- Email: [email protected]
Spatiotemporal dynamics of soybean crop in the Matopiba region, Brazil (1990–2015)
Land Use Policy 80 (2019) 57–67
Contents lists available at ScienceDirect
Land Use Policy journal homepage: www.elsevier.com/locate/landusepol
Spatiotemporal dynamics of soybean crop in the Matopiba region, Brazil (1990–2015)
T
⁎
Mayara Lucyanne Santos de Araújoa, , Edson Eyji Sanob, Édson Luis Bolfec, Jessflan Rafael Nascimento Santosd, Juliana Sales dos Santosd, Fabrício Brito Silvad a
Graduate Program in Applied Geoscience, University of Brasília, Brazil Embrapa Cerrados, Brazil c Embrapa Secretaria de Inteligência e Relações Estratégicas, Brazil d Universidade Ceuma, Brazil b
A R T I C LE I N FO
A B S T R A C T
Keywords: Agricultural expansion Soybean Moran index Geostatistics
Brazil is a world leader in the production and export of grains, particularly soybeans. The newest agricultural frontier in Brazil is the Matopiba region, which is a continuous zone formed by the states of Maranhão, Tocantins, Piauí, and Bahia, located mostly within the Cerrado biome. The objective of this study was to analyze the spatiotemporal dynamics of soybean production and yield in the Matopiba region. We analyzed municipality-based planted areas and production data obtained by the Brazilian Institute of Geography and Statistics during 1990–2015. Yield was estimated from the production and planted area, and the data were analyzed using global and local Moran indices. The results showed that soybean production in the Matopiba region does not occur randomly. Positive and significant autocorrelation was found at the beginning of the time series among those municipalities located in the west of Bahia. This region influenced the soybean expansion from south to north. Currently, high-production areas are concentrated in two autocorrelated blocks: one in western Bahia and the other in the central Matopiba region. Analysis of spatial autocorrelation involving yield showed a decreasing trend at the end of the time series. The presence of municipalities with high yield surrounded by others with low yield, and vice-versa, were observed. The findings of this study could assist local and regional agricultural planning in the Matopiba region, and support related analyses in other fields of agriculture, the environment, and logistics.
1. Introduction Because of its considerable territorial extent and its favorable climate, topography, and soil physical properties that support extensive rainfed crop production, Brazil has become one of the main exporters of agricultural commodities, such as soybean, corn, coffee, sugarcane, and cotton (MAPA, 2016a; Mueller and Mueller, 2016). In 2015, the annual crops that presented the largest planted areas in Brazil were soybeans (32 million ha), maize (16 million ha), and sugarcane (10 million ha) (IBGE, 2015). Currently, Brazil and the United States are the leading producers of soybeans (MAPA, 2015). In 2015, Brazil exported approximately 57 million tons of soybeans (An and Ouyang, 2016), primarily to China (Lima et al., 2017). In 2016/2017, Brazil produced 114 million tons of soybeans (USDA, 2017). Analysis by the United Nations Food and Agriculture Organization has indicated that soybean production in Brazil will increase by 37% over the next 10 years (OCDE/
⁎
FAO, 2015). Soybean planting began in Brazil in the 1940s as an option for crop rotation with wheat (Brown et al., 2005), and the crop adapted well in the southernmost part of the country because of the temperate climate. According to Paludzyszyn Filho et al. (1993), soybean became important in Rio Grande do Sul State because of the exportation. Gradually, because of investments by the Brazilian government in research institutions, soybean plants were genetically modified to improve adaptation in other regions of the country with tropical climate (Guimarães and Leme, 1997; Andersen et al., 2002). In addition, the use of chemical fertilizers was implemented to correct the predominantly acidic soils with low natural fertility found mainly in central parts of the country (Delgado, 1985). Brazilian Agricultural Research Institute (Embrapa Cerrados), created in 1973 by the Ministry of Agriculture, played a key role here since its research allowed remarkable increase in the soybean production in Brazil (Mueller and Mueller, 2016).
Corresponding author at: Universidade de Brasília, Campus Universitário Darcy Ribeiro, CEP: 70910-900, Brasília, DF, Brazil. E-mail address: [email protected] (M.L.S.d. Araújo).
https://doi.org/10.1016/j.landusepol.2018.09.040 Received 21 April 2018; Received in revised form 23 July 2018; Accepted 29 September 2018 0264-8377/ © 2018 Elsevier Ltd. All rights reserved.
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 1. Map of the Matopiba region in Brazil, subdivided into 10 mesoregions. Each mesoregion corresponds to a subdivision of a Brazilian state that aggregates a varying number of municipalities with high levels of economic and social similarity. State identification: MA = Maranhão, TO = Tocantins, PI = Piauí, and BA = Bahia.
regions of western Bahia State, southeastern Goiás State (municipalities of Jataí and Rio Verde), and the central region of Mato Grosso State (municipalities of Lucas do Rio Verde, Sinop, and Sorriso). Agricultural frontier is defined as a region dominated by natural vegetation that started to face intensive agriculture-related land occupation. The most recent agricultural frontier of the Cerrado is the Matopiba region, a continuous zone formed by the states of Maranhão, Tocantins, Piauí, and Bahia (Miranda et al., 2014). In this region, infrastructure is poor, land prices are cheap, and the climate and topographic relief are favorable for rainfed agriculture. Currently, soybean is the main agricultural crop of Matopiba (MAPA, 2017). Following the rapid agricultural expansion in Matopiba, the Brazilian government issued Federal Decree No. 8,447 on May 6, 2015, establishing an Agricultural Development Plan for Matopiba. The purpose of this decree was to promote and coordinate public policies for economic and sustainable development of agricultural and livestock activities in the Matopiba region. The plan proposes guidelines for federal programs, projects, and actions to be undertaken with the objectives of improving both the living standards of the local population and the economic growth of the country. For this plan to succeed, it is of great relevance to understand the spatiotemporal dynamics of the crops produced in this region, with emphasis on the spatial clusters that
At the same time, because of the approved agrarian reform legislation, large landowners started to prevent against land loss for small farmers, rural workers, and landless peasants by increasing their production and investing in mechanization (Mueller and Mueller, 2016). Other relevant factors that contributed to the expansion of soybean in Central Brazil included the public tax incentives to open new areas for soybean and for establishment of companies for grain storage and processing, the availability of large areas with flat topography (plateaus), that is, favorable for mechanization, the relatively high precipitation conditions for rainfed agriculture, and the relatively good economic and technological levels of the farmers from southern part of the country that migrated to Cerrado (Dall’Agnol (2008); Campos, 2010). Since 1990, the Brazilian government reduced its direct involvement in agriculture. On the other hand, private sector started to invest strongly in this sector (Alston et al., 2016). The country also implemented a series of more open and predictable political and economic institutional arrangements (Alves and Pastore, 1978). This lower need for direct political intervention in areas such as credit and price management allowed the country to grow even more in the agricultural sector (Mueller, 2009). New agricultural frontiers were created in the Cerrado biome, e.g., 58
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
neighbors, evaluating whether the distribution is grouped or random (Anselin, 2005). Positive values (between 0 and +1) show direct correlation and negative values (between 0 and −1) reflect an inverse correlation (Câmara et al., 2002). Spatial autocorrelation is computed only for neighbors of the first order in space, as established by the wi,j weights of Eq. (1) (Anselin, 1995):
might eventually appear over time. This understanding will assist the agricultural, environmental, and economic planning activities of the region. Despite the recent increase in the number of studies focused on the Matopiba region, the use of statistics such as the Moran index for spatial analysis has not been explored yet, leading to some unanswered questions. 1) Does soybean production and yield follow spatial patterns that reflect local or regional aptitude? 2) If spatial patterns of production and yield occur, do they vary systematically over time, demonstrating the migration of aptitude to soybean cultivation? 3) Is there any spatial correlation between soybean production and yield? The answers to these questions will assist both the preparation of local and regional agricultural planning and the ongoing research and analysis in fields such as agriculture, the environment, and logistics. The primary objective of this research was to analyze the spatiotemporal dynamics of soybean production in the Matopiba region (1990–2015) by applying spatial statistics techniques on time series data.
n
I=
n
n ∑i = 1 ∑ j = 1 wi, j z i zj → so = n so ∑i = 1 z i 2
n
n
∑ ∑ wi,j (1)
i=1 j=1
where zi is the deviation of an attribute for the resource i from its average, zj is the deviation of an attribute for resource j from its average, wi,j is the weight assigned according to the connection between areas i and j, n is the total number of attributes, and So is the aggregate of all spatial weights. The local Moran index (Ii) is a measure of the spatial arrangement that assesses the correlation between an observation and its neighborhood (Anselin, 1995). It can capture possible local patterns of spatial autocorrelation and identify the presence of spatial clusters of similar values or anomalous objects (outliers). Derived values that are significantly high and positive indicate the presence of clusters of similar values, whereas significantly low values indicate inequality within the region (Zhang and Lin, 2016). The result is expressed as clusters of high values (HH) and of low values (LL), and as outliers where a high value is surrounded by low values (HL) or a low value is surrounded by high values (LH). The Ii can be defined by the following equation:
2. Materials and methods 2.1. Study area The selected study area was the Matopiba region, which extends over parts of the states of Maranhão, Tocantins, Piauí, and Bahia (Miranda et al., 2014; Santos Filho et al., 2016). The Matopiba region comprises 10 mesoregions (5 in the state of Maranhão, 2 in Tocantins, 1 in Piauí, and 2 in Bahia) and 337 municipalities (Fig. 1). A mesoregion corresponds to a subdivision of a Brazilian state and it can encompass a varying number of municipalities with high levels of economic and social similarity. The study region occupies approximately 73 million ha located in the area within 2°30′–15°15′S latitude, 42°00′–50°00′W longitude. In 2014/2015, Matopiba contributed 9.4% of the 209.5 million tons of grains produced in Brazil (Portal Brasil, 2015). This region encompasses 50 federal, state, and municipal conservation units (7.2 million ha) and 23 indigenous lands (3.6 million ha). Latossolo (Oxisols) is the dominant soil type, covering 27.8 million ha (38% of the Matopiba region) (Magalhães and Miranda, 2014). Latossolos are welldeveloped deep soils, which are highly weathered with low soil fertility, good permeability, and high porosity. The second largest class of soils in the Matopiba region is Neossolo Quartzarênico (deep, sandy soils, composed largely of quartz), which occupies an area of 18 million ha (25% of the region).
