Next Article in Journal
Integrating Product Stewardship into the Clothing and Textile Industry: Perspectives of New Zealand Stakeholders
Previous Article in Journal
Numerical Investigation of Recommended Operating Parameters Considering Movement of Polymetallic Nodule Particles during Hydraulic Lifting of Deep-Sea Mining Pipeline
Previous Article in Special Issue
Effects of Different Land Use Types on Soil Surface Temperature in the Heihe River Basin
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

A Worrying Future for River Flows in the Brazilian Cerrado Provoked by Land Use and Climate Changes

by
Yuri Botelho Salmona
1,2,*,
Eraldo Aparecido Trondoli Matricardi
1,*,
David Lewis Skole
3,
João Flávio Andrade Silva
4,
Osmar de Araújo Coelho Filho
5,
Marcos Antonio Pedlowski
6,
James Matos Sampaio
7,
Leidi Cahola Ramírez Castrillón
5,
Reuber Albuquerque Brandão
1,
Andréa Leme da Silva
8 and
Saulo Aires de Souza
9
1
Graduate Program in Forest Sciences, University of Brasilia, Brasília 70900-910, DF, Brazil
2
Instituto Cerrados, Brasília 70737-530, DF, Brazil
3
Global Observatory for Ecosystem Services, Department of Forestry, Michigan State University, East Lansing, MI 48824-6402, USA
4
Interinstitutional Graduate Program in Statistics UFSCar-USP (PIPGEs), Universidade Federal de São Carlos, Rodovia Washington Luís, km 235—SP-310, São Carlos 13565-905, SP, Brazil
5
Graduate Program in Environmental Technologies and Hydric Resources, University of Brasilia, Brasília 70900-910, DF, Brazil
6
Laboratório de Estudos do Espaço Antrópico (LEEA), Centro de Ciências do Homem (CCH), Universidade Estadual do Norte Fluminense Darcy Ribeiro (UENF), Campos dos Goytacazes 28013-602, RJ, Brazil
7
Department of Statistic, University of Brasília, Campus Darcy Ribeiro, Brasília 70900-910, DF, Brazil
8
Graduate Program in Environment and Rural Development, University of Brasília, Brasília 73345-010, DF, Brazil
9
Agência Nacional de Águas—ANA, Brasília 70610-200, DF, Brazil
*
Authors to whom correspondence should be addressed.
Sustainability 2023, 15(5), 4251; https://doi.org/10.3390/su15054251
Submission received: 7 October 2022 / Revised: 24 January 2023 / Accepted: 31 January 2023 / Published: 27 February 2023
(This article belongs to the Special Issue Climate Change, Land Use Change and Water Resources)

Abstract

:
In this study, we assessed the impacts of land use and climate changes on the river flows of 81 watersheds within the Cerrado biome, Brazil, based on a comprehensive analysis of field and secondary data acquired between 1985 and 2018. Complementarily, we projected a future deforestation and climate scenario up to 2050 and predicted their impacts on surface water in the study area. We observed that direct impacts by large-scale deforestation oriented to the production of irrigated agricultural commodities have more significantly impacted river flows than climate changes. We estimated an average decrease of 8.7% and 6.7% in the streamflow due to deforestation and climate changes, respectively. Most of the observed changes (56.7%) were due to land use and land cover changes and occurred in recent decades. Climate and land use and land cover changes combined were responsible for a total surface water reduction of −19,718 m³/s within the Cerrado watersheds. By assuming the current deforestation rates, we predicted a total water reduction of 23,653 m³/s by 2050, equivalent to a decrease of 33.9% of the river flows in the study region. It will cause severe streamflow discontinuity in many rivers and strongly affect agricultural, electric power production, biodiversity, and water supply, especially during dry seasons in that region.

1. Introduction

Areas of extensive agricultural production have been increasingly more vulnerable to the amount of surface water availability due to large-scale deforestation [1], and the effects of climate change [2] are issues demanding a closer look at how much each of those drivers impacts water supplies in tropical regions subjected to high rates of deforestation and climate changes [3].
Different countries have emerged as major exporters of agricultural commodities in recent decades, especially those expanding croplands over large and relatively flat areas with well-defined periods of rain, high suitability to intensive crop production, and agricultural mechanization, such as the case of the Cerrado biome in the central part of Brazil [4,5]. Currently, Brazil is responsible for more than 42% of the soybeans [6] and 20% of the beef consumed in the world [6]. The Cerrado biome plays an important role in the growth of agricultural production in Brazil [7], where 44% (87 M ha) of its territory is occupied by field crops and pastures [8], responsible for 44% of the country’s beef production [9] and for 48% of its soybean exportation [10].
A number of factors are at play: the increasing global demand for agricultural products [11], the rise in internal commodity prices due to the weakening of the Brazilian currency [12], the lack of environmental command and control policies [13], and agricultural occupation policies, such as the one implemented in the MATOPIBA region (acronym for the states of Maranhão, Tocantins, Piauí, and Bahia) have all steadily stimulated the expansion of the agricultural areas in new deforestation frontiers in the Brazilian Amazon and Cerrado regions [14]. The Cerrado itself is the most biodiverse savannah in the world [15]. Unlike the Amazon, prolonged dry seasons and more frequent drought episodes are observed in that region [16,17]. This fluctuation in water availability induces farmers to invest heavily in irrigation systems, leading to the concentration of 80% of the country’s central irrigation systems in that region [18].
Different studies reported changes in river flows (both superficial and groundwater) in the Cerrado biome [19,20,21,22], reporting different variables causing streamflow reduction. The increase in deforestation, irrigated agricultural production [18,21,23], climatic factors such as the decrease in rainfall, and the increase in evapotranspiration [24] are some of the factors discussed. However, detailed hydrological surveys have been restricted to small study sites and time periods, raising questions as to the role of drivers and their implications for future scenarios in surface water changes.
The decrease in river flows in the Cerrado biome has been the center of conflicts over access to water for different uses [25,26], such as water for human consumption, natural ecosystem maintenance, agricultural use, and energy production, especially during years of severe drought episodes. Water scarcity is a growing problem in Brazil due to most of the water reservoirs and hydropower dams within the Brazilian territory being spatially located in the Cerrado biome [27,28]. In addition, this biome is affected by leakages of deforestation effects from neighboring biomes, such as the Brazilian Amazon [29,30] and the Pantanal biomes [31].
River flows are generated by a complex mechanism involving water–energy interactions between the land surface and the atmosphere. Those flows depend on the maintenance of the hydrological cycle, including ecosystem services such as the vegetation function of reducing soil erosion and runoff during rainy seasons, water infiltration in deep and porous soils regulated by the long roots of the Cerrado vegetation, groundwater supply of the base flows during dry seasons [24], and the moisture availability in the atmosphere provided by the processes of evapotranspiration [19].
The hydrological cycles include atmospheric and terrestrial components, which makes it difficult to distinguish the consequences of river flow changes that could be attributed to climatic phenomena, large-scale deforestation [3], and forest degradation. For example, the deforestation effects on rainfall depend on the actual local evapotranspiration [19]. However, direct effects on river flows can be estimated along with their direct causes and drivers [3]. In this analysis, we estimated the effects of changes in potential evapotranspiration, precipitation, and local land uses on streamflow based on a time series of field data for each individual watershed in the study area.
From the climatic point of view, changes in rainfall or potential evapotranspiration can directly affect the river flows [32]. Likewise, the increase in deforestation tends to reduce soil water infiltration [20], evapotranspiration [19], and, consequently, reduce river flow regulation based on groundwater [33] during dry periods. These are crucial elements in seasonal environments such as in the Cerrado biome. Additionally, deforestation can increase water consumption by agricultural irrigation systems, because it demands higher amounts of water during the period of lowest surface water availability and causes even more water scarcity [34]. However, assessing the sustainability of the coupled human–nature system requires the distinction of the potential drivers of surface water changes, which can enable the improvement in water resource utilization [35]. In this analysis, we assessed the impacts of climate and land use changes in the flow of rivers within the Cerrado biome. We analyzed field datasets of 81 watersheds acquired in the last 34 years and developed future scenarios of climate and deforestation changes up to the year 2050. We analyzed their effects, separately and jointly, on the streamflow of the entire study region. Our study results call attention to potential water flow discontinuity soon, which may directly affect agricultural production, electric power production, and water supply in the Cerrado region.

2. Materials and Methods

Our methodological approach consisted of three types of analyses: (I) Time Series Analyses, (II) an Estimate of Contributing Effects, and (III) Future Scenario Analyses of the river flows of 81 watersheds within the Cerrado biome, as shown in Figure 1.

2.1. Time Series Analyses

In the Time Series Analysis, we assessed the contribution from land use changes (Lc), which refer mostly to deforestation to create pastures and agricultural lands, and climate changes (Cc) from historical data series. In addition, we projected estimates of deforestation effects on river flows based on future scenarios of deforestation between now and the year 2050.
Fluviometric and climatic data were collected from stations showing data gap rate of <12% and rainfall data gap rate of less than 7%. In the case of overlapping drainage areas, we selected the smallest one to better detect the direct effects. This process resulted in 81 selected watersheds, covering 30,042,423 hectares, representing 15.1% of the Cerrado biome.
Kalman Filter [36] estimates were used to fill fluviometric gaps, a non-linear procedure widely applied to streamflows [37], as is the case for streamflow. The average precipitation and potential evapotranspiration for each watershed were estimated using the inverse distance weighting (IDW) [38] technique and the data collected by the rainfall stations from the Brazilian national water agency (ANA). The estimates for evapotranspiration, calculated by Penman–Monteith equation, were retrieved from the database of Xavier, A.C., King, C.W., and Scanlon, B.R. (1980–2013) [39].
Climatic variables (rainfall and estimated evapotranspiration) were isolated from land use and cover changes to distinguish from the contribution and effects of Lc and Cc using lasso regression [40]. We then applied the Pettitt test [41] to identify the break pattern in these variables. This allowed us to distinguish two periods: pre-disruption (P1) and post-disruption (P2).
The historical land use data [8] were resampled to 500 m × 500 m grids, clipped by individual watershed boundaries, and grouped into the following classes: forest, savanna, countryside, agriculture, pasture, agricultural mosaic, forestry, urban areas, mining, and water.
Once corrected, the historical data series were analyzed by season and year to check for series stability [42,43], trends [44], break patterns [41,45], and the runoff coefficient (RC), represented by the rate between mean streamflow and precipitation [46].

