Tropical Forest Remnants: Ecology, Management, and Conservation of Fragmented Communities
Edited by William F. Laurance and Richard O. Bierregaard, Jr.

Chapter Abstracts


Chapter 1
Tropical Forest Disturbance, Disappearance, and Species Loss
T. C. Whitmore

Tropical forests are currently disappearing at roughly 15.4 million ha (0.8%) per year. This rate, though obviously alarming, has increased little since the 1960s. The rates of deforestation differ and have differed between regions and forest types, as do the main threats and amounts of forest remaining. Substantial areas of the remaining forest are being degraded by fragmentation and timber extraction. There is so far no evidence for the massive species extinctions predicted by the species-area curve. However, the long lag-time before extinction, especially of trees that live to over a century, makes it probable that many species are doomed to eventual extinction.
Forest fragments of a few tens of hectares in area may contain an unexpectedly high fraction of the regional flora. Many plant and animal species can survive in timber-production forest. Some animals can utilize secondary or plantation forest surrounding fragments of primary forest. These factors all aid species survival. The likely future landscape of many rainforest countries--which will provide the living space for the native fauna and flora--will contain patches of conserved primary forest, larger tracts of managed production forest, and small remnants of primary or disturbed forest persisting as patches or riparian strips in agricultural lands.

Chapter 2
The Role of Economic Factors in Tropical Deforestation
James R. Kahn and Judith A. McDonald

Analyses of the economic causes of deforestation have typically focused on microeconomic factors such as poorly constructed timber leasing agreements and incentives for deforestation being imbedded in public policies. However, more recent work has begun to emphasize the importance of macroeconomic factors, such as the role of national external debt. Here we focus on the macroeconomic causes of deforestation, developing a conceptual model which suggests that macroeconomic factors can cause forested countries to engage in excessive deforestation to meet their short-term needs. We focus particularly on the role of external debt and test its importance using data from a cross-section of developing countries. Our results indicate that debt is positively correlated with rates of deforestation, which suggests that macroeconomic factors need further consideration in the policy process. The potential of debt-for-nature swaps for conserving tropical forests is discussed.

Chapter 3
Edge-related Changes in Environment and Plant Responses due to Forest Fragmentation in Central Amazonia
Valerie Kapos, Elisa Wandelli, José Luis Camargo, and Gislene Ganade

We investigated changes during the five years following edge creation in the nature and extent of edge effects in a single isolated reserve of the Biological Dynamics of Forest Fragments project near Manaus, Brazil. Air humidity and soil moisture, which had increased along simple gradients with distance into the forest from the newly cut edge, varied in complex patterns with distance from the same edge five years later.
One understory plant species, Astrocaryum sociale (Palmae) was significantly less abundant near edges, while another, Duguetia aff. flagellaris (Annonaceae), showed no consistent variation in frequency with distance from the edge. Neither of these species exhibited any sign of severe water shortage at any location in the forest. Although their leaf conductances frequently declined after reaching late morning maxima of 80-120 mmol m-2 s-1, these patterns did not vary consistently with distance from the forest edge. Predawn water potentials were never low, and minimum leaf water potentials never approached the turgor loss point. New leaves of D. flagellaris expanded at similar rates regardless of location. Foliar 13C analysis gave no evidence of changed water-use efficiency in canopy trees near the edge, and increased foliar 13C near the edge in understory species was due to variation in understory air composition related to mixing with air from outside the forest.
Canopy gaps were more frequent in the outer 70 m of the reserve than in control areas. The resulting complexity of vegetation structure near the edge may explain the complexity of environmental variation and plant responses in relation to the older edge. Understanding the long-term extent of changes in vegetation structure and the mechanisms behind particular edge effects is important for making appropriate management decisions.

Chapter 4
Edge and Aspect Effects on the Microclimate of a Small Tropical Forest Remnant on the Atherton Tableland, Northeast Australia
Stephen M. Turton and Heidi Jo Freiburger

This study examines horizontal microclimate gradients within a 20-ha upland rainforest remnant on the Atherton Tableland in northeastern Australia. Microclimate parameters and dicotyledonous seedling densities were considered in relation to distance from the forest edge and edge aspect under dry and wet conditions in summer. Canopy openness was generally low throughout the forest remnant, with a clearly defined and mature forest edge. Microclimate edge effects were found to penetrate about 30 m into the remnant under both dry and wet conditions. The magnitude of edge effects varied not only with distance from the edge, but in some cases with edge aspect; there also were some differences under dry and wet conditions. Under wet conditions, soil temperatures were strongly affected by both distance from the forest edge (at 0, 5, and 10 cm depths) and edge aspect (at 0 and 10 cm depths). In comparison, under both wet and dry conditions, ambient temperatures at seedling (20 cm) and sapling (150 cm) heights varied significantly with edge aspect but were not affected by edge-distance. Vapor densities at seedling and sapling heights were strongly affected by edge aspect but unaffected by edge-distance under wet conditions, while under dry conditions, neither aspect nor edge-distance influenced vapor densities. The density of dicotyledonous seedlings increased steadily from the edge to about 30 m into the remnant, after which it leveled off. Seedling density also varied significantly with edge aspect. The implications of these results for forest remnant conservation and management in upland regions of the seasonally wet tropics are discussed.

Chapter 5
The Influence of Edge Effects and Forest Fragmentation on Leal-litter Invertebrates in Central Amazonia
Raphael K. Didham

Few studies have focused on the impact of habitat fragmentation on invertebrates. In experimental forest isolates from the Biological Dynamics of Forest Fragments Project near Manaus, Brazil, I sampled a total of 920 m2 of forest leaf litter for invertebrates at seven distances from the forest edge in different-sized fragments and continuous forest. Invertebrate abundance showed some increase with reduction in fragment size. This increase may be explained as an edge-driven process. Sampling for invertebrates along edge-to-interior transects failed to support the widely held notion of a monotonic change in abundance with distance from the forest edge. Instead, invertebrate abundance was elevated at the edge, but also showed evidence of a second, mid-distance peak in abundance at 50-100 m from the edge.
Preliminary analysis of beetle community structure suggested that species composition in small forest fragments was influenced more strongly by fragment area than by edge effects. Many species characteristic of undisturbed Amazonian forest were absent even from the interiors of relatively large (100 ha) forest fragments. Beetle species composition was modified up to 105-210 m into continuous forest and up to 420 m into 100-ha fragments. These results challenge not only the way conservation managers view the extent of edge-driven processes in forest fragments, but also concepts regarding the minimum size of fragments required for maintenance of an intact terrestrial invertebrate assemblage.

