The Expansion of Agriculture and Its Effects of Evapotranspiration in Brazil's Newest Agricultural Frontier

Although approximately half of the Brazilian Cerrado has been deforested, mostly in the name of export-oriented mechanized agriculture, the effects of this deforestation on the regional hydrologic cycle have been unexamined. Will it affect precipitation in the region? This question is important as future agricultural production and ecosystem sustainability rely on a stable precipitation regime. Thus, understanding the nature of these feedbacks in a complex, rapidly developing landscape is essential for determining the consequences of these land-use changes and engendering effective environmental policy.

We focus our analysis on Brazil's newest–and potentially final–agricultural frontier, the “Mapitoba” region, so named because it spans the states of Maranhão (MA), Piauí (PI), Tocantins (TO) and Bahia (BA). We use Moderate Resolution Imaging Spectroradiomter (MODIS) Enhanced Vegetation Index (EVI) data (MOD13Q1) to map land use change, specifically row-crop agriculture, within Mapitoba between 2003 and 2013. We do so with 87% accuracy. During our study period, row-crop agriculture more than doubled, expanding from 1.2 to 2.5 million ha, and 75% of this row-crop agriculture originated from natural Cerrado vegetation.

We use MODIS evapotranspiration (ET) data and quantify the impacts of agricultural conversion on evapotranspiration (ET) in Mapitoba. Monthly ET was, on average, 26 mm mo-1 higher over areas of natural vegetation for all months except January, February and March. These 26 mm mo-1 equate to a 22% (April) - 77% (September) difference in ET rates depending on the month. In January and February, during the height of agricultural (soy) growing season, evapotranspiration occurs at similar (~100%) or slightly higher (~107%) rates over row-crop agriculture than natural vegetation. The annual difference in average ET between row-crop agriculture and natural vegetation decreases over time: during the 2003 growing season, an average square kilometer of row-crop agriculture transpired only 64% of the total water the same area of Cerrado transpired; by the 2013 growing season, this number had increased to almost 83%. One potential explanation for this is the increase in double-cropping area, as double-cropping, unlike single-cropping, was significantly and positively correlated with ET during the onset of the dry season in April, May and June. Adding a second crop to the rotation extends the growing season and increases the period of high ET relative to native vegetation. Although annually the difference in dry season ET is being compensated for by row-crop agriculture during the wet season, the differences in ET during the dry season where single-cropping dominates may be more important. If the dry season extends, farmers may be unable to intensify agricultural production through double-cropping.

The change in dry season (June-August) Cerrado ET is directly and linearly related (R2 = 0.81) to the cumulative amount of land converted to row-crop agriculture. For every new 1000 ha of row-crops, 1.7 million m3 less water is recycled back to the atmosphere. If row-crop were to expand onto the 8.2 million ha of remnant natural Cerrado deemed suitable for soy agriculture in Mapitoba, we estimate an additional reduction of 14 km3 of water each year during these three months alone. Air masses traveling over the Cerrado recycle evapotranspired water into the Cerrado itself and the Amazon Basin as precipitation (Spracken et al. 2012). Feedbacks between land use changes and climate have the potential to reduce precipitation and ultimately threaten both the health of the Cerrado and Amazon biomes and sustainability of agricultural production in the region.

Unlike within the Amazon, there is a lack of market pressure to prevent Cerrado deforestation, relatively lower land rent, and ease of clearing, all of which suggest that agricultural expansion will likely persist if no changes in governance occur. We demonstrate that the 450,000 km2 that comprise the Mapitoba region has experienced land-use changes that are directly affecting the region's water balance, and will continue to do so should governance in the region not change.