2.2. Estimate of Contribution and Effects

By assigning an observed break point for each watershed, we assessed the contribution and effects of each driver on each river’s streamflow. The Estimates of Contributor Effects analyses looked at the contribution of land use changes (Lc), climate changes (Cc), and future land change scenarios. Above, we mentioned the climatic variables contributing effects were separated into pre-disruption (P1) and post-pre-disruption (P2) periods. However, assuming the land use and land cover data were kept unchanged, the water flow disruptions in the P1 scenario would be evidenced. The P’ scenario reflects the isolated effect of the climatic variables (Figure 2).
Based on the assigned break points for P1, P2, and P2′, we estimated the contribution and effects of climate variables (Cc), land use, and land cover (Lc) on changes in streamflows in the study area. We estimated the contribution impact (%) by calculating the change a driver had on the observed pre-disturbance river flows (P1). Thus, Q is the effect in m3/s a driver may impact the river flows. The relative change ( Δ Q) is the average difference of river flow pre- and post-disturbance, estimated using the following equations:
Δ   t o t a l = x ¯ p 2 x ¯ p 1 ,
Δ Q L c = x ¯ p 2 x ¯ p 2 ,
Δ Q C c = x ¯ p 2 x ¯   p 1 ,
  C L c = Δ L c / Δ L c + Δ Q C c × 100 ,
C C c % = Δ Q C c / Δ Q L c + Δ Q C c × 100   or   1 C L c % ,
Total relative change (1); effect of changes in land use (Lc) on river flow (m3/s) (2); effects of climate changes (Cc) on river flow (m3/s) (3); contributions of land use changes (Lc) to streamflow (%) (4); contributions of land use changes (Lc) to streamflow (%) (5), where ΔQ is the average variation of streamflow.
We applied a model approach to estimate the river flows under the contributions of Lc and Cc to the P2 post-disruption period and P2′ scenarios. Initially, we applied an empirical hydrological model (type rain–river flow), which explicitly represented the water balance components, mathematically modeled by applying Lasso’s regression [40]. This regression model was chosen based on its simplicity, flexibility, and suitability to integrate data and models. Lasso´s regression model fitting was carried out by decreasing variance, increasing bias, and adding penalties on the absolute values of coefficients, thus reducing the model’s complexity through the elimination of some variables. It simplifies and adjusts the model to those variables that can statistically explain at a significant level the dependent variable, as is the case of the streamflow.
The penalties are controlled by a hyperparameter ( λ ), selected by K-Fold Cross-Validation. This parameter speeds up the individualization of the model, prevents excessively weighing the variables, and avoids overfitting. The self and individualized parametrization, based on individual watersheds, enabled us to apply the flexible regression models, optimizing the group analysis. After normalizing the data, the coefficient of each dependent variable retains some intelligibility, suggesting that each variable contributes to the solution that describes the behavior of the dependent variable (streamflow). The equation that describes Lasso´s regression is presented as follows [47].
i = 1 n y i β 0 j = 1 p β j x i j 2 + λ j = 1 p β j ,
Lasso´s regression equation (least absolute shrinkage and selection operator) (6) and the retraction operator and absolute minimum selection. It shows the sum of squares of the errors added to the sum of the absolute values of the dependent variable parameters, multiplied by the lick hyperparameter.
To generate a function capable of predicting the river flows, this model was based on the definition of weights for each of its coefficients: the climatic variables (rainfall and monthly potential evapotranspiration), months of the year, and land use classes (forest, savanna, grassland, pastures, agriculture, silviculture, agricultural mosaic, urban areas, mining, and bodies of water). In our methodological approach, we were able to integrate the historical time-series dataset, the future climate model, and the spatially explicit model of LULC change.
This approach was applied to each watershed, aiming to determine the best lambda from the K-Fold Cross-Validation with the cv.glmnet function, nfolds = 10. We randomly partitioned the sample (P1) into 10 parts, using 9 parts for training and 1 for validation for all possible combinations, which allowed us to select the most accurate lambda value. Subsequently, we generated 100 models, randomly sampling 25 years at a time, to generate the predictive intervals for each basin. We adopted the mean regression of those models and validated them by applying objective functions [48,49] to compare the observed river flow for period P2 with the river flow modeled for that period.
F 1 = i = 1 n l n Q s i m , i l n Q O b s , i 2 i = 1 n l n Q O b s , i l n Q O b s ¯ 2 ,
F 2 = i = 1 n Q s i m , i Q O b s , i 2 i = 1 n Q O b s , i Q O b s ¯ 2 ,
F 3 = 1 i = 1 n Q s i m , i Q s i m i ¯ Q o b s , i Q O b s ¯ i = 1 n Q s i m , , i Q s i m ¯ 2 i = 1 n Q o b s , , i Q o b s ¯ 2
  F 4 = l n   i = 1 n Q s i m , i i = 1 n Q o b s , i ,
F1 (7) is a logarithmic function and the error weights at low flows rather than the errors at high flows; F2 (8) emphasizes errors at high flows; F3 (9) focuses on time changes between flow rates; F4 (10) evaluates the overall difference between the simulated and observed flow during the calibration period.

2.3. Future Land Use and Climate Change Scenarios

Future estimates of LULC change were carried out using artificial neural network (ANN), available on the Land Change Modeler [50]. The estimates were applied to the land use and land cover dataset from 2015 to 2017 [8] and combined with land use spatial variables. These variables were selected by applying V’Cramer technique for each LC transition when greater than 0.16 or when directly related to the change in question [51]. The neural network uses spatial variables as criteria to search for the most significant land use class transition relationships by learning and replicating the identified major trends (Figure 3).
A large set of land use-related variables was tested with different transformations (e.g., natural log, square root of the distance, and evidence likelihood). The most adherent ones, showing highest explanatory power (V’Cramer), were selected. We also included constraints (weights) to different levels of land use change, such as Permanent preservation areas (APP) [52], legal reserves (RL) [53], integral protection conservation units (UCPI), sustainable use conservation units (UCUS) [54], quilombo areas and indigenous lands. A sample size sufficiently large with respect to the number of pixels in the transition from classes t1 to t2 was necessary so the ANN analysis could identify the main land use change patterns. A large enough sample size could not be found for the minority of classes such as forestry, mining, water, urban, agricultural mosaic, and others since it represented a total area smaller than 2.5% of the entire Cerrado biome. Those class areas were therefore held static, and were not evaluated by ANN.
A set of artificial neurons (neurons) interconnected by a multi-layer perceptron (MLPNN) acquired information about a possible transition of land use pattern, which was then incorporated/processed and later made available to the rest of the neuron network [55]. As the input data fed the nodes with information (signal), the information advanced through the network, accumulating and propagating information from other nodes through the layers until they reached a certain activation state change at the output node layer. A final solution was found by the ANN when the network passed through the following phases: training, incorporation of learning, correction of errors and weights (delta rule), and the calculation of the mean square error (root mean square—RMS).
Initially, we used pixel training samples representing land use change. This procedure helped us identify the relationship between the selected variables and the activation levels of the network’s output nodes. At the end of the process, the pixels with the highest level of activation are represented as those where land changes should be assigned.
Since there is an accumulation of learning during the training phase, the network uses the technique of Reverse Propagation (back propagation), which consists of a forward and a backward step. Changes are made to the state of the nodes in each step, according to the weights assigned by the system. Each sample is fed by the input layer and the weighted signals of all nodes to the receiving node and then connected to the previous layer [52].
Through training, the neural network is able to generate several values (link weights) for the relationships between the variables, which are incorporated into the model (learning), helping it to simulate the process of land use changes [51].
The information learned from the training data was transmitted through the network before propagating errors and weights. The error is then propagated backward, using weights for connections, and corrected based on a relationship known as delta rule.
Repeated back-and-forth iterations were applied until the network had learned the characteristics related to the correct solution for all classes. At the end of the training phase, the network obtains the appropriate weights for the connections among the nodes in the multiple layers, bringing greater accuracy to the output data.
The output data generated represented the change in state according to the nodes with the highest activation level. It is not expected for the output data to be the same as the observed data. It is very likely there were errors in the network causing the difference between the output and the observed data [51,56]. The acceptable error rate is calculated according to the error associated with the network learning, calculated by the errors’ root mean square (RMS).
The land use changes predicted by the ANN (artificial neural network) were distributed using the current trend of deforestation rate. Based on it, we estimated that an average of 6.3 thousand km2/year will be deforested by 2030, an average of 3.4 thousand km2/year between 2030 and 2040, and finally, around 2.5 thousand km²/year between 2040 and 2050.
The land use projection model validation process [57] consisted of verifying the model’s agreement with observed land use for 2018. Subsequently, we observed the agreement between the model prediction and the observed data for that year and from the previous period (2015), which we called the null model [51]. This verification used several indices, including the random agreement—N(n), quantity agreement—N(m), standard agreement—M(m), and adjusted location agreement—P(m), in addition to the standard Kappa and the Kappa for no information (Kno).
The indicators N(m), M(m), and P(m) provide information about the correctness between the reference map (observed data) and the modeled map regarding the location of pixels of each category and the corresponding values by category [57,58].
N(m) refers to the agreement between observed and modeled maps, where pixels were randomly located in the reference map. M(m) indicates the correct portion between the reference and modeled maps, without any change, neither in quantity nor in pixel location. P(m) is the agreement between the reference map and a modified comparison map, where the modification is to rearrange the cell locations within the map as closely as possible to maximize the agreement between the modified modeled map and the reference map [51].
The Kappa index (K) is a measure of agreement between two datasets and measures the degree of agreement beyond what would be expected by chance alone. The K index measure varies from 1 to 0, where 1 represents a total agreement between the datasets, 0 (zero) indicates an expected agreement by chance (random), and below 0 indicates an agreement lesser than expected by chance.
Transition models were generated and validated among the land use and land cover classes, capable of supporting a construction scenario for land use changes during the study period and using above-mentioned validation metrics [57]. This allowed us to project future scenarios of land use changes by applying current deforestation rates up to 2030 and systematically decreasing them between 2030 and 2050 (final year of the projection). The projection fitting was based on two trend scenarios: “Conservation” and “Business as Usual”.
We used the dataset provided by the statistically downscaled earth system models CCSM4 from National Aeronautics and Space Administration’s (NASA) Earth Exchange Global Daily Downscaled Projections (NEX-GDDP) [59] to project the rainfall and potential evapotranspiration. These climate projections showed the best adherence to regional climate patterns of the RCP 4.5 [60,61] scenario, which is considered a medium emission intensity scenario, assuming global stability of carbon emissions. The river flow estimates for the Cerrado biome were modeled and compared for the future P2 post-disruption period and for the P2′ scenarios by using a data series of land use and land cover, potential evapotranspiration, and complete rainfall up to year 2050.