Ii =
x i−x¯ si2
n
n
∑
wi, j (x j−x¯) → si2 =
j = 1, j ≠ i
∑ j = 1, j ≠ i (x j−x¯)² n−1
−x¯² (2)
where xi is the value of the attribute considered in area i, xj is the value of the attribute considered in area j, x¯ is the average of the corresponding attribute, wi,j is the weight assigned according to the connection between areas i and j, and n is total number of attributes. 3. Results 3.1. Spatiotemporal dynamics The Matopiba region showed significant growth in soybean production during the time series (Fig. 2): from 260,624 t in 1990 to 10,758,927 t in 2015, an increase of 4028% in 25 years. The municipality of São Desidério (Extremo Oeste Baiano mesoregion) presented the highest increase in production, rising from 1999 t in 1990 to 1,134,000 t in 2015 (an increase of more than 28,000%). This municipality presents average mean precipitation of 1145 mm (Tropical Rainfall Measuring Mission data, 1998–2015 time series), relatively extensive plateaus with average elevation of 778 m and average slope of 1.8% (Shuttle Radar Topography Mission data), dystrophic yellow-red Latossolo (deep, acidic, low to intermediate levels of soil fertility) (IBGE, 2017). Such combination of precipitation, topography and soil type favors the implementing of large-scale grain production. At the beginning of the time series, the highest production was concentrated in the Extremo Oeste Baiano mesoregion. Subsequently, there was an expansion from south to north, involving not only the municipalities of Extremo Oeste Baiano but also the municipalities of the Sul Maranhense and Sudoeste Piauiense mesoregions. However, we must be aware that agricultural expansion over these areas is not so straightforward since their biophysical conditions are not so similar to those from western Bahia region. Overall, in the Sul Maranhense and Sudoeste Piauiense mesoregions, the crop production is more sensitive to drought effects (less rainfall since they are closer to the semi-arid Caatinga), there are more occurrence of Neossolos Quartzarênicos (sandy soils) and the extension of plateaus are smaller. The municipality of São Desidério presented linear growth in soybean production during the time series (Fig. 3). It has had the second
2.2. Approach The Municipality Agricultural Production (PAM – Produção Agrícola Municipal) time series data obtained by the Brazilian Institute of Geography and Statistics (IBGE) during 1990–2015 comprised the basic data for this study. These data are intended to provide the basis for statistical analysis on the production, planting, harvesting areas, and price of Brazilian agricultural products at the municipality level on an annual basis for the entire country. The data are obtained through questionnaires submitted by IBGE agents to producers and technicians involved in the agricultural sector (IBGE, 2017; IBGE, 2018). Here, we must be aware that questionnaire-based data present some limitations since they are dependent on information provided, for example, by farmers and seed sellers, there may have lack of data in some municipalities during the time series, and some farms can be located in-between two or more municipalities. The spatiotemporal analyses in this study were based on soybean production data (tons) and the planted and harvested areas (ha). The yield (tons ha−1) was estimated based on the ratio between production and planted area. These data were analyzed based on global and local Moran indices (Moran, 1948). The main purpose of the global Moran index (I) is to verify the degree of spatial autocorrelation between the 59
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 2. Spatiotemporal dynamics of soybean production in the Matopiba region based on IBGE Municipality Agricultural Production (PAM) data: (a) 1990, (b) 1995, (c) 2000, (d) 2005, (e) 2010, and (f) 2015. TP = total production; Mt = million tons. See Fig. 1 for mesoregion identification.
yield in the region (2.1 t ha−1) was found in the municipality of Guaraí. By 2015, maximum yield had increased to 4.0 t ha−1 in the municipality of Serra do Ramalho, Bahia State.
largest territorial expansion in Bahia State of approximately 15,166 km² (IBGE, 2016), and it is in the lead of the municipalities with the highest agricultural production in the country. Initially, soybean yield within the study area was most prominent in the municipality of Guaraí (Tocantins State), Extremo Oeste Baiano mesoregion, southwest Tocantins State and the central Matopiba region (Fig. 4). At the end of the time series, yield had increased in different areas of the Matopiba region, particularly in the Extremo Oeste Baiano mesoregion, southwest Tocantins State, and western parts of Maranhão State, an Amazon–Cerrado transition area. In 1990, maximum soybean
3.2. Spatial statistics The global Moran index applied to the data of soybean production, yield, planted area, and harvested area produced values greater than zero, indicating the existence of spatial patterns among the municipalities belonging to the study area (Table 1). The p-values were < 0.01 60
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 3. Evolution of soybean production in the municipality of São Desidério, Bahia State (1990–2015).
showed statistically significant HH, HL, LL, and LH clusters (Fig. 6). In 1990, there was indication of the presence of two HH zones with high yield. One was located in the Extremo Oeste Baiano mesoregion (Formosa do Rio Preto, Riachão das Neves, Barreiras, São Desidério and Correntina municipalities), while the other was located in the Sudoeste Tocantinense (Dueré, Aliança do Tocantins, Gurupi, Peixe, Figueirópolis and Araguaçu municipalities) mesoregion, together with the Pedro Afonso municipality in the central part of Tocantins State. In 1990, the municipality of Goiatins (Tocantins State) presented an HL cluster. However, by 1995, both the Pedro Afonso and the Goiatins municipalities were unable to maintain their high yield levels. In 1995, another HH cluster appeared in the center of the Matopiba region. In this year, the municipality of Ribeiro Gonçalves (Sudoeste Piauiense mesoregion) presented low yield autocorrelation with its neighboring municipalities with high yield (i.e., an LH cluster). However, in 2000, this changed to an HH cluster, indicating increased soybean yield in this municipality. The same characteristic was observed in the São Félix de Balsas municipality (Maranhão State) in 2000. This result confirms the importance of the relationship between neighboring municipalities regarding the growth of agriculture. In 2000, spatial autocorrelation of the municipalities was concentrated in the Extremo Oeste Baiano, Sul Maranhense, and Sudoeste Piauiense mesoregions (HH cluster). In 2005, 2010, and 2015, two LL clusters were produced, involving the municipalities of the Centro Maranhense, Leste Maranhense, Oeste Maranhense, and Ocidental do Tocantins mesoregions. In 2015, an LL cluster was also observed in the municipality of Wanderley (Bahia State). It can be seen from Fig. 7 that production and yield data were not correlated over time. Soybean production presented smaller numbers of spatial groupings compared with soybean yield. This indicates that municipalities with greater production were not necessarily the municipalities with the largest areas of cultivation. The number of municipalities with high soybean productions that were surrounded by other municipalities with high soybean productions varied from 5–14. This indicates, despite the considerable territorial extension of the Matopiba region and its importance to Brazilian agriculture, the productive zones were concentrated in only a few municipalities. The concentration remained constant throughout the time series. In 1990, the HH-rated municipalities accounted for 83.9% of total soybean production (260,624 t). In 2015, they represented 61.2% of the total production (10,758,927 t). The four spatial groupings of yield (HH, HL, LL, and LH), which ranged from 1 to 71 municipalities during the time series (1990–2015), confirmed the high numbers of municipalities with large areas of soybean plantations. However, it is important to highlight that the numbers of municipalities in the Matopiba region rated within LL clusters are growing and that they account for more municipalities than in HH clusters.
for the entire time series, indicating statistical significance for the spatial autocorrelation. Moran index for planted areas and harvested areas were quite similar, indicating that there were no significant losses of planted area during the crop cycles because of some disease or fire. The results of the local Moran index for production and yield data are presented in Figs. 5 and 6, respectively. HH indicates higher production/yield in a certain municipality and in its neighboring municipalities; HL indicates high yield in a certain municipality and low yield in the neighboring municipalities; LH indicates low yield in a certain municipality and high production/yield in the surrounding municipalities; and LL indicates lower yield in a certain municipality and in the surrounding municipalities. The local Moran index for soybean production presented an HH cluster comprising the municipalities of Barreiras, Correntina, Formosa do Rio Preto, Riachão das Neves, and São Desidério (Extremo Oeste Baiano mesoregion) (Fig. 5). This cluster appears from the beginning of the time series (1990), consolidated over time, and expanded into the north, starting in 2000 in the southern portions of the Sul Maranhense and Sudoeste Piauiense mesoregions. The highest value of the local Moran index was observed in the municipality of São Desidério, which suggests high probability of it having influenced other municipalities to consolidate soybean production within the study area, mainly in terms of transferring new knowledge and technologies to neighboring municipalities. In 1995, the municipality of Jaborandi joined the HH cluster formed in the Extremo Oeste Baiano mesoregion. In 2005, the Luís Eduardo Magalhães municipality also joined the HH cluster. In 2000, in addition to the positive autocorrelation cluster located in the Extremo Oeste Baiano, another HH cluster formed in the Sul Maranhense mesoregion, which comprised the municipalities of Balsas and Tasso Fragoso. In 2005, the municipalities of Riachão (Maranhão State), Campos Lindos (Tocantins State), and the municipalities of Ribeiro Gonçalves and Baixa Grande do Ribeiro (Piauí State) joined this cluster. In 2010, the municipality of Santa Filomena (Piauí State) also joined this cluster. The consolidation of this cluster in the central part of the Matopiba region was confirmed in 2015. The results identified the presence and consolidation of two zones of high soybean production in the Matopiba region. The presence of the Jalapão State Park and the Nascentes do Parnaíba National Park is mainly responsible for the geographical discontinuity between these two areas. The result indicates that soybean production in the Matopiba region has not occurred randomly but it has integrated in well-defined spatial patterns during the time series. The most relevant factor controlling such spatial pattern is that the soybean expansion is dependent on the availability of a combination of flat terrains, favorable soil physical properties (texture) as well as enough rainfall conditions during the crop cycle. Therefore, expansion of soybean in Matopiba is likely to happen only in areas with such combination of biophysical properties. The results also confirm statistically the occurrence of the south–north expansion of soybean production in this region. The local Moran index for soybean yield in the Matopiba region 61
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 4. Spatiotemporal dynamics of soybean yield in the Matopiba region based on IBGE Municipality Agricultural Production (PAM) data: (a) 1990, (b) 1995, (c) 2000, (d) 2005, (e) 2010, and (f) 2015. PT = total production. AY = average yield. See Fig. 1 for mesoregion identification.
4. Discussion
and shown effective in identifying spatial clusters produced over time. Thus, these indices constitute important tools in environmental and agricultural studies, as exemplified by Javari (2017); Clemente et al. (2017), and Yuan et al. (2017). Matopiba has been in the center of discussion of dominant model of Cerrado occupation (based on the complete removal of native vegetation, that is, deforestation) since it is part of largest region where native vegetation of Cerrado is still preserved. In fact, Matopiba is a very interesting region for implementation of 11 components defined by the Programa Cerrado Sustentável (MMA, 2006), which includes actions of biodiversity conservation, sustainable use of biodiversity, water resources management, and environmentally sustainable agriculture.
