Ratio of coccolith CaCO3 to foraminifera CaCO3 in late Holocene deep sea sediments



[1] On the basis of measurements of the relative amounts of CaCO3 in the less than 20-μm and the greater than 20-μm size fractions in open ocean core tops, we find that the coccoliths contribute about half the calcite present in late Holocene deep sea sediments which have experienced little or no dissolution. Although this ratio is of importance to the understanding of the ocean's CaCO3 cycle, we can find only a few quantitative estimates of their relative contribution to currently forming marine sediments. As dissolution of foraminifera calcite takes place more rapidly than that of coccolith calcite, coccoliths dominate the CaCO3 in sediments which have experienced sizable dissolution. Although coccoliths contribute 40–60 of the CaCO3 in tropical sediments, higher-latitude sediments and those adjacent to continental margins often have larger proportions of coccolith CaCO3.

1. Introduction

[2] Although the literature is filled with papers dealing with foraminifera and also with those dealing with coccoliths, surprisingly, references to the relative contribution of these two entities to the CaCO3 present in deep sea sediments are rare. A paper by Frenz et al. [2005] stands out. Using a combination of grain size distribution and CaCO3 content, these authors analyzed 72 core top samples from the South Atlantic's mid-ocean ridge (equator to 47°S). For samples from less than 4 km depth, where CaCO3 dissolution is minimal, they found that CaCO3 in sand-sized fraction (mainly foraminifera) exceeded that in the silt-sized fraction (mainly coccoliths) by a factor of 1.3 ± 0.3. Research conducted by Tzu-Chien Chiu while a postdoctoral fellow at Lamont-Doherty created a means to improve this situation. She demonstrated that a clean separation between these two calcite entities can be made using a 20 μm sieve. The calcite which passes through the sieve is dominated by coccoliths and that caught by the sieve is dominated by foraminifera shells. However, as she did not exploit her separation method to survey the relative abundances of these two types of CaCO3 in sediment core tops, we present here a first-order global survey of foraminifera and coccolith abundance in Holocene sediment using two approaches.

2. Estimates Based on Size Index Measurements

[3] It turns out that our previously published size index results [Broecker and Clark, 1999] for low-latitude cores can be used for this purpose. They are based on the fraction of the CaCO3 in the greater than 63 μm fraction (or more precisely, as we observed that the 63 μm fraction was uniformly rich in CaCO3, on the total weight of this fraction rather than on the actual weight of its CaCO3). As shown by Chiu and Broecker [2008] in an Ontong Java Plateau core top from 2.3 km water depth, 23% of the foraminifera CaCO3 is in the 20–38 μm fraction and in a Ceara Rise core top from 3.3 km water depth, 27% of the CaCO3 is in this size range (see Table 1). So, in order to convert the Broecker and Clark size index results to percent foraminifera calcite, this contribution to the foraminifera CaCO3 has to be taken into account. Also, a correction has to be made for the non CaCO3 contribution to the greater than 63 μm weight. As summarized in Figure 1 for the above samples 12 and 8 units, respectively, must be subtracted from the size index to get the percent foraminifera CaCO3. On the basis of these two measurements, we adopt a 10 unit conversion factor. While this conversion has an error of approximately ±6 units, nevertheless the corrected size index results provide useful information. As can be seen in Figure 2, nearly all of the equatorial core top results lie in the range 40–60% coccolith calcite with a median value of 47%.

Figure 1.

The factor used to convert our published size index results [Broecker and Clark, 1999] to percent coccolith CaCO3 is based on two analyses made by Chiu and Broecker [2008]. It involves adding in the amount of foraminifera CaCO3 in the size range 20–63 μm and taking into account the non-CaCO3 contribution to the weight of the greater than 63 μm fraction. The uncertainty in this 10-unit correction factor is approximately ±6 units.

Figure 2.

Summary of the percent coccolith CaCO3 results based on the corrected size index equatorial core top results published by Broecker and Clark [1999]. Each box represents an individual sample. The uncertainty in each result is on the order of 6%.

Table 1. Comparison Between the Results on Holocene Samples From a Shallow and a Deep Core From the Equator on the Ontong Java Plateaua
 Bulk CaCO3 Content (%)Foraminifera CaCO3 Content (%)Coccolith CaCO3 Content (%)Foraminifera Coccolith20–38 μm CaCO3/>20 μm CaCO3
  • a

    Data are from Chiu and Broecker [2008]. On the basis of radiocarbon ages, the accumulation rate for the deep core is roughly half that for the shallow core. This difference is mainly the result of preferential dissolution of the foraminifera CaCO3. Although this dissolution must result in shell fragmentation, it does not significantly alter the fraction of foraminifera CaCO3 in the 20–38 μm size range.

