Biostratigraphic, paleomagnetic, and other data, if available, were used to construct age-depth models for Leg 202 sites. The models are tabulated in the site chapters and plotted along with all reliable age control points from which the models were derived. The mcd depths were used for each site (see "Composite Section").
Linear sedimentation rates (LSRs) were calculated for regular time intervals, and corrected LSRs were computed to compensate for the growth rates inherent in the mcd depth scale. Dry density data were used to convert corrected LSRs to total MARs. Carbonate concentration data (and in one case TOC and biogenic opal concentrations) were used to estimate carbonate (and TOC and opal) MARs.
Plots of all age-depth models, LSRs, and MARs were prepared for each site that had sufficient age-depth control datums to allow construction of an age model (Sites 1233 and 1236-1242). The results were briefly discussed in the site chapters and in the leg summary.
Shipboard age models are constructed under considerable time pressure and are afflicted with uncertainties that mainly result from a limited number of age control points, the quality and depth range of a datum, and the age itself. Each datum event has an age uncertainty that may vary from a few thousand years to several hundred thousands of years. The ages of individual datums calibrated elsewhere may not necessarily apply to a given site and this introduces an additional age uncertainty for some datums. Biostratigraphic sampling is often limited to one per core (i.e., a sample interval of ~10 m), which results in significant depth uncertainties for the datums. Reworking and dissolution may render some microfossil groups less reliable than others as a function of sediment composition and diagenesis. Thus, inconsistencies are frequent. Precise paleomagnetic datums were provided for some of the sites where chrons and polarity reversal boundaries could be clearly identified and calibrated with biostratigraphic datums.
The common biostratigraphic and polarity reversal approach provided no age control points for the upper Pleistocene to Holocene high-resolution records at Sites 1233-1235. However, Site 1233 provided an unprecedented record of centennial to millennial scale variability in paleomagnetic intensities, which are a powerful tool of global extent for stratigraphy. Additional Holocene age control points at Site 1233 were derived from correlation with a nearby sediment core, using magnetic susceptibility core logging data, and mapping of the 14C datums obtained for that sediment core to Site 1233.
Given all these circumstances, the construction of age-depth models is to some degree a subjective process and represents a preliminary first approximation of the true age-depth relationship. We applied a procedure resulting in conservative shipboard age models that satisfy as many constraints as possible without introducing artificial or speculative features in the model. The starting point was a plot of all control points from biostratigraphy, paleomagnetism, or other sources if available. Age and depth uncertainties were presented with error bars. Age ranges are rarely reported for biostratigraphic datums, but depth ranges can be calculated from the samples that were analyzed. Obvious outliers and conflicting datums were then masked until the line connecting the remaining control points was reasonably smooth. A smooth curve was fit to these points using locally weighted (10% of data) least-squares (Stineman function). The curve fit eliminates artificial kinks in the model that are functions of where a tie point was chosen within a depth range. A further step was used at some sites, including depth shifting of some control points to make the curve fit pass through more data points (their areas of uncertainty) or to constrain hiatuses. The final shipboard age model was as conservative as possible, preserving inflections in the age-depth curve that are supported by the data and smoothing those that are likely artifacts.
Paleomagnetic data, if available, were usually given first priority in constraining the age-depth model in case of conflicting datums, followed by calcareous nannofossils, planktonic foraminifers, and diatoms. In cases of conflicting microfossil datums we also took into account the reliability of individual datums as global dating tools, the stratigraphic reliability of fossil groups or specific datums, abundance, preservation, and reworking, if indicated, and the uncertainties associated with the first or last occurrence of a datum (first occurrence datums are generally preferred).
The age models were presented as plots of original datums with error bars overlain by the final shipboard age model, and tabulated as age-depth series, in each site chapter. These age models will undoubtedly be refined and replaced by detailed postcruise analysis.
Once the final shipboard age model is defined, calculating and plotting LSRs and MARs is a process that involves a number of simple interpolations and computations. The only subjective part of this process is the initial selection of time intervals at which the smooth age model is sampled. Instead of calculating LSRs and MARs over intervals given by the original stratigraphic control points, we decided to choose regular and conservative time intervals. As a rule of thumb, a time interval should include and average at least three age control points. For Site 1236-1239 and 1241, a 1-m.y. window was used so the results could be compared conveniently between sites. A 0.4-m.y. window was used for the high-resolution Sites 1240 and 1242, and a 10-k.y. window was used for Site 1233, which has an extremely high sedimentation rate. It should be kept in mind that the time series is an arbitrary choice of an average sampling rate and does not reflect the resolution of the age model equally in all intervals. However, in conjunction with a smooth age model, it presents the most realistic and conservative trends in LSRs and MARs.
First, the smooth age model was sampled at the age series points to obtain interpolated depths in mcd for those ages. LSRs in meters per million years were then calculated as the first derivative of the age-depth series over the selected interval. The LSRs were subsequently corrected to compensate for the expansion of the recovered sediment sequence represented by the mcd scale (see "Composite Section"). This correction was made by dividing the LSR by the appropriate growth factor for each depth interval.
In preparation for computing the MARs, dry density obtained from moisture and density measurements (see "Physical Properties") and carbonate concentration data (see "Geochemistry") were plotted against mcd and outliers were removed. Ages were interpolated for each depth point using the shipboard age model. The dry density and carbonate values were averaged over the regular age intervals. This averaging step is important to avoid a commonly introduced bias that occurs when dry density data at relatively high age/depth resolution are multiplied by LSRs calculated at a much lower resolution. Such a bias gives the impression that we know the MAR variations at great age/time and amplitude resolution when, in fact, associated LSR variations may tend to compensate for the dry density variations (higher dry density typically results in lower LSR). The generally low resolution of LSRs determines the resolution of MAR variations, even if the available dry density record has a much higher resolution.
After correcting LSR values for the growth factor and averaging dry densities over the appropriate interval, MARs were calculated using the equation
Component MARs were calculated by multiplying the bulk MAR by the concentration of the appropriate component (such as calcium carbonate).
Finally, a step plot of LSR, total MAR, and carbonate MAR summarizes the data. Although a smooth plot through the midpoints of the age intervals would convey the same information, we preferred the step plot because it illustrates clearly the selected age resolution that we considered to be reasonable and conservative.