DISCUSSION

Two fundamental questions arise from the Leg 199 organic carbon records: the first is why the organic carbon values are so low for Eocene sediments at Sites 1218 and 1219, and the second is whether relevant geologic information can be extracted from the organic carbon variations, especially for the observed coherence between C_org MAR and the MARs of the biogenic components (Fig. F5). The near absence of organic carbon in Eocene sediments could reflect one or more of the following mechanisms: (1) primary productivity, and therefore organic carbon production, was much lower than modern rates; (2) organic carbon production, sedimentation, and preservation were higher than measured, but diagenetic remineralization, perhaps long after deposition, recycled the organic carbon; and (3) organic carbon production was much higher than observed but some process "short-circuited" its sedimentation such that it was never preserved in the sediments. Our conclusions support the third of these three scenarios, and we present evidence which eliminates the first two.

The C_org content of Holocene surface sediments in the pelagic Pacific are low (range = 0.2–0.4 wt% C_org) relative to continental margin sediments (range = 1–6 wt% C_org), yet they are an order of magnitude higher than Eocene sediments at Sites 1218 and 1219. Only two possibilities can explain this discrepancy: (1) the initial C_org flux to surface sediments was higher when deposited (similar to modern?), and the low values we observe reflect C_org degradation over geologic time; or (2) the initial C_org flux to surface sediments was very low, resulting in the concentrations we observe. If the former case is true, then it follows that there has been postdepositional remineralization, and further studies of these sediments would provide insight into the many unresolved questions regarding organic matter degradation and preservation in marine sediments. If the latter scenario is true, then it implies that conditions in the equatorial Pacific Ocean that controlled the C_org rain to the seafloor were very different during the Eocene

Upper limits on the initial organic carbon content and subsequent diagenesis of the sediments can be inferred from the pore water chemistry (Shipboard Scientific Party, 2002). Diagenesis has been primarily oxic as evidenced by sulfate concentrations in the interstitial waters from both sites. Sulfate concentrations are similar to modern seawater (range = 25–32 mM SO₄) and persists down to basement, suggesting that sulfate reduction was not a significant process that oxidized the labile fraction of organic matter. Assuming that the organic carbon rain and burial rate in the sediments remained below the sulfate reduction "threshold," the competing effects between C_org MAR and oxygen availability within the framework of oxic diagenesis provide boundary conditions. Indeed, it appears that throughout the Eocene, oxygen availability in equatorial Pacific sediments region was as high, if not higher than, modern sediments beneath deep waters in the equatorial Atlantic and Pacific Oceans.

The dissolved oxygen content of Eocene eastern equatorial Pacific seawater was estimated using the oxygen depletion history of North Atlantic surface-to-deep waters over an equivalent path length and modern temperature-oxygen saturation data of surface waters. Bottom water pathways are still debated (Corfield and Norris, 1996) but there was likely a North Atlantic source component to the equatorial Pacific (Bice et al., 2000; Huber and Sloan, 2001). Both the Drake Passage and South Australia pathways in the Southern Ocean were closed during the Eocene, and Atlantic-derived waters likely traveled through the Panama Gateway, which was still open to exchange (Frakes and Kemp, 1972).

We used the distance of the modern pathway from the North Atlantic to the South Atlantic to select an oxygen depletion value for the same distance to the equatorial Pacific via the Panama Gateway. For the modern case, the oxygen saturation value of high-latitude, North Atlantic surface waters is ~350 µmol/kg and is depleted to ~250 µmol/kg O₂ when it reaches the South Atlantic Deep Waters (Millero, 1996).

The oxygen concentration of surface waters depends upon temperature. We know that global Eocene deep-sea bottom waters were persistently warm, ranging between 12° and 5°C (Lear et al., 2000; Zachos et al., 2001). This epoch is characterized by a cooling trend from a high of 12°C at 52 Ma to 5°C during the latest Eocene. Warm bottom waters (10°–12°C) also characterize Leg 199 sites in the equatorial Pacific as evidenced by the ¹⁸O values of benthic foraminifers (Tripati et al., unpubl. data). Recent work by Thomas (2004) argues that the source of deep ocean waters to the equatorial Pacific during the early Cenozoic (prior to ~41 Ma) was the North Pacific, when bottom water temperatures were consistently above 7°C. After 41 Ma, temperatures cooled below 7°C and the source of deep water changed to the Atlantic Ocean. Differences in source latitude and path length for the source waters should then be reflected in the oxygen content of eastern equatorial Pacific deep water. The total path length from the North Pacific vs. the North Atlantic would have increased by ~20% after 41 Ma. Because other factors cannot be constrained at this time, we simplified our estimate by using the North Atlantic source (Millero, 1996).