Chapter 6
Hyper-disturbed Parks: Edge Effects and the Ecology of Isolated Rainforest Reserves in Tropical Australia
William F. Laurance

Forest fragmentation causes a sharp increase in the amount of habitat edge. Relative to continuous forest, the edges of forest fragments are exposed to winds of increased speed, vorticity, and turbulence, which often lead to elevated rates of windthrow and forest structural damage. I measured physiognomic, edaphic, and landscape features in two large (ca 500 ha) rainforest fragments in tropical Queensland, Australia, to determine whether the fragments were more heavily disturbed than nearby continuous forest.
Both fragments exhibited moderate reductions in canopy cover and sharply elevated abundances of disturbance-adapted rattans (Calamus spp.) and lianas, relative to continuous forest. Surprisingly, there was no increase in the density or basal area of treefalls and snapped boles in fragments, despite other indications of elevated structural damage. Multiple regression models suggested that the distance of plots to forest edge and topographic factors (elevation, slope, and topographic position) were significant determinants of structural damage.
As illustrated by studies of nonflying mammals, elevated disturbance may be an important structuring force for faunal communities in forest remnants, and could exacerbate the impacts of fragmentation upon forest-interior species. Many tropical and subtropical areas are subjected to strong prevailing winds or periodic windstorms, and the patterns observed here could be typical of other regions.

Chapter 7
Disturbance, Fragmentation, and the Dynamics of Diversity in Amazonian Forest Butterflies
Keith S. Brown, Jr. and Roger W. Hutchings

From 1980-1995, butterflies were censused in 25 reserves of 1-1,000 ha in the Biological Dynamics of Forest Fragments Project north of Manaus, Brazil. Of 455 species recorded, 108 were typical of forest understory shade, 205 of understory sun-patches, 67 of large clearings or the forest canopy, and 75 of sun-bathed edges of isolated reserves. Many species were rare; 78 species were known from only one individual, and only 133 were present in more than half the reserves. However, a single Ithomiine species made up almost 30% of all butterflies recorded.
Cumulative reserve lists showed small to negligible effects of reserve area. Most of the variation between sampling days and reserve lists could be related to the heterogeneity of the reserve environment; areas with large internal clearings, pronounced topographic variation, or overgrown edges were much richer than flat, homogeneous areas inside continuous forest or surrounded by pasture.
Community turnover between samples was greater in continuous than fragmented forests, and also increased with greater intervals between successive censuses. Many butterfly groups were faithful indicators of their usual resources, associations, and preferred habitats. Over-all, these results suggest that forest reserve systems should be large and stratified across different habitats. In addition, special attention should be paid to environmental microheterogeneity, second-growth forest, corridors, and soft edges, to ensure that the diversity of species, communities, and ecological processes will be maintained in these forests.

Chapter 8
Phenetic Variation in Insular Populations of a Rainforest Centipede
John F. Weishampel, Herman H. Shugart, and Walter E. Westman

Morphometric analyses were performed on the exoskeleta of the centipede Rhysida nuda, collected from 14 rainforest fragments (2.5-430 ha in area) and two continuous rainforest tracts on the Atherton Tableland in north-east Queensland, Australia. The relationships of within-population and within-individual phenotypic variation to landscape-scale habitat attibutes were explored with multiple regression analyses. Within-individual variation was assessed by measuring the fluctuating asymmetry of individuals, I.e., slight, random deviations from bilateral deviations from bilateral symmetry.
Patch area was positively correlated with within-population phenotypic variation, accounting for 52% of the total variance. Although population sizes of the centipede should be too large to result in inbreeding even in the smaller remnants, reduced population- or metapopulation-level phenotypic variation is thought to reflect a reduction in niche width corresponding to smaller rainforest remnant size.
Within-individual variation was positively correlated with a fractal index, defined as a ratio of patch edge to patch area, which explained 37% of the variance. Because smaller habitats consist largely of edge habitat, seasonal variations in moisture and temperature are more extreme, possibly leading to developmental instability in desiccation-sensitive species such as centipedes.
These results suggest that fragmentation could increase the probability of local extinction of sedentary rainforest invertebrates by decreasing within-population phenotypic variation. Such a decrease may reduce the ability of a population to track environmental change which in turn could be manifested by an increase in fluctuating asymmetry.

Chapter 9
Fragmentation Effects on A Central Amazonian Frog Community: A Ten-year Study
Mandy D. Tocher, Claude Gascon, and Barbara L. Zimmerman

Base-line data on frog community structure in intact forests were collected for seven years in Central Amazonia, and were compared with recent surveys of frog richness, abundance, and breeding success in forest fragments of varying sizes. The main goal of the study was to detect changes in frog community and population parameters attributable to deforestation and habitat fragmentation.
Overall, there was a positive relationship between fragment size and species richness. Surprisingly, however, fragments surveyed both before and after isolation consistently showed a increase in species richness by seven years after isolation. Remnants ranging from 1-100 ha in area exhibited a mean increase of ten species after isolation, independent of remnant size. Few species were lost from the fragments. Most of the increase in species richness resulted from invasions of species associated with the modified habitats surrounding fragments.
Several frog species were studied intensively, and these exhibited varied responses to fragmentation. The abundance of Eleutherodactylus fenestratus, a terrestrial breeder, increased significantly as fragment size decreased, and its abundance was significantly higher in both large and small fragments than in continuous forest. Colostethus stepheni, a semi-terrestrial breeder, was less abundant in fragments than continuous forest. Finally, Eleutherodactylus zimmermaneae and Osteocephalus sp. did not differ significantly in abundance among fragments, or between fragments and continuous forest. Multiple regression analysis indicated that variation in litter depth and canopy cover may explain the observed increases in E. fenestratus abundance in small fragments. Breeding success of pool breeders attracted to artificial pools was variable, but there was no evidence of reduced breeding success in fragments relative to primary forest.
Overall, fragmentation appeared to affect the frog community less severely than other taxonomic groups. Species richness tended to increase as a result of fragmentation, and only one of the four intensively-studied species exhibited reduced abundance in fragments.