3. Results

The temporal analysis of the river flows in 81 watersheds studied indicated significant (p value < 0.5) rainfall reduction in two watersheds (2.4%); another 46 watersheds showed a non-significant decrease (56.7%); and 28 watersheds (35%) showed a significant increase in rainfall.
During the dry period, 12 watersheds (14.8%) showed significant rainfall reduction. The potential evapotranspiration showed a significant ascending trend for 77 watersheds (95%) and over 75 watersheds (92.6%) during dry periods. We observed a significant average reduction in water flow over 77 watersheds (95%) during the study period and a significant descending trend of water flow over 74 watersheds (91.3%) during dry periods. Only three watersheds (3.7%) showed an ascending trend of water flow. Concerning the average runoff coefficient for watersheds between 1985 and 2018, 57 watersheds (70%) showed a reduction ranging from 3.3 to 2.7 (Figure 4).
The water flow estimate models indicated average accuracies of 99% for the global accuracy (F1), 70% for the dry season (F2), and 70% for the rainy season (F3) and the estimated average adherence to the delay was 84% (Figure 5).
The explanatory power analysis of Cerrado’s model estimates for future land use highlighted the distance from previous deforestation, the travel cost matrix, slope, and soil type, among other variables related to the main land use transitions (Table 1). The accuracy of the land transitions was satisfactory, ranging from 61.01% to 98.04% (Table 2), and the general accuracy indicated that our model was appropriate to carry out the projections for future scenarios (Table 2).
Our land use model allowed us to estimate deforestation rates of approximately 6400 km2/year by 2030, 3400 km2/year by 2040, and 2500 km2/year by 2050. The suggested decreasing trend of projected deforestation rates suggests new croplands could become scarcer in the study area. In this scenario, between 2018 and 2050, the projected native Cerrado vegetation could be reduced from 54.3% to 47.5%, replacing pastures for agricultural lands (Figure 6, Figure 7 and Figure 8 and Table 3).
Our assessment of the contribution and effects of climate drivers (Cc) and Land changes (Lc) between 1985 and 2022 indicates that 71 (87.7%) of the total analyzed watersheds showed a decreasing trend of water flows due to land use and land cover changes, and 69 (85.2%) watersheds showed a decreasing trend of streamflow due to climate effect (Figure 9). On average, the climate change contribution (mainly as potential evapotranspiration) affecting streamflow was −43.3%, while the effects of land use change were −56.7%. Comparing the river flows observed during the pre-disturbance period (P1), the average estimated total effect of land use and land cover changes on river flow was −8.7% and −6.7%, which were attributed to climate changes. The total average reduction in the estimated basin water flows was −19,718 m3/s, of which −11,190 m m³/s was related to the effects of land use and land cover changes and −8528 m³/s was related to the climate change effects.
Based on our predicted future scenarios, we estimated that 75 (92.6%) watersheds would decrease their water flows compared to the pre-disturbance period, where the water flow of 69 (85.1%) watersheds would be decreased due to the effects of the expansion of agricultural lands and the water flow of 71 (87.6%) watersheds would be decreased due to the effects of climate changes (Figure 10).
We estimated an average water flow change of −65.8% attributed to the effects of land use changes and −34.2% attributed to the effects of climate changes. Overall, those drivers tend to affect future water flows of the watersheds in the Cerrado biome generating an average decrease of 33.9%, where 23.4% would be attributed to land use expansion, especially agriculture, and 12.8% would be attributed to climate changes (Figure 11). We predicted a decrease of 23,653 m3/s for analyzed watersheds between 2022 and 2050 compared to the water flows observed during the pre-disturbance period.
Our predicted future scenario indicates decreasing rates of streamflow in most of the watersheds in the study area resulting from the combined effects of climate and land use changes (Figure 9). In other words, based on our future estimates of streamflow, we can predict that if agricultural expansion stagnated in the pre-disturbance period, none of the watersheds would show significant streamflow changes in the future after 2016. Based on currently observed trends of land use, in the future scenario of land use changes (P2), more than 90% of the watersheds showed a significant decrease in streamflow (Figure 10). We observed a few watersheds that showed significant increases in streamflow in the study region. Most of these watersheds were spatially related to fewer land use changes as predicted by the future scenarios.
By comparing observed streamflow observed during pre-disturbance (P1) and post-disturbance (P2) periods under these two different scenarios, (a) zero deforestation in future scenario (P2′) and post-disturbance (P2) and (b) a future scenario of continuing deforestation (P2), we predicted a worse outcome for scenario (b), with the greatest water scarcity due to keeping the current pace of deforestation rates (Figure 12).
We had some emblematic cases illustrating specific contexts. The case where the difference between the P2 and P2′ scenarios is smaller is due to a lesser intense occupation trend (example, Quebra Anzol River). Or the case of areas with combined monoculture and deforestation rates can be attributed to the western state of Bahia, which has been intensely occupied by large-scale monoculture agriculture, mostly soybeans. This area aggregates the municipalities with the highest deforestation rates in the Cerrado biome, and a significant streamflow loss (−24.2%) was attributed to the 54.6% expansion of agricultural lands.
An example of streamflow loss is the Arrojado river, showing intermediate streamflow capacity, which has lost 18.2% of its average streamflow, equivalent to 328.5 m3/s. About 56% of this loss was due to agricultural occupation and 44% due to climate changes. Future trends indicate an accumulated loss of 36.2% of its flow (651.5 m3/s) by 2050. Another example is the Ondas River, which has lost 25% of its average streamflow, 391.5 m3/s, mostly due to the expansion of agricultural lands. For this river, we predicted an accumulated streamflow loss of 56% (864.32 m3/s) by 2050. Lastly, in the Corda River, spatially located in the Northern e Cerrado biome in the state of Maranhão, we observed an average streamflow of 778 m3/s during the initial period of our analysis (P1), decreasing to 570 m3/s in P2 (−27% of the streamflow). The streamflow could have increased to 716 m3/s if no deforestation had occurred in that watershed. Future estimates indicate if the average flow reaches 266 m3/s (−66% of the streamflow), the river will become intermittent, drying out during the dry season by 2047 (Figure 13).

