DISCUSSION

The different response of CaCO₃ compared to the other biogenic components of the Eocene sediment section is strong evidence of significant changes in ocean carbon chemistry over relatively short periods of time. We investigated change in the CCD between 38 and 41.6 Ma (base of Site 1218) in detail. We found that production of CaCO₃ was an important factor in deepening the CCD during CAE-3 but that dissolution was the most important factor ending CAE-3 at ~40.5 Ma. CAE-3 and CAE-4 are actually part of a single high-productivity interval divided by an interval of high CaCO₃ dissolution (Fig. F4). The dissolution between the events coincides with a major ¹³C anomaly followed by warming conditions at the poles, as indicated by oxygen isotopes. The increases of CaCO₃ MAR and biogenic SiO₂ MAR were about equivalent at the start of CAE-3, and the CCD deepened relatively slowly. At the end of CAE-3, the CCD dropped abruptly by more than 500 m in 100 k.y. The decrease in deposition of other biogenic components occurred much more slowly and indicates that dissolution amplified the loss of CaCO₃ relative to biogenic Si and C_org. The total CCD shallowing from the peak of CAE-3 deposition at 41 Ma to the CCD minimum between 40.5 and 40.0 Ma is at least 800 m, about two-thirds the size but in the opposite sense as the CCD change at the Eocene–Oligocene transition.

The initiation of CAE-3 coincided with the end of the MECO event, just before 42 Ma (Bohaty and Zachos, 2003). Low levels of CaCO₃ were deposited during the MECO interval at Site 1219, so the equatorial Pacific CCD was relatively deep for the Eocene during the MECO event. We estimate that the CCD was probably in the range of 3900–4000 m then, slightly deeper than the paleodepth of Site 1219 at that time (Table T4). Because biogenic SiO₂ MAR increased by a factor of 4 at the start of CAE-3 and the CCD prior to the event was relatively deep, it is clear that a major factor in the CaCO₃ event was increased production of CaCO₃ in a productivity event.

CaCO₃ MAR increases after 42 Ma are also linked with a strong increase in ¹⁸O of bulk CaCO₃ (Fig. F7). The oxygen isotope increase of 1.2 was divided into two steps, roughly half before 41.5 Ma and half afterward. The total event thus involved either ~6°C cooling of surface waters or ~120 m of sea level change caused by ice growth or a combination of the two. The first step coincides with the beginning of CAE-3. Benthic oxygen isotopes (Fig. F8A) from Site 1218 and the bulk oxygen isotope record from Site 1219 show that the end of the event began while oxygen isotopes remained heavy. Significant loss of CaCO₃ thus began prior to a sea level rise associated with deglaciation.

The CCD minimum between CAE-3 and CAE-4 is one of the most extreme minima in the record. Biogenic SiO₂ MAR remains high over the interval, ~50% higher than the background biogenic SiO₂ MAR, so we expected that CaCO₃ MAR should have been ~0.2–0.4 g CaCO₃/cm²/k.y. if the dissolution rate remained constant. Instead, all CaCO₃ was dissolved. A primary cause of the CaCO₃ minimum must be changes in the carbon system within the ocean, making deep waters much more corrosive to solid CaCO₃. Figure F8B shows how the first disappearance of CaCO₃ between CAE-3 and CAE-4 apparently coincided with a large ¹³C minimum. The final disappearance of CaCO₃ coincided with the appearance of lighter oxygen isotopes, indicating warming and probably ice melting as well. The ¹³C minimum, if further work confirms its existence, should mark a large new addition of an isotopically light C_org reservoir to the oceans (e.g., additional C_org from the continents or a methane release), which would cause warming and increased carbonate dissolution in the oceans. Expansion of shallow seas from glacial melting may have further trapped CaCO₃ on the shelf regions and driven the pelagic CCD to extremely shallow depths. For a time near 40 Ma, the CCD rose above the level of Site 1218, or to a depth shallower than ~3250 m. Feedback within the carbon cycle thus ended the development of ice sheets in Antarctica in the middle Eocene and returned the Eocene to warm conditions.