Chapter 10
Understory Birds and Dynamic Habitat Mosaics in Amazonian Rainforests
Richard O. Bierregaard, Jr. and Philip C. Stouffer

We used a mark-recapture program to monitor Amazonian forest birds in the understory of experimental forest fragments before and after their isolation by clearing for cattle pastures. Most insectivorous and frugivorous species disappeared from recently isolated 1- and 10-ha forest remnants. Obligate army-ant followers were the first species to abandon small isolates. Mixed-species insectivorous flocks disintegrated within two years after isolation. Understory hummingbirds, however, were particularly insensitive to substantial habitat alteration.
Many understory frugivorous and insectivorous birds typical of primary forest frequented a mosaic of forest patches interspersed with agriculture, second-growth (>6 year-old) vegetation, and primary forest, foraging in both the remnants and surrounding second growth. In so doing, they recolonized remnants from which they had disappeared when the remnants were previously surrounded by clearcuts or pasture. Army ant-following birds were quick to recolonize remnants, while some ground-foraging insectivores were the least likely to use the mosaic of forest patches and second-growth vegetation.
The type of second-growth vegetation surrounding forest remnants affected the likelihood that certain species would recolonize the remnants. With few exceptions, population and activity levels of primary forest frugivorous and insectivorous species declined in small remnants with little or no compensation by second-growth specialists.

Chapter 11
Avian Extinction and Persistence Mechanisms in Lowland Panama
Kathryn E. Sieving and James R. Karr

Demographic, natural history, genetic, and environmental data for nine species of understory birds that forage on or near the ground are examined to provide insight about mechanisms determining extinction or persistence on Barro Colorado Island, Panama. Dispersal between island and mainland populations of the study species is not evident, therefore we assume recolonization does not influence island extinction probability. For each species, the mechanistic interaction between intrinsic (autecological) and extrinsic (environmental) factors determines the fate of its island population. Interactions between environmentally induced mortality and species-specific survival and fecundity rates emerge as the key mechanism defining insular extinction and persistence among our species.

Chapter 12
Frugivorous Birds in Fragmented Neotropical Montane Forests: Landscape Pattern and Body-mass Distribution
Carla Restrepo, Luis Miguel Renjifo, and Paul Marples

We investigated how changes in landscape pattern influence assemblages of frugivorous birds in the northern Andes of South America. Specifically, we looked for aggregations, or lumps in the distribution of body mass of frugivorous birds among (1) elevational zones, and (2) sites within elevational zones differing in their degree of disturbance by human activities. These comparisons represent two different scales of inquiry, yet produced similar results.
With decreasing landscape complexity the number of lumps decreased, partially as the result of the loss of lumps representing the largest and smallest species. Nevertheless, this pattern was not clear for sites within the upper montane zone where lumps containing large birds persisted across the four sites investigated. Dramatic changes in lump structure among sites within the same elevational zone were always associated with dramatic changes in landscape pattern.
Our results suggest that (1) lump structure in body mass of Neotropical montane frugivorous birds is influenced more by landscape pattern than by species composition or site conditions, (2) provided that landscape pattern is not critically altered, assemblages of Neotropical montane frugivorous birds appear robust to human disturbance, and (3) the numbers of lumps and species richness are related in important ways.

Chapter 13
Structure and Conservation of Forest Avifauna in Isolated Rainforest Remnants in Tropical Australia
Neil H. Warburton

Rainforest on the Atherton Tableland, northeastern Australia, has been reduced to isolated habitat patches surrounded by mosaics of agricultural land, urban development, and secondary forests. Over a 4.5-year period, I censused 30 remnant patches of rainforest (from 0.5-620 ha in area) and two control sites in continuous forest (>100,000 ha in area) to obtain comprehensive lists of rainforest bird species for each site.
The distribution of avifauna in these fragments was shown to be highly nested using a nested subsets analysis (Patterson 1987). Locally nomadic and migratory species showed a less strongly nested pattern than sedentary species. The relative merits of preserving one or a few large tracts of habitat compared to a multiple, smaller areas are discussed in the light of this analysis.
Incidence functions are a way of graphically representing the relationship between the occurrence of a particular species and the size of habitat patches. Incidence functions of all species were produced, and it was noted that bird species fell into one of five categories according to their occurrence in different-sized classes of habitat patches. These categories reflected varying responses of species to forest fragmentation.
Ecological traits of species associated with their response to fragmentation were assessed using correlation and regression analyses. The presence of species in "corridors" of remnant vegetation along streams and in planted windbreaks, and their natural abundance in rainforest, were found to be the best predictors of vulnerability to fragmentation. In general, tolerance of the matrix appeared to be a key determinant of species' vulnerability.
I also searched for environmental features of sites that were good predictors of bird species richness. The major determinant of species richness was the natural log of fragment area. A number of other variables were significantly correlated with avian richness but none were significant when effects of log-area were removed with a partial correlation analysis.
I conclude that while even small (<20 ha) patches of rainforest can support a significant fraction of the local avifauna, even the largest forest remnant examined (620 ha) failed to conserve a few sensitive species. The relatively high bird diversity in small patches is probably at least partly due to immigration of birds from other forest areas, and a high proportion of generalist species in the Australian rainforest avifauna.