4. Discussion

Our study results of 81 watersheds indicate macro trends of decreasing streamflow in the Cerrado biome due to agricultural expansion and climate changes. From the climatic point of view, the historical series observed in the Cerrado region showed a clear pattern of increase in the potential evapotranspiration associated with a rise of temperature and radiation in the region presented in the results above and in other studies [61,62]. The climate variables studied have become more intensive during dry seasons when the rainfall substantially decreases. On the other hand, annual rainfall showed variable behavior, mostly showing no significant negative trends, such as the studies conducted by [23]. Therefore, we cannot attribute changes in annual precipitation as the leading factor in decreasing streamflow in the Cerrado biome, as previously mentioned in our analysis of driver contributions and effects.
Current scientific knowledge related to the effects of replacing native vegetation with other land use types (e.g., agriculture and pasture) has been controversial [63,64,65,66,67,68]. Nevertheless, most of the current scientific literature indicates that deforestation increases streamflow [65], mainly maximum and mean flows [63,66]. Those studies’ results were affected by several factors such as climate, causes of surface runoff, physiographic characteristics of watersheds, vegetation characteristics (i.e., age, species, and regrowth), landscape characteristics, and their environmental interactions, soil conservation practices, and the type of change in soil conditions [65]. However, the Cerrado biome shows particular physical and biological characteristics that make any assessment of the impact of eventual land use changes very complex and difficult to compare with previous results reported by other scientific studies and analyses carried out in other regions of the planet.
We developed a promising methodological approach to assess the effects of land use, land cover, and climate change by innovating and adjusting available technical approaches. Based on [3,69], we estimated and distinguished the effects of climate change and land use change in several river flows within the Cerrado biome by adjusting those available approaches to the Lasso regression and by combining future climate, land use, and land cover modeling. Our approach substantially contributed to integrating previously developed modeling approaches into more integrated models that enabled us to better understand natural systems. We observed that the period analysis may cause model limitations to capture macro climatic cycles, masking out eventual intra-cycle trends, such as in the following cases: (1) by considering that parts of the Cerrado have been occupied prior to 1985, the streamflow during the pre-disturbance period would have been affected by previous expansion of agricultural or pasture lands expansion, implying an underestimate in land use changes’ effect; (2) the coefficient values of the variables in the lasso modeling may lose part of their intelligibility due to the Cerrado’s climatic seasonality; (3) there are different delays between the interaction of climate variables and land use changes, which may or may not be properly considered for each analyzed basin, reflecting in a loss of the regression accuracy; (4) land use changes observed in other regions may have an impact on our studied watersheds, such as deforestation in the Amazon, by reducing the humidity that reaches the Cerrado region through the flying river phenomena; (5) our model of land use changes assumes that the main arable lands have already been occupied and, therefore, deforestation rates should be lower over the next decades; and (6) our methodological approach has enough elements to indicate the role of each driver in reducing the streamflow of Cerrado rivers, although we have observed model constraints distinguishing some dynamics between climate and land use changes.
Our methodological approach considered the systemic complexity by assuming that each assessed watershed shows some physical and biological characteristics that make it unique. We carried out individual analyses, taking advantage of the available long time-series data available for each watershed, and applied a methodological framework that allowed the integration of the individual results of the watersheds. We applied the necessary adjustments and inferences from the regression model of each of the 81 hydrographically analyzed watersheds. As a result, the fit quality (Figure 5 and Figure 13) of the parameters representing the variability of streamflow indicates a strong adequacy, especially when using the time-series data showing those more pronounced trends. Based on it, we discussed the contribution or “effects” of individual variables used in our analysis.
The statistical regression approach allowed us to distinguish the effects of land use and climate change on the streamflow in the study area. We applied the regression model analyses to assess the relationships among variables using the long time-series data acquired in the field. These models produced values from a sum of contributing variables, where the predicted streamflow was the net sum of all variable contributions. The results of the regression parameters indicated that although the climate is an important factor that affects streamflow, it is no longer the most important in explaining the recently observed decreasing trends of streamflow in the study region. For example, practically all regression models (>95% of the 81 analyzed watersheds) showed at least one significant negative parameter associated with variables related to land use. A total of 70% of the analyzed models showed that the agricultural use parameter was negative, and in more than 50% of those cases, it was a dominant predictor of that decreasing streamflow trend. In addition, pasture and forestry showed a relevant percentage (>50%) of regression models showing a negative parameter.
Although our results are not in agreement with some previously reported results by (19, 20, 22, 33, 97) for the Cerrado region, we argue that our analysis differs from previous work due to the applied space and time scales, the uncertainties associated with data collection, and the quantification of the vegetation changes, which are associated with our applied methodological approach that was able to distinguish effects of climate and LULC change. We also observed that the time-lapse data of the streamflow varied in previous studies. A short streamflow record may fail and only partly capture long-term cumulative hydrological responses following changes in native vegetation cover [70]. For example, in [71], the study area encompassed only a small portion of the Cerrado, and the time-series data of streamflow were analyzed up to 2012, which was a relevant period impact of climate and LULC change in this biome. In the most recent research conducted by [72], which included the entire country of Brazil, their results agree with our findings in the Cerrado biome. [72] used the same empirical statistical context by applying regression model analyses and observed trends toward drier climate conditions and water scarcity in the Cerrado, mainly driven by changes in the climate component. In several regions of Brazil, they observed that the expansion of intensive agricultural lands was the most significant component of streamflow changes, especially in the Cerrado biome.
We call attention to land use changes caused by the expansion of agricultural lands and pastures, which is the main cause of greenhouse gas emissions in Brazil (46%). Agricultural activity alone directly emits a total of 27% of CO2 emissions for the whole country [73]. Therefore, some effects of climate change on surface water availability are derived or originated from agricultural activities. Although there are some relationships between climate and land use changes that cannot be isolated, it did not prevent us from estimating the impacts of deforestation on streamflow.
Ecosystems such as the Cerrado biome, which show a well-defined [16] and high water baseflow contribution to the streamflow during the dry season, deforestation, and land cover conversion from native vegetation to soybean plantations or other types of monoculture, affect a wide chain of hydrological components at medium and long-term time lapses. Those hydrological effects include the reduction in real evapotranspiration [19], reduction in water infiltration capacity and aquifer recharge [28], the increase in surface flow during the rainy period, and the decrease in stream flow during dry periods in the medium and long term. Together they may decrease the ability of aquifer recharge during the rainy season and the ability to sustain high water consumption for crop irrigation during dry seasons [74]. Therefore, it is very likely that the observed increased streamflow in the short-term just after deforestation and at the beginning of agricultural production [20,22] can be considered responses to the reduction in the observed evapotranspiration. In those cases, the streamflow will tend to decrease at medium and long-term time lapses. It is important to note that the trend of increasing potential evapotranspiration in the Cerrado biome reported by [39,62] is related to the climatic parameters only. As for, the studies by [19,20] observed a decrease in real evapotranspiration in environments occupied by agricultural systems.
Consequently, the trends are of equal concern for biodiversity conservation not only for the Cerrado aquatic biota [75], which relies on both tree survival [76] but also on the availability of small streams and lentic habitats in that landscape for survival. The Cerrado borders several other South American biomes, contributing to large continental water basins. The water shortage forecasted by our analysis will likely affect the resilience of those biomes to fire and climate changes. The Pantanal biome, for example, largely depends on the water flow from the Cerrado rivers for its hydroperiod dynamics, and has already experienced unexpected fires that deeply threaten its biota [77].
The extensive and continued expansion of agricultural lands in the Cerrado leads to the conclusion that the production of agricultural commodities and the scarcity of water are intrinsically linked. The increasing export of intensive water-consuming commodities has changed water governance, shifting the water control from local, regional, and national actors to those who dominate global agricultural production chains [78]. Some studies in Latin America have observed that political alliances between the private sector and the state have driven environmental policy reforms that have enabled the concentration of water rights to agro-industrial companies, heavily affecting the social equity and sustainability of water use [78,79,80]. Approximately 80% of central pivots in Brazil are spatially located in the Matopiba region, western Bahia state, where the environmental policy reform after 2012 allowed industrial agriculture to access water supplies in a region showing increasing water scarcity and social conflicts related to water access [81]. Some studies have enhanced the concept of “water justice” combined with demands for an equitable distribution of water access rights and political water decision-making rights [82].
The role that producers, exporters, and consumers of agricultural commodities have is of great concern, as well as the role of definition and enforcement of public policies of command and control in the mitigation of and adaptation to ongoing and future climate changes. In this context, Brazil plays a major role in global agribusiness exportations, reaching more than 120 billion dollars a year exporting agricultural commodities to China, the European Union, and the USA [83].
However, the expansion of agricultural lands has critically affected surface water availability in the Cerrado biome, which may heavily impact the international economy and food security, especially under the recent scenario of weakening environmental policies in Brazil [52,84,85]. The importation policies of several countries that depend on Brazilian agricultural commodities should be appropriately reviewed to prevent the importation of any agricultural commodity related to new deforestation in the Cerrado biome. In this case, environmental impacts, such as increasing water scarcity, carbon emissions, biodiversity lost, etc., should be commercially shared to those importers and informed to consumers.
The future trend scenario includes a worsening of surface water availability in almost the entire Cerrado biome. It will require rigorous evaluation and monitoring of farming production, which is responsible for 78.3% of surface water consumption in Brazil [86] and 96.6% of deforestation in the Cerrado biome [8]. In tune with other studies we have shown, the Cerrado Biome acts as an exporter of virtual water [78,87,88] attached to the exported agricultural commodities from Brazil. The high water consumption required to sustain agricultural commodity exports is mainly responsible for increasing surface water scarcity to supply human livelihood and biological processes in the study region [89]. There is a perception that the agricultural producers are “shooting themselves on the foot” by destroying important ecosystem services, which ultimately will affect both environmental and economic sustainability and the water supply, which has been increasingly critical to maintain local communities in the Cerrado biome [89].
Therefore, any expansion of agricultural lands in the Cerrado biome should be properly and rigorously evaluated based on the water requirements of each user. It would also require strengthening participation in the water management process (e.g., watershed committees), aiming to prevent unequal or overloaded water consumption in the Cerrado biome. We foresee that legal reserves on private properties and other protected areas can substantially contribute to the conservation of water resources and limit the expansion of agricultural lands by preserving a native vegetation [90] quota per watershed capable of maintaining water flows at satisfactory levels.

5. Conclusions

Our results suggest that land use and climate changes have directly affected surface water availability in the Cerrado biome in Brazil, and this is likely to be aggravated in the coming years, assuming the future land use and climate changes used in this study. The same results suggest the trends in our scenarios will threaten agricultural production, public and private water supplies, and energy supply, as was experienced in Central Brazil in 2021.
Future scenarios indicate that more than 90% of watersheds within the Cerrado biome tend to decrease their water flow due to the expansion of agricultural lands. This may lead to recurrent critical levels of water scarcity during dry periods in the study region. The watersheds spatially located in the west of Bahia state, for example, showed clear decreasing trends and significant reductions in river flows, which will potentially worsen surface water availability, especially during dry periods, as found in some study cases [18].
Our analysis indicates we are embracing an uncertain future of surface water availability in the Cerrado biome. Most of the studied watersheds are subjected to high rates of deforestation linked to the equally high expansion of agricultural lands. For the medium and long term, we predict future critical levels of surface water supply and intensification in the current water conflicts [26] in the medium and long term. We conclude that Cerrado’s water capacity has been substantially weakened by agricultural activities and climate changes. Yet, the Cerrado biome has been considered “Brazil’s water reservoir”, although it is now heavily “leaking” and showing future scenarios of brief water scarcity.
Stakeholders overusing water resources, or those with insufficient access to water in the study region, will be directly affected by a decreasing surface water availability, and will be challenged to find alternatives to sustainable use of water and land. It is critical to heed nature’s warnings and reorient processes of land occupation and production in the Cerrado biome. Improving the production chains of non-timber forest products and production systems more adapted to local conditions could substantially contribute to protecting the native vegetation in the Cerrado biome.
We concluded that land use changes and the additional effects of climate change are impacting water availability in the Cerrado biome. Climate and land use changes are closely linked to the international consumption of certain agricultural commodities, which ultimately impacts local communities and increases water conflicts at local and regional levels. Finally, our study results can support the definition of political strategies and the legal measures needed to mitigate the negative effects of water overuse, and avoid an overwhelming expansion of agricultural lands into Cerrado.

Author Contributions

Conceptualization, Y.B.S.; methodology and validation Y.B.S., J.F.A.S. and J.M.S.; formal analysis, Y.B.S., S.A.d.S., L.C.R.C., D.L.S., E.A.T.M. and M.A.P.; investigation, Y.B.S.; Resources, Y.B.S. and J.F.A.S.; data curation, S.A.d.S. and Y.B.S.; writing—preparation of the original draft, Y.B.S.; writing—review and editing, D.L.S., E.A.T.M., R.A.B., A.L.d.S., O.d.A.C.F. and M.A.P. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Institute for Society, Population and Nature—ISPN and was funded in 2019.

Informed Consent Statement

Not applicable.

Data Availability Statement

Os principais dados utilizados nesta pesquisa estão disponíveis no Hidroweb, Mapbiomas e NASA Center for Climate Simulation (NCCS).

