A plot of Ce/Yb vs. Ba/Ce, for locality averages, effectively separates mid-ocean ridge basalts (MORB) (Ce/Yb <10, Ba/Ce 1-4.2), oceanic island volcanics (OIV) (Ce/Yb >10, Ba/Ce <6), which are generally hotspot related, and island arc volcanics (IAV) (Ce/Yb <23, Ba/Ce >4.2). The conventional interpretation is that these three types of volcanic environments involve oceanic rift-related, large-volume partial melts (˜20-30%) of a depleted source (MORB), small volume melts (˜5% for alkalic volcanics) of enriched sources related to plumes (OIV), and melts of hydrous-enriched sources during subduction, especially for Ba (IAV). Three OIV sites, however, have average ratios that fall in the MORB field (e.g., Krafla Volcano, Iceland), and these localities also tend to have other geochemical data similar to MORB. Average ratios of Hawaiian tholeiitic shield basalts of Mauna Kea and Koolau volcanoes occupy a restricted field on a plot of Ce/Yb vs. Ba/Ce of 10-18 for Ce/Yb and 2.8-3.1 for Ba/Ce, a field toward which other shield basalts and cone-building volcanics regress. In general, post-shield alkalic rocks have higher values of Ce/Yb than do tholeiites. Peralkalic basalts (basanites, melilitites, and phonolites) have even higher values of Ce/Yb, reflecting smaller degrees of partial melting (perhaps 1-2%) and melting of sources containing phlogopite that were enriched by CO₂-dominated fluids. The minor post-erosion nephelinitic suites of Hawaii (e.g., the Honolulu Series on Oahu, and the Koloa suite on Kauai) generally have values both greater than IAV for Ce/Yb and greater than other kinds of OIV for Ba/Ce in a part of the plot previously not found to be occupied by data. Alkali basalts of both these nephelinitic series have the lowest and similar ratios (Ce/Yb ˜ 25; Ba/Ce ˜ 10). In the Hawaiian Islands, there are two trends. One (a), where phlogopite has been interpreted to remain in the source, generally has Ba/Ce decrease away from the alkali basalts as Ce/Yb increases. The other (b), where phlogopite has been interpreted to enter the melt, occupies a field that is high in both Ce/Yb (>30) relative to IAV and in Ba/Ce (>8) relative to the OIV field.

There are some exceptions, also, for IAV that plot outside the IAV field. The values of Ce/Yb in Mariana Islands samples, for example, are exceptionally low for the IAV (Ce/Yb <5 with many samples <2). Examples of two cross-chain Kasuga Islands, however, have average values of Ce/Yb considerably greater than for any other Mariana Islands data, and individual samples extend from within the IAV field into the OIV field, which may indicate a mixture of IAV and OIV sources (rather than involvement of a hotspot, these island volcanics have been interpreted as magma of OIV entrapped "plums" in an IAV "pudding" by Stern et al., 1993).

Not surprisingly, continental arc volcanics (CAV) are generally similar to IAV, but with somewhat greater dispersion in Ce/Yb, perhaps representing a larger contribution of continental materials to the volcanics. Continental rift volcanics (CRV) are complex. The Antarctic rift data fall in the OIV field, and clearly define a hotspot origin for the rift with little contamination in the continental lithosphere, but most CRV data fall in the IAV field (Rio Grande rift tholeiites, Yellowstone Plateau basalts, Columbia River basalts, East African rift basalts). The Yellowstone basalt samples judged to be least crustally contaminated from other considerations (e.g., through Pb and Sr isotopes) approach closest to the OIV or hotspot field in the Ce/Yb vs. Ba/Ce plot, compatible with a hotspot origin with variable continental lithosphere interactions. The data from the Rio Grande rift have no such trend in Ce/Yb vs. Ba/Ce. Other trace element and isotopic data are suggestive of a different kind of origin, perhaps melting in the continental lithosphere from pressure release or other causes as suggested in the literature.

Carbonatites, kimberlites, and ultrapotassic rocks form extreme end members for the peralkalic rocks on the continents with Ce/Yb values in the hundreds and even exceeding 1,000 in natrocarbonatite. Carbonatites and kimberlite type I, however, have Ba/Ce <8 with few exceptions. Ultrapo tassic rocks and kimberlite type II also have Ce/Yb values in the hundreds but with Ba/Ce >9. These rocks, although rare in the ocean basins (e.g., carbonatite on São Vicente Island in the Cape Verdes archipelago, Indian Ocean) plot similarly to their continental cousins. For Hawaii, the nephelinitic suites of both the Honolulu and Koloa series trend from alkali and alkali olivine basalt ratios toward higher signatures for Ce/Yb for other rock types. The Honolulu series, however, progresses towards smaller values of Ba/Ce for nephelinite-melilitite (Ce/Yb ˜ 85; Ba/Ce ˜ 5-7) near the low end of Ce/Yb found in carbonatite/kimberlite type I, whereas the Koloa series progresses toward higher Ba/Ce (Ce/Yb ˜ 65; Ba/Ce ˜ 14-15) for nephelinite-melilitite with Ce/Yb values near the lower end of kimberlite type II/ultrapotassic rocks. Carbonated phlogopitic sources have been proposed for peralkalic rocks of both oceans and continents. Carbonatites and/or kimberlites are suggested to possibly be present at depth under the Hawaiian nephelinitic series and in other OIV environments containing peralkalic suites.