13C can be affected by detrital input such as from volcanogenic sources; however, such inputs were negligible in our study area at the time of the carbonate crash, as indicated by the nearby Caribbean sites (Fig.
F4).
Stable oxygen and carbon isotopic measurements of single foraminiferal species have long been used in paleoceanographic and paleoclimatic studies because of the simpler interpretation of these data. In contrast, bulk carbonate isotopic ratios reflect weighted-average signals of different source materials, which complicates the interpretation and compromises the application of these data. Under certain circumstances, however, when the geological and sedimentary settings rule out enough such complications, bulk carbonate isotopic records can faithfully reproduce trends from a single foraminiferal species (Shackleton et al., 1993; Schrag et al., 1995); bulk carbonate has also been used for stable isotope analysis. Moreover, bulk isotopic analysis becomes the sole resort where the samples are too well lithified to allow separation of foraminifers, where foraminifers are sparse (which is the case here), and/or where very high sample resolution is desired over long intervals.
Site 1256 on the Cocos plate is separated from the west coast of Central America by the Middle American Trench, which traps most terrestrial sediments shed from the continent. The sedimentary sources, therefore, are mainly biogenic calcite and silica as indicated by the relationship of MARs between bulk sediments and the main sedimentary components (Fig. F3). These biogenic materials are produced almost entirely in surface waters, and the benthic component is minor (e.g., van Andel et al., 1975). Furthermore, smear slide examinations have shown that the sediments contain highly abundant calcareous nannofossils, abundant siliceous microfossils, and extremely rare foraminifers (Shipboard Scientific Party, 2003; Jiang and Wise, this volume). The survival and preservation of calcareous nannofossils are likely linked to fecal pellet transport through the water column, where organic coatings mitigate coccolith dissolution in the water column and at the seafloor (Honjo, 1976). Thus, calcareous nannofossils contributed the great majority of the carbonate and consequent stable isotopic signals. We recognize that bulk 13C can be affected by detrital input such as from volcanogenic sources; however, such inputs were negligible in our study area at the time of the carbonate crash, as indicated by the nearby Caribbean sites (Fig.
F4).
The influence of different vital effects of calcareous nannofossil species is expected to be minor. Stoll (2005) showed that nearly monogeneric nannofossil isotopic records closely parallel those from bulk carbonate, that the interspecific vital effect in nannofossil isotopic measurements is small, and that variable nannofossil assemblages do not significantly bias the isotopic records. Schrag et al. (1995) further demonstrated that bulk isotope data are less sensitive to compositional variations. In such a case where the nannofossil assemblage is dominated by reticulofenestrids before, during, and shortly after the carbonate crash (Jiang and Wise, this volume), bulk carbonate isotopes can be reliably used to investigate the cause of this event.
The dominant lithology in Hole 1256B is unconsolidated calcareous nannofossil ooze (Shipboard Scientific Party, 2003), consisting mainly of nannofossil skeletons of low-Mg calcium carbonate resistant to dissolution. Foraminifers are very rare and so are insufficient for isotopic analysis. Low-Mg carbonate is thermodynamically stable in deep, cold seawater and resistant to diagenetic alteration after deposition (Schlanger and Douglass, 1974); therefore, pelagic oozes remain virtually unlithified until buried to a certain depth, at which point CaCO3 is released to the pore waters and reprecipitates as intraparticle fill, exterior overgrowth, and intergranular cement (Garrison, 1981).
Diagenetic alteration, including dissolution and postdepositional diagenesis, has potential for altering isotopic signals originally preserved in carbonates. Dissolution preferentially removes 13C, resulting in a depletion of 0.2
, as observed in benthic foraminifers (McCorkle et al., 1995), though theoretically higher depletion may occur under Rayleigh distillation. Burial diagenesis, increasing with increasing CaCO3 content and burial depth, and diagenetic alteration by underlying basalts tend to progressively deplete heavier 18O (Frank and Bernet, 2000; Schrag et al., 1992). Thus, severe diagenetic alteration produces isotopically lighter calcite with well-coupled stable oxygen and carbon values.