We estimate that Eocene surface waters originating at 65°N and temperatures between 7° and 12°C would have contained between 250 and 300 µmol/kg O₂, based on saturation data in Millero (1996). Deep waters formed from this mass would have contained between 175 and 200 µmol/kg O₂, all other factors remaining equal, when they reached the eastern equatorial Pacific. This concentration is greater than deep waters in the modern Eastern equatorial Pacific (110 µmol/kg O₂) (Emerson and Bender, 1981) between 3.1 and 3.5 km depth and North Pacific Deep Waters (~150 µmol/kg O₂) between 3 and 5 km, the end-point of deepwater circulation. After 41 Ma, oxygen concentrations would have been lower than 175 µmol/kg O₂ if the source changed to the more distant Atlantic. Regardless of source, deep waters of the Eocene equatorial Pacific probably contained <175 µmol/kg given the longer path length but not <150 µmol/kg O₂, the modern value for North Pacific deep waters between 2 and 3 km.

The estimated oxygen content of Eocene deep waters provides a framework for understanding organic carbon degradation during early sediment diagenesis. Two modern analogs of organic carbon burial relative to their respective oxygen exposure were considered for comparison with Leg 199 sediments. The modern examples record the organic carbon deposition and burial in both the Atlantic and Pacific equatorial upwelling regions (Lyle, 1988) over the past 300 k.y. Surprisingly, we found that the organic carbon profiles beneath these upwelling regions, in well-oxygenated waters, are identical to each other (Lyle, 1988) despite a fourfold difference in oxygen exposure time for Atlantic sediments vs. the Pacific. The C_org MAR in the equatorial Atlantic is lower by a factor of two, yet the deepwater oxygen saturation is higher by a factor of two (Levitus and Boyer, 1994) relative to the equatorial Pacific. Organic carbon sediment profiles are the same for both regions, ranging between 0.2 and 0.4 wt% C_org over the past 300 k.y. (Lyle, 1988). Bulk MARs at equatorial Pacific Sites 1218 and 1219 are 5 g/(cm² x k.y.), equal to modern Atlantic equatorial sediments (Lyle, 1988). These data, combined with the estimated oxygen saturation in Pacific equatorial deep water (175 µmol/kg O₂) during the Eocene, predicts an oxygen exposure time that is lower than the modern equatorial Atlantic. The implication of this lower oxygen exposure time is that the order of magnitude lower organic carbon content in Leg 199 sediments does not result from diagenetic remobilization in the top few meters of the sediment column. To the contrary, it can be argued that lower oxygen exposure times should have produced higher than modern organic carbon contents in Leg 199 sediments. The question, then, is whether the low organic carbon content reflects significantly lower primary productivity, and here we turned to the barium content in Leg 199 sediments.

High pore water sulfate profiles suggest that barite was never under-saturated but was well preserved in the sediments (Shipboard Scientific Party, 2002) and can be used as a good proxy indicator of primary productivity in the overlying water column. A significant body of research suggests that barite is formed in the water column during the simultaneous decay of organic matter as it rains out of the euphotic zone (Dehairs et al., 1980; Bishop, 1988; Stroobants et al., 1991; Dymond et al., 1992; Gingele and Dahmke, 1994; Francois et al., 1995). Sulfate, released from the rain of decaying organic matter, likely forms sulfate-rich microenvironments, which combine with barium in seawater to ultimately form dissolution-resistant barium sulfate crystals that are deposited in the sediments.

Dymond et al. (1992) quantitatively inferred surface water productivity from the particulate biogenic-barium rain collected in sediment traps near the ocean floor. Pfeifer et al. (2001) tested this against other models and compared the predicted primary productivity results to actual measurements in the South Atlantic Ocean. Their analysis of surface sediment barium concentrations shows that the model of Dymond et al. (1992) produced the best fit to independent measurements of modern primary productivity in the surface waters. Another important result of the Pfeifer et al. (2001) study is that the predicted level of productivity agreed more closely to measured values when the selected Bio-Ba/C_org ratio, a required model parameter, reflected the basin waters of the study area, rather than a "global average." Because barite crystals that form in the cells of xenophyophores, now commonly considered to be an unusual group of benthic foraminifera, are very similar to those formed in the water column (Hopwood et al., 1997), we cannot at this stage deconvolve our prediction of surface water productivity signal from this potential source of barium (K. Faul, pers. comm., 2004).

