Landscape Ecology Approach in the Conservation

Salinas Chávez, Eduardo; Middleton, John. 1998. La ecología del paisaje como base para el desarrollo sustentable en América Latina / Landscape ecology as a tool for sustainable development in Latin America. http://www.brocku.ca/epi/lebk/lebk.html

Landscape Ecology as a Tool for Sustainable Development in Latin America
edited by Eduardo Salinas Chavéz and John Middleton

(1) Jean Paul Metzger is an assistant professor at the University of São Paulo. He did his PhD in landscape ecology at the "Centre d'Ecologie des Systèmes Aquatiques Continentaux" (CNRS, Toulouse, France), under the supervision of professor Henri Décamps. At the moment, his researches focus on the impact of different landuse and agricultural practices in the spatial distribution of primary and secondary forests in Brazilian Amazonia. His work deals also with the influence of landscape connectivity and matrix structure on community diversity in the Amazonian and Atlantic landscapes .

Laboratory of Landscape Ecology and Conservation, Department of Ecology, Institute of Biosciences, University of São Paulo, 05508-900, São Paulo SP, Brazil. E-mail: jpm@usp.br

(2) Vânia R. Pivello graduated in Biological Sciences by the University of São Paulo, Brazil. In the same University, she did her MSc in Ecology, related to fire effects on cerrados (Brazilian savannahs). Her PhD research was developed at the Imperial College Centre for Environmental Technology, University of London, concerning the use of fire to manage cerrado protected areas through a decision support system. Her present research interests focus on fire effects on savannahs, nutrient cycling influenced by fire, qualitative modelling (especially expert systems), the conservation of terrestrial ecosystems and sustainability, plant invasion in fragmented habitats, and the management of protected areas.

Laboratory of Landscape Ecology and Conservation, Department of Ecology, Institute of Biosciences, University of São Paulo, 05508-900, São Paulo SP, Brazil. E-mail: vrpivel@usp.br

(3) Carlos Joly is a Plant Ecology professor and the Director of the Botanical Garden "Hermogenes de Freitas Leitao Filho" of the University of Campinas (UNICAMP). He coordinated the project "Study, preservation and rehabilitation of gallery forest areas in the Jacaré-Pepira watershed", from 1986 to December 1992. During the last years, he has dedicate himself to the development of research related to the conservation and sustainable use of São Paulo state biodiversity (www.bdt.org.br/bdt/biotasp).

Department of Botany, Institute of Biology ,State University of Campinas, C.P. 6109, 13083-970 Campinas, Brazil. E-Mail: cjoly@obelix.unicamp.br

The ideal of economic development followed by Western countries in this century conflicted with a rational use of natural resources and has resulted in serious environmental losses. From the 70s, environmental awareness became stronger and new alternative ways to reach development started to be sought, trying to bring together economic expansion and environmental conservation. In Brazil, the concern on sustainability came to influence decision making mainly after 1986, with new environmental legislation. In this context, the landscape ecology approach, integrating social, biological and physical environmental elements at scales compatible with the management of territories, shows a great potential for planning towards sustainability. Analysis at the landscape level has recently emerged in Brazil, either in academic research or in applied projects for territorial planning and environmental impact assessment. In this chapter, we present an example of the use of landscape ecology concepts for the sustainable development of the Jacaré-Pepira river basin, in SE Brazil (São Paulo State). This river lays in a region which had been highly deforested in the second half of last century and beginning of this century, and today, it presents serious problems of erosion, river filling, agricultural pollution and biodiversity loss. The main purpose of this project was to create a model for riparian forest rehabilitation, aiming at reestablishing the ecological functions of these ecotones, particularly relating to the protection of river margins, the purification of the water table and the establishment of corridors for regional flora and fauna dispersion. Landscape ecology concepts were used to analyze spatial variations in the floristic composition and species diversity of riparian forest fragments and their relationship with the landscape structure. Fifteen riparian forest fragments under similar hydrogeological, topographic and edaphic conditions but of different landscape types - especially related to landscape composition and to degree of connectivity and isolation - were studied. The results show important variations in the species composition and diversity in the fragments according to the landscape type. One of the most relevant relationships was obtained between species diversity and forest connectivity. This relationship, always positive, was even more significant for the zoochoric (the most abundant) and climax (typical of the interior forest) species. Still, the results show it is necessary to adapt the rehabilitation action to each landscape type. Landscapes with higher connectivity may indicate advantageous conditions for a quicker rehabilitation or, inversely, the recovery of riparian corridors may be used to favour an increase in landscape connectivity. This study also adds indicator species to be used in reforestation projects and the assignment of adequate forest widths for maintaining species diversity. In this study, the main obstacles to the application of the results obtained were due to unstable political and financial support. The understanding and the usage of landscape ecology concepts for sustainable development need long term research, carried out at the river basin level, and with local population partnership.

El modelo de desarrolo económico adoptado por los países occidentales en este siglo ha ido de encuentro a una utilización racional de los recursos naturales, resultando en serios perjuícios ambientales. A partir de los años 70, surgió una mayor conscientización ambiental y se han buscado modelos alternativos de desarrollo, aliando crescimiento económico y conservación ambiental. Las preocupaciones con la sustentabilidad emplezan a reflejarse en las tomadas de decisión en Brasil, principalmente a partir de 1986. En este contexto, la ecología del paisage, como disciplina integradora de abordage social, biológico y físico en escalas compatíbles con el manejo de territorios, aparente tener grán potencialidad para el planejamento de un desarrollo sustentable. Análises en la escala del paisage se han difundido recientemente en el Brasil, tanto en centros de investigación como en proyectos aplicados de ordenación territorial o de estudios de impacto ambiental. En este capítulo, presentamos un ejemplo de aplicación de conceptos de ecología del paisage en el desarrollo sustentable de una región del SE del Brasil, lacuenca del rio Jacaré-Pepira (Estado de São Paulo). Este rio se situa en uma región intensamente desflorestada en la segunda mitad del siglo XIX y en el inicio del siglo XX y que apresenta serios problemas de erosión, colmatage del rio, polución agrícola y pierda de la diversidad biológica. El principal objetivo de este proyecto fué de crear un modelo de reabilitación de bosques riparios para restabelecer las funciones de estos ecotonos, en particular en lo que dice respecto a la protección de los margenes de los rios, a la depuración de las aguas de la capa freática y al establecimiento de passillos de dispersión para la fauna y flora local. Conceptos de la ecología del paisage fueron utilizados para analisar las variaciones espaciales en la composición y diversidad florística de parches de bosques ripários y suas relaciones con la estrutura del paisage al rededor de estos parches. Se estudiaron 15 parches de bosques ripários, situados en condiciones hidrogeologicas, edaficas y topograficas semejantes, y en tipos de paisages diferentes, particularmente cuanto a la composición del paisage y al grado de isolamiento y conectividad de los parches. Los resultados muestran importantes variaciones en la composición y en la diversidad específica de los parches en función del tipo de paisage. Una de las relaciones más significativas se obtuvo entre la diversidad específica y la conectividad florestal. Esta relacción, que es siempre positiva, es particularmente significativa para los grupos de especies zoocóricas (el grupo más abundante) y climácicas (típicas del interior del bosque). El conjunto de los resultados muestra que es necessario adaptar la acción de reabilitación a cada tipo de paisage. Paisages con mayor conectividad pueden constituir condiciones favorables para una reabilitación más rápida o, inversamente, la reabilitación de passillos ripários puede ser utilizada de manera a favorecer el aumento de la conectividad del paisage. Este trabajo resulta todavia en indicaciones sobre la elección de las especies a seren utilizadas en un reflorestamiento y sobre el anchor de bosque necesario para la manutención de una mayor diversidad específica. En este proyecto, el principal obstáculo para la aplicación de los resultados alcanzados fué una instabilidad en el soporte financiero y político. La comprensión y aplicación de conceptos de ecología del paisage en el desarrollo sustentable necesita investigación de largo plazo, desarrollada al nível de las cuencas hidrográficas, con participación de la población local.

