DISCUSSION

In the late middle Miocene, dramatic change in the nature of water masses filling the Caribbean basins was recorded by five episodes of greatly reduced carbonate accumulation (Fig. 11, Fig. 12). Because these five episodes, identified by a decline in the carbonate content and MAR, are characterized by a decreased preservation of planktonic foraminifers and usually a smaller coarse-fraction proportion, they are interpreted to correspond to carbonate-dissolution episodes triggered by a major oceanic perturbation in the Caribbean that lasted ~2 m.y. (from ~12 to 10 Ma). The interval including the five carbonate-dissolution episodes is referred to as the Caribbean carbonate crash. While the term "carbonate crash" was borrowed from ODP Leg 138 published results (Lyle et al., 1995), we also acknowledge here that the respective timing of the carbonate-dissolution interval observed in the Pacific and Caribbean at the middle to late Miocene transition did not fully overlap. The third of the five carbonate Caribbean dissolution episodes (Fig. 11, Fig. 12) appears to be contemporaneous with the onset of the main carbonate dissolution in the eastern equatorial Pacific Ocean (Fig. 1, Fig. 15). In this discussion, we will attempt to show that the interplay between the seaway opening along the northern Nicaraguan Rise and the gradual closure of the Central American Seaway resulted in the initiation of the Caribbean and Loop Currents, the strengthening of the Gulf Stream, and, as a consequence, the re-establishment and intensification of the NADW production. We will try to demonstrate that this scenario can explain the simultaneous occurrence of the carbonate crash on either side of the Isthmus of Panama.

The results of this study show that the Caribbean carbonate crash reaches its zenith during an interval between 12 and 10 Ma when the connection between the southern and northern Caribbean basins (Colombian and Yucatan Basins) was finally established through the opening of several seaways along the northern Nicaraguan Rise and the contemporaneous gradual closing of the Central American Seaway to intermediate depths (Fig. 2, Fig. 4, Fig. 5). The temporary complete closure of the Central American Seaway between ~10.0 and 9.5 Ma (Fig. 2), possibly tied to a major lowering of sea level during this interval (Haq et al., 1987) (Fig. 13A, B), appears to correspond to the zenith of the carbonate crash in the eastern equatorial Pacific (Fig. 1, Fig. 15D) and, in the deep Caribbean basins, to some relatively heavy ¹⁸O values (Fig. 13), the lightest benthic ¹³C (Fig. 14), and an interval characterized by the sustained lowest aragonite MAR in Hole 1000A (Fig. 12D). Surprisingly, during the same time interval, the carbonate system had already fully recovered in the Caribbean basins. In this discussion, the previous models to explain the carbonate crash in the eastern equatorial Pacific Ocean (Lyle et al., 1995) are summarized. Then, a model modified from the one published by Lyle et al. (1995) will be proposed to explain the occurrence of the carbonate crash on both sides of the Isthmus of Panama. This model is based on an analogy with a carbonate model proposed by Haddad and Droxler (1996) to explain the late Quaternary glacial/interglacial carbonate preservation pattern in the Caribbean basins.

Lyle et al. (1995) attributed the carbonate crash observed in ODP Leg 138 Neogene pelagic sequences of the eastern equatorial Pacific (Fig. 1, Fig. 15D) to the emergence of the Isthmus of Panama. Although the isthmus was still below sea level for most of the time of the crash, they calculated that a restriction of 2 Sv (1 Sv = 1 x 10⁶ m³/s) of carbonate-rich deep and intermediate water masses from the Atlantic to the Pacific would account for the loss of carbonate accumulation on the equatorial part of the East Pacific Rise and west of the isthmus. However, because this scenario does not restrict the flux of carbonate-rich deep and intermediate water masses in the Caribbean Sea, the occurrence of the carbonate crash in the Caribbean basins, discovered subsequent to Lyle et al. (1995) by Sigurdsson, Leckie, Acton, et al. (1997), cannot be explained by this model.