Chapter 14
Biomass and Diversity of Small Mammals in Amazonian Forest Fragments
Jay R. Malcolm

Live-trapping on the ground and in the forest canopy were used to census small mammals in 1- and 10-ha primary forest fragments and continuous, undisturbed forest at the Biological Dynamics of Forest Fragments site near Manaus, Brazil. Fragments had been isolated from continuous forest for 1-8 years by 100-800 m of pasture or young second-growth forest.
Analyses of three measures of community structure (abundance, biomass, and number of species) revealed that fragmentation strongly affected the terrestrial small mammal fauna, but not the canopy fauna. Fragments (especially 1-ha) had a more abundant and diverse terrestrial small mammal community than sites in continuous forest. As a result, the vertical distribution of small mammal biomass was altered; terrestrial biomass exceeded arboreal biomass in fragments, especially 1-ha fragments, whereas arboreal and terrestrial biomass differed little in continuous forest.
In fragments, the observed increase in terrestrial fauna resulted from a general increase in abundance of most primary forest species. There were few species trapped in fragments that were not also found in primary forest. The fragmentation effect is attributed to increased understory resource-productivity along fragment edges and in secondary habitats surrounding fragments. The responses of small mammals to fragmentation contrasted markedly with that of other faunal groups such as birds, primates, and coprophagous beetles, and this may reflect different degrees of utilization of secondary habitats by these groups.

Chapter 15
Rapid Decline of Small Mammal Diversity in Monsoon Evergreen Forest Fragments in Thailand
Antony J. Lynam

Determining the changes in species composition that result from habitat fragmentation is difficult because knowledge of pre-fragmentation communities is usually lacking. In 1987, about 100 islands of tropical forest were created when the Saeng River Valley was flooded to create the Chiew Larn Hydroelectric Reservoir, in southern Thailand. Exhaustive surveys for small mammals were conducted in continuous forest on the adjacent mainland to determine baseline species distributions.
Using this information, comparative surveys across 12 islands (ranging from 0.7-109 ha in size) suggested that species loss has occurred rapidly on islands, especially smaller (<10 ha) ones, during the seven years following island isolation. Conversion of evergreen forest to dry forest by annual fires and logging has exacerbated the natural loss of native mammals on islands, especially of primary forest specialists which are common on the mainland. In contrast, some disturbance-tolerant species which are rare in unfragmented forest have exploited the altered ecological milieu on islands.
Many forest reserves in southern Thailand are of a comparable size to those studied here, and experience similar types of disturbances. Unless the natural composition and heterogeneity of forest vegetation is maintained, small mammal assemblages in Thailand reserves and forest remnants are likely to collapse, as they have done on the study islands at Chiew Larn.

Chapter 16
Internal Fragmentation: The Effects of Roads, Highways and Powerline Clearings on Movements and Mortality of Rainforest Vertebrates
Miriam Goosem

"Internal fragmentation" occurs when wildlife populations are subdivided by linear clearings such as roads and powerline clearings. I review the literature and aspects of my own research on linear barrier effects, with a particular focus on the few studies in tropical rainforest. Wildlife mortality statistics and live-trapping studies demonstrate that roads and powerline clearings through tropical rainforests do cause internal fragmentation of populations.
During weekly surveys, I found more than 1,300 vertebrates killed each year along two kilometers of highway traversing rainforest in northeastern Australia. Such high levels of mortality on roads may reduce the size of faunal populations, particularly those of rare or seasonally susceptible species. Fewer roadkills were recorded on wider parts of the road, indicating that for many faunal groups, increased clearing widths may reduce the rate of attempted road-crossings. Unless able to avoid traffic or use other crossing routes, species that are entirely absent from roadkill statistics may avoid the vicinity of highways and thus have severely fragmented populations.
Crossing movements of small mammals and birds were significantly inhibited in three trapping studies along roads in tropical rainforest. In another study, powerline clearings severely inhibited crossings of small mammals. However, the extent and potential barrier mechanism differed between roads and powerline clearings. Small mammals could easily be induced to cross most roads ranging from 6-22 m width, but no crossings were recorded along a 60 m-wide powerline clearing, even under the inducement of attractive baits placed only on one side of the clearing. Thus, powerline clearings appear to cause highly significant fragmentation of populations. The invasion of grassland-adapted rodents along the powerline clearing may have contributed to the exclusion of rainforest species.
Partial mitigation of linear barrier effects on ground-dwelling vertebrates occurred when crossing routes beneath roads were provided by culverts or underpasses. Strips of regrowth forest spanning powerline clearings also can facilitate movements of forest animals.

Chapter 17
Transitory States in Relaxing Ecosystems of Land-bridge Islands
John Terborgh, Lawrence Lopez, José Tello, Douglas Yu, and Ana Rita Bruni

We are investigating the faunas of 12 land-bridge islands created by a 4,300 km2 hydroelectric impoundment, Lago Guri, in east-central Venezuela. The islands were four years old when our work began in 1990, and are now ten years old. The faunas of "small" (1 ha) and "medium" (ca 10 ha) islands are already drastically impoverished, lacking numerous birds, mammals, reptiles, amphibians, and invertebrates found on a 350-ha "large" island and on the nearby mainland. Many of the species that have survived on small and medium Lago Guri islands have increased in abundance relative to the mainland, and in some instances, the increases have been by more than an order of magnitude.
The loss of many species and the hyperabundance of others has resulted in gross ecological distortions. In relative terms, there are too many herbivores and seed-predators, and too few predators, parasites, and seed-dispersers. The resulting ecological imbalances are likely to generate further impacts, particularly on the tree recruitment process.
We therefore speculate that fragmentation leads, first, to the loss of critical species, particularly top predators, and that the corresponding loss of key ecological functions then results in ecological imbalances, which, in turn, generate further distortions that drive a continuing loss of species. The system passes through a series of transitory states until, at some remote time, a biologically simplified steady-state is attained.

