port in the xylem from root to canopy, and result in
nutrient-starved leaves (Odum 1970, Leigh 1975).
These hypotheses implicate metabolically deter-
mined slow growth. Another hypothesis targets the
pruning of trees. Vigorously growing plants may be
kept small by assiduous pruning by wind (Gleason
and Cook 1927, Howard 1968), drought (Shreve 1914,
Seifriz 1953), and/or grazing by insects (Howard 1969).
At a larger spatial scale, the stature of a forest is
the result of a balance between the growth of its con-
stituent trees and their deaths. Forests recovering from
past disturbances change in structure and composi-
tion through time, and forests may experience epi-
sodes of decline (Jacobi 1983, Stewart and Veblen
1983). They may also achieve a balance at the land-
scape level in a shifting mosaic steady state (Bormann
and Likens 1979) or quasi equilibrium (Shugart and
West 1981). Differences in the rates of disturbance
across the landscape may influence forest stature by
limiting tree lifespans, even if the forest is at quasi
equilibrium. Elfin forest trees may simply not live
long enough to get very large before they are blown
over.
All of the preceding hypotheses are equilibrium
based, in which the stresses imposed on trees are
viewed as chronic. At the extreme, catastrophic dis-
turbances such as major hurricanes or earthquake-
induced landslides may result in nonequilibrium
patches of forest that are short simply because they
have not had time to grow larger since the last catas-
trophe (Gomez-Pompa 1973). The "cyclone forests"
of Queensland, Australia (Webb 1958), and the palm
forests of the lesser Antilles (Beard 1946, 1949) are
examples of this phenomenon.
Trees and forests might be short for adaptive rea-
sons. If storms are a regular feature of the environ-
ment, the population of trees of the area might be
adapted to the mechanical stresses of life there, and
acclimate as individuals to their particular circum-
stances. Increased wood strength is a successful ad-
justment to mechanically stressful sites, and such
responses are widespread among plants (Grace 1977,
Wilson and Archer 1977, Jaffe 1980, Niklas 1993).
Energy and material invested in strength are diverted
from leaf production; investing in strength slows
growth.
These conjectures are not mutally exclusive;
several or all might act together to determine forest
stature. Furthermore, forests may be dwarfed in dif-
ferent places for different reasons. In the most shel-
tered ravines of the Brillante area, trees can grow to
30-35 m tall; on the adjacent wind-exposed ridge-
crests (only about 100 m away), elfin forest thickets
with 5-8-m tall canopy trees occur.
To examine this issue, Lawton (1980, 1984, 1990)
established a permanent study area in a 12-ha water-
shed on the southeastern side of the summit of Cerro
de las Centinelas to monitor environmental condi-
tions, natural disturbances, forest regeneration, and
vegetation structure (Fig. 9.1). Most of the watershed
(10.6 ha) was surveyed with transit and range pole and
gridded into contiguous 20 x 20 m plots. The south-
ern and western margins of the study watershed are
the Continental Divide. The watershed is drained by
a small creek; it runs throughout the wet and windy
seasons, becoming intermittent in the driest months.
The geomorphology of this watershed is described in
Chapter 2, Physical Environment.
Forest structure changes dramatically within the
watershed (Figs. 9.2, 9.3). Canopy trees are generally
15-23 m tall on the lower slopes of the watershed
bowl; the tallest are 25-27 m tall Sapium pachy-
stachys trees growing at the bottom of the ravine on
slopes adjacent to the deep incision of the creek chan-
nel. On the ridgecrests that bound the watershed,
however, the canopy trees are 5-8 m tall and are squat
in architecture. The largest are Clusia sp., with mul-
tiple trunks (each up to 60 cm diameter) rising from
dense arrays of rigid prop roots arching down in a man-
ner similar to mangrove trees. Their trunks branch spar-
ingly; massive limbs support dense crowns as broad
as the trees are tall. The gradient of forest structure
within these Brillante watersheds (ca. 100 horizontal
m and ca. 50 vertical m) encompasses Beard's (1955)
montane rain forest (cloud forest) through montane
thicket almost to elfin woodland. In the schema of
Holdridge (1967), the habitat would be considered
facies of a wind-exposed atmospheric association of
tropical lower montane rain forest. The taller forest
is described as windward cloud forest and the shorter
as elfin forest by Lawton and Dryer (1980).
The spatial scale at which forest dwarfing occurs
is quite clear. But what of the correlates and causes?
A year of climate monitoring along this local gradi-
ent revealed no differences in light availability or tem-
perature at the forest canopy surface (Lawton 1980).
There are no striking differences in soil macronutrient
availability. Only extractable calcium is more abun-
dant in the soils lower in the ravine, but the volcani-
cally derived soils throughout the study area are re-
markably fertile by agricultural standards (Lawton
1980; see Chap. 2, Physical Environment). Symptoms
of nutrient starvation are not conspicuous; chlorotic
discolorations are not common in young or mature
leaves, and trees flower and fruit prolifically.
The only striking environmental difference be-
tween the sheltered Brillante coves with their tall
cloud forest and the adjacent elfin-forested ridge-
crests is their exposure to wind. This is apparent at
the Ventana, the pass across the Continental Divide305 Ecosystem Ecology and Forest Dynamics