To make reliable decisions regarding regional agricultural planning and development, knowledge of the spatiotemporal patterns of crops and their performance over time is essential. Therefore, the analyses performed in this study, evidencing the spatial patterns of soybean cultivation in the Matopiba region over 25 years, could provide important support for the development of agriculture in the region, particularly within the context of environmentally sustainable progress. The spatial statistical techniques adopted here have been used in other agricultural studies (Cho and Newman, 2005; Yu et al., 2014; Donfouet et al., 2017). The global and local Moran indices have been used widely 62
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
The results of the local Moran index for soybean yield in the Matopiba region showed that soybean is produced in areas with defined spatial patterns. It is important to highlight the presence of areas with LL clusters, which were identified mainly in central eastern parts of Maranhão, western Maranhão, and northern Tocantins from 2005 to 2015, as well as in the municipality of Wanderley (Bahia State) in 2015. These areas, which will require public policy support and financial incentives to increase production, tend to be located in transitional areas between biomes. The central eastern area of Maranhão, western Maranhão, and northern Tocantins regions are located in the Amazon–Cerrado transition region, while the municipality of Wanderley is located in the Cerrado–Caatinga transition region. According to Silva et al. (2016), a transition area between different biomes presents highly diverse ecosystems and climatic conditions, as well as areas with lower natural fertility, making them particularly vulnerable to climate change. Another important issue in the Matopiba region is the sensitivity to the El Niño events. This region faced an El Niño event during the period of 2012–2015 that led to a strong decrease in the crop yield, especially in municipalities with worse edaphoclimatic conditions. During the 2015/2016 drought event, there was an overall reduction of grain production of approximately 35% because of the 47-day dry spell occurrence (Canal Rural, 2016). Although our statistical results showed unclear influence of El Niño in the 2015 data set, such climatic event may cause some relevant impact in the local Moran Index. It was also found that the most productive zones were concentrated in only a few areas of the Matopiba region. Figueiredo (2016) reported the growth in private research by large multinational companies in productive areas. The consequence of this process has been regional concentration of soybean production in areas that have received the greatest technological contributions from the investments of in situ research, mainly in relation to the testing of the performance of cultivar suitability to the climatic conditions of the region. The future of soybean expansion in Matopiba seems to be dependent on the development of new crop varieties more resistant to drought and dry spells. Brazilian farmers use a threshold of mean annual average rainfall of 1000 mm to determine if an area is suitable for rainfed agriculture. Development of new varieties either by research institutions such as Embrapa or by private companies should lower this threshold for 800 mm or so. Transportation network is another important issue since road condition in this region is very poor and railroad does not exist yet. The rapid deforestation in Matopiba is also arousing attention in terms of compliance with the Brazilian Forest Code. Though the New Forest Code determine that only 20% of farms located in the Cerrado biome must be preserved by native vegetation (Reserva Legal), farmers also must protect their areas considered as permanently preserved area (e.g., riparian forests along streams). Currently, farmers who practice illegal deforestation have their area embargoed by the Brazilian Institute for Environment and Renewable Natural Resources (IBAMA) and are not allowed to sell their products until the area is restored. The increasing number of traders committed to no-deforestation grain production and the increasing number of protests from civil society organizations such as the Cerrado Manifesto will also contribute significantly to the development of environmentally sustainable agriculture in Matopiba. Brazilian government is also starting a program to monitor Cerrado deforestation at an annual basis and using Landsat-like, moderate spatial resolution satellite data. This will be another important tool to monitor the rate of deforestation not only in Matopiba but also in other important regions of Cerrado in terms of grain production.
Table 1 Global Moran index for production, yield, planted area, and harvested area variables in 1990, 1995, 2000, 2005, 2010, and 2015. Period
Moran index
z-score
p-value
Result
Production (t) 1990 1995 2000 2005 2010 2015
0.073 0.076 0.082 0.132 0.105 0.129
4.396 4.382 4.680 6.752 5.366 6.373
< 0.01 < 0.01 < 0.01 < 0.01 < 0.01 < 0.01
Grouped Grouped Grouped Grouped Grouped Grouped
Yield (t ha−1) 1990 1995 2000 2005 2010 2015
0.072 0.190 0.188 0.444 0.428 0.389
3.506 8.601 8.516 19.794 19.088 17.363
< 0.1 0 0 0 0 0
Grouped Grouped Grouped Grouped Grouped Grouped
Planted area (ha) 1990 0.074 1995 0.078 2000 0.081 2005 0.132 2010 0.109 2015 0.135
4.480 4.447 4.665 6.764 5.521 6.608
< 0.01 < 0.01 < 0.01 0 0 0
Grouped Grouped Grouped Grouped Grouped Grouped
Harvested area (ha) 1990 0.074 1995 0.078 2000 0.081 2005 0.132 2010 0.109 2015 0.135
4.474 4.447 4.665 6.764 5.520 6.606
< 0.01 < 0.01 < 0.01 0 0 0
Grouped Grouped Grouped Grouped Grouped Grouped
The results of the global spatial autocorrelations applied to the variables of soybean production and yield within the study area showed positive autocorrelation between municipalities. The results demonstrated tendencies of expansion and consolidation of the use of soybean in the Matopiba region. This region is now considered the major agricultural frontier for agribusiness investment in Brazil (Anderson et al., 2016). According to Espíndola and Cunha (2015), several economic and political incentives relating to agricultural production have been directed toward the central part of the country. For example, according to the 2016–2017 Agricultural and Livestock Plan (MAPA, 2016b), a total of approximately US$55 billion has been made available to rural producers and their cooperatives to finance their activities. However, the large expansion of soybean that was verified in the Extremo Oeste Baiano is unlikely to occur in other mesoregions because of the biophysical conditions, that is, smaller extension of plateaus, higher occurrence of sandy soils (Neossolos Quartzarênicos) and higher sensitivity to the occurrence of dry spells (border to the semi-arid Caatinga biome). The local Moran index revealed that soybean production has been consolidated in the Extremo Oeste Baiano. According to Spagnolo (2011), the agricultural expansion of soybean cultivation in this mesoregion was consolidated in 2000, following the availability of low-tax loans offered by the federal government as part of various programs to increase agricultural production in Bahia State. One example is the AGRINVEST Investment Program for the Modernization of Agriculture in Bahia. From 2000, the Sul Maranhense mesoregion presented high and statistically significant autocorrelation; a characteristic that was also identified from 2005 in the Sudoeste Piauiense mesoregion. In Maranhão, the greatest incentives have been the improvement of the state’s road infrastructure (Frota and Campelo, 1999; Studte, 2008), valorization of soybean prices in international markets, credit incentives (Mesquita, 2011), and technological investments, which have helped transform the traditional subsistence agrarian structure into a technified agrarian structure (Paludzyszyn Filho, 1995; Bolfe et al., 2016).
5. Conclusions This study identified important zones of soybean production in the Matopiba region of Brazil based on the spatial statistics of the global and local Moran indices. The most productive areas of soybean 63
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 5. Spatiotemporal dynamics of the local Moran index applied to soybean production in the Matopiba region: (a) 1990, (b) 1995, (c) 2000, (d) 2005, (e) 2010, and (f) 2015. NP = Nascentes do Parnaíba National Park; JSP = Jalapão State Park.
investments formulated for the region. The spatial groupings formed during the time series revealed consolidation of production in the Extremo Oeste Baiano, Sul Maranhense, and Sudoeste Piauiense mesoregions, highlighting the focus of soybean production in only 14 of the 337 municipalities that constitute the Matopiba region. The high numbers of municipalities forming groups with low soybean yield in the Matopiba region were also revealed. Therefore, in the case of a region with high agricultural potential, it is recommended that regional agricultural planning be undertaken with the objectives of reversing low yield groups and increasing the numbers of municipalities with high soybean production. The expansion of soybean in Matopiba is unlikely to grow indefinitely, since it depends on the combination of favorable topography, soil physics and climate, that is, a combination of simultaneous availability of flat terrains (plateaus), Latossolos, and relatively high levels of precipitation (mean annual rainfall > 1000 mm). The future of
cultivation were found in the Extremo Oeste Baiano, Sul Maranhense, and Sudoeste Piauiense mesoregions. Areas of low soybean yield were found located in the central eastern parts of Maranhão, western Maranhão, and northern Tocantins, as well as in the municipality of Wanderley in Bahia State, all of which are located in transition areas between biomes. Based on this spatiotemporal analysis, it was possible to conclude that soybean production and yield in the Matopiba region follow spatial patterns that reflect local or regional aptitudes. However, there is no spatial correlation between these two parameters. The results of this study could be useful in assisting with local and regional agricultural planning in the Matopiba region, as well as supporting studies in other fields of agriculture, the environment, and logistics. During the studied period, soybean production and yield showed high rates of growth in the Matopiba region. However, this growth has had considerable dependence on public policies and technological 64
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 6. Spatiotemporal dynamics of the local Moran index applied to soybean yield in the study area: (a) 1990, (b) 1995, (c) 2000, (d) 2005, (e) 2010, and (f) 2015. In 1990, cluster A is formed by the municipalities of Formosa do Rio Preto, Riachão das Neves, Barreiras, São Desidério, and Correntina and cluster B is formed by the municipalities of Dueré, Aliança do Tocantins, Gurupi, Peixe, Figueirópolis, and Araguaçu.
Conflicts of interest
grain production in Matopiba also seems to be environment-oriented, that is, committed to the Brazilian Forest Code and to the traders demanding for no-deforestation-related grain production. It is recommended that further studies be undertaken to analyze the spatial correlation between the variables of public policies and technological investments on the spatiotemporal dynamics of soybean cultivation in the Matopiba region. Analysis of the environmental variables in the Matopiba region is also recommended to assess the spatiotemporal effects of agricultural expansion on natural resources, such as soil erosion and surface and groundwater pollution.
None.
