Table 1. Summary of direct land-use changes associated with industrial-scale feedstock expansion in the case study sites.

Site Year operations began Land or concession area secured
(ha)
Area developed
(ha)
Area deforested
(ha)
Feedstock expansion causing forest loss
(%)
Forest type
Brazil: Mato Grosso State§ Various Various 5,075,079
(by 2007)
540,000
(2001–2004)
13–18¦ Dry forest (cerrado)
Ghana: Pru District, Brong Ahafo 2008 14,500 780
(by 2010)
379 forest, 240 fallow
(by 2010)
47
(77 including fallow)
Dry forest (forest-savannah transition zone)
Indonesia: Kubu Raya, West Kalimantan 1994 13,605 5,350 4,949
(as of May 2009)
94 Secondary peat swamp forest
Indonesia: Manokwari, West Papua 1982 12,049 10,207 5,260
(as of Jun/Aug 2006)
96 Primary humid tropical rain forest
Indonesia: Boven Digoel, Papua 1998 34,000 18,804 20,709
(as of Dec 2008)
99 Primary humid tropical rain forest
Malaysia: Sabah 1987 6,861# 6,861 5,329 75 Scrub forest and logged forest
Mexico: Yucatán 2007 12,000
(2009)
2,350
(2009)
Unavailable†† Secondary dry forest (acahual)
Land-use change data for Mato Grosso encompass two of the research sites: Sorriso and northern Mato Grosso (Guarantã do Norte/Alta Floresta). Similar data are unavailable for the Santarem site.
With the exception of Ghana, only a portion of this area is from the biofuel industry; much of oil palm and soybean production is destined for food and feed markets.
§This case is unique in capturing trends within an entire state rather than a specific plantation investment.
¦This is the total due to soy expansion, but the authors estimate between 0.8 and 5.9% to be attributable specifically to the biodiesel component, depending on the food-fuel allocation approach used (Lima et al. 2011).
These figures correspond with oil palm-induced deforestation; total deforestation was found to be 7100, 6833, and 36,666 ha, respectively, in the three sites (Obidzinski et al. unpublished manuscript).
#Area figures are for Sapi 1 and Sapi 2 estates; total landholdings of PPB Wilmar are about three times this.
††This analysis was attempted but is left unreported because of the high level of patchiness of the displaced vegetation, the difficulties of clearly differentiating vegetation at different stages of regeneration, and the uncertainties therefore introduced in producing an unambiguous land cover classification.

Sources: Achten and Verchot 2011, German et al. 2011a, Lima et al. 2011, Schoneveld et al. 2011, Skutsch et al. 2011, Dayang Norwana et al. in press, Obidzinski et al. unpublished manuscript.