Supplementary online material for Comparing glacial and Holocene opal fluxes in the Pacific sector of the Southern Ocean, [Paper # 2008PA001693]: Methods of core selection and age model construction The appendix seeks to describe in greater detail the methods used to select cores for study, to create chronologies for those cores, and to reject cores later found to be outside the time period of interest. The first section will explore the proxies mentioned in Section 3 of the manuscript in greater depth. Magnetic susceptibility (MS) in sediment reflects three primary variables: the mineral composition of the sediment, the flux of magnetic minerals, and the flux of non-magnetic (largely biogenic) minerals. The magnetic susceptibility of sediment therefore increases with increasing magnetic mineral content and flux, and with decreasing biogenic flux (the reverse is also true). Since biogenic fluxes south of the APF were lower and the supply of lithogenic material was greater during glacial periods [Chase et al., 2003; Frank et al., 2000; Kumar et al., 1995], magnetic susceptibility records there are reasonably easy to interpret, as increased magnetic susceptibility corresponds to glacial periods. North of the polar front, where both lithogeinc and biogenic fluxes were greater during glacial periods [Chase et al., 2003; Frank et al., 2000; Kumar et al., 1995], interpretation of magnetic susceptibility is not as straightforward because the glacial to interglacial change in MS will depend on whether the decreasing flux of magnetic minerals or the decreasing dilution by biogenic minerals had the larger effect. However, MS can still be used to identify systematic patterns in downcore records north of the APF, after which point other proxies can be used to constrain the age model. The percent abundance of the diatom Eucampia antarctica can be used to establish glacial versus interglacial periods within a core of established Quaternary age. Glacial periods show E. antarctica abundance of up to 60% as compared to interglacials (<10%) [Burckle and Cooke, 1983; Burckle and Burak, 1995]. This difference is driven by changes in the absolute number of open ocean diatoms, and therefore the relative abundance of E. antarctica has also been interpreted as a sea ice proxy. That is, as sea ice extent increased during glacial periods open ocean diatom growth was inhibited, increasing the relative abundance of E. antarctica. It has also been shown that % E. antarctica correlates closely with planktonic foraminiferal delta18O records, enhancing its use in distinguishing between glacial and interglacial periods [Burckle and Burak, 1995]. Relative abundances of calcium carbonate and opal can also be used to help identify glacial and interglacial periods in the Southern Ocean. Carbonate minima occur during glacial periods throughout most of the Southern Ocean due to a combination of corrosive, low CO32- bottom waters and low production [Howard and Prell, 1992; 1994]. This method is more effective in cores with slightly higher overall carbonate content (~10%), and less effective in cores where the entire range of variability lies between the detection limit and 1 or 2 percent. Percent opal content must be interpreted regionally, much like MS. South of the APF, previous studies have shown a consistent decrease in opal burial during the last glacial period [Chase et al., 2003; Dezileau et al., 2003; Francois et al., 1997; Frank et al., 2000; Kumar et al., 1995], and so in this region we can interpret opal minima as approximating the glacial maximum. Glacial opal fluxes north of the APF were greater than Holocene fluxes (see above references), but interpreting downcore patterns of opal abundance in this region is more complicated due to an increase in lithogenic flux during the glacial period. Therefore, the opal content of glacial sediments north of the APF reflects both increased opal flux (increasing percent opal abundance) and increased lithogenic flux (decreasing percent opal abundance by dilution). Since the difference in the relative glacial increase in opal and lithogenic material varies at each site, north of the APF percent opal can only be used to identify cycles to target for investigation with other methods. The following paragraphs describe results from three cores to illustrate how sediment composition was used to constrain the age model. Composition data referred to here, as well as composition data from all other cores included in the study can be found in Figure 2. For E11-3 (56.903degS, 115.243degW, 4023m, 5.287degN of the APF) all four composition proxies (percent E. antarctica, percent opal, percent carbonate and magnetic susceptibility; see section 3) were used in addition to 14C dating in constructing the age model. Radiocarbon dates (Table 2) at 18cm (10 200 +/- 45 radiocarbon years) and 48cm (26 400 +/- 170 radiocarbon years) identify the upper 18cm of core as roughly Holocene in age, and constrain the LGP to sediments above 48cm. The relative abundance of carbonate drops from 60% at the top of the core to 35% at 36cm, and remains between 30% and 40% through 80cm. The CaCO3 minimum below 36cm is interpreted to reflect low-CO32- bottom waters found in the glacial SO (see above), and therefore indicates sediment of glacial age; the CaCO3 maximum at the coretop is consistent with Holocene sediment. The relative abundance of E. antarctica peaks at 36cm (13.7%), which is also an indication of glacial sediments