ALBERT

Notes: Abstract The rate earth element chemistry of a large suite of samples from the Central American volcanic front has been determined to understand their petrogenesis. Different segments of the volcanic front are compared on the basis of their calculated source rare earth patterns as deduced from inverse modeling. The results yield a range in the extent of light rare earth enrichment of the source, as well as in source mineralogy. Moderateto-strong LREE enrichment and high modal garnet contents are observed for the sources of both Guatemala and central Costa Rica, whereas slight LREE depletion and little or no garnet occurs in the Nicaraguan source region. Although distinct source regions beneath each Central American segment are permitted by the modeling, it is more likely that the mantle wedge is broadly homogeneous but locally heterogeneous. Small volume, relatively enriched, garnet-bearing veins surrounded by a matrix of isotopically depleted mantle periodotite could exist throughout the mantle wedge. Apparently distinct sources occur due to the variation in partial melting beneath the different segments, controlled by the amount of subduction-generated flux per unit wedge volume, which in turn is a function of the dip of the subducted lithosphere.

Notes: Abstract During the Mauna Ulu flank eruption on Kilauea, Hawaii, the concentrations in the lavas of the minor elements K, P, Na and Ti, and the incompatible trace elements (analyzed by isotope dilution) K, Rb, Cs, Ba, Sr, and the REE (except Yb) decreased monotonically and linearly with the time (or date) of the eruption. At the same time, the concentrations of the major elements and of Yb, and the ratios of K/Rb, K/Cs, Ba/Rb, 87Sr/86Sr and 143Nd/144Nd remained constant. Most of the scatter in the raw concentration data is removed by a simple correction for olivine (plus chromite) fractionation previously established by Wright et al. (1975). These results are explained by simple equilibrium partial melting of a uniform source. The degree of melting increased by about 20% of the initial value during the course of the eruption. The trace element data are inverted by the method originated by Minster and Allègre (1978) and simplified by Hofmann and Feigenson (1983). The source has the following element (or isotope) ratios: K/Rb=501±7, Ba/Rb=14.0±0.5, Rb/Cs=95±7, Rb/Sr=0.0193 (+0.0045, −0.0090), (Ce/Ba)CN= 1.1±0.1, (Sr/Ba)CN=1.19 (+0.30, −0.19), 87Sr/86Sr=0.703521±0.000016, and 143Nd/144Nd=0.512966±0.000008. The REE pattern of the source has a nearly flat or slightly negative slope (=relative LREE enrichment) between Ce and Dy and a strongly positive slope between Dy and Yb. However, this relative HREE enrichment is poorly constrained by the analytical data, is highly model dependent and may not be a true source feature. The Yb concentration in the source is particularly poorly constrained because it is essentially constant in the melts. On the other hand, this special feature demonstrates that Yb must be buffered by a mineral phase with a high partition coefficient for Yb, namely garnet. The calculated clinopyroxene/garnet ratio in the source is roughly equal to one. In contrast, the source of Kohala volcano had previously been found to contain little or no garnet.

Notes: Abstract A geochemical traverse across Honduras reveals the heterogeneity of the mantle underneath Central America. Alkali basalts from Lake Yojoa (170 km behind the front) have low 87Sr/86Sr but high La/Yb, and elevated incompatible trace element abundances, consistent with derivation from a normal mid-ocean ridge basalt source mantle via low degrees of melting. These lavas lack evidence for an enriched source thought to be intermingled with normal mid-ocean ridge basalt source mantle beneath most of Central America. The amplitude of the subducted slab signature decreases smoothly with distance from the volcanic front. Lavas from Zacate Grande, the area nearest to the volcanic front (17 km behind the arc), display large ion lithophile element enrichment and high field strength element depletion indicating the involvement of subducted material in magma genesis. Components of subducted material are not evident in lavas from Lake Yojoa, the area furthest from the arc. Basalts and basaltic andesites from Tegucigalpa, 102 km behind the volcanic front, are geochemically intermediate between those of Lake Yojoa and Zacate Grande. The lavas from Tegucigalpa show a decreased influence of the subduction component, and are affected by assimilation-fractional crystallization processes at shallow depths. The gradual decrease in the subducted component from the volcanic front to Zacate Grande, Tegucigalpa and finally Lake Yojoa contrasts with the abrupt decrease documented for southeast Guatemala, the only other area in Central America where a cross-arc transect has been studied.

