Review TRENDS in Ecology and Evolution Vol.20 No.

10 October 2005

Tropical forests in a changing

environment
S. Joseph Wright
Smithsonian Tropical Research Institute, Apartado 0843 – 03092, Balboa, Ancón, Panamá, República de Panamá

Understanding and mitigating the impact of an ever- these forests might respond to increasing anthropogenic
increasing population and global economic activity on pressure. I evaluate local effects first because they are
tropical forests is one of the great challenges currently unequivocally detectable, vary widely among countries,
facing biologists, conservationists and policy makers. and set the stage for potential synergisms with global
Tropical forests currently face obvious regional changes, effects. The evidence for global effects suggests that a
both negative and positive, and uncertain global massive reorganization of the structure and dynamics of
changes. Although deforestation rates have increased tropical forests is already underway. The evidence comes
to unprecedented levels, natural secondary succession almost exclusively from repeated censuses of tree plots;
has reclaimed approximately 15% of the area deforested however, and additional independent evidence is needed.
during the 1990s. Governments have also protected
18% of the remaining tropical moist forest; however,
unsustainable hunting continues to threaten many Local anthropogenic influences on tropical forests
keystone mammal and bird species. The structure and Deforestation
dynamics of old-growth forests appear to be rapidly Land use and forest cover are changing throughout the
changing, suggesting that there is a pantropical response tropics (Box 1). Approximately half of the potential
to global anthropogenic forcing, although the evidence tropical closed-canopy forest has already been removed
comes almost exclusively from censuses of tree plots and and the land converted to other uses (Table 1). For
is controversial. Here, I address ongoing anthropogenic comparison, the extant forest cover of the eastern USA
change in tropical forests and suggest how these forests has never fallen below w40% of its potential area [4], a
might respond to increasing anthropogenic pressure. threshold that has already been passed in tropical Asia
and Africa (Table 1). The highest rates of tropical
deforestation yet recorded occurred during the 1980s
Introduction and 1990s [5], and have continued to increase in some
Tropical forest landscapes are changing rapidly as human areas. Across tropical Asia, deforestation rates increased
populations and economies grow. Tropical forests also significantly from 1.8!106 ha yK1 during the 1980s to
have a disproportionate role in global carbon and energy 2.6!106 ha yK1 during the 1990s [6]. In the Brazilian
cycles [1] and support w50% of described species and an Amazon, deforestation rates increased from 1.7!106 ha y–1
even larger number of undescribed species [2]. An under- during the 1990s to 1.8, 2.3 and 2.4!106 ha y–1 in 2001,
standing of anthropogenic change in tropical forests is 2002, and 2003, respectively [7]. The immense deforested
thus crucial to understanding global climate change and area, larger than the combined areas of Canada and
the conservation of nature. France, is just part of the story.
Anthropogenic effects on tropical forests can be grouped Deforestation creates a patchwork landscape, with
into two broad categories. Local effects include local land- forest fragments scattered among towns, pastures and
cover change, invasive species, and timber and bush meat fields and abrupt edges separating forests from other land
extraction. Global effects include changes to the atmo- uses. Each year, deforestation results in 20 000 km of new
sphere and climate caused largely by fossil-fuel consump- forest edge in the Brazilian Amazon alone [8], and the area
tion and remote land-cover change. The challenge is to of forest in fragments of !100 km2 or within 1 km of the
understand the overall impact of this mix of anthropogenic nearest forest edge is greater than the area deforested [9].
drivers, particularly because anthropogenic forcing is Globally, there was 45.3!106 ha of fragmented tropical
likely to only intensify. The population of tropical forest in 1997 [10]. Active deforestation is underway in
countries increased from 1.8 billion in 1950 to 4.9 billion close proximity (defined as 1% clearing in 106 ha) to 15%,
in 2000 and is projected to grow by a further 2 billion 12%, and 8% of the remaining closed-canopy forest in
before 2030 [3]. Tropical and global economies are both tropical Asia, America, and Africa, respectively [6]. This
projected to grow even more rapidly. brings loggers and colonists closer to previously remote
Here, I address ongoing anthropogenic change in forests and contributed to the degradation of an additional
tropical forests with an eye toward understanding how 2.3!106 ha of forest each year during the 1990s [10].
Corresponding author: Wright, S.J. (wrightj@si.edu).
Thus, deforestation fragments and facilitates the degra-
Available online 8 August 2005 dation of remaining forests.
www.sciencedirect.com 0169-5347/$ - see front matter Published by Elsevier Ltd. doi:10.1016/j.tree.2005.07.009
554 Review TRENDS in Ecology and Evolution Vol.20 No.10 October 2005

