For slowly accumulating sediments, a major contrast exists in the radiocarbon-age differences among coexisting shells of planktic foraminifera between those experiencing little dissolution and those experiencing significant dissolution. In the former, the ages generally agree to within a couple of hundred years. In the latter, age differences as large as 1000 years are common. The most likely explanation appears to be the Barker Effect, which involves the preferential fragmentation of dissolution-prone G. sacculifer and G. ruber. The whole shells of these species picked for radiocarbon dating have shorter residence times in the bioturbation zone than those for dissolution-resistant species (including benthics). As low accumulation rate sediment cores often fail to yield reliable radiocarbon-based ocean ventilation ages, where possible, such studies should be conducted on high accumulation rate cores.
 Our observational window into the rate of ventilation of the deep ocean during glacial time is based largely on the contrast between radiocarbon ages for coexisting benthic and planktic foraminifera shells. In a recently published paper [Broecker and Clark, 2010], we show that, for a sediment core from 4.4 km depth in equatorial Pacific, large age differences are found between the age of G. sacculifer and those for other planktics. Following Barker et al. , we attribute these differences to preferential fragmentation of the dissolution-prone G. sacculifer. If this explanation is correct, then the best estimate of abyssal ocean to surface ocean 14C/C difference should be based on measurements on dissolution-resistant planktic foraminifera. An alternate explanation for these large age differences is that they reflect a steep vertical gradient in 14C to C ratio in the upper ocean water column. If so, then reconstructions of the surface to deep 14C/C difference should be based on foraminifera shells formed in the surface-mixed layer, i.e., G. sacculifer or G. ruber. As we find that G sacculifer (and G. ruber) ages are much younger than those of coexisting species, if this is the correct explanation, then the reconstructed apparent age difference between deep and surface water would be considerably larger. In order to account for the large drop in the atmosphere's 14C to C ratio during the deglacial time interval, it is tempting to opt for this latter explanation. However, as shown here, the former explanation is very likely the correct one. Our case is based on the results from seven equatorial Pacific cores (see Figure 1).
2. Barker Effect
 In a paper published in 2007, Stephen Barker made a case that differential dissolution and fragmentation occurring within the sedimentary bioturbated zone should create offsets in radiocarbon ages among planktonic foraminifera. His idea was that, as only whole shells are picked for dating, fragmentation would produce a bias. The reason is that, as individual entities have a range in residence times, if fragmentation occurs, then those which have shorter than average residence times will have a lower probability of breakup than those which have longer than average residence times. Because of this, for sediments deposited in water undersaturated with respect to calcite, dissolution-prone planktic species should have younger radiocarbon ages than dissolution-resistant species (see illustration in Figure 2).
 It has been well established that G. sacculifer shells are among the most dissolution prone. Hence, our observation that they have younger ages than the more dissolution-resistant shells of P. obliquiloculata and N. dutertrei could be explained in this way. It is important to keep in mind that the magnitude of the G. sacculifer offset should depend on the mean residence time of entities in the bioturbated zone. Hence, given the same intensity of dissolution, this offset should be related to the ratio of the thickness of the bioturbation zone to the accumulation rate of the sediment. In recognition of the origin of this idea, we refer to this offset as the Barker effect.
 Instead of doing 14C-age measurements on only one planktic entity, we have made a point of doing two or more. Early on, Broecker et al.  made G. sacculifer–N. dutertrei14C-age comparison on samples from the Ceara Rise. For the two middepth cores (see Table 1), out of nine such comparisons, except for one, the ages agreed within the stated measurement error (average about 200 years). However, in the case of the core from 4.0 km depth, for two out of five samples, the G. sacculifer ages averaged 500 years younger than those for N. dutertrei. As during glacial time, the seafloor at 4 km depth was very likely subject to calcite dissolution; in hindsight, it is tempting to attribute these differences to the Barker effect.
Table 1. Comparisons of Radiocarbon Ages for Coexisting Dissolution-Prone G. sacculifer and Dissolution-Resistant N. dutertrei in Cores From the Ceara Risea
 In our latest paper [Broecker and Clark, 2010], we made analyses on four planktic species and also on coccoliths. As listed in Table 2, this paid off. We found that in all nine samples, G. sacculifer yielded ages far younger than those for the other three planktics and, in six samples, also for coccoliths. Our interpretation was that dissolution-prone G. sacculifer were anomalously young because the whole shells picked for analysis had shorter residence times in the core top mixed layer than did the dissolution-resistant planktics. If this were the case, then reconstructions based on surface to deep 14C to C ratio differences using G. sacculifer would be anomalously large.
