0198566123.pdf

another than is the case for the other archipelagos,
so that the spread of newly evolved endemics
across the archipelago is hampered. For instance,
several islands of the Galápagos archipelago (Santa
Cruz, Santiago, Baltra, Pinzón, Rábida, Santa Fe,
Isabela, and Fernandina) were merged into a single
entity during the last glaciation event, as also hap-
pened in the Cape Verde islands. However, this was
not the case for the archipelago showing the great-
est tendency to multiple-island occupancy, the
Azores, where only the Faial–Pico fusion occurred
(F. García-Talavera, personal communication).
Alternatively, and perhaps of greater relevance, the
high proportion of SIEs on the Canaries may reflect
greater opportunities for within-island speciation
and radiation provided by the high topographic
complexity and habitat diversity of the western
islands of the group, especially within Tenerife, with
its steeply dissected Tertiary age Anaga and Teno
peninsulars, and the central Quaternary age massif,
which is capped bythe 3718 m peak of El Teide.

Island-hopping allopatric radiations: do clades

respond to islands or to habitats?

Oceanic island archipelagos have been described
as ‘speciation machines’ (Rosenzweig 1995). But,
as we have seen, the proportions of single-and multi-
ple-island endemics may be highly variable between
archipelagos, indicating that these ‘machines’

can function in rather different ways. As highlighted

in Box 8.2, the application of the label ‘archipelago

speciation’ does not tell us whether speciation is

adaptive or non-adaptive. Advances in the applica-

tion of molecular analyses in recent decades allow us

to approach these phenomena in rather different

ways, and to ask different questions (Emerson 2002;

Silvertownet al. 2005), such as: (1) do clades respond

to islands or to habitats? (2) do they move across

oceans by routes determined by the shortest dis-

tances? (3) do they move back and forth rather than

unidirectionally?

Within an archipelago we may posit two hypo-

thetical extreme scenarios for the speciation

machine that may be distinguishable in the phylo-

geographical structure of lineages (clades). First, a

particular taxon may respond to islands more than

to habitats, i.e. species inhabiting an island are

genetically closer to other species on the same

island than to species exploiting the same ecosys-

tem on different islands. In this case, a single

founder event per island is involved in the process,

followed by an adaptive radiation driven by

natural selection, which can be more or less com-

plex according to the island’s habitat diversity.

Second, we can think of clades responding to habi-

tats instead of to islands if the species in a certain

habitat are taxonomically closer to other species

occupying the same habitat all over the archipel-

ago, than they are to their island congenerics. This

FROM VALLEY ISOLATES TO ISLAND-HOPPING RADIATIONS 233

Table 9.6 Endemic plant species distribution in the Azores, the Cape Verde and the Galápagos archipelagos (Azores and Cape Verde data
include both bryophytes and vascular plants, whilst Galápagos data only include vascular plants). The data shown are the number and percentage
of plant species in each distributional class. The largest class is highlighted in bold for each archipelago (Sources: Borges et al. 2005 for Azores;
Arechavaleta et al. 2005 for Cape Verde; Lawesson et al. 1987 for Galápagos)

Archipelago 1 2 3 4 5 6 7 8 9 or Total
island islands islands islands islands islands islands islands more endemic
islands species

Azores 4 5 2 5 8 8 6 13 17 77
(%) 5.2 6.5 2.6 6.5 10.4 10.4 7.8 17.0 22.1 (100)
Cape Verde 16 6 12281081972
(%) 22.2 8.3 16.7 2.8 11.1 14.0 11.1 1.4 12.5 (100)
Galápagos 32 16 14 17 9 10 13 8 29 148
(%) 21.6 10.8 9.5 11.5 6.1 6.8 8.8 5.4 19.6 (100)