O modelo de desenvolvimento econômico adotado pelos países ocidentais neste século vem de encontro com uma utilização racional dos recursos naturais e vem acarretando sérios prejuízos ambientais. A partir dos anos 70, começa a surgir uma maior conscientização ambiental e tem-se procurado modelos alternativos de desenvolvimento, aliando crescimento econômico e conservação ambiental. As preocupações com a sustentabilidade começam a se refletir nas tomadas de decisão no Brasil, principalmente a partir de 1986. Neste contexto, a ecologia da paisagem, como disciplina integradora de abordagens social, biológica e física, em escalas compatíveis com o manejo de territórios, aparenta ter grande potencialidade para o planejamento de um desenvolvimento sustentável. Análises na escala da paisagem vêm se difundindo recentemente no Brasil, tanto em centros de pesquisa quanto em projetos aplicados de ordenamento territorial ou de estudo de impacto ambiental. Neste capítulo, apresentamos um exemplo de aplicação de conceitos de ecologia da paisagem no desenvolvimento sustentável de uma região do SE do Brasil, a bacia do rio Jacaré-Pepira (Estado de São Paulo). Este rio situa-se numa região que foi intensamente desmatada na segunda metade do século XIX e no início do século XX, e que apresenta sérios problemas de erosão, colmatagem do rio, poluição agrícola e perda da diversidade biológica. O principal objetivo deste projeto foi o de criar um modelo de reabilitação de matas ripárias afim de restabelecer as funções destes ecótonos, em particular no que tange à proteção das margens dos rios, à depuração das águas do lençol freático e ao estabelecimento de corredores de dispersão para a fauna e a flora locais. Conceitos da ecologia da paisagem foram utilizados para se analisar as variações espacias na composição e diversidade florísticas de fragmentos de matas ripárias, e suas relações com a estrutura da paisagem no entorno destes fragmentos. Foram estudados 15 fragmentos de matas ripárias, situados em condições hidrogeológicas, edáficas e topográficas semelhantes, e em tipos de paisagens diferentes, particularmente quanto à composição da paisagem e ao grau de isolamento e conectividade dos fragmentos. Os resultados mostram importantes variações na composição e na diversidade específica dos fragmentos em função do tipo de paisagem. Uma das relações mais significativas foi obtida entre a diversidade específica e a conectividade florestal. Esta relação, que é sempre positiva, é particularmente significativa para os grupos de espécies zoocóricas (o grupo mais abundante) e climácicas (típicas do interior da mata). O conjunto dos resultados mostra que é necessário adaptar a ação de reabilitação a cada tipo de paisagem. Paisagens com maior conectividade podem constituir condições favoráveis para uma reabilitação mais rápida ou, inversamente, a reabilitação de corredores ripários pode ser utilizada de forma a favorecer o aumento da conectividade da paisagem. Este trabalho resulta ainda em indicações sobre a escolha das espécies a serem utilizadas num reflorestamento e sobre a largura de mata necessária para a manutenção de uma maior diversidade específica. Neste projeto, o principal obstáculo para a aplicação dos resultados alcançados foi a instabilidade no suporte financeiro e político. A compreensão e aplicação de conceitos de ecologia da paisagem no desenvolvimento sustentável necessita pesquisas de longo prazo, desenvolvidas ao nível de bacias hidrográficas, com a participação da população local.

The holistic and transdisciplinary approach adopted by landscape ecology distinguishes it as a unifying science, able to promote the integration and inter-relationship of both natural and social aspects of the environment, and the understanding of interactions, leading to a global perception of the situation and to comprehensive conclusions.

The landscape, earlier conceived as a "mosaic where the local ecosystem or land uses are repeated in similar forms throughout" (Forman & Godron 1986) comes to an essentially spatial definition, however, a very fortunate perception, since mosaics are a constant pattern in nature, at different scales. Subsequent additions to the landscape approach, including the cultural aspects which impelled land sharpening (Naveh 1995) broadened the concept.

The distinction of landscape patterns, associated with the possibility of dealing with different scales and the emphasis on social influences, make the landscape a very suitable unit for land planning and management (Forman 1995, Lyle 1985).

Planning has always been essential to adequately drive efforts towards an achievement. However, when the efforts relate to natural resources, not just planning but an ecological planning is necessary. The wasteful use of resources in the last decades, imposed by a technifyed and highly productive society is leading to resource scarcity. Man has changed the original landscape structure and now it is necessary to restructure the land in order to obtain perhaps a less productive but a more sustainable society, where resources are available on a long term basis, without compromising environmental integrity and quality. Sustainability considers "productivity in accordance to balanced economic and ecological benefits" (Falkenmark & Suprapto 1992).