Box 1. Estimates of tropical forest cover, deforestation and reforestation

Table 1 (main text) compares estimates of potential forest cover, 60% [6]. The principal advantage is annual, global coverage, and the
extant forest cover and recent change in forest cover for the tropics. principal limitation is the coarse 8-km spatial resolution of the AVHRR.
These best available estimates must be compared cautiously because Forest degradation is undetectable and deforestation and the small
methods and forest definitions differ. clearings made for shifting agriculture are difficult to detect [6]. Only
Potential forest cover was estimated from a 1-km resolution the abrupt change when continuous forest is replaced by extensive
satellite-based land-cover data set supplemented by the BIOME3 agriculture is readily detected.
vegetation model [5]. The vegetation model was used to identify the The third estimate of tropical forest cover comes from Landsat
potential vegetation cover wherever crops or natural vegetation Thematic Mapper imagery [10]. Forest was defined to include the
covered O50% or !20% of a 5-min, latitude-longitude grid cell, ‘evergreen and seasonal forests of the tropical humid bioclimatic
respectively [5]. Tropical forest included the ‘tropical evergreen and zone’. The principal advantage is a 30-m resolution, which enables the
tropical deciduous forest/woodland biomes’. The principal limitation quantification of forest degradation, small clearings and the recovery
is the use of 5-min latitude–longitude grid cells [5] and the difficulty of of forests through natural secondary succession. The principal
matching biomes with the definitions of forest used in the three limitation is incomplete spatial and infrequent temporal coverage.
estimates of extant forest cover [14]. Estimates of forest cover, deforestation and reforestation were based
The best known estimate of extant forest cover is compiled each on a random 6.5% sample of the appropriate lands stratified by an
decade from national forest inventories by the Forestry Department of independent assessment of previous deforestation [10].
the UN Food and Agricultural Organization (FAO) [64]. The FAO defines Two robust results emerge in spite of differences in methods and
‘all forests’ as land where O10% of the area is covered by the crowns of forest definitions among the studies summarized in Table 1 (main text)
trees O5-m tall [64]. This definition includes large areas of tropical . First, secondary forest succession offset one of each six to seven
savannah and is therefore not discussed here. The FAO also defines hectares deforested during the 1990s, contributing to the widespread
‘closed forest’ as land where O40% of the area is covered by tree crowns replacement of old-growth forest by secondary forest throughout the
and a continuous grass layer is absent. The FAO estimates extant ‘closed tropics. Second, cumulative deforestation is reaching critical levels at
forest’ cover [64] but not changes in ‘closed forest’ cover. The principal the scale of tropical Africa and Asia. In spite of ongoing secondary
limitations are that the working definition of forest, the dates and the succession, extant forest is likely to be !40% of potential forest cover
quality of national inventories vary among countries. for the tropical portions of both continents. The situation varies widely
The satellite-based Advanced Very High Resolution Radiometer among countries and biomes, and the local situation can be much
(AVHRR) has also been used to estimate closed forest cover [6]. Closed better or much worse than suggested by these continental-scale
forest was defined to include land where tree crown coverage exceeds compilations [14,19].

Secondary forest succesion intense urbanization of most tropical countries, former

Natural secondary forest succession occurs where land is agricultural lands are being abandoned, and secondary
abandoned or temporarily fallowed [11,12], and this site- forests are becoming more widespread [13,14].
specific mix will determine the long-term prospects of Secondary forests are important for global carbon
secondary forests. Secondary forests reclaimed one cycles [10] and could have great conservation value.
hectare for every six to seven hectares deforested in the Tropical secondary forests recover animal species diver-
tropics during the 1990s (Box 1). Land might be sity in 20–40 years [15], but support fewer tree species
abandoned because agricultural yields fall, government than do old-growth forests [16,17]. The species compo-
subsidies change, or greater economic opportunity lies sition of plant and animal communities often differs
elsewhere. In the Brazilian Amazon, for example, the between secondary and old-growth forests [15–17]. In
rural:urban income ratio averages 0.6, the percentage of particular, weedy species can become more widespread as
the population living in rural settings declined from 57% secondary forests become more important, homogenizing
to 40% in 16 years, the number of rural establishments species composition over large areas. The long-term
declined by 23% in 11 years, and secondary forest has conservation value of secondary forest will vary with the
reclaimed 31% of the once deforested land [12]. More proportion of species whose distributions are restricted to
generally, rural–urban migration is contributing to the old-growth forest.

Table 1. Potential and extant forest cover and recent rates of deforestation, reforestation and forest degradation in the tropics

a
DISCover 1-km resolution satellite imagery supplemented by the BIOME3 vegetation model [5] for countries whose geographical centers fall within the tropics [14].
b
National forest inventories compiled by the FAO [64] for countries whose geographical centers fall within the tropics [14].
c
Advanced Very High Resolution Radiometer 8-km resolution satellite imagery for all lands between the tropics [6].
d
LandSat 30-m resolution satellite imagery for the ‘evergreen and seasonal forest of the tropical humid bioclimatic zone’ for all continents plus the ‘dry biome of continental
Southeast Asia’ [10]. Excludes Mexico and the Atlantic coastal forest of Brazil.
e
Rates of deforestation, secondary forest succession and forest degradation are for 1990–1997.

www.sciencedirect.com
 Review TRENDS in Ecology and Evolution Vol.20 No.10 October 2005 555

