composites. Again they are locally available, are
very dispersive, and produce large numbers of
propagules. They are adapted to a highly disper-
sive life style and are very capable of ‘finding out’
such open environments, on which their survival
depends (they are presumed to be not good com-
petitors). As with the strand species, they are in
essence a subset of those found on the other islands
post-1883.
3 The third set of species, the second phase colonists
of interior habitats, only managed to become estab-
lished in the most recent flora. It will be interesting
to establish the extent to which it may repeat
following any further wipeout. Comparisons with
the older islands suggest that in relation to the
sequence on the older islands this group of species
is likely to be less predictable in colonization
sequence because of the more variable patterns of
seed production and dispersal (mostly by birds or
bats) within the archipelago. For instance, some of
the contemporarily mostabundant tree species on
the other islands (e.g. Arthrophyllum javanicum,
Timonius compressicaulis, and Dysoxylum gaudichaudi-
anum) were recorded on Anak Krakatau at anearlier
stage of successional development than in the
post-1883 sequence.
Whittaker and Jones (1994a) have formulated
these and other observations from Krakatau into an
informal ‘rule table’ in which the dispersal, succes-
sional, turnover, and compositional characteristics
of five such functional plant guilds were tentatively
put forward should the opportunity arise to evalu-
ate them elsewhere. Within plant successions there
may be both autogenic ‘facilitation’ effects and
diffuse competitive effects between guilds, culmi-
nating in later successional communities shading
out earlier ones. Yet much of the structure may be
attributable to a form of relay floristics, mediated
through dispersal attributes of the plants and
dependent, for some guilds, on hierarchical links
between plant and animal communities (Bush and
Whittaker 1991; Whittaker 1992; Thornton 1996;
and compare with Thornton et al. 2001).
Just as in other island systems, as well as there
being structure as to which species assemble and in
what sequence, there may be particular sets of
species which fail to colonize. Whittaker and Jones
(1994a) highlight those types for which they regard
Krakatau as probably undersampling the regional
species pool (Fig. 5.6). More formal comparison of
the Krakatau floras with other sites in the region
has provided support for these observations
(Whittakeret al. 1997). Later successional species
with poor dispersal adaptations, those which have
large, winged, wind-scattered propagules, those
dispersed by terrestrial mammals, and large-
seeded bat-fruits (lacking diplochory) remain
deficient on Krakatau, whereas highly dispersive
forms such as ferns and orchids have been ‘over-
sampled’.
The Krakatau plant and animal recolonization
data thus suggest a general trend in the degree of
predictability through time. While the draw of
early pioneers is fairly predictable as a function in
large measure of their superior dispersability,
precisely which later successional species happen
to establish breeding populations is much less
predictable. Thornton (1996) has made a further,
interesting point regarding the elements of chance
and determinism. He draws the analogy between
community assembly and the construction of a
jigsaw, arguing that the further on in the process a
species joins the system the less influence it can
have over subsequent events, as it operates within
successively narrow bands set by what has gone
before. Thus a late-joining species may be unpre-
dictable in its identity, yet may be predicted to have
relatively little impact on community trajectories.
The parallel with the assembly rules of Diamond is
clear. At early stages of the recovery process, a
number of species combinations or community
pathways are possible, but as the system becomes
more complex, and the jigsaw more complete, the
number of pieces that will fit in to a given gap will
decline, eventually to just one. This form of analogy
has some intuitive value for groups such as birds,
but is harder to sustain for plants. Successional
pathways are not as discrete and limited as such a
picture implies. The islands are continually subject
to the vagaries of a varied environment, ranging
from volcanic eruptions, through storms and
landslides, to droughts. In such a disturbed setting
competitive replacement within natural plant142 COMMUNITY ASSEMBLY AND DYNAMICS