Comparing the trends and strength of determinants to deforestation in the Brazilian Amazon in consideration of biofuel policies in Brazil and the United States

Recent assessments of the lifecycle greenhouse gas (GHG) emissions associated with increased production of liquid biofuels in the U.S. have found that a relatively large share of these emissions is related to land-use change in other countries, with Brazil playing a prominent role. However, most of the existing analyses including EPA's RFS assumptions about indirect land-use change (ILUC) do not reflect recent data on deforestation rates in Brazil, which have been declining. There seems to be evidence that agricultural producers in Brazil are intensifying crop and livestock production and incorporating new land at lower rates than in the recent past. Additionally, the competition for forestry areas, and in particular between agriculture, pastures, and planted forests is poorly understood, and not explicitly taken into account in previous assessments in several modeling exercises. This paper provides a review of several of the major factors that will determine the need to incorporate additional land to production in response to a demand increase, for example as a result of biofuel policies. This additional land that may need to be brought into production is critical as it will affect significantly the environmental credentials and in particular carbon footprints of different biofuels. Among the factors reviewed are the potential for yield intensification in response to higher returns (intensification effects), and the limited existing evidence in yield drags as areas are incorporate to crop production (extensification effects). Scenario analysis using an augmented version of the CARD/FAPRI agricultural modeling system (augmented to include planted forests in Brazil) seem to provide evidence, that intensification of crops and livestock production in countries like Brazil, and of competition with planted forests reduces the pressure for deforestation of natural areas. We also highlight that the explicit modeling of planted forests as a user of land, allows for the inclusion of the competition of this activity (and its resistance to give away area) with the more traditionally modeled crops and pastures.