THREATS TO THE CERRADO REMNANTS OF THE STATE OF SÃO PAULO , BRAZIL

Patches of natural vegetation remaining in landscapes occupied by man are continuously under threat due to the edge effects and also to land use types around these remnants. The most frequent threats and land use types in the vicinity of 81 Cerrado (tropical savanna type) fragments in the State of São Paulo, Brazil, were analyzed in order to verify if the frequency of every type of disturbance to the natural ecosystem depends on the neighboring land use. The hypothesis of the study assumes that environmental threats are correlated with land use around protected areas. From the 81 areas, the most frequent human-induced land cover types around the Cerrado remnants were: pasture (recorded in 78% of the areas), sugarcane plantations (26%), roads (19%), annual crops and reforestation (14% each). The most frequent sources of threats were invasive grasses (35% of the areas partially or totally invaded) and cattle (observed in 32% of the areas), followed by deforestation (21%), and fire (21%). The chi-square analysis revealed that, with the exception of deforestation, which does not depend on land use, all other threats are influenced by the neighboring land use. The occurrence of invasive grasses and fires are strongly favored by the presence of roads and urban areas. Sugarcane, reforestation, and permanent crops were the less impacting land use types found in the study area, when only considering impact frequency. These land use types have fire and weed control, and also exclude cattle, indirectly protecting natural ecosystems.


INTRODUCTION
The Cerrado vegetation occupies more than 2 million km 2 in the central part of South America and is the largest, richest and possibly most threatened tropical savanna in the world (Silva & Bates, 2002), and has been highlighted as a global hotspot for biodiversity conservation (Myers et al., 2000).The causes of destruction and the present situation of the Cerrado biome was analyzed by Alho & Martins (1995), Ratter et al. (1997), Cavalcanti & Joly (2002), Klink & Moreira (2002), Machado et al. (2004), Alho (2005), Felfili et al. (2005), Pivello (2005), Ribeiro et al. (2005) and Ratter et al. (2006).These studies show that from an almost no impacting low density cattle raising some 40 years ago, the land use of the Cerrado domain has changed to cultivated pastures with African grasses or modern mechanized agriculture, with their techniques for the rapid clearing of extensive landscapes and for the improvement of the soil fertility through liming and fertilization.Current estimates (Machado et al., 2004) place the remaining more or less intact areas of Cerrado vegetation in ca.34% of its original surface in Brazil.
The Cerrado vegetation, which covered originally 14% of São Paulo State (São Paulo, 1997) has been even more drastically reduced, the remnants corresponding to only 0.81% of the state (Kronka et al., 2005).The current Cerrado vegetation in the state (less than 7% of the original cover) has been split into thousands of small areas, and surrounded by pasture, sugarcane, soybean, reforestation, perennial crops, and urban zones.
The main threats to the Cerrado biodiversity, (Alho & Martins, 1995;Pivello, 2005) are the deforestation, fire and invasive grasses, which are subject of the present study.Every change in land use around natural areas brings its own consequences to the ecosystem, and yet this relationship has rarely been studied.
Fragments of the Cerrado vegetation in São Paulo State (public natural reserves not included) were the subject of the present study, in order to verify whether the frequency of different types of impacts on the fragments depends on the land use around it.This study also aimed to determine the most frequent threats to the Cerrado vegetation and the most frequent land use types around the remnants in the state.