Acknowledgments

We thank the Institute for Society, Population and Nature—ISPN; the Department of Forest Sciences at the University of Brasília; the Global Observatory for Ecosystem Services Lab, Department of Forestry, Michigan State University; and Climate scenarios used, which were from the NEX-GDDP dataset, prepared by the Climate Analytics Group and NASA Ames Research Center using the NASA Earth Exchange, and distributed by the NASA Center for Climate Simulation (NCCS), for the support that made this study possible. Skole acknowledges the NASA LCLUC Program for funding support, NASA Grant 80NSSC20K0263, and NASA Grant 80NSSC21K0315 to Michigan State University. RAB thanks CNPq for productivity grant (process # 306644/2020-7) We wish to acknowledge Katie James for assistance with international coordination. The authors express their gratitude to Suzana Müller for her total and careful review of the manuscript.

Conflicts of Interest

The authors declare no conflict of interest. The funders had no role in the design of this study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

References

  1. Hoekstra, A.Y.; Hung, P.Q. Virtual Water Trade a Quantification of Virtual Water Flows between Nations in Relation to International Crop Trade; IHE Delft: Delft, The Netherlands, 2002; pp. 1–120. Available online: https://waterfootprint.org/media/downloads/Report11.pdf (accessed on 10 January 2020).
  2. Langenbrunner, B. Water, water not everywhere. Nat. Clim. Chang. 2021, 11, 650. [Google Scholar] [CrossRef]
  3. Ashok Mishra, P.D. Separating the impacts of climate change and human activities on streamflow: A review of methodologies and critical assumptions. J. Hydrol. 2017, 548, 278–290. [Google Scholar]
  4. Sano, E.E.; Rosa, R.; Brito, J.L.S.; Ferreira, L.G. Mapeamento semidetalhado do uso da terra do Bioma Cerrado. Pesqui. Agropecu. Bras. 2008, 43, 153–156. [Google Scholar] [CrossRef]
  5. Denis Castilho, M.P. Cerrados Perspectivas e Olhares; Vieira: Goiânia, Brazil, 2010. [Google Scholar]
  6. Trase Yearbook 2020—The State of Forest Risk Supply Chains, Brazilian Soybean. 2020. Available online: https://insights.trase.earth/yearbook/contexts/brazil-soy/ (accessed on 5 July 2021).
  7. Jepson, W.; Brannstrom, C.; Filippi, A. Access Regimes and Regional Land Change in the Brazilian Cerrado, 1972–2002. Ann. Assoc. Am. Geogr. 2010, 100, 87–111. [Google Scholar] [CrossRef]
  8. Mapbiomas Collection 4. 2020. Available online: https://mapbiomas.org/ (accessed on 1 July 2020).
  9. Trase. Trase Yearbook 2020—The State of Forest Risk Supply Chains, Brazilian Beef. 2020. Available online: https://insights.trase.earth/yearbook/contexts/brazil-beef (accessed on 20 November 2020).
  10. IBGE. Produção Agrícola Municipal. 2020. Available online: https://sidra.ibge.gov.br/pesquisa/pam/tabelas (accessed on 2 November 2020).
  11. FAO. World Agriculture: Towards 2015/2030. 2002. Available online: https://www.fao.org/3/y3557e/y3557e06.htm (accessed on 19 November 2019).
  12. IPEA. Boletim de Taxa de Cambio. 2022. Available online: http://www.ipeadata.gov.br/ExibeSerie.aspx?serid=31924 (accessed on 29 July 2022).
  13. Pereira, E.J.D.A.L.; Ribeiro, L.C.D.S.; Freitas, L.F.D.S.; Pereira, H.B.D.B. Brazilian policy and agribusiness damage the Amazon rainforest. Land Use Policy 2020, 92, 104491. [Google Scholar] [CrossRef]
  14. Sauer, S. Soy expansion into the agricultural frontiers of the Brazilian Amazon: The agribusiness economy and its social and environmental conflicts. Land Use Policy 2018, 79, 326–338. [Google Scholar] [CrossRef]
  15. Myers, N.; Mittermeier, R.A.; Mittermeier, C.G.; da Fonseca, G.A.B.; Kent, J. Biodiversity hotspots for conservation priorities. Nature 2000, 403, 853–858. [Google Scholar] [CrossRef]
  16. Sano, S.M.; Almeida, S.P.d.A.; Ribeiro, J.F. Cerrado: Ecologia e Flora; Embrapa: Brasília, Brazil, 2008; Volume 1. [Google Scholar]
  17. Almagro, A.; Oliveira, P.T.S.; Neto, A.A.M.; Roy, T.; Troch, P. CABra: A novel large-sample dataset for Brazilian catchments. Hydrol. Earth Syst. Sci. 2021, 25, 3105–3135. [Google Scholar] [CrossRef]
  18. Silva, A.; Souza, S.; Filho, O.C.; Eloy, L.; Salmona, Y.; Passos, C. Water Appropriation on the Agricultural Frontier in Western Bahia and Its Contribution to Streamflow Reduction: Revisiting the Debate in the Brazilian Cerrado. Water 2021, 13, 1054. [Google Scholar] [CrossRef]
  19. Spera, S.A.; Galford, G.L.; Coe, M.T.; Macedo, M.N.; Mustard, J.F. Land-use change affects water recycling in Brazil’s last agricultural frontier. Glob. Chang. Biol. 2016, 22, 3405–3413. [Google Scholar] [CrossRef]
  20. Nóbrega, R.L.B.; Guzha, A.C.; Torres, G.N.; Kovacs, K.; Lamparter, G.; Amorim, R.S.S.; Couto, E.; Gerold, G. Effects of conversion of native cerrado vegetation to pasture on soil hydro-physical properties, evapotranspiration and streamflow on the Amazonian agricultural frontier. PLoS ONE 2017, 12, e0179414. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  21. Gonçalves, R.D.; Engelbrecht, B.Z.; Chang, H.K. Análise hidrológica de séries históricas da Bacia do Rio Grande (BA): Contribuição do Sistema Aquífero Urucuia. Águas Subterrâneas 2016, 30, 190. [Google Scholar] [CrossRef] [Green Version]
  22. Dias, L.C.P.; Macedo, M.N.; Costa, M.H.; Coe, M.T.; Neill, C. Effects of land cover change on evapotranspiration and streamflow of small catchments in the Upper Xingu River Basin, Central Brazil. J. Hydrol. Reg. Stud. 2015, 4, 108–122. [Google Scholar] [CrossRef] [Green Version]
  23. Marengo, J.A.; Alves, L.M. Tendências hidrológicas da bacia do rio Paraíba do Sul. Rev. Bras. Meteorol. 2005, 20, 215–226. [Google Scholar]
  24. Pousa, R.; Costa, M.H.; Pimenta, F.M.; Fontes, V.C.; De Brito, V.F.A.; Castro, M. Climate Change and Intense Irrigation Growth in Western Bahia, Brazil: The Urgent Need for Hydroclimatic Monitoring. Water 2019, 11, 933. [Google Scholar] [CrossRef] [Green Version]
  25. Favareto, A.; Nakagawa, L.; Pó, M.; Seifer, P.; Kleeb, S. Entre Chapadas e Baixões do Matopiba: Dinâmicas Territoriais e Impactos Socioeconômicos Na Fronteira da Expansão Agropecuária. Prefixo Editor. 2019, 8, 92545. [Google Scholar]
  26. Os Pivôs da Discórdia e a Digna Raiva: Uma Análise dos Conflitos por Terra, Água e Território em Correntina—BA. Doc. De Trab. Inédito 2018.
  27. Strassburg, B.B.N.; Brooks, T.; Feltran-Barbieri, R.; Iribarrem, A.; Crouzeilles, R.; Loyola, R.; Latawiec, A.E.; Filho, F.J.B.O.; Scaramuzza, C.A.D.M.; Scarano, F.R.; et al. Moment of truth for the Cerrado hotspot. Nat. Ecol. Evol. 2017, 1, 99. [Google Scholar] [CrossRef]
  28. Oliveira, P.T.S.; Wendland, E.; Nearing, M.A.; Scott, R.L.; Rosolem, R.; da Rocha, H.R. The water balance components of undisturbed tropical woodlands in the Brazilian cerrado. Hydrol. Earth Syst. Sci. 2015, 19, 2899–2910. [Google Scholar] [CrossRef] [Green Version]
  29. Marengo, J.A.; Jimenez, J.C.; Espinoza, J.-C.; Cunha, A.P.; Aragão, L.E.O. Increased climate pressure on the agricultural frontier in the Eastern Amazonia–Cerrado transition zone. Sci. Rep. 2022, 12, 457. [Google Scholar] [CrossRef]
  30. Soterroni, A.C.; Ramos, F.M.; Mosnier, A.; Fargione, J.; Andrade, P.R.; Baumgarten, L.; Pirker, J.; Obersteiner, M.; Kraxner, F.; Câmara, G.; et al. Expanding the Soy Moratorium to Brazil’s Cerrado. Sci. Adv. 2019, 5, eaav7336. [Google Scholar] [CrossRef] [Green Version]
  31. Timpe, K.; Kaplan, D. The changing hydrology of a dammed Amazon. Sci. Adv. 2017, 3, e1700611. [Google Scholar] [CrossRef] [Green Version]
  32. Arora, V.K. The use of the aridity index to assess climate change effect on annual runoff. J. Hydrol. 2002, 265, 164–177. [Google Scholar] [CrossRef]
  33. Luo, X.; Li, H.-Y.; Leung, L.R.; Tesfa, T.K.; Getirana, A.; Papa, F.; Hess, L.L. Modeling surface water dynamics in the Amazon Basin using MOSART-Inundation v1.0: Impacts of geomorphological parameters and river flow representation. Geosci. Model Dev. 2017, 10, 1233–1259. [Google Scholar] [CrossRef] [Green Version]
  34. Calijuri, M.L.; Castro, J.D.S.; Costa, L.S.; Assemany, P.P.; Alves, J.E.M. Impact of land use/land cover changes on water quality and hydrological behavior of an agricultural subwatershed. Environ. Earth Sci. 2015, 74, 5373–5382. [Google Scholar] [CrossRef]
  35. Li, C.; Wang, L.; Wanrui, W.; Qi, J.; Linshan, Y.; Zhang, Y.; Lei, W.; Cui, X.; Wang, P. An analytical approach to separate climate and human contributions to basin streamflow variability. J. Hydrol. 2018, 559, 30–42. [Google Scholar] [CrossRef]
  36. Kalman, R.E. A New Approach to Linear Filtering and Prediction Problems. J. Basic Eng. 1960, 82, 35–45. [Google Scholar] [CrossRef] [Green Version]
  37. Chui, C.K.; Chen, G. Kalman Filtering with Real-Time Applications; Springer: Berlin, Germany, 2009. [Google Scholar]
  38. Li, J.; Heap, A.D. A review of comparative studies of spatial interpolation methods in environmental sciences: Performance and impact factors. Ecol. Inform. 2011, 6, 228–241. [Google Scholar] [CrossRef]
  39. Xavier, A.C.; King, C.W.; Scanlon, B.R. Daily gridded meteorological variables in Brazil (1980–2013). Int. J. Climatol. 2016, 36, 2644–2659. [Google Scholar] [CrossRef] [Green Version]
  40. Tibshirani, R. Regression Shrinkage and Selection Via the Lasso. J. R. Stat. Soc. Ser. B Methodol. 1996, 58, 267–288. [Google Scholar] [CrossRef]
  41. Pettitt, A.N. A Non-Parametric Approach to the Change-Point Problem. J. R. Stat. Soc. Ser. C (Appl. Stat.) 1979, 28, 126–135. [Google Scholar] [CrossRef]
  42. Bradley, J.V. Distribution-Free Statistical Tests; United States Air Force: Washington, DC, USA, 1968. [Google Scholar]
  43. Thom, H.C.S. Same Methods of Climatological Analyses; World Meteorological Organization: Geneva, Switzerland, 1966; p. 16. [Google Scholar]
  44. Sen, P.K. Estimates of the Regression Coefficient Based on Kendall’s Tau. J. Am. Stat. Assoc. 1968, 63, 1379–1389. [Google Scholar] [CrossRef]
  45. Rosin, C.; Amorim, R.; Morais, T. Análise de tendências hidrológicas na bacia do rio das Mortes/analysis of hydrological trends in the Rio das Mortes watershed. Rev. Bras. Recur. Hídricos 2015, 20, 991–998. [Google Scholar] [CrossRef]
  46. Sriwongsitanon, N.; Taesombat, W. Effects of land cover on runoff coefficient. J. Hydrol. 2011, 410, 226–238. [Google Scholar] [CrossRef]
  47. Hastie, T.; Tibshirani, R.; Friedman, J.H. The Elements of Statistical Learning: Data Mining, Inference, and Prediction; Springer: Berlin/Heidelberg, Germany, 2009. [Google Scholar]
  48. Duan, Q.; Sorooshian, S.; Gupta, V. Effective and efficient global optimization for conceptual rainfall-runoff models. Water Resour. Res. 1992, 28, 1015–1031. [Google Scholar] [CrossRef]
  49. Zhang, J.; Zhang, Y.; Song, J.; Cheng, L. Evaluating relative merits of four baseflow separation methods in Eastern Australia. J. Hydrol. 2017, 549, 252–263. [Google Scholar] [CrossRef]
  50. Clark Labs. The Land Change Modeler for Ecological Sustainability. IDRISI Focus Pap. 2007, 2, 234–256. [Google Scholar]
  51. Eastman, J.R. IDRISI Taiga Guide to GIS and Image Processing; Clark Labs Clark University: Worcester, MA, USA, 2009. [Google Scholar]
  52. Rajão, R.; Nobre, A.D.; Cunha, E.L.; Duarte, T.R.; Marcolino, C.; Soares-Filho, B.; Sparovek, G.; Rodrigues, R.R.; Valera, C.; Bustamante, M.; et al. The risk of fake controversies for Brazilian environmental policies. Biol. Conserv. 2022, 266, 109447. [Google Scholar] [CrossRef]
  53. Azevedo, A.A.; Saito, C.H. O perfil dos desmatamentos em Mato Grosso, após implementação do licenciamento ambiental em propriedades rurais. CERNE 2013, 19, 111–122. [Google Scholar] [CrossRef] [Green Version]
  54. Françoso, R.D.; Brandão, R.; Nogueira, C.C.; Salmona, Y.B.; Machado, R.B.; Colli, G.R. Habitat loss and the effectiveness of protected areas in the Cerrado Biodiversity Hotspot. Nat. Conserv. 2015, 13, 35–40. [Google Scholar] [CrossRef] [Green Version]