Spatial variation in land-use change Still, protected forests suffer less degradation than do
The local loss of forest cover can be extreme. Myers et al. nearby unprotected forests throughout the tropics [24].
[18] identified 25 endemicity ‘hotspots’ that include the
entire ranges of 44% and 35% of all vascular plants and Poachers
terrestrial vertebrates, respectively. Of these, the 16 Few tropical protected areas exclude poachers and, as a
hotspots that support tropical forest have lost significant consequence, a pantropical bush meat crisis threatens
percentages of their original vegetation cover (Figure 1) many keystone species throughout the tropics. Forest
[19]. Simple species–area projections incorporating this ungulates, primates and other large vertebrates provide a
level of habitat loss indicate that extinction now threatens significant proportion of protein consumed by humans in
w50% of the tropical hotspot endemics that require old- many tropical settings [25]. Hunters routinely extirpate
growth forest cover [19]. Tropical dry forests are the most the more sensitive species, such as large primates and
endangered forests worldwide with, for example, !2% of cats, which are characterized by delayed maturation and
the original tropical dry forest remaining in Central few young, and reduce the abundances of less sensitive
America [20]. species, such as deer and forest pigs [26]. Game species
By contrast, old-growth forests have largely survived include granivores that eat large palatable seeds,
where human population density is low. Rural population browsers that eat palatable leaves, and frugivores that
density explains 66–83% of the among-country variation disperse seeds and are sometimes the only primary
in the percentage of potential closed-canopy forest dispersal agents of large-seeded plants. The primates,
remaining in tropical Africa, America and Indo-Malaya carnivores and rodents that disperse and consume the
[14]. For example, Bangladesh (975 people kmK2) and seeds of many coccosoid palms provide an example [26].
Haiti (293 people kmK2) support !10% of their potential Thus, hunting also indirectly alters the species compo-
closed-canopy forest, whereas French Guiana (2.0 people sition of regenerating plants, favoring species with
kmK2), Gabon (4.6 people kmK2), Papua New Guinea (10.4 palatable foliage, palatable seeds, and seeds that are
people kmK2) and Suriname (2.7 people kmK2) support dispersed by wind, bats and small birds [26]. Habitat loss
O90%. and unsustainable hunting could therefore result in
Old-growth forests also survive in protected areas. species extinctions that permanently alter remaining
Globally, 18% of all tropical and subtropical moist forests tropical forests.
and 9% of all tropical dry forests are nominally protected
[21]. Thus, tropical moist forests are potentially among the Invasive species
best protected biomes; of 14 major biomes, only temperate Africanized bees and feral pigs have successfully invaded
coniferous forest receives a greater level of global species-rich, old-growth tropical forests throughout the
protection [21]. Nevertheless, large numbers of tropical Neotropics and Australia, respectively. Invasive plants are
species have small ranges outside existing protected areas also important components of many secondary tropical
and tropical forests also comprise 65% of the global area forests [17,27]. With these notable exceptions, there are
prioritized for future protection [22]. Protected areas are relatively few reports of invasive species in tropical
also only as effective as the governments that protect them forests, particularly old-growth forests on continental
and can collapse during political and economic crises [23]. land masses. For plants, this might reflect the tendency
for humans to transport light-demanding species associ-
ated with agriculture and open habitats that are unable to
10
regenerate in closed-canopy forest [27]. However, planta-
tions of shade-tolerant, tropical hard woods are increas-
ingly being established outside their native range and will
8
Number of endemicity hot spots

increasingly provide sources of exotic propagules appro-

priate to conditions in old-growth forests [27]. Old-growth
forests will also become more vulnerable to invasion as
6 fragmentation and degradation intensify. Thus, invasive
species are likely to become more important in future
tropical forests.
4
Global anthropogenic effects on lowland tropical forests
The evidence for biotic responses to global climate change
2 is clear for high latitudes, but sparse and controversial for
tropical latitudes. Many temperate and boreal species
reproduce earlier, experience longer growing seasons and
migrate or have extended their ranges to higher latitudes
0
or elevations [28]. Similar changes have occurred on
0 20 40 60 80 100
tropical mountains, where the elevations characterized by
Original vegetation lost (%)
persistent cloud formation are increasing with global
TRENDS in Ecology & Evolution warming [28]. Lowland tropical climates are also changing
Figure 1. The percentage of original vegetation cover lost from hotspots of
in ways that are likely to affect forest organisms (Box 2).
endemicity that support tropical forest. Median Z90% cover loss. Data from [19]. However, to my knowledge, there are no studies
www.sciencedirect.com
556 Review TRENDS in Ecology and Evolution Vol.20 No.10 October 2005

Box 2. The changing tropical environment 7

Autotrophic plants require CO2, light, water and nutrients and their
metabolism varies with ambient temperature. Anthropogenic change 6
affects each of these factors either globally (CO2, temperature and light)

Relative change (mean % y–1)