Carbon isotope data likely have been little affected by diagenetic alteration based on the following observations:
Oxygen isotope values are prone to diagenetic alteration during burial diagenesis because oxygen isotopes show significant temperature-dependent fractionation (Anderson and Arthur, 1983; Marshall, 1992). However, Schrag et al. (1995) demonstrated that the effect of rapid calcite precipitation is small for the biogenic carbonates from mid-latitude Atlantic because primary oxygen isotope values in carbonates are close to isotopic equilibrium with cold pore fluids. As today, Site 1256 was located during the middle Miocene beneath the equatorial divergence and under a strong influence of upwelling of cold deep water, so the sea-surface temperatures (SSTs) are considered to be comparable to those of the temperate Atlantic as indicated by ocean SST pattern. The sediments at this site are shallowly buried (250.7 m), and hydrothermal circulation is no longer a major mechanism of heat transport (Shipboard Scientific Party, 2003). Therefore, oxygen isotope data mainly reflect environmental signals.
Paleoproductivity reconstruction is a major area of paleoceanographic research because it places important constraints on past ocean circulation, nutrient distribution, and oceanic carbon–cycle history. Important proxies have been developed for past productivity and interpreted in terms of organic carbon export. These proxies include direct measurements of organic carbon in sediments, biogenic opal and calcium carbonate accumulation, foraminiferal assemblage data, geochemical tracers, and 13C fluctuations (e.g., Müller and Suess, 1979; Berger et al., 1989; Herguera and Berger, 1991; Paytan et al., 1996; Tappan, 1968; respectively). In this study we employ CaCO3 MARs and 13C values as proxies to assess changes in paleoproductivity (Fig. F3).
As the major components in the sediments are biogenic skeletons of organisms dwelling in the uppermost water column (Fig. F3), the geochemical composition of the sediments predominantly reflects a surface water signal. Thus, CaCO3 MARs reflect carbonate production by surface-dwelling, calcite-secreting organisms when dissolution is a minor effect.
Interpretation of 13C, however, depends on the specific geological and oceanographic setting. The bulk carbonate in this study was produced nearly exclusively in the surface water; therefore, variations in the bulk 13C values should reflect fluctuations therein, which are the product of the interplay between phytoplankton photosynthesis and surface water chemistry as influenced by seawater alkalinity and the dissolved carbon 13C/12C ratio. Phytoplankton dwell in the oceanic euphotic layer and preferentially take up 12C during photosynthesis, producing organic matter with 13C values from –20 to ~23 and a relatively 13C enriched, dissolved inorganic carbon pool in surface waters. Seawater pH influences the concentrations of different forms of dissolved inorganic carbon, the 13C values of which decrease with increasing seawater CO32– (Spero et al., 1997).
In the present study, the close coupling between the CaCO3 MARs and 13C values prior to 5 Ma and decoupling thereafter (Fig. F2B) reflect a switch of dominance between carbonate production and dissolution. Prior to 5 Ma, carbonate production controlled the sedimentary patterns at Site 1256 and 13C variations predominantly reflected changes in the standing stock of carbonate-producing organisms in the surface waters (Figs. F2, F3). After 5 Ma, dissolution dominated this site, probably resulting from the effective blocking of the Panama Seaway since then (Haug and Tiedemann, 1998).
The large excursions observed in 13C values could not have arisen from enhanced dissolution. Dissolution occurs when there is a reduction in CO32– in the surface water, the water column that the carbonates travel through, and the bottom water. Changes in seawater CO32– in intermediate and deep waters, however, do not significantly alter the 13C signal of carbonates produced, as observed in foraminifers (McCorkle et al., 1995). If such reduction in CO32– occurred in surface waters, it would have greatly increased the 13C and 18O values of carbonate (Spero et al., 1997), a situation that is opposite from the observations here (Fig. F2). Therefore, no matter where severe dissolution occurs, it could not produce the negative excursions in 13C and 18O values observed in this study. 13C values faithfully represent variations in the surface water standing stock of calcite producers.
CaCO3 sedimentation at Site 1256 shows several extreme lows between 12 and 8 Ma, which were initiated at ~11.3 and 10.6 Ma, geologically synchronous with the onset of the carbon isotope excursions (Fig. F2). The deepest drop is at 9.6 Ma, with carbonate accumulation virtually ceasing. This is temporally comparable with the "nadir" of the carbonate crash at other ODP/Deep Sea Drilling Project (DSDP) sites in this region (Lyle et al., 1995; Farrell et al., 1995), whereas the five carbonate minima in the Caribbean occurred ~1 m.y. earlier (12–10 Ma) (Roth et al., 2000). These phenomena were previously ascribed to enhanced carbonate dissolution as a consequence of changes in deepwater circulation (e.g., Lyle et al., 1995; Roth et al., 2000).
