90 Campbell O. Webbetal.
a number of taxa by historical mixin gof the
Southeast Asian flora, but there is also a strong
seasonality gradient from north to south. We note
first that the Euclidean distance (based on species
presence/absence) between HKK and Pasoh is
0.83 times the distance between Pasoh and
Lambir, and HKK–Lambir is 0.97 times Pasoh–
Lambir, indicatin grelatively lowspeciessimilarity
between Pasoh and Lambir, and overlap of com-
mon species amon gall three plots. However, the
meannearesttaxondistance(thephylogeneticdis-
tance, or age, between most closely related taxa;
Webbet al. 2002) between HKK and Pasoh is
3.0 times the distance between Pasoh and Lam-
bir, and HKK–Lambir is 4.4 times Pasoh–Lambir,
indicatin gmore phylo genetic and therefore eco-
logical similarity between Pasoh and Lambir,
which are both aseasonal rainforests. Lambir also
showed a significant association with rainforest
clades, such as annonaceae/, sapindaceae/, and
anacardiaceae/ (Figure 6.2), and a significant
under-representation of taxa more common in
seasonal forests: phyllanthaceae/, myrtales/, and
asteraceae/.
Global biogeography and assembly of
continental-scale biota
Most of the species diversity within continents
is generated by intra-continental speciation, and
the phylogenetic structure of continental sam-
ples on a global pool will show clustering of taxa
within separate clades, which does not necessarily
reflect any ecological signal. The larger question
is, are deeper angiosperm clades distributed more
evenly around the globe than expected by chance?
Clearly, some deep clades are restricted geograph-
ically (e.g., dipterocarpaceae/ in Southeast Asia).
However, in an influential paper, Gentry (1988)
noted that the family-level taxonomic structure
of rainforest plots was very similar across the
three tropical zones. This could occur by (1) a
more similar than expected phylogenetic com-
position of continental pools or (2) continental-
to-regional, regional-to-local, or local-to-habitat
processes that selectively create habitat-scale plots
with a globally similar deep phylogenetic struc-
ture. The tricky part of addressin gthis question
is decidin gwhat the appropriate null models
should be. An appropriate null model for the
global assortment of plant lineages must take
into account the history of land movements and
climatic zones (e.g., Morley 2000), the emerg-
in gunderstandin gof the bio geo graphic tracks of
major lineages (e.g., Daviset al. 2002), and the
possibility of extensive intercontinental dispersal
(Pennington and Dick 2004). The time is nearing
when a grand review of the movement of higher
plant lineages in space–time will be possible, an
update of Raven and Axelrod’s (1974) landmark
paper. However, we know of no processes that
would cause an over-dispersed (or even) global
distribution of major clades. The stems of clades
that compose tropical rainforests are old enough
that there has been time for members of most
majorcladestodisperseglobally,butweexpectand
observe significant variation among continents in
the phylogenetic composition of each clade.
Any similarity in composition of forest plots
around the world must therefore lie either
in selective intra-continental diversification pat-
terns, such that continental species pools have a
more similar than expected clade composition, or
in continental-to-local processes that may cause
local plot composition to be globally more sim-
ilar than expected. The former is a possibility,
given the conservatism of reproductive charac-
ters and their potential influence on speciation
rates in different clades. Alternatively, what is
the possibility that the taxonomic components
of a forest “fit together” – that, for example, a
“rubiaceae/ and lauraceae/ and sapotaceae/” set
is more stable in some way than a “rubiaceae/ and
rubiaceae/ and rubiaceae/” set? This resembles
the old question of whether there are hundreds
of niches in a forest so that hundreds of species
can coexist. Rather than hundreds of niches,
there may be a smaller number of niches that
do occur regularly on the scale of a forest sam-
ple plot (Valenciaet al. 2004), which, combined
withclade-wideecologicalcharacters,stabilizethe
higher-level taxonomic and phylogenetic struc-
ture of a forest. For example, we might eventually
attempt to deconstruct forest phylogenetic com-
position in terms of conserved functional char-
acters (Wilson 1999), for example: (1) N-fixers
(fabaceae/), (2) emergents (some clades of