at that depth, while lower values at the coretop (4.3%) are consistent with Holocene sediments. Magnetic susceptibility is more than twice as high from 36 - 48cm (1.9 - 2.2 SI units) than at the coretop (0.2 - 0.8 SI), which further supports the idea that 38 - 48cm represents the last glacial period, while the uppermost 18cm of core represent the Holocene. Percent opal increases down core from 20% at the coretop to 42% at 36cm and 45% at 48cm, and remains near 45% through 80cm. As this core lies north of the APF, the higher opal content below 36cm cannot strictly be interpreted as representing the LGP, but in the context of other proxies it can be considered supporting evidence for glacial-aged sediments. In summary, radiocarbon dates constrain the age of the LGP to between 18cm and 48cm, and indicate that the upper 18cm of the core are likely Holocene in age. This is in agreement with the four other proxies, all of which show maxima (except carbonate, which shows a minimum) between 36 - 48cm, indicating the LGP. Core E17-7 (61.083degS, 134.35degW, 4435m, -2.583degS of the APF) is an example of a core for which magnetic susceptibility data was unavailable, and so the age model was constrained using the remaining proxies (percent E. antarctica, percent opal, percent carbonate and 14C dating; see section 3). A radiocarbon date at 40cm yields an age of 13 500 +/- 50 radiocarbon years, constraining the Holocene to sediments above 40cm, and the LGP to sediments below. Percent carbonate varies between 3% and 5% near the coretop, then peaks at 45cm (14%) before reaching a minimum at 60cm (0.2%); carbonate values below 60cm do not exceed 0.1%. This period of extremely low carbonate is interpreted to encompass the LGP, as in E11-3 described above. Percent abundance of E. antarctica varies between 2 - 3% from 20 - 40cm, peaks at 60cm (10.8%), and steadily decreases through the core to values of 2 - 3% from 120 - 140cm. This suggests that the LGP lies somewhere between 60cm and 100cm. Opal abundance is slightly less at 60cm (65%) than at 15cm (70%), but reaches 76% at 80cm, and 85% at 100cm. Since E17-7 lies south of the modern APF, this helps to constrain the LGP to 60cm since glacial opal abundance was generally less than the Holocene in that region. Again, E. antarctica, opal and carbonate were able to consistently refine the age model around the single datum provided by radiocarbon dating. Core E19-7 (62.163degS, 109.088degW, 5051m, 0.483degS of the APF) has extremely low carbonate content (below the detection limit) throughout, and therefore neither percent carbonate abundance nor radiocarbon data are available. However, it was still possible to constrain the age model using opal abundance, magnetic susceptibility and E. Antarctica abundance. The percent abundance of E. antarctica varies between 4 - 6% in the upper 20cm, peaks between 42 - 52cm (9%), and decreases to approximately 6% in the deepest samples measured (82 - 92cm). This peak is interpreted to represent part of the LGP. Magnetic susceptibility in the upper 20cm of the core ranges between 0 and 1.5 SI units, peaks between 45 - 60cm (5 SI), and decreases to values of 2 - 3 SI between 75 - 95cm. This peak is consistent with sediments between 42 - 60cm representing the LGP, and with coretop sediments representing the Holocene. Opal abundance ranges from 69 - 71% in the upper 20cm of the core, decreases to 37% at 42cm, and increases to values between 60 - 72% between 62 - 92cm. Since E19-7 is south of the polar front, the opal minimum at 42cm, coinciding with maxima in MS and in the abundance of E. antarctica, can be interpreted as the depth of the LGP. Therefore, despite the lack of carbonate data from this core, it is still possible to constrain the LGP (42 - 52cm) and the Holocene (2 - 12cm). Using the methods described above, age models were constructed for each core. In general, the identified Holocene section of the core was assigned an age range of 0-10ka, unless radiocarbon evidence suggested that coretop sediment was older than modern. Sections of transitional composition stratigraphically below the Holocene were assigned ages between 10 and 18ka, again relying on any available radiocarbon data for further refinement. Samples identified as glacial in age were assigned ages between 18 and 28ka, pending radiocarbon or compositional evidence to the contrary. We recognize the potential for error in this method, and it is important to consider the effects of these errors on 230Th-normalized fluxes. In 230Th-normalization, the age assigned to a given sample is used to decay-correct the 230Th present in that sample. An underestimate of sample age results in 230Th-normalized fluxes that are too high, and vice verse. However, the errors associated with incorrectly estimating the age by 10ka are fairly small, approximately 9%. Additionally, for the purpose of comparing the LGP and Holocene, we will for the most part use the average of several samples within each period, further minimizing errors. Therefore, while we recognize the limits inherent in this method of age control, we feel that it is a useful tool for the comparison of glacial and Holocene data, as well as the comparison of our data with other studies. Once glacial and interglacial periods have been identified in a core, 231Pa/230Th ratios and 230Th-normalized fluxes of lithogenic material can be used to determine if the glacial-interglacial cycle at the top of the core is in fact the most