  55. Clark Labs. Land Change Modeler: An Application within IDRISI for Analyzing and Predicting Land Cover Change and Assessing the Implications of that Change for Biodiversity. Available online: https://clarklabs.org/terrset/land-change-modeler/ (accessed on 30 January 2023).
  56. Almeida, C.M.; Gleriani, J.M.; Castejon, E.F.; Soares-Filho, B.S. Using neural networks and cellular automata for modelling intra-urban land-use dynamics. Int. J. Geogr. Inf. Sci. 2008, 22, 943–963. [Google Scholar] [CrossRef]
  57. Pontius, R.; Huffaker, D.; Denman, K. Useful techniques of validation for spatially explicit land-change models. Ecol. Model. 2004, 179, 445–461. [Google Scholar] [CrossRef]
  58. Pontius Júnior, R.G.; Chen, H. GEOMOD Modeling Land-Use & Cover Change Modeling; Clark University: Worcester, MA, USA, 2006; 44p. [Google Scholar]
  59. Thrasher, B.; Maurer, E.P.; McKellar, C.; Duffy, P.B. Technical Note: Bias correcting climate model simulated daily temperature extremes with quantile mapping. Hydrol. Earth Syst. Sci. 2012, 16, 3309–3314. [Google Scholar] [CrossRef] [Green Version]
  60. Avila-Diaz, A.; Benezoli, V.; Justino, F.; Torres, R.; Wilson, A. Assessing current and future trends of climate extremes across Brazil based on reanalyses and earth system model projections. Clim. Dyn. 2020, 55, 1403–1426. [Google Scholar] [CrossRef]
  61. Avila-Diaz, A.; Abrahão, G.; Justino, F.; Torres, R.; Wilson, A. Extreme climate indices in Brazil: Evaluation of downscaled earth system models at high horizontal resolution. Clim. Dyn. 2020, 54, 5065–5088. [Google Scholar] [CrossRef]
  62. Salviano, M.F.; Groppo, J.D.; Pellegrino, G.Q. Análise de Tendências em Dados de Precipitação e Temperatura no Brasil. Rev. Bras. Meteorol. 2016, 31, 64–73. [Google Scholar] [CrossRef] [Green Version]
  63. Patterson, L.A.; Lutz, B.; Doyle, M.W. Climate and direct human contributions to changes in mean annual streamflow in the South Atlantic, USA. Water Resour. Res. 2013, 49, 7278–7291. [Google Scholar] [CrossRef]
  64. Zhou, G.; Wei, X.; Chen, X.; Zhou, P.; Liu, X.; Xiao, Y.; Sun, G.; Scott, D.F.; Zhou, S.; Han, L.; et al. Global pattern for the effect of climate and land cover on water yield. Nat. Commun. 2015, 6, 5918. [Google Scholar] [CrossRef] [Green Version]
  65. Zhang, M.; Liu, N.; Harper, R.; Li, Q.; Liu, K.; Wei, X.; Ning, D.; Hou, Y.; Liu, S. A global review on hydrological responses to forest change across multiple spatial scales: Importance of scale, climate, forest type and hydrological regime. J. Hydrol. 2017, 546, 44–59. [Google Scholar] [CrossRef] [Green Version]
  66. Creed, I.F.; Noordwijk, M. Forest and Water on a Changing Planet: Vulnerability, Adaptation and Governance Opportunities; International Union of Forestry Research Organizations—IUFRO: Viena, Austria, 2018; Volume 38, 188p, Available online: https://www.cabdirect.org/cabdirect/abstract/20193124755 (accessed on 12 November 2020).
  67. Coble, A.A.; Barnard, H.; Du, E.; Johnson, S.; Jones, J.; Keppeler, E.; Kwon, H.; Link, T.E.; Penaluna, B.E.; Reiter, M.; et al. Long-term hydrological response to forest harvest during seasonal low flow: Potential implications for current forest practices. Sci. Total. Environ. 2020, 730, 138926. [Google Scholar] [CrossRef]
  68. Wierik, S.A.T.; Cammeraat, E.L.H.; Gupta, J.; Artzy-Randrup, Y.A. Reviewing the Impact of Land Use and Land-Use Change on Moisture Recycling and Precipitation Patterns. Water Resour. Res. 2021, 57, e2020WR029234. [Google Scholar] [CrossRef]
  69. Zhao, F.; Zhang, L.; Xu, Z.; Scott, D.F. Evaluation of methods for estimating the effects of vegetation change and climate variability on streamflow: Effects of vegetation and climate variability. Water Resour. Res. 2010, 46, W03505. [Google Scholar] [CrossRef]
  70. Brown, A.E.; Zhang, L.; McMahon, T.A.; Western, A.W.; Vertessy, R.A. A review of paired catchment studies for determining changes in water yield resulting from alterations in vegetation. J. Hydrol. 2005, 310, 28–61. [Google Scholar] [CrossRef]
  71. Levy, M.C.; Lopes, A.V.; Cohn, A.; Larsen, L.G.; Thompson, S.E. Land Use Change Increases Streamflow Across the Arc of Deforestation in Brazil. Geophys. Res. Lett. 2018, 45, 3520–3530. [Google Scholar] [CrossRef]
  72. Chagas, V.B.P.; Chaffe, P.L.B.; Blöschl, G. Climate and land management accelerate the Brazilian water cycle. Nat. Commun. 2022, 13, 5136. [Google Scholar] [CrossRef]
  73. Potenza, R.; Alencar, A.; Azevedo, T.; Shimbo, J. Análise das Emissões Brasileiras de e Suas Implicações para as Metas Climáticas do Brasil 1970–2020 Gases de Efeito Estufa. 2022. Available online: https://seeg-br.s3.amazonaws.com/Documentos%20Analiticos/SEEG_9/OC_03_relatorio_2021_FINAL.pdf (accessed on 15 November 2022).
  74. Oliveira, R.S.; Bezerra, L.; Davidson, E.; Pinto, F.; Klink, C.A.; Nepstad, D.C.; Moreira, A. Deep root function in soil water dynamics in cerrado savannas of central Brazil. Funct. Ecol. 2005, 19, 574–581. [Google Scholar] [CrossRef]
  75. Brasil, L.S.; Ferreira, V.R.S.; de Resende, B.O.; Juen, L.; Batista, J.D.; de Castro, L.A.; Giehl, N.F.D.S. Dams Change Beta Diversity of Aquatic Communities in the Veredas of the Brazilian Cerrado. Front. Ecol. Evol. 2021, 9, 612642. [Google Scholar] [CrossRef]
  76. Brasil, L.S.; Juen, L.; Cabette, H.S.R. The effects of environmental integrity on the diversity of mayflies, Leptophlebiidae (Ephemeroptera), in tropical streams of the Brazilian Cerrado. Ann. Limnol.-Int. J. Limnol. 2014, 50, 325–334. [Google Scholar] [CrossRef]
  77. Tomas, W.M.; Berlinck, C.N.; Chiaravalloti, R.M.; Faggioni, G.P.; Strüssmann, C.; Libonati, R.; Abrahão, C.R.; Alvarenga, G.D.V.; Bacellar, A.E.D.F.; Batista, F.R.D.Q.; et al. Distance sampling surveys reveal 17 million vertebrates directly killed by the 2020′s wildfires in the Pantanal, Brazil. Sci. Rep. 2021, 11, 23547. [Google Scholar] [CrossRef]
  78. Vos, J.; Hinojosa, L. Virtual water trade and the contestation of hydrosocial territories. Water Int. 2016, 41, 37–53. [Google Scholar] [CrossRef] [Green Version]
  79. Velez-Torres, I. Water Grabbing in the Cauca Basin: The Capitalist Exploitation of Water and Dispossession of Afro-Descendant Communities. Water Altern. 2012, 5, 431–449. [Google Scholar]
  80. López, R.R.; Hoogendam, P.; Vos, J.; Boelens, R. Transforming hydrosocial territories and changing languages of water rights legitimation: Irrigation development in Bolivia’s Pucara watershed. Geoforum 2019, 102, 202–213. [Google Scholar] [CrossRef]
  81. Leme, A.S.; Eloy, L.; Coelho Filho, O.; Santos, M.R.B. Environmental policy reform and water grabbing in an agricultural frontier in the Brazilian Cerrado. In Frontier Territories: Countering the Green Revolution Legacy in the Brazilian Cerrado; Cabral, L., Sauer, S., Shankland, A., Eds.; in press; Available online: https://bulletin.ids.ac.uk/index.php/idsbo/article/view/3194/3261 (accessed on 15 November 2022).
  82. Boelens, R.; Hoogesteger, J.; Swyngedouw, E.; Vos, J.; Wester, P. Hydrosocial territories: A political ecology perspective. Water Int. 2016, 41, 1–14. [Google Scholar] [CrossRef] [Green Version]
  83. AGROSTAT—Estatísticas de Comércio Exterior do Agronegócio Brasileiro. 2022. Available online: https://indicadores.agricultura.gov.br/agrostat/index.htm (accessed on 17 November 2022).
  84. Barbosa, L.G.; Alves, M.A.S.; Grelle, C.E.V. Actions against sustainability: Dismantling of the environmental policies in Brazil. Land Use Policy 2021, 104, 105384. [Google Scholar] [CrossRef]
  85. Milhorance, C. Policy dismantling and democratic regression in Brazil under Bolsonaro: Coalition politics, ideas, and underlying discourses. Rev. Policy Res. 2022, 39, 752–770. [Google Scholar] [CrossRef]
  86. de Águas, A.N. Conjuntura dos Recursos Hídricos No Brasil 2018: Informe Annual; Versão Atualizada: Brasília, Brazil, 2019; p. 72. Available online: https://www.snirh.gov.br/portal/centrais-de-conteudos/conjuntura-dos-recursos-hidricos/informe_conjuntura_2018.pdf (accessed on 15 November 2022).
  87. Allan, J.A. Virtual Water: A Strategic Resource Global Solutions to Regional Deficits. Groundwater 1998, 36, 545–546. [Google Scholar] [CrossRef]
  88. Oliveira, G.; Hecht, S. Sacred groves, sacrifice zones and soy production: Globalization, intensification and neo-nature in South America. J. Peasant Stud. 2016, 43, 251–285. [Google Scholar] [CrossRef] [Green Version]
  89. FAO. The Future of Food and Agriculture. Trends and Challenges; Food and Agriculture Organization of the United Nations: Rome, Italy, 2017. [Google Scholar]
  90. Rausch, L.L.; Gibbs, H.K.; Schelly, I.; Jr, A.B.; Morton, D.C.; Filho, A.C.; Strassburg, B.; Walker, N.; Noojipady, P.; Barreto, P.; et al. Soy expansion in Brazil’s Cerrado. Conserv. Lett. 2019, 12, e12671. [Google Scholar] [CrossRef]
Figure 1. Methodological approach to estimate future effects of land use and climate change on the river flows in the Cerrado biome: Time Series Analyses, Estimates of Contributions and Effects, and Future Scenario Analyses.
Figure 1. Methodological approach to estimate future effects of land use and climate change on the river flows in the Cerrado biome: Time Series Analyses, Estimates of Contributions and Effects, and Future Scenario Analyses.
Sustainability 15 04251 g001
Figure 2. Components of each scenario, namely, P1, pre-disturbance period: the observed water flows previously to the rupture pattern of water flows; P2, post-disturbance period: the observed water flows during the period after the rupture pattern of water flows by considering variables of land use and land cover (LULC) changes and the observed climate changes and predicted future trends of land use and climate changes; P2′, predicted disturbance period: the river flow modeled assuming a scenario of fixed land use variables on the detected date using the Pettitt test.
Figure 2. Components of each scenario, namely, P1, pre-disturbance period: the observed water flows previously to the rupture pattern of water flows; P2, post-disturbance period: the observed water flows during the period after the rupture pattern of water flows by considering variables of land use and land cover (LULC) changes and the observed climate changes and predicted future trends of land use and climate changes; P2′, predicted disturbance period: the river flow modeled assuming a scenario of fixed land use variables on the detected date using the Pettitt test.
Sustainability 15 04251 g002
Figure 3. Methodological approach for predicting LULC change using artificial neural networks (ANN).
Figure 3. Methodological approach for predicting LULC change using artificial neural networks (ANN).
Sustainability 15 04251 g003
Figure 4. Boxplot showing the Theil–Sen slope trend analysis of 81 watersheds, post-trend-free pre-whitening (TFPW) technique for autocorrelation correction test to support the application of the trend test significant Mann–Kendall approach to assess variations of evapotranspiration (mm), streamflow (m3/s), and rainfall (m3/s) data. The slope of the linear regression of the annual indices was applied to assess the runoff index.
Figure 4. Boxplot showing the Theil–Sen slope trend analysis of 81 watersheds, post-trend-free pre-whitening (TFPW) technique for autocorrelation correction test to support the application of the trend test significant Mann–Kendall approach to assess variations of evapotranspiration (mm), streamflow (m3/s), and rainfall (m3/s) data. The slope of the linear regression of the annual indices was applied to assess the runoff index.
Sustainability 15 04251 g004
Figure 5. Boxplot of accuracy assessment of the regression models of 81 watersheds in the study area. Global accuracy verifies how similar the projected and observed water flows are. The global estimate (F1) indicates the global accuracy, F2 the dry season, F3 the rainy season, and F4 the mismatch or delay. The graphs show the percentage hit, where 1 indicates the total correct and 0 indicates the total error.