2.31 km Core (MW91-9-GGC15)
Late Holocene8757292.00.23
4.04 km Core (MW91-9-BC56)
Late Holocene7827510.530.24
Early Holocene8235470.740.21

[4] As summarized in Figure 3, on the Ontong Java Plateau the Broecker and Clark [1999] size index decreases linearly with water depth and with decreasing pressure-corrected carbonate ion concentration in all three equatorial oceans. Rather than being the result of fragmentation as originally suggested by Broecker and Clark [1999], Chiu and Broecker [2008] show that it is rather the result of preferential dissolution of foraminifera calcite. On the Ontong Java Plateau, foraminifera calcite dissolves an order of magnitude more rapidly than coccolith calcite. Although dissolution results in breakup of foraminifera shells, even in moderately dissolved sediments none of the foraminifers' fragments are smaller than 20 μm. Further, as shown in Table 1 in two Holocene samples from a core from 4.04 km water depth on the Ontong Java Plateau [Chiu and Broecker, 2008], the split between the amount of foraminifera CaCO3 in the 20–63 μm fraction and that in the greater than 63 μm fraction does not change significantly as dissolution proceeds. In one sample from the 4-km depth core with a size index of 26 units, 24% of the greater than 20 μm CaCO3 was in the 20–63 μm fraction and in the other with a size index of 34 units, 21% of the greater than 20 μm fraction CaCO3 was in the 20–63 μm fraction. Although dissolution must result in fragmentation, it does not significantly alter the split between the amount of CaCO3 in the 20–63 μm size fraction and that in the greater than 63 μm size fraction.

Figure 3.

(a) Summary of size fraction results on core top samples as a function of water depth on the Ontong Java Plateau. (b) CaCO3 content of the >63 μm fraction as a function of the size fraction. (c) Size fraction as a function of pressure-normalized carbonate ion concentration for tropical Indian Ocean and Pacific Ocean core tops. (d) Same for tropical Atlantic core tops. These results are reproduced from a paper by Broecker and Clark [1999], who conclude that, for sediments experiencing little dissolution, the size index is approximately 55 units. On the basis of the conversion factor of nine units given in Figure 1, this is equivalent to 47% coccolith calcite.

3. New Measurements

[5] Our new measurements are on one gram samples from the Lamont-Doherty sediment core collection. The samples were taken from the upper 8 cm of both piston cores and their companion trigger weight cores. In order to minimize the impact of dissolution, which preferentially removes foraminifera CaCO3, we have restricted our survey to sediment cores from water depths shallower than 3.5 km. While this restriction is adequate in the Atlantic Ocean (where significant CaCO3 dissolution is observed only below about 4 km), it is suspect in the Pacific and Indian Oceans (where dissolution commences at depths in the range 2.8–3.5 km). Unfortunately, as Pacific and Indian open ocean cores from shallower than 2.8 km are rare in the Lamont collection, we examined cores as deep as 3.4 km.

[6] An aliquot of the bulk sediment was analyzed for CaCO3 using a coulometer. The remainder of the sample was then weighed and its greater than 20 μm fraction was isolated by sieving. This fraction was then dried and weighed. Its CaCO3 content was determined by a coulometer with an accuracy of about 1%. The weight of CaCO3 in the less than 20 μm fraction was then calculated by subtracting the weight of CaCO3 in the greater than 20 μm fraction from that in the bulk sample.

[7] This method is less labor intensive than that employed by Frenz et al. [2005], who conducted a grain size analysis of the 2–63 μm size fraction using a SediGraph 5100 unit. The minimum observed at 8 μm was selected as the boundary between the coccolith and foraminifera fractions. Further, their CaCO3 measurements were based on weight loss as the result of acidification. Our method involving a 20 μm sieve and a coulometer is certainly faster and we feel equally accurate. However, if there is interest in examining the specifics of the separated coccolith-rich fraction and of the clay-rich fraction, as was the case in the Frenz et al. study, then our method is inadequate.

