Recent and Glacial Age Organic Carbon and Biogenic Silica Accumulation in Marine Sediments

Sedimentation rate data for the late Pleistocene and Holocene was compiled and mapped along with bulk sediment accumulation rate data estimated from the surface calcium carbonate concentration. This data was combined with surface organic carbon and opal (biogenic silica) concentration data in order to calculate the recent rate of acc umulation of these biogenically derived sedimentary components. The maps of organic carbon and opal accumulation rates showed similar trends, being highest in known regions of upwelling and high productivity. Annual organic carbon burial was estimated by multiplying the accumulation rates by the areas between the contours and found to be 0.21xl0 14 gC/yr for the deep-sea (pelagi c and hemipelagic) and 0.04xl0 14 gC/yr for the shelves exclusive of deltaic sediments. Burial of organic carbon in deltaic sediments is very large, h owever the necessary data is not available to calcu late the burial of o r ganic carbon in all the worlds major river deltas. For this reason an estimate of 1.04xl014 gC/yr from Berner (1982) is assumed to be correct for delta deposits, yield ing a global organic carbon burial rate of 1.29xlo14 gC/yr. Organic carbon burial in the most recent Mediterranean sapropel is 0.016xl0 14 and herefore had little or no affect on t he global carbon cycle. Glacial versus interglacial organic carbon accumulation is compared at 10 sites, showing glacial rates higher in areas of present upwelling. Organic carbon ii

Dauphin. Much of the plotting and contou r ing was done by me as was the writing of the initial draft of the text, however the text was heavily revised and the unpublished data used, had been obtained prior to my beginni n g wo rk on that project. Appendix I, while not an integral part of this thesis, provided data and experience in plotting and handling large data sets, used in the preparation of this thesis. Appendix 2 and 3 contain all of the raw data used to generate both t he maps for my thesis a nd the maps of Leinen, et al, 1986. Appe ndix 4 contains the calculated values used to generate the sedimentation rate, bulk sediment, organic carbon and opal accumulation rate maps. vi App endi ces 78

INTRODUCTION
Orga nic carbon is usually a minor constituent of marine sediment s , however its burial plays a very important role in biogeochemical c y cl e s. Marine sediments represent a significant sink for carbon in the marine carbon cycle that has been estimated by many researchers (eg. Garrels and Perry, 1974;Kempe, 1979;Berner, 1982), but has not yet been studied in great detail.
In this study we combine maps of the global distribution of organic carbon in deep sea sediments (Premuzic, et al, 1982;Cwienk and Leinen, 1986) with sedimentation and mass accumulation rat es to make a quantitative estimate of recent and glacial age organic carbon accumulation in marine sediments.
The main sources of organic carbon to marine sediments are terrestrial organic matter transported to the ocean via rivers and photosynthetic production of marine phytoplanktic organic matter. Chemosynthetic production of organic carbo n compounds occurs in hydrothermal vent communities (Corliss, et al, 1979). The net carbon production of these unique areas is probably small a n d will be ignored for the purpos es of this study.
The amount of marine autochth on ous organic carbon fixed in near-surfa ce waters by photos yn t hesis is dependent, in part, o n nutrient availability in surface waters. During respiration a large proportion of this photosynthetic organic carbon is metabolically oxidized and a smaller fraction of it is reformed into different organic carbon compounds. Carbon (C0 2 ) is cycled through this system of photosynthetic fixation and metabolic liberation. A portion of the organic carbon in this cyc l e survives o x idation in the water column and is transport e d to the sediments where most of it is recycled by benthic faunal respiration. A similarly small fraction (ca. 0.5% of the primary production) of the organic carbon reaching the sediments is preserved by burial (Muller and Suess, 1979). Previous studies suggest that on a regional and local basis (e.g. Lisitzin, 1972;Suess, 1980) the rate of organic c a rbon burial reflects primary productivity in spit e of the many processes which diminish carbon fluxes to sediments. Such a relationship between sedimentary organic carbon accumulation and primary producti v ity (photosynthesis) would allow us to make inferences about surface productivity in the past from sediment accumulation, despite the fact that only 0.01% to 20% (typically 0.5%) of the organic carbon pro duced is preserved in the sediments (Muller and Suess, 1979) Whereas river input and primary production (photosynthesis) in surface waters control the amount of organic carbon supplied to the seafloor, many f actors modify the concentration of organic carbon buried in the sediment s.
For example, because o rganic carb o n fluxes may be modified by oxidation in the water column and at the sediment water interface organic carbon distribution can be altered by depth-dependent oxygen concentrations. In areas where the midwater oxy gen-minimum zone intersects the seafloor along continental margins, organic carbon is preferentially preserved in surface sediments (e.g. the Peru margin, Froelich, 1979;Demaison and Moore, 1980). Organic carbon concentrations and accumulation rates may also be influenced by varying burial rates. Higher s e di mentation rates remove organic carbon from the zone o f oxic b enthic metabolism (reduced residence time at the sediment/water interface) The relationship between the rat e of o rganic carbon burial and sedimentation rate is quantitative for specific areas (eg. northeast Pacific, Heath et al, 1977;West Africa, Jansen, et al, 1984). Muller and Suess (1979) have suggested that the relationship between the rate of organic carbon burial and sedimentation rate vari e s in a systematic fashion in all sediments.
Organic carbon di st ribution may also be affected by the grain size and mineralogy of the accomp a nying sediments.
Some organic carbon compounds are sorbed onto clays (Weiss, 1969; Morris and Calvert, 1975), thereby apparently increasing organic carbon preservation in the fine fractio n .
Fine-grained sediments have been found to be higher in organic carbon (Emery and Uchupi, 1972;Bordovski y , 1965; and others) . Of course, this relationship may be the result of depositional energy and higher organic carbon fluxes to the seafloor over the contine ntal slope where finer grained sediments preferentially accumulate. Muller and Suess (1977) suggested that organic carbon preservation is also related to calcium carbonate concentration, due to sorption of dissolved organic matter onto the surface of calcium carbonate grains. Direct evidence for these effects is lacking.
It has been proposed that primary production in many low latitude oceanic regions was higher during glacial periods than during interglacial periods (e.g. Muller, et al, 1983;CLIMAP Project Members, 1976, andBroecker, 1982). Such changes in productivity and the amount of carbon buried would also have affected the co 2 content of the atmosphere. Evidence from Antarctic ice cores suggests that the co 2 content of the atmosphere was approximately 90 ppm lower during the last glacial maximum (18 ka) (Neftel, et al, 1982). It has been shown that increasing the concentration of co 2 in the atmosphere would cause a warming of the atmosphere due to the retention of short wave radiation being back-radiated by the earth (Schneider and Kellogg, 1973). This phenomenon is known as the "greenhouse effect".

