Uranium series and beryllium isotope evidence for an extended history of subduction modification of the mantle below Nicaragua

Abstract

U-series nuclides, beryllium, and lead isotopes have been measured on historic lavas from eight volcanoes in Nicaragua. Low-Ti samples from northern Nicaragua have ( 230Th) ( 232Th) ratios from 2.23 to 2.56 and are enriched in 238U and 234U over 230Th by 1–16%. Those from southern Nicaragua have ( 230Th) ( 232Th) ≈ 2.1 and have ( 238U) ( 230Th) = 1.0–0.9. High-Ti samples have intermediate ( 230Th) ( 232Th) ratios and are strongly enriched in 230Th. Lead isotopic data for all samples plot in the mantle array with 206Pb 204Pb = 18.50–18.63 . Th isotopic ratios for all samples imply source U/Th ratios that are significantly higher than those implied by lead isotopic ratios. All historic samples have ( 210Po) ( 230Th) and thus ( 226Ra) ( 230Th) > 1 , suggesting that lavas erupt less than 8,000 y after generation. Thorium isotopic ratios for all volcanic samples correlate well with 10Be 9Be and B/Be as well as with 87Sr 86Sr and Ba/La ratios from the literature. ( 238U) ( 230Th) ratios do not correlate well with any of these ratios but rather anti-correlate with Th concentrations and published La Yb ratios. These observations, and inferences therefrom, lead to the following preferred, although not completely unique, interpretations. Comparison of thorium, beryllium, and strontium isotopic ratios of the Cocos plate sediments at DSDP 495 with the low-Ti volcanic regression trends shows that the subducted component transferred to the mantle has isotopic compositions similar to the bulk sediments. B/Be and Ba/La ratios of the volcanic samples are well correlated with the isotopes, but the inferred ratios in the subduction component are at least 10 and 5 times higher, respectively, than values in the bulk sediment. These relationships suggest that the subduction component combines the element-transport properties of both a melt and a hydrous fluid. The ( 230Th) ( 232Th) and 87Sr 86Sr ratios corresponding to a 10Be 9Be ratio of zero are about 2.0 and 0.704, significantly higher than expected in the postulated MORB source for the lavas. We attribute these characteristics to subduction related addition of strontium, uranium, plus or minus thorium, and related tracers to the mantle prior to about 4.5 Ma, such that the 10Be signal of this event has decayed away. The observed good correlations are generated primarily by addition of 10Be, U > Th, B and Ba between 4.5 Ma. and ca. 200–300 ka. The most recent event occurred since 200–300 ka and perhaps more recently than 8 ka. It transported only a fraction of the total slab-derived flux of 10Be, B, Ba, and probably thorium stored within the mantle, but added U > Th and may have raised the partition coefficient of uranium over thorium. The magnitude of the uranium excess thus relates to the extent of partial melting of the mantle source during this event. The U-series and beryllium isotope systematics of the high-Ti lavas are consistent with their derivation from the mantle residue after extraction of low-Ti melts.