[8] We did not conduct electron microscope checks on any of these samples. Rather, we assume that the clean separation between coccoliths and foraminifera found by Chiu and Broecker [2008] for Ontong Java Plateau and Ceara Rise samples applies to those from other parts of the ocean. Electron microscope photographs of the less than 20 μm fraction from both well-preserved and highly dissolved samples from Ontong Java Plateau cores showed that the CaCO3 was entirely from coccoliths. No foraminifera fragments were present. We did, however, conduct optical microscope checks on the greater than 20 μm fraction and found that in a few of the northern Atlantic cores pteropods were present. In these samples, pteropods calcite is included in the foraminifera fraction. Further, Frenz et al. [2005] come to a similar conclusion. Although no check was made, calcareous dinocysts may make a small contribution to the less than 20 μm fraction.

[9] A listing of the locations, water depths, CaCO3 contents and fraction of the CaCO3 coccoliths (i.e., in the less than 20 μm fraction) can be found in Table 2. The fractions of coccolith calcite and core depths for the Atlantic samples are shown in the map in Figure 4 and those for the Pacific and Indian oceans in the map in Figure 5.

Figure 4.

Summary of the percent coccolith CaCO3 results for core top samples from various locales in the Atlantic Ocean. Where both piston and trigger weight samples were measured, the results are averaged. The red numbers are water depths in kilometers, and the blue numbers are percentages of coccolith CaCO3. The core numbers, CaCO3 contents, and coccolith percentages for individual samples are listed in Table 2.

Figure 5.

Summary of the percent coccolith CaCO3 results for core top samples from the Pacific and Indian oceans. Where both piston and trigger weight samples were measured, the results were averaged. The red numbers are water depths in kilometers, and the blue numbers are percentages of coccolith CaCO3. The core numbers, CaCO3 contents, and coccolith percentages for individual samples are listed in Table 2.

Table 2. Summary of the Measurements of the Split Between Less Than 20 μm CaCO3 and greater than 20 μm CaCO3 on Core Top Material From Trigger Weight and Piston Cores From the Lamont-Doherty Sediment Core Collectiona
Core NumberLatitudeLongitudeWater Depth (km)Bulk CaCO3 (%)Foraminifera CaCO3 (%)Coccolith CaCO3 (%)equation image × 100
  • a

    Here <20 μm CaCO3, coccolith; >20 μm CaCO3, foraminifera. TW, trigger weight; P, piston. Also included is a single measurement on an Ontong-Java Plateau giant gravity core top provided by WHOI's Dan McCorkle and published previously by Chiu and Broecker [2008].

Atlantic Ocean 35°–40°N
   P  3.285265969
RC9-189 P37°N20°W3.44764187
   P   41103176
   P   18.62.516.187
Atlantic Ocean 22°N–30°N
   P   8047695
   P   79196076
   P   80503038
   P   89593034
   P   90315966
VM22-218 P28°N43°W3.983384554
   P   86355159
   P   91415055
   P   82226073
   P   50193162
   P   50163468
   P   86345260
VM30-63 TW26°N16°W3.048173165
   P   63293454
   P   66244264
Gulf of Mexico 22°N–27°N
   P   36   
   P   43151842
VM3-36 TW27°N85°W2.648163267
Caribbean Sea 12°N–15°N
   P   51114078
RC8-107 TW15°N73°W3.252292344
   P   49272245
Equatorial Atlantic 6°N–9°S
VM31-139 TW6°N49°W3.330141653
   P   35221337
   P   74423243
   P   36181850
   P   80443645
   P   62372540
   P   82552733
   P   93593436
Atlantic Ocean 30°S–39°S
   P   79   
   P   31181342
RC11-37 P32°S35°W2.682542834
   P   268.517.567
   P   57322544
   P   44212352
   P   308.721.371
   P   55441120
VM13-2 TW39°S70°W2.6297.121.975
North Indian Ocean
VM34-91 TW21°N64°E3.4477.539.584
Indian Ocean 1°S–6°S
   P   84453946
   P   82463644
   P   81473442
   P   80542633
Indian Ocean 21°S–27°S
VM18-200 TW21°S64°E3.384572732
   P   87464143
VM20-173 P22°S69°E3.385345160
   P   90612932
RC17-94 P22°S69°E2.789554449
   P   88583034
   P   87592832
   P   80285265
   P   86275969
North Pacific 24°N–29°N
VM21-93 TW25°N142°E2.913.73.410.375
VM24-96 TW28°N179°W3.387434451
VM28-304 TW29°N134°E2.94183278
RC12-365 TW24°N126°E2.853312242
Pacific Ocean 1°N–5°S
   P   84503440
   P   81433847
RC17-176 TW4°N159°E3.281522936
South Pacific 25°S–29°S
   P   87256271
   P   78245469
   P   72254765
RC9-99 TW25°S116°W2.783612227