Lyle et al. (1995) also proposed a second model to explain the carbonate crash in the eastern Pacific. This latter model involves changing the global deep-ocean circulation triggered by the onset of deep- water production in the high latitudes of the North Atlantic Ocean, a scenario analogous to the modern oceanographic setting of deep-water circulation. Accordingly, the initiation of NADW would cause a reorganization of deep-water circulation and would affect the carbonate preservation in the eastern equatorial Pacific. The initiation and the strengthening of NADW flow would compete with the Antarctic Bottom Water (AABW) in the South Atlantic, resulting in the displacement of some of the more corrosive AABW toward the Pacific, triggering more dissolution in the eastern equatorial Pacific. Lyle et al. (1995) were in favor of the closing of the Central American Seaway as the mechanism responsible for the carbonate crash at the middle to late Miocene transition. The role of the NADW establishment and intensification in influencing the carbonate preservation and accumulation in the eastern equatorial Pacific Ocean was not clear prior to the discovery that the carbonate crash occurred at about the same time on both the Caribbean and Pacific sides of the Isthmus of Panama (Sigurdsson, Leckie, Acton, et al., 1997). The partial and perhaps temporary complete closure of the Central American Seaway at the middle to late middle Miocene transition, in addition to decreasing the flow of carbonate-rich intermediate waters from the Atlantic to the Pacific, may also have directly influenced the NADW production. Constraining the timing and observing the pattern of carbonate accumulation on both sides of the Isthmus of Panama should help us to develop a scenario to explain the occurrence of the carbonate crash at the middle to late Miocene transition.

The model proposed here to explain the occurrence of the Caribbean carbonate crash draws on the circulation changes induced by the re-establishment and intensification of the NADW (or its precursor, NCW). We propose that the carbonate crash in the Caribbean Sea and the eastern equatorial Pacific Ocean resulted from a global reorganization of the thermohaline oceanic circulation at the middle to late Miocene transition. The re-establishment (Fig. 15E) (Wright and Miller, 1996) and/or the initiation (Wei and Peleo-Alampay, 1997) of the NADW production at this time caused an influx of corrosive AAIW entering the Caribbean basins and ultimately resulted in dramatic seafloor dissolution of calcareous sediments. This hypothesis is built upon an analogy with the late Quaternary glacial to interglacial perturbations of the global thermohaline circulation and their related results in terms of carbonate sediment accumulation in the Caribbean basins (Haddad and Droxler, 1996). Moreover, we further propose that the middle Miocene drowning of carbonate banks in the northern part of the Pedro Channel and Walton Basin along the northern Nicaraguan Rise (Fig. 4, Fig. 5) was contemporaneous to the partial closing of the Central American Seaway, and thus would have played a significant role in triggering the global reorganization of the oceanic circulation (Droxler et al., 1998).

Our model for the Caribbean carbonate crash proposes that the re-establishment and intensification of the NADW in the late middle Miocene (Wright and Miller, 1996) were triggered by the opening of seaways along the northern Nicaraguan Rise, creating a connection between the Colombian and Yucatan Basins (Fig. 2B, Fig. 5) (Droxler et al., 1998). These events were contemporaneous with the closure at intermediate- and deep-water levels of low-latitude seaways connecting the Atlantic and the Pacific oceans (Fig. 2) (Duque-Caro, 1990). Prior to significant NADW production in the early middle Miocene (Fig. 15E) (Wright and Miller, 1996), AABW was the main source of deep water and, because it formed in the Southern Ocean. AABW could directly fill all three major ocean basins as far north as the north latitudes of the Atlantic and Pacific oceans. Once the production of NADW was established and subsequently intensified, large volumes of deep waters were formed in the North Atlantic. Without a low-latitude connection at bathyal and abyssal depths between the Atlantic and the Pacific, the NADW had to travel the length of the Atlantic and around the perimeter of Antarctica before entering the Pacific. This quasi-unidirectional flow, as the modern thermohaline conveyor, resulted in a case of basin-to-basin fractionation between the carbonate-rich Atlantic and carbonate-poor Pacific oceans (Berger, 1970a; Gordon, 1986; Broecker et al., 1990).

Despite NADW's chemistry, which promotes calcium carbonate preservation, carbonate sediment is dissolving in the Caribbean today when the production of NADW is at its maximum (Fig. 6A, Fig. 16A). We propose that this is as it was during the Caribbean late middle Miocene carbonate crash, a time when NADW was newly being reestablished (Wright and Miller, 1996). The changes of oceanic circulation at the transition from the middle to late Miocene can be illustrated by using the late Quaternary glacial-interglacial circulation and carbonate preservation patterns in the Caribbean as an analogue to "pre-carbonate crash"-"height of carbonate crash" periods. During the late Pleistocene glacial stages (Gordon, 1986; Broecker et al., 1990; Raymo et al., 1990), as it may have been during the early middle Miocene (Wright et al., 1992; Wright and Miller, 1996), the production of NADW was significantly reduced or had completely ceased and calcium carbonate was preserved in both the eastern equatorial Pacific Ocean and the Caribbean Sea (Fig. 16A) (Lyle et al., 1995; Le et al., 1995; Sigurdsson, Leckie, Acton, et al., 1997; Haddad and Droxler, 1996). The water mass entering the Caribbean over bathyal sill depths during the last glacial maximum and perhaps during the pre-crash early middle Miocene interval was a well-oxygenated and relatively heavy ¹³C glacial North Atlantic Intermediate Water (Slowey and Curry, 1995; Haddad and Droxler, 1996) (Fig. 6B, Fig. 16). With the NADW being significantly reduced, the Southern Ocean then dominates deep-water production (Fig. 6B). As a consequence, the eastern equatorial Pacific deep waters were relatively young and well oxygenated and consequently noncorrosive to carbonate (i.e., Le et al., 1995, and references herein)

