HISTORICAL LAND-COVER/USE IN DIFFERENT SLOPE AND RIPARIAN BUFFER ZONES IN WATERSHEDS OF THE STATE OF SÃO PAULO, BRAZIL

Information about the land cover of a region it is a key information for several purposes. This paper aimed to elaborate land-cover maps using digital satellite images obtained in 1997 from seven watersheds (Piracicaba, Moji-Guaçu, Alto Paranapanema, Turvo Aguapeí, Peixe, and São José dos Dourados) located in the State of São Paulo, southeastern Brazil. Additionaly, this study evaluated the relationship between land-cover and slopes of the terrain of the seven watersheds. A third objective was to estimate the percentage of riparian vegetation currently remaining along the streams in a 30meter width buffer zone. Three research questions were posed: i) What is the dominant land-cover of these watersheds? ii) Is the riparian vegetation well preserved in the 30m width buffer zone? If not, iii) what is the dominant land-cover in these areas and what would be the cost of recovering such areas? Pasture was the predominant land-cover, occurring in approximately 50% of the entire study area, while sugar cane (Saccharum officinarum) (14%) constituted the second most frequent land-cover. Approximately 50% of the area of the seven basins is considered flat (40%) or smoothly rolling (10%). The terrain only becomes hillier in the Piracicaba and Alto Paranapanema basins, where a little less than 50% have slopes higher than 8%. The total riparian buffer strip zone occupied an area equivalent to approximately 6,200 km. From this total, only 25% is preserved. Pasture is the main land-cover of the riparian buffer strip zone.


INTRODUCTION
Through multiple land-use changes, biological systems support human needs regarding food, fiber and energy (Vitousek et al., 1997).Therefore, land-use changes are crucial to human well-being (Lambin et al., 2003).On the other hand, land-use changes often alter the structure and functioning of ecosystems by modifying the composition of plant and animal communities as well as nutrients, water cycles, and local climate (Webb et al., 2005;2006;Lambin et al., 2003;Daniel et al., 2002;Mander et al., 2000;Ojima et al., 1994, Houghton, 1994).In addition, several studies show that land use indices are strong predictors of river water quality (e.g.Ometto et al., 2000).Therefore, information on how land-cover is historically altered is essential for improved management of aquatic and terrestrial resources of a watershed.
The State of São Paulo, located in southeastern Brazil, is the most industrialized and populated region of Brazil, being responsible for 40% of the gross domestic product of the country.Some regions of the State of São Paulo face the same problems as other over populated and industrialized areas of the developing world: increasing deterioration of water resources coupled to decreased water availability (Martinelli et al., 2002;Groppo et al., 2006).In order to meet such challenges, at the beginning of the Nineties, the State was divided in twenty-one hydrologic units corresponding to one or more meso scale watersheds, decentralizing the water management of the State.Consequently, it is fundamental that a body of environmental, social and economic information be readily available to water managers of these watersheds.Currently, this detailed information on land-cover and land use of watersheds is lacking.
Based on this lack of information, the first objective of this paper was to describe the land-cover and land use in 1997 of seven meso scale watersheds located in the State of São Paulo and to investigate the relationships between the land-cover and the slopes of the terrains of the study area at the time.Since the area of remaining original forests is already mapped, the second objective of this study was to estimate the percentage of riparian vegetation currently remaining along the streams of the seven basins of this study in a 30 m-width buffer zone.
This study considers: i) what is the dominant land-cover of these watersheds?ii) is the riparian vegetation well preserved in the 30 m-width buffer zone?If not, iii) what is the dominant land-cover in these areas and what it cost to recover such areas?