or regionally in the tropics (precipitation, nutrients and possibly light)
and must, therefore, also affect plant function [39]. 5
Atmospheric CO2 concentrations have increased by 30% since the
19th century. This might increase photosynthetic carbon uptake,
particularly where other resources (i.e. water, nutrients and light) are 4
not limiting. This will also decrease transpirational water loss when
plants respond by reducing stomatal aperture.
3
Tropical temperatures have increased by 0.268C per decade since the
1970s, with minimum night-time temperatures increasing more
rapidly than day-time temperatures [65]. Global climate models driven 2
by anthropogenic greenhouse gas emissions predict overall increases
in tropical temperatures of 1–48C in the 21st century [1]. Higher
temperatures increase respiratory CO2 losses by all plants [55], and 1
higher day-time temperatures might also affect photosynthesis [39].
Solar irradiance penetrating the atmosphere decreased by 4% to
6% from w1960 until the late 1980s and has increased by a similar 0
ty t y s
si en lit as ea ity
amount since then [66,67]. The cause is believed to be anthropo- n ta ar ns
de itm or om l de
genic loading of the atmosphere with particulates followed by a ru M Bi sa
an ec Ba em
effective air pollution control, economic change in Eastern Europe,
Li R St
and recovery from the 1991 eruption of Mt Pinatubo [67]. Solar
TRENDS in Ecology & Evolution
irradiance limits net primary production by closed-canopy tropical
forests because leaf area indices (m2 of leaf area per m2 of ground Figure 2. Relative change (mean % yK1 G95% CI) in various indices of the structure
area) range from four to six or seven and heavy shade limits and dynamics of old-growth Neotropical forests [29–33]. All values are for the most
photosynthetic carbon uptake by most leaves. Experimental inclusive data set and incorporate the original author’s corrections for possible
augmentation of light increased carbon uptake, stem elongation biases. I annualized recruitment and mortality rate changes assuming exponential
and reproduction by a tropical canopy tree [58], confirming that change between mean values reported for two census intervals in [31].
changes in solar irradiance will have large effects on forest carbon
balance and dynamics. usually 10 cm in diameter at breast height (dbh), and
Precipitation is unchanged since the 1970s in tropical America, has median plot area is w1 ha. Significant changes in tree
decreased significantly in tropical Asia, and has decreased more
strongly in Africa, particularly across the northern margin of the
species composition have been detected from repeated
African moist tropics [65]. Global climate models predict that tropical censuses of such plots in the central Amazon [29]. The
precipitation will change regionally with local increases of as much RAINFOR network brings together investigators with
as 8% and local decreases of as much as 35% [1]. Reduced similar data from across the Amazon (http://www.geog.
precipitation will limit seasonally dry tropical forests by extending
leeds.ac.uk/projects/rainfor/). RAINFOR authors have
the seasonal drought.
Nitrogen deposition falls off with distance from human sources recently reported significant increases in recruitment
and is now relatively modest (1–7.5 kg haK1 yK1) over most extant rates, mortality rates, aboveground biomass, basal area,
tropical forests [68]. Much higher levels (10–50 kg haK1 yK1) are stem density and liana stem density for old-growth forests
projected for 2050 as industry and fertilizer use intensifies [68]. [30–33]. The reported rates of change are quite astonish-
Deposition is greatest in tropical Asia, intermediate in tropical Africa,
ing (Figure 2). A hectare of forest that initially supported
and lowest in tropical America [68]. The deposition of other
pollutants is often well correlated with nitrogen deposition [39]. 200 Mg haK1 and 600 stems haK1 would just one decade
Multiple pollutants (e.g. nitrogen, ozone, sulfates, etc.) are also likely later support 210 Mg haK1 and 611 stems haK1. Changes
to stress tropical forests. of this magnitude involve globally significant amounts of
To summarize, remote anthropogenic drivers have altered all carbon and should be of central importance to most biolo-
aspects of the abiotic environment that influence plant function in
the tropics. Although forest plants are likely to respond to these
gists working in tropical forests because their study
changes, the size, and even the direction, of the net response is unclear. organisms are adjusting to rapid changes in forest
structure and dynamics.

addressing the response of a resident lowland tropical

The evidence questioned
forest animal to climate change. The evidence for plants is
The first report of rapid change from repeated censuses of
also limited and sometimes contradictory [29–40]. Here, I
lowland tropical forest plots was published in 1994 [34,35],
emphasize limitations, contradictions and alternative
and possible methodological biases were soon being
interpretations in the hope of stimulating new analyses
discussed [36–38]. RAINFOR authors have evaluated
of the responses of lowland tropical forests to global
many of these biases and conclude that rapid change is
change.
real [31–33,39,40]. This discussion has, however, over-
looked two possibilities.
The changing structure and dynamics of lowland tropical First, increases in recruitment, stem density, basal area
forests and aboveground biomass could reflect the discovery of
The evidence for change in old-growth tropical forests previously overlooked trees as census personnel changed.
comes mainly from censuses of the woody stems of trees Trees, particularly small ones, are inevitably overlooked
and lianas that are larger than a threshold size in fully when plots are first censused. Known trees are system-
enumerated plots [29–33]. The threshold stem size is atically relocated in subsequent censuses, but newly
www.sciencedirect.com
 Review TRENDS in Ecology and Evolution Vol.20 No.10 October 2005 557