In the Holocene, as may also have been the case at the nadir of the middle to late Miocene carbonate crash, especially during the five episodes of massive carbonate dissolution in the Caribbean basins (Fig. 11, Fig. 12), NADW was well developed, and carbonate dissolution is observed on both sides of the Isthmus of Panama in the eastern equatorial Pacific and in the Caribbean Sea during an interval from ~11 to 10 Ma (Fig. 15). Using a composite dissolution index based on metastable carbonate, Haddad and Droxler (1996) linked late Quaternary increases of NADW production with greater dissolution of calcium carbonate in the Caribbean Sea (Fig. 16A). The Caribbean carbonate crash between 12 and 10 Ma is interpreted to be related to a period of greater NADW production (Wright and Miller, 1996). When NADW is being produced, large volumes of water, equivalent to the amount sinking in the Norwegian and Labrador Seas and exiting the North Atlantic Ocean toward the Southern Ocean, are pulled northward through the Caribbean in the upper 1500 m of the water column to replenish the sinking water (Fig. 6A) (Schmitz and McCartney, 1993). This brings southern-sourced intermediate waters farther north during interglacial times than during glacial stages (Fig. 6). The carbonate-corrosive AAIW entrains the upper NADW when entering over the Caribbean sill depth. These waters fill the basins, causing increased dissolution in the Caribbean (Fig. 16) (Haddad and Droxler, 1996). In times of increased NADW production, such as during the Holocene-last interglacial Stage 5 and possibly at the middle to late Miocene transition, the eastern equatorial Pacific is farther from the northern source of deep-water production, causing its carbonate chemistry to be dominated by older, corrosive waters enriched in CO₂. NADW production can be associated, therefore, with dissolution in the eastern equatorial Pacific Ocean (corrosive bottom/deep waters) as well as in the Caribbean basins (corrosive intermediate waters).

Maier-Reimer et al. (1990) use general circulation models (GCM) to show how changes in thermohaline circulation and NADW production can be tied to partial or full closure of the Isthmus of Panama. In their modern model with an emergent Isthmus of Panama, NADW is produced. In a second scenario, with a modeled open Central American Seaway, NADW production is inhibited. The GCM of Maier-Reimer et al. (1990) appears to demonstrate that a deeply submerged isthmus would be linked with a lack of NADW production. Mikolajewicz and Crowley (1997) supplemented those results with more GCM experiments that support NADW production to some degree already with a partially emergent isthmus, as is thought to be the case in the middle to late Miocene transition. So far, no models have been developed to demonstrate the possible effect of channel opening across the northern Nicaraguan Rise.

The modern thermohaline circulation pattern, characterized by deep-water formation in high latitudes of the Atlantic Ocean or North Component Water (NCW), was initiated as early as the late early Miocene (Fig. 15E) (Wright and Miller, 1996), but certainly by the late middle Miocene (Woodruff and Savin, 1989; Wei, 1995; Wei and Peleo-Alampay, 1997), with an interval in the early middle Miocene when the NCW was temporarily inhibited (Wright and Miller, 1996). The onset of the Caribbean carbonate crash, characterized by dramatic decreases in carbonate accumulation at Sites 998, 999, and 1000, appears to be synchronous with the main intensification (re-establishment) of the modern thermohaline circulation and may be directly linked to the reorganization of the oceanic water masses flowing through the Caribbean basins (Fig. 15, Fig. 17). An intensification of NADW at this time may be the result of the opening and partial closure of gateways in the Caribbean Sea during the middle Miocene.

