The large-scale atmospheric circulation in the Caribbean area coupled with the abundance of Miocene silicic ignimbrites in Central America provide strong evidence that the source of the western Caribbean tephra fall layers was to the west of the Leg 165 sites during the Miocene to Holocene. Giant co-ignimbrite plumes generated from pyroclastic flow-generating eruptions injected tephra to stratospheric levels with subsequent dispersal both to the east and west of the Tertiary ignimbrite province. Many of the eruptions were likely to be similar to or larger than the great Campanian ignimbrite eruption in Italy that ejected 80 km3 of magma. The crystal size data from Sites 998, 999, and 1000 also indicate a significant coarsening of the tephra at ~18 Ma, or coincident with the Miocene peak in ash accumulation rate (Fig. 3, Fig. 6, Fig. 9). This coarsening could be attributed to (1) the increased probability of sampling coarser layers in intervals that contain a higher number of tephra layers, (2) a strengthening of the winds that were dispersing material from a range of eruptions that remained of equal size over time, or (3) a period of more energetic eruptions that were dispersed over a larger area because of higher eruption column heights.
First, we consider the influence of sampling frequency. Tephra layers were selected randomly throughout the sequences and the frequency of sampling ranges from 1 layer/m.y. to a maximum of 6 layers/m.y. There does not appear to a be a strong correlation between the sampling frequency and the maximum crystal size found within a 1-m.y. interval (Fig. 17). There is a slight tendency for the coarsest layers at any given site to be found within the interval with the most layers, but there are many instances of relatively coarse layers being sampled in intervals where only one layer was selected. We feel, therefore, that sampling frequency alone cannot account for the significant increase in grain size found in layers near the peak in ash accumulations at Sites 998, 999, and 1000.
The next possibility to be considered is whether the coarsening of the Miocene tephra layers reflects more vigorous atmospheric circulation. Such conditions might lead to both increased grain size of individual layers and a higher frequency of layers. The past history of atmospheric circulation has been assessed in a semi-quantitative manner using the record of eolian components in deep-sea sediments (Rea, 1994). In particular, changes in the grain size of atmospherically transported dust has been used to make inferences about changes in the paleointensity of wind. These records predominantly reflect changes in tropospheric winds, as dust transport is restricted to such elevations. One of the best records of Cenozoic atmospheric circulation has been obtained from Core LL44-GPC3 from the central gyre region of the Pacific (Fig. 18). This core provides a nearly continuous record of dust transport that extends back 70 m.y. The most significant change in Cenozoic circulation occurred at the time of the Paleocene/Eocene boundary, when there was an abrupt reduction in the grain size of dust. This reduction has been attributed to a decrease in wind speed of a factor of three or four (Janecek and Rea, 1983). Throughout the Eocene the dust remained small but began to coarsen in the early Oligocene and then again during much of the middle Miocene. Continued coarsening is also associated with the onset of Northern Hemispheric glaciation in the late Pliocene. There does not appear to be any correlation between the western Caribbean Miocene peak in explosive volcanism and any distinct coarsening peak in the dust grain-size record (Fig. 18). The coarsest dust in the upper Cenozoic is actually deposited after the peak in explosive volcanism. Furthermore, there is also no correlation between the lower upper Eocene peak in explosive volcanism recorded at Site 999 and any significant coarsening in the dust record. This period corresponds to some of the finest grained dust deposition in the central Pacific.
The dust record of Core LL44-GPC3 provides information for winds at latitudes >20°N during its migration to the north on the Pacific plate and thus lies further north than the latitudinal belt associated with Caribbean tephra transport. There is evidence, however, that the current position of the boundary between the higher latitude westerlies and the lower latitude trade winds might have been different during the early Miocene. Rea (1994) has suggested that the decrease in grain size of the dust record from Core LL44-GPC3 in the lower Miocene may reflect passage of the site beneath a northerly displaced intertropical convergence zone (ITCZ) that was located at ~22°-24°N. The northward displacement of the ITCZ is attributed to an asymmetry of the thermal gradients between hemispheres, with the Southern Hemisphere exhibiting a strong gradient as a result of ice buildup on Antarctica and the Northern Hemisphere being characterized by warm and equitable climates with weaker thermal gradients. Under these conditions the boundary between the westerlies and trade winds in the northern hemisphere may have been somewhere between 40°N and 50°N (Rea, 1994). The southern trade winds would have been stronger during the early Miocene, owing to the higher pole-to-equator thermal gradient in the southern hemisphere. This configuration would be expected to enhance westerly transport of tephra from Central America as a result of low-level transport in the troposphere and perhaps diminish easterly transport into the Caribbean. However, the tephra record indicates the opposite, thus supporting the contention that the early to mid-Miocene peak in explosive volcanism is not a primary signal of major atmospheric circulation changes at that time.
We believe that the best explanation for the coarsening of the tephra layers is that eruption intensity, or column height, was greater for the event at the peak of the tephra accumulation episode. Eruption column height has a major impact on the dispersal of tephra because it defines the nature and extent of the umbrella region from which tephra is subsequently advected by prevailing winds (Sparks, 1986). The relationship between the dispersal area for a particular particle size and eruption column height is nonlinear, and, consequently, the variations in eruption column height can translate into large changes in dispersal efficiency (Carey and Sparks, 1986). Thus, the Miocene peak in explosive volcanism likely represented not only an increase in the frequency of eruptions but also an increase in the intensity of events, as reflected in higher eruption column heights.