Deeply incised and backfilled paleocanyons in early Paleogene shelf strata along the western and northern Gulf of Mexico margin attest to large relative sea-level fluctuations, but they predate the accepted age for onset of Cenozoic continental glaciation. Using Pleistocene canyons as a crude yardstick, the scale of these paleocanyons suggests relative sea-level changes at least as large as Pleistocene fluctuations. Therefore, we speculate that water level in the Gulf of Mexico was drawn down while the Gulf was isolated from the world's oceans during the late Paleocene/early Eocene interval. We suggest that the cause for isolation was the progressive collision of the Cuban arc with the Yucatn and Bahamas carbonate platforms, which temporarily closed off the southeastern Gulf of Mexico. In Miocene Mediterranean and Holocene Black Sea examples of marine-basin isolation, evaporation greatly exceeded rainfall and runoff, and our examination of the Gulf of Mexico case suggests that water level may have dropped below the level of the world's oceans at least once by several hundred meters, and possibly much more.

Implications for geology and hydrocarbon exploration in the Gulf may include:

• bypass of enormous quantities of coarse detritus into the deep basin
• seaward collapse of exposed clastic shelf margins
• triggering and/or acceleration of salt evacuation (basinward squeegee effect of slumping sediments)
• release of gas hydrates from sediments under shallower and warmer water, thereby contributing to the 100,000-year-long worldwide Paleocene/Eocene boundary heating event
• development of secondary porosity in both platform and deep-water carbonate sections by dissolution and phreatic diagenesis, e.g., in the Golden Lane/Poza Rica area of Mexico
• hypersalinity and possible sea-bottom stagnation with source-rock deposition in areas that remained marine
• deposition of fine-grained condensed sections (seal and source rock) during flooding period(s) when connection with the world's oceans was reestablished, creating stratigraphic traps at canyon flanks and turbidite reservoirs in the canyons.

Recognition that early Paleogene relative sea-level changes seen in the Gulf may pertain to basin isolation is grounds for treating eustatic curves derived for or from the Gulf with suspicion. In addition, catastrophic basinward transfer and collapse of mass near the shelf edges would have caused isostatic unloading (rebound) of shelf margins that was proportional to the mass transfer. In the case of a discreet slumping event, such as the Lavaca Megaslump event of south Texas, this effect may have caused uplift of several to a few tens of meters of footwall areas within about 100 km from the slump. Larger downslope movements such as those related to the collective Wilcox fault province would have caused far larger isostatic rebounds on the shelf, perhaps in excess of 100 m if sedimentation did not keep pace with faulting.

A body of circumstantial evidence continues to grow in support of this hypothesis; its potential implications, both academic and commercial, merit further investigation. Integration of information from Cuba, Mexico, the United States, and the Bahamas will be required to fully test the hypothesis.