Study area
The land-cover of seven watersheds located in the State of São Paulo, southeastern Brazil was mapped.They comprise 36% of the area of the State of São Paulo (Figure 1), which is the most industrialized and populated state of the country.In 2002, 22% of the Brazilian population lives in the State of São Paulo, which is responsible for 40% of the Brazilian Gross Domestic Product (IBGE, 2002).The land uses of the following watersheds were mapped: Piracicaba, Moji-Guaçu, Alto Paranapanema, Turvo Aguapeí, Peixe, and São José dos Dourados.Most of these watersheds are located entirely in the State of São Paulo.Exceptions are the Moji-Guaçu and the Piracicaba watersheds, of which approximately 10% is located in the State of Minas Gerais.These were selected due to their economic importance, as well as their importance as under-surveyed areas in relation to aquatic biodiversity.
The area of the basins varied from 5,800 km 2 (São José dos Dourados) to 22,804 km 2 (Alto Paranapanema).In addition to these extremes, the size of the basins varied from approximately 11,000 to 22,000 km 2 (Table 1).The average annual rainfall is similar in all of them, varying from a minimum of 1,220 mm per year in the Aguapeí basin up to a maximum of 1,460 mm per year in the Moji-Guaçu basin (Table 1).The smallest average discharge was observed for the São José dos Dourados River (52 m 3 s -1 ) and the largest for the Paranapanema River (310 m 3 s -1 ) (Table 1).The most populated basin is by far Piracicaba, with almost 4 million people, followed by the Moji-Guaçu basin with 1.3 million people (Table 1).All other basins have less than 1 million people, notabley the São José dos Dourados basin that has only 210,000 inhabitants (Table 1).The population density in the Piracicaba basin was also the largest: from 3 to 5 times larger than those of other basins, where the population density varied from 34 to 110 inhabitants per km 2 (Table 1).Most of the population is concentrated in urban centers; as a consequence, the urbanization rates are higher than 80-90%, with the exception of the Alto Paranapanema basin, where the rate is only 75% (Table 1).Finally, the water use for the Turvo and the Moji-Guaçu basins is unavailable.In most of the other basins, water is mainly used for irrigation (Table 1).An exception was in the Piracicaba basin, where water is mainly used by industries, followed by population (Table 1).
The classification accuracy was checked through a field survey.In June 2002, we visited 230 randomly selected points.In this visit, the land-cover around these points (in a radius of approximately 250 meters) was recorded, and the geographic coordinates were checked with a GPS receiver.The Kappa index of agreement was calculated using the "Errmat" command of the Idrisi (Eastman, 2001).The overall accuracy of the classification was 90% and the Kappa index of agreement was 0.85.
The land-cover in the map legends are: pasture, remnant forest, urban, perennial crops, annual crops, water bodies, bare soil, reforestation, sugar cane, and a general class called "other".The landcover classification was conducted for all watersheds, except for the Piracicaba.For this specific watershed, the land-cover map, which is already available, can be found at: www.cena.usp.br/piracena.This land-cover map (referring to land-cover situation of the region in 1997) was converted to a format equal to the other, and the step to generate the land use maps and date was also computed for this watershed.The land use maps were created by reclassification (command "Reclass" of the Idrisi) based on the land-cover maps.Pasture, perennial crops, annual crops, sugar cane, bare soil, and reforestation were reclassified into one class: "agrosilvipasture".The class "remnant forest" was reclassified as "conservation -relics", the class "urban" was reclassified as "urbanization", and finally the classes water bodies and "other" were classified as "water-bodies/other".

Digital Terrain Models and Slope Maps
The Digital Terrain Models (DEM) were purchased from Imagem Sensoriamento Remoto (www.img.com.br) for almost all watersheds, except the DEM for the Piracicaba River basin which was acquired from the Piracena project database (www.cena.usp.br/piracena/).All of these DEMs have a 20 m spatial resolution.The slope maps were gen-Table 1 -Basic characteristics of the seven watersheds investigated in this study.Source: São Paulo (1990).SJD -São José dos Dourados, AGUA -Aguapeí, PEI -Peixe, TUR -Turvo, ALPA -Alto Paranapanema, MOG -Moji-Guaçu, PIRA -Piracicaba.64, n.4, p.325-335, July/August 2007 erated by the Slope command of Idrisi (Eastman, 2001).After the generation of the slope maps for all watersheds, they were converted into a 30m spatial resolution throughout the command "resample" of the Idrisi, to permit cross-tabulation among the slope maps and land-cover maps.Before the cross-tabulation, the slope maps were reclassified into four slope categories (Santos et al., 2005): flat (<3%); smoothly rolling (3 to 8%); rolling (8 to 20%); and hilly (>20%).The slope and land-cover maps were cross-tabulated to generate thematic information showing the land-cover class according to slope class.

