PROTECTED AREAS IN BRAZIL: EVOLUTION, LAND USE AND COVER, AND IMPACT ON EMISSIONS INVENTORY

Faced with population growth and intensification of the use of natural resources, Protected Areas (PA) and Indigenous Lands (IL) play a fundamental role in environmental conservation. The carbon increment of preserved vegetation within these areas is accounted for as CO2 removal in the Land Use, Land-Use Change, and Forestry (LULUCF) sector of the National Emissions Inventory. A topological evaluation of overlaps was made of the databases of PAs (full protection and sustainable use) and ILs, showing that they represent 6% of the protected areas. Of the total of PAs and ILs areas, 43% were established by 1994, 26% by 2002, 26% by 2010, and 4% between 2011 and 2016. This evaluation showed that 30% of the territory is within PAs and ILs. In terms of area, 84% is distributed in the Amazon, 9% in the Cerrado, 4% in the Atlantic Forest, 3% in the Caatinga, 0,2% in the Pantanal, and 0,3% in the Pampa. In assessing the use and cover of 2010 of the PAs and ILs created up to this year, it was possible to notice that 68% of the areas within the PAs have preserved vegetation, whereas in ILs, this percentage increases to 89%. Under the Third Emissions Inventory, the removal of vegetation protected in PAs and ILs decreased by 16% of the national net emissions of the LULUCF sector between 2002 and 2010, showing the importance of preserving forests and grasslands.


INTRODUCTION
Population growth is linked to an increased demand for natural resources to meet human needs. As a particular natural resource becomes scarce, the use of other products increases. This replacement dynamics generates a process of successive depletion of several natural resources (HASSLER, 2005).
Unchecked deforestation affects environmental conditions, such as humidity and heat. Therefore, the creation of protected areas is crucial to reestablish ecosystem services in disturbed landscapes (MAACK, 1981). The legal provisions that regulate the creation of protected areas (PAs) were established in the National System of Protected Areas (SNUC; BRAZIL, 2000). These PAs contribute to the protection of native biodiversity and to the restoration of degraded ecosystems, among other functions (ICMBIO, 2018).
In addition to the PAs, the Federal Union owns portions of the national territory inhabited by one or more indigenous peoples, which are designated as Indigenous Lands (ILs). These are used for the productive activities of indigenous peoples, and their demarcation is the responsibility of the National Indigenous Foundation (FUNAI). The National Policy for Territorial and Environmental Management of Indigenous Lands (PNGATI) ensures Details of each methodological step will be provided below. FLORESTA,Curitiba,PR,v. 51,n. 1,jan

Data acquisition
Spatialized files (in shapefile format) of PAs and ILs were acquired on July 19 and 31, 2018, respectively, from the databases of the Ministry of the Environment (MMA, 2018) and the National Indigenous Foundation (FUNAI, 2018). For this work, only the PAs and ILs created until 2016 are considered, as they reflect the database used in the scope of the IV Emissions Inventory, part of the Fourth National Communication. The Private Reserves of Natural Heritage (RPPNs) were not considered in this analysis; although they present amounts higher than those of all categories of PAs, they are not significantly representative in relation to the total area of PAs (0.005%, 0.0006%, 0.31%, 0.47 % and 8.96% of the total area of PAs, in the case of the Pampa, Amazon, Cerrado, Atlantic Forest and Pantanal biomes, respectively). In addition, some of them overlap with other PAs. Despite expressive representation of the Pantanal biome, it was decided to maintain national consistency.
The indigenous land demarcation process (Article 231 of the 1988 Federal Constitution; Law 6,001; Decree 1,775/1996) is carried out in 5 stages, ranging from identification and delimitation of the area to registration as Federal Union property. This study focuses on the first stage and, when this information was not available, the following was considered.
The information on land use and cover for the year 2010 was acquired from the database of the 3 rd Emissions Inventory (BRASIL, 2016a). This map was generated by means of visual interpretation of medium resolution images (TM/Landsat and LISS3/ResouceSat-1) for the whole of Brazil, on a 1: 250,000 scale and minimum area of 6.25 ha. The maps database of the 3 rd Emissions Inventory also includes maps of the entire territory for the years 1994, 2002 and 2010, and specifically of the Amazon biome for 2005. As previously mentioned, it was decided to use only official data for this study, although there are other data sources, including more recent ones (MAPBIOMAS, 2019).
The matrices of gross vegetation emissions, vegetation removals, soil emissions and removals and net emissions, by period and biome, were obtained from the LULUCF Sector Reference Report of the 3 rd Emissions Inventory (BRASIL, 2016b).

