The significance of anagenesis as a model of island
evolutionary change may well have been over-
looked given the obvious fascination provided by
spectacularly branching lineages like the Hawaiian
honeycreeper-finches. In the absence of systematic
molecular data, a simple yet fairly conservative basis
for estimating the proportion of speciation events
fitting the anagenesis model is to count the number
of cases where a genus is represented on an island/
archipelago by a single endemic species. On this
basis, Stuessy et al. (2006) estimate that for 2640
endemic angiosperm species from 13 island systems,
about one quarter of speciation events match the
anagenesis model.
Anagenesis describes an emergent phylogenetic
pattern and does not imply a tightly specified
explanatory model. For instance, a single endemic
on an island might arise through a combination of a
pronounced founder effect followed by genetic drift
of the newly isolated island population. On the
other hand, the colonist populations may be subject
to strong directional selection pressure in the novel
biotic and abiotic environment provided by a
remote island, and show apparently clear adaptive
features in response. There is insufficient genetic
data to reveal whether species conforming to this
model typically show unusual levels of genetic dif-
ference and thus whether chance or selection domi-
nates. Where anagenesis is important on an island it
indicates that intraisland vicariant events and com-
petitive speciation have each been relatively infre-
quent, such that the motor for speciation is the
further arrival of new colonists combined with envi-
ronmental novelty and possibly environmental (e.g.
climatic) change through time. It seems likely that
anagenesis will be of greatest relative importance on
islands of relatively limited environmental ampli-
tude (Stuessy et al. 2006), and especially where such
islands are themselves isolated (as the case of the
Juan Fernández archipelago) so that opportunities
for repeated phases of island hopping leading to
multiple speciation events are limited.
9.2 The taxon cycle
Until recently it was generally held that islands rep-
resent a form of evolutionary blind alley, i.e. the flow
of colonization events was considered essentially
unidirectional from continents to islands, implying
a generally lower fitness of island endemic forms.
We now have evidence that sometimes island forms
successfully colonize mainlands (e.g. Nicholson
et al. 2005), but the general trend is undoubtedly
mainland to island, and the taxon cycle helps
explain why this is so.
The term ‘taxon cycle’ was coined by Wilson
(1961) in his studies of Pacific ants. The model has
been adapted to fit circumstances by different
authors working on different taxa and regions, and
hence it is not a discrete model allowing easy refu-
tation. The shared features of different invocations
appear to be that the evolution takes place within
an island archipelago, in which immigrant species
undergo niche shifts, which are in part driven by
competitive interactions with later arrivals, such
that the later arrivals ultimately may drive the ear-
lier colonizing inhabitants towards extinction. The
methodology employed in all the early studies was
that patterns of geographic distribution and tax-
onomic differentiation provided the empirical basis
from which were inferred cycles of expansion and
contraction in the geographical distribution, habitat
distribution, and population density of species in
island groups (Ricklefs and Cox 1978).Melanesian ants
As developed by Wilson (1959, 1961), the taxon
cycle described ‘the inferred cyclical evolution of
species [of Melanesian ants], from the ability to live
in marginal habitats and disperse widely, to prefer-
ence for more central, species-rich habitats with an
associated loss of dispersal ability, and back again’
(MacArthur and Wilson 1967, glossary).
Wilson (1959, 1961) recognized differences in the
ranges of the ponerine ants as a function of their
habitat affinities (Fig. 9.1). Marginal habitats (littoral
and savanna habitats) were found to contain both
small absolute numbers of species and higher per-
centages of widespread species. These data were
interpreted as a function of changes in ecology and
dispersal capability as ants moved from the
Oriental region, and particularly its rain forests,
through the continental islands of Indonesia,
through New Guinea, then out across the Bismarck
and Solomon islands, on to Vanuatu, Fiji, andTHE TAXON CYCLE 209