244 EMERGENT MODELS OF ISLAND EVOLUTION
figure. Further, although both SandImay each
decline during the phase of island
subsidence/destruction, the relative importance of
Smay be greater than in the very earliest stages
(thousands of years) of the island’s history,
during which (as argued by Heaney 2000) most
of the increase in richness comes from
immigration.
To complete the model, we must consider the
loss of SIEs. This can occur through two
mechanisms: either colonizing another island, or
simply going extinct.
●Colonizing another island. When islands exist
for lengthy periods, opportunities exist for some
of their new endemics to island hop to other,
younger islands, thus becoming (at least for a
while) an MIE and no longer an SIE. Many more
lineages follow such a progression rule (see text)
than back-colonize.
●Extinction. We have not attempted to sketch an
extinction-rate curve in the figure, but we anti-
cipate that in terms of bioticdrivers of extinction,
Eshould increase as R approaches K (see figure),
although remaining very low in absolute terms.
However, these biotic drivers of extinction are
likely to be dwarfed by abiotic drivers; at early
stages by mega-disturbance phenomena (damag-
ing eruptions, mega-landslides), in the later stages
of the island life cycle by the gradual attrition of
topographic relief and area (Stuessy et al. 1998),
and over the last 10 000 years, by ‘relaxation’
associated with the loss of area due to postglacial
sea-level rise.
Over the life-cycle of an island as sketched in
the figure, we would therefore anticipate that in
extreme youth ISE, from youth towards island
maturitySIE, interrupted by phases of high
E(catastrophe), and that in advanced old age,
E(S I). If IandSdo come into approximate
balance on an ageing island, then the attrition of
SIE status within the biota through the
progression rule should result in a decline in both
the number and the proportionof SIEs on theoldest islands. This means that for archipelagos
with a sufficiently complete range of island ages,
we expect to see a humped relationship between
speciation rate and island age, anda humped
relationship between the proportionof SIEs and
island age.
From our analyses (not illustrated) of Canarian
arthropod and plant data taken from Izquierdo et
al. (2004) we found that the proportion of SIEs on
the islands is a humped function of island age.
This is consistent with the logic that volcanic
oceanic islands undergo a burst of speciation
relatively early on in their evolution, reflecting the
opportunities provided by the vacant niche space,
and topographic complexity.
If we accept the logic that the inherent carrying
capacity of volcanic islands describes a humped
curve over time, it follows that older islands
(which have longest to reach their Kvalue) are
likely to have greater competitive pressures
operating over longer periods of time than
younger islands. Hence, if the ‘diversity begets
diversity’ logic holds, these islands (e.g. Lanzarote)
should not have lower proportions of SIEs than
islands of younger age (e.g. Tenerife), yet they
show precisely this pattern. We therefore
conclude that the ‘diversity begets diversity’
model (Emerson and Kolm 2005a,b) is at best an
incomplete representation of the data, because it
ignores the non-linearity of relationships between
island life history and the key biological
parameters and rates.
Unfortunately, the empirical evaluation of these
ideas depends on simple exploratory regression
models, based on several simplifying assumptions
and tested on a necessarily small number of
islands. This makes for a blunt instrument given
the complexities of the environmental histories of
archipelagos like the Canaries. The most
profitable route to testing the model proposed
herein would therefore seem to be via analyses of
phylogenies of multiple taxa.
Acknowledgement. This box is a shortened
version of an as yet unpublished manuscript,
written with our colleagues R. J. Ladle, M. B.
Araújo, J. D. Delgado and J. R. Arévalo.