On the ocean floor, dissolution of calcium carbonate in seawater is determined by seawater pH values influenced by temperature, pressure, and the partial pressure of CO2. The depth of the CCD varies in different ocean basins depending on bottom water chemistry and carbonate supply from surface water (Wise, 2003). The former explains a present-day CCD in the Pacific shallower than in the Atlantic, the latter a shallower CCD in high latitudes where carbonate production is low. High biological productivity tends to stimulate higher rates of carbonate production and to depress the CCD. However, if high productivity is engendered by extremely high concentrations of nutrients, a condition that favors diatoms and dinoflagellates, the CCD tends to shoal as a result of addition of CO2 to the bottom water and/or acid production resulting from the degradation of organic matter (Emerson and Bender, 1981; Archer, 1991a, 1991b). This scenario may occur in response to changes in climate (Dymond and Lyle, 1985).
Multiple causative mechanisms have been proposed for the carbonate crash. Severe dissolution in the equatorial Pacific and Caribbean has been attributed to changes in bottom water chemistry. The underlying hypotheses are either
Although the "nadir" of the carbonate crash occurred ~1 m.y. earlier in the Caribbean than elsewhere, the comparable nature and time overlap of its occurrences suggest a common cause associated with changing oceanic circulation (Roth et al., 2000). This does not exclude other competitive mechanisms, such as "dilution" by terrigenous sediments (Diester-Haass et al., 2004) or an opal component (Westerhold et al., 2003; Böhm and Dullo, 2000) for those "carbonate crashes" reported elsewhere (i.e., the southern and eastern South Atlantic, southern Indian Ocean).
Phytoplankton community restructuring against calcite-producing organisms could also produce severe reduction in CaCO3 MARs, even though the overall surface water productivity remains constant (Dymond and Lyle, 1985). At Site 1256, the negative excursions in 13C and 18O coincide prior to the carbonate crash. At this time, an oxygen isotope signal is supposed to record a combination of surface temperature, ice volume, and seawater chemistry, the latter two of which tend to enrich heavier isotopes in response to the development of the East Antarctic Ice Sheet (EAIS) (Zachos et al., 2001) and, if any, the dissolution-related decrease in seawater CO32–. Thus, 13C excursions observed at Site 1256 likely represent sudden SST increases, a condition that favors warm-water specialists/calcite-producing organisms (McIntyre and Bé, 1967; McIntyre et al., 1970). Similar observations, especially the coeval excursion in 13C and 18O values at 11.3 Ma, have also been documented in other nearby ODP/DSDP sites (Shackleton and Hall, 1984, 1995) and in the Caribbean (Mutti, 2000). In other words, these 13C excursions represent a sharp decrease in standing stock of calcareous phytoplankton in surface waters.
Evidence for increased dissolution across the carbonate crash includes a decreased coarse calcareous fraction, increased benthic/planktonic foraminifer ratios, and deteriorated preservation of calcareous fossils. These indirect proxies should be cautiously placed into a sedimentary and geochemical context. The drop in the sand-sized fraction recorded at Sites 998 and 999 was attributed to enhanced fragmentation of foraminiferal tests in more corrosive water columns (Roth et al., 2000), whereas the same phenomenon observed in the southeast Atlantic (Sites 1085 and 1087) was related to sea level regression and consequent increased terrestrial input (Diester-Haass et al., 2004). The increased corrosiveness of the water column itself, however, does not necessarily demand an influx of corrosive waters because a reduction in carbonate production in the surface water would have the same effect. The increase in the ratios of benthic to planktonic foraminifers in the southwest Atlantic seems to be controlled more by nutrient condition than dissolution (Diester-Haass et al., 2004). At Site 1256, the calcareous nannofossil assemblages exhibit their best preservation just prior to the carbonate crash nadir (Jiang and Wise, this volume). This does not necessarily mean no dissolution occurred during the two dramatic drops in carbonate MARs because the preservation is closely associated with the presence and abundance of diatoms, which release silica to pore waters and thus inhibit dissolution and/or precipitation of calcite during diagenesis (Wise, 1977).
Constriction of the Panama Seaway in the late middle Miocene limited communication between the Atlantic and Pacific at intermediate- and deep-water levels (Duque-Caro, 1990) and could have caused basin-to-basin isotopic fractionation (Lyle et al., 1995; Roth et al., 2000), a scenario seen today. The modern Atlantic is filled with young, well-oxygenated water as a result of the production of NADW, whereas the Pacific has older water originating from the North Atlantic and southern high latitudes. As these waters age, the organic matter therein decomposes and releases 12C-depleted carbon into seawater, which is responsible for the more negative 13C values in eastern Pacific relative to the western Atlantic (Fig.