recruited trees and previously overlooked trees might not smaller suppressed trees gradually die and the survivors
be located. The RAINFOR plots were also established by grow.
many different investigators with many different objec- The use of minimum size thresholds when investi-
tives. For example, zoologists interested in stand-level gating closed-canopy forests could complicate this signal.
fruit production or botanists interested in local diversity Several small trees are likely to grow past the minimum
might have tended to overlook sterile trees near the size threshold and ‘recruit’ into the censused population
minimum size threshold. Many plots were subsequently for each canopy tree that dies and falls to form a forest gap.
recensused by RAINFOR personnel. The possible bias that Thus, recruitment will exceed mortality at the scale of a
might result could be evaluated by tracking changes in single treefall gap. At equilibrium, this localized burst of
census personnel and apparent plant community responses. recruitment would be offset by mortality elsewhere as
The implicit assumption of equilibrium and the use of recruits died in older, recovering gaps. When past
minimum size thresholds raises a second possible bias. disturbance has instead established an even-aged cohort
Forest dynamics and structure should be stable at equili- of canopy trees, suitably old gaps will be relatively rare
brium. Thus, any spatially consistent temporal trend and the rate of new gap formation and subsequent
becomes evidence for an effect of global change. Alter- recruitment will outpace mortality as the even-aged
natively, most forests might not be at equilibrium but cohort thins. Hence, recruitment will appear to increase
might instead be recovering from past disturbance. In this before mortality. This is a robust outcome of the RAINFOR
case, a spatially consistent temporal trend would be analyses [31]. Stem density will also increase because
consistent with widespread disturbance. Have sites several small recruits replace each canopy tree that fell to
scattered widely across the Neotropics been disturbed create a gap. Thus, the use of minimum size thresholds in
recently? More precisely, have disturbances occurred in closed-canopy forests complicates the familiar pattern of
the relevant past, but not since tree plots were first change expected for a thinning stand.
established during the 1970s (only one of 97 RAINFOR RAINFOR authors have previously discounted past
plots was established earlier, in 1967 [31])? The relevant disturbance as a cause of observed changes in forest
dynamics because increases in recruitment and stem
past reaches back several centuries because tropical trees
density could not be reconciled with recovering, thinning
routinely survive this long. For example, when 20
stands. This possibility now needs to be reevaluated.
emergent trees from the central Amazon were 14C dated,
RAINFOR authors have also discounted El Niño events as
the youngest individual was O200-years old, six indivi-
a source of disturbance because two very strong events
duals were O600 years old, and the oldest individual was
occurred during 1982–1983 and 1997–1998 after most tree
O1400 years old [41].
plots were established [31,39]. The most severe regional
At least three types of widespread disturbance have
drought coincided with the 1925–1926 El Niño event,
occurred over the past few centuries. First, the Spanish
however, when fires were widespread and the Amazon
Conquest precipitated the collapse of many Amerindian
reached the lowest level yet recorded at Manaus [48,49].
cultures and initiated widespread secondary forest succes-
More importantly, multiple sources of past disturbance
sion beginning 500 years ago in Mesoamerica and as
need to be evaluated. Soil charcoal deposits left by
recently as 300 years ago in parts of the Amazon [42–44]. prehistoric fires have been found near three RAINFOR
Second, climate fluctuations associated with a w200–year sites [42,50,51], which suggests that these, and possibly
periodicity in solar activity repeatedly destabilized pre- other sites, might be recovering from past disturbance.
Colombian cultures across the Neotropics and would have The implications for forest structure and dynamics could
impacted forests most recently during the early 1800s be evaluated by contrasting forest plots where the
[45]. Third, severe El Niño events have brought drought disturbance history is known. There are additional
that can increase tree mortality and the risk of forest fires reasons to question the RAINFOR estimates of rates of
across southern Mesoamerica, the Guiana Shield and the change in old-growth tropical forests (Box 3) and, thus,
northern and eastern Amazon Basin [46,47]. These three independent assessments of the stability of lowland
widespread disturbances combined could provide a suit- tropical forests are needed.
able composite of continental-scale disturbances, indicat-
ing that widely scattered forest plots might indeed be Independent evidence of change in lowland tropical
recovering from disturbances that occurred decades if not forests
centuries ago. To my knowledge, there have been four independent
Disturbance introduces cohorts of trees of similar age. attempts to detect long-term trends that might reflect
The even-aged cohort might include an entire stand after a global change in lowland tropical forests. On Barro
severe disturbance (i.e. fire or agricultural clearing) or a Colorado Island (BCI), Panama, the old-growth forest
modest proportion of the trees in a stand after a moderate has escaped fire and agriculture for at least 1500 years
disturbance (i.e. drought or flooding). In either case, the [52]. Here, aboveground biomass was almost constant
even-aged cohort joins the forest canopy and will thin between 1985 and 2000 over 50 hectares where all stems
through time. Given the long lifetimes of tropical trees O1 cm in dbh were censused at five-year intervals [38].
[41], the familiar signature of a thinning stand should be The contribution of lianas to total leaf litter production
detectable for centuries. Aboveground biomass and basal increased, however, from w9% to 13% between 1986 and
area should increase, recruitment should be infrequent, 2002, respectively [53]. In the Kibale National Park,
and stem density and mortality should decrease as the Uganda, a 30-year record compiled from four shorter
www.sciencedirect.com
558 Review TRENDS in Ecology and Evolution Vol.20 No.10 October 2005