Notes: Abstract The sedimentary section (at DSDP Site 495) on the subducting Cocos Plate has large stratigraphic changes in incompatible elements and element ratios, the result of early carbonate deposition followed by late hemipelagic deposition. Lavas from Central America define both local and regional geochemical trends that reflect the strong influence of the two Cocos Plate sediment units. Element ratios with large stratigraphic variations on the Cocos Plate (e.g. Ba/Th, U/La) define local variations within individual volcanic centers in Central America, indicating that marine stratigraphy controls some geochemical characteristics of the lavas. These local trends can be explained by changing the proportions of hemipelagic sediment input into the magma generation process. These local trends are observed in all the segments of the arc, regardless of the intensity of the slab signature. Regional variations are most clearly seen in element ratios that are nearly constant through the Cocos Plate sediment stratigraphy (e.g. Ba/La, U/Th), suggesting that regional variations reflect differences in the intensity of the flux from the subducting slab. The slab signal is strongest in Nicaragua and along the volcanic front. The signal decreases to the northwest and southeast of Nicaragua and toward the back arc. The large slab signature in the lavas from western Nicaragua occurs in the area with the thinnest continental crust and steepest dip of the slab. The mass flux of incompatible elements into the system is easily estimated, except for elements, like Pb, that have high and variable abundances in the basaltic oceanic crust section. The mass flux of elements out of the system depends on eruption rates, which are variable along the arc and only approximately known. Comparison of input and output fluxes for five different segments of the arc reveals that some elements (K, B, Cs, and Rb) are very efficiently delivered to the volcanoes from the subducted slab. Other elements (Sr, Ba, and U) are returned to the surface with moderate efficiency, whereas some elements (REEs) may come mostly from the mantle wedge with minor slab contribution. The relative order of recycling efficiencies of incompatible elements implies that a hydrous fluid dominates the transfer of material from the slab to mantle.

Notes: Abstract The simplified model of basalt genesis described in Part I of this series, equilibrium partial melting followed by Rayleigh-type fractional crystallization, is applied to a stratigraphically controlled sequence of basalt flows from Kohala volcano. Major-element compositions were determined for 52 samples and show a time-stratigraphic progression from tholeiites through transitional basalts to alkali basalts. Twenty-six of these samples were analyzed by isotope dilution for K, Rb, Cs, Sr, Ba and the REE, 13 for87Sr/86Sr, and 19 for Co, Cr, Ni and V by atomic absorption. After a simple, first-order correction for the effects of fractional crystallization (involving mostly olivine and aluminous clinopyroxene), the major element concentrations cluster tightly, and the incompatible trace elements show monotonic increases in concentration as a function of stratigraphic height. The process identification plot shows that all the (fractionation corrected) melt compositions can be explained by equilibrium partial melting of compositionally identical batches of source material. The REE and Sr are fractionated because of the presence of residual clinopyroxene. Garnet may also be present but in much smaller amounts. In this respect our results differ significantly from those of Leeman et al. (1980). The calculated chondrite-normalized REE patterns of the source are nearly flat to slightly convex upward. Therefore there is no need to invoke special mechanisms, such as metasomatic REE preenrichment of the source, in order to explain the petrogenesis of the suite of lavas. Specifically, Ce concentrations ranging from 20 to 250 times chondritic are all explained by the same calculated source pattern having a chondrite-normalized ratio of Ce/Sm=0.9±0.2. However, the normalized ratio Ce/Ba≅2 shows that the source is not simply primitive mantle.

