138 COMMUNITY ASSEMBLY AND DYNAMICS
300(a) (b)2001001883 1903 1923
YearCumulativeImmigrationExtinctionNumber of speciesSpecies per annumNrec1943 1963 1983 1883 1903 1923
Year1943 1963 198305.04.02.01.03.00Figure 5.8Plant recolonization data for Rakata Island (Krakatau) 1883–1989 (redrawn from Bush and Whittaker 1991). Surveys are grouped
into survey periods, and the mean year of each period is used to estimate rates. Shaded lines represent the difference in rate according to
whether minimum turnover is assumed or if ‘recorded’ immigration/extinction is assumed. (a) Cumulative species totals and number of species
recorded at each survey (Nrec), curves for higher plants. Nrec figures are on the basis of the assumption of minimum turnover, i.e. species are
only counted as going extinct if they fail to be found in all subsequent surveys. This assumes that temporary absences are artefacts of sampling
rather than the true extinction and subsequent re-invasion of the species. (b) Immigration/ extinction curves for spermatophyte species. Rates are
calculated as follows: (i) assuming recorded turnover, i.e. that all extinction and immigration records are real (immigration ratenumber of
species not recorded at time 1 but recorded at time 2/time 2 time 1; extinction ratenumber of species recorded at time 1 not found at time
2/time 2 time 1); (ii) assuming minimum turnover (immigration ratenumber of species not previously recorded, but found at time 2/time
2 time 1; extinction ratenumber of species recorded at time 1 not found at or after time 2/time 2 time 1).
found within the first 50 years of recolonization
(1883–1934), and to examine their persistence in the
contemporary flora using first the 1979–1983 datum
adopted in their earlier work and second the more
exhaustive 1979–1994 data. The improved survey
data reduced the number of ‘extinct’ species by
one-third to 94, of which some 41 species could be
regarded as ephemerals that never really estab-
lished breeding populations in the first place, i.e.
the true number of extinctions could be 50 or fewer.
Although the estimates of extinction rate were thus
shown to be unreliable, structural features in the
extinction data were rather more robust. Losses
relate to (1) the original abundance of a species as
recorded in the 1883–1934 period (Fig 5.9a), (2) the
number of islands on which a species occurred
(another indicator of abundance) (Fig. 5.9b), and to
a lesser degree (3) the primary dispersal mode of
the species (indicative of the autoecology of the
species). In short, exactly as we might anticipate, it
is the ephemeral (not proper ‘colonists’) and the
rare which have failed to persist.
Animal-dispersed species have the lowest overall
rate of loss, whilst among sea-dispersed species
there is an interesting contrast between core mem-
bers of the regional strandline flora, which were
found on all three islands and have shown high
persistence, and those species found initially on
only one island, the majority of which have disap-
peared having either never established populations
in the first place, or having been swept away by
coastal erosion (Fig 5.9b). To sum up, the first
110 years of the succession have involved huge
transformations in the environment and habitat
types, and in the character of the forest, but even so,
few of the species that established themselves in the
first 50 years have subsequently failed to persist.
Those that have failed are a statistically non-ran-
dom subset made up mostly of the initially rare
species.