Chapter 18
Tropical Forest Disruption and Stochastic Biodiversity Losses
Mark Andersen, Alan Thornhill, and Harold Koopowitz

Patterns and rates of deforestation influence the geographic distributions of species persisting in forest remnants. We discuss a stochastic model based on species distribution profiles (the frequency distribution of the number of sites at which species in a taxon or region are found) to study the effects of habitat loss on rare endemic species. We also assess how deforestation turns species with moderate-sized geographic ranges into rare endemics. The model assumes a random spatial pattern of deforestation and a given distribution profile for the taxa in question.
We apply the model to a distribution profile for over 5,000 Neotropical plant species. The shape of the distribution profile determines the rate of extinction of species with restricted geographic ranges, but does not affect the rate at which species' ranges become restricted as a result of deforestation. We conclude that policy-makers should focus on in situ conservation (e.g. forest preserves) in regions with high levels of endemism, and on ex situ conservation (e.g. gene banking) for taxa with high levels of endemism. We also recommend that researchers try to discover which ecological and evolutionary characteristics determine a taxon's distribution profile.

Chapter 19
Regeneration of Large-seeded Trees in Australian Rainforest Fragments: A Study of Higher-order Interactions
Graham N. Harrington, Anthony K. Irvine, Francis H. J. Crome, and Les A. Moore

The regeneration of trees is a fundamental process in forest dynamics, and is essential for forest maintenance. Large (>2 cm diameter) seeds are usually dispersed by medium- to large-sized terrestrial animals, and the population density and movements of such animals are often strongly affected by forest fragmentation. In tropical Queensland, we compared forest fragments and continuous forest to assess the density and size-structure of selected large-seeded tree species, the rate of predation on their seeds, and the density of seed predators or dispersers.
We found that the large-seeded species could be divided into two main classes: those that were eaten by a large (600-1,000 g) rodent (Uromys caudimaculatus), and those that were not. Palatable seeds were subject to intense predation (100% of 971 seeds of known fate were eaten). Such predation was the probable cause of very low numbers of juvenile plants per adult in the palatable species; the ratio was <1 for palatable species but >50 for unpalatable species. Surprisingly, however, no effect of fragmentation on the density of juveniles of palatable tree species was detected, despite the fact that Uromys abundance was significantly lower in small (<20 ha) fragments. This apparently occurred because, even when at low densities, Uromys removed all of the palatable seeds available. Nevertheless, Uromys cached about half of all seeds before they were eaten, and although all the cached seeds studied were eventually eaten, a few survived up to 125 days. Because of the high predation rates, caching may be essential for palatable seeds to survive to germination.
The unpalatable seeds we studied were dispersed by frugivorous birds and mammals such as cassowaries (Casuarius casuarius). Cassowaries are territorial, flightless birds standing up to 2 m in height which are reputed to be critical seed dispersers for some rainforest tree species. Cassowaries did not occur in the smaller forest fragments we studied, but we were unable to detect any effects of fragmentation on the distribution or density of juvenile trees. This may have occurred for several reasons. First, seed dispersal by cassowaries is mostly localized, and other vectors also disperse the same seeds over short distances. Second, fundamental changes in the distribution and demography of long-lived trees may require centuries to become manifest. Finally, confounding factors such as environmental heterogeneity, climatic variation, and changes in land use may obscure the effects of loss of seed dispersers, rendering such changes undetectable over a period of a century or more.
We conclude that there probably are effects of forest fragmentation on tree regeneration in the system we studied, mediated via changes in assemblages of seed dispersers and predators. However, such higher-order interactions are affected by so many factors that measurement and prediction of these effects are impractical.

Chapter 20
Dispersal and the Dynamics of Genetic Structure in Fragmented Tropical Tree Populations
John D. Nason, Preston R. Aldrich, and J. L. Hamrick

The genetic fate of tropical forest fragments depends on processes operating both prior and subsequent to fragmentation. Chief among these is pollen and seed dispersal, which, in tropical trees, is typically animal mediated. Pollen and seed dispersal distances strongly influence the spatial distribution of genetic variation in continuous forest stands. Consequently, at the time of fragmentation, levels of genetic diversity within and among forest remnants will be determined not only by the number of individuals within a patch, but also by the spatial scale and patterning of fragmentation relative to pre-existing genetic structure.
Over time, changes in fragment- and species-level genetic variation will be determined by the effects of fragment isolation on the presence and behavior of pollinators and frugivores, and by resultant rates of pollen and seed migration versus extinction and recolonization. Although there is currently little information on the genetic responses of tropical trees to fragmentation, the presence of large neighborhood areas in continuous forests suggests that surviving species may persist as metapopulations of interacting fragments worthy of conservation.

Chapter 21
Plant Dispersal in Fragmented Landscapes: A Field Study of Woody Colonization in Rainforest Remnants of the Mascarene Archipelago
Christophe Thébaud and Dominique Strasberg

The ability of a species to colonize matrix habitats is widely thought to influence its long-term persistence in fragmented landscapes. Here we quantify the patterns and rates of matrix-colonization in 32 woody plant species within a system characterized by small rainforest remnants embedded in a matrix of recent lava flows on the island of La Réunion, Mascarene Archipelago.
For most of the 32 species, estimated matrix-colonization rates were found to be low (<1 m/yr). However, species differed markedly in their ability to move across the matrix. Differences among species in colonization rates primarily reflected differences in dispersal mode. On a per-unit-time basis, wind-dispersed species moved farther from forest remnants than did fleshy-fruited species dispersed by vertebrates. In other systems, this pattern is usually reversed. Among fleshy-fruited species, colonization rates declined significantly with increasing fruit size, with species having fruit >20 mm in length exhibiting colonization rates that were consistently low (<0.1 m/yr).
The Mascarene Archipelago has suffered massive animal extinctions during the past 300 years, with all but one seed-disperser going extinct on the island of La Réunion. We suggest that the very low rates of colonization in fleshy-fruited species may reflect inability of plants to adjust to disperser loss. We briefly discuss the implications of this pattern for conservation.

Chapter 22
Long-term Survival in Tropical Forest Remnants in Singapore and Hong Kong
Richard. T. Corlett and I. M. Turner

The long-term effects of forest fragmentation can only be assessed by studying existing, long-isolated forest fragments. In Singapore, the majority of existing fragments were isolated at least 150 years ago, although most have been linked by secondary forest subsequently, while in Hong Kong fragmentation was completed more than 350 years ago. At least 50% of the forest-specialist bird species have become extinct in Singapore and none have survived in Hong Kong. Mammalian extinctions show a similar pattern.
In contrast, vascular plant extinctions have been far fewer than expected. A rich forest flora has persisted in both Singapore and Hong Kong in fragments of less than 50 ha, although extinctions of both plants and animals continue. Exotic plant invasions have not been a problem in either area. Secondary forests accrete plant species slowly and selectively, so active intervention will be necessary to restore a diverse, continuous forest and to prevent further extinctions.