Figure 5. Boxplot of accuracy assessment of the regression models of 81 watersheds in the study area. Global accuracy verifies how similar the projected and observed water flows are. The global estimate (F1) indicates the global accuracy, F2 the dry season, F3 the rainy season, and F4 the mismatch or delay. The graphs show the percentage hit, where 1 indicates the total correct and 0 indicates the total error.
Sustainability 15 04251 g005
Figure 6. Changes in land use and land cover observed from 2015 to 2020 and predicted until 2050. Highlight for the reduction of native vegetation up to 47.5% in 2050. The categories with smaller area in the biome appear expanded in the graph box.
Figure 6. Changes in land use and land cover observed from 2015 to 2020 and predicted until 2050. Highlight for the reduction of native vegetation up to 47.5% in 2050. The categories with smaller area in the biome appear expanded in the graph box.
Sustainability 15 04251 g006
Figure 7. Land use and land cover map showing the observed (1985 to 2020) and predicted (2030 to 2050) transitions between pastures and agricultural lands.
Figure 7. Land use and land cover map showing the observed (1985 to 2020) and predicted (2030 to 2050) transitions between pastures and agricultural lands.
Sustainability 15 04251 g007
Figure 8. Each hexagon represents a watershed and the respective observed and modeled data, identified by the name and code of the fluviometric station, with each hexagon being as close as possible to its real spatial location. Shades of red and yellow, respectively, represent the proportion occupied by agricultural land and pasture in different years and future estimates; light blue and dark blue represent the magnitude of the trend (Theil–Sen) of water flow and precipitation; the black lines represent the overall accuracy of the model.
Figure 8. Each hexagon represents a watershed and the respective observed and modeled data, identified by the name and code of the fluviometric station, with each hexagon being as close as possible to its real spatial location. Shades of red and yellow, respectively, represent the proportion occupied by agricultural land and pasture in different years and future estimates; light blue and dark blue represent the magnitude of the trend (Theil–Sen) of water flow and precipitation; the black lines represent the overall accuracy of the model.
Sustainability 15 04251 g008
Figure 9. Spatial locations of the impacts on water flow of individual watersheds observed during the pre-disturbance (P1) and post-disturbance periods. The circle sizes indicate the percentage change. Blue indicates an increasing water flow; red indicates a decreasing water flow. Pink portion within the circles indicates the estimated contribution attributed to land use changes, and the yellow portion indicates the estimated contribution attributed to climate changes to the river water flow changes.
Figure 9. Spatial locations of the impacts on water flow of individual watersheds observed during the pre-disturbance (P1) and post-disturbance periods. The circle sizes indicate the percentage change. Blue indicates an increasing water flow; red indicates a decreasing water flow. Pink portion within the circles indicates the estimated contribution attributed to land use changes, and the yellow portion indicates the estimated contribution attributed to climate changes to the river water flow changes.
Sustainability 15 04251 g009
Figure 10. Predicted changes in stream flow of watersheds in the Cerrado biome up to 2050 compared to the pre-disturbance period (P1). The circle sizes indicate the percentage of stream flow changes, blue indicates increase in stream flows, red indicates decrease in stream flows. The pink portion within the circles indicates the estimated contribution of land use changes to stream flow changes, and the yellow portion indicates the estimated contribution of climate changes.
Figure 10. Predicted changes in stream flow of watersheds in the Cerrado biome up to 2050 compared to the pre-disturbance period (P1). The circle sizes indicate the percentage of stream flow changes, blue indicates increase in stream flows, red indicates decrease in stream flows. The pink portion within the circles indicates the estimated contribution of land use changes to stream flow changes, and the yellow portion indicates the estimated contribution of climate changes.
Sustainability 15 04251 g010
Figure 11. Scatter plot showing the contribution of each driver (climate, land use, and land cover) to the respective changes in streamflow of each watershed.
Figure 11. Scatter plot showing the contribution of each driver (climate, land use, and land cover) to the respective changes in streamflow of each watershed.
Sustainability 15 04251 g011
Figure 12. Boxplots of each scenario analyzed showing the flow of the set of 81 basins. From left to right: observed river flow before the period indicated by the Pettitt test; post-period indicated by the Pettitt test, with the parameters of land use change stagnant on the respective date; post-period indicated by the Pettitt test, but with land use changes occurring up to 2022; future estimate up to 2050, in a hypothetical scenario if only the land use change occurred up to the date detected by the Pettitt test had been maintained; future estimate up to 2050, in a trend scenario of land use change as modeled in this study.
Figure 12. Boxplots of each scenario analyzed showing the flow of the set of 81 basins. From left to right: observed river flow before the period indicated by the Pettitt test; post-period indicated by the Pettitt test, with the parameters of land use change stagnant on the respective date; post-period indicated by the Pettitt test, but with land use changes occurring up to 2022; future estimate up to 2050, in a hypothetical scenario if only the land use change occurred up to the date detected by the Pettitt test had been maintained; future estimate up to 2050, in a trend scenario of land use change as modeled in this study.
Sustainability 15 04251 g012
Figure 13. Examples of hydrograms: (A)—Quebra Anzol river, in Minas Gerais; (B)—Arrojado river, in Bahia; (C)—Ondas River, also in Bahia; (D)—Corda River, in Maranhão. The dashed line indicates when the change in the pattern of the variables was detected; the blue portion indicates the modeled flow if the change in land use had ended at the time when the change in the pattern of variables was identified; red shows the flow trend in a scenario of changes in land use and climate trends.
Figure 13. Examples of hydrograms: (A)—Quebra Anzol river, in Minas Gerais; (B)—Arrojado river, in Bahia; (C)—Ondas River, also in Bahia; (D)—Corda River, in Maranhão. The dashed line indicates when the change in the pattern of the variables was detected; the blue portion indicates the modeled flow if the change in land use had ended at the time when the change in the pattern of variables was identified; red shows the flow trend in a scenario of changes in land use and climate trends.
Sustainability 15 04251 g013
Table 1. Variables related to land use changes and their respective explanatory powers (V’Cramer), showing values close to 1, highly related to land use changes, and values below 0.16, a relationship close to null.
Table 1. Variables related to land use changes and their respective explanatory powers (V’Cramer), showing values close to 1, highly related to land use changes, and values below 0.16, a relationship close to null.
VariablesGeneralForestPastureSavannaAgricultureGrasslandSilvicultureWaterOthersUrbanMosaicMining
Deforestation distance (natural log)0.380.890.720.530.480.470.170.10.060.080.070.007
Friction (travel cost)0.260.630.30.370.350.340.140.360.060.010.030.001
Slope0.280.870.490.410.370.330.120.10.060.050.040.006
Altimetry0.280.870.490.420.360.330.140.070.060.050.040.006
Rainfall of the rainiest month0.280.870.50.430.310.340.140.070.060.050.060.007
Rainfall of the driest month0.240.670.510.250.30.180.160.050.040.050.110
Soils0.270.870.480.480.30.330.110.060.060.050.040.005
Soils (evidence likelihood)0.270.770.460.420.340.350.230.090.080.060.050.007
Protected areas (natural log)0.290.870.510.410.310.40.110.070.080.060.060.005
Legal reserves (natural Log)0.260.80.480.330.310.290.110.080.060.070.040.006
Permanent preservation area (square root distance)0.280.850.490.40.310.340.110.060.060.050.070.006
Valley (square root distance)0.110.590.320.270.320.190.10.10.040.040.020.003
Hydrography (distance)0.050.010.0030.00040.0060.00190.0010.070.140.020.0004
Limite constrange290.870.510.410.370.320.120.070.060.050.0010.07
Table 2. On the left: the accuracy of the land use transitions. On the right: the model accuracy as a whole, namely, random agreement—N(n), quantity agreement—N(m), standard agreement—M(m), and adjusted location agreement—P (m), in addition to standard Kappa and Kappa without information (Kno).
Table 2. On the left: the accuracy of the land use transitions. On the right: the model accuracy as a whole, namely, random agreement—N(n), quantity agreement—N(m), standard agreement—M(m), and adjusted location agreement—P (m), in addition to standard Kappa and Kappa without information (Kno).
Accuracy of Land Use TransitionsModel Accuracy
TransitionAccuracyTransitionAccuracyOverall Model Accuracy MetricsModelNull Model
Forest to Pasture71.82%Grassland to Mining77.78%
Forest to agriculture71.82%Others to savanna67.51%P(N)0.300.30
Savanna to pasture75.20%Others to grassland68.00%P(M)0.990.99
Savanna to silviculture83.25%Others to pasture65.60%P(P)1.001.00
Savanna to mosaic88.98%Others to agriculture76.38%M(n)0.300.28
Savanna to urban89.62%Others to mosaic100%M(m)0.990.96
Savanna to agriculture73.65%Pasture to grassland63.50%M(p)0.990.96
Grassland to pasture78.50%Pasture to agriculture57.54%N(n)0.090.09
Grassland to agriculture80.04%Pasture to mosaic77.00%N(m)0.630.63
Grassland to silviculture86.69%Forest to silviculture98.04%N(p)0.630.63
Grassland to mosaic88.89%Forest to mosaic98.44%Kno0.990.96
Grassland to urban92.45%Savanna to forest61.01%Klocation0.990.90
Table 3. Land use and land cover changes observed by 2015, 2018, and 2020 and predicted by 2030, 2040, and 2050. Land use, land cover classes, and total projected deforestation and remnants of native vegetation in scenario Business as Usual.
Table 3. Land use and land cover changes observed by 2015, 2018, and 2020 and predicted by 2030, 2040, and 2050. Land use, land cover classes, and total projected deforestation and remnants of native vegetation in scenario Business as Usual.
Land Use201520182020203020402050
Forest18.40%18.70%18.50%17.70%17.20%16.90%
Savanna22.00%21.50%21.30%20.10%19.40%18.90%
Grassland14.40%14.10%13.90%12.80%12.20%11.70%
Others0.50%0.50%0.50%0.40%0.40%0.30%
Pastures30.00%30.00%29.80%29.40%29.20%29.10%
Agriculture11.70%12.20%13.00%16.60%18.60%20.00%
Silviculture1.70%1.70%1.70%1.70%1.70%1.70%
Agricultural mosaic0.30%0.30%0.30%0.30%0.30%0.30%
Urban areas0.30%0.40%0.40%0.40%0.40%0.40%
Mining0.00%0.00%0.00%0.00%0.00%0.00%
Water areas0.70%0.60%0.60%0.60%0.60%0.60%
Deforestation(km2)10.63511.74763.95234.10825.581
Deforestation(km2/year)3.5455.8746.3953.4112.558
Native remaining54.80%54.30%53.70%50.50%48.80%47.50%
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Salmona, Y.B.; Matricardi, E.A.T.; Skole, D.L.; Silva, J.F.A.; Coelho Filho, O.d.A.; Pedlowski, M.A.; Sampaio, J.M.; Castrillón, L.C.R.; Brandão, R.A.; Silva, A.L.d.; et al. A Worrying Future for River Flows in the Brazilian Cerrado Provoked by Land Use and Climate Changes. Sustainability 2023, 15, 4251. https://doi.org/10.3390/su15054251

