Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Introduction
Tropical forests are being cleared at a rate of over
150,000 km2 per year (Whitmore 1997), causing extensive loss and fragmentation of existing wildlife habitats.
Fragmentation has myriad impacts on the dynamics of
tropical ecosystems (e.g., Laurance & Bierregaard 1997)
but its effects on plant communities have received only
limited attention (e.g., Williams-Linera 1990; Laurance
1991, 1997; Malcolm 1994; Turner et al. 1996).
We describe the frequency of mortality and damage in
trees of the family Myrtaceae in fragmented and continuous Amazonian rainforests. By assessing the relative importance of edge and area effects and fragment age, we
can better understand the mechanisms of ecological
change in recently fragmented forests.
Methods
Study Area
This study is part of the Biological Dynamics of Forest
Fragments Project (Fig. 1), a long-term experimental
study of Amazonian forest fragmentation (Lovejoy et al.
1986; Bierregaard et al. 1992). The study area is located
70 km north of Manaus in central Amazonia (28309S,
608W), at 100150 m elevation. Local soils are nutrientpoor. Rainfall ranges from 19002500 mm annually with
a pronounced dry season from June to October. The forest canopy is 3037 m tall, with emergents to 55 m. The
local flora is remarkably diverse in tree species (Rankinde Merona et al. 1992).
The study area is surrounded by large expanses of continuous forest. In the early 1980s, a series of 1-, 10-, and
100-ha fragments (Fig. 1) were isolated by distances of
701000 m from surrounding forest by clearing and often by burning the intervening vegetation to establish
Paper submitted May 17, 1996; revised manuscript accepted October
23, 1996.
798
since isolation). We contrasted the two edge-distance predictors because our study area receives prevailing easterly
winds (V. Kapos, personal communication) that could
cause increased windthrow along east-facing edges. Edgedistances were measured from the center of each plot.
Plots were pooled into discrete categories to facilitate
statistical analysis. There were four categories for the
edge-distance measures (1 5 #60 m; 2 5 61100 m; 3 5
101500 m; 4 5 .500 m) and for fragment area (1 5 1 ha;
2 5 10 ha; 3 5 100 ha; 4 5 control) and three categories
for fragment age (1 5 3 years; 2 5 4 years; 3 5 7 years).
The rationale for the edge-distance categories is that microclimatic changes can occur within 60 m of forest
edges in the study area (Kapos 1989), whereas wind-disturbance may be pronounced within 100200 m of
edges and detectable up to 500 m from edges, at least in
some Australasian forests (Laurance 1991).
Statistical Analysis
Tree mortality and damage were patchy in nature, yielding a strongly positively skewed (approximately negative binomial) data distribution. We ranked the dependent variables, which reduced both skewness and
heteroscedasticity among samples. Ranked data were
compared between treatments using one-way ANOVAs,
followed where appropriate by Tukeys tests to contrast
sample means.
The factors describing fragment area and edge-distance were intercorrelated (rs 5 0.660.87; Spearman
rank correlations). To assess effects of area independently of edge effects, all plots ,150 m from any edge
were excluded, and the remaining plots were used to
contrast damage levels in fragmented and continuous
forest. The two edge-distance measures also were inter-
Conservation Biology
Volume 11, No. 3, June 1997
Results
Nearly half (25/56) of the plots had no damage, whereas
the remainder had 657% dead or damaged trees. Annual
mortality rates were low (0.56 6 0.26%, x 6 SE) in forestinterior plots (.100 m from edge) but were nearly seven
times higher (3.85 6 0.72%) in edge plots. Annual rates
of tree damage were over eight times higher in edge
(2.96 6 1.06%) than interior (0.32 6 0.32%) plots.
799
Table 1. Summaries of one-way ANOVAs used to assess effects of landscape features on tree mortality and damage in fragmented and
continuous forests in central Amazonia.
All dead or
damaged trees
Factor tested *
Dist. nearest edge
Dist. eastern edge
Fragment area
Fragment age
Standing dead
trees only
Fallen or
damaged trees
14.39
6.38
0.00
2.71
,0.0001
0.0009
0.9920
0.0840
6.33
8.56
0.77
1.09
0.0010
0.0001
0.3900
0.3490
12.87
2.46
1.33
8.24
0.0003
0.0730
0.2600
0.0016
* Sample sizes varied between treatments and comparisons, but were reasonably large (n 5 730 plots) for all treatments except non-edge plots
in fragments (n 5 4).
Variation in census interval appeared to have little effect on mortality and damage estimates. When both
dead and damaged trees were considered, the length of
the sampling interval accounted for only 5.3% of the total variation in the data set (F 5 3.03, p 5 0.087; leastsquares regression analysis).