Processing Topological evaluation
The topological evaluation consisted of identifying doubling up of information in the same area, i.e., overlap between polygons in the shapefile.
Once the database had all the information needed for the evaluations, a topological evaluation of the PA and IL files was carried out. The ArcGis Topology tools were used to identify and quantify overlaps in the PA and IL shapefiles. In the topological analysis, performed in ArcGis, a cluster tolerance of 0.000000009 was maintained, which is automatically set by the software. When necessary, to simplify the analysis of the overlaps, the ArcGis dissolve tool was used to join the identified fragments.

Merging of PA and IT bases, overlap identification and correction
The QGis Intersection tool was used to identify possible overlaps between PAs and ILs, which have different sources of information. Next, the overlaps were adjusted, keeping the original information of the source files.

Intersection of protected areas and land use and cover
In order to evaluate land use and land cover in Brazilian protected areas, the IL and PA areas were intersected with the 2010 land use and land cover map generated in the scope of the 3 rd Emissions Inventory. The QGis Intersection tool was used for this purpose.

Evaluation of the impact of protected vegetation CO2 removal in PAs and ILs in LULUCF sector emissions
Based on the matrices of CO2 emission from land use and cover changes presented in the LULUCF sector reference report of the 3 rd Emissions Inventory, managed forest and grassland removals (FM and GM, respectively, according to the Emissions Inventory key) were specifically considered. For the purposes of the Brazilian inventory, "managed" forest or grassland means that this vegetation is protected within a PA or IL. The proportion of these removals in relation to the country's gross emissions was calculated in order to quantify the impact of accounting for these CO2 removals in Brazil's net emissions.

Qualitative and quantitative evaluations on information management
The results of the topological evaluations with respect to overlaps between protected areas, per biome, are presented in Table 1 Regarding the topological evaluation of national data on PAs provided by ICMBio, it was noted that overlaps occur mostly when considering different government spheres (municipal, state and federal). Overlaps were also found in the official data on ILs provided by FUNAI. When merging the PA and IL bases, the overlap between ICMBio and FUNAI data was also significant.
The overlaps in the Amazon biome amounted to 11,888,980 ha, which represents 6% of the biome's total protected area. The Cerrado and Atlantic Forest presented overlaps of 1,766,343 ha and 1,077,399 ha, respectively, equivalent to 8% and 11% of the total protected areas in each of these biomes. For Caatinga and Pampa, the overlapping areas represented 0.4% and 1% of the total protected areas of these biomes (25,081 ha and 2,605 ha). As for the Pantanal, an overlap area between ILs was adjusted according to a re-study of the area and only one overlap between PAs and ILs was found, which represents 0.3% biome (2,252 ha).
These overlaps represent 6% of all protected areas in the country by 2016. The biomes where these overlaps were most significant are the Amazon and the Cerrado.
At the national level, the overlap areas that were found represent 6% of all areas of PAs and ILs in the country, considering all those created up to 2016.