Its possible consequences have caused much concern because
Of the progressive increase in atmospheric C0 2 concentrations since the industrial revolution, as the result of the burning of fossil fuels.
Any attempt to study changes in the carbon cycle through time or to assess possible responses to anthropogenic influences (e.g. Garrels and Perry, 1974;Kempe, 1979;Arthur, 1982;etc.) must include an estimate of changes in the rate of removal of organic carbon from the system by sedimentation on the sea floor so that a baseline is available against which to compare estimates for older time periods. Organic carbon accumulati on rates are related to productivity, the rates determined here also provide data with which such a relationship can be quantified for estimation of rates of productivity during other time intervals may be made.
Other approaches to estimating rates of organic carbon burial and its role in the carbon cycle are based on net flux measurements at a few localities, which are calculated from surface carbon flux measured by sediment traps and benthic flux measured by benthic respiration and pore water studies. Such studies are necessarily short term and therefore are influenced by seasonal and short-term events, such as organic floe falls (Honjo, 1982) . Surface flux minus benthic flux equals the burial flux or accumulation rate. The average organic carbon accumulation rate in sediments integrates the burial flux over longer periods of time (a few kyrs.) and mediates the e ffects of seasonality, short-term and localized phenomena. By mapping the distribution of organic carbon accumul at ion on the sea floor, local studies of fluxes may be placed into the context of global or regional trends. I n this way such flux studies can be extrapolated globally without having to study fluxes everywhere in the world for extended periods of time.
Biogenic silica (opal) is often a major constituent of marine sedime nts and its importance has long been recognized. Opal con si sts primarily of the tests of siliceous marine phytoplankton and zooplankton, which are most abundant in zones of high nutrient availability and therefore it is another possible indicator of surface water biological productivity. Its distribution in ancient sediments has been used by many autho r s to infer paleoproductivit y and paleocirculation (e. g . Molina-Cruz, 1976; . Like organic carbon, opal produce d in the surface waters does not reach the seafloor wit h out dissolution.
Opal is subject to dissolution both in the water column and on the seafloor because seawater is everywhere undersaturated with respect to opaline silica (Miskell, et al, 1985) . For deep water masses it is a general rule that older water masses (i.e. ones which have b e en separated from the surface water longer) are less under saturated (Berger, 1970;Edmond, 1974). This is because dissolved silica is added to the bottom waters by progressive dissolution of opal falling from surface waters and a net upward flux of dissolved silica from sediments to deepwater masses. Unlike calcium carbonate, which is subject to different ial dissolution wi th depth, preservation of opal is largely independent of depth.
Opal and organic carbon are similar i n t hat preservation is partially dependent o n sed imentation r a te, however since other factors may af fe ct the preservation of either or both to varying degrees, any major differences in their distribution should distinguish areas in which variations are due to prese r v ation rather the n to actual variations in productivity. A

Organic Carbon
There are several techniques for determining carbon in deep sea sediments. Sediment geochemists analyze marine carbon in three categories: 1) carbonate carbon, which in deep-sea sediments is generally the remains of calcareous organisms, 2) organic carbon, which is t he remains of living tissue, and 3) total carbon, the carbonate plus the organic carbon. Because these three components are related, only two need to be determined, and the third can be calculated.
Some investigators measure both t he organic carbon and the carbonate carbon (Weliky, et al, 1983), while others measur e the total carbon and either organic carbon or carbo nate and find the third value by difference (Kolpack and Bell, 1968).
Carbonate carbon is measured by adding a strong acid such as hydrochloric or phosphoric acid, which dissolves the carbonate, generating carbon dioxide (Weliky, et al, 1983).
The amount of co 2 generated is measured and the results are compared to standards to determine the concentration.
Measurements of carbonate c a rbon using a carbonate bomb have an accuracy of ca. ±5% (Dunn, 1980), while measurements made with a coulometer a re d e pendent on the accurac y of the sample weight, (Engleman, et al, 1985) .
Organic carbon is g e nerally measured by heating the sample to extremely high temperatures and measuring either the weight loss (for samples with high concentrations) or measuring the co 2 generated by the oxidation of the organic carbon (for samples with low organic carbon concentrations, e.g. LECO, CHN and coulometry techniques) (Gibbs, 1977). Weliky, et al (1983) used a slightly different technique for measuring organic carbon, which c on si sted of measuring the co 2 generated by dichromate oxidati on of t he sample. Carbon of various types is usually measured by instruments such as the LECO Carbon Analyzer or automated CHN (carbon-hydrogennitrogen) analyzer, which determine carbon by combusting the sample at high temperatures with purified o 2 and measuring the C0 2 generated (Kolpack and Bell, 1968). The determinations of organic carbon concentration in the literature generally have an accuracy of ±0.02 weight percent. High precision and accuracy are required in partiti oning total carbon into carbonate and organic carbon because organic carbon is found in concentrations of only 5 % to <0.1% in open ocean sediments (Heath et al, 1977).
Concentrations are somewhat higher on the continental shelves and slopes (Premuzic, et al, 1982) .
There are some problems with the analyti ca l techniques f o r carbon, which cause inaccuracies in the measurements.
When a sample is a c idified in order to measure the carbonat e content, some of the volatile/soluble portion of the organi c carbon can be dissolved by the acid and measured with the carbonate. Similarly when a sample is ashed at high temperatures in order to measure the organic carbon, some of the carbonate c arbon is oxidized by the extreme heat (Lyle, written communication) . Attempts to reduce the error due to these effects by using a weaker acid or lower temperature may result in incomplete dissolution or oxidation of the desired form of carbon (Froelich, 1980;Weliky et al, 1983).
For these reasons the measured organic c arbon and carbonat e carbon rarely add up to 100% of the total carbon when all three are measured. Froelich (1980) attempted to solve this problem by acidifying samples to remove the c a rbonate, then analyzing the insoluble residue with a CHN analyzer. The acid-dissolved solution (filtrate) wa s also analyzed using a modified diss ol ved organic c a rbon (DOC) method (Menzel and Vaccaro, 196 4 ;Kerr, 1977 where they were acidifed with phosphoric acid. The co 2 gen~rated was then automatically titrated by the coulometer to provide the percentage of calcium carbonate. A second split was placed in an oven at 980°c where the co 2 generated was swept by a constant flow of ultrapure o 2 into the same automated titrator to determine the total carbon (Huffman, 1977). Standards of pure calcium carbonate were run to ensure that all carbonate carbon was being dissolved or combusted during each of the analyses. The organic carbon was calculated by subtract ion.

Biogenic Silica
Opal concentrations used in this study a r e from the compilation of Leinen and others (1986). Most of the opal values reported were determined by x-ray diffractometry using alumina as an internal standard (Calvert, 1966;Ellis and Moore, 1973). Although the absolute values indicated by opal determinations have been suspect, recent work (Leinen and King, 1981a, b;Leinen, 1985) shows that the relative abundances indicated by x-ray diff ra ctometry are valid. Leinen and others (1986) also used several techniques to intercalibrate the x-ray diffraction data with other data sets, including 1) comparison of samples analyzed by more then one author, 2) normative partitioning of sediment geochemistry, and 3) standard additions.
Their results indicate that the accuracy of uncalibrated x-ray opal determinations is <±10 wt%.