One-way ANOVAs revealed highly significant effects
of edge-distance on most mortality and damage parameters (Table 1). Edge effects appeared to have a major influence on tree mortality and damage (Fig. 2). Distance
to the nearest east-facing edge, however, was not a better predictor than simple edge-distance. For overall mortality and damage, edge-distance accounted for more of
the variation in the data set (45.3%) than did distance to
nearest eastern edge (26.9%; based on sums of squares
from the ANOVAs).
Tukeys tests suggested tree mortality and damage
were significantly elevated within 100 m of fragment
margins. When total mortality and damage were considered, samples from both 060 m and 61100 m from
edges had significantly ( p , 0.01) higher values than
those farther from edges. Results were identical when
standing dead and fallen or damaged trees were analyzed separately (Fig. 2).
Discussion
Edge and Area Effects
Our findings suggest edge effects can cause sharply elevated tree mortality and damage in recently fragmented
forests, at least among species of Myrtaceae in Amazonia. Rates of tree death and damage were seven to eight
times higher in edge than non-edge sites. When all dead
or damaged trees were considered, nearly half (45%) of
Conservation Biology
Volume 11, No. 3, June 1997
800
the total variation in the data set (using ranked data) was
explained by the distance of plots to forest edge (Fig. 2).
The distance of plots to the nearest easterly-facing
edge (which receives persistent tradewinds) was a less
effective predictor, explaining 27% of the total variation.
This may occur because windstorms that damage trees
can come from virtually any direction (V. Kapos, pers.
comm.) and may cause complex patterns of forest disturbance (Boose et al. 1994) that obviate any simple relationship between edge aspect and tree mortality. In addition, microclimatic changes in fragments, such as
reduced humidity and increased temperature variability
near edges, may not vary greatly between easterly edges
and those with other aspects.
Edge and area effects are rarely discriminated in studies of fragmented ecosystems (Temple 1986; Laurance &
Yensen 1991; Didham 1997), but in this study we demonstrated that area effects per se had little apparent effect on tree mortality and damage, accounting for only
4% of variation in the data set. Area effects that might influence tree persistence in fragments include population-level processes, such as losses of small populations
via random genetic or demographic events (Shafer 1981),
and community-level phenomena, such as declines in reproduction following losses of specialized pollinators or
seed-dispersers (Powell & Powell 1987; Aizen & Feinsinger
1994). In general, however, we suspect such changes require longer time-scales to be manifested than in the recent (3- to 7-year-olds) fragments examined in this study.
At least for this data-set, edge effects apparently swamped
area effects as the proximate cause of tree damage and
mortality.
Conservation Biology
Volume 11, No. 3, June 1997
Acknowledgments
C. Gascon, S. Lewis, W. Magnusson, J. Chambers, and
three anonymous referees commented on earlier drafts
of the manuscript. This study was supported by World
Wildlife Fund-U.S., National Institute for Research in the
Amazon, Ministerio de Ciencia e Tecnologia (MCT-Brazil), Smithsonian Institution, and the Mellon Foundation.
This is publication number 175 in the BDFFP technical
series.
Literature Cited
Aizen, M. A., and P. Feinsinger. 1994. Habitat fragmentation, native insect pollinators, and feral honey bees in Argentine Chaco Serrano. Ecological Applications 4:378392.
Bierregaard, R. O., Jr., T. E. Lovejoy, V. Kapos, A. A. dos Santos, and
R. W. Hutchings. 1992. The biological dynamics of tropical rainforest fragments. Bioscience 42:859866.
Boose, E. R., D. R. Foster, and M. Fluet. 1994. Hurricane impacts to
tropical and temperate forest landscapes. Ecological Monographs
64:369400.
Bull, G. A. D., and E. R. C. Reynolds. 1968. Wind turbulence generated
by vegetation and its implications. Forestry (Suppl.) 41:28-37.
Chen, J., J. F. Franklin, and T. A. Spies. 1992. Vegetation responses to
edge environments in old-growth Douglas-fir forests. Ecological Applications 2:387396.
Didham, R. K. 1997. The influence of edge effects and forest fragmentation on leaf-litter invertebrates in central Amazonia. Pages 5570
in W. F. Laurance and R. O. Bierregaard, Jr., editors. Tropical forest
remnants: ecology, management and conservation of fragmented
communities. University of Chicago Press, Chicago.
Kapos, V. 1989. Effects of isolation on the water status of forest
patches in the Brazilian Amazon. Journal of Tropical Ecology 5:
173185.
Kapos, V., G. Ganade, E. Matsui, and R. L. Victoria. 1993. 13C as an indicator of edge effects in tropical rainforest reserves. Journal of Ecology 81:425432.
Laurance, W. F. 1991. Edge effects in tropical forest fragments: application of a model for the design of nature reserves. Biological Conservation 57:205219.
801
Conservation Biology
Volume 11, No. 3, June 1997