Riparian buffer zone
The development and analysis of riparian buffer zones required the implementation of several preliminary steps.First, a contemporary 1:50,000-land-cover map was generated from a Landsat TM scene.Second, stream network data (vector format) were extracted from the satellite images through the on-screen digitalization procedure.Third, the dimension of the buffer zone was determined according to the Brazilian Forest Code (www.ipef.br)and type of material available.Fourth, the buffer zones were defined using the stream network through the "Buffer" routine of the Idrisi 32 software (Eastman, 2001).Fifth, the maps containing the buffer zones and the land-cover were overlaid ("Overlay" routine of the Idrisi, with the option multiplication) in order to generate a new map containing the land-cover of the buffer zones.Finally, the areas of each land-cover class were calculated ("Area" routine of the Idrisi).
Additionally, the cost of the complete riparian vegetation restoration program was calculated for the detected areas with no riparian vegetation.The cost of the restoration per hectare was provided by Dr. Ricardo Ribeiro Rodrigues (Biological Sciences Department, Escola Superior de Agricultura "Luiz de Queiroz", University of São Paulo).The total cost of restoration is approximately R$ 7000/ha, which is equivalent to approximately U$ 3350/ha (dollar to real ratio 2.0885; February 23, 2007), and involves three main phases.The implementation phase, which is the most extensive, equivalent to 75% of the total cost of the project, and a first and a second year of maintenance, equivalent to 15% an 10% of the total cost of the project, respectively.The most costly restoration item is labor, which accounts for almost 47% of the cost.The cost involved in applying pesticides and fertilizers, including machinery used in these operations, is 36% of the total cost.Finally, the cost of seedlings is only 12% of the total cost.This estimate is a preliminary step in estimating the cost of restoration of riparian areas.It involves several assumptions such as extrapolating the budget shown above to the entire State of São Paulo, which may not be accurate.Additionally, our estimates do not account for economies of scale, which may accrue when large areas are reforested.
In 1999, the Brazilian Forest Legislation established that the riparian vegetation must be preserved in a: i) 30m width of each margin for streams up to 10 m width; ii) 50 m width of each margin for streams from 10 to 50 m width; iii) 100 m width for rivers from 50 to 200 m width; iv) 200 m width for rivers from 200 to 500 m width; and v) 500 m for rivers higher than 600 m width (www.ipef.br).In order to facilitate our estimates, we adopted a buffer width of 30m of each margin independent the river width.Therefore, our estimate is the minimum width required by law, which under estimates the area needing to be preserved.

