400 Robin L. Chazdon
many old-growth forests recruit poorly, even in
canopy gaps (Aubréville 1938, Jones 1956). More
than 20% of the tree species in old-growth
forests of southern Cameroon showed a prefer-
ence for recruitment in shiftin gcultivation fields.
The presence of charcoal in almost a third of
the areas sampled supports the view that these
forests are currently undergoing late stages of
succession (van Gemerdenet al.2003). Evidence
from other studies confirms that large-scale dis-
turbances in rainforest areas throughout the
world have been caused by widespread histori-
cal human impact (Denevan 1992, White and
Oates 1999, Bayliss-Smithet al.2003, Williset al.
2004).
If there is no stable endpoint to the succes-
sional process, we are forced to view all forests
as points alon ga successional continuum. We
must also recognize that we may never be able
to reconstruct the initial (pre-disturbance) species
composition of a successional forest. The chal-
lenge is then to identify how biotic and abiotic
factors at a range of spatial scales influence
the successional trajectory of particular forests.
This task may ultimately require experimental
approaches at the scale of entire communities
and landscapes, but such large-scale experiments
will be challenging to execute and manage over
lon gtime periods. A mixed approach involves
conductin gexperimental studies combined with
monitoring of long-term changes in vegetation
dynamics in sets of replicated stands that initially
spanarangeof successionalagesbutsharesimilar
abiotic conditions.
SUCCESSION IN RELATION TO
LANDSCAPE PATTERN
Forest succession occurs within the context of
the surroundin glandscape. As tropical landscapes
become more deforested and fragmented over
time, these landscape patterns will influence both
the pattern and the processes of secondary for-
est succession. In shiftin gcultivation fields of
Belize, composition of woody and herbaceous
species was significantly influenced by distance
to older forest, but species richness and even-
ness were not significantly affected (Kupferet al.
2004). Abandoned fields close to intact forest
had greater densities of successional woody taxa
that are common in seasonally dry, subtropical
forests.
Landscape-level studies clearly show that sec-
ondary forests more frequently develop in areas
close to or borderin gexistin gforest patches and
that species diversity and composition recover
more quickly in areas close to large forest patches
(Tomlinsonet al.1996). Although comparative
studies are greatly needed, these trends may be
more representative of neotropical regions than
in East Asian forests, where mature forest species
often fail to recruit, even in adjacent second-
growth forests (Turneret al.1997). In this case,
recruitment failure may be due, at least in part, to
the extinction or rarity of large frugivores, which
are important dispersal agents for large-seeded
mature forest species (Turneret al.1997). In mon-
taneCostaRica,secondaryforestsweremorelikely
to occur near old-growth forests, at increased
elevation, on steeper slopes, further from roads,
in areas of lower population density, and within
forest reserves (Helmer 2000). Distance to older
forest was a key predictor of species richness and
diversity in a landscape-scale study of secondary
forests in Puerto Rico (Chinea 2002). Chinea
and Helmer (2003) examined the effect of land-
scape pattern on species composition in secondary
forests in Puerto Rico, based on a series of 167 for-
est inventory plots (each approximately 120 m^2 )
that varied in climate, land-use history, and land-
scape structure. Canonical correspondence analy-
sis based on nine variables explained only 16% of
the total variance in species abundances. Land use
covaried with elevation and substrate, so variation
in species composition of secondary forests was
generatedbyinteractionsbetweenbiophysicaland
socioeconomic forces (Chinea and Helmer 2003).
Species composition of abandoned coffee planta-
tions (at higher elevations) remained distinct from
that of abandoned pastures (at lower elevations).
Distance to large forest patches (at least 3800 ha)
was also a significant factor explaining variation
in total and native species richness, although the
effect was small in this large-scale study (Chinea
and Helmer 2003).
Tropical forests are amon gthe most complex
and diverse ecosystems in the world. It should