Guia para o Meio Ambiente

Trabalhos Acadêmicos #1- FRAGMENTATION OF THE MATA ATLÂNTICA, #2 - 


MESTRADO no Canadá - Concordia University





Fragmentation of Tropical Rain Forest





 Rain forests, usually recognized for their range of biodiversity are gradually and ultimately disappearing. They are known as the most productive ecosystem due to the tight coupling of production and decomposition, part of which is the result of high diversity in mutualist, symbiotic and parasitic relationships being developed within the ecosystems. This paper discusses the effects of fragmentation on forest with its variety of short and long-term effects. Some species benefit from fragmentation while most are at risk. Although biodiversity can be maintained in part by taking those species most in danger and reversing the process, species not in danger of extinction must continue to be studied before the habitat alterations change their status.

The Nature of Ecosystems

An ecosystem must be in equilibrium to prosper, if not, the habitat is expected to collapse. The equilibrium theory focuses on the idea that there is a stable point, the balance of nature, which implies a closed system with a self-regulating structure and function. Opposing this theory, is the dynamic non-equilibrium ecological perspective, which is based on the idea that diversity within natural areas is not exclusive, but related to occurrences in the surrounding areas as well. In this theory, the genetic composition is prone to change whether due to genetic drift, migration, or natural selection. The objective is not to stop change but to ensure that populations continue to respond to changes in an adaptive way. This must be kept in mind when relating ecosystem to fragmentation. This is the non-equilibrium ecological world within a dynamic system represented by external processes in the form of disturbances, like fires, floods, storms, and disease out-brakes. Species interactions continue to be important (Botkin, 1990).


History of Fragmentation SCIENCE

 starts in 1778 with the theories developed by Johann Forster (Forster 1778 in Browne, 1983). Other theories dealing with conservation were developed only in the beginning of the last century. Mainly two schools of thought emerged, one advocating appreciation of nature, and the other having a utilitarian view. Pinchot, F.G. (1947) and Aldo Leopold, who saw his own views as inadequate, present examples of this last view (Browne 1983). In the 1960’s, as forests became more fragmented, the theory of island biogeography created by MacArthur and Wilson came to the forefront. This theory focuses on the balance between migration and extinction of species. Starting in the 1970’s, research showed that nature is very complex and dynamic, and fragmentation is very hard to predict.


Importance of Fragmentation

            Fragmentation is an important issue influencing biodiversity conservation. It impacts adversely on habitats and destroys biological diversity through activities such as forestry and agriculture, which modify the landscape and generates destruction (Harris, 1984; Wilcove et al.; Sounders et al., 1991). Fragmentation precedes the destruction of entire ecosystems, and only forest value loss will stop it for the well being of both, the planet and humanity (Soulé, 1985; Ehrenfeld, 1988; Regan, 1983; Taylor, 1986; Rolston, 1988; and Johnson, 1991). Fragmentation can cause immeasurable damage and should be studied in depth, especially in light of increases in population and the demand for natural resources.


What is Fragmentation

            It can be defined in different ways, but essentially it is the disruption of continuity in patterns and processes (Lord and Norton, 1990). Fragmentation starts with a forest gap or perforation of the ecosystem matrix (the most common and protected area of the habitat). Initially, population patterns are barely affected. When the gap enlarges, it becomes the most common part of the area. This is the time when the matrix shifts to an anthropogenic habitat, affecting flora and fauna, as well as climate and the physical environment (Wiens, 1989). Large forest fragment patches are known as forest remnants, that is, large unused parts of the old forest that are kept for watershed protection, wildlife conservation and other benefits. Another type of fragment is the managed forest, formed as a result of extraction. Natural forest plantations are yet another type that keep native trees while producing agricultural products and other commodities. The last type of fragmented forest is riparian, which is a linear strip of a few trees to hundreds of meters wide used for the protection of streams from erosion (Schellas and Greenberg, 1995). In addition there are a number of smaller formations known as habitat shredding. Although these fragments described above lack biological diversity, they still contain many endemic species and ecosystems. The destruction of original forest habitat causes a great reduction in available resources changing the environment species history (Meffe and C. R. Carroll, 1994). Fragmentation destroys corridors between ecosystems that take hundreds of years to occur naturally. Interchanges between flora and fauna of these ecosystems are suppressed as a result, and consequently disrupts evolution because of the loss of predators, pollinators, seed dispersal and nutrient cycling. Another damage to forests is caused by the edge effects of fragmentation, which produce light level increase, invasion of open country species, and damaging winds that change the physical structure of the ecosystem (Bierregaard and Dale, 1995). In this case, forests will remain in a secondary state continuously. Species that are restricted to a single patch not large enough to meet their needs will become endangered locally. Other species require a mix of habitat types with different resources in order to meet their survival (Bierregaard and Dale, 1995).


Natural Fragmentation

            Natural fragmentation also exists in forest ecosystems, which is a normal process of change, and not so disruptive as a fragmented landscape. Natural fragmentation is a mosaic of fragments with rich internal structure within layers of vegetation, largely because of small contrasts between neighboring patches due to undetected differences in structure. Consequently, there are fewer edge effects in naturally patchy habitats (Meffe and Carroll, 1994).


Major Theories of Fragmentation

            The most famous explanation for fragmentation is the equilibrium theory of island biogeography, which centers on the idea of a balance between immigration and extinction within island habitats (MacArthur and Wilson, 1963, 1967). The rate of immigration is determined by species mobility, and surplus of population at the origin, and the rate of extinction depends on the size of the island of destination, population dynamics, and the biological characteristics of the species involved. Island species continuously go extinct over-time, however, others species fill their niches. Islands close to the main land experience greater immigration, and large islands contain larger populations. Islands that were connected to the main land in the past are known as land-bridge islands because they were not dependent on immigration and extinction to provide species diversity (Harris, 1984). Land-bridge islands lost diversity over-time when the isthmus was inundated and they were isolated (Diamond, 1972; Terborg, 1974; Faaborg, 1979). This theory also explains the lack of diversity in isolated terrestrial fragments surrounded by degraded habitat. However, the matrix of the habitat in landscapes clearly distinguishes these patches from ocean islands because disturbed land may pose greater threats than water.

With the use of island biogeography, the first related theory is that of species area effect. This is so because it deals with the size of the fragment and its carrying capacity (Lack, 1976; Sinberloff, 1991). Additional studies have also concluded that habitat diversity is possibly a better predictor of species richness than area size (Power, 1972; Johnson, 1975). However, for increase species richness, both area and habitat diversity is very important, suggesting that forest patches must be both large and heterogeneous (Freemark and Merrian, 1986).

One may expect that patch size requirements would differ between birds, mammals, and insects because of differences in their ability to disperse between patches and across hostile territory. It was proven that there is no significant difference in their response to fragmentation (Andren, 1994).

Another theory is that of resources concentration. It attempts to explain that habitat patches with a large amount of resources. This hypothesis suggests that population density should be correlated with patch area.


Measuring Fragmentation

            Each ecosystem has its unique structure, and ecologists have suggested a number of ways to measure change, including measurements of fractal dimension (size and shape complexity), contagion (positive/ negative association between patch types), and other types of spatial statistics (O’Neill et al., 1988; Turner, 1989; Mladennof et al., 1993). A time series analysis can define change in species composition and other expressions of biological diversity, while providing a monitoring approach (Noss, 1990). Another way of measuring fragmentation is through the use of gap analysis. It is a technological intensive analysis demanding time because it uses remote sensing data to construct a set of overlaid maps. It identifies a number of parameters to show the overall changes and spatial gaps in an area. The amount of contrast between habitat fragments and the matrix in which they exist is another measure of fragmentation. However, the point in fragmentation at which there is a collapse in biological integrity has not yet been defined.

            An important indicator of forest health is the number of species and individuals of that species in a forest area. Two mechanisms account for species-area relationships, that is, the area per se hypothesis, and the passive sampling hypothesis. The first requires that the average total of species be larger on large fragments, and the second that the colonist arriving is proportionally higher on larger than on small islands (Connor and  McCoy, 1979).


Ecosystem/Biological Consequences

            The most noticeable effect of a fragmented landscape is the increase of the “edge to interior” effect, which refers to the amount of forest border interacting with the anthropogenic clearing that comprises the fragment (Forman, 1995). The edge provides a different environment, with greater light availability and higher temperatures that encourages different species compositions (Kopos et al.,1997). The microclimatic edge effect on biodiversity is strongly dependent on the size and type of forest patch and favors some species while being unsuitable to others (Turner,1996; Kopos, 1999). As the fragment size decreases, the importance of the edge increases, and the edge effect can be influential (Turner, 1996). In any habitat, the outer part is a zone of influence of negative impacts on the habitat and species within. Sunlight, wind, shade, humidity, and tree height all affect the ecosystem by elevating tree mortality, reducing stocking density, affecting regeneration, declining forest bird density, nest predation, and invasions (Chen et al., 1992). The effect of fragmentation can be seen in several levels of biological organization, from changes in gene frequency to wide reaching changes in distribution of species. The first effect is isolation of habitat due to movement of barriers that may cause population decline (Stouffer and Bierregaard, 1995). Dispersal can also affect the population in patches as crowding can induce intra and inter specific interactions, competition, and predation (Hanski et al., 1994). Another consequence of fragmentation is genetic alterations to species not attributed to adaptation or natural selection (Simberloff and Cox, 1987; Simberloff, 1988). The influx of new species from surrounding areas is another problem (Schieck et al., 1995). There is a potential for disease as wildlife and domesticated animals are in close proximity. Further, there is a loss of predators, which are important to the food chain.

Some species are at greater risk in fragmented landscape than others. Many rare species are endemic with small populations found in few patches, what makes them susceptible to fragmentation. Davies et al. (2000) developed a test of relationship between population decline and traits (2000), predicting the increase of extinction risks for species in a fragmented forest landscape. It has been found that body size or spatial scale is not directly linked to extinction risks because its relationship with the variables (abundance, population fluctuation and growth) is uncertain (Blackburn and Gaston, 1994, 1995,1996; Cyr et al., 1997). As for trophic groups, theory predicts that species at the top end of the food chain are more prone to extinction because they tend to have more unstable population dynamics (Lawton, 1995; Holt, 1996). Species especially vulnerable are listed on page 33. Also included in this paper is a reflection on the Amazon and the Brazilian Atlantic Forest fragmented forest ecosystems.    

To conclude, it can be stated that in order to escape the potentially dangerous effects of fragmentation, species have three options for surviving highly fragmented ecosystems: first, a specie might survive or even thrive in the matrix; secondly, a species might survive in a fragmented landscape by maintaining viable populations within individual habitat fragments, and a third option is for a species to be highly mobile, integrating a number of home range habitat patches or an interbreeding population. Species incapable of pursuing one or more of these options is bound for extinction. 




For a variety of reasons, rain forests, usually recognized for their broad range of biological diversity, are gradually disappearing as parts of forested areas are removed. Only small fragments of rain forest remain and are surrounded by farms and grazing areas. This fragmentation may be one factor in the decline of biodiversity. Tropical rain forests are one of the most productive and diverse ecosystems on the planet. Their high productivity is due to a very tight coupling between production and decomposition. The “turnover” rates are very high, and minerals are recycled quickly and efficiently. Part of this productivity is a result of high biodiversity, an incredible host of organisms that have adapted to form a tight interconnecting loop, coupling production and decomposition. There are many mutualistic, symbiotic and parasitic relationships that take advantage of this energy flow (e.g. lianas, bromeliads, mycorhizae). As a result of these relationships above, tropical forests are very productive and their biological diversity must be protected with the suppression of fragmentation.
The purpose of this paper is to discuss the effects of fragmentation on rain forest environments. This paper reviews evidence that fragmentation occurs at many spatial scales and may have a variety of short and long-term effects. A few species benefit from fragmentation while most others are at increased risk. Global biodiversity can be maintained in part by taking the species and ecosystems most in danger and reversing the processes that place them at risk. However, species not currently in great danger of extinction due to fragmentation must continue to be studied before the habitat alteration changes their status. Increasing the number of animals in danger makes conservation more difficult. There is also a great need for strategies to protect ecosystems already fragmented, including the study of the internal dynamics of remnant natural areas and the external influences on those areas. This paper looks into the problems and effects of fragmentation, the theories that can be applied to fragmentation, and ways of measuring it.
The Nature of Ecosystems
According to the Equilibrium of Nature Theory, an ecosystem must be in equilibrium for, if not, the habitat will probably collapse. Also the rate of increase must equal the rate of decrease in species diversity, population size and other spatial statistics. The equilibrium paradigm focuses on the idea that there exists a definable stable point such as a “climax community”. This implies a closed system with a self-regulating structure and function, that is, the ”balance of nature”. Recently, however, a dynamic non-equilibrium ecological perspective has replaced the equilibrium theory. This equilibrium view was changed considering that the biological diversity within natural areas is not only mutually exclusive, but strongly related to what happens in the surrounding areas as well. The evolution process and natural selection are the main reasons why non-equilibrium dynamism has become dominant. The genetic composition of most populations is likely to change over time whether due to drifts in small populations, immigration from other populations or due to natural selection. The goal is not to stop genetic change and conserve the status quo, but rather to ensure that the population may continue to respond to the environmental changes and in an adaptive way. This goal has to be kept in mind when relating to problems with fragmentation. The ecological world is a dynamic system in a non-equilibrium state. This contemporary view of ecology (Botkin 1990) recognizes that ecological systems are generally not in a dynamic equilibrium, at least definitely, and have no stable point.  The structure and function of ecosystems are not generated internally. External processes in the form of natural disturbances such as fires, floods, drought, storms, earth movements and outbreaks of diseases and parasites are frequently responsible for their structure. This does not mean that species interaction are unimportant. Ecosystems are not a chaotic assemblage of species, rater, they are structured. Within the environments are clusters of species that have strong interactions, and in many cases they have an evolutionary history.  Nevertheless, this does not prove that community structure or species compositions do not change at some scale in space and time. 
Ecosystems are open systems with fluxes of species, materials and energy. This dynamism shows that fragmentation studies must anticipate isolation and make sure that fragments are open to interaction.
History of Fragmentation Research  
The first potential negative effect of habitat fragmentation on biodiversity was acknowledged in 1855 when Swiss phyto-geographer Alphonse de Candolle predicted that “the break up of a large land mass into smaller units would necessarily lead to the extinction of one or more species… “ (Browne 1983).  Along the same line of research, more than 80 years before, in 1778, Johann Reinhold Forster first mentioned species-area effect. During this time Forster noted that “regions only produce a greater or lesser number of species as their circumference is more or less extensive” (Forster 1778, in Browne 1983). During this time the main idea considered was that species in national parks were to be preserved for human enjoyment and recreation. However, since the beginning of this century, several other theories dealing with conservation have been developed. This second phase began with Forester Gifford Pinchot, who believed in the anthropocentric value of nature. According to Pinchot, nature was to be used by man in the most economic way possible and land used for many simultaneous purposes such as logging, grazing, wilderness preservation, recreation and watershed protection (Gifford Pinchot, 1947). During this phase emerged two schools of thought, one advocating pure wilderness and appreciation of nature, and the other having an utilitarian view.  Out of the utilitarian tradition came Aldo Leopold who saw his own views as inadequate and scientifically inaccurate. According to him, nature could not be seen as a collection of independent parts but as a complicated and integrated system of interdependent processes and components. In the 60’s, as forests became more fragmented, the study by MacArthur and Willson on the Theory of Island Biogeography came to the forefront. This theory is important because it focus on the balance between migration and extinction of species. In the last three decades, ecological research has shown that nature is very complex and dynamic. Fragmentation does not follow statistical models or theories. It is a very hard process to predict. 
Why Fragmentation is an Issue
There are many reasons justifying why the study of fragmentation is important. As it is known, The fragmentation process has an adverse impact on forest habitats and destroys biological diversity. It also causes disruption of extensive habitats, turning them into many isolated and smaller areas (Harris 1984; Wilcove et al.; Sounders et al. 1991). Fragmentation impacts plant and animal species to a great extent due to the loss of pristine habitat and the reduction of resources.  Activities such as forestry and agriculture have modified the landscape into many pieces of fragmented forest, and have generated much environmental destruction. Forest clearing however, is rarely complete, and often not permanent. This leads to great numbers of fragments throughout the world. Fragmentation cannot continue at the present rate indefinitely because it precedes the destruction of entire ecosystems. If fragmentation of a whole biome is taken to its conclusion, it will completely disintegrate the whole, creating an “empty” ecosystem, and In this case, an “empty” ecosystem will have no value to humanity. The loss of economic value makes stopping fragmentation important for the well being of both, the planet and people. If the purpose is to conserve resources on behalf of economics, one must note that natural resources are not inexhaustible. There are many people that wish to conserve nature in order to profit from its resources in the future. However, there is also a widely held belief that nature and all living beings have an intrinsic value. This is not a simple concept to advocate. It depends upon different principals and beliefs of a psychological, philosophical and religious character. Soulé (1985) and Ehrenfeld (1988) have expressly affirmed that all biotic diversity has intrinsic value. Thus, intrinsic valuing of nature has broken new ground and in some instances breached with the Western Religious and philosophical tradition. Contemporary philosophers have attributed intrinsic value to conscious or sentient animals (Regan, 1983); all living things (Taylor 1986); species, ecosystems (Rolston 1988, Johnson 1991) and evolutionary processes (Rolston 1988). Others believe that intrinsic value exists objectively. Biological entities have a purpose that must be respected. They have a self-organized and self-directed goal (Fox 1990) to strive consciously, unconsciously or in evolutionary sense in the achievement of a certain basic predetermined goals. These goals are to at least: grow, reach maturity and reproduce (Taylor 1986). As a result of all these factors mentioned above, fragmentation can be recognized as a disrupter of ecological, climatic and physical environments. It can cause incalculable damage and should be studied in depth specially in light of human population increases and the demand for natural resources.
What is Fragmentation
Fragmentation can be defined in a few different ways. Essentially, “fragmentation is the disruption of continuity“ in patterns or processes (Lord and Norton 1990). It typically starts with the formation of a forest gap or perforation of the ecosystem matrix (the most common and protected part of the habitat). For a while, the matrix remains with its natural vegetation. Species composition and population patterns are barely affected (Wiens 1989). When the gaps become larger or greater in number, they eventually become the most common part of the habitat area.  At the point in which this happens and the original habitat has been broken, the landscape matrix shifts from forest to anthropogenic habitat (Wiens 1989).  
This is a process with unpredictable thresholds, not simply an either/or condition. The patterns fragmentation disrupts are ecological, adversely affecting flora and fauna, as well as the climatic and physical environment. Fragmentation is most closely associated with a reduction in the total area of the original habitat and the creation of an edge between original forest habitat and highly altered landscapes. To put fragmentation in context, Larry D. Harris (1984) may be quoted as he makes a very clear argument:
“A natural community that occurs as a central portion of a larger regional habitat will contain numerous rare species that rely on the larger system for existence. As progressively more of the surrounding area is allocated to other uses, the distinctiveness of the habitat island patch is accentuated. As the habitat island becomes progressively more isolated from surrounding vegetation of similar form, the rare species are quickly lost.”
In other words, regions once connected by wide expanses of natural habitat that are now isolated, block genetic interchanges of both flora and fauna species. Blocking these interchanges has potential dramatic effects on highly complex ecological processes (Lord and Norton 1990), especially with regards to back system functioning. Microorganisms and insects within forest soils, for example, undertake these underlying processes. Back systems are more likely disrupted at small scales of fragmentation, even though, the organisms affected and the overall process is usually not noticeable (Lord and Norton 1990). 
There are many types of fragments that have taken shape for many different reasons. Larger forest fragments are usually known as forest remnants. These remnants are largely unused fragments of old-growth forest, including forest intentionally left unclear for watershed protection, wildlife conservation, and other benefits. Some times they have not yet been cleared because of inaccessibility. Another fragment type is the managed forest. These forests are formed as a result of extraction. They are used for extraction of timber products or for the extraction of different forest resources. Another fragment type is the natural forest plantation. These forests are systems that integrate the use of native trees with the production of agricultural and other commodities, as well as used for fuel wood and other household needs. The last type of fragment is the gallery or riparian. This fragment is a linear strip of forest habitat that can have a dimension of a few trees to hundreds of meters wide, protecting streams from erosion and insolation (Schelhas and Greenberg, 1995). In addition to all the types mentioned above, there are a number of smaller formations known as habitat shredding. These are agricultural landscapes created by changing the original habitat into long narrow strips that may be all edge habitat.  These strips include living fences, farm field borders and shade patches for livestock (Schelhas and Greenberg, 1995).  
The complexity of fragmentation issues require that in the future, ecologists develop new or modified models for population dynamics, demography, dispersal and genetics (Forman and Gordon 1986). These different models are imperative in identifying the damage caused by fragmentation. 
Fragments then are the result of changes to the physical, climatic and biotic environment, creating patches of primary, secondary and managed forests. These resulting forests however are incomplete, for many of the most common organisms are lost at even moderate levels of fragmentation and/or modification of forest habitats. Although these forest fragments lack biological diversity, they still contain many endemic species and habitats. Forest fragments are in a state of constant change themselves and in many cases they are also disappearing under pressure. The destruction of original forest habitat causes a great reduction in available resources, invariably changing the environment of plant and animal species history (Meffe, G.K. and C.R. Carroll, 1994).
Fragmentation leads to destruction of transition corridors between ecosystems that would take hundreds of years to occur naturally. Interchanges between the flora and fauna of these ecosystems, as a result, are repressed and may have great consequences to evolution. Fragmentation within natural areas has potential and dramatic effects on local biodiversity and ecological processes because of the loss of top predators, pollinators, seed dispersal, and nutrient cycling. Additionally, there is uncertainty related to the loss of genetic diversity. Fragmentation also exposes the forest to edge effects. These edges related effects produce increased light levels, invasion by open country species and damaging winds (Bierregaard and Dale). These effects influence on fragmented forests change their physical structure. With change in physical structure, arrested succession is most likely to occur, and the forest will remain continuously in a secondary state, lacking the original structure of the primary forest. Animal species that are restricted to a single patch not large enough to meet the needs of individuals or groups will become endangered locally. Other species are dependent on a constellation of habitat patches in relatively close proximity because they require a mix of habitat types with distinct resources in order to meet their life history (Meffe, G.K. and C.R. Carroll (1994).
Natural Fragmentation of Heterogeneous Habitat
Not all interruption of forest continuity is bad. A heterogeneous landscape mosaic is not equivalent to, or as disruptive, as a fragmented landscape. Heterogeneity is very important within ecosystems. A naturally fragmented landscape has rich internal patch structure within different layers of vegetation, largely because of small contrasts between adjacent patches due to small structural differences. There is less contrast between patches that have formed naturally (e.g., trees to tall herbs) than between patches formed through fragmentation (e.g., trees to asphalt), and therefore less intense edge effects in naturally patchy habitat. Naturally patchy habitats do not contain features of specific threat (e.g., roads, fences, people) (Meffe, G.K. and C.R. Carroll (1994). In a landscape scale analysis, the distribution of vegetation types typically corresponds to change in elevation and slope aspect. Natural disturbances also create considerable heterogeneity in forests and other vegetation. The landscape is determined by spatial scale of disturbance, that is, by the size and distribution of disturbance-generated patches. Spatial scale is highly disturbed by larger size disturbances (fire, flood, drought, etc) but these are infrequent occurrences that happen maybe every 20, 50 or 100 years.  Tree fall gaps are also very important for ecosystem function. Death and fall of individual trees or small groups of trees create only small patterns of disturbance history (Meffe, G.K. and C.R. Carroll (1994). These openings in the forest canopy are widely recognized as important for the establishment and growth of trees, especially rain forest trees. Hartshorn (1989) suggests that perhaps as much as 75% of tree species in rain forest ecosystems are dependent on canopy opening for seed germination and for the growth of tree sapling.  There is greater growth survival and reproduction rates when plants occur in or near canopy openings (Denslow, 1987). These examples include canopy, sub-canopy and understory species whether the trees are shade tolerant or light demanding. Habitat in gap areas represents an environment that changes temporally and spatially. The most important effect of canopy opening is increased time and intensity of direct sun light exposure to the lower strata of the forest. As a consequence, there is more heterogeneity in light availability for seedling and sapling. Forest turnover rates are also dependent on tree fall gaps. This gap in forests is important because they represent a niche for bacteria and a place where anaerobic decomposition of detritus occurs. In simple terms, naturally heterogeneous landscapes have rich internal patches (e.g., wind-fall gaps, varied vegetation) whereas fragmented landscapes have simplified gaps (e.g., parking lots, farms, clear-cuts). 
Major Theories 
The most famous and controversial explanation applied to fragmentation is the Equilibrium theory of island Biogeography (MacArthur and Wilson 1963, 1967). This theory focuses on the balance between immigration and extinction in island habitats. Species mobility, distance from a colonizing source, and surplus population at the source determine the rates of immigration, and the rates of extinction depend on island size, population dynamics and biological characteristics of the species involved. Over a period of time, island species continuously go extinct, but other species fill similar niches as they migrate from elsewhere. Rationally, islands near the mainland experience higher rates of immigration than remote islands. Large islands contain larger populations and should therefore suffer lower rates of extinction. Size may effect immigration rates as well, as larger islands have a higher chance of intercepting individuals. By similar logic, extinction rates are lower as immigrants of the same species can bolster declining populations. Consequently, the theory of island biogeography predicts that, all else being equal, larger islands closer to the mainland will contain the most species. 
A common distinction that should be made with regards to island biogeography studies is related to oceanic and land-bridge islands (MacArthur 1972). In the Pleistocene, islands that were connected to the mainland when sea levels were on average 100 meters lower, today are known as land-bridge islands. Species richness on land-bridge islands is usually higher than on similar size oceanic islands (Harris 1984) because during the Pleistocene they were not dependent on immigration and extinction for species diversity. Since becoming isolated again, land-bridge islands apparently lost species overtime (relaxation phenomenon) with some species being more extinction-prone than others (Diamond 1972; Terborgh 1974; Faaborg 1979). 
The analogy of a land-bridge island surrounded by ocean and an isolated terrestrial fragment surrounded by degraded habitat areas is very persuasive for explaining lack of diversity in fragments. However, the matrix surrounding habitat patches in terrestrial landscapes clearly distinguish these patches from ocean islands. Disturbed lands may pose far greater threats than water. Some studies have found that a forest bordered on at least one side by water had lower species predation rates than those surrounded completely by land (Small and Hunter 1988).  Usually, the greater the structural contrast between adjacent terrestrial habitats, the more intense the edge effects. Among other reasons, there is weak evidence that relaxation is related to island size (Abele and Connor 1979; Faeth and Connor 1979). Another consideration is that island biogeogaphy does not address the influence of edge effect on the number of species in a given island. Often researchers claim support for Island Biogeography Theory without considering alternatives or null hypothesis. Island Biogeography Theory has however led to major advances in habitat conservation. Above all, this theory expanded the focus of scientists to a collection of sites and on the potential effects of habitat area. The Equilibrium Theory of Island Biogeography makes assumptions about population size, population density, and area in order to explain species-area relationships via the area per se hypothesis (MacArthur and Wilson 1967, Simberloff 1976, Connor and McCoy 1979). It assumes that the number of individuals within a taxon increases linearly with the area of an island (MacArthur and Wilson 1967:13). In other words, the theory presumes that population density remains constant with increasing area. While it is not explicit in their text, MacArthur and Wilson’s (1967) assumption has been interpreted to apply to whole biotas, faunas, and communities, as well as to individual species (MacArthur and Wilson 1967, Simberloff 1988). The implications of MacArthur and Wilson’s (1967) assumption are quite different if they refer to groups of species (e.g., faunas) rather than to individual species. For individual species, their assumption implies that larger areas would contain more individuals of a species, but that the number per unit area would remain constant. However, since species richness (or group size) is a positive function of area, if we assume that the density of a group of species is constant for all areas, then the densities of individual species must, on average, decline in larger areas (Schoener 1986).
With the use of Island Biogeography, the first theory that comes to mind regarding fragmentation is related to species-area effect. This is so because it discusses the size of fragments and its carrying capacity.  As area decreases, so should the diversity of physical habitats and resources that in turn support a lesser number of species (Lack 1976). Species-area effects have been confirmed for many island groups (ocean islands) and have been extended to “habitat islands in terrestrial landscapes”. Many scientists have sought this single theory to explain the relationship between species richness and area.  Simberloff (1991) concluded that  “probably, on all but very small sets of sites, the majority of species-area relationships are accounted for by the fact that larger sites on average have more habitat than small ones”.
Additional studies, have concluded that habitat diversity is possibly a better predictor of species diversity than area size (Power 1972; Johnson 1975). They also show that for increased species richness, both area and habitat diversity are very important. This suggests that forest tracts must be both large and naturally heterogeneous (Freemark and Merriam 1986). Although habitat diversity provides the best explanation for species-area relationship, other theories are often at work. Distinguishing between the effects of area, habitat diversity and other factors contributing to species richness has proven difficult (Meffe, G.K. and C.R. Carroll (1994). 
There is without any doubt a reduction in average patch size in forests throughout the world. In patchy or fragmented landscapes, patch size effects should be relatively simple to detect. Consider the following: habitat loss is expected to produce a proportional decline in the number of animals living in a particular landscape. For example, if a piece of forest habitat supports a large population of some animal, and 50% of that forest is removed, then one might expect a decline in animal abundance of 50%. However, it often has been found that species abundance declines beyond that predicted by habitat loss alone (Edward F. Connor, et al., 2000). This difference stems from the effects of reduced mean patch size and decreased connectivity in the landscape (i.e., a reduction in the rate of successful dispersal, Merriam 1984; see also Venier and Fahrig 1996). Habitat association explains most of the variation in effect sizes. 
Patch size effects are commonly observed for edge and interior species, but are not common for habitat generalists. This result can be anticipated because animal densities are calculated in a biased fashion by many investigators ( Darren J. Bender, 1998). This bias occurs when the density of a species within a patch is calculated using total patch area, rather than the area of the inhabited portion of the patch. When the animal densities are calculated using total patch size, values for edge and interior species are always underestimated. This underestimate is most pronounced in large patches for edge species, and in small patches for interior species, which is called by many conservation biologists the “geometric” effect. Because the degree of underestimation depends on patch size, there will always be an apparent relationship between density and patch size for edge and interior species. This effect has been reported numerous times in the literature (e.g., Whitcomb et al. 1981, Lynch and Whigham 1984, Freemark and Merriam 1986, Blake and Karr 1987, Merriam and Wegner 1992, Johns 1993, McGarigal and McComb 1995).
Evidence indicates that other variables, such as migration strategy or animal taxa, are related to the occurrence of patch size effects, which suggests that the relationship is more complex. This question has received remarkably little attention in the literature, and certainly warrants further investigation.
Patch size effect is a general and predictable effect that occurs for a broad set of edge and interior species living in patchy landscapes. Regardless of the mechanism(s), this result predicts that habitat loss and fragmentation will greatly affect the abundance of edge and interior species. In situations in which habitat loss and fragmentation create a greater number of smaller patches from pieces of previously contiguous habitat, interior species should always suffer a decline in population, attributable to these patch size effects, that occurs in addition to the decline attributable to habitat loss (Edward F. Connor et al., 2000). This is because the actual density of the species within patches will be predicted to decline as patches get smaller and smaller. The opposite effect should be seen in edge species, for which population densities may actually increase as patches become smaller and proportional amounts of edge habitat increase. The pattern will continue as patches decline in size, until each patch no longer contains any interior habitat, and is all edge habitat. This increase in density will offset the decline in population size associated with habitat loss that occurs when habitat is destroyed (Edward F. Connor, et al., 2000).
One must stress that these predictions are contingent upon the pattern of habitat destruction. They are based on the assumption that habitat destruction will subdivide existing habitat patches to form new patches that are (necessarily) smaller than those previously in the landscape (i.e., the process of fragmentation). Habitat destruction that only removes habitat and has little effect on the fragmentation of patches in the landscape will not produce the predicted effects. In fact, certain patterns of loss could produce the opposite effect. For example, one can envision a pattern of habitat destruction that removes all the small patches from a landscape, but nothing else. In this case, the layout of habitat patches within the landscape has not changed much, but the effect on edge and interior species will be opposite to our predictions (Edward F. Connor, et al., 2000). The removal of small patches will have more of an impact on edge species than on interior species because small patches contain proportionally more edge habitat. Consequently, observed declines in regional abundance potentially will be greater for edge species than for interior species in such a case. Therefore, for edge and interior species, the decline in population size associated with habitat destruction will depend both on habitat fragmentation per se and on the pattern of habitat loss if large or small patches are preferentially removed ( Edward F. Connor, Aaron C Courtney, James M. Yoder , 2000).
A number of different factors relating to landscape characteristics or species life history traits might explain when patch size effects are important determinants of population density. According to Darren J. Bender (1998) there is no relationship between effect size and COVER, so it appears that the proportion of habitat in the landscape does not determine the emergence of patch size effects. This result apparently contradicts the conclusions of Andren (1994) that demonstrated a tendency for patch size and isolation effects to emerge as percent cover decreased. Andren predicted that, as the extent of habitat fragmentation within a landscape increases, patch size and isolation effects would emerge and contribute to the decline of species richness and abundance that occurs due to pure habitat loss alone. Andren also suggested the presence of a threshold value of percent cover below which these patch size and isolation effects would begin to emerge, but there is no evidence of such a relationship in the studies reviewed here.
There are two possible explanations for the apparent disagreement between Andren’s (1994) and Bender’s (1998) results. First, Andren used a vote-counting method to summarize the findings of each study he reviewed, which is prone to error. His response variable was simply a “yes” or a “no” vote that described whether the study rejected the random-sample hypothesis. This type of response is very coarse in comparison to the continuous response variable. Also, this method does not account for direction of the effect. In meta-analysis, positive and negative effects tend to cancel each other out, two opposing effects would appear to support one another. Thus, the two methods can yield different results even when similar data are used.
A second explanation for the discrepancy is that Andren’s (1994) review used studies that examined the effect of both patch size and patch isolation, whereas Bender is concerned only with patch size effects. Andren (1994) did assess isolation, but he did not differentiate between studies that reported significant patch size-density relationships and those that reported patch isolation-density relationships. Therefore, it is possible that many of the significant results encountered by Andren may have been due to isolation effects and not to patch size effects. An important, unanswered question is which effect is more important: patch size or isolation? Given that it is difficult to gather enough field data to assess this issue, spatially explicit population modeling is likely to be the most successful approach to answering this question.
The most significant life history predictor of the patch size effect was migratory status. Many migrant species are thought to be more area sensitive, whereas resident species are said to be more “tolerant” of fragmentation effects because of differences in life history traits. For example, residents are reported to exhibit differences in nest-building behaviors that make them less susceptible to predators (Weins 1989b, Hansen and Urban 1992, Bohning-Gaese et al. 1993). However, according to Bender (1998), patch size effects were contingent upon both migratory status and habitat association. On average, migrant species have significantly lower effect sizes than residents. Therefore, as habitat loss and fragmentation occur and habitat patches are reduced in area, patch size effects should produce a greater decline for resident interior species than for migrant interior species. Because the patch size effects are negative for edge species, migrant edge species will be predicted to show a greater increase in density due to patch size effects than will resident edge species. In both cases, the population decline associated with patch size effects is predicted to be greater for resident than for migrant species that is contrary to current conviction. It is likely that the current view has arisen from confusions in the use of the term habitat fragmentation. The term has often been used to imply both habitat loss and fragmentation even though these are two different effects. Such a definition is dangerous because one can always expect an effect of habitat loss even when there is no effect of habitat fragmentation. If one cannot separate these two effects, one will conclude that there is a “fragmentation” effect even when habitat loss is the only factor affecting population decline (Fahrig 1997).
One may expect that patch size effects would differ between birds, mammals, and insects because of differences in their ability to disperse between patches, and across hostile territory. Presumably, flying animals should be better equipped to move around the landscape and to exploit the maximum amount of available habitat. Andren (1994) demonstrated that, surprisingly, there was no significant difference between birds and mammals in their responses to habitat fragmentation. Results by Bender’s study suggest that fragmentation may actually have a tendency to increase densities of herbivore edge species more than for carnivores. It is possible that herbivore densities may be more closely linked to food production in edge habitat. It has been shown that the diversity and productivity of edge plant species is greater in small than in large patches (Levenson 1981). Relatively greater levels of primary productivity may result in greater amounts of food for edge herbivores, particularly small, frugivorous mammals (Santos and Telleria 1994). Food levels for carnivores will not necessarily follow this same pattern. According to the above evidence, the species responded different to the species-area affect according to their preferred niche in the forest ecosystem. To simplify, the following characteristics of the species summarized were either generalists, interior or edge dwellers.
For a generalist species that are not associated with only the edge or only the interior habitat, the decline in population size associated with habitat destruction should be accounted for by pure habitat loss alone. In other words, patch size effects are not expected to be an important factor in determining the population size of generalist species in fragmented landscapes ( Edward F. Connor, Aaron C Courtney, James M. Yoder , 2000).
For an interior species, the decline in population size associated with habitat fragmentation per se will be greater than that predicted from pure habitat loss alone. – This will always occur, because the ratio of interior habitat to total patch size declines as patches become smaller following habitat fragmentation and loss ( Edward F. Connor, Aaron C Courtney, James M. Yoder , 2000).
For an edge species, the decline in population size will be less than that predicted by pure habitat loss alone. In fact, relative abundances of edge species may actually increase in the landscape following fragmentation, especially if fragmentation serves to increase the total amount of edge habitat for these species ( Edward F. Connor, Aaron C Courtney, James M. Yoder , 2000).
A theory related to species-area that tries to explain this correlation is the Phenomenon of Density Compensation Theory. It attempts to account for observations that the summed density of animal species on small patches equals that of fauna on larger  fragments, with the result that the average population density of each species is greater on smaller fragments (MacArthur et al. 1972). This pattern is consistent with MacArthur and Wilson’s (1967) assumption that density of animal groups is independent of area. Because there are fewer species on small islands, density compensation leads to the conclusion that, on average, island species have higher population densities than do mainland species. This pattern originally was interpreted to suggest that species on small islands were less subject to interspecific competition and predation, which allowed their populations to increase (MacArthur et al. 1972). However, Williamson (1981:240) and Schoener (1986) point out that if total faunal density is independent of island size (as assumed by MacArthur and Wilson 1967) and if a species-area relationship exists, experts need not infer competitive release to account for density compensation. In any event, if density compensation is a widespread phenomena, one would expect that on average animal population-densities should be inversely related to patch area.
Another theory related to species-area is that of resource concentration. It attempts to explain the often-observed phenomenon that habitat patches with large amounts of resources (e.g., monocultures, areas of high plant density, or large patches) have higher densities of insects (Root 1973, Kareiva 1983). Therefore, this hypothesis conjectures that population density should be positively correlated with patch area. The comparison of small and large habitat patches rather than homogeneous and heterogeneous habitat patches has provided a stronger test for the effect of resource concentration, since it is free from the confounding influences of habitat heterogeneity on food plant quality (Risch 1981, Kareiva 1983). Root (1973) conjectured that the higher density of animals in larger patches might be solely a consequence of movement behavior (the movement hypothesis). Herbivores are more likely to find and remain in large, monospecific stands of their host plant than in small or heterogeneous patches. Other explanations include the enemie hypothesis that suggests that predators are more effective in smaller patches than in large ones (Root 1973, Raupp and Denno 1979, Denno et al. 1981, Risch 1981, Kareiva 1983).
The relationship between animal population density and area has been addressed explicitly by the phenomenon of density compensation (MacArthur et al. 1972), the resource concentration hypothesis (Root 1973), the Island Biogeography and the Species-area Relationship theories. These were initially motivated by studies of islands and by studies of habitat patches, each of these theories has been interpreted and applied broadly to a variety of spatial scales and taxa (Janzen 1968, Simberloff 1974, Futuyma and Wasserman 1980, Faeth 1984, Dooley and Bowers 1996).
Measuring Fragmentation
Each landscape at any point in time has its own unique structure. Consequently, ecologists have suggested a number of ways to measure changes in landscape, including measurement of fractal dimension (a measure of patch size and shape complexity), contagion (the positive or negative association between patch types) and other types of spatial statistics (O’Neill et al. 1988; Turner 1989; Mladennof et al.1993). Time series analysis of landscape pattern can be accomplished by these spatial measures. The measuring of time series can establish changes in species composition and other expressions of biodiversity, while providing an effective monitoring approach (Noss 1990). Fragmentation can also be ascertained through indicator species. These species are used to gauge the conditions within the particular fragment, and they are important actors within the ecosystem for the equilibrium to be kept at acceptable limits.
Gap Analysis
Another important way of measuring fragmentation is through the use of the gap analysis. This type of analysis is technologically intensive because it uses several remote sensing data to construct a set of overlaid maps. It identifies a number of parameters (vegetation, soil, protected areas and species distribution) to identify spatial gaps in a certain physical area. The amount of structural contrast between habitat fragments and the matrix in which they exist is one measure of fragmentation (Harris 1984).  Although all these gauges are successful in ascertaining many factors and indicators of fragmentation, they have not yet been successful in discovering the point in fragmentation at which there is a collapse of biological integrity. No study has been carried out over a long period of time to identify the collapse point and other intervals in the fragmentation process.
Structural Health  - Species - Area Relationship Measurement
An important indication of forest fragmentation and structural health is determined by the number of species and individuals of that species within a forest area. The greater the diversity of healthy species population, the healthier the forest. Two mechanisms proposed to account for species-area relationships are the Area Per Se Hypothesis and the Passive Sampling Hypothesis. In the former case, it is required that the average total abundance of individual species populations be larger on large islands and, in the latter case, that the number of colonists arriving is proportionately higher on large than on small islands (Connor and McCoy 1979). If the higher density of animals on large patches is due to local reproductive recruitment, then The Area Per Se mechanism leads to larger population sizes, thus lowering extinction rates and permitting more species to occupy large patches (Simberloff 1976, Connor et al. 1983). If the higher density of animals on large patches is generated by differential movement into large patches or higher movement rates out of small patches with equal rates of local reproductive recruitment in patches of all sizes, then the passive sampling hypothesis would account for the arrival and persistence of more species on large patches. Because movement may be more common among habitat islands than among true islands, the Area Per Se Hypothesis combined with the Habitat Diversity Hypothesis is more likely to explain species-area relationships on true islands (Connor and McCoy 1979). Although this may not be the case for migratory birds that re-colonize habitat patches annually (Haila et al. 1993). A common way to test for these hypothesis is through patch size effects. One can compare the relative densities of organisms within different-sized patches. If patch size effects account for any additional decline in abundance, then one would expect density to be positively related to patch size (Darren J. Bender, 1998; Edward F Connor, 2000). Therefore, the strength of the relationship between patch size and density can be used to index the strength of the patch size effect. It could be hypothesized that edge and interior species are more prone to patch size effects for reasons other than the geometric effect, i.e., that other factors were also contributing to the effect sizes observed. One could test this hypothesis by using an unbiased density measure calculated by using the area of the patch that is actually inhabited by any given species rather than the entire patch area. Thus, a measure of the strength of the relationship between usable patch area and density would not be subject to the geometric effect, and a significant patch size effect would be attributable to something inherent in edge and/or interior species themselves. The difficulty in applying such an approach results from the fact that it is usually not possible or practical to determine the actual proportion of each patch that is used by a particular species. Although boundary distances have been reported for a few species, the actual boundaries that separate edge and interior habitat are difficult to define. They are “fuzzy”, meaning that there is no distinct threshold beyond which the distribution of an animal abruptly begins or ends. The vast majority of studies reviewed for this paper made no attempts to distinguish between usable habitat and total habitat in a patch. It is also not appropriate to make simple assumptions about patch shape and the position of the boundary between edge and interior habitat within the patch to estimate the proportion of a patch that is usable ( Edward F. Connor, Aaron C Courtney, James M. Yoder, 2000). Such boundary distances are most certainly site specific because they are determined by local conditions (e.g., microclimate, vegetation composition, and presence of ecological enemies). Therefore, boundary distance estimates may be relevant only to the population under study and may not be reliably extrapolated to other populations within a species. For these reasons, no attempts were made to measure patch area at the resolution of usable patch area (Edward F. Connor, et al., 2000).
Measuring Animal Density
For the estimation of fauna, there is an absence in the literature of studies with both small sample sizes and small effect sizes. This suggests, first of all, that a bias exists against publishing statistically non-significant values of the correlation between faunal density and area. However, a selection of studies, regardless of sample size, taxa, or the nature of the habitat patch, should be a sufficient guard against unforeseen bias. Second, for studies of birds based on point counts, the radius of detection of many species may be greater than the radius of small patches. Therefore, in a region sampled, a circle determined by the location of the sample point and a species’ radius of detection may not be entirely forested. As a result, there might be a tendency to underestimate densities on small patches what generates a positive bias in estimates of the correlation between bird population density and patch area (Haila 1988, Haila et al. 1993). Even in studies such as those prepared by Haila (1981, 1983) and Haila et al. (1983, 1987), in which density estimates are made using transects that cover the entire area of small patches the correlation between population density and patch area remains positive. Third, a large proportion of the effect size estimates obtained were based on avian studies employing point count or transect methods that rely on detecting calling or singing birds. If there is a systematic bias for particular bird species, or bird species in general, to be more or less detectable as a function of patch area, then the patterns could simply reflect such a bias. However, the literature suggests that point and transect counts tend to underestimate abundance, not that the detectability of calling or singing birds depends on patch area (but see McShea and Rappole 1997). Fourth, Haila (1988) also suggests that because small patches contain more species that forage in regions outside the putative habitat patch than do large patches, density estimates that do not account for this additional foraging area tend to be overestimated. Therefore, Haila (1988) claims that the null expectation for the relationship between density and patch area should be negative.  Haila’s argument would imply that effect sizes may be underestimated. Fifth, the repeatability of estimates within species in both, the sign of the correlation and its magnitude, suggests that the effect sizes estimated are largely attributes of each species rather than idiosyncratic measures heavily dependent on site characteristics and study methodology. 
Ecosystem Consequences - Edge effects 
The most noticeable characteristic of a fragmented landscape is the significant increase in the forest edge to interior. “Edge to interior” refers to the relative amount of forest border interacting with anthropogenic clearing to the amount of forest that composes the fragment (Forman 1995). The edge of a forest patch provides a different environment compared to the tropical forest interior, typically having greater light availability and higher temperatures (Kapos et al 1997, Saunders et al. 1991, turner 1996, Turton 1997). Therefore, it encourages different species composition (Kapos et al. 1997). The altered microclimate of the edge has been found to be unsuitable for some species, while promoting an increase abundance of others (Turner 1996). Lovejoy et al. (1986) accredits changes in butterfly community composition in tropical forest fragments partially to the increased isolation within small forest patches. The microclimatic difference between the forest’s edge and its interior limits their habitat, and resource availability, leading to local extinction of a certain butterfly population (Lovejoy et al.1986). The microclimate of forest edges, however, provides appropriate habitat for disturbance-associated species. Studies in other biomes have found that edges create “windows for invasion” of non-native species that alter the ecosystem (Brothers and Spingarn 1992).
The magnitude of microclimatic edge effects on biodiversity is strongly dependent upon the size and shape of the forest patch (Kopos 1997). The relative importance of edges increases as fragment size decreases, and edge effects may become highly influential” (Turner 1996, 204). Small or thin fragments may experience microclimatic shifts throughout the entire patch encouraging the presence of species better adapted to the new environment. Along with the change in microclimate, these newly established species altering resource availability, can potentially cause local extinction of other species within the patch.
The elimination of surrounding forest vegetation and the edge increase causes forest patches to become vulnerable to hot, dry tropical winds. These winds damage the vegetation by creating tree falls on the edge and tree fall gaps in the forest interior (Lovejoy et al 1986). Laurence et al. (1998), in a study examining fragmentation effects on Amazonian tree communities, contend that “a sudden increase in gap-phase vegetation could help drive local extinction of disturbance-sensitive species in fragments”. By opening up the forest floor to more light, and as a result to higher temperatures, tree fall gaps can prevent the regeneration of shade-tolerant, moisture-adapted species, and instead promote the establishment of shade-intolerant species that do not require moist soils. Species in small or thin fragments, with higher edge to interior ratios, are particularly vulnerable to the effects of wind damage because wind has the potential to penetrate deep into a forest (Laurence 1998). Due to this fact, small fragments will typically have a higher proportion of their area in gaps (Laurence 1998). In cleared areas, day-time temperatures are higher and night time temperatures lower. Remnants can also receive significantly more solar radiation. Within forests, often less than 1% of sun rays and energy penetrates to the forest floor, where in remnants, with large deforested areas, they can be completely open to solar penetration. Edge zones are usually drier and have increased incidence of fire as they are drier, warmer and windier than interior forests. Fragments are also less buffered against run-off.
In any habitat island, the outer part is not a line but rather a zone of influence that varies in width depending on what is measured. This area of influence has negative impacts on the interior conditions of a habitat or species within that habitat. Sunlight and wind move into the forest from the edge and alter the microclimate, and as a result, edge zones are usually drier and less shady than forest interiors favoring shade-intolerant plants rather than typical forest plants. Increased rates of blowdown, reduced humidity and other physical edge effects may extend the edge zone from the edge of the fragment two or three tree-heights, or over 200 meters into a forest (Harris 1984, Franklin and Forman 1987, Chen and Franklin 1990). The values thus vary with forest types. 10-20m for Eastern Hardwood, 15-25 m in semi-dry tropical forests, 50-70 m for Pacific Northwest, over 200 m for Old-growth Douglas fir, 100- 150 m for tropical rain forests. Physical edge effects have been shown to increase growth rates, elevate rates of mortality, reduce stocking density, and differentially affect regeneration (Chen et al. 1992). Edge effects are a major reason for the decline of forest birds in heavily fragmented landscape (Brittingham and Temple 1983). There are also the increased rates of nest predation by opportunistic predators and problems with weedy species invading natural disturbed sites.  Especially in pristine forests, problems with weedy species are greater when the forests are surrounded by succession in habitat rich in persistent weedy species. The high number of edge effects implies that habitat patches below a certain size will lack the “core” habitat that some species require. 
Biological Consequences 
The effects of fragmentation can be seen at several levels of the biological organization, from changes in gene frequency within populations to wide reaching changes in distribution of species and ecosystems. An important effect of fragmentation is the isolation of habitats by movement of barriers. Species that are restricted to a certain kind of habitat may depend on a constellation of habitat patches at relatively close proximity, if no single patch is large enough to meet the needs of individuals or groups. Many animal species require a mix of habitat types, with distinct resources, in order to meet their life history requirements. If these critical areas become separate by barriers, populations may decline to extinction. Barrier effects are both relative and cumulative. The cumulative effect of many barriers is probably what destroys populations in many cases. In a multiple species context, the landscape matrix will allow individuals of some species, but not of others, to pass through. The long-term effects of dispersal barriers on population dynamics are generally unknown. Habitat fragmentation causes indisputable declines in populations of species isolated on fragments, a finding supported by many fragmentation studies in which, movement through or presence in the matrix was measured (Klein 1989, Diffendorfer et al. 1995, Sarre et al. 1995, Stouffer and Bierregaard 1995b). Wind dispersed or vagile species are more likely to arrive at isolated remnants as there is the difficulty of crossing the matrix surrounding the remnant. Removal of vegetation creating an unexpected amount of gaps may influence colonizing ability and therefore isolating areas. The density of Bachman’s sparrows for example, decreases with the distance from a source habitat (Dunning et al.1995). Also, similar but sub-optimal matrix can become sink habitat (e.g., spotted owls). In these cases, dissimilar matrix may be preferable.
By definition (Davies, Margules, Lawrence, 2000; Connor, Courtney, Yoder, 2000), for species to be completely isolated on a fragment, two conditions are necessary: (1) the species do not occur in the matrix and (2) the species do not disperse between fragments. More recent formulations of Metapopulation Theory show that dispersal can affect the abundance of populations in patches (Hanski et al. 1994). Species do not disperse along the ground through the matrix, or rarely do so. Unfortunately, one can say little about other modes of dispersal. Thus, it is possible that some species categorized as isolated meet the first condition, but are able to disperse, undetected, between fragments. An alternative explanation is that habitat specialists declined because the quality of their habitat on fragments was reduced in contrast to isolated species. Species present in the matrix are mostly unaffected by fragmentation. Extending the definition, species can be non-isolated in two ways: (1) species move freely through the matrix between fragments, but not inhabit the matrix, or (2) populations inhabit the matrix so that, even if individuals move only relatively small distances, the overall effect is a mixed population at the between-fragment or landscape scale. Some species probably perceive the matrix as better quality habitat than the fragments, resulting in a net flow of individuals from the matrix into the fragments. Thus, depending on a species’ perception of the quality of the matrix habitat, a landscape that is fragmented to one species may be continuous to another (see Diffe ndorfer et al. 1995, Stouffer and Bierregaard 1995a, Ingham and Samways 1996).
When an area is isolated by destruction of surrounding natural habitat, removal of the surrounding vegetation renders the remnant the only area suitable for the biota. This may lead to crowding, which in turn would induce intra- and inter specific interactions, competition or predation. As a result, there may be an initial increase of population density in animal species. This may occur initially in the fragment as animals are displaced from their former homes. The crowding effect has been described for both tropical (Leck 1979) and temperate (Noss 1981) forest reserves. The initial rise in population in isolated fragments is followed by population collapse. Longer-term crowding effects are likely in many cases but has not been proven. The subsequent biological consequence of crowding effects is not known.
Another problem caused by fragmentation is genetic alterations to species not attributed to adaptation or natural selection. Long term genetic effects are uncertain, usually inferred. Low rates of genetic interchange between populations appear sufficient to prevent inbreeding depression and other genetic problems (recall that one immigrant per generation is required to prevent genetic drift). Although inbreeding depression and genetic drift may increase extinction risk in small, isolated populations, in other cases fragmentation may increase the among-population component of genetic diversity (Simberloff and Cox 1987; Simberloff 1988). The creation of fragmented distribution and population bottlenecks by human activities has apparently increased genetic differentiation in the populations of some species when the species are still in proximity to each other and can have genetic exchange (Leberg 1991).  However, these genetic differentiation within the populations that are no longer connected causes genetic drift. 
As shown above, there are many changes to an ecological community due to fragmentation. One consequence is non-random loss of species, and another is that fragmentation is usually accompanied by an influx of new species from surrounding, altered landscape. This invasion is shown in the composition of bird communities in old-growth forests patches (Schieck et al 1995). More invasive exotic species can establish themselves in modified forest ecosystems. This is a danger because alien species may out-compete the native species and become dominant where it does not belong. This is an important issue and was one of the main topics in the agenda of the sixth annual SUBSTTA (Subsidiary Body of Scientific Technical and Technological Advice) meeting of the Convention on Biological Diversity (March 2001).  There is also potential for disease as wildlife and domesticated animals are in closer proximity in the distressed forest community. A further consequence of fragmentation is the loss of predators, which are important as the top of the food chain, completely disrupting it. Microclimate changes in edge habitat are other alterations that, can adversely affect invertebrates, therefore decomposition rates, pollination, seed predation, insect parasitism, etc. Fire suppression in grazing and agricultural fields around forest habitat also results in loss of animal and plant species. Lost to competition are the short, small-seeded, and nitrogen-fixing plants from the wettest and most productive areas. In conclusion, many native plant and animal species are lost in the homogenization of the landscape.
In order to escape these potentially dangerous effects, animal species have essentially three options for surviving a highly fragmented landscape. First, a species might survive or even thrive in the matrix of human landscape (a number of species fit this description throughout the world). Second, a species might survive in a fragmented landscape by maintaining viable populations within individual habitat fragments (this option is only suitable for species with small home ranges or with few life history requirements). A third way to survive in a fragmented landscape is to be a highly mobile species that can integrate a number of home range habitat patches or an interbreeding population.  A species incapable of pursuing one or more of these options is bound for extinction. Besides the above options for species survival, there is a need to ask which species are most vulnerable to local and regional extinction as a result of fragmentation. There are some hypothesis about the relative vulnerability of species to fragmentation.
Vulnerable Species 
Some species are at greater risk in fragmented landscapes than are others. Identifying these species would be useful. Theory predicts that species with particular traits may be at greater risk of extinction. Many rare species are endemic with small distribution and are found in one or a few patches of suitable habitat, making them very susceptive to fragmentation. Whether rare or even declining species are headed toward extinction in most cases is uncertain, however they are definitely at great risk. Large populations typically become small before going extinct (Hanski et al. 1994). In what follows, assume that there is a greater decline in abundance on fragments equating with increased extinction risk. A test of the relationships between population declines and traits was developed by Davies et al. (2000) who predict the increase extinction risk for species in an experimentally fragmented forest landscape ( Davies, Kendi,  C. Margules and  J. Lawrence 2000). In their study traits of species were considered and linked, in theory, to extinction risk. Two of these traits that were deemed important by this study are also found in other studies. These traits and characteristic are the body size and the trophic level membership of species.
In the large body of literature, a range of findings including positive, negative, and no relationship between extinction risk and body size have been reported (e.g., Terborgh and Winter 1980, Pimm et al. 1988, Soule et al. 1988, Burbidge and McKenzie 1989, Laurance 1991, Rosenweig and Clark 1994, Angermeier 1995, Gaston and Blackburn 1996b; for review, see Gaston and Blackburn 1996a). Perhaps it is not surprising that a clear pattern has failed to emerge given the complexity of the relationship between extinction risk and body size.
Body size is not directly linked to extinction risk, but because it is correlated with three variables (abundance, population fluctuations, and population growth rate), it can be linked with extinction risk. However, the relationships between body size and these variables are uncertain because, first, the relationship between body size and abundance has received much attention that reveals no clear pattern (Blackburn et al. 1992, 1993, 1994, Cotgreave 1993, Blackburn and Gaston 1994, Gaston and Blackburn 1995, 1996a, c, d, Cyr et al. 1997). Recent evidence suggests that the spatial scale of the study may be crucial to establish the relationship. Studies at local spatial scales rarely find a relationship between body size and abundance, whereas studies at regional scales often find negative linear relationships (Blackburn and Gaston 1997). This may explain the results by Davies et al.  Their study was carried out at a local scale and there was no relationship between abundance and body size. Second, the relationship between population fluctuations and extinction risk is also contentious (McArdle and Gaston 1993). Some theories assume that populations of large-bodied taxa fluctuate less than populations of small-bodied taxa, and therefore predicts that large species are less likely to fluctuate to extinction (reviewed in Pimm 1991). On the other hand, others argue that neither empirical evidence nor logic supports this theory (Schoener and Spiller 1992, Tracy and George 1992), and that fluctuations are only likely to lead to extinction when populations are unrealistically small (Chesson 1991). Third, because small species recover faster from low numbers than do large species, they could be at lower risk of extinction (Goodman 1987, Pimm et al. 1988). On the other hand, large body size is also associated with high longevity, which could also lower extinction risk (Pimm et al. 1988).Finally, the action of these variables in combination probably contributes to the fuzziness of the body size vs. population decline relationship. That is, positive correlations between body size and extinction risk due to population abundance and rate of recovery (Cotgreave 1993, Lawton 1994), are offset by the negative correlation between body size and extinction risk due to population fluctuations (Pimm 1991).
As for trophic group membership, theory predicts that species at the top end of food chains are more prone to extinction than species at lower levels because the former tend to have more unstable population dynamics, and are less likely to persist in a fluctuating environment (Pimm and Lawton 1977, Lawton 1995, Holt 1996). An important test to assert  prediction that species in trophic groups at the top end of food chains are at the greatest risk of extinction in experimentally fragmented forest would be very helpful to this discussion. The predicted greater vulnerability of species at higher trophic levels is due to unstable population dynamics and vulnerability to fluctuating environments, at least among species with populations that had become small and isolated on fragments. Isolated predators were no more vulnerable than herbivores or detritus feeders. However, among species that declined, predators declined the most. Only a few empirical studies have linked extinction risk and trophic group, and their results have been mixed. In laboratory microcosms of protozoa and bacteria, prey went extinct more often than predators (Lawler 1993). For small mammals in boreal forest fragments, the proportion of predators declined with decreasing species richness, but the proportion of insectivores remained constant, and the proportion of herbivores increased (Patterson 1984). In a study incorporating data from five different surveys of vertebrates and plants, there was no tendency for species in one trophic group to go extinct more often than any other (Mikkelson 1993). Perhaps these results are mixed because other traits also determine extinction risk (Lawton 1994). For example, species at higher trophic levels are often large in body size, but populations of large species are thought to fluctuate less, and therefore, to be less extinction prone (Pimm 1991). Conversely, species at higher trophic levels are usually the taxa with the lowest population densities and, thus, the highest risk of extinction (Gard 1984).
Species Especially Vulnerable 
The following are the species especially vulnerable:
a) Species with Short Life Cycles - These species are dependent upon a certain niche for their survival. Due to their short life history, they have to be able to adapt quickly enough so that they can grow and thrive. They have to be able to survive till the reproductive mature stage is reached (e. g., extirpated shrubs, climbers, and epiphytes). 
b) Species with Variable Population Sizes - Are species using patchy or unpredictable resources, and therefore are very dependent on the health of the fragment (e.g., ephemeral wetland species, mast-dependent species, the Bay checker-spot butterfly). 
c) Naturally Rare Species - Terborgh and Winter (1980) concluded that rarity is the best predictor of population vulnerability. Notwithstanding, there are many potential reasons why a species is rare ( Rabinowitz et al. 1986). Some plants and animals are rare because humans have driven them to that condition. Other species are rare naturally and have limited and patchy geographic distribution, very narrow distributions and/or low population densities (e. g., demographic, genetic, or environmental stochasticity).
d) Wide-ranging Species - Some animals, such as large carnivores and migratory ungulates, roam a large area in the course of their daily or seasonal movements. Even large fragments will not be able to provide enough area for viable populations and they must travel widely and often attempt to move in heavily fragmented landscapes. As a result, they may encounter dangerous situations that will lead to their death.
e) Nonvagile Species - Species with poor dispersal ability will not move very far from where they were born. Some species of birds have very low colonizing abilities and will not cross areas of unsuitable habitat (Diamond 1975; Opdam et al. 1984; van Dorp and Opdam 1987). Without the occasional arrival of immigrants to provide a rescue effect on genetic diversity, these species will not persist long in a habitat fragment.
f) Species with Low Fecundity - A species with low reproductive capacity cannot rebuild its population after a severe reduction. These species may be prone to genetic deterioration because they are unable to recover adequate levels of genetic variability after large losses (Carroll 1994).
g) Species with Specialized Habitat - Species dependent on patchy or unpredictable resources or otherwise highly variable population size are very susceptible to extinction. When the resource fluctuation is seasonal or annual, species dependent on these resources also fluctuate. Populations may also fluctuate in response to weather extremes or other variations in the physical environment. The higher the level of fluctuation, the greater the chance of extinction (Karr 1982; Pimm et al. 1988). 
h) Ground Nesters - Nesting on or near the ground is another life history trait not well suited to ecological conditions in fragmented landscapes. Ground nesting birds and other animals are highly vulnerable to various mesopredators present in areas with high levels of edge habitat (Wilcove 1985).
i) Interior Species - Some species avoid habitat edges. They occur only in the interior of forests, prairies or other habitats and are absent from small habitat patches with little or no true interior habitat.
j) Species Exploited by Humans - Some species are actively sought by people for food, fur and other needs. Most habitat fragments are highly accessible to humans due to high levels of edge-interior ratio in habitat fragments. In travelling between habitat fragments, animals may be visible and easily killed or collected by people.
 Examples of Fragmented Ecosystems and their Study
Examples of fragmented ecosystems can be found throughout the globe, one being the Amazon Forest and the other the Brazilian Atlantic Forest. Fragmentation can specially be found in the Atlantic Forest where the entirety of the forest ecosystem has been fragmented. 
a) Amazon Forest - There is fragmentation throughout the Amazon Forest, however it is not highly fragmented as a rule. The forest becomes highly fragmented near cities in the region, and the largest is Manaus. Near this city of 2 million people a study on fragmentation is presently under way (the BDFF-The Biological Dynamics of Forest Fragment Project). This project began as a minimal critical size of ecosystem project. The defined goal was to identify a minimum size of tropical forest habitat that would maintain most of the biotic diversity found in an intact ecosystem. The research involves study of plant and animal communities in forest plots before and after isolation due to fragmentation. Additionally the plots are to be compared to control studies in adjacent continuos forest, through time.  With the data gathered, specific predictions and correlation are to be made between forest fragment size and retention of diversity. During this study (between 1980 to 1984) 1-, 10- and 100-ha fragments where isolated.  In 1990, other 100-ha reserves were isolated, 200-ha remnant, isolated in 1979, was included in the sampling and data collecting (Smithsonian Institute, 1998).
After 20 years of observations, BDFF researchers realized the complexity of their task. It was an oversimplification of their part to expect to predict the carrying capacity based on the size of the reserve alone. Species/area relations are insufficient to understand all the processes that determine how many and which species will be present in a given fragment. Rather, species-specific habitat requirements, structural changes in the reserve originating from the forest edge, and changes in vegetation are key in determining ecosystem structure in fragments. A few of the results of this 20-year research are:
1) Fragmentation caused a decrease in species richness of primates, birds and some insects such as bees, ants and termites. 
2) Fragmentation caused an increase in species diversity of small mammals, amphibians and butterflies with some changes in species composition. 
3) Within taxonomic groups, fragment size has proven to be an important variable on community evolution. 
4) Herbivore species play an important role in fragmentation. They have a strong negative impact on the regeneration of degraded lands.
5) In the study of fragments, trees and other plant species have been subject to serious structural changes causing an alteration in resource distribution, mainly attributed to fruit and seed.  Another negative effect of this fragmentation is on tree survival and leaf fall.
6) Edge effects are shown to have caused many of the problems attributed to fragments in the BDFF study. Animal response to edge habitat and barriers are variable. The importance of a matrix habitat for the understanding of population and community dynamics in fragments is re-iterated. 
The research results demonstrate that there is a complex myriad of factors affecting habitat fragmentation in the Amazon region near Manaus. The most important variables are the relationship of forest fragment size and species number, the dynamics of the forest edges, both biotic and physical, as well as the interaction between the forest islands and the matrix habitat around them  (Smithsonian Institute, 1998).
b. Atlantic Forest - The Atlantic Forest is an even better example than the Amazon in showing the problems of fragmented ecosystems. The entire Atlantic Forest ecosystem is in the verge of collapse due to fragmentation. Known in Brazil as “Mata Atlântica”, this forest is one of the most endangered ecosystems on Earth. Most of its native plant and animal species are endemic to the region (Mittermeier et al., 1982; Myers, 1984; da Fonseca, 1985; Alves 1990; Viana and Tabanez, 1995). In the Atlantic Forest, on the Southern part of the state of Bahia, there are 450 plant species per hectare, three out of four are endemic (Conservation International, 1993). Of the 63 brazilian primates, 24 live in the Atlantic Forest and 19 are endemic to the region (Conservation International, 1990). The Atlantic Forest is also home to 199 birds, 260 reptile and amphibian plus 73 mammal species (Conservation International, 1993). The Mata Atlântica originally extended throughout the Brazilian coast, from the Northeastern to Southeastern coast of Brazil. All together, it covered about one million square kilometers (Dean, 1995). Today, the Atlantic Forest holds 171 of the 202 known endangered animal species in Brazil (Conservation International, 1993) and covers less than 10% of its original extent (da Fonseca, 1985; Conservation International, 1993; Algers, 1994; Dean, 1995), and less then 5% of the primary forest vegetation remains (Conservation International, 1993).   Animal species in the region are in great danger of extinction including the Atlantic marmoset, the thin-spined porcupine, the red necked tanager, the muruqui or wooly spider monkey, the red billed currasau and the golden lion tamarin.
The human inhabited part of the Atlantic Forest region contain large industries and highly urbanized areas. Only very small pockets remain as sanctuary for wildlife. Eighty million people living in the coastal areas of the region create enormous problems. Most areas of the forest have been cleared for crops like coffee and sugar cane, for pastureland and for supply of fuelwood. Of the original extent, 15% of the forest remains in the state of Bahia, 7.5% in Espirito Santo, 5% in Rio de Janeiro, and 4% in Sao Paulo. Minas Gerais is estimated to have between 1 and 5% of their remaining forest while the states of Paraná, Santa Catarina and Rio Grande do Sul hold together 10% of their original forest. Most of the remaining forest is located in areas poorly suited for agriculture. 
A description of the Atlantic Forest ecosystem is important for the understanding of the regional fragmentation. The description is mainly drawn from the book “With Broadax and Firebrand: The Destruction of The Brazilian Atlantic Forest” By Warren Dean. The Atlantic Forest is a complex extending inland from the coast, from 100 to 500 kilometers in certain places. The Mata Atlântica occurs along the coast due to the topography. Along most of its length, a few kilometers inland, there is a mountain chain known as the “Serra do Mar”. It reaches 1,000 meters in height. It is predominant in the Central to Southern reaches of the coast. Behind the first wall, there is a parallel arrangement of still higher mountain chains. These mountains add as much as another one thousand meters in height to the coastal profile. Against all these barriers, there is a steady Eastern trade wind that draws moisture from the warm sea. As this air current is lifted, it cools down and rain clouds are formed. Overall rainfall in the forest is of about 1,500 millimeters a year, with some of the Eastern faces of the mountain receiving precipitation of more than 4,000 millimeters a year.  The forest is divided by the mountains and classified as coastal or landward facing.
The main vegetation type of the Atlantic forest is of tropical mesophytic broad-leaf evergreens on the coastal forest. This type consists of trees 30 to 40 meters tall with trunk 12 meters or more in circumference. Underneath these evergreens are four more stories of trees consisting of shorter broad-leaved trees, bamboo and giant ferns. Epiphytes cover some of the branches, vines cover the trunks and lianas fill the space between trees. The coastal forest covers the Eastern slopes of the coastal mountains and extends throughout the coastline. As the forest reaches the ocean and other watercourses, the forest gallery becomes more open and studded with palm trees. It forms a coastal plain forest. 
On the Western landward faces of the mountains, there is a tropical semi-deciduous mesophytic broadleaf forest. The coastal vegetation gives away to this second type of forest composed of both evergreen and semi-deciduous species. There is an increasing proportion of semi-deciduous species appearing toward the Western face where tree heights are lower and lianas and epiphytes are rarer. A highland forest occurs among the inter-montane valleys, where the vegetation is structurally and floristically similar to the coastal plain forest.
In the Southern highlands of the states of Paraná, Santa Catarina and Rio Grande do Sul, the Atlantic forest acquires another character. It is dominated by a primitive hardy pine - Araucaria Augustifolia. The Araucaria forest is an open forest formation. Inland, where rainfall becomes too scarce and seasonal to support forests, the “Campo” begins in the South, and the “Cerrado” in the Central to the Northern regions.
 To describe the fragments and fragmentation processes of the Atlantic Forest, the Santa Rita forest study area is used to illustrate the problem. This forest shows many of the difficulties in keeping fragments of forest viable. The forest is a privately owned forest fragment that has been studied since 1990 by the University of Sao Paulo’s Laboratory of Tropical Silviculture as part of their Biology and Management of Forest Fragments Project. 
The study site is located in the Central part of the State of Sao Paulo. It is characterized as a “Plateau Forest” and is found in the Western part of the Atlantic moist forest. The forest is semi-deciduous with high plant species diversity and emergent trees up to 55 meter in height. The soil has moderate to high fertility and good drainage. The climate is seasonal, with an average rainfall of 1,250 millimeters per year, and is characterized by a pronounced dry season during the winter. The fragment is of 9.5 hectares, well protected, with no record of logging. The former owner practiced intensive agriculture on the remainder of the area but chose not to plant sugarcane around the fragment. The area was recently sold and the new owner has planted sugarcane to the forest edge and a fire destroyed part of it in mid-1994. 
The prediction set forth by the forest fragment project estimates that, since forest fragments are degrading, poor or non-existent tree regeneration and critically small populations of several tree species should be found (Viana and Tabanez, 1995).  Another indicator might be a poor forest structure dominated by low diversity eco-units in an arrested forest succession process (Viana et al., 1995). 
The Santa Rita Forest has a relatively high diversity with 101 tree species. In comparison with other fragments being studied in the region, the Santa Rita forest is relatively diverse and well protected (Viana et al, l995). This forest, much like any other plateau forests, is a mosaic of patches, ecological heterogeneous units of forest patches within forest fragments that have similar structure and successional stage. Four different units were found in the forest: low forest (“capoeira baixa”), high forest (“capoeira alta”), old growth forest (“mata madura”), and a bamboo dominated forest. These units vary in diversity and structure.
All these ecological units are dominated by high light capoeira eco-units where pioneer species are expected to have high density (Viana et al, 1995). However, only 13 individuals of pioneer species favored by large-gap environments were found at the site. This is equivalent to 2% of the total density of all species. 
In the study, two possible explanations for the low number of pioneers were examined. There is either a lack of availability of seeds in the soil or competition with climbers and vines. The soil seed bank was found to have a high seed density of pioneers, showing that the lack of seed availability is not a barrier to regeneration success. Relative success of climbers and vines was analyzed using leaf area indices. 
Fifty percent of the total leaf area in the forest eco-units was attributed to vines and climbers (Viana et al., 1995). Some climbers in the region have been found to over compete with pioneer trees, and end up dominating large patches of the fragment. These climbers put their weight on small trees, which are very frequently found bent and broken. This may explain the low density of large-gap pioneers, that cannot occupy these habitats favorable to them in environmental conditions (Viana, 1995). The pioneer species are not regenerating well because they are losing the competition with vines and climbers that occupy the low forest units (Viana et al., 1995). 
Some other tree species, which are not pioneer, also have poor regeneration rates associated with vine and climber competition. There are cases of large emergent species that have a barrier at the sampling stage. Many of the emergent trees species show discontinuities in population structure. Other barriers to tree regeneration in the forest fragments include lack of pollinators and seed dispersers, excessive predation by lianas and vines and unfavorable microclimatic conditions (Nepstad et al., 1990). The apparent lack of regeneration of some species, if not a result of sampling limitations, posses an important threat for regeneration ecology. A limiting factor in the fragmented landscape of the Atlantic forest is the self-sustainability of the small population size of several species (Viana. 1995). This is likely to be a problem in the long-term sustainability of the species due to the fact that these populations are completely isolated. This isolation is a serious problem for the Atlantic Forest ecosystem. Another adverse factor is that several species of pollinators may have gone extinct locally, further reducing genetic flow and creating barriers for self-sustainability. There is evidence that forest succession in small forest fragments in the plateau forest is not likely to recover forest diversity, as evidenced by the low diversity of pioneers and the overabundance of climbers. A critical issue in forest fragmentation is the reduction of population size (Soulé, 1987).
Management Suggestions  
To manage fragmentation, several propositions can be made and are shown below: 
a)  Identify and concentrate on vulnerable species. 
b) Keep remnants as large as possible. The larger the reserve, the more species (the species/area relationship), and the more likely minimum viable populations are being maintained. Smaller the remnant, the more influenced by external forces, edge effect. (From 0 - 22% of current reserves are expected to support large mammalian carnivores (10-100kg) for 100 years. None are expected to support them for 1000 years. From 4 - 100% of reserves are expected to support large herbivores for 100 years. From 0 - 22% of reserves are expected to support large herbivores for 1000 years. For large mammals (more them >50 kg) to persist for 100,000 to 1,000,000 years, reserves 1,000,000 to 1,000,000,000 km2 are needed. The majority of the reserves though, are under 100,000 km2) (Belovsky 1987). 
c) Heterogeneous areas are generally superior to homogeneous areas. Natural patches (metapopulations) may accommodate change and disturbance better than homogeneous areas. 
d) Consider remnant shape. Shape determines the area/perimeter ratio, i.e. edge versus interior. It has been suggested to grow an artificial edge around a fragment to “seal” it. A long thin strip will have more edge than a round or square piece. However, it is possible that a long strip might contain more species and will be less sensitive to further habitat loss. 
e) The position in the landscape might be important to determine the impact of fragmentation. Effect of the matrix on the reserve, buffer zones, etc. 
f)  A single remnant is more vulnerable than several ones. Corridors either designed as a matrix or as stepping stones have often been proposed. They would allow periodic movement between habitat types and immigration and emigration. There are three scales for this: the Fencerow scale, that can connect small, close patches that are entirely edge, e.g., line corridors such as hedgerows, cut lines, highway medians of trees or shrubs for mice, birds, chipmunks. The landscape mosaic scale, that can connect major habitat features, and be used for movement of interior and edge species, e.g., strip corridors, mountainous ridges, forested streams. The regional scale, that can connect reserves into networks. The corridor design should follow the species characteristics, e.g., culverts and tunnels under roads for amphibians. Also, there are both costs and benefits to corridors (Simberloff et al. 1992). However, they could facilitate the spread of pests and disease. Their establishment could also drain resources that might otherwise enlarge existing reserves or establish new ones. Finally, one should consider linking north and south, high and low altitude habitats as a precaution against global warming.
Fragmentation, loss and isolation of natural habitats are the greatest threats to diversity. Where natural disturbances and other processes create heterogeneous landscapes rich in native species, human land uses often create islands of natural habitat embedded in a hostile matrix. This fragmentation reduces and prevents normal dispersal of species, and increases edge effects and other threats. Fragmentation acts to reduce biodiversity through four major mechanisms that are: first, because remaining fragments represent only a sample of the original habitat, many species will be eliminated by chance (initial exclusion); second, the modified landscape in which fragments exist is often inhospitable for many native species, preventing their normal movement and dispersal (isolation); third, small fragments contain less habitat, support smaller populations and are less likely to intercept paths of dispersing individuals (island-area effect). 
Another mechanism that must be considered is related to climatic influences and opportunistic predators and competitors from the disturbed landscape that penetrate into fragments reducing the core area of suitable habitat (edge effect). There are a number of types of species that are very vulnerable to fragmentation as well. Throughout the world, there are a large number of examples of fragmentation and two of them are well illustrated by the Amazon and the Atlantic Forest region ecosystems in Brazil. To conclude, there is a pressing need to stop fragmentation if ecosystems are to be protected from this destruction process.
Final Words 
To protect the biodiversity of a modified tropical landscape, fragmentation effects must be recognized and controlled. In order to do this, the ecological value of individual forest fragments should be considered and efforts should focus on the protection of flora and fauna within these individual patches. Often this is accomplished through the creation of forest reserves where the fragmentation effects are managed on a continual basis (Wiens 1994). A landscape approach to management, considering the external influences created by the surrounding matrix on the native vegetation, is taken in order to control these effects (Saunders 1991). Acquiring a representation of community types or ecosystems through a network of reserved fragments may potentially increase the likelihood of preservation of biodiversity at the landscape level. However, the surrounding matrix lands should not be overlooked as a contributor to this effort (Holl 1999). Restoration and recovery of land that has been removed from agricultural areas could connect forest patches, potentially reducing edge effects and isolation, allowing increased habitat for fauna and improved seed dispersal. 
A review of the current literature on tropical landscapes suggests that still more research on tropical fragmentation and its impact on ecological processes are needed in order to create the best conservation strategies and to ensure the protection of tropical biota. The accomplishments in research so far, however, provide managers with the foundation for effective land management plans that control some of the consequences of fragmentation such as edge effects and isolation.