy data suggest a weakening of the AMOC25,26,28 that is further supported by a comparison of Nd and Pb isotopes from the South and North Atlantic4343. Between ~4 and ~3 Ma both isotope signatures diverged between the North and the South Atlantic pointing to a weakened North Atlantic Deep Water circulation43.

Driving mechanisms for the Pliocene weakening of the AMOC are complex and include climatic changes in the North Atlantic and/or the Southern Hemisphere and might be a combination of the global cooling trend and/or tectonic changes in the Indonesian region or even the Bering Strait (Fig. 4b). For instance, as Pliocene cooling was more pronounced at high latitudes41 cooling of the Southern Hemisphere high latitudes would have initiated the northward displacement of the westerly wind belts which in turn, would have substantially weakened the AMOC44 (Fig. 4b).

The constriction of the Indonesian Seaway with the related emergence of the Maritime Continent with a gain in landmasses of ~60% including many islands contributed to the general global cooling trend since ~5 Ma45. Intensified weathering of basaltic rocks along with the rising of the Maritime Continent might have contributed to a long-term CO2 drawdown45. Further, the slight tectonic-induced decrease of eastern tropical Pacific SST might have led to a stronger east-west Pacific SST gradient causing a stronger Walker circulation with far reaching climatic effects on North America45. The tectonic-induced distinct freshening of the subsurface eastern tropical Indian Ocean due to the restriction of the Indonesian Throughflow waters during ~3.5–2.95 Ma13,15 (Fig. 4b) might have affected the Agulhas Current via the “warm water” route13,46. Freshening at the subsurface level in this current system was simulated to result in a substantial weakening of the AMOC47. Other temperature and salinity reconstructions along the “warm water route” in the South Atlantic also witnessed the climatic effects of the constriction of the Indonesian Seaway in combination with global cooling showing cooling and/or freshening since ~4 Ma (sites 1090, 1264 and 1084; refs 48, 49 and 17). Model simulations50,51,52 proposed the reduced poleward heat flux resulting from the constriction/closure of the Indonesian Seaway including the cooling of the Leeuwin Current with dramatic consequences for the climate of western Australia (drying), and a cooling of the southern South Atlantic and the Benguela upwelling region (Site 1084; ref. 17), respectively. This was largely confirmed by proxy data in those ocean regions15,17.

During the same time at ~3.6 Ma, a significant change in the migration pattern of Pacific molluscs into the Atlantic Ocean evidenced the opening of the Bering Strait21. This opening seaway was simulated to significantly lower SST in the North Atlantic and weaken the AMOC circulation20,22. However, the exact timing of this event was recently questioned and estimates now range from the early Pliocene until the late Pliocene21,22,53.

Conclusions

We show that the long-term Pliocene changes in the Atlantic interhemispheric temperature and δ18Oseawater gradients presented here reacted synchronously with a published benthic δ13C record from the Southeast Atlantic indicative of changes in North Atlantic Deep Water29,30. This similarity supports the hypothesis that the SST and SSS gradients between the North and South Atlantic closely reflect AMOC changes25,26,28. Overall, the proxy data allow to test the impacts of tectonic reorganisations of ocean gateways on the Pliocene climate. We suggest an early reduction of the AMOC at ~5.3 Ma, possibly related to the end of the Messinian Salinity Crisis. Between ~4.8–3.8 Ma, the reduced SSTMg/Ca and δ18Oseawater gradients between the North and the South Atlantic support hypotheses claiming that the CAS closure strengthened AMOC with prominent climatic effects on both hemispheres5,6,7,8,9,12,22 (Fig. 4a). During ~3.8–3 Ma, our surface proxy data in combination with the benthic δ13C record from the Southeast Atlantic29 suggest the weakening of the AMOC (Fig. 4b) that might be a complex climatic effect of global cooling possibly supported by tectonic changes in the Indonesian region (Fig. 4b).

Data Availability

Data of this study are available electronically at the World Data Center for Paleoclimatology (WDC Paleo), https://www.ncdc.noaa.gov/paleo/wdc-paleo.html”.

Additional Information

How to cite this article: Karas, C. et al. Pliocene oceanic seaways and global climate. Sci. Rep. 7, 39842; doi: 10.1038/srep39842 (2017).

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.