Land-cover and Land Use Maps
An area approximately equivalent to 88,300 km 2 was mapped in this study, corresponding to almost 36% of the entire area of the State of São Paulo.Pasture is the main land-cover (approximately 50%) of the entire mapped area (Table 2).It is not the main land-cover only in the Moji-Guaçu basin, and in the central part of the Piracicaba basin (Figure 2).On the other hand, in some watersheds, such as the São José dos Dourados, the Aguapeí, and the Peixe, pasture is by far the main land use (Figure 2).
Sugar cane, one of the most cultivated and economically important crops in the State of São Paulo, covers approximately 14% of the study area.The Moji-Guaçu basin has the greatest area covered by sugar cane (39%), and the Piracicaba basin has the second greatest area (34%) (Table 2).This crop covers most parts of the Moji-Guaçu basin, while in the Piracicaba basin, sugar cane is more concentrated in the western portion of the basin (Figure 3), where the topography is gentler than in the eastern portion of the basin (www.cena.usp.br/piracena).In other watersheds, such as the Alto Paranapanema, sugar cane occupies less than 1% of the area and is clustered in the final portion of the basin (Table 2).The bare soil occupied approximately 7% of the mapped area and is associated to sugar cane in two periods: (i) immediately after harvesting or (ii) immediately before replanting the crop (Table 2).
The remnant of natural vegetation corresponds to 15.5% of the entire study area.This class of landcover includes patches of the Atlantic Forest biome, such as the Ombrophyllus Dense Forest, and the Semi Deciduous Mesophytic Forest, and several physiognomies of the Cerrado, a tropical savanna-like vegetation.The Alto Paranapanema has the highest area covered with natural vegetation (26%), which is mainly concentrated in the southern portion of this basin (Figure 4), while in São José dos Dourados only 7% of the area is covered with natural vegetation (Table 2).
Reforestation is composed of Eucalyptus sp. and Pinus sp.species farms (Kronka, 2002).Both occupy an area equivalent to 4% of the entire area, which is a little lower than the average value for the entire State of São Paulo (5%) (IEA web site, see references).Reforestation is important mainly in the Alto Paranapanema (11%), the Moji-Guaçu (4%), and the Piracicaba basins (4%).In the remaining watersheds, reforestation occupies less than 1% of the area (Table 2 and Figure 5).
The perennial crops cover approximately 4% of the entire mapped area (Table 2).Perennial crops are especially important in the watersheds located in    Other annual crops such as cotton (Gossypium sp.), zucchini (Curcubita sp.), and beans (Phaseolus sp.) were also observed, but in small proportion as compared to the first three crops.During the field visit, it was observed that corn is spread along several watersheds, while wheat and bean fields are mostly concentrated on the Alto Paranapanema watershed, which had the highest proportion of this type of land-cover (4%), while Peixe had the smallest proportion (0.13%).
Urban centers comprise approximately 2% of the entire study area.The watershed with the highest percentage of urbanization is Piracicaba (6%) due to a significant number of densely-populated cities such as Campinas, Rio Claro, Americana, Piracicaba, and Limeira.These cities are all located in the central part of the basin, and form a continuous urbanized area from Campinas to Piracicaba (Figure 7).The Piracicaba watershed is followed by Moji-Guaçu and Turvo that have almost 2% of urbanized area (Table 2).The basin that presents the smallest percentage of urban centers was Alto Paranapanema (0.4%).
The water bodies comprise 1.3% of the study area.The watershed with the largest body of water is Alto Paranapanema with the Jurumirim reservoir, a dam built on the Paranapanema River in the late 1960s.The Piracicaba basin has the second highest percentage of area covered by water, mainly due to the Barra Bonita reservoir.The class "other" is represented by patches that were not identified by the classifier and represents 0.3% of the study area.The Moji-Guaçu watershed presents the highest value for this landcover class, perhaps due to the natural complexity of land-cover along this basin.

Slope classes and land use
Approximately 40% of the area of the seven watersheds fall in a slope class considered as flat (<3%) (Figure 8).If the next slope class is considered (3-8%, smoothly rolling), more than 50% of all basins are either flat or smoothly rolling.This is especially true in the watersheds of western São Paulo State   -Turvo, São José dos Dourados, Aguapeí, and Peixe -where more than 70% of the area of these basins have slopes <8% (Figure 8).The terrain only becomes hillier in Piracicaba and Alto Paranapanema, where a little less than 50% of the area has slopes higher than 8% (Figure 8).Pasture is the main land-cover of the watersheds occurring in most slope classes, especially in Piracicaba, Moji-Guaçu, and Turvo, where pasture is also present oin higher slopes (Figure 9).Sugar cane occupies mainly the flat areas of the terrain in the Moji-Guaçu and Piracicaba watersheds (Figure 9).Finally, the remaining original vegetation occurs in different slope classes (Figure 9).However, a significant proportion of this vegetation is still preserved in high slopes in the Piracicaba basin, and especially in the Alto Paranapanema, with several state parks that encompass a large area of preserved Atlantic Forest.