It is also our view that sustainability cannot be reached without changing some modern society values and aspirations. Goodland (1995) contrasts "sustainable economic growth" with "environmental sustainability", stating that the former is a controversial term, once growth implies quantitative physical or material increase, that is, additional use of resources, and it is not possible to grow into sustainability. On the other hand, he stresses the need for "environmental sustainability", where levels of consumption must not overtake production, seeking to sustain global life-support systems indefinitely, which can only be reached through a better division of resources, or redistribution.

Planning and managing for sustainability also requires the understanding of the territory structure, functioning, the cultural forces which dictate them and all sorts of interactions and relationships (Sharitz et al. 1992, Naveh 1995), the basis of landscape ecology. As Forman says (1995), all ecological and human processes are spatially differentiated in the mosaic and thus, mosaic stability is a key element of sustainable development.

Brazil occupies an extensive area, mostly under tropical climate, where rain forests have developed, but also including seasonal climate in the Central Plateau, covered with savannahs, temperate conditions in the South and over the mountains, where mesophyllous or coniferous forests grow, and dry lands in the Northeast, covered with deciduous forests or scrubs. Thus, Brazilian territory comprises a tremendous variety of biomes (Figure 1) and still retains a great deal of its natural resources. These features award the country the highest diversity in the world, in terms of biomes or habitats, species, and genetic resources.

On the other hand, regional development is very heterogeneous in Brazil, as is income distribution. The technologically developed South and Southeast show the best social and economic figures in the country, but their natural resources have been severely damaged. Natural ecosystems have been very fragmented due to human activities and the few remnants are scattered, being very small in area, some of them protected as nature preserves. The North presents an opposite situation, sheltering the exuberant Amazonian forest but, together with the Northeast, exhibiting bad social indices. Also in the North, the original forest has gradually been giving place to pastures and agriculture. The Central region shows an intermediate situation, but the advance of grain cultures and pasture plantations has been fast displacing the natural environments.

Besides regional difficulties, the lack of planning for the use of land and of natural resources or the lack of integration among organizational sectors are usually verified in the whole country, resulting in inadequate resource utilization and distribution, or in a poor establishment of priorities, which sometimes aggravates the regional contrasts.

Facing this panorama, the country urgently needs a strong policy for integrated regional land use planning, as well as for rational use of its natural resources, particularly to keep the still existing biodiversity. On one side, the poor people need essential goods for living and, having few options and no environmental awareness, overexploit the natural resources in the rural areas or aggravate an already adverse environmental condition in the increasingly swelling urban centres. On the other side, big enterprises also promote destruction, although nowadays they are under greater control, due to restrictive environmental legislation. The large environmental catastrophes, such as vegetation downfall or burning, pollution and habitat destruction follow, in a great extent, the lack of an effective integrated national land use plan and weak legislation control.

The landscape approach in Brazilian science started with biogeographers, around 1960, and remained restricted to a few research groups for 2-3 decades. At that time, environmental problems, such as resource scarcity, pollution, and overpopulation, started to worry the world and a more global conception of problems was required. A new science, linking physiographical, biological and social issues gradually started to gain force and, by the 80s, landscape ecology was flourishing worldwide, feeding scientific undertaking in Brazil.

In 1986, significant additions occurred in Brazilian environmental legislation, when environmental impact assessment studies became compulsory for every civil project considered potentially damaging to the environment, taking as an example the most developed countries. Although the multidisciplinary approach had been used in many proposals before that time, the emphasis had been restricted mainly to the economic and technological aspects, often not including social and environmental questions. With the new legislation, the exercise of multi and transdisciplinarity became much more effective and environmental consciousness started to grow among all social sectors.

Landscape ecology revealed itself to be an excellent tool for promoting the required integration towards practical applications, and this science, which had been restricted to the academic world, became more widespread since then, coming out of the research centres to practical life, especially at the regional planning scale.

Today, in Brazil, approaches at the level of landscape are rooted in several domains, being found in regional and local planning design, environmental conservation and natural resource management projects. Academically, landscape ecology concepts can be found implicit in many areas of science, however, only a few research centres are particularly dedicated to the subject.

Table 1 and figure 2 present some of the research centres where people deal with landscape ecology, mostly in the fields of ecology, geography and architecture. Most of these centres are located in SE Brazil. It is important to emphasize that the list presented is not complete and, at the moment, a more complete survey is being carried out to achieve a more detailed inventory. The main sources of the presented data combined data banks of governmental agencies, universities, and the authors’ personal knowledge.

Tab. 1 Some research groups which deal with landscape ecology approach in Brazil (preliminary information). The numbers in the first column are related to figure 2.

UNESP (campus Pres. Prudente) = Universidade Estadual de São Paulo, Presidente Prudente, SP

EMBRAPA/ CNPS = Empresa Brasileira de Pesquisa Agropecuária, Rio de Janeiro, RJ

The development of the State of São Paulo in the second half of the 19th and the beginning of the 20th century was characterized by a fast expansion of coffee plantations and an inversely high decrease in natural forested areas. Nowadays, only 7% of the State is forested while at the beginning of the 19th century, the figure was around 80% (Fundação SOS Mata Atlântica & INPE 1992). The riparian area was also affected by this deforestation process and is still now under severe pressure from the expansion of crops, as sugar cane and orange. The destruction of this land/inland water ecotone results in a steep increase in soil erosion, with the obvious consequence of silting up rivers, lakes and water reservoirs, reducing the area occupied by natural vegetation and, in consequence, reducing the regional fauna and flora biodiversity (Joly 1994). In 1987 alone, São Paulo State lost 200 million tons of soil due to erosion, resulting in increased flood frequency and disruption of the regional economy (IAC 1993).

This catastrophic situation induced the creation, in 1987, of the "Intermunicipal Consortium for the Preservation of the Watershed of the Jacaré-Pepira River", uniting the mayors of the 13 municipalities of the Jacaré-Pepira watershed. The Jacaré-Pepira is a sixth order river (according to Strahler 1957) in the Paraná watershed, and drains about 2500 km² (Fig. 3). It was the first time in Brazil that local governments decided to unite efforts toward the preservation and management of their natural resources. This Consortium has had, since its creation, the support of the environmental agency of the State of São Paulo (CETESB) and of scientists from the State University of Campinas (UNICAMP), who were asked to help in establishing a model for the recovery of degraded riparian forest areas. This research project was developed within the Mab/Unesco scientific program "The role of land/inland water ecotone in landscape management and restoration" (Naiman et al. 1989). The final aim of the project was to promote the recovery of the Jacaré-Pepira riparian ecological functions, including soil and bank stability, nutrient and sediment trapping, biological filter against agricultural pollution, habitat for species conservation and regeneration, and as a corridor for species dissemination (Joly 1994).