Northern Nicaraguan Rise

One feature in the Caribbean thought to be prominent during the Oligocene to early Miocene is a series of carbonate banks and barrier reefs that spanned the distance from Nicaragua and Honduras to Jamaica (Fig. 4, Fig. 5) (Droxler et al, 1989, 1992, 1998; Lewis and Draper, 1990; Cunningham, 1998). This is the location of the modern northern Nicaraguan Rise (Fig. 3). The presence of these neritic banks during the early Miocene would have served as a barrier to northward water transport and would have also enhanced westward tropical flow between the Caribbean and the eastern Pacific (Fig. 5). Coccolith assemblages at Sites 998 and 999, north (Yucatan Basin) and south (Colombian Basin) of the northern Nicaraguan Rise, respectively, show minimal connection in the surface circulation between those two basins during nannozones CN3 and CN4 (16.2-13.57 Ma) (Fig. 2B) (Kameo and Bralower, Chap. 1, this volume; Kameo and Sato, in press). This observation supports the idea of a barrier impeding any significant surface flow over the northern Nicaraguan Rise in the early middle Miocene. Partial foundering of this barrier because of faulting, linked to rifting intensification in the Cayman Trough, may have started in the mid-Oligocene but mostly occurred in the early middle Miocene. Cunningham (1998) places the initiation of tectonic activity and mini-basin formation in the Pedro Channel area at 16-11 Ma (ages of Raffi and Flores, 1995). This activity may also be related to the change from a relatively long period of quiescence on the northern Nicaraguan Rise to the uplift of Jamaica in the late middle Miocene (Leroy et al., 1996). The demise of carbonate neritic banks in the northern part of Pedro Channel and the central part of Walton Basin has led to the observed modern configuration of shallow, carbonate banks segmented by north-south oriented channels (Fig. 4) (Cunningham, 1998; Droxler et al., 1998). The merging of coccolith assemblages between Sites 998 and 999 (Fig. 2B) (Kameo and Bralower, Chap. 1, this volume; Kameo and Sato, in press), was first initiated during nannozone CN5 (13.57-10.71 Ma) and was fully completed during nannozones CN6 and CN7 (10.71-9.36 Ma) and supports the estimated timing of a seaway opening along the northern Nicaraguan Rise.

Central American Seaway

Because of the episodic uplift history of the Panamanian isthmus (see Farrell et al., 1995), it is difficult to constrain the timing of the Central American Seaway closure. Duque-Caro (1990) dates the uplift to 1000 m (upper bathyal depths) at 12.9-11.8 Ma, based on benthic foraminiferal assemblages from onshore Colombian basins and the time scale of Keller and Barron (1983), which is equivalent to the 12-10.2 Ma of Raffi and Flores (1995) (Fig. 2A, Fig. 7). Kameo and Sato (in press) show that eastern Pacific coccolith assemblages remained identical to those of Site 999 in the Colombian Basin from 16.21-13.57 Ma, started to diverge at 13.57 Ma, and became very different between 10.71 and 9.36 Ma (Fig. 2B). Based upon the results of Duque-Caro (1990) and Kameo and Sato (in press), the flow of the upper water column was restricted through the Central American Seaway during the late middle Miocene while the surface mixed-layer flow through the seaway might have been temporarily closed for the first time sometime in the interval between 10.71 and 9.36 Ma at the middle to late Miocene transition. This temporary first closure of the Isthmus of Panama is recorded by the first intermingling of terrestrial faunas between the South and North American continents (Fig. 17) (Webb, 1985).

Establishment of the Caribbean Current

With the barrier across the northern part of the Pedro Channel subsided and the Isthmus of Panama partially uplifted, the Caribbean Current became established. Saline water from the southern Caribbean basins was for the first time transported northward into the Gulf of Mexico. The Loop Current, which connects the Caribbean water with the Gulf of Mexico, the Gulf Stream, and ultimately the northern North Atlantic (Fig. 6A), also became established at this time (15-12 Ma, Mullins et al., 1987; ages of Berggren et al., 1985, corresponding to ~14-11.5 Ma in the Raffi and Flores, 1995, time scale). The timing of the onset of the Loop Current in the Gulf of Mexico appears to correspond to the base of a large sediment drift in Santaren Channel, dated at ~12.4 Ma by Eberli et al. (1998) and thought to be linked to a major Gulf Stream intensification. Moreover, farther downstream along the western boundary current of the North Atlantic, Popenoe (1985) noted a pronounced unconformity in the late middle Miocene sediments of the Blake Plateau. The initiation of the sedimentary drift in the Santaren Channel and the unconformity on the Blake Plateau would correspond to the strengthening of the Gulf Stream in response to contemporaneous establishment of the Caribbean/Loop Current system. Timing of the re-establishment and intensification of NADW in the late middle Miocene appears to be contemporaneous to those changes along the North Atlantic western boundary current (Wright and Miller, 1996; Woodruff and Savin, 1989; Wei, 1995; Wei and Peleo-Alampay, 1997). The formation of the Caribbean/Loop Current system introduced for the first time a large volume of warm, saline waters from the Caribbean and the Gulf of Mexico to the high latitudes of the North Atlantic via the Florida Current and Gulf Stream. As established from the modern thermohaline circulation, these conditions are necessary for the formation of a deep-water mass in the high latitudes of the North Atlantic Ocean (Gordon, 1986; Schmitz and McCartney, 1993). If this scenario is correct, the Caribbean carbonate crash and the initiation of the Caribbean Current should be contemporaneous with the intensification of NADW.

