Abstract Devonian reef systems are thought to represent the greatest phase of global reef development in the Phanerozoic. Despite this, ecological and environmental controls on the sedimentary nature of these vast systems have scarcely been investigated and remain enigmatic. The Late Devonian (Frasnian) Alexandra Reef System, exposed in the Northwest Territories of Canada, developed on a ramp that was situated on the western margin of Laurussia. The system consists of two reef complexes. The second reef complex developed basinwards of the first after sea level fell ~ 17 m. In contrast to stromatoporoid (± coral)-dominated reef facies in the first reef complex and the upper part of the second reef complex, reef facies in the lower part of the second reef complex are dominated by stromatoporoid-microbe associations. These include significant renalcid boundstone and stromatolite accumulations that are not found elsewhere in the reef system. It is concluded that the occurrence of the stromatoporoid-microbe reef facies indicates that a shift in the reef environment from oligotrophic to mesotrophic conditions took place. The mechanisms of nutrification were linked to the platform geometry, sea-level position, and oceanographic system, indicating that on carbonate ramps, systems tracts of falling sea level (forced regression) and sea-level lowstand may be particularly susceptible to nutrification. A nutrient-gradient model developed to explain different types of reef facies in the Alexandra Reef System indicates that trophic resources were an important control on the composition of Devonian reef-building communities, and that Devonian reefs and carbonate platforms were not highly susceptible to nutrient-invoked drowning.

Cores recovered on Deep Sea Drilling Program leg 43 and on Bermuda itself, together with geophysical data (anomalies in basement depth, geoid, and heatflow) and modeling have long suggested that the uplift forming the Bermuda Rise, as well as the initial igneous activity that produced the Bermuda volcanoes, began ca. 47–40 Ma, during the early to middle part of the Middle Eocene. Some authors attribute 65 Ma igneous activity in Mississippi and 115 Ma activity in Kansas to a putative “Bermuda hotspot” or plume fixed in the mantle below a moving North America plate. While this is more or less consistent with hotspot traces computed from “absolute motion” models, the hotspot or plume must resemble a blob in a lava lamp that is turned off for up to 25 million years at a time, and/or be heavily influenced by lithosphere structure. Moreover, Cretaceous igneous activity in Texas and Eocene intrusions in Virginia then require separate mantle “blobs.” The pillow lavas forming the original Bermuda shield volcano have not been reliably dated, and the three associated smaller edifices have not been drilled or dated. A well-dated (ca. 33–34 Ma) episode of unusually titaniferous sheet intrusion in the Bermuda edifice was either triggered by platewide stress changes or reflects local volcanogenic events deep in the mantle source region. The high Ti and Fe of the Bermuda intrusive sheets probably relate to the very high-amplitude magnetic anomalies discovered on the islands. Numerical models constrained by available geophysical data attribute the Bermuda Rise to some combination of lithospheric reheating and dynamic uplift. While the relative contributions of these two processes cannot yet be wholly separated, three features of the rise clearly distinguish it from the Hawaiian swell: (1) the Bermuda Rise is elongated at right angles to the direction of plate motion; (2) there has been little or no subsidence of the rise and the volcanic edifice since its formation—in fact, rise uplift continued at the same site from the late Middle Eocene into the Miocene; and (3) the Bermuda Rise lacks a clear, age-progressive chain. We infer that the Bermuda Rise and other Atlantic midplate rises are supported by anomalous asthenosphere, upwelling or not, that penetrates the thermal boundary layer and travels with the overlying plate. The elongation along crustal isochrons of both the Bermuda volcanoes and the Bermuda Rise and rise development mostly within a belt of rougher, thinner crust and seismically “slower” upper mantle—implying retention of gabbroic melts at the ancient Mid-Atlantic Ridge axis—suggest that the mantle lithosphere may have helped localize rise development, in contradiction to plume models. The Bermuda Rise area is seismically more active than its oceanic surroundings, preferentially along old transform traces, possibly reflecting a weaker upper mantle lithosphere. We attribute the “Bermuda event” to a global plate kinematic reorganization triggered by the closing of the Tethys and/or the associated gravitational collapse into the lower mantle of subducted slabs that had been temporarily stagnant near the 660 km mantle discontinuity. The widespread onset of sinking slabs required simultaneous up-welling for mass balance. In addition, the global plate kinematic reorganization was accompanied by increased stress in some plate interiors, favoring magma ascent along fractures at structurally weak sites. We suggest that the Bermuda event and concomitant igneous activity in Virginia, West Antarctica, Africa, and other regions were among such upwellings, but structurally influenced by the lithosphere, and probably originated in the upper mantle. Drilling a transect of boreholes across and along the Bermuda Rise to elucidate turbidite offlap during rise formation might discriminate between a widely distributed mantle source (such as a previously subducted slab) and a narrow plume whose head (or melt root) spreads out quasi-radially over time, generating an upward and outward expanding swell.

Considerable uncertainty exists as to whether the last interglacial was relatively “short” (~10 ka) or “long” (∼20–60 ka), although most investigators generally agree that the last interglacial correlates with all or part of deep-sea oxygen-isotope stage 5. A compilation of reliable U-series ages of marine terrace corals from deposits that have been correlated with isotope stage 5 indicates that there were three relatively high sea-level stands at ca 125–120 ka, ca. 105 ka, and ca. 85–80 ka, and these ages agree with the times of high sea level predicted by the Milankovitch orbital-forcing theory. At a number of localities, however, there are apparently reliable coral ages of ca. 145–135 ka and ca. 70 ka, and the Milankovitch theory would not predict high sea levels at these times. These ages are at present unexplained and require further study. The issue of whether the last interglacial was “short” or “long” can be addressed by examining the evidence for how high sea level was during the stands at ca. 125 ka, ca. 105 ka, and ca. 80 ka, because sea level is inversely proportional to global ice volume. In technically stable areas such as Bermuda, the Bahamas, the Yucatan peninsula, and Florida, there is clear evidence that sea level at ca. 125 ka was +3 to +10 m higher than present. During the ca. 105 ka and ca. 80 ka high sea-level stands, there is conflicting evidence for how high sea levels were. Studies of uplifted terraces on Barbados and Haiti and most studies of terraces on New Guinea indicate sea levels considerably lower than present. Studies of the terraces and deposits on the east and west coasts of North America, Bermuda, and the Bahamas, however, indicate sea levels close to, or only slightly below, the present at these times. Thus, data from Barbados, Haiti, and New Guinea indicate a “short” last interglacial centering ca. 125 ka, but data from the other localities indicate that sea level was high during much of the period from 125 to 80 ka, and that there were two minor ice advances in that period. If it is accepted that the last interglacial period was relatively “long” and ended sometime after ca. 80 ka, then coastal deposits on the California Channel Islands record a shift in the nature of sedimentation at the interglacial/glacial transition. Marine terraces that are ca. 80 ka are overlain by two eolianite units separated by paleosols. U-series ages of the terrace corals and carbonate rhizoliths indicate that eolian sedimentation occurred between ca. 80 and 49 ka, and again between ca. 27 and 14 ka. Eolian sands were apparently derived from carbonate-rich shelf sediments during glacially-lowered sea levels, because there are not sufficient beach sources for calcareous sediment at present. The times of eolian sedimentation agree well with times of glaciation predicted by the Milankovitch model of climatic change.