Fig. 3 Geographical location of Jacaré-Pepira watershed and detail of the study area.

A floristic survey of the remnant riparian forests in the study area started in 1986, under the coordination of Dr. Carlos A. Joly, as well as seed collection for seedling production. In January 1987, the recovering of degraded areas started by planting experimental plots in private properties, authorized and supported by the owners. In this first step, seedlings were produced in the UNICAMP, but by February 1988, seedlings were being produced in a plant nursery built in the Jacaré-Pepira watershed, with the financial support of the Ministry of the Environment. By the end of that year, seedling production of 125 native species reached 100,00 seedlings/year. The first results of experimental plots showed that a considerable amount of information was needed to set up a model of riparian forest rehabilitation, especially: riparian forest floristic composition and structure; population dynamics of the most important species; water, light and nutrient requirements of seedlings; flooding tolerance; and interaction between fauna and flora (pollination and dispersion processes, for example). Thereafter, a scientific project was designed to allow a synchronic development of applied and basic research. Most of the scientific knowledge was obtained by studying well preserved riparian forest fragments. These studies included: floristic and phytosociological "relevées" of two riparian fragments (Kotchetkoff-Henriques & Joly 1994, Salino 1993, Salis et al. 1994), demography of some key riparian species, typical of waterlogged areas, such as Inga affinis DC (Lieberg 1992), Calophyllum brasiliense Camb. (Marques 1993), Talauma ovata St.Hil. (Lobo 1993) and Geonoma schottiana Mart. (Mertz, unpublished data); demography and ecophysiology of key secondary species of well drained areas (Centrolobium tomentosum Guillem. ex Benth.Hook.) (Aidar 1992); faunistic surveys (Barrela 1989 for fishes, Haddad unpublished for amphibian, Almeida 1997 for birds); and water and nutrient balance (Novoa & Silva unpublished). For the first time in Brazil such a wide range of information on auto and synecology of native populations and communities was used in a riparian rehabilitation project.

However, most of these studies were undertaken in only one or two fragments and one can contest the use of these data for the establishment of a general model for riparian rehabilitation on the whole watershed, which shows a wide range of environmental conditions and degrees of human pressure. To overcome part of this limitation, the landscape ecology approach was undertaken.

Introduction The landscape approach was stated to answer two questions: (1) to what extent is the floristic composition and diversity of the Jacaré-Pepira riparian forest fragments affected by the neighbouring landscape structure (i.e., by the landscape composition and spatial configuration)?; and (2) how may this landscape approach help in the choice of a riparian rehabilitation and conservation strategy?

Methods In order to answer these two questions we analyzed 15 riparian forest fragments situated in a 62 km reach of the Jacaré-Pepira river, between 48°08’W, 22°21’S and 48°30’W, 22°00’S (Fig. 3). The regional climate is classified as Cwa (Köppen 1948), i.e. moist sub-tropical climate with dry winter and wet summer. The mean annual temperature varies between 21 to 23°C and average annual rainfall goes from 1300 to 1560 mm. The study reach is characterized by a relatively deep valley with a very rapid river flow, reduced floodplain, large sandstone formations (Bistrichi et al. 1981) and predominantly poor soils (Typic Quartzipsamment, Almeida et al. 1981). This reach was chosen primarily because of its homogeneous hydro-geomorphology, in order to minimize the effects of some abiotic variables in floristic composition and species diversity. Secondly, the reach was chosen for its heterogeneous landscape structure, in order to test the influence of that heterogeneity on tree species composition and diversity.

The fragments were chosen in four different landscape types (Tab. 2) in order to take into account all the landscape heterogeneity of the study site. These landscapes were defined by the landuse/cover and by forest unit spatial arrangement (e.g., degree of fragmentation and connectivity). In these landscapes, forests appear as remnant patches of mesophytic submontane semideciduous forest (Veloso et al. 1991), and as corridors of deciduous forest, in the low tableland slopes, or seasonally flooded forest, in the riparian areas. The size of riparian studied fragments ranged from about 1 to 70 ha. These fragments included areas of seasonally flooded forest in strips of 10 to 30 m adjacent to the river but most of the fragment area presented well drained forest, i.e. mesophytic submontane semideciduous forest.

The point-centred-quarter method (Cottam & Curtis 1956) was used for the floristic survey of trees with diameter at breast height (1.3 m) greater than 3 cm. The sampling design was undertaken in each fragment to account for the whole environmental gradient orthogonal to the river, as well as to obtain the same floristic sufficiency. This sufficiency was considered to be reached when a duplication of the number of trees added less than 20% of new species.

Landscape structure was quantified by digital analysis of a classified Landsat Thematic Mapper image. In a supervised K-Nearest-Neighbour classification, ten landuse/cover units were distinguished: urban areas, water, wet meadows, dry meadows, culture, reforested areas, scrubs and savannahs, riparian forests, semideciduous forests and deciduous forests. Overall accuracy and landuse/cover unit accuracies were higher than 93% (Metzger 1997). From the classified image, we analyzed parameters (or features) of landscape at the level of: (1) the studied fragments, (2) the semideciduous forest (i.e., the main habitat of the studied tree species) and (3) the overall landscape, taking into account all of the 10 landuse/cover units. The structure of the fragments was characterized by fragment size, shape (simple boundary proportion), boundary complexity (Metzger & Muller 1996), isolation (Metzger 1997) and connectivity (Metzger 1997). Semideciduous forest structure was characterized by 5 parameters: forest proportion, boundary complexity (Metzger 1997), fragmentation (Li et al. 1993, Zipperer 1993, Kleinn et al. 1993), isolation (Metzger 1997) and connectivity (Metzger 1997). Overall landscape structure was characterized by landscape diversity (O'Neill et al. 1988, Turner et al. 1989) and boundary complexity (Metzger 1997). More details about the indices used are presented in Metzger (1995). Overall landscape and forest parameters were evaluated in the neighbourhood of the studied fragments for three environment sizes (2 by 2, 4 by 4 and 8 by 8 km) in order to take into account different species' dispersal capabilities. The scale of analysis is so defined by a TM image resolution (30 by 30 metres) and by the three landscape extents tested.

The relationships involving landscape structure indices and tree diversity were quantified by Pearson (linear relations) and Spearman (non-parametric relations) simple correlation coefficients. These correlations were evaluated separately for each environment.