Acknowledgments We acknowledge two anonymous reviewers for valuable comments. The authors thank the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for providing the Master of Science scholarship for the first author. 65
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Donfouet, H.P.P., Barczak, A., Dessendre, C.D., Maigné, E., 2017. Crop production and crop diversity in France: a spatial analysis. Ecol. Econ. 134, 29–39. Espíndola, C.J., Cunha, R.C.C., 2015. A dinâmica geoeconômica recente da cadeia produtiva de soja no Brasil e no Mundo. Geotextos 11 (1), 217–238. Figueiredo, P.N., 2016. New challenges for public research organisations in agricultural innovation in developing economies: evidence from Embrapa in Brazil’s soybean industry. Q. Rev. Econ. Financ. 62, 21–32. Frota, A.B., Campelo, G.J.A., 1999. Evolução e perspectivas da produção de soja na região Meio-Norte do Brasil. In: Queiroz, M.A., Goedert, C.O., Ramos, S.R.R. (Eds.), Recursos Genéticos e Melhoramento de Plantas para o Nordeste brasileiro. Embrapa Semiárido, Petrolina, pp. 48–60. Guimarães, E.N., Leme, H.J.C., 1997. Caracterização histórica e configuração espacial da estrutura produtiva do Centro-Oeste. Redistribuição da população e meio ambiente. Campinas: Unicamp/Nepo, São Paulo e Centro-Oeste, pp. 25–65. IBGE, 2015. Instituto Brasileiro de Geografia e Estatística. IBGE; Produto Interno Bruto dos Municípios, Rio de Janeiro. Available at: https://ww2.ibge.gov.br/home/ estatistica/economia/pibmunicipios/2012/, Accessed: 28 Mar. 2017. IBGE, 2016. Instituto Brasileiro de Geografia e Estatística. IBGE; Cidades, Rio de Janeiro. Available at: https://cidades.ibge.gov.br/, Accessed: 28 Mar. 2017. IBGE, 2017. Instituto Brasileiro de Geografia e Estatística. Available at:. IBGE Pedologia, Rio de Janeiro Accessed: 13 Jul. 2018. https://downloads.ibge.gov.br/downloads_ geociencias.htm. IBGE, 2018. Instituto Brasileiro de Geografia e Estatística. METABD. Banco de metadados. Available at: https://metadados.ibge.gov.br/consulta/prnRelatorio Pesquisa.aspx? codPesquisa=PA. Accessed: 10 Mar. 2018. Javari, M., 2017. Spatial monitoring and variability of daily rainfall in Iran. Int. J. Appl. Environ. Sci. 12 (5), 801–824. Lima, D.P., Fioriolli, J.C., Padula, A.D., Pumi, G., 2017. The impact of Chinese imports of soybean on port infrastructure in Brazil: a study based on the concept of the Bullwhip Effect. J. Commod. Mark. 9 (November) available online. Magalhães, L.A., Miranda, E.E., 2014. MATOPIBA: Quadro Natural. Nota técnica 5. EMBRAPA. Available at:. Grupo de Inteligência Territorial Estratégica (GITE) (Accessed 12 January 2017). https://www.embrapa.br/gite/publicacoes/NT5_ Matopiba_Quadro_Natural.pdf. MAPA, 2015. Ministério da Agricultura, Pecuária e Abastecimento. Câmara setorial avalia situação do mercado de biodiesel. Available at: https://www.agrolink.com.br/ noticias/camara-setorial-avalia-situacao-do-mercado-de-biodiesel_220547.html (Accessed 20 December 2017). MAPA, 2016a. Ministério da Agricultura, Pecuária e Abastecimento. Estatísticas. Brasília: MAPA Available at: http://www.agricultura.gov.br/ assuntos/ sustentabilidade/ agroenergia/estatisticas. (Accessed 28 March 2017). MAPA, 2016b. Ministério da Agricultura, Pecuária e Abastecimento. Plano Agrícola e Pecuário 2016-2017. MAPA/SPA, Brasília 46 p. Available at: http://www.agricultura.gov.br/assuntos/sustentabilidade/plano-agricola-e-pecuario/ arquivospap/livropap1617.pdf (Accessed 5 March 2017). MAPA, 2017. Ministério da Agricultura, Pecuária e Abastecimento. Projeções do Agronegócio. Brasil 2016/17 a 2026/27. Projeções de longo prazo. Brasília: MAPA. 8ª ed., 125 p. Mesquita, B.A., 2011. O Desenvolvimento Desigual da Agricultura: A Dinâmica do Agronegócio e da Agricultura Familiar. EDUFMA, São Luís 110 p. Miranda, E.E., Magalhães, L.A., Carvalho, C.A., 2014. Proposta de delimitação territorial do Matopiba. Nota técnica 1. EMBRAPA. Grupo de Inteligência Territorial Estratégica (GITE) Available at: https://www.embrapa.br/gite/ publicacoes/ NT1_DelimitacaoMatopiba.pdf. (Accessed 12 January 2017). MMA, 2006. Ministério do Meio Ambiente. Programa Nacional de Conservação e Uso Sustentável do Bioma Cerrado. Programa Cerrado Sustentável. Núcleo dos Biomas Cerrado e Pantanal, Brasília: Distrito Federal. 56 p. Moran, P.A.P., 1948. The interpretation of statistical maps. J. R. Stat. Soc. 10, 243–251. Mueller, C.C., 2009. Agricultural, agrarian and environmental policy formation under Lula: the role of policy networks. In: Love, J., Werner, B. (Eds.), Brazil Under Lula. Economy, Politics and Society Under the Worker-President. Palgrave Macmillan, New York, pp. 135–150. Mueller, B., Mueller, C., 2016. The political economy of the Brazilian model of agricultural development: institutions versus sectoral policy. Q. Rev. Econ. Financ. 62, 12–20. OCDE/FAO, 2015. Food and agriculture organization of the United nations. OCDE-FAO Perspectivas Agrícolas 2015-2024. OECD Publishing, Paris Available at: https://doi. org/10.1787/agr_outlook-2015-es Accessed: 2 Feb. 2017. Paludzyszyn Filho, E., 1995. A cultura da soja no Maranhão. Londrina: Embrapa-CNPSo. 34 p. (Documentos, 84.). . Paludzyszyn Filho, F., Kiihl, R.A.S., Almeida, L.A., 1993. Desenvolvimento de cultivares de soja na região Norte e Nordeste do Brasil. In: Arantes, N.E., Souza, P.I.M. (Eds.), Cultura da soja nos Cerrados. Piracicaba: Potáfos, pp. 255–263. Portal Brasil, 2015. Brasília: Portal Brasil; “Matopiba” se consolida como nova fronteira agrícola do País. Available at: http://www.brasil.gov.br/economia-e-emprego/2015/ 10/matopiba-se-consolida-como-nova-fronteira-agricola-do-pais, Accessed: 25 Mar. 2017. Santos Filho, J.I.S., Talamini, D.J.D., Scheuermann, G.N., Bertol, T.M., 2016. Potencial do Matopiba na produção de aves e suínos. Revista de Política Agrícola 25 (2), 90–102. Silva, F.B., Santos, J.R.N., Feitosa, F.E.C.S., Silva, I.D.C., Araújo, M.L.S., Guterres, C.E., Santos, J.S., Ribeiro, C.V., Bezerra, D.S., Neres, R.L., 2016. Evidências de mudanças climáticas na região de transição Amazônia-Cerrado no estado do Maranhão. Rev. Bras. Meteorol. 31, 330–336. Spagnolo, T.F.O., 2011. Análise da dinâmica espacial da expansão agrícola no Oeste baiano entre 1984 e 2008: estudo de caso do município de São Desidério-BA. Dissertação (Mestrado) – Brasília. Universidade de Brasília 69 f.
Fig. 7. Number of municipalities with spatial groupings of (a) production and (b) yield of soybean in the Matopiba region for HH, HL, LL, and LH clusters (1990–2015).
References Alston, L., Melo, M., Mueller, B., Pereira, C., 2016. Brazil in Transition: Beliefs, Leadership and Institutional Change. Princeton University Press, Princeton. Alves, E.R.A., Pastore, A.C., 1978. Import substitution and implicit taxation of agriculture in Brazil. Am. J. Agric. Econ. 60 (5), 865–871. An, K., Ouyang, Y., 2016. Robust grain supply chain design considering post-harvest loss and harvest timing equilibrium. Transp. Res. Part E Logist. Transp. Rev. 88, 110–128. Andersen, L.E., Granger, C.W.J., Reis, E.J., Weinhold, D., Wunder, S., 2002. The Dynamics of Deforestation and Economic Growth in the Brazilian Amazon. Cambridge University Press, Cambridge. Anderson, M.C., Zolin, C.A., Sentelhas, P.C., Hain, C.R., Semmens, K., Yilmaz, M.T., Gao, F., Otkin, J.A., Tetrault, R., 2016. The evaporative stress index as an indicator of agricultural drought in Brazil: an assessment based on crop yield impacts. Remote Sens. Environ. 174, 82–99. Anselin, L., 1995. Local indicators of spatial association - LISA. Geogr. Anal. 27 (2), 93–115. Anselin, L., 2005. Exploring Spatial Data With GeoDa: a Workbook. Illinois: Urbanachampaign. 244 p.. . Bolfe, E., Victória, D., Contini, E., Bayma-Silva, G., Spinelli-Araújo, L., Gomes, D., 2016. MATOPIBA em crescimento agrícola: aspectos territoriais e socioeconômicos. Revista de Política Agrícola 25 (4), 38–62. Brown, J.C., Koeppe, M., Coles, B., Price, K.P., 2005. Soybean production and conversion of tropical forest in the Brazilian Amazon: the case of Vilhena, Rondônia. Ambio A J. Hum. Environ. 34, 462–469. Câmara, G., Carvalho, M.S., Cruz, O.G., Correa, V., 2002. Análise espacial de áreas. In: Fuks, S.D., Carvalho, M.S., Câmara, G., Monteiro, A.M.V. (Eds.), Análise espacial de dados geográficos. São José dos Campos: INPE, 44 p. Campos, M.C., 2010. A Embrapa Soja em Londrina – PR: a pesquisa agrícola de um país moderno (Ph.D. dissertation). Universidade Federal de Santa Catarina, Florianópolis 123 p. Cho, S., Newman, D.H., 2005. Spatial analysis of rural land development. For. Policy Econ. 7 (5), 732–744. Clemente, A.M., Carvalho Júnior, O.A., Guimarães, R.F., McManus, C., Turazi, C.M.V., Hermuche, P.M., 2017. Spatial-temporal patterns of bean crop in Brazil over the period 1990–2013. ISPRS Int. J. Geoinf. 6 (107), 18. Dall’Agnol, A., 2008. Soja: o fenômeno brasileiro. EMBRAPA, Londrina Available at: https://www.agrolink.com.br/colunistas/coluna/soja–o-fenomeno-brasileiro_383 966.html. Accessed: 10 Jul. 2018. Delgado, G.C., 1985. Capital financeiro e agricultura no Brasil. Unicamp/Cone, Campinas 240 p.
66
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
spatial and temporal variation in agricultural output elasticities in Turkey. Agric. Econ. 45, 279–290. Yuan, Y., Cave, M., Zhang, C., 2017. Using local Moran’s I to identify contamination hotspots of rare earth elements in urban soils of London. Appl. Geochem. 88 (Part B), 167–178. Zhang, T., Lin, G., 2016. On Moran’s I coefficient under heterogeneity. Comput. Stat. Data Anal. 95, 83–94.
