The results of drilling during Leg 165, with the recovery of thousands of tephra fall layers, volcaniclastic turbidites, and submarine basaltic lavas of the Caribbean oceanic crust, provide new insights into the timing and character of volcanism in the Caribbean region. The results are reviewed here in the context of the new dating and their implications with respect to the tectonic evolution of the region are discussed.
With the discovery of numerous datable tephra layers in the sections cored at ODP Sites 998-1001, it was hoped that these Caribbean strata would provide a rare opportunity to test and refine the Cenozoic time scale. A revised Cenozoic geochronology and chronostratigraphy was recently published by Berggren et al. (1995a, 1995b). This time scale is based on a modified version of Cande and Kent's (1992, 1995) magnetochronology, which itself is based primarily on South Atlantic seafloor magnetic anomalies. Nine age calibration points, anchored by Chron C34n(y) at 83.0 Ma plus the zero-age ridge axis, and interpolated by a cubic spline function provide the numerical basis for the global polarity time scale. All calibration points are constrained directly by calcareous microfossil datums for purposes of correlation (Berggren et al., 1995b). The application of Milankovitch climate cyclicity to geochronology, corroborated by high precision 40Ar/39Ar dating, has provided a powerful tool in the effort to calibrate polarity boundaries and biostratigraphic datums in the younger part of the record (e.g., Shackleton et al., 1995a, 1995b). In their revised time scale, Berggren et al. (1995a, 1995b) accepted the astronomical time scale values of polarity events from the present to 5.23 Ma (Shackleton et al., 1990; Hilgen, 1991; Langereis et al., 1994). The magnetic polarity time scale used for ODP Leg 165 adopted the astronomically tuned time scale of Shackleton et al. (1995b) and Shackleton and Crowhurst (1997) for the interval from 5 to 14 Ma (Sigurdsson, Leckie, Acton, et al., 1997).
Astronomical calibration of calcareous nannofossil and planktonic foraminiferal first and last occurrences (FO and LO) has been applied back to at least 13.5 Ma (Shackleton et al., 1995a; Raffi and Flores, 1995; Berggren et al., 1995a; Backman and Raffi, 1997; Chaisson and Pearson, 1997; Pearson and Chaisson, 1997); biostratigraphic datums older than 13.5 Ma are from Berggren et al., (1995b, appendices 1 and 2). It is within this framework that we compare the ages of the dated tephra levels based on biostratigraphy to the 40Ar/39Ar ages of the individual tephras (Fig. 4). We present several biostratigraphic age estimates for each of the tephra layers based on one or more pairs of calcareous nannofossil and planktonic foraminiferal datums, and/or different age models for these biostratigraphic datum pairs (Table 2). We ranked the tephra 40Ar/39Ar ages according to magnitude of analytical error and relative congruence with the biostratigraphic age: 1 = congruent age with small error (<0.5 Ma); 2 = congruent age with larger error (0.5-1.5 Ma); 3 = reasonably congruent mean but very large error (>1.5 Ma), or suspect (incongruent?) age; and 4 = an incongruent age because of diagenetic alteration or sedimentary reworking.
Our findings demonstrate that there is excellent agreement between the astrochronologically tuned nannofossil ages and the radiometric ages of the tephra layers for the interval from 4.8 to 15.3 Ma; the interpolated planktonic foraminiferal ages yield less congruent ages (Fig. 4). Strata older than the early middle Miocene yield mixed results. Some of the radiometric ages probably reflect diagenetic alteration, and a few of the pre-middle Miocene dates may prove to be useful in future recalibrations of the Cenozoic time scale. For purposes of this discussion we consider some of the incongruent results in the context of possible diagenetic influences. For example, there is a significant disparity between the 40Ar/39Ar ages and the biostratigraphic ages of two lower middle Miocene-upper lower Miocene tephras from Hole 999A (Samples 165-999A-41X-4, 41-42 cm, and 45X-2, 94-95 cm; 378.61 and 414.54 mbsf, respectively); yet the upper lower Miocene tephra in Sample 165-999A-48X-6, 42-43 cm (449.02 mbsf) yields an excellent age that is identical to the nannofossil age. However, we also note that although the biotite age of Sample 165-999A-48X-6, 42-43 cm, is congruent with the biostratigraphic age, the sanidine age is ~1.6 m.y. younger. Another example of within-site variability comes from the lower Miocene tephras in Hole 998A. The tephra layer in Sample 165-998A-27X-1, 120-122 cm (244.40 mbsf), yields an age that is in excellent agreement with the nannofossil age, whereas the tephra at 29X-3, 68-69 cm (266.18 mbsf), is ~3 m.y. older than the biostratigraphic age. A replicate analysis from the latter tephra gives an age that is 10 Ma younger. The lower Miocene at Site 1000 also shows significant disparities between the radiometric and biostratigraphic ages. In this case, the three lower Miocene tephras analyzed in this study yield ages that are inverted; the stratigraphically highest of the three (Sample 165-1000B-14R-2, 139-140 cm; 611.77 mbsf) yields the oldest radiometric age (18.31 Ma), whereas the stratigraphically lowest of these tephras (19R-2, 99-101 cm; 659.89 mbsf) yields the youngest radiometric age (17.02 Ma; Table 1).