recent one (i.e., LGM and Holocene). This is not always the case! Using the age model described above, it is possible to calculate initial unsupported 231Pa/230Th ratios and 230Th-normalized detrital fluxes. As 231Pa has a half-life of 32 800 years, sediments older than the LGP have noticeably low 231Pa/230Th ratios, which can help to identify a missing glacial cycle. As the 231Pa/230Th ratio of SO sediment varies spatially to some extent, it is important to compare results from a particular core to the average of the surrounding area in order to determine if 231Pa/230Th ratios in that core are anomalously low. Thorium-normalized fluxes also depend on decay correction, as the 230Th-normalized flux of a given sample is inversely related to xs230Th0 (see methods). As such, if the decay correction is incorrect (i.e. if we undercorrect for decay by a large amount), then fluxes will appear to be much larger than is realistic. For example, if a core contains a hiatus and a sample from Marine Isotope Stage 6 (e.g. 140 ka) is mistakenly identified as a 25,000 year old sample, xs230Th0 for this sample will be anomalously low for the (incorrectly) assigned age due to the extra ~1.5 half lives of decay time unaccounted for in the age model. This is turn will result in calculated 230Th- normalized fluxes much greater than are reasonable. We have used detrital fluxes to identify this phenomenon, and to eliminate cores with suspected hiatuses. Fluxes of lithogenic material, rather than opal, were used to verify age models because we found relatively little variability in lithogenic flux over fairly large areas (see section 5). In the paragraphs that follow, we will illustrate the process of rejecting a core based on radionuclide data using core E11-11 (64.843degS, 114.492degW, 4819m, 2.67degS of the APF). Compositional data (Appendix Figure 1) show a peak in MS between 20cm and 35cm, and a corresponding minimum in percent opal (17cm to 37cm). Increased MS often identifies the LGP (see above), and decreased percent opal content can be used to identify the LGP south of the APF, where glacial opal burial is known to have been lower. Percent E. antarctica is increased in this interval, which is another indicator of glacial sediments. Therefore, we assigned the sediments between 30cm and 45cm ages consistent with the LGP (18-28ka), and the sediments at 7 and 12cm ages of 5 and 10ka, respectively. Using this age model, we calculated 230Th-normalized fluxes of opal and detrital material, as well as decay-corrected 231Pa/230Th ratios for all samples (see Appendix Figure 2). Two aspects of the radionuclide data led us to the conclusion that the core contains a hiatus, and that the glacial cycle present is not the last glacial period. First, 231Pa/230Th ratios are extremely low, particularly in the bottom of the core. As discussed in Section 3, this is often a sign of sediment older than the LGP, due to the relatively short half- life of 231Pa (32 800 years). Secondly, detrital fluxes are extremely high (1- 4.5g/cm2/ka), which suggests that the 230Th-normalized fluxes for these samples have been under-corrected for decay- that is to say, the samples are actually much older than the ages assigned to the for the age correction. This evidence led us to eliminate this core from consideration. Burckle, L. H., and D. W. Cooke (1983), Late Pleistocene Eucampia antarctica abundance stratigraphy in the Atlantic sector of the Southern-Ocean 29, 6-10., Micropaleontology, 29, 6-10. Burckle, L. H., and R. W. Burak (1995), Relative abundance of Eucampia antarctica as a close proxy to delta18O in upper Quaternary sediments of the Southern Ocean, in Landscapes and Life. Studies in Honour of Urve Miller, edited by A. M. Robertsson, S. Hicks, A. Akerlund, J. Risberg and T. Hackens, pp. 15- 22, Conseil de l'Europe, Rixensart, Belgium. Chase, Z., R. F. Anderson, M. Q. Fleisher, and P. W. Kubik (2003), Accumulation of biogenic and lithogenic material in the Pacific sector of the Southern Ocean during the past 40,000 years, Deep-Sea Research Part II-Topical Studies in Oceanography, 50(3-4), 799-832. Dezileau, L., J. L. Reyss, and F. Lemoine (2003), Late Quaternary changes in biogenic opal fluxes in the southern Indian Ocean, Marine Geology, 202(3-4), 143-158. Francois, R., M. A. Altabet, E. F. Yu, D. M. Sigman, M. P. Bacon, M. Frank, G. Bohrmann, G. Bareille, and L. D. Labeyrie (1997), Contribution of Southern Ocean surface-water stratification to low atmospheric CO2 concentrations during the last glacial period, Nature, 389(6654), 929-935. Frank, M., R. Gersonde, M. R. van der Loeff, G. Bohrmann, C. C. Nurnberg, P. W. Kubik, M. Suter, and A. Mangini (2000), Similar glacial and interglacial export bioproductivity in the Atlantic sector of the Southern Ocean: Multiproxy evidence and implications for glacial atmospheric CO2, Paleoceanography, 15(6), 642-658. Howard, W. R., and W. L. Prell (1992), Late Quaternary surface circulation of the southern Indian Ocean and its relationship to orbital variations, Paleoceanography, 7, 79-117. Howard, W. R., and W. L. Prell (1994), Late Quaternary CaCO3 production and preservation in the Southern Ocean- Implications for oceanic and atmospheric carbon cycling Paleoceanography, 9(3), 453-482. Kumar, N., R. F. Anderson, R. A. Mortlock, P. N. Froelich, P. Kubik, B. Dittrichhannen, and M. Suter (1995), Increased Biological Productivity and Export Production in the Glacial Southern-Ocean, Nature, 378(6558), 675-680.