in light availability [58] and increases in CO2 concen-

Box 3. Limits on net carbon uptake by old-growth tropical
trations for individual branches of canopy trees [59] and
forests
for understorey seedlings [60]. The tropics lack a single
Recent reports of net carbon uptake by old growth tropical forests forest-level elevated CO2 experiment and have very few, if
sometimes exceed the size of the missing global carbon sink [69].
any, long-term experiments in secondary forest. Until new
Estimates range from w0.5 Mg C haK1 yK1 for repeated censuses of
forest inventory plots [33] to 1–6 Mg C haK1 yK1 for eddy flux data become available, we must accept the available
measurements of net ecosystem CO2 exchange [70,71]. Three evidence indicating that global change is causing a
types of evidence now suggest that these values are too high. massive restructuring of lowland tropical forests with
The eddy flux measurements are suspect for two reasons. The rapid increases in aboveground biomass, basal area, stem
method models air movement and CO2 concentrations between
density, mortality, recruitment and the relative import-
the vegetation and the atmosphere. The model assumptions fail at
the low wind speeds commonly observed at night in tropical forests, ance of lianas and canopy tree species at the expense of
which alone could fully explain the reports of substantial net carbon understorey tree species [29–33].
uptake by tropical forests [70]. The second possibility is that early
eddy flux studies might have been conducted in maturing forests, The future of tropical forests
where net CO2 uptake is to be expected. Large net CO2 losses have
recently been reported from an Amazonian old-growth forest [71]
The broad outline of the future of tropical landscapes is
A forest dynamics model indicates that the more modest net clear. As the populations of tropical countries increase by
uptake values of w0.5 Mg C haK1 yK1 might also be too high [72]. two billion over the next 25 years, agriculture will replace
The model incorporated tree mortality, growth and recruitment rates extensive tracts of old-growth forest [61]. Elsewhere,
observed for the central Amazon. Growth rates were then increased
secondary forests will become more important as degraded
by 25%, which is the maximum increase observed when CO2 is
experimentally doubled in the presence of ample light, water and land is abandoned and urbanization intensifies [13,14].
nutrients [72]. Model simulations demonstrated that the maximum Given that 18% of moist and humid tropical forests and 9%
potential increase in aboveground biomass attributable to increased of dry tropical forests have governmental protection [21],
atmospheric CO2 was much less that the increase observed from future tropical landscapes will also include islands of
repeated plot censuses [72]. Either a more potent limiting factor has
secondary forest and protected old-growth forest isolated
increased (light is a possibility [39]) or the central Amazonian forests
are recovering from past disturbance [72,73]. in a sea of land converted to agriculture and other human
The final line of evidence that suggests limited potential for uses. This resembles modern temperate landscapes except
increased carbon sequestration in old-growth tropical forests more old-growth forest is likely to survive in the tropics.
concerns levels of carbon storage [69]. Trees store carbon in their The future of those surviving tropical forests is
tissues as non-structural carbohydrates, and the tropical trees that
have been examined have consistently high non-structural carbo-
uncertain. There is no information about the possible
hydrate concentrations [69,74]. This suggests that current photo- responses of secondary tropical forests to global change,
synthesis levels meet, or even exceed, the carbon requirements of and old-growth forests might already be undergoing a
maintenance and growth. Forest trees with large carbon stores at massive restructuring (Figure 2) [29–33]. Positive feed-
current atmospheric CO2 concentrations and light levels are unlikely back might intensify global change effects; for example, an
to sequester still more carbon in response to future increases in
either resource unless some additional limiting resource is also
increase in tree mortality might favor the proliferation of
augmented [69,74]. lianas that then further increase tree mortality [30,62]. It
is equally easy, however, to imagine negative feedback
mechanisms. For example, the proliferation of lianas
records suggests changing levels of reproductive activity might favor tree species with straight trunks, smooth
by forest trees [54]; and, at La Selva, Costa Rica, diameter bark and other traits that inhibit lianas. In my opinion,
growth rates decreased among surviving individuals for the present understanding of long-term change in tropical
cohorts of nine species measured annually for 17 years forests is too limited to speculate about possible feedback
[55]. To summarize, the four independent studies confirm mechanisms.
that lianas are increasing in importance on BCI and Other direct effects can be anticipated, however. Many
suggest that trees are reproducing more often at Kibale; of the remaining forests will be degraded and valuable
they also indicate that aboveground biomass is almost timber will often be removed. The well known influences of
constant on BCI and that tree growth rates are declining edges and fragmentation will alter the dynamics and
at the La Selva Biological Station. structure of many surviving forests as land converted to
These mixed outcomes necessitate additional indepen- human use comes to dominate the landscape [8]. More
dent assessment of the stability of old-growth tropical exotic species will invade as sources of propagules are
forests. At least three assessments are possible based on established ever closer to these remaining forests [27].
existing long-term records. First, tree plot data and/or Game species will be removed wherever human poverty
annual tree rings could be used to explore temporal trends and forests meet [25,26].
in the growth rates of individual trees. Second, the many The scientific community has an important role to play
published studies of plant phenology could be used to as the future of the tropics unfolds [63]. Basic research will
explore possible changes in levels of plant reproduction help to understand the dimensions and mechanisms of
[54]. Finally, the many published studies of fine litter forest responses to anthropogenic forcing. Conservation
production could be used to explore possible changes in scientists will help to mitigate the number of species lost
this major component of net primary production. Experi- to extinction by enhancing the effectiveness of the network
ments must also have a crucial role. They have been used of protected areas [21,22]. Other applied research will help
to explore the responses of old-growth forests to increases to rehabilitate degraded lands and to improve agricultural
[56] and decreases [57] in moisture availability, increases yields and living standards. The tropics support over half
www.sciencedirect.com
 Review TRENDS in Ecology and Evolution Vol.20 No.10 October 2005 559

