Eight Centuries of North Atlantic Ocean Atmosphere Variability

See allHide authors and affiliations

Science  26 Nov 1999:
Vol. 286, Issue 5445, pp. 1709-1713
DOI: 10.1126/science.286.5445.1709


Climate in the tropical North Atlantic is controlled largely by variations in the strength of the trade winds, the position of the Intertropical Convergence Zone, and sea surface temperatures. A high-resolution study of Caribbean sediments provides a subdecadally resolved record of tropical upwelling and trade wind variability spanning the past 825 years. These results confirm the importance of a decadal (12- to 13-year) mode of Atlantic variability believed to be driven by coupled tropical ocean-atmosphere dynamics. Although a well-defined interdecadal mode of variability does not appear to be characteristic of the tropical Atlantic, there is evidence that century-scale variability is substantial. The tropical Atlantic may also have been involved in a major shift in Northern Hemisphere climate variability that took place about 700 years ago.

Although short-term variability in Atlantic climate is thought to be relatively well understood (1), both the patterns and the mechanisms of variability on decadal to century scales are as yet poorly known. In the southern Caribbean, the anoxic Cariaco Basin is well positioned to provide century- to millennium-length histories of trade wind–induced coastal upwelling (Fig. 1) as well as surface ocean changes that result from variations in the large-scale circulation of the Atlantic (2–4). Varved high-deposition-rate sediments [up to >100 cm per thousand years (ky)] and an abundance of microfossils result in one of the few marine records capable of preserving evidence of interannual- to century-scale climate variability in the tropical Atlantic. Between January and March, when the Intertropical Convergence Zone (ITCZ) moves close to the equator, strong easterly trade winds along the coast of Venezuela create intense Ekman upwelling and peak primary productivity over the Cariaco Basin and the continental margin (3,5). Beginning in June or July, as the ITCZ moves north to a position near the Venezuelan coast, the trade winds diminish, and upwelling over the basin weakens. Studies of Cariaco Basin microplankton have shown that local populations dominated by the planktic foraminifer Globigerina bulloides reflect the cooler, more nutrient-rich waters found along the coast and that seasonal assemblage changes mirror the changes in upwelling intensity, and thus trade winds, over the year (6).

Figure 1

Location and correlation map showing the spatial relation between variations in the abundance of the planktic foraminifer G. bulloides in the Cariaco Basin (southern Caribbean) and SST anomalies in the Atlantic Ocean. Mapped correlation coefficients (r) indicate the correlations between boxcar decadal-averaged G. bulloides data and historical SST anomaly data (12) for the time period 1856–1990 A.D. Correlations between G. bulloides abundance and SST anomaly time series are highest (–0.8 < r < −0.9, significant at the 99.9% level) in the North Atlantic region between 50o and 60oN, indicating thatG. bulloides abundance, and hence trade wind–induced upwelling over the Cariaco Basin, are tightly coupled to SST anomalies in the northern North Atlantic region.

We sampled sediments from box core PL07-71 BC (395 m water depth), recovered from the gentle northern slope of the eastern Cariaco Basin, at continuous 1-mm intervals (7). The age model for core PL07–71 BC (Fig. 2) consists of two independently derived, mutually supporting age data sets. Age control for the upper 12.6 cm of the core (1880 to 1990 A.D.) is based on robust faunal correlations with a nearby box core already possessing a well-constrained 210Pb and varve chronology (8). Below a depth of 12.6 cm, sediment laminations are less distinct and are difficult to count reliably. For this reason, below 12.6 cm (1165 to 1879 A.D.) the age model is based on a series of accelerator mass spectrometry (AMS) dates from monospecific samples of G. bulloides. Radiocarbon-based sedimentation rates for samples immediately at and below 12.7 cm are the same as the 210Pb/varve-based estimates above 12.7 cm. The resulting 825-year record has a sample resolution of about one sample per year near the top of the record, decreasing to one sample per 2.5 years at the base.