F5) (Kroopnick, 1985). However, the following perceptions raise several questions regarding this mechanism.
It takes ~1500 yr for deep water sinking in the North Atlantic to reach the North Pacific (Manighetti, 2001), which cannot explain the 1-m.y. lead for the carbonate crash in the Caribbean. The water brought up to the surface by equatorial divergence-driven upwelling has a shallow-water source above the thermocline with a mean depth of 50 m (Vossepoel et al., 1999). In fact, no matter where the ultimate source of these waters is, the North Atlantic or circumpolar Antarctic, the 13C difference between these sources today and the eastern Pacific is no more than 0.7 (Fig.
F5). This alone could not have caused the 13C excursions of >1.6 accompanying the carbonate crash documented at Site 1256. The carbonate crash was observed mostly from records retrieved from >3000 m water depth, a depth level that is highly sensitive to fluctuations in the CCD in the eastern equatorial basins (Lyle et al., 1995). The previously proposed mechanisms can hardly account for its occurrences at very shallow sites (e.g., Sites 1000 and 1241) in water depths well above the modern CCD, hence, the least susceptible to the influence of shoaling CCDs.
As a matter of fact, carbonate production played a dominant role over dissolution as evidenced by the close parallelism in CaCO3 MARs and 13C values in the 14- to 5-Ma interval and by the absence of deterioration in preservation of calcareous nannofossils during the crash (Raffi and Flores, 1995; Jiang and Wise, this volume). In other words, the crash was not a dissolution event but a low-productivity event. This conclusion is in line with the lack of basinwide occurrences of the crash in the Pacific (Lyle, 2003).
Based on these observations and discussions above, we speculate that a reduction in the standing stock of carbonate-producing organisms, likely induced by a reduction in nutrient availability, triggered the widespread carbonate crash at the middle/late Miocene boundary.
The model proposed here emphasizes that carbonate supply to intermediate and deep waters is primarily a function of changes in carbonate production and, as a consequence, affects the corrosiveness of these waters on carbonate and preservation of carbonate. The change of carbonate production is associated with changing global surface water circulation, which, in turn, is induced by the seaway-controlled interocean exchange of water masses.
During the late Cenozoic, prior to the segmentation of the once virtually continuous circumequatorial current, nutrient-rich intermediate water was tapped out to the surface, where it sustained high biological productivity along the equator. This flow began to fragment because of the uplift of the Panama Isthmus at 12.9–11.8 Ma (Duque-Caro, 1990) and the closure of the Indonesian Seaway at 12–11 Ma (Keller, 1985; Romine and Lombari, 1985; Kennett et al., 1985). The latter occurred when subduction developed on the east and west sides of New Guinea (Hall, 2001). The difference in timing can explain the ~1-m.y. lead of the carbonate crash in the Caribbean relative to the Pacific and Indian Oceans.
Although the Indonesian Seaway was effectively blocked at ~17–15 Ma, Indo-Pacific communication was still possible through small passages during times of high sea level (Nishimura and Suparka, 1997). This transport of surface water, termed the Indonesian Throughflow (ITF) today, carries warm and fresh tropical-Pacific surface water through this passage into the Indian Ocean and creates a warm pool in the western Indian Ocean. Global sea level drop at the middle/late Miocene boundary (Haq et al., 1987; Sen et al., 1999) may have completely blocked this Pacific-to-Indian surface water transport, switching the locale for warm water accumulation to the western Pacific, creating a greater warm pool there. The eastward spread of the warm-pool water strengthened the EUC system (Kennett et al., 1985), resulting in warm SSTs in the central and eastern Pacific and reduced equatorial upwelling of colder subsurface water, both of which contribute to the creation of a 18O excursion. The switch of dominance to nutrient-poor warm surface water caused a sudden reduction in biological productivity, which is represented by a 13C excursion. This event is analogous to today's El Niño. A similar scenario may have occurred in the Caribbean and Atlantic when the Panama Seaway suddenly became restricted at 12.9–11.8 Ma.