In general, there is less agreement between the tephra ages and the biostratigraphic ages in the Oligocene part of the section. Two of the three Oligocene tephras analyzed in this study, one from Sample 165-999B-15R-2, 45-46 cm (661.05 mbsf) and the other from 1R-7, 53-54 cm (567.30 mbsf), produced radiometric ages that are close to the biostratigraphic ages, but the means suggest some differences. Provided these tephras have not been diagenetically altered, the results may indicate that either the nannofossil datums used for the Oligocene need further refinement or the polarity time scale needs additional age calibration points in the Oligocene. There is good congruence between radiometric and biostratigraphic ages in the lower Eocene of Site 999 (Sample 165-999B-48R-3, 6-7 cm), yet there is significant disparity between the two geochronologies in another lower Eocene sample from the same site (Sample 165-999B-47R-6, 4-5 cm). The three late Paleocene tephra layers from Site 1001 analyzed in this study are highly incongruent with the biostratigraphic ages. The most likely explanation is that these tephras have been diagenetically altered (see below).
Interstitial pore fluid movement through some of these coarse-grained tephra layers may be at least partially responsible for the incongruent results between radiometric ages and biostratigraphic ages. The distinctive diagenetic profiles of Sites 998-1001 illustrate the migration of chemical species through the sedimentary column (Sigurdsson, Leckie, Acton, et al., 1997; Lyons et al., Chap. 19, this volume). For example, microbially mediated redox diagenesis characterizes the lower and middle Miocene at Site 1000 (Lyons et al., Chap. 19, this volume). In addition, elevated levels of dissolved silica characterize the middle Miocene and older sections at Sites 998, 999, and 1001 indicating diagenetic changes and movement of pore fluids in the deeper parts of each section (Lyons et al., Chap. 19, this volume). Note for example that the radiometric ages are highly congruent with the biostratigraphic ages at Site 998 down to a level near the highest occurrence of diagenetic chert at ~248 mbsf. Mineral grains analyzed from three tephra layers at Site 1001 were most likely altered by diagenesis based on the highly incongruent results. Redeposition of volcanic debris is an unlikely explanation for the spurious ages because these particular layers resemble all the other numerous thick tephras cored during Leg 165. Although the tephra layers often do contain coarser mineral grains at the base thereby resembling a graded bed, the tephras do not show the distinctive scoured basal contact or the rippled to parallel laminations expected in the upper part of turbidite deposits. Fabricius (Chap. 10, this volume) concluded that ~400 m of section has been removed beneath the middle Miocene-Eocene unconformity at this site, based on a study of the physical properties and the presence of microstylolites in these deeply buried carbonate-rich rocks.
The most voluminous volcanism in the Caribbean region occurred in the submarine environment, producing an oceanic plateau. Most of the information on the oceanic plateau has come from studies at its exposed subaerial margins on the perimeter of the Caribbean plate (Duncan and Hargraves, 1984; Sinton et al., 1998). This oceanic plateau is up to 20 km thick, and at least 600,000 km2 in lateral extent. During Leg 165, the oceanic crust of the plateau was sampled at Site 1001 on the lower Nicaraguan Rise, with a penetration of 38 m into a succession of submarine basalt lavas. Some of these basalts are exceptionally fresh, as evidenced by the preservation of basaltic glass. The 40Ar/39Ar dating of these basalts yielded an age of 81 ± 1 Ma, in agreement with earlier dating of the plateau (Sinton et al., Chap. 15, this volume). These dates are particularly important, in that the dated rocks are demonstrably submarine lavas that predate the overlying mid-Campanian calcareous sediment, and thus provide a reliable age of the beginning of burial of the oceanic plateau in this region (Sigurdsson, Leckie, Acton, et al., 1997). Benthic foraminiferal evidence suggests that the plateau subsided from upper bathyal depths (200-500 m) in early to mid-Campanian time to middle or lower bathyal depths (1000-2500 m) by the late Maastrichtian.
