The vertical and seasonal
structure of gradients in temperature, salinity, nutrients, and light
penetration in the Cariaco Basin will govern the distribution of haptophyte
production in the upper water column and determine the value of the Uk´37
recorded by underlying sediments. Consideration of modern hydrography helps
clarify the meaning of late Holocene values of Uk´37.
Hydrography also provides a lens to interpret past variations in Uk´37
and other indices of paleo-SST such as planktonic 
18O
and faunal assemblages. The important considerations for interpreting variations
in the organic index include the vertical range of haptophyte growth in relation
to thermal gradients and seasonal variations in the flux of haptophyte-produced
alkenones from the euphotic zone to the seafloor.
Time-series data provided by F. Müller-Karger (pers. comm., 1997) in the Cariaco Basin not far from ODP Site 1002 characterize mean annual and seasonal variations in temperature and salinity in the basin. These data are consistent with previous "snapshot" pictures of Cariaco Basin hydrography (cf. Herrera and Febres-Ortega, 1975, cited by Peterson et al. [Chap. 4, this volume]; Lin et al., 1997) but provide a more representative picture of seasonal variance through the upper water column. We modeled the mean annual and seasonal variations in temperature and salinity by fitting the data at 10-m depth intervals to an equation representing a mean plus annual and first harmonic of the annual cycles (cosine wave) with mean amplitudes and phases of the cycles determined by least-squares regression. The addition of the first harmonic term made little difference to the range of temperatures calculated at most depths, but it significantly improved the fit to salinity. Figure 1 shows the application of this model to temperature at 2 m, which we consider SST. Clearly, because of the limited duration of the data set, we cannot model interannual variations in hydrography that might affect paleoceanographic proxy data. Mean annual and seasonal variations in temperature and salinity with depth are presented in Figure 2. The sill depth at 146 m very nearly coincides with the base of the thermocline.
Our Hole 1002C core-top temperature estimate of 24.9°C (Table 1) based on the alkenone unsaturation index is in striking agreement with the mean annual SST in the Cariaco Basin, which we determined to average 25.1°C from the oceanographic time series. We also determined the Uk´37 of eight surficial samples from box cores collected by the Plume-07 cruise to the Cariaco Basin in 1990. These data should represent the most recent sediment deposition in the basin and hence be more closely related temporally to historical hydrographic data than the ~600-yr-old near-surface sample from Hole 1002C. The eight box core samples give a mean Uk´37 of 0.906, equivalent to 25.5°C, with a standard deviation equivalent to ±0.4°C. Alkenone production must therefore occur very near the sea surface under present conditions in the Cariaco Basin.
The coincidence of the alkenone temperature signal with mean annual conditions does not prove that there is no seasonality of production and sedimentation of these organic compounds or no potential depth bias to their temperature of production. The annual range at the sea surface is estimated to be 5.5°C, and at times subsurface temperatures reach core-top Uk´37 temperatures. However, if we look at an enlarged view of the upper 100 m of the Cariaco site, we see that it is hard to reconcile alkenone temperature estimates with production below 25 m at any time of the year (Fig. 3). It also seems unlikely that the alkenone signal is produced dominantly during the upwelling season of winter to early spring, which should result in core-top temperatures of ~23°C. Mixed-layer depths are quite shallow according to the Müller-Karger time series: we calculated an average mixed-layer depth of 15.8 ± 7.2 m over the length of the series. The alkenone core-top temperature estimate, therefore, can be explained most simply as a reflection of production either within or just below the mixed layer, with perhaps a slight bias to postupwelling growth.
These considerations of present-day relationships set the stage to consider hydrographic factors that might influence the interpretation of Uk´37 temperatures and the relationship of these temperature estimates to foraminiferal faunas and isotopic time series. First, it seems likely that alkenone paleotemperatures in the Cariaco location will not have a strong upwelling signal embedded in them. Changes in alkenone paleotemperatures at our sample resolution of ~500 yr should reflect decadal- to century-scale climate variations instead of seasonal changes. Second, of the three foraminiferal species analyzed by Lin et al. (1997) (Neogloboquadrina dutertrei, Globigerina bulloides, and Globigerinoides ruber), the conditions of alkenone production most closely approximate those of the foraminifer G. ruber in the modern environment, whose isotopic signal appears to represent production under mean annual mixed-layer conditions (Overpeck et al., 1989; Lin et al., 1997).