The sites
In 1998, the Brazilian Ministry of the Environment promoted a national workshop to define conservation priority areas for conservation of the biodiversity in the Brazilian Cerrado (Brasil, 1999).A number of priority areas were identified, based on biodiversity richness and threat.Some of the selected areas, located in São Paulo State, had been previously identified in a workshop held in 1995 by the government of São Paulo State (São Paulo, 1997), identifying 23 areas of high priority for conservation.
In both national and state meetings, every priority area included some large and/or clustered fragments with high biodiversity, frequently under threat, and generally located in regions where none or few reserves existed.
A set of 81 fragments, containing cerrado or ecotonal vegetation Cerrado / Atlantic forest, inserted in those priority areas were evaluated in this study (Table 1).The fragments were selected and surveyed on the following principles: all large fragments in the priority areas; fragments of different size (large -400 ha or over, median -100 to 400 ha, small -less than 100 ha) were surveyed in each area, and sites representing all different Cerrado physiognomies in an area.
These fragments were submitted to a rapid biological assessment and to a phytogeographical analysis (Durigan et al., 2003) that showed distinct floristic patterns in the different regions.
Every fragment had its physical attributes (coordinates, limits, Cerrado physiognomies, topography, water bodies etc.) previously mapped based on satellite images (Landsat 5 and 7, Thematic Mapper sensor, years 2000-2001) using Arc View Geographical Information System (ESRI).Using these maps, and field surveys, the fragments were evaluated according to their biophysical attributes, integrity of natural resources and external influences, and were classified according to the conservation priority criterium (Durigan et al., 2006).
Some of these records were used in the present analysis: land uses around the Cerrado fragment and threats to the ecosystems, including incidence of fire, biological invasion by African grasses, presence of cattle, and deforestation.

Land use
Land use along the perimeter of the Cerrado remnants was classified in the field, according to the following categories: Pasture: including native grasses, but all surveyed areas used for cattle raising planted with African grasses, especially those from the genus Urochloa ("braquiárias"); Annual crops: all crops of short cycles, such as soybean, corn, cassava, peanuts, etc., which imply in soil preparation and planting at least once a year; Perennial crops: woody plants which do not require soil preparation and annual planting, and that also do not depend on cutting or pruning, such as fruit, coffee and rubber tree plantations; Sugarcane: considered as a separate category since its management usually includes human-induced fires, supposed to increase fire incidence in Cerrado    A number of areas were surrounded by more than one land use, and all were recorded as present or absent in the vicinity of every fragment.

Threats
Threats to the Cerrado ecosystem were classified as follows: Invasive grasses: four categories were considered for the diagnosis: absent; only at the edges of the natural vegetation; area partially invaded; area totally invaded; Fire: three categories were considered: no evidence of fire; area partially burned; area totally burned;  Cattle: present; absent inside the natural area; Deforestation: classified in two categories: trees sparsely cut or area partially deforested after the last mapping (Kronka et al., 1993) about ten years before this evaluation; no cutting recorded.

Chi-square analysis
Applied to verify weather the disturbance frequency is correlated to land use in the vicinity of the Cerrado fragments, by comparing the observed against the expected frequency of each type of disturbance.The observed frequency was the number of fragments in which each kind of disturbance was recorded in the field, for every land use.
The expected frequency was calculated as the number of fragments which were expected to have suffered impact if its frequency was the same for all land uses.If the observed and the expected frequency were not significantly different, the null hypothesis would be confirmed and the type of disturbance, and its recurrence would be independent of the land use around the Cerrado area.
For the chi-square analysis, areas were classified according to invasion as: partially or totally invaded; invasive grasses absent or just at the edges.For fire, areas were classified under two categories: fire recorded (area partially or totally burned); fire not recorded.
Two types of land uses were excluded from the chi-square analyses: water bodies and railroads, since they were both recorded in only two areas, a number insufficient for inclusion within the chi-square analysis.Seven land uses, with two categories each, resulted in six degrees of freedom for every land use.

