The “New Transamazonian Highway”: BR-319 and Its Current Environmental Degradation

: The Brazilian government intends to complete the paving of the BR-319 highway, which connects Porto Velho in the deforestation arc region with Manaus in the middle of the Amazon Forest. This paving is being planned despite environmental legislation, and there is concern that its effectiveness will cause additional deforestation, threatening large portions of forest, conservation units (CUs), and indigenous lands (ILs) in the surrounding areas. In this study, we evaluated environmental degradation along the BR-319 highway from 2008 to 2020 and veriﬁed whether highway maintenance has contributed to deforestation. For this purpose, we created a 20 km buffer adjacent to the BR-319 highway and evaluated variables extracted from remote sensing information between 2008 and 2020. Fire foci, burned areas, and rainfall data were used to calculate a drought index using statistical tests for a time series. Furthermore, these were related to data on deforestation, CUs, and ILs using principal component analysis and Pearson’s correlation. Our results showed that 743 km 2 of forest was deforested during the period evaluated, most of which occurred in the last four years. A total of 16,472 ﬁre foci were identiﬁed. Both deforestation and ﬁre foci occurred mainly outside the CUs and ILs. The most affected areas were close to capital cities, and after resuming road maintenance in 2015, deforestation increased outside the capital cities. Current government policy for Amazon occupation promotes deforestation and will compromise Brazil’s climate goals of reducing greenhouse gas (GHG) emissions and deforestation.


Introduction
At the beginning of the 26th UN Conference on Climate Change (COP 26), Brazil surprisingly changed its discourse on climate change, which was initially one of denial, just

Study Area
The study was conducted along BR-319, officially known as Rodovia Álvaro Ma Rodovia Manaus-Porto Velho, an 885 km-long highway ( Figure 1). In Amazonas highway begins in the capital city of Manaus and passes through the municipalitie Careiro da Várzea, Careiro, Manaquiri, Borba, Beruri, Manicoré, Tapauá, Humaitá, Canutama. It ends in the capital city of Porto Velho, the only municipality within the S of Rondônia. BR-319 was built in the Amazon biome [24,25], and a 20 km buffer was sidered, resulting in a total area of 35,697.52 km 2 ( Figure 1). More than 2500 tree spe (one-third of the world's tropical wood) and 30,000 plant species grow in the Ama biome [26]. According to the Köppen classification, updated by Alvares et al. [27] climate type is "A" and is categorized into four main sub-climates: tropical mons ("Am"), dry and humid tropical ("Aw"), rainy equatorial ("Af") and hot tropical and ("AS"). Forest and anthropization are presented in the previous figure using a methodo developed by Souza Jr et al. [28] based on the normalized difference fraction index (ND The NDFI values range from −1 to 1, with high values (close to 1) indicating intact fo This is due to the combination of a high pattern of healthy vegetation and low valu photosynthetically inactive vegetation and soil. The methodology for logged area collection is presented in Section 2.3. We utilized data from the PRODES project [7] fo calculation of deforestation along the buffer created between 2008 and 2020. Forest and anthropization are presented in the previous figure using a methodology developed by Souza Jr et al. [28] based on the normalized difference fraction index (NDFI). The NDFI values range from −1 to 1, with high values (close to 1) indicating intact forest. This is due to the combination of a high pattern of healthy vegetation and low values of photosynthetically inactive vegetation and soil. The methodology for logged area data collection is presented in Section 2.3. We utilized data from the PRODES project [7] for the calculation of deforestation along the buffer created between 2008 and 2020.

Fire Foci Analysis
Fire foci data were calculated using the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor product MCD14DL (TERRA/AQUA). Near real-time (NRT) MODIS Thermal Anomalies/Fire locations-Collection 6 processed by NASA's Land, Atmosphere NRT Capability for EOS (LANCE) Fire Information for Resource Management System (FIRMS), using swath products (MOD14/MYD14) rather than the tiled MOD14A1 and MYD14A1 products, was used to obtain solid data that are also used by the Brazilian program BDQueimadas [29]. The thermal anomalies/active fire represent the center of a 1 km pixel that is flagged by the MODIS MOD14/MYD14 fire and thermal anomaly algorithm [30], which contains one or more fires within the pixel. This characteristic is the most basic fire product in which active fires and other thermal anomalies, such as volcanoes, are identified. Data were downloaded directly from FIRMS (https://firms.modaps.eosdis.nasa.gov/, accessed on 15 November 2021) and changed to shapefile format (https://earthdata.nasa.gov/active-fire-data, accessed on 15 November 2021). The FIRMS Fire Map allows users to interactively browse the full archive of global active fire detection from MODIS and VIIRS. NRT fire data are available within approximately 3 h of satellite overpass and imagery within 4-5 h. In this case, we used only processed data and excluded NRT data because the foci were calculated for the BR-319 buffer between 2008 and 2020 with MODIS data only.