In addition, prior to 12.1 Ma, the time of the onset of the Caribbean carbonate crash, a reduction of carbonate percent and accumulation rates is observed in the two southerly sites, Site 999 (Colombian Basin) and Site 1000 (Pedro Channel, northern Nicaraguan Rise) (Fig. 11, Fig. 12). We consider this early carbonate reduction as a precursor to the carbonate crash. The conspicuous absence of a carbonate-crash precursor in the northern Caribbean Site 998 (Yucatan Basin) may illustrate the isolation of the Yucatan Basin relative to the Colombian Basin and Pedro Channel prior to the carbonate crash itself. As noted earlier in our discussion, the presence of different coccolith assemblages on either side of the northern Nicaraguan Rise during the early middle Miocene (Fig. 2B) is another line of evidence to argue that the Yucatan Basin was isolated from the Colombian Basin and Pedro Channel prior to the onset of the Caribbean carbonate crash and that the Caribbean Current was not established at that time. The connection between the Colombian Basin, Pedro Channel, and the Yucatan Basin was probably only established by the first major carbonate-dissolution episode at ~12.1-12.0 Ma with the final subsidence of the reefal barrier in the northern part of Pedro Channel (Fig. 4, Fig. 5).

From 12.1 to 10.1 Ma, five episodes of carbonate dissolution are observed in all three Caribbean sites. During most of the five episodes, CO₃ MARs reach zero or slightly above zero in Holes 998A and 999A. These carbonate-dissolution episodes are likely caused by the influx of a southern-sourced, CO₂-rich, ¹³C-depleted intermediate water mass, equivalent to the current AAIW, into the Caribbean over bathyal sill depth.

Benthic ¹³C Water Mass Signature

The ¹³C interbasinal gradient that exists today reflects the sources and pathways of intermediate- and deep-water masses. As intermediate- and deep-water masses age, the lighter ¹²C isotopes (preferentially stored in organic tissues) are released in the waters through oxidation of the organic matter accumulating on the seafloor. This mechanism progressively lightens the ¹³C composition of intermediate- and deep-water masses until they are upwelled. Figure 18A shows the depth variations of the ¹³C values for the different oceans and in particular for the western Atlantic Ocean (Fig. 18B) (Kroopnick, 1980; 1985). A comparison between Figure 18B and Figure 6A shows that each individual intermediate- and deep-water mass displays characteristic ¹³C values. The oxygen-enriched and nutrient-poor NADW is characterized by relatively heavy ¹³C (>1.0) when compared to the relatively light ¹³C (<0.8) oxygen-poor and nutrient-enriched AAIW (Fig. 18).

Predating and postdating the carbonate crash and during the intervening intervals of relatively high carbonate accumulation between the five carbonate-dissolution episodes, heavier ¹³C values (usually >1.0) are generally observed (Fig. 14). These heavier ¹³C values perhaps reflect the influence of oxygen- and carbonate-enriched and nutrient-poor intermediate-water masses formed in North Atlantic Ocean. This suggestion is based upon the analogy we are making between the carbonate-preservation pattern observed during the carbonate crash at the middle to late Miocene transition and the late Quaternary glacial-interglacial cycles. Carbonate sediments in the Caribbean basins are preferentially preserved during late Quaternary glacial stages when benthic ¹³C values are generally heavy and production of NADW reaches minimum values (Fig. 16B) (Haddad and Droxler, 1996).