Figure 2

The age model for Cariaco Basin box core PL07-71 BC is based on a combination of210Pb, varve, and AMS 14C chronologies. For the upper 12.6 cm of PL07-71 BC, the age model is based on detailed correlations to PL07-81 BC (A), a nearby box core that is already well dated by 210Pb dating and varve counts (8), through an intermediate box core PL07-64 BC (B); dpm, disintegrations per minute. Cores PL07-81 BC and PL07-64 BC lie deeper in the basin and contain identical interbedded turbidites. Varve counts and 210Pb values both indicate that the turbidites are not erosive and that a continuous sediment sequence is present in both cores. Sediments from PL07-64 BC (B) above and between the turbidites were sampled at continuous 1-mm intervals and analyzed for their planktic foraminiferal content. These data were used to develop faunal correlations between PL07-64 BC (B) and PL07-71 BC (C) and to transfer the established chronology from PL07-81 BC (A) to the latter. An example of this type of faunal correlation is shown for the foraminifer Orbulina universa (data shown as a three-point smooth); multiple taxa were used to check and verify the correlations. For sediment in core PL07-71 BC deposited at a depth between 12.6 and 56.4 cm (D), an interval where laminae were too faint to count, the age model is based on a series of 12 AMS 14C dates measured on monospecific samples of G. bulloides(Table 1). Radiocarbon dates were converted to the calendar ages shown in (D) using a calibration model for marine samples (35); a local reservoir correction of 420 years, shown to be constant over periods of large climatic change, was used to account for radiocarbon differences between the surface waters and the atmosphere (4). Error bars represent the 1σ deviation; the unevenness of the error bars is a function of the marine calibration curve. A third-order regression line passing through both sets of age control points yields an r 2 of 0.98 and was used to assign calendar year ages to all samples in the box core deeper than 12.6 cm.

The recent (1880 to 1990 A.D.) abundance record for G. bulloides illustrates substantial variability that appears to be the result of changes in trade wind–induced upwelling intensity (Fig. 3A). Comparisons of G. bulloides abundance to a record of interhemispheric Atlantic sea surface temperature (SST) anomalies for the period 1901–1985 (Fig. 3B) (9) show that both records reflect well-known decadal-scale variations in tropical Atlantic surface conditions (10, 11). Trade winds are more intense over the Cariaco Basin (as measured by increased G. bulloidesabundance) when SSTs in the North Atlantic are colder than average, a pattern also seen when the G. bulloides data are compared directly to a record (12) of North Atlantic SST anomalies (45° to 65°N, 15° to 55°W; Fig. 3C). The overall relation between trade wind intensity and upwelling over the Cariaco Basin and Atlantic SSTs can be further illustrated in a spatial comparison ofG. bulloides abundance to Atlantic SSTs (Fig. 1). Correlations are highest (significant at the 99.9% level) in the northern North Atlantic region between 50° to 60°N, confirming a robust association between intensification of the Atlantic trade winds and cold conditions in the northern North Atlantic region.

Figure 3

(A) Comparison of G. bulloides abundance data (solid line; data are plotted with increasing values down- ward) from box core PL07-71 BC to Comprehensive Ocean-Atmosphere Data Set zonal wind speed data (dotted line) (36) for the grid square centered on the Cariaco Basin (11°N, 65°W). Wind speed and G. bulloidesabundance data in this panel are presented as 3-year boxcar averages for the period 1880–1990 A.D. Higher abundances of the upwelling-sensitive G. bulloides are associated with stronger zonal (trade) winds (r = 0.66, significant above the 99.9% level). (B) Comparison ofG. bulloides abundance data (solid line) from PL07-71 BC to interhemispheric SST anomalies (vertical bars) over the Atlantic basin calculated from (9) for the time period 1901–1985 A.D. High abundances of G. bulloides are interpreted as evidence of strong upwelling and intensified trade winds in the southern Caribbean. Negative temperature anomalies indicate a pattern in which the North Atlantic is cool relative to a warm South Atlantic, with variations in this index describing the well-known Atlantic thermal dipole (10, 11). The G. bulloides abundance record from the Cariaco Basin shows a clear decadal-scale correlation to the interhemispheric SST anomalies, with high G. bulloides abundances (strong trade winds and upwelling) associated with cooler North Atlantic temperatures. (C) Comparison of G. bulloides abundance data (solid line) with an index of North Atlantic SST anomalies (dashed line) (12), here calculated for the region of the Atlantic from 45° to 65°N and 15° to 55°W.