Table 2. Summary of the Radiocarbon Measurements on TTN13-18 and TTN13-61, Cores With Very Low Accumulation Rates, From the Central Equatorial Pacifica
TTN13-18 (33 cm)
TTN13-18 (39 cm)
TTN13-18 (43 cm)
TTN13-18 (46 cm)
TTN13-18 (47 cm)
TTN13-18 (51 cm)
TTN13-18 (55 cm)
TTN13-18 (59 cm)
TTN13-61 (30 cm)
TTN13-18 (2°S, 140°W, 4.4 km) and TTN13-61 (1°S, 140°W, 4.2 km); accumulations rates of (2 cm/kyr) [Broecker and Clark, 2010]. As can be seen, a regularity exists between the age differences for planktic entities. G. sacculifer is the youngest, and G. tumida and coccoliths are the oldest. Note that the results from the sample at 55 cm have been excluded from the average.
 Since the publication of this paper, several colleagues interested in this problem have challenged our interpretation. They maintain that instead, the differences reflect a steep vertical gradient in 14C to C ratio in the glacial-age eastern equatorial Pacific upper water column. Were this the case, then the preferred species would be shallow calcifying G. sacculifer. The resulting much greater benthic-planktic age difference for glacial time would be consistent with the greater than present 14C to C difference required to explain the large drop in atmosphere and surface ocean 14C to C ratio which occurred during the deglacial time interval.
 In order to decide which of these hypotheses is more likely to be correct, we present here the results on six additional equatorial Pacific cores. As shown on the map in Figure 1, they represent a large range in both longitude and water depth (and also in accumulation rate). The results are listed in Tables 3, 4, and 5, and a summary is provided in Table 6.
Table 3. Summary of the Radiocarbon Measurements on MD1-2386, a Core With an Extremely High Accumulation Rate, From the Western Equatorial Pacifica
Table 6. Summary of Age Differences Between P. obliquiloculata and G. sacculifer, Between Benthics and P. obliquiloculata, and Between Benthics and G. sacculifer for Seven Sediment Cores From the Equatorial Pacific, Ranging in Depth From 1.4 to 4.4 km and in Accumulation Rate From 2 to 75 cm/kyr
Accumulation Rate (cm/kyr)
Δ Age, obl.-sac. (years)
Δ Age, Ben.-obl. (years)
Δ Age, Ben.-sac. (years)
 As summarized in Table 6, for four of the seven cores, the G. sacculifer shells have ages close to those for P. obliquilaculata (or N. dutertrei). The remaining three have G. sacculifer ages averaging 900 to 1400 years younger than those for P. obliquiloculata (or N. dutertrei). The latter group of cores come from water depths greater than 3 km at sites that are currently overlain by calcite undersaturated water. As during the last glaciation, there was little difference in the carbonate ion concentration in the deep equatorial Pacific Ocean [Yu et al., 2010], while in the bioturbation zone these forams were subject to dissolution. Three of the cores in the former group came from shallow enough water depths that their bioturbation zones, during glacial time, were bathed in water supersaturated with respect to calcite. Although the fourth core (i.e., the one from 2.8 km) is deep enough to have experienced dissolution, the sedimentation rate (40 cm/kyr) is so large that residence time of shells in an 8 or so cm thick bioturbation zone is too small to give rise to a significant Barker effect.
 It is important to note that the core from 1.4 km which shows no significant age difference among the planktic entities comes from the easternmost Pacific and that the core from 3.1 km which has a large G. sacculifer–P. obliquiloculata age difference comes from the westernmost Pacific. Taken together, these results allow us to reject the hypothesis that the G. sacculifer offset is the result of a very steep 14C/C gradient in the glacial eastern Pacific (but not in the glacial western Pacific). Further, cores experiencing dissolution show large age differences both on the eastern and western sides of the equatorial Pacific.
 According to Berger , the planktonic species G. ruber is even more susceptible to dissolution that G. saculifer. As G. ruber are rare in most of the cores we studied, only in a few cases was it measured. In core TTN13-18, we made a G. ruber–G. sacculifer comparison. As can be seen in Table 2, their ages agreed to within 50 years. Both were on the order of 2000 years younger than the other planktic entities. In core V21-40 (Table 5), two G. ruber–G. sacculifer age comparisons were made. In one, G. ruber was 210 years younger and in the other, 510 years older than G. sacculifer. In both cases G. ruber was much younger than the companion planktics. In V28-238, Table 5, one comparison yields a G. ruber age 240 years younger than G. sacculifer and both are far younger than the companion N. dutertrei. As these two cores are at depths of 3.1 and 3.2 km respectively, they were subject to dissolution. Finally, two comparisons were made in V19-27 from 1.4 km. In these the ages of G. ruber and G. sacculifer agreed to within the measurement error with each other and with those of the companion planktics.
 Having provided evidence that the Barker effect is operative, the question arises as to whether it biases the age differences for other species of foraminifera. If so, which species (or entity) provides the best match to benthics? In Table 7, we summarize the age differences between benthic foraminifera and those for coccoliths and four species of planktic foraminifera for samples from the two abyssal cores. Putting aside several anomalies, the average age differences (from benthics) for P. obliquiloculata, N. dutertrei and coccoliths agree to within ±150 years. However, the benthic–G. tumida age differences are on the average 500 years smaller.