A number of volcaniclastic turbidites and volcanic ash layers are found in the Paleocene carbonate sediment 158-255 m above the submarine basalt formation at Site 1001, and biotites were extracted from three of these layers for 40Ar/39Ar dating (Table 2). However, the biotites yield apparent ages that are well in excess of the biostratigraphic age of the adjacent sediment. This discrepancy is attributed to cryptic alteration of the biotites, as they occur in ash layers where virtually all of the volcanic glass has been altered to smectite.
One of the earliest known episodes of explosive volcanism in the circum-Caribbean region is represented by a late Paleocene to middle Eocene island arc that extended from the submerged Cayman Ridge in the west to the now-exposed Sierra Maestra province of eastern Cuba (Draper and Barros, 1994; Iturralde-Vinent, 1996). Drilling at Site 998 on the north flank of the Cayman Rise during Leg 165 led to the discovery of a major lower to middle Eocene volcanic ash and volcaniclastic turbidite succession that we correlate with the Sierra Maestra arc of Cuba (Sigurdsson, Leckie, Acton, et al., 1997), although a source in the Newcastle and Summerfield formations of Jamaica cannot be ruled out. At this ODP site, the frequency and thickness of ash layers and turbidites increases progressively with depth through the middle to lower Eocene section, and drilling at this site was terminated before the volcaniclastic-rich sequence was fully penetrated. At Site 999, lower Eocene tephra fallout layers form a weak, but significant peak in ash accumulation rate in the 50-Ma range, and biotites from two of these layers were selected for 40Ar/39Ar dating (Table 1, Table 2). They yield ages of 48.6 and 52.5 Ma. Tephra layers and volcaniclastic turbidites in this episode at Site 998 did not contain biotites suitable for dating.
Seismic data suggest that basement rocks at Site 998 are ~210 m below the bottom of the hole, indicating that this part of the Cayman Rise is a volcanic arc, most likely continuous with the Paleocene-Eocene Sierra Maestra arc. Studies of the Sierra Maestra terrain provide clues about the arc structure underlying Site 998, some 750 km to the west. The volcanic Turquino zone is a major part of the Sierra Maestra region in southeast Cuba, and it is flanked to the north by the sedimentary and volcaniclastic Cauto zone, which has been interpreted as a backarc basin (Cobiella-Reguera, 1988, 1997). The Turquino succession, which rests on Cretaceous volcaniclastics, is composed of calc-alkaline basaltic to rhyolitic lavas and pyroclastics, with thin intercalations of marine sediments ranging from Danian to Eocene in age. The volcanic succession, which is several kilometers in thickness, is intruded locally by younger granites that range in age from 46 to 58 Ma (Khudoley and Meyerhoff, 1971).
An integrated picture of evidence from drilling at Site 998, from the Sierra Maestra in Cuba and from dredging of the north wall of the Cayman Trough (Perfit and Heezen, 1978), is of an east-west-trending island arc in Paleocene-Eocene time, shedding off volcaniclastic turbidites and tephra fallout into the Yucatan Basin to the north (Fig. 5). The polarity of subduction beneath the arc has a fundamental bearing on tectonics in the Caribbean region. We have previously proposed that subduction of the leading edge of the Caribbean plate beneath the Cayman Rise was from the south (Sigurdsson, Leckie, Acton, et al., 1997), which is in agreement with tectonic models proposed for the Sierra Maestra segment of the arc in Cuba (Iturralde-Vinent, 1996; Cobiella-Reguera, 1997). The principal implications of this conclusion are, first, that the Caribbean plate had a strong vector of northward motion in the Paleogene and, second, that the leading edge of the plate was subducted and thus relatively thin oceanic crust, as opposed to the thick oceanic plateau that makes up much of the plate interior. It is likely, as inferred in Figure 5, that the Aves Ridge was the eastern and southern continuation of the Cayman Ridge, as radiometric dates of granodiorites dredged from the Aves Ridge yield Late Cretaceous to early Tertiary ages (Fox et al., 1971). We note that a southwest-dipping seismic zone has also been proposed for the Cayman Rise (Pindell and Barrett 1990; Pindell 1994; Draper et al., 1994), and thus, the plate tectonic configuration of this area still remains unresolved.