Burned Area Analysis
The burned area data product from the combined Terra and Aqua satellites MCD64A1 Version 6 is a monthly 500-m grid product containing per-pixel burned area information. The burned area mapping approach employs surface reflectance imagery from MODIS [31]. A burn-sensitive vegetation index was calculated using MODIS time series data from shortwave infrared channels. Dynamic thresholds were then applied to guide the statistical characterization of burn-and non-burn-related changes. Finally, spatial and temporal active fire information was used to create regional probability density functions to classify each pixel as burned or unburned [32].
The MCD64 burned area mapping approach employs MODIS imagery along with 1 km active MODIS fire observations. The hybrid algorithm applies dynamic thresholds to composite images generated from a burn-sensitive index, derived from MODIS shortwave infrared channels 5 and 7, and a temporal texture measure (Equation (1)).
where ρ5 and ρ7 are the atmospherically corrected surface reflectance in bands 5 and 7, respectively.
Mapping was used to identify the recording date of the satellite pass to the nearest record for individual grid cells. The date was then encoded into a single data layer of the astronomical day output. Burn output ranged between 1 and 366, and a value of 0 indicated "no burn" pixels and additional special values were reserved for missing data and water grid cells. For this purpose, mapping was performed along the entire length of BR-319, including a 20 km buffer, and calculated in km 2 between 2008 and 2020. A cloud processing algorithm was adopted using Google Earth Engine (ImageCollection ID: MODIS/006/MCD64A1; see link in Supplementary Material).

Standardized Precipitation Index
The standardized precipitation index (SPI) [33] seeks to quantify the rainfall deficit or excess at different time scales and was calculated on an annual scale (SPI-12) for the 20 km buffer along BR-319 from 1981 to 2020 to identify drought events in the historical time series. The SPI was calculated from the total accumulated rainfall records over 30 years, which was fitted to a probability distribution function and transformed into a normal probability distribution function. This set the average SPI value for a given location and period to zero. To determine the SPI, we initially used the gamma distribution calculation, which was defined by the probability density function (PDF) given by Equation (2). (2) where α (>0; dimensionless) is the shape parameter, β (>0; mm) is the scale parameter, x (>0; mm) is the total rainfall, and Г (α) is the Gamma function: Г(a) = ∞ 0 x a−1 e −x dx. All parameters and the gamma PDF were fitted to the frequency distribution of accumulated rainfall using remote sensing data at the mentioned scale. The gamma PDF parameters α and β estimated at the mentioned scale were calculated. The parameters α and β were estimated using the maximum likelihood method (MLM), which is the most appropriate method [34,35]. Calculations of the α and β parameters were performed to determine the cumulative probability of an observed rainfall event for the adopted scale.
For this purpose, a Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) dataset developed by the U.S. Geological Survey (USGS) and the Climate Hazards Group at the University of California, Santa Barbara, was used. It combines pentadal precipitation climatology, near-global geostationary TIR satellite observations of the CPC and the National Climatic Data Center (NCDC) [36], precipitation fields from the atmospheric model of the NOAA Climate Forecast System (CFSv2) [37], and in situ rainfall observations [38]. In the first step of this study, these resources were used because of the lack of temporal data from in situ stations along the BR-319 and because they are available from 1981 to the present. Their spatial resolution is 0.05 • (±5.3 km) and is provided in monthly, pentadal, and decadal temporal resolution for the entire globe [39].

Mann-Kendall and Pettitt Tests
The Mann-Kendall test was used to verify trends in the timeframe under study (2008-2020) for each variable [40,41]. Then, the Pettitt test was applied to identify change points in the time series when shown as significant in the Mann-Kendall test. For both tests, a 5% significance level was employed.
The Mann-Kendall test is calculated as follows (Equation (4)): ; for S > 0 where Z MK is the Z-index of the Mann-Kendall test, S is the "score" statistic, and Var(S) is the variance of the S statistic.
The Pettitt test is obtained as follows (Equation (5)): Sustainability 2022, 14, 823 The change point is t at which the maximum k(t) occurs. The critical values of K are obtained as follows (Equation (6)):

Principal Components Analysis
Data were subjected to principal component analysis (PCA) to assess the relationship between the variables and the years studied. PCA is a multivariate statistical analysis that transforms an original dataset (X 1 , X 2 , . . . , X p ) into another dataset of the same dimensions (Y 1 , Y 2 , . . . , Y p principal components), reducing data mass with minimum information loss (Equation (7)). The principal components are derived from the linear combination of the original variables, which are independent of each other and retain the maximum amount of information [42].
where a 1 , a 2 , . . . , a p are the eigenvectors of the correlation matrix between variables.