Results and discussion The results obtained showed high tree species richness. From 5,472 trees sampled, 222 species and 56 families were identified (Metzger 1995). Species richness and diversity between fragments also varied greatly (Tab. 3). In fact, despite the distribution of studied fragments in a short and relatively homogeneous river reach, the differences observed in species diversity were as important as those found in mesophytic semideciduous forests over the whole of São Paulo State (i.e., 21 = S = 113, 2.32 = H = 4.36, 0.76 = J = 0.98).

Tab. 3 Total species richness (S), diversity (H) and evenness (J), of the 15 studied forest fragments. Species diversity and evenness were computed using the Shannon-Wiener function based on natural logarithms (Ludwig & Reynolds 1988).


Fragment (1)	River influence (2)	S	H	J


F11	week	65	3.09	0.74
F12	week	48	3.31	0.86
F13	moderate	56	3.48	0.87
F14	moderate	55	3.37	0.84
F15	important	14	1.59	0.60
F16	week	74	3.86	0.90
F21	week	111	4.32	0.92
F22	moderate	80	3.87	0.88
F23	moderate	52	3.52	0.89
F31	week	89	3.61	0.80
F32	week	68	3.58	0.85
F33	important	40	2.87	0.78
F41	week	69	3.25	0.77
F42	week	55	3.08	0.77
F43	important	32	2.81	0.81

(1) Fragments are presented by landscape type and decreasing area (for example, F21 is the largest fragment from landscape type 2, F43 is the smallest fragment from landscape type 4).

(2) A weak influence corresponds to forest areas flooded every 5 years or more; moderate influence corresponds to forest areas flooded every one or two years; important influence means forest areas flooded at least one month per year.

Data analysis showed that this variation between fragments seems to be explained by two factors: the extent of river influences and the effect of landscape structure on fragments. The influence of the river is revealed by a Factorial Correspondence Analysis (FCA, Benzecri 1973), using the species abundance in the 15 studied fragments. The first FCA axis distinguished the fragments most disturbed by river floods (F15, F33 and F43) from the largest and driest ones (F11 and F31) (Fig. 4). In fact, the most frequently flooded fragments were clearly poorer and less diversified than the others. Tree composition in these fragments was always, in every landscape type, dominated by a small number of pioneer species (particularly Inga affinis, Croton urucurana and Mimosa bimucronata ), highly adapted to hydromorphic conditions. In the forests less frequently disturbed by the river (presenting a typical semideciduous structure), soils were not saturated by water, the ground water table was not superficial and species did not need to be adapted to hydromorphic conditions. The soils were dry and deep and, therefore, favourable to a great number of species which require well drained areas.

Fig. 4 Ordination of the 15 studied fragments on the first two FCA axes. The classification shows floristic affinity between fragments according to the K'Means method.

The effects of landscape structure were attested by correlation results between landscape structure and fragment diversity measures (Tab. 4). In this analysis we used only the semideciduous forest areas thus, only 11 fragments were considered. Boundary complexity (particularly the proportion of coverts, i.e. the points where three or more landuse/cover units converge) and forest connectivity (particularly the density of corridors and stepping stones) appeared as the main parameters of landscape structure explaining tree diversity variation. In a regression analysis, the proportion of coverts and the density of corridors and stepping stones explained over 80% of tree species diversity variation (Metzger 1997). The analysis of these correlations by functional groups, according to their dispersal capacity (anemochorous and zoochorous species vs barochorous species) and shade tolerance (pioneer species vs climax species), showed that almost all functional groups presented high correlation with landscape structure parameters. The nature of these relationships changed according to the group and to the landscape extent considered. Only the pioneer species always presented weak correlations. The most important correlations were obtained between: fragment connectivity parameters and climax and zoochorous species richness; forest fragmentation parameter and climax species richness; forest isolation and boundary complexity parameters and barochorous species richness. As expected, species with low dispersal capacity (e.g., barochorous species) were the most related to the landscape structure in the closest environment, while species with high dispersal capacity (e.g., zoochorous and anemochorous) were the most related to landscape structure in the larger environment. Results showed also that the youngest trees (10 years old or less) were more sensitive to landscape structure parameters when compared to the oldest trees, attesting the influence of the present landscape structure on the establishment of new trees.

Tab. 4 Simple correlation coefficients between landscape parameters and the functional groups richness in eleven fragments and landscapes (Pioneer, Secondary, Climax, Anemochorous, Zoochorous and Barochorous). When the correlation is not significant, only the sign of relation is indicated.

The analysis of floristic pattern at the landscape scale clearly demonstrated the influence of river floods and landscape structure parameters on riparian forest composition and diversity. But the question was: how these results may help in the establishment of a model for riparian rehabilitation ?

The landscape approach allowed to suggest several recommendations for the riparian conservation and rehabilitation of the Jacaré-Pepira river and so for the sustainable development of this watershed.

First, the results showed that even a homogeneous hydro-geomorphological river reach may present high variation of tree species composition among fragments (in the well drained areas). These variations must be considered in the choice of species for rehabilitation. In the first steps for a recovery program, preference must be given to pioneer or secondary species with wide distribution patterns or to species frequently found in the recovery plot environment. Preference must also be given to fast-growing species. Among the 222 species identified, 6 presented high values for abundance, frequency and growth and are indicated to forest recovery in the Jacaré-Pepira watershed: Cariniana estrellensis (Raddi) O. Ktze.Glas., Centrolobium tomentosum Guill. ex Benth., Enterolobium contortisiliquum (Vell.) Morong., Luehea divaricata Mart., Syagrus romanzoffiana (Cham.), Croton floribundus Spreng.

Second, our results clearly showed that the semideciduous forest is richer and more diversified than the seasonally flooded forest. Narrow corridors (20 to 50 m) presented only 40% of the total species found. 45% of species were found only in the largest fragments, more than 100 m wide. If the riparian conservation and rehabilitation project intends to maintain regional biodiversity, it is necessary to maintain (or restore) forest riparian corridors wide enough to include areas of semideciduous forest. This drier forest located along rivers may act as a dispersal corridor for semideciduous forest species and may play a key role in the maintenance of species diversity in highly fragmented landscapes.