Riparian buffer zone
The total riparian buffer strip zone (30 m of each stream side) is equal to 6,236 km 2 , which corresponds to approximately 7% of the sum of the area of the seven basins (88,322 km 2 ).From this total of riparian area approximately 4,646 km 2 , which corresponds to approximately 75%, need to be restored to comply with the law, i.e.only 25% of the total riparian area is indeed covered with riparian vegetation.This proportion varied among the basins (Figure 10).For instance, the riparian vegetation covered only 8% of the buffer strip zone in the São José dos Dourados watershed.In contrast, in the Alto Paranapanema basin approximately 40% of the buffer zone was still covered with riparian vegetation (Figure 10).In addition to these two extremes, in the Turvo and in the Peixe basins original riparian vegetation covers an area equivalent to 25 to 30% of the buffer zone (Figure 10).In the Aguapeí basin, this proportion is reduced to approximately 22%, and in the Piracicaba and the Mogi Guaçu, decreases even more to less than 20% (Figure 10).
Pasture is the main land-cover of the riparian buffer strip zone.This is especially true in the Turvo, the São José dos Dourados, the Aguapéi and Peixe basins.The only exception to this pattern is the Moji-Guaçu basin where sugar cane is the main land-cover of the buffer zone, which in turn, is also important in the Piracicaba basin (Figure 10).The current cost to restore 1 hectare of riparian forest in the Piracicaba Basin was estimated to be approximately R$ 7,000/ha (US$ 3,350/ha).Assuming that these values are also valid for other basins, the final cost to restore approximately 4,700 km 2 of riparian area would be equivalent to a cost of 3.28 billions reais, or 1.57 billion dollars.
Our first research question considered what the major land-cover was in 1997 in certain watersheds.Pasture was the dominant land-cover in the watersheds, dominating almost half of the entire area.
The expansion of the agricultural frontier in the State of São Paulo toward the west resulted in a pat-  tern of land-cover where the original vegetation was replaced by coffee and cotton, and in turn, these crops by pasture (Webb et al., 2005;Brannstrom & Oliveira 2000;Brannstrom, 2001).The economic reasons for these changes go beyond the main focus of this study, but the Wall Street crash in 1929 together with the loss of fertility due to the erosion process in the Vale do Paraíba, the northeastern region of the State of São Paulo, are among the main causes for such changes.
Land-cover and land-use changes are frequently associated to the production of food, fiber and energy to meet human needs (Lambin et al., 2003).These are key activities for maintaining life and generating wealth.For instance, in 2004 the gross domestic product for agricultural business in the State of São Paulo was equivalent to approximately 34 billion reais or 16 billion dollars (dollar to real ratio of 2. 0885, February 23, 2007), equivalent to 6% of the whole gross domestic product of the State.This is a positive impact related to land-cover/use changes.On the other hand, land-cover/use changes also have negative environmental impacts that we briefly discuss below.
The current predominance of pastures in the investigated basins affects several ecological processes.These large areas of pasture limit the movement of the wild life among forest patches for many reasons.For instance, Pires et al. (2002) studied the movement of small mammals (rodents and marsupials) among patches of forest surrounded by grasslands and found that the movement between patches was restricted due to presence of pasture areas.The same seems to be true for top predators, such as the population of Panthera onca living in the Morro Diabo State Reserve, located in southwestern the State of São Paulo (Cullen et al., 2005).Another important environmental aspect of pasture land is the risk from the fire, which is intensively used by Brazilian livestock farmers.Pasture fires intensify soil erosion (Bertoni & Lombardi Neto, 1990) and water pollution (Cerri et al., 2001) due the loss of the soil surface cover and the subsequent exposure to erosive rainfall.
The second most important land-cover in the State of São Paulo in 1997 is sugar cane.The main region that sugar cane is cultivated is the central-eastern towards the northwestern regions of the State of São Paulo, mainly in the Piracicaba and Moji-Guaçu watersheds.The sugar cane area in the State of São Paulo comprises approximately 31,600 km 2 , equivalent to 14% the area of the State (Rudorff et al., 2004).In comparison, in 1997 the area planted with sugar cane in the seven watersheds of this study was equal to approximately 12,000 km 2 .
The main ecological and social problem linked to this crop is the burning of sugar cane fields before harvesting each year from May to October (Martinelli & Filoso, 2007).This process leads to serious atmospheric pollution and acid rain (Lara et al., 2001), which may cause further soil acidification (Krusche et al., 2003).In addition to these effects, the amount of air-borne particules also increased with sugar cane burning (Lara et al., 2001;Lara et al., 2005).This increase of air-borne particles caused an increase of respiratory diseases in regions such as Araraquara (Arbex et al., 2000), and Piracicaba (Cançado et al., 2006).In addition, every 5 or 6 years, sugar cane plantations are removed and cultivated again.This type of management requires heavy traffic of machinery in the field, and each year a portion of the planted area is replaced, exposing bare soil for months, increasing the erosion risk (Cerri et al., 2001;Sparovek & Schunug, 2001).Once the erosion risk is higher, it also increases the chances of a higher input of applied fertilizers to streams (Filoso et al., 2003).