RESULTS AND DISCUSSION
The diagnosis concerning to the most common land use types around the Cerrado remnants in São Paulo State between 2000 and 2001 is in Figure 1.Pasture is also the most widespread land use for the Brazilian Cerrado Domain as a whole (Sano et al., 2001).These authors found pastures (native or planted) corresponding to 61% of the deforested areas, followed by annual crops (8%), and other uses such as perennial agriculture, reforestation and urban zones (altogether corresponding to 31%).The main difference from the national scenario is that sugarcane has been reported as replacing the Cerrado vegetation only in the São Paulo State (Kronka et al., 2005;Durigan et al., 2004).In other states of the core Cerrado Domain, soybean has been recently the main cause of deforestation, followed by corn.According to Klink & Moreira (2002), 40% and 22% of national production of soybean and corn, respectively, come from the Cerrado region.
In the São Paulo State, there is a recent tendency of replacing vast areas of pasture by sugarcane, especially in the west, where soil fertility, climate and topography are suitable for this monoculture.The international demand for biofuel, which has been continuously increasing, and the economic and environmental attractiveness of sugar cane as a renewable energy resource for climate change mitigation through substitution of fossil fuels lead sugar cane area to keep growing (Cerri et al., 2007).
Analysing the threats to the Cerrado ecosystem in São Paulo State (Figure 2), the invasion of African grasses (Figure 2A) has been the most common disturbance, recorded in 72% of the fragments (37% only at the edges).African grasses have been reported as one of the most widespread threats to the Cerrado ecosystem, since they proliferate, persist and spread, jeopardizing the ecosystem (Klink, 1994;Berardi, 1994;Pivello et al., 1999a;1999b).They also increase the flammable biomass considerably, intensifying the damage caused by wildfires.
Although considered a very important threat to the Cerrado Biome when intensified by human activities (Pivello, 2005), fire has become rare in the São Paulo State, since its use in natural areas is forbidden by law.Only 9% of all sites in this study had been totally burned, 12% were partially affected by recent fires and 79% had no evidences of fire at all (Figure 2B).There is a debate on the negative effects of fire on the Cerrado ecosystem, since it has been recorded since the arrival of the first inhabitants in South  America (Dean, 1995) and even before that, caused by natural causes (Ramos-Neto & Pivello, 2000;Miranda & Sato, 2005).Some authors point out benefits of fire to the Cerrado ecosystem (e.g.Coutinho, 1990).However, the increase in fire frequency and modification of the fire season can have negative effects on the regeneration of woody plants, favouring the herbaceous layer (Hoffman, 1999;Hoffmann & Moreira, 2002;Klink & Moreira, 2002), thereby changing both plant and animal communities (Durigan et al., 1994;Sato & Miranda, 1996;Hoffmann, 1998;Ramos-Neto & Pivello, 2000;Pivello, 2005).Protection against fire favours the Cerrado physiognomies to become denser, as observed in a number of areas (Goodland & Ferri, 1979;Ratter et al., 1988;Ratter, 1992;Moreira, 2000;Durigan & Ratter, 2006) and these changes seem to be the natural direction towards a new climax for most of the Cerrado remnants, especially in the São Paulo State.
Extensive cattle grazing (Figure 2C) was recorded in 31% of the areas.Whether the presence of cattle inside the natural fragments is negative to the Cerrado ecosystem or whether it helps to control invasive grasses, decreasing the risk of wildfires, is another controversial aspect.In spite of considerably reducing grass biomass, cattle can remarkably change the structure and dynamics of the Cerrado vegetation, by selective herbivory and mechanical impact (Souza et al., 2006).
Deforestation (Figure 2D) includes both cases: partial clear-cutting (habitat reduction) in a ten-year period from the last mapping (Kronka et al., 1993) or trees sparsely cut (changing the community structure without reducing the size of the fragment).This impact was recorded in 21% of the fragments.More than ten years ago (Kronka et al., 1993), the Cerrado was considered one of the most devastated vegetations in São Paulo state.Unfortunately, recent studies (see Kronka et al., 2005) indicate that deforestation is still ongoing.In 1988, a federal law prohibiting any exploitation or deforestation of the Atlantic Forest was issued (Decreto 750, Brasil, 1993).As a consequence, Cerrado and even ecotonal areas started being the only possible areas to be converted into agriculture or pasture in the state, thus accelerating the destruction of the already scarce and unprotected Cerrado vegetation.