Leg 199 sedimentation fluxes and predicted rain fluxes of Bio-Ba and C_org for Eocene and younger ages are listed in Table T4 along with data for modern Pacific equatorial upwelling regions. Rain flux data in Table T4 were calculated from sediment flux measurements using the "recycled rates" from Dymond and Lyle (1994). They measured the "rain rate" ("export flux") and sediment "burial rate" (MAR) of the C_org component to calculate the C_org "recycled flux" for two equatorial Pacific sites: Site C (equatorial Pacific, 1°N) and Site S (11°N). The recycled flux is the difference between the rain rate and burial rate and represents the fraction of the rain not buried.

Figure F6 illustrates the measured Bio-Ba MAR in the sediments as a function of paleolatitude and age for both Leg 199 sites. The maximum flux occurs during the CAE-3 event between 41.4 and 41.1 Ma and also coincides geographically with the location of the paleoequator, consistent with our interpretation that Bio-Ba is a proxy for paleoproductivity. There is a general decrease in productivity away from the equator, as indicated by Bio-Ba, a pattern consistent with the modern regime. At both sites, the estimated C_org sediment MAR is in concert with the measured Bio-Ba MAR throughout the Eocene, except at 41 Ma, when the estimated C_org flux to the sediment roughly equals or exceeds modern fluxes at the equatorial Site C (Table T4; rows 1, 3, 7). At Site 1218, the measured Bio-Ba flux throughout the Eocene is 50% higher than modern (between 38 and 43 Ma) (Table T4; row 3). At Site 1219 between 36 and 49 Ma (Table T4; row 7) background Bio-Ba and Ba-estimated C_org sediment fluxes are approximately equal to modern. In general, the values are lower than those at Site 1218, which may be explained by the fact that Site 1219 was located off the equator, between 1° and 4°S. During the CAE-3 event that occurred between 41.1 and 41.5 Ma, a large flux in Ba-estimated C_org MAR occurred at both sites. At Site 1218 (Table T4; row 4), it is two–three times higher than modern Pacific equatorial Site C (Table T4; row 1), and Bio-Ba sediment MAR is four–six times greater than modern (Table T4; row 4). At Site 1219, the estimated C_org and Bio-Ba burial fluxes are two and three times modern, respectively (Table T4; row 8).

In the modern equatorial Pacific, C_org MAR is 2% of the rain flux and Bio-Ba MAR is 30% of the rain flux (Dymond and Lyle, 1994). We calculated rain fluxes from these results by assuming similar recycling rates during the 41-Ma CAE-3 event. As shown in Table T4 for Site 1218, the Bio-Ba rain flux was six times higher than modern, and the C_org rain flux estimated from Bio-Ba was three times higher than modern (Table T4; row 4: 28 and 400 mg/[cm² x k.y.], respectively). During this 41-Ma event at Site 1219, the estimated C_org rain flux is two times modern, and the Bio-Ba rain flux is three times modern. Again, the estimated rain fluxes are one-half that at Site 1218 but may also be explained by the differences in location relative to the paleoequator: Site 1219 was located 2°S of the equator, whereas Site 1218 was directly beneath the equator.

Collectively, our results suggest that throughout the Eocene there was higher productivity in surface waters as evidenced by the measured Bio-Ba sediment and predicted rain fluxes. The measured C_org MAR was lower than Holocene rates by a factor of 10. Because C_org and Bio-Ba fluxes covary, the C_org signal is probably robust. In contrast, the calculated C_org MAR (from the measured Bio-Ba) indicates fluxes that are roughly equal to modern sedimentation fluxes throughout the Eocene with the exception of the 41-Ma event, when calculated C_org fluxes triple. This 41 Ma event corresponds to CAE-3, the third carbonate accumulation event in a series of seven identified in the Eocene from Leg 199 sediments (Lyle et al., this volume; Tripati et al., unpubl. data). The calculated C_org MAR increases by a factor of four at 41 Ma to a peak value of 7.8 mg/cm² x k.y. This calculated result is complementary to, and fits with, the interpretation of Tripati et al. (unpubl. data), who surmise that the large negative ¹³C isotope excursion that also occurred at 41 Ma is a reflection of increased ocean productivity. Because the bulk sedimentation rates at both sites are the same as or higher than the modern analogs (Site C at 1°N and Site S at 11°N), organic carbon preservation during the Eocene should have been the same as, if not higher than, modern preservation.

Leg 199 organic carbon average concentrations (0.03 wt%) are an order of magnitude lower than modern pelagic sediments of equal or lower sedimentation rates in well-oxygenated seawater under Pacific and Atlantic upwelling regions (Lyle, 1988). Modern sediments exhibit near monotonous C_org concentration profiles ranging between 0.2 and 0.4 wt% over the past 300 k.y. However, a comparison of MARs for modern vs. Eocene sediments shows that the C_org fluxes are roughly equal in magnitude, between 2 and 10 mg/cm² x k.y. (Lyle, 1988). An increased flux occurs during the last glacial maximum at 18 Ka in the Atlantic, when C_org MARs doubled to 20 mg/cm² x k.y. relative to interglacial Stage 3. In the equatorial Pacific, C_org MARs increased by a factor of 3–10, with values ranging from 4 to 100 mg C_org/cm² x k.y. The coherency of the trends for Leg 199 Eocene and modern sites demonstrates that cooling events are associated with increasing C_org mass accumulation rates to the sediments and that some fundamental mechanism has operated throughout the Cenozoic to account for this association.

We suggest that low C_org MARs in Eocene sediments reflect the fact that most of the organic carbon was remineralized in the water column or at the sediment/water interface, thus preempting burial. The notion that the fluxes reflect lower primary productivity is rejected because our evidence shows that the proxy indicator of productivity, Bio-Ba, which resists dissolution, has escaped recycling to become preserved in the sediments. A very simple model for carbon and nutrient utilization is based on the long-known biological "Q₁₀ rule" derived from the Arrhenius relationship for reaction rates as a function of temperature. The Q₁₀ model predicts a doubling to tripling of metabolic rates for ectothermic organisms with every 10°C increase in temperature. Faster recycling of organic carbon (and other nutrients) likely occurred in the very warm Eocene oceans for all organisms based on higher metabolic requirements alone.

The simple Q₁₀ model of metabolism has been expanded and elucidated by Gillooly et al. (2001) to include many life forms such as microbes, ectotherms, endotherms, and plants. They show that the resting metabolism of all organisms is a function of both environmental temperature and body mass over the "biologically relevant" temperature range between 0° and 40°C (i.e., temperatures at which life is possible for these organisms). The lowest metabolic rates are associated exclusively with the smallest unicellular organisms, whereas metabolic rates are greatest for the largest organisms (endothermic vertebrates) by a factor of 20.

Gillooly et al. (2001) proposed a single equation, known as the Universal Temperature Dependence (UTD) of metabolism (Equation 4). The UTD equation predicts metabolic rate as a function of temperature and body mass and applies to all organisms over their biologically relevant temperature range. Specifically, the UTD equation is expressed as:

Q = b₀ M^3/4 e^–E/kT, (4),

where

Q = metabolic rate,

M = body mass,

E = activation energy for the rate-limiting enzyme-catalyzed metabolic reactions,

k = Boltzman constant,

T = absolute temperature (K), and

b₀ = a constant independent of M and T.

The UTD equation is more sophisticated than the Q₁₀ rule because it accounts for the scaling of metabolism with the ³/₄ power of body mass, yet preserves the temperature dependence term derived from classical statistical thermodynamics. In contrast, the Q₁₀ relationship, although relevant, simply quantifies metabolic rate changes for 10°C jumps in temperature.

All cellular metabolic reactions are characterized by two features: (1) the requirement of water and (2) the production of adenosine triphosphate (ATP; C₁₀H₁₆N₅O₁₃P₃), the primary energy currency for all cellular biochemical reactions. Because ATP production uses oxygen as the final electron acceptor for aerobic organisms, oxygen utilization rate is the traditional metric used to quantify metabolic rate.

Basal metabolism, or "maintenance metabolism," is defined as the sum of processes that require ATP. These processes are necessary for the cell to just remain alive, and for an individual organism, basal metabolism refers to the condition such that food intake results in neither an increase nor a decrease in body mass. For some ectotherms, the majority of their lifetime energy is spent on basal metabolism (i.e., it is not invested in growth or reproduction but used exclusively to maintain cellular function) (Clarke and Fraser, 2004). The function of basal metabolism involves protein synthesis, and most of the energy is spent on counteracting the passive leakage of protons across the inner mitochondrial membrane of eukaryotic cells (plants, animals, fungi, algae, and protozoa). This essential proton gradient, known as the proton motive force by Mitchell (1966), is the driving force for ATP synthesis and all subsequent biological functions. An intriguing consequence of our hypothesis is that the Cenozoic marine sedimentary record of organic carbon reflects this fundamental requirement of all living cells to maintain proton gradients across their cell membranes.

Nearly all studies of modern marine organisms show a positive, nonlinear relationship between temperature and resting metabolic rate (Clarke and Fraser, 2004, and references therein). Clarke and Fraser (2004) expanded upon these relationships and upon the simple UTD model proposed by Gilooly et al. (2001). They found that the basal metabolic function of an organism depends not only upon its ecology but also on the organism's evolutionary cost-benefit compromises. For example, in polar environments with severe food shortages, whether they are chronic or seasonal, selection has occurred for organisms with very low basal metabolic rates and efficient metabolisms. Within the polar environment, both benthic and planktonic organisms have evolved different strategies to meet their basal metabolic rates and "winter-over" strategies in the face of these food shortages. The two different strategies in turn reflect fundamentally different lifestyles. Immobile benthic organisms have very low basal metabolic rates and do not store energy reserves for winter food shortages. In contrast, planktonic organisms have much higher energy requirements for two reasons: first, they require energy for staying afloat in the water column, and second, these organisms synthesize lipid reserves for the purpose of using them during the over-wintering period.

Higher basal metabolic function during times of ocean warming also requires a concomitant supply of organic carbon and nutrients in surface waters in order to sustain a given unit of primary production. If the nutrient supply to the euphotic zone during the Eocene was lower than modern values, then it is reasonable to expect total primary productivity to have been lower than modern values, a result that is contradicted by Leg 199 Bio-Ba data. Indeed, both the measured Bio-Ba MAR and the calculated C_org MAR is two–three times higher than modern during the height of the CAE-3 "cooling" event at 41 Ma, despite the fact that ocean temperatures were warmer than modern and proportionately more C_org was required to maintain basal metabolism. This implies that large supplies of carbon and other nutrients to the euphotic zone were sustained.

Pleistocene marine sediments also support our proposal that the UTD theory is a fundamental mechanism controlling the global carbon cycle, at least since the beginning of the Phanerozoic. For example, during the Last Glacial Maximum (LGM) at 18 Ka, C_org MAR peaked in both the equatorial Atlantic and Pacific Oceans (Lyle, 1988). This history in core V19-28 is particularly relevant because of its location (2°S, 85°W) in the eastern equatorial Pacific. During the LGM, when global temperatures dropped, the C_org MAR in core V19-28 peaked to 100 mg/cm² x k.y. This value is 10 times higher than those that preceded and followed this event. This peak MAR is 12 times greater than the value for the CAE-3 cold event at 41 Ma; at Site 1218, the maximum C_org MAR (calculated from Bio-Ba) is 8 mg/cm² x k.y. Inasmuch as global ocean temperatures were colder during the LGM relative to the Eocene, the fivefold increase in organic carbon MAR is consistent with a fundamental temperature dependence on primary production and consequent nutrient supply affecting all organisms in the food web. Note also that the Lyle (1988) data set is itself internally consistent with our hypothesis: the LGM at 18 Ka represents a maxima in C_org MAR relative to warmer periods before and after this event.

In summary, carbon isotopic data, oxygen isotopic data, and the sedimentation patterns of the biogenic calcium carbonate and opal components (Tripati et al., unpubl. data; Lyle et al., this volume) point to a rapid cooling event at 41 Ma, which then rebounded to warm conditions by 40.2 Ma. This transition from relatively cold to warm conditions is reflected in changes in the organic carbon flux at Sites 1218 and 1219, supporting the geologic universality of the UTD theory: because basal metabolic rates and organic carbon recycling rates would have decreased at 41 Ma (colder), a higher proportion of C_org would be expected to reach the sediments and survive early diagenesis until burial. Slower basal metabolic rates during cold periods would not translate into lower export production of the biogenic proxies to the sediments. On the contrary, slower basal metabolism for all organisms would encourage higher overall productivity because the "excess carbon" could be used for reproduction instead of maintaining cell function. Indeed, at both Sites 1218 and 1219, the calculated rain of Bio-Ba and C_org during the 41-Ma event is higher than modern rates; Bio-Ba was higher by factor of six and C_org was higher by a factor of three when Site 1218 was beneath the paleoequator.