Pearson's Correlation
Pearson's correlation coefficients were estimated to verify the interdependence between the analyzed variables. A correlation heatmap was used to graphically express results. Positive correlations are expressed in blue, and negative correlations are expressed in red (Equation (8)).
where x 1 , x 2 , . . . , x n and y 1 , y 2 , . . . , y n are the measured values of both variables and x and y are the arithmetic averages of the variables. The free software Rbio was used to process the Mann-Kendall and Pettitt tests [43]. PCA and correlation analyses were performed using R software [44].

Results
There were 16,472 fires recorded near the BR-319 highway during the period of study. In 2010, 10.97% of the fires (1807) occurred, followed by 2015 with 12.20% (2010), and 2020 with 11.32% (1865). Of all the fires analyzed, 832 occurred in CUs, 44 in ILs, and 2888 in forested areas. The size of the accumulated deforested area in the study region is 743 km 2 . Of this total, the largest area of deforestation occurred in 2020, 149 km 2 (20.05%). The total burned area during the evaluation period was 3597 km 2 . The study results show that areas were affected by fires each year near BR-319, especially in 2015 (542 km 2 ). However, deforestation and fire foci in deforestation were the only significant variables according to the Mann-Kendall test and were shown to increase during the years of study. According to the Pettitt test, none of the variables showed significance, and no point of change was detected along the time series (Table 1).  Figure 2 highlights the burned areas, fires, and deforestation between 2008 and 2020 along the BR-319 highway. The image shows a high concentration of fires around the highway, mainly outside CUs and ILs, but these areas were also affected in smaller proportions, especially in the extreme north and south (Figure 2A,B). There was a larger clustering of burned areas in fragments where a high concentration of fire foci predominated, located in the municipality of Humaitá, AM ( Figure 2B,C). This indicates an advancement in deforestation in areas near the CUs in the south-north direction of the highway ( Figure 2D). Deforestation over the 13-year-long study period has been present in the surrounding area along the highway. It was shown to progress annually, with the highest incidence in 2020.
As shown in Figure 3A,B, fires predominated in deforested areas, CUs, and ILs. Of the analyzed fires, 17.53% occurred in deforested areas and were located near roads and urban areas. Of note, the fires occurring in CUs and ILs were recorded mainly in regions near the capital cities of Porto Velho-RO and Manaus-AM. In Figure 3D, it is shown that most deforestation that occurred in the study area was near the BR-319 highway, a process that occurred because of highway road network expansion. As shown in Figures 3A,B, fires predominated in deforested areas, CUs, and ILs. Of the analyzed fires, 17.53% occurred in deforested areas and were located near roads and urban areas. Of note, the fires occurring in CUs and ILs were recorded mainly in regions near the capital cities of Porto Velho-RO and Manaus-AM. In Figure 3D, it is shown that most deforestation that occurred in the study area was near the BR-319 highway, a process that occurred because of highway road network expansion.  Positive SPI values indicate above-average rainfall, while negative values indicate below-average rainfall [35,45,46]. It is shown in Figure 4 that from June to September were critical periods with minimal rainfall throughout the study period. Similarly, December to April experienced significant rainfall as compared to other months, except for the years 2010 and 2011, when there was less rainfall in the first half than in the other years. The months of May, October, and November represented the average rainfall at the site. Likewise, when comparing the annual SPI values, negative values were observed in almost all years, except for 2013, 2017, and 2019 (Table 1), which indicates a reduction in the average annual rainfall over the years.
x FOR PEER REVIEW 10 of 15 Positive SPI values indicate above-average rainfall, while negative values indicate below-average rainfall [35,45,46]. It is shown in Figure 4 that from June to September were critical periods with minimal rainfall throughout the study period. Similarly, December to April experienced significant rainfall as compared to other months, except for the years 2010 and 2011, when there was less rainfall in the first half than in the other years. The months of May, October, and November represented the average rainfall at the site. Likewise, when comparing the annual SPI values, negative values were observed in almost all years, except for 2013, 2017, and 2019 (Table 1), which indicates a reduction in the average annual rainfall over the years. The sum of the first two components in the PCA analysis explained 78% of the total variation in the data, and as the sum was greater than 70% it was considered adequate [42] (Figure 5A). It can be observed that the deforestation and fire foci in deforestation vectors were close for the years 2008, 2018, 2019, and 2020. This suggests a significant variable contribution and that these years reached a higher index for both variables (Table 1). Moreover, these variables contrasted with the fire foci, burned areas, and fire foci variables in ILs and CUs. This indicates a lower tendency of fire foci and burning in areas with high rainfall. The fire foci vectors in ILs and CUs were close in the years 2010, 2012, and 2015-periods of higher incidence for these variables (Table 1).
Correlations between 0.2 and 0.5 can be considered weak; those from 0.5 to 0.7 have a strong correlation; and those above 0.7 have a very strong correlation [47]. It is possible to observe a strong and positive correlation of 0.8 between fire foci and burned areas, which means that the increase in one influences the increase in another. This is also observed between the burned areas and fire foci in ILs (a strong and positive correlation of 0.7). A strong and negative correlation was observed between fire foci in the CUs and there was an SPI with a magnitude of 0.6. The sum of the first two components in the PCA analysis explained 78% of the total variation in the data, and as the sum was greater than 70% it was considered adequate [42] ( Figure 5A). It can be observed that the deforestation and fire foci in deforestation vectors were close for the years 2008, 2018, 2019, and 2020. This suggests a significant variable contribution and that these years reached a higher index for both variables (Table 1). Moreover, these variables contrasted with the fire foci, burned areas, and fire foci variables in ILs and CUs. This indicates a lower tendency of fire foci and burning in areas with high rainfall. The fire foci vectors in ILs and CUs were close in the years 2010, 2012, and 2015-periods of higher incidence for these variables (Table 1).
Correlations between 0.2 and 0.5 can be considered weak; those from 0.5 to 0.7 have a strong correlation; and those above 0.7 have a very strong correlation [47]. It is possible to observe a strong and positive correlation of 0.8 between fire foci and burned areas, which means that the increase in one influences the increase in another. This is also observed between the burned areas and fire foci in ILs (a strong and positive correlation of 0.7). A strong and negative correlation was observed between fire foci in the CUs and there was an SPI with a magnitude of 0.6.