of all species and over two-thirds of all people. Without an 27 Fine, P.V.A. (2002) The invasibility of tropical forests by exotic plants.
appropriate commitment from the scientific community, J. Trop. Ecol. 18, 687–705
28 Walther, G. (2002) Ecological responses to recent climate change.
the two are unlikely to continue to coexist.
Nature 416, 389–395
29 Laurance, W.F. et al. (2004) Pervasive alteration of tree communities
Acknowledgements in undisturbed Amazonian forests. Nature 428, 171–175
I thank Patrick Jansen, Helene Muller-Landau and three anonymous 30 Phillips, O.L. et al. (2002) Increasing dominance of large lianas in
reviewers for constructive criticism of the article. Amazonian forests. Nature 418, 770–774
31 Phillips, O.L. et al. (2004) Pattern and process in Amazon tree
turnover 1976–2001. Philos. Trans. R. Soc. Lond. B Biol. Sci. 359,
References
381–407
1 IPCC (2002) Climate Change 2001: The Scientific Basis, Cambridge
32 Lewis, S.L. et al. (2004) Concerted changes in tropical forest structure
University Press
and dynamics: evidence from 50 South American long-term plots.
2 Dirzo, R. and Raven, P.H. (2003) Global state of biodiversity and loss.
Philos. Trans. R. Soc. Lond. B Biol. Sci. 359, 421–436
Annu. Rev. Environ. Resources 28, 137–167
33 Baker, T.R. et al. (2004) Increasing biomass in Amazonian forest plots.
3 United Nations (2004) World Urbanization Prospects: The 2003
Philos. Trans. R. Soc. Lond. B Biol. Sci. 359, 353–365
Revision, UN Department of Economic and Social Affairs
34 Phillips, O.L. et al. (1998) Changes in the carbon balance of tropical
4 Pimm, S.L. and Askins, R.A. (1995) Forest losses predict bird
extinctions in eastern North America. Proc. Natl. Acad. Sci. U. S. A. forest: evidence from long-term plots. Science 282, 439–442
92, 9343–9347 35 Phillips, O.L. and Gentry, A.H. (1994) Increasing turnover through
5 Ramankutty, N. and Foley, J. (1999) Estimating historical changes in time in tropical forests. Science 263, 954–958
global land cover: croplands from 1700 to 1992. Global Biogeochem. 36 Clark, D.A. (2002) Are tropical forests an important carbon sink?
Cycles 13, 997–1027 Reanalysis of the long-term plot data. Ecol. Appl. 12, 3–7
6 Hansen, M.C. and DeFries, R.S. (2004) Detecting long-term global 37 Clark, D.A. (2004) Sources or sinks? The responses of tropical forests
forest change using continuous fields of tree-cover maps from 8-km to current and future climate and atmospheric composition. Philos.
Advanced Very High Resolution Radiometer (AVHRR) data for the Trans. R. Soc. Lond. B Biol. Sci. 359, 477–491
tears 1982–99. Ecosystems 7, 695–716 38 Chave, J. et al. (2003) Spatial and temporal variation of biomass in a
7 Fearnside, P.M. and Barbosa, R.I. (2004) Accelerating deforestation in tropical forest: results from a large census plot in Panama. J. Ecol. 91,
Brazilian Amazonia: towards answering open questions. Environ. 240–252
Conserv. 31, 7–10 39 Lewis, S.L. et al. (2004) Fingerprinting the impacts of global change on
8 Laurance, W.F. (2004) Forest–climate interactions in fragmented tropical forests. Philos. Trans. R. Soc. Lond. B Biol. Sci. 359, 437–462
tropical landscapes. Philos. Trans. R. Soc. Lond. B Biol. Sci. 359, 40 Phillips, O.L. et al. (2002) Changes in the biomass of tropical forests:
345–352 evaluating potential biases. Ecol. Appl. 12, 576–587
9 Skole, D. and Tucker, C.J. (1993) Tropical deforestation and habitat 41 Chambers, J.Q. et al. (1998) Ancient trees in Amazonia. Nature 391,
fragmentation in the Amazon: satellite data from 1978 to 1988. 135–136
Science 260, 1905–1910 42 Bush, M.B. et al. (1989) A 6000 year history of Amazonian maize
10 Achard, F. et al. (2002) Determination of deforestation rates of the cultivation. Nature 340, 303–305
world’s humid tropical forests. Science 297, 999–1002 43 Heckenberger, M.J. et al. (2003) Amazonia 1492: pristine forest or
11 Perz, S.G. and Skole, D.L. (2003) Social determinants of secondary cultural parkland? Science 301, 1710–1714
forests in the Brazilian Amazon. Soc. Sci. Res. 32, 25–60 44 Willis, K.J. et al. (2004) How “virgin” is virgin rainforest. Science 304,
12 Perz, S.G. and Skole, D.L. (2003) Secondary forest expansion in the 402–403
Brazilian Amazon and the refinement of forest transition theory. Soc. 45 Schimmelmann, A. et al. (2003) Palaeoclimatic and archaeological
Nat. Res. 16, 277–294 evidence for a w200-yr recurrence of floods and droughts linking
13 Aide, T.M. and Grau, H.R. (2004) Ecology – globalization, migration, California, Mesoamerica and South America over the past 2000 years.
and Latin American ecosystems. Science 305, 1915–1916 Holocene 13, 763–778
14 Wright, S.J. and Muller-Landau, H.C. The future of tropical forest 46 Cochrane, M.A. (2003) Fire science for rainforests. Nature 421, 913–919
species. Biotropica (in press) 47 Nepstad, D. et al. (2004) Amazon drought and its implications for
15 Dunn, R.R. (2004) Recovery of faunal communities during tropical forest flammability and tree growth: a basin-wide analysis. Glob.
forest regeneration. Conserv. Biol. 18, 302–309 Change Biol. 10, 704–717
16 Chazdon, R.L. (2003) Tropical forest recovery: legacies of human
48 Schöngart, J. et al. (2004) Teleconnection between tree growth in the
impact and natural disturbances. Perspect. Plant Ecol. Evol. Syst 6,
Amazonian floodplains and the El Niño-Southern Oscillation effect.
51–71
Glob. Change Biol. 10, 683–692
17 Lugo, A.E. and Helmer, E. (2004) Emerging forests on abandoned
49 Sombroek, W. (2001) Spatial and temporal patterns of Amazon
land: Puerto Rico’s new forests. For. Ecol. Manage. 190, 145–161
rainfall. Ambio 30, 388–396
18 Myers, N. et al. (2000) Biodiversity hotspots for conservation
50 Piperno, D.R. and Becker, P. (1996) Vegetation history of a site in the
priorities. Nature 403, 853–858
central Amazon Basin derived from phytolith and charcoal records
19 Brooks, T.M. et al. (2002) Habitat loss and extinction in the hotspots of
from natural soils. Quat. Res. 45, 202–209
biodiversity. Conserv. Biol. 16, 909–923
20 Janzen, D.H. (1988) Tropical dry forests: the most endangered major 51 Sanford, R.L. et al. (1985) Amazon rainforest fires. Science 227,
tropical ecosystem. In Biodiversity (Wilson, E.O., ed.), pp. 130–137, 53–55
National Academy Press 52 Piperno, D. (1990) Fitolitos, arquelogia y cambios prehistoricos de la
21 Brooks, T.M. (2004) Coverage provided by the global protected-area vegetacion en un lote de cincuenta hectáreas de la isla de Barro
system: is it enough? Bioscience 54, 1081–1091 Colorado. In Ecologia de un Bosque Tropical (Leigh, E.G., Jr. et al.,
22 Rodrigues, A.L. (2004) Global gap analysis: priority regions for expand- eds), pp. 153–156, Smithsonian Institution Press
ing the global protected-area network. Bioscience 54, 1092–1100 53 Wright, S.J. et al. (2004) Are lianas increasing in importance in
23 Curran, L.M. (2004) Lowland forest loss in protected areas of tropical forests? A 17-year record from Panama. Ecology 85, 484–489
Indonesian Borneo. Science 303, 1000–1003 54 Chapman, C.A. (2005) A long-term evaluation of fruiting phenology:
24 Bruner, A.G. et al. (2001) Effectiveness of parks in protecting tropical importance of climate change. J. Trop. Ecol. 21, 31–45
biodiversity. Science 291, 125–128 55 Clark, D.A. et al. (2003) Tropical rain forest tree growth and atmospheric
25 Fa, J.E. et al. (2003) Bushmeat and food security in the Congo Basin: carbon dynamics linked to interannual temperature variation during
linkages between wildlife and people’s future. Environ. Conserv. 30, 1984-2000. Proc. Natl. Acad. Sci. U. S. A. 100, 5852–5857
71–78 56 Cavelier, J. et al. (1999) Effects of irrigation on fine root biomass and
26 Wright, S.J. (2003) The myriad effects of hunting for vertebrates and production, litterfall and trunk growth in a semideciduous lowland
plants in tropical forests. Perspect. Plant Ecol. Evol. Syst. 6, 73–86 forest in Panama. Plant Soil 211, 207–213
www.sciencedirect.com
560 Review TRENDS in Ecology and Evolution Vol.20 No.10 October 2005

