or (5) missing (trees not found again). During the
final census, standing broken stems were checked for
sprouting of foliage. Longevity of standing broken
stems was calculated by counting the number of
months between the time of damage and when we
noted the stem fallen to the ground.
A total of 1403 trees were tagged. The number of
standing broken was 77 (19.3/ha), or 5%. The num-
ber of trees that were damaged or died during the
study period was 147. The mean number of damaged
and dead trees was 15.9 trees/ha (2.8% of the tagged
trees). Tree damage was highest during the 6-month
wet season (47%), intermediate months in the dry
season (31%), and lowest in the windy-misty season
(22%) (Matelson et al. 1995).
Of the damaged trees, 61% were standing broken
stems, 22% were uproots, 7% were knockdowns, 4%
were standing dead trees, and 10% were missing.
Larger trees were more likely to be damaged or die
(Fig. 9.10). Trees that were damaged or died during
the study were in 19 plant families, 20 genera, and
21 species (Matelson et al. 1995). Trees in most of the
plant families died at similar proportions to that ex-
pected from their distribution in the live population.
However, trees in three plant genera (Cecropia, Ham-
pea, Heliocarpus) died at strikingly higher rates than
expected, more than twice their proportion in the total
marked population. These are gap-colonizing, rapidly
growing, short-lived trees.
Of the 91 standing broken stems that were created
during the study, 30 (34%) had resprouted by the final
census. The identified damaged trees that resprouted
were in 10 plant families; those that did not resprout
were in 18 families. Two of the most common genera
Figure 9.10. Size class distribution (percentage of total
stems) of the 147 trees that were damaged or died during
the study divided into five categories (Ml, missing; SD,
standing dead; KD, knockdown; UP, uproot; SBS,
standing broken stem). (From Matelson et al. 1995)in the sample (Ocotea and Guarea) accounted for 50%
of the damaged trees that resprouted (Matelson et al.
1995).
A total of 116 of the 147 damaged trees died dur-
ing the study period. Most of this mortality was due
to standing broken stems that did not resprout (53%),
followed by uproots (28%), knockdowns (7%), stand-
ing dead (3%), and 9% due to missing trees. The mean
annual mortality was 12.7 trees/ha^1 , equivalent to an
annual mortality of 2.2% for all size classes. The ex-
pectation of further life (calculated as the inverse of
the average proportion of trees dying per year; Putz
and Milton 1982) was 43.8 yr overall. Turnover time,
calculated as the number of years necessary for all of
the originally inventoried trees to die (number of origi-
nally tagged trees/[number of dead and snapped trees/
time observed]; Uhl 1982), was 55.5 yr for 10-30 cm
DBH, and 42.4 yr for trees > 30 cm DBH.
Annual mortality rate in this study is similar to
other tropical forests (1-3%; Putz and Milton 1982,
Hartshorn 1990, Lieberman et al. 1990, Swaine et al.
1990), most of which are from lowland tropical re-
gions. The length of expectation of further life, turn-
over rates, and seasonality of tree damage and death
were also within the range of other forests even though
the wet steep slopes of montane sites, coupled with
strong seasonal winds and high epiphyte loads, might
be expected to increase the incidence of tree damage
and tree falls.
The mode of damage and death in our site was also
similar to other forests. Our study area experienced
65% of tree damage and mortality due to snaps and
22% to uprooting of whole trees. On Barro Colorado
Island (BCI), Panama, 52% of the mortality was due
to snapping and 17% to uprooting (Putz and Milton
1982). In La Selva, Costa Rica, fallen dead trees (up-
roots and snaps combined) accounted for 31% of total
mortality (Lieberman and Lieberman 1987). The major
difference between our montane site and lowland
sites was that tree mortality that occurs as standing
dead stems was rare in the Monteverde forest (3% of
total mortality) compared to lowland tropical sites.
La Selva had 26% of mortality as standing dead trees
(Lieberman and Lieberman 1987); BCI, 15% (Putz and
Milton 1982); and tierre firme forest in Venezuela,
10% (Uhl 1982).
One-third of the standing broken stems that formed
during our study resprouted. Few studies have mea-
sured this phenomenon or incorporated it into their
calculations of tree mortality; the rates of annual
"mortality" reported in some studies may thus be
overestimates. Resprouting would be expected to be
more common where wind and/or forceful storms are
frequent, such as in lower montane forests. In the for-
est on BCI, chronic disturbance is thought to lead to319 Ecosystem Ecology and Forest Dynamics