Chapter 23
Dynamics and Restoration of Forest Fragments in the Brazilian Atlantic Moist Forest
Virgílio M. Viana, André A. J. Tabanez, and João Luiz F. Batista

We studied the ecological consequences of fragmentation and the technologies for restoration of forest fragments in the Atlantic Moist Forests of Brazil. We address part of a general question: are small fragments of these forests self-sustainable? Our hypothesis is that forest fragments often are not self-sustainable because of edge effects and recurring physical and anthropogenic disturbances, and will require restoration practices to maintain their ecological viability and functioning in the long run.
If forest fragments are undergoing degradation, our predictions are that (1) tree recruitment will be lower than tree mortality rates; (2) edge effects will increase over time; (3) populations of several tree species will be overly small; and (4) there will be poor forest structure dominated by low-diversity eco-units (here defined as patches of forest at distinct stages of succession or degradation). If forest restoration practices are to be effective, they should facilitate successional processes, decreasing the dominance of vines and increasing tree canopy cover.
Our results indicate an ongoing process of forest degradation on the forest margin. Forest edges have lower densities of live trees and higher mortality rates than the forest interior. Tree basal area did not vary with distance from forest edge, probably because a few large trees survived along edges. Restoration practices aimed at facilitating successional processes in low secondary-forest eco-units produced expected changes in forest structure. Basal area and crown cover were most strongly affected by vine control and by certain combinations of experimental plantings. Collectively, these results suggest that certain restoration practices can help reverse the progressive ecological degradation of small forest remnants.

Chapter 24
Rejoining Habitat Remnants: Restoring Degraded Rainforest Lands
David Lamb, John Parotta, Rod Keenan, and Nigel Tucker

Restoring natural vegetation to cleared or otherwise degraded areas is essential for overcoming the detrimental impacts of habitat fragmentation on native fauna and flora. This chapter identifies the most appropriate locations within fragmented landscapes for restoration activities, and examines different approaches to the restoration process using three case studies from tropical Australia and Puerto Rico.
Two of the case studies are government programs being implemented in northern Queensland in an effort to overcome the effects of past clearing of rainforest for agriculture. One of these aims to restore natural vegetation and wildlife habitat largely through the establishment of "framework" tree species that encourage use of the restored area by frugivorous birds or mammals. These, in turn, facilitate the establishment of other tree and shrub species. The other program uses mixed-species plantations of native trees to re-create a timber industry as well as providing employment for the local community, restoring water quality in degraded streams and rivers and re-establishing wildlife habitat.
In the third case study, evidence from mixed-species plantations in Puerto Rico suggests such plantations also facilitate the establishment of tree and shrub species, including many native species. Under certain management regimes, plantations might provide important refugia for native plant species in heavily degraded landscapes. Some factors affecting this facilitation process and the implications for plantation managers are discussed.

Chapter 25
Measuring Landscape Changes in Remnant Tropical Dry Forests
Elizabeth A. Kramer

As deforestation proceeds, resource-managers in tropical regions will frequently be required to manage landscapes comprised of degraded habitats and mosaics of forest remnants, which must be restored to maintain biodiversity. In this chapter I demonstrate how the new tools of geographic information systems (GIS) and remote sensing can aid in the management of forest recovery. Remotely-sensed data are especially useful as a cost-effective means for viewing habitat change over large spatial scales.
I use these tools to characterize and measure changes in the vegetation of Santa Rosa Park, in northwestern Costa Rica, from 1979 to 1985. Landsat MSS and TM imagery were processed to produce vegetation maps. Further characterization of landscape structure was performed using a GIS and the spatial analysis software, Fragstats. The results of the study indicate that changes in management and land-use practices are strongly influencing landscape patterning. Forested vegetation is rapidly increasing within Santa Rosa Park.
GIS and remote sensing allow ecologists to step back from the forest remnant, and begin to integrate the entire landscape with all its various elements. I assert that conservation ecologists should redirect their research focus from assessing effects of fragmentation on local population declines, to identifying ways of maintaining forest remnants and landscapes for biodiversity. Research should be designed around the notion that isolated forest remnants do not exist in isolation, and should explicitly consider both the surrounding vegetation and biophysical conditions of the landscape. By adopting a more integrative approach, ecologists will increase the relevance of their work to conservation management.

Chapter 26
Quantifying Habitat Fragmentation due to Land-use Change in Amazonia
Virginia H. Dale and Scott M. Pearson

Human use of the landscape has caused fragmentation and loss of tropical forests. In Rondônia, Brazil, forest habitats are being rapidly converted to agricultural uses. These patterns of landscape change can be recorded with remotely sensed data. The dynamics of land-use change can then be explored by building simulation models.
We apply both techniques to analyze forest fragmentation in Rondônia. The study demonstrates that simulation models can accurately project the rate and pattern of deforestation at coarse spatial scales. However, such models may not be able to predict change at the level of spatial detail available from Landsat imagery. In general, models allow investigators to evaluate possible mechanisms driving land-cover changes observed in remotely sensed images.

