●They become unlike enough to be called different
species.
●Often the barrier breaks down and the isolates
overlap but do not interbreed (or they interbreed
with reduced success).
The first step of this model suggests a vicariant
event (barrier formation). This scenario applies to
certain ancient continental fragment islands (e.g.
Madagascar) and to former land-bridge islands. In
such cases, the island population is assumed not to
have experienced a bottleneck founding event.
For true oceanic islands, the starting point is
different: dispersal across a pre-existing barrier.
The founding population may be quite small, and
thus the founding event may involve a bottleneck
effect (Chapter 7). In either scenario, considering
a now remote island population, it develops in
effective isolation from the mainland source popu-
lation. Subsequent environmentally-determined
bottlenecks, genetic drift effects and selective
pressures in the novel island context then provide
the engine for further differentiation from the main-
land form.
The failure of the isolating barrier may not be a
frequent feature for oceanic islands, but can hap-
pen, for instance, as a function of sea-level change
providing ‘stepping stones’ between island and
mainland. This can lead to double invasions,
where a second invasion of the mainland form
occurs notwithstanding the continued existence of
an oceanic barrier. This was the hypothesis put
forward to explain the occurrence of two species
of chaffinches on the Canaries (Lack 1947a). The
endemic blue chaffinch (Fringilla teydea) was con-
sidered to have evolved from an early colonization
event, to a point where it was sufficiently distinct
to survive alongside a later-arriving population
of mainland common chaffinches (F. coelebs).
More recent genetic work based on mitochondrial
DNA has supported this basic idea, with an
interesting twist, as it seems likely that the later
colonizingF. coelebsreached the Canaries via the
Azores and Madeira (Marshall and Baker 1999).
As often the case with such data, the results
and interpretation are not straightforward, but this
was regarded as the most likely sequence, with
strong support also for a back-colonization event
from the Canaries to Madeira. Hence, it seems
likely that the sequence of population movements
amongst the islands was rather more complex than
implied by the double invasion model. Similar
complexity of movements may be necessary to
account for other Atlantic island phylogenies
(e.g. Kvist et al. 2005).
Interestingly, available evidence shows that the
greatest degree of radiation within the Macaronesian
flora can be attributable to monophyletic genera,
with paraphyletic lineages paradoxically producing
fewer endemic species. Silvertown (2004) reported
that 20 monophyletic endemic genera account for
269 endemic plant species, while 20 paraphyletic
lineages account for just 38 species. Double or
multiple invasions occur within the monophyletic
lineages, but the movements occur at the intra-
Macaronesia scale rather than in the sense of
mainland–island colonization events. Silvertown
attempts to explain this paradox in terms of a com-
petitive mechanism involving niche pre-emption
(see discussion in: Silvertown 2004; Saunders and
Gibson 2005; Silvertown et al. 2005).
The barriers that exist between islands within an
archipelago are undoubtedly important features of
remote oceanic island groups. As Rosenzweig
(1995, pp. 88–9) puts it ‘Suppose propagules occa-
sionally cross those barriers, but usually after
enough time for speciation has passed. Then the
region and its barriers act like a speciation machine,
rapidly cranking out new species.’ The radiations
that characterize many taxa on the Hawaiian
islands reflect this, with interpretations for the rela-
tionships among the 700 (or more) species of
Drosophilainvolving a number of phases of interis-
land movements as a part of this most spectacular
of radiations (e.g. Carson et al. 1970; Carson 1992).
Radiation on archipelagos is examined further in
the next chapter.
An illustration of the contrast between the archi-
pelago context and the single remote island is pro-
vided by that most famous of examples of island
evolution, Darwin’s finches. The Geospizinae are
recognized as a distinct subfamily, found in the
Galápagos archipelago, which comprises nine
islands larger than 50 km^2 (and a larger number ofMECHANISMS OF SPECIATION 203