on the forest floor but rather remain intact, the trunk
suspended well above the ground. This often results
in exuberant growth of their woody hemiepiphytes
(Lawton and Putz 1988, Williams-Linera and Lawton
1995; see Putz, "Trees on Trees," p. 70).
Many gap attributes are correlated. In the same
set of gaps, gap area (logarithmically transformed)
was correlated with gap aperture (r = 0.51), gap-maker
diameter (r = 0.72), nurse-log area (log-transformed;
r = 0.83), and the area of disturbed soil (log-trans-
formed; r = 0.47). Bigger gaps are made by bigger
trees, which also leave more nurse logs and create
bigger pits and mounds if they uproot. Nurse-log
area, which is important to a set of shade-intolerant
tree species specialized for sapling establishment on
logs, is independently related to gap area (log-trans-
formed; partial r = 0.60), gap-maker diameter (par-
tial r= 0.54), the height of the break (partial r= -0.23),
and gap aperture (partial r = 0.22) (Lawton and Putz
1988).
Because many gap attributes are related to gap size
in Monteverde and other forests (Denslow 1980,1987,
Brokaw 1985), there is a temptation to use gap area
as a proxy to organize the variation that exists among
gaps. Principal components analysis of the variation
among gaps in our study area illustrated these haz-
ards (Lawton and Putz 1988). The first principal com-
ponent, accounting for half of the variation among
gaps, is a contrast between measures of gap size and
the height at which the gap-maker broke. The second
principal component, accounting for about one-sixth
of the variation, contrasts gap aperture ("openness")
with the position of the gap on the slope, the height
of breakage, and gap area. The size of the tree that fell
and the amount of nurse log area created are both
better correlated with the first principal component
than is the area of the gap. Gap area, gap aperture, gap-
maker diameter, nurse-log area, and disturbed min-
eral soil are all important gap characteristics. Al-
though intercorrelations seem intuitively correct,
each gap presents its own opportunities and hazards
to juveniles and cannot be characterized with a single
measure such as gap area.
Gaps and light availability Penetration of light into
the understory of Mont ever de's forests depends largely
on the nature of canopy disruption by limb fall and
tree fall (Lawton 1990). The amount of light in the
middle of a gap depends logarithmically upon its size.
Bigger gaps receive more light than small ones, but
this increase tapers off as the effect of adjacent trees
shading the gap from oblique light becomes progres-
sively smaller. The aperture of the gap is a better in-
dicator of light availability than is gap area. A small
gap surrounded by short trees can be as "open" to light
as a larger gap surrounded by taller trees (Figs. 9.5,
9.6). In the Cordillera de Tilaran, where forest stat-
ure varies dramatically with exposure to wind, this
relationship has important implications. Elfin forest
gaps, though small, may be well lit.
Gaps may also contribute to the lighting of the ad-
jacent forest interior, which creates an "extended gap"
(Runkle 1982). The impact of canopy disturbance on
light availability on the understory in the forest inte-
rior is complex (Lawton 1990). In the elfin forest
understory, PAR availability decreases with increas-
ing distance from the nearest gap, but the correlation
is not strong. In the taller forests of lower elevations,
light availability is not correlated with distance from
the nearest gap. Light availability in specific gaps
suggests that some gaps are walled off from the adja-
cent forest interior by well-developed edge vegetation
(Figs. 9.5, 9.6). Although all regrowth fosters the de-
velopment of this edge (Williams-Linera 1990), some
gaps produce this boundary from the outset, due
either to past disturbance history or to the vagaries
of understory vegetation structure and gap formation.
Our study demonstrates this pattern (Lawton and
Putz 1988). On the southern edge of gap 206, a large
Hieronyma poasana (57 in Fig. 9.7) had, in an earlier
episode of gap creation, snapped its trunk 6 m above
the ground. The crown portion of that earlier gap was
filled by a large Urera elata (62 in Fig. 9.7), whereas
the butt end (now the edge of Gap 206) filled in with
Hieronyma sprouts and hemiepiphytes growing on
the broken Hieronyma trunk; tree ferns filled the
understory. The result was a wall of vegetation that
effectively blocked the light that entered this gap
from penetrating the understory. Since gaps are ag-
gregated, these local historical effects are common,
and generalizations about the influence of gaps on
the light environment in the neighboring understory
are limited.Gap colonization and the partitioning of regenerative
opportunity. Whether natural disturbances provide
sufficient environmental heterogeneity to contribute
to the evolution and maintenance of tropical tree di-
versity is debated (van Steenis 1956, Ashton 1969,
Grubb 1977, Ricklefs 1977, Connell 1978, Denslow
1980, 1987, Orians 1982, Hubbell and Foster 1986).
In the elfin forests of the Cordillera de Tilaran, two
conspicuous guilds of gap colonizing trees exist: spe-
cies that occupy exposed mineral soil, and those that
occupy nurse logs (Lawton and Putz 1988).
In our study area, of the 557 saplings that colonized
the eight elfin forest gaps (8-78 months old), 56 sap-
lings (8.3 saplings/m^2 ) were on the 3% of gap area
covered by disturbed soil of tip-up pits and mounds,
168 (6.4/m^2 ) were on the 10% of the gaps covered by
nurse logs, and 333 (1.3/m^2 ) were on other substrates,310 Ecosystem Ecology and Forest Dynamics