The temporal resolution of the stable isotope data in Holes 998A, 999A, and 1000A varied due to the absence of the benthic foraminifer Planulina wuellerstorfi in many analyzed samples, especially the ones with very low carbonate-content values. In spite of a limited benthic carbon isotope data set in our Caribbean crash study, light ¹³C values, indicative of a nutrient-enriched intermediate-water mass in the Caribbean basins, appear to occur during the five episodes of massive carbonate dissolution during the carbonate-crash interval (Fig. 12, Fig. 14). In Hole 998A, benthic ¹³C light values of ~0.6 occur at ~12 Ma, the time of carbonate reduction Episode I and between 10.7 and 10.4 Ma (Episode IV) (Fig. 14A). Light benthic ¹³C values in Hole 999A occur at ~11 Ma (Episode III), ~10.6 Ma (Episode IV), and ~10 Ma (Episode IV). When compared to reduced CO₃ MARs, the light values of benthic ¹³C in Hole 1000A (ranging between 0.4 and -0.3) occur at ~12 Ma (Episode I), ~11.4 Ma (Episode II), and ~11.0 Ma (Episode III). Light ¹³C values, tied to the AAIW influx within the Caribbean basins (Fig. 18B), are observed during the Holocene and last interglacial Stage 5e in the Caribbean basins, when severe carbonate dissolution occurred on the seafloor of the Caribbean basins and NADW production was at its maximum (Fig. 16).

Some of the lightest ¹³C values are outside of the Caribbean carbonate-crash interval, when the CO₃ MARs display some of the highest values during the middle to late Miocene transition. The intervals of 9.8-9.5 Ma (Hole 998A) and 10.0-9.5 Ma (Hole 999A) are characterized by ¹³C values as light as 0.1. We interpret the ¹³C fluctuations over the 2-m.y. period of the carbonate crash as driven by water mass changes. The low ¹³C values are tied to episodes when southern-sourced intermediate-water mass, corrosive toward carbonate sediments, fill in the Caribbean basins. Although sea-level fall at the very early part of the late Miocene may have caused the lighter ¹³C values observed between ~10 and 9.5 Ma, the occurrence of the lightest ¹³C values remains an enigma since their occurrence is not contemporaneous with poorly preserved carbonate sediments.

Intensification of NADW and a reorganization of the deep-water circulation would cause sedimentological changes in areas outside of the Caribbean. The link between the Caribbean carbonate crash and the re-establishment of the NADW may be better understood by observing the middle to late Miocene transition in sediments from other locations, such as the Ceara Rise (western tropical Atlantic; ODP Leg 154) and the eastern equatorial Pacific (ODP Leg 138) (Fig. 15). Evaluating how circulation changes may affect these three areas provides a way to test the effect of NADW re-establishment and intensification on the carbonate preservation in the Caribbean basins.

Leg 154 Ceara Rise: Evidence from the Equatorial Atlantic

Studies of ODP sites recovered from the Ceara Rise in the western equatorial Atlantic show that the sediments at the middle to late Miocene transition experienced similar decreases in carbonate content. Sites 925, 926, and 927 are located in water depths occupied by the NADW as determined by the data set of the Geochemical Ocean Sections Study for modern water conditions (King et al., 1997). The re-establishment and the intensification of NADW at the middle to late Miocene transition should produce conditions similar to those of the modern setting. The current overall good preservation of carbonate sediments in the area of the Ceara Rise is linked to the NADW flowing within most of the depth range of the rise and contrasts with the poor preservation of the carbonate sediments observed today in the Caribbean basins. One could expect that the preservation/dissolution pattern in the equatorial western Atlantic during the middle to late Miocene transition should be out of phase with the carbonate-preservation pattern observed in the Caribbean basins. During intervals when the NADW (or NCW) production was minimum or had ceased, such as during the early middle Miocene (Fig. 15E, Fig. 17) (Wright and Miller, 1996), the preservation of the carbonate sediment should have been minimum in the deep equatorial Atlantic Ocean and maximum in the Caribbean basins. In contrast, the opposite carbonate-preservation pattern between the deep equatorial Atlantic Ocean and the Caribbean basins should be observed at times when the NADW production was at its optimum during the late middle Miocene transition (Fig. 15E, Fig. 17) (Wright and Miller, 1996).

King et al. (1997) created synthetic records of the carbonate-content variations for the interval from 14 to 5 Ma of sites recovered from the Ceara Rise. Their results show an overall shoaling of the lysocline from 14.0 to 11.5 Ma, an interval partially predating the Caribbean carbonate crash but contemporaneous with the carbonate-crash precursor observed at Sites 999 and 1000, located south of the northern Nicaraguan Rise (Fig. 11, Fig. 12, Fig. 15). This observed lysocline shoaling contrasts with the contemporaneous overall high carbonate content and MAR in Hole 998A located on the northern side of the northern Nicaraguan Rise.