Table 7. A Summary of the Age Differences Between Benthic and Various Planktic Entities for the Samples From the Two Cores From the Very Deep Central Equatorial Pacifica
TTN13-18 (33 cm)
TTN13-18 (39 cm)
TTN13-18 (43 cm)
TTN13-18 (46 cm)
TTN13-18 (47 cm)
TTN13-18 (51 cm)
TTN13-18 (55 cm)
TTN13-18 (59 cm)
TTN13-61 (30 cm)
Those for G. sacculifer average twice as large as those for P. obliquiloculata, N. dutertrei, and coccoliths and four times as large as that for G. tumida. The results from the sample from 55 cm are excluded from the averages. As explained by Broecker and Clark , in the samples from 51, 55, and 59 cm, the coccoliths were separated from the foraminifera by bioturbation.
 This being the case, should G. tumida or one or more of the other three entities be used to estimate the ventilation age? While the answer is by no means clear, a case can be made that coccoliths should be the winner. The reason is that both coccoliths and benthics appear to be highly resistant to dissolution. A study by Chiu and Broecker , showed that in the deep equatorial Pacific coccoliths dissolve an order of magnitude more slowly than foraminifera shells. Further, as the coccoliths are isolated by sieving, rather than picking, fragmentation is not an issue. But, of course, this would be the case only if benthics are immune to fragmentation. Although the case is weaker, observations reveal that benthic shells tend to be more robust than planktics.
 There is one drawback. As shown by Broecker and Clark , foraminifera and coccoliths can be differentially transported during burrowing. To guard against this, it would be safer to analyze P. obliquiloculata or N. dutertrei in addition to coccoliths. Agreement between the ages for these forams and that for coccoliths would insure that separation during bioturbation had not occurred.
 Of course if this strategy is adopted, then it would have to be assumed that the older ages found for G. tumida reflect a deeper habitat. As N. dutertrei and P. obliquiloculata have symbionts adapted to low light, they must live in the upper thermocline [Gastrich, 1987]. G. tumida have no symbionts. Watkins et al. , based on plankton tows, found that in the equatorial Pacific at 140°W all three of these species were abundant in the upper 100 m and rare below that depth. Based on 18O analysis of core top samples in the tropical Atlantic, Steph et al.  conclude that G. tumida calcifies at depths between 150 and 350 m. Based on this evidence, there is no indication that depth habitat can explain the sizable magnitude of the G. tumida offset.
6. Note of Puzzlement
 Just as we were completing this paper, we received results from an Indian Ocean core which raise questions regarding the above interpretation. This core has an accumulation rate of only 1.5 cm/kyr (see Figure 3). Based on its water depth (3.8 km) and on its size index [Broecker and Clark, 1999], it experienced substantial dissolution (see Figure 3). As listed in Table 8, the radiocarbon ages on planktonics from the last glacial maximum span a very large range. The apparent age of the G. sacculifer shells is 7.5 kyr younger than those for P. obliquiloculata and coccoliths. The dissolution-prone G. ruber shells are 4.8 kyr younger. The disconcerting results are those for the G. tumida and benthic (wuellerstorfi) shells. The G. tumida are 3.0 kyr older than the benthics.
Table 8. Radiocarbon Ages and 13C to 12C Ratios for a Sample From 38 cm Depth in Indian Ocean Piston Core RC14-33a
 Although the order of the differences among the planktics is reminiscent of that found for TTN13-18, the magnitude is too large to be explained by the Barker effect (i.e., the G. tumida–G. sacculifer age difference of 11.5 kyr is much greater than the 5.3 kyr residence time of entities in an 8 cm thick mixed layer. But for such a low accumulation rate, other bioturbation artifacts such as differential abundance gradients [Peng and Broecker, 1984; Duplessy et al., 1986] can be important. Such artifacts are likely to be more prominent at high latitudes where relative abundance changes tend to be larger and during the degalcial time interval when water temperature and salinity are changing. Although these offsets should perhaps be less serious during the relatively stable tropical climate, in low accumulation-rate cores bioturbation often extends from one climate interval to another. Of course, another possible cause of offsets is contamination with secondary 14C.
 One thing is clear from the results presented here. In low accumulation-rate sediments which have experienced dissolution, the radiocarbon ages in G. sacculifer (and G. ruber) are anomalously young. Hence in such situations, radiocarbon age for this species should not be paired with that for benthics. Further, if meaningful information regarding ventilation rates of glacial-age abyssal waters is to be obtained, then, where possible, such studies should be restricted to high accumulation rate sediments. In their absence, more must be learned about the impacts of the Barker effect.
 Advice from Howie Spero was extremely helpful in sorting out the depth habitats of planktic species. Lea Peña pushed us to consider the possible role of depth habitats in the glacial eastern equatorial Pacific. This material is based upon work supported by the National Science Foundation under grant OCE-04-35703. This is Lamont-Doherty Earth Observatory contribution 7457.