The analysis of the types of disturbance to the Cerrado remnants in São Paulo State according to land uses in the vicinity of the natural vegetation showed that, except for deforestation, the frequency of all other disturbances depended on neighbouring land uses (Figure 3).
Invasive grasses (Figure 3A) and fire (Figure 3B) have a very similar relationship with land use and are positively correlated (χ 2 = 9.19, χ 2 0.01(1d.f.) = 6.64), grasses increasing fire and vice-versa.They both increase when the Cerrado has roads or urban zones in the vicinity, and decrease when the land use is refor-   This means even being a land use that requires fire every year, the fire control was efficient in the period of the present study (2000)(2001), so that native vegetation was less frequently burned where the land use around was sugarcane than where it was pasture, annual crops, roads or urban zones.The presence of cattle inside the fragments (Figure 3C) was slightly over the expected when the natural vegetation in the vicinity was pasture or urban zones and significantly reduced if the land use around was reforestation, perennial crops or sugarcane.Reforestation, sugarcane and perennial crops promote the use of or are prone to fire and weed control, and also exclude cattle, thus indirectly protecting the natural ecosystems.The positive role of Eucalyptus plantations to the Cerrado fragments was also pointed out by Shida & Pivello (2005), who noticed a wide variety of mammals using the cerrado/plantation matrix as a habitat.
Although the observed frequency of deforestation (Figure 3D) has been slightly higher than that expected for the Cerrado fragments close to urban zones and sugarcane, the dependence of land use was not significant.This means that the frequency of Cerrado deforestation has been practically the same for all land use types under comparison.The extension of the deforested area for each land use, however, was not analyzed.Other aspects such as the permeability of different land use types for fauna and the relative harm of the threats to the ecosystems were not analyzed as well.Therefore, our discussion is based just on how different land uses in the vicinity of Cerrado fragments can raise the probability of bringing a certain disturbance.
The threats resulting from different land use pressures to the remnants of natural vegetation must be considered in reserve selection and conservation policies.Conservation strategies and costs will be, necessarily, different among Cerrado areas surrounded by urban zones and roads or, in the opposite, by reforestation or perennial crops.Public policies must consider this, so that those land uses which increase disturbance frequency should be discouraged in the neighborhood of the scarce Cerrado remnants, while those which decrease the threats should be stimulated, not disregarding other aspects, such as the permeability of every land use to the fauna.Analyzing the results of the present study as to how they can be applied to select an area for conservation is a controversial issue.The final decision usually takes into consideration the principle: 'if it ain't broke, don't fix it'.In other words, priority has been given to areas under imminent risk of suppression or degradation.Although controversy exists, this has been the criteria for the indication of global hotspots (Myers et al., 2000) and also for priority areas for biodiversity conservation in Brazil (São Paulo, 1997;Brasil, 1999;Brasil, 2000).On the basis of this rationale, Cerrado fragments surrounded by urban zones or highways would be considered priority areas for establishing new reserves.
However, this study indicates that selecting areas for natural reserves must consider also the investment risks and costs, and, except when supported by their uniqueness or irreplaceability, natural areas under permanent threat should not be considered a priority when selecting areas for establishing new reserves.

Figure 1 -
Figure 1 -Frequency of land uses in the vicinity of 81 Cerrado remnants in São Paulo State, Brazil (% means the percentage of the surveyed areas where the land use was recorded).

Figure 2 -
Figure 2 -Threats to the Cerrado remnants in São Paulo State and their frequency.A: invasive grasses; B: fire; C: cattle; D: deforestation.

Table 1 -
Location of 81 Cerrado fragments in the São Paulo State.

355-363, July/August 2007 estation
or sugarcane.The expected high incidence of fire in Cerrado fragments correlated with the yearly burned sugarcane in the vicinity was not confirmed.