Previous work (13) indicates that when the cross-equatorial SST gradient in the Atlantic is reduced, the tropical North Atlantic is cool relative to the South Atlantic, resulting in increased surface pressure over the North Atlantic and a southward shift of the ITCZ. This leads to weakening of the southeasterly trades and intensification of the northeasterly trades, and hence to the stronger upwelling over the Cariaco Basin that is recorded by G. bulloides abundance. On decadal to interdecadal time scales, it has been suggested (14) that variability in the cross-equatorial SST gradient and wind intensity in the North Atlantic may be related to variability in Atlantic thermohaline circulation. Indeed, recent portions of the 1880–1990 A.D. G. bulloides abundance record correspond well to historical and instrumental evidence for North Atlantic circulation changes. During the 1970s, large negative salinity anomalies in the northern North Atlantic [the “Great Salinity Anomaly” (15)] were associated with unusually cold SSTs and substantially reduced deep convection in the Greenland Sea (16), one of the principal source regions of North Atlantic Deep Water. An inferred weakening of Gulf Stream flow during the 1970s accompanied this change (17), which was a time of higher G. bulloides abundance and stronger trade winds over the Cariaco Basin (Fig. 3). In contrast, stronger Gulf Stream flow during the 1950s occurred during a period of warmer North Atlantic SSTs and reducedG. bulloides abundance and trade wind strength in the southern Caribbean. However, the apparent lack of a temporal delay between increased G. bulloides abundance and cold SST anomalies and between decreased G. bulloides abundance and warm SST anomalies in the northern North Atlantic (Fig. 1) argues strongly for atmospherically driven Atlantic linkages rather than oceanic ones, particularly at interannual time scales. It may be that high-latitude oceanic processes provide the initial conditions that set the stage for low latitude–high latitude atmospheric linkages on interannual time scales. Alternatively, it is possible that North Atlantic SSTs and associated changes in trade wind–controlled upwelling within the Cariaco Basin are driven from the tropics and precondition the North Atlantic's thermohaline variability (18).

The complete 825-year record of G. bulloidesabundance and upwelling in the Cariaco Basin (Fig. 4A) exhibits the same level of distinct decadal variability as seen in the record of the past 110 years. Assuming that the recent correlation between wind stress and G. bulloides abundance holds in the older record, the G. bulloides data suggest that abrupt, large-magnitude changes in the North Atlantic trade wind system have occurred over longer time scales as well. Relatively long periods of stronger trade wind–induced upwelling are recorded between the years 1165–1360 A.D., 1410–1590 A.D., and 1640–1880 A.D., punctuated by intervals of weaker winds and lower foraminiferal productivity events centered at approximately 1380, 1620, and 1890 A.D.

Figure 4

Record of G. bulloides abundance variations (1-mm intervals) for the complete box core PL07-71 BC (from 0 to 56.4 cm), spanning the time interval from about 1165 to 1990 A.D. (solid line). Abundance data are plotted with increasing values upward. Given the increased perspective of the longer record, a strong pattern of interdecadal- to century-scale Atlantic variability is clearly seen to be superimposed on the decadal-scale variability of the past century. Spectral analysis (inset) (28) reveals concentrations of variance centered at periods of 140 to 500 years, 12.5 to 13.0 years, 8.7 to 9.0 years, 6.8 to 7.0 years, and 6.1 to 6.3 years (the Nyquist frequency is 1 cycle per 6 years; thus the inset has the entire spectrum). Also shown (main plot) is a visual comparison to calculated changes in 14C production rate (dashed line) (33), which is one measure of changing solar activity in the past. Periods of reduced solar output (the well-known Maunder, Spörer, and Wolf Minima) correlate well with periods of highG. bulloides abundance in the Cariaco Basin (strong upwelling and trade winds), suggesting a possible role for solar forcing on these longer time scales.

For the period before about 1900 A.D., few data exist on trade wind variability in the southern Caribbean. Nevertheless, for the period 1640–1880 A.D., there are hints that support our inference that North Atlantic trade winds were stronger and that the ITCZ was positioned further south on average. A southward shift of the ITCZ between 1600 and 1900 A.D. has been previously postulated, based on scattered historical South American precipitation records (19). Coral δ18O data from the Gulf of Panama have also been interpreted to indicate an intensification and southward shift of the trade winds between 1650 and 1800 A.D. (20).