Evaluation of the evolution of protected areas until 2016
In the case of the Amazon biome, 274 PAs and 331 ILs were identified. It was not possible to find information on the date of creation of 15 ILs, whose total area represented 1,091,233.21 ha. In the Cerrado, 217 PAs and 101 ILs were verified, of which 11 ILs had no creation date and were disregarded (197,506 ha). As for the Atlantic Forest biome, 556 PAs and 145 ILs were identified. Of these, 7 ILs with no date (amounting to 1,477 ha) were disregarded in the analyzes. For the Caatinga biome, 99 PAs and 39 ILs were identified, while 9 indigenous lands (83,475 ha) with no year of creation were disregarded. In the Pampa biome, 17 PAs and 8 ILs were identified, two of which had no creation date (138.4 ha). 7 PAs and 8 ILs were registered in the Pantanal biome. Of the total protected area in the country, 84% is found in the Amazon biome, 9% in the Cerrado, 4% in the Atlantic Forest, 3% in the Caatinga, 0.2% in the Pantanal and 0.3% in the Pampa. Table 2 (BRASIL, 2016b). It is noted that, of the approximately 30% of the territory contained in PAs or ILs until 2016, the biome with the largest share within protected areas is the Amazon, followed by the Cerrado, the Atlantic Forest, the Caatinga, the Pantanal and the Pampa, respectively. The biomes with the largest protected area are also the ones with the highest representativeness of vegetation with forest structure (MTIC, 2016).

Land use and cover in Brazil's protected areas
The results are presented in relation to the percentage of representativeness of each use in 2010, being: i) protected vegetation (referring to water, managed forest and grassland classes in the 3 rd Emissions Inventory); ii) secondary vegetation; iii) agriculture and reforestation; iv) pasture and v) other land (areas not observed due to clouds and shadows, urban area, mining, dunes and artificial reservoirs). For this stage of the evaluation, only the PAs and ILs created until 2010 were considered. Figure 3    Brazil's Emissions Inventory takes into account forests and grasslands preserved within PAs and ILs as "managed", since there is some anthropic intervention on that land, according to the IPCCs concept of "managed". In this case, specifically, it is assumed that this vegetation is preserved thanks to the designation of the land as protected. For the sake of clarity, this concept is not linked to a "Management Plan", for example.
Based on the carbon dioxide (CO2)  The annual increase factors for carbon in primary forests used in the 3 rd Emission Inventory are 0.43 tC/ha for the Amazon, 0.2 tC/ha for the Cerrado and Pantanal, 0.1 tC/ha for the Caatinga and 0.32 tC/ha for the Atlantic Forest. For primary grasslands, the value used was 0.52 tC/ha per year for all biomes (BRASIL, 2016b). In addition, for situations 3 and 4 described above, removals are accounted for during half of the period, i.e., instead of considering removals for the full period (8 years from 2002 to 2010), removals for half the period are considered (4 years). This carbon gain is converted to CO2 by means of the conversion factor (multiplying by 44/12) and, because there is an increase in biomass and, consequently, in carbon removal from the atmosphere, they are subtracted from the total emissions resulting mainly from forest conversions to deforestation. Table 3 presents removals from forests and primary grasslands within PAs and ILs from 2002 to 2010. These values are presented in gigagrams (or 1,000 tons) of carbon dioxide (CO2), the unit used to present results in the emission inventories. In this study, carbon removals resulting from the growth of secondary vegetation or even land use conversions linked to some gain in carbon stock were not considered (for example, pasture converted to reforestation represents a carbon gain and, consequently, CO2 sequestration). The gross emission values are also presented, i.e., without considering any carbon gain, by biome, in the period, and the impact that removals have on these emissions.