The geographic extent of tephra fallout from explosive volcanism of the Cayman Ridge and Sierra Maestra volcanic arc also includes the region of the Nicaraguan Rise, as indicated by recovery of large numbers of tephra layers in upper Paleocene and lower Eocene sediments at Site 1001. At the time of this activity, Site 1001 was probably due south of the Cayman Ridge, if we backtrack the transcurrent motion on the Cayman trench fracture zone since the Eocene. At this time, Jamaica would have been in an intermediate position with respect to the Nicaraguan Rise and the Cayman Ridge, or perhaps somewhat further to the west. Contemporaneous volcaniclastics interbedded with sediments are found in the Summerfield Formation and the Wagwater Formation in Jamaica (Robinson, 1994), and may correlate with the Paleocene-Eocene episode of explosive volcanism documented during Leg 165. In Hispaniola, lower Tertiary formations provide evidence of volcanic arc construction in the Paleocene to Eocene (Draper et al., 1994).
The largest episode of explosive volcanism documented in Leg 165 sediments began in the middle Eocene (~46 Ma) and terminated in the early Oligocene (~32 Ma). This episode is particularly well recorded at Site 999 on the Kogi Rise, Colombian Basin, but is also present in sediments at Site 998 on the Cayman Rise. At Site 999, the accumulation rate of tephra is ~250 cm/m.y., and the frequency of major explosive eruptions in the source region(s) is on the order of one event per 20 k.y. The rhyolitic tephra layers are generally 2-5 cm in thickness. Biotites from two tephra layers at Site 998, representing the late stages of this episode, were dated by the 40Ar/39Ar technique, yielding ages of 30.7 and 31.7 Ma (Table 1, Table 2). Tephra layers at Site 999 from this episode do not contain biotites suitable for dating. Considering that Site 999 is over 600 km from a possible volcanic source, it is evident that these fallout deposits represent major ignimbrite-forming events. The thickness-distance relationship of the tephra layers suggests a magnitude (volume) for these eruptions in the range of the 30-k.y. eruption from the Phlegrean Fields caldera in Italy (100 km3) and the 75-k.y. Toba eruption on Sumatra (2000 km3).
The source of these tephra layers is most likely to be found in Central America. Two major Tertiary ignimbrite formations extend from the Mexican border in the north, through Guatemala, El Salvador and Honduras, into Nicaragua, over a region that is more than 800 km in length and up to 300 km wide. They represent the largest episode of silicic ignimbrite volcanism in Central America. The ignimbrite-rich formations underlie upper Tertiary and Quaternary andesitic and basaltic andesite stratocones of the active Central American volcanic arc (Reynolds, 1980). However, the Tertiary ignimbrite-bearing formations generally extend far east of the active arc. We have earlier proposed that the Eocene tephra fallout episode found during Leg 165 correlates with the ignimbrite-producing events represented by the Matagalpa and Morazan Formations of Central America (Sigurdsson, Leckie, Acton, et al., 1997). The latter formations were tentatively given an early Oligocene age by Ehrenborg (1996) on the basis of meager radiometric data, but the deep-sea evidence from Leg 165, including data presented here, indicates that these ignimbrites were dominantly erupted during the Eocene. A pre-Miocene ignimbrite series has been recognized in Central America for some time, and it is now becoming accepted as a widespread lithostratigraphic unit. In both Guatemala and El Salvador, older Tertiary (pre-Miocene) rhyolitic tuffs are present, but generally unnamed and unmapped, or grouped in the Morazan Formation (El Salvador). In the Nicaraguan Highlands, the Matagalpa Group is unconformably overlain by the Miocene Coyol Group. The Matagalpa Group is dominated by 50-m-thick dacite to rhyolite ignimbrites and associated tephra fall deposits, with a total thickness in excess of 2 km (Ehrenborg, 1996).
The Central American Eocene-early Oligocene volcanic episode recorded in Leg 165 sediments is broadly contemporaneous with the rhyolitic ignimbrite flare-up in the Sierra Madre Occidental province of Mexico (Fig. 8). The Sierra Madre Occidental is Earth's largest ignimbrite province, with an areal extent of ~250,000 km2 and over 1 km in thickness (McDowell and Clabaugh, 1979). The initiation of these contemporaneous episodes of siliceous volcanism in Central America and Mexico may be related to plate tectonic rearrangements in the Pacific Ocean. They coincide broadly with the abrupt change in the direction of motion of the Pacific plate at 43 Ma, from northwesterly to a more westerly direction, as recorded by the bend (60°) in the Emperor-Hawaiian hot spot chain (Atwater, 1970, 1989).
