Figure 12 A: Map of the Central American convergent margin, showing the location of volcanoes and DSDP site 495 and ODP sites 844, 1039, and 1253. Subduction signal is maximum in Nicaragua. Open symbols are back-arc or back-arc like samples? See also Figure 4.
B: Variations of Ba/Th vs. U/La in lavas from the volcanic front of Central America. Lavas from Nicaragua (diamonds), El Salvador (squares), and western Costa Rica (crosses) define binary mixing arrays between carbonate-dominated and hemipelagic-dominated end members. Sources include the carbonate (Carb) and hemipelagic (Hemi) sediment sections on the Cocos Plate and three depleted sources, a depleted MORB source and an E-MORB source from Sun and McDonough (1989) and an altered oceanic crust (AOC) estimate. Any triangle, formed by mixingbetween the two sediments and any of the three depleted sources, includes most but not all the lavas. Fluid extracted from AOC and Hemi should have much higher U/La and thus may expand the triangle over the observed range.
C: Regional variation in Ba/La in lavas from along the Central American arc. Ba/La is another monitor for the subduction component. Filled symbols are samples from along the volcanic front. Open symbols are back-arc samples. The back-arc samples in Honduras, furthest from the volcanic front (open triangles), have low Ba/La and are derived from depleted mantle and lack a subduction signature. Lavas from western Nicaragua (magenta diamonds) show the maximum slab signal (Ba/La up to 140 compared to normal mantle Ba/La =10-20).
D: Regional variation in U/Th along the volcanic front of the arc. This ratio is a useful tracer for sediment subduction in Central America because of the unusually high U content and U/Th of carbon-rich hemipelagic sediments near the top of the sediment section on the Cocos plate, as found at DSDP site 495.