Third, the most important parameters of landscape structure linked to functional group richness and to total diversity were boundary complexity and fragment connectivity. Comparing to these two parameters, other ones commonly used in previous studies, such as fragment area and isolation, presented weaker correlations with the species richness variations. So, the rehabilitation of riparian corridors must consider the landscape context where the recovery plots will be established. Special importance must be given to forest connectivity: even small fragments on highly connected landscapes may present high species diversity. One may suppose that a trial plot on a highly connected landscape may regenerate more rapidly than a plot situated in poorly connected ones. Or inversely, the rehabilitation of riparian forest corridors may be a tool for the increase of local connectivity, favouring the maintenance (or even the increase) of species diversity in the forest fragments present in the surroundings. The landscape permeability also seems to be an important factor, as attested by the high correlation between the proportion of scrub and savannah and the richness of zoochorous species (Tab.4). High forest connectivity or landscape permeability may favour flows of seeds, pollens or animals across the landscape and so, favour the rate of species immigration in the recovery plots (Metzger and Décamps 1997). It is important to note that zoochorous and climax species were highly sensitive to landscape structure parameters, particularly to connectivity indices. The particular sensitivity of the zoochorous species (that represent 60% of the total species found) to connectivity leads to develop tree species conservation strategies that take into account the needs of dispersal species. Whenever we create mechanisms to facilitate the dispersal of animals, we optimize the renewal possibilities of zoochorous fragmented populations. Climax species are considered the most sensitive to habitat fragmentation and thus, need the greatest conservation attention (e.g., Terborgh 1974). Therefore, the rehabilitation of riparian corridors needs not only forest management but also the management of the landscape matrix, by creating, for example, corridors and stepping stones. According to our results, connectivity is an important component of landscape structure to consider when establishing a conservation policy.

The landscape approach also showed that the large number of species found in the riparian forest remnants, the complex pattern of floristic variation among areas and the widely varying degrees of disturbance by human activities make almost impossible to establish a unique model for the rehabilitation of degraded areas (Joly 1994). Rehabilitation actions must be adapted to the local environment and landscape conditions.

The contribution of the landscape approach for the conservation and rehabilitation of riparian areas must be interpreted within the scope and limitation of our investigation. The main limitation of this investigation was the use of a single-year pattern data. It may be argued that the correlation patterns observed, as the strong and positive relationship between connectivity and climax or zoochorous species richness, may be due to coincidence. Only on a long term basis will it be possible to determine causal relationships, supposing we can define the time-lag between landscape structure changes and the community adaptation to these changes (Opdam 1991, Ligon & Stacey 1996). Perhaps the most important limitation (and interest) of landscape ecology is to deal with highly heterogeneous and complex systems. In such a system, it is not possible to reduce the understanding of pattern structure or the effect of pattern over ecological function to a small set of causal relations. A good understanding of pattern/effect relationships, that will lead to robust management guidelines, will require long-term studies. That was not the case with the Jacaré-Pepira project. Unfortunately, the project failed in 1993, mainly due to lack of financial support and to repeated changes in the local governmental structure. Therefore, it was not possible to test or to apply the observed relationships in the rehabilitation sub-project.

What have we learnt with the successes and failures of the landscape approach adopted in Jacaré-Pepira project? This approach allows us to make some recommendations for the rehabilitation of riparian forest corridors in other areas or countries.

(1) Riparian forest recovery is a very expensive action: US $ 2,600 for the recovery of one hectare of riparian forest in the Jacaré-Pepira watershed, not including handling charges (Joly 1994). In undeveloped countries, local government may not be able to afford to spend such a large sum on a recovery project. A cheaper option is to stimulate the natural regeneration processes. The results of our landscape approach may help in this task. Regeneration may be accelerated by creating landscape structures that favour the movement across the area. For example, the establishment, protection or expansion of little fragments or even of isolated trees in pasturelands may favour the movement of dispersal species (birds, small mammals, etc.). Little fragments or even narrow corridors may act as stepping stones and must be protected against cattle pressure. The enrichment of these stepping stones by planting of zoochorous species in their boundaries may be considered as a practical way to favour flows of pollens, seeds and animals, and to stimulate forest regeneration.

(2) The maintenance of biodiversity (i.e., of species diversity) in tropical fragmented landscapes, composed by a mosaic of agricultural and semi-natural patches, does not depend exclusively on the creation of habitat reserves. Landscape analysis showed that the spatial arrangement of forest fragments as well as the landscape mosaic complexity may play a key role in species dispersal possibilities. On agricultural landscapes, the riparian forest network may be a key element for landscape connectivity, as the European or North American hedgerow networks (Forman & Baudry 1984, Burel & Baudry 1989). Thus, a management strategy for biodiversity conservation should consider, at watershed level: i) the conservation of large fragments that may act as species sources; ii) the conservation or rehabilitation of riparian forest corridors (wide enough to include areas of drier forests) and stepping stones among large fragments. The choice of the establishment of a riparian recovery plot should consider the linkage between large fragments as a strategy of regional biodiversity maintenance.

(3) The main limiting factor of such a rehabilitation project may be political. A project with this aim needs support of several segments of society, including the local population (e.g., land owners), local government, State agencies (such as the State Secretary of Environment) and Federal agencies (as the Ministry of Science and Technology or the Ministry of Environment). If one of them fails, all the project fails, as in the Jacaré-Pepira case. To succeed in such a long term applied research and to observe economical results, stable political, human and financial structures are needed. It appears that a good option would be the adoption of something like the French watershed management model ("Basin Agencies") in Brazil. These watershed agencies perceive taxes from water users and apply the money in local water resources conservation. These agencies may act as a stable financial and technical support for long-term applied researches.

(4) A major contribution of the Jacaré-Pepira project was to show the importance on considering the watershed as the working unit in landscape planning. The political process for the establishment of Jacaré-Pepira Consortium can be considered a turning point of watershed management policies in Brazil. It was the first time that all local governments of a watershed decided to put aside political differences to unite efforts towards water resources preservation. As a consequence of this successful conception more than ten other consortia were established from 1987 to 1992 in São Paulo, Goias, Mato Grosso do Sul, Paraná, Rio de Janeiro and Santa Catarina States (Joly 1994).

The major insight of landscape ecology is to perceive ecological and human interactions at the scale of management planning, i.e., the scale of a set of interactive ecosystems. But the understanding of such a heterogeneous and complex systems is also hard and demands a great amount of information to achieve good results. Today, the Brazilian landscape ecologist still does not have a robust background to support rehabilitation actions. For example, we still do not know the value of corridors for a large number of species, or the importance of landscape permeability, boundary complexity or landscape heterogeneity on the population and community dynamics, even because basic aspects of the biology of most species are still lacking. Besides, the huge biodiversity in the tropics makes this task much more difficult. On the other hand, people working at the regional planning sector are becoming aware of the importance of having corridors and buffer zones and of maintaining diversity at the different levels. The first step to apply landscape ecology to land management is to develop comparative long term studies, including a large set of landscape and community types and integrating socio-economic and ecological approaches. A strong knowledge at landscape ecology level would constitute a trustworthy and powerful tool for sustainable development.

