INTRODUCTION

The Caribbean plate currently lies between the North and South American plates at a latitude between 10°N and 18°N (Fig. 1). From a plate tectonic perspective, it is extremely unlikely that the small Caribbean plate has maintained this position over time, particularly since all the larger bounding plates have been in motion.

Recent plate motions, estimated from earthquake slip vectors and seafloor spreading rates, give a broad range of directions and motions for the Caribbean plate. For instance, at a point (18°N, 278°E) near the Cayman Rise, the NUVEL-1A global plate motion model predicts that the Caribbean plate is moving N77°E at a rate of 11 mm/yr relative to North America (DeMets et al., 1994). In contrast, estimates that use data directly from Caribbean plate boundaries indicate that the relative motion is more easterly (N80°E) and two to three times faster (20-37 mm/yr) (e.g., Sykes et al., 1982; Deng and Sykes, 1995). To ascertain the absolute motion of the Caribbean plate, these relative motions can be combined with the motion of North America relative to the hot spots. In the hot-spot reference frame, the North American plate is moving N250°E at a rate of 31 km/m.y. (Gripp and Gordon, 1990). Within the uncertainties of the relative and absolute motion estimates, the Caribbean plate could be moving either east-northeast at rates of up to ~10 km/m.y. or west-southwest at rates of up to 20 km/m.y. over the past few million years.

Global Positioning System (GPS) measurements spanning the past 12 yr have begun to give more precise relative and absolute motion estimates, though these too have interpretational flexibility because the Caribbean GPS sites are limited to islands that in some cases are likely part of a plate boundary zone rather than the Caribbean plate. From the Euler vector given by Dixon et al. (1998), which is at 18.6°N, 107.2°E with a rotation rate of 0.36°/m.y., the Caribbean plate is moving N76°E at 24 mm/yr relative to North America. When placed in the International Terrestrial Reference Frame (ITRF-94; Boucher et al., 1996), the four GPS sites—Capotillo, Hispaniola; Cabo Frances Viejo, Hispaniola; Cabo Rojo, Hispaniola; and Isabela, Puerto Rico—considered to be on the northern Caribbean plate— move on average N47°E at a rate of 8 mm/yr (computed from table 5 of Dixon et al., 1998). The ITRF was designed to agree on average with the absolute plate motions of Argus and Gordon (1991), which assume no net rotation of the lithosphere. Assuming no true polar wander, this estimate would suggest that the Caribbean plate is currently moving northward at 5.4 mm/yr.

The position and motion of the Caribbean plate prior to recent times is even more poorly constrained because the plate boundary zones are complex and mainly destructive (strike-slip boundaries with components of extension and compression and subduction zones) and hot-spot tracks are absent. Paleomagnetic data are also very sparse from within the Caribbean plate (e.g., Kent and Spariosu, 1982; Gose, 1985; MacDonald, 1990), so much so that no Caribbean apparent polar wander path can be constructed and only crude estimates of the paleolatitude of the plate can be made. Many more data come from the margins of the plate (e.g., Gose, 1983, 1985; Beck, 1988; MacDonald, 1990; Frisch et al., 1992; Burmester et al., 1996), but these are from crustal blocks that show a variety of vertical axis rotations and latitudinal translations. Extracting an apparent polar wander path from these crustal blocks would, therefore, be nearly impossible because each crustal block likely has a unique history of moving relative to and docking with the Caribbean plate.

Data are very sparse, however, from directly on the Caribbean plate. In the compilation of Caribbean paleomagnetic data by MacDonald (1990), the sum of all Caribbean basin data of pre-Miocene age came from just 34 specimens from five Deep Sea Drilling Project (DSDP) sites analyzed by Lowrie and Opdyke (1973). These Late Cretaceous specimens indicated a paleolatitude 5°-10° less than that of the present latitude of the DSDP sites, but were too few to establish a convincing paleolatitude estimate. Similarly, a paleomagnetic pole computed from the magnetic anomaly over a presumably Late Cretaceous-age seamount near DSDP Site 145 indicates a near equatorial position for the Caribbean plate (Raff, 1973). Likewise, paleomagnetic inclinations from Late Cretaceous- to Eocene-age rock units in Central America that are now part of the Caribbean plate indicate near-equatorial paleolatitudes (Gose, 1983; Frisch et al., 1992).

Taken together, the recent plate motion estimates and the sparse paleomagnetic data suggest a more southerly position for the Caribbean plate since the Late Cretaceous. The latitude and the rate of northward motion over the interval from Late Cretaceous until recent times are, however, largely unconstrained.

Here we present results from paleomagnetic measurements from Leg 165 cores, including three sites (Sites 999, 1000, and 1001) from the Caribbean plate and one (Site 998) from the Cayman Rise, just to the north of the Caribbean plate. The new data expand the number of paleomagnetic results from pre-Miocene rocks from the Caribbean basin by more than an order of magnitude and constrain the paleolatitude of the Caribbean plate over the past 80 m.y. These constraints are then compared against tectonic models for the evolution of the Caribbean plate.

NEXT