Coarse-fraction data from Site 926 on the Ceara Rise, recovered in a water depth of 3598 m, shows a smaller percentage of sand-sized material at ~13.5, 13.2-13.1, 12.6-12.4, 11.7-11.2, and 10.2-10.1 Ma (ages of Shackleton and Crowhurst, 1997, equivalent to Raffi and Flores, 1995) (Fig. 15A). Since dissolution is usually associated with a small percentage of coarse fraction and because the intervals of low proportion of coarse fraction correspond to intervals with low carbonate content, it appears that a dissolution component is recorded in the reduced carbonate content at Site 926 (Fig. 15A). At least three of the four low-carbonate intervals at 13.5, 13.1, 12.5-12.3, and 11.8 Ma (ages of Shackleton and Crowhurst, 1997), interpreted as punctuated episodes of lysocline shoaling in Hole 926A, match relatively well with intervals of relatively low values in carbonate content and accumulation rates in Hole 999A from the Colombian Basin during the carbonate-crash precursor (Fig. 15A, B). At the location of Site 926 on the Ceara Rise, the more corrosive AABW may have shoaled at times when the NADW had ceased or significantly decreased (Fig. 15E). Although some of these dissolution intervals predate the Caribbean carbonate crash and appear to be in phase with some intermediate carbonate values during the carbonate-crash precursor at Site 999, a long interval from 11.7 to 11.2 Ma characterized by very low coarse fraction and low carbonate content centered at ~11.4 Ma in Hole 926A corresponds to an interval of relatively high carbonate content and MAR at Sites 998 and 999 and is bounded by the two most intense Caribbean dissolution Episodes I and III at 12.0-11.8 and 11.1 Ma, respectively (Fig. 15).

Although the pattern of carbonate preservation between the equatorial Atlantic and the Caribbean basins during the middle and early late Miocene interval is not as clearly out of phase as the model would have predicted, it is still interesting to note that maximum carbonate dissolution occurred in the equatorial Atlantic before 11 Ma. In the middle Miocene interval prior to 12 Ma when the NADW was not fully developed (Wright and Miller, 1996), punctuated episodes of lysocline shoaling on the Ceara Rise match relatively well with intervals of carbonate dissolution in the Colombian Basin (southern Caribbean basins) during the carbonate-crash precursor and suggest that deep and intermediate waters in the equatorial Atlantic were somewhat connected and probably southern sourced and corrosive to carbonates (Fig. 15). The interval between 12 and 11 Ma, the first half of the Caribbean carbonate crash, was a transition period when the pattern of carbonate preservation became out of phase between the equatorial Atlantic and the southern and northern Caribbean basins. The two most intense episodes (I and III) of carbonate dissolution in the Caribbean basins correspond to times of good carbonate preservation in the equatorial Atlantic. Episode II, characterized by intermediate carbonate-dissolution intensity in the Caribbean basins, occurs slightly earlier than the peak dissolution at 11.4 Ma in the equatorial Atlantic. During the second half of the Caribbean carbonate-crash interval from 11 to 10 Ma, the equatorial Atlantic displays a contrasting overall good carbonate preservation. This interval is characterized in the eastern equatorial Pacific by a general decrease of the carbonate preservation, interpreted by Lyle et al. (1995) as the initial shoaling of the lysocline. (Fig. 15D)

Leg 138: Evidence from the Eastern Equatorial Pacific

When the Southern Ocean was the sole source of deep water, as in the early middle Miocene, the eastern equatorial Pacific was only half of an ocean basin away from the source. In this setting, the deep waters in the Pacific are expected to be relatively CO₂ poor and, therefore, to contain better preserved carbonate sediments. When the deep- and intermediate-water connection between the Atlantic and eastern Pacific ceased in the late middle Miocene because of the tectonic uplift of the Isthmus of Panama, the reorganization of the global thermohaline circulation, induced by the re-establishment and intensification of the NADW, located the eastern equatorial Pacific farther from the location of deep-water formation and at the end of the global conveyor belt. During times of NADW production, the establishment of a unidirectional long conveyor belt, such as the one occurring in the Holocene, would subject the eastern region of the Pacific to corrosive waters. In parallel, once deep waters are produced in the high latitudes of the Atlantic and the deep global conveyor belt is established, the Caribbean basins become the pathway for the return flow of the thermohaline circulation and are filled in with southern-sourced intermediate waters corrosive toward carbonate sediments, such as the modern AAIW. In this model and as observed at the middle to late Miocene transition, intervals of carbonate dissolution in the eastern equatorial Pacific and in the Caribbean basins are in phase.