[10] The 17 results for equatorial zone samples are a bit lower than those based on the size index measurements (i.e., they range from 29 to 53% and average 42%). Those for latitudes greater than 20° spread over a much larger range (29–93%). With two exceptions, the 7 cores with coccolith percentages of 75 or greater are located near the margins of the oceans. These high values could be explained in several ways: (1) a high production ratio of coccolith calcite to foraminifera calcite, (2) preferential pore water dissolution of foraminifera calcite in organic-rich continental margin sediments, or (3) sorting by current action. Eight cores from a small area on the Mid-Atlantic Ridge (23–30°N) yield coccolith calcite percentages ranging from 31 to 76. While it is possible that this reflects sorting by currents, it might also be related to the extremely low sedimentation rates found in this area (∼1 cm/103 years). The results for 9 cores from 23 ± 3°S in the Indian Ocean range from 29 to 71% coccolith calcite. Again, either sorting or low accumulation rates might be called upon. We also recognize that significant amounts of fine-grained carbonate may be shed from the shelf along some margins.

4. Discussion

[11] Broadly speaking our results suggest that roughly equal amounts of coccolith and foraminifera calcite rain to the open ocean seafloor. This in itself is an interesting result pointing to some ecologic tie between the productivity of coccolithophorids and that of foraminifera. This being said, it is the departures from this near equality which hold promise for future study. It would be interesting to make measurements on traverses of core tops across zones where coccolith blooms occur to determine whether or not there is a corresponding increase in foraminifera production. Studies of a series of cores from continental margins would perhaps make possible the separation of production-related and dissolution-related variations in the coccolith to foraminifera calcite ratio.

[12] As can be seen in Table 2 and summarized in Figure 6, the agreement between piston and their companion trigger weight core tops is often well outside the measurement error (a few percent). Some of these differences could represent losses during coring (i.e., over penetration of the trigger weight cores or faulty setting of the piston core scope). To avoid such losses, future studies would best be conducted on box cores or multi cores. Also, confirmation that the core top material is of late Holocene age should be provided by 14C dating. In the Pacific and Indian Ocean, in order to avoid dissolution biases, studies should be confined to cores from shallower than 3 km.

Figure 6.

Histogram of the differences between the percent coccolith CaCO3 (i.e., <20 μm CaCO3) as measured in the trigger weight (TW) and piston (P) core tops.

[13] Although little has been written about the rain ratio of coccolith to foraminifera calcite to the seafloor, the results presented here are broadly consistent with three studies which have come to our attention. As already mentioned, Frenz et al. [2005] obtained similar results on South Atlantic sediment. Sediment trap measurements carried out by Ramaswamy and Gaye [2006] in the Arabian Sea, Bay of Bengal and Equatorial Indian Ocean found a similar range of rain rates for foraminifera calcite (6–23 g m−2 a−1) to coccolith calcite (4–24 g m−2 a−1) yielding a rain rate ratio ranging from 0.8 to 2.2. Balch et al. [2007] developed algorithms on the basis of 14C and 45Ca calcification incubations to estimate global coccolith production rates. They obtained an estimate of 1.6. ± 0.0.3 Pg a−1. While somewhat higher than the estimate of 1.0 Pg a−1 for the total calcite rain to the seafloor [Morse and Mackenzie, 1990; Archer and Maier-Reimer, 1994], considering the large uncertainties in the coccolith production rate it does not disagree with our finding. If we accept the estimate of 1.0 Pg a−1 for the total CaCO3 rain, then our ratio would convert to a coccolith rain of about 0.5 Pg a−1.

[14] In conclusion, our preliminary study is an attempt to quantify the relative contribution of CaCO3 produced by coccoliths and that produced by foraminifera to deep sea sediments. It confirms the earlier study by Frenz et al. [2005]. The results raise a question to which we have no ready answer. What is the ecologic connection between the productivity of these two calcifiers that leads to a near equality in their contribution to late Holocene marine sediments?


[15] This research was funded in full or in part under the Cooperative Institute for Climate Applications Research award NA08OAR4320179 from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce. The statements, findings, conclusions, and recommendations are those of the authors and do not necessarily reflect the views of the National Oceanic and Atmospheric Administration or the Department of Commerce. This is LDEO contribution 7283.