Notes: Abstract A comprehensive model is developed to explain the major, trace element and strontium and neodymium isotopic characteristics of alkali basalts from Hawaii. The model is similar to that of Chen and Frey (1983) in that it requires mixing of a small melt fraction of MORB-source material with another component to generate the alkalic suite of a particular Hawaiian volcano. It differs from the Chen and Frey model in that the other end-member must be different from primitive mantle if it is to be consistent with both trace element and isotopic data. Alkali basalts and tholeiites from Kauai analyzed in this study show a nearly complete transition in Sr and Nd isotopes. There is a relatively well-constrained array on a Nd-Sr isotope correlation plot that can be explained by two-component mixing of Kauai tholeiite magma and a small amount of melt of East Pacific Rise source rock. After corrections are made for fractional crystallization (involving primarily clinopyroxene and olivine), the Sr and Ba concentrations of Kauai lavas plot along mixing curves defined by the above sources, providing positive tests of the mixing hypothesis. Implications of this model are: (1) the main source of Hawaiian shield-building tholeiites is a mixture of subducted crust, primitive mantle and depleted asthenosphere that has been homogenized prior to melting, (2) early alkalic volcanism (as at Loihi seamount) will be characterized by greater isotopic heterogeneity than will late-stage alkali basalt production, and (3) there are two fundamentally distinct types of alkalic lavas erupted towards the end of magmatism at a given Hawaiian volcano. One represents smaller degrees of melting of the same source that generated shield-building tholeiites (Kohala-type); the other derives from the mixed source discussed in this paper (Haleakala-, Kauai-type).

Notes: Abstract We present new isotopic data for Sr and Nd in basalts and alkalic volcanics from Kohala volcano, Hawaii, which had previously been described by Feigenson et al. (1983). These data complement our own isotopic data presented in that paper and those given in the companion paper by Lanphere and Frey (1986). We show that in spite of appearances to the contrary, there is no significant analytical bias in our previously published analyses. Accidental sampling bias and one erroneous value prevented us from recognizing the isotopic heterogeneity in our previously published data. The new data both confirm the Sr-isotopic distinction between Pololu and Hawi volcanics discovered by Lanphere and Frey and narrow the gap between them significantly. The two data sets agree for the Hawi samples, but the mean 87Sr/86Sr=0.703651±13 for our Pololu basalts is significantly lower than the mean 87Sr/86Sr= 0.703748±18 found by Lanphere and Frey. The Ndisotopic ratios are also heterogeneous, but they overlap for the two formations. We agree with the assessment of Lanphere and Frey that some of our samples originally classified as belonging to the Hawi Formation are actually derived from the uppermost Pololu Formation, but with some stratigraphic ambiguities remaining. We believe that our previous results of inverse modelling are valid for the tholeiitic and moderately alkalic Pololu Formation despite the isotopic heterogeneity because this heterogeneity does not correlate with the trace element chemistry of the Pololu samples. The severe depletion of Sc, which correlates with decreasing CaO/Al2O3 ratios and increasing Yb concentrations, confirms the importance of clinopyroxene fractionation in the evolved lavas of the Hawi Formation. In addition, apatite precipitation did fractionate the P/Ce ratios in the more evolved Hawi lavas, but its effect on the REE abundances is still uncertain and may not be significant. The MgO — P2O5 plot of Lanphere and Frey does not provide compelling evidence against a simple genetic relationship between Pololu and Hawi lavas. The internal consistency of the (fractionation corrected) trace element ratios such as Ba/Ce indicates that Ba is depleted in both the Hawi and the Pololu sources and that these sources do have similar chemistry. Finally, we show that contrary to the conclusions of Lanphere and Frey the REE patterns of Kohala volcanics can be generated from sources with only slightly negatively sloping REE patterns without involvement of garnet, as was indicated by the formal inversion analysis. Models which include garnet yield more highly anomalous source abundance patterns and calculated bulk-source partition coefficients which are inconsistent with the presence of garnet. The persistence of residual garnet is also inconsistent with the absence of significant heavy-REE fractionation among the Pololu basalts.