One of the most important aspects of landcover studies is to establish how much of the original vegetation remains for a "biophysical baseline", as defined by Brannstrom (2001).This "biophysical baseline" directly influences the distribution of vegetation and animal species, which ultimately defines the biodiversity of a certain region (Webb et al., 2006).According to our results, in the entire study area, only 15% of the original vegetation remained in 1997.
The highest deforestation rate in the State of São Paulo was observed between 1907 and 1934, when approximately 80,000 km 2 of original vegetation was lost in the conversion from forest to coffee (Brannstrom, 2000).In the Fifties, only 25% of the original vegetation remained (Vianna & Pinheiro, 1998).In 1962, using aerial photographs, Borgonovi et al. (1967) estimated that this area decreased to 14%, reaching a minimum of approximately 10% in the Eighties (Vianna & Pinheiro, 1998) erences).After this legislation was approved, deforestation was forbidden in many areas, such as in riparian zones, and in areas with slopes steeper than 25%.This law might have diminished the rate of deforestation and probably slightly increased the area of natural reforestation.To illustrate this situation, Toledo (2001Toledo ( ) found, between 1962Toledo ( -1995, in , in two small watersheds, an increase of natural vegetated areas from 19.5% to 34.0%.The three land-cover classes -pasture, sugar cane, and original forest -discussed above comprised approximately 80% of the entire study area.The first two land classes certainly play an important role in the economy of the State of São Paulo, and the last assumes a fundamental role in several ecological aspects of the landscape.On the other hand, most of the population in the watersheds that were addressed in this study live in urban centers (Table 1).Urban areas cover a small portion of the landscape, but these areas have a disproportional impact in relation to their area either socially or ecologically.Urbanized areas are normally linked to many social-economic and environmental problems (Ojima et al., 1994).Cities become one of the most life-threatening environments on Earth.The concentration of wastes is one of the most important aspects of cities (United Nations, 2003).The 2003 UN report, "Water for People Water for Life" notes that most urban areas have inadequate waste management, especially in developing countries.In the State of São Paulo, although 80% of the domestic sewage is collected, only 17% is effectively treated (Martinelli et al., 2002).Most of the untreated sewage is discarded into rivers and streams seriously impacting ecological and biological integrity (Ometto et al., 2000;Daniel et al., 2002).Therefore, in the most populated areas, such as the Piracicaba and the Mogi Guaçu basins, the water quality of rivers and streams is a major concern.
Our second research question was on how much of the riparian vegetation was still preserved in 1997 in the 30 m-width buffer zone.It is observed that only 25% of this vegetation type was preserved.This lack of conservation in riparian areas is not only occurring in the State of São Paulo.Riparian vegetation has been constantly and directly affected by deforestation worldwide.Riparian zones are important because they frequently harbor a disproportionately higher number of wildlife species and perform a disparate number of ecological functions as compared to most upland habitats (Fischer et al., 2000).Such an ecosystem is among the most productive and richest in biodiversity of the world (Loegering 1999), playing a key role as a corridor for wildlife (Metzger & Décamps, 1997, Metzger 1999).They are usually recognized in facilitating biological fluxes, reducing extinction risks, and providing refuges against disturbances, being essential components of the connectivity of patches in the landscapes among other environmental services (Metzger & Décamps, 1997, Stromberg, 2001, Joly et al., 2001).
Finally our third research question was about the dominant land-cover invading riparian zones.Grasses from pastures are the main type of vegeta-tion invading these areas in most of the basins, except in the Moji-Guaçu watershed, where sugar cane is the dominant invading vegetation into the riparian zone.Currently there is a government program in the State of São Paulo devoted to restoring riparian vegetation in small basins.This program claims that approximately 145 km 2 of riparian forest has already been restored (much less than the 4,700 km 2 that needs to be restored) at an estimated cost of 3.28 billion reais (1.57 billion dollars).In order to put these sums in perspective, we estimated that these costs are equivalent only to 10% of the gross domestic product related to agricultural business of the State of São Paulo in 2004.Not considering the difficulties in restoring riparian areas attributable to the natural complexity plant community structure (Joly et al., 2001), the total cost of restoration is modest as compared to the annual income generated by the agricultural business of the State of São Paulo.

