TURNOVER 103The path to equilibrium
One of the few rigorous tests of the process of
colonization and development of equilibrium was
provided by Simberloff and Wilson’s classic
experiments on mangrove islets in the Florida Keys
(Simberloff and Wilson 1969, 1970; Simberloff
1976). These experiments involved, first, a full
survey of all arthropods on individual isolated
mangrove trees (11–18 m in diameter); secondly,
the elimination of all animal life from islets; and
thirdly, the monitoring of the recolonization.
Although seemingly a satisfactory demonstration
of the achievement of equilibrium and one in
which turnover was occurring, there were some
problems. For instance, certain species groups
were regarded as not treating the trees as an island.
Some of these groups were included in the analysis
and others excluded. Simberloff also had problems
determining which species were proper immigrants
as opposed to transients (Williamson 1981). One
interesting features was that a degree of overshoot
occurred, suggesting that the islands could sup-
port more than their ‘equilibrium’ number of
species while most species were rare, but as popu-
lations approached their carrying capacities, com-
petition and predation eliminated the excess
species. As Simberloff (1976, p. 576) described it:
‘more highly co-adapted species sets find them-
selves by chance on an island and persist longer as
sets’. This development was termed an assortative
equilibrium.
Demonstration of equilibrial turnover demands
the most stringent data. The clearest evidence
inevitably comes from ‘islands’ which are relatively
small and simple, and for organisms with fast life
cycles. The mangrove ‘island’ experiments are the
classic first test and they fit this description. These
tiny patches of mangrove consist of a single habitat
type (albeit not to an insect), the arthropods con-
cerned have very short generation times, and can
be expected to respond very rapidly to changing
opportunities and to exhibit relatively little com-
munity development. Such features may be atypi-
cal of islands in general. It is also notable that
although turnover rates were indeed high in the
early stages of the experiment, a couple of years on,
once the assortative process was completed, and
pseudoturnover or transient species removed from
the analysis, the rates were found to have slowed
greatly. This is not expected if area and isolation
remain constant.
A similar study was conducted by Rey (1984,
1985). His islands were patches of the grass Spartina
alterniflora, varying in size from 56 to 1023 m^2 , but
structurally simpler than the mangroves. Once
again, fumigation was used and arthropod recolo-
nization was monitored on a weekly basis. Initially,
the colonization rate was slow because extinction
rates were high. As the assemblages built up, popu-
lations persisted longer and extinction rates fell. Rey
makes a useful distinction between two extremes in
turnover patterns. If all species are participating
equally in turnover, such that each species is
involved somewhere in the archipelago in extinc-
tion and immigration events, he terms the pattern
homogeneous turnover. If, in contrast, only a small
subset of the species pool is involved in turnover,
this would be heterogeneous turnover. If strongly
heterogeneous, the turnover would be inconsistent
with the EMIB. Rey’s data were intermediate
between these two extremes, neither totally homo-
geneous, nor excessively heterogeneous.
Why is heterogeneous turnover problematic for
the EMIB? The most basic equilibrium model of
MacArthur and Wilson (1967) contains the assump-
tion that each species has a finite probability of
becoming extinct at all times. The rate of extinction
would then depend on how many species there
were. This model assumes entirely homogeneous
turnover. However, their favoured model took into
account that the more species there are, the rarer
each is (on average) and hence an increased number
of species increases the likelihood of any given
species dying out. This results in the curved func-
tion for extinction. This is clearly correct under sto-
chastic birth and death processes, provided you
assume that the more species there are on an island,
the smaller, on average, each population is. As
already established, this provides us with the hol-
low forms of immigration and extinction curve and
the expectation that equilibrium species number
should be approached by a colonization curve of
negative exponential form. It also generates the