Acknowledgements We wish to thank Marcelo Hiromiti Matsumoto for the illustrations and Daniel Dario Cavana for translating the abstract to Spanish.

Ab'Saber, A.N. 1967. Domínios morfoclimáticos e províncias fitogeográficas do Brasil. Bol. USP-IGEOG (orientação 3): 45-48.

Aidar, M. 1992. Ecologia do Arariba (Centrolobium tomentosum Guill ecos Benth. - Fabaceae) e o ecótono mata ciliar do rio Jacaré-Pepira, SP. Master thesis, Universidade de Campinas, Campinas.

Almeida, M.E.C. 1997. Estrutura de comunidades de aves em dois remanescentes florestais na bacia do rio Jacaré-Pepira, SP. Master thesis, Universidade Federal de São Carlos, SP, Brazil.

Almeida, C.L.F., Oliveira, J.B. & Prado, H. 1981. Levantamento pedológico semi-detalhado do Estado de São Paulo, escala 1 : 100,000. Quadrícula de Brotas. Convênio EMBRAPA-Instituto Agronômico de Campinas, Campinas.

Barrela, W. 1989. Estrutura da communidade de peixes da bacia do Rio Jacaré-Pepira (SP) em diferentes biótopos. Master thesis, Universidade de Campinas, Campinas.

Benzecri, J.P. 1973. L'analyse des données, tome 1: la taxinomie. Ed. Dunod, Paris.

Bistrichi, C.A., Carneiro, C.D.R., Dantas, A.S.L., Ponçano, W.L., Campanha, G.A. da C., Nagata, N., Almeida, M.A. de, Stein, D.P., Melo, M.S. de & Cremonini, O.A. 1981. Mapa geológico do Estado de São Paulo. Escala 1 : 500,000. IPT Monografias n°6, Volume II. Mapas.

Burel, F. & Baudry, J. 1989. Hedgerow network patterns and processes in France. In Zonneveld, I.S. & Forman, R.T.T. (eds.). Changing Landscapes. An ecological perspective. Springer Verlag, New York. p. 99-120.

Cottam, G. & Curtis, J.T. 1956. The use of distance measures in phytosociological sampling. Ecology 37: 451-460.

Falkenmark, M. & Suprapto, R.A. 1992. Population-landscape interactions in development: a water perspective to environmental sustainability. Ambio 21: 31-36.

Forman, R.T.T. 1995. Land mosaics - The ecology of landscapes and regions. Cambridge Univ. Press, Great Britain.

Forman, R.T.T. & Baudry, J. 1984. Hedgerows and hedgerow networks in landscape ecology. Environ. Manag. 8: 495-510.

Fundação SOS Mata Atlântica / Instituto Nacional de Pesquisas Espaciais. 1992. Atlas da evolução dos remanescentes florestais e ecossistemas associados do domínio da mata atlântica no Estado de São Paulo no período 1985-1990. Fundação SOS Mata Atlântica / Instituto Nacional de Pesquisas Espaciais. São Paulo, SP.

Goodland, R. 1995. The concept of environmental sustainability. Annual Review of Ecology and Systematics 26: 1-24.

Instituto Agronômico de Campinas (IAC) 1993. Seção de conservação dos solos: meio século de pesquisas. O Agronômico 45: 1-6.

Joly, C.A. 1994. Biodiversity of the gallery forest and its role in soil stability in the Jacaré-Pepira watershed, State of São Paulo, Brazil. In Jenzen, A.E. (ed.). Ecotones at the river basin scale- Global land/water interactions. South Australia Departement of Environment and Natural Resources, Adelaide, Australia. p. 40-54.

Kleinn, C., Dees, M. & Pelz, D.R. 1993. Sampling aspects in the TREES project: global inventory of tropical forests Final Report to the Joint Research Center, Freiburg Universität, Germany.

Kotchetkoff-Henriques, O. & Joly, C.A. 1994. Estudo florístico e fitossociológico de uma mata mesófila semidecídua na cabeceira do rio Cachoeira, Itirapina, SP. Revista Brasileira de Biologia 54: 477-487.

Li, H., Franklin, J.F., Swanson, F.J. & Spies, T.A. 1993. Developing alternative forest cutting patterns: a simulation approach. Landscape Ecology 8: 63-75.

Lieberg, S.A. 1992. Tolerância a inundação e aspectos demográficos de Inga affinis DC. Master thesis, Universidade de Campinas, Campinas.

Ligon, J.D. & Stacey, P.B. 1996. Land use, lag times and the detection of demographic change: the case of the Acorn Woodpecker. Conservation Biology 10: 840-846.

Lobo, P.C. 1993. Tolerância a inundação de Talauma ovata St. Hil. e aspectos de seu comportamento em uma mata ciliar da bacia do rio Jacaré-Pepira, Brotas, SP. Master thesis, Universidade de Campinas, Campinas.

Ludwig, J.A. & Reynolds, J.F. 1988. Statistical ecology. A primer on methods and computing. Wiley & Sons, New York.

Metzger, J.P. 1995. Structure du paysage et diversité des peuplements ligneux fragmentés du rio Jacaré-Pepira (Sud-Est du Brésil). Ph.D. thesis, Université Paul Sabatier, Toulouse, France.

Metzger, J.P. 1997. Relationships between landscape structure and tree species diversity in tropical forests of South-East Brazil. Landscape and Urban Planning (in press)

Metzger, J.P. & Muller, E. 1996. Characterizing the complexity of landscape boundaries by remote sensing. Landscape Ecology 11: 65-77.

Metzger, J.P. & Décamps, H. 1997. The structural connectivity threshold: an hypothesis in conservation biology at the landscape scale. Acta Œcologica 18: (in press)

Naiman, R., Décamps, H. & Fournier, F. 1989. Role of land/inland water ecotones in landscape management and restoration. Proposals for collaborative research. Mab Digest n°4, UNESCO, Paris.

Naveh, Z. 1995. Interactions of landscapes and cultures. Landscape and Urban Planning 32: 43-54.