Chapter 27
Deforestation, Fragmentation, and Reserve Design in Western Madagascar
Andrew P. Smith

From 1950 to 1990, primary forest cover in the Western Domain of Madagascar declined from 12.5% to about 2.8%. Deforestation has been more severe in the west than in the whole of Madagascar, which retains approximately 11% forest cover overall. Deforestation in the Western Domain progressed at an estimated rate of 106,000 ha per annum between 1950 and 1974, and 63,000 ha per annum between 1974 and 1990, comparable with the rate of 110,000 ha per annum in Madagascar's eastern rainforests between 1950 and 1984.
About 712,000 ha of high-quality primary forest remains in the Western Domain. Deforestation has not been random but has been concentrated in areas with more fertile soils and volcanic substrates. Remnant forests are highly fragmented and dispersed. Only two tracts exceed 50,000 ha in area, and patches of less than 600 ha make up 19% of the remaining forest area. The largest remnants persist on inaccessible karst formations and low-fertility sandy-calcareous soils. Only 23% of the remaining forest occurs within existing nature reserves.
Four new reserves are urgently needed to improve representation of lemur habitats, geological substrates, and elevational gradients in the current reserve network. Priority areas for conservation are those of large size (>25,000 ha) with high levels of biodiversity and endemicity, and a low risk of unregulated disturbance. Remnant forests at greatest risk of clearing are those within 4 km of villages or 2 km of roads. The largest remaining patch of Western Dry Forest in Madagascar (67,000 ha) is not adequately protected in an existing reserve. This patch has many features including large size, high biodiversity, a core area at low risk of disturbance, the presence of endemic vertebrate species, and ecotourism potential, all of which make it an ideal priority area for establishment of an integrated conservation and development project.

Chapter 28
Molecular Perspectives on Historical Fragmentation of Australian Tropical and Subtropical Rainforest: Implications for Conservation
Craig Moritz, Leo Joseph, Michael Cunningham, and Chris Schneider

Recent molecular analyses have revealed extensive genetic diversity in rainforest-restricted animals. The approximately congruent distribution of intraspecific variation among diverse species is consistent with vicariant refuge hypotheses. However, genetic differences among disjunct populations are generally much greater than expected for late Pleistocene events. By contrast, interspecific phylogenies show little geographic congruence, interdigitation of east Australian and Papua New Guinea taxa, and much greater sequence divergences, suggesting that most currently recognized species arose earlier than the Pleistocene, when Australasian rainforests were distributed differently than at present. The conservation of this evolutionary diversity and the maintenance of the processes that gave rise to it are important concerns in the design of reserve systems. In particular, our analyses of intraspecific phylogeography confirm the importance of climate through its effect on rainforest distributions, and support suggestions that reserve design should allow for dynamic processes associated with predicted future climate change.

Chapter 29
Diversity, Differentiation, and the Historical Biogeography of Non-volant Small Mammals of the Neotropical Forests
James L. Patton, Maria Nazareth F. da Silva, Márcia C. Lara, and Meika A. Mustrangi

Species richness and uniqueness constitute basic information used by conservationists to define areas for protection in efforts to preserve biodiversity. These also provide the currency by which evolutionary biologists seek to understand the temporal history of species, and of the communities and areas in which they occur. This chapter focuses on the non-volant small mammal fauna of the lowland wet forests of South America, primarily of the Amazon Basin and Mata Atlântica of coastal Brazil. We examine a set of factors, each considered important in efforts to recognize and synthesize the biodiversity of a region, namely: (1) the quality of baseline information on species inventories; (2) the adequacy of current taxonomy in the recognition of species diversity; (3) the importance of phylogenetic relationships among member components of specific lineages; (4) the degree and concordance of diversification patterns for different taxa across the same geographic region; and (5) the estimation of temporal depth in the origin of diversification.
We use examples from our own studies of several marsupial and rodent groups to show that the current data-base for the lowland wet forests of South America is inadequate in its recognition of the true level of species diversity within and among regions. We also demonstrate that a combination of phylogenetic approaches can identify common patterns of regional diversification that presumably reflect a common history. We conclude by providing recommendations for research that we consider critical for furthering initiatives to preserve the biodiversity of these, and other regions.

Chapter 30
Species Richness and Endemism in South American Birds: Implications for the Design of Networks of Nature Reserves
Jon Fjeldså and Carsten Rahbek

The Amazon rainforest is the world's most species-rich biome, with the greatest concentration of species richness in the sub-Andean zone. Based on taxic diversity measures this area would receive top priority for conservation, and it has already received enormous conservation attention. However, because most Amazonian bird species are widespread, this avian diversity is well covered by the current protected areas network. A Gap Analysis using the WORLDMAP computer program demonstrated that the reserve network in South America is very insufficient for Brazilian Atlantic forests and western Ecuador, areas that are now seriously threatened by a high rate of forest loss. Other gaps occur in the Andes. Unfortunately, because of the ad hoc approach used for selecting conservation areas in the past, existing reserves are often in "wrong" positions relative to areas of high endemism.
Recent studies are reviewed which suggest that the intensive differentiation of new species in the Andes is associated with intrinsic features of specific small areas, and that these characteristics also affect human viability here and in the adjacent lowlands. We suggest, therefore, that natural resource management initiatives be concentrated in such areas. This does not imply that the vast tracts of lowland rainforest are less significant from biological and ecoclimatic perspectives. However, because the species richness in lowland Amazonia appears to be maintained by dynamic processes operating over large areas, we find it difficult to precisely identify areas of high conservation priority there. It is probably more efficient to preserve the biodiversity of Amazonian regions through creation of international agreements and sound regional development policies.

Chapter 31
Researching Tropical Forest Fragmentation: Shall We Keep on Doing What We are Doing?
Francis H. J. Crome

This chapter examines a range of conceptual and methodological issues that students of conservation biology may need to come to terms with. It involves some re-examination of the way we do science and its utility. Hypothetico-deductivism, reliance on theory, and slavish use of significance testing are singled out for analysis. The complexity of fragmentation, the difficulties of generalizing, dealing with connectivity and balance, and coming to terms with differing views of reality are all canvased.

Chapter 32
Tropical Forest Fragmentation: Synthesis of a Diverse and Dynamic Discipline
William F. Laurance, Richard O. Bierregaard, Jr., Claude Gascon, Raphael K. Didham, Andrew P. Smith, Anthony J. Lynam, Virgílio M. Viana, Thomas E. Lovejoy, Katherine Sieving, Jack W. Sites, Jr., Mark Andersen, Mandy Tocher, Elizabeth Kramer, Carla Restrepo, and Craig Moritz

Consensus can be difficult to reach in any scientific discipline. In the study and management of fragmented tropical landscapes, this task is complicated by several factors--the recent and explosive growth of the field, widely varying approaches of different investigators, a diversity of land-use histories, and the enormous natural variation inherent in the tropical forest biome.
Given these challenges, it is hardly surprising that we initially approached our attempt at a general synthesis with more than a little trepidation. Fortunately, many of the volume's contributing authors were able to attend a symposium and workshop on tropical forest fragmentation as part of the 1995 Annual Meeting of the Ecological Society of America in Snowbird, Utah. During the workshop we discussed and debated a wide range of issues, eventually producing a draft of the following synthesis. We discovered that there was much we agreed upon.
The goal of this penultimate chapter is to survey ideas pertaining to the ecology and management of fragmented tropical landscapes. Our principal question is: What do we currently know--or think we know--about fragmented tropical forests? Obviously, a single chapter like this could never be exhaustive, but we have attempted to highlight important concepts then direct the interested reader to pertinent chapters in the volume, and to some strategic references elsewhere.
The chapter is divided into three sections. The first focuses on the vulnerability of tropical biotas to fragmentation, while the second details relevant additional features of tropical ecosystems. The final section focuses on the management and conservation of fragmented landscapes. Each section is divided into a series of themes or concepts, and each of these is supported by a number of specific points. While our emphasis is clearly on tropical forests, we believe that many of the principles described below will apply to fragmented ecosystems in general.

Chapter 33
Key Priorities for the Study of Fragmented Tropical Ecosystems
Richard O. Bierregaard, Jr., William F. Laurance, Jack W. Sites, Jr., Anthony J. Lynam, Raphael Didham, Mark Andersen, Claude Gascon, Mandy Tocher, Andrew P. Smith, Virgílio M. Viana, Thomas E. Lovejoy, Katherine Sieving, Elizabeth Kramer, Carla Restrepo, and Craig Moritz.

The fate of tropical forests and their biotas depends on a bewilderingly complex suite of factors. Historically, changes in global climate patterns have probably had the strongest effect on the distribution of the world's tropical forests. Within the past half-century, however, a growing number of anthropogenic factors has overwhelmed climate as prime determinants of the extent and nature of these ecosystems.
In reality, factors such as international trade relationships and treaties, national development policies, laws and their enforcement, and the local consciousness and education of residents and colonists all interact to hasten or slow the incursion of humans into the tropical forests of the world. One need only compare current activities of such conservation organizations as Conservation International or the Worldwide Fund for Nature to their target species-oriented programs just two decades ago, to see that these are indeed the arenas in which many conservationists now feel they can act most effectively. Prospecting for biomedically active compounds in Costa Rica, ethnobotany, "debt-for-nature swaps," eco-tourism, and the empowerment of local citizens groups now highlight the annual reports of the major conservation foundations.
The essential first step toward slowing the rapid erosion of tropical ecosystems is to create conditions where conservation is desired. Only when diverse sociological, political, and economic factors are aligned in such a way that tropical forests are seen to a majority--or a sufficiently vocal and convincing minority--as worthy of preservation and management, and resources consequently made available to these ends, will managers and conservation biologists be able to begin effective stewardship of the tropical landscape.
In theory, three approaches to conservation management are possible: preserving what we have now, planning the nature of the landscape that development will leave behind, and managing what is left. Ideally, we should set aside vast tracts of forest in national parks of no less than two million hectares each. However, even under the most optimistic scenarios, human populations will continue to expand into tropical forests, seeking material and economic gain or simply a place to live. Costa Rica, the tropical country with the most progressive, conservation-oriented government in the world, has only 25% of its countryside in protected areas. Because we have no reason to believe that other countries will protect more than a small percentage of their land in large reserves, we can safely predict that most of the world's remaining tropical forests will eventually become mosaics of forest remnants surrounded by modified habitats. Accepting this reality, as resource-managers we would like to design the size and configuration of the remnants before development begins, but this is rarely an option. By far the most probable scenario is that we will often have little choice but to manage an existing landscape mosaic created by colonization or economic exploitation, which proceeded with little or no regard for its conservation implications.
When managing existing landscape-mosaics, resources, in the broadest sense, are always likely to be limiting. Such resources include not only money, but also information, trained personnel, and the genetic substrate required for species reintroductions, reforestation, and so forth. Priorities must be set, and inevitably compromises will have to be made in determining how scarce resources are channeled. Hence, we need to have clear goals. If preserving species richness per se is the main objective, a landscape mosaic of pasture, second growth, fragments, and a large tract of primary forest might fulfill the goal better for some species (e.g. frogs) than a single larger tract of primary continuous forest. If preservation of a unique biota and its myriad interactions is the goal, a larger tract of forest, undisturbed by clearing, may be called for.
Whether we are managing an existing landscape or enjoy the luxury of planning development before the fact, successful conservation depends upon our understanding the physical and biological changes that occur in the aftermath of habitat fragmentation, and our ability to mitigate them. This book represents but a sample of the international effort to fathom the process of forest fragmentation at the species and ecosystem levels. In the preceding chapter (Laurance et al., chapter 32), we have attempted to distill some of the hard-won understanding we have already achieved down to a modest number of generalizations. We have also pointed out where and why generalizations seem impossible.
In this chapter, we outline unanswered questions and whole avenues of research that we feel are needed to fill critical gaps in our current knowledge base. In a provocative essay, Crome (chapter 31) addresses many methodological and philosophical challenges confronting students of habitat fragmentation. It is our hope that if researchers address the following questions with the intellectual and statistical rigor that Crome advocates, we will be better able to maximize the compatibility of development and conservation in the tropical landscape.

Tropical Forest Remnants: Ecology, Management, and Conservation of Fragmented Communities
Edited by William F. Laurance and Richard O. Bierregaard, Jr.
(c) 1997, 632 pages, 4 color plates, 12 halftones, 33 maps, 93 line drawings, 85 tables
Cloth ISBN: 0-226-46898-4
Paper ISBN: 0-226-46899-2

For information on purchasing the book--from bookstores or here online--please go to the webpage for Tropical Forest Remnants.

Return to the University of Chicago Press homepage.