57 Nepstad, D.C. et al. (2002) The effects of partial throughfall exclusion the climate of tropical rainforest regions. Philos. Trans. R. Soc. Lond.
on canopy processes, aboveground production, and biogeochemistry of B Biol. Sci. 359, 311–329
an Amazon forest. J. Geophys. Res. 107, 1–18 66 Pinker, R.T. et al. (2005) Do satellites detect trends in surface solar
58 Graham, E.A. et al. (2003) Cloud cover limits net CO2 uptake and radiation? Science 308, 850–854
growth of a rainforest tree during tropical rainy seasons. Proc. Natl. 67 Wild, M. et al. (2005) From dimming to brightening: decadal changes
Acad. Sci. U. S. A. 100, 572–576 in solar radiation at Earth’s surface. Science 308, 847–850
59 Lovelock, C.E. et al. (1999) Effects of elevated CO2 concentrations on 68 Galloway, J.N. et al. (2004) Nitrogen cycles: past, present, and future.
photosynthesis, growth and reproduction of branches of the tropical Biogeochemistry 70, 153–226
canopy tree species, Luehea seemannii Tr. & Planch. Plant Cell 69 Körner, C. (2003) Carbon limitation in trees. J. Ecol. 91, 4–17
Environ. 22, 49–59 70 Martens, C.S. et al. (2004) Radon fluxes in tropical forest ecosystems of
60 Würth, M.K.R. et al. (1998) In situ responses to elevated CO2 in Brazilian Amazonia: night-time CO2 net ecosystem exchange derived
tropical forest understorey plants. Funct. Ecol. 12, 886–895
from radon and eddy covariance methods. Glob. Change Biol. 10,
61 Geist, H.J. and Lambin, E.F. (2002) Proximate causes and underlying
618–629
driving forces of tropical deforestation. Bioscience 52, 143–150
71 Saleska, S.R. et al. (2004) Carbon in Amazon forests: unexpected
62 Körner, C. (2004) Through enhanced tree dynamics carbon dioxide
seasonal fluxes and disturbance-induced losses. Science 302, 1554–1557
enrichment may cause tropical forests to lose carbon. Philos. Trans. R.
72 Chambers, J.Q. et al. (2004) Response of tree biomass and wood litter
Soc. Lond. B Biol. Sci. 359, 493–498
to disturbance in a Central Amazon forest. Oecologia 141, 596–614
63 Bawa, K.S. et al. (2004) Tropical ecosystems into the 21st century.
Science 306, 227–228 73 Chambers, J.Q. and Silver, W.L. (2004) Some aspects of ecophysio-
64 FAO (2000) Global Forest Resource Assessment 2000 – Main Report, logical and biogeochemical responses of tropical forests to atmospheric
FAO Forestry Paper 140, United Nations Food and Agricultural change. Philos. Trans. R. Soc. Lond. B Biol. Sci. 359, 463–476
Organization 74 Würth, M.K.R. et al. (2005) Non-structural carbohydrate pools in a
65 Malhi, Y. and Wright, J. (2004) Spatial patterns and recent trends in tropical forest. Oecologia 143, 11–24

AGORA initiative provides free agriculture journals to developing countries

The Health Internetwork Access to Research Initiative (HINARI) of the WHO has launched a new community scheme with the UN Food
and Agriculture Organization.

As part of this enterprise, Elsevier has given 185 journals to Access to Global Online Research in Agriculture (AGORA). More than 100
institutions are now registered for the scheme, which aims to provide developing countries with free access to vital research that will
ultimately help increase crop yields and encourage agricultural self-sufficiency.

According to the Africa University in Zimbabwe, AGORA has been welcomed by both students and staff. ‘It has brought a wealth of
information to our fingertips’ says Vimbai Hungwe. ‘The information made available goes a long way in helping the learning, teaching
and research activities within the University. Given the economic hardships we are going through, it couldn’t have come at a better time.’

For more information visit:

http://www.healthinternetwork.net

www.sciencedirect.com