Discussion
Government-approved or illegal roads developed in the Amazon have often been at the forefront of deforestation [1,48]. This is demonstrated by the BR-319 highway, which has coincided with an increase in deforestation in the surrounding areas. In just 13 years, 743 km 2 of the forest was cleared. From 2009 to 2016, deforestation remained below the average of 57.15 km 2 per year. At the beginning of the study period, in 2008, the deforested area exceeded the interstitial average. Overall, 42% of the forest area in the study region was deforested by 2016. Approximately 58% of the deforestation occurred between 2017 and 2020. This was concentrated mainly in 2019 and 2020 and is approximately 2.2 and 2.6 times larger than the overall average, respectively. This result is due to two factors. First, the "maintenance" of the highway after 2015, which facilitated its access, and second, the Brazilian government's campaign promises to occupy the region and the commitment of providing enforcement agencies [8,9]. One factor to highlight is the importance of protected areas in avoiding deforestation [49][50][51]. More than 95% of deforestation occurred in unprotected areas, while in CUs, this rate was 4.3% and in ILs only 0.3%.
We counted 16,472 fire foci. Of these, 2888 (17.5%) occurred in areas that had been deforested, 832 (5%) in CUs, and only 44 (0.2%) in Lis. This demonstrates the role that protected areas play in mitigating deforestation and fires. The remaining fire foci focused on areas that had already been anthropized since fire is the least expensive way to clear areas and is widely practiced by Amazonian farmers. The average number of fire foci remained high and without statistical significance in all years. It did not vary even in periods of extreme drought or periods of higher rainfall, thus attesting to human interference in the environment. According to the classification determined by Oliveira-Júnior et al. [32], the SPI values in 2009, 2014, 2016, and 2017 were considered normal. For the positive SPI

