grassland, dominated by Saccharum spontaneumand
Imperata cylindrica, interspersed with small clusters
of young pioneer trees. Ferns dominated only the
higher regions of Rakata and the balance of species
had also shifted in favour of the flowering plants.
By 1906, the woodland species had increased
considerably in the interiors, although remaining
patchy, and the fern communities were gradually
receding upwards. The grasslands were tall and
dense and so difficult to penetrate that proper explo-
ration of the interiors was greatly hampered. The
woodlands of the lowlands continued their rapid
development, with Ficusspp.,Macaranga tanarius,
and other animal-dispersed trees to the fore.
Forest closure took place over most of the interior
of each island during the 1920s, such that by 1930
very little open habitat remained. This key phase of
system development was fortunately the subject of
detailed investigations by Docters van Leeuwen
(1936). As the forests developed, habitat space for
forest-dependent ferns, orchids, and other epiphytic
plants became available, and their numbers
increased rapidly in response. Conversely, the pio-
neering and grassland habitats were reduced,
species populations shrank, and some species disap-
peared. Rakata is a high island, c.735 m, and altitudi-
nal differentiation of forest composition was evident
as early as 1921. The highest altitudes thereafter fol-
lowed a differing successional pathway, in which the
shrub Cyrtandra sulcatawas for many years a key
component. Although most vegetation changes
appear to have been faster in the lowlands, spread-
ing up the mountain of Rakata, one key species, the
wind-dispersed pioneering tree Neonauclea calycina,
first established a stronghold in the upper reaches,
before spreading downwards. By 1951 it had become
the principal canopy tree of Rakata from just below
the summit down to the near-coastal lowlands. At
the close of the century it was clearly in decline in the
lowlands, as the forests developed a more patchy
character, with a number of other large canopy and
subcanopy species varying in importance from place
to place within the interiors.
The patterns of development on the much lower
islands of Panjang and Sertung were broadly
similar up to c.1930, although differences in the
presence and abundance of particular forest species
were noted. Since 1930, both islands have received
substantial quantities (typically in excess of 1 m
depth) of volcanic ashes over the whole of their
land areas. Historical records and studies of ash
stratigraphies demonstrate that some falls of ash
have been very light, but c. 1932/35 and 1952/53,
and possibly on other occasions, the impact has
been highly destructive (Whittaker et al. 1992b). For
instance, Docters van Leeuwen (1936) described
how in March 1931 the most disturbed forests of
Sertung resembled ‘a European wood in winter’,
and how grasses re-invaded (possibly resprouted)
within the stricken woodlands. Since then, forests
dominated to a considerable extent by the animal-
dispersed trees Timonius compressicaulis and
Dysoxylum gaudichaudianumhave become charac-
teristic of large areas of both islands. The interplay
of slight environmental differences between
islands, differential patterns of landfall, and distur-
bance episodes from the volcano, has produced a
shifting successional matrix across each island,
influencing diversity patterns at fine and coarse
scales in complex ways that are difficult to attribute
in other than general terms to the array of potential
causal variables (Schmitt and Whittaker 1998).
Although different forest types have been recog-
nized on the Krakatau islands, these ‘communities’
are not discrete and forest successional pathways
are in practice more complex than simple summary
successional schema (e.g. Whittaker et al. 1989,
Fig. 15) might be taken to imply. It is notable that of
all the vegetation types of the Krakatau islands,
only the pes-caprae formation and Barringtonia
association were assigned with any confidence to
phytosociological types by the earlier plant scien-
tists. As Docters van Leeuwen (1936, pp. 262–3) put
it, ‘all other associations in the Krakatau islands are
of a temporary nature: they change or are crowded
out.’ Although the coastal systems are similar to
those of other locations in the region, those cur-
rently recognized from the interior by Whittaker
et al. (1989) lack documented regional analogues of
which we are aware. The interior forests of the
Krakatau islands continue to accrue new species of
higher plants, and the balance of species in the
canopy is undoubtedly in a state of flux, with
strong directional shifts in the importance ofKRAKATAU—SUCCESSION, DISPERSAL STRUCTURE, AND HIERARCHIES 133