AMA Style

Salmona YB, Matricardi EAT, Skole DL, Silva JFA, Coelho Filho OdA, Pedlowski MA, Sampaio JM, Castrillón LCR, Brandão RA, Silva ALd, et al. A Worrying Future for River Flows in the Brazilian Cerrado Provoked by Land Use and Climate Changes. Sustainability. 2023; 15(5):4251. https://doi.org/10.3390/su15054251

Chicago/Turabian Style

Salmona, Yuri Botelho, Eraldo Aparecido Trondoli Matricardi, David Lewis Skole, João Flávio Andrade Silva, Osmar de Araújo Coelho Filho, Marcos Antonio Pedlowski, James Matos Sampaio, Leidi Cahola Ramírez Castrillón, Reuber Albuquerque Brandão, Andréa Leme da Silva, and et al. 2023. "A Worrying Future for River Flows in the Brazilian Cerrado Provoked by Land Use and Climate Changes" Sustainability 15, no. 5: 4251. https://doi.org/10.3390/su15054251

APA Style

Salmona, Y. B., Matricardi, E. A. T., Skole, D. L., Silva, J. F. A., Coelho Filho, O. d. A., Pedlowski, M. A., Sampaio, J. M., Castrillón, L. C. R., Brandão, R. A., Silva, A. L. d., & Souza, S. A. d. (2023). A Worrying Future for River Flows in the Brazilian Cerrado Provoked by Land Use and Climate Changes. Sustainability, 15(5), 4251. https://doi.org/10.3390/su15054251

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop