atolls, or guyots. The chain extends from the Loihi
Seamount, which is believed to be closest to the
hotspot centre, through the Hawaiian islands them-
selves and along the Emperor Seamount Chain, a
distance of some 6130 km (Keast and Miller 1996).
The oldest of the submerged seamounts is more
than 70 million years old, so that there must have
been a group of islands in this part of the Pacific for
far longer than is indicated by the age of the pres-
ent Hawaiian islands. This may have allowed for
colonization of the present archipelago from popu-
lations established considerably earlier on islands
that have since sunk. However, recent geological
findings indicate that there was a pause in island
building in the Hawaiian chain, and molecular
analyses support the notion that few lineages
exceed 10 million years—that is, the pre-Kauai
signal in the present biota is actually rather limited
(Wagner and Funk 1995; Keast and Miller 1996).
The building of linear island groups such as the
Hawaiian chain is explained by J. T. Wilson’s (1963)
hotspot hypothesis, which postulates that ‘station-
ary’ thermal plumes in the Earth’s upper mantle lie
beneath the active volcanoes of the island chain
(Wagner and Funk 1995). A volcano builds over the
hotspot, then drifts away from it as a result of plate
movement and eventually becomes separated from
the magma source, and is subject to erosion by
waves and subaerial processes, and to subsidence
under its own weight. It has been calculated that
the Hawaiian islands have been sinking at a rate of
approximately 2.6–2.7 mm a year over the last
475 000 years (cited in Whelan and Kelletat 2003).
As each island moves away, a new island begins to
form more directly over the hotspot. This process is
thought to have been operating over some
75–80 million years in the case of the Hawaiian
hotspot.
The change in orientation of the Hawaiian chain
(Fig. 2.3) has been attributed to past changes in the
direction of plate movement about 43 Ma,
although recent work has questioned whether in
fact hotspots are really such fixed reference points,
raising the possibility of a combination of plate
movement and hotspot migration to explain the
distribution of hotspot island chains (Christensen
1999). The Society and Marquesas island groups,
also in the Pacific, provide further examples of
hotspot chains, and Nunn (1994) provides a critique
of several other postulated cases.Clustered groups of islands
Many island clusters, once regarded as hotspot
island chains that had become slightly less regular
than the classic examples, have since been realized
to differ significantly from the hotspot model, such
that different groups and different islands within
the groups may require distinctive models. The tec-
tonic-control model, attributed to Jackson et al.
(1972), postulates that, instead of lying along a sin-
gle lineation, the islands lie along shorter lines of
crustal weakness (termed en echelonlines), which
are sub-parallel to each other. This line of reasoning
may explain clusters of islands and chains in which
there is no neat age–distance relation along the
length of the chain, an example being the Line
Islands in the central Pacific (Nunn 1994).
Two of the largest intraplate island clusters are the
Canary and Cape Verde island groups, both in the
central Atlantic and including a number of active
volcanoes. Until recently, it was held possible that the
Canaries were of mixed origins, with the eastern-
most—Lanzarote and Fuerteventura—being land-
bridge islands, once connected to Africa (Sunding
1979). However, it is now established beyond doubt
that the entire archipelago is oceanic in origin, and
that the gap of 100 km and more than 1500 m depth
between the eastern islands and the African continent
has never been bridged. Although thus clearly
oceanic in the strict sense (Box 2.1), the Canaries do
not appear to conform with the classical hotspot
model as (1) they lack a clear lineal geographic distri-
bution, (2) they do not present a simple age sequence,
and (3) all the islands, with the exception of La
Gomera, have been volcanically active in the last few
thousand of years, with Lanzarote, Tenerife, and La
Palma active within the last two centuries.
One explanation for their origin is the propagat-
ing fracture model, which relates the origin of the
Canaries to fractures generated in the oceanic crust
of this region when the Atlas chain began to form
some 70–80 Ma as a result of the collision of the
African and Eurasian plates (Anguita and Hernán
1975). However, some authors continue to favour
the hotspot concept, suggesting a locus somewhere
between the two youngest islands (La Palma andMODES OF ORIGIN 21