CO₃ MARs in the Neogene sedimentary sequences recovered at ODP Leg 138 sites in the eastern equatorial Pacific (Fig. 1), as at Sites 998, 999, and 1000 in the Caribbean (Fig. 12), dramatically decrease at the middle to late Miocene transition. However, the timing of the carbonate crash in the eastern Pacific is significantly delayed when compared with the timing of the carbonate crash in the Caribbean. Although the CO₃ MAR records are only available from Hole 846B in the Peru Basin past 12.5 Ma, the CO₃ MAR reductions in several sites are centered around 11.5, 11.0-10.8, 10.5-10.3, and 10.2-10.1 Ma, and the largest, most sustained CO₃ MAR decreases occurred from 9.8 to 9.4 Ma (Fig. 1A, Fig. 15D), as well as a pronounced event at 9.0-8.8 Ma. Four of those intervals (11.5, 11.0-10.8, 10.5-10.3, and 10.2-10.1 Ma) correspond to the second, third, fourth, and fifth dissolution episodes of the Caribbean carbonate-crash intervals (Fig. 1A, Fig. 12, Fig. 15). The timing and periodicity of these four episodes of carbonate dissolution in the equatorial eastern Pacific and the Caribbean basins are identical and appear to correspond to peaks of NCW (equivalent to NADW) production (Fig. 15) (Wright and Miller, 1996). However, the high intensity of these dissolution episodes observed between 12 and 10 Ma in the Caribbean sites is only prevalent in the eastern Pacific in the interval between 11 and 9 Ma. The intensity of the carbonate crash in the eastern Pacific, therefore, lags behind by ~1 m.y. when compared with the Caribbean carbonate crash (Fig. 15). It is also curious to note and difficult to explain that the lowest CO₃ MARs in the eastern equatorial Pacific occurred at a time when the NADW production was gradually fading (Fig. 15D, F) (Wright and Miller, 1996), but seem to overlap with the Caribbean's ¹³C lightest values between ~10.1 and 9.5 Ma, after the Caribbean CO₃ MAR had fully recovered (Fig. 14, Fig. 15). This interval is known, as illustrated in Figure 2 and Figure 17, to correspond to a time when the Central American Seaway was for the first time temporarily closed based upon the first recorded intermingling of large mammals between the South and North American continents (Webb, 1985) and the lack of common affinities between the Caribbean and eastern Pacific coccolith assemblages (Fig. 2B) (Kameo and Sato, in press).

It appears that the timing of the most intense dissolution interval in the eastern Pacific is linked to the temporary closure of the Central American Seaway as opposed to the global thermohaline reorganization. Not taking into account the ~1-m.y. delay between the observed maximum dissolution in the eastern equatorial Pacific relative to that in the Caribbean basins, the carbonate-dissolution episodes observed in the Caribbean and the eastern equatorial Pacific appear to be synchronous and, therefore, in phase as predicted in the model. In addition, the timing of these episodes appears to be directly linked to peaks of NADW (NCW) production, with the exception of the first NADW production peak centered at ~12 Ma, which only linked to the first Caribbean dissolution of the carbonate crash and is conspicuously absent in the eastern Pacific CO₃ MAR records.

Figure 17 combines the summaries of Keller and Barron (1983) and Farrell et al. (1995) regarding the paleoceanographic history of the Miocene and the closure of the Isthmus of Panama, respectively. In addition, Figure 15 and Figure 17 include new constraints based upon our analyses of the Caribbean Sites 998, 999, and 1000. Figure 15 and Figure 17 help to place, at the middle to late Miocene transition, the occurrence of the Caribbean carbonate crash relative to the carbonate records from the equatorial Pacific and Atlantic oceans. As already suggested by Droxler et al. (1998), by triggering the subsidence of carbonate reefal barriers in the northern part of the Pedro Channel and the central region of the Walton Basin, tectonic activity in the late middle Miocene along the northern Nicaraguan Rise opened a series of new seaways to allow the establishment of the Caribbean/Loop Current system. This newly developed current system triggered an overall strengthening of the western boundary current, which transported from the Caribbean-Gulf of Mexico a large volume of saline waters, a necessary condition to explain the re-establishment and intensification of the NADW at the middle to late Miocene transition. The opening of the seaways along the northern Nicaraguan Rise, in conjunction with the uplift and partial emergence of the Isthmus of Panama, may have been crucial low-latitude controls that could explain the reorganization of global thermohaline circulation to a pattern more similar to that of today. Although other changes such as a sea-level regression or subsidence of northern North Atlantic ridges (Wright and Miller, 1996) may have played a role in the carbonate crash, the evidence from this study suggests the important factors may well have originated in the Caribbean and Central American Seaway.