Independent pre–20th century climatological records from the North Atlantic with resolution comparable to that of the Cariaco Basin record are sparse, but they do support inferences based on the Cariaco Basin G. bulloides abundance patterns. Fluctuations in records of summer ice core melt from Svalbard and periods of increased sea ice around Iceland match intervals of increased G. bulloides abundance (21), which suggests that the recent relation between colder northern North Atlantic SSTs and increased G. bulloides in the Cariaco Basin was maintained for the earlier period. In contrast, the G. bulloidesabundance record does not appear to correlate well with the three-millennia record of SSTs from the Bermuda Rise (22). However, correlations between Cariaco Basin G. bulloidesabundance and SSTs in the region of the Bermuda Rise (northern Sargasso Sea; Fig. 1) are poor (−0.2 < r < −0.4) for the period since 1856 A.D., so there is probably little reason to expect a good correlation between the Cariaco Basin and Bermuda Rise records.

An important feature of the long-term G. bulloides record is a distinct change in the character of variability observed near the base of the record (Fig. 4). In sediments deposited before about 1320 A.D., G. bulloides abundance values fluctuate dramatically, with rapid high-amplitude changes of 300 to 400% occurring regularly in less than a decade. Since 1320 A.D., G. bulloides abundance values have varied more typically on the order of 100 to 200% on time scales of 10 to 20 years. This more modern-like variability regime came into being with the initial Little Ice Age cooling of the North Atlantic after 1300 A.D. (22,23) and the demise of the Greenland Norse settlements (24). Other millennium-length records from the Northern Hemisphere show a change in the character of climate variability at approximately the same time. Tree ring–derived summer precipitation records from western North America show a change in character before 1400 A.D. at about the same time that extreme floods in the southwestern United States became more frequent (25). Lake sediment data from North Dakota also suggest a pronounced shift in drought regime around 1200 A.D., with droughts before this time characterized by greater magnitude, persistence, and frequency than droughts that have occurred since (26). A recent review of available data (27) supports the existence of a drought regime shift in western North America but notes that the timing may be somewhat later than 1200 A.D. and thus more in accord with the Cariaco Basin record. The implication is that regime shifts are a natural aspect of Atlantic variability and that these shifts may play a role in triggering changes in the frequency and persistence of drought over North America.

Spectral analysis of the G. bulloides time series was performed (28) to identify significant modes of Atlantic variability that might point to possible forcing mechanisms or related linkages (Fig. 4, inset). Results suggest that a century-scale mode of variability, distinct from the secular trend, appears to be an important natural aspect of Atlantic variability. Although the length of our new time series makes it is difficult to resolve the exact nature of this centennial mode, it is most likely related to long-term variations in the coupled ocean-atmosphere system, solar forcing, or a combination of these two. Climate model studies (29) indicate that significant interdecadal- to century-scale variability can result from processes (thermohaline circulation, for example) internal to Earth's climate system alone, but each model suggests a different time scale of dominant variability. The Cariaco Basin record indicates that the most important mode may be centennial rather than interdecadal in nature. This result contrasts with earlier analysis of the much shorter instrumental Atlantic record (30, 31), as well as of paleoclimatic data from land areas adjacent to the Atlantic (32), but it is quite plausible that an interdecadal mode is dominant in regions poorly correlated with Cariaco Basin variability (Fig. 1). Comparison of theG. bulloides record with one record of past solar variability (Fig. 4) (33) suggests that small changes in solar output may influence Atlantic variability on centennial time scales, a possibility supported by climate model experiments that indicate that solar variations can influence Hadley circulation and hence trade wind variability (34). Climate model results also suggest that past solar variability may have been large enough to affect the salinity balance of the North Atlantic and thus the thermohaline circulation, northward heat transport, and SST fields of the North Atlantic (34).

Over decadal time scales, the Cariaco Basin record demonstrates that significant spectral power in the 12.5- to 13-year band is an important long-term feature of natural Atlantic variability. In this case, the Cariaco Basin record confirms findings based on the instrumental record, where a nearly identical mode has been well documented (10, 30). This highlights the importance of tropical air-sea interactions in generating substantial decadal variability (10) and supports the hypothesis that Atlantic variability is linked to other tropical climatic variations [such as rainfall over the Sahel and northeast Brazil (9,11, 13)] in a long-term fundamental way.

  • * Present address: Department of Geological Sciences, University of South Carolina, Columbia, SC 29208, USA.

  • Present address: Institute for the Study of Planet Earth, Department of Geosciences, University of Arizona, Tucson, AZ 85721, USA.


Table 1

Results of AMS radiocarbon dating of Cariaco Basin box core PL07-71 BC at the Center for AMS (CAMS), Lawrence Livermore National Laboratory. Meas., measured; cal., calender.

View this table:
View Abstract

Navigate This Article