Discussion
Government-approved or illegal roads developed in the Amazon have often been at the forefront of deforestation [1,48]. This is demonstrated by the BR-319 highway, which has coincided with an increase in deforestation in the surrounding areas. In just 13 years, 743 km 2 of the forest was cleared. From 2009 to 2016, deforestation remained below the average of 57.15 km 2 per year. At the beginning of the study period, in 2008, the deforested area exceeded the interstitial average. Overall, 42% of the forest area in the study region was deforested by 2016. Approximately 58% of the deforestation occurred between 2017 and 2020. This was concentrated mainly in 2019 and 2020 and is approximately 2.2 and 2.6 times larger than the overall average, respectively. This result is due to two factors. First, the "maintenance" of the highway after 2015, which facilitated its access, and second, the Brazilian government's campaign promises to occupy the region and the commitment of providing enforcement agencies [8,9]. One factor to highlight is the importance of protected areas in avoiding deforestation [49][50][51]. More than 95% of deforestation occurred in unprotected areas, while in CUs, this rate was 4.3% and in ILs only 0.3%.
We counted 16,472 fire foci. Of these, 2888 (17.5%) occurred in areas that had been deforested, 832 (5%) in CUs, and only 44 (0.2%) in Lis. This demonstrates the role that protected areas play in mitigating deforestation and fires. The remaining fire foci focused on areas that had already been anthropized since fire is the least expensive way to clear areas and is widely practiced by Amazonian farmers. The average number of fire foci remained high and without statistical significance in all years. It did not vary even in periods of extreme drought or periods of higher rainfall, thus attesting to human interference in the environment. According to the classification determined by Oliveira-Júnior et al. [32], the SPI values in 2009, 2014, 2016, and 2017 were considered normal. For the positive SPI values, 2013 was a year of high rainfall, and 2019 was a year of severe rainfall. As for the negative SPI values, 2008 and 2018 were considered moderate and severe drought years, respectively, while 2010, 2011, 2012, 2015, and 2020 were extreme drought years (Table 1). However, the results of the Mann-Kendall test [40,41] presented only 70% assertiveness, which could be explained by the absence of fire foci in the ILs even with extreme drought, for example, for the years 2011 and 2020. It was clear that the critical SPI values in 2010 (−4.39) and 2015 (−4.86) were also the highest for burned areas, foci of fire, foci of fire in CUs, and ILs of 487 km 2 , 1807, 157, and 6 for the year 2010, and 542 km 2 , 2010, 129, and 6 for the year 2015, respectively. The critical SPI values observed in 2010 and 2015 matched the highest values for burned areas, fire foci, and fire foci in CUs and LIs, possibly explained by the El Niño climate pattern [52,53].
The Amazon is a Brazilian biome with the largest amount of greenhouse gas (GHG) emissions that compromise the climate agreements signed by Brazil [23]. As long as the Brazilian government does not establish a strong national policy to combat deforestation and fire foci throughout its territory, and especially in the Amazon, Brazil will never achieve its assumed climate goals. BR-319 is a duality of the government. The government announced ambitious targets at COP 26; however, it provides opportunity for Amazonia occupation. A disorderly occupation has serious environmental consequences, such as loss of biodiversity and detrimental GHG emissions. Regarding these emissions, the PCA analysis showed the proximity of the deforestation and fire foci in deforestation variables in 2008. This fact can be explained by the re-opening of the BR-319 highway in 2007, under the Lula government. In 2008, the Army began paving 190 km of road near the city of Humaitá, and another 215 km further north, leading to Manaus, which subsequently caused increased deforestation and fire foci. Due to the lack of an environmental license, BR-319 was again abandoned, which explains the decrease in deforestation and fires in subsequent years. The years 2009, 2011, 2013, 2014, and 2016 showed no relationship with any variables, which indicates that fire outbreaks and deforestation were lower for these years. Again, this is most likely attributed to the abandonment of the BR-319, mainly at 250 and 655 km, called the "Middle stretch." The proximity of the variables deforestation and fire foci in deforestation between 2019 and 2020 can be explained by the new environmental policies of the Brazilian government, coupled with the "maintenance" of the highway after 2015.
In situ studies should be pursued in the future to analyze the occupations and land uses in certain highway sections. These data can then be extrapolated with orbital images. In addition, it is necessary to compare specific paved and unpaved areas in relation to environmental impacts, such as fires, loss of native vegetation cover, and the characteristics of regional occupation impacts.

Conclusions
BR-319 is already a vector of deforestation and occupation of Amazonia without being completely paved. Deforestation has been consistent during the mandate of various governments, but 2019 and 2020 revealed significant increases. The highest deforestation and fire foci were observed near the capital cities of Manaus and Porto Velho, particularly following resumed road paving at its northern and southern extremes. Its intermediate section experienced a rapid increase in deforestation after a government plan to repair the road in 2015, but without paving it. CUs and ILs have proven effective in controlling the advancement of deforestation, although 4.3% of deforestation occurred illegally in the CUs, as shown in our study. Regardless of severe droughts or high rainfall, there was a change in the use and occupation of the soil, evidencing anthropic action along the highway. Deforestation and GHG emissions resulting from fire foci in 2019 and 2020 characterize the regional occupation policy intended by the Brazilian government, which greatly contradicts promises made in the climate agreements at COP 26.