require the presence of fairly specific food plants.
The rate of arrival appears to have peaked during
the period of most rapid floral and habitat diversi-
fication. Extinction also peaked following forest
closure. Bush and Whittaker (1991) attribute four of
the losses to succession, as open habitats were lost,
other losses being possibly due to sampling defi-
ciencies and to the inclusion of likely migratory
species in the calculations. Thornton et al. (1993)
took this view in replotting the data, thus finding a
lower peak in the extinction rate, although the gen-
eral trends in rates were the same.
Figure 5.11 shows the trends in the bird data as
calculated by Thornton et al. (1993). These data are
similar to the butterfly data but have an additional
survey point, for 1951. This survey, by Hoogerwerf
(1953), was missed by MacArthur and Wilson
(1967) who suggested from the three survey colla-
tions for Rakata and Sertung up to 1934 that equi-
librium might already have been reached. The data
in Fig. 5.11 show that numbers subsequently rose
slightly but at a much reduced rate (Thornton et al.
1993). There is evidence in these data of succes-
sional effects involving different trophic levels. The
data also indicate a degree of turnover, the precise
figure involved being highly sensitive to assump-
tions made as to breeding status of birds (Thornton
et al. 1993). The colonization curve is now fairly flat,
but only future survey data can show if the birds of
Rakata have achieved a dynamic equilibrium.
Rawlinson’s analyses of reptile data also indicate
both an important role for abiotic factors and a low
turnover rate. There have been only two extinctions
of stabilized reptile populations, one due to habitat
changes (canopy closure and coastal erosion) and
the other to catastrophic alteration of habitat by
eruptions of Anak Krakatau. No evidence was
found of an approach to a dynamic equilibrium.
Thus, for Krakatau, the fit of the data (Figs 5.8,
5.10, 5.11) to the expectations of the EMIB appears
to be poor, the best relationship being for birds,
although even here the fit is imperfect (Fig. 5.11b).
MacArthur and Wilson’s (1967) stated position was
that the precise shapes of the IandEcurves could
be modified within the confines of the EMIB, pro-
viding that they remained monotonic, ‘When a new
set of curves must be derived for a new situation,
the model loses much of its virtue... ‘ (p. 64).
Departure from monotonicity suggests that the
intersects may be unstable or perhaps that there140 COMMUNITY ASSEMBLY AND DYNAMICS
45(a) (b)35251551883 1903 1923
YearCumulativeImmigrationExtinctionNumber of speciesNrec1943 1963 1983 1883 1903 1923
Year1943 1963 19830Species per annum0.51.00Figure 5.11’Resident’ land-bird recolonization data for Krakatau (redrawn from Thornton et al. 1993). (a) Cumulative total and number of
species recorded at each survey (Nrec), assuming minimal turnover. (b) Immigration and extinction curves, as species per year for inter-survey
periods, as Fig. 5.8.