We documented that there were significant CCD changes in the equatorial Pacific through the middle and late Eocene and showed that the CCD changes are linked to long-term cooling events lasting as long as 2 m.y. We also showed that at least one of these events (CAE-3) was coincident with high-latitude cooling, implying that the CAEs mark glacial periods within the Eocene.

We suggest the cause of these events and their demise was a result of interesting dynamics between different carbon reservoirs in the carbon cycle interacting with climate. The evidence we have amassed suggests an interplay between pelagic carbonate production and changes in carbon reservoirs either internal or external to the ocean. We also note that the CAEs are sufficiently long that changes in the major cation content of the oceans could also have played a role in their evolution but not in their sudden appearance or disappearance.

Rapid changes in the equatorial Pacific CCD can be driven by any or all of several factors:

1. Shifting CaCO₃ depocenters from the deep ocean to shelves and shallow seas;
2. Shifting pelagic CaCO₃ production from regions outside the equator to the equatorial region; or
3. Changes in the dissolved inorganic carbon (DIC) content of the oceans relative to the net weathering flux of anions and cations.

Shifting CaCO₃ burial, either to shelf and shallow seas or from one region of the deep ocean to another, affects the CCD by changing the locus where weathering inputs are balanced. It is not important to the oceanic carbon cycle where the balance occurs. However, moving the locus of CaCO₃ deposition from the deep ocean to the shelf regions results in a shallow CCD because the total DIC of the oceans drops, CO₃^2– activity drops, and more CaCO₃ dissolves at shallower depths. The result is that the increased shallow CaCO₃ burial results in lower pelagic CaCO₃ burial and a shallow CCD. The same scenario can be explained by changes in pelagic biogeochemical cycles. High production in one region results in a relatively large fraction of CaCO₃ surviving at the seafloor to be buried in the sediments. Whereas regions of both high and low CaCO₃ production will suffer from increased dissolution, the region of lower CaCO₃ production will be depleted of CaCO₃ first and, therefore, have a shallower CCD.

Changes in the total DIC content of the oceans can be driven by addition or removal of C_org or hydrocarbons as well as carbonate production. The most relevant example of this is the disappearance of CaCO₃ at the Paleocene/Eocene boundary, perhaps caused by a major release of methane trapped in gas hydrates (Dickens et al., 1995). The resulting increase in DIC drove down ocean pH and caused the CCD to shallow. In addition, high atmospheric CO₂ content is supported in this scenario. If organic matter is rapidly buried in shallow basins, the opposite effect can occur. Drawdown of the DIC pool with respect to alkalinity results in higher burial rates of CaCO₃ in the deep ocean and a deeper CCD, as well as lower atmospheric CO₂.

The indications of higher productivity associated with CAE-3 and CAE-4 imply that drawdown of DIC by organic burial in shallow basins may drive such events only if they are sufficiently decoupled from the pelagic nutrient cycles (i.e., lowering DIC does not lower total nutrient contents of the oceans as well). The shifts in CCD that we observe would not necessarily require huge changes in biogeochemical cycling (e.g., Broecker, 1982). Carbon is buried preferentially to the other nutrients, so burial of carbon in shallow sea basins or along continental margins may not require large losses of phosphorus or nitrogen. In addition, the storage of nutrients in the deep ocean buffers the effect of nutrient changes from productive regions. Changes in the rate of upwelling typically have a far larger effect on nutrient supply to the euphotic zone than changes in nutrient content.

The shelf-basin CaCO₃ fractionation scenario most intimately links changes in the CCD with glaciations. The slow buildup of the CAEs is consistent with a slow buildup of continental ice, but the shelf-basin fractionation cannot stop the continued development of glaciers. Another feedback mechanism must end the event. For CAE-3, the event ended by a transfer of DIC into the oceans as indicated by an abrupt shallowing of the CCD prior to an oxygen isotope change. Additional shallowing of the CCD followed, perhaps associated with the flooding of shelves and shallow seas or a movement of pelagic CaCO₃ depocenters away from equatorial regions. A decrease in productivity is associated with the CaCO₃ decline, but the decline of productivity itself was insufficient to cause the extreme shallowing of the CCD between CAE-3 and CAE-4.