Sedimentation Rates
Sedimentation rates were calculated by dividing the depth to a given stratigraphic marker (in centimeters) by the age of that marker (in thousands of years) . This yielded the sedimentation rate in centimeters per thousands of years.
In the absence of data to the contrary, it was assumed that sedimentation was continuous during the interval between the surface and the stratigraphic datums.
The most reliable stratigraphic markers for determining Holocene and glacial sedimentation rates are the last glacial maximum (at 18 ka)and the end of the last glacial (at 12 ka) as determined from the oxygen isotopic composition of planktonic or benthonic foraminifera (Imbrie, 1 1 14 et al, 1984), and Cage determinations. Many cores were available for which oxygen isotope stratigraphies had been determi n ed by CLIMAP, a study o f the Holocene and glacial global ocean (Moore, et al, 1980;Prell, et al, 1980;CLIMAP Project Members, 1976) . Because this study was designed to investigate the accumulation rates of organic carbon f or t h e Holocene (12 ka to present) and for the last glacial maximum, stratigraphies based on oxygen isotope determinations or carbon-14 ages ar e preferable because no pre-stage 2 sediment a tion i s included in the rate. The oxygen i s otopi c strati g rap hy is v e ry accurate and allows resolution of the desired i n terval (Prell, e t al, 1986). In s ome cas e s t h e ox y gen is ot opic stratigraphy is det a iled enough to identify t h e e n d of the last glacial at 12 ka.
In this case both glacial and interglacial accumulation rates can be determined. In other cases the resolution of the oxygen isotopic stratigraphy was not sufficient to identify the end of the last glacial periods accurately. In this case, if there have been changes in sedimentation rate since the last glacial maximum the differences will be averaged. Average sedimentation rates determined using only the Bruhnes/Matuyama datum represent longer averages and include possible changes that occurred over a number of glacials and interglacials (i.e. average late Pleistocene sedimentation) . Carbon-14 dates based on organic carbon provide absolute ages, however the accuracy of these ages i s altered by reworking and redeposition of the carbon and to some extent by diagenesis (Erlenkeuser, 1980). Most C-14 ages are too old by 1 to 2 kyrs. Sedimentation rates based on ash layer dates are precise and the ages of many of the ash layer datums are very accurate, but in general they are only useful in small areas.

Accumulation Rates
The bulk sediment accumulation rate is the mass burial flux of sediments. It is calculated by multiplying the dry bulk density (in g/cm 3 ) and the sedimentation rate. When the bulk accumulation rate is multiplied by the where, Bd is the uncompressed dry bulk density, Pt is the wet bulk density, X is the water content, a nd C is the w calcium carbonate concentration. Whereas e stimates based on this technique appear to be reasonable in t h e regional examples cited by Lyle and Dymond, the technique is l e ss reliable for estimating downco r e variations in bulk density within a single core (Curry, 1986). Such uncertainty introduces some error into the accumulation rate calculations, but this error is gen e rally smaller than errors in estimating absolute ages for various levels in the cores. Kominz and others (1977)   ) was used to generate the opal accumulat ion rate map. A map of bulk density was not generated because in most areas the bulk density was estimated from the Caco 3 concentration as discussed above.
More complete maps of Caco 3 concentration in surfa c e sediments than could be drawn from this data set have been published (Lisitzin, 1972 .

SEDIMENTATION RATE MAP
One of the most important controls on the thickness of sediment accumulating on the sea-floor per unit time ( (Fig. 2). The e olian or windblown input in areas beneath major wind be lt s, such as the westerlies belt east of Japan, which has desert sources in southeast Asia or the eastern North Atlantic off Northwest Africa, which has sources in the desert Sahara, will a l so increase the sedimentation rate (e.g., Fig. 4 Figure 5 is the map of bulk sediment accumulation rates generated from the bulk densities estimated from Caco 3 concentration using the equation of Lyle and Dymond (1976) and sedimentation rates. Measured bulk densities were available for a small number of sites in the South Pacific.

BULK SEDIMENT ACCUMULATION RATE
For many cores sedimentation rate data were available but Caco 3 concentrations were not. Because there were detailed maps of Caco 3 available , Biscaye, e t al, 1976) and we had Cac o 3 data from cores near the sedimentation rate d a ta, we e st imated the caco 3 concentration for cores for which such data were not available. The error introduced by such a n estimation is at most about 20% which is small compared to the variation in sedimentation rate. The map of bulk sediment accumulation rate (Fig.5) shows the same general features as the sedimentation rate map (Fig.1). This confirms that bulk density, does not vary greatly compared to sedimentation r at e and that the l a tt e r is the dominant control on mass accumulation rates.

ORGANIC CARBON ACCUMULATION
Organic carbon accumulation rates ( Fig. 6) are highest in the equatorial Pacific, along the coasts (particularly the west coast of the United States), south of the Aleutian Islands and in a region extending east of Japan. Along the coasts, the organic carbon accumulation is high as a result of high primary and secondary biological production due to the input of nutrients and as a result of the accumulation of terrestrial organic carbon from the land. The organic carbon accumulation rate is high in the equatorial Pacific because of high primary production in the zone of upwelling (Koblenz-Mishke, et al, 1970). The lobe extending eastward from Japan is a reflection of the high productivity in that area due to upwelling at the confluence of the Kuroshio and Oyashio currents (Koblenz-Mishke, et al, 1970). The high values south of the Aleutian Islands is due to the cumulative effects of slightly higher productivity and high burial rates of terrigenous carbon in turbidites (Koblenz-Mishke, et al, 1970

OPAL ACCUMULATION RATES
The opal accumulation rate map (Fig.7) shows highest accumulation rates in the equato r ial Pacific, east of Japan, off the northwest coast of Africa, around Antarctica, in the Caribbean and south of India. Aside from the Caribbean which is rather shallow, all of the r e gions mentioned are known areas of strong mixing and/or upwelling.
In most of the areas where organic carbon is accumulating rapidly, opal is also accumulating rapidly (see Fig. 8). This suggests that early work calling attention to the relationship between organic ca rb on accumulation, opal accumulation and primary productivity (Lisitzin, 1972;Heath, 1974) over broad areas, is reasonable, although this relationship may not hold for individual sites. This agreement can be seen between a map of estimated primary Productivity (Fig.8) generated by Berger and others (1986) for the Pacific Ocean and the organic carbon accumulation rate for that oce a n basin (Fig.6).

GLOBAL ORGANI C CARBON DEPOSITION
one of the major objectives of this study was to determine the total amount of organic ca r bon being deposite d in sediments during the Holocene . For the Pacific Ocean, we estimated the rate of organic carbon deposition by multiplying the areas between co n tour s of o r g a nic carbon accumulation by the mean value within that area (Table 1) This procedure could not be used for the Atlantic and Indian Oceans because of the poor data coverage for organic carbon contents. Atl a ntic and Indian Oc ean sedimentation is controlled to a large degree by the bottom topography (e.g. province in order to determine its average organic carbon accumulation rate (Fig. 10). The organic ca r bon burial rate was then determined as for that in the P ac i fic Ocean (Table 1). in which, P is productivity in gC / m 2 /yr, C is organic carbon concentration in weight percent, SB is the sedimentation rate in cm/kyr, ..P(l-~) is the dry bulk de n s ity in g/cm 3 , SB-C is the sedimentation rate for the organic carbon-free fraction and Z is the water depth in meters.
Average productivity for the Antarcti c was estimated at 85 gC/m 2 /yr based on Berger and others (1986) (Table   2) . We gave no special considerat ion to deltaic sediments in this study. Berner (1982) determined orga n ic carbon burial in deltaic d epo sit s by multiply ing the suspended sediment transport from rivers by the average organic carbo n content of the deltaic sediments. Arthur and others (1985) arrived at a global organic carbon burial value compar a b le to Berner's (Table 2) by balancing the carbon reservoirs with the global & 13 c. Mopper and Degens (1979) Strick, et al, 1982). The increased vertical stabil i ty, possibly coupled with higher than norma l organic carbon Production (Jenkins and Williams, 1984) caused rapid depletion of oxyge n i n bottom waters a n d enhanced preservation of o r g a nic matter to form the organic-rich sediments (C i ta, e t al, 1973). Because the r a pid depos iti on Of organic-rich layers over a large a rea cou l d have a significant effect on the global marine orga n i c carbon budget we have estimated the total organic carbon burial resulting from the deposition of this sapropel. Stanley (1978) showed that the deposition of sapropelic sediments occurred over much of the Mediterranean, but probably no more than 1.5 million square kilometers.
Sapoprels usually represent a period of 2000 years and have a rate of deposition of 2 cm/kyr (Dominik and Mangini, 1979). Sapropels have an organic carbon concentration of 2 to 7% (Anastasakis and Stanley, 1984

GLACIAL vs INTERGLACIAL ORGANIC CARBON ACCUMULATION
It has been proposed by many authors, that primary Productivity in the surface oceans was higher during glacials then interglacials (Muller et al, 1983;CLIMAP Project Members, 1976). In order to es timate the difference in glacial versus interglacial organic carbon deposition, we determined rates for all CLIMAP cores which had oxygen isotopic stratigraphies and organic carbon analyses of glacial and interglacial sediments. There are only 10 such cores in the Pacific Ocean. The depth to the glacial/interglacial transition (at 12ka) was picked from oxygen or carbon isotope data, thereby allowing the determination of the glacial (18 to 12ka) and interglacial (12ka to present) organic carbon ac c umulation rates ( Table   3) • Glacial organic carbon rates were higher in 6 of the 10 cores ( Fig 11). These data are inadequate to support or refute the proposal that organic carbon accumulation was higher during glacials, as it would be expected to be if primary productivity had been high-particularly 2 to 3 time s higher as proposed by Muller and ot hers (1983). The data d o suggest that glacial primary product ion is higher in areas in which upwelling is presently strong, while it is lower in other areas (Fig. 11). This would agree with Mull e r and others (1983), who found that glacial pr i mary production increased significant ly in areas of pre se nt upwe ll ing.

ESTIMATING ORGANIC CARBON ACCUMULATION
It has been proposed (Heath et al, 1 977) that t here is a correlation betwe en o r gan ic carbon a c cumulation and sedimentation rate (CA=O.Ol*Sed 1 · 4 ), which o ccurs because high sedimentation rates s equester organic carbon from degradation at the sediment surface. Using all of the data Points for which organic carbon accumulation rates were calculated (Pacific Ocean), the sedimentation rat e versus the organic carbon accumulation rate was plotted ( Fig. 12) These two variabl es are well correlated (r 2 =0.81). Muller and Suess (1979) refined this idea by including surface water productivity as an additio n al variable controlling organic carbon accumulation (CA/PP*100=Sed 1 · 30 ).
In order to test Muller and Suess's (1979) hypothesis, the same data points were plotted but with the organic carbon accumulation rate divided by the primary productivity, as estimated from the Berger and others (1986) primary production map (Fig. 13). The correlation of these points is relatively good (r 2 =0.66). The data were also divided into regions to determine whether the correlation is better within regions of similar sedimentation (the Australian/New Zealand high produ ctivit y region, the Kuroshio/Oyashio confluence region, the Peru/Equatorial high productivity region, the Pacific Coast of North America shelf region, and the Central Gyral regions) . The c o rrelation within each of the regions was significantly worse than that for all of the data comb i n e d (r 2 between 0.32 and 0.11) except the North American Shelf region (r 2 =0.67). Our inability to estimate the primary productivity precisely enough may have caused sufficient error to mask the correlation in the regional sets (due to a lower number of observations) while the broader trend was still apparent in the larger data set.

Figure 9:
D~stribution of primary production in surface waters (gC/m /yr) (Berger, et al, 1986). Equal area pr~jection of organic carbon accumulation rates (mg/cm /ky). Regions without data points shown have values estimated from the average of the available data within that region.    Organic carbon accumulation rate data divided b y primary productivity above that site plotted against the sedimentation rate data at that site for all cores containing both in the Pacific Ocean.   Figure 14:

SYNTHETIC PRIMARY PRODUCTIVITY
Regional annual organic carbon accumulation (gC/yr), exclusive of deltaic sediments.

INTRODUCTION
Although quartz is often a small component of deep-sea sediments, its abundance and distribution pattern can be very useful for the interpretation of sediment source areas, sedimentation processes and paleoclimate. The siliceous remains of plankton are generally a major component of deep-sea sediments; the abundance and distribution of this biogenic silica (opal) have been used to infer the paleoproductivity and paleoci rcu lation of the oceans (e.g. Molina-Cruz, 1977;Pisias, 1979;. While detailed maps of the other major biological component of sediments, calcium carbonate, have been published (Berger, 1976;Biscaye ruL, 1976), world maps of similar detail for quartz or opal, which have been analyzed by many investigators over the years, have not been published. We have compiled all of the published quartz and opal data determined by the X-ray diffraction technique and have includc::d all of o ur unpub li shed ::malyses as well.

QC ARTZ
Since 1955 a great many quantitative analyses of the quartz content of deep-sea sediments have been made by X-ray diffractometry. After Till and Spears ( 1969) refined the X-ray technique for quartz determination using alumina as an internal standard, it was used routinely for sediment analyses by many oceanographic laboratories. Detaikd maps of the quartz distribution in the Atlantic Ocean (Kolla, ~. 1979), Indian Ocean (Kolla and Biscaye, 1977) and North Pacific Ocean (Moore and Heath, 1978;Heath, tl..l!L., 1983) are available, but maps of similar data density for the entire Pacific have not been made, although many data are available. In addition, the various data sets have not been intercalibrated.
Quartz is stable at ocean bottom conditions and does not form authigenically in recent sediments. Therefore its source is, with rare exception (Peterson and Goldberg, 1982), continental. Being resistant to abrasion and dissolution, quartz arrives at the sea floor in much the same condition that it reached the sea surface . Most quartz in pelagic sediments occurs as chips and shards and although the grain size varies with its origin and transport mechanism, most grains are in the 5-10 µm range (Rex and Goldberg , 1958). The flux of fluvial and hemipelagic sediments decreases rapidly with distance from land. Such sediments are not sources of quartz far from the continents.
For the e reasons, quartz in pelagic sediment far from land and in locations protected from turbidity current deposition has been inferred to be eolian (Rex and Goldberg, 1958).
Quartz is common in the mineral aerosol transported by wind (Prospero and 8onatti, 1969;Prospero and Carlson, 1972;Blank, tl..l!L.. 1985 ). Recent studies ha ve shown that the proportion of quartz in aerosols matches that in sediments accumulating beneath the aerosol collection site (e.g. Blank, tl..al.., 1985 ) , furt her supporting its inferred eolian ori gin in pelagic sediments.

\tfethods
We collected all available published and unpublished X-ray diffraction determinations of quartz. Because the X-ray diffraction technique for quartz and opal analysis requires calcium carbonate removal, all data are on a calcium carbonate-free bas is and do not reflect influence by this important diluent. (The data sources are listed in Table 1; core identifications, locations, quartz, opal, and carbonate concentrations are listed in Appendix 1* .) A few areas have some overlap in data and one, the South Atlantic , has been studied in detail by two sets of investigators (Ellis and Moore, 1973;Kolla il.a1. 1979). The two independent studies showed the same distribution pattern.
The analysis of several samples by both sets of investigators allowed us to intercalibrate the quartz analyses done at Lamont-Doherty Geological Observatory (Biscaye, Kolla) with those done at Oregon State University (Ellis, Heath, Molina-Cruz, Thiede, Dauphin ) and the University of Rhode Island (Leinen). A comparison of the samples that had been analyzed by both Kolla , tl1!L. ( 1979) and Ellis and Moore (1973) indicated that the Ellis and Moore values were systematically greater by 5.8 wt% quartz (r2 = 0.68). The Ellis and Moore samples were calibrated to a standard curve of varying percents of quartz in a clay matrix and the details of the conversion factors were reported in Ellis ( 1972). This curve was used fo r all subsequent Oregon State University analyses . The exact factors for • Note to Reviewers: We are including the data Appendix in the manuscript sent out for review . We will request that the appendix not be published , but be included in a data archive that is available by mail request from the Society.
converting peak area ratios to weight percent quartz were not given by Kolla tl.i!L ( 1979 ), an d they did not publish a standard curve . As a result, we chose co increase the Kolla tlilL ( 1979) quartz values for the Atlantic Ocean and the Kolla and Biscaye (1977) quartz val ue s for the Indian Ocean by 5.8 wt% to match the Oregon State data. The University of Rhode Island data were also intercalibrated with the Oregon State University data . The calibrated data were plotted and contoured.

Distribution
The distribution of quartz in pelagic sediments of eolian origin should reflect dominant wind systems and major arid regions (Griffin, .e..t..a1, 1968 ;Kolla and Biscaye, 1977;Moore and Heath, 1978 ;Kolla ~. 1979;Thiede, 1979). If quartz particles settled to the seafloor by Stokesian settling, the fine grain size of the material would lead to slow settling rates. This action by currents would smear or obliterate any pattern of distribution by wind transport. Such is not the case; quartz distributions reflect wind patterns in many regions. Sediment trap research suggests that filter-feeding plankton concentrate inorganic particles from the surface water into fecal pellets causing them to sink rapidly (Honjo, ~ 1982). Large organic aggregates ("marine snow") also increase the sinking rate (Honjo, tl1!.L 1982).
There is a strong latitudinal pattern of quartz distribution in the North Pacific ( Fig. l) that .has been related to the mean position of the westerly wind system over this region (Rex and Goldberg, 1958;Griffin,.-e.Llll.., 1968;Moore and Heath , 1978). Numerous sediment and aerosol studies have demonstrated that this sediment is derived from Asia (Rex and Goldberg, 1958;Griffin, ~. 1968;Windom, 1969;Duce. ~. 1980Duce. ~. , 1983Shaw, 1980;Parrington, 1981). The South Pacific has low quartz concentrations reflecting smaller input from continental sediment sources (Thiede, 1979;Dymond, 1981;Schramm and Leinen, in press;Bloomstine and Rea, in press). Although the influence of Australia as a source region for quartz in the southwest Pacific has been documented by Th iede (1979), this material does not extend across the entire South Pacific .
The Sahara is an important source of eolian sediment in the eastern Atlantic (Kolla and Biscaye, 1977;Sarnthein, 1979). Its relative influence on deep-sea sedimentation drops off markedly to the west, however, and does not extend across the sub-equatorial Atlantic. In the equatorial Atlantic Ocean there is some suggestion of latitudinal banding, but the quartz distribution has been strongly modified by turbidite deposition and bottom processes which are discussed in detail by Kolla, tl..aL ( 1979).
Quartz is diluted by opal in areas of strong upwelling and along the equatorial divergences. This is very apparent in the eastern equatorial Pacific where carbonate-free sediments are dominated by biogenic silica (Molina-Cruz and Price, 1977, Heath,~.

1983).
OPAL Biogenic silica (opal) in deep-sea sediments is dominated by the remains of marine plankton, and is closely related to surface productivity Molina-Cruz; because siliceous sediments are not affected by differential dissolution with depth, as are calcareous sediments. Earlier maps of opal distribution in pelagic sediments (Lisitzin, 1972;Heath , ~. 1983) focussed on the Antarctic and the North Pacific .
The remainder of the world ocean was highly generalized. Unfortunately, no data points were shown on the Lisitzin map and the data have not been published. The importance of biogenic silica ·has long been recognized, but its surface distribution has not be mapped in detail previously because of the great difficulty of analyzing opal quantitatively in deepsea sediments (Leinen, 1977;Eggiman, tl..al. 1980;DeMaster, 1980).

Methods
Opal can be determined by X-ray diffractometry using alumina as an internal standard (Calvert, 1966;Ellis and Moore, 1973), and in fact most of the investigators who have analyzed quartz in sediments also detennined the opal contents of their samples by the X-r:.iy diffraction technique (e.g. Molina-Cruz and Price, 1977). Most of these region:il data sets .have never been published, however, because the absolute value of the opal concentrations were suspect. Recent work (Leinen and King, 1981;Leinen, in press) suggests that the relative abundances indicated by the X-ray diffraction analyses are valid. After compiling and contouring the opal data for this study (all data are listed in Appendix I), it was clear that the regional patterns of opal distribution detennined by X-ray diffractometry did indeed make sense oceanographically, but needed calibration.
X-ray opal values were calibrated using data from other techniques. These calibrations yielded results that were very consistent for individual deep-sea regions.
Data for the equatorial Pacific were calibrated by comparing opal values detennined by Heath using X-ray diffractometry with those determined by normative p:initioning of sediment geochemistry (Leinen, 1977). Several subtropical South Pacific samples analyzed by Molina-Cruz and Price (1977) also had bulk sediment geochemical analyses.
Data for this area were calibrated by comparison with estimates of opal content from geochemical partitioning (Dymond, 1981;_ Central North and South Pacific data (Moore and Heath, 1978) and Indian Ocean data (Kolla and Biscaye, 1977;Moll, et al.. in press) were calibrated by the standard additions technique of Leinen and King ( 1981). Northwest Pacific data were also calibrated by the standard additions technique and by comparison with quantitative microfossil counts (Leinen, in press) . Finally, the overlapping analyses of South Atlantic samples (Kolla, tl.fil., 1979; Ellis and \foore , 1973) allowed a direct comparison. Because the Ellis and Moore ( 1973) data were calibrated using a standard curve and microfossil counts, they were accepted and the Kolla, ~ ( 1979) data were multiplied by a factor of 1.2 to bring them into agreement with the Ellis and Moore data. We did have to adjust the opal values by up to ±5 wt% in areas of overlap between regions calibrated by different techniques.
Because the concentration of opal in the sediments was generally large, this adjustment intrcx:iuced an error of <10% of measured value. At the 10 wt% contou r ·nterval used on the map, suc h error would move contours to one side or the other of a point, but would not affect the pattern of the contours. We chose not to include the contours published by Lisitzin ( 1972) because we could not obtain the core locations or calibrate the data to our x-ray opal analyses.

Distribution
Although the opal data are not ideal and there is certainly room for improvement in the calibration of different data sets , the distribution pattern of the opal data and the gocx:i agreement between the X-ray diffraction data and other indicators of biogenic silica concentration (such as rnicrofossil counts and geochemical partitioning) suggest that the data reflect real differences in opal concentration in deep-sea sediments ( Windom, 1969;Leinen and Heath, 1981 ). The distribution pattern in the North Pacific reflects both eolian transport by the major wind systems and differences in the quartz content of terrigenous sources. The latitudinal patterns are much clearer in the opal-free data. In the North Pacific the opal-free data also show a lobe of higher quartz concentrations in the eastern North Pacific at about 20°N which has been attributed to eolian transport from the arid desert regions of Central America and southwestern North America (Moore and Heath, 1978). The map suggests that eolian material transported across the North Pacific from Asia by the westerlies has a similar quartz content to that supplied to subtropical northeast Pacific sediments from the desert regions of North America by the trade winds. The concentration of quartz (calculated on an opal-free basis) also is influenced by the weathering style on the continent. For example, deep-sea sediments contain high concentrations of quartz downwind of major deserts like the Sahara (Sarnthein, 1979;Dauphin, 1982).
Eolian processes are not the only processes controlling quartz distribution in the ocean, however, and in some areas they are not the dominant processes. In the Atlantic Ocean :rny eolian pattern of latitudinal banding is strongly mcxiified by turbidite depos ition on abyssal plains and by contour current winnowing. It is obvious from our results th at great care must be taken when choosing where quartz determinations can be used as reliable indicators of eolian transport.
The gross pattern of opal distribution mirrors the surface primary prcxiuctivity (Koblentz-Mishke, e.LaL., 1970; Heath, .e..t.Ji.L., 1983) even in areas like the equatorial and nonhwest Pac ific, where the siliceous sediment is dominated by the skeletal remains of radiolari;..ms wh ic h are not primary producers. The similarity between primary productivity and opal distribution is strongest in the Pacific where sediments are dominantly pelagic. Absolute concentrations must be used with care because of problems in calibrating opal techniques discussed earlier, however the map clearly reflects firstorder features like the equatorial high productivity belts and some second-order features like the productivity associated with the confluence of the Kuroshio and Oyashio currents in the northwest Pacific.

ACKNOWLEDGE1\1ENTS
We would like to thank the many technicians and students who contributed to the data sets which are combined in this paper.  ( 1977) Kolla, tL.lL ( 1979) Thiede ( 1979) Ellis (

Appen dix II: Core Information and Concentration of Sedimentary Components
The following is a list of the table headings and their definitions: and -is west) . LAT = latitude (+ is north and -is south). CORE ID = the standard core designations (explained on the following page) . ACCESS = sample n umber used by analyzing laboratory. DEPTH = water depth above core site (in meters) . SAMP DEPTH = dept h in core from which sample was taken (in centimeters) .
QTZ % = quartz concentration (in carbonate-free sampl e ) OPAL % = opal concentration (in carbonate-free sample) CAC03 % = calcium carbonate concentration. SOURCE = data source listed on a following page.

Key for Core Identification Codes:
Cores are identified by standard codes used in the literature or the code used by the author from which the data was obtained.