204 Stefan A. Schnitzeretal.
Barker 2000, Schnitzeret al. 2000, Schnitzer
and Bongers 2002), overall, lianas may cre-
ate conditions in gaps that are inimical to
shade-tolerant species.
To test this liana competition hypothesis, we
monitored recruitment, growth, and survivorship
of all woody species for 8 years in experimen-
tal liana removal and control gaps in central
Panama (Schnitzer and Carson unpublished). We
found that the presence of lianas reduced recruit-
ment of shade-tolerant trees (≥1.3 m tall) in
all five sampling years, and by a total of 51%
after 8 years (n =9 removal,n =8 control
gaps). Lianas also substantially reduced relative
growth rates of shade-tolerant trees (Schnitzer
and Carson unpublished). Mortality of shade-
tolerant trees, however, was not significantly
altered by lianas, suggesting that lianas reduce
shade-tolerant tree density by limiting recruit-
ment and growth rather than increasing their
mortality. Nonetheless, shade-tolerant tree mor-
tality was slightly higher where lianas were
present, and thus the total accumulation of shade-
tolerant trees (recruitment minus mortality) after
8 years was 70% higher in gaps where we had
removed lianas. Thus, gaps may fail to maintain
shade-tolerant tree diversity because lianas sub-
stantially restrict shade-tolerant tree recruitment
into gaps. Previously, the lack of shade-tolerant
tree recruitment in gaps has been interpreted as
evidence for seed limitation or dispersal limita-
tion (reviewed by Brokaw and Busing 2000). Our
data, however, demonstrate that lianas also play
a role in limiting shade-tolerant tree recruitment
in gaps. To fully explain why gaps fail to main-
tain shade-tolerant tree diversity, further research
is necessary to determine the relative importance
of plant competition versus seed and dispersal
limitation.
VARIATION IN THE IMPACT OF
GAPS ACROSS BROAD
ENVIRONMENTAL GRADIENTS
The impact of gaps on the maintenance of
species diversity and forest regeneration likely
varies over large-scale environmental gradients.
Theoretically, the influence of gaps should be
greatest in forests where treefalls create steep
resource gradients from the ga pcenter to the
intact understory because these stee pgradi-
ents provide the highest potential for resource
partitioning (Ricklefs 1977). Gaps should be most
important in aseasonal tropical wet forests, which
tend to have dark understories (Asneret al. 2003)
and relatively poor soils (Denslow and Hartshorn
1994). These dark understories are caused by a
combination of high year-round cloud cover, few
deciduous trees, and multiple understory layers
that efficiently intercept light before it reaches the
forest floor. In contrast, gradients in light levels
from a ga pcenter to the intact forest are likely to
be much lower in seasonally moist and dry forests,
which tend to have lower cloud cover and a much
higher proportion of deciduous trees (Conditet al.
2000), and thus allow far more light to penetrate
the intact canopy into the understory, especially
during the dry season. Dry forests, in particular,
tend to have shorter stature, less complex struc-
ture, and lower leaf area, resulting in greater
year-round light penetration.
The steepness of nutrient gradients in treefall
gapsoftenparallelsthatof lightgradients:increas-
ing with greater rainfall. Upon tree death, the
nutrients in the phytomass move into the soil
where they are quickly assimilated by resident
vegetation or leached from the soil. The availabil-
ityof nutrientsinatreefallgapmayprovideasteep
albeit ephemeral gradient, especially beneath the
fallen crown where leaves release a pulse of nutri-
ents into the soil (Brokaw 1985a, Vitousek and
Denslow 1986, Ostertag 1998). Nutrient gradi-
ents should be steepest in aseasonal wet forests,
which may have lower nutrient levels because
heavy year-round precipitation leaches nutrients
out of the soil, although exceptions to this gen-
eral rule certainly exist (Denslow and Hartshorn
1994).
Treefall gaps may also be more common in
aseasonal forests than in seasonally dry forests
because the year-round prevalence of unstable,
waterloggedsoils tends to increase treefall rates
(Hartshorn 1978, Brandaniet al. 1988). If true,
the greater frequency of gaps may somewhat
ameliorate dispersal limitation in wet forests by
reducing the distance between gaps and propagule
sources, thus providing more opportunities for