Studte, M., 2008. A expansão da fronteira agrícola e a agricultura familiar no Cerrado Maranhense. Dissertação (Mestrado) – São Paulo. Universidade de São Paulo; Berlin: Universidade Técnica de Berlin 55 f. USDA United States Department of Agriculture, 2017. World Agricultural Production. Available at: https://apps.fas.usda.gov/psdonline/circulars/ production.pdf. (Accessed 28 March 2017). . Yu, T.H.E., Cho, S., Koç, A.A., Boluk, G., Kim, S.G., Lambert, D.M., 2014. Evaluating
67
Contents lists available at ScienceDirect
Land Use Policy journal homepage: www.elsevier.com/locate/landusepol
Spatiotemporal dynamics of soybean crop in the Matopiba region, Brazil (1990–2015)
T
⁎
Mayara Lucyanne Santos de Araújoa, , Edson Eyji Sanob, Édson Luis Bolfec, Jessflan Rafael Nascimento Santosd, Juliana Sales dos Santosd, Fabrício Brito Silvad a
Graduate Program in Applied Geoscience, University of Brasília, Brazil Embrapa Cerrados, Brazil c Embrapa Secretaria de Inteligência e Relações Estratégicas, Brazil d Universidade Ceuma, Brazil b
A R T I C LE I N FO
A B S T R A C T
Keywords: Agricultural expansion Soybean Moran index Geostatistics
Brazil is a world leader in the production and export of grains, particularly soybeans. The newest agricultural frontier in Brazil is the Matopiba region, which is a continuous zone formed by the states of Maranhão, Tocantins, Piauí, and Bahia, located mostly within the Cerrado biome. The objective of this study was to analyze the spatiotemporal dynamics of soybean production and yield in the Matopiba region. We analyzed municipality-based planted areas and production data obtained by the Brazilian Institute of Geography and Statistics during 1990–2015. Yield was estimated from the production and planted area, and the data were analyzed using global and local Moran indices. The results showed that soybean production in the Matopiba region does not occur randomly. Positive and significant autocorrelation was found at the beginning of the time series among those municipalities located in the west of Bahia. This region influenced the soybean expansion from south to north. Currently, high-production areas are concentrated in two autocorrelated blocks: one in western Bahia and the other in the central Matopiba region. Analysis of spatial autocorrelation involving yield showed a decreasing trend at the end of the time series. The presence of municipalities with high yield surrounded by others with low yield, and vice-versa, were observed. The findings of this study could assist local and regional agricultural planning in the Matopiba region, and support related analyses in other fields of agriculture, the environment, and logistics.
1. Introduction Because of its considerable territorial extent and its favorable climate, topography, and soil physical properties that support extensive rainfed crop production, Brazil has become one of the main exporters of agricultural commodities, such as soybean, corn, coffee, sugarcane, and cotton (MAPA, 2016a; Mueller and Mueller, 2016). In 2015, the annual crops that presented the largest planted areas in Brazil were soybeans (32 million ha), maize (16 million ha), and sugarcane (10 million ha) (IBGE, 2015). Currently, Brazil and the United States are the leading producers of soybeans (MAPA, 2015). In 2015, Brazil exported approximately 57 million tons of soybeans (An and Ouyang, 2016), primarily to China (Lima et al., 2017). In 2016/2017, Brazil produced 114 million tons of soybeans (USDA, 2017). Analysis by the United Nations Food and Agriculture Organization has indicated that soybean production in Brazil will increase by 37% over the next 10 years (OCDE/
⁎
FAO, 2015). Soybean planting began in Brazil in the 1940s as an option for crop rotation with wheat (Brown et al., 2005), and the crop adapted well in the southernmost part of the country because of the temperate climate. According to Paludzyszyn Filho et al. (1993), soybean became important in Rio Grande do Sul State because of the exportation. Gradually, because of investments by the Brazilian government in research institutions, soybean plants were genetically modified to improve adaptation in other regions of the country with tropical climate (Guimarães and Leme, 1997; Andersen et al., 2002). In addition, the use of chemical fertilizers was implemented to correct the predominantly acidic soils with low natural fertility found mainly in central parts of the country (Delgado, 1985). Brazilian Agricultural Research Institute (Embrapa Cerrados), created in 1973 by the Ministry of Agriculture, played a key role here since its research allowed remarkable increase in the soybean production in Brazil (Mueller and Mueller, 2016).
Corresponding author at: Universidade de Brasília, Campus Universitário Darcy Ribeiro, CEP: 70910-900, Brasília, DF, Brazil. E-mail address: [email protected] (M.L.S.d. Araújo).
https://doi.org/10.1016/j.landusepol.2018.09.040 Received 21 April 2018; Received in revised form 23 July 2018; Accepted 29 September 2018 0264-8377/ © 2018 Elsevier Ltd. All rights reserved.
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 1. Map of the Matopiba region in Brazil, subdivided into 10 mesoregions. Each mesoregion corresponds to a subdivision of a Brazilian state that aggregates a varying number of municipalities with high levels of economic and social similarity. State identification: MA = Maranhão, TO = Tocantins, PI = Piauí, and BA = Bahia.
regions of western Bahia State, southeastern Goiás State (municipalities of Jataí and Rio Verde), and the central region of Mato Grosso State (municipalities of Lucas do Rio Verde, Sinop, and Sorriso). Agricultural frontier is defined as a region dominated by natural vegetation that started to face intensive agriculture-related land occupation. The most recent agricultural frontier of the Cerrado is the Matopiba region, a continuous zone formed by the states of Maranhão, Tocantins, Piauí, and Bahia (Miranda et al., 2014). In this region, infrastructure is poor, land prices are cheap, and the climate and topographic relief are favorable for rainfed agriculture. Currently, soybean is the main agricultural crop of Matopiba (MAPA, 2017). Following the rapid agricultural expansion in Matopiba, the Brazilian government issued Federal Decree No. 8,447 on May 6, 2015, establishing an Agricultural Development Plan for Matopiba. The purpose of this decree was to promote and coordinate public policies for economic and sustainable development of agricultural and livestock activities in the Matopiba region. The plan proposes guidelines for federal programs, projects, and actions to be undertaken with the objectives of improving both the living standards of the local population and the economic growth of the country. For this plan to succeed, it is of great relevance to understand the spatiotemporal dynamics of the crops produced in this region, with emphasis on the spatial clusters that
At the same time, because of the approved agrarian reform legislation, large landowners started to prevent against land loss for small farmers, rural workers, and landless peasants by increasing their production and investing in mechanization (Mueller and Mueller, 2016). Other relevant factors that contributed to the expansion of soybean in Central Brazil included the public tax incentives to open new areas for soybean and for establishment of companies for grain storage and processing, the availability of large areas with flat topography (plateaus), that is, favorable for mechanization, the relatively high precipitation conditions for rainfed agriculture, and the relatively good economic and technological levels of the farmers from southern part of the country that migrated to Cerrado (Dall’Agnol (2008); Campos, 2010). Since 1990, the Brazilian government reduced its direct involvement in agriculture. On the other hand, private sector started to invest strongly in this sector (Alston et al., 2016). The country also implemented a series of more open and predictable political and economic institutional arrangements (Alves and Pastore, 1978). This lower need for direct political intervention in areas such as credit and price management allowed the country to grow even more in the agricultural sector (Mueller, 2009). New agricultural frontiers were created in the Cerrado biome, e.g., 58
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
neighbors, evaluating whether the distribution is grouped or random (Anselin, 2005). Positive values (between 0 and +1) show direct correlation and negative values (between 0 and −1) reflect an inverse correlation (Câmara et al., 2002). Spatial autocorrelation is computed only for neighbors of the first order in space, as established by the wi,j weights of Eq. (1) (Anselin, 1995):
might eventually appear over time. This understanding will assist the agricultural, environmental, and economic planning activities of the region. Despite the recent increase in the number of studies focused on the Matopiba region, the use of statistics such as the Moran index for spatial analysis has not been explored yet, leading to some unanswered questions. 1) Does soybean production and yield follow spatial patterns that reflect local or regional aptitude? 2) If spatial patterns of production and yield occur, do they vary systematically over time, demonstrating the migration of aptitude to soybean cultivation? 3) Is there any spatial correlation between soybean production and yield? The answers to these questions will assist both the preparation of local and regional agricultural planning and the ongoing research and analysis in fields such as agriculture, the environment, and logistics. The primary objective of this research was to analyze the spatiotemporal dynamics of soybean production in the Matopiba region (1990–2015) by applying spatial statistics techniques on time series data.
n
I=
n
n ∑i = 1 ∑ j = 1 wi, j z i zj → so = n so ∑i = 1 z i 2
n
n
∑ ∑ wi,j (1)
i=1 j=1
where zi is the deviation of an attribute for the resource i from its average, zj is the deviation of an attribute for resource j from its average, wi,j is the weight assigned according to the connection between areas i and j, n is the total number of attributes, and So is the aggregate of all spatial weights. The local Moran index (Ii) is a measure of the spatial arrangement that assesses the correlation between an observation and its neighborhood (Anselin, 1995). It can capture possible local patterns of spatial autocorrelation and identify the presence of spatial clusters of similar values or anomalous objects (outliers). Derived values that are significantly high and positive indicate the presence of clusters of similar values, whereas significantly low values indicate inequality within the region (Zhang and Lin, 2016). The result is expressed as clusters of high values (HH) and of low values (LL), and as outliers where a high value is surrounded by low values (HL) or a low value is surrounded by high values (LH). The Ii can be defined by the following equation:
2. Materials and methods 2.1. Study area The selected study area was the Matopiba region, which extends over parts of the states of Maranhão, Tocantins, Piauí, and Bahia (Miranda et al., 2014; Santos Filho et al., 2016). The Matopiba region comprises 10 mesoregions (5 in the state of Maranhão, 2 in Tocantins, 1 in Piauí, and 2 in Bahia) and 337 municipalities (Fig. 1). A mesoregion corresponds to a subdivision of a Brazilian state and it can encompass a varying number of municipalities with high levels of economic and social similarity. The study region occupies approximately 73 million ha located in the area within 2°30′–15°15′S latitude, 42°00′–50°00′W longitude. In 2014/2015, Matopiba contributed 9.4% of the 209.5 million tons of grains produced in Brazil (Portal Brasil, 2015). This region encompasses 50 federal, state, and municipal conservation units (7.2 million ha) and 23 indigenous lands (3.6 million ha). Latossolo (Oxisols) is the dominant soil type, covering 27.8 million ha (38% of the Matopiba region) (Magalhães and Miranda, 2014). Latossolos are welldeveloped deep soils, which are highly weathered with low soil fertility, good permeability, and high porosity. The second largest class of soils in the Matopiba region is Neossolo Quartzarênico (deep, sandy soils, composed largely of quartz), which occupies an area of 18 million ha (25% of the region).
Ii =
x i−x¯ si2
n
n
∑
wi, j (x j−x¯) → si2 =
j = 1, j ≠ i
∑ j = 1, j ≠ i (x j−x¯)² n−1
−x¯² (2)
where xi is the value of the attribute considered in area i, xj is the value of the attribute considered in area j, x¯ is the average of the corresponding attribute, wi,j is the weight assigned according to the connection between areas i and j, and n is total number of attributes. 3. Results 3.1. Spatiotemporal dynamics The Matopiba region showed significant growth in soybean production during the time series (Fig. 2): from 260,624 t in 1990 to 10,758,927 t in 2015, an increase of 4028% in 25 years. The municipality of São Desidério (Extremo Oeste Baiano mesoregion) presented the highest increase in production, rising from 1999 t in 1990 to 1,134,000 t in 2015 (an increase of more than 28,000%). This municipality presents average mean precipitation of 1145 mm (Tropical Rainfall Measuring Mission data, 1998–2015 time series), relatively extensive plateaus with average elevation of 778 m and average slope of 1.8% (Shuttle Radar Topography Mission data), dystrophic yellow-red Latossolo (deep, acidic, low to intermediate levels of soil fertility) (IBGE, 2017). Such combination of precipitation, topography and soil type favors the implementing of large-scale grain production. At the beginning of the time series, the highest production was concentrated in the Extremo Oeste Baiano mesoregion. Subsequently, there was an expansion from south to north, involving not only the municipalities of Extremo Oeste Baiano but also the municipalities of the Sul Maranhense and Sudoeste Piauiense mesoregions. However, we must be aware that agricultural expansion over these areas is not so straightforward since their biophysical conditions are not so similar to those from western Bahia region. Overall, in the Sul Maranhense and Sudoeste Piauiense mesoregions, the crop production is more sensitive to drought effects (less rainfall since they are closer to the semi-arid Caatinga), there are more occurrence of Neossolos Quartzarênicos (sandy soils) and the extension of plateaus are smaller. The municipality of São Desidério presented linear growth in soybean production during the time series (Fig. 3). It has had the second
2.2. Approach The Municipality Agricultural Production (PAM – Produção Agrícola Municipal) time series data obtained by the Brazilian Institute of Geography and Statistics (IBGE) during 1990–2015 comprised the basic data for this study. These data are intended to provide the basis for statistical analysis on the production, planting, harvesting areas, and price of Brazilian agricultural products at the municipality level on an annual basis for the entire country. The data are obtained through questionnaires submitted by IBGE agents to producers and technicians involved in the agricultural sector (IBGE, 2017; IBGE, 2018). Here, we must be aware that questionnaire-based data present some limitations since they are dependent on information provided, for example, by farmers and seed sellers, there may have lack of data in some municipalities during the time series, and some farms can be located in-between two or more municipalities. The spatiotemporal analyses in this study were based on soybean production data (tons) and the planted and harvested areas (ha). The yield (tons ha−1) was estimated based on the ratio between production and planted area. These data were analyzed based on global and local Moran indices (Moran, 1948). The main purpose of the global Moran index (I) is to verify the degree of spatial autocorrelation between the 59
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 2. Spatiotemporal dynamics of soybean production in the Matopiba region based on IBGE Municipality Agricultural Production (PAM) data: (a) 1990, (b) 1995, (c) 2000, (d) 2005, (e) 2010, and (f) 2015. TP = total production; Mt = million tons. See Fig. 1 for mesoregion identification.
yield in the region (2.1 t ha−1) was found in the municipality of Guaraí. By 2015, maximum yield had increased to 4.0 t ha−1 in the municipality of Serra do Ramalho, Bahia State.
largest territorial expansion in Bahia State of approximately 15,166 km² (IBGE, 2016), and it is in the lead of the municipalities with the highest agricultural production in the country. Initially, soybean yield within the study area was most prominent in the municipality of Guaraí (Tocantins State), Extremo Oeste Baiano mesoregion, southwest Tocantins State and the central Matopiba region (Fig. 4). At the end of the time series, yield had increased in different areas of the Matopiba region, particularly in the Extremo Oeste Baiano mesoregion, southwest Tocantins State, and western parts of Maranhão State, an Amazon–Cerrado transition area. In 1990, maximum soybean
3.2. Spatial statistics The global Moran index applied to the data of soybean production, yield, planted area, and harvested area produced values greater than zero, indicating the existence of spatial patterns among the municipalities belonging to the study area (Table 1). The p-values were < 0.01 60
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 3. Evolution of soybean production in the municipality of São Desidério, Bahia State (1990–2015).
showed statistically significant HH, HL, LL, and LH clusters (Fig. 6). In 1990, there was indication of the presence of two HH zones with high yield. One was located in the Extremo Oeste Baiano mesoregion (Formosa do Rio Preto, Riachão das Neves, Barreiras, São Desidério and Correntina municipalities), while the other was located in the Sudoeste Tocantinense (Dueré, Aliança do Tocantins, Gurupi, Peixe, Figueirópolis and Araguaçu municipalities) mesoregion, together with the Pedro Afonso municipality in the central part of Tocantins State. In 1990, the municipality of Goiatins (Tocantins State) presented an HL cluster. However, by 1995, both the Pedro Afonso and the Goiatins municipalities were unable to maintain their high yield levels. In 1995, another HH cluster appeared in the center of the Matopiba region. In this year, the municipality of Ribeiro Gonçalves (Sudoeste Piauiense mesoregion) presented low yield autocorrelation with its neighboring municipalities with high yield (i.e., an LH cluster). However, in 2000, this changed to an HH cluster, indicating increased soybean yield in this municipality. The same characteristic was observed in the São Félix de Balsas municipality (Maranhão State) in 2000. This result confirms the importance of the relationship between neighboring municipalities regarding the growth of agriculture. In 2000, spatial autocorrelation of the municipalities was concentrated in the Extremo Oeste Baiano, Sul Maranhense, and Sudoeste Piauiense mesoregions (HH cluster). In 2005, 2010, and 2015, two LL clusters were produced, involving the municipalities of the Centro Maranhense, Leste Maranhense, Oeste Maranhense, and Ocidental do Tocantins mesoregions. In 2015, an LL cluster was also observed in the municipality of Wanderley (Bahia State). It can be seen from Fig. 7 that production and yield data were not correlated over time. Soybean production presented smaller numbers of spatial groupings compared with soybean yield. This indicates that municipalities with greater production were not necessarily the municipalities with the largest areas of cultivation. The number of municipalities with high soybean productions that were surrounded by other municipalities with high soybean productions varied from 5–14. This indicates, despite the considerable territorial extension of the Matopiba region and its importance to Brazilian agriculture, the productive zones were concentrated in only a few municipalities. The concentration remained constant throughout the time series. In 1990, the HH-rated municipalities accounted for 83.9% of total soybean production (260,624 t). In 2015, they represented 61.2% of the total production (10,758,927 t). The four spatial groupings of yield (HH, HL, LL, and LH), which ranged from 1 to 71 municipalities during the time series (1990–2015), confirmed the high numbers of municipalities with large areas of soybean plantations. However, it is important to highlight that the numbers of municipalities in the Matopiba region rated within LL clusters are growing and that they account for more municipalities than in HH clusters.
for the entire time series, indicating statistical significance for the spatial autocorrelation. Moran index for planted areas and harvested areas were quite similar, indicating that there were no significant losses of planted area during the crop cycles because of some disease or fire. The results of the local Moran index for production and yield data are presented in Figs. 5 and 6, respectively. HH indicates higher production/yield in a certain municipality and in its neighboring municipalities; HL indicates high yield in a certain municipality and low yield in the neighboring municipalities; LH indicates low yield in a certain municipality and high production/yield in the surrounding municipalities; and LL indicates lower yield in a certain municipality and in the surrounding municipalities. The local Moran index for soybean production presented an HH cluster comprising the municipalities of Barreiras, Correntina, Formosa do Rio Preto, Riachão das Neves, and São Desidério (Extremo Oeste Baiano mesoregion) (Fig. 5). This cluster appears from the beginning of the time series (1990), consolidated over time, and expanded into the north, starting in 2000 in the southern portions of the Sul Maranhense and Sudoeste Piauiense mesoregions. The highest value of the local Moran index was observed in the municipality of São Desidério, which suggests high probability of it having influenced other municipalities to consolidate soybean production within the study area, mainly in terms of transferring new knowledge and technologies to neighboring municipalities. In 1995, the municipality of Jaborandi joined the HH cluster formed in the Extremo Oeste Baiano mesoregion. In 2005, the Luís Eduardo Magalhães municipality also joined the HH cluster. In 2000, in addition to the positive autocorrelation cluster located in the Extremo Oeste Baiano, another HH cluster formed in the Sul Maranhense mesoregion, which comprised the municipalities of Balsas and Tasso Fragoso. In 2005, the municipalities of Riachão (Maranhão State), Campos Lindos (Tocantins State), and the municipalities of Ribeiro Gonçalves and Baixa Grande do Ribeiro (Piauí State) joined this cluster. In 2010, the municipality of Santa Filomena (Piauí State) also joined this cluster. The consolidation of this cluster in the central part of the Matopiba region was confirmed in 2015. The results identified the presence and consolidation of two zones of high soybean production in the Matopiba region. The presence of the Jalapão State Park and the Nascentes do Parnaíba National Park is mainly responsible for the geographical discontinuity between these two areas. The result indicates that soybean production in the Matopiba region has not occurred randomly but it has integrated in well-defined spatial patterns during the time series. The most relevant factor controlling such spatial pattern is that the soybean expansion is dependent on the availability of a combination of flat terrains, favorable soil physical properties (texture) as well as enough rainfall conditions during the crop cycle. Therefore, expansion of soybean in Matopiba is likely to happen only in areas with such combination of biophysical properties. The results also confirm statistically the occurrence of the south–north expansion of soybean production in this region. The local Moran index for soybean yield in the Matopiba region 61
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 4. Spatiotemporal dynamics of soybean yield in the Matopiba region based on IBGE Municipality Agricultural Production (PAM) data: (a) 1990, (b) 1995, (c) 2000, (d) 2005, (e) 2010, and (f) 2015. PT = total production. AY = average yield. See Fig. 1 for mesoregion identification.
4. Discussion
and shown effective in identifying spatial clusters produced over time. Thus, these indices constitute important tools in environmental and agricultural studies, as exemplified by Javari (2017); Clemente et al. (2017), and Yuan et al. (2017). Matopiba has been in the center of discussion of dominant model of Cerrado occupation (based on the complete removal of native vegetation, that is, deforestation) since it is part of largest region where native vegetation of Cerrado is still preserved. In fact, Matopiba is a very interesting region for implementation of 11 components defined by the Programa Cerrado Sustentável (MMA, 2006), which includes actions of biodiversity conservation, sustainable use of biodiversity, water resources management, and environmentally sustainable agriculture.
To make reliable decisions regarding regional agricultural planning and development, knowledge of the spatiotemporal patterns of crops and their performance over time is essential. Therefore, the analyses performed in this study, evidencing the spatial patterns of soybean cultivation in the Matopiba region over 25 years, could provide important support for the development of agriculture in the region, particularly within the context of environmentally sustainable progress. The spatial statistical techniques adopted here have been used in other agricultural studies (Cho and Newman, 2005; Yu et al., 2014; Donfouet et al., 2017). The global and local Moran indices have been used widely 62
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
The results of the local Moran index for soybean yield in the Matopiba region showed that soybean is produced in areas with defined spatial patterns. It is important to highlight the presence of areas with LL clusters, which were identified mainly in central eastern parts of Maranhão, western Maranhão, and northern Tocantins from 2005 to 2015, as well as in the municipality of Wanderley (Bahia State) in 2015. These areas, which will require public policy support and financial incentives to increase production, tend to be located in transitional areas between biomes. The central eastern area of Maranhão, western Maranhão, and northern Tocantins regions are located in the Amazon–Cerrado transition region, while the municipality of Wanderley is located in the Cerrado–Caatinga transition region. According to Silva et al. (2016), a transition area between different biomes presents highly diverse ecosystems and climatic conditions, as well as areas with lower natural fertility, making them particularly vulnerable to climate change. Another important issue in the Matopiba region is the sensitivity to the El Niño events. This region faced an El Niño event during the period of 2012–2015 that led to a strong decrease in the crop yield, especially in municipalities with worse edaphoclimatic conditions. During the 2015/2016 drought event, there was an overall reduction of grain production of approximately 35% because of the 47-day dry spell occurrence (Canal Rural, 2016). Although our statistical results showed unclear influence of El Niño in the 2015 data set, such climatic event may cause some relevant impact in the local Moran Index. It was also found that the most productive zones were concentrated in only a few areas of the Matopiba region. Figueiredo (2016) reported the growth in private research by large multinational companies in productive areas. The consequence of this process has been regional concentration of soybean production in areas that have received the greatest technological contributions from the investments of in situ research, mainly in relation to the testing of the performance of cultivar suitability to the climatic conditions of the region. The future of soybean expansion in Matopiba seems to be dependent on the development of new crop varieties more resistant to drought and dry spells. Brazilian farmers use a threshold of mean annual average rainfall of 1000 mm to determine if an area is suitable for rainfed agriculture. Development of new varieties either by research institutions such as Embrapa or by private companies should lower this threshold for 800 mm or so. Transportation network is another important issue since road condition in this region is very poor and railroad does not exist yet. The rapid deforestation in Matopiba is also arousing attention in terms of compliance with the Brazilian Forest Code. Though the New Forest Code determine that only 20% of farms located in the Cerrado biome must be preserved by native vegetation (Reserva Legal), farmers also must protect their areas considered as permanently preserved area (e.g., riparian forests along streams). Currently, farmers who practice illegal deforestation have their area embargoed by the Brazilian Institute for Environment and Renewable Natural Resources (IBAMA) and are not allowed to sell their products until the area is restored. The increasing number of traders committed to no-deforestation grain production and the increasing number of protests from civil society organizations such as the Cerrado Manifesto will also contribute significantly to the development of environmentally sustainable agriculture in Matopiba. Brazilian government is also starting a program to monitor Cerrado deforestation at an annual basis and using Landsat-like, moderate spatial resolution satellite data. This will be another important tool to monitor the rate of deforestation not only in Matopiba but also in other important regions of Cerrado in terms of grain production.
Table 1 Global Moran index for production, yield, planted area, and harvested area variables in 1990, 1995, 2000, 2005, 2010, and 2015. Period
Moran index
z-score
p-value
Result
Production (t) 1990 1995 2000 2005 2010 2015
0.073 0.076 0.082 0.132 0.105 0.129
4.396 4.382 4.680 6.752 5.366 6.373
< 0.01 < 0.01 < 0.01 < 0.01 < 0.01 < 0.01
Grouped Grouped Grouped Grouped Grouped Grouped
Yield (t ha−1) 1990 1995 2000 2005 2010 2015
0.072 0.190 0.188 0.444 0.428 0.389
3.506 8.601 8.516 19.794 19.088 17.363
< 0.1 0 0 0 0 0
Grouped Grouped Grouped Grouped Grouped Grouped
Planted area (ha) 1990 0.074 1995 0.078 2000 0.081 2005 0.132 2010 0.109 2015 0.135
4.480 4.447 4.665 6.764 5.521 6.608
< 0.01 < 0.01 < 0.01 0 0 0
Grouped Grouped Grouped Grouped Grouped Grouped
Harvested area (ha) 1990 0.074 1995 0.078 2000 0.081 2005 0.132 2010 0.109 2015 0.135
4.474 4.447 4.665 6.764 5.520 6.606
< 0.01 < 0.01 < 0.01 0 0 0
Grouped Grouped Grouped Grouped Grouped Grouped
The results of the global spatial autocorrelations applied to the variables of soybean production and yield within the study area showed positive autocorrelation between municipalities. The results demonstrated tendencies of expansion and consolidation of the use of soybean in the Matopiba region. This region is now considered the major agricultural frontier for agribusiness investment in Brazil (Anderson et al., 2016). According to Espíndola and Cunha (2015), several economic and political incentives relating to agricultural production have been directed toward the central part of the country. For example, according to the 2016–2017 Agricultural and Livestock Plan (MAPA, 2016b), a total of approximately US$55 billion has been made available to rural producers and their cooperatives to finance their activities. However, the large expansion of soybean that was verified in the Extremo Oeste Baiano is unlikely to occur in other mesoregions because of the biophysical conditions, that is, smaller extension of plateaus, higher occurrence of sandy soils (Neossolos Quartzarênicos) and higher sensitivity to the occurrence of dry spells (border to the semi-arid Caatinga biome). The local Moran index revealed that soybean production has been consolidated in the Extremo Oeste Baiano. According to Spagnolo (2011), the agricultural expansion of soybean cultivation in this mesoregion was consolidated in 2000, following the availability of low-tax loans offered by the federal government as part of various programs to increase agricultural production in Bahia State. One example is the AGRINVEST Investment Program for the Modernization of Agriculture in Bahia. From 2000, the Sul Maranhense mesoregion presented high and statistically significant autocorrelation; a characteristic that was also identified from 2005 in the Sudoeste Piauiense mesoregion. In Maranhão, the greatest incentives have been the improvement of the state’s road infrastructure (Frota and Campelo, 1999; Studte, 2008), valorization of soybean prices in international markets, credit incentives (Mesquita, 2011), and technological investments, which have helped transform the traditional subsistence agrarian structure into a technified agrarian structure (Paludzyszyn Filho, 1995; Bolfe et al., 2016).
5. Conclusions This study identified important zones of soybean production in the Matopiba region of Brazil based on the spatial statistics of the global and local Moran indices. The most productive areas of soybean 63
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 5. Spatiotemporal dynamics of the local Moran index applied to soybean production in the Matopiba region: (a) 1990, (b) 1995, (c) 2000, (d) 2005, (e) 2010, and (f) 2015. NP = Nascentes do Parnaíba National Park; JSP = Jalapão State Park.
investments formulated for the region. The spatial groupings formed during the time series revealed consolidation of production in the Extremo Oeste Baiano, Sul Maranhense, and Sudoeste Piauiense mesoregions, highlighting the focus of soybean production in only 14 of the 337 municipalities that constitute the Matopiba region. The high numbers of municipalities forming groups with low soybean yield in the Matopiba region were also revealed. Therefore, in the case of a region with high agricultural potential, it is recommended that regional agricultural planning be undertaken with the objectives of reversing low yield groups and increasing the numbers of municipalities with high soybean production. The expansion of soybean in Matopiba is unlikely to grow indefinitely, since it depends on the combination of favorable topography, soil physics and climate, that is, a combination of simultaneous availability of flat terrains (plateaus), Latossolos, and relatively high levels of precipitation (mean annual rainfall > 1000 mm). The future of
cultivation were found in the Extremo Oeste Baiano, Sul Maranhense, and Sudoeste Piauiense mesoregions. Areas of low soybean yield were found located in the central eastern parts of Maranhão, western Maranhão, and northern Tocantins, as well as in the municipality of Wanderley in Bahia State, all of which are located in transition areas between biomes. Based on this spatiotemporal analysis, it was possible to conclude that soybean production and yield in the Matopiba region follow spatial patterns that reflect local or regional aptitudes. However, there is no spatial correlation between these two parameters. The results of this study could be useful in assisting with local and regional agricultural planning in the Matopiba region, as well as supporting studies in other fields of agriculture, the environment, and logistics. During the studied period, soybean production and yield showed high rates of growth in the Matopiba region. However, this growth has had considerable dependence on public policies and technological 64
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Fig. 6. Spatiotemporal dynamics of the local Moran index applied to soybean yield in the study area: (a) 1990, (b) 1995, (c) 2000, (d) 2005, (e) 2010, and (f) 2015. In 1990, cluster A is formed by the municipalities of Formosa do Rio Preto, Riachão das Neves, Barreiras, São Desidério, and Correntina and cluster B is formed by the municipalities of Dueré, Aliança do Tocantins, Gurupi, Peixe, Figueirópolis, and Araguaçu.
Conflicts of interest
grain production in Matopiba also seems to be environment-oriented, that is, committed to the Brazilian Forest Code and to the traders demanding for no-deforestation-related grain production. It is recommended that further studies be undertaken to analyze the spatial correlation between the variables of public policies and technological investments on the spatiotemporal dynamics of soybean cultivation in the Matopiba region. Analysis of the environmental variables in the Matopiba region is also recommended to assess the spatiotemporal effects of agricultural expansion on natural resources, such as soil erosion and surface and groundwater pollution.
None.