CONCLUSIONS
The land-cover of the State of São Paulo in 1997 was a consequence of economic and social processes that occurred in the past.These processes have led to a clear dominance of pastures and sugar cane plantations in this State, with few areas of original vegetation remaining.These changes in land use had a wide range of environmental impact that was briefly described in this study, including an increase in soil erosion, water pollution, and loss of biodiversity mainly due to landscape fragmentation.Developing countries, such as Brazil, are characterized by dynamic economies, with very rapid and significant changes.These changes lead to important changes in the land-cover of the State of São Paulo.This study, which encompasses an area equivalent to almost 40% of the entire state, opens the interesting possibility of comparing the land-cover in 1997 and the future land-cover of seven major watersheds of the State of São Paulo

Figure 1 -
Figure 1 -Map of the São Paulo state showing the seven studied basins in dark grey contours.

Figure 2 -
Figure 2 -Spatial distribution of pasture in seven basins of the State of São Paulo.

Figure 3 -
Figure 3 -Spatial distribution of sugar cane in seven river basins of the State of São Paulo.

Figure 4 -
Figure 4 -Spatial distribution of original vegetation in seven river basins of the State of São Paulo.

Figure 5 -
Figure 5 -Spatial distribution of reforestation in seven watersheds of the State of São Paulo.

Figure 6 -
Figure 6 -Spatial distribution of perennial crops in seven watersheds of the State of São Paulo.

Figure 7 -
Figure 7 -Spatial distribution of urban settlements in seven watersheds of the State of São Paulo.
Figure 10 -Land use, expressed in percentage, occurring in the riparian preservation area (30m buffer strip zone for each stream side).

Table 2 -
Land-cover classes and their respective areas (km 2 ) and proportions.
. More recently the Instituto Florestal (2005) estimated that approximately 15% of the area of the State of São Paulo is covered with natural vegetation.It is important to note that in September 15, 1965 the Brazilian Forest Legislation was implemented (Instituto Florestal web site, see ref-