The Cariaco Basin is an anoxic marine basin off the coast of Venezuela (Figure 1) that is separated from the open Caribbean Sea by shallow sills (<146 m), and possesses varved sediments with the potential for continuous, high-resolution AMS ¹⁴C dating due to high concentrations of planktonic foraminifera. The climate cycle in the Cariaco Basin region consists of a dry season with strong trade winds and coastal upwelling, alternating with a rainy season with weaker winds and no upwelling. This climatic regime results in the annual deposition of laminae couplets of light-colored, plankton-rich and dark-colored, terrigenous mineral grain-rich layers. The annual nature of the laminae couplets was investigated in well-laminated surface sediments containing two distinct turbidites, using ²¹⁰Pb dating and historical records of earthquakes (which would be expected to result in turbidity currents) in the neighboring region. The independent dating methods yielded ages of 58 ± 4 and 89 ± 5 years (²¹⁰Pb), versus 61 and 90 years (historical) for the two turbidites, in good agreement with paired-laminae counts of 60 and 90 years, respectively. This agreement using multiple dating methods demonstrates that the laminae couplets are annually deposited varves.

The Cariaco Basin has distinct, thick laminae at depths corresponding to ages of 12.6-9.0 ¹⁴C kyr BP, after which they become thinner and less pronounced toward the surface. Four sediment cores were cross-correlated on the basis of distinct, millimeter-scale ‘marker laminae’, and laminae couplets were counted to create a floating annual chronology 5500 varve-years long, covering the period of deglaciation from approximately 8.0 to 12.6 ¹⁴C kyr BP. Accelerator Mass Spectrometer (AMS) ¹⁴C dates were obtained on sixty samples of the planktonic foraminifer Globigerina bulloides, hand picked from 1.5-to-2 cm thick samples, each corresponding to 10-15 varve years. The samples were taken from cores PL07-56PC and PL07-39PC, over the same sediment interval in which the varves were counted. The floating varve chronology was anchored to absolute calendar age by 'wiggle-matching' radiocarbon versus calendar age variations to those measured in the German pine dendrochronology. Recently, the German pines themselves have been securely anchored to German oaks and thus now constitute an absolute and continuous calendar time scale. The Cariaco Basin and German pine ¹⁴C variations were matched by incrementally adjusting the calendar-age offset until the correlation between the data sets was maximized (r = 0.99), anchoring the varve chronology in absolute calendar time. The uncertainty of the wiggle-match was determined by the amount of offset between the data sets that still yielded a correlation coefficient of at least r=0.99. This resulted in a ± 20 calendar year uncertainty during the period covered by the match itself. Cumulative errors in varve counting were constrained to accrue only within the 3000 years older than tree rings and provide an additional 1-2% uncertainty during that period.

The tight linkage between Cariaco Basin and Greenland ice core paleoclimate records was used to assess the accuracy of the anchored varve chronology. Cariaco Basin grey scale and accumulation from the GISP2 ice core are plotted during the period of deglaciation, each versus its individual annual chronology (Figure 2). In addition to the large changes at the Glacial/Bølling boundary and the beginning and end of the Younger Dryas (14.7, 13 and 11.7 cal kyr BP, respectively), the two records show similar events and patterns of change at decade-to-century scales. A least-squares procedure allowing manual identification of similar, large climate events as constraints (dashed lines, Figure 2) was used to align the GISP2 accumulation and Cariaco Basin grey scale records along their entire lengths. The resulting correlation between the two records is good (r = 0.69) and was used to assign the GISP2 layer-age chronology to the Cariaco Basin grey scale record. In this way, the two independently derived chronologies could be compared directly and differences between them quantified. The two chronologies consistently agree with less than 0.7% error. The pattern of increasing disagreement with age may reflect errors in either chronology compounding downcore, the direction in which the varves and annual ice layers were counted. For most of the period from 10-14.7 cal kyr BP, the difference is less than 20 years, and reaches 100 years only in the oldest part of the chronology (i.e. Glacial/Bølling boundary). The age difference meanders randomly about zero, rather than remaining consistently positive or negative, and is always well within the combined independent errors (~1-2 %) of the two chronologies. This assessment of Cariaco Basin varve ages, independent of ¹⁴C considerations, provides strong evidence for the accuracy of the Cariaco Basin calendar chronology.