DISCUSSION
The overlaps found both in the institutions' own data (in this case, ICMBio and FUNAI), as well as in the process of gathering information, indicated the need to adjust not only the bases themselves, geometrically speaking, but also aligning the hierarchy of the different government spheres and entities. These overlaps were also discussed and presented by Ricardo (2004). As mentioned earlier, overlaps can generate a doubling up of CO2 removal count linked to preserved forests and grasslands in protected areas. The adjustment of the official database represents a meticulous processing step within the scope of the Emissions Inventory, since this base cannot simply be incorporated into the processing in the format provided by the bodies responsible for generating this information. In addition, as CO2 removals are counted from the date of creation of the PA or IL, it is always necessary to allocate a year to that overlap area. Therefore, this association is made subjectively, since, for most cases, the years of creation are not coincident and there is no consensus on the hierarchy in relation to the creation of areas, as also discussed by Ferreira (2018). These overlaps also occur in PAs and ILs that do not allow the same level of intervention in the areas, which also hinders the enforcement of legislation on protected areas.
Regarding the evaluation of land use and cover within PAs for 2010, Sustainable Use (SU) and Full Protection (FP) PAs are being considered. The SU PAs do not, therefore, reflect the main objective of the FP PAs, which is to preserve vegetation. In the case of the Pantanal, due to the characteristics of the biome, there are limitations/restrictions to exploitation of the vegetation which contribute to its preservation. In the case of the Pampa, many pasture areas (20%) possibly represent grassland areas used for grazing, but that would not necessarily be connected to a conversion of use. The same seems to occur in the Cerrado biome. In the case of the Atlantic Forest, practically 50% of the areas would in fact be protected, with a significant share of the PAs being used for agricultural crops and reforestation. In the case of the Caatinga, 10% of the "Other land" class is represented by areas not observed in the map prepared for the 3 rd Emissions Inventory. The same is true for the Amazon biome, where the unobserved areas of 2010 represent 6% of the 7% of the "Other land" class". The difficulty in these cases is to associate these areas with preservation or anthropic use. The results presented by Mattar et al. (2018) show vegetation conservation scenarios greater than those presented in this study, as they only consider federal PAs. The authors do not make it clear whether only FP PAs were considered, but they also do not consider RPPNs. In the case of the Emissions Inventory, the different types of management established in protected areas (i.e., full protection, sustainable use) are not considered, but the entire dynamics of land use and cover is considered. The Emissions Inventory considers the carbon increase from preserved vegetation in the protected areas, which is the only difference in relation to the rest of the territory, i.e., outside the PAs and ILs.
The land use and land cover evaluation within ILs for 2010 showed that most of these areas are preserved. With respect to the Pantanal and Pampa, possibly, areas of natural grassland used for grazing were included in the pasture class. The areas within the IL of the Cerrado and Amazon biomes are practically all preserved. Regarding the Atlantic Forest and Caatinga biomes, practically 60% of the areas would in fact be preserved, with a significant share being used for pasture, agricultural crops and reforestation. In this sense, it is worth questioning whether the PNGATI monitoring is taking place efficiently in these regions. The legislation applied to ILs is being reviewed to allow more intensive use of the areas, so it is essential to ascertain the position of the communities in relation to the current legislation and possible changes, in order to balance their need for production with the need for preservation. Even though the areas within the ILs are preserved, it is important to evaluate their surroundings in terms of the regional and economic context (BRONDIZIO et al., 2009).
The preservation of vegetation plays an important role in accounting for Brazil's net emissions. Nogueira et al. (2018) correlate protected areas and their carbon stock, in order to quantify the associated carbon losses. Likewise, the authors highlight the importance of preserving vegetation in these areas.  also evaluated land use and cover in protected areas, highlighting the need for preservation not only in these areas, but also the importance of private landowners in the scope of the Forest Code. From the point of view of the Emissions Inventory, however, it is still necessary to question whether considering removals from preserved forests and grasslands in the long term will not lead to a false impression that the country has effectively protected its vegetation.

CONCLUSIONS
• It is essential to ensure interaction and harmonization of data from government agencies responsible for data on PAs (ICMBio) and ILs (FUNAI), since these data should complement one another and not overlap. The management of the use of protected areas is directly impacted by the instrument that governs them. For the Emissions Inventory, overlaps may represent double counting of removals by carbon increment from preserved forests and grasslands in these areas. important, since they have different objectives, as well as different rules and levels of restrictions regarding land use and occupation. Therefore, this distinction should be considered in future analyzes. • The protection of forests and grasslands within PAs and ILs has a significant impact on Brazil's GHG emissions inventory, and it is important to assess whether the inventory methodology employed to account for removals throughout the inventory period is in fact consistent with the reality of forests that have already reached the climax stage. • This research has limitations in terms of time cutoff because official data on land use and land cover across the country has not yet been updated. Evaluations could be carried out in a subsequent step by separating SU PAs from FP PAs. Although this limitation exists, this work represents a key contribution to the discussion on the accounting and monitoring of national GHG emissions, with a focus on improving analysis methods and criteria. • This study is aligned with Sustainable Development Objective (SDG) number 13, which concerns action against global climate change, part of the 2030 UN Agenda for Sustainable Development.