Sites 998, 999, and 1000 recovered a large number of tephra fall layers that constitute a major episode of Miocene silicic explosive volcanism in the Caribbean region, comparable in magnitude and character to the Eocene event, although individual tephra layers tend to be thicker in the Miocene episode (Fig. 6). At all three sites, the siliceous tephra layers have the characteristics of co-ignimbrite fallout deposits (Sigurdsson, Leckie, Acton, et al., 1997; Carey and Sigurdsson, Chap. 5, this volume). The distribution and thickness of the Miocene tephra at these three sites indicates a general fallout axis trending easterly out of Central America, the nearest major source of ignimbrite volcanism in the late Tertiary (Fig. 7).
As shown in Figure 2, the Miocene tephra accumulation at Site 999 defines a rather sharp peak at ~19 Ma, beginning near the Oligocene/Miocene boundary (~23 Ma) and terminating in the middle Miocene at ~13 Ma. Biotites from 18 Miocene tephra layers from all three sites were dated by the 40Ar/39Ar technique (Table 1, Table 2) and sanidine crystals were 40Ar/39Ar dated in one layer (Table 3). Ages of most of the dated layers fall within the range of the peak in accumulation rate, as this part of the record contains the most abundant and thickest layers suitable for crystal extraction. As shown in Figure 2, a subsidiary peak in ash accumulation occurs in the range of 8-10 Ma.
The Miocene tephra episode recorded in the Caribbean sediments is likely to represent the distal fallout equivalent of thick ignimbrite formations that are present throughout Central America. The major formations include silicic welded tuffs of the Chalatenango Formation in south-central Guatemala and El Salvador (Reynolds, 1987; Wiesemann, 1975), thick rhyolitic ignimbrites of the Padre Miguel Group in southeast Guatemala and Honduras (Reynolds, 1980), and the thick siliceous ignimbrites of the Coyol Group in Nicaragua (Ehrenborg, 1996). Reliable 40Ar/39Ar single crystal biotite ages for ignimbrites within the Coyol Group cluster from 12.3 to 18.4 Ma, defining an early-middle Miocene volcanic episode. Many of the formations are linked to caldera structures and have aggregate thicknesses of several hundred meters. Some of the co-ignimbrite tephra from eruptions of this ignimbrite province were transported into the Pacific, where it was recovered off Guatemala during Deep Sea Drilling Program (DSDP) Leg 67 (Cadet et al., 1982a), and near the Middle America Trench off Mexico during Leg 66 (Cadet et al., 1982b). However, as we would predict from atmospheric circulation (Carey and Sigurdsson, Chap. 5, this volume, the dispersal and accumulation rate to the west is minor, compared to the easterly fallout pattern in the Caribbean.
A relationship between a change in the plate tectonic configuration and origin of the Central American ignimbrite episodes is as yet unclear, but some factors suggest a connection. The onset of the Miocene episode at ~23 Ma coincides with a major tectonic reconfiguration in the eastern Pacific, adjacent to the Central American arc. Magnetic anomalies on the Pacific and Nazca (Farallon) plates show an abrupt change in trend between anomalies generated during the Miocene and the Oligocene, requiring a major reorientation of spreading dynamics (Handschumacher, 1976). Subduction of the Farallon plate during the Oligocene was nearly due east below Central America, at a rate of 6.5 cm/yr. Because of the subduction of the Farallon-Pacific spreading ridge below North America, a direct coupling of the American-Pacific plate occurred at ~26 Ma, resulting in a radical change in spreading kinematics in the eastern Pacific. This brought about north-south spreading from a new east-west-trending Cocos-Nazca spreading center, beginning between 20 and 25 Ma, dividing the Farallon plate into the Nazca and Cocos plates (Handschumacher, 1976; Duncan and Hargraves, 1984; Meschede et al. 1998). This led to a change in subduction of the Cocos plate beneath Central America, to a more northeasterly direction, at a higher rate of ~7 cm/yr. We speculate that this is what initiated the onset of the large Miocene ignimbrite flare-up on the Chortis block in Central America.