Acknowledgments We acknowledge two anonymous reviewers for valuable comments. The authors thank the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for providing the Master of Science scholarship for the first author. 65
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
Donfouet, H.P.P., Barczak, A., Dessendre, C.D., Maigné, E., 2017. Crop production and crop diversity in France: a spatial analysis. Ecol. Econ. 134, 29–39. Espíndola, C.J., Cunha, R.C.C., 2015. A dinâmica geoeconômica recente da cadeia produtiva de soja no Brasil e no Mundo. Geotextos 11 (1), 217–238. Figueiredo, P.N., 2016. New challenges for public research organisations in agricultural innovation in developing economies: evidence from Embrapa in Brazil’s soybean industry. Q. Rev. Econ. Financ. 62, 21–32. Frota, A.B., Campelo, G.J.A., 1999. Evolução e perspectivas da produção de soja na região Meio-Norte do Brasil. In: Queiroz, M.A., Goedert, C.O., Ramos, S.R.R. (Eds.), Recursos Genéticos e Melhoramento de Plantas para o Nordeste brasileiro. Embrapa Semiárido, Petrolina, pp. 48–60. Guimarães, E.N., Leme, H.J.C., 1997. Caracterização histórica e configuração espacial da estrutura produtiva do Centro-Oeste. Redistribuição da população e meio ambiente. Campinas: Unicamp/Nepo, São Paulo e Centro-Oeste, pp. 25–65. IBGE, 2015. Instituto Brasileiro de Geografia e Estatística. IBGE; Produto Interno Bruto dos Municípios, Rio de Janeiro. Available at: https://ww2.ibge.gov.br/home/ estatistica/economia/pibmunicipios/2012/, Accessed: 28 Mar. 2017. IBGE, 2016. Instituto Brasileiro de Geografia e Estatística. IBGE; Cidades, Rio de Janeiro. Available at: https://cidades.ibge.gov.br/, Accessed: 28 Mar. 2017. IBGE, 2017. Instituto Brasileiro de Geografia e Estatística. Available at:. IBGE Pedologia, Rio de Janeiro Accessed: 13 Jul. 2018. https://downloads.ibge.gov.br/downloads_ geociencias.htm. IBGE, 2018. Instituto Brasileiro de Geografia e Estatística. METABD. Banco de metadados. Available at: https://metadados.ibge.gov.br/consulta/prnRelatorio Pesquisa.aspx? codPesquisa=PA. Accessed: 10 Mar. 2018. Javari, M., 2017. Spatial monitoring and variability of daily rainfall in Iran. Int. J. Appl. Environ. Sci. 12 (5), 801–824. Lima, D.P., Fioriolli, J.C., Padula, A.D., Pumi, G., 2017. The impact of Chinese imports of soybean on port infrastructure in Brazil: a study based on the concept of the Bullwhip Effect. J. Commod. Mark. 9 (November) available online. Magalhães, L.A., Miranda, E.E., 2014. MATOPIBA: Quadro Natural. Nota técnica 5. EMBRAPA. Available at:. Grupo de Inteligência Territorial Estratégica (GITE) (Accessed 12 January 2017). https://www.embrapa.br/gite/publicacoes/NT5_ Matopiba_Quadro_Natural.pdf. MAPA, 2015. Ministério da Agricultura, Pecuária e Abastecimento. Câmara setorial avalia situação do mercado de biodiesel. Available at: https://www.agrolink.com.br/ noticias/camara-setorial-avalia-situacao-do-mercado-de-biodiesel_220547.html (Accessed 20 December 2017). MAPA, 2016a. Ministério da Agricultura, Pecuária e Abastecimento. Estatísticas. Brasília: MAPA Available at: http://www.agricultura.gov.br/ assuntos/ sustentabilidade/ agroenergia/estatisticas. (Accessed 28 March 2017). MAPA, 2016b. Ministério da Agricultura, Pecuária e Abastecimento. Plano Agrícola e Pecuário 2016-2017. MAPA/SPA, Brasília 46 p. Available at: http://www.agricultura.gov.br/assuntos/sustentabilidade/plano-agricola-e-pecuario/ arquivospap/livropap1617.pdf (Accessed 5 March 2017). MAPA, 2017. Ministério da Agricultura, Pecuária e Abastecimento. Projeções do Agronegócio. Brasil 2016/17 a 2026/27. Projeções de longo prazo. Brasília: MAPA. 8ª ed., 125 p. Mesquita, B.A., 2011. O Desenvolvimento Desigual da Agricultura: A Dinâmica do Agronegócio e da Agricultura Familiar. EDUFMA, São Luís 110 p. Miranda, E.E., Magalhães, L.A., Carvalho, C.A., 2014. Proposta de delimitação territorial do Matopiba. Nota técnica 1. EMBRAPA. Grupo de Inteligência Territorial Estratégica (GITE) Available at: https://www.embrapa.br/gite/ publicacoes/ NT1_DelimitacaoMatopiba.pdf. (Accessed 12 January 2017). MMA, 2006. Ministério do Meio Ambiente. Programa Nacional de Conservação e Uso Sustentável do Bioma Cerrado. Programa Cerrado Sustentável. Núcleo dos Biomas Cerrado e Pantanal, Brasília: Distrito Federal. 56 p. Moran, P.A.P., 1948. The interpretation of statistical maps. J. R. Stat. Soc. 10, 243–251. Mueller, C.C., 2009. Agricultural, agrarian and environmental policy formation under Lula: the role of policy networks. In: Love, J., Werner, B. (Eds.), Brazil Under Lula. Economy, Politics and Society Under the Worker-President. Palgrave Macmillan, New York, pp. 135–150. Mueller, B., Mueller, C., 2016. The political economy of the Brazilian model of agricultural development: institutions versus sectoral policy. Q. Rev. Econ. Financ. 62, 12–20. OCDE/FAO, 2015. Food and agriculture organization of the United nations. OCDE-FAO Perspectivas Agrícolas 2015-2024. OECD Publishing, Paris Available at: https://doi. org/10.1787/agr_outlook-2015-es Accessed: 2 Feb. 2017. Paludzyszyn Filho, E., 1995. A cultura da soja no Maranhão. Londrina: Embrapa-CNPSo. 34 p. (Documentos, 84.). . Paludzyszyn Filho, F., Kiihl, R.A.S., Almeida, L.A., 1993. Desenvolvimento de cultivares de soja na região Norte e Nordeste do Brasil. In: Arantes, N.E., Souza, P.I.M. (Eds.), Cultura da soja nos Cerrados. Piracicaba: Potáfos, pp. 255–263. Portal Brasil, 2015. Brasília: Portal Brasil; “Matopiba” se consolida como nova fronteira agrícola do País. Available at: http://www.brasil.gov.br/economia-e-emprego/2015/ 10/matopiba-se-consolida-como-nova-fronteira-agricola-do-pais, Accessed: 25 Mar. 2017. Santos Filho, J.I.S., Talamini, D.J.D., Scheuermann, G.N., Bertol, T.M., 2016. Potencial do Matopiba na produção de aves e suínos. Revista de Política Agrícola 25 (2), 90–102. Silva, F.B., Santos, J.R.N., Feitosa, F.E.C.S., Silva, I.D.C., Araújo, M.L.S., Guterres, C.E., Santos, J.S., Ribeiro, C.V., Bezerra, D.S., Neres, R.L., 2016. Evidências de mudanças climáticas na região de transição Amazônia-Cerrado no estado do Maranhão. Rev. Bras. Meteorol. 31, 330–336. Spagnolo, T.F.O., 2011. Análise da dinâmica espacial da expansão agrícola no Oeste baiano entre 1984 e 2008: estudo de caso do município de São Desidério-BA. Dissertação (Mestrado) – Brasília. Universidade de Brasília 69 f.
Fig. 7. Number of municipalities with spatial groupings of (a) production and (b) yield of soybean in the Matopiba region for HH, HL, LL, and LH clusters (1990–2015).
References Alston, L., Melo, M., Mueller, B., Pereira, C., 2016. Brazil in Transition: Beliefs, Leadership and Institutional Change. Princeton University Press, Princeton. Alves, E.R.A., Pastore, A.C., 1978. Import substitution and implicit taxation of agriculture in Brazil. Am. J. Agric. Econ. 60 (5), 865–871. An, K., Ouyang, Y., 2016. Robust grain supply chain design considering post-harvest loss and harvest timing equilibrium. Transp. Res. Part E Logist. Transp. Rev. 88, 110–128. Andersen, L.E., Granger, C.W.J., Reis, E.J., Weinhold, D., Wunder, S., 2002. The Dynamics of Deforestation and Economic Growth in the Brazilian Amazon. Cambridge University Press, Cambridge. Anderson, M.C., Zolin, C.A., Sentelhas, P.C., Hain, C.R., Semmens, K., Yilmaz, M.T., Gao, F., Otkin, J.A., Tetrault, R., 2016. The evaporative stress index as an indicator of agricultural drought in Brazil: an assessment based on crop yield impacts. Remote Sens. Environ. 174, 82–99. Anselin, L., 1995. Local indicators of spatial association - LISA. Geogr. Anal. 27 (2), 93–115. Anselin, L., 2005. Exploring Spatial Data With GeoDa: a Workbook. Illinois: Urbanachampaign. 244 p.. . Bolfe, E., Victória, D., Contini, E., Bayma-Silva, G., Spinelli-Araújo, L., Gomes, D., 2016. MATOPIBA em crescimento agrícola: aspectos territoriais e socioeconômicos. Revista de Política Agrícola 25 (4), 38–62. Brown, J.C., Koeppe, M., Coles, B., Price, K.P., 2005. Soybean production and conversion of tropical forest in the Brazilian Amazon: the case of Vilhena, Rondônia. Ambio A J. Hum. Environ. 34, 462–469. Câmara, G., Carvalho, M.S., Cruz, O.G., Correa, V., 2002. Análise espacial de áreas. In: Fuks, S.D., Carvalho, M.S., Câmara, G., Monteiro, A.M.V. (Eds.), Análise espacial de dados geográficos. São José dos Campos: INPE, 44 p. Campos, M.C., 2010. A Embrapa Soja em Londrina – PR: a pesquisa agrícola de um país moderno (Ph.D. dissertation). Universidade Federal de Santa Catarina, Florianópolis 123 p. Cho, S., Newman, D.H., 2005. Spatial analysis of rural land development. For. Policy Econ. 7 (5), 732–744. Clemente, A.M., Carvalho Júnior, O.A., Guimarães, R.F., McManus, C., Turazi, C.M.V., Hermuche, P.M., 2017. Spatial-temporal patterns of bean crop in Brazil over the period 1990–2013. ISPRS Int. J. Geoinf. 6 (107), 18. Dall’Agnol, A., 2008. Soja: o fenômeno brasileiro. EMBRAPA, Londrina Available at: https://www.agrolink.com.br/colunistas/coluna/soja–o-fenomeno-brasileiro_383 966.html. Accessed: 10 Jul. 2018. Delgado, G.C., 1985. Capital financeiro e agricultura no Brasil. Unicamp/Cone, Campinas 240 p.
66
Land Use Policy 80 (2019) 57–67
M.L.S.d. Araújo et al.
spatial and temporal variation in agricultural output elasticities in Turkey. Agric. Econ. 45, 279–290. Yuan, Y., Cave, M., Zhang, C., 2017. Using local Moran’s I to identify contamination hotspots of rare earth elements in urban soils of London. Appl. Geochem. 88 (Part B), 167–178. Zhang, T., Lin, G., 2016. On Moran’s I coefficient under heterogeneity. Comput. Stat. Data Anal. 95, 83–94.
Studte, M., 2008. A expansão da fronteira agrícola e a agricultura familiar no Cerrado Maranhense. Dissertação (Mestrado) – São Paulo. Universidade de São Paulo; Berlin: Universidade Técnica de Berlin 55 f. USDA United States Department of Agriculture, 2017. World Agricultural Production. Available at: https://apps.fas.usda.gov/psdonline/circulars/ production.pdf. (Accessed 28 March 2017). . Yu, T.H.E., Cho, S., Koç, A.A., Boluk, G., Kim, S.G., Lambert, D.M., 2014. Evaluating
67