O'Neill, R.V., Krummel, J.R., Gardner, R.H., Sugihara, G., Jackson, B., DeAngelis, D.L., Milne, B.T., Turner, M.G., Zigmunt, B., Christensen, S.W., Dale, V.H. & Graham, R.L. 1988. Indices of landscape pattern. Landscape Ecology 1: 153-162.

Opdam, P. 1991. Metapopulation theory and habitat fragmentation: a review of holarctic breeding bird studies. Landscape Ecology 5: 93-106.

Salino, A. 1993. Flora pteridofítica das matas ciliares da bacia do rio Jacaré-Pepira, Estado de São Paulo, Brasil. Master thesis, Universidade de Campinas, Campinas.

Salis, S.M., Tamashiro, J.Y. & Joly, C.A. 1994. Florística e fitossociologia do estrato arbóreo de um remanescente de mata ciliar do rio Jacaré-Pepira, Brotas, SP. Revista Brasileira de Botânica 17: 93-103.

Sharitz, R.R., Boring, L.R., Van Lear, D.H. & Pinder, J.E. 1992. Integrating ecological concepts with natural resource management of Southern forests. Ecological Applications 2: 226-237.

Strahler, A.M. 1957. Quantitative analysis of watershed geomorphology. Transactions, American Geophysical Union 38: 913-920.

Terborgh, J. 1974. Preservation of natural diversity: the problem of extinction prone species. Bioscience 24: 715-722.

Turner, M.G., O'Neill, R.V., Gardner, R.H. & Milne, B.T. 1989. Effects of changing spatial scale on the analysis of landscape pattern. Landscape Ecology 3: 153-162.

Veloso, H.P., Rangel Filho, A.L.R. & Lima, J.C.A. 1991. Classificação da vegetação brasileira adaptada a um sistema universal. IBGE, Rio de Janeiro.

Zipperer, W.C. 1993. Deforestation patterns and their effects on forest patches. Landscape Ecology 8: 177-184.

Local	Institution	Research Field	Tools					Spatial domain
			GIS	aero- photo	satel. image	indices	models
1	USP/ BiosciencesInstitute Environ. Analysis & Monitoring Lab	biol. conservation, environ. planning, ecosystem ecology	X	X	X	X	X	cerrado,Atlantic & Amazonian forests
1	USP/ Architecture & Urbanism Fac. Design Department	urban and regional planning						SE Brazil
1	USP Geography Dept.	geomorphology, environ. planning, resouces management		X				SE Brazil
2	USP - ESALQ Forestry Dept.	biol. conservation						cerrado & SE forests
3	UFSCar/ PPGERN (Ecology & Nat. Resources Program)	environ. management	X	X	X	X	X	cerrado & SP forests
4	UNESP (campus Rio Claro) Inst. of Geosciences (IGCE)	environ. planning, landscape structure	X	X	X	X		SE Brazil
4	UNESP (campus Rio Claro) Ecology Dept.	biological conservation, forestry						rainforests in Brazil
5	UNESP (campus Pres. Prudente) Design Dept.	urban and regional planning	X	X				urban areas in Brazil
6	UNIVAP Social Sciences Dept.	environmental planning	X					wetlands in SE Brazil
6	INPE Remote sensing Dept.	environ. monitoring,, biol. conservation	X	X	X		X	Amazonia, SE Brazil
7	UNICAMP/ Civil Engineer Inst. Sanitation and Environment Dept.	environmental planning	X	X	X		X	SE Brazil
7	EMBRAPA/ NMA (Environ. Monitoring Nucleous)	agroecological and regional planning	X	X	X			agroecosystems
8	UFPR Forestry School	biol. conservation, urban planning						S Brazil
9	UEL	biol. conservation		X		X		S Brazil
10	UEPG Soil Sciences Dept.	biol. conservation, environ. planning		X	X			PR State
11	UFRGS Landscape Ecology Lab	environ. planning, pollution	X		X		X	S Brazil
11	UFRGS Ecology and Palinology Lab	plant communities, pollution, palinology	X				X	S Brazil
11	FURG Systems Ecology Dept.	environ. planning, sustainability					X	coastal zones in Brazil
11	FURG Oceanography Dept.	systems ecology			X			coastal zones in Brazil
12	UFRJ Architecture Dept.	urban and regional planning						urban areas in Brazil
12	UFRJ Geography Dept.	environmental planning	X	X	X	X		SE Brazil
12	UFF Geography Dept.	environmental planning						Atlantic forest in RJ State
12	UERJ Ecology Sector	fauna and water resource conservation	X					Atlantic forest
12	EMBRAPA/ CNPS (National Centre for Soil Research)	environ. planning/ biol. conservation						Atlantic region
13	UFMG Ecology Dept.	mammal conservation						Central Brazil
13	UFMG Architecture School	urban and regional planning						urban areas
14	UFJF Architecture Dept.	urban and regional landscapes						urban areas in Brazil
15	UFU Documentation and Socio Economic Res. Centre	environ. planning, natural resources economy	X		X			cerrado & agroecosystems
16	UFV Forestry Dept.	biol. conservation	X	X				Atlantic forest
17	UFC Geography Dept.	environ. planning, ecol. management	X		X			coastal zones in Brazil
18	INPA Ecology Dept.	biol. conservation	X		X		X	Amazon region
19	EMBRAPA/ CPATU (Humid Tropic Research Centre)	sustainability	X		X			Amazon region
20	UnB Zoology Dept.	fauna conservation	X		X		X	Central Brazil
20	UnB Ecology Dept.	environmental and agricultal planning	X		X			cerrado & agroecosystems
21	PUC Goiás Ancient History Dept.	landscape structure					X	cerrado in central Brazil (Goisas)
22	UFMT Biosciences Institute	community ecology						Pantanal

Landscape type	Matrix (predominant unit)	Main patches	Spatial arrangement of forests	Riparian forest corridor
1	Dry meadow.	Eucalyptus plantations and sugar cane.	Small fragments (<40 ha) of semideciduous forest badly connected.	Narrow or non-existent.
2	Scrub, savannah and meadow.	Forest fragments.	Large fragments of semideciduous forest well connected.	Large, dense and regular.
3	Dry meadow and cultures (sugar cane and Citrus).	Forest fragments and corridors.	One large fragment of semideciduous forest and wide corridors of deciduous forest along the tableland slopes.	Irregular, generally quite large and not dense.
4	Dry meadow and sugar cane.	Eucalyptus plantations, forest fragments and savannah.	Small fragments of semideciduous forest and wide areas of riparian and deciduous forests.	Irregular, generally wide and dense.