Because blockage of the ITF cut off Pacific-to-Indian Ocean heat transport, this should have triggered an overall warming in the tropical Pacific and cooling in the southern Indian Ocean. This is evidenced by an abrupt disappearance of foraminiferal and radiolarian provincialism across the equatorial Pacific (Keller, 1985; Romine and Lombari, 1985; Kennett et al., 1985) and a positive 18O excursion in shallow-dwelling planktonic foraminifers at DSDP Sites 216 and 237 (Vincent et al., 1985). A temporally similar warming of the entire water column was recorded at DSDP Site 289 (Gasperi and Kennett, 1993). The eastern spread of warm surface water warmed not only the east equatorial Pacific, but also the southeast Pacific off Chile (Tsuchi, 1997). These perceptions are consistent with the results of a near-global ocean general circulation model for the circulation and thermal structure of the Pacific and Indian Oceans with open and closed Indonesian passages from 1981 to 1997 proposed by Lee et al. (2002).
It is worth mentioning that the evolution of global climate entered a full ice-house mode during the Neogene (Zachos et al., 2001). Cold climate narrows carbonate producing areas, rendering carbonate production in tropical oceans more important; small changes in production therein greatly affect total carbonate supply to deep waters. After full development of the EAIS by the middle Miocene (Shackleton and Kennett, 1975; Kennett et al., 1985; Woodruff and Savin, 1991; Zachos et al., 2001) (North Hemisphere glaciations did not begin until ~7 Ma [Fronval and Jansen, 1996]), the southern trade winds began to prevail over their northern counterpart, causing a northward shift of the ITCZ, an analog of the initiation of modern North Hemisphere summers. The intensification of southeastern trade winds during the Neogene is consistent with the coarsening of eolian grain size recorded from the subtropical South Pacific (Rea and Bloomstine, 1986). Positions farther north than present (5°N) of 10°–12°N, ~22°–24°N, and 13°–14°N have been suggested for the ITCZ around the middle/late Miocene boundary by Flöhn (1981), Rea (1994), and Shipboard Scientific Party (2002), respectively. This northward shift of the ITCZ can potentially wash out eolian dust at the northern mid-latitudes, mitigating the iron-limited condition and stimulating biological production. This scenario is consistent with the records from North Pacific (Snoeckx et al., 1995).
The eastern equatorial Pacific is presently an HNLC region because of iron limitation on phytoplankton (Martin et al., 1991). This region did not suffer from iron deficiency during the early and middle Miocene when there was intense activity in circum-Caribbean and Central American volcanism as evidenced by tephra accumulation rates (Sigurdsson et al., 2000), but it has since then. Once exposed to seawater, volcanic ash can fertilize the open ocean by releasing large amounts of macronutrients and bioactive trace metals (Frogner et al., 2001).
According to our model, with the closure of the Indonesian Seaway by sea level drop at the middle/late Miocene boundary, warm surface water began to pile up in the western Pacific. The formation of a larger warm pool strengthened the EUC system. The eastward spread of this water triggered a prominent El Niño, reducing upwelling and then nutrient availability. At the same time, prevailing southeastern trade winds across the equator deflected the delivery of volcanic ash from Central America and the circum-Caribbean, where coincidently volcanism was reduced sharply (Sigurdsson et al., 2000). The decrease in availability of macronutrients and micronutrients resulted in a drastic reduction in primary productivity in the eastern and central equatorial Pacific as well as the Caribbean, as evidenced by the carbon isotope excursions at Site 1256 and the synchronous variations in MARs of carbonate and volcanic ash throughout the Neogene at Sites 998 and 999 (Fig. F4).
The CCD in the Neogene Pacific is a product of the balance between regional production and basinwide dissolution (Lyle, 2003). The modern deep Pacific below 1500 m is essentially a single water mass (Joyce et al., 1986; Talley and Roemmich, 1991); therefore, the entire Pacific Ocean floor is bathed by the same water mass and carbonate dissolves primarily at the ocean floor (Edmond, 1974; Walsh et al., 1988). On one hand, dissolution rates should be similar everywhere and there should be a base-level CCD for all regions in the Pacific. The variable depth of the CCD in the Pacific, however, originates from biogeographic differences (Lyle, 2003). That is, although there is a base level, the CCD may be depressed or elevated depending on surface productivity. On the other hand, although there has always been during the Neogene an Antarctic deepwater source into the Pacific Basin, no convincing evidence exists for stronger flow at the middle/late Miocene boundary (Lyle et al., 1995). The consequent near-constant deepwater chemistry, plus the close correlation between CaCO3 MARs and biological productivity, suggest that the carbonate crash in the eastern equatorial Pacific and likely in other regions is not a dissolution event, but one of low productivity.
