Mantle plumes, most geochemists believe, provide a sample of the
deep mantle, though the exact depth from which plumes rise
remains uncertain. Schilling [1991] used the distinctive
isotope and trace element ratios in plumes as ``tracers of upper
mantle dynamics, much as dyes are used to follow flow patterns.''
He investigated 13 plumes located sufficiently close to spreading
centers that they influenced the composition of basalt erupted on
those spreading centers. Schilling used the width of the
geochemical anomaly and the excess ridge elevation along the ridge axis in
a simple plume source---ridge sink model to estimate the volume fluxes
of these plumes. His estimates of fluxes for these plumes agreed within
a factor of 5 with completely independent estimates made by Sleep
[1990]. Schilling [1991] also calculated the excess
temperature of the plumes and found they fell in the range of 160-280
K, again in reasonable agreement with other independent estimates.
Given the uncertainty in Schilling's temperature estimates (
K), plume temperatures appear to be remarkably uniform. Furthermore,
excess temperature and plume geochemistry were unrelated, implying that
all plumes had a common depth of origin with a regulated source of
heat, which Schilling felt was the D
layer (a region of
anomalous seismic velocities at the base of the mantle).
Plumes generally produce linear island chains with only a few,
and sometimes just one, volcanos active at any given time. The
Galapagos are an exception, with at least 13 active volcanos arranged in
a rectilinear pattern over an area of 
km
. This
fortunate circumstance allowed White et al. [1993] to produce a
2 dimensional geochemical map of the upper mantle in the region. They
found isotope and incompatible element ratios defined a horseshoe
pattern, with the most depleted values in the center of the
Galapagos Archipelago and the more enriched values on the eastern,
northern, and southern periphery. They concluded this pattern
reflected thermal entrainment of asthenosphere by the plume as it
undergoes velocity shear in the uppermost asthenosphere. Desonie et
al. [1993] also invoked thermal entrainment by a sheared plume
to explain isotopic variations in the Marquesas Archipelago.
Hawaii continues to be an area of intense study. Kennedy et
al. [1991] found that post-shield lavas do not define a temporal
trend to more depleted compositions, as has been found in other
Hawaiian volcanos. Leeman et al. [1994] also found no
simple temporal trend in the lavas of Kahoolawe, but did find evidence of
a cyclic pattern of geochemical variation, which may reflect variations
in the contribution of plume and lithosphere to Kahoolawe magmas.
In contrast, Kurz and Kammer [1991] found distinct temporal
trends in 
Sr/
Sr, 
Pb/
Pb, and
He/
He in Mauna Loa lavas over the past 30,000 yrs and
concluded 3 distinct mantle sources are involved. Temporal changes in
trace element ratios such as Pb/Ce and Nb/La suggest a systematic change
in the composition of the source of Lanai Volcano [ West et al.,
1992].
Cheng et al. [1993] and Duncan et al. [1994] found that the temporal evolution of Tahiti follows the Hawaiian pattern in which eruption of alkalic magmas with depleted isotopic signatures follows a shield-building phase that consists of eruption of tholeiitic magmas with enriched isotopic signatures. The old shield-building lavas were created by the largest degrees of melting and have the most enriched isotopic signatures. As time passed, degree of melting decreased and isotopic signatures became more MORB-like. Duncan et al. [1994] argued that the temporal, compositional and isotopic variations reflect a progressive change from large degree melts derived from the hot plume core in the shield-building stage to smaller degree melts of the cooler plume sheath in late stages. The plume sheath in their view consists of material with more depleted composition that is viscously entrained during plume rise. In constrast, Hoernle et al. [1991] found no simple temporal pattern in data from Gran Canaria in the Canary Islands.
Plume magma composition may be influenced by a nearby ridge, as well as visa versa. That is the conclusion Weis and Frey [1991] from a study of the basalts from the Ninetyeast Ridge, produced by the Kerguelen mantle plume. The Southern Kerguelen Plateau was created 115 Ma ago by the same plume. Because the Antarctic Plate has been nearly stationary, the Plateau was probably never far from the hotspot. Throughout this long 115 Ma history, the plume has maintained its distinctive isotopic characteristics [ Weis et al., 1993].
Current theory holds that plumes initiate with large bulbous heads that produce flood basalt episodes when the heads reach the surface. Futher evidence to support this idea was reported by Mahoney et al. [1991]. Madagascar was over the Marion/Prince Edward plume when voluminous basalts erupted there about 90 million years ago, and Madagascar basalts bear some isotopic affinities to the Marion ones. Oceanic plateaus also appear to be produced in this way. Isotopic affinities of Ontong-Java basalts and those from the Louisville Ridge seamounts are consistent with the Louisville plume, currently located near the intersection of the Eltanin Fracture Zone and the Pacific-Antarctic Rise, having created the Ontong-Java and Manihiki Plateaus [ Mahoney and Spencer, 1991; Mahoney et al., 1993]. Cretaceous basalts from the Nauru Basin have isotopic compositions displaced toward MORB compared to Ontong-Java basalts. Castillo et al. [1991] suggested that the former were produced at a spreading center influenced by the Ontong-Java/Louisville plume.
The South-Central Pacific is an unusual region in several respects. Mantle plumes appear to be particularly common, and anomalously shallow bathymetry suggests high mantle temperatures. Staudigel et al. [1991] found that extreme isotopic compositions, which characterize modern basalts from this region, also characterized Cretaceous basalts of the Magellan, Marshall and Wake seamounts, suggesting this ``South Pacific Isotopic and Thermal Anomaly'' has existed at least since the Cretaceous. The Line Islands, also mostly Cretaceous, show a particularly complex history [ Garcia et al., 1993]. In some parts of the chain, three phases of volcanism occurred. The earliest basalts can be temporally, spatially, and compositionally associated with the Easter Island mantle plume. However, the other phases of volcanism, which have similar isotopic characteristics, cannot be backtracked to a known hotspot.
It has been apparent for some time that there is an
underlying simplicity to the isotope systematics of oceanic basalts:
they can be divided into 4 or 5 ``species'' on the basis of isotope
ratios [ White, 1985]. Weaver [1991] found that these
groups can also be distinguished on the basis of trace element ratios.
As with isotope ratios, no single ratio is diagnostic, but combinations
of ratios are. Weaver found HIMU (HIMU derives from ``high-
'',
where 
is 
U/
Pb) can be distinguished from EM I and
EM II by K/Nb < 180, Ba/Nb 
, Ba/La < 9 and Ba/Th < 80
(``EM'' stands for ``enriched mantle''). EM I has Ba/Th between 100 and
150, while EM II has Ba/Th <85.